From 2b90141c42574cedf48aa285b24e7d53b142ede8 Mon Sep 17 00:00:00 2001 From: serxa Date: Tue, 28 Feb 2023 13:17:31 +0000 Subject: [PATCH 001/433] Integrate DiskObjectStorage with IO Scheduler --- src/Common/CurrentThread.h | 8 ++++ src/Disks/DiskLocal.cpp | 4 +- .../registerDiskAzureBlobStorage.cpp | 4 +- .../ObjectStorages/DiskObjectStorage.cpp | 47 +++++++++++++++++-- src/Disks/ObjectStorages/DiskObjectStorage.h | 11 ++++- .../ObjectStorages/HDFS/registerDiskHDFS.cpp | 5 +- .../ObjectStorages/S3/registerDiskS3.cpp | 4 +- .../Web/registerDiskWebServer.cpp | 4 +- src/IO/IResourceManager.h | 2 +- src/Interpreters/Context.cpp | 4 +- src/Interpreters/Context.h | 4 ++ 11 files changed, 85 insertions(+), 12 deletions(-) diff --git a/src/Common/CurrentThread.h b/src/Common/CurrentThread.h index f36b92e319d..1571f52558c 100644 --- a/src/Common/CurrentThread.h +++ b/src/Common/CurrentThread.h @@ -93,6 +93,14 @@ public: return current_thread->getQueryId(); } + /// Returns attached query context + static ContextPtr getQueryContext() + { + if (isInitialized()) + return get().getQueryContext(); + return {}; + } + /// Non-master threads call this method in destructor automatically static void detachQuery(); static void detachQueryIfNotDetached(); diff --git a/src/Disks/DiskLocal.cpp b/src/Disks/DiskLocal.cpp index a3b7e413014..088e44357b3 100644 --- a/src/Disks/DiskLocal.cpp +++ b/src/Disks/DiskLocal.cpp @@ -613,7 +613,9 @@ DiskObjectStoragePtr DiskLocal::createDiskObjectStorage() metadata_storage, object_storage, false, - /* threadpool_size */16 + /* threadpool_size */ 16, + /* read_resource_name */ "", + /* write_resource_name */ "" ); } diff --git a/src/Disks/ObjectStorages/AzureBlobStorage/registerDiskAzureBlobStorage.cpp b/src/Disks/ObjectStorages/AzureBlobStorage/registerDiskAzureBlobStorage.cpp index 562b2b2fec0..e0c0a525d49 100644 --- a/src/Disks/ObjectStorages/AzureBlobStorage/registerDiskAzureBlobStorage.cpp +++ b/src/Disks/ObjectStorages/AzureBlobStorage/registerDiskAzureBlobStorage.cpp @@ -43,7 +43,9 @@ void registerDiskAzureBlobStorage(DiskFactory & factory, bool global_skip_access std::move(metadata_storage), std::move(azure_object_storage), send_metadata, - copy_thread_pool_size + copy_thread_pool_size, + config.getString(config_prefix + ".read_resource", ""), + config.getString(config_prefix + ".write_resource", "") ); bool skip_access_check = global_skip_access_check || config.getBool(config_prefix + ".skip_access_check", false); diff --git a/src/Disks/ObjectStorages/DiskObjectStorage.cpp b/src/Disks/ObjectStorages/DiskObjectStorage.cpp index d55b1c91c07..825ed03b1ba 100644 --- a/src/Disks/ObjectStorages/DiskObjectStorage.cpp +++ b/src/Disks/ObjectStorages/DiskObjectStorage.cpp @@ -6,6 +6,7 @@ #include #include #include +#include #include #include #include @@ -109,7 +110,9 @@ DiskObjectStorage::DiskObjectStorage( MetadataStoragePtr metadata_storage_, ObjectStoragePtr object_storage_, bool send_metadata_, - uint64_t thread_pool_size_) + uint64_t thread_pool_size_, + const String & read_resource_name_, + const String & write_resource_name_) : IDisk(name_, getAsyncExecutor(log_name, thread_pool_size_)) , object_storage_root_path(object_storage_root_path_) , log (&Poco::Logger::get("DiskObjectStorage(" + log_name + ")")) @@ -117,6 +120,8 @@ DiskObjectStorage::DiskObjectStorage( , object_storage(std::move(object_storage_)) , send_metadata(send_metadata_) , threadpool_size(thread_pool_size_) + , read_resource_name(read_resource_name_) + , write_resource_name(write_resource_name_) , metadata_helper(std::make_unique(this, ReadSettings{})) {} @@ -517,7 +522,9 @@ DiskObjectStoragePtr DiskObjectStorage::createDiskObjectStorage() metadata_storage, object_storage, send_metadata, - threadpool_size); + threadpool_size, + getReadResourceName(), + getWriteResourceName()); } void DiskObjectStorage::wrapWithCache(FileCachePtr cache, const FileCacheSettings & cache_settings, const String & layer_name) @@ -546,6 +553,32 @@ NameSet DiskObjectStorage::getCacheLayersNames() const return cache_layers; } +template +static inline Settings updateResourceLink(const Settings & settings, const String & resource_name) +{ + if (resource_name.empty()) + return settings; + if (auto query_context = CurrentThread::getQueryContext()) + { + Settings result(settings); + result.resource_link = query_context->getClassifier()->get(resource_name); + return result; + } + return settings; +} + +String DiskObjectStorage::getReadResourceName() const +{ + std::unique_lock lock(resource_mutex); + return read_resource_name; +} + +String DiskObjectStorage::getWriteResourceName() const +{ + std::unique_lock lock(resource_mutex); + return write_resource_name; +} + std::unique_ptr DiskObjectStorage::readFile( const String & path, const ReadSettings & settings, @@ -554,7 +587,7 @@ std::unique_ptr DiskObjectStorage::readFile( { return object_storage->readObjects( metadata_storage->getStorageObjects(path), - object_storage->getAdjustedSettingsFromMetadataFile(settings, path), + object_storage->getAdjustedSettingsFromMetadataFile(updateResourceLink(settings, getReadResourceName()), path), read_hint, file_size); } @@ -572,7 +605,7 @@ std::unique_ptr DiskObjectStorage::writeFile( path, buf_size, mode, - object_storage->getAdjustedSettingsFromMetadataFile(settings, path)); + object_storage->getAdjustedSettingsFromMetadataFile(updateResourceLink(settings, getWriteResourceName()), path)); return result; } @@ -585,6 +618,12 @@ void DiskObjectStorage::applyNewSettings( if (AsyncThreadPoolExecutor * exec = dynamic_cast(&getExecutor())) exec->setMaxThreads(config.getInt(config_prefix + ".thread_pool_size", 16)); + + std::unique_lock lock(resource_mutex); + if (String new_read_resource_name = config.getString(config_prefix + ".read_resource", ""); new_read_resource_name != read_resource_name) + read_resource_name = new_read_resource_name; + if (String new_write_resource_name = config.getString(config_prefix + ".write_resource", ""); new_write_resource_name != write_resource_name) + write_resource_name = new_write_resource_name; } void DiskObjectStorage::restoreMetadataIfNeeded( diff --git a/src/Disks/ObjectStorages/DiskObjectStorage.h b/src/Disks/ObjectStorages/DiskObjectStorage.h index a24acc270c0..f5b9fdf54d3 100644 --- a/src/Disks/ObjectStorages/DiskObjectStorage.h +++ b/src/Disks/ObjectStorages/DiskObjectStorage.h @@ -34,7 +34,9 @@ public: MetadataStoragePtr metadata_storage_, ObjectStoragePtr object_storage_, bool send_metadata_, - uint64_t thread_pool_size_); + uint64_t thread_pool_size_, + const String & read_resource_name_, + const String & write_resource_name_); /// Create fake transaction DiskTransactionPtr createTransaction() override; @@ -211,6 +213,9 @@ private: /// execution. DiskTransactionPtr createObjectStorageTransaction(); + String getReadResourceName() const; + String getWriteResourceName() const; + const String object_storage_root_path; Poco::Logger * log; @@ -226,6 +231,10 @@ private: const bool send_metadata; size_t threadpool_size; + mutable std::mutex resource_mutex; + String read_resource_name; + String write_resource_name; + std::unique_ptr metadata_helper; }; diff --git a/src/Disks/ObjectStorages/HDFS/registerDiskHDFS.cpp b/src/Disks/ObjectStorages/HDFS/registerDiskHDFS.cpp index 693b966caf2..b7f703b278d 100644 --- a/src/Disks/ObjectStorages/HDFS/registerDiskHDFS.cpp +++ b/src/Disks/ObjectStorages/HDFS/registerDiskHDFS.cpp @@ -54,7 +54,10 @@ void registerDiskHDFS(DiskFactory & factory, bool global_skip_access_check) std::move(metadata_storage), std::move(hdfs_storage), /* send_metadata = */ false, - copy_thread_pool_size); + copy_thread_pool_size, + config.getString(config_prefix + ".read_resource", ""), + config.getString(config_prefix + ".write_resource", "") + ); disk->startup(context, skip_access_check); return disk; diff --git a/src/Disks/ObjectStorages/S3/registerDiskS3.cpp b/src/Disks/ObjectStorages/S3/registerDiskS3.cpp index 1c192a0d89c..bce1e640a7c 100644 --- a/src/Disks/ObjectStorages/S3/registerDiskS3.cpp +++ b/src/Disks/ObjectStorages/S3/registerDiskS3.cpp @@ -163,7 +163,9 @@ void registerDiskS3(DiskFactory & factory, bool global_skip_access_check) std::move(metadata_storage), std::move(s3_storage), send_metadata, - copy_thread_pool_size); + copy_thread_pool_size, + config.getString(config_prefix + ".read_resource", ""), + config.getString(config_prefix + ".write_resource", "")); s3disk->startup(context, skip_access_check); diff --git a/src/Disks/ObjectStorages/Web/registerDiskWebServer.cpp b/src/Disks/ObjectStorages/Web/registerDiskWebServer.cpp index 8a54de81815..19005727fa7 100644 --- a/src/Disks/ObjectStorages/Web/registerDiskWebServer.cpp +++ b/src/Disks/ObjectStorages/Web/registerDiskWebServer.cpp @@ -53,7 +53,9 @@ void registerDiskWebServer(DiskFactory & factory, bool global_skip_access_check) metadata_storage, object_storage, /* send_metadata */false, - /* threadpool_size */16); + /* threadpool_size */16, + config.getString(config_prefix + ".read_resource", ""), + config.getString(config_prefix + ".write_resource", "")); disk->startup(context, skip_access_check); return disk; }; diff --git a/src/IO/IResourceManager.h b/src/IO/IResourceManager.h index f084a903cb1..ff372698ed5 100644 --- a/src/IO/IResourceManager.h +++ b/src/IO/IResourceManager.h @@ -23,7 +23,7 @@ class IClassifier : private boost::noncopyable public: virtual ~IClassifier() {} - /// Returns ResouceLink that should be used to access resource. + /// Returns ResourceLink that should be used to access resource. /// Returned link is valid until classifier destruction. virtual ResourceLink get(const String & resource_name) = 0; }; diff --git a/src/Interpreters/Context.cpp b/src/Interpreters/Context.cpp index 33505e41789..72a472fcb4a 100644 --- a/src/Interpreters/Context.cpp +++ b/src/Interpreters/Context.cpp @@ -1165,7 +1165,9 @@ ResourceManagerPtr Context::getResourceManager() const ClassifierPtr Context::getClassifier() const { auto lock = getLock(); - return getResourceManager()->acquire(getSettingsRef().workload); + if (!classifier) + classifier = getResourceManager()->acquire(getSettingsRef().workload); + return classifier; } diff --git a/src/Interpreters/Context.h b/src/Interpreters/Context.h index 19bb6868331..9ea8d253439 100644 --- a/src/Interpreters/Context.h +++ b/src/Interpreters/Context.h @@ -402,6 +402,10 @@ private: /// Temporary data for query execution accounting. TemporaryDataOnDiskScopePtr temp_data_on_disk; + /// Resource classifier for a query, holds smart pointers required for ResourceLink + /// NOTE: all resource links became invalid after `classifier` destruction + mutable ClassifierPtr classifier; + public: /// Some counters for current query execution. /// Most of them are workarounds and should be removed in the future. From 21b6ccc677ba040acc62fc0c63cadbe53634ce3f Mon Sep 17 00:00:00 2001 From: serxa Date: Thu, 2 Mar 2023 15:35:28 +0000 Subject: [PATCH 002/433] fix comment --- src/IO/ResourceRequest.h | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/src/IO/ResourceRequest.h b/src/IO/ResourceRequest.h index 989349148cf..3d2230746f9 100644 --- a/src/IO/ResourceRequest.h +++ b/src/IO/ResourceRequest.h @@ -45,8 +45,7 @@ class ResourceRequest { public: /// Cost of request execution; should be filled before request enqueueing. - /// NOTE: If cost is not known in advance, credit model can be used: - /// NOTE: for the first request use 1 and + /// NOTE: If cost is not known in advance, ResourceBudget should be used (note that every ISchedulerQueue has it) ResourceCost cost; /// Request outcome From af66d44106a40bd48d5320098de416fa40fe6455 Mon Sep 17 00:00:00 2001 From: alexX512 Date: Mon, 10 Apr 2023 13:51:39 +0000 Subject: [PATCH 003/433] Add settings for real-time updates during query execution --- src/Client/ClientBase.cpp | 32 +++++++++++++- src/Client/ClientBase.h | 6 +++ src/Core/BlockInfo.h | 8 +++- src/Core/Settings.h | 3 ++ src/Processors/Chunk.cpp | 17 ++++---- src/Processors/Chunk.h | 18 ++++++-- .../PullingAsyncPipelineExecutor.cpp | 1 + src/Processors/Formats/IOutputFormat.cpp | 20 ++++++++- src/Processors/Formats/IOutputFormat.h | 2 + .../Formats/Impl/PrettyBlockOutputFormat.cpp | 6 ++- .../Impl/PrettyCompactBlockOutputFormat.cpp | 3 +- src/Processors/LimitTransform.cpp | 5 +++ src/Processors/QueryPlan/SortingStep.cpp | 6 ++- src/Processors/QueryPlan/SortingStep.h | 2 + .../Transforms/LimitsCheckingTransform.cpp | 2 +- .../Transforms/MergeSortingTransform.cpp | 42 ++++++++++++++++++- .../Transforms/MergeSortingTransform.h | 11 ++++- 17 files changed, 162 insertions(+), 22 deletions(-) diff --git a/src/Client/ClientBase.cpp b/src/Client/ClientBase.cpp index 6a2961d69da..aa413c4d749 100644 --- a/src/Client/ClientBase.cpp +++ b/src/Client/ClientBase.cpp @@ -440,7 +440,9 @@ void ClientBase::onData(Block & block, ASTPtr parsed_query) if (!block) return; - processed_rows += block.rows(); + if (!block.info.has_partial_result) + processed_rows += block.rows(); + /// Even if all blocks are empty, we still need to initialize the output stream to write empty resultset. initOutputFormat(block, parsed_query); @@ -450,14 +452,35 @@ void ClientBase::onData(Block & block, ASTPtr parsed_query) if (block.rows() == 0 || (query_fuzzer_runs != 0 && processed_rows >= 100)) return; + if (received_first_full_result && block.info.has_partial_result) + throw Exception(ErrorCodes::LOGICAL_ERROR, "Server shouldn't send partial results after the first block with a non-empty full result"); + + if (!is_interactive && block.info.has_partial_result) + return; + + bool first_full_result = false; + if (!received_first_full_result && !block.info.has_partial_result) + { + received_first_full_result = true; + first_full_result = true; + } + /// If results are written INTO OUTFILE, we can avoid clearing progress to avoid flicker. if (need_render_progress && tty_buf && (!select_into_file || select_into_file_and_stdout)) progress_indication.clearProgressOutput(*tty_buf); + if (is_interactive && first_full_result && has_partial_result_setting) + std::cout << "Full result:" << std::endl; + try { + /// Clear previous partial results to write new partial results if needed + if (!received_first_full_result && written_first_block) + output_format->clearLastLines(prev_block_rows + 2); + output_format->write(materializeBlock(block)); written_first_block = true; + prev_block_rows = block.rows(); } catch (const Exception &) { @@ -865,6 +888,10 @@ void ClientBase::processOrdinaryQuery(const String & query_to_execute, ASTPtr pa const auto & settings = global_context->getSettingsRef(); const Int32 signals_before_stop = settings.stop_reading_on_first_cancel ? 2 : 1; + has_partial_result_setting = settings.partial_result_update_duration_ms.totalMilliseconds() > 0; + + if (is_interactive && has_partial_result_setting) + std::cout << "Partial result:" << std::endl; int retries_left = 10; while (retries_left) @@ -1614,6 +1641,9 @@ void ClientBase::processParsedSingleQuery(const String & full_query, const Strin } processed_rows = 0; + prev_block_rows = 0; + has_partial_result_setting = false; + received_first_full_result = false; written_first_block = false; progress_indication.resetProgress(); profile_events.watch.restart(); diff --git a/src/Client/ClientBase.h b/src/Client/ClientBase.h index faf3fa8653a..1a785cc390f 100644 --- a/src/Client/ClientBase.h +++ b/src/Client/ClientBase.h @@ -259,6 +259,12 @@ protected: size_t processed_rows = 0; /// How many rows have been read or written. bool print_num_processed_rows = false; /// Whether to print the number of processed rows at + /// If `received_first_full_result` set to true then + /// client has already received all updates with a partial result + bool received_first_full_result = false; + bool has_partial_result_setting = false; + size_t prev_block_rows = 0; /// How many rows were in the previously received block of data + bool print_stack_trace = false; /// The last exception that was received from the server. Is used for the /// return code in batch mode. diff --git a/src/Core/BlockInfo.h b/src/Core/BlockInfo.h index d431303ca39..a0357b292cd 100644 --- a/src/Core/BlockInfo.h +++ b/src/Core/BlockInfo.h @@ -28,9 +28,15 @@ struct BlockInfo * Otherwise -1. */ + /** has_partial_result: + * If user wants to receive updates containing partial results during query execution, + * then data will be sent with has_partial_result flag set to true. + */ + #define APPLY_FOR_BLOCK_INFO_FIELDS(M) \ M(bool, is_overflows, false, 1) \ - M(Int32, bucket_num, -1, 2) + M(Int32, bucket_num, -1, 2) \ + M(bool, has_partial_result, false, 3) #define DECLARE_FIELD(TYPE, NAME, DEFAULT, FIELD_NUM) \ TYPE NAME = DEFAULT; diff --git a/src/Core/Settings.h b/src/Core/Settings.h index eaed800ae88..5099803c60c 100644 --- a/src/Core/Settings.h +++ b/src/Core/Settings.h @@ -281,6 +281,9 @@ class IColumn; M(Bool, final, false, "Query with the FINAL modifier by default. If the engine does not support final, it does not have any effect. On queries with multiple tables final is applied only on those that support it. It also works on distributed tables", 0) \ \ M(Bool, stop_reading_on_first_cancel, false, "Allows query to return a partial result after cancel.", 0) \ + \ + M(Milliseconds, partial_result_update_duration_ms, 0, "Duration of time in milliseconds between real-time updates of result table sent to the client during query execution.", 0) \ + M(UInt64, max_rows_in_partial_result, 10, "Max rows displayed to user after each real-time update of output table during query execution.", 0) \ /** Settings for testing hedged requests */ \ M(Milliseconds, sleep_in_send_tables_status_ms, 0, "Time to sleep in sending tables status response in TCPHandler", 0) \ M(Milliseconds, sleep_in_send_data_ms, 0, "Time to sleep in sending data in TCPHandler", 0) \ diff --git a/src/Processors/Chunk.cpp b/src/Processors/Chunk.cpp index bbfa1683cf6..040beceeb7d 100644 --- a/src/Processors/Chunk.cpp +++ b/src/Processors/Chunk.cpp @@ -12,13 +12,14 @@ namespace ErrorCodes extern const int POSITION_OUT_OF_BOUND; } -Chunk::Chunk(DB::Columns columns_, UInt64 num_rows_) : columns(std::move(columns_)), num_rows(num_rows_) +Chunk::Chunk(DB::Columns columns_, UInt64 num_rows_, bool has_partial_result_) + : columns(std::move(columns_)), num_rows(num_rows_), has_partial_result(has_partial_result_) { checkNumRowsIsConsistent(); } -Chunk::Chunk(Columns columns_, UInt64 num_rows_, ChunkInfoPtr chunk_info_) - : columns(std::move(columns_)), num_rows(num_rows_), chunk_info(std::move(chunk_info_)) +Chunk::Chunk(Columns columns_, UInt64 num_rows_, ChunkInfoPtr chunk_info_, bool has_partial_result_) + : columns(std::move(columns_)), num_rows(num_rows_), chunk_info(std::move(chunk_info_)), has_partial_result(has_partial_result_) { checkNumRowsIsConsistent(); } @@ -33,21 +34,21 @@ static Columns unmuteColumns(MutableColumns && mut_columns) return columns; } -Chunk::Chunk(MutableColumns columns_, UInt64 num_rows_) - : columns(unmuteColumns(std::move(columns_))), num_rows(num_rows_) +Chunk::Chunk(MutableColumns columns_, UInt64 num_rows_, bool has_partial_result_) + : columns(unmuteColumns(std::move(columns_))), num_rows(num_rows_), has_partial_result(has_partial_result_) { checkNumRowsIsConsistent(); } -Chunk::Chunk(MutableColumns columns_, UInt64 num_rows_, ChunkInfoPtr chunk_info_) - : columns(unmuteColumns(std::move(columns_))), num_rows(num_rows_), chunk_info(std::move(chunk_info_)) +Chunk::Chunk(MutableColumns columns_, UInt64 num_rows_, ChunkInfoPtr chunk_info_, bool has_partial_result_) + : columns(unmuteColumns(std::move(columns_))), num_rows(num_rows_), chunk_info(std::move(chunk_info_)), has_partial_result(has_partial_result_) { checkNumRowsIsConsistent(); } Chunk Chunk::clone() const { - return Chunk(getColumns(), getNumRows(), chunk_info); + return Chunk(getColumns(), getNumRows(), chunk_info, has_partial_result); } void Chunk::setColumns(Columns columns_, UInt64 num_rows_) diff --git a/src/Processors/Chunk.h b/src/Processors/Chunk.h index 15d91431b68..f9cc4a4e087 100644 --- a/src/Processors/Chunk.h +++ b/src/Processors/Chunk.h @@ -38,14 +38,16 @@ public: : columns(std::move(other.columns)) , num_rows(other.num_rows) , chunk_info(std::move(other.chunk_info)) + , has_partial_result(other.has_partial_result) { other.num_rows = 0; + other.has_partial_result = false; } - Chunk(Columns columns_, UInt64 num_rows_); - Chunk(Columns columns_, UInt64 num_rows_, ChunkInfoPtr chunk_info_); - Chunk(MutableColumns columns_, UInt64 num_rows_); - Chunk(MutableColumns columns_, UInt64 num_rows_, ChunkInfoPtr chunk_info_); + Chunk(Columns columns_, UInt64 num_rows_, bool has_partial_result_ = false); + Chunk(Columns columns_, UInt64 num_rows_, ChunkInfoPtr chunk_info_, bool has_partial_result_ = false); + Chunk(MutableColumns columns_, UInt64 num_rows_, bool has_partial_result_ = false); + Chunk(MutableColumns columns_, UInt64 num_rows_, ChunkInfoPtr chunk_info_, bool has_partial_result_ = false); Chunk & operator=(const Chunk & other) = delete; Chunk & operator=(Chunk && other) noexcept @@ -54,6 +56,8 @@ public: chunk_info = std::move(other.chunk_info); num_rows = other.num_rows; other.num_rows = 0; + has_partial_result = other.has_partial_result; + other.has_partial_result = false; return *this; } @@ -64,6 +68,7 @@ public: columns.swap(other.columns); chunk_info.swap(other.chunk_info); std::swap(num_rows, other.num_rows); + std::swap(has_partial_result, other.has_partial_result); } void clear() @@ -71,6 +76,7 @@ public: num_rows = 0; columns.clear(); chunk_info.reset(); + has_partial_result = false; } const Columns & getColumns() const { return columns; } @@ -103,10 +109,14 @@ public: void append(const Chunk & chunk); + void changePartialResultStatus(bool has_partial_result_) { has_partial_result = has_partial_result_; } + bool hasPartialResult() const { return has_partial_result; } + private: Columns columns; UInt64 num_rows = 0; ChunkInfoPtr chunk_info; + bool has_partial_result = false; void checkNumRowsIsConsistent(); }; diff --git a/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp b/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp index abe0cd6320b..ad068a01636 100644 --- a/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp +++ b/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp @@ -160,6 +160,7 @@ bool PullingAsyncPipelineExecutor::pull(Block & block, uint64_t milliseconds) } block = lazy_format->getPort(IOutputFormat::PortKind::Main).getHeader().cloneWithColumns(chunk.detachColumns()); + block.info.has_partial_result = chunk.hasPartialResult(); if (auto chunk_info = chunk.getChunkInfo()) { diff --git a/src/Processors/Formats/IOutputFormat.cpp b/src/Processors/Formats/IOutputFormat.cpp index 88a6fb1e92f..cd292d7795e 100644 --- a/src/Processors/Formats/IOutputFormat.cpp +++ b/src/Processors/Formats/IOutputFormat.cpp @@ -113,7 +113,7 @@ void IOutputFormat::flush() void IOutputFormat::write(const Block & block) { writePrefixIfNeeded(); - consume(Chunk(block.getColumns(), block.rows())); + consume(Chunk(block.getColumns(), block.rows(), block.info.has_partial_result)); if (auto_flush) flush(); @@ -130,4 +130,22 @@ void IOutputFormat::finalize() finalized = true; } +void IOutputFormat::clearLastLines(size_t lines_number) +{ + /// http://en.wikipedia.org/wiki/ANSI_escape_code + #define MOVE_TO_PREV_LINE "\033[A" + #define CLEAR_TO_END_OF_LINE "\033[K" + + static const char * clear_prev_line = MOVE_TO_PREV_LINE \ + CLEAR_TO_END_OF_LINE; + + /// Move cursor to the beginning of line + writeCString("\r", out); + + for (size_t line = 0; line < lines_number; ++line) + { + writeCString(clear_prev_line, out); + } +} + } diff --git a/src/Processors/Formats/IOutputFormat.h b/src/Processors/Formats/IOutputFormat.h index 02e91d5b28b..80403c49912 100644 --- a/src/Processors/Formats/IOutputFormat.h +++ b/src/Processors/Formats/IOutputFormat.h @@ -102,6 +102,8 @@ public: } } + void clearLastLines(size_t lines_number); + protected: friend class ParallelFormattingOutputFormat; diff --git a/src/Processors/Formats/Impl/PrettyBlockOutputFormat.cpp b/src/Processors/Formats/Impl/PrettyBlockOutputFormat.cpp index cefe3ee4a98..208b851b7df 100644 --- a/src/Processors/Formats/Impl/PrettyBlockOutputFormat.cpp +++ b/src/Processors/Formats/Impl/PrettyBlockOutputFormat.cpp @@ -140,7 +140,8 @@ void PrettyBlockOutputFormat::write(Chunk chunk, PortKind port_kind) { if (total_rows >= format_settings.pretty.max_rows) { - total_rows += chunk.getNumRows(); + if (!chunk.hasPartialResult()) + total_rows += chunk.getNumRows(); return; } if (mono_block) @@ -321,7 +322,8 @@ void PrettyBlockOutputFormat::writeChunk(const Chunk & chunk, PortKind port_kind } writeString(bottom_separator_s, out); - total_rows += num_rows; + if (!chunk.hasPartialResult()) + total_rows += num_rows; } diff --git a/src/Processors/Formats/Impl/PrettyCompactBlockOutputFormat.cpp b/src/Processors/Formats/Impl/PrettyCompactBlockOutputFormat.cpp index 2ba9ec725e2..bf6e4cb21f1 100644 --- a/src/Processors/Formats/Impl/PrettyCompactBlockOutputFormat.cpp +++ b/src/Processors/Formats/Impl/PrettyCompactBlockOutputFormat.cpp @@ -194,7 +194,8 @@ void PrettyCompactBlockOutputFormat::writeChunk(const Chunk & chunk, PortKind po writeBottom(max_widths); - total_rows += num_rows; + if (!chunk.hasPartialResult()) + total_rows += num_rows; } diff --git a/src/Processors/LimitTransform.cpp b/src/Processors/LimitTransform.cpp index 2feee7e65b1..a6a0e8ab18a 100644 --- a/src/Processors/LimitTransform.cpp +++ b/src/Processors/LimitTransform.cpp @@ -180,6 +180,11 @@ LimitTransform::Status LimitTransform::preparePair(PortsData & data) return Status::NeedData; data.current_chunk = input.pull(true); + if (data.current_chunk.hasPartialResult()) + { + output.push(std::move(data.current_chunk)); + return Status::PortFull; + } auto rows = data.current_chunk.getNumRows(); diff --git a/src/Processors/QueryPlan/SortingStep.cpp b/src/Processors/QueryPlan/SortingStep.cpp index 0ab8e091e05..2643c315785 100644 --- a/src/Processors/QueryPlan/SortingStep.cpp +++ b/src/Processors/QueryPlan/SortingStep.cpp @@ -33,6 +33,8 @@ SortingStep::Settings::Settings(const Context & context) max_bytes_before_external_sort = settings.max_bytes_before_external_sort; tmp_data = context.getTempDataOnDisk(); min_free_disk_space = settings.min_free_disk_space_for_temporary_data; + partial_result_limit = settings.max_rows_in_partial_result; + partial_result_duration_ms = settings.partial_result_update_duration_ms.totalMilliseconds(); } SortingStep::Settings::Settings(size_t max_block_size_) @@ -214,7 +216,9 @@ void SortingStep::mergeSorting( sort_settings.remerge_lowered_memory_bytes_ratio, sort_settings.max_bytes_before_external_sort, std::move(tmp_data_on_disk), - sort_settings.min_free_disk_space); + sort_settings.min_free_disk_space, + sort_settings.partial_result_limit, + sort_settings.partial_result_duration_ms); }); } diff --git a/src/Processors/QueryPlan/SortingStep.h b/src/Processors/QueryPlan/SortingStep.h index eed1f26e110..0c616c7ce23 100644 --- a/src/Processors/QueryPlan/SortingStep.h +++ b/src/Processors/QueryPlan/SortingStep.h @@ -27,6 +27,8 @@ public: size_t max_bytes_before_external_sort = 0; TemporaryDataOnDiskScopePtr tmp_data = nullptr; size_t min_free_disk_space = 0; + UInt64 partial_result_limit = 0; + UInt64 partial_result_duration_ms = 0; explicit Settings(const Context & context); explicit Settings(size_t max_block_size_); diff --git a/src/Processors/Transforms/LimitsCheckingTransform.cpp b/src/Processors/Transforms/LimitsCheckingTransform.cpp index 02d2fef808c..7edab0d417c 100644 --- a/src/Processors/Transforms/LimitsCheckingTransform.cpp +++ b/src/Processors/Transforms/LimitsCheckingTransform.cpp @@ -38,7 +38,7 @@ void LimitsCheckingTransform::transform(Chunk & chunk) return; } - if (chunk) + if (chunk && !chunk.hasPartialResult()) { info.update(chunk); diff --git a/src/Processors/Transforms/MergeSortingTransform.cpp b/src/Processors/Transforms/MergeSortingTransform.cpp index efd9249066c..36de7165344 100644 --- a/src/Processors/Transforms/MergeSortingTransform.cpp +++ b/src/Processors/Transforms/MergeSortingTransform.cpp @@ -85,13 +85,18 @@ MergeSortingTransform::MergeSortingTransform( double remerge_lowered_memory_bytes_ratio_, size_t max_bytes_before_external_sort_, TemporaryDataOnDiskPtr tmp_data_, - size_t min_free_disk_space_) + size_t min_free_disk_space_, + UInt64 partial_result_limit_, + UInt64 partial_result_duration_ms_) : SortingTransform(header, description_, max_merged_block_size_, limit_, increase_sort_description_compile_attempts) , max_bytes_before_remerge(max_bytes_before_remerge_) , remerge_lowered_memory_bytes_ratio(remerge_lowered_memory_bytes_ratio_) , max_bytes_before_external_sort(max_bytes_before_external_sort_) , tmp_data(std::move(tmp_data_)) , min_free_disk_space(min_free_disk_space_) + , partial_result_limit(partial_result_limit_) + , partial_result_duration_ms(partial_result_duration_ms_) + , watch(CLOCK_MONOTONIC) { } @@ -124,6 +129,35 @@ Processors MergeSortingTransform::expandPipeline() return std::move(processors); } +void MergeSortingTransform::updatePartialResult() +{ + /// Sort all input data + remerge(); + /// Add a copy of the first `partial_result_limit` rows to a generated_chunk + /// to send it as a partial result in the next prepare stage + auto generated_columns = chunks[0].cloneEmptyColumns(); + size_t total_rows = 0; + for (const auto & merged_chunk : chunks) + { + size_t rows = std::min(merged_chunk.getNumRows(), partial_result_limit - total_rows); + if (rows == 0) + break; + + for (size_t position = 0; position < generated_columns.size(); ++position) + { + auto column = merged_chunk.getColumns()[position]; + generated_columns[position]->insertRangeFrom(*column, 0, rows); + } + + total_rows += rows; + } + + generated_chunk.setColumns(std::move(generated_columns), total_rows); + generated_chunk.setChunkInfo(chunks[0].getChunkInfo()); + generated_chunk.changePartialResultStatus(true /*is_chunk_partial*/); + enrichChunkWithConstants(generated_chunk); +} + void MergeSortingTransform::consume(Chunk chunk) { /** Algorithm: @@ -159,6 +193,12 @@ void MergeSortingTransform::consume(Chunk chunk) remerge(); } + if (partial_result_duration_ms && partial_result_duration_ms < watch.elapsedMilliseconds() && !chunks.empty()) + { + updatePartialResult(); + watch.restart(); + } + /** If too many of them and if external sorting is enabled, * will merge blocks that we have in memory at this moment and write merged stream to temporary (compressed) file. * NOTE. It's possible to check free space in filesystem. diff --git a/src/Processors/Transforms/MergeSortingTransform.h b/src/Processors/Transforms/MergeSortingTransform.h index c64c93393ce..2b2db6fae16 100644 --- a/src/Processors/Transforms/MergeSortingTransform.h +++ b/src/Processors/Transforms/MergeSortingTransform.h @@ -30,7 +30,9 @@ public: double remerge_lowered_memory_bytes_ratio_, size_t max_bytes_before_external_sort_, TemporaryDataOnDiskPtr tmp_data_, - size_t min_free_disk_space_); + size_t min_free_disk_space_, + UInt64 partial_result_limit_ = 0, + UInt64 partial_result_duration_ms_ = 0); String getName() const override { return "MergeSortingTransform"; } @@ -51,6 +53,11 @@ private: size_t sum_rows_in_blocks = 0; size_t sum_bytes_in_blocks = 0; + UInt64 partial_result_limit; + UInt64 partial_result_duration_ms; + + Stopwatch watch; + Poco::Logger * log = &Poco::Logger::get("MergeSortingTransform"); /// If remerge doesn't save memory at least several times, mark it as useless and don't do it anymore. @@ -59,6 +66,8 @@ private: /// Merge all accumulated blocks to keep no more than limit rows. void remerge(); + void updatePartialResult(); + ProcessorPtr external_merging_sorted; }; From df1601bfa9c09f9b4b42cad71ed48aa9396f5611 Mon Sep 17 00:00:00 2001 From: alexX512 Date: Mon, 10 Apr 2023 14:06:19 +0000 Subject: [PATCH 004/433] Change to correct name of setting partial_result_on_first_cancel --- src/Core/Settings.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/Core/Settings.h b/src/Core/Settings.h index f1732000a5e..d02ca58de82 100644 --- a/src/Core/Settings.h +++ b/src/Core/Settings.h @@ -283,7 +283,7 @@ class IColumn; \ M(Bool, final, false, "Query with the FINAL modifier by default. If the engine does not support final, it does not have any effect. On queries with multiple tables final is applied only on those that support it. It also works on distributed tables", 0) \ \ - M(Bool, stop_reading_on_first_cancel, false, "Allows query to return a partial result after cancel.", 0) \ + M(Bool, partial_result_on_first_cancel, false, "Allows query to return a partial result after cancel.", 0) \ \ M(Milliseconds, partial_result_update_duration_ms, 0, "Duration of time in milliseconds between real-time updates of result table sent to the client during query execution.", 0) \ M(UInt64, max_rows_in_partial_result, 10, "Max rows displayed to user after each real-time update of output table during query execution.", 0) \ From 9754e4e9b34948a359ca99f7350c1ad24318572f Mon Sep 17 00:00:00 2001 From: alexX512 Date: Tue, 11 Apr 2023 09:20:12 +0000 Subject: [PATCH 005/433] Add supporting of partial result in LimitTransform --- src/Client/ClientBase.cpp | 2 +- src/Processors/LimitTransform.cpp | 14 ++++++++++++-- 2 files changed, 13 insertions(+), 3 deletions(-) diff --git a/src/Client/ClientBase.cpp b/src/Client/ClientBase.cpp index 18912d22497..bf8ab8be626 100644 --- a/src/Client/ClientBase.cpp +++ b/src/Client/ClientBase.cpp @@ -885,7 +885,7 @@ void ClientBase::processOrdinaryQuery(const String & query_to_execute, ASTPtr pa } const auto & settings = global_context->getSettingsRef(); - const Int32 signals_before_stop = settings.stop_reading_on_first_cancel ? 2 : 1; + const Int32 signals_before_stop = settings.partial_result_on_first_cancel ? 2 : 1; has_partial_result_setting = settings.partial_result_update_duration_ms.totalMilliseconds() > 0; if (is_interactive && has_partial_result_setting) diff --git a/src/Processors/LimitTransform.cpp b/src/Processors/LimitTransform.cpp index 4e517ef6064..07de5795938 100644 --- a/src/Processors/LimitTransform.cpp +++ b/src/Processors/LimitTransform.cpp @@ -187,8 +187,9 @@ LimitTransform::Status LimitTransform::preparePair(PortsData & data) } auto rows = data.current_chunk.getNumRows(); + bool is_current_data_partial = data.current_chunk.hasPartialResult(); - if (rows_before_limit_at_least && !data.input_port_has_counter) + if (rows_before_limit_at_least && !data.input_port_has_counter && !is_current_data_partial) rows_before_limit_at_least->add(rows); /// Skip block (for 'always_read_till_end' case). @@ -212,6 +213,9 @@ LimitTransform::Status LimitTransform::preparePair(PortsData & data) if (rows_read <= offset) { + if (is_current_data_partial) + rows_read = 0; + data.current_chunk.clear(); if (input.isFinished()) @@ -240,9 +244,15 @@ LimitTransform::Status LimitTransform::preparePair(PortsData & data) bool may_need_more_data_for_ties = previous_row_chunk || rows_read - rows <= offset + limit; /// No more data is needed. - if (!always_read_till_end && !limit_is_unreachable && rows_read >= offset + limit && !may_need_more_data_for_ties) + if (!always_read_till_end && !limit_is_unreachable && rows_read >= offset + limit && !may_need_more_data_for_ties && !is_current_data_partial) input.close(); + if (is_current_data_partial) + { + rows_read = 0; + previous_row_chunk = {}; + } + output.push(std::move(data.current_chunk)); return Status::PortFull; From e94285d4683065fc21949ebb3adad72d3ce07eb3 Mon Sep 17 00:00:00 2001 From: alexX512 Date: Mon, 1 May 2023 09:25:15 +0000 Subject: [PATCH 006/433] Send partial result with the new architectuin the special pipeline --- src/Client/ClientBase.cpp | 54 ++++---- src/Client/ClientBase.h | 12 +- src/Core/BlockInfo.h | 8 +- src/Processors/Chunk.cpp | 18 +-- src/Processors/Chunk.h | 18 +-- src/Processors/Executors/ExecutorTasks.cpp | 4 +- .../PullingAsyncPipelineExecutor.cpp | 1 - src/Processors/Formats/IOutputFormat.cpp | 105 +++++++++++--- src/Processors/Formats/IOutputFormat.h | 9 +- .../Formats/Impl/PrettyBlockOutputFormat.cpp | 13 +- .../Formats/Impl/PrettyBlockOutputFormat.h | 3 + .../Impl/PrettyCompactBlockOutputFormat.cpp | 2 +- src/Processors/Formats/LazyOutputFormat.h | 1 + src/Processors/IProcessor.h | 8 ++ src/Processors/LimitTransform.cpp | 28 ++-- src/Processors/LimitTransform.h | 3 + .../QueryPlan/BuildQueryPipelineSettings.cpp | 7 +- .../QueryPlan/BuildQueryPipelineSettings.h | 3 + src/Processors/QueryPlan/QueryPlan.cpp | 4 + src/Processors/QueryPlan/SortingStep.cpp | 6 +- .../Transforms/ExpressionTransform.cpp | 6 + .../Transforms/ExpressionTransform.h | 3 + .../Transforms/LimitPartialResultTransform.h | 53 +++++++ .../Transforms/LimitsCheckingTransform.cpp | 9 +- .../Transforms/LimitsCheckingTransform.h | 3 + .../Transforms/MergeSortingTransform.cpp | 122 ++++++++++------ .../Transforms/MergeSortingTransform.h | 15 +- .../Transforms/PartialResultTransform.cpp | 67 +++++++++ .../Transforms/PartialResultTransform.h | 51 +++++++ src/QueryPipeline/Pipe.cpp | 131 +++++++++++++----- src/QueryPipeline/Pipe.h | 22 ++- src/QueryPipeline/QueryPipeline.cpp | 49 ++++++- src/QueryPipeline/QueryPipeline.h | 4 +- src/QueryPipeline/QueryPipelineBuilder.h | 6 + 34 files changed, 639 insertions(+), 209 deletions(-) create mode 100644 src/Processors/Transforms/LimitPartialResultTransform.h create mode 100644 src/Processors/Transforms/PartialResultTransform.cpp create mode 100644 src/Processors/Transforms/PartialResultTransform.h diff --git a/src/Client/ClientBase.cpp b/src/Client/ClientBase.cpp index bf8ab8be626..3fe2670fc98 100644 --- a/src/Client/ClientBase.cpp +++ b/src/Client/ClientBase.cpp @@ -438,7 +438,18 @@ void ClientBase::onData(Block & block, ASTPtr parsed_query) if (!block) return; - if (!block.info.has_partial_result) + if (block.rows() == 0 && partial_result_mode == PartialResultMode::Active) + { + partial_result_mode = PartialResultMode::Inactive; + if (is_interactive) + { + progress_indication.clearProgressOutput(*tty_buf); + std::cout << "Full result:" << std::endl; + progress_indication.writeProgress(*tty_buf); + } + } + + if (partial_result_mode == PartialResultMode::Inactive) processed_rows += block.rows(); /// Even if all blocks are empty, we still need to initialize the output stream to write empty resultset. @@ -450,35 +461,21 @@ void ClientBase::onData(Block & block, ASTPtr parsed_query) if (block.rows() == 0 || (query_fuzzer_runs != 0 && processed_rows >= 100)) return; - if (received_first_full_result && block.info.has_partial_result) - throw Exception(ErrorCodes::LOGICAL_ERROR, "Server shouldn't send partial results after the first block with a non-empty full result"); - - if (!is_interactive && block.info.has_partial_result) + if (!is_interactive && partial_result_mode == PartialResultMode::Active) return; - bool first_full_result = false; - if (!received_first_full_result && !block.info.has_partial_result) - { - received_first_full_result = true; - first_full_result = true; - } - /// If results are written INTO OUTFILE, we can avoid clearing progress to avoid flicker. if (need_render_progress && tty_buf && (!select_into_file || select_into_file_and_stdout)) progress_indication.clearProgressOutput(*tty_buf); - if (is_interactive && first_full_result && has_partial_result_setting) - std::cout << "Full result:" << std::endl; - try { - /// Clear previous partial results to write new partial results if needed - if (!received_first_full_result && written_first_block) - output_format->clearLastLines(prev_block_rows + 2); + if (partial_result_mode == PartialResultMode::Active) + output_format->writePartialResult(materializeBlock(block)); + else + output_format->write(materializeBlock(block)); - output_format->write(materializeBlock(block)); written_first_block = true; - prev_block_rows = block.rows(); } catch (const Exception &) { @@ -541,6 +538,9 @@ void ClientBase::onProfileInfo(const ProfileInfo & profile_info) void ClientBase::initOutputFormat(const Block & block, ASTPtr parsed_query) try { + if (partial_result_mode == PartialResultMode::NotInit) + partial_result_mode = PartialResultMode::Active; + if (!output_format) { /// Ignore all results when fuzzing as they can be huge. @@ -886,10 +886,14 @@ void ClientBase::processOrdinaryQuery(const String & query_to_execute, ASTPtr pa const auto & settings = global_context->getSettingsRef(); const Int32 signals_before_stop = settings.partial_result_on_first_cancel ? 2 : 1; - has_partial_result_setting = settings.partial_result_update_duration_ms.totalMilliseconds() > 0; + bool has_partial_result_setting = settings.partial_result_update_duration_ms.totalMilliseconds() > 0; - if (is_interactive && has_partial_result_setting) - std::cout << "Partial result:" << std::endl; + if (has_partial_result_setting) + { + partial_result_mode = PartialResultMode::NotInit; + if (is_interactive) + std::cout << "Partial result:" << std::endl; + } int retries_left = 10; while (retries_left) @@ -1645,9 +1649,7 @@ void ClientBase::processParsedSingleQuery(const String & full_query, const Strin } processed_rows = 0; - prev_block_rows = 0; - has_partial_result_setting = false; - received_first_full_result = false; + partial_result_mode = PartialResultMode::Inactive; written_first_block = false; progress_indication.resetProgress(); profile_events.watch.restart(); diff --git a/src/Client/ClientBase.h b/src/Client/ClientBase.h index 1c128c0700d..dee1761fa43 100644 --- a/src/Client/ClientBase.h +++ b/src/Client/ClientBase.h @@ -259,11 +259,13 @@ protected: size_t processed_rows = 0; /// How many rows have been read or written. bool print_num_processed_rows = false; /// Whether to print the number of processed rows at - /// If `received_first_full_result` set to true then - /// client has already received all updates with a partial result - bool received_first_full_result = false; - bool has_partial_result_setting = false; - size_t prev_block_rows = 0; /// How many rows were in the previously received block of data + enum class PartialResultMode: UInt8 + { + NotInit, /// Query doesn't show partial result before the first block with 0 rows. + Active, /// Query shows partial result after the first and before the second block with 0 rows. + Inactive /// Query doesn't show partial result at all. + }; + PartialResultMode partial_result_mode = PartialResultMode::Inactive; bool print_stack_trace = false; /// The last exception that was received from the server. Is used for the diff --git a/src/Core/BlockInfo.h b/src/Core/BlockInfo.h index a0357b292cd..d431303ca39 100644 --- a/src/Core/BlockInfo.h +++ b/src/Core/BlockInfo.h @@ -28,15 +28,9 @@ struct BlockInfo * Otherwise -1. */ - /** has_partial_result: - * If user wants to receive updates containing partial results during query execution, - * then data will be sent with has_partial_result flag set to true. - */ - #define APPLY_FOR_BLOCK_INFO_FIELDS(M) \ M(bool, is_overflows, false, 1) \ - M(Int32, bucket_num, -1, 2) \ - M(bool, has_partial_result, false, 3) + M(Int32, bucket_num, -1, 2) #define DECLARE_FIELD(TYPE, NAME, DEFAULT, FIELD_NUM) \ TYPE NAME = DEFAULT; diff --git a/src/Processors/Chunk.cpp b/src/Processors/Chunk.cpp index d63e4c14db0..ccb72cb977f 100644 --- a/src/Processors/Chunk.cpp +++ b/src/Processors/Chunk.cpp @@ -12,14 +12,14 @@ namespace ErrorCodes extern const int POSITION_OUT_OF_BOUND; } -Chunk::Chunk(DB::Columns columns_, UInt64 num_rows_, bool has_partial_result_) - : columns(std::move(columns_)), num_rows(num_rows_), has_partial_result(has_partial_result_) +Chunk::Chunk(DB::Columns columns_, UInt64 num_rows_) + : columns(std::move(columns_)), num_rows(num_rows_) { checkNumRowsIsConsistent(); } -Chunk::Chunk(Columns columns_, UInt64 num_rows_, ChunkInfoPtr chunk_info_, bool has_partial_result_) - : columns(std::move(columns_)), num_rows(num_rows_), chunk_info(std::move(chunk_info_)), has_partial_result(has_partial_result_) +Chunk::Chunk(Columns columns_, UInt64 num_rows_, ChunkInfoPtr chunk_info_) + : columns(std::move(columns_)), num_rows(num_rows_), chunk_info(std::move(chunk_info_)) { checkNumRowsIsConsistent(); } @@ -34,21 +34,21 @@ static Columns unmuteColumns(MutableColumns && mutable_columns) return columns; } -Chunk::Chunk(MutableColumns columns_, UInt64 num_rows_, bool has_partial_result_) - : columns(unmuteColumns(std::move(columns_))), num_rows(num_rows_), has_partial_result(has_partial_result_) +Chunk::Chunk(MutableColumns columns_, UInt64 num_rows_) + : columns(unmuteColumns(std::move(columns_))), num_rows(num_rows_) { checkNumRowsIsConsistent(); } -Chunk::Chunk(MutableColumns columns_, UInt64 num_rows_, ChunkInfoPtr chunk_info_, bool has_partial_result_) - : columns(unmuteColumns(std::move(columns_))), num_rows(num_rows_), chunk_info(std::move(chunk_info_)), has_partial_result(has_partial_result_) +Chunk::Chunk(MutableColumns columns_, UInt64 num_rows_, ChunkInfoPtr chunk_info_) + : columns(unmuteColumns(std::move(columns_))), num_rows(num_rows_), chunk_info(std::move(chunk_info_)) { checkNumRowsIsConsistent(); } Chunk Chunk::clone() const { - return Chunk(getColumns(), getNumRows(), chunk_info, has_partial_result); + return Chunk(getColumns(), getNumRows(), chunk_info); } void Chunk::setColumns(Columns columns_, UInt64 num_rows_) diff --git a/src/Processors/Chunk.h b/src/Processors/Chunk.h index e3f9b17dfe9..6f2097b71f1 100644 --- a/src/Processors/Chunk.h +++ b/src/Processors/Chunk.h @@ -38,16 +38,14 @@ public: : columns(std::move(other.columns)) , num_rows(other.num_rows) , chunk_info(std::move(other.chunk_info)) - , has_partial_result(other.has_partial_result) { other.num_rows = 0; - other.has_partial_result = false; } - Chunk(Columns columns_, UInt64 num_rows_, bool has_partial_result_ = false); - Chunk(Columns columns_, UInt64 num_rows_, ChunkInfoPtr chunk_info_, bool has_partial_result_ = false); - Chunk(MutableColumns columns_, UInt64 num_rows_, bool has_partial_result_ = false); - Chunk(MutableColumns columns_, UInt64 num_rows_, ChunkInfoPtr chunk_info_, bool has_partial_result_ = false); + Chunk(Columns columns_, UInt64 num_rows_); + Chunk(Columns columns_, UInt64 num_rows_, ChunkInfoPtr chunk_info_); + Chunk(MutableColumns columns_, UInt64 num_rows_); + Chunk(MutableColumns columns_, UInt64 num_rows_, ChunkInfoPtr chunk_info_); Chunk & operator=(const Chunk & other) = delete; Chunk & operator=(Chunk && other) noexcept @@ -56,8 +54,6 @@ public: chunk_info = std::move(other.chunk_info); num_rows = other.num_rows; other.num_rows = 0; - has_partial_result = other.has_partial_result; - other.has_partial_result = false; return *this; } @@ -68,7 +64,6 @@ public: columns.swap(other.columns); chunk_info.swap(other.chunk_info); std::swap(num_rows, other.num_rows); - std::swap(has_partial_result, other.has_partial_result); } void clear() @@ -76,7 +71,6 @@ public: num_rows = 0; columns.clear(); chunk_info.reset(); - has_partial_result = false; } const Columns & getColumns() const { return columns; } @@ -110,14 +104,10 @@ public: void append(const Chunk & chunk); void append(const Chunk & chunk, size_t from, size_t length); // append rows [from, from+length) of chunk - void changePartialResultStatus(bool has_partial_result_) { has_partial_result = has_partial_result_; } - bool hasPartialResult() const { return has_partial_result; } - private: Columns columns; UInt64 num_rows = 0; ChunkInfoPtr chunk_info; - bool has_partial_result = false; void checkNumRowsIsConsistent(); }; diff --git a/src/Processors/Executors/ExecutorTasks.cpp b/src/Processors/Executors/ExecutorTasks.cpp index e61d225a968..496e859746a 100644 --- a/src/Processors/Executors/ExecutorTasks.cpp +++ b/src/Processors/Executors/ExecutorTasks.cpp @@ -108,8 +108,8 @@ void ExecutorTasks::pushTasks(Queue & queue, Queue & async_queue, ExecutionThrea { context.setTask(nullptr); - /// Take local task from queue if has one. - if (!queue.empty() && !context.hasAsyncTasks()) + /// Take local task from queue if has one and it's not a processor which generates partial result. + if (!queue.empty() && !queue.front()->processor->isPartialResultProcessor() && !context.hasAsyncTasks()) { context.setTask(queue.front()); queue.pop(); diff --git a/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp b/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp index e568ec21dac..73a3142d459 100644 --- a/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp +++ b/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp @@ -160,7 +160,6 @@ bool PullingAsyncPipelineExecutor::pull(Block & block, uint64_t milliseconds) } block = lazy_format->getPort(IOutputFormat::PortKind::Main).getHeader().cloneWithColumns(chunk.detachColumns()); - block.info.has_partial_result = chunk.hasPartialResult(); if (auto chunk_info = chunk.getChunkInfo()) { diff --git a/src/Processors/Formats/IOutputFormat.cpp b/src/Processors/Formats/IOutputFormat.cpp index cd292d7795e..d3e0ed2fcc6 100644 --- a/src/Processors/Formats/IOutputFormat.cpp +++ b/src/Processors/Formats/IOutputFormat.cpp @@ -6,7 +6,7 @@ namespace DB { IOutputFormat::IOutputFormat(const Block & header_, WriteBuffer & out_) - : IProcessor({header_, header_, header_}, {}), out(out_) + : IProcessor({header_, header_, header_, header_}, {}), out(out_) { } @@ -15,26 +15,13 @@ IOutputFormat::Status IOutputFormat::prepare() if (has_input) return Status::Ready; - for (auto kind : {Main, Totals, Extremes}) - { - auto & input = getPort(kind); + auto status = prepareMainAndPartialResult(); + if (status != Status::Finished) + return status; - if (kind != Main && !input.isConnected()) - continue; - - if (input.isFinished()) - continue; - - input.setNeeded(); - - if (!input.hasData()) - return Status::NeedData; - - current_chunk = input.pull(true); - current_block_kind = kind; - has_input = true; - return Status::Ready; - } + status = prepareTotalsAndExtremes(); + if (status != Status::Finished) + return status; finished = true; @@ -44,6 +31,65 @@ IOutputFormat::Status IOutputFormat::prepare() return Status::Finished; } +IOutputFormat::Status IOutputFormat::prepareMainAndPartialResult() +{ + bool need_data = false; + for (auto kind : {Main, PartialResult}) + { + auto & input = getPort(kind); + + if (input.isFinished()) + continue; + + if (kind == PartialResult && was_main_input) + { + input.close(); + continue; + } + + input.setNeeded(); + need_data = true; + + if (!input.hasData()) + continue; + + setCurrentChunk(input, kind); + return Status::Ready; + } + + if (need_data) + return Status::NeedData; + + return Status::Finished; +} + +IOutputFormat::Status IOutputFormat::prepareTotalsAndExtremes() +{ + for (auto kind : {Totals, Extremes}) + { + auto & input = getPort(kind); + + if (!input.isConnected() || input.isFinished()) + continue; + + input.setNeeded(); + if (!input.hasData()) + return Status::NeedData; + + setCurrentChunk(input, kind); + return Status::Ready; + } + + return Status::Finished; +} + +void IOutputFormat::setCurrentChunk(InputPort & input, PortKind kind) +{ + current_chunk = input.pull(true); + current_block_kind = kind; + has_input = true; +} + static Chunk prepareTotals(Chunk chunk) { if (!chunk.hasRows()) @@ -83,8 +129,16 @@ void IOutputFormat::work() case Main: result_rows += current_chunk.getNumRows(); result_bytes += current_chunk.allocatedBytes(); + if (!was_main_input && current_chunk.hasRows()) + { + consume(Chunk(current_chunk.cloneEmptyColumns(), 0)); + was_main_input = true; + } consume(std::move(current_chunk)); break; + case PartialResult: + consumePartialResult(std::move(current_chunk)); + break; case Totals: writeSuffixIfNeeded(); if (auto totals = prepareTotals(std::move(current_chunk))) @@ -113,7 +167,16 @@ void IOutputFormat::flush() void IOutputFormat::write(const Block & block) { writePrefixIfNeeded(); - consume(Chunk(block.getColumns(), block.rows(), block.info.has_partial_result)); + consume(Chunk(block.getColumns(), block.rows())); + + if (auto_flush) + flush(); +} + +void IOutputFormat::writePartialResult(const Block & block) +{ + writePrefixIfNeeded(); + consumePartialResult(Chunk(block.getColumns(), block.rows())); if (auto_flush) flush(); diff --git a/src/Processors/Formats/IOutputFormat.h b/src/Processors/Formats/IOutputFormat.h index 04cf38b3038..2c598e2620b 100644 --- a/src/Processors/Formats/IOutputFormat.h +++ b/src/Processors/Formats/IOutputFormat.h @@ -23,7 +23,7 @@ class WriteBuffer; class IOutputFormat : public IProcessor { public: - enum PortKind { Main = 0, Totals = 1, Extremes = 2 }; + enum PortKind { Main = 0, Totals = 1, Extremes = 2, PartialResult = 3 }; IOutputFormat(const Block & header_, WriteBuffer & out_); @@ -54,6 +54,7 @@ public: /// TODO: separate formats and processors. void write(const Block & block); + void writePartialResult(const Block & block); void finalize(); @@ -120,6 +121,7 @@ protected: virtual void consume(Chunk) = 0; virtual void consumeTotals(Chunk) {} virtual void consumeExtremes(Chunk) {} + virtual void consumePartialResult(Chunk) {} virtual void finalizeImpl() {} virtual void finalizeBuffers() {} virtual void writePrefix() {} @@ -168,6 +170,7 @@ protected: Chunk current_chunk; PortKind current_block_kind = PortKind::Main; + bool was_main_input = false; bool has_input = false; bool finished = false; bool finalized = false; @@ -182,6 +185,10 @@ protected: Statistics statistics; private: + IOutputFormat::Status prepareMainAndPartialResult(); + IOutputFormat::Status prepareTotalsAndExtremes(); + void setCurrentChunk(InputPort & input, PortKind kind); + size_t rows_read_before = 0; bool are_totals_written = false; diff --git a/src/Processors/Formats/Impl/PrettyBlockOutputFormat.cpp b/src/Processors/Formats/Impl/PrettyBlockOutputFormat.cpp index 208b851b7df..3e468ceaf75 100644 --- a/src/Processors/Formats/Impl/PrettyBlockOutputFormat.cpp +++ b/src/Processors/Formats/Impl/PrettyBlockOutputFormat.cpp @@ -140,7 +140,7 @@ void PrettyBlockOutputFormat::write(Chunk chunk, PortKind port_kind) { if (total_rows >= format_settings.pretty.max_rows) { - if (!chunk.hasPartialResult()) + if (port_kind != PortKind::PartialResult) total_rows += chunk.getNumRows(); return; } @@ -322,7 +322,7 @@ void PrettyBlockOutputFormat::writeChunk(const Chunk & chunk, PortKind port_kind } writeString(bottom_separator_s, out); - if (!chunk.hasPartialResult()) + if (port_kind != PortKind::PartialResult) total_rows += num_rows; } @@ -396,6 +396,15 @@ void PrettyBlockOutputFormat::consumeExtremes(Chunk chunk) write(std::move(chunk), PortKind::Extremes); } +void PrettyBlockOutputFormat::consumePartialResult(Chunk chunk) +{ + if (prev_partial_block_rows > 0) + clearLastLines(prev_partial_block_rows + 2); + + prev_partial_block_rows = chunk.getNumRows(); + write(std::move(chunk), PortKind::PartialResult); +} + void PrettyBlockOutputFormat::writeMonoChunkIfNeeded() { diff --git a/src/Processors/Formats/Impl/PrettyBlockOutputFormat.h b/src/Processors/Formats/Impl/PrettyBlockOutputFormat.h index 95c72d15fa9..f9e69ebad51 100644 --- a/src/Processors/Formats/Impl/PrettyBlockOutputFormat.h +++ b/src/Processors/Formats/Impl/PrettyBlockOutputFormat.h @@ -27,9 +27,11 @@ protected: void consume(Chunk) override; void consumeTotals(Chunk) override; void consumeExtremes(Chunk) override; + void consumePartialResult(Chunk) override; size_t total_rows = 0; size_t terminal_width = 0; + size_t prev_partial_block_rows = 0; size_t row_number_width = 7; // "10000. " @@ -57,6 +59,7 @@ protected: void resetFormatterImpl() override { total_rows = 0; + prev_partial_block_rows = 0; } private: diff --git a/src/Processors/Formats/Impl/PrettyCompactBlockOutputFormat.cpp b/src/Processors/Formats/Impl/PrettyCompactBlockOutputFormat.cpp index bf6e4cb21f1..3a04d86b1ad 100644 --- a/src/Processors/Formats/Impl/PrettyCompactBlockOutputFormat.cpp +++ b/src/Processors/Formats/Impl/PrettyCompactBlockOutputFormat.cpp @@ -194,7 +194,7 @@ void PrettyCompactBlockOutputFormat::writeChunk(const Chunk & chunk, PortKind po writeBottom(max_widths); - if (!chunk.hasPartialResult()) + if (port_kind != PortKind::PartialResult) total_rows += num_rows; } diff --git a/src/Processors/Formats/LazyOutputFormat.h b/src/Processors/Formats/LazyOutputFormat.h index b539a8494c7..81b9d3e464a 100644 --- a/src/Processors/Formats/LazyOutputFormat.h +++ b/src/Processors/Formats/LazyOutputFormat.h @@ -49,6 +49,7 @@ protected: void consumeTotals(Chunk chunk) override { totals = std::move(chunk); } void consumeExtremes(Chunk chunk) override { extremes = std::move(chunk); } + void consumePartialResult(Chunk chunk) override { consume(std::move(chunk)); } private: diff --git a/src/Processors/IProcessor.h b/src/Processors/IProcessor.h index c9dd7d8d77d..cb819c00ff2 100644 --- a/src/Processors/IProcessor.h +++ b/src/Processors/IProcessor.h @@ -234,6 +234,14 @@ public: { throw Exception(ErrorCodes::NOT_IMPLEMENTED, "Method 'expandPipeline' is not implemented for {} processor", getName()); } + + virtual bool isPartialResultProcessor() const { return false; } + virtual bool supportPartialResultProcessor() const { return false; } + + virtual ProcessorPtr getPartialResultProcessor(ProcessorPtr /*current_processor*/, UInt64 /*partial_result_limit*/, UInt64 /*partial_result_duration_ms*/) + { + throw Exception(ErrorCodes::NOT_IMPLEMENTED, "Method 'getPartialResultProcessor' is not implemented for {} processor", getName()); + } /// In case if query was cancelled executor will wait till all processors finish their jobs. /// Generally, there is no reason to check this flag. However, it may be reasonable for long operations (e.g. i/o). diff --git a/src/Processors/LimitTransform.cpp b/src/Processors/LimitTransform.cpp index 07de5795938..0d43b1e4fac 100644 --- a/src/Processors/LimitTransform.cpp +++ b/src/Processors/LimitTransform.cpp @@ -1,5 +1,5 @@ #include - +#include namespace DB { @@ -180,16 +180,9 @@ LimitTransform::Status LimitTransform::preparePair(PortsData & data) return Status::NeedData; data.current_chunk = input.pull(true); - if (data.current_chunk.hasPartialResult()) - { - output.push(std::move(data.current_chunk)); - return Status::PortFull; - } - auto rows = data.current_chunk.getNumRows(); - bool is_current_data_partial = data.current_chunk.hasPartialResult(); - if (rows_before_limit_at_least && !data.input_port_has_counter && !is_current_data_partial) + if (rows_before_limit_at_least && !data.input_port_has_counter) rows_before_limit_at_least->add(rows); /// Skip block (for 'always_read_till_end' case). @@ -213,9 +206,6 @@ LimitTransform::Status LimitTransform::preparePair(PortsData & data) if (rows_read <= offset) { - if (is_current_data_partial) - rows_read = 0; - data.current_chunk.clear(); if (input.isFinished()) @@ -244,15 +234,9 @@ LimitTransform::Status LimitTransform::preparePair(PortsData & data) bool may_need_more_data_for_ties = previous_row_chunk || rows_read - rows <= offset + limit; /// No more data is needed. - if (!always_read_till_end && !limit_is_unreachable && rows_read >= offset + limit && !may_need_more_data_for_ties && !is_current_data_partial) + if (!always_read_till_end && !limit_is_unreachable && rows_read >= offset + limit && !may_need_more_data_for_ties) input.close(); - if (is_current_data_partial) - { - rows_read = 0; - previous_row_chunk = {}; - } - output.push(std::move(data.current_chunk)); return Status::PortFull; @@ -382,5 +366,11 @@ bool LimitTransform::sortColumnsEqualAt(const ColumnRawPtrs & current_chunk_sort return true; } +ProcessorPtr LimitTransform::getPartialResultProcessor(ProcessorPtr /*current_processor*/, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) +{ + const auto & header = inputs.front().getHeader(); + return std::make_shared(header, partial_result_limit, partial_result_duration_ms, limit, offset); +} + } diff --git a/src/Processors/LimitTransform.h b/src/Processors/LimitTransform.h index 33ff968985f..7ad358c5d41 100644 --- a/src/Processors/LimitTransform.h +++ b/src/Processors/LimitTransform.h @@ -55,6 +55,9 @@ private: ColumnRawPtrs extractSortColumns(const Columns & columns) const; bool sortColumnsEqualAt(const ColumnRawPtrs & current_chunk_sort_columns, UInt64 current_chunk_row_num) const; + bool supportPartialResultProcessor() const override { return true; } + ProcessorPtr getPartialResultProcessor(ProcessorPtr current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) override; + public: LimitTransform( const Block & header_, UInt64 limit_, UInt64 offset_, size_t num_streams = 1, diff --git a/src/Processors/QueryPlan/BuildQueryPipelineSettings.cpp b/src/Processors/QueryPlan/BuildQueryPipelineSettings.cpp index fb3ed7f80fc..60ac30389a1 100644 --- a/src/Processors/QueryPlan/BuildQueryPipelineSettings.cpp +++ b/src/Processors/QueryPlan/BuildQueryPipelineSettings.cpp @@ -9,7 +9,12 @@ namespace DB BuildQueryPipelineSettings BuildQueryPipelineSettings::fromContext(ContextPtr from) { BuildQueryPipelineSettings settings; - settings.actions_settings = ExpressionActionsSettings::fromSettings(from->getSettingsRef(), CompileExpressions::yes); + + const auto & context_settings = from->getSettingsRef(); + settings.partial_result_limit = context_settings.max_rows_in_partial_result; + settings.partial_result_duration_ms = context_settings.partial_result_update_duration_ms.totalMilliseconds(); + + settings.actions_settings = ExpressionActionsSettings::fromSettings(context_settings, CompileExpressions::yes); settings.process_list_element = from->getProcessListElement(); settings.progress_callback = from->getProgressCallback(); return settings; diff --git a/src/Processors/QueryPlan/BuildQueryPipelineSettings.h b/src/Processors/QueryPlan/BuildQueryPipelineSettings.h index 3b5e4e06953..0410bf925d1 100644 --- a/src/Processors/QueryPlan/BuildQueryPipelineSettings.h +++ b/src/Processors/QueryPlan/BuildQueryPipelineSettings.h @@ -19,6 +19,9 @@ struct BuildQueryPipelineSettings QueryStatusPtr process_list_element; ProgressCallback progress_callback = nullptr; + UInt64 partial_result_limit = 0; + UInt64 partial_result_duration_ms = 0; + const ExpressionActionsSettings & getActionsSettings() const { return actions_settings; } static BuildQueryPipelineSettings fromContext(ContextPtr from); }; diff --git a/src/Processors/QueryPlan/QueryPlan.cpp b/src/Processors/QueryPlan/QueryPlan.cpp index 3fbe3d89845..1e31ce0711d 100644 --- a/src/Processors/QueryPlan/QueryPlan.cpp +++ b/src/Processors/QueryPlan/QueryPlan.cpp @@ -168,6 +168,7 @@ QueryPipelineBuilderPtr QueryPlan::buildQueryPipeline( QueryPipelineBuilderPtr last_pipeline; + bool has_partial_result_setting = build_pipeline_settings.partial_result_duration_ms > 0; std::stack stack; stack.push(Frame{.node = root}); @@ -195,6 +196,9 @@ QueryPipelineBuilderPtr QueryPlan::buildQueryPipeline( } else stack.push(Frame{.node = frame.node->children[next_child]}); + + if (last_pipeline && has_partial_result_setting) + last_pipeline->activatePartialResult(build_pipeline_settings.partial_result_limit, build_pipeline_settings.partial_result_duration_ms); } last_pipeline->setProgressCallback(build_pipeline_settings.progress_callback); diff --git a/src/Processors/QueryPlan/SortingStep.cpp b/src/Processors/QueryPlan/SortingStep.cpp index ad09afa6ea2..db44da5a0fc 100644 --- a/src/Processors/QueryPlan/SortingStep.cpp +++ b/src/Processors/QueryPlan/SortingStep.cpp @@ -33,8 +33,6 @@ SortingStep::Settings::Settings(const Context & context) max_bytes_before_external_sort = settings.max_bytes_before_external_sort; tmp_data = context.getTempDataOnDisk(); min_free_disk_space = settings.min_free_disk_space_for_temporary_data; - partial_result_limit = settings.max_rows_in_partial_result; - partial_result_duration_ms = settings.partial_result_update_duration_ms.totalMilliseconds(); } SortingStep::Settings::Settings(size_t max_block_size_) @@ -219,9 +217,7 @@ void SortingStep::mergeSorting( sort_settings.remerge_lowered_memory_bytes_ratio, sort_settings.max_bytes_before_external_sort, std::move(tmp_data_on_disk), - sort_settings.min_free_disk_space, - sort_settings.partial_result_limit, - sort_settings.partial_result_duration_ms); + sort_settings.min_free_disk_space); }); } diff --git a/src/Processors/Transforms/ExpressionTransform.cpp b/src/Processors/Transforms/ExpressionTransform.cpp index 49988932947..5ec76dc95b8 100644 --- a/src/Processors/Transforms/ExpressionTransform.cpp +++ b/src/Processors/Transforms/ExpressionTransform.cpp @@ -25,6 +25,12 @@ void ExpressionTransform::transform(Chunk & chunk) chunk.setColumns(block.getColumns(), num_rows); } +ProcessorPtr ExpressionTransform::getPartialResultProcessor(ProcessorPtr /*current_processor*/, UInt64 /*partial_result_limit*/, UInt64 /*partial_result_duration_ms*/) +{ + const auto & header = getInputPort().getHeader(); + return std::make_shared(header, expression); +} + ConvertingTransform::ConvertingTransform(const Block & header_, ExpressionActionsPtr expression_) : ExceptionKeepingTransform(header_, ExpressionTransform::transformHeader(header_, expression_->getActionsDAG())) , expression(std::move(expression_)) diff --git a/src/Processors/Transforms/ExpressionTransform.h b/src/Processors/Transforms/ExpressionTransform.h index 791c7d7ba73..6f9b585ae1f 100644 --- a/src/Processors/Transforms/ExpressionTransform.h +++ b/src/Processors/Transforms/ExpressionTransform.h @@ -29,6 +29,9 @@ public: protected: void transform(Chunk & chunk) override; + bool supportPartialResultProcessor() const override { return true; } + ProcessorPtr getPartialResultProcessor(ProcessorPtr current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) override; + private: ExpressionActionsPtr expression; }; diff --git a/src/Processors/Transforms/LimitPartialResultTransform.h b/src/Processors/Transforms/LimitPartialResultTransform.h new file mode 100644 index 00000000000..8a919615088 --- /dev/null +++ b/src/Processors/Transforms/LimitPartialResultTransform.h @@ -0,0 +1,53 @@ +#pragma once + +#include + +namespace DB +{ + +class LimitPartialResultTransform : public PartialResultTransform +{ +public: + LimitPartialResultTransform( + const Block & header, + UInt64 partial_result_limit_, + UInt64 partial_result_duration_ms_, + UInt64 limit_, + UInt64 offset_) + : PartialResultTransform(header, partial_result_limit_, partial_result_duration_ms_) + , limit(limit_) + , offset(offset_) + {} + + String getName() const override { return "MergeSortingPartialResult"; } + + void transformPartialResult(Chunk & chunk) override + { + UInt64 num_rows = chunk.getNumRows(); + if (num_rows < offset || limit == 0) + { + chunk = {}; + return; + } + + UInt64 length = std::min(limit, num_rows - offset); + + /// Check if some rows should be removed + if (length < num_rows) + { + auto columns = chunk.detachColumns(); + UInt64 num_columns = chunk.getNumColumns(); + + for (UInt64 i = 0; i < num_columns; ++i) + columns[i] = columns[i]->cut(offset, limit); + + chunk.setColumns(std::move(columns), length); + } + } + +private: + UInt64 limit; + UInt64 offset; +}; + +} diff --git a/src/Processors/Transforms/LimitsCheckingTransform.cpp b/src/Processors/Transforms/LimitsCheckingTransform.cpp index 7edab0d417c..b426d050543 100644 --- a/src/Processors/Transforms/LimitsCheckingTransform.cpp +++ b/src/Processors/Transforms/LimitsCheckingTransform.cpp @@ -1,4 +1,5 @@ #include +#include #include namespace DB @@ -38,7 +39,7 @@ void LimitsCheckingTransform::transform(Chunk & chunk) return; } - if (chunk && !chunk.hasPartialResult()) + if (chunk) { info.update(chunk); @@ -74,4 +75,10 @@ void LimitsCheckingTransform::checkQuota(Chunk & chunk) } } +ProcessorPtr LimitsCheckingTransform::getPartialResultProcessor(ProcessorPtr /*current_processor*/, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) +{ + const auto & header = inputs.front().getHeader(); + return std::make_shared(header, partial_result_limit, partial_result_duration_ms); +} + } diff --git a/src/Processors/Transforms/LimitsCheckingTransform.h b/src/Processors/Transforms/LimitsCheckingTransform.h index 2f96a17c17b..663eb774edf 100644 --- a/src/Processors/Transforms/LimitsCheckingTransform.h +++ b/src/Processors/Transforms/LimitsCheckingTransform.h @@ -36,6 +36,9 @@ public: protected: void transform(Chunk & chunk) override; + bool supportPartialResultProcessor() const override { return true; } + ProcessorPtr getPartialResultProcessor(ProcessorPtr current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) override; + private: StreamLocalLimits limits; diff --git a/src/Processors/Transforms/MergeSortingTransform.cpp b/src/Processors/Transforms/MergeSortingTransform.cpp index c44633fb80c..c4c6743c775 100644 --- a/src/Processors/Transforms/MergeSortingTransform.cpp +++ b/src/Processors/Transforms/MergeSortingTransform.cpp @@ -1,4 +1,5 @@ #include +#include #include #include #include @@ -26,6 +27,11 @@ namespace ProfileEvents namespace DB { +namespace ErrorCodes +{ + extern const int LOGICAL_ERROR; +} + class BufferingToFileTransform : public IAccumulatingTransform { public: @@ -75,6 +81,58 @@ private: Poco::Logger * log; }; +class MergeSortingPartialResultTransform : public PartialResultTransform +{ +public: + using MergeSortingTransformPtr = std::shared_ptr; + + MergeSortingPartialResultTransform(const Block & header, MergeSortingTransformPtr merge_sorting_transform_, UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_) + : PartialResultTransform(header, partial_result_limit_, partial_result_duration_ms_) + , merge_sorting_transform(std::move(merge_sorting_transform_)) + { + } + + String getName() const override { return "MergeSortingPartialResultTransform"; } + + ShaphotResult getRealProcessorSnapshot() override + { + std::lock_guard lock(merge_sorting_transform->snapshot_mutex); + if (merge_sorting_transform->generated_prefix) + return {{}, SnaphotStatus::Stopped}; + + if (merge_sorting_transform->chunks.empty()) + return {{}, SnaphotStatus::NotReady}; + + /// Sort all input data + merge_sorting_transform->remerge(); + /// Add a copy of the first `partial_result_limit` rows to a generated_chunk + /// to send it as a partial result in the next prepare stage + auto generated_columns = merge_sorting_transform->chunks[0].cloneEmptyColumns(); + size_t total_rows = 0; + for (const auto & merged_chunk : merge_sorting_transform->chunks) + { + size_t rows = std::min(merged_chunk.getNumRows(), partial_result_limit - total_rows); + if (rows == 0) + break; + + for (size_t position = 0; position < generated_columns.size(); ++position) + { + auto column = merged_chunk.getColumns()[position]; + generated_columns[position]->insertRangeFrom(*column, 0, rows); + } + + total_rows += rows; + } + + auto partial_result = Chunk(std::move(generated_columns), total_rows, merge_sorting_transform->chunks[0].getChunkInfo()); + merge_sorting_transform->enrichChunkWithConstants(partial_result); + return {std::move(partial_result), SnaphotStatus::Ready}; + } + +private: + MergeSortingTransformPtr merge_sorting_transform; +}; + MergeSortingTransform::MergeSortingTransform( const Block & header, const SortDescription & description_, @@ -85,18 +143,13 @@ MergeSortingTransform::MergeSortingTransform( double remerge_lowered_memory_bytes_ratio_, size_t max_bytes_before_external_sort_, TemporaryDataOnDiskPtr tmp_data_, - size_t min_free_disk_space_, - UInt64 partial_result_limit_, - UInt64 partial_result_duration_ms_) + size_t min_free_disk_space_) : SortingTransform(header, description_, max_merged_block_size_, limit_, increase_sort_description_compile_attempts) , max_bytes_before_remerge(max_bytes_before_remerge_) , remerge_lowered_memory_bytes_ratio(remerge_lowered_memory_bytes_ratio_) , max_bytes_before_external_sort(max_bytes_before_external_sort_) , tmp_data(std::move(tmp_data_)) , min_free_disk_space(min_free_disk_space_) - , partial_result_limit(partial_result_limit_) - , partial_result_duration_ms(partial_result_duration_ms_) - , watch(CLOCK_MONOTONIC) { } @@ -129,35 +182,6 @@ Processors MergeSortingTransform::expandPipeline() return std::move(processors); } -void MergeSortingTransform::updatePartialResult() -{ - /// Sort all input data - remerge(); - /// Add a copy of the first `partial_result_limit` rows to a generated_chunk - /// to send it as a partial result in the next prepare stage - auto generated_columns = chunks[0].cloneEmptyColumns(); - size_t total_rows = 0; - for (const auto & merged_chunk : chunks) - { - size_t rows = std::min(merged_chunk.getNumRows(), partial_result_limit - total_rows); - if (rows == 0) - break; - - for (size_t position = 0; position < generated_columns.size(); ++position) - { - auto column = merged_chunk.getColumns()[position]; - generated_columns[position]->insertRangeFrom(*column, 0, rows); - } - - total_rows += rows; - } - - generated_chunk.setColumns(std::move(generated_columns), total_rows); - generated_chunk.setChunkInfo(chunks[0].getChunkInfo()); - generated_chunk.changePartialResultStatus(true /*is_chunk_partial*/); - enrichChunkWithConstants(generated_chunk); -} - void MergeSortingTransform::consume(Chunk chunk) { /** Algorithm: @@ -169,6 +193,8 @@ void MergeSortingTransform::consume(Chunk chunk) /// If there were only const columns in sort description, then there is no need to sort. /// Return the chunk as is. + std::lock_guard lock(snapshot_mutex); + if (description.empty()) { generated_chunk = std::move(chunk); @@ -193,12 +219,6 @@ void MergeSortingTransform::consume(Chunk chunk) remerge(); } - if (partial_result_duration_ms && partial_result_duration_ms < watch.elapsedMilliseconds() && !chunks.empty()) - { - updatePartialResult(); - watch.restart(); - } - /** If too many of them and if external sorting is enabled, * will merge blocks that we have in memory at this moment and write merged stream to temporary (compressed) file. * NOTE. It's possible to check free space in filesystem. @@ -251,6 +271,8 @@ void MergeSortingTransform::serialize() void MergeSortingTransform::generate() { + std::lock_guard lock(snapshot_mutex); + if (!generated_prefix) { size_t num_tmp_files = tmp_data ? tmp_data->getStreams().size() : 0; @@ -311,4 +333,22 @@ void MergeSortingTransform::remerge() sum_bytes_in_blocks = new_sum_bytes_in_blocks; } +ProcessorPtr MergeSortingTransform::getPartialResultProcessor(ProcessorPtr current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) +{ + if (getName() != current_processor->getName() || current_processor.get() != this) + throw Exception( + ErrorCodes::LOGICAL_ERROR, + "To create partial result processor variable current_processor should use " \ + "the same class and pointer as in the original processor with class {} and pointer {}. " \ + "But current_processor has another class {} or pointer {} then original.", + getName(), + static_cast(this), + current_processor->getName(), + static_cast(current_processor.get())); + + const auto & header = inputs.front().getHeader(); + const auto & merge_sorting_processor = std::static_pointer_cast(current_processor); + return std::make_shared(header, std::move(merge_sorting_processor), partial_result_limit, partial_result_duration_ms); +} + } diff --git a/src/Processors/Transforms/MergeSortingTransform.h b/src/Processors/Transforms/MergeSortingTransform.h index 2b2db6fae16..0b9766674a5 100644 --- a/src/Processors/Transforms/MergeSortingTransform.h +++ b/src/Processors/Transforms/MergeSortingTransform.h @@ -30,9 +30,7 @@ public: double remerge_lowered_memory_bytes_ratio_, size_t max_bytes_before_external_sort_, TemporaryDataOnDiskPtr tmp_data_, - size_t min_free_disk_space_, - UInt64 partial_result_limit_ = 0, - UInt64 partial_result_duration_ms_ = 0); + size_t min_free_disk_space_); String getName() const override { return "MergeSortingTransform"; } @@ -43,6 +41,9 @@ protected: Processors expandPipeline() override; + bool supportPartialResultProcessor() const override { return true; } + ProcessorPtr getPartialResultProcessor(ProcessorPtr current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) override; + private: size_t max_bytes_before_remerge; double remerge_lowered_memory_bytes_ratio; @@ -53,11 +54,6 @@ private: size_t sum_rows_in_blocks = 0; size_t sum_bytes_in_blocks = 0; - UInt64 partial_result_limit; - UInt64 partial_result_duration_ms; - - Stopwatch watch; - Poco::Logger * log = &Poco::Logger::get("MergeSortingTransform"); /// If remerge doesn't save memory at least several times, mark it as useless and don't do it anymore. @@ -66,7 +62,8 @@ private: /// Merge all accumulated blocks to keep no more than limit rows. void remerge(); - void updatePartialResult(); + friend class MergeSortingPartialResult; + std::mutex snapshot_mutex; ProcessorPtr external_merging_sorted; }; diff --git a/src/Processors/Transforms/PartialResultTransform.cpp b/src/Processors/Transforms/PartialResultTransform.cpp new file mode 100644 index 00000000000..4fda35085c3 --- /dev/null +++ b/src/Processors/Transforms/PartialResultTransform.cpp @@ -0,0 +1,67 @@ +#include + +namespace DB +{ + +PartialResultTransform::PartialResultTransform(const Block & header, UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_) + : IProcessor({header}, {header}) + , input(inputs.front()) + , output(outputs.front()) + , partial_result_limit(partial_result_limit_) + , partial_result_duration_ms(partial_result_duration_ms_) + , watch(CLOCK_MONOTONIC) + {} + +IProcessor::Status PartialResultTransform::prepare() +{ + if (output.isFinished()) + { + input.close(); + return Status::Finished; + } + + if (finished_getting_snapshots) + { + output.finish(); + return Status::Finished; + } + + if (input.hasData()) + partial_result = {input.pull(), SnaphotStatus::Ready}; + + /// Send partial result from real processor snapshot or from previous partial result processor if possible + if (partial_result.snapshot_status == SnaphotStatus::Ready && output.canPush()) + { + transformPartialResult(partial_result.chunk); + partial_result.snapshot_status = SnaphotStatus::NotReady; + if (partial_result.chunk.getNumRows() > 0) + { + output.push(std::move(partial_result.chunk)); + return Status::PortFull; + } + } + + /// If input data from previous partial result processor is finished then + /// PartialResultTransform ready to create snapshots and send them as a partial result + if (input.isFinished()) + { + return Status::Ready; + } + + input.setNeeded(); + return Status::NeedData; +} + +void PartialResultTransform::work() +{ + if (partial_result_duration_ms < watch.elapsedMilliseconds()) + { + partial_result = getRealProcessorSnapshot(); + if (partial_result.snapshot_status == SnaphotStatus::Stopped) + finished_getting_snapshots = true; + + watch.restart(); + } +} + +} diff --git a/src/Processors/Transforms/PartialResultTransform.h b/src/Processors/Transforms/PartialResultTransform.h new file mode 100644 index 00000000000..0e3c503968f --- /dev/null +++ b/src/Processors/Transforms/PartialResultTransform.h @@ -0,0 +1,51 @@ +#pragma once + +#include + +namespace DB +{ + +class PartialResultTransform : public IProcessor +{ +public: + PartialResultTransform(const Block & header, UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_); + + String getName() const override { return "PartialResultTransform"; } + + Status prepare() override; + void work() override; + + bool isPartialResultProcessor() const override { return true; } + +protected: + enum class SnaphotStatus + { + NotReady, + Ready, + Stopped, + }; + + struct ShaphotResult + { + Chunk chunk; + SnaphotStatus snapshot_status; + }; + + InputPort & input; + OutputPort & output; + + UInt64 partial_result_limit; + UInt64 partial_result_duration_ms; + + ShaphotResult partial_result = {{}, SnaphotStatus::NotReady}; + + bool finished_getting_snapshots = false; + + virtual void transformPartialResult(Chunk & /*chunk*/) {} + virtual ShaphotResult getRealProcessorSnapshot() { return {{}, SnaphotStatus::Stopped}; } + +private: + Stopwatch watch; +}; + +} diff --git a/src/QueryPipeline/Pipe.cpp b/src/QueryPipeline/Pipe.cpp index 91ba01c479f..b84cb279acd 100644 --- a/src/QueryPipeline/Pipe.cpp +++ b/src/QueryPipeline/Pipe.cpp @@ -12,6 +12,7 @@ #include #include #include +#include namespace DB { @@ -167,12 +168,9 @@ Pipe::Pipe(ProcessorPtr source) { checkSource(*source); - if (collected_processors) - collected_processors->emplace_back(source); - output_ports.push_back(&source->getOutputs().front()); header = output_ports.front()->getHeader(); - processors->emplace_back(std::move(source)); + addProcessor(std::move(source)); max_parallel_streams = 1; } @@ -319,6 +317,19 @@ Pipe Pipe::unitePipes(Pipes pipes, Processors * collected_processors, bool allow res.processors->insert(res.processors->end(), pipe.processors->begin(), pipe.processors->end()); res.output_ports.insert(res.output_ports.end(), pipe.output_ports.begin(), pipe.output_ports.end()); + if (res.isPartialResultActive() && pipe.isPartialResultActive()) + { + res.partial_result_ports.insert( + res.partial_result_ports.end(), + pipe.partial_result_ports.begin(), + pipe.partial_result_ports.end()); + } + else + { + res.is_partial_result_active = false; + res.partial_result_ports.clear(); + } + res.max_parallel_streams += pipe.max_parallel_streams; if (pipe.totals_port) @@ -352,11 +363,11 @@ void Pipe::addSource(ProcessorPtr source) else assertBlocksHaveEqualStructure(header, source_header, "Pipes"); - if (collected_processors) - collected_processors->emplace_back(source); - output_ports.push_back(&source->getOutputs().front()); - processors->emplace_back(std::move(source)); + if (isPartialResultActive()) + partial_result_ports.push_back(nullptr); + + addProcessor(std::move(source)); max_parallel_streams = std::max(max_parallel_streams, output_ports.size()); } @@ -374,11 +385,9 @@ void Pipe::addTotalsSource(ProcessorPtr source) assertBlocksHaveEqualStructure(header, source_header, "Pipes"); - if (collected_processors) - collected_processors->emplace_back(source); - totals_port = &source->getOutputs().front(); - processors->emplace_back(std::move(source)); + + addProcessor(std::move(source)); } void Pipe::addExtremesSource(ProcessorPtr source) @@ -394,11 +403,20 @@ void Pipe::addExtremesSource(ProcessorPtr source) assertBlocksHaveEqualStructure(header, source_header, "Pipes"); - if (collected_processors) - collected_processors->emplace_back(source); - extremes_port = &source->getOutputs().front(); - processors->emplace_back(std::move(source)); + + addProcessor(std::move(source)); +} + +void Pipe::activatePartialResult(UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_) +{ + if (!is_partial_result_active) + { + is_partial_result_active = true; + partial_result_limit = partial_result_limit_; + partial_result_duration_ms = partial_result_duration_ms_; + partial_result_ports.assign(output_ports.size(), nullptr); + } } static void dropPort(OutputPort *& port, Processors & processors, Processors * collected_processors) @@ -426,6 +444,15 @@ void Pipe::dropExtremes() dropPort(extremes_port, *processors, collected_processors); } +void Pipe::dropPartialResult() +{ + for (auto & port : partial_result_ports) + dropPort(port, *processors, collected_processors); + + is_partial_result_active = false; + partial_result_ports.clear(); +} + void Pipe::addTransform(ProcessorPtr transform) { addTransform(std::move(transform), static_cast(nullptr), static_cast(nullptr)); @@ -506,10 +533,7 @@ void Pipe::addTransform(ProcessorPtr transform, OutputPort * totals, OutputPort if (extremes_port) assertBlocksHaveEqualStructure(header, extremes_port->getHeader(), "Pipes"); - if (collected_processors) - collected_processors->emplace_back(transform); - - processors->emplace_back(std::move(transform)); + addProcessor(std::move(transform)); max_parallel_streams = std::max(max_parallel_streams, output_ports.size()); } @@ -595,14 +619,41 @@ void Pipe::addTransform(ProcessorPtr transform, InputPort * totals, InputPort * if (extremes_port) assertBlocksHaveEqualStructure(header, extremes_port->getHeader(), "Pipes"); - if (collected_processors) - collected_processors->emplace_back(transform); + addProcessor(std::move(transform)); - processors->emplace_back(std::move(transform)); max_parallel_streams = std::max(max_parallel_streams, output_ports.size()); } +void Pipe::addPartialResultTransformIfNeeded(ProcessorPtr transform, size_t partial_result_port_id) +{ + if (isPartialResultActive()) + { + auto & partial_result_port = partial_result_ports[partial_result_port_id]; + + if (!transform->supportPartialResultProcessor()) + { + dropPort(partial_result_port, *processors, collected_processors); + return; + } + + if (partial_result_port == nullptr) + { + auto source = std::make_shared(getHeader()); + partial_result_port = &source->getPort(); + + addProcessor(std::move(source)); + } + + auto partial_result_transform = transform->getPartialResultProcessor(std::move(transform), partial_result_limit, partial_result_duration_ms); + + connect(*partial_result_port, partial_result_transform->getInputs().front()); + partial_result_port = &partial_result_transform->getOutputs().front(); + + addProcessor(std::move(partial_result_transform)); + } +} + void Pipe::addSimpleTransform(const ProcessorGetterWithStreamKind & getter) { if (output_ports.empty()) @@ -610,7 +661,7 @@ void Pipe::addSimpleTransform(const ProcessorGetterWithStreamKind & getter) Block new_header; - auto add_transform = [&](OutputPort *& port, StreamType stream_type) + auto add_transform = [&](OutputPort *& port, size_t partial_result_port_id, StreamType stream_type) { if (!port) return; @@ -646,19 +697,22 @@ void Pipe::addSimpleTransform(const ProcessorGetterWithStreamKind & getter) { connect(*port, transform->getInputs().front()); port = &transform->getOutputs().front(); + if (stream_type == StreamType::Main) + addPartialResultTransformIfNeeded(transform, partial_result_port_id); - if (collected_processors) - collected_processors->emplace_back(transform); - - processors->emplace_back(std::move(transform)); + addProcessor(std::move(transform)); } }; + size_t partial_result_port_id = 0; for (auto & port : output_ports) - add_transform(port, StreamType::Main); + { + add_transform(port, partial_result_port_id, StreamType::Main); + ++partial_result_port_id; + } - add_transform(totals_port, StreamType::Totals); - add_transform(extremes_port, StreamType::Extremes); + add_transform(totals_port, 0, StreamType::Totals); + add_transform(extremes_port, 0, StreamType::Extremes); header = std::move(new_header); } @@ -697,18 +751,21 @@ void Pipe::addChains(std::vector chains) auto added_processors = Chain::getProcessors(std::move(chains[i])); for (auto & transform : added_processors) - { - if (collected_processors) - collected_processors->emplace_back(transform); - - processors->emplace_back(std::move(transform)); - } + addProcessor(std::move(transform)); } header = std::move(new_header); max_parallel_streams = std::max(max_parallel_streams, max_parallel_streams_for_chains); } +void Pipe::addProcessor(ProcessorPtr processor) +{ + if (collected_processors) + collected_processors->emplace_back(processor); + + processors->emplace_back(std::move(processor)); +} + void Pipe::resize(size_t num_streams, bool force, bool strict) { if (output_ports.empty()) diff --git a/src/QueryPipeline/Pipe.h b/src/QueryPipeline/Pipe.h index 09931e38578..10800434678 100644 --- a/src/QueryPipeline/Pipe.h +++ b/src/QueryPipeline/Pipe.h @@ -48,6 +48,9 @@ public: OutputPort * getOutputPort(size_t pos) const { return output_ports[pos]; } OutputPort * getTotalsPort() const { return totals_port; } OutputPort * getExtremesPort() const { return extremes_port; } + OutputPort * getPartialResultPort(size_t pos) const { return partial_result_ports.empty() ? nullptr : partial_result_ports[pos]; } + + bool isPartialResultActive() { return is_partial_result_active; } /// Add processor to list, add it output ports to output_ports. /// Processor shouldn't have input ports, output ports shouldn't be connected. @@ -58,9 +61,13 @@ public: void addTotalsSource(ProcessorPtr source); void addExtremesSource(ProcessorPtr source); - /// Drop totals and extremes (create NullSink for them). + /// Activate sending partial result during main pipeline execution + void activatePartialResult(UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_); + + /// Drop totals, extremes and partial result (create NullSink for them). void dropTotals(); void dropExtremes(); + void dropPartialResult(); /// Add processor to list. It should have size() input ports with compatible header. /// Output ports should have same headers. @@ -69,6 +76,8 @@ public: void addTransform(ProcessorPtr transform, OutputPort * totals, OutputPort * extremes); void addTransform(ProcessorPtr transform, InputPort * totals, InputPort * extremes); + void addPartialResultTransformIfNeeded(ProcessorPtr transform, size_t partial_result_port_id); + enum class StreamType { Main = 0, /// Stream for query data. There may be several streams of this type. @@ -109,10 +118,17 @@ private: Block header; std::shared_ptr processors; - /// Output ports. Totals and extremes are allowed to be empty. + /// If is true, then on each addition of processor also try + /// to add processor which will send partial result from original processor + bool is_partial_result_active = false; + UInt64 partial_result_limit = 0; + UInt64 partial_result_duration_ms = 0; + + /// Output ports. Totals, extremes and partial results are allowed to be empty. OutputPortRawPtrs output_ports; OutputPort * totals_port = nullptr; OutputPort * extremes_port = nullptr; + OutputPortRawPtrs partial_result_ports; /// It is the max number of processors which can be executed in parallel for each step. /// Usually, it's the same as the number of output ports. @@ -127,6 +143,8 @@ private: bool isCompleted() const { return !empty() && output_ports.empty(); } static Pipe unitePipes(Pipes pipes, Processors * collected_processors, bool allow_empty_header); void setSinks(const Pipe::ProcessorGetterWithStreamKind & getter); + + void addProcessor(ProcessorPtr processor); friend class QueryPipelineBuilder; friend class QueryPipeline; diff --git a/src/QueryPipeline/QueryPipeline.cpp b/src/QueryPipeline/QueryPipeline.cpp index f060f2f508f..5acc80dcb18 100644 --- a/src/QueryPipeline/QueryPipeline.cpp +++ b/src/QueryPipeline/QueryPipeline.cpp @@ -63,7 +63,8 @@ static void checkPulling( Processors & processors, OutputPort * output, OutputPort * totals, - OutputPort * extremes) + OutputPort * extremes, + OutputPort * partial_result) { if (!output || output->isConnected()) throw Exception( @@ -80,9 +81,15 @@ static void checkPulling( ErrorCodes::LOGICAL_ERROR, "Cannot create pulling QueryPipeline because its extremes port is connected"); + if (partial_result && partial_result->isConnected()) + throw Exception( + ErrorCodes::LOGICAL_ERROR, + "Cannot create pulling QueryPipeline because its partial_result port is connected"); + bool found_output = false; bool found_totals = false; bool found_extremes = false; + bool found_partial_result = false; for (const auto & processor : processors) { for (const auto & in : processor->getInputs()) @@ -96,6 +103,8 @@ static void checkPulling( found_totals = true; else if (extremes && &out == extremes) found_extremes = true; + else if (partial_result && &out == partial_result) + found_partial_result = true; else checkOutput(out, processor); } @@ -113,6 +122,10 @@ static void checkPulling( throw Exception( ErrorCodes::LOGICAL_ERROR, "Cannot create pulling QueryPipeline because its extremes port does not belong to any processor"); + if (partial_result && !found_partial_result) + throw Exception( + ErrorCodes::LOGICAL_ERROR, + "Cannot create pulling QueryPipeline because its partial_result port does not belong to any processor"); } static void checkCompleted(Processors & processors) @@ -315,14 +328,16 @@ QueryPipeline::QueryPipeline( std::shared_ptr processors_, OutputPort * output_, OutputPort * totals_, - OutputPort * extremes_) + OutputPort * extremes_, + OutputPort * partial_result_) : resources(std::move(resources_)) , processors(std::move(processors_)) , output(output_) , totals(totals_) , extremes(extremes_) + , partial_result(partial_result_) { - checkPulling(*processors, output, totals, extremes); + checkPulling(*processors, output, totals, extremes, partial_result); } QueryPipeline::QueryPipeline(Pipe pipe) @@ -333,9 +348,10 @@ QueryPipeline::QueryPipeline(Pipe pipe) output = pipe.getOutputPort(0); totals = pipe.getTotalsPort(); extremes = pipe.getExtremesPort(); + partial_result = pipe.getPartialResultPort(0); processors = std::move(pipe.processors); - checkPulling(*processors, output, totals, extremes); + checkPulling(*processors, output, totals, extremes, partial_result); } else { @@ -367,6 +383,7 @@ QueryPipeline::QueryPipeline(std::shared_ptr format) auto & format_main = format->getPort(IOutputFormat::PortKind::Main); auto & format_totals = format->getPort(IOutputFormat::PortKind::Totals); auto & format_extremes = format->getPort(IOutputFormat::PortKind::Extremes); + auto & format_partial_result = format->getPort(IOutputFormat::PortKind::PartialResult); if (!totals) { @@ -382,12 +399,21 @@ QueryPipeline::QueryPipeline(std::shared_ptr format) processors->emplace_back(std::move(source)); } + if (!partial_result) + { + auto source = std::make_shared(format_partial_result.getHeader()); + partial_result = &source->getPort(); + processors->emplace_back(std::move(source)); + } + connect(*totals, format_totals); connect(*extremes, format_extremes); + connect(*partial_result, format_partial_result); input = &format_main; totals = nullptr; extremes = nullptr; + partial_result = nullptr; output_format = format.get(); @@ -415,6 +441,7 @@ void QueryPipeline::complete(std::shared_ptr sink) drop(totals, *processors); drop(extremes, *processors); + drop(partial_result, *processors); connect(*output, sink->getPort()); processors->emplace_back(std::move(sink)); @@ -430,6 +457,7 @@ void QueryPipeline::complete(Chain chain) drop(totals, *processors); drop(extremes, *processors); + drop(partial_result, *processors); processors->reserve(processors->size() + chain.getProcessors().size() + 1); for (auto processor : chain.getProcessors()) @@ -455,6 +483,7 @@ void QueryPipeline::complete(Pipe pipe) pipe.resize(1); pipe.dropExtremes(); pipe.dropTotals(); + pipe.dropPartialResult(); connect(*pipe.getOutputPort(0), *input); input = nullptr; @@ -483,11 +512,13 @@ void QueryPipeline::complete(std::shared_ptr format) addMaterializing(output, *processors); addMaterializing(totals, *processors); addMaterializing(extremes, *processors); + addMaterializing(partial_result, *processors); } auto & format_main = format->getPort(IOutputFormat::PortKind::Main); auto & format_totals = format->getPort(IOutputFormat::PortKind::Totals); auto & format_extremes = format->getPort(IOutputFormat::PortKind::Extremes); + auto & format_partial_result = format->getPort(IOutputFormat::PortKind::PartialResult); if (!totals) { @@ -503,13 +534,22 @@ void QueryPipeline::complete(std::shared_ptr format) processors->emplace_back(std::move(source)); } + if (!partial_result) + { + auto source = std::make_shared(format_partial_result.getHeader()); + partial_result = &source->getPort(); + processors->emplace_back(std::move(source)); + } + connect(*output, format_main); connect(*totals, format_totals); connect(*extremes, format_extremes); + connect(*partial_result, format_partial_result); output = nullptr; totals = nullptr; extremes = nullptr; + partial_result = nullptr; initRowsBeforeLimit(format.get()); output_format = format.get(); @@ -642,6 +682,7 @@ void QueryPipeline::convertStructureTo(const ColumnsWithTypeAndName & columns) addExpression(output, actions, *processors); addExpression(totals, actions, *processors); addExpression(extremes, actions, *processors); + addExpression(partial_result, actions, *processors); } std::unique_ptr QueryPipeline::getReadProgressCallback() const diff --git a/src/QueryPipeline/QueryPipeline.h b/src/QueryPipeline/QueryPipeline.h index 55c78ca78ed..177cf0af822 100644 --- a/src/QueryPipeline/QueryPipeline.h +++ b/src/QueryPipeline/QueryPipeline.h @@ -75,7 +75,8 @@ public: std::shared_ptr processors_, OutputPort * output_, OutputPort * totals_ = nullptr, - OutputPort * extremes_ = nullptr); + OutputPort * extremes_ = nullptr, + OutputPort * partial_result_ = nullptr); bool initialized() const { return !processors->empty(); } /// When initialized, exactly one of the following is true. @@ -147,6 +148,7 @@ private: OutputPort * output = nullptr; OutputPort * totals = nullptr; OutputPort * extremes = nullptr; + OutputPort * partial_result = nullptr; QueryStatusPtr process_list_element; diff --git a/src/QueryPipeline/QueryPipelineBuilder.h b/src/QueryPipeline/QueryPipelineBuilder.h index 3a5d65d4388..ed9aa0c3035 100644 --- a/src/QueryPipeline/QueryPipelineBuilder.h +++ b/src/QueryPipeline/QueryPipelineBuilder.h @@ -79,6 +79,12 @@ public: /// Pipeline will be completed after this transformation. void setSinks(const Pipe::ProcessorGetterWithStreamKind & getter); + /// Activate building separate pipeline for sending partial result. + void activatePartialResult(UInt64 partial_result_limit, UInt64 partial_result_duration_ms) { pipe.activatePartialResult(partial_result_limit, partial_result_duration_ms); } + + /// Check if building of a pipeline for sending partial result active. + bool isPartialResultActive() { return pipe.isPartialResultActive(); } + /// Add totals which returns one chunk with single row with defaults. void addDefaultTotals(); From 6a32bbfcd85f428cf21729778f5f2231e60c148c Mon Sep 17 00:00:00 2001 From: alexX512 Date: Mon, 1 May 2023 13:27:08 +0000 Subject: [PATCH 007/433] Fix style check --- src/Processors/IProcessor.h | 2 +- src/Processors/Transforms/LimitPartialResultTransform.h | 2 +- src/Processors/Transforms/MergeSortingTransform.h | 2 +- src/Processors/Transforms/PartialResultTransform.cpp | 4 ++-- src/Processors/Transforms/PartialResultTransform.h | 2 +- src/QueryPipeline/Pipe.cpp | 7 +++++++ 6 files changed, 13 insertions(+), 6 deletions(-) diff --git a/src/Processors/IProcessor.h b/src/Processors/IProcessor.h index cb819c00ff2..26a1ce3cddc 100644 --- a/src/Processors/IProcessor.h +++ b/src/Processors/IProcessor.h @@ -234,7 +234,7 @@ public: { throw Exception(ErrorCodes::NOT_IMPLEMENTED, "Method 'expandPipeline' is not implemented for {} processor", getName()); } - + virtual bool isPartialResultProcessor() const { return false; } virtual bool supportPartialResultProcessor() const { return false; } diff --git a/src/Processors/Transforms/LimitPartialResultTransform.h b/src/Processors/Transforms/LimitPartialResultTransform.h index 8a919615088..e5d772db289 100644 --- a/src/Processors/Transforms/LimitPartialResultTransform.h +++ b/src/Processors/Transforms/LimitPartialResultTransform.h @@ -43,7 +43,7 @@ public: chunk.setColumns(std::move(columns), length); } - } + } private: UInt64 limit; diff --git a/src/Processors/Transforms/MergeSortingTransform.h b/src/Processors/Transforms/MergeSortingTransform.h index 0b9766674a5..0c42798c0a5 100644 --- a/src/Processors/Transforms/MergeSortingTransform.h +++ b/src/Processors/Transforms/MergeSortingTransform.h @@ -62,7 +62,7 @@ private: /// Merge all accumulated blocks to keep no more than limit rows. void remerge(); - friend class MergeSortingPartialResult; + friend class MergeSortingPartialResultTransform; std::mutex snapshot_mutex; ProcessorPtr external_merging_sorted; diff --git a/src/Processors/Transforms/PartialResultTransform.cpp b/src/Processors/Transforms/PartialResultTransform.cpp index 4fda35085c3..7cf2473684a 100644 --- a/src/Processors/Transforms/PartialResultTransform.cpp +++ b/src/Processors/Transforms/PartialResultTransform.cpp @@ -19,7 +19,7 @@ IProcessor::Status PartialResultTransform::prepare() input.close(); return Status::Finished; } - + if (finished_getting_snapshots) { output.finish(); @@ -40,7 +40,7 @@ IProcessor::Status PartialResultTransform::prepare() return Status::PortFull; } } - + /// If input data from previous partial result processor is finished then /// PartialResultTransform ready to create snapshots and send them as a partial result if (input.isFinished()) diff --git a/src/Processors/Transforms/PartialResultTransform.h b/src/Processors/Transforms/PartialResultTransform.h index 0e3c503968f..3f9089f8cce 100644 --- a/src/Processors/Transforms/PartialResultTransform.h +++ b/src/Processors/Transforms/PartialResultTransform.h @@ -38,7 +38,7 @@ protected: UInt64 partial_result_duration_ms; ShaphotResult partial_result = {{}, SnaphotStatus::NotReady}; - + bool finished_getting_snapshots = false; virtual void transformPartialResult(Chunk & /*chunk*/) {} diff --git a/src/QueryPipeline/Pipe.cpp b/src/QueryPipeline/Pipe.cpp index b84cb279acd..b1a740b3390 100644 --- a/src/QueryPipeline/Pipe.cpp +++ b/src/QueryPipeline/Pipe.cpp @@ -483,6 +483,9 @@ void Pipe::addTransform(ProcessorPtr transform, OutputPort * totals, OutputPort if (extremes) extremes_port = extremes; + /// TODO: Add support for partial result in multithreading mode + dropPartialResult(); + size_t next_output = 0; for (auto & input : inputs) { @@ -570,6 +573,9 @@ void Pipe::addTransform(ProcessorPtr transform, InputPort * totals, InputPort * extremes_port = nullptr; } + /// TODO: Add support for partial result in multithreading mode + dropPartialResult(); + bool found_totals = false; bool found_extremes = false; @@ -733,6 +739,7 @@ void Pipe::addChains(std::vector chains) dropTotals(); dropExtremes(); + dropPartialResult(); size_t max_parallel_streams_for_chains = 0; From c0c8fa7b8e419fd8696d23da2783c4e350664cba Mon Sep 17 00:00:00 2001 From: alexX512 Date: Mon, 1 May 2023 13:42:26 +0000 Subject: [PATCH 008/433] style fix --- src/QueryPipeline/Pipe.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/QueryPipeline/Pipe.h b/src/QueryPipeline/Pipe.h index 10800434678..7200cf4944a 100644 --- a/src/QueryPipeline/Pipe.h +++ b/src/QueryPipeline/Pipe.h @@ -143,7 +143,7 @@ private: bool isCompleted() const { return !empty() && output_ports.empty(); } static Pipe unitePipes(Pipes pipes, Processors * collected_processors, bool allow_empty_header); void setSinks(const Pipe::ProcessorGetterWithStreamKind & getter); - + void addProcessor(ProcessorPtr processor); friend class QueryPipelineBuilder; From 2cd1b37f9b5b6576478a44846f5fa4b78de8f381 Mon Sep 17 00:00:00 2001 From: alexX512 Date: Tue, 2 May 2023 07:26:36 +0000 Subject: [PATCH 009/433] Add support for connection of partial result processors with multiple ports --- .../PullingAsyncPipelineExecutor.cpp | 5 +- .../Executors/PullingAsyncPipelineExecutor.h | 2 +- src/Processors/Formats/IOutputFormat.cpp | 2 +- src/Processors/Formats/IOutputFormat.h | 4 + .../Transforms/LimitPartialResultTransform.h | 1 + src/QueryPipeline/Pipe.cpp | 82 +++++++++++++++---- src/QueryPipeline/Pipe.h | 4 +- src/Server/TCPHandler.cpp | 3 +- 8 files changed, 82 insertions(+), 21 deletions(-) diff --git a/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp b/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp index 73a3142d459..0b28ed67cd1 100644 --- a/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp +++ b/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp @@ -41,12 +41,15 @@ struct PullingAsyncPipelineExecutor::Data } }; -PullingAsyncPipelineExecutor::PullingAsyncPipelineExecutor(QueryPipeline & pipeline_) : pipeline(pipeline_) +PullingAsyncPipelineExecutor::PullingAsyncPipelineExecutor(QueryPipeline & pipeline_, bool /*has_partial_result_setting*/) : pipeline(pipeline_) { if (!pipeline.pulling()) throw Exception(ErrorCodes::LOGICAL_ERROR, "Pipeline for PullingAsyncPipelineExecutor must be pulling"); lazy_format = std::make_shared(pipeline.output->getHeader()); + // if (has_partial_result_setting) + // lazy_format->activatePartialResultProtocol(); + pipeline.complete(lazy_format); } diff --git a/src/Processors/Executors/PullingAsyncPipelineExecutor.h b/src/Processors/Executors/PullingAsyncPipelineExecutor.h index 361bcc0155c..202ecbf281b 100644 --- a/src/Processors/Executors/PullingAsyncPipelineExecutor.h +++ b/src/Processors/Executors/PullingAsyncPipelineExecutor.h @@ -21,7 +21,7 @@ struct ProfileInfo; class PullingAsyncPipelineExecutor { public: - explicit PullingAsyncPipelineExecutor(QueryPipeline & pipeline_); + explicit PullingAsyncPipelineExecutor(QueryPipeline & pipeline_, bool has_partial_result_setting = false); ~PullingAsyncPipelineExecutor(); /// Get structure of returned block or chunk. diff --git a/src/Processors/Formats/IOutputFormat.cpp b/src/Processors/Formats/IOutputFormat.cpp index d3e0ed2fcc6..6496f5357e7 100644 --- a/src/Processors/Formats/IOutputFormat.cpp +++ b/src/Processors/Formats/IOutputFormat.cpp @@ -129,7 +129,7 @@ void IOutputFormat::work() case Main: result_rows += current_chunk.getNumRows(); result_bytes += current_chunk.allocatedBytes(); - if (!was_main_input && current_chunk.hasRows()) + if (is_partial_result_protocol_active && !was_main_input && current_chunk.hasRows()) { consume(Chunk(current_chunk.cloneEmptyColumns(), 0)); was_main_input = true; diff --git a/src/Processors/Formats/IOutputFormat.h b/src/Processors/Formats/IOutputFormat.h index 2c598e2620b..6e5e21055a3 100644 --- a/src/Processors/Formats/IOutputFormat.h +++ b/src/Processors/Formats/IOutputFormat.h @@ -105,6 +105,8 @@ public: void clearLastLines(size_t lines_number); + void activatePartialResultProtocol() { is_partial_result_protocol_active = true; } + protected: friend class ParallelFormattingOutputFormat; @@ -192,6 +194,8 @@ private: size_t rows_read_before = 0; bool are_totals_written = false; + bool is_partial_result_protocol_active = false; + /// Counters for consumed chunks. Are used for QueryLog. size_t result_rows = 0; size_t result_bytes = 0; diff --git a/src/Processors/Transforms/LimitPartialResultTransform.h b/src/Processors/Transforms/LimitPartialResultTransform.h index e5d772db289..8ed60d20da3 100644 --- a/src/Processors/Transforms/LimitPartialResultTransform.h +++ b/src/Processors/Transforms/LimitPartialResultTransform.h @@ -5,6 +5,7 @@ namespace DB { +/// Currently support only single thread implementation with one input and one output ports class LimitPartialResultTransform : public PartialResultTransform { public: diff --git a/src/QueryPipeline/Pipe.cpp b/src/QueryPipeline/Pipe.cpp index b1a740b3390..ff86d8df776 100644 --- a/src/QueryPipeline/Pipe.cpp +++ b/src/QueryPipeline/Pipe.cpp @@ -483,8 +483,7 @@ void Pipe::addTransform(ProcessorPtr transform, OutputPort * totals, OutputPort if (extremes) extremes_port = extremes; - /// TODO: Add support for partial result in multithreading mode - dropPartialResult(); + addPartialResultTransform(transform); size_t next_output = 0; for (auto & input : inputs) @@ -573,8 +572,7 @@ void Pipe::addTransform(ProcessorPtr transform, InputPort * totals, InputPort * extremes_port = nullptr; } - /// TODO: Add support for partial result in multithreading mode - dropPartialResult(); + addPartialResultTransform(transform); bool found_totals = false; bool found_extremes = false; @@ -627,11 +625,10 @@ void Pipe::addTransform(ProcessorPtr transform, InputPort * totals, InputPort * addProcessor(std::move(transform)); - max_parallel_streams = std::max(max_parallel_streams, output_ports.size()); } -void Pipe::addPartialResultTransformIfNeeded(ProcessorPtr transform, size_t partial_result_port_id) +void Pipe::addPartialResultSimpleTransform(ProcessorPtr transform, size_t partial_result_port_id) { if (isPartialResultActive()) { @@ -643,23 +640,75 @@ void Pipe::addPartialResultTransformIfNeeded(ProcessorPtr transform, size_t part return; } - if (partial_result_port == nullptr) - { - auto source = std::make_shared(getHeader()); - partial_result_port = &source->getPort(); - - addProcessor(std::move(source)); - } - auto partial_result_transform = transform->getPartialResultProcessor(std::move(transform), partial_result_limit, partial_result_duration_ms); - connect(*partial_result_port, partial_result_transform->getInputs().front()); + connectPartialResultPort(partial_result_port, partial_result_transform->getInputs().front()); + partial_result_port = &partial_result_transform->getOutputs().front(); addProcessor(std::move(partial_result_transform)); } } +void Pipe::addPartialResultTransform(ProcessorPtr transform) +{ + if (isPartialResultActive()) + { + size_t new_outputs_size = transform->getOutputs().size(); + + if (!transform->supportPartialResultProcessor()) + { + for (auto & partial_result_port : partial_result_ports) + dropPort(partial_result_port, *processors, collected_processors); + + partial_result_ports.assign(new_outputs_size, nullptr); + return; + } + + auto partial_result_transform = transform->getPartialResultProcessor(std::move(transform), partial_result_limit, partial_result_duration_ms); + auto & inputs = partial_result_transform->getInputs(); + + if (inputs.size() != partial_result_ports.size()) + throw Exception( + ErrorCodes::LOGICAL_ERROR, + "Cannot add transform {} to Pipe because it has {} input ports, but {} expected", + partial_result_transform->getName(), + inputs.size(), + partial_result_ports.size()); + + size_t next_port = 0; + for (auto & input : inputs) + { + connectPartialResultPort(partial_result_ports[next_port], input); + ++next_port; + } + + partial_result_ports.assign(new_outputs_size, nullptr); + + next_port = 0; + for (auto & new_partial_result_port : partial_result_transform->getOutputs()) + { + partial_result_ports[next_port] = &new_partial_result_port; + ++next_port; + } + + addProcessor(std::move(partial_result_transform)); + } +} + +void Pipe::connectPartialResultPort(OutputPort * partial_result_port, InputPort & partial_result_transform_port) +{ + if (partial_result_port == nullptr) + { + auto source = std::make_shared(getHeader()); + partial_result_port = &source->getPort(); + + addProcessor(std::move(source)); + } + + connect(*partial_result_port, partial_result_transform_port); +} + void Pipe::addSimpleTransform(const ProcessorGetterWithStreamKind & getter) { if (output_ports.empty()) @@ -704,7 +753,7 @@ void Pipe::addSimpleTransform(const ProcessorGetterWithStreamKind & getter) connect(*port, transform->getInputs().front()); port = &transform->getOutputs().front(); if (stream_type == StreamType::Main) - addPartialResultTransformIfNeeded(transform, partial_result_port_id); + addPartialResultSimpleTransform(transform, partial_result_port_id); addProcessor(std::move(transform)); } @@ -832,6 +881,7 @@ void Pipe::setSinks(const Pipe::ProcessorGetterWithStreamKind & getter) add_transform(totals_port, StreamType::Totals); add_transform(extremes_port, StreamType::Extremes); + dropPartialResult(); output_ports.clear(); header.clear(); diff --git a/src/QueryPipeline/Pipe.h b/src/QueryPipeline/Pipe.h index 7200cf4944a..650e4d664a5 100644 --- a/src/QueryPipeline/Pipe.h +++ b/src/QueryPipeline/Pipe.h @@ -76,7 +76,9 @@ public: void addTransform(ProcessorPtr transform, OutputPort * totals, OutputPort * extremes); void addTransform(ProcessorPtr transform, InputPort * totals, InputPort * extremes); - void addPartialResultTransformIfNeeded(ProcessorPtr transform, size_t partial_result_port_id); + void addPartialResultTransform(ProcessorPtr transform); + void addPartialResultSimpleTransform(ProcessorPtr transform, size_t partial_result_port_id); + void connectPartialResultPort(OutputPort * partial_result_port, InputPort & partial_result_transform_port); enum class StreamType { diff --git a/src/Server/TCPHandler.cpp b/src/Server/TCPHandler.cpp index a608219ed63..67ed1d071e4 100644 --- a/src/Server/TCPHandler.cpp +++ b/src/Server/TCPHandler.cpp @@ -815,7 +815,8 @@ void TCPHandler::processOrdinaryQueryWithProcessors() std::unique_lock progress_lock(task_callback_mutex, std::defer_lock); { - PullingAsyncPipelineExecutor executor(pipeline); + bool has_partial_result_setting = query_context->getSettingsRef().partial_result_update_duration_ms.totalMilliseconds() > 0; + PullingAsyncPipelineExecutor executor(pipeline, has_partial_result_setting); CurrentMetrics::Increment query_thread_metric_increment{CurrentMetrics::QueryThread}; Block block; From 40c0b461bf7a421e71076f042414e3eff60fa722 Mon Sep 17 00:00:00 2001 From: alexX512 Date: Tue, 2 May 2023 07:37:01 +0000 Subject: [PATCH 010/433] Style fix --- src/QueryPipeline/Pipe.cpp | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/QueryPipeline/Pipe.cpp b/src/QueryPipeline/Pipe.cpp index ff86d8df776..21eb5c92d4f 100644 --- a/src/QueryPipeline/Pipe.cpp +++ b/src/QueryPipeline/Pipe.cpp @@ -660,11 +660,11 @@ void Pipe::addPartialResultTransform(ProcessorPtr transform) { for (auto & partial_result_port : partial_result_ports) dropPort(partial_result_port, *processors, collected_processors); - + partial_result_ports.assign(new_outputs_size, nullptr); return; } - + auto partial_result_transform = transform->getPartialResultProcessor(std::move(transform), partial_result_limit, partial_result_duration_ms); auto & inputs = partial_result_transform->getInputs(); From 9fa2c8f652ff19c9b82fa0b359011adfc2f6f990 Mon Sep 17 00:00:00 2001 From: alexX512 Date: Tue, 2 May 2023 18:16:15 +0000 Subject: [PATCH 011/433] Update test with a new pipeline format --- tests/queries/0_stateless/02210_processors_profile_log.reference | 1 + 1 file changed, 1 insertion(+) diff --git a/tests/queries/0_stateless/02210_processors_profile_log.reference b/tests/queries/0_stateless/02210_processors_profile_log.reference index 181022d2421..4592d13c9a5 100644 --- a/tests/queries/0_stateless/02210_processors_profile_log.reference +++ b/tests/queries/0_stateless/02210_processors_profile_log.reference @@ -38,4 +38,5 @@ LazyOutputFormat 1 1 1 0 0 LimitsCheckingTransform 1 1 1 1 1 NullSource 1 0 0 0 0 NullSource 1 0 0 0 0 +NullSource 0 0 0 0 0 SourceFromSingleChunk 1 0 0 1 1 From ecd93bee6b1e10e6783a430809563ece313dcd87 Mon Sep 17 00:00:00 2001 From: alexX512 Date: Wed, 3 May 2023 06:44:32 +0000 Subject: [PATCH 012/433] Change copy of shared pointer to const reference of shared pointer --- src/Processors/IProcessor.h | 2 +- src/Processors/LimitTransform.cpp | 2 +- src/Processors/LimitTransform.h | 2 +- src/Processors/Transforms/ExpressionTransform.cpp | 2 +- src/Processors/Transforms/ExpressionTransform.h | 2 +- src/Processors/Transforms/LimitsCheckingTransform.cpp | 2 +- src/Processors/Transforms/LimitsCheckingTransform.h | 2 +- src/Processors/Transforms/MergeSortingTransform.cpp | 4 ++-- src/Processors/Transforms/MergeSortingTransform.h | 2 +- src/QueryPipeline/Pipe.cpp | 8 ++++---- src/QueryPipeline/Pipe.h | 4 ++-- 11 files changed, 16 insertions(+), 16 deletions(-) diff --git a/src/Processors/IProcessor.h b/src/Processors/IProcessor.h index 26a1ce3cddc..dd08075fd01 100644 --- a/src/Processors/IProcessor.h +++ b/src/Processors/IProcessor.h @@ -238,7 +238,7 @@ public: virtual bool isPartialResultProcessor() const { return false; } virtual bool supportPartialResultProcessor() const { return false; } - virtual ProcessorPtr getPartialResultProcessor(ProcessorPtr /*current_processor*/, UInt64 /*partial_result_limit*/, UInt64 /*partial_result_duration_ms*/) + virtual ProcessorPtr getPartialResultProcessor(const ProcessorPtr & /*current_processor*/, UInt64 /*partial_result_limit*/, UInt64 /*partial_result_duration_ms*/) { throw Exception(ErrorCodes::NOT_IMPLEMENTED, "Method 'getPartialResultProcessor' is not implemented for {} processor", getName()); } diff --git a/src/Processors/LimitTransform.cpp b/src/Processors/LimitTransform.cpp index 0d43b1e4fac..b2bf3c28eee 100644 --- a/src/Processors/LimitTransform.cpp +++ b/src/Processors/LimitTransform.cpp @@ -366,7 +366,7 @@ bool LimitTransform::sortColumnsEqualAt(const ColumnRawPtrs & current_chunk_sort return true; } -ProcessorPtr LimitTransform::getPartialResultProcessor(ProcessorPtr /*current_processor*/, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) +ProcessorPtr LimitTransform::getPartialResultProcessor(const ProcessorPtr & /*current_processor*/, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) { const auto & header = inputs.front().getHeader(); return std::make_shared(header, partial_result_limit, partial_result_duration_ms, limit, offset); diff --git a/src/Processors/LimitTransform.h b/src/Processors/LimitTransform.h index 7ad358c5d41..b0ec7600406 100644 --- a/src/Processors/LimitTransform.h +++ b/src/Processors/LimitTransform.h @@ -56,7 +56,7 @@ private: bool sortColumnsEqualAt(const ColumnRawPtrs & current_chunk_sort_columns, UInt64 current_chunk_row_num) const; bool supportPartialResultProcessor() const override { return true; } - ProcessorPtr getPartialResultProcessor(ProcessorPtr current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) override; + ProcessorPtr getPartialResultProcessor(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) override; public: LimitTransform( diff --git a/src/Processors/Transforms/ExpressionTransform.cpp b/src/Processors/Transforms/ExpressionTransform.cpp index 5ec76dc95b8..da03e882931 100644 --- a/src/Processors/Transforms/ExpressionTransform.cpp +++ b/src/Processors/Transforms/ExpressionTransform.cpp @@ -25,7 +25,7 @@ void ExpressionTransform::transform(Chunk & chunk) chunk.setColumns(block.getColumns(), num_rows); } -ProcessorPtr ExpressionTransform::getPartialResultProcessor(ProcessorPtr /*current_processor*/, UInt64 /*partial_result_limit*/, UInt64 /*partial_result_duration_ms*/) +ProcessorPtr ExpressionTransform::getPartialResultProcessor(const ProcessorPtr & /*current_processor*/, UInt64 /*partial_result_limit*/, UInt64 /*partial_result_duration_ms*/) { const auto & header = getInputPort().getHeader(); return std::make_shared(header, expression); diff --git a/src/Processors/Transforms/ExpressionTransform.h b/src/Processors/Transforms/ExpressionTransform.h index 6f9b585ae1f..bf8425b2300 100644 --- a/src/Processors/Transforms/ExpressionTransform.h +++ b/src/Processors/Transforms/ExpressionTransform.h @@ -30,7 +30,7 @@ protected: void transform(Chunk & chunk) override; bool supportPartialResultProcessor() const override { return true; } - ProcessorPtr getPartialResultProcessor(ProcessorPtr current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) override; + ProcessorPtr getPartialResultProcessor(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) override; private: ExpressionActionsPtr expression; diff --git a/src/Processors/Transforms/LimitsCheckingTransform.cpp b/src/Processors/Transforms/LimitsCheckingTransform.cpp index b426d050543..487592f9809 100644 --- a/src/Processors/Transforms/LimitsCheckingTransform.cpp +++ b/src/Processors/Transforms/LimitsCheckingTransform.cpp @@ -75,7 +75,7 @@ void LimitsCheckingTransform::checkQuota(Chunk & chunk) } } -ProcessorPtr LimitsCheckingTransform::getPartialResultProcessor(ProcessorPtr /*current_processor*/, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) +ProcessorPtr LimitsCheckingTransform::getPartialResultProcessor(const ProcessorPtr & /*current_processor*/, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) { const auto & header = inputs.front().getHeader(); return std::make_shared(header, partial_result_limit, partial_result_duration_ms); diff --git a/src/Processors/Transforms/LimitsCheckingTransform.h b/src/Processors/Transforms/LimitsCheckingTransform.h index 663eb774edf..6e8d5547cc5 100644 --- a/src/Processors/Transforms/LimitsCheckingTransform.h +++ b/src/Processors/Transforms/LimitsCheckingTransform.h @@ -37,7 +37,7 @@ protected: void transform(Chunk & chunk) override; bool supportPartialResultProcessor() const override { return true; } - ProcessorPtr getPartialResultProcessor(ProcessorPtr current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) override; + ProcessorPtr getPartialResultProcessor(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) override; private: StreamLocalLimits limits; diff --git a/src/Processors/Transforms/MergeSortingTransform.cpp b/src/Processors/Transforms/MergeSortingTransform.cpp index c4c6743c775..23096f785ad 100644 --- a/src/Processors/Transforms/MergeSortingTransform.cpp +++ b/src/Processors/Transforms/MergeSortingTransform.cpp @@ -333,7 +333,7 @@ void MergeSortingTransform::remerge() sum_bytes_in_blocks = new_sum_bytes_in_blocks; } -ProcessorPtr MergeSortingTransform::getPartialResultProcessor(ProcessorPtr current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) +ProcessorPtr MergeSortingTransform::getPartialResultProcessor(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) { if (getName() != current_processor->getName() || current_processor.get() != this) throw Exception( @@ -347,7 +347,7 @@ ProcessorPtr MergeSortingTransform::getPartialResultProcessor(ProcessorPtr curre static_cast(current_processor.get())); const auto & header = inputs.front().getHeader(); - const auto & merge_sorting_processor = std::static_pointer_cast(current_processor); + auto merge_sorting_processor = std::dynamic_pointer_cast(current_processor); return std::make_shared(header, std::move(merge_sorting_processor), partial_result_limit, partial_result_duration_ms); } diff --git a/src/Processors/Transforms/MergeSortingTransform.h b/src/Processors/Transforms/MergeSortingTransform.h index 0c42798c0a5..17f79efe5d6 100644 --- a/src/Processors/Transforms/MergeSortingTransform.h +++ b/src/Processors/Transforms/MergeSortingTransform.h @@ -42,7 +42,7 @@ protected: Processors expandPipeline() override; bool supportPartialResultProcessor() const override { return true; } - ProcessorPtr getPartialResultProcessor(ProcessorPtr current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) override; + ProcessorPtr getPartialResultProcessor(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) override; private: size_t max_bytes_before_remerge; diff --git a/src/QueryPipeline/Pipe.cpp b/src/QueryPipeline/Pipe.cpp index 21eb5c92d4f..6de45d67133 100644 --- a/src/QueryPipeline/Pipe.cpp +++ b/src/QueryPipeline/Pipe.cpp @@ -628,7 +628,7 @@ void Pipe::addTransform(ProcessorPtr transform, InputPort * totals, InputPort * max_parallel_streams = std::max(max_parallel_streams, output_ports.size()); } -void Pipe::addPartialResultSimpleTransform(ProcessorPtr transform, size_t partial_result_port_id) +void Pipe::addPartialResultSimpleTransform(const ProcessorPtr & transform, size_t partial_result_port_id) { if (isPartialResultActive()) { @@ -640,7 +640,7 @@ void Pipe::addPartialResultSimpleTransform(ProcessorPtr transform, size_t partia return; } - auto partial_result_transform = transform->getPartialResultProcessor(std::move(transform), partial_result_limit, partial_result_duration_ms); + auto partial_result_transform = transform->getPartialResultProcessor(transform, partial_result_limit, partial_result_duration_ms); connectPartialResultPort(partial_result_port, partial_result_transform->getInputs().front()); @@ -650,7 +650,7 @@ void Pipe::addPartialResultSimpleTransform(ProcessorPtr transform, size_t partia } } -void Pipe::addPartialResultTransform(ProcessorPtr transform) +void Pipe::addPartialResultTransform(const ProcessorPtr & transform) { if (isPartialResultActive()) { @@ -665,7 +665,7 @@ void Pipe::addPartialResultTransform(ProcessorPtr transform) return; } - auto partial_result_transform = transform->getPartialResultProcessor(std::move(transform), partial_result_limit, partial_result_duration_ms); + auto partial_result_transform = transform->getPartialResultProcessor(transform, partial_result_limit, partial_result_duration_ms); auto & inputs = partial_result_transform->getInputs(); if (inputs.size() != partial_result_ports.size()) diff --git a/src/QueryPipeline/Pipe.h b/src/QueryPipeline/Pipe.h index 650e4d664a5..8b26435fd30 100644 --- a/src/QueryPipeline/Pipe.h +++ b/src/QueryPipeline/Pipe.h @@ -76,8 +76,8 @@ public: void addTransform(ProcessorPtr transform, OutputPort * totals, OutputPort * extremes); void addTransform(ProcessorPtr transform, InputPort * totals, InputPort * extremes); - void addPartialResultTransform(ProcessorPtr transform); - void addPartialResultSimpleTransform(ProcessorPtr transform, size_t partial_result_port_id); + void addPartialResultTransform(const ProcessorPtr & transform); + void addPartialResultSimpleTransform(const ProcessorPtr & transform, size_t partial_result_port_id); void connectPartialResultPort(OutputPort * partial_result_port, InputPort & partial_result_transform_port); enum class StreamType From 7e3b0c5f86046a03f30b484cc99481269d4dfebd Mon Sep 17 00:00:00 2001 From: Boris Kuschel Date: Thu, 8 Jun 2023 08:43:22 -0700 Subject: [PATCH 013/433] Add new exceptions to 4xx error --- src/Server/HTTPHandler.cpp | 16 +++++++++++++--- 1 file changed, 13 insertions(+), 3 deletions(-) diff --git a/src/Server/HTTPHandler.cpp b/src/Server/HTTPHandler.cpp index 5a2bf0bad6c..09063d93a59 100644 --- a/src/Server/HTTPHandler.cpp +++ b/src/Server/HTTPHandler.cpp @@ -70,12 +70,15 @@ namespace ErrorCodes extern const int CANNOT_PARSE_INPUT_ASSERTION_FAILED; extern const int CANNOT_OPEN_FILE; extern const int CANNOT_COMPILE_REGEXP; - + extern const int DUPLICATE_COLUMN; + extern const int ILLEGAL_COLUMN; + extern const int THERE_IS_NO_COLUMN; extern const int UNKNOWN_ELEMENT_IN_AST; extern const int UNKNOWN_TYPE_OF_AST_NODE; extern const int TOO_DEEP_AST; extern const int TOO_BIG_AST; extern const int UNEXPECTED_AST_STRUCTURE; + extern const int VALUE_IS_OUT_OF_RANGE_OF_DATA_TYPE; extern const int SYNTAX_ERROR; @@ -186,7 +189,9 @@ static Poco::Net::HTTPResponse::HTTPStatus exceptionCodeToHTTPStatus(int excepti { return HTTPResponse::HTTP_FORBIDDEN; } - else if (exception_code == ErrorCodes::CANNOT_PARSE_TEXT || + else if (exception_code == ErrorCodes::BAD_ARGUMENTS || + exception_code == ErrorCodes::CANNOT_COMPILE_REGEXP || + exception_code == ErrorCodes::CANNOT_PARSE_TEXT || exception_code == ErrorCodes::CANNOT_PARSE_ESCAPE_SEQUENCE || exception_code == ErrorCodes::CANNOT_PARSE_QUOTED_STRING || exception_code == ErrorCodes::CANNOT_PARSE_DATE || @@ -196,14 +201,19 @@ static Poco::Net::HTTPResponse::HTTPStatus exceptionCodeToHTTPStatus(int excepti exception_code == ErrorCodes::CANNOT_PARSE_IPV4 || exception_code == ErrorCodes::CANNOT_PARSE_IPV6 || exception_code == ErrorCodes::CANNOT_PARSE_INPUT_ASSERTION_FAILED || + exception_code == ErrorCodes::CANNOT_PARSE_UUID || + exception_code == ErrorCodes::DUPLICATE_COLUMN || + exception_code == ErrorCodes::ILLEGAL_COLUMN || exception_code == ErrorCodes::UNKNOWN_ELEMENT_IN_AST || exception_code == ErrorCodes::UNKNOWN_TYPE_OF_AST_NODE || + exception_code == ErrorCodes::THERE_IS_NO_COLUMN || exception_code == ErrorCodes::TOO_DEEP_AST || exception_code == ErrorCodes::TOO_BIG_AST || exception_code == ErrorCodes::UNEXPECTED_AST_STRUCTURE || exception_code == ErrorCodes::SYNTAX_ERROR || exception_code == ErrorCodes::INCORRECT_DATA || - exception_code == ErrorCodes::TYPE_MISMATCH) + exception_code == ErrorCodes::TYPE_MISMATCH || + exception_code == ErrorCodes::VALUE_IS_OUT_OF_RANGE_OF_DATA_TYPE) { return HTTPResponse::HTTP_BAD_REQUEST; } From ae28c549a76f500a72426211a75ef2cfbb7eb7cd Mon Sep 17 00:00:00 2001 From: Boris Kuschel Date: Fri, 9 Jun 2023 09:10:06 -0700 Subject: [PATCH 014/433] Fix style --- src/Server/HTTPHandler.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/Server/HTTPHandler.cpp b/src/Server/HTTPHandler.cpp index 09063d93a59..c99134c86d2 100644 --- a/src/Server/HTTPHandler.cpp +++ b/src/Server/HTTPHandler.cpp @@ -212,7 +212,7 @@ static Poco::Net::HTTPResponse::HTTPStatus exceptionCodeToHTTPStatus(int excepti exception_code == ErrorCodes::UNEXPECTED_AST_STRUCTURE || exception_code == ErrorCodes::SYNTAX_ERROR || exception_code == ErrorCodes::INCORRECT_DATA || - exception_code == ErrorCodes::TYPE_MISMATCH || + exception_code == ErrorCodes::TYPE_MISMATCH || exception_code == ErrorCodes::VALUE_IS_OUT_OF_RANGE_OF_DATA_TYPE) { return HTTPResponse::HTTP_BAD_REQUEST; From 003565d08e1a4bf0d29711edbf8fe32a2d7ca36a Mon Sep 17 00:00:00 2001 From: Boris Kuschel Date: Mon, 12 Jun 2023 07:28:21 -0700 Subject: [PATCH 015/433] Add undefs --- src/Server/HTTPHandler.cpp | 2 ++ 1 file changed, 2 insertions(+) diff --git a/src/Server/HTTPHandler.cpp b/src/Server/HTTPHandler.cpp index c99134c86d2..9906438f995 100644 --- a/src/Server/HTTPHandler.cpp +++ b/src/Server/HTTPHandler.cpp @@ -57,6 +57,7 @@ namespace DB namespace ErrorCodes { + extern const int BAD_ARGUMENTS; extern const int LOGICAL_ERROR; extern const int CANNOT_PARSE_TEXT; extern const int CANNOT_PARSE_ESCAPE_SEQUENCE; @@ -67,6 +68,7 @@ namespace ErrorCodes extern const int CANNOT_PARSE_DOMAIN_VALUE_FROM_STRING; extern const int CANNOT_PARSE_IPV4; extern const int CANNOT_PARSE_IPV6; + extern const int CANNOT_PARSE_UUID; extern const int CANNOT_PARSE_INPUT_ASSERTION_FAILED; extern const int CANNOT_OPEN_FILE; extern const int CANNOT_COMPILE_REGEXP; From 25053323d9f88e3794bfb0dbafbc72370548aea3 Mon Sep 17 00:00:00 2001 From: Nikolai Kochetov Date: Mon, 26 Jun 2023 15:43:55 +0000 Subject: [PATCH 016/433] Test libunwind changes. --- contrib/libunwind | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/contrib/libunwind b/contrib/libunwind index e48aa13f67d..09f93176072 160000 --- a/contrib/libunwind +++ b/contrib/libunwind @@ -1 +1 @@ -Subproject commit e48aa13f67dc722511b5af33a32ba9b7748176b5 +Subproject commit 09f931760727a200c24e11f06ff0ad254502a304 From 16ef88c346f4c79e74a978c4cfb95d5879600068 Mon Sep 17 00:00:00 2001 From: Nikolai Kochetov Date: Tue, 27 Jun 2023 15:11:18 +0000 Subject: [PATCH 017/433] Update libunwind --- contrib/libunwind | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/contrib/libunwind b/contrib/libunwind index 09f93176072..ad41616855c 160000 --- a/contrib/libunwind +++ b/contrib/libunwind @@ -1 +1 @@ -Subproject commit 09f931760727a200c24e11f06ff0ad254502a304 +Subproject commit ad41616855ce75b768a5ebb6975003a6ac959250 From 62feaf6e54cc00e5efc33e04e90432d46705b2e2 Mon Sep 17 00:00:00 2001 From: Nikolai Kochetov Date: Tue, 27 Jun 2023 18:48:19 +0000 Subject: [PATCH 018/433] Update contrib. --- contrib/libunwind | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/contrib/libunwind b/contrib/libunwind index ad41616855c..2a82f19272f 160000 --- a/contrib/libunwind +++ b/contrib/libunwind @@ -1 +1 @@ -Subproject commit ad41616855ce75b768a5ebb6975003a6ac959250 +Subproject commit 2a82f19272f37d345001f6568ee32655aeb2e54d From 48c3c5a347236f741a449cf1414aac3850bc877e Mon Sep 17 00:00:00 2001 From: CuiShuoGuo <129303239+bakam412@users.noreply.github.com> Date: Thu, 20 Jul 2023 11:25:18 +0800 Subject: [PATCH 019/433] Update drop.md --- docs/en/sql-reference/statements/drop.md | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/docs/en/sql-reference/statements/drop.md b/docs/en/sql-reference/statements/drop.md index b6208c2fd52..a8658dd49fc 100644 --- a/docs/en/sql-reference/statements/drop.md +++ b/docs/en/sql-reference/statements/drop.md @@ -32,6 +32,14 @@ Syntax: DROP [TEMPORARY] TABLE [IF EXISTS] [db.]name [ON CLUSTER cluster] [SYNC] ``` +You can also drop multiple tables at the same time. + +Syntax: + +```sql +DROP [TEMPORARY] TABLE [IF EXISTS] [db.]name_1[,[db.]name_2][,[db.]name_3]... [ON CLUSTER cluster] [SYNC] +``` + ## DROP DICTIONARY Deletes the dictionary. From e851be78bce89b5cf652c710714277824c4f85e8 Mon Sep 17 00:00:00 2001 From: zvonand Date: Mon, 31 Jul 2023 23:56:07 +0200 Subject: [PATCH 020/433] fix fs_error on attempt to read unavailable dirs --- src/Storages/StorageFile.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/Storages/StorageFile.cpp b/src/Storages/StorageFile.cpp index cbd32460f7e..7e214b76e3d 100644 --- a/src/Storages/StorageFile.cpp +++ b/src/Storages/StorageFile.cpp @@ -122,7 +122,7 @@ void listFilesWithFoldedRegexpMatchingImpl(const std::string & path_for_ls, return; const fs::directory_iterator end; - for (fs::directory_iterator it(path_for_ls); it != end; ++it) + for (fs::directory_iterator it(path_for_ls, std::filesystem::directory_options::skip_permission_denied); it != end; ++it) { const std::string full_path = it->path().string(); const size_t last_slash = full_path.rfind('/'); From c8d995e42e4e2595981fed21191a20da00f80192 Mon Sep 17 00:00:00 2001 From: UnamedRus Date: Tue, 1 Aug 2023 01:21:46 +0300 Subject: [PATCH 021/433] add optimize to bitmap write method --- src/AggregateFunctions/AggregateFunctionGroupBitmapData.h | 1 + 1 file changed, 1 insertion(+) diff --git a/src/AggregateFunctions/AggregateFunctionGroupBitmapData.h b/src/AggregateFunctions/AggregateFunctionGroupBitmapData.h index 7ea1ebe7749..f92f8c1b5e5 100644 --- a/src/AggregateFunctions/AggregateFunctionGroupBitmapData.h +++ b/src/AggregateFunctions/AggregateFunctionGroupBitmapData.h @@ -151,6 +151,7 @@ public: } else if (BitmapKind::Bitmap == kind) { + roaring_bitmap->runOptimize(); auto size = roaring_bitmap->getSizeInBytes(); writeVarUInt(size, out); std::unique_ptr buf(new char[size]); From 8e3296a44a49efe5285e5f0151ef642b385c788f Mon Sep 17 00:00:00 2001 From: alexX512 Date: Tue, 1 Aug 2023 13:31:18 +0000 Subject: [PATCH 022/433] Add support of local progress in scheduler --- .../Executors/CompletedPipelineExecutor.cpp | 4 ++-- src/Processors/Executors/ExecutingGraph.cpp | 5 +++- .../Executors/ExecutionThreadContext.h | 13 +++++++++- src/Processors/Executors/ExecutorTasks.cpp | 21 ++++++++++++---- src/Processors/Executors/ExecutorTasks.h | 2 +- src/Processors/Executors/PipelineExecutor.cpp | 5 ++-- src/Processors/Executors/PipelineExecutor.h | 5 +++- .../PullingAsyncPipelineExecutor.cpp | 8 +++---- .../Executors/PullingPipelineExecutor.cpp | 2 +- .../PushingAsyncPipelineExecutor.cpp | 2 +- .../Executors/PushingPipelineExecutor.cpp | 2 +- src/Processors/Executors/TasksQueue.h | 1 + src/Processors/Formats/IOutputFormat.cpp | 1 + src/Processors/ISource.cpp | 24 +++++++++++++++++++ .../Transforms/LimitPartialResultTransform.h | 2 +- .../Transforms/PartialResultTransform.cpp | 12 ++++++++++ src/QueryPipeline/Pipe.cpp | 16 ++++++++----- src/QueryPipeline/Pipe.h | 1 + src/QueryPipeline/QueryPipeline.cpp | 1 + src/QueryPipeline/QueryPipeline.h | 3 +++ src/QueryPipeline/QueryPipelineBuilder.cpp | 21 +++++++++++++++- 21 files changed, 124 insertions(+), 27 deletions(-) diff --git a/src/Processors/Executors/CompletedPipelineExecutor.cpp b/src/Processors/Executors/CompletedPipelineExecutor.cpp index b0f842dec1b..217b397360e 100644 --- a/src/Processors/Executors/CompletedPipelineExecutor.cpp +++ b/src/Processors/Executors/CompletedPipelineExecutor.cpp @@ -74,7 +74,7 @@ void CompletedPipelineExecutor::execute() if (interactive_timeout_ms) { data = std::make_unique(); - data->executor = std::make_shared(pipeline.processors, pipeline.process_list_element); + data->executor = std::make_shared(pipeline.processors, pipeline.process_list_element, pipeline.partial_result_duration_ms); data->executor->setReadProgressCallback(pipeline.getReadProgressCallback()); /// Avoid passing this to lambda, copy ptr to data instead. @@ -100,7 +100,7 @@ void CompletedPipelineExecutor::execute() } else { - PipelineExecutor executor(pipeline.processors, pipeline.process_list_element); + PipelineExecutor executor(pipeline.processors, pipeline.process_list_element, pipeline.partial_result_duration_ms); executor.setReadProgressCallback(pipeline.getReadProgressCallback()); executor.execute(pipeline.getNumThreads()); } diff --git a/src/Processors/Executors/ExecutingGraph.cpp b/src/Processors/Executors/ExecutingGraph.cpp index 27f6a454b24..e639cea75ea 100644 --- a/src/Processors/Executors/ExecutingGraph.cpp +++ b/src/Processors/Executors/ExecutingGraph.cpp @@ -1,6 +1,7 @@ #include #include #include +#include namespace DB { @@ -60,6 +61,7 @@ bool ExecutingGraph::addEdges(uint64_t node) /// Add backward edges from input ports. auto & inputs = from->getInputs(); auto from_input = nodes[node]->back_edges.size(); + LOG_DEBUG(&Poco::Logger::get("ExecutingGraph::addEdges"), "Node id {}. Node name {}. Backward edges size {}. Inputs size {}. Direct edges size {}. Outputs size {}", node, from->getName(), from_input, inputs.size(), nodes[node]->direct_edges.size(), from->getOutputs().size()); if (from_input < inputs.size()) { @@ -220,6 +222,7 @@ bool ExecutingGraph::updateNode(uint64_t pid, Queue & queue, Queue & async_queue updated_processors.push(pid); std::shared_lock read_lock(nodes_mutex); + // LOG_DEBUG(&Poco::Logger::get("ExecutingGraph::updateNode"), "Original node for update. Node id {}. Node name {}", pid, nodes[pid]->processor->getName()); while (!updated_processors.empty() || !updated_edges.empty()) { @@ -260,7 +263,6 @@ bool ExecutingGraph::updateNode(uint64_t pid, Queue & queue, Queue & async_queue { pid = updated_processors.top(); updated_processors.pop(); - /// In this method we have ownership on node. auto & node = *nodes[pid]; @@ -282,6 +284,7 @@ bool ExecutingGraph::updateNode(uint64_t pid, Queue & queue, Queue & async_queue IProcessor::Status last_status = node.last_processor_status; IProcessor::Status status = processor.prepare(node.updated_input_ports, node.updated_output_ports); node.last_processor_status = status; + // LOG_DEBUG(&Poco::Logger::get("ExecutingGraph::updateNode"), "Node id {}. Node name {}. Status {}", pid, nodes[pid]->processor->getName(), status); if (profile_processors) { diff --git a/src/Processors/Executors/ExecutionThreadContext.h b/src/Processors/Executors/ExecutionThreadContext.h index eb048f8ab09..85788a70771 100644 --- a/src/Processors/Executors/ExecutionThreadContext.h +++ b/src/Processors/Executors/ExecutionThreadContext.h @@ -30,6 +30,12 @@ private: /// Callback for read progress. ReadProgressCallback * read_progress_callback = nullptr; + /// Timer that stops optimization of running local tasks instead of queuing them. + /// It provides local progress for each IProcessor task, allowing the partial result of the request to be always sended to the user. + Stopwatch watch; + /// Time period that limits the maximum allowed duration for optimizing the scheduling of local tasks within the executor + const UInt64 partial_result_duration_ms; + public: #ifndef NDEBUG /// Time for different processing stages. @@ -62,8 +68,13 @@ public: void setException(std::exception_ptr exception_) { exception = exception_; } void rethrowExceptionIfHas(); - explicit ExecutionThreadContext(size_t thread_number_, bool profile_processors_, bool trace_processors_, ReadProgressCallback * callback) + bool needWatchRestartForPartialResultProgress() { return partial_result_duration_ms != 0 && partial_result_duration_ms < watch.elapsedMilliseconds(); } + void restartWatch() { watch.restart(); } + + explicit ExecutionThreadContext(size_t thread_number_, bool profile_processors_, bool trace_processors_, ReadProgressCallback * callback, UInt64 partial_result_duration_ms_) : read_progress_callback(callback) + , watch(CLOCK_MONOTONIC) + , partial_result_duration_ms(partial_result_duration_ms_) , thread_number(thread_number_) , profile_processors(profile_processors_) , trace_processors(trace_processors_) diff --git a/src/Processors/Executors/ExecutorTasks.cpp b/src/Processors/Executors/ExecutorTasks.cpp index 496e859746a..9cbebb6996c 100644 --- a/src/Processors/Executors/ExecutorTasks.cpp +++ b/src/Processors/Executors/ExecutorTasks.cpp @@ -107,9 +107,18 @@ void ExecutorTasks::tryGetTask(ExecutionThreadContext & context) void ExecutorTasks::pushTasks(Queue & queue, Queue & async_queue, ExecutionThreadContext & context) { context.setTask(nullptr); + // if (!queue.empty()) + // LOG_DEBUG(&Poco::Logger::get("ExecutorTasks::pushTasks"), "First task to add in queue. Processor id {}. Processor name {}", queue.front()->processors_id, queue.front()->processor->getName()); - /// Take local task from queue if has one and it's not a processor which generates partial result. - if (!queue.empty() && !queue.front()->processor->isPartialResultProcessor() && !context.hasAsyncTasks()) + /// If sending partial results is allowed and local tasks scheduling optimization is repeated longer than the limit + /// or new task need to send partial result later, skip optimization for this iteration. + /// Otherwise take local task from queue if has one. + if ((!queue.empty() && queue.front()->processor->isPartialResultProcessor()) + || context.needWatchRestartForPartialResultProgress()) + { + context.restartWatch(); + } + else if (!queue.empty() && !context.hasAsyncTasks()) { context.setTask(queue.front()); queue.pop(); @@ -117,6 +126,8 @@ void ExecutorTasks::pushTasks(Queue & queue, Queue & async_queue, ExecutionThrea if (!queue.empty() || !async_queue.empty()) { + // LOG_DEBUG(&Poco::Logger::get("ExecutorTasks::pushTasks"), "After local optimizations push new tasks to queue"); + // task_queue.print(); std::unique_lock lock(mutex); #if defined(OS_LINUX) @@ -137,9 +148,11 @@ void ExecutorTasks::pushTasks(Queue & queue, Queue & async_queue, ExecutionThrea /// Wake up at least one thread that will wake up other threads if required tryWakeUpAnyOtherThreadWithTasks(context, lock); } + // LOG_DEBUG(&Poco::Logger::get("ExecutorTasks::pushTasks"), "After push new tasks to queue"); + // task_queue.print(); } -void ExecutorTasks::init(size_t num_threads_, size_t use_threads_, bool profile_processors, bool trace_processors, ReadProgressCallback * callback) +void ExecutorTasks::init(size_t num_threads_, size_t use_threads_, bool profile_processors, bool trace_processors, ReadProgressCallback * callback, UInt64 partial_result_duration_ms) { num_threads = num_threads_; use_threads = use_threads_; @@ -151,7 +164,7 @@ void ExecutorTasks::init(size_t num_threads_, size_t use_threads_, bool profile_ executor_contexts.reserve(num_threads); for (size_t i = 0; i < num_threads; ++i) - executor_contexts.emplace_back(std::make_unique(i, profile_processors, trace_processors, callback)); + executor_contexts.emplace_back(std::make_unique(i, profile_processors, trace_processors, callback, partial_result_duration_ms)); } } diff --git a/src/Processors/Executors/ExecutorTasks.h b/src/Processors/Executors/ExecutorTasks.h index d35f8de94d1..ab6d5e91411 100644 --- a/src/Processors/Executors/ExecutorTasks.h +++ b/src/Processors/Executors/ExecutorTasks.h @@ -58,7 +58,7 @@ public: void tryGetTask(ExecutionThreadContext & context); void pushTasks(Queue & queue, Queue & async_queue, ExecutionThreadContext & context); - void init(size_t num_threads_, size_t use_threads_, bool profile_processors, bool trace_processors, ReadProgressCallback * callback); + void init(size_t num_threads_, size_t use_threads_, bool profile_processors, bool trace_processors, ReadProgressCallback * callback, UInt64 partial_result_duration_ms); void fill(Queue & queue); void upscale(size_t use_threads_); diff --git a/src/Processors/Executors/PipelineExecutor.cpp b/src/Processors/Executors/PipelineExecutor.cpp index f523e7b7cf9..29c16f31aa3 100644 --- a/src/Processors/Executors/PipelineExecutor.cpp +++ b/src/Processors/Executors/PipelineExecutor.cpp @@ -36,8 +36,9 @@ namespace ErrorCodes } -PipelineExecutor::PipelineExecutor(std::shared_ptr & processors, QueryStatusPtr elem) +PipelineExecutor::PipelineExecutor(std::shared_ptr & processors, QueryStatusPtr elem, UInt64 partial_result_duration_ms_) : process_list_element(std::move(elem)) + , partial_result_duration_ms(partial_result_duration_ms_) { if (process_list_element) { @@ -309,7 +310,7 @@ void PipelineExecutor::initializeExecution(size_t num_threads) Queue queue; graph->initializeExecution(queue); - tasks.init(num_threads, use_threads, profile_processors, trace_processors, read_progress_callback.get()); + tasks.init(num_threads, use_threads, profile_processors, trace_processors, read_progress_callback.get(), partial_result_duration_ms); tasks.fill(queue); if (num_threads > 1) diff --git a/src/Processors/Executors/PipelineExecutor.h b/src/Processors/Executors/PipelineExecutor.h index 1e7d52d8290..fe424114bee 100644 --- a/src/Processors/Executors/PipelineExecutor.h +++ b/src/Processors/Executors/PipelineExecutor.h @@ -33,7 +33,7 @@ public: /// During pipeline execution new processors can appear. They will be added to existing set. /// /// Explicit graph representation is built in constructor. Throws if graph is not correct. - explicit PipelineExecutor(std::shared_ptr & processors, QueryStatusPtr elem); + explicit PipelineExecutor(std::shared_ptr & processors, QueryStatusPtr elem, UInt64 partial_result_duration_ms_ = 0); ~PipelineExecutor(); /// Execute pipeline in multiple threads. Must be called once. @@ -89,6 +89,9 @@ private: QueryStatusPtr process_list_element; ReadProgressCallbackPtr read_progress_callback; + + /// Duration between sending partial result through the pipeline + const UInt64 partial_result_duration_ms; using Queue = std::queue; diff --git a/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp b/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp index 581220fe26c..8875ba85a07 100644 --- a/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp +++ b/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp @@ -41,14 +41,14 @@ struct PullingAsyncPipelineExecutor::Data } }; -PullingAsyncPipelineExecutor::PullingAsyncPipelineExecutor(QueryPipeline & pipeline_, bool /*has_partial_result_setting*/) : pipeline(pipeline_) +PullingAsyncPipelineExecutor::PullingAsyncPipelineExecutor(QueryPipeline & pipeline_, bool has_partial_result_setting) : pipeline(pipeline_) { if (!pipeline.pulling()) throw Exception(ErrorCodes::LOGICAL_ERROR, "Pipeline for PullingAsyncPipelineExecutor must be pulling"); lazy_format = std::make_shared(pipeline.output->getHeader()); - // if (has_partial_result_setting) - // lazy_format->activatePartialResultProtocol(); + if (has_partial_result_setting) + lazy_format->activatePartialResultProtocol(); pipeline.complete(lazy_format); } @@ -105,7 +105,7 @@ bool PullingAsyncPipelineExecutor::pull(Chunk & chunk, uint64_t milliseconds) if (!data) { data = std::make_unique(); - data->executor = std::make_shared(pipeline.processors, pipeline.process_list_element); + data->executor = std::make_shared(pipeline.processors, pipeline.process_list_element, pipeline.partial_result_duration_ms); data->executor->setReadProgressCallback(pipeline.getReadProgressCallback()); data->lazy_format = lazy_format.get(); diff --git a/src/Processors/Executors/PullingPipelineExecutor.cpp b/src/Processors/Executors/PullingPipelineExecutor.cpp index cbf73c5cb07..f79f15c19bf 100644 --- a/src/Processors/Executors/PullingPipelineExecutor.cpp +++ b/src/Processors/Executors/PullingPipelineExecutor.cpp @@ -44,7 +44,7 @@ bool PullingPipelineExecutor::pull(Chunk & chunk) { if (!executor) { - executor = std::make_shared(pipeline.processors, pipeline.process_list_element); + executor = std::make_shared(pipeline.processors, pipeline.process_list_element, pipeline.partial_result_duration_ms); executor->setReadProgressCallback(pipeline.getReadProgressCallback()); } diff --git a/src/Processors/Executors/PushingAsyncPipelineExecutor.cpp b/src/Processors/Executors/PushingAsyncPipelineExecutor.cpp index 59d33cbffed..c01fcc634b7 100644 --- a/src/Processors/Executors/PushingAsyncPipelineExecutor.cpp +++ b/src/Processors/Executors/PushingAsyncPipelineExecutor.cpp @@ -166,7 +166,7 @@ void PushingAsyncPipelineExecutor::start() started = true; data = std::make_unique(); - data->executor = std::make_shared(pipeline.processors, pipeline.process_list_element); + data->executor = std::make_shared(pipeline.processors, pipeline.process_list_element, pipeline.partial_result_duration_ms); data->executor->setReadProgressCallback(pipeline.getReadProgressCallback()); data->source = pushing_source.get(); diff --git a/src/Processors/Executors/PushingPipelineExecutor.cpp b/src/Processors/Executors/PushingPipelineExecutor.cpp index 696932932df..f2b018792c7 100644 --- a/src/Processors/Executors/PushingPipelineExecutor.cpp +++ b/src/Processors/Executors/PushingPipelineExecutor.cpp @@ -87,7 +87,7 @@ void PushingPipelineExecutor::start() return; started = true; - executor = std::make_shared(pipeline.processors, pipeline.process_list_element); + executor = std::make_shared(pipeline.processors, pipeline.process_list_element, pipeline.partial_result_duration_ms); executor->setReadProgressCallback(pipeline.getReadProgressCallback()); if (!executor->executeStep(&input_wait_flag)) diff --git a/src/Processors/Executors/TasksQueue.h b/src/Processors/Executors/TasksQueue.h index bb8996fc1a6..bf8e8a1cde4 100644 --- a/src/Processors/Executors/TasksQueue.h +++ b/src/Processors/Executors/TasksQueue.h @@ -2,6 +2,7 @@ #include #include #include +#include namespace DB { diff --git a/src/Processors/Formats/IOutputFormat.cpp b/src/Processors/Formats/IOutputFormat.cpp index 6496f5357e7..7dba4b22eab 100644 --- a/src/Processors/Formats/IOutputFormat.cpp +++ b/src/Processors/Formats/IOutputFormat.cpp @@ -1,5 +1,6 @@ #include #include +#include namespace DB diff --git a/src/Processors/ISource.cpp b/src/Processors/ISource.cpp index 6a88d3973a1..34448186fa8 100644 --- a/src/Processors/ISource.cpp +++ b/src/Processors/ISource.cpp @@ -1,5 +1,6 @@ #include #include +#include namespace DB @@ -21,38 +22,61 @@ ISource::ISource(Block header, bool enable_auto_progress) ISource::Status ISource::prepare() { + const auto name = getName(); + if (name == "NullSource") + LOG_DEBUG(&Poco::Logger::get("ISource::prepare"), "KEK {}", getName()); if (finished) { + if (name == "NullSource") + LOG_DEBUG(&Poco::Logger::get("ISource::prepare"), "finished"); output.finish(); return Status::Finished; } /// Check can output. if (output.isFinished()) + { + if (name == "NullSource") + LOG_DEBUG(&Poco::Logger::get("ISource::prepare"), "output.isFinished()"); return Status::Finished; + } if (!output.canPush()) + { + if (name == "NullSource") + LOG_DEBUG(&Poco::Logger::get("ISource::prepare"), "!output.canPush()"); return Status::PortFull; + } if (!has_input) + { + if (name == "NullSource") + LOG_DEBUG(&Poco::Logger::get("ISource::prepare"), "!has_input"); return Status::Ready; + } output.pushData(std::move(current_chunk)); has_input = false; if (isCancelled()) { + if (name == "NullSource") + LOG_DEBUG(&Poco::Logger::get("ISource::prepare"), "isCancelled()"); output.finish(); return Status::Finished; } if (got_exception) { + if (name == "NullSource") + LOG_DEBUG(&Poco::Logger::get("ISource::prepare"), "got_exception"); finished = true; output.finish(); return Status::Finished; } + if (name == "NullSource") + LOG_DEBUG(&Poco::Logger::get("ISource::prepare"), "Status::PortFull"); /// Now, we pushed to output, and it must be full. return Status::PortFull; } diff --git a/src/Processors/Transforms/LimitPartialResultTransform.h b/src/Processors/Transforms/LimitPartialResultTransform.h index 8ed60d20da3..4b1f0e1026a 100644 --- a/src/Processors/Transforms/LimitPartialResultTransform.h +++ b/src/Processors/Transforms/LimitPartialResultTransform.h @@ -20,7 +20,7 @@ public: , offset(offset_) {} - String getName() const override { return "MergeSortingPartialResult"; } + String getName() const override { return "LimitPartialResultTransform"; } void transformPartialResult(Chunk & chunk) override { diff --git a/src/Processors/Transforms/PartialResultTransform.cpp b/src/Processors/Transforms/PartialResultTransform.cpp index 7cf2473684a..c3f38e77205 100644 --- a/src/Processors/Transforms/PartialResultTransform.cpp +++ b/src/Processors/Transforms/PartialResultTransform.cpp @@ -1,4 +1,5 @@ #include +#include namespace DB { @@ -14,28 +15,36 @@ PartialResultTransform::PartialResultTransform(const Block & header, UInt64 part IProcessor::Status PartialResultTransform::prepare() { + // LOG_DEBUG(&Poco::Logger::get("PartialResultTransform::prepare"), "KEK {}", getName()); if (output.isFinished()) { + // LOG_DEBUG(&Poco::Logger::get("PartialResultTransform::prepare"), "output.isFinished()"); input.close(); return Status::Finished; } if (finished_getting_snapshots) { + // LOG_DEBUG(&Poco::Logger::get("PartialResultTransform::prepare"), "finished_getting_snapshots"); output.finish(); return Status::Finished; } if (input.hasData()) + { + // LOG_DEBUG(&Poco::Logger::get("PartialResultTransform::prepare"), "input.hasData()"); partial_result = {input.pull(), SnaphotStatus::Ready}; + } /// Send partial result from real processor snapshot or from previous partial result processor if possible if (partial_result.snapshot_status == SnaphotStatus::Ready && output.canPush()) { + // LOG_DEBUG(&Poco::Logger::get("PartialResultTransform::prepare"), "partial_result.snapshot_status == SnaphotStatus::Ready && output.canPush()"); transformPartialResult(partial_result.chunk); partial_result.snapshot_status = SnaphotStatus::NotReady; if (partial_result.chunk.getNumRows() > 0) { + // LOG_DEBUG(&Poco::Logger::get("PartialResultTransform::prepare"), "partial_result.chunk.getNumRows() > 0"); output.push(std::move(partial_result.chunk)); return Status::PortFull; } @@ -45,15 +54,18 @@ IProcessor::Status PartialResultTransform::prepare() /// PartialResultTransform ready to create snapshots and send them as a partial result if (input.isFinished()) { + // LOG_DEBUG(&Poco::Logger::get("PartialResultTransform::prepare"), "input.isFinished()"); return Status::Ready; } input.setNeeded(); + // LOG_DEBUG(&Poco::Logger::get("PartialResultTransform::prepare"), "input.setNeeded()"); return Status::NeedData; } void PartialResultTransform::work() { + // LOG_DEBUG(&Poco::Logger::get("PartialResultTransform::prepare"), "Name {}. Duration_ms {}. Elapsed {}", getName(), partial_result_duration_ms, watch.elapsedMilliseconds()); if (partial_result_duration_ms < watch.elapsedMilliseconds()) { partial_result = getRealProcessorSnapshot(); diff --git a/src/QueryPipeline/Pipe.cpp b/src/QueryPipeline/Pipe.cpp index 6de45d67133..454f14d5626 100644 --- a/src/QueryPipeline/Pipe.cpp +++ b/src/QueryPipeline/Pipe.cpp @@ -411,12 +411,11 @@ void Pipe::addExtremesSource(ProcessorPtr source) void Pipe::activatePartialResult(UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_) { if (!is_partial_result_active) - { - is_partial_result_active = true; - partial_result_limit = partial_result_limit_; - partial_result_duration_ms = partial_result_duration_ms_; partial_result_ports.assign(output_ports.size(), nullptr); - } + + is_partial_result_active = true; + partial_result_limit = partial_result_limit_; + partial_result_duration_ms = partial_result_duration_ms_; } static void dropPort(OutputPort *& port, Processors & processors, Processors * collected_processors) @@ -881,7 +880,9 @@ void Pipe::setSinks(const Pipe::ProcessorGetterWithStreamKind & getter) add_transform(totals_port, StreamType::Totals); add_transform(extremes_port, StreamType::Extremes); - dropPartialResult(); + + for (auto & port : partial_result_ports) + add_transform(port, StreamType::PartialResult); output_ports.clear(); header.clear(); @@ -892,6 +893,9 @@ void Pipe::transform(const Transformer & transformer, bool check_ports) if (output_ports.empty()) throw Exception(ErrorCodes::LOGICAL_ERROR, "Cannot transform empty Pipe"); + /// TODO: Add functionality to work with partial result ports in transformer. + dropPartialResult(); + auto new_processors = transformer(output_ports); /// Create hash table with new processors. diff --git a/src/QueryPipeline/Pipe.h b/src/QueryPipeline/Pipe.h index 8b26435fd30..cbb7a8cbf5c 100644 --- a/src/QueryPipeline/Pipe.h +++ b/src/QueryPipeline/Pipe.h @@ -85,6 +85,7 @@ public: Main = 0, /// Stream for query data. There may be several streams of this type. Totals, /// Stream for totals. No more than one. Extremes, /// Stream for extremes. No more than one. + PartialResult, /// Stream for partial result data. There may be several streams of this type. }; using ProcessorGetter = std::function; diff --git a/src/QueryPipeline/QueryPipeline.cpp b/src/QueryPipeline/QueryPipeline.cpp index 190cc855918..8fca23dba7d 100644 --- a/src/QueryPipeline/QueryPipeline.cpp +++ b/src/QueryPipeline/QueryPipeline.cpp @@ -343,6 +343,7 @@ QueryPipeline::QueryPipeline( } QueryPipeline::QueryPipeline(Pipe pipe) + : partial_result_duration_ms(pipe.partial_result_duration_ms) { if (pipe.numOutputPorts() > 0) { diff --git a/src/QueryPipeline/QueryPipeline.h b/src/QueryPipeline/QueryPipeline.h index b93bc8647cf..cabdc8b3858 100644 --- a/src/QueryPipeline/QueryPipeline.h +++ b/src/QueryPipeline/QueryPipeline.h @@ -160,6 +160,9 @@ private: size_t num_threads = 0; + UInt64 partial_result_limit = 0; + UInt64 partial_result_duration_ms = 0; + friend class PushingPipelineExecutor; friend class PullingPipelineExecutor; friend class PushingAsyncPipelineExecutor; diff --git a/src/QueryPipeline/QueryPipelineBuilder.cpp b/src/QueryPipeline/QueryPipelineBuilder.cpp index 553b18dd57b..7da22b02ac6 100644 --- a/src/QueryPipeline/QueryPipelineBuilder.cpp +++ b/src/QueryPipeline/QueryPipelineBuilder.cpp @@ -110,6 +110,16 @@ void QueryPipelineBuilder::init(QueryPipeline & pipeline) pipe.header = {}; } + if (pipeline.partial_result) + { + /// Set partial result ports only after activation because when activated, it is set to nullptr + pipe.activatePartialResult(pipeline.partial_result_limit, pipeline.partial_result_duration_ms); + pipe.partial_result_ports = {pipeline.partial_result}; + } + + if (!pipeline.partial_result) + pipe.dropPartialResult(); + pipe.totals_port = pipeline.totals; pipe.extremes_port = pipeline.extremes; pipe.max_parallel_streams = pipeline.num_threads; @@ -347,6 +357,10 @@ std::unique_ptr QueryPipelineBuilder::joinPipelinesYShaped { left->checkInitializedAndNotCompleted(); right->checkInitializedAndNotCompleted(); + + /// TODO: Support joining of partial results from different pipelines. + left->pipe.dropPartialResult(); + right->pipe.dropPartialResult(); left->pipe.dropExtremes(); right->pipe.dropExtremes(); @@ -360,6 +374,7 @@ std::unique_ptr QueryPipelineBuilder::joinPipelinesYShaped auto joining = std::make_shared(join, inputs, out_header, max_block_size); + /// TODO: Support partial results in merge pipelines after joining support above. return mergePipelines(std::move(left), std::move(right), std::move(joining), collected_processors); } @@ -380,6 +395,10 @@ std::unique_ptr QueryPipelineBuilder::joinPipelinesRightLe left->pipe.dropExtremes(); right->pipe.dropExtremes(); + /// TODO: Support joining of partial results from different pipelines. + left->pipe.dropPartialResult(); + right->pipe.dropPartialResult(); + left->pipe.collected_processors = collected_processors; /// Collect the NEW processors for the right pipeline. @@ -634,7 +653,7 @@ PipelineExecutorPtr QueryPipelineBuilder::execute() if (!isCompleted()) throw Exception(ErrorCodes::LOGICAL_ERROR, "Cannot execute pipeline because it is not completed"); - return std::make_shared(pipe.processors, process_list_element); + return std::make_shared(pipe.processors, process_list_element, pipe.partial_result_duration_ms); } Pipe QueryPipelineBuilder::getPipe(QueryPipelineBuilder pipeline, QueryPlanResourceHolder & resources) From f9d7fe9ab9a3cd36bcc40ec9f529c0bd4980cd63 Mon Sep 17 00:00:00 2001 From: avogar Date: Tue, 1 Aug 2023 14:32:10 +0000 Subject: [PATCH 023/433] Fix misleading error message in OUTFILE with CapnProto/Protobuf --- src/Client/ClientBase.cpp | 4 +++- .../02842_capn_proto_outfile_without_schema.reference | 1 + .../02842_capn_proto_outfile_without_schema.sh | 10 ++++++++++ 3 files changed, 14 insertions(+), 1 deletion(-) create mode 100644 tests/queries/0_stateless/02842_capn_proto_outfile_without_schema.reference create mode 100755 tests/queries/0_stateless/02842_capn_proto_outfile_without_schema.sh diff --git a/src/Client/ClientBase.cpp b/src/Client/ClientBase.cpp index 06dabf96c28..1434498e549 100644 --- a/src/Client/ClientBase.cpp +++ b/src/Client/ClientBase.cpp @@ -1037,7 +1037,9 @@ void ClientBase::receiveResult(ASTPtr parsed_query, Int32 signals_before_stop, b } catch (const LocalFormatError &) { - local_format_error = std::current_exception(); + /// Remember the first exception. + if (!local_format_error) + local_format_error = std::current_exception(); connection->sendCancel(); } } diff --git a/tests/queries/0_stateless/02842_capn_proto_outfile_without_schema.reference b/tests/queries/0_stateless/02842_capn_proto_outfile_without_schema.reference new file mode 100644 index 00000000000..d00491fd7e5 --- /dev/null +++ b/tests/queries/0_stateless/02842_capn_proto_outfile_without_schema.reference @@ -0,0 +1 @@ +1 diff --git a/tests/queries/0_stateless/02842_capn_proto_outfile_without_schema.sh b/tests/queries/0_stateless/02842_capn_proto_outfile_without_schema.sh new file mode 100755 index 00000000000..fde08484026 --- /dev/null +++ b/tests/queries/0_stateless/02842_capn_proto_outfile_without_schema.sh @@ -0,0 +1,10 @@ +#!/usr/bin/env bash + +CUR_DIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) +# shellcheck source=../shell_config.sh +. "$CUR_DIR"/../shell_config.sh + + +$CLICKHOUSE_LOCAL -q "select * from numbers(10) into outfile '$CLICKHOUSE_TEST_UNIQUE_NAME.capnp'" 2>&1 | grep "The format CapnProto requires a schema" -c +rm $CLICKHOUSE_TEST_UNIQUE_NAME.capnp + From 9ac2d12f3bd81cae5f2c449b70ada1c30f7421cf Mon Sep 17 00:00:00 2001 From: alexX512 Date: Wed, 2 Aug 2023 13:07:02 +0000 Subject: [PATCH 024/433] Add test for partial sorting result --- src/Processors/Executors/ExecutingGraph.cpp | 4 - src/Processors/Executors/ExecutorTasks.cpp | 6 - src/Processors/Executors/PipelineExecutor.h | 2 +- src/Processors/Executors/TasksQueue.h | 1 - src/Processors/ISource.cpp | 24 -- .../Transforms/PartialResultTransform.cpp | 12 - src/QueryPipeline/Pipe.cpp | 2 +- src/QueryPipeline/QueryPipelineBuilder.cpp | 4 +- .../0_stateless/02010_lc_native.python | 283 ++---------------- .../02458_insert_select_progress_tcp.python | 236 ++------------- .../02750_settings_alias_tcp_protocol.python | 211 +------------ ...rting_result_during_query_execution.python | 49 +++ ...ng_result_during_query_execution.reference | 21 ++ ...l_sorting_result_during_query_execution.sh | 8 + .../queries/0_stateless/helpers/tcp_client.py | 274 +++++++++++++++++ 15 files changed, 415 insertions(+), 722 deletions(-) create mode 100755 tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python create mode 100644 tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.reference create mode 100755 tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.sh create mode 100644 tests/queries/0_stateless/helpers/tcp_client.py diff --git a/src/Processors/Executors/ExecutingGraph.cpp b/src/Processors/Executors/ExecutingGraph.cpp index e639cea75ea..6a946b4a4b9 100644 --- a/src/Processors/Executors/ExecutingGraph.cpp +++ b/src/Processors/Executors/ExecutingGraph.cpp @@ -1,7 +1,6 @@ #include #include #include -#include namespace DB { @@ -61,7 +60,6 @@ bool ExecutingGraph::addEdges(uint64_t node) /// Add backward edges from input ports. auto & inputs = from->getInputs(); auto from_input = nodes[node]->back_edges.size(); - LOG_DEBUG(&Poco::Logger::get("ExecutingGraph::addEdges"), "Node id {}. Node name {}. Backward edges size {}. Inputs size {}. Direct edges size {}. Outputs size {}", node, from->getName(), from_input, inputs.size(), nodes[node]->direct_edges.size(), from->getOutputs().size()); if (from_input < inputs.size()) { @@ -222,7 +220,6 @@ bool ExecutingGraph::updateNode(uint64_t pid, Queue & queue, Queue & async_queue updated_processors.push(pid); std::shared_lock read_lock(nodes_mutex); - // LOG_DEBUG(&Poco::Logger::get("ExecutingGraph::updateNode"), "Original node for update. Node id {}. Node name {}", pid, nodes[pid]->processor->getName()); while (!updated_processors.empty() || !updated_edges.empty()) { @@ -284,7 +281,6 @@ bool ExecutingGraph::updateNode(uint64_t pid, Queue & queue, Queue & async_queue IProcessor::Status last_status = node.last_processor_status; IProcessor::Status status = processor.prepare(node.updated_input_ports, node.updated_output_ports); node.last_processor_status = status; - // LOG_DEBUG(&Poco::Logger::get("ExecutingGraph::updateNode"), "Node id {}. Node name {}. Status {}", pid, nodes[pid]->processor->getName(), status); if (profile_processors) { diff --git a/src/Processors/Executors/ExecutorTasks.cpp b/src/Processors/Executors/ExecutorTasks.cpp index 9cbebb6996c..08920592391 100644 --- a/src/Processors/Executors/ExecutorTasks.cpp +++ b/src/Processors/Executors/ExecutorTasks.cpp @@ -107,8 +107,6 @@ void ExecutorTasks::tryGetTask(ExecutionThreadContext & context) void ExecutorTasks::pushTasks(Queue & queue, Queue & async_queue, ExecutionThreadContext & context) { context.setTask(nullptr); - // if (!queue.empty()) - // LOG_DEBUG(&Poco::Logger::get("ExecutorTasks::pushTasks"), "First task to add in queue. Processor id {}. Processor name {}", queue.front()->processors_id, queue.front()->processor->getName()); /// If sending partial results is allowed and local tasks scheduling optimization is repeated longer than the limit /// or new task need to send partial result later, skip optimization for this iteration. @@ -126,8 +124,6 @@ void ExecutorTasks::pushTasks(Queue & queue, Queue & async_queue, ExecutionThrea if (!queue.empty() || !async_queue.empty()) { - // LOG_DEBUG(&Poco::Logger::get("ExecutorTasks::pushTasks"), "After local optimizations push new tasks to queue"); - // task_queue.print(); std::unique_lock lock(mutex); #if defined(OS_LINUX) @@ -148,8 +144,6 @@ void ExecutorTasks::pushTasks(Queue & queue, Queue & async_queue, ExecutionThrea /// Wake up at least one thread that will wake up other threads if required tryWakeUpAnyOtherThreadWithTasks(context, lock); } - // LOG_DEBUG(&Poco::Logger::get("ExecutorTasks::pushTasks"), "After push new tasks to queue"); - // task_queue.print(); } void ExecutorTasks::init(size_t num_threads_, size_t use_threads_, bool profile_processors, bool trace_processors, ReadProgressCallback * callback, UInt64 partial_result_duration_ms) diff --git a/src/Processors/Executors/PipelineExecutor.h b/src/Processors/Executors/PipelineExecutor.h index fe424114bee..63524efc6d4 100644 --- a/src/Processors/Executors/PipelineExecutor.h +++ b/src/Processors/Executors/PipelineExecutor.h @@ -89,7 +89,7 @@ private: QueryStatusPtr process_list_element; ReadProgressCallbackPtr read_progress_callback; - + /// Duration between sending partial result through the pipeline const UInt64 partial_result_duration_ms; diff --git a/src/Processors/Executors/TasksQueue.h b/src/Processors/Executors/TasksQueue.h index bf8e8a1cde4..bb8996fc1a6 100644 --- a/src/Processors/Executors/TasksQueue.h +++ b/src/Processors/Executors/TasksQueue.h @@ -2,7 +2,6 @@ #include #include #include -#include namespace DB { diff --git a/src/Processors/ISource.cpp b/src/Processors/ISource.cpp index 34448186fa8..6a88d3973a1 100644 --- a/src/Processors/ISource.cpp +++ b/src/Processors/ISource.cpp @@ -1,6 +1,5 @@ #include #include -#include namespace DB @@ -22,61 +21,38 @@ ISource::ISource(Block header, bool enable_auto_progress) ISource::Status ISource::prepare() { - const auto name = getName(); - if (name == "NullSource") - LOG_DEBUG(&Poco::Logger::get("ISource::prepare"), "KEK {}", getName()); if (finished) { - if (name == "NullSource") - LOG_DEBUG(&Poco::Logger::get("ISource::prepare"), "finished"); output.finish(); return Status::Finished; } /// Check can output. if (output.isFinished()) - { - if (name == "NullSource") - LOG_DEBUG(&Poco::Logger::get("ISource::prepare"), "output.isFinished()"); return Status::Finished; - } if (!output.canPush()) - { - if (name == "NullSource") - LOG_DEBUG(&Poco::Logger::get("ISource::prepare"), "!output.canPush()"); return Status::PortFull; - } if (!has_input) - { - if (name == "NullSource") - LOG_DEBUG(&Poco::Logger::get("ISource::prepare"), "!has_input"); return Status::Ready; - } output.pushData(std::move(current_chunk)); has_input = false; if (isCancelled()) { - if (name == "NullSource") - LOG_DEBUG(&Poco::Logger::get("ISource::prepare"), "isCancelled()"); output.finish(); return Status::Finished; } if (got_exception) { - if (name == "NullSource") - LOG_DEBUG(&Poco::Logger::get("ISource::prepare"), "got_exception"); finished = true; output.finish(); return Status::Finished; } - if (name == "NullSource") - LOG_DEBUG(&Poco::Logger::get("ISource::prepare"), "Status::PortFull"); /// Now, we pushed to output, and it must be full. return Status::PortFull; } diff --git a/src/Processors/Transforms/PartialResultTransform.cpp b/src/Processors/Transforms/PartialResultTransform.cpp index c3f38e77205..7cf2473684a 100644 --- a/src/Processors/Transforms/PartialResultTransform.cpp +++ b/src/Processors/Transforms/PartialResultTransform.cpp @@ -1,5 +1,4 @@ #include -#include namespace DB { @@ -15,36 +14,28 @@ PartialResultTransform::PartialResultTransform(const Block & header, UInt64 part IProcessor::Status PartialResultTransform::prepare() { - // LOG_DEBUG(&Poco::Logger::get("PartialResultTransform::prepare"), "KEK {}", getName()); if (output.isFinished()) { - // LOG_DEBUG(&Poco::Logger::get("PartialResultTransform::prepare"), "output.isFinished()"); input.close(); return Status::Finished; } if (finished_getting_snapshots) { - // LOG_DEBUG(&Poco::Logger::get("PartialResultTransform::prepare"), "finished_getting_snapshots"); output.finish(); return Status::Finished; } if (input.hasData()) - { - // LOG_DEBUG(&Poco::Logger::get("PartialResultTransform::prepare"), "input.hasData()"); partial_result = {input.pull(), SnaphotStatus::Ready}; - } /// Send partial result from real processor snapshot or from previous partial result processor if possible if (partial_result.snapshot_status == SnaphotStatus::Ready && output.canPush()) { - // LOG_DEBUG(&Poco::Logger::get("PartialResultTransform::prepare"), "partial_result.snapshot_status == SnaphotStatus::Ready && output.canPush()"); transformPartialResult(partial_result.chunk); partial_result.snapshot_status = SnaphotStatus::NotReady; if (partial_result.chunk.getNumRows() > 0) { - // LOG_DEBUG(&Poco::Logger::get("PartialResultTransform::prepare"), "partial_result.chunk.getNumRows() > 0"); output.push(std::move(partial_result.chunk)); return Status::PortFull; } @@ -54,18 +45,15 @@ IProcessor::Status PartialResultTransform::prepare() /// PartialResultTransform ready to create snapshots and send them as a partial result if (input.isFinished()) { - // LOG_DEBUG(&Poco::Logger::get("PartialResultTransform::prepare"), "input.isFinished()"); return Status::Ready; } input.setNeeded(); - // LOG_DEBUG(&Poco::Logger::get("PartialResultTransform::prepare"), "input.setNeeded()"); return Status::NeedData; } void PartialResultTransform::work() { - // LOG_DEBUG(&Poco::Logger::get("PartialResultTransform::prepare"), "Name {}. Duration_ms {}. Elapsed {}", getName(), partial_result_duration_ms, watch.elapsedMilliseconds()); if (partial_result_duration_ms < watch.elapsedMilliseconds()) { partial_result = getRealProcessorSnapshot(); diff --git a/src/QueryPipeline/Pipe.cpp b/src/QueryPipeline/Pipe.cpp index 454f14d5626..6ff2cb6a4fd 100644 --- a/src/QueryPipeline/Pipe.cpp +++ b/src/QueryPipeline/Pipe.cpp @@ -893,7 +893,7 @@ void Pipe::transform(const Transformer & transformer, bool check_ports) if (output_ports.empty()) throw Exception(ErrorCodes::LOGICAL_ERROR, "Cannot transform empty Pipe"); - /// TODO: Add functionality to work with partial result ports in transformer. + /// TODO: Add functionality to work with partial result ports in transformer. dropPartialResult(); auto new_processors = transformer(output_ports); diff --git a/src/QueryPipeline/QueryPipelineBuilder.cpp b/src/QueryPipeline/QueryPipelineBuilder.cpp index 7da22b02ac6..e22b0077d1c 100644 --- a/src/QueryPipeline/QueryPipelineBuilder.cpp +++ b/src/QueryPipeline/QueryPipelineBuilder.cpp @@ -116,7 +116,7 @@ void QueryPipelineBuilder::init(QueryPipeline & pipeline) pipe.activatePartialResult(pipeline.partial_result_limit, pipeline.partial_result_duration_ms); pipe.partial_result_ports = {pipeline.partial_result}; } - + if (!pipeline.partial_result) pipe.dropPartialResult(); @@ -357,7 +357,7 @@ std::unique_ptr QueryPipelineBuilder::joinPipelinesYShaped { left->checkInitializedAndNotCompleted(); right->checkInitializedAndNotCompleted(); - + /// TODO: Support joining of partial results from different pipelines. left->pipe.dropPartialResult(); right->pipe.dropPartialResult(); diff --git a/tests/queries/0_stateless/02010_lc_native.python b/tests/queries/0_stateless/02010_lc_native.python index a197d32a3b9..5054ad3d454 100755 --- a/tests/queries/0_stateless/02010_lc_native.python +++ b/tests/queries/0_stateless/02010_lc_native.python @@ -1,226 +1,26 @@ #!/usr/bin/env python3 # -*- coding: utf-8 -*- -import socket import os -import uuid +import sys -CLICKHOUSE_HOST = os.environ.get("CLICKHOUSE_HOST", "127.0.0.1") -CLICKHOUSE_PORT = int(os.environ.get("CLICKHOUSE_PORT_TCP", "900000")) -CLICKHOUSE_DATABASE = os.environ.get("CLICKHOUSE_DATABASE", "default") +CURDIR = os.path.dirname(os.path.realpath(__file__)) +sys.path.insert(0, os.path.join(CURDIR, "helpers")) - -def writeVarUInt(x, ba): - for _ in range(0, 9): - byte = x & 0x7F - if x > 0x7F: - byte |= 0x80 - - ba.append(byte) - - x >>= 7 - if x == 0: - return - - -def writeStringBinary(s, ba): - b = bytes(s, "utf-8") - writeVarUInt(len(s), ba) - ba.extend(b) - - -def readStrict(s, size=1): - res = bytearray() - while size: - cur = s.recv(size) - # if not res: - # raise "Socket is closed" - size -= len(cur) - res.extend(cur) - - return res - - -def readUInt(s, size=1): - res = readStrict(s, size) - val = 0 - for i in range(len(res)): - val += res[i] << (i * 8) - return val - - -def readUInt8(s): - return readUInt(s) - - -def readUInt16(s): - return readUInt(s, 2) - - -def readUInt32(s): - return readUInt(s, 4) - - -def readUInt64(s): - return readUInt(s, 8) - - -def readVarUInt(s): - x = 0 - for i in range(9): - byte = readStrict(s)[0] - x |= (byte & 0x7F) << (7 * i) - - if not byte & 0x80: - return x - - return x - - -def readStringBinary(s): - size = readVarUInt(s) - s = readStrict(s, size) - return s.decode("utf-8") - - -def sendHello(s): - ba = bytearray() - writeVarUInt(0, ba) # Hello - writeStringBinary("simple native protocol", ba) - writeVarUInt(21, ba) - writeVarUInt(9, ba) - writeVarUInt(54449, ba) - writeStringBinary("default", ba) # database - writeStringBinary("default", ba) # user - writeStringBinary("", ba) # pwd - s.sendall(ba) - - -def receiveHello(s): - p_type = readVarUInt(s) - assert p_type == 0 # Hello - server_name = readStringBinary(s) - # print("Server name: ", server_name) - server_version_major = readVarUInt(s) - # print("Major: ", server_version_major) - server_version_minor = readVarUInt(s) - # print("Minor: ", server_version_minor) - server_revision = readVarUInt(s) - # print("Revision: ", server_revision) - server_timezone = readStringBinary(s) - # print("Timezone: ", server_timezone) - server_display_name = readStringBinary(s) - # print("Display name: ", server_display_name) - server_version_patch = readVarUInt(s) - # print("Version patch: ", server_version_patch) - - -def serializeClientInfo(ba, query_id): - writeStringBinary("default", ba) # initial_user - writeStringBinary(query_id, ba) # initial_query_id - writeStringBinary("127.0.0.1:9000", ba) # initial_address - ba.extend([0] * 8) # initial_query_start_time_microseconds - ba.append(1) # TCP - writeStringBinary("os_user", ba) # os_user - writeStringBinary("client_hostname", ba) # client_hostname - writeStringBinary("client_name", ba) # client_name - writeVarUInt(21, ba) - writeVarUInt(9, ba) - writeVarUInt(54449, ba) - writeStringBinary("", ba) # quota_key - writeVarUInt(0, ba) # distributed_depth - writeVarUInt(1, ba) # client_version_patch - ba.append(0) # No telemetry - - -def sendQuery(s, query): - ba = bytearray() - query_id = uuid.uuid4().hex - writeVarUInt(1, ba) # query - writeStringBinary(query_id, ba) - - ba.append(1) # INITIAL_QUERY - - # client info - serializeClientInfo(ba, query_id) - - writeStringBinary("", ba) # No settings - writeStringBinary("", ba) # No interserver secret - writeVarUInt(2, ba) # Stage - Complete - ba.append(0) # No compression - writeStringBinary(query, ba) # query, finally - s.sendall(ba) - - -def serializeBlockInfo(ba): - writeVarUInt(1, ba) # 1 - ba.append(0) # is_overflows - writeVarUInt(2, ba) # 2 - writeVarUInt(0, ba) # 0 - ba.extend([0] * 4) # bucket_num - - -def sendEmptyBlock(s): - ba = bytearray() - writeVarUInt(2, ba) # Data - writeStringBinary("", ba) - serializeBlockInfo(ba) - writeVarUInt(0, ba) # rows - writeVarUInt(0, ba) # columns - s.sendall(ba) - - -def assertPacket(packet, expected): - assert packet == expected, packet - - -def readHeader(s): - packet_type = readVarUInt(s) - if packet_type == 2: # Exception - raise RuntimeError(readException(s)) - assertPacket(packet_type, 1) # Data - - readStringBinary(s) # external table name - # BlockInfo - assertPacket(readVarUInt(s), 1) # 1 - assertPacket(readUInt8(s), 0) # is_overflows - assertPacket(readVarUInt(s), 2) # 2 - assertPacket(readUInt32(s), 4294967295) # bucket_num - assertPacket(readVarUInt(s), 0) # 0 - columns = readVarUInt(s) # rows - rows = readVarUInt(s) # columns - print("Rows {} Columns {}".format(rows, columns)) - for _ in range(columns): - col_name = readStringBinary(s) - type_name = readStringBinary(s) - print("Column {} type {}".format(col_name, type_name)) - - -def readException(s): - code = readUInt32(s) - name = readStringBinary(s) - text = readStringBinary(s) - readStringBinary(s) # trace - assertPacket(readUInt8(s), 0) # has_nested - return "code {}: {}".format(code, text.replace("DB::Exception:", "")) +from tcp_client import TCPClient, CLICKHOUSE_DATABASE, writeVarUInt, writeStringBinary, serializeBlockInfo, assertPacket def insertValidLowCardinalityRow(): - with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: - s.settimeout(30) - s.connect((CLICKHOUSE_HOST, CLICKHOUSE_PORT)) - sendHello(s) - receiveHello(s) - sendQuery( - s, + with TCPClient() as client: + client.sendQuery( "insert into {}.tab settings input_format_defaults_for_omitted_fields=0 format TSV".format( CLICKHOUSE_DATABASE ), ) # external tables - sendEmptyBlock(s) - readHeader(s) + client.sendEmptyBlock() + client.readHeader() # Data ba = bytearray() @@ -239,31 +39,25 @@ def insertValidLowCardinalityRow(): writeStringBinary("hello", ba) # key ba.extend([1] + [0] * 7) # num_indexes ba.extend([0] * 8) # UInt64 index (0 for 'hello') - s.sendall(ba) + client.send(ba) # Fin block - sendEmptyBlock(s) + client.sendEmptyBlock() - assertPacket(readVarUInt(s), 5) # End of stream - s.close() + assertPacket(client.readVarUInt(), 5) # End of stream def insertLowCardinalityRowWithIndexOverflow(): - with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: - s.settimeout(30) - s.connect((CLICKHOUSE_HOST, CLICKHOUSE_PORT)) - sendHello(s) - receiveHello(s) - sendQuery( - s, + with TCPClient() as client: + client.sendQuery( "insert into {}.tab settings input_format_defaults_for_omitted_fields=0 format TSV".format( CLICKHOUSE_DATABASE ), ) # external tables - sendEmptyBlock(s) - readHeader(s) + client.sendEmptyBlock() + client.readHeader() # Data ba = bytearray() @@ -282,29 +76,23 @@ def insertLowCardinalityRowWithIndexOverflow(): writeStringBinary("hello", ba) # key ba.extend([1] + [0] * 7) # num_indexes ba.extend([0] * 7 + [1]) # UInt64 index (overflow) - s.sendall(ba) + client.send(ba) - assertPacket(readVarUInt(s), 2) - print(readException(s)) - s.close() + assertPacket(client.readVarUInt(), 2) # Exception + print(client.readException()) def insertLowCardinalityRowWithIncorrectDictType(): - with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: - s.settimeout(30) - s.connect((CLICKHOUSE_HOST, CLICKHOUSE_PORT)) - sendHello(s) - receiveHello(s) - sendQuery( - s, + with TCPClient() as client: + client.sendQuery( "insert into {}.tab settings input_format_defaults_for_omitted_fields=0 format TSV".format( CLICKHOUSE_DATABASE ), ) # external tables - sendEmptyBlock(s) - readHeader(s) + client.sendEmptyBlock() + client.readHeader() # Data ba = bytearray() @@ -323,29 +111,23 @@ def insertLowCardinalityRowWithIncorrectDictType(): writeStringBinary("hello", ba) # key ba.extend([1] + [0] * 7) # num_indexes ba.extend([0] * 8) # UInt64 index (overflow) - s.sendall(ba) + client.send(ba) - assertPacket(readVarUInt(s), 2) - print(readException(s)) - s.close() + assertPacket(client.readVarUInt(), 2) # Exception + print(client.readException()) def insertLowCardinalityRowWithIncorrectAdditionalKeys(): - with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: - s.settimeout(30) - s.connect((CLICKHOUSE_HOST, CLICKHOUSE_PORT)) - sendHello(s) - receiveHello(s) - sendQuery( - s, + with TCPClient() as client: + client.sendQuery( "insert into {}.tab settings input_format_defaults_for_omitted_fields=0 format TSV".format( CLICKHOUSE_DATABASE ), ) # external tables - sendEmptyBlock(s) - readHeader(s) + client.sendEmptyBlock() + client.readHeader() # Data ba = bytearray() @@ -364,11 +146,10 @@ def insertLowCardinalityRowWithIncorrectAdditionalKeys(): writeStringBinary("hello", ba) # key ba.extend([1] + [0] * 7) # num_indexes ba.extend([0] * 8) # UInt64 index (0 for 'hello') - s.sendall(ba) + client.send(ba) - assertPacket(readVarUInt(s), 2) - print(readException(s)) - s.close() + assertPacket(client.readVarUInt(), 2) # Exception + print(client.readException()) def main(): diff --git a/tests/queries/0_stateless/02458_insert_select_progress_tcp.python b/tests/queries/0_stateless/02458_insert_select_progress_tcp.python index 696eb01ff7e..1fe2363d7d5 100644 --- a/tests/queries/0_stateless/02458_insert_select_progress_tcp.python +++ b/tests/queries/0_stateless/02458_insert_select_progress_tcp.python @@ -1,187 +1,23 @@ #!/usr/bin/env python3 -import socket -import os -import uuid import json +import os +import sys -CLICKHOUSE_HOST = os.environ.get("CLICKHOUSE_HOST", "127.0.0.1") -CLICKHOUSE_PORT = int(os.environ.get("CLICKHOUSE_PORT_TCP", "900000")) -CLICKHOUSE_DATABASE = os.environ.get("CLICKHOUSE_DATABASE", "default") +CURDIR = os.path.dirname(os.path.realpath(__file__)) +sys.path.insert(0, os.path.join(CURDIR, "helpers")) - -def writeVarUInt(x, ba): - for _ in range(0, 9): - byte = x & 0x7F - if x > 0x7F: - byte |= 0x80 - - ba.append(byte) - - x >>= 7 - if x == 0: - return - - -def writeStringBinary(s, ba): - b = bytes(s, "utf-8") - writeVarUInt(len(s), ba) - ba.extend(b) - - -def readStrict(s, size=1): - res = bytearray() - while size: - cur = s.recv(size) - # if not res: - # raise "Socket is closed" - size -= len(cur) - res.extend(cur) - - return res - - -def readUInt(s, size=1): - res = readStrict(s, size) - val = 0 - for i in range(len(res)): - val += res[i] << (i * 8) - return val - - -def readUInt8(s): - return readUInt(s) - - -def readUInt16(s): - return readUInt(s, 2) - - -def readUInt32(s): - return readUInt(s, 4) - - -def readUInt64(s): - return readUInt(s, 8) - - -def readVarUInt(s): - x = 0 - for i in range(9): - byte = readStrict(s)[0] - x |= (byte & 0x7F) << (7 * i) - - if not byte & 0x80: - return x - - return x - - -def readStringBinary(s): - size = readVarUInt(s) - s = readStrict(s, size) - return s.decode("utf-8") - - -def sendHello(s): - ba = bytearray() - writeVarUInt(0, ba) # Hello - writeStringBinary("simple native protocol", ba) - writeVarUInt(21, ba) - writeVarUInt(9, ba) - writeVarUInt(54449, ba) - writeStringBinary(CLICKHOUSE_DATABASE, ba) # database - writeStringBinary("default", ba) # user - writeStringBinary("", ba) # pwd - s.sendall(ba) - - -def receiveHello(s): - p_type = readVarUInt(s) - assert p_type == 0 # Hello - server_name = readStringBinary(s) - # print("Server name: ", server_name) - server_version_major = readVarUInt(s) - # print("Major: ", server_version_major) - server_version_minor = readVarUInt(s) - # print("Minor: ", server_version_minor) - server_revision = readVarUInt(s) - # print("Revision: ", server_revision) - server_timezone = readStringBinary(s) - # print("Timezone: ", server_timezone) - server_display_name = readStringBinary(s) - # print("Display name: ", server_display_name) - server_version_patch = readVarUInt(s) - # print("Version patch: ", server_version_patch) - - -def serializeClientInfo(ba, query_id): - writeStringBinary("default", ba) # initial_user - writeStringBinary(query_id, ba) # initial_query_id - writeStringBinary("127.0.0.1:9000", ba) # initial_address - ba.extend([0] * 8) # initial_query_start_time_microseconds - ba.append(1) # TCP - writeStringBinary("os_user", ba) # os_user - writeStringBinary("client_hostname", ba) # client_hostname - writeStringBinary("client_name", ba) # client_name - writeVarUInt(21, ba) - writeVarUInt(9, ba) - writeVarUInt(54449, ba) - writeStringBinary("", ba) # quota_key - writeVarUInt(0, ba) # distributed_depth - writeVarUInt(1, ba) # client_version_patch - ba.append(0) # No telemetry - - -def sendQuery(s, query): - ba = bytearray() - query_id = uuid.uuid4().hex - writeVarUInt(1, ba) # query - writeStringBinary(query_id, ba) - - ba.append(1) # INITIAL_QUERY - - # client info - serializeClientInfo(ba, query_id) - - writeStringBinary("", ba) # No settings - writeStringBinary("", ba) # No interserver secret - writeVarUInt(2, ba) # Stage - Complete - ba.append(0) # No compression - writeStringBinary(query, ba) # query, finally - s.sendall(ba) - - -def serializeBlockInfo(ba): - writeVarUInt(1, ba) # 1 - ba.append(0) # is_overflows - writeVarUInt(2, ba) # 2 - writeVarUInt(0, ba) # 0 - ba.extend([0] * 4) # bucket_num - - -def sendEmptyBlock(s): - ba = bytearray() - writeVarUInt(2, ba) # Data - writeStringBinary("", ba) - serializeBlockInfo(ba) - writeVarUInt(0, ba) # rows - writeVarUInt(0, ba) # columns - s.sendall(ba) - - -def assertPacket(packet, expected): - assert packet == expected, packet +from tcp_client import TCPClient class Progress: - def __init__(self): + def __init__(self, read_rows=0, read_bytes=0, total_rows_to_read=0, written_rows=0, written_bytes=0): # NOTE: this is done in ctor to initialize __dict__ - self.read_rows = 0 - self.read_bytes = 0 - self.total_rows_to_read = 0 - self.written_rows = 0 - self.written_bytes = 0 + self.read_rows = read_rows + self.read_bytes = read_bytes + self.total_rows_to_read = total_rows_to_read + self.written_rows = written_rows + self.written_bytes = written_bytes def __str__(self): return json.dumps(self.__dict__) @@ -194,13 +30,6 @@ class Progress: self.written_bytes += b.written_bytes return self - def readPacket(self, s): - self.read_rows += readVarUInt(s) - self.read_bytes += readVarUInt(s) - self.total_rows_to_read += readVarUInt(s) - self.written_rows += readVarUInt(s) - self.written_bytes += readVarUInt(s) - def __bool__(self): return ( self.read_rows > 0 @@ -211,52 +40,25 @@ class Progress: ) -def readProgress(s): - packet_type = readVarUInt(s) - if packet_type == 2: # Exception - raise RuntimeError(readException(s)) - - if packet_type == 5: # End stream - return None - - assertPacket(packet_type, 3) # Progress - - progress = Progress() - progress.readPacket(s) - return progress - - -def readException(s): - code = readUInt32(s) - name = readStringBinary(s) - text = readStringBinary(s) - readStringBinary(s) # trace - assertPacket(readUInt8(s), 0) # has_nested - return "code {}: {}".format(code, text.replace("DB::Exception:", "")) - - def main(): - with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: - s.settimeout(30) - s.connect((CLICKHOUSE_HOST, CLICKHOUSE_PORT)) - sendHello(s) - receiveHello(s) + with TCPClient() as client: # For 1 second sleep and 1000ms of interactive_delay we definitelly should have non zero progress packet. # NOTE: interactive_delay=0 cannot be used since in this case CompletedPipelineExecutor will not call cancelled callback. - sendQuery( - s, + client.sendQuery( "insert into function null('_ Int') select sleep(1) from numbers(2) settings max_block_size=1, interactive_delay=1000", ) # external tables - sendEmptyBlock(s) + client.sendEmptyBlock() summary_progress = Progress() non_empty_progress_packets = 0 while True: - progress = readProgress(s) - if progress is None: + progress_info = client.readProgress() + if progress_info is None: break + + progress = Progress(*progress_info) summary_progress += progress if progress: non_empty_progress_packets += 1 @@ -267,8 +69,6 @@ def main(): # - 1 or 2 for each SELECT block assert non_empty_progress_packets in (3, 4), f"{non_empty_progress_packets=:}" - s.close() - if __name__ == "__main__": main() diff --git a/tests/queries/0_stateless/02750_settings_alias_tcp_protocol.python b/tests/queries/0_stateless/02750_settings_alias_tcp_protocol.python index 768fb2144e3..1736807410f 100644 --- a/tests/queries/0_stateless/02750_settings_alias_tcp_protocol.python +++ b/tests/queries/0_stateless/02750_settings_alias_tcp_protocol.python @@ -1,216 +1,23 @@ #!/usr/bin/env python3 -import socket + import os -import uuid -import json +import sys -CLICKHOUSE_HOST = os.environ.get("CLICKHOUSE_HOST", "127.0.0.1") -CLICKHOUSE_PORT = int(os.environ.get("CLICKHOUSE_PORT_TCP", "900000")) -CLICKHOUSE_DATABASE = os.environ.get("CLICKHOUSE_DATABASE", "default") +CURDIR = os.path.dirname(os.path.realpath(__file__)) +sys.path.insert(0, os.path.join(CURDIR, "helpers")) - -def writeVarUInt(x, ba): - for _ in range(0, 9): - byte = x & 0x7F - if x > 0x7F: - byte |= 0x80 - - ba.append(byte) - - x >>= 7 - if x == 0: - return - - -def writeStringBinary(s, ba): - b = bytes(s, "utf-8") - writeVarUInt(len(s), ba) - ba.extend(b) - - -def readStrict(s, size=1): - res = bytearray() - while size: - cur = s.recv(size) - # if not res: - # raise "Socket is closed" - size -= len(cur) - res.extend(cur) - - return res - - -def readUInt(s, size=1): - res = readStrict(s, size) - val = 0 - for i in range(len(res)): - val += res[i] << (i * 8) - return val - - -def readUInt8(s): - return readUInt(s) - - -def readUInt16(s): - return readUInt(s, 2) - - -def readUInt32(s): - return readUInt(s, 4) - - -def readUInt64(s): - return readUInt(s, 8) - - -def readVarUInt(s): - x = 0 - for i in range(9): - byte = readStrict(s)[0] - x |= (byte & 0x7F) << (7 * i) - - if not byte & 0x80: - return x - - return x - - -def readStringBinary(s): - size = readVarUInt(s) - s = readStrict(s, size) - return s.decode("utf-8") - - -def sendHello(s): - ba = bytearray() - writeVarUInt(0, ba) # Hello - writeStringBinary("simple native protocol", ba) - writeVarUInt(21, ba) - writeVarUInt(9, ba) - writeVarUInt(54449, ba) - writeStringBinary(CLICKHOUSE_DATABASE, ba) # database - writeStringBinary("default", ba) # user - writeStringBinary("", ba) # pwd - s.sendall(ba) - - -def receiveHello(s): - p_type = readVarUInt(s) - assert p_type == 0 # Hello - _server_name = readStringBinary(s) - _server_version_major = readVarUInt(s) - _server_version_minor = readVarUInt(s) - _server_revision = readVarUInt(s) - _server_timezone = readStringBinary(s) - _server_display_name = readStringBinary(s) - _server_version_patch = readVarUInt(s) - - -def serializeClientInfo(ba, query_id): - writeStringBinary("default", ba) # initial_user - writeStringBinary(query_id, ba) # initial_query_id - writeStringBinary("127.0.0.1:9000", ba) # initial_address - ba.extend([0] * 8) # initial_query_start_time_microseconds - ba.append(1) # TCP - writeStringBinary("os_user", ba) # os_user - writeStringBinary("client_hostname", ba) # client_hostname - writeStringBinary("client_name", ba) # client_name - writeVarUInt(21, ba) - writeVarUInt(9, ba) - writeVarUInt(54449, ba) - writeStringBinary("", ba) # quota_key - writeVarUInt(0, ba) # distributed_depth - writeVarUInt(1, ba) # client_version_patch - ba.append(0) # No telemetry - - -def sendQuery(s, query, settings): - ba = bytearray() - query_id = uuid.uuid4().hex - writeVarUInt(1, ba) # query - writeStringBinary(query_id, ba) - - ba.append(1) # INITIAL_QUERY - - # client info - serializeClientInfo(ba, query_id) - - # Settings - for key, value in settings.items(): - writeStringBinary(key, ba) - writeVarUInt(1, ba) # is_important - writeStringBinary(str(value), ba) - writeStringBinary("", ba) # End of settings - - writeStringBinary("", ba) # No interserver secret - writeVarUInt(2, ba) # Stage - Complete - ba.append(0) # No compression - writeStringBinary(query, ba) # query, finally - s.sendall(ba) - - -def serializeBlockInfo(ba): - writeVarUInt(1, ba) # 1 - ba.append(0) # is_overflows - writeVarUInt(2, ba) # 2 - writeVarUInt(0, ba) # 0 - ba.extend([0] * 4) # bucket_num - - -def sendEmptyBlock(s): - ba = bytearray() - writeVarUInt(2, ba) # Data - writeStringBinary("", ba) - serializeBlockInfo(ba) - writeVarUInt(0, ba) # rows - writeVarUInt(0, ba) # columns - s.sendall(ba) - - -def assertPacket(packet, expected): - assert packet == expected, "Got: {}, expected: {}".format(packet, expected) - - -def readResponse(s): - packet_type = readVarUInt(s) - if packet_type == 2: # Exception - raise RuntimeError(readException(s)) - - if packet_type == 1: # Data - return None - if packet_type == 3: # Progress - return None - if packet_type == 5: # End stream - return None - - raise RuntimeError("Unexpected packet: {}".format(packet_type)) - - -def readException(s): - code = readUInt32(s) - _name = readStringBinary(s) - text = readStringBinary(s) - readStringBinary(s) # trace - assertPacket(readUInt8(s), 0) # has_nested - return "code {}: {}".format(code, text.replace("DB::Exception:", "")) +from tcp_client import TCPClient def main(): - with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: - s.settimeout(30) - s.connect((CLICKHOUSE_HOST, CLICKHOUSE_PORT)) - sendHello(s) - receiveHello(s) - sendQuery(s, "select 1", {"replication_alter_partitions_sync": 1}) + with TCPClient() as client: + client.sendQuery("select 1", {"replication_alter_partitions_sync": 1}) # external tables - sendEmptyBlock(s) + client.sendEmptyBlock() - while readResponse(s) is not None: + while client.readResponse() is not None: pass - - s.close() print("OK") diff --git a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python new file mode 100755 index 00000000000..00c11d0d233 --- /dev/null +++ b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python @@ -0,0 +1,49 @@ +#!/usr/bin/env python3 +# -*- coding: utf-8 -*- + + +import os +import sys + +CURDIR = os.path.dirname(os.path.realpath(__file__)) +sys.path.insert(0, os.path.join(CURDIR, "helpers")) + +from tcp_client import TCPClient, assertPacket + + +def main(): + with TCPClient() as client: + client.sendQuery( + f"SELECT number FROM numbers_mt(1e7+1) ORDER BY -number LIMIT 15 SETTINGS max_threads = 1, partial_result_update_duration_ms=1, max_rows_in_partial_result=10" + ) + + # external tables + client.sendEmptyBlock() + client.readHeader() + + # Partial result + _, partial_result = client.readHeader(need_read_data=True)[0] + + assert_message = "There should be at least one block of data with partial result" + assert len(partial_result) > 0, assert_message + + while True: + _, new_partial_result = client.readHeader(need_read_data=True, need_print_info=False)[0] + if (len(new_partial_result) == 0): + break + + data_size = len(partial_result) + assert_message = f"New block contains more info about the full data so sorted results should not be less then in the previous iteration. New result {new_partial_result}. Previous result {partial_result}" + assert all(partial_result[i] <= new_partial_result[i] for i in range(data_size)), assert_message + + partial_result = new_partial_result + + # Full result + _, full_result = client.readHeader(need_read_data=True)[0] + for result in full_result: + print(result) + + +if __name__ == "__main__": + main() + diff --git a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.reference b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.reference new file mode 100644 index 00000000000..4d706474b22 --- /dev/null +++ b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.reference @@ -0,0 +1,21 @@ +Rows 0 Columns 1 +Column number type UInt64 +Rows 10 Columns 1 +Column number type UInt64 +Rows 15 Columns 1 +Column number type UInt64 +10000000 +9999999 +9999998 +9999997 +9999996 +9999995 +9999994 +9999993 +9999992 +9999991 +9999990 +9999989 +9999988 +9999987 +9999986 diff --git a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.sh b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.sh new file mode 100755 index 00000000000..1ed15197dbf --- /dev/null +++ b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.sh @@ -0,0 +1,8 @@ +#!/usr/bin/env bash + +CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) +# shellcheck source=../shell_config.sh +. "$CURDIR"/../shell_config.sh + +# We should have correct env vars from shell_config.sh to run this test +python3 "$CURDIR"/02833_partial_sorting_result_during_query_execution.python diff --git a/tests/queries/0_stateless/helpers/tcp_client.py b/tests/queries/0_stateless/helpers/tcp_client.py new file mode 100644 index 00000000000..d1c6cb4894c --- /dev/null +++ b/tests/queries/0_stateless/helpers/tcp_client.py @@ -0,0 +1,274 @@ +import socket +import os +import uuid + +CLICKHOUSE_HOST = os.environ.get("CLICKHOUSE_HOST", "127.0.0.1") +CLICKHOUSE_PORT = int(os.environ.get("CLICKHOUSE_PORT_TCP", "900000")) +CLICKHOUSE_DATABASE = os.environ.get("CLICKHOUSE_DATABASE", "default") + + +def writeVarUInt(x, ba): + for _ in range(0, 9): + byte = x & 0x7F + if x > 0x7F: + byte |= 0x80 + + ba.append(byte) + + x >>= 7 + if x == 0: + return + + +def writeStringBinary(s, ba): + b = bytes(s, "utf-8") + writeVarUInt(len(s), ba) + ba.extend(b) + + +def serializeClientInfo(ba, query_id): + writeStringBinary("default", ba) # initial_user + writeStringBinary(query_id, ba) # initial_query_id + writeStringBinary("127.0.0.1:9000", ba) # initial_address + ba.extend([0] * 8) # initial_query_start_time_microseconds + ba.append(1) # TCP + writeStringBinary("os_user", ba) # os_user + writeStringBinary("client_hostname", ba) # client_hostname + writeStringBinary("client_name", ba) # client_name + writeVarUInt(21, ba) + writeVarUInt(9, ba) + writeVarUInt(54449, ba) + writeStringBinary("", ba) # quota_key + writeVarUInt(0, ba) # distributed_depth + writeVarUInt(1, ba) # client_version_patch + ba.append(0) # No telemetry + + +def serializeBlockInfo(ba): + writeVarUInt(1, ba) # 1 + ba.append(0) # is_overflows + writeVarUInt(2, ba) # 2 + writeVarUInt(0, ba) # 0 + ba.extend([0] * 4) # bucket_num + + +def assertPacket(packet, expected): + assert packet == expected, "Got: {}, expected: {}".format(packet, expected) + + +class TCPClient(object): + def __init__(self, timeout=30): + self.timeout = timeout + self.socket = None + + def __enter__(self): + self.socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM) + self.socket.settimeout(self.timeout) + self.socket.connect((CLICKHOUSE_HOST, CLICKHOUSE_PORT)) + + self.sendHello() + self.receiveHello() + + return self + + def __exit__(self, exc_type, exc_value, traceback): + if self.socket: + self.socket.close() + + def readStrict(self, size=1): + res = bytearray() + while size: + cur = self.socket.recv(size) + # if not res: + # raise "Socket is closed" + size -= len(cur) + res.extend(cur) + + return res + + def readUInt(self, size=1): + res = self.readStrict(size) + val = 0 + for i in range(len(res)): + val += res[i] << (i * 8) + return val + + def readUInt8(self): + return self.readUInt() + + def readUInt16(self): + return self.readUInt(2) + + def readUInt32(self): + return self.readUInt(4) + + def readUInt64(self): + return self.readUInt(8) + + def readVarUInt(self): + x = 0 + for i in range(9): + byte = self.readStrict()[0] + x |= (byte & 0x7F) << (7 * i) + + if not byte & 0x80: + return x + + return x + + def readStringBinary(self): + size = self.readVarUInt() + s = self.readStrict(size) + return s.decode("utf-8") + + def send(self, byte_array): + self.socket.sendall(byte_array) + + def sendHello(self): + ba = bytearray() + writeVarUInt(0, ba) # Hello + writeStringBinary("simple native protocol", ba) + writeVarUInt(21, ba) + writeVarUInt(9, ba) + writeVarUInt(54449, ba) + writeStringBinary(CLICKHOUSE_DATABASE, ba) # database + writeStringBinary("default", ba) # user + writeStringBinary("", ba) # pwd + self.send(ba) + + def receiveHello(self): + p_type = self.readVarUInt() + assert p_type == 0 # Hello + _server_name = self.readStringBinary() + _server_version_major = self.readVarUInt() + _server_version_minor = self.readVarUInt() + _server_revision = self.readVarUInt() + _server_timezone = self.readStringBinary() + _server_display_name = self.readStringBinary() + _server_version_patch = self.readVarUInt() + + def sendQuery(self, query, settings=None): + if settings == None: + settings = {} # No settings + + ba = bytearray() + query_id = uuid.uuid4().hex + writeVarUInt(1, ba) # query + writeStringBinary(query_id, ba) + + ba.append(1) # INITIAL_QUERY + + # client info + serializeClientInfo(ba, query_id) + + # Settings + for key, value in settings.items(): + writeStringBinary(key, ba) + writeVarUInt(1, ba) # is_important + writeStringBinary(str(value), ba) + writeStringBinary("", ba) # End of settings + + writeStringBinary("", ba) # No interserver secret + writeVarUInt(2, ba) # Stage - Complete + ba.append(0) # No compression + writeStringBinary(query, ba) # query, finally + self.send(ba) + + def sendEmptyBlock(self): + ba = bytearray() + writeVarUInt(2, ba) # Data + writeStringBinary("", ba) + serializeBlockInfo(ba) + writeVarUInt(0, ba) # rows + writeVarUInt(0, ba) # columns + self.send(ba) + + def readException(self): + code = self.readUInt32() + _name = self.readStringBinary() + text = self.readStringBinary() + self.readStringBinary() # trace + assertPacket(self.readUInt8(), 0) # has_nested + return "code {}: {}".format(code, text.replace("DB::Exception:", "")) + + def readResponse(self): + packet_type = self.readVarUInt() + if packet_type == 2: # Exception + raise RuntimeError(self.readException()) + + if packet_type == 1: # Data + return None + if packet_type == 3: # Progress + return None + if packet_type == 5: # End stream + return None + + raise RuntimeError("Unexpected packet: {}".format(packet_type)) + + def readData(self): + packet_type = self.readVarUInt() + if packet_type == 2: # Exception + raise RuntimeError(self.readException()) + + if packet_type == 5: # End stream + return None + + assertPacket(packet_type, 1) # Data + + self.readStringBinary() # external table name + + + def readProgress(self): + packet_type = self.readVarUInt() + if packet_type == 2: # Exception + raise RuntimeError(self.readException()) + + if packet_type == 5: # End stream + return None + + assertPacket(packet_type, 3) # Progress + + read_rows = self.readVarUInt() + read_bytes = self.readVarUInt() + total_rows_to_read = self.readVarUInt() + written_rows = self.readVarUInt() + written_bytes = self.readVarUInt() + return read_rows, read_bytes, total_rows_to_read, written_rows, written_bytes + + def readRow(self, row_type, rows): + if (row_type == 'UInt64'): + row = [self.readUInt64() for _ in range(rows)] + return row + else: + raise RuntimeError("Currently python version of tcp client doesn't support the following type of row: {}".format(row_type)) + + def readHeader(self, need_read_data=False, need_print_info=True): + packet_type = self.readVarUInt() + if packet_type == 2: # Exception + raise RuntimeError(self.readException()) + assertPacket(packet_type, 1) # Data + + self.readStringBinary() # external table name + # BlockInfo + assertPacket(self.readVarUInt(), 1) # field number 1 + assertPacket(self.readUInt8(), 0) # is_overflows + assertPacket(self.readVarUInt(), 2) # field number 2 + assertPacket(self.readUInt32(), 4294967295) # bucket_num + assertPacket(self.readVarUInt(), 0) # 0 + columns = self.readVarUInt() # rows + rows = self.readVarUInt() # columns + + data = [] if need_read_data else None + if need_print_info: + print("Rows {} Columns {}".format(rows, columns)) + + for _ in range(columns): + col_name = self.readStringBinary() + type_name = self.readStringBinary() + if need_print_info: + print("Column {} type {}".format(col_name, type_name)) + + if need_read_data: + data.append((col_name, self.readRow(type_name, rows))) + + return data From c0f9dbbacb34fabbb2ce8431476f53d3086a1eb9 Mon Sep 17 00:00:00 2001 From: zvonand Date: Wed, 2 Aug 2023 15:51:27 +0200 Subject: [PATCH 025/433] shorten --- src/Storages/StorageFile.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/Storages/StorageFile.cpp b/src/Storages/StorageFile.cpp index 7e214b76e3d..c6c9fefac97 100644 --- a/src/Storages/StorageFile.cpp +++ b/src/Storages/StorageFile.cpp @@ -122,7 +122,7 @@ void listFilesWithFoldedRegexpMatchingImpl(const std::string & path_for_ls, return; const fs::directory_iterator end; - for (fs::directory_iterator it(path_for_ls, std::filesystem::directory_options::skip_permission_denied); it != end; ++it) + for (fs::directory_iterator it(path_for_ls, fs::directory_options::skip_permission_denied); it != end; ++it) { const std::string full_path = it->path().string(); const size_t last_slash = full_path.rfind('/'); From 06e187e58a68ae5e68cc4022b578933986d87dc6 Mon Sep 17 00:00:00 2001 From: alexX512 Date: Wed, 2 Aug 2023 18:15:25 +0000 Subject: [PATCH 026/433] Split mergeSortingTransform and mergeSortingPartialResultTransform into 2 files --- .../PullingAsyncPipelineExecutor.cpp | 4 +- src/Processors/Formats/IOutputFormat.cpp | 58 ++++++++++--------- src/Processors/Formats/IOutputFormat.h | 8 +-- src/Processors/Formats/LazyOutputFormat.h | 4 +- .../LimitPartialResultTransform.cpp | 42 ++++++++++++++ .../Transforms/LimitPartialResultTransform.h | 36 +++--------- .../MergeSortingPartialResultTransform.cpp | 49 ++++++++++++++++ .../MergeSortingPartialResultTransform.h | 25 ++++++++ .../Transforms/MergeSortingTransform.cpp | 54 +---------------- .../Transforms/PartialResultTransform.cpp | 35 ++++++----- .../Transforms/PartialResultTransform.h | 6 +- src/QueryPipeline/Pipe.cpp | 5 +- src/QueryPipeline/Pipe.h | 2 +- src/QueryPipeline/QueryPipeline.cpp | 2 +- 14 files changed, 190 insertions(+), 140 deletions(-) create mode 100644 src/Processors/Transforms/LimitPartialResultTransform.cpp create mode 100644 src/Processors/Transforms/MergeSortingPartialResultTransform.cpp create mode 100644 src/Processors/Transforms/MergeSortingPartialResultTransform.h diff --git a/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp b/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp index 8875ba85a07..4b2e8e55904 100644 --- a/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp +++ b/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp @@ -46,9 +46,7 @@ PullingAsyncPipelineExecutor::PullingAsyncPipelineExecutor(QueryPipeline & pipel if (!pipeline.pulling()) throw Exception(ErrorCodes::LOGICAL_ERROR, "Pipeline for PullingAsyncPipelineExecutor must be pulling"); - lazy_format = std::make_shared(pipeline.output->getHeader()); - if (has_partial_result_setting) - lazy_format->activatePartialResultProtocol(); + lazy_format = std::make_shared(pipeline.output->getHeader(), /*is_partial_result_protocol_active*/ has_partial_result_setting); pipeline.complete(lazy_format); } diff --git a/src/Processors/Formats/IOutputFormat.cpp b/src/Processors/Formats/IOutputFormat.cpp index 7dba4b22eab..6104037a72b 100644 --- a/src/Processors/Formats/IOutputFormat.cpp +++ b/src/Processors/Formats/IOutputFormat.cpp @@ -6,30 +6,18 @@ namespace DB { -IOutputFormat::IOutputFormat(const Block & header_, WriteBuffer & out_) - : IProcessor({header_, header_, header_, header_}, {}), out(out_) +IOutputFormat::IOutputFormat(const Block & header_, WriteBuffer & out_, bool is_partial_result_protocol_active_) + : IProcessor({header_, header_, header_, header_}, {}) + , out(out_) + , is_partial_result_protocol_active(is_partial_result_protocol_active_) { } -IOutputFormat::Status IOutputFormat::prepare() +void IOutputFormat::setCurrentChunk(InputPort & input, PortKind kind) { - if (has_input) - return Status::Ready; - - auto status = prepareMainAndPartialResult(); - if (status != Status::Finished) - return status; - - status = prepareTotalsAndExtremes(); - if (status != Status::Finished) - return status; - - finished = true; - - if (!finalized) - return Status::Ready; - - return Status::Finished; + current_chunk = input.pull(true); + current_block_kind = kind; + has_input = true; } IOutputFormat::Status IOutputFormat::prepareMainAndPartialResult() @@ -42,7 +30,7 @@ IOutputFormat::Status IOutputFormat::prepareMainAndPartialResult() if (input.isFinished()) continue; - if (kind == PartialResult && was_main_input) + if (kind == PartialResult && main_input_activated) { input.close(); continue; @@ -84,11 +72,25 @@ IOutputFormat::Status IOutputFormat::prepareTotalsAndExtremes() return Status::Finished; } -void IOutputFormat::setCurrentChunk(InputPort & input, PortKind kind) +IOutputFormat::Status IOutputFormat::prepare() { - current_chunk = input.pull(true); - current_block_kind = kind; - has_input = true; + if (has_input) + return Status::Ready; + + auto status = prepareMainAndPartialResult(); + if (status != Status::Finished) + return status; + + status = prepareTotalsAndExtremes(); + if (status != Status::Finished) + return status; + + finished = true; + + if (!finalized) + return Status::Ready; + + return Status::Finished; } static Chunk prepareTotals(Chunk chunk) @@ -130,10 +132,12 @@ void IOutputFormat::work() case Main: result_rows += current_chunk.getNumRows(); result_bytes += current_chunk.allocatedBytes(); - if (is_partial_result_protocol_active && !was_main_input && current_chunk.hasRows()) + if (is_partial_result_protocol_active && !main_input_activated && current_chunk.hasRows()) { + /// Sending an empty block signals to the client that partial results are terminated, + /// and only data from the main pipeline will be forwarded. consume(Chunk(current_chunk.cloneEmptyColumns(), 0)); - was_main_input = true; + main_input_activated = true; } consume(std::move(current_chunk)); break; diff --git a/src/Processors/Formats/IOutputFormat.h b/src/Processors/Formats/IOutputFormat.h index 6e5e21055a3..fc85b0d063b 100644 --- a/src/Processors/Formats/IOutputFormat.h +++ b/src/Processors/Formats/IOutputFormat.h @@ -25,7 +25,7 @@ class IOutputFormat : public IProcessor public: enum PortKind { Main = 0, Totals = 1, Extremes = 2, PartialResult = 3 }; - IOutputFormat(const Block & header_, WriteBuffer & out_); + IOutputFormat(const Block & header_, WriteBuffer & out_, bool is_partial_result_protocol_active_ = false); Status prepare() override; void work() override; @@ -105,8 +105,6 @@ public: void clearLastLines(size_t lines_number); - void activatePartialResultProtocol() { is_partial_result_protocol_active = true; } - protected: friend class ParallelFormattingOutputFormat; @@ -172,7 +170,7 @@ protected: Chunk current_chunk; PortKind current_block_kind = PortKind::Main; - bool was_main_input = false; + bool main_input_activated = false; bool has_input = false; bool finished = false; bool finalized = false; @@ -187,9 +185,9 @@ protected: Statistics statistics; private: + void setCurrentChunk(InputPort & input, PortKind kind); IOutputFormat::Status prepareMainAndPartialResult(); IOutputFormat::Status prepareTotalsAndExtremes(); - void setCurrentChunk(InputPort & input, PortKind kind); size_t rows_read_before = 0; bool are_totals_written = false; diff --git a/src/Processors/Formats/LazyOutputFormat.h b/src/Processors/Formats/LazyOutputFormat.h index 3b3dd365458..bbcfdbb7193 100644 --- a/src/Processors/Formats/LazyOutputFormat.h +++ b/src/Processors/Formats/LazyOutputFormat.h @@ -14,8 +14,8 @@ class LazyOutputFormat : public IOutputFormat { public: - explicit LazyOutputFormat(const Block & header) - : IOutputFormat(header, out), queue(2) {} + explicit LazyOutputFormat(const Block & header, bool is_partial_result_protocol_active = false) + : IOutputFormat(header, out, is_partial_result_protocol_active), queue(2) {} String getName() const override { return "LazyOutputFormat"; } diff --git a/src/Processors/Transforms/LimitPartialResultTransform.cpp b/src/Processors/Transforms/LimitPartialResultTransform.cpp new file mode 100644 index 00000000000..fdaa95aac6c --- /dev/null +++ b/src/Processors/Transforms/LimitPartialResultTransform.cpp @@ -0,0 +1,42 @@ +#include +#include + +namespace DB +{ + +LimitPartialResultTransform::LimitPartialResultTransform( + const Block & header, + UInt64 partial_result_limit_, + UInt64 partial_result_duration_ms_, + UInt64 limit_, + UInt64 offset_) + : PartialResultTransform(header, partial_result_limit_, partial_result_duration_ms_) + , limit(limit_) + , offset(offset_) + {} + +void LimitPartialResultTransform::transformPartialResult(Chunk & chunk) +{ + UInt64 num_rows = chunk.getNumRows(); + if (num_rows < offset || limit == 0) + { + chunk = {}; + return; + } + + UInt64 length = std::min(limit, num_rows - offset); + + /// Check if some rows should be removed + if (length < num_rows) + { + auto columns = chunk.detachColumns(); + UInt64 num_columns = chunk.getNumColumns(); + + for (UInt64 i = 0; i < num_columns; ++i) + columns[i] = columns[i]->cut(offset, limit); + + chunk.setColumns(std::move(columns), length); + } +} + +} diff --git a/src/Processors/Transforms/LimitPartialResultTransform.h b/src/Processors/Transforms/LimitPartialResultTransform.h index 4b1f0e1026a..0f5549323c0 100644 --- a/src/Processors/Transforms/LimitPartialResultTransform.h +++ b/src/Processors/Transforms/LimitPartialResultTransform.h @@ -5,50 +5,30 @@ namespace DB { +class LimitTransform; + /// Currently support only single thread implementation with one input and one output ports class LimitPartialResultTransform : public PartialResultTransform { public: + using LimitTransformPtr = std::shared_ptr; + LimitPartialResultTransform( const Block & header, UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_, UInt64 limit_, - UInt64 offset_) - : PartialResultTransform(header, partial_result_limit_, partial_result_duration_ms_) - , limit(limit_) - , offset(offset_) - {} + UInt64 offset_); String getName() const override { return "LimitPartialResultTransform"; } - void transformPartialResult(Chunk & chunk) override - { - UInt64 num_rows = chunk.getNumRows(); - if (num_rows < offset || limit == 0) - { - chunk = {}; - return; - } - - UInt64 length = std::min(limit, num_rows - offset); - - /// Check if some rows should be removed - if (length < num_rows) - { - auto columns = chunk.detachColumns(); - UInt64 num_columns = chunk.getNumColumns(); - - for (UInt64 i = 0; i < num_columns; ++i) - columns[i] = columns[i]->cut(offset, limit); - - chunk.setColumns(std::move(columns), length); - } - } + void transformPartialResult(Chunk & chunk) override; private: UInt64 limit; UInt64 offset; + + LimitTransformPtr limit_transform; }; } diff --git a/src/Processors/Transforms/MergeSortingPartialResultTransform.cpp b/src/Processors/Transforms/MergeSortingPartialResultTransform.cpp new file mode 100644 index 00000000000..21513fe1faf --- /dev/null +++ b/src/Processors/Transforms/MergeSortingPartialResultTransform.cpp @@ -0,0 +1,49 @@ +#include +#include + +namespace DB +{ + +MergeSortingPartialResultTransform::MergeSortingPartialResultTransform( + const Block & header, MergeSortingTransformPtr merge_sorting_transform_, + UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_) + : PartialResultTransform(header, partial_result_limit_, partial_result_duration_ms_) + , merge_sorting_transform(std::move(merge_sorting_transform_)) + {} + +PartialResultTransform::ShaphotResult MergeSortingPartialResultTransform::getRealProcessorSnapshot() +{ + std::lock_guard lock(merge_sorting_transform->snapshot_mutex); + if (merge_sorting_transform->generated_prefix) + return {{}, SnaphotStatus::Stopped}; + + if (merge_sorting_transform->chunks.empty()) + return {{}, SnaphotStatus::NotReady}; + + /// Sort all input data + merge_sorting_transform->remerge(); + /// Add a copy of the first `partial_result_limit` rows to a generated_chunk + /// to send it later as a partial result in the next prepare stage of the current processor + auto generated_columns = merge_sorting_transform->chunks[0].cloneEmptyColumns(); + size_t total_rows = 0; + for (const auto & merged_chunk : merge_sorting_transform->chunks) + { + size_t rows = std::min(merged_chunk.getNumRows(), partial_result_limit - total_rows); + if (rows == 0) + break; + + for (size_t position = 0; position < generated_columns.size(); ++position) + { + auto column = merged_chunk.getColumns()[position]; + generated_columns[position]->insertRangeFrom(*column, 0, rows); + } + + total_rows += rows; + } + + auto partial_result = Chunk(std::move(generated_columns), total_rows, merge_sorting_transform->chunks[0].getChunkInfo()); + merge_sorting_transform->enrichChunkWithConstants(partial_result); + return {std::move(partial_result), SnaphotStatus::Ready}; +} + +} diff --git a/src/Processors/Transforms/MergeSortingPartialResultTransform.h b/src/Processors/Transforms/MergeSortingPartialResultTransform.h new file mode 100644 index 00000000000..9ba937d9109 --- /dev/null +++ b/src/Processors/Transforms/MergeSortingPartialResultTransform.h @@ -0,0 +1,25 @@ +#pragma once + +#include + +namespace DB +{ + +class MergeSortingPartialResultTransform : public PartialResultTransform +{ +public: + using MergeSortingTransformPtr = std::shared_ptr; + + MergeSortingPartialResultTransform( + const Block & header, MergeSortingTransformPtr merge_sorting_transform_, + UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_); + + String getName() const override { return "MergeSortingPartialResultTransform"; } + + ShaphotResult getRealProcessorSnapshot() override; + +private: + MergeSortingTransformPtr merge_sorting_transform; +}; + +} diff --git a/src/Processors/Transforms/MergeSortingTransform.cpp b/src/Processors/Transforms/MergeSortingTransform.cpp index 23114244c7a..d21cb179b0a 100644 --- a/src/Processors/Transforms/MergeSortingTransform.cpp +++ b/src/Processors/Transforms/MergeSortingTransform.cpp @@ -1,5 +1,5 @@ #include -#include +#include #include #include #include @@ -82,58 +82,6 @@ private: Poco::Logger * log; }; -class MergeSortingPartialResultTransform : public PartialResultTransform -{ -public: - using MergeSortingTransformPtr = std::shared_ptr; - - MergeSortingPartialResultTransform(const Block & header, MergeSortingTransformPtr merge_sorting_transform_, UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_) - : PartialResultTransform(header, partial_result_limit_, partial_result_duration_ms_) - , merge_sorting_transform(std::move(merge_sorting_transform_)) - { - } - - String getName() const override { return "MergeSortingPartialResultTransform"; } - - ShaphotResult getRealProcessorSnapshot() override - { - std::lock_guard lock(merge_sorting_transform->snapshot_mutex); - if (merge_sorting_transform->generated_prefix) - return {{}, SnaphotStatus::Stopped}; - - if (merge_sorting_transform->chunks.empty()) - return {{}, SnaphotStatus::NotReady}; - - /// Sort all input data - merge_sorting_transform->remerge(); - /// Add a copy of the first `partial_result_limit` rows to a generated_chunk - /// to send it as a partial result in the next prepare stage - auto generated_columns = merge_sorting_transform->chunks[0].cloneEmptyColumns(); - size_t total_rows = 0; - for (const auto & merged_chunk : merge_sorting_transform->chunks) - { - size_t rows = std::min(merged_chunk.getNumRows(), partial_result_limit - total_rows); - if (rows == 0) - break; - - for (size_t position = 0; position < generated_columns.size(); ++position) - { - auto column = merged_chunk.getColumns()[position]; - generated_columns[position]->insertRangeFrom(*column, 0, rows); - } - - total_rows += rows; - } - - auto partial_result = Chunk(std::move(generated_columns), total_rows, merge_sorting_transform->chunks[0].getChunkInfo()); - merge_sorting_transform->enrichChunkWithConstants(partial_result); - return {std::move(partial_result), SnaphotStatus::Ready}; - } - -private: - MergeSortingTransformPtr merge_sorting_transform; -}; - MergeSortingTransform::MergeSortingTransform( const Block & header, const SortDescription & description_, diff --git a/src/Processors/Transforms/PartialResultTransform.cpp b/src/Processors/Transforms/PartialResultTransform.cpp index 7cf2473684a..4703eceddda 100644 --- a/src/Processors/Transforms/PartialResultTransform.cpp +++ b/src/Processors/Transforms/PartialResultTransform.cpp @@ -26,29 +26,38 @@ IProcessor::Status PartialResultTransform::prepare() return Status::Finished; } - if (input.hasData()) - partial_result = {input.pull(), SnaphotStatus::Ready}; - - /// Send partial result from real processor snapshot or from previous partial result processor if possible - if (partial_result.snapshot_status == SnaphotStatus::Ready && output.canPush()) + if (!output.canPush()) { - transformPartialResult(partial_result.chunk); - partial_result.snapshot_status = SnaphotStatus::NotReady; - if (partial_result.chunk.getNumRows() > 0) - { - output.push(std::move(partial_result.chunk)); - return Status::PortFull; - } - } + input.setNotNeeded(); + return Status::PortFull; + } /// If input data from previous partial result processor is finished then /// PartialResultTransform ready to create snapshots and send them as a partial result if (input.isFinished()) { + if (partial_result.snapshot_status == SnaphotStatus::Ready) + { + partial_result.snapshot_status = SnaphotStatus::NotReady; + output.push(std::move(partial_result.chunk)); + return Status::PortFull; + } + return Status::Ready; } input.setNeeded(); + if (!input.hasData()) + return Status::NeedData; + + partial_result.chunk = input.pull(); + transformPartialResult(partial_result.chunk); + if (partial_result.chunk.getNumRows() > 0) + { + output.push(std::move(partial_result.chunk)); + return Status::PortFull; + } + return Status::NeedData; } diff --git a/src/Processors/Transforms/PartialResultTransform.h b/src/Processors/Transforms/PartialResultTransform.h index 3f9089f8cce..7d4b9026ce1 100644 --- a/src/Processors/Transforms/PartialResultTransform.h +++ b/src/Processors/Transforms/PartialResultTransform.h @@ -20,9 +20,9 @@ public: protected: enum class SnaphotStatus { - NotReady, - Ready, - Stopped, + NotReady, // Waiting for data from the previous partial result processor or awaiting a timer before creating the snapshot. + Ready, // Current partial result processor has received a snapshot from the processor in the main pipeline. + Stopped, // The processor from the main pipeline has started sending data, and the pipeline for partial results should use data from the next processors of the main pipeline. }; struct ShaphotResult diff --git a/src/QueryPipeline/Pipe.cpp b/src/QueryPipeline/Pipe.cpp index 6ff2cb6a4fd..adf06dde53b 100644 --- a/src/QueryPipeline/Pipe.cpp +++ b/src/QueryPipeline/Pipe.cpp @@ -325,10 +325,7 @@ Pipe Pipe::unitePipes(Pipes pipes, Processors * collected_processors, bool allow pipe.partial_result_ports.end()); } else - { - res.is_partial_result_active = false; - res.partial_result_ports.clear(); - } + res.dropPartialResult(); res.max_parallel_streams += pipe.max_parallel_streams; diff --git a/src/QueryPipeline/Pipe.h b/src/QueryPipeline/Pipe.h index cbb7a8cbf5c..70e933bcfd2 100644 --- a/src/QueryPipeline/Pipe.h +++ b/src/QueryPipeline/Pipe.h @@ -121,7 +121,7 @@ private: Block header; std::shared_ptr processors; - /// If is true, then on each addition of processor also try + /// If the variable is true, then each time a processor is added pipe will try /// to add processor which will send partial result from original processor bool is_partial_result_active = false; UInt64 partial_result_limit = 0; diff --git a/src/QueryPipeline/QueryPipeline.cpp b/src/QueryPipeline/QueryPipeline.cpp index 8fca23dba7d..741729cbd48 100644 --- a/src/QueryPipeline/QueryPipeline.cpp +++ b/src/QueryPipeline/QueryPipeline.cpp @@ -127,7 +127,7 @@ static void checkPulling( if (partial_result && !found_partial_result) throw Exception( ErrorCodes::LOGICAL_ERROR, - "Cannot create pulling QueryPipeline because its partial_result port does not belong to any processor"); + "Cannot create pulling QueryPipeline because its partial result port does not belong to any processor"); } static void checkCompleted(Processors & processors) From ee01d643506ea9231619b88caeeade5d753c7ab9 Mon Sep 17 00:00:00 2001 From: alexX512 Date: Wed, 2 Aug 2023 18:42:57 +0000 Subject: [PATCH 027/433] Style fix --- .../MergeSortingPartialResultTransform.cpp | 2 +- .../MergeSortingPartialResultTransform.h | 2 +- .../Transforms/PartialResultTransform.cpp | 4 ++-- .../0_stateless/02010_lc_native.python | 9 +++++++- .../02458_insert_select_progress_tcp.python | 9 +++++++- ...rting_result_during_query_execution.python | 23 +++++++++++-------- .../queries/0_stateless/helpers/tcp_client.py | 21 +++++++++-------- 7 files changed, 46 insertions(+), 24 deletions(-) diff --git a/src/Processors/Transforms/MergeSortingPartialResultTransform.cpp b/src/Processors/Transforms/MergeSortingPartialResultTransform.cpp index 21513fe1faf..2c4d6bea3c5 100644 --- a/src/Processors/Transforms/MergeSortingPartialResultTransform.cpp +++ b/src/Processors/Transforms/MergeSortingPartialResultTransform.cpp @@ -5,7 +5,7 @@ namespace DB { MergeSortingPartialResultTransform::MergeSortingPartialResultTransform( - const Block & header, MergeSortingTransformPtr merge_sorting_transform_, + const Block & header, MergeSortingTransformPtr merge_sorting_transform_, UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_) : PartialResultTransform(header, partial_result_limit_, partial_result_duration_ms_) , merge_sorting_transform(std::move(merge_sorting_transform_)) diff --git a/src/Processors/Transforms/MergeSortingPartialResultTransform.h b/src/Processors/Transforms/MergeSortingPartialResultTransform.h index 9ba937d9109..f08f74dad6b 100644 --- a/src/Processors/Transforms/MergeSortingPartialResultTransform.h +++ b/src/Processors/Transforms/MergeSortingPartialResultTransform.h @@ -11,7 +11,7 @@ public: using MergeSortingTransformPtr = std::shared_ptr; MergeSortingPartialResultTransform( - const Block & header, MergeSortingTransformPtr merge_sorting_transform_, + const Block & header, MergeSortingTransformPtr merge_sorting_transform_, UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_); String getName() const override { return "MergeSortingPartialResultTransform"; } diff --git a/src/Processors/Transforms/PartialResultTransform.cpp b/src/Processors/Transforms/PartialResultTransform.cpp index 4703eceddda..5e9704d38a2 100644 --- a/src/Processors/Transforms/PartialResultTransform.cpp +++ b/src/Processors/Transforms/PartialResultTransform.cpp @@ -30,7 +30,7 @@ IProcessor::Status PartialResultTransform::prepare() { input.setNotNeeded(); return Status::PortFull; - } + } /// If input data from previous partial result processor is finished then /// PartialResultTransform ready to create snapshots and send them as a partial result @@ -48,7 +48,7 @@ IProcessor::Status PartialResultTransform::prepare() input.setNeeded(); if (!input.hasData()) - return Status::NeedData; + return Status::NeedData; partial_result.chunk = input.pull(); transformPartialResult(partial_result.chunk); diff --git a/tests/queries/0_stateless/02010_lc_native.python b/tests/queries/0_stateless/02010_lc_native.python index 5054ad3d454..219fdf04472 100755 --- a/tests/queries/0_stateless/02010_lc_native.python +++ b/tests/queries/0_stateless/02010_lc_native.python @@ -7,7 +7,14 @@ import sys CURDIR = os.path.dirname(os.path.realpath(__file__)) sys.path.insert(0, os.path.join(CURDIR, "helpers")) -from tcp_client import TCPClient, CLICKHOUSE_DATABASE, writeVarUInt, writeStringBinary, serializeBlockInfo, assertPacket +from tcp_client import ( + TCPClient, + CLICKHOUSE_DATABASE, + writeVarUInt, + writeStringBinary, + serializeBlockInfo, + assertPacket, +) def insertValidLowCardinalityRow(): diff --git a/tests/queries/0_stateless/02458_insert_select_progress_tcp.python b/tests/queries/0_stateless/02458_insert_select_progress_tcp.python index 1fe2363d7d5..929107093d0 100644 --- a/tests/queries/0_stateless/02458_insert_select_progress_tcp.python +++ b/tests/queries/0_stateless/02458_insert_select_progress_tcp.python @@ -11,7 +11,14 @@ from tcp_client import TCPClient class Progress: - def __init__(self, read_rows=0, read_bytes=0, total_rows_to_read=0, written_rows=0, written_bytes=0): + def __init__( + self, + read_rows=0, + read_bytes=0, + total_rows_to_read=0, + written_rows=0, + written_bytes=0 + ): # NOTE: this is done in ctor to initialize __dict__ self.read_rows = read_rows self.read_bytes = read_bytes diff --git a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python index 00c11d0d233..a341cbd3106 100755 --- a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python +++ b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python @@ -20,24 +20,30 @@ def main(): # external tables client.sendEmptyBlock() client.readHeader() - + # Partial result _, partial_result = client.readHeader(need_read_data=True)[0] - assert_message = "There should be at least one block of data with partial result" + assert_message = ( + "There should be at least one block of data with partial result" + ) assert len(partial_result) > 0, assert_message - + while True: - _, new_partial_result = client.readHeader(need_read_data=True, need_print_info=False)[0] - if (len(new_partial_result) == 0): + _, new_partial_result = client.readHeader( + need_read_data=True, need_print_info=False + )[0] + if len(new_partial_result) == 0: break - + data_size = len(partial_result) assert_message = f"New block contains more info about the full data so sorted results should not be less then in the previous iteration. New result {new_partial_result}. Previous result {partial_result}" - assert all(partial_result[i] <= new_partial_result[i] for i in range(data_size)), assert_message + assert all( + partial_result[i] <= new_partial_result[i] for i in range(data_size) + ), assert_message partial_result = new_partial_result - + # Full result _, full_result = client.readHeader(need_read_data=True)[0] for result in full_result: @@ -46,4 +52,3 @@ def main(): if __name__ == "__main__": main() - diff --git a/tests/queries/0_stateless/helpers/tcp_client.py b/tests/queries/0_stateless/helpers/tcp_client.py index d1c6cb4894c..ca10f74b812 100644 --- a/tests/queries/0_stateless/helpers/tcp_client.py +++ b/tests/queries/0_stateless/helpers/tcp_client.py @@ -17,7 +17,7 @@ def writeVarUInt(x, ba): x >>= 7 if x == 0: - return + return def writeStringBinary(s, ba): @@ -68,7 +68,7 @@ class TCPClient(object): self.sendHello() self.receiveHello() - + return self def __exit__(self, exc_type, exc_value, traceback): @@ -85,7 +85,7 @@ class TCPClient(object): res.extend(cur) return res - + def readUInt(self, size=1): res = self.readStrict(size) val = 0 @@ -182,7 +182,7 @@ class TCPClient(object): writeVarUInt(0, ba) # rows writeVarUInt(0, ba) # columns self.send(ba) - + def readException(self): code = self.readUInt32() _name = self.readStringBinary() @@ -217,7 +217,6 @@ class TCPClient(object): self.readStringBinary() # external table name - def readProgress(self): packet_type = self.readVarUInt() if packet_type == 2: # Exception @@ -236,11 +235,15 @@ class TCPClient(object): return read_rows, read_bytes, total_rows_to_read, written_rows, written_bytes def readRow(self, row_type, rows): - if (row_type == 'UInt64'): + if row_type == 'UInt64': row = [self.readUInt64() for _ in range(rows)] return row else: - raise RuntimeError("Currently python version of tcp client doesn't support the following type of row: {}".format(row_type)) + raise RuntimeError( + "Currently python version of tcp client doesn't support the following type of row: {}".format( + row_type + ) + ) def readHeader(self, need_read_data=False, need_print_info=True): packet_type = self.readVarUInt() @@ -257,11 +260,11 @@ class TCPClient(object): assertPacket(self.readVarUInt(), 0) # 0 columns = self.readVarUInt() # rows rows = self.readVarUInt() # columns - + data = [] if need_read_data else None if need_print_info: print("Rows {} Columns {}".format(rows, columns)) - + for _ in range(columns): col_name = self.readStringBinary() type_name = self.readStringBinary() From ff8027b74b1bbf65111a134e80fa14296afacbbc Mon Sep 17 00:00:00 2001 From: alexX512 Date: Wed, 2 Aug 2023 19:01:08 +0000 Subject: [PATCH 028/433] Style fix --- .../0_stateless/02458_insert_select_progress_tcp.python | 2 +- ...02833_partial_sorting_result_during_query_execution.python | 2 +- tests/queries/0_stateless/helpers/tcp_client.py | 4 ++-- 3 files changed, 4 insertions(+), 4 deletions(-) diff --git a/tests/queries/0_stateless/02458_insert_select_progress_tcp.python b/tests/queries/0_stateless/02458_insert_select_progress_tcp.python index 929107093d0..fdc64a8dba8 100644 --- a/tests/queries/0_stateless/02458_insert_select_progress_tcp.python +++ b/tests/queries/0_stateless/02458_insert_select_progress_tcp.python @@ -17,7 +17,7 @@ class Progress: read_bytes=0, total_rows_to_read=0, written_rows=0, - written_bytes=0 + written_bytes=0, ): # NOTE: this is done in ctor to initialize __dict__ self.read_rows = read_rows diff --git a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python index a341cbd3106..ce31431ba97 100755 --- a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python +++ b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python @@ -23,7 +23,7 @@ def main(): # Partial result _, partial_result = client.readHeader(need_read_data=True)[0] - + assert_message = ( "There should be at least one block of data with partial result" ) diff --git a/tests/queries/0_stateless/helpers/tcp_client.py b/tests/queries/0_stateless/helpers/tcp_client.py index ca10f74b812..5b1e312b229 100644 --- a/tests/queries/0_stateless/helpers/tcp_client.py +++ b/tests/queries/0_stateless/helpers/tcp_client.py @@ -214,7 +214,7 @@ class TCPClient(object): return None assertPacket(packet_type, 1) # Data - + self.readStringBinary() # external table name def readProgress(self): @@ -235,7 +235,7 @@ class TCPClient(object): return read_rows, read_bytes, total_rows_to_read, written_rows, written_bytes def readRow(self, row_type, rows): - if row_type == 'UInt64': + if row_type == "UInt64": row = [self.readUInt64() for _ in range(rows)] return row else: From 9335a494dae52e7652d26d5cb0983fcf5c02bca9 Mon Sep 17 00:00:00 2001 From: alexX512 Date: Thu, 3 Aug 2023 07:51:25 +0000 Subject: [PATCH 029/433] Read data without progress for partial result test --- ...rting_result_during_query_execution.python | 8 +- .../queries/0_stateless/helpers/tcp_client.py | 89 ++++++++++--------- 2 files changed, 53 insertions(+), 44 deletions(-) diff --git a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python index ce31431ba97..14f5a912672 100755 --- a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python +++ b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python @@ -22,7 +22,7 @@ def main(): client.readHeader() # Partial result - _, partial_result = client.readHeader(need_read_data=True)[0] + _, partial_result = client.readDataWithoutProgress()[0] assert_message = ( "There should be at least one block of data with partial result" @@ -30,8 +30,8 @@ def main(): assert len(partial_result) > 0, assert_message while True: - _, new_partial_result = client.readHeader( - need_read_data=True, need_print_info=False + _, new_partial_result = client.readDataWithoutProgress( + need_print_info=False )[0] if len(new_partial_result) == 0: break @@ -45,7 +45,7 @@ def main(): partial_result = new_partial_result # Full result - _, full_result = client.readHeader(need_read_data=True)[0] + _, full_result = client.readDataWithoutProgress()[0] for result in full_result: print(result) diff --git a/tests/queries/0_stateless/helpers/tcp_client.py b/tests/queries/0_stateless/helpers/tcp_client.py index 5b1e312b229..1c8b7ae0498 100644 --- a/tests/queries/0_stateless/helpers/tcp_client.py +++ b/tests/queries/0_stateless/helpers/tcp_client.py @@ -191,11 +191,15 @@ class TCPClient(object): assertPacket(self.readUInt8(), 0) # has_nested return "code {}: {}".format(code, text.replace("DB::Exception:", "")) - def readResponse(self): + def readPacketType(self): packet_type = self.readVarUInt() if packet_type == 2: # Exception raise RuntimeError(self.readException()) + return packet_type + + def readResponse(self): + packet_type = self.readPacketType() if packet_type == 1: # Data return None if packet_type == 3: # Progress @@ -205,35 +209,46 @@ class TCPClient(object): raise RuntimeError("Unexpected packet: {}".format(packet_type)) - def readData(self): - packet_type = self.readVarUInt() - if packet_type == 2: # Exception - raise RuntimeError(self.readException()) - - if packet_type == 5: # End stream - return None - - assertPacket(packet_type, 1) # Data - - self.readStringBinary() # external table name - - def readProgress(self): - packet_type = self.readVarUInt() - if packet_type == 2: # Exception - raise RuntimeError(self.readException()) - - if packet_type == 5: # End stream - return None - - assertPacket(packet_type, 3) # Progress - + def readProgressData(self): read_rows = self.readVarUInt() read_bytes = self.readVarUInt() total_rows_to_read = self.readVarUInt() written_rows = self.readVarUInt() written_bytes = self.readVarUInt() + return read_rows, read_bytes, total_rows_to_read, written_rows, written_bytes + def readProgress(self): + packet_type = self.readPacketType() + if packet_type == 5: # End stream + return None + assertPacket(packet_type, 3) # Progress + return self.readProgressData() + + def readHeaderInfo(self): + self.readStringBinary() # external table name + # BlockInfo + assertPacket(self.readVarUInt(), 1) # field number 1 + assertPacket(self.readUInt8(), 0) # is_overflows + assertPacket(self.readVarUInt(), 2) # field number 2 + assertPacket(self.readUInt32(), 4294967295) # bucket_num + assertPacket(self.readVarUInt(), 0) # 0 + columns = self.readVarUInt() # rows + rows = self.readVarUInt() # columns + + return columns, rows + + def readHeader(self): + packet_type = self.readPacketType() + assertPacket(packet_type, 1) # Data + + columns, rows = self.readHeaderInfo() + print("Rows {} Columns {}".format(rows, columns)) + for _ in range(columns): + col_name = self.readStringBinary() + type_name = self.readStringBinary() + print("Column {} type {}".format(col_name, type_name)) + def readRow(self, row_type, rows): if row_type == "UInt64": row = [self.readUInt64() for _ in range(rows)] @@ -245,23 +260,18 @@ class TCPClient(object): ) ) - def readHeader(self, need_read_data=False, need_print_info=True): - packet_type = self.readVarUInt() - if packet_type == 2: # Exception - raise RuntimeError(self.readException()) + def readDataWithoutProgress(self, need_print_info=True): + packet_type = self.readPacketType() + while packet_type == 3: # Progress + self.readProgressData() + packet_type = self.readPacketType() + + if packet_type == 5: # End stream + return None assertPacket(packet_type, 1) # Data - self.readStringBinary() # external table name - # BlockInfo - assertPacket(self.readVarUInt(), 1) # field number 1 - assertPacket(self.readUInt8(), 0) # is_overflows - assertPacket(self.readVarUInt(), 2) # field number 2 - assertPacket(self.readUInt32(), 4294967295) # bucket_num - assertPacket(self.readVarUInt(), 0) # 0 - columns = self.readVarUInt() # rows - rows = self.readVarUInt() # columns - - data = [] if need_read_data else None + columns, rows = self.readHeaderInfo() + data = [] if need_print_info: print("Rows {} Columns {}".format(rows, columns)) @@ -271,7 +281,6 @@ class TCPClient(object): if need_print_info: print("Column {} type {}".format(col_name, type_name)) - if need_read_data: - data.append((col_name, self.readRow(type_name, rows))) + data.append((col_name, self.readRow(type_name, rows))) return data From de62239a5db53534ad8be9cb39ed7174388f9bc2 Mon Sep 17 00:00:00 2001 From: alexX512 Date: Thu, 3 Aug 2023 08:13:30 +0000 Subject: [PATCH 030/433] Style fix --- tests/queries/0_stateless/helpers/tcp_client.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tests/queries/0_stateless/helpers/tcp_client.py b/tests/queries/0_stateless/helpers/tcp_client.py index 1c8b7ae0498..2f0680b4eae 100644 --- a/tests/queries/0_stateless/helpers/tcp_client.py +++ b/tests/queries/0_stateless/helpers/tcp_client.py @@ -262,7 +262,7 @@ class TCPClient(object): def readDataWithoutProgress(self, need_print_info=True): packet_type = self.readPacketType() - while packet_type == 3: # Progress + while packet_type == 3: # Progress self.readProgressData() packet_type = self.readPacketType() From e797d877480386cb38dcdd5034129b9f4a469ef1 Mon Sep 17 00:00:00 2001 From: chen768959 <934103231@qq.com> Date: Thu, 3 Aug 2023 22:43:08 +0800 Subject: [PATCH 031/433] More precise Integer type inference --- src/DataTypes/DataTypesNumber.h | 16 ++++ src/DataTypes/FieldToDataType.cpp | 22 ++++- src/DataTypes/IDataType.h | 2 + src/DataTypes/getLeastSupertype.cpp | 132 ++++++++++++++++++++++++++++ src/DataTypes/getLeastSupertype.h | 2 + 5 files changed, 171 insertions(+), 3 deletions(-) diff --git a/src/DataTypes/DataTypesNumber.h b/src/DataTypes/DataTypesNumber.h index 5843086248c..8511955e193 100644 --- a/src/DataTypes/DataTypesNumber.h +++ b/src/DataTypes/DataTypesNumber.h @@ -1,6 +1,7 @@ #pragma once #include +#include #include #include #include @@ -9,10 +10,20 @@ namespace DB { +using DataTypes = std::vector; + template class DataTypeNumber final : public DataTypeNumberBase { public: + DataTypeNumber() = default; + + explicit DataTypeNumber(DataTypes data_types) + : DataTypeNumberBase() + , possible_data_types(std::move(data_types)) + { + } + bool equals(const IDataType & rhs) const override { return typeid(rhs) == typeid(*this); } bool canBeUsedAsVersion() const override { return true; } @@ -32,6 +43,11 @@ public: { return std::make_shared>(); } + + DataTypes getPossiblePtr() const override { return possible_data_types; } + +private: + DataTypes possible_data_types; }; using DataTypeUInt8 = DataTypeNumber; diff --git a/src/DataTypes/FieldToDataType.cpp b/src/DataTypes/FieldToDataType.cpp index 210dab9921e..0a48c43ceb2 100644 --- a/src/DataTypes/FieldToDataType.cpp +++ b/src/DataTypes/FieldToDataType.cpp @@ -33,18 +33,34 @@ DataTypePtr FieldToDataType::operator() (const Null &) const template DataTypePtr FieldToDataType::operator() (const UInt64 & x) const { + if (x <= std::numeric_limits::max()) return std::make_shared(DataTypes{ std::make_shared() }); if (x <= std::numeric_limits::max()) return std::make_shared(); + if (x <= std::numeric_limits::max()) return std::make_shared(DataTypes{ std::make_shared() }); if (x <= std::numeric_limits::max()) return std::make_shared(); + if (x <= std::numeric_limits::max()) return std::make_shared(DataTypes{ std::make_shared() }); if (x <= std::numeric_limits::max()) return std::make_shared(); + if (x <= std::numeric_limits::max()) return std::make_shared(DataTypes{ std::make_shared() }); return std::make_shared(); } template DataTypePtr FieldToDataType::operator() (const Int64 & x) const { - if (x <= std::numeric_limits::max() && x >= std::numeric_limits::min()) return std::make_shared(); - if (x <= std::numeric_limits::max() && x >= std::numeric_limits::min()) return std::make_shared(); - if (x <= std::numeric_limits::max() && x >= std::numeric_limits::min()) return std::make_shared(); + if (x >= 0) + { + if (x <= std::numeric_limits::max()) return std::make_shared(); + if (x <= std::numeric_limits::max()) return std::make_shared(DataTypes{ std::make_shared() }); + if (x <= std::numeric_limits::max()) return std::make_shared(); + if (x <= std::numeric_limits::max()) return std::make_shared(DataTypes{ std::make_shared() }); + if (x <= std::numeric_limits::max()) return std::make_shared(); + if (x <= std::numeric_limits::max()) return std::make_shared(DataTypes{ std::make_shared() }); + } + else + { + if (x >= std::numeric_limits::min()) return std::make_shared(); + if (x >= std::numeric_limits::min()) return std::make_shared(); + if (x >= std::numeric_limits::min()) return std::make_shared(); + } return std::make_shared(); } diff --git a/src/DataTypes/IDataType.h b/src/DataTypes/IDataType.h index 4adafe5d212..330836f9d92 100644 --- a/src/DataTypes/IDataType.h +++ b/src/DataTypes/IDataType.h @@ -73,6 +73,8 @@ public: DataTypePtr getPtr() const { return shared_from_this(); } + virtual DataTypes getPossiblePtr() const { return { shared_from_this() }; } + /// Name of data type family (example: FixedString, Array). virtual const char * getFamilyName() const = 0; /// Name of corresponding data type in MySQL (exampe: Bigint, Blob, etc) diff --git a/src/DataTypes/getLeastSupertype.cpp b/src/DataTypes/getLeastSupertype.cpp index 9d42d82ce91..3a7a1560af1 100644 --- a/src/DataTypes/getLeastSupertype.cpp +++ b/src/DataTypes/getLeastSupertype.cpp @@ -592,6 +592,7 @@ DataTypePtr getLeastSupertype(const DataTypes & types) /// For numeric types, the most complicated part. { + optimizeTypeIds(types, type_ids); auto numeric_type = getNumericType(type_ids); if (numeric_type) return numeric_type; @@ -601,6 +602,137 @@ DataTypePtr getLeastSupertype(const DataTypes & types) return throwOrReturn(types, "", ErrorCodes::NO_COMMON_TYPE); } +void optimizeTypeIds(const DataTypes & types, TypeIndexSet & type_ids) +{ + // Determine whether the type_id is UInt + auto is_unsigned = [](const TypeIndex & type_id) + { + switch (type_id) + { + case TypeIndex::UInt8: + case TypeIndex::UInt16: + case TypeIndex::UInt32: + case TypeIndex::UInt64: + return true; + default: + return false; + } + }; + + auto maximize = [](size_t & what, size_t value, bool & only_unsigned, bool & only_signed, bool & both) + { + if (value > what) + { + what = value; + only_unsigned = false; + only_signed = false; + both = false; + return true; + }else if (value == what) + { + return true; + } + + return false; + }; + + size_t max_bits_of_integer = 0; + bool only_unsigned = false; + bool only_signed = false; + bool both = false; + + // Determine the distribution of maximum signed and unsigned, Example: + // Int64, Int64 = only_signed. + // UInt64, UInt64 = only_unsigned. + // UInt64(possible: Int64), Int64(possible: UInt64) = both. + // UInt64(possible: Int64), Int64 = both, only_signed. + for (const auto & type : types) + { + TypeIndex type_id = type->getTypeId(); + bool is_max_bits = false; + if (type_id == TypeIndex::UInt8 || type_id == TypeIndex::Int8) + is_max_bits = maximize(max_bits_of_integer, 8, only_unsigned, only_signed, both); + else if (type_id == TypeIndex::UInt16 || type_id == TypeIndex::Int16) + is_max_bits = maximize(max_bits_of_integer, 16, only_unsigned, only_signed, both); + else if (type_id == TypeIndex::UInt32 || type_id == TypeIndex::Int32) + is_max_bits = maximize(max_bits_of_integer, 32, only_unsigned, only_signed, both); + else if (type_id == TypeIndex::UInt64 || type_id == TypeIndex::Int64) + is_max_bits = maximize(max_bits_of_integer, 64, only_unsigned, only_signed, both); + + if (is_max_bits) + { + bool type_is_unsigned = is_unsigned(type_id); + bool type_is_both = false; + for (const auto & possible_type : type->getPossiblePtr()) + { + if (type_is_unsigned != is_unsigned(possible_type->getTypeId())) + { + type_is_both = true; + break; + } + } + + if (type_is_both) + both = true; + else if (type_is_unsigned) + only_unsigned = true; + else + only_signed = true; + } + } + + auto optimize_type_id = [&is_unsigned](const DataTypePtr & type, bool try_change_unsigned) + { + switch (type_id) + { + case TypeIndex::UInt8: + case TypeIndex::UInt16: + case TypeIndex::UInt32: + case TypeIndex::UInt64: + if (try_change_unsigned) + return type_id; + case TypeIndex::Int8: + case TypeIndex::Int16: + case TypeIndex::Int32: + case TypeIndex::Int64: + if (!try_change_unsigned) + return type_id; + default: + return type_id; + } + + for (const auto & other_type : type->getPossiblePtr()) + { + TypeIndex other_type_id = other_type->getTypeId(); + if ((try_change_unsigned && is_unsigned(other_type_id)) + || (!try_change_unsigned && !is_unsigned(other_type_id))) + { + return other_type_id; + } + } + + return type_id; + }; + + // optimize type_ids, Example: + // if only_signed. UInt64(possible: Int64), Int64 = Int64, Int64 + // if only_unsigned. Int64(possible: UInt64), UInt64 = UInt64, UInt64 + if (!(only_unsigned && only_signed) && (both || only_unsigned || only_signed)) { + type_ids.clear(); + for (const auto & type : types) + { + if (only_unsigned) + { + type_ids.insert(optimize_type_id(type, true)); + } + else if (both || only_signed) + { + type_ids.insert(optimize_type_id(type, false)); + } + } + } +} + DataTypePtr getLeastSupertypeOrString(const DataTypes & types) { return getLeastSupertype(types); diff --git a/src/DataTypes/getLeastSupertype.h b/src/DataTypes/getLeastSupertype.h index 2ef4a0e6850..0a3fa3c2536 100644 --- a/src/DataTypes/getLeastSupertype.h +++ b/src/DataTypes/getLeastSupertype.h @@ -29,6 +29,8 @@ DataTypePtr tryGetLeastSupertype(const DataTypes & types); using TypeIndexSet = std::unordered_set; +void optimizeTypeIds(const DataTypes & types, TypeIndexSet & type_ids); + template DataTypePtr getLeastSupertype(const TypeIndexSet & types); From 446465307d78ddaa99b8c625f0d6a8584ebccd67 Mon Sep 17 00:00:00 2001 From: chen768959 <934103231@qq.com> Date: Fri, 4 Aug 2023 09:26:42 +0800 Subject: [PATCH 032/433] add test. fix style. --- src/DataTypes/getLeastSupertype.cpp | 4 +++- .../0_stateless/02832_integer_type_inference.reference | 1 + tests/queries/0_stateless/02832_integer_type_inference.sql | 1 + 3 files changed, 5 insertions(+), 1 deletion(-) create mode 100644 tests/queries/0_stateless/02832_integer_type_inference.reference create mode 100644 tests/queries/0_stateless/02832_integer_type_inference.sql diff --git a/src/DataTypes/getLeastSupertype.cpp b/src/DataTypes/getLeastSupertype.cpp index 3a7a1560af1..07a3de4760b 100644 --- a/src/DataTypes/getLeastSupertype.cpp +++ b/src/DataTypes/getLeastSupertype.cpp @@ -683,6 +683,7 @@ void optimizeTypeIds(const DataTypes & types, TypeIndexSet & type_ids) auto optimize_type_id = [&is_unsigned](const DataTypePtr & type, bool try_change_unsigned) { + TypeIndex type_id = type->getTypeId(); switch (type_id) { case TypeIndex::UInt8: @@ -717,7 +718,8 @@ void optimizeTypeIds(const DataTypes & types, TypeIndexSet & type_ids) // optimize type_ids, Example: // if only_signed. UInt64(possible: Int64), Int64 = Int64, Int64 // if only_unsigned. Int64(possible: UInt64), UInt64 = UInt64, UInt64 - if (!(only_unsigned && only_signed) && (both || only_unsigned || only_signed)) { + if (!(only_unsigned && only_signed) && (both || only_unsigned || only_signed)) + { type_ids.clear(); for (const auto & type : types) { diff --git a/tests/queries/0_stateless/02832_integer_type_inference.reference b/tests/queries/0_stateless/02832_integer_type_inference.reference new file mode 100644 index 00000000000..2d072437c90 --- /dev/null +++ b/tests/queries/0_stateless/02832_integer_type_inference.reference @@ -0,0 +1 @@ +[-4741124612489978151,-3236599669630092879,5607475129431807682] diff --git a/tests/queries/0_stateless/02832_integer_type_inference.sql b/tests/queries/0_stateless/02832_integer_type_inference.sql new file mode 100644 index 00000000000..9734bc2966f --- /dev/null +++ b/tests/queries/0_stateless/02832_integer_type_inference.sql @@ -0,0 +1 @@ +select [-4741124612489978151, -3236599669630092879, 5607475129431807682]; From 57d69a10e360fd4cccb5acd0759aa9f44a21f7f6 Mon Sep 17 00:00:00 2001 From: chen768959 <934103231@qq.com> Date: Fri, 4 Aug 2023 09:55:22 +0800 Subject: [PATCH 033/433] add switch break --- src/DataTypes/getLeastSupertype.cpp | 2 ++ 1 file changed, 2 insertions(+) diff --git a/src/DataTypes/getLeastSupertype.cpp b/src/DataTypes/getLeastSupertype.cpp index 07a3de4760b..5a24a59a60e 100644 --- a/src/DataTypes/getLeastSupertype.cpp +++ b/src/DataTypes/getLeastSupertype.cpp @@ -692,12 +692,14 @@ void optimizeTypeIds(const DataTypes & types, TypeIndexSet & type_ids) case TypeIndex::UInt64: if (try_change_unsigned) return type_id; + break ; case TypeIndex::Int8: case TypeIndex::Int16: case TypeIndex::Int32: case TypeIndex::Int64: if (!try_change_unsigned) return type_id; + break ; default: return type_id; } From 5cb856d167a37c46ac31d441e5a688b45ddb0434 Mon Sep 17 00:00:00 2001 From: Antonio Andelic Date: Fri, 4 Aug 2023 11:53:22 +0000 Subject: [PATCH 034/433] Try batching multiple flush requests in Keeper --- src/Coordination/Changelog.cpp | 78 +++++++++++++++++++------ src/Coordination/KeeperStateManager.cpp | 3 + 2 files changed, 63 insertions(+), 18 deletions(-) diff --git a/src/Coordination/Changelog.cpp b/src/Coordination/Changelog.cpp index 94062140bac..7f972f67846 100644 --- a/src/Coordination/Changelog.cpp +++ b/src/Coordination/Changelog.cpp @@ -1014,8 +1014,60 @@ void Changelog::writeThread() { WriteOperation write_operation; bool batch_append_ok = true; - while (write_operations.pop(write_operation)) + size_t pending_appends = 0; + bool try_batch_flush = false; + + /// turn into setting + static constexpr size_t max_flush_batch_size = 1000; + + const auto flush_logs = [&](const auto & flush) { + LOG_INFO(log, "Flushing {} logs", pending_appends); + + { + std::lock_guard writer_lock(writer_mutex); + current_writer->flush(); + } + + { + std::lock_guard lock{durable_idx_mutex}; + last_durable_idx = flush.index; + } + + pending_appends = 0; + }; + + const auto notify_append_completion = [&] + { + durable_idx_cv.notify_all(); + + // we need to call completion callback in another thread because it takes a global lock for the NuRaft server + // NuRaft will in some places wait for flush to be done while having the same global lock leading to deadlock + // -> future write operations are blocked by flush that cannot be completed because it cannot take NuRaft lock + // -> NuRaft won't leave lock until its flush is done + if (!append_completion_queue.push(batch_append_ok)) + LOG_WARNING(log, "Changelog is shut down"); + }; + + while (true) + { + if (try_batch_flush) + { + try_batch_flush = false; + if (!write_operations.tryPop(write_operation)) + { + chassert(batch_append_ok); + const auto & flush = std::get(write_operation); + flush_logs(flush); + notify_append_completion(); + continue; + } + } + else if (!write_operations.pop(write_operation)) + { + break; + } + assert(initialized); if (auto * append_log = std::get_if(&write_operation)) @@ -1027,6 +1079,7 @@ void Changelog::writeThread() assert(current_writer); batch_append_ok = current_writer->appendRecord(buildRecord(append_log->index, append_log->log_entry)); + ++pending_appends; } else { @@ -1034,30 +1087,19 @@ void Changelog::writeThread() if (batch_append_ok) { + /// we can try batching more logs for flush + if (pending_appends < max_flush_batch_size) { - std::lock_guard writer_lock(writer_mutex); - current_writer->flush(); - } - - { - std::lock_guard lock{durable_idx_mutex}; - last_durable_idx = flush.index; + try_batch_flush = true; + continue; } + flush_logs(flush); } else { *flush.failed = true; } - - durable_idx_cv.notify_all(); - - // we need to call completion callback in another thread because it takes a global lock for the NuRaft server - // NuRaft will in some places wait for flush to be done while having the same global lock leading to deadlock - // -> future write operations are blocked by flush that cannot be completed because it cannot take NuRaft lock - // -> NuRaft won't leave lock until its flush is done - if (!append_completion_queue.push(batch_append_ok)) - LOG_WARNING(log, "Changelog is shut down"); - + notify_append_completion(); batch_append_ok = true; } } diff --git a/src/Coordination/KeeperStateManager.cpp b/src/Coordination/KeeperStateManager.cpp index cf1bad8c5fa..cc13c755629 100644 --- a/src/Coordination/KeeperStateManager.cpp +++ b/src/Coordination/KeeperStateManager.cpp @@ -184,6 +184,9 @@ KeeperStateManager::parseServersConfiguration(const Poco::Util::AbstractConfigur total_servers++; } + /// this will only apply to fresh clusters + result.cluster_config->set_async_replication(true); + if (!result.config && !allow_without_us) throw Exception(ErrorCodes::RAFT_ERROR, "Our server id {} not found in raft_configuration section", my_server_id); From 41e8345169616ea052e09dcd5f54aea8c5253c62 Mon Sep 17 00:00:00 2001 From: chen768959 <934103231@qq.com> Date: Sat, 5 Aug 2023 17:35:25 +0800 Subject: [PATCH 035/433] Add the optimize_type_ids parameter to control whether optimization is enabled. --- src/DataTypes/FieldToDataType.cpp | 2 +- src/DataTypes/getLeastSupertype.cpp | 7 ++++--- src/DataTypes/getLeastSupertype.h | 2 +- 3 files changed, 6 insertions(+), 5 deletions(-) diff --git a/src/DataTypes/FieldToDataType.cpp b/src/DataTypes/FieldToDataType.cpp index 0a48c43ceb2..9b8e476067e 100644 --- a/src/DataTypes/FieldToDataType.cpp +++ b/src/DataTypes/FieldToDataType.cpp @@ -155,7 +155,7 @@ DataTypePtr FieldToDataType::operator() (const Array & x) const for (const Field & elem : x) element_types.emplace_back(applyVisitor(*this, elem)); - return std::make_shared(getLeastSupertype(element_types)); + return std::make_shared(getLeastSupertype(element_types, true)); } template diff --git a/src/DataTypes/getLeastSupertype.cpp b/src/DataTypes/getLeastSupertype.cpp index 5a24a59a60e..f3b396ba895 100644 --- a/src/DataTypes/getLeastSupertype.cpp +++ b/src/DataTypes/getLeastSupertype.cpp @@ -201,7 +201,7 @@ DataTypePtr getNumericType(const TypeIndexSet & types) } template -DataTypePtr getLeastSupertype(const DataTypes & types) +DataTypePtr getLeastSupertype(const DataTypes & types, bool optimize_type_ids) { /// Trivial cases @@ -592,7 +592,8 @@ DataTypePtr getLeastSupertype(const DataTypes & types) /// For numeric types, the most complicated part. { - optimizeTypeIds(types, type_ids); + if (optimize_type_ids) + optimizeTypeIds(types, type_ids); auto numeric_type = getNumericType(type_ids); if (numeric_type) return numeric_type; @@ -798,7 +799,7 @@ DataTypePtr tryGetLeastSupertype(const TypeIndexSet & types) return getLeastSupertype(types); } -template DataTypePtr getLeastSupertype(const DataTypes & types); +template DataTypePtr getLeastSupertype(const DataTypes & types, bool optimize_type_ids); template DataTypePtr getLeastSupertype(const TypeIndexSet & types); } diff --git a/src/DataTypes/getLeastSupertype.h b/src/DataTypes/getLeastSupertype.h index 0a3fa3c2536..ba43b3966f9 100644 --- a/src/DataTypes/getLeastSupertype.h +++ b/src/DataTypes/getLeastSupertype.h @@ -18,7 +18,7 @@ enum class LeastSupertypeOnError * Examples: there is no least common supertype for Array(UInt8), Int8. */ template -DataTypePtr getLeastSupertype(const DataTypes & types); +DataTypePtr getLeastSupertype(const DataTypes & types, bool optimize_type_ids = false); /// Same as above but return String type instead of throwing exception. /// All types can be casted to String, because they can be serialized to String. From ad49798f4550338c314d0620e283aee800fa7664 Mon Sep 17 00:00:00 2001 From: chen768959 <934103231@qq.com> Date: Sat, 5 Aug 2023 20:17:28 +0800 Subject: [PATCH 036/433] fix optimizeTypeIds, Some scenarios do not require optimization: if UInt64(possible: Int64), UInt64(possible: Int64) = UInt64, UInt64 if Int64(possible: UInt32), Int64(possible: UInt32) = Int64, Int64 --- src/DataTypes/getLeastSupertype.cpp | 32 ++++++++++++++++++----------- 1 file changed, 20 insertions(+), 12 deletions(-) diff --git a/src/DataTypes/getLeastSupertype.cpp b/src/DataTypes/getLeastSupertype.cpp index f3b396ba895..c534f3c6edc 100644 --- a/src/DataTypes/getLeastSupertype.cpp +++ b/src/DataTypes/getLeastSupertype.cpp @@ -641,6 +641,8 @@ void optimizeTypeIds(const DataTypes & types, TypeIndexSet & type_ids) bool only_unsigned = false; bool only_signed = false; bool both = false; + bool has_unsigned = false; + bool has_signed = false; // Determine the distribution of maximum signed and unsigned, Example: // Int64, Int64 = only_signed. @@ -673,6 +675,11 @@ void optimizeTypeIds(const DataTypes & types, TypeIndexSet & type_ids) } } + if (type_is_unsigned) + has_unsigned = true; + else + has_signed = true; + if (type_is_both) both = true; else if (type_is_unsigned) @@ -718,22 +725,23 @@ void optimizeTypeIds(const DataTypes & types, TypeIndexSet & type_ids) return type_id; }; - // optimize type_ids, Example: - // if only_signed. UInt64(possible: Int64), Int64 = Int64, Int64 - // if only_unsigned. Int64(possible: UInt64), UInt64 = UInt64, UInt64 - if (!(only_unsigned && only_signed) && (both || only_unsigned || only_signed)) + // optimize type_ids + if (both) { - type_ids.clear(); - for (const auto & type : types) + // Example: UInt64(possible: Int64), Int64 = Int64, Int64 + if (only_unsigned && !only_signed) { - if (only_unsigned) - { + type_ids.clear(); + for (const auto & type : types) type_ids.insert(optimize_type_id(type, true)); - } - else if (both || only_signed) - { + } + // Example: Int64(possible: UInt32), UInt64 = UInt32, UInt64 + // Int64(possible: UInt32), UInt64(possible: Int64) = Int64, Int64 + else if ((only_signed && !only_unsigned) || (has_unsigned && has_signed && !only_signed && !only_unsigned)) + { + type_ids.clear(); + for (const auto & type : types) type_ids.insert(optimize_type_id(type, false)); - } } } } From 75a8512017027d92a2ef7311e89d84ff627629c5 Mon Sep 17 00:00:00 2001 From: chen768959 <934103231@qq.com> Date: Sun, 6 Aug 2023 09:52:53 +0800 Subject: [PATCH 037/433] modify comments --- src/DataTypes/getLeastSupertype.cpp | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/DataTypes/getLeastSupertype.cpp b/src/DataTypes/getLeastSupertype.cpp index c534f3c6edc..c04130a58e4 100644 --- a/src/DataTypes/getLeastSupertype.cpp +++ b/src/DataTypes/getLeastSupertype.cpp @@ -728,14 +728,14 @@ void optimizeTypeIds(const DataTypes & types, TypeIndexSet & type_ids) // optimize type_ids if (both) { - // Example: UInt64(possible: Int64), Int64 = Int64, Int64 + // Example: Int64(possible: UInt32), UInt64 = UInt32, UInt64 if (only_unsigned && !only_signed) { type_ids.clear(); for (const auto & type : types) type_ids.insert(optimize_type_id(type, true)); } - // Example: Int64(possible: UInt32), UInt64 = UInt32, UInt64 + // Example: UInt64(possible: Int64), Int64 = Int64, Int64 // Int64(possible: UInt32), UInt64(possible: Int64) = Int64, Int64 else if ((only_signed && !only_unsigned) || (has_unsigned && has_signed && !only_signed && !only_unsigned)) { From 1540c02ccabaf4921ca8327d5e3779bce3da03d4 Mon Sep 17 00:00:00 2001 From: chen768959 <934103231@qq.com> Date: Mon, 7 Aug 2023 17:56:51 +0800 Subject: [PATCH 038/433] temporarily disable the new Integer type inference feature to verify if it is causing the zk validation failure issue. --- src/DataTypes/FieldToDataType.cpp | 2 +- .../queries/0_stateless/02832_integer_type_inference.reference | 2 +- tests/queries/0_stateless/02832_integer_type_inference.sql | 2 +- 3 files changed, 3 insertions(+), 3 deletions(-) diff --git a/src/DataTypes/FieldToDataType.cpp b/src/DataTypes/FieldToDataType.cpp index 9b8e476067e..5947eff7ea2 100644 --- a/src/DataTypes/FieldToDataType.cpp +++ b/src/DataTypes/FieldToDataType.cpp @@ -155,7 +155,7 @@ DataTypePtr FieldToDataType::operator() (const Array & x) const for (const Field & elem : x) element_types.emplace_back(applyVisitor(*this, elem)); - return std::make_shared(getLeastSupertype(element_types, true)); + return std::make_shared(getLeastSupertype(element_types, false)); } template diff --git a/tests/queries/0_stateless/02832_integer_type_inference.reference b/tests/queries/0_stateless/02832_integer_type_inference.reference index 2d072437c90..e1bfb315d40 100644 --- a/tests/queries/0_stateless/02832_integer_type_inference.reference +++ b/tests/queries/0_stateless/02832_integer_type_inference.reference @@ -1 +1 @@ -[-4741124612489978151,-3236599669630092879,5607475129431807682] +[-4741124612489978151,-3236599669630092879] diff --git a/tests/queries/0_stateless/02832_integer_type_inference.sql b/tests/queries/0_stateless/02832_integer_type_inference.sql index 9734bc2966f..9c558b3ae67 100644 --- a/tests/queries/0_stateless/02832_integer_type_inference.sql +++ b/tests/queries/0_stateless/02832_integer_type_inference.sql @@ -1 +1 @@ -select [-4741124612489978151, -3236599669630092879, 5607475129431807682]; +select [-4741124612489978151, -3236599669630092879]; From 4b3e399f7d450ed50b01741543e2888d5f28234c Mon Sep 17 00:00:00 2001 From: Antonio Andelic Date: Mon, 7 Aug 2023 09:05:21 +0000 Subject: [PATCH 039/433] Wait for read requests --- src/Coordination/KeeperDispatcher.cpp | 92 +++++++++++++++---------- src/Coordination/KeeperDispatcher.h | 4 +- src/Coordination/KeeperStateMachine.cpp | 7 +- src/Coordination/KeeperStateMachine.h | 2 +- src/Coordination/KeeperStateManager.cpp | 2 - 5 files changed, 65 insertions(+), 42 deletions(-) diff --git a/src/Coordination/KeeperDispatcher.cpp b/src/Coordination/KeeperDispatcher.cpp index 99c28674273..4608e918042 100644 --- a/src/Coordination/KeeperDispatcher.cpp +++ b/src/Coordination/KeeperDispatcher.cpp @@ -12,6 +12,7 @@ #include #include +#include #include #include #include @@ -73,7 +74,6 @@ void KeeperDispatcher::requestThread() auto coordination_settings = configuration_and_settings->coordination_settings; uint64_t max_wait = coordination_settings->operation_timeout_ms.totalMilliseconds(); - uint64_t max_batch_size = coordination_settings->max_requests_batch_size; uint64_t max_batch_bytes_size = coordination_settings->max_requests_batch_bytes_size; /// The code below do a very simple thing: batch all write (quorum) requests into vector until @@ -136,12 +136,9 @@ void KeeperDispatcher::requestThread() return false; }; - /// TODO: Deprecate max_requests_quick_batch_size and use only max_requests_batch_size and max_requests_batch_bytes_size - size_t max_quick_batch_size = coordination_settings->max_requests_quick_batch_size; - while (!shutdown_called && !has_read_request && - !has_reconfig_request && - current_batch.size() < max_quick_batch_size && current_batch_bytes_size < max_batch_bytes_size && - try_get_request()) + size_t max_batch_size = coordination_settings->max_requests_batch_size; + while (!shutdown_called && current_batch.size() < max_batch_size && !has_reconfig_request + && current_batch_bytes_size < max_batch_bytes_size && try_get_request()) ; const auto prev_result_done = [&] @@ -152,10 +149,8 @@ void KeeperDispatcher::requestThread() }; /// Waiting until previous append will be successful, or batch is big enough - while (!shutdown_called && !has_read_request && - !has_reconfig_request && !prev_result_done() && - current_batch.size() <= max_batch_size - && current_batch_bytes_size < max_batch_bytes_size) + while (!shutdown_called && !has_reconfig_request && !prev_result_done() && current_batch.size() <= max_batch_size + && current_batch_bytes_size < max_batch_bytes_size) { try_get_request(); } @@ -166,9 +161,10 @@ void KeeperDispatcher::requestThread() if (shutdown_called) break; + nuraft::ptr result_buf = nullptr; /// Forcefully process all previous pending requests if (prev_result) - forceWaitAndProcessResult(prev_result, prev_batch); + result_buf = forceWaitAndProcessResult(prev_result, prev_batch); /// Process collected write requests batch if (!current_batch.empty()) @@ -177,13 +173,7 @@ void KeeperDispatcher::requestThread() auto result = server->putRequestBatch(current_batch); - if (result) - { - /// If we will execute read or reconfig next, we have to process result now - if (has_read_request || has_reconfig_request) - forceWaitAndProcessResult(result, current_batch); - } - else + if (!result) { addErrorResponses(current_batch, Coordination::Error::ZCONNECTIONLOSS); current_batch.clear(); @@ -194,6 +184,28 @@ void KeeperDispatcher::requestThread() prev_result = result; } + /// If we will execute read or reconfig next, we have to process result now + if (has_read_request || has_reconfig_request) + { + if (prev_result) + result_buf = forceWaitAndProcessResult(prev_result, current_batch); + + if (result_buf) + { + nuraft::buffer_serializer bs(result_buf); + auto log_idx = bs.get_u64(); + + while (true) + { + auto current_last_committed_idx = last_committed_log_idx.load(std::memory_order_relaxed); + if (current_last_committed_idx >= log_idx) + break; + + last_committed_log_idx.wait(current_last_committed_idx); + } + } + } + if (has_reconfig_request) server->getKeeperStateMachine()->reconfigure(request); @@ -360,28 +372,33 @@ void KeeperDispatcher::initialize(const Poco::Util::AbstractConfiguration & conf snapshots_queue, keeper_context, snapshot_s3, - [this](const KeeperStorage::RequestForSession & request_for_session) + [this](uint64_t log_idx, const KeeperStorage::RequestForSession & request_for_session) { - /// check if we have queue of read requests depending on this request to be committed - std::lock_guard lock(read_request_queue_mutex); - if (auto it = read_request_queue.find(request_for_session.session_id); it != read_request_queue.end()) { - auto & xid_to_request_queue = it->second; - - if (auto request_queue_it = xid_to_request_queue.find(request_for_session.request->xid); - request_queue_it != xid_to_request_queue.end()) + /// check if we have queue of read requests depending on this request to be committed + std::lock_guard lock(read_request_queue_mutex); + if (auto it = read_request_queue.find(request_for_session.session_id); it != read_request_queue.end()) { - for (const auto & read_request : request_queue_it->second) - { - if (server->isLeaderAlive()) - server->putLocalReadRequest(read_request); - else - addErrorResponses({read_request}, Coordination::Error::ZCONNECTIONLOSS); - } + auto & xid_to_request_queue = it->second; - xid_to_request_queue.erase(request_queue_it); + if (auto request_queue_it = xid_to_request_queue.find(request_for_session.request->xid); + request_queue_it != xid_to_request_queue.end()) + { + for (const auto & read_request : request_queue_it->second) + { + if (server->isLeaderAlive()) + server->putLocalReadRequest(read_request); + else + addErrorResponses({read_request}, Coordination::Error::ZCONNECTIONLOSS); + } + + xid_to_request_queue.erase(request_queue_it); + } } } + + last_committed_log_idx.store(log_idx, std::memory_order_relaxed); + last_committed_log_idx.notify_all(); }); try @@ -636,7 +653,7 @@ void KeeperDispatcher::addErrorResponses(const KeeperStorage::RequestsForSession } } -void KeeperDispatcher::forceWaitAndProcessResult(RaftAppendResult & result, KeeperStorage::RequestsForSessions & requests_for_sessions) +nuraft::ptr KeeperDispatcher::forceWaitAndProcessResult(RaftAppendResult & result, KeeperStorage::RequestsForSessions & requests_for_sessions) { if (!result->has_result()) result->get(); @@ -647,8 +664,11 @@ void KeeperDispatcher::forceWaitAndProcessResult(RaftAppendResult & result, Keep else if (result->get_result_code() != nuraft::cmd_result_code::OK) addErrorResponses(requests_for_sessions, Coordination::Error::ZCONNECTIONLOSS); + auto result_buf = result->get(); + result = nullptr; requests_for_sessions.clear(); + return result_buf; } int64_t KeeperDispatcher::getSessionID(int64_t session_timeout_ms) diff --git a/src/Coordination/KeeperDispatcher.h b/src/Coordination/KeeperDispatcher.h index ae592ae3fa1..31a8f80f252 100644 --- a/src/Coordination/KeeperDispatcher.h +++ b/src/Coordination/KeeperDispatcher.h @@ -102,11 +102,13 @@ private: /// Forcefully wait for result and sets errors if something when wrong. /// Clears both arguments - void forceWaitAndProcessResult(RaftAppendResult & result, KeeperStorage::RequestsForSessions & requests_for_sessions); + nuraft::ptr forceWaitAndProcessResult(RaftAppendResult & result, KeeperStorage::RequestsForSessions & requests_for_sessions); public: std::mutex read_request_queue_mutex; + std::atomic last_committed_log_idx = 0; + /// queue of read requests that can be processed after a request with specific session ID and XID is committed std::unordered_map> read_request_queue; diff --git a/src/Coordination/KeeperStateMachine.cpp b/src/Coordination/KeeperStateMachine.cpp index 8f2e3c3ac0e..5a8cf88ea73 100644 --- a/src/Coordination/KeeperStateMachine.cpp +++ b/src/Coordination/KeeperStateMachine.cpp @@ -167,7 +167,10 @@ nuraft::ptr KeeperStateMachine::pre_commit(uint64_t log_idx, nur request_for_session->zxid = log_idx; preprocess(*request_for_session); - return nullptr; + auto result = nuraft::buffer::alloc(8); + nuraft::buffer_serializer ss(result); + ss.put_u64(log_idx); + return result; } std::shared_ptr KeeperStateMachine::parseRequest(nuraft::buffer & data, bool final, ZooKeeperLogSerializationVersion * serialization_version) @@ -433,7 +436,7 @@ nuraft::ptr KeeperStateMachine::commit(const uint64_t log_idx, n last_committed_idx = log_idx; if (commit_callback) - commit_callback(*request_for_session); + commit_callback(log_idx, *request_for_session); return nullptr; } diff --git a/src/Coordination/KeeperStateMachine.h b/src/Coordination/KeeperStateMachine.h index 116fa9257a0..aad5d3aafd4 100644 --- a/src/Coordination/KeeperStateMachine.h +++ b/src/Coordination/KeeperStateMachine.h @@ -20,7 +20,7 @@ using SnapshotsQueue = ConcurrentBoundedQueue; class KeeperStateMachine : public nuraft::state_machine { public: - using CommitCallback = std::function; + using CommitCallback = std::function; KeeperStateMachine( ResponsesQueue & responses_queue_, diff --git a/src/Coordination/KeeperStateManager.cpp b/src/Coordination/KeeperStateManager.cpp index cc13c755629..902c2ef595f 100644 --- a/src/Coordination/KeeperStateManager.cpp +++ b/src/Coordination/KeeperStateManager.cpp @@ -184,9 +184,7 @@ KeeperStateManager::parseServersConfiguration(const Poco::Util::AbstractConfigur total_servers++; } - /// this will only apply to fresh clusters result.cluster_config->set_async_replication(true); - if (!result.config && !allow_without_us) throw Exception(ErrorCodes::RAFT_ERROR, "Our server id {} not found in raft_configuration section", my_server_id); From 080b4badbd59aafa24d00349cb4338a8b2e9d0c2 Mon Sep 17 00:00:00 2001 From: alexX512 Date: Mon, 7 Aug 2023 15:58:14 +0000 Subject: [PATCH 040/433] Support partial result for aggregating transform during query execution --- src/Interpreters/Aggregator.cpp | 23 +++++ src/Interpreters/Aggregator.h | 2 + src/Processors/IProcessor.cpp | 5 ++ src/Processors/IProcessor.h | 12 +-- src/Processors/LimitTransform.h | 3 +- .../AggregatingPartialResultTransform.cpp | 42 +++++++++ .../AggregatingPartialResultTransform.h | 26 ++++++ .../Transforms/AggregatingTransform.cpp | 12 +++ .../Transforms/AggregatingTransform.h | 7 ++ .../Transforms/ExpressionTransform.h | 4 +- .../Transforms/LimitsCheckingTransform.h | 3 +- .../MergeSortingPartialResultTransform.cpp | 1 - .../MergeSortingPartialResultTransform.h | 1 + .../Transforms/MergeSortingTransform.cpp | 11 --- .../Transforms/MergeSortingTransform.h | 7 +- .../Transforms/PartialResultTransform.cpp | 6 +- .../Transforms/PartialResultTransform.h | 1 + src/QueryPipeline/Pipe.cpp | 4 +- ...rting_result_during_query_execution.python | 25 ++++-- ...ating_result_during_query_execution.python | 88 +++++++++++++++++++ ...ng_result_during_query_execution.reference | 53 +++++++++++ ...gregating_result_during_query_execution.sh | 8 ++ .../queries/0_stateless/helpers/tcp_client.py | 26 +++++- 23 files changed, 332 insertions(+), 38 deletions(-) create mode 100644 src/Processors/Transforms/AggregatingPartialResultTransform.cpp create mode 100644 src/Processors/Transforms/AggregatingPartialResultTransform.h create mode 100644 tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python create mode 100644 tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.reference create mode 100755 tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.sh diff --git a/src/Interpreters/Aggregator.cpp b/src/Interpreters/Aggregator.cpp index 91cd574708a..e310c2a8315 100644 --- a/src/Interpreters/Aggregator.cpp +++ b/src/Interpreters/Aggregator.cpp @@ -2273,6 +2273,29 @@ Block Aggregator::prepareBlockAndFillWithoutKey(AggregatedDataVariants & data_va return block; } +Block Aggregator::prepareBlockAndFillWithoutKeySnapshot(AggregatedDataVariants & data_variants) const +{ + size_t rows = 1; + bool final = true; + + auto && out_cols + = prepareOutputBlockColumns(params, aggregate_functions, getHeader(final), data_variants.aggregates_pools, final, rows); + auto && [key_columns, raw_key_columns, aggregate_columns, final_aggregate_columns, aggregate_columns_data] = out_cols; + + AggregatedDataWithoutKey & data = data_variants.without_key; + + /// Always single-thread. It's safe to pass current arena from 'aggregates_pool'. + for (size_t insert_i = 0; insert_i < params.aggregates_size; ++insert_i) + aggregate_functions[insert_i]->insertResultInto( + data + offsets_of_aggregate_states[insert_i], + *final_aggregate_columns[insert_i], + data_variants.aggregates_pool); + + Block block = finalizeBlock(params, getHeader(final), std::move(out_cols), final, rows); + + return block; +} + template Aggregator::ConvertToBlockRes Aggregator::prepareBlockAndFillSingleLevel(AggregatedDataVariants & data_variants, bool final) const diff --git a/src/Interpreters/Aggregator.h b/src/Interpreters/Aggregator.h index 4f2c86606c5..6bfaa76f9b3 100644 --- a/src/Interpreters/Aggregator.h +++ b/src/Interpreters/Aggregator.h @@ -1210,6 +1210,7 @@ private: friend class ConvertingAggregatedToChunksSource; friend class ConvertingAggregatedToChunksWithMergingSource; friend class AggregatingInOrderTransform; + friend class AggregatingPartialResultTransform; /// Data structure of source blocks. Block header; @@ -1391,6 +1392,7 @@ private: std::atomic * is_cancelled = nullptr) const; Block prepareBlockAndFillWithoutKey(AggregatedDataVariants & data_variants, bool final, bool is_overflows) const; + Block prepareBlockAndFillWithoutKeySnapshot(AggregatedDataVariants & data_variants) const; BlocksList prepareBlocksAndFillTwoLevel(AggregatedDataVariants & data_variants, bool final, ThreadPool * thread_pool) const; template diff --git a/src/Processors/IProcessor.cpp b/src/Processors/IProcessor.cpp index 8b160153733..2f294a32531 100644 --- a/src/Processors/IProcessor.cpp +++ b/src/Processors/IProcessor.cpp @@ -40,5 +40,10 @@ std::string IProcessor::statusToName(Status status) UNREACHABLE(); } +ProcessorPtr IProcessor::getPartialResultProcessorPtr(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) +{ + return current_processor->getPartialResultProcessor(current_processor, partial_result_limit, partial_result_duration_ms); +} + } diff --git a/src/Processors/IProcessor.h b/src/Processors/IProcessor.h index 140e5c13f27..a156da7bad2 100644 --- a/src/Processors/IProcessor.h +++ b/src/Processors/IProcessor.h @@ -164,6 +164,8 @@ public: static std::string statusToName(Status status); + static ProcessorPtr getPartialResultProcessorPtr(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms); + /** Method 'prepare' is responsible for all cheap ("instantaneous": O(1) of data volume, no wait) calculations. * * It may access input and output ports, @@ -238,11 +240,6 @@ public: virtual bool isPartialResultProcessor() const { return false; } virtual bool supportPartialResultProcessor() const { return false; } - virtual ProcessorPtr getPartialResultProcessor(const ProcessorPtr & /*current_processor*/, UInt64 /*partial_result_limit*/, UInt64 /*partial_result_duration_ms*/) - { - throw Exception(ErrorCodes::NOT_IMPLEMENTED, "Method 'getPartialResultProcessor' is not implemented for {} processor", getName()); - } - /// In case if query was cancelled executor will wait till all processors finish their jobs. /// Generally, there is no reason to check this flag. However, it may be reasonable for long operations (e.g. i/o). bool isCancelled() const { return is_cancelled.load(std::memory_order_acquire); } @@ -377,6 +374,11 @@ public: protected: virtual void onCancel() {} + virtual ProcessorPtr getPartialResultProcessor(const ProcessorPtr & /*current_processor*/, UInt64 /*partial_result_limit*/, UInt64 /*partial_result_duration_ms*/) + { + throw Exception(ErrorCodes::NOT_IMPLEMENTED, "Method 'getPartialResultProcessor' is not implemented for {} processor", getName()); + } + private: /// For: /// - elapsed_us diff --git a/src/Processors/LimitTransform.h b/src/Processors/LimitTransform.h index b0ec7600406..ebdcbe49670 100644 --- a/src/Processors/LimitTransform.h +++ b/src/Processors/LimitTransform.h @@ -55,7 +55,6 @@ private: ColumnRawPtrs extractSortColumns(const Columns & columns) const; bool sortColumnsEqualAt(const ColumnRawPtrs & current_chunk_sort_columns, UInt64 current_chunk_row_num) const; - bool supportPartialResultProcessor() const override { return true; } ProcessorPtr getPartialResultProcessor(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) override; public: @@ -76,6 +75,8 @@ public: void setRowsBeforeLimitCounter(RowsBeforeLimitCounterPtr counter) override { rows_before_limit_at_least.swap(counter); } void setInputPortHasCounter(size_t pos) { ports_data[pos].input_port_has_counter = true; } + + bool supportPartialResultProcessor() const override { return true; } }; } diff --git a/src/Processors/Transforms/AggregatingPartialResultTransform.cpp b/src/Processors/Transforms/AggregatingPartialResultTransform.cpp new file mode 100644 index 00000000000..fbfb9be572b --- /dev/null +++ b/src/Processors/Transforms/AggregatingPartialResultTransform.cpp @@ -0,0 +1,42 @@ +#include + +namespace DB +{ + +AggregatingPartialResultTransform::AggregatingPartialResultTransform( + const Block & input_header, const Block & output_header, AggregatingTransformPtr aggregating_transform_, + UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_) + : PartialResultTransform(input_header, output_header, partial_result_limit_, partial_result_duration_ms_) + , aggregating_transform(std::move(aggregating_transform_)) + {} + +PartialResultTransform::ShaphotResult AggregatingPartialResultTransform::getRealProcessorSnapshot() +{ + std::lock_guard lock(aggregating_transform->snapshot_mutex); + + auto & params = aggregating_transform->params; + /// Currently not supported cases + /// TODO: check that insert results from prepareBlockAndFillWithoutKey return values without changing of the aggregator state + if (params->params.keys_size != 0 /// has at least one key for aggregation + || params->aggregator.hasTemporaryData() /// use external storage for aggregation + || aggregating_transform->many_data->variants.size() > 1) /// use more then one stream for aggregation + return {{}, SnaphotStatus::Stopped}; + + if (aggregating_transform->is_generate_initialized) + return {{}, SnaphotStatus::Stopped}; + + if (aggregating_transform->variants.empty()) + return {{}, SnaphotStatus::NotReady}; + + auto & aggregator = params->aggregator; + + auto prepared_data = aggregator.prepareVariantsToMerge(aggregating_transform->many_data->variants); + AggregatedDataVariantsPtr & first = prepared_data.at(0); + + aggregator.mergeWithoutKeyDataImpl(prepared_data); + auto block = aggregator.prepareBlockAndFillWithoutKeySnapshot(*first); + + return {convertToChunk(block), SnaphotStatus::Ready}; +} + +} diff --git a/src/Processors/Transforms/AggregatingPartialResultTransform.h b/src/Processors/Transforms/AggregatingPartialResultTransform.h new file mode 100644 index 00000000000..e5e28dc7f7a --- /dev/null +++ b/src/Processors/Transforms/AggregatingPartialResultTransform.h @@ -0,0 +1,26 @@ +#pragma once + +#include +#include + +namespace DB +{ + +class AggregatingPartialResultTransform : public PartialResultTransform +{ +public: + using AggregatingTransformPtr = std::shared_ptr; + + AggregatingPartialResultTransform( + const Block & input_header, const Block & output_header, AggregatingTransformPtr aggregating_transform_, + UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_); + + String getName() const override { return "AggregatingPartialResultTransform"; } + + ShaphotResult getRealProcessorSnapshot() override; + +private: + AggregatingTransformPtr aggregating_transform; +}; + +} diff --git a/src/Processors/Transforms/AggregatingTransform.cpp b/src/Processors/Transforms/AggregatingTransform.cpp index 4bd000797a6..b4d2785bed2 100644 --- a/src/Processors/Transforms/AggregatingTransform.cpp +++ b/src/Processors/Transforms/AggregatingTransform.cpp @@ -1,3 +1,4 @@ +#include #include #include @@ -657,6 +658,8 @@ void AggregatingTransform::consume(Chunk chunk) src_rows += num_rows; src_bytes += chunk.bytes(); + std::lock_guard lock(snapshot_mutex); + if (params->params.only_merge) { auto block = getInputs().front().getHeader().cloneWithColumns(chunk.detachColumns()); @@ -676,6 +679,7 @@ void AggregatingTransform::initGenerate() if (is_generate_initialized) return; + std::lock_guard lock(snapshot_mutex); is_generate_initialized = true; /// If there was no data, and we aggregate without keys, and we must return single row with the result of empty aggregation. @@ -806,4 +810,12 @@ void AggregatingTransform::initGenerate() } } +ProcessorPtr AggregatingTransform::getPartialResultProcessor(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) +{ + const auto & input_header = inputs.front().getHeader(); + const auto & output_header = outputs.front().getHeader(); + auto aggregating_processor = std::dynamic_pointer_cast(current_processor); + return std::make_shared(input_header, output_header, std::move(aggregating_processor), partial_result_limit, partial_result_duration_ms); +} + } diff --git a/src/Processors/Transforms/AggregatingTransform.h b/src/Processors/Transforms/AggregatingTransform.h index 38baa4d0394..649b90ae74a 100644 --- a/src/Processors/Transforms/AggregatingTransform.h +++ b/src/Processors/Transforms/AggregatingTransform.h @@ -170,9 +170,13 @@ public: void work() override; Processors expandPipeline() override; + bool supportPartialResultProcessor() const override { return true; } + protected: void consume(Chunk chunk); + ProcessorPtr getPartialResultProcessor(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) override; + private: /// To read the data that was flushed into the temporary data file. Processors processors; @@ -212,6 +216,9 @@ private: bool is_consume_started = false; + friend class AggregatingPartialResultTransform; + std::mutex snapshot_mutex; + void initGenerate(); }; diff --git a/src/Processors/Transforms/ExpressionTransform.h b/src/Processors/Transforms/ExpressionTransform.h index bf8425b2300..a7b5ef4f51b 100644 --- a/src/Processors/Transforms/ExpressionTransform.h +++ b/src/Processors/Transforms/ExpressionTransform.h @@ -26,13 +26,15 @@ public: static Block transformHeader(Block header, const ActionsDAG & expression); + bool supportPartialResultProcessor() const override { return true; } + protected: void transform(Chunk & chunk) override; - bool supportPartialResultProcessor() const override { return true; } ProcessorPtr getPartialResultProcessor(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) override; private: + ExpressionActionsPtr expression; }; diff --git a/src/Processors/Transforms/LimitsCheckingTransform.h b/src/Processors/Transforms/LimitsCheckingTransform.h index 6e8d5547cc5..e8635384129 100644 --- a/src/Processors/Transforms/LimitsCheckingTransform.h +++ b/src/Processors/Transforms/LimitsCheckingTransform.h @@ -33,10 +33,11 @@ public: void setQuota(const std::shared_ptr & quota_) { quota = quota_; } + bool supportPartialResultProcessor() const override { return true; } + protected: void transform(Chunk & chunk) override; - bool supportPartialResultProcessor() const override { return true; } ProcessorPtr getPartialResultProcessor(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) override; private: diff --git a/src/Processors/Transforms/MergeSortingPartialResultTransform.cpp b/src/Processors/Transforms/MergeSortingPartialResultTransform.cpp index 2c4d6bea3c5..e4a2af2cdd8 100644 --- a/src/Processors/Transforms/MergeSortingPartialResultTransform.cpp +++ b/src/Processors/Transforms/MergeSortingPartialResultTransform.cpp @@ -1,4 +1,3 @@ -#include #include namespace DB diff --git a/src/Processors/Transforms/MergeSortingPartialResultTransform.h b/src/Processors/Transforms/MergeSortingPartialResultTransform.h index f08f74dad6b..1eaebf9d902 100644 --- a/src/Processors/Transforms/MergeSortingPartialResultTransform.h +++ b/src/Processors/Transforms/MergeSortingPartialResultTransform.h @@ -1,5 +1,6 @@ #pragma once +#include #include namespace DB diff --git a/src/Processors/Transforms/MergeSortingTransform.cpp b/src/Processors/Transforms/MergeSortingTransform.cpp index d21cb179b0a..af4a51b81ee 100644 --- a/src/Processors/Transforms/MergeSortingTransform.cpp +++ b/src/Processors/Transforms/MergeSortingTransform.cpp @@ -285,17 +285,6 @@ void MergeSortingTransform::remerge() ProcessorPtr MergeSortingTransform::getPartialResultProcessor(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) { - if (getName() != current_processor->getName() || current_processor.get() != this) - throw Exception( - ErrorCodes::LOGICAL_ERROR, - "To create partial result processor variable current_processor should use " \ - "the same class and pointer as in the original processor with class {} and pointer {}. " \ - "But current_processor has another class {} or pointer {} then original.", - getName(), - static_cast(this), - current_processor->getName(), - static_cast(current_processor.get())); - const auto & header = inputs.front().getHeader(); auto merge_sorting_processor = std::dynamic_pointer_cast(current_processor); return std::make_shared(header, std::move(merge_sorting_processor), partial_result_limit, partial_result_duration_ms); diff --git a/src/Processors/Transforms/MergeSortingTransform.h b/src/Processors/Transforms/MergeSortingTransform.h index bf4bfec72fb..e8a1f7a5853 100644 --- a/src/Processors/Transforms/MergeSortingTransform.h +++ b/src/Processors/Transforms/MergeSortingTransform.h @@ -33,6 +33,8 @@ public: String getName() const override { return "MergeSortingTransform"; } + bool supportPartialResultProcessor() const override { return true; } + protected: void consume(Chunk chunk) override; void serialize() override; @@ -40,7 +42,6 @@ protected: Processors expandPipeline() override; - bool supportPartialResultProcessor() const override { return true; } ProcessorPtr getPartialResultProcessor(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) override; private: @@ -61,10 +62,10 @@ private: /// Merge all accumulated blocks to keep no more than limit rows. void remerge(); + ProcessorPtr external_merging_sorted; + friend class MergeSortingPartialResultTransform; std::mutex snapshot_mutex; - - ProcessorPtr external_merging_sorted; }; } diff --git a/src/Processors/Transforms/PartialResultTransform.cpp b/src/Processors/Transforms/PartialResultTransform.cpp index 5e9704d38a2..97ff79dee54 100644 --- a/src/Processors/Transforms/PartialResultTransform.cpp +++ b/src/Processors/Transforms/PartialResultTransform.cpp @@ -3,8 +3,12 @@ namespace DB { + PartialResultTransform::PartialResultTransform(const Block & header, UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_) - : IProcessor({header}, {header}) + : PartialResultTransform(header, header, partial_result_limit_, partial_result_duration_ms_) {} + +PartialResultTransform::PartialResultTransform(const Block & input_header, const Block & output_header, UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_) + : IProcessor({input_header}, {output_header}) , input(inputs.front()) , output(outputs.front()) , partial_result_limit(partial_result_limit_) diff --git a/src/Processors/Transforms/PartialResultTransform.h b/src/Processors/Transforms/PartialResultTransform.h index 7d4b9026ce1..2bcf494400d 100644 --- a/src/Processors/Transforms/PartialResultTransform.h +++ b/src/Processors/Transforms/PartialResultTransform.h @@ -9,6 +9,7 @@ class PartialResultTransform : public IProcessor { public: PartialResultTransform(const Block & header, UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_); + PartialResultTransform(const Block & input_header, const Block & output_header, UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_); String getName() const override { return "PartialResultTransform"; } diff --git a/src/QueryPipeline/Pipe.cpp b/src/QueryPipeline/Pipe.cpp index adf06dde53b..15f6bea70a7 100644 --- a/src/QueryPipeline/Pipe.cpp +++ b/src/QueryPipeline/Pipe.cpp @@ -636,7 +636,7 @@ void Pipe::addPartialResultSimpleTransform(const ProcessorPtr & transform, size_ return; } - auto partial_result_transform = transform->getPartialResultProcessor(transform, partial_result_limit, partial_result_duration_ms); + auto partial_result_transform = IProcessor::getPartialResultProcessorPtr(transform, partial_result_limit, partial_result_duration_ms); connectPartialResultPort(partial_result_port, partial_result_transform->getInputs().front()); @@ -661,7 +661,7 @@ void Pipe::addPartialResultTransform(const ProcessorPtr & transform) return; } - auto partial_result_transform = transform->getPartialResultProcessor(transform, partial_result_limit, partial_result_duration_ms); + auto partial_result_transform = IProcessor::getPartialResultProcessorPtr(transform, partial_result_limit, partial_result_duration_ms); auto & inputs = partial_result_transform->getInputs(); if (inputs.size() != partial_result_ports.size()) diff --git a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python index 14f5a912672..ce80b0f29a3 100755 --- a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python +++ b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python @@ -8,13 +8,13 @@ import sys CURDIR = os.path.dirname(os.path.realpath(__file__)) sys.path.insert(0, os.path.join(CURDIR, "helpers")) -from tcp_client import TCPClient, assertPacket +from tcp_client import TCPClient def main(): with TCPClient() as client: client.sendQuery( - f"SELECT number FROM numbers_mt(1e7+1) ORDER BY -number LIMIT 15 SETTINGS max_threads = 1, partial_result_update_duration_ms=1, max_rows_in_partial_result=10" + "SELECT number FROM numbers_mt(1e7+1) ORDER BY -number LIMIT 15 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 10" ) # external tables @@ -23,13 +23,13 @@ def main(): # Partial result _, partial_result = client.readDataWithoutProgress()[0] - - assert_message = ( - "There should be at least one block of data with partial result" - ) - assert len(partial_result) > 0, assert_message + assert len(partial_result) > 0, "Expected at least one block with a non-empty partial result before getting the full result" while True: + assert all( + a >= b for a, b in zip(partial_result, partial_result[1:]) + ), "Partial result always should be sorted for this test" + _, new_partial_result = client.readDataWithoutProgress( need_print_info=False )[0] @@ -37,15 +37,22 @@ def main(): break data_size = len(partial_result) - assert_message = f"New block contains more info about the full data so sorted results should not be less then in the previous iteration. New result {new_partial_result}. Previous result {partial_result}" + assert all( partial_result[i] <= new_partial_result[i] for i in range(data_size) - ), assert_message + ), f"New partial result values should always be greater then old one because a new block contains more information about the full data. New result {new_partial_result}. Previous result {partial_result}" + partial_result = new_partial_result # Full result _, full_result = client.readDataWithoutProgress()[0] + + data_size = len(partial_result) + assert all( + partial_result[i] <= full_result[i] for i in range(data_size) + ), f"Full result values should always be greater then partial result values. Full result {full_result}. Partial result {partial_result}" + for result in full_result: print(result) diff --git a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python new file mode 100644 index 00000000000..6fd778c77e7 --- /dev/null +++ b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python @@ -0,0 +1,88 @@ +#!/usr/bin/env python3 +# -*- coding: utf-8 -*- + + +import os +import sys + +CURDIR = os.path.dirname(os.path.realpath(__file__)) +sys.path.insert(0, os.path.join(CURDIR, "helpers")) + +from tcp_client import TCPClient, assertPacket + +def run_query_without_errors(query, support_partial_result, invariants=None): + if invariants is None: + invariants = {} + + with TCPClient() as client: + client.sendQuery(query) + + # external tables + client.sendEmptyBlock() + client.readHeader() + + # Partial result + partial_results = client.readDataWithoutProgress() + if support_partial_result: + assert len(partial_results[0][1]) > 0, "Expected at least one block with a non-empty partial result before getting the full result" + while True: + new_partial_results = client.readDataWithoutProgress(need_print_info=False) + if len(new_partial_results[0][1]) == 0: + break + + for new_result, old_result in zip(new_partial_results, partial_results): + assert new_result[0] == old_result[0], "Keys in blocks should be in the same order" + key = new_result[0] + if key in invariants: + old_value = old_result[1] + new_value = new_result[1] + assert invariants[key](old_value, new_value), f"Problem with the invariant between old and new versions of a partial result for key: {key}. Old value {old_value}, new value {new_value}" + else: + assert len(partial_results[0][1]) == 0, "Expected no non-empty partial result blocks before getting the full result" + + # Full result + full_results = client.readDataWithoutProgress() + if support_partial_result: + for full_result, partial_result in zip(full_results, partial_results): + assert full_result[0] == partial_result[0], "Keys in blocks should be in the same order" + key = full_result[0] + if key in invariants: + full_value = full_result[1] + partial_value = partial_result[1] + assert invariants[key](partial_value, full_value), f"Problem with the invariant between full and partial result for key: {key}. Partial value {partial_value}. Full value {full_value}" + + for key, value in full_results: + if isinstance(value[0], int): + print(key, value) + + +def supported_scenarios(): + query = "select median(number), stddevSamp(number), stddevPop(number), max(number), min(number), any(number), count(number), avg(number), sum(number) from numbers_mt(1e7+1) settings max_threads = 1, partial_result_update_duration_ms = 1" + invariants = { + "median(number)": lambda old_value, new_value: old_value <= new_value, + "max(number)": lambda old_value, new_value: old_value <= new_value, + "min(number)": lambda old_value, new_value: old_value >= new_value, + "count(number)": lambda old_value, new_value: old_value <= new_value, + "avg(number)": lambda old_value, new_value: old_value <= new_value, + "sum(number)": lambda old_value, new_value: old_value <= new_value, + } + run_query_without_errors(query, support_partial_result=True, invariants=invariants) + + +def unsupported_scenarios(): + # Currently aggregator for partial result supports only single thread aggregation without key + # Update test when multithreading or aggregation with GROUP BY will be supported for partial result updates + multithread_query = "select sum(number) from numbers_mt(1e7+1) settings max_threads = 2, partial_result_update_duration_ms = 1" + run_query_without_errors(multithread_query, support_partial_result=False) + + group_with_key_query = "select mod2, sum(number) from numbers_mt(1e7+1) group by number % 2 as mod2 settings max_threads = 1, partial_result_update_duration_ms = 1" + run_query_without_errors(group_with_key_query, support_partial_result=False) + + +def main(): + supported_scenarios() + unsupported_scenarios() + + +if __name__ == "__main__": + main() diff --git a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.reference b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.reference new file mode 100644 index 00000000000..a97ee3f8d79 --- /dev/null +++ b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.reference @@ -0,0 +1,53 @@ +Rows 0 Columns 9 +Column median(number) type Float64 +Column stddevSamp(number) type Float64 +Column stddevPop(number) type Float64 +Column max(number) type UInt64 +Column min(number) type UInt64 +Column any(number) type UInt64 +Column count(number) type UInt64 +Column avg(number) type Float64 +Column sum(number) type UInt64 +Rows 1 Columns 9 +Column median(number) type Float64 +Column stddevSamp(number) type Float64 +Column stddevPop(number) type Float64 +Column max(number) type UInt64 +Column min(number) type UInt64 +Column any(number) type UInt64 +Column count(number) type UInt64 +Column avg(number) type Float64 +Column sum(number) type UInt64 +Rows 1 Columns 9 +Column median(number) type Float64 +Column stddevSamp(number) type Float64 +Column stddevPop(number) type Float64 +Column max(number) type UInt64 +Column min(number) type UInt64 +Column any(number) type UInt64 +Column count(number) type UInt64 +Column avg(number) type Float64 +Column sum(number) type UInt64 +max(number) [10000000] +min(number) [0] +any(number) [0] +count(number) [10000001] +sum(number) [50000005000000] +Rows 0 Columns 1 +Column sum(number) type UInt64 +Rows 0 Columns 1 +Column sum(number) type UInt64 +Rows 1 Columns 1 +Column sum(number) type UInt64 +sum(number) [50000005000000] +Rows 0 Columns 2 +Column mod2 type UInt8 +Column sum(number) type UInt64 +Rows 0 Columns 2 +Column mod2 type UInt8 +Column sum(number) type UInt64 +Rows 2 Columns 2 +Column mod2 type UInt8 +Column sum(number) type UInt64 +mod2 [0, 1] +sum(number) [25000005000000, 25000000000000] diff --git a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.sh b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.sh new file mode 100755 index 00000000000..e70a3c53ec4 --- /dev/null +++ b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.sh @@ -0,0 +1,8 @@ +#!/usr/bin/env bash + +CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) +# shellcheck source=../shell_config.sh +. "$CURDIR"/../shell_config.sh + +# We should have correct env vars from shell_config.sh to run this test +python3 "$CURDIR"/02834_partial_aggregating_result_during_query_execution.python diff --git a/tests/queries/0_stateless/helpers/tcp_client.py b/tests/queries/0_stateless/helpers/tcp_client.py index 2f0680b4eae..bab449d71fe 100644 --- a/tests/queries/0_stateless/helpers/tcp_client.py +++ b/tests/queries/0_stateless/helpers/tcp_client.py @@ -1,6 +1,7 @@ import socket import os import uuid +import struct CLICKHOUSE_HOST = os.environ.get("CLICKHOUSE_HOST", "127.0.0.1") CLICKHOUSE_PORT = int(os.environ.get("CLICKHOUSE_PORT_TCP", "900000")) @@ -105,6 +106,15 @@ class TCPClient(object): def readUInt64(self): return self.readUInt(8) + def readFloat16(self): + return struct.unpack("e", self.readStrict(2)) + + def readFloat32(self): + return struct.unpack("f", self.readStrict(4)) + + def readFloat64(self): + return struct.unpack("d", self.readStrict(8)) + def readVarUInt(self): x = 0 for i in range(9): @@ -250,12 +260,22 @@ class TCPClient(object): print("Column {} type {}".format(col_name, type_name)) def readRow(self, row_type, rows): - if row_type == "UInt64": - row = [self.readUInt64() for _ in range(rows)] + supported_row_types = { + "UInt8": self.readUInt8, + "UInt16": self.readUInt16, + "UInt32": self.readUInt32, + "UInt64": self.readUInt64, + "Float16": self.readFloat16, + "Float32": self.readFloat32, + "Float64": self.readFloat64, + } + if row_type in supported_row_types: + read_type = supported_row_types[row_type] + row = [read_type() for _ in range(rows)] return row else: raise RuntimeError( - "Currently python version of tcp client doesn't support the following type of row: {}".format( + "Current python version of tcp client doesn't support the following type of row: {}".format( row_type ) ) From 51372d841dfa0f2060201160b6d59b890ccec4d7 Mon Sep 17 00:00:00 2001 From: alexX512 Date: Mon, 7 Aug 2023 16:52:38 +0000 Subject: [PATCH 041/433] Style fix --- .../Transforms/MergeSortingTransform.cpp | 5 --- ...rting_result_during_query_execution.python | 10 ++--- ...ating_result_during_query_execution.python | 41 ++++++++++++++----- 3 files changed, 35 insertions(+), 21 deletions(-) diff --git a/src/Processors/Transforms/MergeSortingTransform.cpp b/src/Processors/Transforms/MergeSortingTransform.cpp index af4a51b81ee..e801e5e16d5 100644 --- a/src/Processors/Transforms/MergeSortingTransform.cpp +++ b/src/Processors/Transforms/MergeSortingTransform.cpp @@ -28,11 +28,6 @@ namespace ProfileEvents namespace DB { -namespace ErrorCodes -{ - extern const int LOGICAL_ERROR; -} - class BufferingToFileTransform : public IAccumulatingTransform { public: diff --git a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python index ce80b0f29a3..38a9d82d8b6 100755 --- a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python +++ b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python @@ -23,7 +23,9 @@ def main(): # Partial result _, partial_result = client.readDataWithoutProgress()[0] - assert len(partial_result) > 0, "Expected at least one block with a non-empty partial result before getting the full result" + assert ( + len(partial_result) > 0 + ), "Expected at least one block with a non-empty partial result before getting the full result" while True: assert all( @@ -37,12 +39,10 @@ def main(): break data_size = len(partial_result) - assert all( partial_result[i] <= new_partial_result[i] for i in range(data_size) ), f"New partial result values should always be greater then old one because a new block contains more information about the full data. New result {new_partial_result}. Previous result {partial_result}" - partial_result = new_partial_result # Full result @@ -50,8 +50,8 @@ def main(): data_size = len(partial_result) assert all( - partial_result[i] <= full_result[i] for i in range(data_size) - ), f"Full result values should always be greater then partial result values. Full result {full_result}. Partial result {partial_result}" + partial_result[i] <= full_result[i] for i in range(data_size) + ), f"Full result values should always be greater then partial result values. Full result {full_result}. Partial result {partial_result}" for result in full_result: print(result) diff --git a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python index 6fd778c77e7..e4fe9dcfad0 100644 --- a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python +++ b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python @@ -8,7 +8,12 @@ import sys CURDIR = os.path.dirname(os.path.realpath(__file__)) sys.path.insert(0, os.path.join(CURDIR, "helpers")) -from tcp_client import TCPClient, assertPacket +from tcp_client import TCPClient + + +def get_keys(result): + return [key for key, _ in rasults] + def run_query_without_errors(query, support_partial_result, invariants=None): if invariants is None: @@ -24,32 +29,46 @@ def run_query_without_errors(query, support_partial_result, invariants=None): # Partial result partial_results = client.readDataWithoutProgress() if support_partial_result: - assert len(partial_results[0][1]) > 0, "Expected at least one block with a non-empty partial result before getting the full result" + assert ( + len(partial_results[0][1]) > 0 + ), "Expected at least one block with a non-empty partial result before getting the full result" while True: - new_partial_results = client.readDataWithoutProgress(need_print_info=False) + new_partial_results = client.readDataWithoutProgress( + need_print_info=False + ) if len(new_partial_results[0][1]) == 0: break for new_result, old_result in zip(new_partial_results, partial_results): - assert new_result[0] == old_result[0], "Keys in blocks should be in the same order" + assert ( + new_result[0] == old_result[0] + ), "Keys in blocks should be in the same order" + key = new_result[0] if key in invariants: old_value = old_result[1] new_value = new_result[1] assert invariants[key](old_value, new_value), f"Problem with the invariant between old and new versions of a partial result for key: {key}. Old value {old_value}, new value {new_value}" else: - assert len(partial_results[0][1]) == 0, "Expected no non-empty partial result blocks before getting the full result" + assert ( + len(partial_results[0][1]) == 0 + ), "Expected no non-empty partial result blocks before getting the full result" # Full result full_results = client.readDataWithoutProgress() if support_partial_result: for full_result, partial_result in zip(full_results, partial_results): - assert full_result[0] == partial_result[0], "Keys in blocks should be in the same order" - key = full_result[0] - if key in invariants: - full_value = full_result[1] - partial_value = partial_result[1] - assert invariants[key](partial_value, full_value), f"Problem with the invariant between full and partial result for key: {key}. Partial value {partial_value}. Full value {full_value}" + assert ( + full_result[0] == partial_result[0] + ), f"Keys in blocks should be in the same order. Full results keys {get_keys(full_results)}. Partial results keys {get_keys(partial_results)}" + + key = full_result[0] + if key in invariants: + full_value = full_result[1] + partial_value = partial_result[1] + assert ( + invariants[key](partial_value, full_value) + ), f"Problem with the invariant between full and partial result for key: {key}. Partial value {partial_value}. Full value {full_value}" for key, value in full_results: if isinstance(value[0], int): From d5c1106790d927f7ae90f3d6cf85a8504292c586 Mon Sep 17 00:00:00 2001 From: alexX512 Date: Mon, 7 Aug 2023 17:10:54 +0000 Subject: [PATCH 042/433] Style fix --- ...rtial_aggregating_result_during_query_execution.python | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) diff --git a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python index e4fe9dcfad0..854c36748a2 100644 --- a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python +++ b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python @@ -48,7 +48,9 @@ def run_query_without_errors(query, support_partial_result, invariants=None): if key in invariants: old_value = old_result[1] new_value = new_result[1] - assert invariants[key](old_value, new_value), f"Problem with the invariant between old and new versions of a partial result for key: {key}. Old value {old_value}, new value {new_value}" + assert ( + invariants[key](old_value, new_value) + ), f"Problem with the invariant between old and new versions of a partial result for key: {key}. Old value {old_value}, new value {new_value}" else: assert ( len(partial_results[0][1]) == 0 @@ -66,8 +68,8 @@ def run_query_without_errors(query, support_partial_result, invariants=None): if key in invariants: full_value = full_result[1] partial_value = partial_result[1] - assert ( - invariants[key](partial_value, full_value) + assert invariants[key]( + partial_value, full_value ), f"Problem with the invariant between full and partial result for key: {key}. Partial value {partial_value}. Full value {full_value}" for key, value in full_results: From bbbf7f4c161f7bd2e6455bad164b2681f18dd5ab Mon Sep 17 00:00:00 2001 From: alexX512 Date: Mon, 7 Aug 2023 17:26:34 +0000 Subject: [PATCH 043/433] Style fix --- ...4_partial_aggregating_result_during_query_execution.python | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python index 854c36748a2..25e9e6b73c9 100644 --- a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python +++ b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python @@ -48,8 +48,8 @@ def run_query_without_errors(query, support_partial_result, invariants=None): if key in invariants: old_value = old_result[1] new_value = new_result[1] - assert ( - invariants[key](old_value, new_value) + assert invariants[key]( + old_value, new_value ), f"Problem with the invariant between old and new versions of a partial result for key: {key}. Old value {old_value}, new value {new_value}" else: assert ( From 12d262ed7ad779e6f2ed9961ee26719eded8d91f Mon Sep 17 00:00:00 2001 From: chen768959 <934103231@qq.com> Date: Tue, 8 Aug 2023 11:48:08 +0800 Subject: [PATCH 044/433] getPossiblePtr no longer has a default value, optimize the optimizeTypeIds method. --- src/DataTypes/FieldToDataType.cpp | 2 +- src/DataTypes/IDataType.h | 2 +- src/DataTypes/getLeastSupertype.cpp | 67 +++++-------------- .../02832_integer_type_inference.reference | 3 +- .../02832_integer_type_inference.sql | 3 +- 5 files changed, 24 insertions(+), 53 deletions(-) diff --git a/src/DataTypes/FieldToDataType.cpp b/src/DataTypes/FieldToDataType.cpp index 5947eff7ea2..9b8e476067e 100644 --- a/src/DataTypes/FieldToDataType.cpp +++ b/src/DataTypes/FieldToDataType.cpp @@ -155,7 +155,7 @@ DataTypePtr FieldToDataType::operator() (const Array & x) const for (const Field & elem : x) element_types.emplace_back(applyVisitor(*this, elem)); - return std::make_shared(getLeastSupertype(element_types, false)); + return std::make_shared(getLeastSupertype(element_types, true)); } template diff --git a/src/DataTypes/IDataType.h b/src/DataTypes/IDataType.h index 330836f9d92..52962462d7e 100644 --- a/src/DataTypes/IDataType.h +++ b/src/DataTypes/IDataType.h @@ -73,7 +73,7 @@ public: DataTypePtr getPtr() const { return shared_from_this(); } - virtual DataTypes getPossiblePtr() const { return { shared_from_this() }; } + virtual DataTypes getPossiblePtr() const { return {}; } /// Name of data type family (example: FixedString, Array). virtual const char * getFamilyName() const = 0; diff --git a/src/DataTypes/getLeastSupertype.cpp b/src/DataTypes/getLeastSupertype.cpp index c04130a58e4..d8be5fab91e 100644 --- a/src/DataTypes/getLeastSupertype.cpp +++ b/src/DataTypes/getLeastSupertype.cpp @@ -620,24 +620,6 @@ void optimizeTypeIds(const DataTypes & types, TypeIndexSet & type_ids) } }; - auto maximize = [](size_t & what, size_t value, bool & only_unsigned, bool & only_signed, bool & both) - { - if (value > what) - { - what = value; - only_unsigned = false; - only_signed = false; - both = false; - return true; - }else if (value == what) - { - return true; - } - - return false; - }; - - size_t max_bits_of_integer = 0; bool only_unsigned = false; bool only_signed = false; bool both = false; @@ -652,41 +634,28 @@ void optimizeTypeIds(const DataTypes & types, TypeIndexSet & type_ids) for (const auto & type : types) { TypeIndex type_id = type->getTypeId(); - bool is_max_bits = false; - if (type_id == TypeIndex::UInt8 || type_id == TypeIndex::Int8) - is_max_bits = maximize(max_bits_of_integer, 8, only_unsigned, only_signed, both); - else if (type_id == TypeIndex::UInt16 || type_id == TypeIndex::Int16) - is_max_bits = maximize(max_bits_of_integer, 16, only_unsigned, only_signed, both); - else if (type_id == TypeIndex::UInt32 || type_id == TypeIndex::Int32) - is_max_bits = maximize(max_bits_of_integer, 32, only_unsigned, only_signed, both); - else if (type_id == TypeIndex::UInt64 || type_id == TypeIndex::Int64) - is_max_bits = maximize(max_bits_of_integer, 64, only_unsigned, only_signed, both); - - if (is_max_bits) + bool type_is_unsigned = is_unsigned(type_id); + bool type_is_both = false; + for (const auto & possible_type : type->getPossiblePtr()) { - bool type_is_unsigned = is_unsigned(type_id); - bool type_is_both = false; - for (const auto & possible_type : type->getPossiblePtr()) + if (type_is_unsigned != is_unsigned(possible_type->getTypeId())) { - if (type_is_unsigned != is_unsigned(possible_type->getTypeId())) - { - type_is_both = true; - break; - } + type_is_both = true; + break; } - - if (type_is_unsigned) - has_unsigned = true; - else - has_signed = true; - - if (type_is_both) - both = true; - else if (type_is_unsigned) - only_unsigned = true; - else - only_signed = true; } + + if (type_is_unsigned) + has_unsigned = true; + else + has_signed = true; + + if (type_is_both) + both = true; + else if (type_is_unsigned) + only_unsigned = true; + else + only_signed = true; } auto optimize_type_id = [&is_unsigned](const DataTypePtr & type, bool try_change_unsigned) diff --git a/tests/queries/0_stateless/02832_integer_type_inference.reference b/tests/queries/0_stateless/02832_integer_type_inference.reference index e1bfb315d40..5a01bd4cd11 100644 --- a/tests/queries/0_stateless/02832_integer_type_inference.reference +++ b/tests/queries/0_stateless/02832_integer_type_inference.reference @@ -1 +1,2 @@ -[-4741124612489978151,-3236599669630092879] +[-4741124612489978151,-3236599669630092879,5607475129431807682] +[100,-100,5607475129431807682,5607475129431807683] diff --git a/tests/queries/0_stateless/02832_integer_type_inference.sql b/tests/queries/0_stateless/02832_integer_type_inference.sql index 9c558b3ae67..221e929d705 100644 --- a/tests/queries/0_stateless/02832_integer_type_inference.sql +++ b/tests/queries/0_stateless/02832_integer_type_inference.sql @@ -1 +1,2 @@ -select [-4741124612489978151, -3236599669630092879]; +select [-4741124612489978151, -3236599669630092879, 5607475129431807682]; +select [100, -100, 5607475129431807682, 5607475129431807683]; From 1dbe007ee58337b7ae8894f8fbf1f8d9078bec27 Mon Sep 17 00:00:00 2001 From: Antonio Andelic Date: Tue, 8 Aug 2023 13:12:07 +0000 Subject: [PATCH 045/433] Add coordination setting for async replication --- src/Coordination/Changelog.cpp | 11 +++++------ src/Coordination/Changelog.h | 8 ++++++++ src/Coordination/CoordinationSettings.h | 6 ++++-- src/Coordination/KeeperLogStore.cpp | 5 ++--- src/Coordination/KeeperLogStore.h | 2 +- src/Coordination/KeeperServer.cpp | 13 +++++++++---- src/Coordination/KeeperStateManager.cpp | 18 +++++++++++++----- src/Coordination/KeeperStateManager.h | 4 ++-- 8 files changed, 44 insertions(+), 23 deletions(-) diff --git a/src/Coordination/Changelog.cpp b/src/Coordination/Changelog.cpp index 7f972f67846..2af68173588 100644 --- a/src/Coordination/Changelog.cpp +++ b/src/Coordination/Changelog.cpp @@ -586,13 +586,15 @@ private: std::unique_ptr read_buf; }; -Changelog::Changelog(Poco::Logger * log_, LogFileSettings log_file_settings, KeeperContextPtr keeper_context_) +Changelog::Changelog( + Poco::Logger * log_, LogFileSettings log_file_settings, FlushSettings flush_settings_, KeeperContextPtr keeper_context_) : changelogs_detached_dir("detached") , rotate_interval(log_file_settings.rotate_interval) , log(log_) , write_operations(std::numeric_limits::max()) , append_completion_queue(std::numeric_limits::max()) , keeper_context(std::move(keeper_context_)) + , flush_settings(flush_settings_) { if (auto latest_log_disk = getLatestLogDisk(); log_file_settings.force_sync && dynamic_cast(latest_log_disk.get()) == nullptr) @@ -1017,12 +1019,9 @@ void Changelog::writeThread() size_t pending_appends = 0; bool try_batch_flush = false; - /// turn into setting - static constexpr size_t max_flush_batch_size = 1000; - const auto flush_logs = [&](const auto & flush) { - LOG_INFO(log, "Flushing {} logs", pending_appends); + LOG_TRACE(log, "Flushing {} logs", pending_appends); { std::lock_guard writer_lock(writer_mutex); @@ -1088,7 +1087,7 @@ void Changelog::writeThread() if (batch_append_ok) { /// we can try batching more logs for flush - if (pending_appends < max_flush_batch_size) + if (pending_appends < flush_settings.max_flush_batch_size) { try_batch_flush = true; continue; diff --git a/src/Coordination/Changelog.h b/src/Coordination/Changelog.h index 9789a6b03dd..dddcb9aa218 100644 --- a/src/Coordination/Changelog.h +++ b/src/Coordination/Changelog.h @@ -82,6 +82,11 @@ struct LogFileSettings uint64_t overallocate_size = 0; }; +struct FlushSettings +{ + uint64_t max_flush_batch_size = 1000; +}; + /// Simplest changelog with files rotation. /// No compression, no metadata, just entries with headers one by one. /// Able to read broken files/entries and discard them. Not thread safe. @@ -91,6 +96,7 @@ public: Changelog( Poco::Logger * log_, LogFileSettings log_file_settings, + FlushSettings flush_settings, KeeperContextPtr keeper_context_); Changelog(Changelog &&) = delete; @@ -229,6 +235,8 @@ private: KeeperContextPtr keeper_context; + const FlushSettings flush_settings; + bool initialized = false; }; diff --git a/src/Coordination/CoordinationSettings.h b/src/Coordination/CoordinationSettings.h index 220265b6cb0..80a7c245de3 100644 --- a/src/Coordination/CoordinationSettings.h +++ b/src/Coordination/CoordinationSettings.h @@ -38,8 +38,9 @@ struct Settings; M(UInt64, stale_log_gap, 10000, "When node became stale and should receive snapshots from leader", 0) \ M(UInt64, fresh_log_gap, 200, "When node became fresh", 0) \ M(UInt64, max_request_queue_size, 100000, "Maximum number of request that can be in queue for processing", 0) \ - M(UInt64, max_requests_batch_size, 100, "Max size of batch of requests that can be sent to RAFT", 0) \ + M(UInt64, max_requests_batch_size, 1000, "Max size of batch of requests that can be sent to RAFT", 0) \ M(UInt64, max_requests_batch_bytes_size, 100*1024, "Max size in bytes of batch of requests that can be sent to RAFT", 0) \ + M(UInt64, max_flush_batch_size, 1000, "Max size of batch of requests that can be flushed together", 0) \ M(UInt64, max_requests_quick_batch_size, 100, "Max size of batch of requests to try to get before proceeding with RAFT. Keeper will not wait for requests but take only requests that are already in queue" , 0) \ M(Bool, quorum_reads, false, "Execute read requests as writes through whole RAFT consesus with similar speed", 0) \ M(Bool, force_sync, true, "Call fsync on each change in RAFT changelog", 0) \ @@ -48,7 +49,8 @@ struct Settings; M(UInt64, configuration_change_tries_count, 20, "How many times we will try to apply configuration change (add/remove server) to the cluster", 0) \ M(UInt64, max_log_file_size, 50 * 1024 * 1024, "Max size of the Raft log file. If possible, each created log file will preallocate this amount of bytes on disk. Set to 0 to disable the limit", 0) \ M(UInt64, log_file_overallocate_size, 50 * 1024 * 1024, "If max_log_file_size is not set to 0, this value will be added to it for preallocating bytes on disk. If a log record is larger than this value, it could lead to uncaught out-of-space issues so a larger value is preferred", 0) \ - M(UInt64, min_request_size_for_cache, 50 * 1024, "Minimal size of the request to cache the deserialization result. Caching can have negative effect on latency for smaller requests, set to 0 to disable", 0) + M(UInt64, min_request_size_for_cache, 50 * 1024, "Minimal size of the request to cache the deserialization result. Caching can have negative effect on latency for smaller requests, set to 0 to disable", 0) \ + M(Bool, async_replication, false, "Enable async replication. All write and read guarantees are preserved while better performance is achieved. Settings is disabled by default to not break backwards compatibility.", 0) DECLARE_SETTINGS_TRAITS(CoordinationSettingsTraits, LIST_OF_COORDINATION_SETTINGS) diff --git a/src/Coordination/KeeperLogStore.cpp b/src/Coordination/KeeperLogStore.cpp index a9153475dbc..b55b083cb5c 100644 --- a/src/Coordination/KeeperLogStore.cpp +++ b/src/Coordination/KeeperLogStore.cpp @@ -6,9 +6,8 @@ namespace DB { -KeeperLogStore::KeeperLogStore(LogFileSettings log_file_settings, KeeperContextPtr keeper_context) - : log(&Poco::Logger::get("KeeperLogStore")) - , changelog(log, log_file_settings, keeper_context) +KeeperLogStore::KeeperLogStore(LogFileSettings log_file_settings, FlushSettings flush_settings, KeeperContextPtr keeper_context) + : log(&Poco::Logger::get("KeeperLogStore")), changelog(log, log_file_settings, flush_settings, keeper_context) { if (log_file_settings.force_sync) LOG_INFO(log, "force_sync enabled"); diff --git a/src/Coordination/KeeperLogStore.h b/src/Coordination/KeeperLogStore.h index 6e71d8c55cf..de9205241bd 100644 --- a/src/Coordination/KeeperLogStore.h +++ b/src/Coordination/KeeperLogStore.h @@ -14,7 +14,7 @@ namespace DB class KeeperLogStore : public nuraft::log_store { public: - KeeperLogStore(LogFileSettings log_file_settings, KeeperContextPtr keeper_context); + KeeperLogStore(LogFileSettings log_file_settings, FlushSettings flush_settings, KeeperContextPtr keeper_context); /// Read log storage from filesystem starting from last_commited_log_index void init(uint64_t last_commited_log_index, uint64_t logs_to_keep); diff --git a/src/Coordination/KeeperServer.cpp b/src/Coordination/KeeperServer.cpp index 88b9f1cedb4..06d396e46ce 100644 --- a/src/Coordination/KeeperServer.cpp +++ b/src/Coordination/KeeperServer.cpp @@ -208,28 +208,33 @@ void KeeperServer::loadLatestConfig() { auto latest_snapshot_config = state_machine->getClusterConfig(); auto latest_log_store_config = state_manager->getLatestConfigFromLogStore(); + auto async_replication = coordination_settings->async_replication; if (latest_snapshot_config && latest_log_store_config) { if (latest_snapshot_config->get_log_idx() > latest_log_store_config->get_log_idx()) { LOG_INFO(log, "Will use config from snapshot with log index {}", latest_snapshot_config->get_log_idx()); + latest_snapshot_config->set_async_replication(async_replication); state_manager->save_config(*latest_snapshot_config); } else { - LOG_INFO(log, "Will use config from log store with log index {}", latest_snapshot_config->get_log_idx()); + LOG_INFO(log, "Will use config from log store with log index {}", latest_log_store_config->get_log_idx()); + latest_log_store_config->set_async_replication(async_replication); state_manager->save_config(*latest_log_store_config); } } else if (latest_snapshot_config) { LOG_INFO(log, "No config in log store, will use config from snapshot with log index {}", latest_snapshot_config->get_log_idx()); + latest_snapshot_config->set_async_replication(async_replication); state_manager->save_config(*latest_snapshot_config); } else if (latest_log_store_config) { LOG_INFO(log, "No config in snapshot, will use config from log store with log index {}", latest_log_store_config->get_log_idx()); + latest_log_store_config->set_async_replication(async_replication); state_manager->save_config(*latest_log_store_config); } else @@ -413,7 +418,7 @@ void KeeperServer::startup(const Poco::Util::AbstractConfiguration & config, boo loadLatestConfig(); - last_local_config = state_manager->parseServersConfiguration(config, true).cluster_config; + last_local_config = state_manager->parseServersConfiguration(config, true, coordination_settings->async_replication).cluster_config; launchRaftServer(config, enable_ipv6); @@ -837,12 +842,12 @@ bool KeeperServer::applyConfigUpdate(const ClusterUpdateAction & action) ClusterUpdateActions KeeperServer::getRaftConfigurationDiff(const Poco::Util::AbstractConfiguration & config) { - auto diff = state_manager->getRaftConfigurationDiff(config); + auto diff = state_manager->getRaftConfigurationDiff(config, coordination_settings); if (!diff.empty()) { std::lock_guard lock{server_write_mutex}; - last_local_config = state_manager->parseServersConfiguration(config, true).cluster_config; + last_local_config = state_manager->parseServersConfiguration(config, true, coordination_settings->async_replication).cluster_config; } return diff; diff --git a/src/Coordination/KeeperStateManager.cpp b/src/Coordination/KeeperStateManager.cpp index 902c2ef595f..f52d0749b10 100644 --- a/src/Coordination/KeeperStateManager.cpp +++ b/src/Coordination/KeeperStateManager.cpp @@ -10,6 +10,7 @@ #include #include #include +#include "Coordination/CoordinationSettings.h" namespace DB { @@ -74,7 +75,7 @@ std::unordered_map getClientPorts(const Poco::Util::Abstrac /// 4. No duplicate IDs /// 5. Our ID present in hostnames list KeeperStateManager::KeeperConfigurationWrapper -KeeperStateManager::parseServersConfiguration(const Poco::Util::AbstractConfiguration & config, bool allow_without_us) const +KeeperStateManager::parseServersConfiguration(const Poco::Util::AbstractConfiguration & config, bool allow_without_us, bool enable_async_replication) const { const bool hostname_checks_enabled = config.getBool(config_prefix + ".hostname_checks_enabled", true); @@ -184,7 +185,8 @@ KeeperStateManager::parseServersConfiguration(const Poco::Util::AbstractConfigur total_servers++; } - result.cluster_config->set_async_replication(true); + result.cluster_config->set_async_replication(enable_async_replication); + if (!result.config && !allow_without_us) throw Exception(ErrorCodes::RAFT_ERROR, "Our server id {} not found in raft_configuration section", my_server_id); @@ -221,6 +223,7 @@ KeeperStateManager::KeeperStateManager(int server_id_, const std::string & host, , secure(false) , log_store(nuraft::cs_new( LogFileSettings{.force_sync = false, .compress_logs = false, .rotate_interval = 5000}, + FlushSettings{}, keeper_context_)) , server_state_file_name("state") , keeper_context(keeper_context_) @@ -243,7 +246,7 @@ KeeperStateManager::KeeperStateManager( : my_server_id(my_server_id_) , secure(config.getBool(config_prefix_ + ".raft_configuration.secure", false)) , config_prefix(config_prefix_) - , configuration_wrapper(parseServersConfiguration(config, false)) + , configuration_wrapper(parseServersConfiguration(config, false, coordination_settings->async_replication)) , log_store(nuraft::cs_new( LogFileSettings { @@ -253,6 +256,10 @@ KeeperStateManager::KeeperStateManager( .max_size = coordination_settings->max_log_file_size, .overallocate_size = coordination_settings->log_file_overallocate_size }, + FlushSettings + { + .max_flush_batch_size = coordination_settings->max_flush_batch_size, + }, keeper_context_)) , server_state_file_name(server_state_file_name_) , keeper_context(keeper_context_) @@ -452,9 +459,10 @@ nuraft::ptr KeeperStateManager::read_state() return nullptr; } -ClusterUpdateActions KeeperStateManager::getRaftConfigurationDiff(const Poco::Util::AbstractConfiguration & config) const +ClusterUpdateActions KeeperStateManager::getRaftConfigurationDiff( + const Poco::Util::AbstractConfiguration & config, const CoordinationSettingsPtr & coordination_settings) const { - auto new_configuration_wrapper = parseServersConfiguration(config, true); + auto new_configuration_wrapper = parseServersConfiguration(config, true, coordination_settings->async_replication); std::unordered_map new_ids, old_ids; for (const auto & new_server : new_configuration_wrapper.cluster_config->get_servers()) diff --git a/src/Coordination/KeeperStateManager.h b/src/Coordination/KeeperStateManager.h index 5abeea604b5..e402143c179 100644 --- a/src/Coordination/KeeperStateManager.h +++ b/src/Coordination/KeeperStateManager.h @@ -93,7 +93,7 @@ public: ClusterConfigPtr getLatestConfigFromLogStore() const; // TODO (myrrc) This should be removed once "reconfig" is stabilized - ClusterUpdateActions getRaftConfigurationDiff(const Poco::Util::AbstractConfiguration & config) const; + ClusterUpdateActions getRaftConfigurationDiff(const Poco::Util::AbstractConfiguration & config, const CoordinationSettingsPtr & coordination_settings) const; private: const String & getOldServerStatePath(); @@ -131,7 +131,7 @@ private: public: /// Parse configuration from xml config. - KeeperConfigurationWrapper parseServersConfiguration(const Poco::Util::AbstractConfiguration & config, bool allow_without_us) const; + KeeperConfigurationWrapper parseServersConfiguration(const Poco::Util::AbstractConfiguration & config, bool allow_without_us, bool enable_async_replication) const; }; } From d8904ffa694500676734c039fb04ddda3b7e00d0 Mon Sep 17 00:00:00 2001 From: slvrtrn Date: Tue, 8 Aug 2023 23:48:23 +0200 Subject: [PATCH 046/433] Simplified prepared statements handling for MySQL interface --- src/Core/MySQL/PacketsGeneric.h | 3 + src/Core/MySQL/PacketsPreparedStatements.cpp | 40 +++++++++++ src/Core/MySQL/PacketsPreparedStatements.h | 35 +++++++++ src/Server/MySQLHandler.cpp | 76 +++++++++++++++++++- src/Server/MySQLHandler.h | 10 +++ 5 files changed, 163 insertions(+), 1 deletion(-) create mode 100644 src/Core/MySQL/PacketsPreparedStatements.cpp create mode 100644 src/Core/MySQL/PacketsPreparedStatements.h diff --git a/src/Core/MySQL/PacketsGeneric.h b/src/Core/MySQL/PacketsGeneric.h index cf990bbf15c..969716dfc7a 100644 --- a/src/Core/MySQL/PacketsGeneric.h +++ b/src/Core/MySQL/PacketsGeneric.h @@ -54,6 +54,9 @@ enum Command COM_CHANGE_USER = 0x11, COM_BINLOG_DUMP = 0x12, COM_REGISTER_SLAVE = 0x15, + COM_STMT_PREPARE = 0x16, + COM_STMT_EXECUTE = 0x17, + COM_STMT_CLOSE = 0x19, COM_RESET_CONNECTION = 0x1f, COM_DAEMON = 0x1d, COM_BINLOG_DUMP_GTID = 0x1e diff --git a/src/Core/MySQL/PacketsPreparedStatements.cpp b/src/Core/MySQL/PacketsPreparedStatements.cpp new file mode 100644 index 00000000000..eddded585d5 --- /dev/null +++ b/src/Core/MySQL/PacketsPreparedStatements.cpp @@ -0,0 +1,40 @@ +#include +#include +#include +#include + +namespace DB +{ +namespace MySQLProtocol +{ + namespace PreparedStatements + { + size_t PrepareStatementResponseOK::getPayloadSize() const + { + return 13; + } + + void PrepareStatementResponseOK::writePayloadImpl(WriteBuffer & buffer) const + { + buffer.write(reinterpret_cast(&status), 1); + buffer.write(reinterpret_cast(&statement_id), 4); + buffer.write(reinterpret_cast(&num_columns), 2); + buffer.write(reinterpret_cast(&num_params), 2); + buffer.write(reinterpret_cast(&reserved_1), 1); + buffer.write(reinterpret_cast(&warnings_count), 2); + buffer.write(0x0); // RESULTSET_METADATA_NONE + } + + void PrepareStatementResponseOK::readPayloadImpl([[maybe_unused]] ReadBuffer & payload) + { + throw Exception(ErrorCodes::NOT_IMPLEMENTED, "PrepareStatementResponseOK::readPayloadImpl is not implemented"); + } + + PrepareStatementResponseOK::PrepareStatementResponseOK( + uint32_t statement_id_, uint16_t num_columns_, uint16_t num_params_, uint16_t warnings_count_) + : statement_id(statement_id_), num_columns(num_columns_), num_params(num_params_), warnings_count(warnings_count_) + { + } + } +} +} diff --git a/src/Core/MySQL/PacketsPreparedStatements.h b/src/Core/MySQL/PacketsPreparedStatements.h new file mode 100644 index 00000000000..7f180d4d142 --- /dev/null +++ b/src/Core/MySQL/PacketsPreparedStatements.h @@ -0,0 +1,35 @@ +#pragma once + +#include +#include + +namespace DB +{ +namespace MySQLProtocol +{ + namespace PreparedStatements + { + // https://dev.mysql.com/doc/dev/mysql-server/latest/page_protocol_com_stmt_prepare.html#sect_protocol_com_stmt_prepare_response_ok + class PrepareStatementResponseOK : public IMySQLWritePacket, public IMySQLReadPacket + { + public: + uint8_t status = 0x00; + uint32_t statement_id; + uint16_t num_columns; + uint16_t num_params; + uint8_t reserved_1 = 0; + uint16_t warnings_count; + + protected: + size_t getPayloadSize() const override; + + void readPayloadImpl(ReadBuffer & payload) override; + + void writePayloadImpl(WriteBuffer & buffer) const override; + + public: + PrepareStatementResponseOK(uint32_t statement_id_, uint16_t num_columns_, uint16_t num_params_, uint16_t warnings_count_); + }; + } +} +} diff --git a/src/Server/MySQLHandler.cpp b/src/Server/MySQLHandler.cpp index f98b86e6cf8..a0018d0dc27 100644 --- a/src/Server/MySQLHandler.cpp +++ b/src/Server/MySQLHandler.cpp @@ -4,6 +4,7 @@ #include #include #include +#include #include #include #include @@ -40,6 +41,7 @@ using namespace MySQLProtocol; using namespace MySQLProtocol::Generic; using namespace MySQLProtocol::ProtocolText; using namespace MySQLProtocol::ConnectionPhase; +using namespace MySQLProtocol::PreparedStatements; #if USE_SSL using Poco::Net::SecureStreamSocket; @@ -181,6 +183,15 @@ void MySQLHandler::run() case COM_PING: comPing(); break; + case COM_STMT_PREPARE: + comStmtPrepare(payload); + break; + case COM_STMT_EXECUTE: + comStmtExecute(payload); + break; + case COM_STMT_CLOSE: + comStmtClose(payload); + break; default: throw Exception(ErrorCodes::NOT_IMPLEMENTED, "Command {} is not implemented.", command); } @@ -254,7 +265,8 @@ void MySQLHandler::authenticate(const String & user_name, const String & auth_pl { try { - // For compatibility with JavaScript MySQL client, Native41 authentication plugin is used when possible (if password is specified using double SHA1). Otherwise SHA256 plugin is used. + // For compatibility with JavaScript MySQL client, Native41 authentication plugin is used when possible + // (if password is specified using double SHA1). Otherwise, SHA256 plugin is used. if (session->getAuthenticationTypeOrLogInFailure(user_name) == DB::AuthenticationType::SHA256_PASSWORD) { authPluginSSL(); @@ -371,6 +383,68 @@ void MySQLHandler::comQuery(ReadBuffer & payload) } } +void MySQLHandler::comStmtPrepare(DB::ReadBuffer & payload) +{ + String query; + readStringUntilEOF(query, payload); + + uint32_t statement_id = current_prepared_statement_id; + if (current_prepared_statement_id == std::numeric_limits::max()) + { + current_prepared_statement_id = 0; + } + else + { + current_prepared_statement_id++; + } + + // Key collisions should not happen here, as we remove the elements from the map with COM_STMT_CLOSE, + // and we have quite a big range of available identifiers with 32-bit unsigned integer + if (prepared_statements_map.contains(statement_id)) [[unlikely]] + { + LOG_ERROR( + log, + "Failed to store a new statement `{}` with id {}; it is already taken by `{}`", + query, + statement_id, + prepared_statements_map.at(statement_id)); + packet_endpoint->sendPacket(ERRPacket(), true); + return; + } + prepared_statements_map.emplace(statement_id, query); + + packet_endpoint->sendPacket(PrepareStatementResponseOK(statement_id, 0, 0, 0), true); +} + +void MySQLHandler::comStmtExecute(ReadBuffer & payload) +{ + uint32_t statement_id; + payload.readStrict(reinterpret_cast(&statement_id), 4); + + if (!prepared_statements_map.contains(statement_id)) [[unlikely]] + { + LOG_ERROR(log, "Could not find prepared statement with id {}", statement_id); + packet_endpoint->sendPacket(ERRPacket(), true); + return; + } + + // Temporary workaround as we work only with queries that do not bind any parameters atm + ReadBufferFromString com_query_payload(prepared_statements_map.at(statement_id)); + MySQLHandler::comQuery(com_query_payload); +}; + +void MySQLHandler::comStmtClose([[maybe_unused]] ReadBuffer & payload) { + uint32_t statement_id; + payload.readStrict(reinterpret_cast(&statement_id), 4); + + if (prepared_statements_map.contains(statement_id)) { + prepared_statements_map.erase(statement_id); + } + + // https://dev.mysql.com/doc/dev/mysql-server/latest/page_protocol_com_stmt_close.html + // No response packet is sent back to the client. +}; + void MySQLHandler::authPluginSSL() { throw Exception(ErrorCodes::SUPPORT_IS_DISABLED, diff --git a/src/Server/MySQLHandler.h b/src/Server/MySQLHandler.h index 3366e8792c9..6b8cc56a46e 100644 --- a/src/Server/MySQLHandler.h +++ b/src/Server/MySQLHandler.h @@ -56,6 +56,12 @@ protected: void authenticate(const String & user_name, const String & auth_plugin_name, const String & auth_response); + void comStmtPrepare(ReadBuffer & payload); + + void comStmtExecute(ReadBuffer & payload); + + void comStmtClose(ReadBuffer & payload); + virtual void authPluginSSL(); virtual void finishHandshakeSSL(size_t packet_size, char * buf, size_t pos, std::function read_bytes, MySQLProtocol::ConnectionPhase::HandshakeResponse & packet); @@ -76,6 +82,10 @@ protected: using Replacements = std::unordered_map; Replacements replacements; + uint32_t current_prepared_statement_id = 0; + using PreparedStatementsMap = std::unordered_map; + PreparedStatementsMap prepared_statements_map; + std::unique_ptr auth_plugin; std::shared_ptr in; std::shared_ptr out; From ef3e0f7b836864d1ddcc9d114d319f639654e661 Mon Sep 17 00:00:00 2001 From: Antonio Andelic Date: Wed, 9 Aug 2023 06:25:47 +0000 Subject: [PATCH 047/433] Use async replication in tests --- tests/config/config.d/keeper_port.xml | 2 ++ .../helpers/0_common_enable_keeper_async_replication.xml | 7 +++++++ tests/integration/helpers/cluster.py | 2 ++ tests/integration/helpers/keeper_config1.xml | 2 ++ tests/integration/helpers/keeper_config2.xml | 2 ++ tests/integration/helpers/keeper_config3.xml | 2 ++ tests/jepsen.clickhouse/resources/keeper_config.xml | 1 + tests/jepsen.clickhouse/resources/keeper_config_solo.xml | 1 + 8 files changed, 19 insertions(+) create mode 100644 tests/integration/helpers/0_common_enable_keeper_async_replication.xml diff --git a/tests/config/config.d/keeper_port.xml b/tests/config/config.d/keeper_port.xml index 7db174c5419..beac507304f 100644 --- a/tests/config/config.d/keeper_port.xml +++ b/tests/config/config.d/keeper_port.xml @@ -19,6 +19,8 @@ 0 0 0 + + 1 diff --git a/tests/integration/helpers/0_common_enable_keeper_async_replication.xml b/tests/integration/helpers/0_common_enable_keeper_async_replication.xml new file mode 100644 index 00000000000..4ecada09444 --- /dev/null +++ b/tests/integration/helpers/0_common_enable_keeper_async_replication.xml @@ -0,0 +1,7 @@ + + + + 1 + + + \ No newline at end of file diff --git a/tests/integration/helpers/cluster.py b/tests/integration/helpers/cluster.py index eff44de842a..1f3cd473e29 100644 --- a/tests/integration/helpers/cluster.py +++ b/tests/integration/helpers/cluster.py @@ -4258,6 +4258,8 @@ class ClickHouseInstance: if len(self.custom_dictionaries_paths): write_embedded_config("0_common_enable_dictionaries.xml", self.config_d_dir) + write_embedded_config("0_common_enable_keeper_async_replication.xml", self.config_d_dir) + logging.debug("Generate and write macros file") macros = self.macros.copy() macros["instance"] = self.name diff --git a/tests/integration/helpers/keeper_config1.xml b/tests/integration/helpers/keeper_config1.xml index f40ed9ac6fa..7702aecba9c 100644 --- a/tests/integration/helpers/keeper_config1.xml +++ b/tests/integration/helpers/keeper_config1.xml @@ -20,6 +20,8 @@ false 2000 4000 + + 1 diff --git a/tests/integration/helpers/keeper_config2.xml b/tests/integration/helpers/keeper_config2.xml index d5bdb92a79d..2a1a1c1003c 100644 --- a/tests/integration/helpers/keeper_config2.xml +++ b/tests/integration/helpers/keeper_config2.xml @@ -20,6 +20,8 @@ false 2000 4000 + + 1 diff --git a/tests/integration/helpers/keeper_config3.xml b/tests/integration/helpers/keeper_config3.xml index aa69b554660..035da1bbd22 100644 --- a/tests/integration/helpers/keeper_config3.xml +++ b/tests/integration/helpers/keeper_config3.xml @@ -20,6 +20,8 @@ false 2000 4000 + + 1 diff --git a/tests/jepsen.clickhouse/resources/keeper_config.xml b/tests/jepsen.clickhouse/resources/keeper_config.xml index 1972ef6b917..6bc4ad89839 100644 --- a/tests/jepsen.clickhouse/resources/keeper_config.xml +++ b/tests/jepsen.clickhouse/resources/keeper_config.xml @@ -47,6 +47,7 @@ {snapshot_distance} {stale_log_gap} {reserved_log_items} + 1 diff --git a/tests/jepsen.clickhouse/resources/keeper_config_solo.xml b/tests/jepsen.clickhouse/resources/keeper_config_solo.xml index b20592545cd..0054cad8f85 100644 --- a/tests/jepsen.clickhouse/resources/keeper_config_solo.xml +++ b/tests/jepsen.clickhouse/resources/keeper_config_solo.xml @@ -21,6 +21,7 @@ 1000 2000 4000 + 1 From 7fdb414793e533444808d111fd29f8fda904e15e Mon Sep 17 00:00:00 2001 From: slvrtrn Date: Wed, 9 Aug 2023 15:18:05 +0200 Subject: [PATCH 048/433] One more [[unlikely]] --- src/Server/MySQLHandler.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/Server/MySQLHandler.cpp b/src/Server/MySQLHandler.cpp index a0018d0dc27..868575b701f 100644 --- a/src/Server/MySQLHandler.cpp +++ b/src/Server/MySQLHandler.cpp @@ -389,7 +389,7 @@ void MySQLHandler::comStmtPrepare(DB::ReadBuffer & payload) readStringUntilEOF(query, payload); uint32_t statement_id = current_prepared_statement_id; - if (current_prepared_statement_id == std::numeric_limits::max()) + if (current_prepared_statement_id == std::numeric_limits::max()) [[unlikely]] { current_prepared_statement_id = 0; } From efb041c4d5482a9acc26aa718d5c9b47e1ad95ab Mon Sep 17 00:00:00 2001 From: chen768959 <934103231@qq.com> Date: Wed, 9 Aug 2023 23:03:50 +0800 Subject: [PATCH 049/433] Optimize the code and streamline the functionality by retaining only the feature of automatically recognizing UInt64 as Int64. --- src/Common/ErrorCodes.cpp | 3 + src/DataTypes/DataTypesNumber.h | 19 +++-- src/DataTypes/FieldToDataType.cpp | 23 +----- src/DataTypes/IDataType.cpp | 6 ++ src/DataTypes/IDataType.h | 11 ++- src/DataTypes/getLeastSupertype.cpp | 113 ++++++---------------------- 6 files changed, 61 insertions(+), 114 deletions(-) diff --git a/src/Common/ErrorCodes.cpp b/src/Common/ErrorCodes.cpp index ae8d5f8796d..3df92c941e8 100644 --- a/src/Common/ErrorCodes.cpp +++ b/src/Common/ErrorCodes.cpp @@ -582,6 +582,9 @@ M(697, CANNOT_RESTORE_TO_NONENCRYPTED_DISK) \ M(698, INVALID_REDIS_STORAGE_TYPE) \ M(699, INVALID_REDIS_TABLE_STRUCTURE) \ + M(700, USER_SESSION_LIMIT_EXCEEDED) \ + M(701, CLUSTER_DOESNT_EXIST) \ + M(702, OPPOSITE_SIGN_DATA_TYPE_NOT_FOUND) \ \ M(999, KEEPER_EXCEPTION) \ M(1000, POCO_EXCEPTION) \ diff --git a/src/DataTypes/DataTypesNumber.h b/src/DataTypes/DataTypesNumber.h index 8511955e193..6be07db1062 100644 --- a/src/DataTypes/DataTypesNumber.h +++ b/src/DataTypes/DataTypesNumber.h @@ -18,9 +18,10 @@ class DataTypeNumber final : public DataTypeNumberBase public: DataTypeNumber() = default; - explicit DataTypeNumber(DataTypes data_types) + explicit DataTypeNumber(DataTypePtr opposite_sign_data_type_) : DataTypeNumberBase() - , possible_data_types(std::move(data_types)) + , opposite_sign_data_type(std::move(opposite_sign_data_type_)) + , has_opposite_sign_data_type(true) { } @@ -39,15 +40,23 @@ public: return std::make_shared(); } + bool hasOppositeSignDataType() const override { return has_opposite_sign_data_type; } + DataTypePtr oppositeSignDataType() const override + { + if (!has_opposite_sign_data_type) + IDataType::oppositeSignDataType(); + + return opposite_sign_data_type; + } + SerializationPtr doGetDefaultSerialization() const override { return std::make_shared>(); } - DataTypes getPossiblePtr() const override { return possible_data_types; } - private: - DataTypes possible_data_types; + DataTypePtr opposite_sign_data_type; + bool has_opposite_sign_data_type = false; }; using DataTypeUInt8 = DataTypeNumber; diff --git a/src/DataTypes/FieldToDataType.cpp b/src/DataTypes/FieldToDataType.cpp index 9b8e476067e..2f759e79661 100644 --- a/src/DataTypes/FieldToDataType.cpp +++ b/src/DataTypes/FieldToDataType.cpp @@ -33,34 +33,19 @@ DataTypePtr FieldToDataType::operator() (const Null &) const template DataTypePtr FieldToDataType::operator() (const UInt64 & x) const { - if (x <= std::numeric_limits::max()) return std::make_shared(DataTypes{ std::make_shared() }); if (x <= std::numeric_limits::max()) return std::make_shared(); - if (x <= std::numeric_limits::max()) return std::make_shared(DataTypes{ std::make_shared() }); if (x <= std::numeric_limits::max()) return std::make_shared(); - if (x <= std::numeric_limits::max()) return std::make_shared(DataTypes{ std::make_shared() }); if (x <= std::numeric_limits::max()) return std::make_shared(); - if (x <= std::numeric_limits::max()) return std::make_shared(DataTypes{ std::make_shared() }); + if (x <= std::numeric_limits::max()) return std::make_shared(std::make_shared()); return std::make_shared(); } template DataTypePtr FieldToDataType::operator() (const Int64 & x) const { - if (x >= 0) - { - if (x <= std::numeric_limits::max()) return std::make_shared(); - if (x <= std::numeric_limits::max()) return std::make_shared(DataTypes{ std::make_shared() }); - if (x <= std::numeric_limits::max()) return std::make_shared(); - if (x <= std::numeric_limits::max()) return std::make_shared(DataTypes{ std::make_shared() }); - if (x <= std::numeric_limits::max()) return std::make_shared(); - if (x <= std::numeric_limits::max()) return std::make_shared(DataTypes{ std::make_shared() }); - } - else - { - if (x >= std::numeric_limits::min()) return std::make_shared(); - if (x >= std::numeric_limits::min()) return std::make_shared(); - if (x >= std::numeric_limits::min()) return std::make_shared(); - } + if (x <= std::numeric_limits::max() && x >= std::numeric_limits::min()) return std::make_shared(); + if (x <= std::numeric_limits::max() && x >= std::numeric_limits::min()) return std::make_shared(); + if (x <= std::numeric_limits::max() && x >= std::numeric_limits::min()) return std::make_shared(); return std::make_shared(); } diff --git a/src/DataTypes/IDataType.cpp b/src/DataTypes/IDataType.cpp index 4ffe82039b2..221254e575e 100644 --- a/src/DataTypes/IDataType.cpp +++ b/src/DataTypes/IDataType.cpp @@ -23,6 +23,7 @@ namespace ErrorCodes { extern const int LOGICAL_ERROR; extern const int DATA_TYPE_CANNOT_BE_PROMOTED; + extern const int OPPOSITE_SIGN_DATA_TYPE_NOT_FOUND; extern const int ILLEGAL_COLUMN; } @@ -71,6 +72,11 @@ DataTypePtr IDataType::promoteNumericType() const throw Exception(ErrorCodes::DATA_TYPE_CANNOT_BE_PROMOTED, "Data type {} can't be promoted.", getName()); } +DataTypePtr IDataType::oppositeSignDataType() const +{ + throw Exception(ErrorCodes::OPPOSITE_SIGN_DATA_TYPE_NOT_FOUND, "Opposite sign data type not found for {}.", getName()); +} + size_t IDataType::getSizeOfValueInMemory() const { throw Exception(ErrorCodes::LOGICAL_ERROR, "Value of type {} in memory is not of fixed size.", getName()); diff --git a/src/DataTypes/IDataType.h b/src/DataTypes/IDataType.h index 52962462d7e..867ed6df5e3 100644 --- a/src/DataTypes/IDataType.h +++ b/src/DataTypes/IDataType.h @@ -73,8 +73,6 @@ public: DataTypePtr getPtr() const { return shared_from_this(); } - virtual DataTypes getPossiblePtr() const { return {}; } - /// Name of data type family (example: FixedString, Array). virtual const char * getFamilyName() const = 0; /// Name of corresponding data type in MySQL (exampe: Bigint, Blob, etc) @@ -160,6 +158,15 @@ public: */ virtual DataTypePtr promoteNumericType() const; + /** The data type has an opposite sign DataTypePtr type. + * Data types that can have an opposite sign are typically signed or unsigned types. + */ + virtual bool hasOppositeSignDataType() const { return false; } + + /** Return the opposite sign data type of the current data type. Throw an exception if `hasOppositeSignDataType() == false`. + */ + virtual DataTypePtr oppositeSignDataType() const; + /** Directly insert default value into a column. Default implementation use method IColumn::insertDefault. * This should be overridden if data type default value differs from column default value (example: Enum data types). */ diff --git a/src/DataTypes/getLeastSupertype.cpp b/src/DataTypes/getLeastSupertype.cpp index d8be5fab91e..4c399e99caa 100644 --- a/src/DataTypes/getLeastSupertype.cpp +++ b/src/DataTypes/getLeastSupertype.cpp @@ -605,113 +605,50 @@ DataTypePtr getLeastSupertype(const DataTypes & types, bool optimize_type_ids) void optimizeTypeIds(const DataTypes & types, TypeIndexSet & type_ids) { - // Determine whether the type_id is UInt - auto is_unsigned = [](const TypeIndex & type_id) + auto is_signed_int = [](const TypeIndex & type_id) { switch (type_id) { - case TypeIndex::UInt8: - case TypeIndex::UInt16: - case TypeIndex::UInt32: - case TypeIndex::UInt64: + case TypeIndex::Int8: + case TypeIndex::Int16: + case TypeIndex::Int32: + case TypeIndex::Int64: return true; default: return false; } }; - bool only_unsigned = false; - bool only_signed = false; - bool both = false; - bool has_unsigned = false; - bool has_signed = false; + bool has_signed_int = false; + bool has_uint64_and_has_opposite = false; + TypeIndexSet opposite_type_ids; - // Determine the distribution of maximum signed and unsigned, Example: - // Int64, Int64 = only_signed. - // UInt64, UInt64 = only_unsigned. - // UInt64(possible: Int64), Int64(possible: UInt64) = both. - // UInt64(possible: Int64), Int64 = both, only_signed. + // Determine whether UInt64 in type_ids needs to change its sign. for (const auto & type : types) { - TypeIndex type_id = type->getTypeId(); - bool type_is_unsigned = is_unsigned(type_id); - bool type_is_both = false; - for (const auto & possible_type : type->getPossiblePtr()) + auto type_id = type->getTypeId(); + + if (!has_signed_int) + has_signed_int = is_signed_int(type_id); + + if (type_id == TypeIndex::UInt64) { - if (type_is_unsigned != is_unsigned(possible_type->getTypeId())) + if (!type->hasOppositeSignDataType()) { - type_is_both = true; - break; + has_uint64_and_has_opposite = false; + break ; + }else + { + has_uint64_and_has_opposite = true; + opposite_type_ids.insert(type->oppositeSignDataType()->getTypeId()); } } - - if (type_is_unsigned) - has_unsigned = true; - else - has_signed = true; - - if (type_is_both) - both = true; - else if (type_is_unsigned) - only_unsigned = true; - else - only_signed = true; } - auto optimize_type_id = [&is_unsigned](const DataTypePtr & type, bool try_change_unsigned) + if (has_uint64_and_has_opposite && has_signed_int) { - TypeIndex type_id = type->getTypeId(); - switch (type_id) - { - case TypeIndex::UInt8: - case TypeIndex::UInt16: - case TypeIndex::UInt32: - case TypeIndex::UInt64: - if (try_change_unsigned) - return type_id; - break ; - case TypeIndex::Int8: - case TypeIndex::Int16: - case TypeIndex::Int32: - case TypeIndex::Int64: - if (!try_change_unsigned) - return type_id; - break ; - default: - return type_id; - } - - for (const auto & other_type : type->getPossiblePtr()) - { - TypeIndex other_type_id = other_type->getTypeId(); - if ((try_change_unsigned && is_unsigned(other_type_id)) - || (!try_change_unsigned && !is_unsigned(other_type_id))) - { - return other_type_id; - } - } - - return type_id; - }; - - // optimize type_ids - if (both) - { - // Example: Int64(possible: UInt32), UInt64 = UInt32, UInt64 - if (only_unsigned && !only_signed) - { - type_ids.clear(); - for (const auto & type : types) - type_ids.insert(optimize_type_id(type, true)); - } - // Example: UInt64(possible: Int64), Int64 = Int64, Int64 - // Int64(possible: UInt32), UInt64(possible: Int64) = Int64, Int64 - else if ((only_signed && !only_unsigned) || (has_unsigned && has_signed && !only_signed && !only_unsigned)) - { - type_ids.clear(); - for (const auto & type : types) - type_ids.insert(optimize_type_id(type, false)); - } + type_ids.erase(TypeIndex::UInt64); + type_ids.insert(opposite_type_ids.begin(), opposite_type_ids.end()); } } From da2a041d6891058bee5bbed621a312526be8ef21 Mon Sep 17 00:00:00 2001 From: chen768959 <934103231@qq.com> Date: Thu, 10 Aug 2023 11:23:43 +0800 Subject: [PATCH 050/433] Temporarily modify ErrorCodes to avoid conflicts. --- src/Common/ErrorCodes.cpp | 3 --- src/DataTypes/IDataType.cpp | 4 ++-- 2 files changed, 2 insertions(+), 5 deletions(-) diff --git a/src/Common/ErrorCodes.cpp b/src/Common/ErrorCodes.cpp index 3df92c941e8..ae8d5f8796d 100644 --- a/src/Common/ErrorCodes.cpp +++ b/src/Common/ErrorCodes.cpp @@ -582,9 +582,6 @@ M(697, CANNOT_RESTORE_TO_NONENCRYPTED_DISK) \ M(698, INVALID_REDIS_STORAGE_TYPE) \ M(699, INVALID_REDIS_TABLE_STRUCTURE) \ - M(700, USER_SESSION_LIMIT_EXCEEDED) \ - M(701, CLUSTER_DOESNT_EXIST) \ - M(702, OPPOSITE_SIGN_DATA_TYPE_NOT_FOUND) \ \ M(999, KEEPER_EXCEPTION) \ M(1000, POCO_EXCEPTION) \ diff --git a/src/DataTypes/IDataType.cpp b/src/DataTypes/IDataType.cpp index 221254e575e..092b9ed8e0e 100644 --- a/src/DataTypes/IDataType.cpp +++ b/src/DataTypes/IDataType.cpp @@ -23,7 +23,6 @@ namespace ErrorCodes { extern const int LOGICAL_ERROR; extern const int DATA_TYPE_CANNOT_BE_PROMOTED; - extern const int OPPOSITE_SIGN_DATA_TYPE_NOT_FOUND; extern const int ILLEGAL_COLUMN; } @@ -74,7 +73,8 @@ DataTypePtr IDataType::promoteNumericType() const DataTypePtr IDataType::oppositeSignDataType() const { - throw Exception(ErrorCodes::OPPOSITE_SIGN_DATA_TYPE_NOT_FOUND, "Opposite sign data type not found for {}.", getName()); + // TODO: Should use ErrorCodes::OPPOSITE_SIGN_DATA_TYPE_NOT_FOUND. + throw Exception(702, "Opposite sign data type not found for {}.", getName()); } size_t IDataType::getSizeOfValueInMemory() const From e4a1780bd74ad9de0dbce1e97509c396911a500b Mon Sep 17 00:00:00 2001 From: zvonand Date: Thu, 10 Aug 2023 15:29:27 +0300 Subject: [PATCH 051/433] update docs according to the PR + fix the same for HDFS --- docs/en/sql-reference/table-functions/file.md | 3 ++- docs/en/sql-reference/table-functions/hdfs.md | 2 +- docs/ru/sql-reference/table-functions/file.md | 2 +- docs/ru/sql-reference/table-functions/hdfs.md | 2 +- src/Storages/HDFS/StorageHDFS.cpp | 8 ++++++-- 5 files changed, 11 insertions(+), 6 deletions(-) diff --git a/docs/en/sql-reference/table-functions/file.md b/docs/en/sql-reference/table-functions/file.md index 4db9494502e..508e58ed3ca 100644 --- a/docs/en/sql-reference/table-functions/file.md +++ b/docs/en/sql-reference/table-functions/file.md @@ -141,7 +141,8 @@ Multiple path components can have globs. For being processed file must exist and - `*` — Substitutes any number of any characters except `/` including empty string. - `?` — Substitutes any single character. -- `{some_string,another_string,yet_another_one}` — Substitutes any of strings `'some_string', 'another_string', 'yet_another_one'`, including `/`. +- `{some_string,another_string,yet_another_one}` — Substitutes any of strings `'some_string', 'another_string', 'yet_another_one'`, including `/`. In case at least one of strings contains `/`, `'permission denied'` errors may be ignored. + - `{N..M}` — Substitutes any number in range from N to M including both borders. - `**` - Fetches all files inside the folder recursively. diff --git a/docs/en/sql-reference/table-functions/hdfs.md b/docs/en/sql-reference/table-functions/hdfs.md index 680ac54ee78..b4f63b35592 100644 --- a/docs/en/sql-reference/table-functions/hdfs.md +++ b/docs/en/sql-reference/table-functions/hdfs.md @@ -45,7 +45,7 @@ Multiple path components can have globs. For being processed file should exists - `*` — Substitutes any number of any characters except `/` including empty string. - `?` — Substitutes any single character. -- `{some_string,another_string,yet_another_one}` — Substitutes any of strings `'some_string', 'another_string', 'yet_another_one'`. +- `{some_string,another_string,yet_another_one}` — Substitutes any of strings `'some_string', 'another_string', 'yet_another_one'`, including `/`. In case at least one of strings contains `/`, `'permission denied'` errors may be ignored. - `{N..M}` — Substitutes any number in range from N to M including both borders. Constructions with `{}` are similar to the [remote table function](../../sql-reference/table-functions/remote.md)). diff --git a/docs/ru/sql-reference/table-functions/file.md b/docs/ru/sql-reference/table-functions/file.md index 83ef115aacd..d7107ba5950 100644 --- a/docs/ru/sql-reference/table-functions/file.md +++ b/docs/ru/sql-reference/table-functions/file.md @@ -79,7 +79,7 @@ SELECT * FROM file('test.csv', 'CSV', 'column1 UInt32, column2 UInt32, column3 U - `*` — заменяет любое количество любых символов кроме `/`, включая отсутствие символов. - `?` — заменяет ровно один любой символ. -- `{some_string,another_string,yet_another_one}` — заменяет любую из строк `'some_string', 'another_string', 'yet_another_one'`, причём строка может содержать `/`. +- `{some_string,another_string,yet_another_one}` — заменяет любую из строк `'some_string', 'another_string', 'yet_another_one'`, причём строка может содержать `/`. В случае, если в какой-либо из строк содержится `/`, то ошибки доступа (permission denied) к существующим, но недоступным директориям/файлам могут быть проигнорированы. - `{N..M}` — заменяет любое число в интервале от `N` до `M` включительно (может содержать ведущие нули). Конструкция с `{}` аналогична табличной функции [remote](remote.md). diff --git a/docs/ru/sql-reference/table-functions/hdfs.md b/docs/ru/sql-reference/table-functions/hdfs.md index b0d182eef14..350d483b592 100644 --- a/docs/ru/sql-reference/table-functions/hdfs.md +++ b/docs/ru/sql-reference/table-functions/hdfs.md @@ -43,7 +43,7 @@ LIMIT 2 - `*` — Заменяет любое количество любых символов кроме `/`, включая отсутствие символов. - `?` — Заменяет ровно один любой символ. -- `{some_string,another_string,yet_another_one}` — Заменяет любую из строк `'some_string', 'another_string', 'yet_another_one'`. +- `{some_string,another_string,yet_another_one}` — Заменяет любую из строк `'some_string', 'another_string', 'yet_another_one'`, причём строка может содержать `/`. В случае, если в какой-либо из строк содержится `/`, то ошибки доступа (permission denied) к существующим, но недоступным директориям/файлам могут быть проигнорированы. - `{N..M}` — Заменяет любое число в интервале от `N` до `M` включительно (может содержать ведущие нули). Конструкция с `{}` аналогична табличной функции [remote](remote.md). diff --git a/src/Storages/HDFS/StorageHDFS.cpp b/src/Storages/HDFS/StorageHDFS.cpp index aa99917d533..da7b1e3fe73 100644 --- a/src/Storages/HDFS/StorageHDFS.cpp +++ b/src/Storages/HDFS/StorageHDFS.cpp @@ -92,9 +92,13 @@ namespace HDFSFileInfo ls; ls.file_info = hdfsListDirectory(fs.get(), path_for_ls.data(), &ls.length); - if (ls.file_info == nullptr && errno != ENOENT) // NOLINT + if (ls.file_info == nullptr && errno != ENOENT && errno != EACCES) // NOLINT { - // ignore file not found exception, keep throw other exception, libhdfs3 doesn't have function to get exception type, so use errno. + // ignore: + // file not found (as in LSWithRegexpMatching) + // permission denied (there is no easy way to determine + // if we really need access or just scanning all dirs while doing recursive search), + // keep throw other exception, libhdfs3 doesn't have function to get exception type, so use errno. throw Exception( ErrorCodes::ACCESS_DENIED, "Cannot list directory {}: {}", path_for_ls, String(hdfsGetLastError())); } From 086bf6cb8ce6912e3d19afa3b2cc0f1842c1af91 Mon Sep 17 00:00:00 2001 From: zvonand Date: Fri, 11 Aug 2023 15:49:21 +0300 Subject: [PATCH 052/433] added ignore_eacces_multidirectory_globs --- src/Core/Settings.h | 6 ++- src/Storages/HDFS/StorageHDFS.cpp | 38 +++++++++-------- src/Storages/StorageFile.cpp | 33 +++++++++------ ...771_complex_globs_in_storage_file_path.sql | 12 ------ ...ltidirectory_globs_storage_file.reference} | 3 ++ ...02771_multidirectory_globs_storage_file.sh | 42 +++++++++++++++++++ 6 files changed, 91 insertions(+), 43 deletions(-) delete mode 100644 tests/queries/0_stateless/02771_complex_globs_in_storage_file_path.sql rename tests/queries/0_stateless/{02771_complex_globs_in_storage_file_path.reference => 02771_multidirectory_globs_storage_file.reference} (76%) create mode 100755 tests/queries/0_stateless/02771_multidirectory_globs_storage_file.sh diff --git a/src/Core/Settings.h b/src/Core/Settings.h index 6c3d339b4be..aa2a78f00af 100644 --- a/src/Core/Settings.h +++ b/src/Core/Settings.h @@ -785,8 +785,10 @@ class IColumn; M(UInt64, extract_kvp_max_pairs_per_row, 1000, "Max number pairs that can be produced by extractKeyValuePairs function. Used to safeguard against consuming too much memory.", 0) \ M(Timezone, session_timezone, "", "This setting can be removed in the future due to potential caveats. It is experimental and is not suitable for production usage. The default timezone for current session or query. The server default timezone if empty.", 0) \ M(Bool, allow_create_index_without_type, false, "Allow CREATE INDEX query without TYPE. Query will be ignored. Made for SQL compatibility tests.", 0)\ - M(Bool, create_index_ignore_unique, false, "Ignore UNIQUE keyword in CREATE UNIQUE INDEX. Made for SQL compatibility tests.", 0)\ - // End of COMMON_SETTINGS + M(Bool, create_index_ignore_unique, false, "Ignore UNIQUE keyword in CREATE UNIQUE INDEX. Made for SQL compatibility tests.", 0) \ + M(Bool, ignore_eacces_multidirectory_globs, false, "Ignore access denied errors when processing multi-directory globs.", 0)\ + +// End of COMMON_SETTINGS // Please add settings related to formats into the FORMAT_FACTORY_SETTINGS and move obsolete settings to OBSOLETE_SETTINGS. #define MAKE_OBSOLETE(M, TYPE, NAME, DEFAULT) \ diff --git a/src/Storages/HDFS/StorageHDFS.cpp b/src/Storages/HDFS/StorageHDFS.cpp index da7b1e3fe73..1163a200bf7 100644 --- a/src/Storages/HDFS/StorageHDFS.cpp +++ b/src/Storages/HDFS/StorageHDFS.cpp @@ -67,7 +67,8 @@ namespace /// Forward-declared to use in LSWithFoldedRegexpMatching w/o circular dependency. std::vector LSWithRegexpMatching(const String & path_for_ls, const HDFSFSPtr & fs, - const String & for_match); + const String & for_match, + bool ignore_eacces_multidirectory_globs); /* * When `{...}` has any `/`s, it must be processed in a different way: @@ -83,7 +84,8 @@ namespace const String & suffix_with_globs, re2::RE2 & matcher, const size_t max_depth, - const size_t next_slash_after_glob_pos) + const size_t next_slash_after_glob_pos, + bool ignore_eacces_multidirectory_globs) { /// We don't need to go all the way in every directory if max_depth is reached /// as it is upper limit of depth by simply counting `/`s in curly braces @@ -92,15 +94,14 @@ namespace HDFSFileInfo ls; ls.file_info = hdfsListDirectory(fs.get(), path_for_ls.data(), &ls.length); - if (ls.file_info == nullptr && errno != ENOENT && errno != EACCES) // NOLINT + if (ls.file_info == nullptr && errno != ENOENT) // NOLINT { - // ignore: - // file not found (as in LSWithRegexpMatching) - // permission denied (there is no easy way to determine - // if we really need access or just scanning all dirs while doing recursive search), + // ignore file not found (as in LSWithRegexpMatching) // keep throw other exception, libhdfs3 doesn't have function to get exception type, so use errno. - throw Exception( - ErrorCodes::ACCESS_DENIED, "Cannot list directory {}: {}", path_for_ls, String(hdfsGetLastError())); + // ignore permission denied if ignore_eacces_multidirectory_globs is true + if (!(ignore_eacces_multidirectory_globs && errno == EACCES)) + throw Exception( + ErrorCodes::ACCESS_DENIED, "Cannot list directory {}: {}", path_for_ls, String(hdfsGetLastError())); } std::vector result; @@ -126,15 +127,16 @@ namespace else { std::vector result_part = LSWithRegexpMatching( - fs::path(full_path) / "" , fs, suffix_with_globs.substr(next_slash_after_glob_pos)); + fs::path(full_path) / "" , fs, suffix_with_globs.substr(next_slash_after_glob_pos), + ignore_eacces_multidirectory_globs); std::move(result_part.begin(), result_part.end(), std::back_inserter(result)); } } else if (is_directory) { std::vector result_part = LSWithFoldedRegexpMatching( - fs::path(full_path), fs, processed_suffix + dir_or_file_name, - suffix_with_globs, matcher, max_depth - 1, next_slash_after_glob_pos); + fs::path(full_path), fs, processed_suffix + dir_or_file_name, suffix_with_globs, + matcher, max_depth - 1, next_slash_after_glob_pos, ignore_eacces_multidirectory_globs); std::move(result_part.begin(), result_part.end(), std::back_inserter(result)); } } @@ -147,7 +149,8 @@ namespace std::vector LSWithRegexpMatching( const String & path_for_ls, const HDFSFSPtr & fs, - const String & for_match) + const String & for_match, + bool ignore_eacces_multidirectory_globs) { const size_t first_glob_pos = for_match.find_first_of("*?{"); const bool has_glob = first_glob_pos != std::string::npos; @@ -189,8 +192,8 @@ namespace if (slashes_in_glob) { - return LSWithFoldedRegexpMatching(fs::path(prefix_without_globs), fs, "", suffix_with_globs, - matcher, slashes_in_glob, next_slash_after_glob_pos); + return LSWithFoldedRegexpMatching(fs::path(prefix_without_globs), fs, "", suffix_with_globs, matcher, + slashes_in_glob, next_slash_after_glob_pos, ignore_eacces_multidirectory_globs); } HDFSFileInfo ls; @@ -223,7 +226,8 @@ namespace { if (re2::RE2::FullMatch(file_name, matcher)) { - std::vector result_part = LSWithRegexpMatching(fs::path(full_path) / "", fs, suffix_with_globs.substr(next_slash_after_glob_pos)); + std::vector result_part = LSWithRegexpMatching(fs::path(full_path) / "", fs, + suffix_with_globs.substr(next_slash_after_glob_pos), ignore_eacces_multidirectory_globs); /// Recursion depth is limited by pattern. '*' works only for depth = 1, for depth = 2 pattern path is '*/*'. So we do not need additional check. std::move(result_part.begin(), result_part.end(), std::back_inserter(result)); } @@ -251,7 +255,7 @@ namespace HDFSBuilderWrapper builder = createHDFSBuilder(uri_without_path + "/", context->getGlobalContext()->getConfigRef()); HDFSFSPtr fs = createHDFSFS(builder.get()); - return LSWithRegexpMatching("/", fs, path_from_uri); + return LSWithRegexpMatching("/", fs, path_from_uri, context->getSettingsRef().ignore_eacces_multidirectory_globs); } } diff --git a/src/Storages/StorageFile.cpp b/src/Storages/StorageFile.cpp index a70095cf808..3b0646039ac 100644 --- a/src/Storages/StorageFile.cpp +++ b/src/Storages/StorageFile.cpp @@ -100,6 +100,7 @@ void listFilesWithRegexpMatchingImpl( const std::string & path_for_ls, const std::string & for_match, size_t & total_bytes_to_read, + bool ignore_eacces_multidirectory_globs, std::vector & result, bool recursive = false); @@ -118,13 +119,17 @@ void listFilesWithFoldedRegexpMatchingImpl(const std::string & path_for_ls, size_t & total_bytes_to_read, const size_t max_depth, const size_t next_slash_after_glob_pos, + bool ignore_eacces_multidirectory_globs, std::vector & result) { if (!max_depth) return; const fs::directory_iterator end; - for (fs::directory_iterator it(path_for_ls, fs::directory_options::skip_permission_denied); it != end; ++it) + fs::directory_iterator it = ignore_eacces_multidirectory_globs + ? fs::directory_iterator(path_for_ls, fs::directory_options::skip_permission_denied) + : fs::directory_iterator(path_for_ls); + for (; it != end; ++it) { const std::string full_path = it->path().string(); const size_t last_slash = full_path.rfind('/'); @@ -141,14 +146,15 @@ void listFilesWithFoldedRegexpMatchingImpl(const std::string & path_for_ls, { listFilesWithRegexpMatchingImpl(fs::path(full_path) / "" , suffix_with_globs.substr(next_slash_after_glob_pos), - total_bytes_to_read, result); + total_bytes_to_read, ignore_eacces_multidirectory_globs, result); } } else if (it->is_directory()) { listFilesWithFoldedRegexpMatchingImpl(fs::path(full_path), processed_suffix + dir_or_file_name, suffix_with_globs, matcher, total_bytes_to_read, - max_depth - 1, next_slash_after_glob_pos, result); + max_depth - 1, next_slash_after_glob_pos, + ignore_eacces_multidirectory_globs, result); } } @@ -161,6 +167,7 @@ void listFilesWithRegexpMatchingImpl( const std::string & path_for_ls, const std::string & for_match, size_t & total_bytes_to_read, + bool ignore_eacces_multidirectory_globs, std::vector & result, bool recursive) { @@ -203,7 +210,7 @@ void listFilesWithRegexpMatchingImpl( re2::RE2 matcher(regexp); if (!matcher.ok()) throw Exception(ErrorCodes::CANNOT_COMPILE_REGEXP, - "Cannot compile regex from glob ({}): {}", for_match, matcher.error()); + "Cannot compile regex from gllistFilesWithFoldedRegexpMatchingImplob ({}): {}", for_match, matcher.error()); bool skip_regex = current_glob == "/*" ? true : false; if (!recursive) @@ -218,9 +225,9 @@ void listFilesWithRegexpMatchingImpl( if (slashes_in_glob) { - listFilesWithFoldedRegexpMatchingImpl(fs::path(prefix_without_globs), "", suffix_with_globs, - matcher, total_bytes_to_read, slashes_in_glob, - next_slash_after_glob_pos, result); + listFilesWithFoldedRegexpMatchingImpl(fs::path(prefix_without_globs), "", suffix_with_globs, matcher, + total_bytes_to_read, slashes_in_glob, next_slash_after_glob_pos, + ignore_eacces_multidirectory_globs, result); return; } @@ -246,11 +253,12 @@ void listFilesWithRegexpMatchingImpl( { listFilesWithRegexpMatchingImpl(fs::path(full_path).append(it->path().string()) / "", looking_for_directory ? suffix_with_globs.substr(next_slash_after_glob_pos) : current_glob, - total_bytes_to_read, result, recursive); + total_bytes_to_read, ignore_eacces_multidirectory_globs, result, recursive); } else if (looking_for_directory && re2::RE2::FullMatch(file_name, matcher)) /// Recursion depth is limited by pattern. '*' works only for depth = 1, for depth = 2 pattern path is '*/*'. So we do not need additional check. - listFilesWithRegexpMatchingImpl(fs::path(full_path) / "", suffix_with_globs.substr(next_slash_after_glob_pos), total_bytes_to_read, result); + listFilesWithRegexpMatchingImpl(fs::path(full_path) / "", suffix_with_globs.substr(next_slash_after_glob_pos), + total_bytes_to_read, ignore_eacces_multidirectory_globs, result); } } } @@ -258,10 +266,11 @@ void listFilesWithRegexpMatchingImpl( std::vector listFilesWithRegexpMatching( const std::string & path_for_ls, const std::string & for_match, - size_t & total_bytes_to_read) + size_t & total_bytes_to_read, + bool ignore_eacces_multidirectory_globs) { std::vector result; - listFilesWithRegexpMatchingImpl(path_for_ls, for_match, total_bytes_to_read, result); + listFilesWithRegexpMatchingImpl(path_for_ls, for_match, total_bytes_to_read, ignore_eacces_multidirectory_globs, result); return result; } @@ -450,7 +459,7 @@ Strings StorageFile::getPathsList(const String & table_path, const String & user else { /// We list only non-directory files. - paths = listFilesWithRegexpMatching("/", path, total_bytes_to_read); + paths = listFilesWithRegexpMatching("/", path, total_bytes_to_read, context->getSettingsRef().ignore_eacces_multidirectory_globs); can_be_directory = false; } diff --git a/tests/queries/0_stateless/02771_complex_globs_in_storage_file_path.sql b/tests/queries/0_stateless/02771_complex_globs_in_storage_file_path.sql deleted file mode 100644 index 1d125920220..00000000000 --- a/tests/queries/0_stateless/02771_complex_globs_in_storage_file_path.sql +++ /dev/null @@ -1,12 +0,0 @@ --- Tags: no-replicated-database, no-parallel - -SELECT *, _file FROM file('02771/dir{?/subdir?1/da,2/subdir2?/da}ta/non_existing.csv', CSV); -- {serverError CANNOT_EXTRACT_TABLE_STRUCTURE} - -INSERT INTO TABLE FUNCTION file('02771/dir1/subdir11/data1.csv', 'CSV', 's String') SELECT 'This is file data1' SETTINGS engine_file_truncate_on_insert=1; -INSERT INTO TABLE FUNCTION file('02771/dir2/subdir22/data2.csv', 'CSV', 's String') SELECT 'This is file data2' SETTINGS engine_file_truncate_on_insert=1; - -SELECT *, _file FROM file('02771/dir{?/subdir?1/da,2/subdir2?/da}ta1.csv', CSV); -SELECT *, _file FROM file('02771/dir{?/subdir?1/da,2/subdir2?/da}ta2.csv', CSV); - -SELECT *, _file FROM file('02771/dir?/{subdir?1/data1,subdir2?/data2}.csv', CSV) WHERE _file == 'data1.csv'; -SELECT *, _file FROM file('02771/dir?/{subdir?1/data1,subdir2?/data2}.csv', CSV) WHERE _file == 'data2.csv'; diff --git a/tests/queries/0_stateless/02771_complex_globs_in_storage_file_path.reference b/tests/queries/0_stateless/02771_multidirectory_globs_storage_file.reference similarity index 76% rename from tests/queries/0_stateless/02771_complex_globs_in_storage_file_path.reference rename to tests/queries/0_stateless/02771_multidirectory_globs_storage_file.reference index e1b420ecf37..a2fa4d7fe3d 100644 --- a/tests/queries/0_stateless/02771_complex_globs_in_storage_file_path.reference +++ b/tests/queries/0_stateless/02771_multidirectory_globs_storage_file.reference @@ -1,4 +1,7 @@ +OK This is file data1 data1.csv This is file data2 data2.csv This is file data1 data1.csv This is file data2 data2.csv +OK +This is file data1 data1.csv diff --git a/tests/queries/0_stateless/02771_multidirectory_globs_storage_file.sh b/tests/queries/0_stateless/02771_multidirectory_globs_storage_file.sh new file mode 100755 index 00000000000..f8f7e69e524 --- /dev/null +++ b/tests/queries/0_stateless/02771_multidirectory_globs_storage_file.sh @@ -0,0 +1,42 @@ +#!/usr/bin/env bash +# Tags: no-replicated-database, no-parallel + +#SELECT *, _file FROM file('02771/dir?/{subdir?1/data1,subdir2?/data2}.csv', CSV) WHERE _file == 'data1.csv'; +#SELECT *, _file FROM file('02771/dir?/{subdir?1/data1,subdir2?/data2}.csv', CSV) WHERE _file == 'data2.csv'; + +set -eu + +CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) +# shellcheck source=../shell_config.sh +. "$CURDIR"/../shell_config.sh + +# Data preparation. +# Now we can get the user_files_path by use the table file function for trick. also we can get it by query as: +# "insert into function file('exist.txt', 'CSV', 'val1 char') values ('aaaa'); select _path from file('exist.txt', 'CSV', 'val1 char')" +user_files_path=$(clickhouse-client --query "select _path,_file from file('nonexist.txt', 'CSV', 'val1 char')" 2>&1 | grep Exception | awk '{gsub("/nonexist.txt","",$9); print $9}') + +rm -rf ${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/* +mkdir -p ${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/ + +${CLICKHOUSE_CLIENT} --query "SELECT *, _file FROM file('${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/dir{?/subdir?1/da,2/subdir2?/da}ta/non_existing.csv', CSV);" 2>&1 | grep -q "CANNOT_EXTRACT_TABLE_STRUCTURE" && echo 'OK' || echo 'FAIL' + +# Create two files in different directories +mkdir -p ${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/dir1/subdir11/ +mkdir -p ${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/dir2/subdir22/ + +echo 'This is file data1' > ${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/dir1/subdir11/data1.csv +echo 'This is file data2' > ${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/dir2/subdir22/data2.csv + +${CLICKHOUSE_CLIENT} --query "SELECT *, _file FROM file('${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/dir{?/subdir?1/da,2/subdir2?/da}ta1.csv', CSV);" +${CLICKHOUSE_CLIENT} --query "SELECT *, _file FROM file('${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/dir{?/subdir?1/da,2/subdir2?/da}ta2.csv', CSV);" + +${CLICKHOUSE_CLIENT} --query "SELECT *, _file FROM file('${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/dir?/{subdir?1/data1,subdir2?/data2}.csv', CSV) WHERE _file == 'data1.csv';" +${CLICKHOUSE_CLIENT} --query "SELECT *, _file FROM file('${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/dir?/{subdir?1/data1,subdir2?/data2}.csv', CSV) WHERE _file == 'data2.csv';" + +# Add a directory to test against permission_denied +rm -rf ${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/dir_inaccessible/ && mkdir -p ${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/dir_inaccessible/ && chmod 000 ${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/dir_inaccessible/ + +${CLICKHOUSE_CLIENT} --query "SELECT *, _file FROM file('${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/dir{?/subdir?1/da,2/subdir2?/da}ta1.csv', CSV) SETTINGS ignore_eacces_multidirectory_globs = 0;" 2>&1 | grep -q "Permission denied" && echo 'OK' || echo 'FAIL' +${CLICKHOUSE_CLIENT} --query "SELECT *, _file FROM file('${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/dir{?/subdir?1/da,2/subdir2?/da}ta1.csv', CSV) SETTINGS ignore_eacces_multidirectory_globs = 1;" + +rm -rf ${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME:?} From 1983c7ce956621a88b4677a1c167d6498f518add Mon Sep 17 00:00:00 2001 From: zvonand Date: Fri, 11 Aug 2023 16:53:17 +0300 Subject: [PATCH 053/433] update docs --- docs/en/operations/settings/settings.md | 40 ++++++++++++++++++ docs/en/sql-reference/table-functions/file.md | 7 ++-- docs/en/sql-reference/table-functions/hdfs.md | 5 ++- docs/ru/operations/settings/settings.md | 41 +++++++++++++++++++ docs/ru/sql-reference/table-functions/file.md | 3 +- docs/ru/sql-reference/table-functions/hdfs.md | 6 ++- .../aspell-ignore/en/aspell-dict.txt | 2 + 7 files changed, 95 insertions(+), 9 deletions(-) diff --git a/docs/en/operations/settings/settings.md b/docs/en/operations/settings/settings.md index de3cc00e4c9..9411ec84e78 100644 --- a/docs/en/operations/settings/settings.md +++ b/docs/en/operations/settings/settings.md @@ -4614,3 +4614,43 @@ SELECT toFloat64('1.7091'), toFloat64('1.5008753E7') SETTINGS precise_float_pars │ 1.7091 │ 15008753 │ └─────────────────────┴──────────────────────────┘ ``` + +## ignore_eacces_multidirectory_globs {#ignore_eacces_multidirectory_globs} + +Allows to ignore 'permission denied' errors when using multi-directory `{}` globs for [File](../../sql-reference/table-functions/file.md#globs_in_path) and [HDFS](../../sql-reference/table-functions/hdfs.md) storages. + +Possible values: `0`, `1`. + +Default value: `0`. + +### Example + +Having the following structure in `user_files`: +``` +my_directory/ +├── data1 +│ ├── f1.csv +├── data2 +│ ├── f2.csv +└── test_root +``` +where `data1`, `data2` directories are accessible, but one has no rights to read `test_root` directories. + +For a query like `SELECT *, _path, _file FROM file('my_directory/{data1/f1,data2/f2}.csv', CSV)` an exception will be thrown: +`Code: 1001. DB::Exception: std::__1::__fs::filesystem::filesystem_error: filesystem error: in directory_iterator::directory_iterator(...): Permission denied`. +It happens because a multi-directory glob requires a recursive search in _all_ available directories under `my_directory`. + +If this setting is on, all inaccessible directories will be silently skipped, even if they are explicitly specified inside `{}`. + +```sql +SELECT _path, _file FROM file('my_directory/{data1/f1,data2/f2}.csv', CSV) SETTINGS ignore_eacces_multidirectory_globs = 0; + +Code: 1001. DB::Exception: std::__1::__fs::filesystem::filesystem_error: filesystem error: in directory_iterator::directory_iterator(...): Permission denied +``` +```sql +SELECT _path, _file FROM file('my_directory/{data1/f1,data2/f2}.csv', CSV) SETTINGS ignore_eacces_multidirectory_globs = 1; + +┌─_path───────────────────┬─_file───────┐ +│ │ │ +└─────────────────────────┴─────────────┘ +``` diff --git a/docs/en/sql-reference/table-functions/file.md b/docs/en/sql-reference/table-functions/file.md index 508e58ed3ca..e3ee9c4f999 100644 --- a/docs/en/sql-reference/table-functions/file.md +++ b/docs/en/sql-reference/table-functions/file.md @@ -135,14 +135,13 @@ Getting data from table in table.csv, located in archive1.zip or/and archive2.zi SELECT * FROM file('user_files/archives/archive{1..2}.zip :: table.csv'); ``` -## Globs in Path +## Globs in Path {#globs_in_path} Multiple path components can have globs. For being processed file must exist and match to the whole path pattern (not only suffix or prefix). - `*` — Substitutes any number of any characters except `/` including empty string. - `?` — Substitutes any single character. -- `{some_string,another_string,yet_another_one}` — Substitutes any of strings `'some_string', 'another_string', 'yet_another_one'`, including `/`. In case at least one of strings contains `/`, `'permission denied'` errors may be ignored. - +- `{some_string,another_string,yet_another_one}` — Substitutes any of strings `'some_string', 'another_string', 'yet_another_one'`. In case at least one of strings contains `/`, `'permission denied'` errors may be ignored using [ignore_eacces_multidirectory_globs](/docs/en/operations/settings/settings.md#ignore_eacces_multidirectory_globs) setting. - `{N..M}` — Substitutes any number in range from N to M including both borders. - `**` - Fetches all files inside the folder recursively. @@ -211,7 +210,7 @@ SELECT count(*) FROM file('big_dir/**/file002', 'CSV', 'name String, value UInt3 - [engine_file_allow_create_multiple_files](/docs/en/operations/settings/settings.md#engine_file_allow_create_multiple_files) - allows to create a new file on each insert if format has suffix. Disabled by default. - [engine_file_skip_empty_files](/docs/en/operations/settings/settings.md#engine_file_skip_empty_files) - allows to skip empty files while reading. Disabled by default. - [storage_file_read_method](/docs/en/operations/settings/settings.md#engine-file-emptyif-not-exists) - method of reading data from storage file, one of: read, pread, mmap (only for clickhouse-local). Default value: `pread` for clickhouse-server, `mmap` for clickhouse-local. - +- [ignore_eacces_multidirectory_globs](/docs/en/operations/settings/settings.md#ignore_eacces_multidirectory_globs) - allows to ignore permission denied errors for multi-directory globs. **See Also** diff --git a/docs/en/sql-reference/table-functions/hdfs.md b/docs/en/sql-reference/table-functions/hdfs.md index b4f63b35592..0e4ee632b71 100644 --- a/docs/en/sql-reference/table-functions/hdfs.md +++ b/docs/en/sql-reference/table-functions/hdfs.md @@ -39,13 +39,13 @@ LIMIT 2 └─────────┴─────────┴─────────┘ ``` -**Globs in path** +## Globs in path {#globs_in_path} Multiple path components can have globs. For being processed file should exists and matches to the whole path pattern (not only suffix or prefix). - `*` — Substitutes any number of any characters except `/` including empty string. - `?` — Substitutes any single character. -- `{some_string,another_string,yet_another_one}` — Substitutes any of strings `'some_string', 'another_string', 'yet_another_one'`, including `/`. In case at least one of strings contains `/`, `'permission denied'` errors may be ignored. +- `{some_string,another_string,yet_another_one}` — Substitutes any of strings `'some_string', 'another_string', 'yet_another_one'`. In case at least one of strings contains `/`, `'permission denied'` errors may be ignored using [ignore_eacces_multidirectory_globs](/docs/en/operations/settings/settings.md#ignore_eacces_multidirectory_globs) setting. - `{N..M}` — Substitutes any number in range from N to M including both borders. Constructions with `{}` are similar to the [remote table function](../../sql-reference/table-functions/remote.md)). @@ -102,6 +102,7 @@ FROM hdfs('hdfs://hdfs1:9000/big_dir/file{0..9}{0..9}{0..9}', 'CSV', 'name Strin - [hdfs_truncate_on_insert](/docs/en/operations/settings/settings.md#hdfs-truncate-on-insert) - allows to truncate file before insert into it. Disabled by default. - [hdfs_create_multiple_files](/docs/en/operations/settings/settings.md#hdfs_allow_create_multiple_files) - allows to create a new file on each insert if format has suffix. Disabled by default. - [hdfs_skip_empty_files](/docs/en/operations/settings/settings.md#hdfs_skip_empty_files) - allows to skip empty files while reading. Disabled by default. +- [ignore_eacces_multidirectory_globs](/docs/en/operations/settings/settings.md#ignore_eacces_multidirectory_globs) - allows to ignore permission denied errors for multi-directory globs. **See Also** diff --git a/docs/ru/operations/settings/settings.md b/docs/ru/operations/settings/settings.md index d3db890ad7a..a230efcd8c6 100644 --- a/docs/ru/operations/settings/settings.md +++ b/docs/ru/operations/settings/settings.md @@ -4206,3 +4206,44 @@ SELECT toFloat64('1.7091'), toFloat64('1.5008753E7') SETTINGS precise_float_pars │ 1.7091 │ 15008753 │ └─────────────────────┴──────────────────────────┘ ``` + +## ignore_eacces_multidirectory_globs {#ignore_eacces_multidirectory_globs} + +Позволяет игнорировать ошибку 'permission denied', возникающую при использовании шаблона `{}`, содержащего `/` внутри себя. +Работает для [File](../../sql-reference/table-functions/file.md#globs_in_path) и [HDFS](../../sql-reference/table-functions/hdfs.md). + +Возможные значения: `0`, `1`. + +Значение по умолчанию: `0`. + +### Пример + +Пусть в `user_files` имеется следующая структура: +``` +my_directory/ +├── data1 +│ ├── f1.csv +├── data2 +│ ├── f2.csv +└── test_root +``` +Пусть также директории `data1`, `data2` могут быть прочитаны, но прав на чтение `test_root` нет. + +На запрос `SELECT *, _path, _file FROM file('my_directory/{data1/f1,data2/f2}.csv', CSV)` будет выброшено исключение: +`Code: 1001. DB::Exception: std::__1::__fs::filesystem::filesystem_error: filesystem error: in directory_iterator::directory_iterator(...): Permission denied`. +Это происходит, так как для обработки такого шаблона необходимо выполнить рекурсивный поиск по _всем_ директориям, находящимся внутри `my_directory`. + +Если данная настройка имеет значение 1, то недоступные директории будут тихо пропущены, даже если они явно указаны внутри `{}`. + +```sql +SELECT _path, _file FROM file('my_directory/{data1/f1,data2/f2}.csv', CSV) SETTINGS ignore_eacces_multidirectory_globs = 0; + +Code: 1001. DB::Exception: std::__1::__fs::filesystem::filesystem_error: filesystem error: in directory_iterator::directory_iterator(...): Permission denied +``` +```sql +SELECT _path, _file FROM file('my_directory/{data1/f1,data2/f2}.csv', CSV) SETTINGS ignore_eacces_multidirectory_globs = 1; + +┌─_path───────────────────┬─_file───────┐ +│ │ │ +└─────────────────────────┴─────────────┘ +``` diff --git a/docs/ru/sql-reference/table-functions/file.md b/docs/ru/sql-reference/table-functions/file.md index d7107ba5950..32d98bd19ee 100644 --- a/docs/ru/sql-reference/table-functions/file.md +++ b/docs/ru/sql-reference/table-functions/file.md @@ -79,7 +79,7 @@ SELECT * FROM file('test.csv', 'CSV', 'column1 UInt32, column2 UInt32, column3 U - `*` — заменяет любое количество любых символов кроме `/`, включая отсутствие символов. - `?` — заменяет ровно один любой символ. -- `{some_string,another_string,yet_another_one}` — заменяет любую из строк `'some_string', 'another_string', 'yet_another_one'`, причём строка может содержать `/`. В случае, если в какой-либо из строк содержится `/`, то ошибки доступа (permission denied) к существующим, но недоступным директориям/файлам могут быть проигнорированы. +- `{some_string,another_string,yet_another_one}` — заменяет любую из строк `'some_string', 'another_string', 'yet_another_one'`. В случае, если в какой-либо из строк содержится `/`, то ошибки доступа (permission denied) к существующим, но недоступным директориям/файлам могут быть проигнорированы при помощи настройки [ignore_eacces_multidirectory_globs](/docs/ru/operations/settings/settings.md#ignore_eacces_multidirectory_globs). - `{N..M}` — заменяет любое число в интервале от `N` до `M` включительно (может содержать ведущие нули). Конструкция с `{}` аналогична табличной функции [remote](remote.md). @@ -123,6 +123,7 @@ SELECT count(*) FROM file('big_dir/file{0..9}{0..9}{0..9}', 'CSV', 'name String, - `_path` — путь к файлу. - `_file` — имя файла. + **Смотрите также** - [Виртуальные столбцы](index.md#table_engines-virtual_columns) diff --git a/docs/ru/sql-reference/table-functions/hdfs.md b/docs/ru/sql-reference/table-functions/hdfs.md index 350d483b592..d74248bd93c 100644 --- a/docs/ru/sql-reference/table-functions/hdfs.md +++ b/docs/ru/sql-reference/table-functions/hdfs.md @@ -39,11 +39,11 @@ LIMIT 2 └─────────┴─────────┴─────────┘ ``` -**Шаблоны в пути** +## Шаблоны поиска в компонентах пути {#globs-in-path} - `*` — Заменяет любое количество любых символов кроме `/`, включая отсутствие символов. - `?` — Заменяет ровно один любой символ. -- `{some_string,another_string,yet_another_one}` — Заменяет любую из строк `'some_string', 'another_string', 'yet_another_one'`, причём строка может содержать `/`. В случае, если в какой-либо из строк содержится `/`, то ошибки доступа (permission denied) к существующим, но недоступным директориям/файлам могут быть проигнорированы. +- `{some_string,another_string,yet_another_one}` — Заменяет любую из строк `'some_string', 'another_string', 'yet_another_one'`. В случае, если в какой-либо из строк содержится `/`, то ошибки доступа (permission denied) к существующим, но недоступным директориям/файлам могут быть проигнорированы при помощи настройки [ignore_eacces_multidirectory_globs](/docs/ru/operations/settings/settings.md#ignore_eacces_multidirectory_globs). - `{N..M}` — Заменяет любое число в интервале от `N` до `M` включительно (может содержать ведущие нули). Конструкция с `{}` аналогична табличной функции [remote](remote.md). @@ -61,3 +61,5 @@ LIMIT 2 **Смотрите также** - [Виртуальные столбцы](index.md#table_engines-virtual_columns) +- Параметр [ignore_eacces_multidirectory_globs](/docs/ru/operations/settings/settings.md#ignore_eacces_multidirectory_globs) + diff --git a/utils/check-style/aspell-ignore/en/aspell-dict.txt b/utils/check-style/aspell-ignore/en/aspell-dict.txt index aefb14d098f..a65c294b1d8 100644 --- a/utils/check-style/aspell-ignore/en/aspell-dict.txt +++ b/utils/check-style/aspell-ignore/en/aspell-dict.txt @@ -1385,6 +1385,7 @@ dragonbox dropoff dumpColumnStructure durations +eacces ecto embeddings emptyArray @@ -1793,6 +1794,7 @@ monthName moscow msgpack msgpk +multidirectory multiFuzzyMatchAllIndices multiFuzzyMatchAny multiFuzzyMatchAnyIndex From e28e4be0feec36a16702e9c2361eab89bb7311d2 Mon Sep 17 00:00:00 2001 From: zvonand Date: Fri, 11 Aug 2023 18:21:03 +0300 Subject: [PATCH 054/433] update tests according to style --- .../02771_multidirectory_globs_storage_file.reference | 1 - .../0_stateless/02771_multidirectory_globs_storage_file.sh | 7 +++---- 2 files changed, 3 insertions(+), 5 deletions(-) diff --git a/tests/queries/0_stateless/02771_multidirectory_globs_storage_file.reference b/tests/queries/0_stateless/02771_multidirectory_globs_storage_file.reference index a2fa4d7fe3d..b0746e0a1be 100644 --- a/tests/queries/0_stateless/02771_multidirectory_globs_storage_file.reference +++ b/tests/queries/0_stateless/02771_multidirectory_globs_storage_file.reference @@ -3,5 +3,4 @@ This is file data1 data1.csv This is file data2 data2.csv This is file data1 data1.csv This is file data2 data2.csv -OK This is file data1 data1.csv diff --git a/tests/queries/0_stateless/02771_multidirectory_globs_storage_file.sh b/tests/queries/0_stateless/02771_multidirectory_globs_storage_file.sh index f8f7e69e524..4901e716be3 100755 --- a/tests/queries/0_stateless/02771_multidirectory_globs_storage_file.sh +++ b/tests/queries/0_stateless/02771_multidirectory_globs_storage_file.sh @@ -15,7 +15,7 @@ CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) # "insert into function file('exist.txt', 'CSV', 'val1 char') values ('aaaa'); select _path from file('exist.txt', 'CSV', 'val1 char')" user_files_path=$(clickhouse-client --query "select _path,_file from file('nonexist.txt', 'CSV', 'val1 char')" 2>&1 | grep Exception | awk '{gsub("/nonexist.txt","",$9); print $9}') -rm -rf ${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/* +rm -rf ${user_files_path:?}/${CLICKHOUSE_TEST_UNIQUE_NAME:?}/* mkdir -p ${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/ ${CLICKHOUSE_CLIENT} --query "SELECT *, _file FROM file('${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/dir{?/subdir?1/da,2/subdir2?/da}ta/non_existing.csv', CSV);" 2>&1 | grep -q "CANNOT_EXTRACT_TABLE_STRUCTURE" && echo 'OK' || echo 'FAIL' @@ -34,9 +34,8 @@ ${CLICKHOUSE_CLIENT} --query "SELECT *, _file FROM file('${user_files_path}/${CL ${CLICKHOUSE_CLIENT} --query "SELECT *, _file FROM file('${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/dir?/{subdir?1/data1,subdir2?/data2}.csv', CSV) WHERE _file == 'data2.csv';" # Add a directory to test against permission_denied -rm -rf ${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/dir_inaccessible/ && mkdir -p ${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/dir_inaccessible/ && chmod 000 ${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/dir_inaccessible/ +rm -rf ${user_files_path:?}/${CLICKHOUSE_TEST_UNIQUE_NAME:?}/dir_inaccessible/ && mkdir -p ${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/dir_inaccessible/ && chmod 000 ${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/dir_inaccessible/ -${CLICKHOUSE_CLIENT} --query "SELECT *, _file FROM file('${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/dir{?/subdir?1/da,2/subdir2?/da}ta1.csv', CSV) SETTINGS ignore_eacces_multidirectory_globs = 0;" 2>&1 | grep -q "Permission denied" && echo 'OK' || echo 'FAIL' ${CLICKHOUSE_CLIENT} --query "SELECT *, _file FROM file('${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/dir{?/subdir?1/da,2/subdir2?/da}ta1.csv', CSV) SETTINGS ignore_eacces_multidirectory_globs = 1;" -rm -rf ${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME:?} +rm -rf ${user_files_path:?}/${CLICKHOUSE_TEST_UNIQUE_NAME:?} From 74485f3e52c96702a8c85ec98a5b9999bf7db764 Mon Sep 17 00:00:00 2001 From: chen768959 <934103231@qq.com> Date: Tue, 15 Aug 2023 03:25:32 +0800 Subject: [PATCH 055/433] Addressed review feedback --- src/DataTypes/DataTypesNumber.h | 27 +---------------- src/DataTypes/FieldToDataType.cpp | 5 ++-- src/DataTypes/IDataType.cpp | 6 ---- src/DataTypes/IDataType.h | 9 ------ src/DataTypes/getLeastSupertype.cpp | 45 +++++++++-------------------- src/DataTypes/getLeastSupertype.h | 2 +- 6 files changed, 19 insertions(+), 75 deletions(-) diff --git a/src/DataTypes/DataTypesNumber.h b/src/DataTypes/DataTypesNumber.h index 6be07db1062..b48c5ff1f5d 100644 --- a/src/DataTypes/DataTypesNumber.h +++ b/src/DataTypes/DataTypesNumber.h @@ -1,7 +1,6 @@ #pragma once #include -#include #include #include #include @@ -10,21 +9,10 @@ namespace DB { -using DataTypes = std::vector; - template -class DataTypeNumber final : public DataTypeNumberBase +class DataTypeNumber : public DataTypeNumberBase { public: - DataTypeNumber() = default; - - explicit DataTypeNumber(DataTypePtr opposite_sign_data_type_) - : DataTypeNumberBase() - , opposite_sign_data_type(std::move(opposite_sign_data_type_)) - , has_opposite_sign_data_type(true) - { - } - bool equals(const IDataType & rhs) const override { return typeid(rhs) == typeid(*this); } bool canBeUsedAsVersion() const override { return true; } @@ -40,23 +28,10 @@ public: return std::make_shared(); } - bool hasOppositeSignDataType() const override { return has_opposite_sign_data_type; } - DataTypePtr oppositeSignDataType() const override - { - if (!has_opposite_sign_data_type) - IDataType::oppositeSignDataType(); - - return opposite_sign_data_type; - } - SerializationPtr doGetDefaultSerialization() const override { return std::make_shared>(); } - -private: - DataTypePtr opposite_sign_data_type; - bool has_opposite_sign_data_type = false; }; using DataTypeUInt8 = DataTypeNumber; diff --git a/src/DataTypes/FieldToDataType.cpp b/src/DataTypes/FieldToDataType.cpp index 2f759e79661..52a46bf55bc 100644 --- a/src/DataTypes/FieldToDataType.cpp +++ b/src/DataTypes/FieldToDataType.cpp @@ -3,6 +3,7 @@ #include #include #include +#include #include #include #include @@ -36,7 +37,7 @@ DataTypePtr FieldToDataType::operator() (const UInt64 & x) const if (x <= std::numeric_limits::max()) return std::make_shared(); if (x <= std::numeric_limits::max()) return std::make_shared(); if (x <= std::numeric_limits::max()) return std::make_shared(); - if (x <= std::numeric_limits::max()) return std::make_shared(std::make_shared()); + if (x <= std::numeric_limits::max()) return std::make_shared(std::make_shared()); return std::make_shared(); } @@ -140,7 +141,7 @@ DataTypePtr FieldToDataType::operator() (const Array & x) const for (const Field & elem : x) element_types.emplace_back(applyVisitor(*this, elem)); - return std::make_shared(getLeastSupertype(element_types, true)); + return std::make_shared(getLeastSupertype(element_types)); } template diff --git a/src/DataTypes/IDataType.cpp b/src/DataTypes/IDataType.cpp index 092b9ed8e0e..4ffe82039b2 100644 --- a/src/DataTypes/IDataType.cpp +++ b/src/DataTypes/IDataType.cpp @@ -71,12 +71,6 @@ DataTypePtr IDataType::promoteNumericType() const throw Exception(ErrorCodes::DATA_TYPE_CANNOT_BE_PROMOTED, "Data type {} can't be promoted.", getName()); } -DataTypePtr IDataType::oppositeSignDataType() const -{ - // TODO: Should use ErrorCodes::OPPOSITE_SIGN_DATA_TYPE_NOT_FOUND. - throw Exception(702, "Opposite sign data type not found for {}.", getName()); -} - size_t IDataType::getSizeOfValueInMemory() const { throw Exception(ErrorCodes::LOGICAL_ERROR, "Value of type {} in memory is not of fixed size.", getName()); diff --git a/src/DataTypes/IDataType.h b/src/DataTypes/IDataType.h index d49c163072e..54cb3d0d5c2 100644 --- a/src/DataTypes/IDataType.h +++ b/src/DataTypes/IDataType.h @@ -159,15 +159,6 @@ public: */ virtual DataTypePtr promoteNumericType() const; - /** The data type has an opposite sign DataTypePtr type. - * Data types that can have an opposite sign are typically signed or unsigned types. - */ - virtual bool hasOppositeSignDataType() const { return false; } - - /** Return the opposite sign data type of the current data type. Throw an exception if `hasOppositeSignDataType() == false`. - */ - virtual DataTypePtr oppositeSignDataType() const; - /** Directly insert default value into a column. Default implementation use method IColumn::insertDefault. * This should be overridden if data type default value differs from column default value (example: Enum data types). */ diff --git a/src/DataTypes/getLeastSupertype.cpp b/src/DataTypes/getLeastSupertype.cpp index 4c399e99caa..30bbf8e2786 100644 --- a/src/DataTypes/getLeastSupertype.cpp +++ b/src/DataTypes/getLeastSupertype.cpp @@ -16,6 +16,7 @@ #include #include #include +#include #include #include @@ -201,7 +202,7 @@ DataTypePtr getNumericType(const TypeIndexSet & types) } template -DataTypePtr getLeastSupertype(const DataTypes & types, bool optimize_type_ids) +DataTypePtr getLeastSupertype(const DataTypes & types) { /// Trivial cases @@ -592,8 +593,7 @@ DataTypePtr getLeastSupertype(const DataTypes & types, bool optimize_type_ids) /// For numeric types, the most complicated part. { - if (optimize_type_ids) - optimizeTypeIds(types, type_ids); + optimizeTypeIds(types, type_ids); auto numeric_type = getNumericType(type_ids); if (numeric_type) return numeric_type; @@ -603,49 +603,32 @@ DataTypePtr getLeastSupertype(const DataTypes & types, bool optimize_type_ids) return throwOrReturn(types, "", ErrorCodes::NO_COMMON_TYPE); } +// Convert the UInt64 type to Int64 in order to cover other signed_integer types and obtain the least super type of all ints. +// Example, UInt64(both Int64), Int8 = Int64, Int8. void optimizeTypeIds(const DataTypes & types, TypeIndexSet & type_ids) { - auto is_signed_int = [](const TypeIndex & type_id) - { - switch (type_id) - { - case TypeIndex::Int8: - case TypeIndex::Int16: - case TypeIndex::Int32: - case TypeIndex::Int64: - return true; - default: - return false; - } - }; + if ((!type_ids.contains(TypeIndex::Int8) && !type_ids.contains(TypeIndex::Int16) && !type_ids.contains(TypeIndex::Int32) && !type_ids.contains(TypeIndex::Int64)) || !type_ids.contains(TypeIndex::UInt64)) + return; - bool has_signed_int = false; - bool has_uint64_and_has_opposite = false; + bool has_opposite = false; TypeIndexSet opposite_type_ids; - // Determine whether UInt64 in type_ids needs to change its sign. for (const auto & type : types) { auto type_id = type->getTypeId(); - if (!has_signed_int) - has_signed_int = is_signed_int(type_id); - if (type_id == TypeIndex::UInt64) { - if (!type->hasOppositeSignDataType()) + if (const auto * uint64_with_opposite = typeid_cast(type.get())) { - has_uint64_and_has_opposite = false; - break ; + has_opposite = true; + opposite_type_ids.insert(uint64_with_opposite->getOppositeSignDataType()->getTypeId()); }else - { - has_uint64_and_has_opposite = true; - opposite_type_ids.insert(type->oppositeSignDataType()->getTypeId()); - } + return; } } - if (has_uint64_and_has_opposite && has_signed_int) + if (has_opposite) { type_ids.erase(TypeIndex::UInt64); type_ids.insert(opposite_type_ids.begin(), opposite_type_ids.end()); @@ -713,7 +696,7 @@ DataTypePtr tryGetLeastSupertype(const TypeIndexSet & types) return getLeastSupertype(types); } -template DataTypePtr getLeastSupertype(const DataTypes & types, bool optimize_type_ids); +template DataTypePtr getLeastSupertype(const DataTypes & types); template DataTypePtr getLeastSupertype(const TypeIndexSet & types); } diff --git a/src/DataTypes/getLeastSupertype.h b/src/DataTypes/getLeastSupertype.h index ba43b3966f9..0a3fa3c2536 100644 --- a/src/DataTypes/getLeastSupertype.h +++ b/src/DataTypes/getLeastSupertype.h @@ -18,7 +18,7 @@ enum class LeastSupertypeOnError * Examples: there is no least common supertype for Array(UInt8), Int8. */ template -DataTypePtr getLeastSupertype(const DataTypes & types, bool optimize_type_ids = false); +DataTypePtr getLeastSupertype(const DataTypes & types); /// Same as above but return String type instead of throwing exception. /// All types can be casted to String, because they can be serialized to String. From 34b1811a53f0299c6a2a2845137a7558000d0c9d Mon Sep 17 00:00:00 2001 From: Andrey Zvonov <32552679+zvonand@users.noreply.github.com> Date: Tue, 15 Aug 2023 19:10:14 +0300 Subject: [PATCH 056/433] Update src/Storages/StorageFile.cpp seems like typo Co-authored-by: SmitaRKulkarni <64093672+SmitaRKulkarni@users.noreply.github.com> --- src/Storages/StorageFile.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/Storages/StorageFile.cpp b/src/Storages/StorageFile.cpp index 3b0646039ac..ddb6b8a3844 100644 --- a/src/Storages/StorageFile.cpp +++ b/src/Storages/StorageFile.cpp @@ -210,7 +210,7 @@ void listFilesWithRegexpMatchingImpl( re2::RE2 matcher(regexp); if (!matcher.ok()) throw Exception(ErrorCodes::CANNOT_COMPILE_REGEXP, - "Cannot compile regex from gllistFilesWithFoldedRegexpMatchingImplob ({}): {}", for_match, matcher.error()); + "Cannot compile regex from glob ({}): {}", for_match, matcher.error()); bool skip_regex = current_glob == "/*" ? true : false; if (!recursive) From d161a87a34123591c0ac7412f3189d1e60be318d Mon Sep 17 00:00:00 2001 From: Andrey Zvonov <32552679+zvonand@users.noreply.github.com> Date: Tue, 15 Aug 2023 19:10:40 +0300 Subject: [PATCH 057/433] Update docs/en/sql-reference/table-functions/file.md Co-authored-by: SmitaRKulkarni <64093672+SmitaRKulkarni@users.noreply.github.com> --- docs/en/sql-reference/table-functions/file.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/docs/en/sql-reference/table-functions/file.md b/docs/en/sql-reference/table-functions/file.md index e3ee9c4f999..f12ac732c22 100644 --- a/docs/en/sql-reference/table-functions/file.md +++ b/docs/en/sql-reference/table-functions/file.md @@ -141,7 +141,7 @@ Multiple path components can have globs. For being processed file must exist and - `*` — Substitutes any number of any characters except `/` including empty string. - `?` — Substitutes any single character. -- `{some_string,another_string,yet_another_one}` — Substitutes any of strings `'some_string', 'another_string', 'yet_another_one'`. In case at least one of strings contains `/`, `'permission denied'` errors may be ignored using [ignore_eacces_multidirectory_globs](/docs/en/operations/settings/settings.md#ignore_eacces_multidirectory_globs) setting. +- `{some_string,another_string,yet_another_one}` — Substitutes any of strings `'some_string', 'another_string', 'yet_another_one'`. In case at least one of strings contains `/`, `'permission denied'` errors may be ignored using [ignore_eacces_multidirectory_globs](/docs/en/operations/settings/settings.md#ignore_eacces_multidirectory_globs) setting for file & HDFS. - `{N..M}` — Substitutes any number in range from N to M including both borders. - `**` - Fetches all files inside the folder recursively. From 72574f8376529be17c69dfd36f9775054edd3ada Mon Sep 17 00:00:00 2001 From: Andrey Zvonov <32552679+zvonand@users.noreply.github.com> Date: Tue, 15 Aug 2023 19:10:49 +0300 Subject: [PATCH 058/433] Update src/Core/Settings.h Co-authored-by: SmitaRKulkarni <64093672+SmitaRKulkarni@users.noreply.github.com> --- src/Core/Settings.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/Core/Settings.h b/src/Core/Settings.h index 84e4053ca80..cec400a2ada 100644 --- a/src/Core/Settings.h +++ b/src/Core/Settings.h @@ -786,7 +786,7 @@ class IColumn; M(Timezone, session_timezone, "", "This setting can be removed in the future due to potential caveats. It is experimental and is not suitable for production usage. The default timezone for current session or query. The server default timezone if empty.", 0) \ M(Bool, allow_create_index_without_type, false, "Allow CREATE INDEX query without TYPE. Query will be ignored. Made for SQL compatibility tests.", 0)\ M(Bool, create_index_ignore_unique, false, "Ignore UNIQUE keyword in CREATE UNIQUE INDEX. Made for SQL compatibility tests.", 0) \ - M(Bool, ignore_eacces_multidirectory_globs, false, "Ignore access denied errors when processing multi-directory globs.", 0)\ + M(Bool, ignore_eacces_multidirectory_globs, false, "Ignore access denied errors when processing multi-directory globs for file & HDFS.", 0)\ // End of COMMON_SETTINGS // Please add settings related to formats into the FORMAT_FACTORY_SETTINGS and move obsolete settings to OBSOLETE_SETTINGS. From 26833856ca6b86cab6f350d71f369eb5e6c746c0 Mon Sep 17 00:00:00 2001 From: chen768959 <934103231@qq.com> Date: Wed, 16 Aug 2023 02:38:25 +0800 Subject: [PATCH 059/433] Perform optimization of LeastSupertype within FieldToDataType, eliminating modifications to DataType. --- src/DataTypes/DataTypesNumber.h | 2 +- src/DataTypes/FieldToDataType.cpp | 74 +++++++++++++++++++++++++++-- src/DataTypes/getLeastSupertype.cpp | 34 ------------- src/DataTypes/getLeastSupertype.h | 2 - 4 files changed, 70 insertions(+), 42 deletions(-) diff --git a/src/DataTypes/DataTypesNumber.h b/src/DataTypes/DataTypesNumber.h index b48c5ff1f5d..5843086248c 100644 --- a/src/DataTypes/DataTypesNumber.h +++ b/src/DataTypes/DataTypesNumber.h @@ -10,7 +10,7 @@ namespace DB { template -class DataTypeNumber : public DataTypeNumberBase +class DataTypeNumber final : public DataTypeNumberBase { public: bool equals(const IDataType & rhs) const override { return typeid(rhs) == typeid(*this); } diff --git a/src/DataTypes/FieldToDataType.cpp b/src/DataTypes/FieldToDataType.cpp index 52a46bf55bc..3a99552089c 100644 --- a/src/DataTypes/FieldToDataType.cpp +++ b/src/DataTypes/FieldToDataType.cpp @@ -3,7 +3,6 @@ #include #include #include -#include #include #include #include @@ -37,7 +36,6 @@ DataTypePtr FieldToDataType::operator() (const UInt64 & x) const if (x <= std::numeric_limits::max()) return std::make_shared(); if (x <= std::numeric_limits::max()) return std::make_shared(); if (x <= std::numeric_limits::max()) return std::make_shared(); - if (x <= std::numeric_limits::max()) return std::make_shared(std::make_shared()); return std::make_shared(); } @@ -138,8 +136,36 @@ DataTypePtr FieldToDataType::operator() (const Array & x) const DataTypes element_types; element_types.reserve(x.size()); + auto checkIfConversionSigned = [&](bool& has_signed_int, bool& has_uint64, bool& uint64_could_opposite, + const DataTypePtr & type, const Field & elem) { + if (uint64_could_opposite) { + has_signed_int |= WhichDataType(type).isNativeInt(); + + if (type->getTypeId() == TypeIndex::UInt64) { + has_uint64 = true; + uint64_could_opposite &= (elem.template get() <= std::numeric_limits::max()); + } + } + }; + + bool has_signed_int = false; + bool has_uint64 = false; + bool uint64_could_opposite = true; for (const Field & elem : x) - element_types.emplace_back(applyVisitor(*this, elem)); + { + DataTypePtr type = applyVisitor(*this, elem); + element_types.emplace_back(type); + checkIfConversionSigned(has_signed_int, has_uint64, uint64_could_opposite, type, elem); + } + + // Convert the UInt64 type to Int64 in order to cover other signed_integer types + // and obtain the least super type of all ints. + if (has_signed_int && has_uint64 && uint64_could_opposite) + { + for (auto & type : element_types) + if (type->getTypeId() == TypeIndex::UInt64) + type = std::make_shared(); + } return std::make_shared(getLeastSupertype(element_types)); } @@ -167,14 +193,52 @@ DataTypePtr FieldToDataType::operator() (const Map & map) const key_types.reserve(map.size()); value_types.reserve(map.size()); + auto checkIfConversionSigned = [&](bool& has_signed_int, bool& has_uint64, bool& uint64_could_opposite, + const DataTypePtr & type, const Field & elem) { + if (uint64_could_opposite) { + has_signed_int |= WhichDataType(type).isNativeInt(); + + if (type->getTypeId() == TypeIndex::UInt64) { + has_uint64 = true; + uint64_could_opposite &= (elem.template get() <= std::numeric_limits::max()); + } + } + }; + + auto updateUInt64Types = [](DataTypes types) { + for (auto& type : types) { + if (type->getTypeId() == TypeIndex::UInt64) { + type = std::make_shared(); + } + } + }; + + bool k_has_signed_int = false; + bool k_has_uint64 = false; + bool k_uint64_could_opposite = true; + bool v_has_signed_int = false; + bool v_has_uint64 = false; + bool v_uint64_could_opposite = true; for (const auto & elem : map) { const auto & tuple = elem.safeGet(); assert(tuple.size() == 2); - key_types.push_back(applyVisitor(*this, tuple[0])); - value_types.push_back(applyVisitor(*this, tuple[1])); + DataTypePtr k_type = applyVisitor(*this, tuple[0]); + key_types.push_back(k_type); + checkIfConversionSigned(k_has_signed_int, k_has_uint64, k_uint64_could_opposite, k_type, tuple[0]); + DataTypePtr v_type = applyVisitor(*this, tuple[1]); + value_types.push_back(v_type); + checkIfConversionSigned(v_has_signed_int, v_has_uint64, v_uint64_could_opposite, v_type, tuple[1]); } + // Convert the UInt64 type to Int64 in order to cover other signed_integer types + // and obtain the least super type of all ints. + if (k_has_signed_int && k_has_uint64 && k_uint64_could_opposite) + updateUInt64Types(key_types); + + if (v_has_signed_int && v_has_uint64 && v_uint64_could_opposite) + updateUInt64Types(value_types); + return std::make_shared( getLeastSupertype(key_types), getLeastSupertype(value_types)); diff --git a/src/DataTypes/getLeastSupertype.cpp b/src/DataTypes/getLeastSupertype.cpp index 30bbf8e2786..9d42d82ce91 100644 --- a/src/DataTypes/getLeastSupertype.cpp +++ b/src/DataTypes/getLeastSupertype.cpp @@ -16,7 +16,6 @@ #include #include #include -#include #include #include @@ -593,7 +592,6 @@ DataTypePtr getLeastSupertype(const DataTypes & types) /// For numeric types, the most complicated part. { - optimizeTypeIds(types, type_ids); auto numeric_type = getNumericType(type_ids); if (numeric_type) return numeric_type; @@ -603,38 +601,6 @@ DataTypePtr getLeastSupertype(const DataTypes & types) return throwOrReturn(types, "", ErrorCodes::NO_COMMON_TYPE); } -// Convert the UInt64 type to Int64 in order to cover other signed_integer types and obtain the least super type of all ints. -// Example, UInt64(both Int64), Int8 = Int64, Int8. -void optimizeTypeIds(const DataTypes & types, TypeIndexSet & type_ids) -{ - if ((!type_ids.contains(TypeIndex::Int8) && !type_ids.contains(TypeIndex::Int16) && !type_ids.contains(TypeIndex::Int32) && !type_ids.contains(TypeIndex::Int64)) || !type_ids.contains(TypeIndex::UInt64)) - return; - - bool has_opposite = false; - TypeIndexSet opposite_type_ids; - - for (const auto & type : types) - { - auto type_id = type->getTypeId(); - - if (type_id == TypeIndex::UInt64) - { - if (const auto * uint64_with_opposite = typeid_cast(type.get())) - { - has_opposite = true; - opposite_type_ids.insert(uint64_with_opposite->getOppositeSignDataType()->getTypeId()); - }else - return; - } - } - - if (has_opposite) - { - type_ids.erase(TypeIndex::UInt64); - type_ids.insert(opposite_type_ids.begin(), opposite_type_ids.end()); - } -} - DataTypePtr getLeastSupertypeOrString(const DataTypes & types) { return getLeastSupertype(types); diff --git a/src/DataTypes/getLeastSupertype.h b/src/DataTypes/getLeastSupertype.h index 0a3fa3c2536..2ef4a0e6850 100644 --- a/src/DataTypes/getLeastSupertype.h +++ b/src/DataTypes/getLeastSupertype.h @@ -29,8 +29,6 @@ DataTypePtr tryGetLeastSupertype(const DataTypes & types); using TypeIndexSet = std::unordered_set; -void optimizeTypeIds(const DataTypes & types, TypeIndexSet & type_ids); - template DataTypePtr getLeastSupertype(const TypeIndexSet & types); From a8cdc9472072e08e76e73ce91597634f0d6ec319 Mon Sep 17 00:00:00 2001 From: zvonand Date: Tue, 15 Aug 2023 23:42:04 +0500 Subject: [PATCH 060/433] update docs acc to review --- docs/en/operations/settings/settings.md | 1 + docs/ru/operations/settings/settings.md | 1 + 2 files changed, 2 insertions(+) diff --git a/docs/en/operations/settings/settings.md b/docs/en/operations/settings/settings.md index 9411ec84e78..4f284aa3354 100644 --- a/docs/en/operations/settings/settings.md +++ b/docs/en/operations/settings/settings.md @@ -4618,6 +4618,7 @@ SELECT toFloat64('1.7091'), toFloat64('1.5008753E7') SETTINGS precise_float_pars ## ignore_eacces_multidirectory_globs {#ignore_eacces_multidirectory_globs} Allows to ignore 'permission denied' errors when using multi-directory `{}` globs for [File](../../sql-reference/table-functions/file.md#globs_in_path) and [HDFS](../../sql-reference/table-functions/hdfs.md) storages. +This setting is only applicable when multi directory `{}` glob is used. Possible values: `0`, `1`. diff --git a/docs/ru/operations/settings/settings.md b/docs/ru/operations/settings/settings.md index a230efcd8c6..3f0bc4ee2ec 100644 --- a/docs/ru/operations/settings/settings.md +++ b/docs/ru/operations/settings/settings.md @@ -4211,6 +4211,7 @@ SELECT toFloat64('1.7091'), toFloat64('1.5008753E7') SETTINGS precise_float_pars Позволяет игнорировать ошибку 'permission denied', возникающую при использовании шаблона `{}`, содержащего `/` внутри себя. Работает для [File](../../sql-reference/table-functions/file.md#globs_in_path) и [HDFS](../../sql-reference/table-functions/hdfs.md). +Работает _только_ для указанных выше шаблонов `{}`. Возможные значения: `0`, `1`. From 3eed7dc3cc59437c5dfb18cf6dfedafb209c6e34 Mon Sep 17 00:00:00 2001 From: chen768959 <934103231@qq.com> Date: Wed, 16 Aug 2023 09:26:39 +0800 Subject: [PATCH 061/433] fix style --- src/DataTypes/FieldToDataType.cpp | 27 ++++++++++++++++----------- 1 file changed, 16 insertions(+), 11 deletions(-) diff --git a/src/DataTypes/FieldToDataType.cpp b/src/DataTypes/FieldToDataType.cpp index 3a99552089c..e25a2f546f2 100644 --- a/src/DataTypes/FieldToDataType.cpp +++ b/src/DataTypes/FieldToDataType.cpp @@ -137,11 +137,14 @@ DataTypePtr FieldToDataType::operator() (const Array & x) const element_types.reserve(x.size()); auto checkIfConversionSigned = [&](bool& has_signed_int, bool& has_uint64, bool& uint64_could_opposite, - const DataTypePtr & type, const Field & elem) { - if (uint64_could_opposite) { + const DataTypePtr & type, const Field & elem) + { + if (uint64_could_opposite) + { has_signed_int |= WhichDataType(type).isNativeInt(); - if (type->getTypeId() == TypeIndex::UInt64) { + if (type->getTypeId() == TypeIndex::UInt64) + { has_uint64 = true; uint64_could_opposite &= (elem.template get() <= std::numeric_limits::max()); } @@ -194,23 +197,25 @@ DataTypePtr FieldToDataType::operator() (const Map & map) const value_types.reserve(map.size()); auto checkIfConversionSigned = [&](bool& has_signed_int, bool& has_uint64, bool& uint64_could_opposite, - const DataTypePtr & type, const Field & elem) { - if (uint64_could_opposite) { + const DataTypePtr & type, const Field & elem) + { + if (uint64_could_opposite) + { has_signed_int |= WhichDataType(type).isNativeInt(); - if (type->getTypeId() == TypeIndex::UInt64) { + if (type->getTypeId() == TypeIndex::UInt64) + { has_uint64 = true; uint64_could_opposite &= (elem.template get() <= std::numeric_limits::max()); } } }; - auto updateUInt64Types = [](DataTypes types) { - for (auto& type : types) { - if (type->getTypeId() == TypeIndex::UInt64) { + auto updateUInt64Types = [](DataTypes types) + { + for (auto& type : types) + if (type->getTypeId() == TypeIndex::UInt64) type = std::make_shared(); - } - } }; bool k_has_signed_int = false; From 349f3b9c02ca38942a57ea8474de9eaff2a4567a Mon Sep 17 00:00:00 2001 From: Mingliang Pan Date: Thu, 17 Aug 2023 15:12:42 +0800 Subject: [PATCH 062/433] Update StorageEmbeddedRocksDB.h add drop method to close file before drop table --- src/Storages/RocksDB/StorageEmbeddedRocksDB.h | 1 + 1 file changed, 1 insertion(+) diff --git a/src/Storages/RocksDB/StorageEmbeddedRocksDB.h b/src/Storages/RocksDB/StorageEmbeddedRocksDB.h index 97fd07626a8..d0cf05f261c 100644 --- a/src/Storages/RocksDB/StorageEmbeddedRocksDB.h +++ b/src/Storages/RocksDB/StorageEmbeddedRocksDB.h @@ -53,6 +53,7 @@ public: void checkMutationIsPossible(const MutationCommands & commands, const Settings & settings) const override; void mutate(const MutationCommands &, ContextPtr) override; + void drop() override; bool supportsParallelInsert() const override { return true; } bool supportsIndexForIn() const override { return true; } From 2db0b4f52afb445a23c9e947858269c8cf57e271 Mon Sep 17 00:00:00 2001 From: Mingliang Pan Date: Thu, 17 Aug 2023 15:14:14 +0800 Subject: [PATCH 063/433] Update StorageEmbeddedRocksDB.cpp add drop function to close file before drop table --- src/Storages/RocksDB/StorageEmbeddedRocksDB.cpp | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/src/Storages/RocksDB/StorageEmbeddedRocksDB.cpp b/src/Storages/RocksDB/StorageEmbeddedRocksDB.cpp index 27e8de78b0f..be322a402ee 100644 --- a/src/Storages/RocksDB/StorageEmbeddedRocksDB.cpp +++ b/src/Storages/RocksDB/StorageEmbeddedRocksDB.cpp @@ -304,6 +304,12 @@ void StorageEmbeddedRocksDB::mutate(const MutationCommands & commands, ContextPt } } +void StorageEmbeddedRocksDB::drop() +{ + rocksdb_ptr->Close(); + rocksdb_ptr = nullptr; +} + void StorageEmbeddedRocksDB::initDB() { rocksdb::Status status; From 7fa72f5cdfd61de0c4b79e4275f142c452ce60e6 Mon Sep 17 00:00:00 2001 From: avogar Date: Fri, 18 Aug 2023 12:47:26 +0000 Subject: [PATCH 064/433] Optimize group by constant keys --- src/Common/BoolArgsToTemplateArgsDispatcher.h | 40 +++++ src/Core/Settings.h | 1 + src/Core/SettingsChangesHistory.h | 3 +- src/Interpreters/Aggregator.cpp | 142 ++++++++++++------ src/Interpreters/Aggregator.h | 8 +- src/Interpreters/InterpreterSelectQuery.cpp | 1 + src/Planner/Planner.cpp | 1 + src/Processors/QueryPlan/AggregatingStep.cpp | 1 + .../TTL/TTLAggregationAlgorithm.cpp | 3 +- .../MergeTree/MergeTreeDataSelectExecutor.cpp | 3 +- .../02845_group_by_constant_keys.reference | 4 + .../02845_group_by_constant_keys.sql | 5 + 12 files changed, 161 insertions(+), 51 deletions(-) create mode 100644 src/Common/BoolArgsToTemplateArgsDispatcher.h create mode 100644 tests/queries/0_stateless/02845_group_by_constant_keys.reference create mode 100644 tests/queries/0_stateless/02845_group_by_constant_keys.sql diff --git a/src/Common/BoolArgsToTemplateArgsDispatcher.h b/src/Common/BoolArgsToTemplateArgsDispatcher.h new file mode 100644 index 00000000000..086b1e4b0f4 --- /dev/null +++ b/src/Common/BoolArgsToTemplateArgsDispatcher.h @@ -0,0 +1,40 @@ +#pragma once +#include + +namespace DB +{ + + +/// Special struct that helps to convert bool variables to function template bool arguments. +/// It can be used to avoid multiple nested if/else on bool arguments. How to use it: +/// Imagine you have template function +/// template return_type foo(...); +/// and bool variables b1, b2, ... bn. To pass these variables as template for foo you can do the following: +/// +/// auto call_foo = []() +/// { +/// return foo(...); +/// } +/// +/// BoolArgsToTemplateArgsDispatcher::call(call_foo, b1, b2, ..., bn); + +template +struct BoolArgsToTemplateArgsDispatcher +{ + template + static auto call(Func f, Args1&&... args) + { + return f.template operator()(std::forward(args)...); + } + + template + static auto call(Func f, bool b, Args1&&... ar1) + { + if (b) + return BoolArgsToTemplateArgsDispatcher::call(f, std::forward(ar1)...); + else + return BoolArgsToTemplateArgsDispatcher::call(f, std::forward(ar1)...); + } +}; + +} diff --git a/src/Core/Settings.h b/src/Core/Settings.h index 6c3d339b4be..418ae08ac7f 100644 --- a/src/Core/Settings.h +++ b/src/Core/Settings.h @@ -650,6 +650,7 @@ class IColumn; M(SetOperationMode, except_default_mode, SetOperationMode::ALL, "Set default mode in EXCEPT query. Possible values: empty string, 'ALL', 'DISTINCT'. If empty, query without mode will throw exception.", 0) \ M(Bool, optimize_aggregators_of_group_by_keys, true, "Eliminates min/max/any/anyLast aggregators of GROUP BY keys in SELECT section", 0) \ M(Bool, optimize_group_by_function_keys, true, "Eliminates functions of other keys in GROUP BY section", 0) \ + M(Bool, optimize_group_by_constant_keys, true, "Optimize GROUP BY when all keys in block are constant", 0) \ M(Bool, legacy_column_name_of_tuple_literal, false, "List all names of element of large tuple literals in their column names instead of hash. This settings exists only for compatibility reasons. It makes sense to set to 'true', while doing rolling update of cluster from version lower than 21.7 to higher.", 0) \ \ M(Bool, query_plan_enable_optimizations, true, "Apply optimizations to query plan", 0) \ diff --git a/src/Core/SettingsChangesHistory.h b/src/Core/SettingsChangesHistory.h index dcb67165add..eaa932a3520 100644 --- a/src/Core/SettingsChangesHistory.h +++ b/src/Core/SettingsChangesHistory.h @@ -80,7 +80,8 @@ namespace SettingsChangesHistory /// It's used to implement `compatibility` setting (see https://github.com/ClickHouse/ClickHouse/issues/35972) static std::map settings_changes_history = { - {"23.8", {{"rewrite_count_distinct_if_with_count_distinct_implementation", false, true, "Rewrite countDistinctIf with count_distinct_implementation configuration"}}}, + {"23.8", {{"rewrite_count_distinct_if_with_count_distinct_implementation", false, true, "Rewrite countDistinctIf with count_distinct_implementation configuration"}, + {"optimize_group_by_constant_keys", false, true, "Optimize group by constant keys by default"}}}, {"23.7", {{"function_sleep_max_microseconds_per_block", 0, 3000000, "In previous versions, the maximum sleep time of 3 seconds was applied only for `sleep`, but not for `sleepEachRow` function. In the new version, we introduce this setting. If you set compatibility with the previous versions, we will disable the limit altogether."}}}, {"23.6", {{"http_send_timeout", 180, 30, "3 minutes seems crazy long. Note that this is timeout for a single network write call, not for the whole upload operation."}, {"http_receive_timeout", 180, 30, "See http_send_timeout."}}}, diff --git a/src/Interpreters/Aggregator.cpp b/src/Interpreters/Aggregator.cpp index 91cd574708a..72ea2c8ec42 100644 --- a/src/Interpreters/Aggregator.cpp +++ b/src/Interpreters/Aggregator.cpp @@ -39,6 +39,7 @@ #include #include #include +#include #include @@ -1036,11 +1037,12 @@ void Aggregator::executeImpl( ColumnRawPtrs & key_columns, AggregateFunctionInstruction * aggregate_instructions, bool no_more_keys, + bool all_keys_are_const, AggregateDataPtr overflow_row) const { #define M(NAME, IS_TWO_LEVEL) \ else if (result.type == AggregatedDataVariants::Type::NAME) \ - executeImpl(*result.NAME, result.aggregates_pool, row_begin, row_end, key_columns, aggregate_instructions, no_more_keys, overflow_row); + executeImpl(*result.NAME, result.aggregates_pool, row_begin, row_end, key_columns, aggregate_instructions, no_more_keys, all_keys_are_const, overflow_row); if (false) {} // NOLINT APPLY_FOR_AGGREGATED_VARIANTS(M) @@ -1060,44 +1062,35 @@ void NO_INLINE Aggregator::executeImpl( ColumnRawPtrs & key_columns, AggregateFunctionInstruction * aggregate_instructions, bool no_more_keys, + bool all_keys_are_const, AggregateDataPtr overflow_row) const { typename Method::State state(key_columns, key_sizes, aggregation_state_cache); + auto call_execute_impl_batch = [&]() + { + executeImplBatch(method, state, aggregates_pool, row_begin, row_end, aggregate_instructions, overflow_row); + }; + + bool use_compiled_functions = false; +#if USE_EMBEDDED_COMPILER + use_compiled_functions = compiled_aggregate_functions_holder && !hasSparseArguments(aggregate_instructions); +#endif + if (!no_more_keys) { /// Prefetching doesn't make sense for small hash tables, because they fit in caches entirely. const bool prefetch = Method::State::has_cheap_key_calculation && params.enable_prefetch && (method.data.getBufferSizeInBytes() > min_bytes_for_prefetch); - -#if USE_EMBEDDED_COMPILER - if (compiled_aggregate_functions_holder && !hasSparseArguments(aggregate_instructions)) - { - if (prefetch) - executeImplBatch( - method, state, aggregates_pool, row_begin, row_end, aggregate_instructions, overflow_row); - else - executeImplBatch( - method, state, aggregates_pool, row_begin, row_end, aggregate_instructions, overflow_row); - } - else -#endif - { - if (prefetch) - executeImplBatch( - method, state, aggregates_pool, row_begin, row_end, aggregate_instructions, overflow_row); - else - executeImplBatch( - method, state, aggregates_pool, row_begin, row_end, aggregate_instructions, overflow_row); - } + BoolArgsToTemplateArgsDispatcher::call(call_execute_impl_batch, no_more_keys, use_compiled_functions, prefetch, all_keys_are_const); } else { - executeImplBatch(method, state, aggregates_pool, row_begin, row_end, aggregate_instructions, overflow_row); + BoolArgsToTemplateArgsDispatcher::call(call_execute_impl_batch, no_more_keys, false, false, all_keys_are_const); } } -template +template void NO_INLINE Aggregator::executeImplBatch( Method & method, typename Method::State & state, @@ -1121,27 +1114,34 @@ void NO_INLINE Aggregator::executeImplBatch( /// For all rows. AggregateDataPtr place = aggregates_pool->alloc(0); - for (size_t i = row_begin; i < row_end; ++i) + if (all_keys_are_const) { - if constexpr (prefetch && HasPrefetchMemberFunc) + state.emplaceKey(method.data, 0, *aggregates_pool).setMapped(place); + } + else + { + for (size_t i = row_begin; i < row_end; ++i) { - if (i == row_begin + prefetching.iterationsToMeasure()) - prefetch_look_ahead = prefetching.calcPrefetchLookAhead(); - - if (i + prefetch_look_ahead < row_end) + if constexpr (prefetch && HasPrefetchMemberFunc) { - auto && key_holder = state.getKeyHolder(i + prefetch_look_ahead, *aggregates_pool); - method.data.prefetch(std::move(key_holder)); - } - } + if (i == row_begin + prefetching.iterationsToMeasure()) + prefetch_look_ahead = prefetching.calcPrefetchLookAhead(); - state.emplaceKey(method.data, i, *aggregates_pool).setMapped(place); + if (i + prefetch_look_ahead < row_end) + { + auto && key_holder = state.getKeyHolder(i + prefetch_look_ahead, *aggregates_pool); + method.data.prefetch(std::move(key_holder)); + } + } + + state.emplaceKey(method.data, i, *aggregates_pool).setMapped(place); + } } return; } /// Optimization for special case when aggregating by 8bit key. - if constexpr (!no_more_keys && std::is_same_v) + if constexpr (!no_more_keys && !all_keys_are_const && std::is_same_v) { /// We use another method if there are aggregate functions with -Array combinator. bool has_arrays = false; @@ -1183,7 +1183,20 @@ void NO_INLINE Aggregator::executeImplBatch( std::unique_ptr places(new AggregateDataPtr[row_end]); /// For all rows. - for (size_t i = row_begin; i < row_end; ++i) + size_t start, end; + /// If all keys are const, key columns contain only 1 row. + if constexpr (all_keys_are_const) + { + start = 0; + end = 1; + } + else + { + start = row_begin; + end = row_end; + } + + for (size_t i = start; i < end; ++i) { AggregateDataPtr aggregate_data = nullptr; @@ -1279,8 +1292,16 @@ void NO_INLINE Aggregator::executeImplBatch( columns_data.emplace_back(getColumnData(inst->batch_arguments[argument_index])); } - auto add_into_aggregate_states_function = compiled_aggregate_functions_holder->compiled_aggregate_functions.add_into_aggregate_states_function; - add_into_aggregate_states_function(row_begin, row_end, columns_data.data(), places.get()); + if constexpr (all_keys_are_const) + { + auto add_into_aggregate_states_function_single_place = compiled_aggregate_functions_holder->compiled_aggregate_functions.add_into_aggregate_states_function_single_place; + add_into_aggregate_states_function_single_place(row_begin, row_end, columns_data.data(), places[0]); + } + else + { + auto add_into_aggregate_states_function = compiled_aggregate_functions_holder->compiled_aggregate_functions.add_into_aggregate_states_function; + add_into_aggregate_states_function(row_begin, row_end, columns_data.data(), places.get()); + } } #endif @@ -1295,12 +1316,24 @@ void NO_INLINE Aggregator::executeImplBatch( AggregateFunctionInstruction * inst = aggregate_instructions + i; - if (inst->offsets) - inst->batch_that->addBatchArray(row_begin, row_end, places.get(), inst->state_offset, inst->batch_arguments, inst->offsets, aggregates_pool); - else if (inst->has_sparse_arguments) - inst->batch_that->addBatchSparse(row_begin, row_end, places.get(), inst->state_offset, inst->batch_arguments, aggregates_pool); + if constexpr (all_keys_are_const) + { + if (inst->offsets) + inst->batch_that->addBatchSinglePlace(inst->offsets[static_cast(row_begin) - 1], inst->offsets[row_end - 1], places[0] + inst->state_offset, inst->batch_arguments, aggregates_pool); + else if (inst->has_sparse_arguments) + inst->batch_that->addBatchSparseSinglePlace(row_begin, row_end, places[0] + inst->state_offset, inst->batch_arguments, aggregates_pool); + else + inst->batch_that->addBatchSinglePlace(row_begin, row_end, places[0] + inst->state_offset, inst->batch_arguments, aggregates_pool); + } else - inst->batch_that->addBatch(row_begin, row_end, places.get(), inst->state_offset, inst->batch_arguments, aggregates_pool); + { + if (inst->offsets) + inst->batch_that->addBatchArray(row_begin, row_end, places.get(), inst->state_offset, inst->batch_arguments, inst->offsets, aggregates_pool); + else if (inst->has_sparse_arguments) + inst->batch_that->addBatchSparse(row_begin, row_end, places.get(), inst->state_offset, inst->batch_arguments, aggregates_pool); + else + inst->batch_that->addBatch(row_begin, row_end, places.get(), inst->state_offset, inst->batch_arguments, aggregates_pool); + } } } @@ -1540,12 +1573,27 @@ bool Aggregator::executeOnBlock(Columns columns, * To make them work anyway, we materialize them. */ Columns materialized_columns; + bool all_keys_are_const = false; + if (params.optimize_group_by_constant_keys) + { + all_keys_are_const = true; + for (size_t i = 0; i < params.keys_size; ++i) + all_keys_are_const &= isColumnConst(*columns.at(keys_positions[i])); + } /// Remember the columns we will work with for (size_t i = 0; i < params.keys_size; ++i) { - materialized_columns.push_back(recursiveRemoveSparse(columns.at(keys_positions[i]))->convertToFullColumnIfConst()); - key_columns[i] = materialized_columns.back().get(); + if (all_keys_are_const) + { + key_columns[i] = assert_cast(*columns.at(keys_positions[i])).getDataColumnPtr().get(); + } + else + { + materialized_columns.push_back(recursiveRemoveSparse(columns.at(keys_positions[i]))->convertToFullColumnIfConst()); + key_columns[i] = materialized_columns.back().get(); + } + if (!result.isLowCardinality()) { @@ -1590,7 +1638,7 @@ bool Aggregator::executeOnBlock(Columns columns, { /// This is where data is written that does not fit in `max_rows_to_group_by` with `group_by_overflow_mode = any`. AggregateDataPtr overflow_row_ptr = params.overflow_row ? result.without_key : nullptr; - executeImpl(result, row_begin, row_end, key_columns, aggregate_functions_instructions.data(), no_more_keys, overflow_row_ptr); + executeImpl(result, row_begin, row_end, key_columns, aggregate_functions_instructions.data(), no_more_keys, all_keys_are_const, overflow_row_ptr); } size_t result_size = result.sizeWithoutOverflowRow(); diff --git a/src/Interpreters/Aggregator.h b/src/Interpreters/Aggregator.h index 4f2c86606c5..dc0391c4289 100644 --- a/src/Interpreters/Aggregator.h +++ b/src/Interpreters/Aggregator.h @@ -1020,6 +1020,8 @@ public: bool enable_prefetch; + bool optimize_group_by_constant_keys; + struct StatsCollectingParams { StatsCollectingParams(); @@ -1057,6 +1059,7 @@ public: size_t max_block_size_, bool enable_prefetch_, bool only_merge_, // true for projections + bool optimize_group_by_constant_keys_, const StatsCollectingParams & stats_collecting_params_ = {}) : keys(keys_) , aggregates(aggregates_) @@ -1077,6 +1080,7 @@ public: , max_block_size(max_block_size_) , only_merge(only_merge_) , enable_prefetch(enable_prefetch_) + , optimize_group_by_constant_keys(optimize_group_by_constant_keys_) , stats_collecting_params(stats_collecting_params_) { } @@ -1276,6 +1280,7 @@ private: ColumnRawPtrs & key_columns, AggregateFunctionInstruction * aggregate_instructions, bool no_more_keys = false, + bool all_keys_are_const = false, AggregateDataPtr overflow_row = nullptr) const; /// Process one data block, aggregate the data into a hash table. @@ -1288,10 +1293,11 @@ private: ColumnRawPtrs & key_columns, AggregateFunctionInstruction * aggregate_instructions, bool no_more_keys, + bool all_keys_are_const, AggregateDataPtr overflow_row) const; /// Specialization for a particular value no_more_keys. - template + template void executeImplBatch( Method & method, typename Method::State & state, diff --git a/src/Interpreters/InterpreterSelectQuery.cpp b/src/Interpreters/InterpreterSelectQuery.cpp index 20fca1b1e76..8839bd8f7e2 100644 --- a/src/Interpreters/InterpreterSelectQuery.cpp +++ b/src/Interpreters/InterpreterSelectQuery.cpp @@ -2579,6 +2579,7 @@ static Aggregator::Params getAggregatorParams( settings.max_block_size, settings.enable_software_prefetch_in_aggregation, /* only_merge */ false, + settings.optimize_group_by_constant_keys, stats_collecting_params }; } diff --git a/src/Planner/Planner.cpp b/src/Planner/Planner.cpp index 9f6c22f90f3..a75031898c4 100644 --- a/src/Planner/Planner.cpp +++ b/src/Planner/Planner.cpp @@ -290,6 +290,7 @@ Aggregator::Params getAggregatorParams(const PlannerContextPtr & planner_context settings.max_block_size, settings.enable_software_prefetch_in_aggregation, /* only_merge */ false, + settings.optimize_group_by_constant_keys, stats_collecting_params); return aggregator_params; diff --git a/src/Processors/QueryPlan/AggregatingStep.cpp b/src/Processors/QueryPlan/AggregatingStep.cpp index eebbfc04304..c68a070f816 100644 --- a/src/Processors/QueryPlan/AggregatingStep.cpp +++ b/src/Processors/QueryPlan/AggregatingStep.cpp @@ -230,6 +230,7 @@ void AggregatingStep::transformPipeline(QueryPipelineBuilder & pipeline, const B transform_params->params.max_block_size, transform_params->params.enable_prefetch, /* only_merge */ false, + transform_params->params.optimize_group_by_constant_keys, transform_params->params.stats_collecting_params}; auto transform_params_for_set = std::make_shared(src_header, std::move(params_for_set), final); diff --git a/src/Processors/TTL/TTLAggregationAlgorithm.cpp b/src/Processors/TTL/TTLAggregationAlgorithm.cpp index cdd6e3e917f..fa3436ec55d 100644 --- a/src/Processors/TTL/TTLAggregationAlgorithm.cpp +++ b/src/Processors/TTL/TTLAggregationAlgorithm.cpp @@ -41,7 +41,8 @@ TTLAggregationAlgorithm::TTLAggregationAlgorithm( settings.min_count_to_compile_aggregate_expression, settings.max_block_size, settings.enable_software_prefetch_in_aggregation, - false /* only_merge */); + false /* only_merge */, + settings.optimize_group_by_constant_keys); aggregator = std::make_unique(header, params); diff --git a/src/Storages/MergeTree/MergeTreeDataSelectExecutor.cpp b/src/Storages/MergeTree/MergeTreeDataSelectExecutor.cpp index ee515106591..8f9d54d942d 100644 --- a/src/Storages/MergeTree/MergeTreeDataSelectExecutor.cpp +++ b/src/Storages/MergeTree/MergeTreeDataSelectExecutor.cpp @@ -328,7 +328,8 @@ QueryPlanPtr MergeTreeDataSelectExecutor::read( settings.min_count_to_compile_aggregate_expression, settings.max_block_size, settings.enable_software_prefetch_in_aggregation, - only_merge); + only_merge, + settings.optimize_group_by_constant_keys); return std::make_pair(params, only_merge); }; diff --git a/tests/queries/0_stateless/02845_group_by_constant_keys.reference b/tests/queries/0_stateless/02845_group_by_constant_keys.reference new file mode 100644 index 00000000000..60e40ea54a7 --- /dev/null +++ b/tests/queries/0_stateless/02845_group_by_constant_keys.reference @@ -0,0 +1,4 @@ +10000000 1 2 3 +10000000 1 2 3 +10000000 1 2 3 +10000000 1 2 3 diff --git a/tests/queries/0_stateless/02845_group_by_constant_keys.sql b/tests/queries/0_stateless/02845_group_by_constant_keys.sql new file mode 100644 index 00000000000..0223cf1df60 --- /dev/null +++ b/tests/queries/0_stateless/02845_group_by_constant_keys.sql @@ -0,0 +1,5 @@ +SELECT count(number), 1 AS k1, 2 as k2, 3 as k3 FROM numbers_mt(10000000) GROUP BY k1, k2, k3 settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=0, compile_aggregate_expressions=0; +SELECT count(number), 1 AS k1, 2 as k2, 3 as k3 FROM numbers_mt(10000000) GROUP BY k1, k2, k3 settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=1, compile_aggregate_expressions = 0; +SELECT count(number), 1 AS k1, 2 as k2, 3 as k3 FROM numbers_mt(10000000) GROUP BY k1, k2, k3 settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=0, compile_aggregate_expressions = 1; +SELECT count(number), 1 AS k1, 2 as k2, 3 as k3 FROM numbers_mt(10000000) GROUP BY k1, k2, k3 settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=1, compile_aggregate_expressions = 1; + From 1f5e209897b4d45dbaa6c9e2588bcc7a536702a4 Mon Sep 17 00:00:00 2001 From: avogar Date: Fri, 18 Aug 2023 12:50:04 +0000 Subject: [PATCH 065/433] Better comment --- src/Common/BoolArgsToTemplateArgsDispatcher.h | 15 +++++++-------- 1 file changed, 7 insertions(+), 8 deletions(-) diff --git a/src/Common/BoolArgsToTemplateArgsDispatcher.h b/src/Common/BoolArgsToTemplateArgsDispatcher.h index 086b1e4b0f4..469bc34cfbf 100644 --- a/src/Common/BoolArgsToTemplateArgsDispatcher.h +++ b/src/Common/BoolArgsToTemplateArgsDispatcher.h @@ -4,36 +4,35 @@ namespace DB { - /// Special struct that helps to convert bool variables to function template bool arguments. /// It can be used to avoid multiple nested if/else on bool arguments. How to use it: /// Imagine you have template function /// template return_type foo(...); -/// and bool variables b1, b2, ... bn. To pass these variables as template for foo you can do the following: +/// and bool variables b1, b2, ..., bool bn. To pass these variables as template for foo you can do the following: /// -/// auto call_foo = []() +/// auto call_foo = []() /// { /// return foo(...); /// } /// /// BoolArgsToTemplateArgsDispatcher::call(call_foo, b1, b2, ..., bn); -template +template struct BoolArgsToTemplateArgsDispatcher { template - static auto call(Func f, Args1&&... args) + static auto call(Functor f, Args1&&... args) { return f.template operator()(std::forward(args)...); } template - static auto call(Func f, bool b, Args1&&... ar1) + static auto call(Functor f, bool b, Args1&&... ar1) { if (b) - return BoolArgsToTemplateArgsDispatcher::call(f, std::forward(ar1)...); + return BoolArgsToTemplateArgsDispatcher::call(f, std::forward(ar1)...); else - return BoolArgsToTemplateArgsDispatcher::call(f, std::forward(ar1)...); + return BoolArgsToTemplateArgsDispatcher::call(f, std::forward(ar1)...); } }; From aac513185952a8d12f6e9e0247623a63d3c39ee4 Mon Sep 17 00:00:00 2001 From: avogar Date: Fri, 18 Aug 2023 13:21:16 +0000 Subject: [PATCH 066/433] Fix for no_more_keys --- src/Common/BoolArgsToTemplateArgsDispatcher.h | 4 ++-- src/Interpreters/Aggregator.cpp | 12 ++++++++++-- src/Storages/StorageFile.cpp | 4 ++-- 3 files changed, 14 insertions(+), 6 deletions(-) diff --git a/src/Common/BoolArgsToTemplateArgsDispatcher.h b/src/Common/BoolArgsToTemplateArgsDispatcher.h index 469bc34cfbf..3e72db49cd2 100644 --- a/src/Common/BoolArgsToTemplateArgsDispatcher.h +++ b/src/Common/BoolArgsToTemplateArgsDispatcher.h @@ -7,8 +7,8 @@ namespace DB /// Special struct that helps to convert bool variables to function template bool arguments. /// It can be used to avoid multiple nested if/else on bool arguments. How to use it: /// Imagine you have template function -/// template return_type foo(...); -/// and bool variables b1, b2, ..., bool bn. To pass these variables as template for foo you can do the following: +/// template return_type foo(...); +/// and bool variables b1, b2, ..., bn. To pass these variables as template for foo you can do the following: /// /// auto call_foo = []() /// { diff --git a/src/Interpreters/Aggregator.cpp b/src/Interpreters/Aggregator.cpp index 72ea2c8ec42..6392428c5e6 100644 --- a/src/Interpreters/Aggregator.cpp +++ b/src/Interpreters/Aggregator.cpp @@ -1180,7 +1180,7 @@ void NO_INLINE Aggregator::executeImplBatch( /// - this affects only optimize_aggregation_in_order, /// - this is just a pointer, so it should not be significant, /// - and plus this will require other changes in the interface. - std::unique_ptr places(new AggregateDataPtr[row_end]); + std::unique_ptr places(new AggregateDataPtr[all_keys_are_const ? 1 : row_end]); /// For all rows. size_t start, end; @@ -1202,7 +1202,7 @@ void NO_INLINE Aggregator::executeImplBatch( if constexpr (!no_more_keys) { - if constexpr (prefetch && HasPrefetchMemberFunc) + if constexpr (prefetch && !all_keys_are_const && HasPrefetchMemberFunc) { if (i == row_begin + prefetching.iterationsToMeasure()) prefetch_look_ahead = prefetching.calcPrefetchLookAhead(); @@ -1267,9 +1267,17 @@ void NO_INLINE Aggregator::executeImplBatch( /// Add only if the key already exists. auto find_result = state.findKey(method.data, i, *aggregates_pool); if (find_result.isFound()) + { aggregate_data = find_result.getMapped(); + } else + { + /// If all keys are constant and this is new key + /// we don't need to do anything and just skip the whole block. + if constexpr (all_keys_are_const) + return; aggregate_data = overflow_row; + } } places[i] = aggregate_data; diff --git a/src/Storages/StorageFile.cpp b/src/Storages/StorageFile.cpp index e3908c75a58..57e04f547d4 100644 --- a/src/Storages/StorageFile.cpp +++ b/src/Storages/StorageFile.cpp @@ -965,13 +965,13 @@ public: { if (virtual_column.name == "_path") { - chunk.addColumn(virtual_column.type->createColumnConst(num_rows, current_path)->convertToFullColumnIfConst()); + chunk.addColumn(virtual_column.type->createColumnConst(num_rows, current_path)); } else if (virtual_column.name == "_file") { size_t last_slash_pos = current_path.find_last_of('/'); auto file_name = current_path.substr(last_slash_pos + 1); - chunk.addColumn(virtual_column.type->createColumnConst(num_rows, file_name)->convertToFullColumnIfConst()); + chunk.addColumn(virtual_column.type->createColumnConst(num_rows, file_name)); } } From 0da8fe2793b88dbc3fd7465644e154b872a75000 Mon Sep 17 00:00:00 2001 From: avogar Date: Fri, 18 Aug 2023 13:22:36 +0000 Subject: [PATCH 067/433] Add missing constexpr --- src/Interpreters/Aggregator.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/Interpreters/Aggregator.cpp b/src/Interpreters/Aggregator.cpp index 6392428c5e6..0d4ebfd1081 100644 --- a/src/Interpreters/Aggregator.cpp +++ b/src/Interpreters/Aggregator.cpp @@ -1114,7 +1114,7 @@ void NO_INLINE Aggregator::executeImplBatch( /// For all rows. AggregateDataPtr place = aggregates_pool->alloc(0); - if (all_keys_are_const) + if constexpr (all_keys_are_const) { state.emplaceKey(method.data, 0, *aggregates_pool).setMapped(place); } From 2d4167c61206e256a36e647c878d3f98f40a2724 Mon Sep 17 00:00:00 2001 From: avogar Date: Fri, 18 Aug 2023 14:50:07 +0000 Subject: [PATCH 068/433] FRemove unneded changes --- src/Storages/StorageFile.cpp | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/Storages/StorageFile.cpp b/src/Storages/StorageFile.cpp index 57e04f547d4..e3908c75a58 100644 --- a/src/Storages/StorageFile.cpp +++ b/src/Storages/StorageFile.cpp @@ -965,13 +965,13 @@ public: { if (virtual_column.name == "_path") { - chunk.addColumn(virtual_column.type->createColumnConst(num_rows, current_path)); + chunk.addColumn(virtual_column.type->createColumnConst(num_rows, current_path)->convertToFullColumnIfConst()); } else if (virtual_column.name == "_file") { size_t last_slash_pos = current_path.find_last_of('/'); auto file_name = current_path.substr(last_slash_pos + 1); - chunk.addColumn(virtual_column.type->createColumnConst(num_rows, file_name)); + chunk.addColumn(virtual_column.type->createColumnConst(num_rows, file_name)->convertToFullColumnIfConst()); } } From c11ed5feb4b8c916ba36b6fe6171e7ab1f76b2b4 Mon Sep 17 00:00:00 2001 From: yariks5s Date: Tue, 22 Aug 2023 10:57:58 +0000 Subject: [PATCH 069/433] support for non-const timezone arguments/strings --- src/Functions/FunctionsConversion.h | 29 ++++++++----------- .../extractTimeZoneFromFunctionArguments.cpp | 4 +-- 2 files changed, 14 insertions(+), 19 deletions(-) diff --git a/src/Functions/FunctionsConversion.h b/src/Functions/FunctionsConversion.h index 79d17d8ac98..102d4339554 100644 --- a/src/Functions/FunctionsConversion.h +++ b/src/Functions/FunctionsConversion.h @@ -1844,20 +1844,20 @@ public: mandatory_args.push_back({"scale", &isNativeInteger, &isColumnConst, "const Integer"}); } - // toString(DateTime or DateTime64, [timezone: String]) + // toString(DateTime or DateTime64, [timezone: String (const/non-const)]) if ((std::is_same_v && !arguments.empty() && (isDateTime64(arguments[0].type) || isDateTime(arguments[0].type))) - // toUnixTimestamp(value[, timezone : String]) + // toUnixTimestamp(value[, timezone : String(const/non-const)]) || std::is_same_v - // toDate(value[, timezone : String]) + // toDate(value[, timezone : String(const/non-const)]) || std::is_same_v // TODO: shall we allow timestamp argument for toDate? DateTime knows nothing about timezones and this argument is ignored below. - // toDate32(value[, timezone : String]) + // toDate32(value[, timezone : String(const/non-const)]) || std::is_same_v - // toDateTime(value[, timezone: String]) + // toDateTime(value[, timezone: String(const/non-const)]) || std::is_same_v - // toDateTime64(value, scale : Integer[, timezone: String]) + // toDateTime64(value, scale : Integer[, timezone: String(const/non-const)]) || std::is_same_v) - { - optional_args.push_back({"timezone", &isString, &isColumnConst, "const String"}); + { + optional_args.push_back({"timezone", &isString, nullptr, "const String/String"}); } validateFunctionArgumentTypes(*this, arguments, mandatory_args, optional_args); @@ -1895,13 +1895,13 @@ public: if (to_datetime64 || scale != 0) /// toDateTime('xxxx-xx-xx xx:xx:xx', 0) return DateTime return std::make_shared(scale, - extractTimeZoneNameFromFunctionArguments(arguments, timezone_arg_position, 0, false)); + extractTimeZoneNameFromFunctionArguments(arguments, timezone_arg_position, 0, true)); - return std::make_shared(extractTimeZoneNameFromFunctionArguments(arguments, timezone_arg_position, 0, false)); + return std::make_shared(extractTimeZoneNameFromFunctionArguments(arguments, timezone_arg_position, 0, true)); } if constexpr (std::is_same_v) - return std::make_shared(extractTimeZoneNameFromFunctionArguments(arguments, timezone_arg_position, 0, false)); + return std::make_shared(extractTimeZoneNameFromFunctionArguments(arguments, timezone_arg_position, 0, true)); else if constexpr (std::is_same_v) throw Exception(ErrorCodes::LOGICAL_ERROR, "Unexpected branch in code of conversion function: it is a bug."); else @@ -1919,12 +1919,7 @@ public: } bool useDefaultImplementationForConstants() const override { return true; } - ColumnNumbers getArgumentsThatAreAlwaysConstant() const override - { - if constexpr (std::is_same_v) - return {2}; - return {1}; - } + bool canBeExecutedOnDefaultArguments() const override { return false; } ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override diff --git a/src/Functions/extractTimeZoneFromFunctionArguments.cpp b/src/Functions/extractTimeZoneFromFunctionArguments.cpp index 7168c68c9c9..38ce0467aee 100644 --- a/src/Functions/extractTimeZoneFromFunctionArguments.cpp +++ b/src/Functions/extractTimeZoneFromFunctionArguments.cpp @@ -30,11 +30,11 @@ std::string extractTimeZoneNameFromColumn(const IColumn * column, const String & } -std::string extractTimeZoneNameFromFunctionArguments(const ColumnsWithTypeAndName & arguments, size_t time_zone_arg_num, size_t datetime_arg_num, bool allow_nonconst_timezone_arguments) +std::string extractTimeZoneNameFromFunctionArguments(const ColumnsWithTypeAndName & arguments, size_t time_zone_arg_num, size_t datetime_arg_num, [[maybe_unused]] bool allow_nonconst_timezone_arguments) { /// Explicit time zone may be passed in last argument. if ((arguments.size() == time_zone_arg_num + 1) - && (!allow_nonconst_timezone_arguments || arguments[time_zone_arg_num].column)) + && (arguments[time_zone_arg_num].column)) { return extractTimeZoneNameFromColumn(arguments[time_zone_arg_num].column.get(), arguments[time_zone_arg_num].name); } From cceb2b2d570bf97de5421c17d8825d9df48139da Mon Sep 17 00:00:00 2001 From: Antonio Andelic Date: Wed, 23 Aug 2023 08:12:05 +0000 Subject: [PATCH 070/433] Disable integration with async replication --- tests/integration/helpers/cluster.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tests/integration/helpers/cluster.py b/tests/integration/helpers/cluster.py index 1f3cd473e29..9f8fa390be3 100644 --- a/tests/integration/helpers/cluster.py +++ b/tests/integration/helpers/cluster.py @@ -4258,7 +4258,7 @@ class ClickHouseInstance: if len(self.custom_dictionaries_paths): write_embedded_config("0_common_enable_dictionaries.xml", self.config_d_dir) - write_embedded_config("0_common_enable_keeper_async_replication.xml", self.config_d_dir) + #write_embedded_config("0_common_enable_keeper_async_replication.xml", self.config_d_dir) logging.debug("Generate and write macros file") macros = self.macros.copy() From ee5d7b1d955b6c2a8d564b1085e0168a0ffbdb14 Mon Sep 17 00:00:00 2001 From: robot-clickhouse Date: Wed, 23 Aug 2023 08:22:54 +0000 Subject: [PATCH 071/433] Automatic style fix --- tests/integration/helpers/cluster.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tests/integration/helpers/cluster.py b/tests/integration/helpers/cluster.py index 9f8fa390be3..35a656aa4a4 100644 --- a/tests/integration/helpers/cluster.py +++ b/tests/integration/helpers/cluster.py @@ -4258,7 +4258,7 @@ class ClickHouseInstance: if len(self.custom_dictionaries_paths): write_embedded_config("0_common_enable_dictionaries.xml", self.config_d_dir) - #write_embedded_config("0_common_enable_keeper_async_replication.xml", self.config_d_dir) + # write_embedded_config("0_common_enable_keeper_async_replication.xml", self.config_d_dir) logging.debug("Generate and write macros file") macros = self.macros.copy() From 625be86dc0d254447429a8ed4737cca4bb7137e3 Mon Sep 17 00:00:00 2001 From: Antonio Andelic Date: Wed, 23 Aug 2023 08:59:41 +0000 Subject: [PATCH 072/433] Fix tests --- src/Coordination/tests/gtest_coordination.cpp | 210 +++++++++++++----- utils/keeper-data-dumper/main.cpp | 4 +- 2 files changed, 162 insertions(+), 52 deletions(-) diff --git a/src/Coordination/tests/gtest_coordination.cpp b/src/Coordination/tests/gtest_coordination.cpp index 08d31bf34f6..759304b40ba 100644 --- a/src/Coordination/tests/gtest_coordination.cpp +++ b/src/Coordination/tests/gtest_coordination.cpp @@ -298,7 +298,9 @@ TEST_P(CoordinationTest, ChangelogTestSimple) setLogDirectory("./logs"); DB::KeeperLogStore changelog( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, + DB::FlushSettings(), + keeper_context); changelog.init(1, 0); auto entry = getLogEntry("hello world", 77); changelog.append(entry); @@ -328,7 +330,9 @@ TEST_P(CoordinationTest, ChangelogTestFile) setLogDirectory("./logs"); DB::KeeperLogStore changelog( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, + DB::FlushSettings(), + keeper_context); changelog.init(1, 0); auto entry = getLogEntry("hello world", 77); changelog.append(entry); @@ -360,7 +364,9 @@ TEST_P(CoordinationTest, ChangelogReadWrite) setLogDirectory("./logs"); DB::KeeperLogStore changelog( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 1000}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 1000}, + DB::FlushSettings(), + keeper_context); changelog.init(1, 0); for (size_t i = 0; i < 10; ++i) @@ -375,7 +381,9 @@ TEST_P(CoordinationTest, ChangelogReadWrite) waitDurableLogs(changelog); DB::KeeperLogStore changelog_reader( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 1000}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 1000}, + DB::FlushSettings(), + keeper_context); changelog_reader.init(1, 0); EXPECT_EQ(changelog_reader.size(), 10); EXPECT_EQ(changelog_reader.last_entry()->get_term(), changelog.last_entry()->get_term()); @@ -398,7 +406,9 @@ TEST_P(CoordinationTest, ChangelogWriteAt) setLogDirectory("./logs"); DB::KeeperLogStore changelog( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 1000}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 1000}, + DB::FlushSettings(), + keeper_context); changelog.init(1, 0); for (size_t i = 0; i < 10; ++i) { @@ -421,7 +431,9 @@ TEST_P(CoordinationTest, ChangelogWriteAt) EXPECT_EQ(changelog.next_slot(), 8); DB::KeeperLogStore changelog_reader( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 1000}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 1000}, + DB::FlushSettings(), + keeper_context); changelog_reader.init(1, 0); EXPECT_EQ(changelog_reader.size(), changelog.size()); @@ -438,7 +450,9 @@ TEST_P(CoordinationTest, ChangelogTestAppendAfterRead) setLogDirectory("./logs"); DB::KeeperLogStore changelog( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, + DB::FlushSettings(), + keeper_context); changelog.init(1, 0); for (size_t i = 0; i < 7; ++i) { @@ -455,7 +469,9 @@ TEST_P(CoordinationTest, ChangelogTestAppendAfterRead) EXPECT_TRUE(fs::exists("./logs/changelog_6_10.bin" + params.extension)); DB::KeeperLogStore changelog_reader( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, + DB::FlushSettings(), + keeper_context); changelog_reader.init(1, 0); EXPECT_EQ(changelog_reader.size(), 7); @@ -520,7 +536,9 @@ TEST_P(CoordinationTest, ChangelogTestCompaction) setLogDirectory("./logs"); DB::KeeperLogStore changelog( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, + DB::FlushSettings(), + keeper_context); changelog.init(1, 0); for (size_t i = 0; i < 3; ++i) @@ -570,7 +588,9 @@ TEST_P(CoordinationTest, ChangelogTestCompaction) EXPECT_EQ(changelog.last_entry()->get_term(), 60); /// And we able to read it DB::KeeperLogStore changelog_reader( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, + DB::FlushSettings(), + keeper_context); changelog_reader.init(7, 0); EXPECT_EQ(changelog_reader.size(), 1); @@ -586,7 +606,9 @@ TEST_P(CoordinationTest, ChangelogTestBatchOperations) setLogDirectory("./logs"); DB::KeeperLogStore changelog( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, + DB::FlushSettings(), + keeper_context); changelog.init(1, 0); for (size_t i = 0; i < 10; ++i) { @@ -602,7 +624,9 @@ TEST_P(CoordinationTest, ChangelogTestBatchOperations) auto entries = changelog.pack(1, 5); DB::KeeperLogStore apply_changelog( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, + DB::FlushSettings(), + keeper_context); apply_changelog.init(1, 0); for (size_t i = 0; i < 10; ++i) @@ -639,7 +663,9 @@ TEST_P(CoordinationTest, ChangelogTestBatchOperationsEmpty) nuraft::ptr entries; { DB::KeeperLogStore changelog( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, + DB::FlushSettings(), + keeper_context); changelog.init(1, 0); for (size_t i = 0; i < 10; ++i) { @@ -658,7 +684,9 @@ TEST_P(CoordinationTest, ChangelogTestBatchOperationsEmpty) ChangelogDirTest test1("./logs1"); setLogDirectory("./logs1"); DB::KeeperLogStore changelog_new( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, + DB::FlushSettings(), + keeper_context); changelog_new.init(1, 0); EXPECT_EQ(changelog_new.size(), 0); @@ -681,7 +709,9 @@ TEST_P(CoordinationTest, ChangelogTestBatchOperationsEmpty) EXPECT_EQ(changelog_new.next_slot(), 11); DB::KeeperLogStore changelog_reader( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, + DB::FlushSettings(), + keeper_context); changelog_reader.init(5, 0); } @@ -693,7 +723,9 @@ TEST_P(CoordinationTest, ChangelogTestWriteAtPreviousFile) setLogDirectory("./logs"); DB::KeeperLogStore changelog( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, + DB::FlushSettings(), + keeper_context); changelog.init(1, 0); for (size_t i = 0; i < 33; ++i) @@ -735,7 +767,9 @@ TEST_P(CoordinationTest, ChangelogTestWriteAtPreviousFile) EXPECT_FALSE(fs::exists("./logs/changelog_31_35.bin" + params.extension)); DB::KeeperLogStore changelog_read( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, + DB::FlushSettings(), + keeper_context); changelog_read.init(1, 0); EXPECT_EQ(changelog_read.size(), 7); EXPECT_EQ(changelog_read.start_index(), 1); @@ -750,7 +784,9 @@ TEST_P(CoordinationTest, ChangelogTestWriteAtFileBorder) setLogDirectory("./logs"); DB::KeeperLogStore changelog( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, + DB::FlushSettings(), + keeper_context); changelog.init(1, 0); for (size_t i = 0; i < 33; ++i) @@ -792,7 +828,9 @@ TEST_P(CoordinationTest, ChangelogTestWriteAtFileBorder) EXPECT_FALSE(fs::exists("./logs/changelog_31_35.bin" + params.extension)); DB::KeeperLogStore changelog_read( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, + DB::FlushSettings(), + keeper_context); changelog_read.init(1, 0); EXPECT_EQ(changelog_read.size(), 11); EXPECT_EQ(changelog_read.start_index(), 1); @@ -807,7 +845,9 @@ TEST_P(CoordinationTest, ChangelogTestWriteAtAllFiles) setLogDirectory("./logs"); DB::KeeperLogStore changelog( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, + DB::FlushSettings(), + keeper_context); changelog.init(1, 0); for (size_t i = 0; i < 33; ++i) { @@ -855,7 +895,9 @@ TEST_P(CoordinationTest, ChangelogTestStartNewLogAfterRead) setLogDirectory("./logs"); DB::KeeperLogStore changelog( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, + DB::FlushSettings(), + keeper_context); changelog.init(1, 0); for (size_t i = 0; i < 35; ++i) @@ -877,7 +919,9 @@ TEST_P(CoordinationTest, ChangelogTestStartNewLogAfterRead) EXPECT_FALSE(fs::exists("./logs/changelog_36_40.bin" + params.extension)); DB::KeeperLogStore changelog_reader( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, + DB::FlushSettings(), + keeper_context); changelog_reader.init(1, 0); auto entry = getLogEntry("36_hello_world", 360); @@ -924,7 +968,9 @@ TEST_P(CoordinationTest, ChangelogTestReadAfterBrokenTruncate) setLogDirectory(log_folder); DB::KeeperLogStore changelog( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, + DB::FlushSettings(), + keeper_context); changelog.init(1, 0); for (size_t i = 0; i < 35; ++i) @@ -949,7 +995,9 @@ TEST_P(CoordinationTest, ChangelogTestReadAfterBrokenTruncate) plain_buf.truncate(0); DB::KeeperLogStore changelog_reader( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, + DB::FlushSettings(), + keeper_context); changelog_reader.init(1, 0); changelog_reader.end_of_append_batch(0, 0); @@ -983,7 +1031,9 @@ TEST_P(CoordinationTest, ChangelogTestReadAfterBrokenTruncate) assertBrokenLogRemoved(log_folder, "changelog_31_35.bin" + params.extension); DB::KeeperLogStore changelog_reader2( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, + DB::FlushSettings(), + keeper_context); changelog_reader2.init(1, 0); EXPECT_EQ(changelog_reader2.size(), 11); EXPECT_EQ(changelog_reader2.last_entry()->get_term(), 7777); @@ -996,7 +1046,9 @@ TEST_P(CoordinationTest, ChangelogTestReadAfterBrokenTruncate2) setLogDirectory("./logs"); DB::KeeperLogStore changelog( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 20}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 20}, + DB::FlushSettings(), + keeper_context); changelog.init(1, 0); for (size_t i = 0; i < 35; ++i) @@ -1015,7 +1067,9 @@ TEST_P(CoordinationTest, ChangelogTestReadAfterBrokenTruncate2) plain_buf.truncate(30); DB::KeeperLogStore changelog_reader( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 20}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 20}, + DB::FlushSettings(), + keeper_context); changelog_reader.init(1, 0); EXPECT_EQ(changelog_reader.size(), 0); @@ -1031,7 +1085,9 @@ TEST_P(CoordinationTest, ChangelogTestReadAfterBrokenTruncate2) EXPECT_EQ(changelog_reader.last_entry()->get_term(), 7777); DB::KeeperLogStore changelog_reader2( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 1}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 1}, + DB::FlushSettings(), + keeper_context); changelog_reader2.init(1, 0); EXPECT_EQ(changelog_reader2.size(), 1); EXPECT_EQ(changelog_reader2.last_entry()->get_term(), 7777); @@ -1044,7 +1100,9 @@ TEST_P(CoordinationTest, ChangelogTestLostFiles) setLogDirectory("./logs"); DB::KeeperLogStore changelog( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 20}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 20}, + DB::FlushSettings(), + keeper_context); changelog.init(1, 0); for (size_t i = 0; i < 35; ++i) @@ -1061,7 +1119,9 @@ TEST_P(CoordinationTest, ChangelogTestLostFiles) fs::remove("./logs/changelog_1_20.bin" + params.extension); DB::KeeperLogStore changelog_reader( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 20}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 20}, + DB::FlushSettings(), + keeper_context); /// It should print error message, but still able to start changelog_reader.init(5, 0); assertBrokenLogRemoved("./logs", "changelog_21_40.bin" + params.extension); @@ -1074,7 +1134,9 @@ TEST_P(CoordinationTest, ChangelogTestLostFiles2) setLogDirectory("./logs"); DB::KeeperLogStore changelog( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 10}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 10}, + DB::FlushSettings(), + keeper_context); changelog.init(1, 0); for (size_t i = 0; i < 35; ++i) @@ -1095,7 +1157,9 @@ TEST_P(CoordinationTest, ChangelogTestLostFiles2) fs::remove("./logs/changelog_21_30.bin" + params.extension); DB::KeeperLogStore changelog_reader( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 10}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 10}, + DB::FlushSettings(), + keeper_context); /// It should print error message, but still able to start changelog_reader.init(5, 0); EXPECT_TRUE(fs::exists("./logs/changelog_1_10.bin" + params.extension)); @@ -1555,6 +1619,7 @@ void testLogAndStateMachine( DB::KeeperLogStore changelog( DB::LogFileSettings{ .force_sync = true, .compress_logs = enable_compression, .rotate_interval = settings->rotate_log_storage_interval}, + DB::FlushSettings(), keeper_context); changelog.init(state_machine->last_commit_index() + 1, settings->reserved_log_items); for (size_t i = 1; i < total_logs + 1; ++i) @@ -1599,6 +1664,7 @@ void testLogAndStateMachine( DB::KeeperLogStore restore_changelog( DB::LogFileSettings{ .force_sync = true, .compress_logs = enable_compression, .rotate_interval = settings->rotate_log_storage_interval}, + DB::FlushSettings(), keeper_context); restore_changelog.init(restore_machine->last_commit_index() + 1, settings->reserved_log_items); @@ -1851,7 +1917,9 @@ TEST_P(CoordinationTest, TestRotateIntervalChanges) setLogDirectory("./logs"); { DB::KeeperLogStore changelog( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, + DB::FlushSettings(), + keeper_context); changelog.init(0, 3); for (size_t i = 1; i < 55; ++i) @@ -1870,7 +1938,9 @@ TEST_P(CoordinationTest, TestRotateIntervalChanges) EXPECT_TRUE(fs::exists("./logs/changelog_1_100.bin" + params.extension)); DB::KeeperLogStore changelog_1( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 10}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 10}, + DB::FlushSettings(), + keeper_context); changelog_1.init(0, 50); for (size_t i = 0; i < 55; ++i) { @@ -1887,7 +1957,9 @@ TEST_P(CoordinationTest, TestRotateIntervalChanges) EXPECT_TRUE(fs::exists("./logs/changelog_101_110.bin" + params.extension)); DB::KeeperLogStore changelog_2( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 7}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 7}, + DB::FlushSettings(), + keeper_context); changelog_2.init(98, 55); for (size_t i = 0; i < 17; ++i) @@ -1911,7 +1983,9 @@ TEST_P(CoordinationTest, TestRotateIntervalChanges) EXPECT_TRUE(fs::exists("./logs/changelog_125_131.bin" + params.extension)); DB::KeeperLogStore changelog_3( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 5}, + DB::FlushSettings(), + keeper_context); changelog_3.init(116, 3); for (size_t i = 0; i < 17; ++i) { @@ -1961,7 +2035,9 @@ TEST_P(CoordinationTest, TestCompressedLogsMultipleRewrite) ChangelogDirTest logs("./logs"); setLogDirectory("./logs"); DB::KeeperLogStore changelog( - DB::LogFileSettings{.force_sync = true, .compress_logs = test_params.enable_compression, .rotate_interval = 100}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = test_params.enable_compression, .rotate_interval = 100}, + DB::FlushSettings(), + keeper_context); changelog.init(0, 3); for (size_t i = 1; i < 55; ++i) @@ -1976,7 +2052,9 @@ TEST_P(CoordinationTest, TestCompressedLogsMultipleRewrite) waitDurableLogs(changelog); DB::KeeperLogStore changelog1( - DB::LogFileSettings{.force_sync = true, .compress_logs = test_params.enable_compression, .rotate_interval = 100}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = test_params.enable_compression, .rotate_interval = 100}, + DB::FlushSettings(), + keeper_context); changelog1.init(0, 3); for (size_t i = 55; i < 70; ++i) { @@ -1990,7 +2068,9 @@ TEST_P(CoordinationTest, TestCompressedLogsMultipleRewrite) waitDurableLogs(changelog1); DB::KeeperLogStore changelog2( - DB::LogFileSettings{.force_sync = true, .compress_logs = test_params.enable_compression, .rotate_interval = 100}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = test_params.enable_compression, .rotate_interval = 100}, + DB::FlushSettings(), + keeper_context); changelog2.init(0, 3); for (size_t i = 70; i < 80; ++i) { @@ -2056,7 +2136,10 @@ TEST_P(CoordinationTest, ChangelogInsertThreeTimesSmooth) setLogDirectory("./logs"); { LOG_INFO(log, "================First time====================="); - DB::KeeperLogStore changelog(DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, keeper_context); + DB::KeeperLogStore changelog( + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, + DB::FlushSettings(), + keeper_context); changelog.init(1, 0); auto entry = getLogEntry("hello_world", 1000); changelog.append(entry); @@ -2068,7 +2151,9 @@ TEST_P(CoordinationTest, ChangelogInsertThreeTimesSmooth) { LOG_INFO(log, "================Second time====================="); DB::KeeperLogStore changelog( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, + DB::FlushSettings(), + keeper_context); changelog.init(1, 0); auto entry = getLogEntry("hello_world", 1000); changelog.append(entry); @@ -2080,7 +2165,9 @@ TEST_P(CoordinationTest, ChangelogInsertThreeTimesSmooth) { LOG_INFO(log, "================Third time====================="); DB::KeeperLogStore changelog( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, + DB::FlushSettings(), + keeper_context); changelog.init(1, 0); auto entry = getLogEntry("hello_world", 1000); changelog.append(entry); @@ -2092,7 +2179,9 @@ TEST_P(CoordinationTest, ChangelogInsertThreeTimesSmooth) { LOG_INFO(log, "================Fourth time====================="); DB::KeeperLogStore changelog( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, + DB::FlushSettings(), + keeper_context); changelog.init(1, 0); auto entry = getLogEntry("hello_world", 1000); changelog.append(entry); @@ -2112,7 +2201,9 @@ TEST_P(CoordinationTest, ChangelogInsertMultipleTimesSmooth) { LOG_INFO(log, "================First time====================="); DB::KeeperLogStore changelog( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, + DB::FlushSettings(), + keeper_context); changelog.init(1, 0); for (size_t j = 0; j < 7; ++j) { @@ -2124,7 +2215,9 @@ TEST_P(CoordinationTest, ChangelogInsertMultipleTimesSmooth) } DB::KeeperLogStore changelog( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, + DB::FlushSettings(), + keeper_context); changelog.init(1, 0); EXPECT_EQ(changelog.next_slot(), 36 * 7 + 1); } @@ -2137,7 +2230,9 @@ TEST_P(CoordinationTest, ChangelogInsertThreeTimesHard) { LOG_INFO(log, "================First time====================="); DB::KeeperLogStore changelog1( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, + DB::FlushSettings(), + keeper_context); changelog1.init(1, 0); auto entry = getLogEntry("hello_world", 1000); changelog1.append(entry); @@ -2149,7 +2244,9 @@ TEST_P(CoordinationTest, ChangelogInsertThreeTimesHard) { LOG_INFO(log, "================Second time====================="); DB::KeeperLogStore changelog2( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, + DB::FlushSettings(), + keeper_context); changelog2.init(1, 0); auto entry = getLogEntry("hello_world", 1000); changelog2.append(entry); @@ -2161,7 +2258,9 @@ TEST_P(CoordinationTest, ChangelogInsertThreeTimesHard) { LOG_INFO(log, "================Third time====================="); DB::KeeperLogStore changelog3( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, + DB::FlushSettings(), + keeper_context); changelog3.init(1, 0); auto entry = getLogEntry("hello_world", 1000); changelog3.append(entry); @@ -2173,7 +2272,9 @@ TEST_P(CoordinationTest, ChangelogInsertThreeTimesHard) { LOG_INFO(log, "================Fourth time====================="); DB::KeeperLogStore changelog4( - DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100}, + DB::FlushSettings(), + keeper_context); changelog4.init(1, 0); auto entry = getLogEntry("hello_world", 1000); changelog4.append(entry); @@ -2235,7 +2336,9 @@ TEST_P(CoordinationTest, TestLogGap) setLogDirectory("./logs"); DB::KeeperLogStore changelog( - DB::LogFileSettings{.force_sync = true, .compress_logs = test_params.enable_compression, .rotate_interval = 100}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = test_params.enable_compression, .rotate_interval = 100}, + DB::FlushSettings(), + keeper_context); changelog.init(0, 3); for (size_t i = 1; i < 55; ++i) @@ -2248,7 +2351,9 @@ TEST_P(CoordinationTest, TestLogGap) } DB::KeeperLogStore changelog1( - DB::LogFileSettings{.force_sync = true, .compress_logs = test_params.enable_compression, .rotate_interval = 100}, keeper_context); + DB::LogFileSettings{.force_sync = true, .compress_logs = test_params.enable_compression, .rotate_interval = 100}, + DB::FlushSettings(), + keeper_context); changelog1.init(61, 3); /// Logs discarded @@ -2597,6 +2702,7 @@ TEST_P(CoordinationTest, ChangelogTestMaxLogSize) DB::KeeperLogStore changelog( DB::LogFileSettings{ .force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 20, .max_size = 50 * 1024 * 1024}, + DB::FlushSettings(), keeper_context); changelog.init(1, 0); @@ -2616,6 +2722,7 @@ TEST_P(CoordinationTest, ChangelogTestMaxLogSize) DB::KeeperLogStore changelog( DB::LogFileSettings{ .force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100'000, .max_size = 4000}, + DB::FlushSettings(), keeper_context); changelog.init(1, 0); @@ -2637,6 +2744,7 @@ TEST_P(CoordinationTest, ChangelogTestMaxLogSize) DB::KeeperLogStore changelog( DB::LogFileSettings{ .force_sync = true, .compress_logs = params.enable_compression, .rotate_interval = 100'000, .max_size = 4000}, + DB::FlushSettings(), keeper_context); changelog.init(1, 0); ASSERT_EQ(changelog.entry_at(last_entry_index)->get_term(), (i - 1 + 44) * 10); diff --git a/utils/keeper-data-dumper/main.cpp b/utils/keeper-data-dumper/main.cpp index 9f089a26360..aa8c0efbb26 100644 --- a/utils/keeper-data-dumper/main.cpp +++ b/utils/keeper-data-dumper/main.cpp @@ -74,7 +74,9 @@ int main(int argc, char *argv[]) LOG_INFO(logger, "Last committed index: {}", last_commited_index); DB::KeeperLogStore changelog( - LogFileSettings{.force_sync = true, .compress_logs = settings->compress_logs, .rotate_interval = 10000000}, keeper_context); + LogFileSettings{.force_sync = true, .compress_logs = settings->compress_logs, .rotate_interval = 10000000}, + FlushSettings(), + keeper_context); changelog.init(last_commited_index, 10000000000UL); /// collect all logs if (changelog.size() == 0) LOG_INFO(logger, "Changelog empty"); From 59ca2374b583efdeaf81e30417b756213850fae8 Mon Sep 17 00:00:00 2001 From: avogar Date: Wed, 23 Aug 2023 19:39:45 +0000 Subject: [PATCH 073/433] Better --- src/Common/BoolArgsToTemplateArgsDispatcher.h | 39 -------------- src/Interpreters/Aggregator.cpp | 54 +++++++++++-------- src/Interpreters/Aggregator.h | 3 +- .../02845_group_by_constant_keys.reference | 36 +++++++++++++ .../02845_group_by_constant_keys.sql | 32 +++++++++-- 5 files changed, 98 insertions(+), 66 deletions(-) delete mode 100644 src/Common/BoolArgsToTemplateArgsDispatcher.h diff --git a/src/Common/BoolArgsToTemplateArgsDispatcher.h b/src/Common/BoolArgsToTemplateArgsDispatcher.h deleted file mode 100644 index 3e72db49cd2..00000000000 --- a/src/Common/BoolArgsToTemplateArgsDispatcher.h +++ /dev/null @@ -1,39 +0,0 @@ -#pragma once -#include - -namespace DB -{ - -/// Special struct that helps to convert bool variables to function template bool arguments. -/// It can be used to avoid multiple nested if/else on bool arguments. How to use it: -/// Imagine you have template function -/// template return_type foo(...); -/// and bool variables b1, b2, ..., bn. To pass these variables as template for foo you can do the following: -/// -/// auto call_foo = []() -/// { -/// return foo(...); -/// } -/// -/// BoolArgsToTemplateArgsDispatcher::call(call_foo, b1, b2, ..., bn); - -template -struct BoolArgsToTemplateArgsDispatcher -{ - template - static auto call(Functor f, Args1&&... args) - { - return f.template operator()(std::forward(args)...); - } - - template - static auto call(Functor f, bool b, Args1&&... ar1) - { - if (b) - return BoolArgsToTemplateArgsDispatcher::call(f, std::forward(ar1)...); - else - return BoolArgsToTemplateArgsDispatcher::call(f, std::forward(ar1)...); - } -}; - -} diff --git a/src/Interpreters/Aggregator.cpp b/src/Interpreters/Aggregator.cpp index d9e9c66e4e2..90f6fc7b978 100644 --- a/src/Interpreters/Aggregator.cpp +++ b/src/Interpreters/Aggregator.cpp @@ -39,7 +39,6 @@ #include #include #include -#include #include @@ -1066,30 +1065,40 @@ void NO_INLINE Aggregator::executeImpl( { typename Method::State state(key_columns, key_sizes, aggregation_state_cache); - auto call_execute_impl_batch = [&]() - { - executeImplBatch(method, state, aggregates_pool, row_begin, row_end, aggregate_instructions, overflow_row); - }; - - bool use_compiled_functions = false; -#if USE_EMBEDDED_COMPILER - use_compiled_functions = compiled_aggregate_functions_holder && !hasSparseArguments(aggregate_instructions); -#endif - if (!no_more_keys) { /// Prefetching doesn't make sense for small hash tables, because they fit in caches entirely. const bool prefetch = Method::State::has_cheap_key_calculation && params.enable_prefetch && (method.data.getBufferSizeInBytes() > min_bytes_for_prefetch); - BoolArgsToTemplateArgsDispatcher::call(call_execute_impl_batch, no_more_keys, use_compiled_functions, prefetch, all_keys_are_const); + +#if USE_EMBEDDED_COMPILER + if (compiled_aggregate_functions_holder && !hasSparseArguments(aggregate_instructions)) + { + if (prefetch) + executeImplBatch( + method, state, aggregates_pool, row_begin, row_end, aggregate_instructions, all_keys_are_const, overflow_row); + else + executeImplBatch( + method, state, aggregates_pool, row_begin, row_end, aggregate_instructions, all_keys_are_const, overflow_row); + } + else +#endif + { + if (prefetch) + executeImplBatch( + method, state, aggregates_pool, row_begin, row_end, aggregate_instructions, all_keys_are_const, overflow_row); + else + executeImplBatch( + method, state, aggregates_pool, row_begin, row_end, aggregate_instructions, all_keys_are_const, overflow_row); + } } else { - BoolArgsToTemplateArgsDispatcher::call(call_execute_impl_batch, no_more_keys, false, false, all_keys_are_const); + executeImplBatch(method, state, aggregates_pool, row_begin, row_end, aggregate_instructions, all_keys_are_const, overflow_row); } } -template +template void NO_INLINE Aggregator::executeImplBatch( Method & method, typename Method::State & state, @@ -1097,6 +1106,7 @@ void NO_INLINE Aggregator::executeImplBatch( size_t row_begin, size_t row_end, AggregateFunctionInstruction * aggregate_instructions, + bool all_keys_are_const, AggregateDataPtr overflow_row) const { using KeyHolder = decltype(state.getKeyHolder(0, std::declval())); @@ -1113,7 +1123,7 @@ void NO_INLINE Aggregator::executeImplBatch( /// For all rows. AggregateDataPtr place = aggregates_pool->alloc(0); - if constexpr (all_keys_are_const) + if (all_keys_are_const) { state.emplaceKey(method.data, 0, *aggregates_pool).setMapped(place); } @@ -1140,7 +1150,7 @@ void NO_INLINE Aggregator::executeImplBatch( } /// Optimization for special case when aggregating by 8bit key. - if constexpr (!no_more_keys && !all_keys_are_const && std::is_same_v) + if constexpr (!no_more_keys && std::is_same_v) { /// We use another method if there are aggregate functions with -Array combinator. bool has_arrays = false; @@ -1153,7 +1163,7 @@ void NO_INLINE Aggregator::executeImplBatch( } } - if (!has_arrays && !hasSparseArguments(aggregate_instructions)) + if (!has_arrays && !hasSparseArguments(aggregate_instructions) && !all_keys_are_const) { for (AggregateFunctionInstruction * inst = aggregate_instructions; inst->that; ++inst) { @@ -1184,7 +1194,7 @@ void NO_INLINE Aggregator::executeImplBatch( /// For all rows. size_t start, end; /// If all keys are const, key columns contain only 1 row. - if constexpr (all_keys_are_const) + if (all_keys_are_const) { start = 0; end = 1; @@ -1201,7 +1211,7 @@ void NO_INLINE Aggregator::executeImplBatch( if constexpr (!no_more_keys) { - if constexpr (prefetch && !all_keys_are_const && HasPrefetchMemberFunc) + if constexpr (prefetch && HasPrefetchMemberFunc) { if (i == row_begin + prefetching.iterationsToMeasure()) prefetch_look_ahead = prefetching.calcPrefetchLookAhead(); @@ -1273,7 +1283,7 @@ void NO_INLINE Aggregator::executeImplBatch( { /// If all keys are constant and this is new key /// we don't need to do anything and just skip the whole block. - if constexpr (all_keys_are_const) + if (all_keys_are_const) return; aggregate_data = overflow_row; } @@ -1299,7 +1309,7 @@ void NO_INLINE Aggregator::executeImplBatch( columns_data.emplace_back(getColumnData(inst->batch_arguments[argument_index])); } - if constexpr (all_keys_are_const) + if (all_keys_are_const) { auto add_into_aggregate_states_function_single_place = compiled_aggregate_functions_holder->compiled_aggregate_functions.add_into_aggregate_states_function_single_place; add_into_aggregate_states_function_single_place(row_begin, row_end, columns_data.data(), places[0]); @@ -1323,7 +1333,7 @@ void NO_INLINE Aggregator::executeImplBatch( AggregateFunctionInstruction * inst = aggregate_instructions + i; - if constexpr (all_keys_are_const) + if (all_keys_are_const) { if (inst->offsets) inst->batch_that->addBatchSinglePlace(inst->offsets[static_cast(row_begin) - 1], inst->offsets[row_end - 1], places[0] + inst->state_offset, inst->batch_arguments, aggregates_pool); diff --git a/src/Interpreters/Aggregator.h b/src/Interpreters/Aggregator.h index dc0391c4289..d5b5abddd84 100644 --- a/src/Interpreters/Aggregator.h +++ b/src/Interpreters/Aggregator.h @@ -1297,7 +1297,7 @@ private: AggregateDataPtr overflow_row) const; /// Specialization for a particular value no_more_keys. - template + template void executeImplBatch( Method & method, typename Method::State & state, @@ -1305,6 +1305,7 @@ private: size_t row_begin, size_t row_end, AggregateFunctionInstruction * aggregate_instructions, + bool all_keys_are_const, AggregateDataPtr overflow_row) const; /// For case when there are no keys (all aggregate into one row). diff --git a/tests/queries/0_stateless/02845_group_by_constant_keys.reference b/tests/queries/0_stateless/02845_group_by_constant_keys.reference index 60e40ea54a7..67cbdf0c025 100644 --- a/tests/queries/0_stateless/02845_group_by_constant_keys.reference +++ b/tests/queries/0_stateless/02845_group_by_constant_keys.reference @@ -2,3 +2,39 @@ 10000000 1 2 3 10000000 1 2 3 10000000 1 2 3 +10 +10 +10 +10 +10 +10 +10 +10 +10 +10 +10 +10 +10 data.Parquet +10 data.2.Parquet +10 data.1.Parquet +10 data.Parquet +10 data.2.Parquet +10 data.1.Parquet +10 data.Parquet +10 data.2.Parquet +10 data.1.Parquet +10 data.Parquet +10 data.2.Parquet +10 data.1.Parquet +10 +10 +10 +10 +10 +10 +10 +10 +10 +10 +10 +10 diff --git a/tests/queries/0_stateless/02845_group_by_constant_keys.sql b/tests/queries/0_stateless/02845_group_by_constant_keys.sql index 0223cf1df60..109a5a9a730 100644 --- a/tests/queries/0_stateless/02845_group_by_constant_keys.sql +++ b/tests/queries/0_stateless/02845_group_by_constant_keys.sql @@ -1,5 +1,29 @@ -SELECT count(number), 1 AS k1, 2 as k2, 3 as k3 FROM numbers_mt(10000000) GROUP BY k1, k2, k3 settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=0, compile_aggregate_expressions=0; -SELECT count(number), 1 AS k1, 2 as k2, 3 as k3 FROM numbers_mt(10000000) GROUP BY k1, k2, k3 settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=1, compile_aggregate_expressions = 0; -SELECT count(number), 1 AS k1, 2 as k2, 3 as k3 FROM numbers_mt(10000000) GROUP BY k1, k2, k3 settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=0, compile_aggregate_expressions = 1; -SELECT count(number), 1 AS k1, 2 as k2, 3 as k3 FROM numbers_mt(10000000) GROUP BY k1, k2, k3 settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=1, compile_aggregate_expressions = 1; +select count(number), 1 AS k1, 2 as k2, 3 as k3 from numbers_mt(10000000) group by k1, k2, k3 settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=0, compile_aggregate_expressions=0; +select count(number), 1 AS k1, 2 as k2, 3 as k3 from numbers_mt(10000000) group by k1, k2, k3 settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=1, compile_aggregate_expressions = 0; +select count(number), 1 AS k1, 2 as k2, 3 as k3 from numbers_mt(10000000) group by k1, k2, k3 settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=0, compile_aggregate_expressions = 1; +select count(number), 1 AS k1, 2 as k2, 3 as k3 from numbers_mt(10000000) group by k1, k2, k3 settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=1, compile_aggregate_expressions = 1; + +drop table if exists test; +create table test (x UInt64) engine=File(Parquet); +set engine_file_allow_create_multiple_files = 1; +insert into test select * from numbers(10); +insert into test select * from numbers(10); +insert into test select * from numbers(10); + +select count() from test group by _file settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=0, compile_aggregate_expressions=0; +select count() from test group by _file settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=1, compile_aggregate_expressions=0; +select count() from test group by _file settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=0, compile_aggregate_expressions=1; +select count() from test group by _file settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=1, compile_aggregate_expressions=1; + +select count(), _file from test group by _file settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=0, compile_aggregate_expressions=0; +select count(), _file from test group by _file settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=1, compile_aggregate_expressions=0; +select count(), _file from test group by _file settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=0, compile_aggregate_expressions=1; +select count(), _file from test group by _file settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=1, compile_aggregate_expressions=1; + +select count() from test group by _file, _path settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=0, compile_aggregate_expressions=0; +select count() from test group by _file, _path settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=1, compile_aggregate_expressions=0; +select count() from test group by _file, _path settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=0, compile_aggregate_expressions=1; +select count() from test group by _file, _path settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=1, compile_aggregate_expressions=1; + +drop table test; From 27fb1b5ced804a37fee6c9fd110e029bb6cc01c5 Mon Sep 17 00:00:00 2001 From: Kruglov Pavel <48961922+Avogar@users.noreply.github.com> Date: Wed, 23 Aug 2023 22:18:30 +0200 Subject: [PATCH 074/433] Fix test --- tests/queries/0_stateless/02845_group_by_constant_keys.sql | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tests/queries/0_stateless/02845_group_by_constant_keys.sql b/tests/queries/0_stateless/02845_group_by_constant_keys.sql index 109a5a9a730..e16fd3fd1d2 100644 --- a/tests/queries/0_stateless/02845_group_by_constant_keys.sql +++ b/tests/queries/0_stateless/02845_group_by_constant_keys.sql @@ -4,7 +4,7 @@ select count(number), 1 AS k1, 2 as k2, 3 as k3 from numbers_mt(10000000) group select count(number), 1 AS k1, 2 as k2, 3 as k3 from numbers_mt(10000000) group by k1, k2, k3 settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=1, compile_aggregate_expressions = 1; drop table if exists test; -create table test (x UInt64) engine=File(Parquet); +create table test (x UInt64) engine=File(JSON); set engine_file_allow_create_multiple_files = 1; insert into test select * from numbers(10); insert into test select * from numbers(10); From 5154c1b9d0eb6fd484d2ca1f7920747bfc3de84e Mon Sep 17 00:00:00 2001 From: Vitaly Baranov Date: Wed, 23 Aug 2023 22:55:46 +0200 Subject: [PATCH 075/433] Use same UUIDs for tables on replicas created by RESTORE ON CLUSTER. --- src/Backups/BackupUtils.cpp | 25 ++++++++ src/Backups/BackupUtils.h | 6 +- src/Backups/DDLAdjustingForBackupVisitor.cpp | 3 - src/Backups/IRestoreCoordination.h | 6 +- src/Backups/RestoreCoordinationLocal.cpp | 35 ++++++++++ src/Backups/RestoreCoordinationLocal.h | 7 ++ src/Backups/RestoreCoordinationRemote.cpp | 40 ++++++++++++ src/Backups/RestoreCoordinationRemote.h | 4 ++ src/Backups/RestorerFromBackup.cpp | 67 ++++++++++---------- src/Interpreters/InterpreterCreateQuery.cpp | 29 +++------ src/Parsers/ASTCreateQuery.cpp | 44 +++++++++++++ src/Parsers/ASTCreateQuery.h | 12 ++++ 12 files changed, 221 insertions(+), 57 deletions(-) diff --git a/src/Backups/BackupUtils.cpp b/src/Backups/BackupUtils.cpp index 89b75a103c2..6efca053f05 100644 --- a/src/Backups/BackupUtils.cpp +++ b/src/Backups/BackupUtils.cpp @@ -1,6 +1,9 @@ #include +#include #include #include +#include +#include #include #include @@ -95,4 +98,26 @@ AccessRightsElements getRequiredAccessToBackup(const ASTBackupQuery::Elements & return required_access; } +bool compareRestoredTableDef(const IAST & restored_table_create_query, const IAST & create_query_from_backup, const ContextPtr & global_context) +{ + auto adjust_before_comparison = [&](const IAST & query) -> ASTPtr + { + auto new_query = query.clone(); + adjustCreateQueryForBackup(new_query, global_context, nullptr); + ASTCreateQuery & create = typeid_cast(*new_query); + create.setUUID({}); + create.if_not_exists = false; + return new_query; + }; + + ASTPtr query1 = adjust_before_comparison(restored_table_create_query); + ASTPtr query2 = adjust_before_comparison(create_query_from_backup); + return serializeAST(*query1) == serializeAST(*query2); +} + +bool compareRestoredDatabaseDef(const IAST & restored_database_create_query, const IAST & create_query_from_backup, const ContextPtr & global_context) +{ + return compareRestoredTableDef(restored_database_create_query, create_query_from_backup, global_context); +} + } diff --git a/src/Backups/BackupUtils.h b/src/Backups/BackupUtils.h index 3dc0a58d304..7976de818e2 100644 --- a/src/Backups/BackupUtils.h +++ b/src/Backups/BackupUtils.h @@ -1,6 +1,7 @@ #pragma once #include +#include namespace DB @@ -12,8 +13,11 @@ class DDLRenamingMap; /// Initializes a DDLRenamingMap from a BACKUP or RESTORE query. DDLRenamingMap makeRenamingMapFromBackupQuery(const ASTBackupQuery::Elements & elements); - /// Returns access required to execute BACKUP query. AccessRightsElements getRequiredAccessToBackup(const ASTBackupQuery::Elements & elements); +/// Checks the definition of a restored table - it must correspond to the definition from the backup. +bool compareRestoredTableDef(const IAST & restored_table_create_query, const IAST & create_query_from_backup, const ContextPtr & global_context); +bool compareRestoredDatabaseDef(const IAST & restored_database_create_query, const IAST & create_query_from_backup, const ContextPtr & global_context); + } diff --git a/src/Backups/DDLAdjustingForBackupVisitor.cpp b/src/Backups/DDLAdjustingForBackupVisitor.cpp index 8223e08f127..0bff3cc9f4e 100644 --- a/src/Backups/DDLAdjustingForBackupVisitor.cpp +++ b/src/Backups/DDLAdjustingForBackupVisitor.cpp @@ -81,9 +81,6 @@ namespace void visitCreateQuery(ASTCreateQuery & create, const DDLAdjustingForBackupVisitor::Data & data) { - create.uuid = UUIDHelpers::Nil; - create.to_inner_uuid = UUIDHelpers::Nil; - if (create.storage) visitStorage(*create.storage, data); } diff --git a/src/Backups/IRestoreCoordination.h b/src/Backups/IRestoreCoordination.h index 2f9e8d171f6..fd6f014c326 100644 --- a/src/Backups/IRestoreCoordination.h +++ b/src/Backups/IRestoreCoordination.h @@ -7,6 +7,7 @@ namespace DB { class Exception; enum class UserDefinedSQLObjectType; +class ASTCreateQuery; /// Replicas use this class to coordinate what they're reading from a backup while executing RESTORE ON CLUSTER. /// There are two implementation of this interface: RestoreCoordinationLocal and RestoreCoordinationRemote. @@ -40,10 +41,13 @@ public: /// The function returns false if user-defined function at a specified zk path are being already restored by another replica. virtual bool acquireReplicatedSQLObjects(const String & loader_zk_path, UserDefinedSQLObjectType object_type) = 0; + /// Generates a new UUID for a table. The same UUID must be used for a replicated table on each replica, + /// (because otherwise the macro "{uuid}" in the ZooKeeper path will not work correctly). + virtual void generateUUIDForTable(ASTCreateQuery & create_query) = 0; + /// This function is used to check if concurrent restores are running /// other than the restore passed to the function virtual bool hasConcurrentRestores(const std::atomic & num_active_restores) const = 0; - }; } diff --git a/src/Backups/RestoreCoordinationLocal.cpp b/src/Backups/RestoreCoordinationLocal.cpp index 068c4fe7e52..6e9a05e06c2 100644 --- a/src/Backups/RestoreCoordinationLocal.cpp +++ b/src/Backups/RestoreCoordinationLocal.cpp @@ -1,4 +1,5 @@ #include +#include #include @@ -51,6 +52,40 @@ bool RestoreCoordinationLocal::acquireReplicatedSQLObjects(const String &, UserD return true; } +void RestoreCoordinationLocal::generateUUIDForTable(ASTCreateQuery & create_query) +{ + String query_str = serializeAST(create_query); + + auto find_in_map = [&] + { + auto it = create_query_uuids.find(query_str); + if (it != create_query_uuids.end()) + { + create_query.setUUID(it->second); + return true; + } + return false; + }; + + { + std::lock_guard lock{mutex}; + if (find_in_map()) + return; + } + + create_query.setUUID({}); + auto new_uuids = create_query.generateRandomUUID(); + String new_query_str = serializeAST(create_query); + + { + std::lock_guard lock{mutex}; + if (find_in_map()) + return; + create_query_uuids[query_str] = new_uuids; + create_query_uuids[new_query_str] = new_uuids; + } +} + bool RestoreCoordinationLocal::hasConcurrentRestores(const std::atomic & num_active_restores) const { if (num_active_restores > 1) diff --git a/src/Backups/RestoreCoordinationLocal.h b/src/Backups/RestoreCoordinationLocal.h index e27f0d1ef88..339b754fca5 100644 --- a/src/Backups/RestoreCoordinationLocal.h +++ b/src/Backups/RestoreCoordinationLocal.h @@ -1,6 +1,7 @@ #pragma once #include +#include #include #include #include @@ -39,6 +40,10 @@ public: /// The function returns false if user-defined function at a specified zk path are being already restored by another replica. bool acquireReplicatedSQLObjects(const String & loader_zk_path, UserDefinedSQLObjectType object_type) override; + /// Generates a new UUID for a table. The same UUID must be used for a replicated table on each replica, + /// (because otherwise the macro "{uuid}" in the ZooKeeper path will not work correctly). + void generateUUIDForTable(ASTCreateQuery & create_query) override; + bool hasConcurrentRestores(const std::atomic & num_active_restores) const override; private: @@ -46,6 +51,8 @@ private: std::set> acquired_tables_in_replicated_databases; std::unordered_set acquired_data_in_replicated_tables; + std::unordered_map create_query_uuids; + mutable std::mutex mutex; }; diff --git a/src/Backups/RestoreCoordinationRemote.cpp b/src/Backups/RestoreCoordinationRemote.cpp index 37abebb26b7..077921f4d32 100644 --- a/src/Backups/RestoreCoordinationRemote.cpp +++ b/src/Backups/RestoreCoordinationRemote.cpp @@ -2,6 +2,8 @@ #include #include #include +#include +#include #include #include #include @@ -87,6 +89,7 @@ void RestoreCoordinationRemote::createRootNodes() ops.emplace_back(zkutil::makeCreateRequest(zookeeper_path + "/repl_tables_data_acquired", "", zkutil::CreateMode::Persistent)); ops.emplace_back(zkutil::makeCreateRequest(zookeeper_path + "/repl_access_storages_acquired", "", zkutil::CreateMode::Persistent)); ops.emplace_back(zkutil::makeCreateRequest(zookeeper_path + "/repl_sql_objects_acquired", "", zkutil::CreateMode::Persistent)); + ops.emplace_back(zkutil::makeCreateRequest(zookeeper_path + "/table_uuids", "", zkutil::CreateMode::Persistent)); zk->tryMulti(ops, responses); }); } @@ -231,6 +234,43 @@ bool RestoreCoordinationRemote::acquireReplicatedSQLObjects(const String & loade return result; } +void RestoreCoordinationRemote::generateUUIDForTable(ASTCreateQuery & create_query) +{ + String query_str = serializeAST(create_query); + + create_query.setUUID({}); + String new_uuids_str = create_query.generateRandomUUID().toString(); + String new_query_str = serializeAST(create_query); + + auto holder = with_retries.createRetriesControlHolder("generateUUIDForTable"); + holder.retries_ctl.retryLoop( + [&, &zk = holder.faulty_zookeeper]() + { + with_retries.renewZooKeeper(zk); + + String path = zookeeper_path + "/table_uuids/" + escapeForFileName(query_str); + String path2 = zookeeper_path + "/table_uuids/" + escapeForFileName(new_query_str); + + Coordination::Requests ops; + ops.emplace_back(zkutil::makeCreateRequest(path, new_uuids_str, zkutil::CreateMode::Persistent)); + ops.emplace_back(zkutil::makeCreateRequest(path2, new_uuids_str, zkutil::CreateMode::Persistent)); + + Coordination::Responses responses; + Coordination::Error res = zk->tryMulti(ops, responses); + + if (res == Coordination::Error::ZOK) + return; + + if ((res == Coordination::Error::ZNODEEXISTS) && (responses[0]->error == Coordination::Error::ZNODEEXISTS)) + { + create_query.setUUID(ASTCreateQuery::UUIDs::fromString(zk->get(path))); + return; + } + + zkutil::KeeperMultiException::check(res, ops, responses); + }); +} + void RestoreCoordinationRemote::removeAllNodes() { /// Usually this function is called by the initiator when a restore operation is complete so we don't need the coordination anymore. diff --git a/src/Backups/RestoreCoordinationRemote.h b/src/Backups/RestoreCoordinationRemote.h index eb0fcff9c2d..22d0c0ed6df 100644 --- a/src/Backups/RestoreCoordinationRemote.h +++ b/src/Backups/RestoreCoordinationRemote.h @@ -46,6 +46,10 @@ public: /// The function returns false if user-defined function at a specified zk path are being already restored by another replica. bool acquireReplicatedSQLObjects(const String & loader_zk_path, UserDefinedSQLObjectType object_type) override; + /// Generates a new UUID for a table. The same UUID must be used for a replicated table on each replica, + /// (because otherwise the macro "{uuid}" in the ZooKeeper path will not work correctly). + void generateUUIDForTable(ASTCreateQuery & create_query) override; + bool hasConcurrentRestores(const std::atomic & num_active_restores) const override; private: diff --git a/src/Backups/RestorerFromBackup.cpp b/src/Backups/RestorerFromBackup.cpp index 8a0908c6b67..026671edd6a 100644 --- a/src/Backups/RestorerFromBackup.cpp +++ b/src/Backups/RestorerFromBackup.cpp @@ -571,12 +571,14 @@ void RestorerFromBackup::createDatabase(const String & database_name) const if (database_info.is_predefined_database) return; - auto create_database_query = database_info.create_database_query; - if (restore_settings.create_table == RestoreTableCreationMode::kCreateIfNotExists) - { - create_database_query = create_database_query->clone(); - create_database_query->as().if_not_exists = true; - } + auto create_database_query = typeid_cast>(database_info.create_database_query->clone()); + + /// Generate a new UUID for a database. + /// The generated UUID will be ignored if the database does not support UUIDs. + restore_coordination->generateUUIDForTable(*create_database_query); + + /// Add the clause `IF NOT EXISTS` if that is specified in the restore settings. + create_database_query->if_not_exists = (restore_settings.create_table == RestoreTableCreationMode::kCreateIfNotExists); LOG_TRACE(log, "Creating database {}: {}", backQuoteIfNeed(database_name), serializeAST(*create_database_query)); @@ -605,17 +607,17 @@ void RestorerFromBackup::checkDatabase(const String & database_name) if (!restore_settings.allow_different_database_def && !database_info.is_predefined_database) { /// Check that the database's definition is the same as expected. - ASTPtr create_database_query = database->getCreateDatabaseQuery(); - adjustCreateQueryForBackup(create_database_query, context->getGlobalContext(), nullptr); - ASTPtr expected_create_query = database_info.create_database_query; - if (serializeAST(*create_database_query) != serializeAST(*expected_create_query)) + + ASTPtr existing_database_def = database->getCreateDatabaseQuery(); + ASTPtr database_def_from_backup = database_info.create_database_query; + if (!compareRestoredDatabaseDef(*existing_database_def, *database_def_from_backup, context->getGlobalContext())) { throw Exception( ErrorCodes::CANNOT_RESTORE_DATABASE, "The database has a different definition: {} " "comparing to its definition in the backup: {}", - serializeAST(*create_database_query), - serializeAST(*expected_create_query)); + serializeAST(*existing_database_def), + serializeAST(*database_def_from_backup)); } } } @@ -714,20 +716,23 @@ void RestorerFromBackup::createTable(const QualifiedTableName & table_name) if (table_info.is_predefined_table) return; - auto create_table_query = table_info.create_table_query; - if (restore_settings.create_table == RestoreTableCreationMode::kCreateIfNotExists) - { - create_table_query = create_table_query->clone(); - create_table_query->as().if_not_exists = true; - } + auto create_table_query = typeid_cast>(table_info.create_table_query->clone()); + + /// Generate a new UUID for a table (the same table on different hosts must use the same UUID, `restore_coordination` will make it so). + /// The generated UUID will be ignored if the database does not support UUIDs. + restore_coordination->generateUUIDForTable(*create_table_query); + + /// Add the clause `IF NOT EXISTS` if that is specified in the restore settings. + create_table_query->if_not_exists = (restore_settings.create_table == RestoreTableCreationMode::kCreateIfNotExists); LOG_TRACE( log, "Creating {}: {}", tableNameWithTypeToString(table_name.database, table_name.table, false), serializeAST(*create_table_query)); try { - DatabasePtr database = DatabaseCatalog::instance().getDatabase(table_name.database); - table_info.database = database; + if (!table_info.database) + table_info.database = DatabaseCatalog::instance().getDatabase(table_name.database); + DatabasePtr database = table_info.database; /// Execute CREATE TABLE query (we call IDatabase::createTableRestoredFromBackup() to allow the database to do some /// database-specific things). @@ -747,37 +752,33 @@ void RestorerFromBackup::createTable(const QualifiedTableName & table_name) void RestorerFromBackup::checkTable(const QualifiedTableName & table_name) { auto & table_info = table_infos.at(table_name); - auto database = table_info.database; try { - if (!database) - { - database = DatabaseCatalog::instance().getDatabase(table_name.database); - table_info.database = database; - } - auto resolved_id = (table_name.database == DatabaseCatalog::TEMPORARY_DATABASE) ? context->resolveStorageID(StorageID{"", table_name.table}, Context::ResolveExternal) : context->resolveStorageID(StorageID{table_name.database, table_name.table}, Context::ResolveGlobal); + if (!table_info.database) + table_info.database = DatabaseCatalog::instance().getDatabase(table_name.database); + DatabasePtr database = table_info.database; + StoragePtr storage = database->getTable(resolved_id.table_name, context); table_info.storage = storage; table_info.table_lock = storage->lockForShare(context->getInitialQueryId(), context->getSettingsRef().lock_acquire_timeout); if (!restore_settings.allow_different_table_def && !table_info.is_predefined_table) { - ASTPtr create_table_query = database->getCreateTableQuery(resolved_id.table_name, context); - adjustCreateQueryForBackup(create_table_query, context->getGlobalContext(), nullptr); - ASTPtr expected_create_query = table_info.create_table_query; - if (serializeAST(*create_table_query) != serializeAST(*expected_create_query)) + ASTPtr existing_table_def = database->getCreateTableQuery(resolved_id.table_name, context); + ASTPtr table_def_from_backup = table_info.create_table_query; + if (!compareRestoredTableDef(*existing_table_def, *table_def_from_backup, context->getGlobalContext())) { throw Exception( ErrorCodes::CANNOT_RESTORE_TABLE, "The table has a different definition: {} " "comparing to its definition in the backup: {}", - serializeAST(*create_table_query), - serializeAST(*expected_create_query)); + serializeAST(*existing_table_def), + serializeAST(*table_def_from_backup)); } } } diff --git a/src/Interpreters/InterpreterCreateQuery.cpp b/src/Interpreters/InterpreterCreateQuery.cpp index 92d74f4f18a..6da99a79f91 100644 --- a/src/Interpreters/InterpreterCreateQuery.cpp +++ b/src/Interpreters/InterpreterCreateQuery.cpp @@ -219,10 +219,12 @@ BlockIO InterpreterCreateQuery::createDatabase(ASTCreateQuery & create) else { bool is_on_cluster = getContext()->getClientInfo().query_kind == ClientInfo::QueryKind::SECONDARY_QUERY; - if (create.uuid != UUIDHelpers::Nil && !is_on_cluster) + if (create.uuid != UUIDHelpers::Nil && !is_on_cluster && !internal) throw Exception(ErrorCodes::INCORRECT_QUERY, "Ordinary database engine does not support UUID"); - /// Ignore UUID if it's ON CLUSTER query + /// The database doesn't support UUID so we'll ignore it. The UUID could be set here because of either + /// a) the initiator of `ON CLUSTER` query generated it to ensure the same UUIDs are used on different hosts; or + /// b) `RESTORE from backup` query generated it to ensure the same UUIDs are used on different hosts. create.uuid = UUIDHelpers::Nil; metadata_path = metadata_path / "metadata" / database_name_escaped; } @@ -983,19 +985,6 @@ void InterpreterCreateQuery::setEngine(ASTCreateQuery & create) const setDefaultTableEngine(*create.storage, getContext()->getSettingsRef().default_table_engine.value); } -static void generateUUIDForTable(ASTCreateQuery & create) -{ - if (create.uuid == UUIDHelpers::Nil) - create.uuid = UUIDHelpers::generateV4(); - - /// If destination table (to_table_id) is not specified for materialized view, - /// then MV will create inner table. We should generate UUID of inner table here, - /// so it will be the same on all hosts if query in ON CLUSTER or database engine is Replicated. - bool need_uuid_for_inner_table = !create.attach && create.is_materialized_view && !create.to_table_id; - if (need_uuid_for_inner_table && create.to_inner_uuid == UUIDHelpers::Nil) - create.to_inner_uuid = UUIDHelpers::generateV4(); -} - void InterpreterCreateQuery::assertOrSetUUID(ASTCreateQuery & create, const DatabasePtr & database) const { const auto * kind = create.is_dictionary ? "Dictionary" : "Table"; @@ -1028,17 +1017,19 @@ void InterpreterCreateQuery::assertOrSetUUID(ASTCreateQuery & create, const Data kind_upper, create.table); } - generateUUIDForTable(create); + create.generateRandomUUID(); } else { bool is_on_cluster = getContext()->getClientInfo().query_kind == ClientInfo::QueryKind::SECONDARY_QUERY; bool has_uuid = create.uuid != UUIDHelpers::Nil || create.to_inner_uuid != UUIDHelpers::Nil; - if (has_uuid && !is_on_cluster) + if (has_uuid && !is_on_cluster && !internal) throw Exception(ErrorCodes::INCORRECT_QUERY, "{} UUID specified, but engine of database {} is not Atomic", kind, create.getDatabase()); - /// Ignore UUID if it's ON CLUSTER query + /// The database doesn't support UUID so we'll ignore it. The UUID could be set here because of either + /// a) the initiator of `ON CLUSTER` query generated it to ensure the same UUIDs are used on different hosts; or + /// b) `RESTORE from backup` query generated it to ensure the same UUIDs are used on different hosts. create.uuid = UUIDHelpers::Nil; create.to_inner_uuid = UUIDHelpers::Nil; } @@ -1619,7 +1610,7 @@ void InterpreterCreateQuery::prepareOnClusterQuery(ASTCreateQuery & create, Cont /// For CREATE query generate UUID on initiator, so it will be the same on all hosts. /// It will be ignored if database does not support UUIDs. - generateUUIDForTable(create); + create.generateRandomUUID(); /// For cross-replication cluster we cannot use UUID in replica path. String cluster_name_expanded = local_context->getMacros()->expand(cluster_name); diff --git a/src/Parsers/ASTCreateQuery.cpp b/src/Parsers/ASTCreateQuery.cpp index 196681a8801..bb0e4008491 100644 --- a/src/Parsers/ASTCreateQuery.cpp +++ b/src/Parsers/ASTCreateQuery.cpp @@ -6,6 +6,8 @@ #include #include #include +#include +#include namespace DB @@ -460,4 +462,46 @@ bool ASTCreateQuery::isParameterizedView() const return false; } + +ASTCreateQuery::UUIDs::UUIDs(const ASTCreateQuery & query) + : uuid(query.uuid) + , to_inner_uuid(query.to_inner_uuid) +{ +} + +String ASTCreateQuery::UUIDs::toString() const +{ + WriteBufferFromOwnString out; + out << "{" << uuid << "," << to_inner_uuid << "}"; + return out.str(); +} + +ASTCreateQuery::UUIDs ASTCreateQuery::UUIDs::fromString(const String & str) +{ + ReadBufferFromString in{str}; + ASTCreateQuery::UUIDs res; + in >> "{" >> res.uuid >> "," >> res.to_inner_uuid >> "}"; + return res; +} + +ASTCreateQuery::UUIDs ASTCreateQuery::generateRandomUUID() +{ + if (uuid == UUIDHelpers::Nil) + uuid = UUIDHelpers::generateV4(); + + /// If destination table (to_table_id) is not specified for materialized view, + /// then MV will create inner table. We should generate UUID of inner table here. + bool need_uuid_for_inner_table = !attach && is_materialized_view && !to_table_id; + if (need_uuid_for_inner_table && (to_inner_uuid == UUIDHelpers::Nil)) + to_inner_uuid = UUIDHelpers::generateV4(); + + return UUIDs{*this}; +} + +void ASTCreateQuery::setUUID(const UUIDs & uuids) +{ + uuid = uuids.uuid; + to_inner_uuid = uuids.to_inner_uuid; +} + } diff --git a/src/Parsers/ASTCreateQuery.h b/src/Parsers/ASTCreateQuery.h index ae45a244a03..25d910913a4 100644 --- a/src/Parsers/ASTCreateQuery.h +++ b/src/Parsers/ASTCreateQuery.h @@ -146,6 +146,18 @@ public: QueryKind getQueryKind() const override { return QueryKind::Create; } + struct UUIDs + { + UUID uuid = UUIDHelpers::Nil; + UUID to_inner_uuid = UUIDHelpers::Nil; + UUIDs() = default; + explicit UUIDs(const ASTCreateQuery & query); + String toString() const; + static UUIDs fromString(const String & str); + }; + UUIDs generateRandomUUID(); + void setUUID(const UUIDs & uuids); + protected: void formatQueryImpl(const FormatSettings & settings, FormatState & state, FormatStateStacked frame) const override; From 7b9d6c4b9645fa3a9c726dd967ad71236c86c5e5 Mon Sep 17 00:00:00 2001 From: Vitaly Baranov Date: Wed, 23 Aug 2023 22:56:01 +0200 Subject: [PATCH 076/433] Add test. --- .../test_backup_restore_on_cluster/test.py | 31 +++++++++++++++++++ 1 file changed, 31 insertions(+) diff --git a/tests/integration/test_backup_restore_on_cluster/test.py b/tests/integration/test_backup_restore_on_cluster/test.py index a9ebbeb66c2..dfce2f15413 100644 --- a/tests/integration/test_backup_restore_on_cluster/test.py +++ b/tests/integration/test_backup_restore_on_cluster/test.py @@ -276,6 +276,37 @@ def test_table_with_parts_in_queue_considered_non_empty(): ) +def test_replicated_table_with_uuid_in_zkpath(): + node1.query( + "CREATE TABLE tbl ON CLUSTER 'cluster' (" + "x UInt8, y String" + ") ENGINE=ReplicatedMergeTree('/clickhouse/tables/{uuid}','{replica}')" + "ORDER BY x" + ) + + node1.query("INSERT INTO tbl VALUES (1, 'AA')") + node2.query("INSERT INTO tbl VALUES (2, 'BB')") + + backup_name = new_backup_name() + node1.query(f"BACKUP TABLE tbl ON CLUSTER 'cluster' TO {backup_name}") + + # The table `tbl2` is expected to have a different UUID so it's ok to have both `tbl` and `tbl2` at the same time. + node2.query(f"RESTORE TABLE tbl AS tbl2 ON CLUSTER 'cluster' FROM {backup_name}") + + node1.query("INSERT INTO tbl2 VALUES (3, 'CC')") + + node1.query("SYSTEM SYNC REPLICA ON CLUSTER 'cluster' tbl") + node1.query("SYSTEM SYNC REPLICA ON CLUSTER 'cluster' tbl2") + + for instance in [node1, node2]: + assert instance.query("SELECT * FROM tbl ORDER BY x") == TSV( + [[1, "AA"], [2, "BB"]] + ) + assert instance.query("SELECT * FROM tbl2 ORDER BY x") == TSV( + [[1, "AA"], [2, "BB"], [3, "CC"]] + ) + + def test_replicated_table_with_not_synced_insert(): node1.query( "CREATE TABLE tbl ON CLUSTER 'cluster' (" From 734ffd916c6786ffd27479cace1298cab8e4cda7 Mon Sep 17 00:00:00 2001 From: slvrtrn Date: Fri, 25 Aug 2023 20:31:21 +0200 Subject: [PATCH 077/433] WIP prepared statements --- src/Core/MySQL/PacketsGeneric.cpp | 475 +++++++++--------- src/Core/MySQL/PacketsProtocolBinary.cpp | 230 +++++++++ src/Core/MySQL/PacketsProtocolBinary.h | 45 ++ src/Core/MySQL/PacketsProtocolText.cpp | 388 +++++++------- src/DataTypes/DataTypesDecimal.cpp | 26 +- src/Formats/FormatSettings.h | 16 +- .../InterpreterShowColumnsQuery.cpp | 2 +- .../Formats/Impl/MySQLOutputFormat.cpp | 95 ++-- .../Formats/Impl/MySQLOutputFormat.h | 3 +- src/Server/MySQLHandler.cpp | 142 ++++-- src/Server/MySQLHandler.h | 32 +- .../InformationSchema/key_column_usage.sql | 38 ++ .../referential_constraints.sql | 25 + .../System/InformationSchema/schemata.sql | 55 +- .../System/InformationSchema/tables.sql | 50 +- .../System/attachInformationSchemaTables.cpp | 5 +- 16 files changed, 1051 insertions(+), 576 deletions(-) create mode 100644 src/Core/MySQL/PacketsProtocolBinary.cpp create mode 100644 src/Core/MySQL/PacketsProtocolBinary.h create mode 100644 src/Storages/System/InformationSchema/key_column_usage.sql create mode 100644 src/Storages/System/InformationSchema/referential_constraints.sql diff --git a/src/Core/MySQL/PacketsGeneric.cpp b/src/Core/MySQL/PacketsGeneric.cpp index af80797d5c1..88183890399 100644 --- a/src/Core/MySQL/PacketsGeneric.cpp +++ b/src/Core/MySQL/PacketsGeneric.cpp @@ -8,254 +8,263 @@ namespace DB namespace MySQLProtocol { -namespace Generic -{ - -static const size_t MYSQL_ERRMSG_SIZE = 512; - -void SSLRequest::readPayloadImpl(ReadBuffer & buf) -{ - buf.readStrict(reinterpret_cast(&capability_flags), 4); - buf.readStrict(reinterpret_cast(&max_packet_size), 4); - buf.readStrict(reinterpret_cast(&character_set), 1); -} - -OKPacket::OKPacket(uint32_t capabilities_) - : header(0x00), capabilities(capabilities_), affected_rows(0x00), last_insert_id(0x00), status_flags(0x00) -{ -} - -OKPacket::OKPacket( - uint8_t header_, uint32_t capabilities_, uint64_t affected_rows_, uint32_t status_flags_, int16_t warnings_, - String session_state_changes_, String info_) - : header(header_), capabilities(capabilities_), affected_rows(affected_rows_), last_insert_id(0), warnings(warnings_), - status_flags(status_flags_), session_state_changes(std::move(session_state_changes_)), info(std::move(info_)) -{ -} - -size_t OKPacket::getPayloadSize() const -{ - size_t result = 2 + getLengthEncodedNumberSize(affected_rows); - - if (capabilities & CLIENT_PROTOCOL_41) + namespace Generic { - result += 4; - } - else if (capabilities & CLIENT_TRANSACTIONS) - { - result += 2; - } - if (capabilities & CLIENT_SESSION_TRACK) - { - result += getLengthEncodedStringSize(info); - if (status_flags & SERVER_SESSION_STATE_CHANGED) - result += getLengthEncodedStringSize(session_state_changes); - } - else - { - result += info.size(); - } + static const size_t MYSQL_ERRMSG_SIZE = 512; - return result; -} - -void OKPacket::readPayloadImpl(ReadBuffer & payload) - -{ - payload.readStrict(reinterpret_cast(&header), 1); - affected_rows = readLengthEncodedNumber(payload); - last_insert_id = readLengthEncodedNumber(payload); - - if (capabilities & CLIENT_PROTOCOL_41) - { - payload.readStrict(reinterpret_cast(&status_flags), 2); - payload.readStrict(reinterpret_cast(&warnings), 2); - } - else if (capabilities & CLIENT_TRANSACTIONS) - { - payload.readStrict(reinterpret_cast(&status_flags), 2); - } - - if (capabilities & CLIENT_SESSION_TRACK) - { - readLengthEncodedString(info, payload); - if (status_flags & SERVER_SESSION_STATE_CHANGED) + void SSLRequest::readPayloadImpl(ReadBuffer & buf) { - readLengthEncodedString(session_state_changes, payload); + buf.readStrict(reinterpret_cast(&capability_flags), 4); + buf.readStrict(reinterpret_cast(&max_packet_size), 4); + buf.readStrict(reinterpret_cast(&character_set), 1); } - } - else - { - readString(info, payload); - } -} -void OKPacket::writePayloadImpl(WriteBuffer & buffer) const + OKPacket::OKPacket(uint32_t capabilities_) + : header(0x00), capabilities(capabilities_), affected_rows(0x00), last_insert_id(0x00), status_flags(0x00) + { + } -{ - buffer.write(header); - writeLengthEncodedNumber(affected_rows, buffer); - writeLengthEncodedNumber(last_insert_id, buffer); /// last insert-id + OKPacket::OKPacket( + uint8_t header_, + uint32_t capabilities_, + uint64_t affected_rows_, + uint32_t status_flags_, + int16_t warnings_, + String session_state_changes_, + String info_) + : header(header_) + , capabilities(capabilities_) + , affected_rows(affected_rows_) + , last_insert_id(0) + , warnings(warnings_) + , status_flags(status_flags_) + , session_state_changes(std::move(session_state_changes_)) + , info(std::move(info_)) + { + } - if (capabilities & CLIENT_PROTOCOL_41) - { - buffer.write(reinterpret_cast(&status_flags), 2); - buffer.write(reinterpret_cast(&warnings), 2); - } - else if (capabilities & CLIENT_TRANSACTIONS) - { - buffer.write(reinterpret_cast(&status_flags), 2); - } + size_t OKPacket::getPayloadSize() const + { + size_t result = 2 + getLengthEncodedNumberSize(affected_rows); - if (capabilities & CLIENT_SESSION_TRACK) - { - writeLengthEncodedString(info, buffer); - if (status_flags & SERVER_SESSION_STATE_CHANGED) - writeLengthEncodedString(session_state_changes, buffer); - } - else - { - writeString(info, buffer); - } -} - -EOFPacket::EOFPacket() : warnings(0x00), status_flags(0x00) -{ -} - -EOFPacket::EOFPacket(int warnings_, int status_flags_) - : warnings(warnings_), status_flags(status_flags_) -{ -} - -size_t EOFPacket::getPayloadSize() const -{ - return 5; -} - -void EOFPacket::readPayloadImpl(ReadBuffer & payload) -{ - payload.readStrict(reinterpret_cast(&header), 1); - assert(header == 0xfe); - payload.readStrict(reinterpret_cast(&warnings), 2); - payload.readStrict(reinterpret_cast(&status_flags), 2); -} - -void EOFPacket::writePayloadImpl(WriteBuffer & buffer) const -{ - buffer.write(header); // EOF header - buffer.write(reinterpret_cast(&warnings), 2); - buffer.write(reinterpret_cast(&status_flags), 2); -} - -void AuthSwitchPacket::readPayloadImpl(ReadBuffer & payload) -{ - payload.readStrict(reinterpret_cast(&header), 1); - assert(header == 0xfe); - readStringUntilEOF(plugin_name, payload); -} - -ERRPacket::ERRPacket() : error_code(0x00) -{ -} - -ERRPacket::ERRPacket(int error_code_, String sql_state_, String error_message_) - : error_code(error_code_), sql_state(std::move(sql_state_)), error_message(std::move(error_message_)) -{ -} - -size_t ERRPacket::getPayloadSize() const -{ - return 4 + sql_state.length() + std::min(error_message.length(), MYSQL_ERRMSG_SIZE); -} - -void ERRPacket::readPayloadImpl(ReadBuffer & payload) -{ - payload.readStrict(reinterpret_cast(&header), 1); - assert(header == 0xff); - - payload.readStrict(reinterpret_cast(&error_code), 2); - - /// SQL State [optional: # + 5bytes string] - UInt8 sharp = static_cast(*payload.position()); - if (sharp == 0x23) - { - payload.ignore(1); - sql_state.resize(5); - payload.readStrict(reinterpret_cast(sql_state.data()), 5); - } - readString(error_message, payload); -} - -void ERRPacket::writePayloadImpl(WriteBuffer & buffer) const -{ - buffer.write(header); - buffer.write(reinterpret_cast(&error_code), 2); - buffer.write('#'); - buffer.write(sql_state.data(), sql_state.length()); - buffer.write(error_message.data(), std::min(error_message.length(), MYSQL_ERRMSG_SIZE)); -} - -ResponsePacket::ResponsePacket(UInt32 server_capability_flags_) - : ok(OKPacket(server_capability_flags_)) -{ -} - -ResponsePacket::ResponsePacket(UInt32 server_capability_flags_, bool is_handshake_) - : ok(OKPacket(server_capability_flags_)), is_handshake(is_handshake_) -{ -} - -void ResponsePacket::readPayloadImpl(ReadBuffer & payload) -{ - UInt16 header = static_cast(*payload.position()); - switch (header) - { - case PACKET_OK: - packetType = PACKET_OK; - ok.readPayloadWithUnpacked(payload); - break; - case PACKET_ERR: - packetType = PACKET_ERR; - err.readPayloadWithUnpacked(payload); - break; - case PACKET_EOF: - if (is_handshake) + if (capabilities & CLIENT_PROTOCOL_41) { - packetType = PACKET_AUTH_SWITCH; - auth_switch.readPayloadWithUnpacked(payload); + result += 4; + } + else if (capabilities & CLIENT_TRANSACTIONS) + { + result += 2; + } + + if (capabilities & CLIENT_SESSION_TRACK) + { + result += getLengthEncodedStringSize(info); + if (status_flags & SERVER_SESSION_STATE_CHANGED) + result += getLengthEncodedStringSize(session_state_changes); } else { - packetType = PACKET_EOF; - eof.readPayloadWithUnpacked(payload); + result += info.size(); } - break; - case PACKET_LOCALINFILE: - packetType = PACKET_LOCALINFILE; - break; - default: - packetType = PACKET_OK; - column_length = readLengthEncodedNumber(payload); + + return result; + } + + void OKPacket::readPayloadImpl(ReadBuffer & payload) + + { + payload.readStrict(reinterpret_cast(&header), 1); + affected_rows = readLengthEncodedNumber(payload); + last_insert_id = readLengthEncodedNumber(payload); + + if (capabilities & CLIENT_PROTOCOL_41) + { + payload.readStrict(reinterpret_cast(&status_flags), 2); + payload.readStrict(reinterpret_cast(&warnings), 2); + } + else if (capabilities & CLIENT_TRANSACTIONS) + { + payload.readStrict(reinterpret_cast(&status_flags), 2); + } + + if (capabilities & CLIENT_SESSION_TRACK) + { + readLengthEncodedString(info, payload); + if (status_flags & SERVER_SESSION_STATE_CHANGED) + { + readLengthEncodedString(session_state_changes, payload); + } + } + else + { + readString(info, payload); + } + } + + void OKPacket::writePayloadImpl(WriteBuffer & buffer) const + + { + buffer.write(header); + writeLengthEncodedNumber(affected_rows, buffer); + writeLengthEncodedNumber(last_insert_id, buffer); /// last insert-id + + if (capabilities & CLIENT_PROTOCOL_41) + { + buffer.write(reinterpret_cast(&status_flags), 2); + buffer.write(reinterpret_cast(&warnings), 2); + } + else if (capabilities & CLIENT_TRANSACTIONS) + { + buffer.write(reinterpret_cast(&status_flags), 2); + } + + if (capabilities & CLIENT_SESSION_TRACK) + { + writeLengthEncodedString(info, buffer); + if (status_flags & SERVER_SESSION_STATE_CHANGED) + writeLengthEncodedString(session_state_changes, buffer); + } + else + { + writeString(info, buffer); + } + } + + EOFPacket::EOFPacket() : warnings(0x00), status_flags(0x00) + { + } + + EOFPacket::EOFPacket(int warnings_, int status_flags_) : warnings(warnings_), status_flags(status_flags_) + { + } + + size_t EOFPacket::getPayloadSize() const + { + return 5; + } + + void EOFPacket::readPayloadImpl(ReadBuffer & payload) + { + payload.readStrict(reinterpret_cast(&header), 1); + assert(header == 0xfe); + payload.readStrict(reinterpret_cast(&warnings), 2); + payload.readStrict(reinterpret_cast(&status_flags), 2); + } + + void EOFPacket::writePayloadImpl(WriteBuffer & buffer) const + { + buffer.write(header); // EOF header + buffer.write(reinterpret_cast(&warnings), 2); + buffer.write(reinterpret_cast(&status_flags), 2); + } + + void AuthSwitchPacket::readPayloadImpl(ReadBuffer & payload) + { + payload.readStrict(reinterpret_cast(&header), 1); + assert(header == 0xfe); + readStringUntilEOF(plugin_name, payload); + } + + ERRPacket::ERRPacket() : error_code(0x00) + { + } + + ERRPacket::ERRPacket(int error_code_, String sql_state_, String error_message_) + : error_code(error_code_), sql_state(std::move(sql_state_)), error_message(std::move(error_message_)) + { + } + + size_t ERRPacket::getPayloadSize() const + { + return 4 + sql_state.length() + std::min(error_message.length(), MYSQL_ERRMSG_SIZE); + } + + void ERRPacket::readPayloadImpl(ReadBuffer & payload) + { + payload.readStrict(reinterpret_cast(&header), 1); + assert(header == 0xff); + + payload.readStrict(reinterpret_cast(&error_code), 2); + + /// SQL State [optional: # + 5bytes string] + UInt8 sharp = static_cast(*payload.position()); + if (sharp == 0x23) + { + payload.ignore(1); + sql_state.resize(5); + payload.readStrict(reinterpret_cast(sql_state.data()), 5); + } + readString(error_message, payload); + } + + void ERRPacket::writePayloadImpl(WriteBuffer & buffer) const + { + buffer.write(header); + buffer.write(reinterpret_cast(&error_code), 2); + buffer.write('#'); + buffer.write(sql_state.data(), sql_state.length()); + buffer.write(error_message.data(), std::min(error_message.length(), MYSQL_ERRMSG_SIZE)); + } + + ResponsePacket::ResponsePacket(UInt32 server_capability_flags_) : ok(OKPacket(server_capability_flags_)) + { + } + + ResponsePacket::ResponsePacket(UInt32 server_capability_flags_, bool is_handshake_) + : ok(OKPacket(server_capability_flags_)), is_handshake(is_handshake_) + { + } + + void ResponsePacket::readPayloadImpl(ReadBuffer & payload) + { + UInt16 header = static_cast(*payload.position()); + switch (header) + { + case PACKET_OK: + packetType = PACKET_OK; + ok.readPayloadWithUnpacked(payload); + break; + case PACKET_ERR: + packetType = PACKET_ERR; + err.readPayloadWithUnpacked(payload); + break; + case PACKET_EOF: + if (is_handshake) + { + packetType = PACKET_AUTH_SWITCH; + auth_switch.readPayloadWithUnpacked(payload); + } + else + { + packetType = PACKET_EOF; + eof.readPayloadWithUnpacked(payload); + } + break; + case PACKET_LOCALINFILE: + packetType = PACKET_LOCALINFILE; + break; + default: + packetType = PACKET_OK; + column_length = readLengthEncodedNumber(payload); + } + } + + LengthEncodedNumber::LengthEncodedNumber(uint64_t value_) : value(value_) + { + } + + size_t LengthEncodedNumber::getPayloadSize() const + { + return getLengthEncodedNumberSize(value); + } + + void LengthEncodedNumber::writePayloadImpl(WriteBuffer & buffer) const + { + writeLengthEncodedNumber(value, buffer); + } + } -} - -LengthEncodedNumber::LengthEncodedNumber(uint64_t value_) : value(value_) -{ -} - -size_t LengthEncodedNumber::getPayloadSize() const -{ - return getLengthEncodedNumberSize(value); -} - -void LengthEncodedNumber::writePayloadImpl(WriteBuffer & buffer) const -{ - writeLengthEncodedNumber(value, buffer); -} - -} } diff --git a/src/Core/MySQL/PacketsProtocolBinary.cpp b/src/Core/MySQL/PacketsProtocolBinary.cpp new file mode 100644 index 00000000000..287dda269e6 --- /dev/null +++ b/src/Core/MySQL/PacketsProtocolBinary.cpp @@ -0,0 +1,230 @@ +#include +#include +#include +#include +#include +#include +#include "Columns/ColumnLowCardinality.h" +#include "Columns/ColumnVector.h" +#include "DataTypes/DataTypeLowCardinality.h" +#include "DataTypes/DataTypeNullable.h" +#include "Formats/FormatSettings.h" +#include "IO/WriteBufferFromString.h" +#include "base/types.h" + +namespace DB +{ + +namespace MySQLProtocol +{ + + namespace ProtocolBinary + { + ResultSetRow::ResultSetRow( + const Serializations & serializations_, const DataTypes & data_types_, const Columns & columns_, int row_num_) + : row_num(row_num_), columns(columns_), data_types(data_types_), serializations(serializations_) + { + /// See https://dev.mysql.com/doc/dev/mysql-server/8.1.0/page_protocol_binary_resultset.html#sect_protocol_binary_resultset_row + payload_size = 1 + null_bitmap_size; + // LOG_TRACE(&Poco::Logger::get("ResultSetRow"), "Null bitmap size: {}", null_bitmap_size); + FormatSettings format_settings; + for (size_t i = 0; i < columns.size(); ++i) + { + ColumnPtr col = columns[i]; + + if (col->isNullAt(row_num)) + { + null_bitmap[i / 8] |= 1 << i % 8; + } + + TypeIndex type_index = removeNullable(removeLowCardinality(data_types[i]))->getTypeId(); + switch (type_index) + { + case TypeIndex::Int8: + case TypeIndex::UInt8: + payload_size += 1; + break; + case TypeIndex::Int16: + case TypeIndex::UInt16: + payload_size += 2; + break; + case TypeIndex::Int32: + case TypeIndex::UInt32: + case TypeIndex::Float32: + payload_size += 4; + break; + case TypeIndex::Int64: + case TypeIndex::UInt64: + case TypeIndex::Float64: + payload_size += 8; + break; + case TypeIndex::Date: { + UInt64 value = col->get64(row_num); + if (value == 0) + { + payload_size += 1; // length only, no other fields + } + else + { + payload_size += 5; + } + break; + } + case TypeIndex::DateTime: { + UInt64 value = col->get64(row_num); + if (value == 0) + { + payload_size += 1; // length only, no other fields + } + else + { + Poco::DateTime dt = Poco::DateTime(Poco::Timestamp(value * 1000 * 1000)); + if (dt.second() == 0 && dt.minute() == 0 && dt.hour() == 0) + { + payload_size += 5; + } + else + { + payload_size += 8; + } + } + break; + } + default: + WriteBufferFromOwnString ostr; + serializations[i]->serializeText(*columns[i], row_num, ostr, format_settings); + payload_size += getLengthEncodedStringSize(ostr.str()); + serialized[i] = std::move(ostr.str()); + break; + } + } + } + + void ResultSetRow::writePayloadImpl(WriteBuffer & buffer) const + { + buffer.write(static_cast(0x00)); + buffer.write(null_bitmap.data(), null_bitmap_size); + for (size_t i = 0; i < columns.size(); ++i) + { + ColumnPtr col = columns[i]; + if (col->isNullAt(row_num)) + { + continue; // NULLs are stored in the null bitmap only + } + + TypeIndex type_index = removeNullable(removeLowCardinality(data_types[i]))->getTypeId(); + switch (type_index) + { + case TypeIndex::UInt8: { + UInt64 value = col->get64(row_num); + buffer.write(reinterpret_cast(&value), 1); + break; + } + case TypeIndex::UInt16: { + UInt64 value = col->get64(row_num); + buffer.write(reinterpret_cast(&value), 2); + break; + } + case TypeIndex::UInt32: { + UInt64 value = col->get64(row_num); + buffer.write(reinterpret_cast(&value), 4); + break; + } + case TypeIndex::UInt64: { + UInt64 value = col->get64(row_num); + buffer.write(reinterpret_cast(&value), 8); + break; + } + case TypeIndex::Int8: { + UInt64 value = col->get64(row_num); + buffer.write(reinterpret_cast(&value), 1); + break; + } + case TypeIndex::Int16: { + UInt64 value = col->get64(row_num); + buffer.write(reinterpret_cast(&value), 2); + break; + } + case TypeIndex::Int32: { + UInt64 value = col->get64(row_num); + buffer.write(reinterpret_cast(&value), 4); + break; + } + case TypeIndex::Int64: { + UInt64 value = col->get64(row_num); + buffer.write(reinterpret_cast(&value), 8); + break; + } + case TypeIndex::Float32: { + Float32 value = col->getFloat32(row_num); + buffer.write(reinterpret_cast(&value), 4); + break; + } + case TypeIndex::Float64: { + Float64 value = col->getFloat64(row_num); + buffer.write(reinterpret_cast(&value), 8); + break; + } + case TypeIndex::Date: { + UInt64 value = col->get64(row_num); + if (value != 0) + { + Poco::DateTime dt = Poco::DateTime(Poco::Timestamp(value * 1000 * 1000)); + buffer.write(static_cast(4)); // bytes_following + int year = dt.year(); + int month = dt.month(); + int day = dt.day(); + buffer.write(reinterpret_cast(&year), 2); + buffer.write(reinterpret_cast(&month), 1); + buffer.write(reinterpret_cast(&day), 1); + } + else + { + buffer.write(static_cast(0)); + } + break; + } + case TypeIndex::DateTime: { + UInt64 value = col->get64(row_num); + if (value != 0) + { + Poco::DateTime dt = Poco::DateTime(Poco::Timestamp(value * 1000 * 1000)); + bool is_date_time = !(dt.hour() == 0 && dt.minute() == 0 && dt.second() == 0); + size_t bytes_following = is_date_time ? 7 : 4; + buffer.write(reinterpret_cast(&bytes_following), 1); + int year = dt.year(); + int month = dt.month(); + int day = dt.day(); + buffer.write(reinterpret_cast(&year), 2); + buffer.write(reinterpret_cast(&month), 1); + buffer.write(reinterpret_cast(&day), 1); + if (is_date_time) + { + int hour = dt.hourAMPM(); + int minute = dt.minute(); + int second = dt.second(); + buffer.write(reinterpret_cast(&hour), 1); + buffer.write(reinterpret_cast(&minute), 1); + buffer.write(reinterpret_cast(&second), 1); + } + } + else + { + buffer.write(static_cast(0)); + } + break; + } + default: + writeLengthEncodedString(serialized[i], buffer); + break; + } + } + } + + size_t ResultSetRow::getPayloadSize() const + { + return payload_size; + }; + } +} +} diff --git a/src/Core/MySQL/PacketsProtocolBinary.h b/src/Core/MySQL/PacketsProtocolBinary.h new file mode 100644 index 00000000000..69936e527c1 --- /dev/null +++ b/src/Core/MySQL/PacketsProtocolBinary.h @@ -0,0 +1,45 @@ +#pragma once + +#include +#include +#include +#include +#include "DataTypes/IDataType.h" +#include "DataTypes/Serializations/ISerialization.h" + +namespace DB +{ + +namespace MySQLProtocol +{ + + namespace ProtocolBinary + { + class ResultSetRow : public IMySQLWritePacket + { + private: + TypeIndex getTypeIndex(DataTypePtr data_type, const ColumnPtr & col) const; + + protected: + int row_num; + const Columns & columns; + const DataTypes & data_types; + const Serializations & serializations; + + std::vector serialized = std::vector(columns.size()); + + size_t null_bitmap_size = (columns.size() + 7) / 8; + std::vector null_bitmap = std::vector(null_bitmap_size, 0); + + size_t payload_size = 0; + + size_t getPayloadSize() const override; + + void writePayloadImpl(WriteBuffer & buffer) const override; + + public: + ResultSetRow(const Serializations & serializations_, const DataTypes & data_types_, const Columns & columns_, int row_num_); + }; + } +} +} diff --git a/src/Core/MySQL/PacketsProtocolText.cpp b/src/Core/MySQL/PacketsProtocolText.cpp index 728e8061e87..9c5bf6b6e05 100644 --- a/src/Core/MySQL/PacketsProtocolText.cpp +++ b/src/Core/MySQL/PacketsProtocolText.cpp @@ -1,7 +1,8 @@ #include -#include #include +#include #include +#include "Core/MySQL/IMySQLWritePacket.h" namespace DB { @@ -9,197 +10,212 @@ namespace DB namespace MySQLProtocol { -namespace ProtocolText -{ - -ResultSetRow::ResultSetRow(const Serializations & serializations, const Columns & columns_, int row_num_) - : columns(columns_), row_num(row_num_) -{ - for (size_t i = 0; i < columns.size(); ++i) + namespace ProtocolText { - if (columns[i]->isNullAt(row_num)) + + ResultSetRow::ResultSetRow(const Serializations & serializations, const Columns & columns_, int row_num_) + : columns(columns_), row_num(row_num_) { - payload_size += 1; - serialized.emplace_back("\xfb"); + for (size_t i = 0; i < columns.size(); ++i) + { + if (columns[i]->isNullAt(row_num)) + { + payload_size += 1; + serialized.emplace_back("\xfb"); + } + else + { + WriteBufferFromOwnString ostr; + serializations[i]->serializeText(*columns[i], row_num, ostr, FormatSettings()); + payload_size += getLengthEncodedStringSize(ostr.str()); + serialized.push_back(std::move(ostr.str())); + } + } } - else + + size_t ResultSetRow::getPayloadSize() const { - WriteBufferFromOwnString ostr; - serializations[i]->serializeText(*columns[i], row_num, ostr, FormatSettings()); - payload_size += getLengthEncodedStringSize(ostr.str()); - serialized.push_back(std::move(ostr.str())); + return payload_size; } + + void ResultSetRow::writePayloadImpl(WriteBuffer & buffer) const + { + for (size_t i = 0; i < columns.size(); ++i) + { + if (columns[i]->isNullAt(row_num)) + buffer.write(serialized[i].data(), 1); + else + writeLengthEncodedString(serialized[i], buffer); + } + } + + void ComFieldList::readPayloadImpl(ReadBuffer & payload) + { + // Command byte has been already read from payload. + readNullTerminated(table, payload); + readStringUntilEOF(field_wildcard, payload); + } + + ColumnDefinition::ColumnDefinition() : character_set(0x00), column_length(0), column_type(MYSQL_TYPE_DECIMAL), flags(0x00) + { + } + + ColumnDefinition::ColumnDefinition( + String schema_, + String table_, + String org_table_, + String name_, + String org_name_, + uint16_t character_set_, + uint32_t column_length_, + ColumnType column_type_, + uint16_t flags_, + uint8_t decimals_, + bool with_defaults_) + : schema(std::move(schema_)) + , table(std::move(table_)) + , org_table(std::move(org_table_)) + , name(std::move(name_)) + , org_name(std::move(org_name_)) + , character_set(character_set_) + , column_length(column_length_) + , column_type(column_type_) + , flags(flags_) + , decimals(decimals_) + , is_comm_field_list_response(with_defaults_) + { + } + + ColumnDefinition::ColumnDefinition( + String name_, uint16_t character_set_, uint32_t column_length_, ColumnType column_type_, uint16_t flags_, uint8_t decimals_) + : ColumnDefinition("", "", "", std::move(name_), "", character_set_, column_length_, column_type_, flags_, decimals_) + { + } + + size_t ColumnDefinition::getPayloadSize() const + { + return 12 + getLengthEncodedStringSize("def") + getLengthEncodedStringSize(schema) + getLengthEncodedStringSize(table) + + getLengthEncodedStringSize(org_table) + getLengthEncodedStringSize(name) + getLengthEncodedStringSize(org_name) + + getLengthEncodedNumberSize(next_length) + is_comm_field_list_response; + } + + void ColumnDefinition::readPayloadImpl(ReadBuffer & payload) + { + String def; + readLengthEncodedString(def, payload); + assert(def == "def"); + readLengthEncodedString(schema, payload); + readLengthEncodedString(table, payload); + readLengthEncodedString(org_table, payload); + readLengthEncodedString(name, payload); + readLengthEncodedString(org_name, payload); + next_length = readLengthEncodedNumber(payload); + payload.readStrict(reinterpret_cast(&character_set), 2); + payload.readStrict(reinterpret_cast(&column_length), 4); + payload.readStrict(reinterpret_cast(&column_type), 1); + payload.readStrict(reinterpret_cast(&flags), 2); + payload.readStrict(reinterpret_cast(&decimals), 1); + payload.ignore(2); + } + + void ColumnDefinition::writePayloadImpl(WriteBuffer & buffer) const + { + writeLengthEncodedString(std::string("def"), buffer); /// always "def" + writeLengthEncodedString(schema, buffer); + writeLengthEncodedString(table, buffer); + writeLengthEncodedString(org_table, buffer); + writeLengthEncodedString(name, buffer); + writeLengthEncodedString(org_name, buffer); + writeLengthEncodedNumber(next_length, buffer); + buffer.write(reinterpret_cast(&character_set), 2); + buffer.write(reinterpret_cast(&column_length), 4); + buffer.write(reinterpret_cast(&column_type), 1); + buffer.write(reinterpret_cast(&flags), 2); + buffer.write(reinterpret_cast(&decimals), 1); + writeChar(0x0, 2, buffer); + if (is_comm_field_list_response) + { + /// We should write length encoded int with string size + /// followed by string with some "default values" (possibly it's column defaults). + /// But we just send NULL for simplicity. + writeChar(0xfb, buffer); + } + } + + ColumnDefinition getColumnDefinition(const String & column_name, const TypeIndex type_index) + { + ColumnType column_type; + CharacterSet charset = CharacterSet::binary; + int flags = 0; + uint8_t decimals = 0; + switch (type_index) + { + case TypeIndex::UInt8: + column_type = ColumnType::MYSQL_TYPE_TINY; + flags = ColumnDefinitionFlags::BINARY_FLAG | ColumnDefinitionFlags::UNSIGNED_FLAG; + break; + case TypeIndex::UInt16: + column_type = ColumnType::MYSQL_TYPE_SHORT; + flags = ColumnDefinitionFlags::BINARY_FLAG | ColumnDefinitionFlags::UNSIGNED_FLAG; + break; + case TypeIndex::UInt32: + column_type = ColumnType::MYSQL_TYPE_LONG; + flags = ColumnDefinitionFlags::BINARY_FLAG | ColumnDefinitionFlags::UNSIGNED_FLAG; + break; + case TypeIndex::UInt64: + column_type = ColumnType::MYSQL_TYPE_LONGLONG; + flags = ColumnDefinitionFlags::BINARY_FLAG | ColumnDefinitionFlags::UNSIGNED_FLAG; + break; + case TypeIndex::Int8: + column_type = ColumnType::MYSQL_TYPE_TINY; + flags = ColumnDefinitionFlags::BINARY_FLAG; + break; + case TypeIndex::Int16: + column_type = ColumnType::MYSQL_TYPE_SHORT; + flags = ColumnDefinitionFlags::BINARY_FLAG; + break; + case TypeIndex::Int32: + column_type = ColumnType::MYSQL_TYPE_LONG; + flags = ColumnDefinitionFlags::BINARY_FLAG; + break; + case TypeIndex::Int64: + column_type = ColumnType::MYSQL_TYPE_LONGLONG; + flags = ColumnDefinitionFlags::BINARY_FLAG; + break; + case TypeIndex::Float32: + column_type = ColumnType::MYSQL_TYPE_FLOAT; + flags = ColumnDefinitionFlags::BINARY_FLAG; + decimals = 31; + break; + case TypeIndex::Float64: + column_type = ColumnType::MYSQL_TYPE_DOUBLE; + flags = ColumnDefinitionFlags::BINARY_FLAG; + decimals = 31; + break; + case TypeIndex::Date: + column_type = ColumnType::MYSQL_TYPE_DATE; + flags = ColumnDefinitionFlags::BINARY_FLAG; + break; + case TypeIndex::DateTime: + column_type = ColumnType::MYSQL_TYPE_DATETIME; + flags = ColumnDefinitionFlags::BINARY_FLAG; + break; + case TypeIndex::Decimal32: + case TypeIndex::Decimal64: + case TypeIndex::Decimal128: + /// MySQL Decimal has max 65 precision and 30 scale. Thus, Decimal256 is reported as a string + column_type = ColumnType::MYSQL_TYPE_DECIMAL; + flags = ColumnDefinitionFlags::BINARY_FLAG; + break; + default: + column_type = ColumnType::MYSQL_TYPE_STRING; + charset = CharacterSet::utf8_general_ci; + break; + } + return ColumnDefinition(column_name, charset, 0, column_type, flags, decimals); + } + } -} - -size_t ResultSetRow::getPayloadSize() const -{ - return payload_size; -} - -void ResultSetRow::writePayloadImpl(WriteBuffer & buffer) const -{ - for (size_t i = 0; i < columns.size(); ++i) - { - if (columns[i]->isNullAt(row_num)) - buffer.write(serialized[i].data(), 1); - else - writeLengthEncodedString(serialized[i], buffer); - } -} - -void ComFieldList::readPayloadImpl(ReadBuffer & payload) -{ - // Command byte has been already read from payload. - readNullTerminated(table, payload); - readStringUntilEOF(field_wildcard, payload); -} - -ColumnDefinition::ColumnDefinition() - : character_set(0x00), column_length(0), column_type(MYSQL_TYPE_DECIMAL), flags(0x00) -{ -} - -ColumnDefinition::ColumnDefinition( - String schema_, String table_, String org_table_, String name_, String org_name_, uint16_t character_set_, uint32_t column_length_, - ColumnType column_type_, uint16_t flags_, uint8_t decimals_, bool with_defaults_) - : schema(std::move(schema_)), table(std::move(table_)), org_table(std::move(org_table_)), name(std::move(name_)), - org_name(std::move(org_name_)), character_set(character_set_), column_length(column_length_), column_type(column_type_), - flags(flags_), decimals(decimals_), is_comm_field_list_response(with_defaults_) -{ -} - -ColumnDefinition::ColumnDefinition( - String name_, uint16_t character_set_, uint32_t column_length_, ColumnType column_type_, uint16_t flags_, uint8_t decimals_) - : ColumnDefinition("", "", "", std::move(name_), "", character_set_, column_length_, column_type_, flags_, decimals_) -{ -} - -size_t ColumnDefinition::getPayloadSize() const -{ - return 12 + - getLengthEncodedStringSize("def") + - getLengthEncodedStringSize(schema) + - getLengthEncodedStringSize(table) + - getLengthEncodedStringSize(org_table) + - getLengthEncodedStringSize(name) + - getLengthEncodedStringSize(org_name) + - getLengthEncodedNumberSize(next_length) + - is_comm_field_list_response; -} - -void ColumnDefinition::readPayloadImpl(ReadBuffer & payload) -{ - String def; - readLengthEncodedString(def, payload); - assert(def == "def"); - readLengthEncodedString(schema, payload); - readLengthEncodedString(table, payload); - readLengthEncodedString(org_table, payload); - readLengthEncodedString(name, payload); - readLengthEncodedString(org_name, payload); - next_length = readLengthEncodedNumber(payload); - payload.readStrict(reinterpret_cast(&character_set), 2); - payload.readStrict(reinterpret_cast(&column_length), 4); - payload.readStrict(reinterpret_cast(&column_type), 1); - payload.readStrict(reinterpret_cast(&flags), 2); - payload.readStrict(reinterpret_cast(&decimals), 1); - payload.ignore(2); -} - -void ColumnDefinition::writePayloadImpl(WriteBuffer & buffer) const -{ - writeLengthEncodedString(std::string("def"), buffer); /// always "def" - writeLengthEncodedString(schema, buffer); - writeLengthEncodedString(table, buffer); - writeLengthEncodedString(org_table, buffer); - writeLengthEncodedString(name, buffer); - writeLengthEncodedString(org_name, buffer); - writeLengthEncodedNumber(next_length, buffer); - buffer.write(reinterpret_cast(&character_set), 2); - buffer.write(reinterpret_cast(&column_length), 4); - buffer.write(reinterpret_cast(&column_type), 1); - buffer.write(reinterpret_cast(&flags), 2); - buffer.write(reinterpret_cast(&decimals), 1); - writeChar(0x0, 2, buffer); - if (is_comm_field_list_response) - { - /// We should write length encoded int with string size - /// followed by string with some "default values" (possibly it's column defaults). - /// But we just send NULL for simplicity. - writeChar(0xfb, buffer); - } -} - -ColumnDefinition getColumnDefinition(const String & column_name, const TypeIndex type_index) -{ - ColumnType column_type; - CharacterSet charset = CharacterSet::binary; - int flags = 0; - switch (type_index) - { - case TypeIndex::UInt8: - column_type = ColumnType::MYSQL_TYPE_TINY; - flags = ColumnDefinitionFlags::BINARY_FLAG | ColumnDefinitionFlags::UNSIGNED_FLAG; - break; - case TypeIndex::UInt16: - column_type = ColumnType::MYSQL_TYPE_SHORT; - flags = ColumnDefinitionFlags::BINARY_FLAG | ColumnDefinitionFlags::UNSIGNED_FLAG; - break; - case TypeIndex::UInt32: - column_type = ColumnType::MYSQL_TYPE_LONG; - flags = ColumnDefinitionFlags::BINARY_FLAG | ColumnDefinitionFlags::UNSIGNED_FLAG; - break; - case TypeIndex::UInt64: - column_type = ColumnType::MYSQL_TYPE_LONGLONG; - flags = ColumnDefinitionFlags::BINARY_FLAG | ColumnDefinitionFlags::UNSIGNED_FLAG; - break; - case TypeIndex::Int8: - column_type = ColumnType::MYSQL_TYPE_TINY; - flags = ColumnDefinitionFlags::BINARY_FLAG; - break; - case TypeIndex::Int16: - column_type = ColumnType::MYSQL_TYPE_SHORT; - flags = ColumnDefinitionFlags::BINARY_FLAG; - break; - case TypeIndex::Int32: - column_type = ColumnType::MYSQL_TYPE_LONG; - flags = ColumnDefinitionFlags::BINARY_FLAG; - break; - case TypeIndex::Int64: - column_type = ColumnType::MYSQL_TYPE_LONGLONG; - flags = ColumnDefinitionFlags::BINARY_FLAG; - break; - case TypeIndex::Float32: - column_type = ColumnType::MYSQL_TYPE_FLOAT; - flags = ColumnDefinitionFlags::BINARY_FLAG; - break; - case TypeIndex::Float64: - column_type = ColumnType::MYSQL_TYPE_DOUBLE; - flags = ColumnDefinitionFlags::BINARY_FLAG; - break; - case TypeIndex::Date: - column_type = ColumnType::MYSQL_TYPE_DATE; - flags = ColumnDefinitionFlags::BINARY_FLAG; - break; - case TypeIndex::DateTime: - column_type = ColumnType::MYSQL_TYPE_DATETIME; - flags = ColumnDefinitionFlags::BINARY_FLAG; - break; - case TypeIndex::String: - case TypeIndex::FixedString: - column_type = ColumnType::MYSQL_TYPE_STRING; - charset = CharacterSet::utf8_general_ci; - break; - default: - column_type = ColumnType::MYSQL_TYPE_STRING; - charset = CharacterSet::utf8_general_ci; - break; - } - return ColumnDefinition(column_name, charset, 0, column_type, flags, 0); -} - -} } diff --git a/src/DataTypes/DataTypesDecimal.cpp b/src/DataTypes/DataTypesDecimal.cpp index fa044d4ac9c..2af216529e5 100644 --- a/src/DataTypes/DataTypesDecimal.cpp +++ b/src/DataTypes/DataTypesDecimal.cpp @@ -1,13 +1,13 @@ #include #include -#include #include #include #include #include #include #include +#include #include @@ -31,6 +31,12 @@ std::string DataTypeDecimal::doGetName() const template std::string DataTypeDecimal::getSQLCompatibleName() const { + /// See https://dev.mysql.com/doc/refman/8.0/en/precision-math-decimal-characteristics.html + /// DECIMAL(M,D) + /// M is the maximum number of digits (the precision). It has a range of 1 to 65. + /// D is the number of digits to the right of the decimal point (the scale). It has a range of 0 to 30 and must be no larger than M. + if (this->precision > 65 || this->scale > 30) + return "TEXT"; return fmt::format("DECIMAL({}, {})", this->precision, this->scale); } @@ -75,14 +81,14 @@ SerializationPtr DataTypeDecimal::doGetDefaultSerialization() const static DataTypePtr create(const ASTPtr & arguments) { if (!arguments || arguments->children.size() != 2) - throw Exception(ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH, - "Decimal data type family must have exactly two arguments: precision and scale"); + throw Exception( + ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH, "Decimal data type family must have exactly two arguments: precision and scale"); const auto * precision = arguments->children[0]->as(); const auto * scale = arguments->children[1]->as(); - if (!precision || precision->value.getType() != Field::Types::UInt64 || - !scale || !(scale->value.getType() == Field::Types::Int64 || scale->value.getType() == Field::Types::UInt64)) + if (!precision || precision->value.getType() != Field::Types::UInt64 || !scale + || !(scale->value.getType() == Field::Types::Int64 || scale->value.getType() == Field::Types::UInt64)) throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Decimal data type family must have two numbers as its arguments"); UInt64 precision_value = precision->value.get(); @@ -95,13 +101,15 @@ template static DataTypePtr createExact(const ASTPtr & arguments) { if (!arguments || arguments->children.size() != 1) - throw Exception(ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH, - "Decimal32 | Decimal64 | Decimal128 | Decimal256 data type family must have exactly one arguments: scale"); + throw Exception( + ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH, + "Decimal32 | Decimal64 | Decimal128 | Decimal256 data type family must have exactly one arguments: scale"); const auto * scale_arg = arguments->children[0]->as(); if (!scale_arg || !(scale_arg->value.getType() == Field::Types::Int64 || scale_arg->value.getType() == Field::Types::UInt64)) - throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, - "Decimal32 | Decimal64 | Decimal128 | Decimal256 data type family must have a one number as its argument"); + throw Exception( + ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, + "Decimal32 | Decimal64 | Decimal128 | Decimal256 data type family must have a one number as its argument"); UInt64 precision = DecimalUtils::max_precision; UInt64 scale = scale_arg->value.get(); diff --git a/src/Formats/FormatSettings.h b/src/Formats/FormatSettings.h index 2c283dcc2b7..56359fd0bea 100644 --- a/src/Formats/FormatSettings.h +++ b/src/Formats/FormatSettings.h @@ -1,7 +1,7 @@ #pragma once -#include #include +#include #include #include @@ -48,9 +48,9 @@ struct FormatSettings enum class DateTimeInputFormat { - Basic, /// Default format for fast parsing: YYYY-MM-DD hh:mm:ss (ISO-8601 without fractional part and timezone) or NNNNNNNNNN unix timestamp. - BestEffort, /// Use sophisticated rules to parse whatever possible. - BestEffortUS /// Use sophisticated rules to parse American style: mm/dd/yyyy + Basic, /// Default format for fast parsing: YYYY-MM-DD hh:mm:ss (ISO-8601 without fractional part and timezone) or NNNNNNNNNN unix timestamp. + BestEffort, /// Use sophisticated rules to parse whatever possible. + BestEffortUS /// Use sophisticated rules to parse American style: mm/dd/yyyy }; DateTimeInputFormat date_time_input_format = DateTimeInputFormat::Basic; @@ -282,6 +282,14 @@ struct FormatSettings uint32_t client_capabilities = 0; size_t max_packet_size = 0; uint8_t * sequence_id = nullptr; /// Not null if it's MySQLWire output format used to handle MySQL protocol connections. + /** + * COM_QUERY uses Text ResultSet + * https://dev.mysql.com/doc/dev/mysql-server/latest/page_protocol_com_query_response_text_resultset.html + * COM_STMT_EXECUTE uses Binary Protocol ResultSet + * https://dev.mysql.com/doc/dev/mysql-server/latest/page_protocol_com_stmt_execute_response.html + * By default, use Text ResultSet. + */ + bool use_binary_result_set = false; } mysql_wire; struct diff --git a/src/Interpreters/InterpreterShowColumnsQuery.cpp b/src/Interpreters/InterpreterShowColumnsQuery.cpp index c86d3c753c4..922f9887a82 100644 --- a/src/Interpreters/InterpreterShowColumnsQuery.cpp +++ b/src/Interpreters/InterpreterShowColumnsQuery.cpp @@ -37,7 +37,7 @@ String InterpreterShowColumnsQuery::getRewrittenQuery() SELECT name AS field, type AS type, - startsWith(type, 'Nullable') AS null, + if(startsWith(type, 'Nullable'), 'YES', 'NO') AS null, trim(concatWithSeparator(' ', if (is_in_primary_key, 'PRI', ''), if (is_in_sorting_key, 'SOR', ''))) AS key, if (default_kind IN ('ALIAS', 'DEFAULT', 'MATERIALIZED'), default_expression, NULL) AS default, '' AS extra )"; diff --git a/src/Processors/Formats/Impl/MySQLOutputFormat.cpp b/src/Processors/Formats/Impl/MySQLOutputFormat.cpp index f2157f63c25..3dafe560281 100644 --- a/src/Processors/Formats/Impl/MySQLOutputFormat.cpp +++ b/src/Processors/Formats/Impl/MySQLOutputFormat.cpp @@ -1,11 +1,12 @@ -#include #include +#include #include #include #include #include #include - +#include +#include "Common/logger_useful.h" namespace DB { @@ -13,17 +14,18 @@ namespace DB using namespace MySQLProtocol; using namespace MySQLProtocol::Generic; using namespace MySQLProtocol::ProtocolText; - +using namespace MySQLProtocol::ProtocolBinary; MySQLOutputFormat::MySQLOutputFormat(WriteBuffer & out_, const Block & header_, const FormatSettings & settings_) - : IOutputFormat(header_, out_) - , client_capabilities(settings_.mysql_wire.client_capabilities) + : IOutputFormat(header_, out_), client_capabilities(settings_.mysql_wire.client_capabilities) { /// MySQlWire is a special format that is usually used as output format for MySQL protocol connections. /// In this case we have a correct `sequence_id` stored in `settings_.mysql_wire`. /// But it's also possible to specify MySQLWire as output format for clickhouse-client or clickhouse-local. /// There is no `sequence_id` stored in `settings_.mysql_wire` in this case, so we create a dummy one. sequence_id = settings_.mysql_wire.sequence_id ? settings_.mysql_wire.sequence_id : &dummy_sequence_id; + /// Switch between Text (COM_QUERY) and Binary (COM_EXECUTE_STMT) ResultSet + use_binary_result_set = settings_.mysql_wire.use_binary_result_set; const auto & header = getPort(PortKind::Main).getHeader(); data_types = header.getDataTypes(); @@ -54,7 +56,7 @@ void MySQLOutputFormat::writePrefix() packet_endpoint->sendPacket(getColumnDefinition(column_name, data_types[i]->getTypeId())); } - if (!(client_capabilities & Capability::CLIENT_DEPRECATE_EOF)) + if (!(client_capabilities & Capability::CLIENT_DEPRECATE_EOF) && !use_binary_result_set) { packet_endpoint->sendPacket(EOFPacket(0, 0)); } @@ -63,39 +65,67 @@ void MySQLOutputFormat::writePrefix() void MySQLOutputFormat::consume(Chunk chunk) { - for (size_t i = 0; i < chunk.getNumRows(); ++i) + if (!use_binary_result_set) { - ProtocolText::ResultSetRow row_packet(serializations, chunk.getColumns(), static_cast(i)); - packet_endpoint->sendPacket(row_packet); + for (size_t i = 0; i < chunk.getNumRows(); ++i) + { + ProtocolText::ResultSetRow row_packet(serializations, chunk.getColumns(), static_cast(i)); + packet_endpoint->sendPacket(row_packet); + } + } + else + { + for (size_t i = 0; i < chunk.getNumRows(); ++i) + { + ProtocolBinary::ResultSetRow row_packet(serializations, data_types, chunk.getColumns(), static_cast(i)); + packet_endpoint->sendPacket(row_packet); + } } } void MySQLOutputFormat::finalizeImpl() { - size_t affected_rows = 0; - std::string human_readable_info; - if (QueryStatusPtr process_list_elem = getContext()->getProcessListElement()) + if (!use_binary_result_set) { - CurrentThread::finalizePerformanceCounters(); - QueryStatusInfo info = process_list_elem->getInfo(); - affected_rows = info.written_rows; - double elapsed_seconds = static_cast(info.elapsed_microseconds) / 1000000.0; - human_readable_info = fmt::format( - "Read {} rows, {} in {} sec., {} rows/sec., {}/sec.", - info.read_rows, - ReadableSize(info.read_bytes), - elapsed_seconds, - static_cast(info.read_rows / elapsed_seconds), - ReadableSize(info.read_bytes / elapsed_seconds)); - } + size_t affected_rows = 0; + std::string human_readable_info; + if (QueryStatusPtr process_list_elem = getContext()->getProcessListElement()) + { + CurrentThread::finalizePerformanceCounters(); + QueryStatusInfo info = process_list_elem->getInfo(); + affected_rows = info.written_rows; + double elapsed_seconds = static_cast(info.elapsed_microseconds) / 1000000.0; + human_readable_info = fmt::format( + "Read {} rows, {} in {} sec., {} rows/sec., {}/sec.", + info.read_rows, + ReadableSize(info.read_bytes), + elapsed_seconds, + static_cast(info.read_rows / elapsed_seconds), + ReadableSize(info.read_bytes / elapsed_seconds)); + } - const auto & header = getPort(PortKind::Main).getHeader(); - if (header.columns() == 0) - packet_endpoint->sendPacket(OKPacket(0x0, client_capabilities, affected_rows, 0, 0, "", human_readable_info), true); - else if (client_capabilities & CLIENT_DEPRECATE_EOF) - packet_endpoint->sendPacket(OKPacket(0xfe, client_capabilities, affected_rows, 0, 0, "", human_readable_info), true); + const auto & header = getPort(PortKind::Main).getHeader(); + if (header.columns() == 0) + packet_endpoint->sendPacket(OKPacket(0x0, client_capabilities, affected_rows, 0, 0, "", human_readable_info), true); + else if (client_capabilities & CLIENT_DEPRECATE_EOF) + packet_endpoint->sendPacket(OKPacket(0xfe, client_capabilities, affected_rows, 0, 0, "", human_readable_info), true); + else + packet_endpoint->sendPacket(EOFPacket(0, 0), true); + } else - packet_endpoint->sendPacket(EOFPacket(0, 0), true); + { + size_t affected_rows = 0; + if (QueryStatusPtr process_list_elem = getContext()->getProcessListElement()) + { + CurrentThread::finalizePerformanceCounters(); + QueryStatusInfo info = process_list_elem->getInfo(); + affected_rows = info.written_rows; + } + if (client_capabilities & CLIENT_DEPRECATE_EOF) + packet_endpoint->sendPacket(OKPacket(0xfe, client_capabilities, affected_rows, 0, 0, "", ""), true); + else + packet_endpoint->sendPacket(EOFPacket(0, 0), true); + } } void MySQLOutputFormat::flush() @@ -107,9 +137,8 @@ void registerOutputFormatMySQLWire(FormatFactory & factory) { factory.registerOutputFormat( "MySQLWire", - [](WriteBuffer & buf, - const Block & sample, - const FormatSettings & settings) { return std::make_shared(buf, sample, settings); }); + [](WriteBuffer & buf, const Block & sample, const FormatSettings & settings) + { return std::make_shared(buf, sample, settings); }); } } diff --git a/src/Processors/Formats/Impl/MySQLOutputFormat.h b/src/Processors/Formats/Impl/MySQLOutputFormat.h index 9481ef67070..6161b6bdc14 100644 --- a/src/Processors/Formats/Impl/MySQLOutputFormat.h +++ b/src/Processors/Formats/Impl/MySQLOutputFormat.h @@ -1,7 +1,7 @@ #pragma once -#include #include +#include #include #include @@ -39,6 +39,7 @@ private: MySQLProtocol::PacketEndpointPtr packet_endpoint; DataTypes data_types; Serializations serializations; + bool use_binary_result_set = false; }; } diff --git a/src/Server/MySQLHandler.cpp b/src/Server/MySQLHandler.cpp index 868575b701f..3715dfea9f7 100644 --- a/src/Server/MySQLHandler.cpp +++ b/src/Server/MySQLHandler.cpp @@ -1,29 +1,29 @@ #include "MySQLHandler.h" #include -#include -#include +#include +#include +#include #include #include -#include #include #include -#include -#include -#include #include #include #include +#include #include #include -#include +#include +#include +#include #include #include -#include -#include -#include -#include #include +#include +#include +#include +#include #include "config_version.h" @@ -67,10 +67,7 @@ static String killConnectionIdReplacementQuery(const String & query); static String selectLimitReplacementQuery(const String & query); MySQLHandler::MySQLHandler( - IServer & server_, - TCPServer & tcp_server_, - const Poco::Net::StreamSocket & socket_, - bool ssl_enabled, uint32_t connection_id_) + IServer & server_, TCPServer & tcp_server_, const Poco::Net::StreamSocket & socket_, bool ssl_enabled, uint32_t connection_id_) : Poco::Net::TCPServerConnection(socket_) , server(server_) , tcp_server(tcp_server_) @@ -78,7 +75,8 @@ MySQLHandler::MySQLHandler( , connection_id(connection_id_) , auth_plugin(new MySQLProtocol::Authentication::Native41()) { - server_capabilities = CLIENT_PROTOCOL_41 | CLIENT_SECURE_CONNECTION | CLIENT_PLUGIN_AUTH | CLIENT_PLUGIN_AUTH_LENENC_CLIENT_DATA | CLIENT_CONNECT_WITH_DB | CLIENT_DEPRECATE_EOF; + server_capabilities = CLIENT_PROTOCOL_41 | CLIENT_SECURE_CONNECTION | CLIENT_PLUGIN_AUTH | CLIENT_PLUGIN_AUTH_LENENC_CLIENT_DATA + | CLIENT_CONNECT_WITH_DB | CLIENT_DEPRECATE_EOF; if (ssl_enabled) server_capabilities |= CLIENT_SSL; @@ -104,8 +102,13 @@ void MySQLHandler::run() try { - Handshake handshake(server_capabilities, connection_id, VERSION_STRING + String("-") + VERSION_NAME, - auth_plugin->getName(), auth_plugin->getAuthPluginData(), CharacterSet::utf8_general_ci); + Handshake handshake( + server_capabilities, + connection_id, + VERSION_STRING + String("-") + VERSION_NAME, + auth_plugin->getName(), + auth_plugin->getAuthPluginData(), + CharacterSet::utf8_general_ci); packet_endpoint->sendPacket(handshake, true); LOG_TRACE(log, "Sent handshake"); @@ -115,8 +118,10 @@ void MySQLHandler::run() client_capabilities = handshake_response.capability_flags; max_packet_size = handshake_response.max_packet_size ? handshake_response.max_packet_size : MAX_PACKET_LENGTH; - LOG_TRACE(log, - "Capabilities: {}, max_packet_size: {}, character_set: {}, user: {}, auth_response length: {}, database: {}, auth_plugin_name: {}", + LOG_TRACE( + log, + "Capabilities: {}, max_packet_size: {}, character_set: {}, user: {}, auth_response length: {}, database: {}, auth_plugin_name: " + "{}", handshake_response.capability_flags, handshake_response.max_packet_size, static_cast(handshake_response.character_set), @@ -160,8 +165,8 @@ void MySQLHandler::run() // For commands which are executed without MemoryTracker. LimitReadBuffer limited_payload(payload, 10000, /* trow_exception */ true, /* exact_limit */ {}, "too long MySQL packet."); - LOG_DEBUG(log, "Received command: {}. Connection id: {}.", - static_cast(static_cast(command)), connection_id); + LOG_DEBUG( + log, "Received command: {}. Connection id: {}.", static_cast(static_cast(command)), connection_id); if (!tcp_server.isOpen()) return; @@ -175,7 +180,7 @@ void MySQLHandler::run() comInitDB(limited_payload); break; case COM_QUERY: - comQuery(payload); + comQuery(payload, false); break; case COM_FIELD_LIST: comFieldList(limited_payload); @@ -227,13 +232,15 @@ void MySQLHandler::finishHandshake(MySQLProtocol::ConnectionPhase::HandshakeResp size_t pos = 0; /// Reads at least count and at most packet_size bytes. - auto read_bytes = [this, &buf, &pos, &packet_size](size_t count) -> void { + auto read_bytes = [this, &buf, &pos, &packet_size](size_t count) -> void + { while (pos < count) { int ret = socket().receiveBytes(buf + pos, static_cast(packet_size - pos)); if (ret == 0) { - throw Exception(ErrorCodes::CANNOT_READ_ALL_DATA, "Cannot read all data. Bytes read: {}. Bytes expected: 3", std::to_string(pos)); + throw Exception( + ErrorCodes::CANNOT_READ_ALL_DATA, "Cannot read all data. Bytes read: {}. Bytes expected: 3", std::to_string(pos)); } pos += ret; } @@ -272,7 +279,8 @@ void MySQLHandler::authenticate(const String & user_name, const String & auth_pl authPluginSSL(); } - std::optional auth_response = auth_plugin_name == auth_plugin->getName() ? std::make_optional(initial_auth_response) : std::nullopt; + std::optional auth_response + = auth_plugin_name == auth_plugin->getName() ? std::make_optional(initial_auth_response) : std::nullopt; auth_plugin->authenticate(user_name, *session, auth_response, packet_endpoint, secure_connection, socket().peerAddress()); } catch (const Exception & exc) @@ -304,8 +312,17 @@ void MySQLHandler::comFieldList(ReadBuffer & payload) for (const NameAndTypePair & column : metadata_snapshot->getColumns().getAll()) { ColumnDefinition column_definition( - database, packet.table, packet.table, column.name, column.name, CharacterSet::binary, 100, ColumnType::MYSQL_TYPE_STRING, 0, 0, true - ); + database, + packet.table, + packet.table, + column.name, + column.name, + CharacterSet::binary, + 100, + ColumnType::MYSQL_TYPE_STRING, + 0, + 0, + true); packet_endpoint->sendPacket(column_definition); } packet_endpoint->sendPacket(OKPacket(0xfe, client_capabilities, 0, 0, 0), true); @@ -318,7 +335,7 @@ void MySQLHandler::comPing() static bool isFederatedServerSetupSetCommand(const String & query); -void MySQLHandler::comQuery(ReadBuffer & payload) +void MySQLHandler::comQuery(ReadBuffer & payload, bool use_binary_protocol_result_set) { String query = String(payload.position(), payload.buffer().end()); @@ -350,20 +367,22 @@ void MySQLHandler::comQuery(ReadBuffer & payload) query_context->setCurrentQueryId(fmt::format("mysql:{}:{}", connection_id, toString(UUIDHelpers::generateV4()))); CurrentThread::QueryScope query_scope{query_context}; - std::atomic affected_rows {0}; + std::atomic affected_rows{0}; auto prev = query_context->getProgressCallback(); - query_context->setProgressCallback([&, my_prev = prev](const Progress & progress) - { - if (my_prev) - my_prev(progress); + query_context->setProgressCallback( + [&, my_prev = prev](const Progress & progress) + { + if (my_prev) + my_prev(progress); - affected_rows += progress.written_rows; - }); + affected_rows += progress.written_rows; + }); FormatSettings format_settings; format_settings.mysql_wire.client_capabilities = client_capabilities; format_settings.mysql_wire.max_packet_size = max_packet_size; format_settings.mysql_wire.sequence_id = &sequence_id; + format_settings.mysql_wire.use_binary_result_set = use_binary_protocol_result_set; auto set_result_details = [&with_output](const QueryResultDetails & details) { @@ -385,11 +404,18 @@ void MySQLHandler::comQuery(ReadBuffer & payload) void MySQLHandler::comStmtPrepare(DB::ReadBuffer & payload) { + if (prepared_statements_map.size() > 10000) /// Shouldn't happen in reality as COM_STMT_CLOSE cleans up the elements + { + LOG_ERROR(log, "Too many prepared statements"); + packet_endpoint->sendPacket(ERRPacket(), true); + return; + } + String query; readStringUntilEOF(query, payload); uint32_t statement_id = current_prepared_statement_id; - if (current_prepared_statement_id == std::numeric_limits::max()) [[unlikely]] + if (current_prepared_statement_id == std::numeric_limits::max()) { current_prepared_statement_id = 0; } @@ -400,7 +426,7 @@ void MySQLHandler::comStmtPrepare(DB::ReadBuffer & payload) // Key collisions should not happen here, as we remove the elements from the map with COM_STMT_CLOSE, // and we have quite a big range of available identifiers with 32-bit unsigned integer - if (prepared_statements_map.contains(statement_id)) [[unlikely]] + if (prepared_statements_map.contains(statement_id)) { LOG_ERROR( log, @@ -411,8 +437,8 @@ void MySQLHandler::comStmtPrepare(DB::ReadBuffer & payload) packet_endpoint->sendPacket(ERRPacket(), true); return; } - prepared_statements_map.emplace(statement_id, query); + prepared_statements_map.emplace(statement_id, query); packet_endpoint->sendPacket(PrepareStatementResponseOK(statement_id, 0, 0, 0), true); } @@ -421,7 +447,7 @@ void MySQLHandler::comStmtExecute(ReadBuffer & payload) uint32_t statement_id; payload.readStrict(reinterpret_cast(&statement_id), 4); - if (!prepared_statements_map.contains(statement_id)) [[unlikely]] + if (!prepared_statements_map.contains(statement_id)) { LOG_ERROR(log, "Could not find prepared statement with id {}", statement_id); packet_endpoint->sendPacket(ERRPacket(), true); @@ -430,14 +456,16 @@ void MySQLHandler::comStmtExecute(ReadBuffer & payload) // Temporary workaround as we work only with queries that do not bind any parameters atm ReadBufferFromString com_query_payload(prepared_statements_map.at(statement_id)); - MySQLHandler::comQuery(com_query_payload); + MySQLHandler::comQuery(com_query_payload, true); }; -void MySQLHandler::comStmtClose([[maybe_unused]] ReadBuffer & payload) { +void MySQLHandler::comStmtClose(ReadBuffer & payload) +{ uint32_t statement_id; payload.readStrict(reinterpret_cast(&statement_id), 4); - if (prepared_statements_map.contains(statement_id)) { + if (prepared_statements_map.contains(statement_id)) + { prepared_statements_map.erase(statement_id); } @@ -447,13 +475,17 @@ void MySQLHandler::comStmtClose([[maybe_unused]] ReadBuffer & payload) { void MySQLHandler::authPluginSSL() { - throw Exception(ErrorCodes::SUPPORT_IS_DISABLED, - "ClickHouse was built without SSL support. Try specifying password using double SHA1 in users.xml."); + throw Exception( + ErrorCodes::SUPPORT_IS_DISABLED, + "ClickHouse was built without SSL support. Try specifying password using double SHA1 in users.xml."); } void MySQLHandler::finishHandshakeSSL( - [[maybe_unused]] size_t packet_size, [[maybe_unused]] char * buf, [[maybe_unused]] size_t pos, - [[maybe_unused]] std::function read_bytes, [[maybe_unused]] MySQLProtocol::ConnectionPhase::HandshakeResponse & packet) + [[maybe_unused]] size_t packet_size, + [[maybe_unused]] char * buf, + [[maybe_unused]] size_t pos, + [[maybe_unused]] std::function read_bytes, + [[maybe_unused]] MySQLProtocol::ConnectionPhase::HandshakeResponse & packet) { throw Exception(ErrorCodes::SUPPORT_IS_DISABLED, "Client requested SSL, while it is disabled."); } @@ -467,10 +499,9 @@ MySQLHandlerSSL::MySQLHandlerSSL( uint32_t connection_id_, RSA & public_key_, RSA & private_key_) - : MySQLHandler(server_, tcp_server_, socket_, ssl_enabled, connection_id_) - , public_key(public_key_) - , private_key(private_key_) -{} + : MySQLHandler(server_, tcp_server_, socket_, ssl_enabled, connection_id_), public_key(public_key_), private_key(private_key_) +{ +} void MySQLHandlerSSL::authPluginSSL() { @@ -478,7 +509,10 @@ void MySQLHandlerSSL::authPluginSSL() } void MySQLHandlerSSL::finishHandshakeSSL( - size_t packet_size, char *buf, size_t pos, std::function read_bytes, + size_t packet_size, + char * buf, + size_t pos, + std::function read_bytes, MySQLProtocol::ConnectionPhase::HandshakeResponse & packet) { read_bytes(packet_size); /// Reading rest SSLRequest. @@ -508,8 +542,8 @@ static bool isFederatedServerSetupSetCommand(const String & query) "|(^(SET AUTOCOMMIT(.*)))" "|(^(SET sql_mode(.*)))" "|(^(SET @@(.*)))" - "|(^(SET SESSION TRANSACTION ISOLATION LEVEL(.*)))" - , std::regex::icase}; + "|(^(SET SESSION TRANSACTION ISOLATION LEVEL(.*)))", + std::regex::icase}; return 1 == std::regex_match(query, expr); } diff --git a/src/Server/MySQLHandler.h b/src/Server/MySQLHandler.h index 6b8cc56a46e..a412b647ae2 100644 --- a/src/Server/MySQLHandler.h +++ b/src/Server/MySQLHandler.h @@ -1,12 +1,12 @@ #pragma once -#include -#include -#include #include -#include #include +#include #include +#include +#include +#include #include "IServer.h" #include "config.h" @@ -19,7 +19,7 @@ namespace CurrentMetrics { - extern const Metric MySQLConnection; +extern const Metric MySQLConnection; } namespace DB @@ -32,11 +32,7 @@ class MySQLHandler : public Poco::Net::TCPServerConnection { public: MySQLHandler( - IServer & server_, - TCPServer & tcp_server_, - const Poco::Net::StreamSocket & socket_, - bool ssl_enabled, - uint32_t connection_id_); + IServer & server_, TCPServer & tcp_server_, const Poco::Net::StreamSocket & socket_, bool ssl_enabled, uint32_t connection_id_); void run() final; @@ -46,7 +42,7 @@ protected: /// Enables SSL, if client requested. void finishHandshake(MySQLProtocol::ConnectionPhase::HandshakeResponse &); - void comQuery(ReadBuffer & payload); + void comQuery(ReadBuffer & payload, bool use_binary_protocol_result_set); void comFieldList(ReadBuffer & payload); @@ -63,7 +59,12 @@ protected: void comStmtClose(ReadBuffer & payload); virtual void authPluginSSL(); - virtual void finishHandshakeSSL(size_t packet_size, char * buf, size_t pos, std::function read_bytes, MySQLProtocol::ConnectionPhase::HandshakeResponse & packet); + virtual void finishHandshakeSSL( + size_t packet_size, + char * buf, + size_t pos, + std::function read_bytes, + MySQLProtocol::ConnectionPhase::HandshakeResponse & packet); IServer & server; TCPServer & tcp_server; @@ -109,8 +110,11 @@ private: void authPluginSSL() override; void finishHandshakeSSL( - size_t packet_size, char * buf, size_t pos, - std::function read_bytes, MySQLProtocol::ConnectionPhase::HandshakeResponse & packet) override; + size_t packet_size, + char * buf, + size_t pos, + std::function read_bytes, + MySQLProtocol::ConnectionPhase::HandshakeResponse & packet) override; RSA & public_key; RSA & private_key; diff --git a/src/Storages/System/InformationSchema/key_column_usage.sql b/src/Storages/System/InformationSchema/key_column_usage.sql new file mode 100644 index 00000000000..43630b8c8b9 --- /dev/null +++ b/src/Storages/System/InformationSchema/key_column_usage.sql @@ -0,0 +1,38 @@ +ATTACH VIEW key_column_usage + ( + `referenced_table_schema` Nullable(String), + `referenced_table_name` Nullable(String), + `referenced_column_name` Nullable(String), + `table_schema` String, + `table_name` String, + `column_name` Nullable(String), + `ordinal_position` UInt32, + `constraint_name` Nullable(String), + `REFERENCED_TABLE_SCHEMA` Nullable(String), + `REFERENCED_TABLE_NAME` Nullable(String), + `REFERENCED_COLUMN_NAME` Nullable(String), + `TABLE_SCHEMA` String, + `TABLE_NAME` String, + `COLUMN_NAME` Nullable(String), + `ORDINAL_POSITION` UInt32, + `CONSTRAINT_NAME` Nullable(String) + ) AS +SELECT NULL AS `referenced_table_schema`, + NULL AS `referenced_table_name`, + NULL AS `referenced_column_name`, + database AS `table_schema`, + table AS `table_name`, + name AS `column_name`, + position AS `ordinal_position`, + 'PRIMARY' AS `constraint_name`, + + `referenced_table_schema` AS `REFERENCED_TABLE_SCHEMA`, + `referenced_table_name` AS `REFERENCED_TABLE_NAME`, + `referenced_column_name` AS `REFERENCED_COLUMN_NAME`, + `table_schema` AS `TABLE_SCHEMA`, + `table_name` AS `TABLE_NAME`, + `column_name` AS `COLUMN_NAME`, + `ordinal_position` AS `ORDINAL_POSITION`, + `constraint_name` AS `CONSTRAINT_NAME` +FROM system.columns +WHERE is_in_primary_key; \ No newline at end of file diff --git a/src/Storages/System/InformationSchema/referential_constraints.sql b/src/Storages/System/InformationSchema/referential_constraints.sql new file mode 100644 index 00000000000..8216b8fff83 --- /dev/null +++ b/src/Storages/System/InformationSchema/referential_constraints.sql @@ -0,0 +1,25 @@ +ATTACH VIEW referential_constraints + ( + `constraint_name` Nullable(String), + `constraint_schema` String, + `table_name` String, + `update_rule` String, + `delete_rule` String, + `CONSTRAINT_NAME` Nullable(String), + `CONSTRAINT_SCHEMA` String, + `TABLE_NAME` String, + `UPDATE_RULE` String, + `DELETE_RULE` String + ) AS +SELECT NULL AS `constraint_name`, + '' AS `constraint_schema`, + '' AS `table_name`, + '' AS `update_rule`, + '' AS `delete_rule`, + + NULL AS `CONSTRAINT_NAME`, + '' AS `CONSTRAINT_SCHEMA`, + '' AS `TABLE_NAME`, + '' AS `UPDATE_RULE`, + '' AS `DELETE_RULE` +WHERE false; \ No newline at end of file diff --git a/src/Storages/System/InformationSchema/schemata.sql b/src/Storages/System/InformationSchema/schemata.sql index 9686fcbf4fa..ca4ad4f7310 100644 --- a/src/Storages/System/InformationSchema/schemata.sql +++ b/src/Storages/System/InformationSchema/schemata.sql @@ -1,26 +1,33 @@ ATTACH VIEW schemata -( - `catalog_name` String, - `schema_name` String, - `schema_owner` String, - `default_character_set_catalog` Nullable(String), - `default_character_set_schema` Nullable(String), - `default_character_set_name` Nullable(String), - `sql_path` Nullable(String), - `CATALOG_NAME` String ALIAS catalog_name, - `SCHEMA_NAME` String ALIAS schema_name, - `SCHEMA_OWNER` String ALIAS schema_owner, - `DEFAULT_CHARACTER_SET_CATALOG` Nullable(String) ALIAS default_character_set_catalog, - `DEFAULT_CHARACTER_SET_SCHEMA` Nullable(String) ALIAS default_character_set_schema, - `DEFAULT_CHARACTER_SET_NAME` Nullable(String) ALIAS default_character_set_name, - `SQL_PATH` Nullable(String) ALIAS sql_path -) AS -SELECT - name AS catalog_name, - name AS schema_name, - 'default' AS schema_owner, - NULL AS default_character_set_catalog, - NULL AS default_character_set_schema, - NULL AS default_character_set_name, - NULL AS sql_path + ( + `catalog_name` String, + `schema_name` String, + `schema_owner` String, + `default_character_set_catalog` Nullable(String), + `default_character_set_schema` Nullable(String), + `default_character_set_name` Nullable(String), + `sql_path` Nullable(String), + `CATALOG_NAME` String, + `SCHEMA_NAME` String, + `SCHEMA_OWNER` String, + `DEFAULT_CHARACTER_SET_CATALOG` Nullable(String), + `DEFAULT_CHARACTER_SET_SCHEMA` Nullable(String), + `DEFAULT_CHARACTER_SET_NAME` Nullable(String), + `SQL_PATH` Nullable(String) + ) AS +SELECT name AS `catalog_name`, + name AS `schema_name`, + 'default' AS `schema_owner`, + NULL AS `default_character_set_catalog`, + NULL AS `default_character_set_schema`, + NULL AS `default_character_set_name`, + NULL AS `sql_path`, + + catalog_name AS `CATALOG_NAME`, + schema_name AS `SCHEMA_NAME`, + schema_owner AS `SCHEMA_OWNER`, + NULL AS `DEFAULT_CHARACTER_SET_CATALOG`, + NULL AS `DEFAULT_CHARACTER_SET_SCHEMA`, + NULL AS `DEFAULT_CHARACTER_SET_NAME`, + NULL AS `SQL_PATH` FROM system.databases diff --git a/src/Storages/System/InformationSchema/tables.sql b/src/Storages/System/InformationSchema/tables.sql index 8eea3713923..b3bbfa72517 100644 --- a/src/Storages/System/InformationSchema/tables.sql +++ b/src/Storages/System/InformationSchema/tables.sql @@ -1,17 +1,35 @@ ATTACH VIEW tables -( - `table_catalog` String, - `table_schema` String, - `table_name` String, - `table_type` Enum8('BASE TABLE' = 1, 'VIEW' = 2, 'FOREIGN TABLE' = 3, 'LOCAL TEMPORARY' = 4, 'SYSTEM VIEW' = 5), - `TABLE_CATALOG` String ALIAS table_catalog, - `TABLE_SCHEMA` String ALIAS table_schema, - `TABLE_NAME` String ALIAS table_name, - `TABLE_TYPE` Enum8('BASE TABLE' = 1, 'VIEW' = 2, 'FOREIGN TABLE' = 3, 'LOCAL TEMPORARY' = 4, 'SYSTEM VIEW' = 5) ALIAS table_type -) AS -SELECT - database AS table_catalog, - database AS table_schema, - name AS table_name, - multiIf(is_temporary, 4, engine like '%View', 2, engine LIKE 'System%', 5, has_own_data = 0, 3, 1) AS table_type -FROM system.tables + ( + `table_catalog` String, + `table_schema` String, + `table_name` String, + `table_type` String, + `table_comment` String, + `table_collation` String, + `TABLE_CATALOG` String, + `TABLE_SCHEMA` String, + `TABLE_NAME` String, + `TABLE_TYPE` String, + `TABLE_COMMENT` String, + `TABLE_COLLATION` String + ) AS +SELECT database AS `table_catalog`, + database AS `table_schema`, + name AS `table_name`, + comment AS `table_comment`, + multiIf( + is_temporary, 'LOCAL TEMPORARY', + engine LIKE '%View', 'VIEW', + engine LIKE 'System%', 'SYSTEM VIEW', + has_own_data = 0, 'FOREIGN TABLE', + 'BASE TABLE' + ) AS `table_type`, + 'utf8mb4_0900_ai_ci' AS `table_collation`, + + table_catalog AS `TABLE_CATALOG`, + table_schema AS `TABLE_SCHEMA`, + table_name AS `TABLE_NAME`, + table_comment AS `TABLE_COMMENT`, + table_type AS `TABLE_TYPE`, + table_collation AS `TABLE_COLLATION` +FROM system.tables \ No newline at end of file diff --git a/src/Storages/System/attachInformationSchemaTables.cpp b/src/Storages/System/attachInformationSchemaTables.cpp index 074a648d235..d4775bf0d4a 100644 --- a/src/Storages/System/attachInformationSchemaTables.cpp +++ b/src/Storages/System/attachInformationSchemaTables.cpp @@ -12,7 +12,8 @@ INCBIN(resource_schemata_sql, SOURCE_DIR "/src/Storages/System/InformationSchema INCBIN(resource_tables_sql, SOURCE_DIR "/src/Storages/System/InformationSchema/tables.sql"); INCBIN(resource_views_sql, SOURCE_DIR "/src/Storages/System/InformationSchema/views.sql"); INCBIN(resource_columns_sql, SOURCE_DIR "/src/Storages/System/InformationSchema/columns.sql"); - +INCBIN(resource_key_column_usage_sql, SOURCE_DIR "/src/Storages/System/InformationSchema/key_column_usage.sql"); +INCBIN(resource_referential_constraints_sql, SOURCE_DIR "/src/Storages/System/InformationSchema/referential_constraints.sql"); namespace DB { @@ -66,6 +67,8 @@ void attachInformationSchema(ContextMutablePtr context, IDatabase & information_ createInformationSchemaView(context, information_schema_database, "tables", std::string_view(reinterpret_cast(gresource_tables_sqlData), gresource_tables_sqlSize)); createInformationSchemaView(context, information_schema_database, "views", std::string_view(reinterpret_cast(gresource_views_sqlData), gresource_views_sqlSize)); createInformationSchemaView(context, information_schema_database, "columns", std::string_view(reinterpret_cast(gresource_columns_sqlData), gresource_columns_sqlSize)); + createInformationSchemaView(context, information_schema_database, "key_column_usage", std::string_view(reinterpret_cast(gresource_key_column_usage_sqlData), gresource_key_column_usage_sqlSize)); + createInformationSchemaView(context, information_schema_database, "referential_constraints", std::string_view(reinterpret_cast(gresource_referential_constraints_sqlData), gresource_referential_constraints_sqlSize)); } } From 2447eb027c37a68f409fec1352e1abce0e6cbf54 Mon Sep 17 00:00:00 2001 From: Andrey Zvonov <32552679+zvonand@users.noreply.github.com> Date: Mon, 28 Aug 2023 15:12:44 +0300 Subject: [PATCH 078/433] Update StorageHDFS.cpp --- src/Storages/HDFS/StorageHDFS.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/Storages/HDFS/StorageHDFS.cpp b/src/Storages/HDFS/StorageHDFS.cpp index dafdba15549..09c205930dc 100644 --- a/src/Storages/HDFS/StorageHDFS.cpp +++ b/src/Storages/HDFS/StorageHDFS.cpp @@ -261,7 +261,7 @@ namespace HDFSBuilderWrapper builder = createHDFSBuilder(uri_without_path + "/", context->getGlobalContext()->getConfigRef()); HDFSFSPtr fs = createHDFSFS(builder.get()); - auto res = LSWithRegexpMatching("/", fs, path_from_uri, context->getSettingsRef().ignore_eacces_multidirectory_globs) + auto res = LSWithRegexpMatching("/", fs, path_from_uri, context->getSettingsRef().ignore_eacces_multidirectory_globs); return res; } } From b6ff1b78c7acbb32d80082d71bf342b346f448f2 Mon Sep 17 00:00:00 2001 From: Andrey Zvonov <32552679+zvonand@users.noreply.github.com> Date: Mon, 28 Aug 2023 21:35:16 +0300 Subject: [PATCH 079/433] Update 02771_multidirectory_globs_storage_file.sh --- .../0_stateless/02771_multidirectory_globs_storage_file.sh | 3 --- 1 file changed, 3 deletions(-) diff --git a/tests/queries/0_stateless/02771_multidirectory_globs_storage_file.sh b/tests/queries/0_stateless/02771_multidirectory_globs_storage_file.sh index 4901e716be3..8c8faafba82 100755 --- a/tests/queries/0_stateless/02771_multidirectory_globs_storage_file.sh +++ b/tests/queries/0_stateless/02771_multidirectory_globs_storage_file.sh @@ -1,9 +1,6 @@ #!/usr/bin/env bash # Tags: no-replicated-database, no-parallel -#SELECT *, _file FROM file('02771/dir?/{subdir?1/data1,subdir2?/data2}.csv', CSV) WHERE _file == 'data1.csv'; -#SELECT *, _file FROM file('02771/dir?/{subdir?1/data1,subdir2?/data2}.csv', CSV) WHERE _file == 'data2.csv'; - set -eu CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) From 1c4d49c46e2f87505f45e30c02648898b3d37849 Mon Sep 17 00:00:00 2001 From: Andrey Zvonov <32552679+zvonand@users.noreply.github.com> Date: Tue, 29 Aug 2023 09:05:03 +0300 Subject: [PATCH 080/433] Update tests/queries/0_stateless/02771_multidirectory_globs_storage_file.sh Co-authored-by: SmitaRKulkarni <64093672+SmitaRKulkarni@users.noreply.github.com> --- .../0_stateless/02771_multidirectory_globs_storage_file.sh | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tests/queries/0_stateless/02771_multidirectory_globs_storage_file.sh b/tests/queries/0_stateless/02771_multidirectory_globs_storage_file.sh index 8c8faafba82..984ae47d222 100755 --- a/tests/queries/0_stateless/02771_multidirectory_globs_storage_file.sh +++ b/tests/queries/0_stateless/02771_multidirectory_globs_storage_file.sh @@ -8,7 +8,7 @@ CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) . "$CURDIR"/../shell_config.sh # Data preparation. -# Now we can get the user_files_path by use the table file function for trick. also we can get it by query as: +# Now we can get the user_files_path by using the file function, we can also get it by this query: # "insert into function file('exist.txt', 'CSV', 'val1 char') values ('aaaa'); select _path from file('exist.txt', 'CSV', 'val1 char')" user_files_path=$(clickhouse-client --query "select _path,_file from file('nonexist.txt', 'CSV', 'val1 char')" 2>&1 | grep Exception | awk '{gsub("/nonexist.txt","",$9); print $9}') From e9b0d3e4a2775bd3a9df5b167452e7971fcbc0ce Mon Sep 17 00:00:00 2001 From: Duc Canh Le Date: Tue, 29 Aug 2023 09:41:18 +0000 Subject: [PATCH 081/433] do not implicitly read pk and version columns in lonely parts if nnot necessary --- .../QueryPlan/ReadFromMergeTree.cpp | 13 ++++---- src/Processors/QueryPlan/ReadFromMergeTree.h | 2 +- ...ross_partitions_final_all_lonely.reference | 21 ++++++++++++ ...rge_across_partitions_final_all_lonely.sql | 32 +++++++++++++++++++ 4 files changed, 60 insertions(+), 8 deletions(-) create mode 100644 tests/queries/0_stateless/02868_no_merge_across_partitions_final_all_lonely.reference create mode 100644 tests/queries/0_stateless/02868_no_merge_across_partitions_final_all_lonely.sql diff --git a/src/Processors/QueryPlan/ReadFromMergeTree.cpp b/src/Processors/QueryPlan/ReadFromMergeTree.cpp index 0b5eb94dbac..02b3d7a07fa 100644 --- a/src/Processors/QueryPlan/ReadFromMergeTree.cpp +++ b/src/Processors/QueryPlan/ReadFromMergeTree.cpp @@ -979,7 +979,7 @@ static void addMergingFinal( Pipe ReadFromMergeTree::spreadMarkRangesAmongStreamsFinal( - RangesInDataParts && parts_with_ranges, size_t num_streams, const Names & column_names, ActionsDAGPtr & out_projection) + RangesInDataParts && parts_with_ranges, size_t num_streams, const Names & origin_column_names, const Names & column_names, ActionsDAGPtr & out_projection) { const auto & settings = context->getSettingsRef(); const auto data_settings = data.getSettings(); @@ -1141,17 +1141,16 @@ Pipe ReadFromMergeTree::spreadMarkRangesAmongStreamsFinal( if (sum_marks_in_lonely_parts < num_streams_for_lonely_parts * min_marks_for_concurrent_read && lonely_parts.size() < num_streams_for_lonely_parts) num_streams_for_lonely_parts = std::max((sum_marks_in_lonely_parts + min_marks_for_concurrent_read - 1) / min_marks_for_concurrent_read, lonely_parts.size()); - auto pipe = read(std::move(lonely_parts), column_names, ReadFromMergeTree::ReadType::Default, + auto pipe = read(std::move(lonely_parts), partition_pipes.empty() ? origin_column_names : column_names, ReadFromMergeTree::ReadType::Default, num_streams_for_lonely_parts, min_marks_for_concurrent_read, info.use_uncompressed_cache); /// Drop temporary columns, added by 'sorting_key_expr' if (!out_projection) out_projection = createProjection(pipe.getHeader()); - pipe.addSimpleTransform([sorting_expr](const Block & header) - { - return std::make_shared(header, sorting_expr); - }); + if (!partition_pipes.empty()) + pipe.addSimpleTransform([sorting_expr](const Block & header) + { return std::make_shared(header, sorting_expr); }); partition_pipes.emplace_back(std::move(pipe)); } @@ -1742,7 +1741,7 @@ Pipe ReadFromMergeTree::spreadMarkRanges( ::sort(column_names_to_read.begin(), column_names_to_read.end()); column_names_to_read.erase(std::unique(column_names_to_read.begin(), column_names_to_read.end()), column_names_to_read.end()); - return spreadMarkRangesAmongStreamsFinal(std::move(parts_with_ranges), num_streams, column_names_to_read, result_projection); + return spreadMarkRangesAmongStreamsFinal(std::move(parts_with_ranges), num_streams, result.column_names_to_read, column_names_to_read, result_projection); } else if (input_order_info) { diff --git a/src/Processors/QueryPlan/ReadFromMergeTree.h b/src/Processors/QueryPlan/ReadFromMergeTree.h index cb2a3a8ddf9..2a4b6022d49 100644 --- a/src/Processors/QueryPlan/ReadFromMergeTree.h +++ b/src/Processors/QueryPlan/ReadFromMergeTree.h @@ -314,7 +314,7 @@ private: const InputOrderInfoPtr & input_order_info); Pipe spreadMarkRangesAmongStreamsFinal( - RangesInDataParts && parts, size_t num_streams, const Names & column_names, ActionsDAGPtr & out_projection); + RangesInDataParts && parts, size_t num_streams, const Names & origin_column_names, const Names & column_names, ActionsDAGPtr & out_projection); ReadFromMergeTree::AnalysisResult getAnalysisResult() const; MergeTreeDataSelectAnalysisResultPtr analyzed_result_ptr; diff --git a/tests/queries/0_stateless/02868_no_merge_across_partitions_final_all_lonely.reference b/tests/queries/0_stateless/02868_no_merge_across_partitions_final_all_lonely.reference new file mode 100644 index 00000000000..9f4677b0f16 --- /dev/null +++ b/tests/queries/0_stateless/02868_no_merge_across_partitions_final_all_lonely.reference @@ -0,0 +1,21 @@ +(Expression) +ExpressionTransform × 16 +Header: max(val) UInt64: max(val) UInt64 UInt64(size = 0) + count() UInt64: count() UInt64 UInt64(size = 0) + (Aggregating) + Resize 4 → 16 + Header × 16 : max(val) UInt64: max(val) UInt64 UInt64(size = 0) + count() UInt64: count() UInt64 UInt64(size = 0) + AggregatingTransform × 4 + Header: max(val) UInt64: max(val) UInt64 UInt64(size = 0) + count() UInt64: count() UInt64 UInt64(size = 0) + StrictResize 4 → 4 + Header × 4 : val UInt64: val UInt64 UInt64(size = 0) + (Expression) + ExpressionTransform × 4 + Header: val UInt64: val UInt64 UInt64(size = 0) + (ReadFromMergeTree) + ExpressionTransform × 4 + Header: val UInt64: val UInt64 UInt64(size = 0) + MergeTreeThread × 4 0 → 1 + Header: val UInt64: val UInt64 UInt64(size = 0) diff --git a/tests/queries/0_stateless/02868_no_merge_across_partitions_final_all_lonely.sql b/tests/queries/0_stateless/02868_no_merge_across_partitions_final_all_lonely.sql new file mode 100644 index 00000000000..d2c40bb002a --- /dev/null +++ b/tests/queries/0_stateless/02868_no_merge_across_partitions_final_all_lonely.sql @@ -0,0 +1,32 @@ +DROP TABLE IF EXISTS all_lonely; + +CREATE TABLE all_lonely +( + `id` UInt64, + `dt` Date, + `val` UInt64, + `version` UInt64 +) +ENGINE = ReplacingMergeTree(version) +PARTITION BY dt +ORDER BY (id); + +INSERT INTO all_lonely SELECT number, '2022-10-28', number*10, 0 FROM numbers(10000); +INSERT INTO all_lonely SELECT number+500000, '2022-10-28', number*10, 1 FROM numbers(10000); +OPTIMIZE TABLE all_lonely PARTITION '2022-10-28' FINAL; + + +INSERT INTO all_lonely SELECT number, '2022-10-29', number*10, 0 FROM numbers(10000); +INSERT INTO all_lonely SELECT number+500000, '2022-10-29', number*10, 1 FROM numbers(10000); +OPTIMIZE TABLE all_lonely PARTITION '2022-10-29' FINAL; + +INSERT INTO all_lonely SELECT number, '2022-10-30', number*10, 0 FROM numbers(10000); +INSERT INTO all_lonely SELECT number+500000, '2022-10-30', number*10, 1 FROM numbers(10000); +OPTIMIZE TABLE all_lonely PARTITION '2022-10-30' FINAL; + + +INSERT INTO all_lonely SELECT number, '2022-10-31', number*10, 0 FROM numbers(10000); +INSERT INTO all_lonely SELECT number+500000, '2022-10-31', number*10, 1 FROM numbers(10000); +OPTIMIZE TABLE all_lonely PARTITION '2022-10-31' FINAL; + +EXPLAIN PIPELINE header=1 SELECT max(val), count(*) FROM all_lonely FINAL SETTINGS do_not_merge_across_partitions_select_final = 1, max_threads = 16; From 47a908dfa8dd1b03ee42d2bebaaba2ec35036baa Mon Sep 17 00:00:00 2001 From: alexX512 Date: Tue, 29 Aug 2023 10:45:34 +0000 Subject: [PATCH 082/433] Add support for nested supqueries for AggregationPartialResult transform and more comments --- src/Client/ClientBase.h | 13 ++- src/Core/Settings.h | 4 +- src/Processors/Formats/IOutputFormat.cpp | 18 ---- src/Processors/Formats/IOutputFormat.h | 2 - .../Formats/Impl/PrettyBlockOutputFormat.cpp | 19 ++++ .../Formats/Impl/PrettyBlockOutputFormat.h | 2 + src/Processors/IProcessor.h | 15 +++- src/Processors/LimitTransform.h | 2 +- src/Processors/QueryPlan/QueryPlan.cpp | 2 +- .../AggregatingPartialResultTransform.cpp | 37 ++++---- .../AggregatingPartialResultTransform.h | 3 + .../Transforms/AggregatingTransform.cpp | 2 + .../Transforms/AggregatingTransform.h | 15 +++- .../Transforms/ExpressionTransform.h | 2 +- .../LimitPartialResultTransform.cpp | 4 +- .../Transforms/LimitPartialResultTransform.h | 2 + .../Transforms/LimitsCheckingTransform.cpp | 6 -- .../Transforms/LimitsCheckingTransform.h | 4 +- .../MergeSortingPartialResultTransform.h | 2 + .../Transforms/MergeSortingTransform.h | 6 +- .../Transforms/PartialResultTransform.h | 9 +- src/QueryPipeline/Pipe.cpp | 29 +++++-- ...rting_result_during_query_execution.python | 70 +++++++++------ ...ng_result_during_query_execution.reference | 51 +++++++---- ...ating_result_during_query_execution.python | 86 ++++++++++--------- ...ng_result_during_query_execution.reference | 35 ++++++++ .../queries/0_stateless/helpers/tcp_client.py | 7 +- 27 files changed, 296 insertions(+), 151 deletions(-) diff --git a/src/Client/ClientBase.h b/src/Client/ClientBase.h index c5e40186962..8aa9fc42de2 100644 --- a/src/Client/ClientBase.h +++ b/src/Client/ClientBase.h @@ -271,9 +271,16 @@ protected: enum class PartialResultMode: UInt8 { - NotInit, /// Query doesn't show partial result before the first block with 0 rows. - Active, /// Query shows partial result after the first and before the second block with 0 rows. - Inactive /// Query doesn't show partial result at all. + /// Query doesn't show partial result before the first block with 0 rows. + /// The first block with 0 rows initializes the output table format using its header. + NotInit, + + /// Query shows partial result after the first and before the second block with 0 rows. + /// The second block with 0 rows indicates that that receiving blocks with partial result has been completed and next blocks will be with the full result. + Active, + + /// Query doesn't show partial result at all. + Inactive, }; PartialResultMode partial_result_mode = PartialResultMode::Inactive; diff --git a/src/Core/Settings.h b/src/Core/Settings.h index 08292034e01..dc69831533d 100644 --- a/src/Core/Settings.h +++ b/src/Core/Settings.h @@ -307,8 +307,8 @@ class IColumn; \ M(Bool, partial_result_on_first_cancel, false, "Allows query to return a partial result after cancel.", 0) \ \ - M(Milliseconds, partial_result_update_duration_ms, 0, "Duration of time in milliseconds between real-time updates of result table sent to the client during query execution.", 0) \ - M(UInt64, max_rows_in_partial_result, 10, "Max rows displayed to user after each real-time update of output table during query execution.", 0) \ + M(Milliseconds, partial_result_update_duration_ms, 0, "Interval (in milliseconds) for sending updates with partial data about the result table to the client (in interactive mode) during query execution. Setting to 0 disables partial results. Only supported for single-threaded GROUP BY without key, ORDER BY, LIMIT and OFFSET.", 0) \ + M(UInt64, max_rows_in_partial_result, 10, "Maximum rows to show in the partial result after every real-time update while the query runs (use partial result limit + OFFSET as a value in case of OFFSET in the query).", 0) \ /** Settings for testing hedged requests */ \ M(Milliseconds, sleep_in_send_tables_status_ms, 0, "Time to sleep in sending tables status response in TCPHandler", 0) \ M(Milliseconds, sleep_in_send_data_ms, 0, "Time to sleep in sending data in TCPHandler", 0) \ diff --git a/src/Processors/Formats/IOutputFormat.cpp b/src/Processors/Formats/IOutputFormat.cpp index 6104037a72b..e691e32a7bc 100644 --- a/src/Processors/Formats/IOutputFormat.cpp +++ b/src/Processors/Formats/IOutputFormat.cpp @@ -198,22 +198,4 @@ void IOutputFormat::finalize() finalized = true; } -void IOutputFormat::clearLastLines(size_t lines_number) -{ - /// http://en.wikipedia.org/wiki/ANSI_escape_code - #define MOVE_TO_PREV_LINE "\033[A" - #define CLEAR_TO_END_OF_LINE "\033[K" - - static const char * clear_prev_line = MOVE_TO_PREV_LINE \ - CLEAR_TO_END_OF_LINE; - - /// Move cursor to the beginning of line - writeCString("\r", out); - - for (size_t line = 0; line < lines_number; ++line) - { - writeCString(clear_prev_line, out); - } -} - } diff --git a/src/Processors/Formats/IOutputFormat.h b/src/Processors/Formats/IOutputFormat.h index fc85b0d063b..470d24e9a22 100644 --- a/src/Processors/Formats/IOutputFormat.h +++ b/src/Processors/Formats/IOutputFormat.h @@ -103,8 +103,6 @@ public: } } - void clearLastLines(size_t lines_number); - protected: friend class ParallelFormattingOutputFormat; diff --git a/src/Processors/Formats/Impl/PrettyBlockOutputFormat.cpp b/src/Processors/Formats/Impl/PrettyBlockOutputFormat.cpp index 6f8461232b1..6fa891297f6 100644 --- a/src/Processors/Formats/Impl/PrettyBlockOutputFormat.cpp +++ b/src/Processors/Formats/Impl/PrettyBlockOutputFormat.cpp @@ -390,9 +390,28 @@ void PrettyBlockOutputFormat::consumeExtremes(Chunk chunk) write(std::move(chunk), PortKind::Extremes); } +void PrettyBlockOutputFormat::clearLastLines(size_t lines_number) +{ + /// http://en.wikipedia.org/wiki/ANSI_escape_code + #define MOVE_TO_PREV_LINE "\033[A" + #define CLEAR_TO_END_OF_LINE "\033[K" + + static const char * clear_prev_line = MOVE_TO_PREV_LINE \ + CLEAR_TO_END_OF_LINE; + + /// Move cursor to the beginning of line + writeCString("\r", out); + + for (size_t line = 0; line < lines_number; ++line) + { + writeCString(clear_prev_line, out); + } +} + void PrettyBlockOutputFormat::consumePartialResult(Chunk chunk) { if (prev_partial_block_rows > 0) + /// number of rows + header line + footer line clearLastLines(prev_partial_block_rows + 2); prev_partial_block_rows = chunk.getNumRows(); diff --git a/src/Processors/Formats/Impl/PrettyBlockOutputFormat.h b/src/Processors/Formats/Impl/PrettyBlockOutputFormat.h index a916a66a620..92466dce3ff 100644 --- a/src/Processors/Formats/Impl/PrettyBlockOutputFormat.h +++ b/src/Processors/Formats/Impl/PrettyBlockOutputFormat.h @@ -27,6 +27,8 @@ protected: void consume(Chunk) override; void consumeTotals(Chunk) override; void consumeExtremes(Chunk) override; + + void clearLastLines(size_t lines_number); void consumePartialResult(Chunk) override; size_t total_rows = 0; diff --git a/src/Processors/IProcessor.h b/src/Processors/IProcessor.h index a156da7bad2..51a0bb1c121 100644 --- a/src/Processors/IProcessor.h +++ b/src/Processors/IProcessor.h @@ -237,8 +237,21 @@ public: throw Exception(ErrorCodes::NOT_IMPLEMENTED, "Method 'expandPipeline' is not implemented for {} processor", getName()); } + enum class PartialResultStatus + { + /// Processor currently doesn't support work with the partial result pipeline. + NotSupported, + + /// Processor can be skipped in the partial result pipeline. + SkipSupported, + + /// Processor creates a light-weight copy of itself in the partial result pipeline. + /// The copy can create snapshots of the original processor or transform small blocks of data in the same way as the original processor + FullSupported, + }; + virtual bool isPartialResultProcessor() const { return false; } - virtual bool supportPartialResultProcessor() const { return false; } + virtual PartialResultStatus getPartialResultProcessorSupportStatus() const { return PartialResultStatus::NotSupported; } /// In case if query was cancelled executor will wait till all processors finish their jobs. /// Generally, there is no reason to check this flag. However, it may be reasonable for long operations (e.g. i/o). diff --git a/src/Processors/LimitTransform.h b/src/Processors/LimitTransform.h index ebdcbe49670..eac5f9e8d6d 100644 --- a/src/Processors/LimitTransform.h +++ b/src/Processors/LimitTransform.h @@ -76,7 +76,7 @@ public: void setRowsBeforeLimitCounter(RowsBeforeLimitCounterPtr counter) override { rows_before_limit_at_least.swap(counter); } void setInputPortHasCounter(size_t pos) { ports_data[pos].input_port_has_counter = true; } - bool supportPartialResultProcessor() const override { return true; } + PartialResultStatus getPartialResultProcessorSupportStatus() const override { return PartialResultStatus::FullSupported; } }; } diff --git a/src/Processors/QueryPlan/QueryPlan.cpp b/src/Processors/QueryPlan/QueryPlan.cpp index a0588455690..824c29cb158 100644 --- a/src/Processors/QueryPlan/QueryPlan.cpp +++ b/src/Processors/QueryPlan/QueryPlan.cpp @@ -197,7 +197,7 @@ QueryPipelineBuilderPtr QueryPlan::buildQueryPipeline( else stack.push(Frame{.node = frame.node->children[next_child]}); - if (last_pipeline && has_partial_result_setting) + if (has_partial_result_setting && last_pipeline && !last_pipeline->isPartialResultActive()) last_pipeline->activatePartialResult(build_pipeline_settings.partial_result_limit, build_pipeline_settings.partial_result_duration_ms); } diff --git a/src/Processors/Transforms/AggregatingPartialResultTransform.cpp b/src/Processors/Transforms/AggregatingPartialResultTransform.cpp index fbfb9be572b..cf8ce72e096 100644 --- a/src/Processors/Transforms/AggregatingPartialResultTransform.cpp +++ b/src/Processors/Transforms/AggregatingPartialResultTransform.cpp @@ -8,33 +8,38 @@ AggregatingPartialResultTransform::AggregatingPartialResultTransform( UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_) : PartialResultTransform(input_header, output_header, partial_result_limit_, partial_result_duration_ms_) , aggregating_transform(std::move(aggregating_transform_)) + , transform_aggregator(input_header, aggregating_transform->params->params) {} +void AggregatingPartialResultTransform::transformPartialResult(Chunk & chunk) +{ + auto & params = aggregating_transform->params->params; + + bool no_more_keys = false; + AggregatedDataVariants variants; + ColumnRawPtrs key_columns(params.keys_size); + Aggregator::AggregateColumns aggregate_columns(params.aggregates_size); + + const UInt64 num_rows = chunk.getNumRows(); + transform_aggregator.executeOnBlock(chunk.detachColumns(), 0, num_rows, variants, key_columns, aggregate_columns, no_more_keys); + + auto transformed_block = transform_aggregator.convertToBlocks(variants, /*final*/ true, /*max_threads*/ 1).front(); + + chunk = convertToChunk(transformed_block); +} + PartialResultTransform::ShaphotResult AggregatingPartialResultTransform::getRealProcessorSnapshot() { std::lock_guard lock(aggregating_transform->snapshot_mutex); - - auto & params = aggregating_transform->params; - /// Currently not supported cases - /// TODO: check that insert results from prepareBlockAndFillWithoutKey return values without changing of the aggregator state - if (params->params.keys_size != 0 /// has at least one key for aggregation - || params->aggregator.hasTemporaryData() /// use external storage for aggregation - || aggregating_transform->many_data->variants.size() > 1) /// use more then one stream for aggregation - return {{}, SnaphotStatus::Stopped}; - if (aggregating_transform->is_generate_initialized) return {{}, SnaphotStatus::Stopped}; if (aggregating_transform->variants.empty()) return {{}, SnaphotStatus::NotReady}; - auto & aggregator = params->aggregator; - - auto prepared_data = aggregator.prepareVariantsToMerge(aggregating_transform->many_data->variants); - AggregatedDataVariantsPtr & first = prepared_data.at(0); - - aggregator.mergeWithoutKeyDataImpl(prepared_data); - auto block = aggregator.prepareBlockAndFillWithoutKeySnapshot(*first); + auto & snapshot_aggregator = aggregating_transform->params->aggregator; + auto & snapshot_variants = aggregating_transform->many_data->variants; + auto block = snapshot_aggregator.prepareBlockAndFillWithoutKeySnapshot(*snapshot_variants.at(0)); return {convertToChunk(block), SnaphotStatus::Ready}; } diff --git a/src/Processors/Transforms/AggregatingPartialResultTransform.h b/src/Processors/Transforms/AggregatingPartialResultTransform.h index e5e28dc7f7a..f7bac3a5394 100644 --- a/src/Processors/Transforms/AggregatingPartialResultTransform.h +++ b/src/Processors/Transforms/AggregatingPartialResultTransform.h @@ -1,5 +1,6 @@ #pragma once +#include #include #include @@ -17,10 +18,12 @@ public: String getName() const override { return "AggregatingPartialResultTransform"; } + void transformPartialResult(Chunk & chunk) override; ShaphotResult getRealProcessorSnapshot() override; private: AggregatingTransformPtr aggregating_transform; + Aggregator transform_aggregator; }; } diff --git a/src/Processors/Transforms/AggregatingTransform.cpp b/src/Processors/Transforms/AggregatingTransform.cpp index b4d2785bed2..b0faf0f96d4 100644 --- a/src/Processors/Transforms/AggregatingTransform.cpp +++ b/src/Processors/Transforms/AggregatingTransform.cpp @@ -664,11 +664,13 @@ void AggregatingTransform::consume(Chunk chunk) { auto block = getInputs().front().getHeader().cloneWithColumns(chunk.detachColumns()); block = materializeBlock(block); + LOG_DEBUG(log, "AggregatingTransform::consume. Merge Block columns {}", block.dumpNames()); if (!params->aggregator.mergeOnBlock(block, variants, no_more_keys)) is_consume_finished = true; } else { + LOG_DEBUG(log, "AggregatingTransform::consume. Execute Block"); if (!params->aggregator.executeOnBlock(chunk.detachColumns(), 0, num_rows, variants, key_columns, aggregate_columns, no_more_keys)) is_consume_finished = true; } diff --git a/src/Processors/Transforms/AggregatingTransform.h b/src/Processors/Transforms/AggregatingTransform.h index 649b90ae74a..b6f32bde389 100644 --- a/src/Processors/Transforms/AggregatingTransform.h +++ b/src/Processors/Transforms/AggregatingTransform.h @@ -170,7 +170,16 @@ public: void work() override; Processors expandPipeline() override; - bool supportPartialResultProcessor() const override { return true; } + PartialResultStatus getPartialResultProcessorSupportStatus() const override { + /// Currently AggregatingPartialResultTransform support only single-thread aggregation without key. + + /// TODO: check that insert results from aggregator.prepareBlockAndFillWithoutKey return values without + /// changing of the aggregator state when aggregation with keys will be supported in AggregatingPartialResultTransform. + bool is_partial_result_supported = params->params.keys_size == 0 /// Aggregation without key. + && many_data->variants.size() == 1; /// Use only one stream for aggregation. + + return is_partial_result_supported ? PartialResultStatus::FullSupported : PartialResultStatus::NotSupported; + } protected: void consume(Chunk chunk); @@ -217,6 +226,10 @@ private: bool is_consume_started = false; friend class AggregatingPartialResultTransform; + /// The mutex protects variables that are used for creating a snapshot of the current processor. + /// The current implementation of AggregatingPartialResultTransform uses the 'is_generate_initialized' variable to check + /// whether the processor has started sending data through the main pipeline, and the corresponding partial result processor should stop creating snapshots. + /// Additionally, the mutex protects the 'params->aggregator' and 'many_data->variants' variables, which are used to get data from them for a snapshot. std::mutex snapshot_mutex; void initGenerate(); diff --git a/src/Processors/Transforms/ExpressionTransform.h b/src/Processors/Transforms/ExpressionTransform.h index a7b5ef4f51b..8250f25f0f8 100644 --- a/src/Processors/Transforms/ExpressionTransform.h +++ b/src/Processors/Transforms/ExpressionTransform.h @@ -26,7 +26,7 @@ public: static Block transformHeader(Block header, const ActionsDAG & expression); - bool supportPartialResultProcessor() const override { return true; } + PartialResultStatus getPartialResultProcessorSupportStatus() const override { return PartialResultStatus::FullSupported; } protected: void transform(Chunk & chunk) override; diff --git a/src/Processors/Transforms/LimitPartialResultTransform.cpp b/src/Processors/Transforms/LimitPartialResultTransform.cpp index fdaa95aac6c..c9eaa9dc7dd 100644 --- a/src/Processors/Transforms/LimitPartialResultTransform.cpp +++ b/src/Processors/Transforms/LimitPartialResultTransform.cpp @@ -29,11 +29,11 @@ void LimitPartialResultTransform::transformPartialResult(Chunk & chunk) /// Check if some rows should be removed if (length < num_rows) { - auto columns = chunk.detachColumns(); UInt64 num_columns = chunk.getNumColumns(); + auto columns = chunk.detachColumns(); for (UInt64 i = 0; i < num_columns; ++i) - columns[i] = columns[i]->cut(offset, limit); + columns[i] = columns[i]->cut(offset, length); chunk.setColumns(std::move(columns), length); } diff --git a/src/Processors/Transforms/LimitPartialResultTransform.h b/src/Processors/Transforms/LimitPartialResultTransform.h index 0f5549323c0..3a0116b624d 100644 --- a/src/Processors/Transforms/LimitPartialResultTransform.h +++ b/src/Processors/Transforms/LimitPartialResultTransform.h @@ -23,6 +23,8 @@ public: String getName() const override { return "LimitPartialResultTransform"; } void transformPartialResult(Chunk & chunk) override; + /// LimitsTransform doesn't have a state which can be snapshoted + ShaphotResult getRealProcessorSnapshot() override { return {{}, SnaphotStatus::Stopped}; } private: UInt64 limit; diff --git a/src/Processors/Transforms/LimitsCheckingTransform.cpp b/src/Processors/Transforms/LimitsCheckingTransform.cpp index 487592f9809..0557f3f291e 100644 --- a/src/Processors/Transforms/LimitsCheckingTransform.cpp +++ b/src/Processors/Transforms/LimitsCheckingTransform.cpp @@ -75,10 +75,4 @@ void LimitsCheckingTransform::checkQuota(Chunk & chunk) } } -ProcessorPtr LimitsCheckingTransform::getPartialResultProcessor(const ProcessorPtr & /*current_processor*/, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) -{ - const auto & header = inputs.front().getHeader(); - return std::make_shared(header, partial_result_limit, partial_result_duration_ms); -} - } diff --git a/src/Processors/Transforms/LimitsCheckingTransform.h b/src/Processors/Transforms/LimitsCheckingTransform.h index e8635384129..eabb988dab6 100644 --- a/src/Processors/Transforms/LimitsCheckingTransform.h +++ b/src/Processors/Transforms/LimitsCheckingTransform.h @@ -33,13 +33,11 @@ public: void setQuota(const std::shared_ptr & quota_) { quota = quota_; } - bool supportPartialResultProcessor() const override { return true; } + PartialResultStatus getPartialResultProcessorSupportStatus() const override { return PartialResultStatus::SkipSupported; } protected: void transform(Chunk & chunk) override; - ProcessorPtr getPartialResultProcessor(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) override; - private: StreamLocalLimits limits; diff --git a/src/Processors/Transforms/MergeSortingPartialResultTransform.h b/src/Processors/Transforms/MergeSortingPartialResultTransform.h index 1eaebf9d902..781aa8e1265 100644 --- a/src/Processors/Transforms/MergeSortingPartialResultTransform.h +++ b/src/Processors/Transforms/MergeSortingPartialResultTransform.h @@ -17,6 +17,8 @@ public: String getName() const override { return "MergeSortingPartialResultTransform"; } + /// MergeSortingTransform always receives chunks in a sorted state, so transformation is not needed + void transformPartialResult(Chunk & /*chunk*/) override {} ShaphotResult getRealProcessorSnapshot() override; private: diff --git a/src/Processors/Transforms/MergeSortingTransform.h b/src/Processors/Transforms/MergeSortingTransform.h index e8a1f7a5853..67f098b4362 100644 --- a/src/Processors/Transforms/MergeSortingTransform.h +++ b/src/Processors/Transforms/MergeSortingTransform.h @@ -33,7 +33,7 @@ public: String getName() const override { return "MergeSortingTransform"; } - bool supportPartialResultProcessor() const override { return true; } + PartialResultStatus getPartialResultProcessorSupportStatus() const override { return PartialResultStatus::FullSupported; } protected: void consume(Chunk chunk) override; @@ -65,6 +65,10 @@ private: ProcessorPtr external_merging_sorted; friend class MergeSortingPartialResultTransform; + /// The mutex protects variables that are used for creating a snapshot of the current processor. + /// The current implementation of MergeSortingPartialResultTransform uses the 'generated_prefix' variable to check + /// whether the processor has started sending data through the main pipeline, and the corresponding partial result processor should stop creating snapshots. + /// Additionally, the mutex protects the 'chunks' variable and all variables in the 'remerge' function, which is used to transition 'chunks' to a sorted state. std::mutex snapshot_mutex; }; diff --git a/src/Processors/Transforms/PartialResultTransform.h b/src/Processors/Transforms/PartialResultTransform.h index 2bcf494400d..4fe87638f38 100644 --- a/src/Processors/Transforms/PartialResultTransform.h +++ b/src/Processors/Transforms/PartialResultTransform.h @@ -5,6 +5,11 @@ namespace DB { +/// Processors of this type are used to construct an auxiliary pipeline with processors corresponding to those in the main pipeline. +/// These processors work in two modes: +/// 1) Creating a snapshot of the corresponding processor from the main pipeline once per partial_result_duration_ms (period in milliseconds), and then sending the snapshot through the partial result pipeline. +/// 2) Transforming small blocks of data in the same way as the original processor and sending the transformed data through the partial result pipeline. +/// All processors of this type rely on the invariant that a new block from the previous processor of the partial result pipeline overwrites information about the previous block of the same previous processor. class PartialResultTransform : public IProcessor { public: @@ -42,8 +47,8 @@ protected: bool finished_getting_snapshots = false; - virtual void transformPartialResult(Chunk & /*chunk*/) {} - virtual ShaphotResult getRealProcessorSnapshot() { return {{}, SnaphotStatus::Stopped}; } + virtual void transformPartialResult(Chunk & /*chunk*/) = 0; + virtual ShaphotResult getRealProcessorSnapshot() = 0; // { return {{}, SnaphotStatus::Stopped}; } private: Stopwatch watch; diff --git a/src/QueryPipeline/Pipe.cpp b/src/QueryPipeline/Pipe.cpp index 15f6bea70a7..293d152ea65 100644 --- a/src/QueryPipeline/Pipe.cpp +++ b/src/QueryPipeline/Pipe.cpp @@ -407,12 +407,13 @@ void Pipe::addExtremesSource(ProcessorPtr source) void Pipe::activatePartialResult(UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_) { - if (!is_partial_result_active) - partial_result_ports.assign(output_ports.size(), nullptr); + if (is_partial_result_active) + throw Exception(ErrorCodes::LOGICAL_ERROR, "Partial result for Pipe should be initialized only once"); is_partial_result_active = true; partial_result_limit = partial_result_limit_; partial_result_duration_ms = partial_result_duration_ms_; + partial_result_ports.assign(output_ports.size(), nullptr); } static void dropPort(OutputPort *& port, Processors & processors, Processors * collected_processors) @@ -629,12 +630,13 @@ void Pipe::addPartialResultSimpleTransform(const ProcessorPtr & transform, size_ if (isPartialResultActive()) { auto & partial_result_port = partial_result_ports[partial_result_port_id]; + auto partial_result_status = transform->getPartialResultProcessorSupportStatus(); - if (!transform->supportPartialResultProcessor()) - { + if (partial_result_status == IProcessor::PartialResultStatus::NotSupported) dropPort(partial_result_port, *processors, collected_processors); + + if (partial_result_status != IProcessor::PartialResultStatus::FullSupported) return; - } auto partial_result_transform = IProcessor::getPartialResultProcessorPtr(transform, partial_result_limit, partial_result_duration_ms); @@ -651,23 +653,34 @@ void Pipe::addPartialResultTransform(const ProcessorPtr & transform) if (isPartialResultActive()) { size_t new_outputs_size = transform->getOutputs().size(); + auto partial_result_status = transform->getPartialResultProcessorSupportStatus(); - if (!transform->supportPartialResultProcessor()) + if (partial_result_status == IProcessor::PartialResultStatus::SkipSupported && new_outputs_size != partial_result_ports.size()) + throw Exception( + ErrorCodes::LOGICAL_ERROR, + "Cannot skip transform {} in the partial result part of the Pipe because it has {} output ports, but the partial result part expects {} output ports", + transform->getName(), + new_outputs_size, + partial_result_ports.size()); + + if (partial_result_status == IProcessor::PartialResultStatus::NotSupported) { for (auto & partial_result_port : partial_result_ports) dropPort(partial_result_port, *processors, collected_processors); partial_result_ports.assign(new_outputs_size, nullptr); - return; } + if (partial_result_status != IProcessor::PartialResultStatus::FullSupported) + return; + auto partial_result_transform = IProcessor::getPartialResultProcessorPtr(transform, partial_result_limit, partial_result_duration_ms); auto & inputs = partial_result_transform->getInputs(); if (inputs.size() != partial_result_ports.size()) throw Exception( ErrorCodes::LOGICAL_ERROR, - "Cannot add transform {} to Pipe because it has {} input ports, but {} expected", + "Cannot add partial result transform {} to Pipe because it has {} input ports, but {} expected", partial_result_transform->getName(), inputs.size(), partial_result_ports.size()); diff --git a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python index 38a9d82d8b6..1fadddb0871 100755 --- a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python +++ b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python @@ -11,51 +11,69 @@ sys.path.insert(0, os.path.join(CURDIR, "helpers")) from tcp_client import TCPClient -def main(): +def run_query_without_errors(query, support_partial_result): with TCPClient() as client: - client.sendQuery( - "SELECT number FROM numbers_mt(1e7+1) ORDER BY -number LIMIT 15 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 10" - ) + client.sendQuery(query) # external tables client.sendEmptyBlock() client.readHeader() # Partial result - _, partial_result = client.readDataWithoutProgress()[0] - assert ( - len(partial_result) > 0 - ), "Expected at least one block with a non-empty partial result before getting the full result" + partial_result = client.readDataWithoutProgress()[0] + if support_partial_result: + assert ( + len(partial_result.value) > 0 + ), "Expected at least one block with a non-empty partial result before getting the full result" - while True: - assert all( - a >= b for a, b in zip(partial_result, partial_result[1:]) - ), "Partial result always should be sorted for this test" + while True: + assert all( + a >= b for a, b in zip(partial_result.value, partial_result.value[1:]) + ), "Partial result always should be sorted for this test" - _, new_partial_result = client.readDataWithoutProgress( - need_print_info=False - )[0] - if len(new_partial_result) == 0: - break + new_partial_result = client.readDataWithoutProgress( + need_print_info=False + )[0] + if len(new_partial_result.value) == 0: + break - data_size = len(partial_result) - assert all( - partial_result[i] <= new_partial_result[i] for i in range(data_size) - ), f"New partial result values should always be greater then old one because a new block contains more information about the full data. New result {new_partial_result}. Previous result {partial_result}" + data_size = len(partial_result.value) + assert all( + partial_result.value[i] <= new_partial_result.value[i] for i in range(data_size) + ), f"New partial result values should always be greater then old one because a new block contains more information about the full data. New result {new_partial_result}. Previous result {partial_result}" + + partial_result = new_partial_result + else: + block_rows = len(partial_result.value) + assert ( + block_rows == 0 + ), f"Expected only empty partial result block before getting the full result, but block has {block_rows} rows" - partial_result = new_partial_result # Full result - _, full_result = client.readDataWithoutProgress()[0] + full_result = client.readDataWithoutProgress()[0] - data_size = len(partial_result) + data_size = len(partial_result.value) assert all( - partial_result[i] <= full_result[i] for i in range(data_size) + partial_result.value[i] <= full_result.value[i] for i in range(data_size) ), f"Full result values should always be greater then partial result values. Full result {full_result}. Partial result {partial_result}" - for result in full_result: + for result in full_result.value: print(result) +def main(): + # Request with partial result limit less then full limit + run_query_without_errors("SELECT number FROM numbers_mt(5e6+1) ORDER BY -number LIMIT 5 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 3", True) + + # Request with partial result limit greater then full limit + run_query_without_errors("SELECT number FROM numbers_mt(5e6+1) ORDER BY -number LIMIT 3 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 5", True) + + # Request with OFFSET + run_query_without_errors("SELECT number FROM numbers_mt(5e6+1) ORDER BY -number LIMIT 3 OFFSET 1 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 5", True) + + # Request with OFFSET greater then partial result limit (partial result pipeline use blocks with less then OFFSET, so there will be no elements in block after LimitPartialResultTransform) + run_query_without_errors("SELECT number FROM numbers_mt(5e6+1) ORDER BY -number LIMIT 3 OFFSET 15 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 5", False) + if __name__ == "__main__": main() diff --git a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.reference b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.reference index 4d706474b22..211a193940a 100644 --- a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.reference +++ b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.reference @@ -1,21 +1,38 @@ Rows 0 Columns 1 Column number type UInt64 -Rows 10 Columns 1 +Rows 3 Columns 1 Column number type UInt64 -Rows 15 Columns 1 +Rows 5 Columns 1 Column number type UInt64 -10000000 -9999999 -9999998 -9999997 -9999996 -9999995 -9999994 -9999993 -9999992 -9999991 -9999990 -9999989 -9999988 -9999987 -9999986 +5000000 +4999999 +4999998 +4999997 +4999996 +Rows 0 Columns 1 +Column number type UInt64 +Rows 3 Columns 1 +Column number type UInt64 +Rows 3 Columns 1 +Column number type UInt64 +5000000 +4999999 +4999998 +Rows 0 Columns 1 +Column number type UInt64 +Rows 3 Columns 1 +Column number type UInt64 +Rows 3 Columns 1 +Column number type UInt64 +4999999 +4999998 +4999997 +Rows 0 Columns 1 +Column number type UInt64 +Rows 0 Columns 1 +Column number type UInt64 +Rows 3 Columns 1 +Column number type UInt64 +4999985 +4999984 +4999983 diff --git a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python index 25e9e6b73c9..4306ae577d0 100644 --- a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python +++ b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python @@ -11,11 +11,31 @@ sys.path.insert(0, os.path.join(CURDIR, "helpers")) from tcp_client import TCPClient -def get_keys(result): - return [key for key, _ in rasults] +def get_keys(results): + return [key for key, _ in results] -def run_query_without_errors(query, support_partial_result, invariants=None): +def check_new_result(new_results, old_results, invariants, rows_limit): + if rows_limit is not None: + assert ( + len(new_results[0].value) <= rows_limit + ), f"Result should have no more then {rows_limit} rows. But it has {len(new_results[0].value)} rows" + + for new_result, old_result in zip(new_results, old_results): + assert ( + new_result.key == old_result.key + ), f"Keys in blocks should be in the same order. Full results keys {get_keys(full_results)}. Partial results keys {get_keys(partial_results)}" + + key = new_result.key + if key in invariants: + new_value = new_result.value + old_value = old_result.value + assert invariants[key]( + old_value, new_value + ), f"Problem with the invariant between new and old result for key: {key}. New value {new_value}. Old value {old_value}" + + +def run_query_without_errors(query, support_partial_result, invariants=None, rows_limit=None): if invariants is None: invariants = {} @@ -30,54 +50,35 @@ def run_query_without_errors(query, support_partial_result, invariants=None): partial_results = client.readDataWithoutProgress() if support_partial_result: assert ( - len(partial_results[0][1]) > 0 + len(partial_results) > 0 and len(partial_results[0].value) > 0 ), "Expected at least one block with a non-empty partial result before getting the full result" while True: new_partial_results = client.readDataWithoutProgress( need_print_info=False ) - if len(new_partial_results[0][1]) == 0: + if len(new_partial_results[0].value) == 0: break - for new_result, old_result in zip(new_partial_results, partial_results): - assert ( - new_result[0] == old_result[0] - ), "Keys in blocks should be in the same order" - - key = new_result[0] - if key in invariants: - old_value = old_result[1] - new_value = new_result[1] - assert invariants[key]( - old_value, new_value - ), f"Problem with the invariant between old and new versions of a partial result for key: {key}. Old value {old_value}, new value {new_value}" + check_new_result(new_partial_results, partial_results, invariants, rows_limit) + partial_results = new_partial_results else: + block_rows = len(partial_results[0].value) assert ( - len(partial_results[0][1]) == 0 - ), "Expected no non-empty partial result blocks before getting the full result" + block_rows == 0 + ), f"Expected only empty partial result block before getting the full result, but block has {block_rows} rows" # Full result full_results = client.readDataWithoutProgress() if support_partial_result: - for full_result, partial_result in zip(full_results, partial_results): - assert ( - full_result[0] == partial_result[0] - ), f"Keys in blocks should be in the same order. Full results keys {get_keys(full_results)}. Partial results keys {get_keys(partial_results)}" + check_new_result(full_results, partial_results, invariants, rows_limit) - key = full_result[0] - if key in invariants: - full_value = full_result[1] - partial_value = partial_result[1] - assert invariants[key]( - partial_value, full_value - ), f"Problem with the invariant between full and partial result for key: {key}. Partial value {partial_value}. Full value {full_value}" - - for key, value in full_results: - if isinstance(value[0], int): - print(key, value) + for data in full_results: + if isinstance(data.value[0], int): + print(data.key, data.value) -def supported_scenarios(): +def supported_scenarios_without_key(): + # Simple aggregation query query = "select median(number), stddevSamp(number), stddevPop(number), max(number), min(number), any(number), count(number), avg(number), sum(number) from numbers_mt(1e7+1) settings max_threads = 1, partial_result_update_duration_ms = 1" invariants = { "median(number)": lambda old_value, new_value: old_value <= new_value, @@ -87,21 +88,28 @@ def supported_scenarios(): "avg(number)": lambda old_value, new_value: old_value <= new_value, "sum(number)": lambda old_value, new_value: old_value <= new_value, } - run_query_without_errors(query, support_partial_result=True, invariants=invariants) + run_query_without_errors(query, support_partial_result=True, invariants=invariants, rows_limit=1) + + # Aggregation query with a nested ORDER BY subquery + query = "select median(number), stddevSamp(number), stddevPop(number), max(number), min(number), any(number), count(number), avg(number), sum(number) FROM (SELECT number FROM numbers_mt(1e7) ORDER BY -number LIMIT 3) settings max_threads = 1, partial_result_update_duration_ms=1" + + # Aggregation receives small partial result blocks from ORDER BY which always sends blocks with bigger values + invariants["min(number)"] = lambda old_value, new_value: old_value <= new_value + run_query_without_errors(query, support_partial_result=True, invariants=invariants, rows_limit=1) def unsupported_scenarios(): # Currently aggregator for partial result supports only single thread aggregation without key # Update test when multithreading or aggregation with GROUP BY will be supported for partial result updates - multithread_query = "select sum(number) from numbers_mt(1e7+1) settings max_threads = 2, partial_result_update_duration_ms = 1" + multithread_query = "select sum(number) from numbers_mt(1e7+1) settings max_threads = 2, partial_result_update_duration_ms = 100" run_query_without_errors(multithread_query, support_partial_result=False) - group_with_key_query = "select mod2, sum(number) from numbers_mt(1e7+1) group by number % 2 as mod2 settings max_threads = 1, partial_result_update_duration_ms = 1" + group_with_key_query = "select mod2, sum(number) from numbers_mt(1e7+1) group by number % 2 as mod2 settings max_threads = 1, partial_result_update_duration_ms = 100" run_query_without_errors(group_with_key_query, support_partial_result=False) def main(): - supported_scenarios() + supported_scenarios_without_key() unsupported_scenarios() diff --git a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.reference b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.reference index a97ee3f8d79..a813b18f24f 100644 --- a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.reference +++ b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.reference @@ -33,6 +33,41 @@ min(number) [0] any(number) [0] count(number) [10000001] sum(number) [50000005000000] +Rows 0 Columns 9 +Column median(number) type Float64 +Column stddevSamp(number) type Float64 +Column stddevPop(number) type Float64 +Column max(number) type UInt64 +Column min(number) type UInt64 +Column any(number) type UInt64 +Column count(number) type UInt64 +Column avg(number) type Float64 +Column sum(number) type UInt64 +Rows 1 Columns 9 +Column median(number) type Float64 +Column stddevSamp(number) type Float64 +Column stddevPop(number) type Float64 +Column max(number) type UInt64 +Column min(number) type UInt64 +Column any(number) type UInt64 +Column count(number) type UInt64 +Column avg(number) type Float64 +Column sum(number) type UInt64 +Rows 1 Columns 9 +Column median(number) type Float64 +Column stddevSamp(number) type Float64 +Column stddevPop(number) type Float64 +Column max(number) type UInt64 +Column min(number) type UInt64 +Column any(number) type UInt64 +Column count(number) type UInt64 +Column avg(number) type Float64 +Column sum(number) type UInt64 +max(number) [9999999] +min(number) [9999997] +any(number) [9999999] +count(number) [3] +sum(number) [29999994] Rows 0 Columns 1 Column sum(number) type UInt64 Rows 0 Columns 1 diff --git a/tests/queries/0_stateless/helpers/tcp_client.py b/tests/queries/0_stateless/helpers/tcp_client.py index bab449d71fe..e0989a000f4 100644 --- a/tests/queries/0_stateless/helpers/tcp_client.py +++ b/tests/queries/0_stateless/helpers/tcp_client.py @@ -57,6 +57,11 @@ def assertPacket(packet, expected): assert packet == expected, "Got: {}, expected: {}".format(packet, expected) +class Data(object): + def __init__(self, key, value): + self.key = key + self.value = value + class TCPClient(object): def __init__(self, timeout=30): self.timeout = timeout @@ -301,6 +306,6 @@ class TCPClient(object): if need_print_info: print("Column {} type {}".format(col_name, type_name)) - data.append((col_name, self.readRow(type_name, rows))) + data.append(Data(col_name, self.readRow(type_name, rows))) return data From 238f6e832773a6546b8a7313bc49922de71ca780 Mon Sep 17 00:00:00 2001 From: alexX512 Date: Tue, 29 Aug 2023 13:15:38 +0000 Subject: [PATCH 083/433] fix style --- .../Transforms/AggregatingTransform.cpp | 2 -- ...rting_result_during_query_execution.python | 27 ++++++++++++++----- ...ating_result_during_query_execution.python | 18 +++++++++---- .../queries/0_stateless/helpers/tcp_client.py | 1 + 4 files changed, 34 insertions(+), 14 deletions(-) diff --git a/src/Processors/Transforms/AggregatingTransform.cpp b/src/Processors/Transforms/AggregatingTransform.cpp index b0faf0f96d4..b4d2785bed2 100644 --- a/src/Processors/Transforms/AggregatingTransform.cpp +++ b/src/Processors/Transforms/AggregatingTransform.cpp @@ -664,13 +664,11 @@ void AggregatingTransform::consume(Chunk chunk) { auto block = getInputs().front().getHeader().cloneWithColumns(chunk.detachColumns()); block = materializeBlock(block); - LOG_DEBUG(log, "AggregatingTransform::consume. Merge Block columns {}", block.dumpNames()); if (!params->aggregator.mergeOnBlock(block, variants, no_more_keys)) is_consume_finished = true; } else { - LOG_DEBUG(log, "AggregatingTransform::consume. Execute Block"); if (!params->aggregator.executeOnBlock(chunk.detachColumns(), 0, num_rows, variants, key_columns, aggregate_columns, no_more_keys)) is_consume_finished = true; } diff --git a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python index 1fadddb0871..ec13a41d07b 100755 --- a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python +++ b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python @@ -28,7 +28,8 @@ def run_query_without_errors(query, support_partial_result): while True: assert all( - a >= b for a, b in zip(partial_result.value, partial_result.value[1:]) + a >= b + for a, b in zip(partial_result.value, partial_result.value[1:]) ), "Partial result always should be sorted for this test" new_partial_result = client.readDataWithoutProgress( @@ -39,7 +40,8 @@ def run_query_without_errors(query, support_partial_result): data_size = len(partial_result.value) assert all( - partial_result.value[i] <= new_partial_result.value[i] for i in range(data_size) + partial_result.value[i] <= new_partial_result.value[i] + for i in range(data_size) ), f"New partial result values should always be greater then old one because a new block contains more information about the full data. New result {new_partial_result}. Previous result {partial_result}" partial_result = new_partial_result @@ -49,7 +51,6 @@ def run_query_without_errors(query, support_partial_result): block_rows == 0 ), f"Expected only empty partial result block before getting the full result, but block has {block_rows} rows" - # Full result full_result = client.readDataWithoutProgress()[0] @@ -64,16 +65,28 @@ def run_query_without_errors(query, support_partial_result): def main(): # Request with partial result limit less then full limit - run_query_without_errors("SELECT number FROM numbers_mt(5e6+1) ORDER BY -number LIMIT 5 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 3", True) + run_query_without_errors( + "SELECT number FROM numbers_mt(5e6+1) ORDER BY -number LIMIT 5 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 3", + support_partial_result=True, + ) # Request with partial result limit greater then full limit - run_query_without_errors("SELECT number FROM numbers_mt(5e6+1) ORDER BY -number LIMIT 3 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 5", True) + run_query_without_errors( + "SELECT number FROM numbers_mt(5e6+1) ORDER BY -number LIMIT 3 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 5", + support_partial_result=True, + ) # Request with OFFSET - run_query_without_errors("SELECT number FROM numbers_mt(5e6+1) ORDER BY -number LIMIT 3 OFFSET 1 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 5", True) + run_query_without_errors( + "SELECT number FROM numbers_mt(5e6+1) ORDER BY -number LIMIT 3 OFFSET 1 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 5", + support_partial_result=True, + ) # Request with OFFSET greater then partial result limit (partial result pipeline use blocks with less then OFFSET, so there will be no elements in block after LimitPartialResultTransform) - run_query_without_errors("SELECT number FROM numbers_mt(5e6+1) ORDER BY -number LIMIT 3 OFFSET 15 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 5", False) + run_query_without_errors( + "SELECT number FROM numbers_mt(5e6+1) ORDER BY -number LIMIT 3 OFFSET 15 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 5", + support_partial_result=False, + ) if __name__ == "__main__": main() diff --git a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python index 4306ae577d0..4d869b05580 100644 --- a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python +++ b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python @@ -35,7 +35,9 @@ def check_new_result(new_results, old_results, invariants, rows_limit): ), f"Problem with the invariant between new and old result for key: {key}. New value {new_value}. Old value {old_value}" -def run_query_without_errors(query, support_partial_result, invariants=None, rows_limit=None): +def run_query_without_errors( + query, support_partial_result, invariants=None, rows_limit=None +): if invariants is None: invariants = {} @@ -59,7 +61,9 @@ def run_query_without_errors(query, support_partial_result, invariants=None, row if len(new_partial_results[0].value) == 0: break - check_new_result(new_partial_results, partial_results, invariants, rows_limit) + check_new_result( + new_partial_results, partial_results, invariants, rows_limit + ) partial_results = new_partial_results else: block_rows = len(partial_results[0].value) @@ -88,14 +92,18 @@ def supported_scenarios_without_key(): "avg(number)": lambda old_value, new_value: old_value <= new_value, "sum(number)": lambda old_value, new_value: old_value <= new_value, } - run_query_without_errors(query, support_partial_result=True, invariants=invariants, rows_limit=1) + run_query_without_errors( + query, support_partial_result=True, invariants=invariants, rows_limit=1 + ) # Aggregation query with a nested ORDER BY subquery query = "select median(number), stddevSamp(number), stddevPop(number), max(number), min(number), any(number), count(number), avg(number), sum(number) FROM (SELECT number FROM numbers_mt(1e7) ORDER BY -number LIMIT 3) settings max_threads = 1, partial_result_update_duration_ms=1" # Aggregation receives small partial result blocks from ORDER BY which always sends blocks with bigger values - invariants["min(number)"] = lambda old_value, new_value: old_value <= new_value - run_query_without_errors(query, support_partial_result=True, invariants=invariants, rows_limit=1) + invariants["min(number)"] = lambda old_value, new_value: old_value <= new_value + run_query_without_errors( + query, support_partial_result=True, invariants=invariants, rows_limit=1 + ) def unsupported_scenarios(): diff --git a/tests/queries/0_stateless/helpers/tcp_client.py b/tests/queries/0_stateless/helpers/tcp_client.py index e9ce01f45ff..fdc4ab28e04 100644 --- a/tests/queries/0_stateless/helpers/tcp_client.py +++ b/tests/queries/0_stateless/helpers/tcp_client.py @@ -63,6 +63,7 @@ class Data(object): self.key = key self.value = value + class TCPClient(object): def __init__(self, timeout=30): self.timeout = timeout From 70f926a75a68417e3ffba7a8c55ea8145d245c80 Mon Sep 17 00:00:00 2001 From: alexX512 Date: Tue, 29 Aug 2023 13:37:46 +0000 Subject: [PATCH 084/433] Style fix --- src/Processors/Transforms/AggregatingTransform.h | 3 ++- .../02833_partial_sorting_result_during_query_execution.python | 1 + 2 files changed, 3 insertions(+), 1 deletion(-) diff --git a/src/Processors/Transforms/AggregatingTransform.h b/src/Processors/Transforms/AggregatingTransform.h index b6f32bde389..791cd12326f 100644 --- a/src/Processors/Transforms/AggregatingTransform.h +++ b/src/Processors/Transforms/AggregatingTransform.h @@ -170,7 +170,8 @@ public: void work() override; Processors expandPipeline() override; - PartialResultStatus getPartialResultProcessorSupportStatus() const override { + PartialResultStatus getPartialResultProcessorSupportStatus() const override + { /// Currently AggregatingPartialResultTransform support only single-thread aggregation without key. /// TODO: check that insert results from aggregator.prepareBlockAndFillWithoutKey return values without diff --git a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python index ec13a41d07b..b3d04d263dc 100755 --- a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python +++ b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python @@ -88,5 +88,6 @@ def main(): support_partial_result=False, ) + if __name__ == "__main__": main() From 02391eafe1dd7a05b4b92f01c05667b207f8d209 Mon Sep 17 00:00:00 2001 From: Anton Popov Date: Tue, 29 Aug 2023 13:38:51 +0000 Subject: [PATCH 085/433] refactoring of reading from MergeTree tables --- .../QueryPlan/ReadFromMergeTree.cpp | 496 +++++++------ src/Processors/QueryPlan/ReadFromMergeTree.h | 42 +- src/Storages/MergeTree/IMergeTreeReadPool.h | 15 +- src/Storages/MergeTree/IMergeTreeReader.h | 1 - .../MergeTreeBaseSelectProcessor.cpp | 696 ------------------ .../MergeTree/MergeTreeBaseSelectProcessor.h | 217 ------ .../MergeTree/MergeTreeBlockReadUtils.cpp | 64 +- .../MergeTree/MergeTreeBlockReadUtils.h | 77 +- src/Storages/MergeTree/MergeTreeData.cpp | 2 +- .../MergeTreeInOrderSelectProcessor.cpp | 76 -- .../MergeTreeInOrderSelectProcessor.h | 33 - .../MergeTree/MergeTreePrefetchedReadPool.cpp | 612 +++++++-------- .../MergeTree/MergeTreePrefetchedReadPool.h | 134 ++-- src/Storages/MergeTree/MergeTreeReadPool.cpp | 325 ++------ src/Storages/MergeTree/MergeTreeReadPool.h | 195 +---- .../MergeTree/MergeTreeReadPoolBase.cpp | 149 ++++ .../MergeTree/MergeTreeReadPoolBase.h | 67 ++ .../MergeTree/MergeTreeReadPoolInOrder.cpp | 74 ++ .../MergeTree/MergeTreeReadPoolInOrder.h | 35 + .../MergeTreeReadPoolParallelReplicas.cpp | 110 +++ .../MergeTreeReadPoolParallelReplicas.h | 39 + ...rgeTreeReadPoolParallelReplicasInOrder.cpp | 106 +++ ...MergeTreeReadPoolParallelReplicasInOrder.h | 40 + src/Storages/MergeTree/MergeTreeReadTask.cpp | 193 +++++ src/Storages/MergeTree/MergeTreeReadTask.h | 156 ++++ .../MergeTreeReverseSelectProcessor.cpp | 134 ---- .../MergeTreeReverseSelectProcessor.h | 44 -- .../MergeTree/MergeTreeSelectAlgorithms.cpp | 87 +++ .../MergeTree/MergeTreeSelectAlgorithms.h | 64 ++ .../MergeTree/MergeTreeSelectProcessor.cpp | 478 ++++++++++-- .../MergeTree/MergeTreeSelectProcessor.h | 172 +++-- src/Storages/MergeTree/MergeTreeSource.cpp | 26 +- src/Storages/MergeTree/MergeTreeSource.h | 8 +- .../MergeTreeThreadSelectProcessor.cpp | 73 -- .../MergeTreeThreadSelectProcessor.h | 54 -- .../01861_explain_pipeline.reference | 6 +- ...02149_read_in_order_fixed_prefix.reference | 24 +- ...mize_aggregation_in_order_prefix.reference | 4 +- ...ct_in_order_optimization_explain.reference | 14 +- ..._distinct_in_order_optimization_explain.sh | 4 +- .../02340_parts_refcnt_mergetree.reference | 4 +- .../02343_aggregation_pipeline.reference | 6 +- .../02343_aggregation_pipeline.sql | 2 + .../02404_memory_bound_merging.reference | 6 +- .../02404_memory_bound_merging.sql | 4 +- ...x_streams_for_merge_tree_reading.reference | 16 +- ...493_max_streams_for_merge_tree_reading.sql | 11 +- .../02521_aggregation_by_partitions.reference | 112 +-- .../02521_aggregation_by_partitions.sql | 3 + ...rouping_sets_predicate_push_down.reference | 4 +- ...2_async_reading_with_small_limit.reference | 2 +- .../02582_async_reading_with_small_limit.sql | 2 + ...stinct_reading_in_order_analyzer.reference | 2 +- ...676_distinct_reading_in_order_analyzer.sql | 2 +- ...inal_streams_data_skipping_index.reference | 8 +- ...explain_settings_not_applied_bug.reference | 20 +- 56 files changed, 2503 insertions(+), 2847 deletions(-) delete mode 100644 src/Storages/MergeTree/MergeTreeBaseSelectProcessor.cpp delete mode 100644 src/Storages/MergeTree/MergeTreeBaseSelectProcessor.h delete mode 100644 src/Storages/MergeTree/MergeTreeInOrderSelectProcessor.cpp delete mode 100644 src/Storages/MergeTree/MergeTreeInOrderSelectProcessor.h create mode 100644 src/Storages/MergeTree/MergeTreeReadPoolBase.cpp create mode 100644 src/Storages/MergeTree/MergeTreeReadPoolBase.h create mode 100644 src/Storages/MergeTree/MergeTreeReadPoolInOrder.cpp create mode 100644 src/Storages/MergeTree/MergeTreeReadPoolInOrder.h create mode 100644 src/Storages/MergeTree/MergeTreeReadPoolParallelReplicas.cpp create mode 100644 src/Storages/MergeTree/MergeTreeReadPoolParallelReplicas.h create mode 100644 src/Storages/MergeTree/MergeTreeReadPoolParallelReplicasInOrder.cpp create mode 100644 src/Storages/MergeTree/MergeTreeReadPoolParallelReplicasInOrder.h create mode 100644 src/Storages/MergeTree/MergeTreeReadTask.cpp create mode 100644 src/Storages/MergeTree/MergeTreeReadTask.h delete mode 100644 src/Storages/MergeTree/MergeTreeReverseSelectProcessor.cpp delete mode 100644 src/Storages/MergeTree/MergeTreeReverseSelectProcessor.h create mode 100644 src/Storages/MergeTree/MergeTreeSelectAlgorithms.cpp create mode 100644 src/Storages/MergeTree/MergeTreeSelectAlgorithms.h delete mode 100644 src/Storages/MergeTree/MergeTreeThreadSelectProcessor.cpp delete mode 100644 src/Storages/MergeTree/MergeTreeThreadSelectProcessor.h diff --git a/src/Processors/QueryPlan/ReadFromMergeTree.cpp b/src/Processors/QueryPlan/ReadFromMergeTree.cpp index 0b5eb94dbac..8c7a4d86e5b 100644 --- a/src/Processors/QueryPlan/ReadFromMergeTree.cpp +++ b/src/Processors/QueryPlan/ReadFromMergeTree.cpp @@ -23,12 +23,12 @@ #include #include #include -#include #include -#include #include +#include +#include +#include #include -#include #include #include #include @@ -251,7 +251,7 @@ ReadFromMergeTree::ReadFromMergeTree( Poco::Logger * log_, MergeTreeDataSelectAnalysisResultPtr analyzed_result_ptr_, bool enable_parallel_reading) - : SourceStepWithFilter(DataStream{.header = IMergeTreeSelectAlgorithm::transformHeader( + : SourceStepWithFilter(DataStream{.header = MergeTreeSelectProcessor::transformHeader( storage_snapshot_->getSampleBlockForColumns(real_column_names_), getPrewhereInfoFromQueryInfo(query_info_), data_.getPartitionValueType(), @@ -268,10 +268,11 @@ ReadFromMergeTree::ReadFromMergeTree( , storage_snapshot(std::move(storage_snapshot_)) , metadata_for_reading(storage_snapshot->getMetadataForQuery()) , context(std::move(context_)) - , max_block_size(max_block_size_) + , block_size{ + .max_block_size_rows = max_block_size_, + .preferred_block_size_bytes = context->getSettingsRef().preferred_block_size_bytes, + .preferred_max_column_in_block_size_bytes = context->getSettingsRef().preferred_max_column_in_block_size_bytes} , requested_num_streams(num_streams_) - , preferred_block_size_bytes(context->getSettingsRef().preferred_block_size_bytes) - , preferred_max_column_in_block_size_bytes(context->getSettingsRef().preferred_max_column_in_block_size_bytes) , sample_factor_column_queried(sample_factor_column_queried_) , max_block_numbers_to_read(std::move(max_block_numbers_to_read_)) , log(log_) @@ -281,7 +282,7 @@ ReadFromMergeTree::ReadFromMergeTree( if (sample_factor_column_queried) { /// Only _sample_factor virtual column is added by ReadFromMergeTree - /// Other virtual columns are added by MergeTreeBaseSelectProcessor. + /// Other virtual columns are added by MergeTreeSelectProcessor. auto type = std::make_shared(); output_stream->header.insert({type->createColumn(), type, "_sample_factor"}); } @@ -325,50 +326,50 @@ ReadFromMergeTree::ReadFromMergeTree( Pipe ReadFromMergeTree::readFromPoolParallelReplicas( RangesInDataParts parts_with_range, Names required_columns, - size_t max_streams, - size_t min_marks_for_concurrent_read, - bool use_uncompressed_cache -) + PoolSettings pool_settings) { const auto & client_info = context->getClientInfo(); + auto extension = ParallelReadingExtension { .all_callback = all_ranges_callback.value(), .callback = read_task_callback.value(), .count_participating_replicas = client_info.count_participating_replicas, .number_of_current_replica = client_info.number_of_current_replica, - .columns_to_read = required_columns + .columns_to_read = required_columns, }; /// We have a special logic for local replica. It has to read less data, because in some cases it should /// merge states of aggregate functions or do some other important stuff other than reading from Disk. - min_marks_for_concurrent_read = static_cast(min_marks_for_concurrent_read * context->getSettingsRef().parallel_replicas_single_task_marks_count_multiplier); + pool_settings.min_marks_for_concurrent_read = static_cast(pool_settings.min_marks_for_concurrent_read * context->getSettingsRef().parallel_replicas_single_task_marks_count_multiplier); + size_t total_rows = parts_with_range.getRowsCountAllParts(); auto pool = std::make_shared( + std::move(extension), + std::move(parts_with_range), storage_snapshot, - max_streams, - extension, - parts_with_range, prewhere_info, actions_settings, reader_settings, required_columns, virt_column_names, - min_marks_for_concurrent_read); + pool_settings, + context); + + auto block_size_copy = block_size; + block_size_copy.min_marks_to_read = pool_settings.min_marks_for_concurrent_read; Pipes pipes; - const auto & settings = context->getSettingsRef(); - size_t total_rows = parts_with_range.getRowsCountAllParts(); - for (size_t i = 0; i < max_streams; ++i) + for (size_t i = 0; i < pool_settings.threads; ++i) { - auto algorithm = std::make_unique( - i, pool, min_marks_for_concurrent_read, max_block_size, - settings.preferred_block_size_bytes, settings.preferred_max_column_in_block_size_bytes, - data, storage_snapshot, use_uncompressed_cache, - prewhere_info, actions_settings, reader_settings, virt_column_names); + auto algorithm = std::make_unique(i); - auto source = std::make_shared(std::move(algorithm)); + auto processor = std::make_unique( + pool, std::move(algorithm), data, prewhere_info, + actions_settings, block_size_copy, reader_settings, virt_column_names); + + auto source = std::make_shared(std::move(processor)); /// Set the approximate number of rows for the first source only /// In case of parallel processing on replicas do not set approximate rows at all. @@ -387,12 +388,8 @@ Pipe ReadFromMergeTree::readFromPoolParallelReplicas( Pipe ReadFromMergeTree::readFromPool( RangesInDataParts parts_with_range, Names required_columns, - size_t max_streams, - size_t min_marks_for_concurrent_read, - bool use_uncompressed_cache) + PoolSettings pool_settings) { - Pipes pipes; - size_t sum_marks = parts_with_range.getMarksCountAllParts(); size_t total_rows = parts_with_range.getRowsCountAllParts(); if (query_info.limit > 0 && query_info.limit < total_rows) @@ -403,11 +400,11 @@ Pipe ReadFromMergeTree::readFromPool( /// round min_marks_to_read up to nearest multiple of block_size expressed in marks /// If granularity is adaptive it doesn't make sense /// Maybe it will make sense to add settings `max_block_size_bytes` - if (max_block_size && !data.canUseAdaptiveGranularity()) + if (block_size.max_block_size_rows && !data.canUseAdaptiveGranularity()) { size_t fixed_index_granularity = data.getSettings()->index_granularity; - min_marks_for_concurrent_read = (min_marks_for_concurrent_read * fixed_index_granularity + max_block_size - 1) - / max_block_size * max_block_size / fixed_index_granularity; + pool_settings.min_marks_for_concurrent_read = (pool_settings.min_marks_for_concurrent_read * fixed_index_granularity + block_size.max_block_size_rows - 1) + / block_size.max_block_size_rows * block_size.max_block_size_rows / fixed_index_granularity; } bool all_parts_are_remote = true; @@ -421,34 +418,30 @@ Pipe ReadFromMergeTree::readFromPool( MergeTreeReadPoolPtr pool; - if ((all_parts_are_remote && settings.allow_prefetched_read_pool_for_remote_filesystem - && MergeTreePrefetchedReadPool::checkReadMethodAllowed(reader_settings.read_settings.remote_fs_method)) - || (all_parts_are_local && settings.allow_prefetched_read_pool_for_local_filesystem - && MergeTreePrefetchedReadPool::checkReadMethodAllowed(reader_settings.read_settings.local_fs_method))) + bool allow_prefetched_remote = all_parts_are_remote + && settings.allow_prefetched_read_pool_for_remote_filesystem + && MergeTreePrefetchedReadPool::checkReadMethodAllowed(reader_settings.read_settings.remote_fs_method); + + bool allow_prefetched_local = all_parts_are_local + && settings.allow_prefetched_read_pool_for_local_filesystem + && MergeTreePrefetchedReadPool::checkReadMethodAllowed(reader_settings.read_settings.local_fs_method); + + if (allow_prefetched_remote || allow_prefetched_local) { pool = std::make_shared( - max_streams, - sum_marks, - min_marks_for_concurrent_read, std::move(parts_with_range), storage_snapshot, prewhere_info, actions_settings, + reader_settings, required_columns, virt_column_names, - settings.preferred_block_size_bytes, - reader_settings, - context, - use_uncompressed_cache, - all_parts_are_remote, - *data.getSettings()); + pool_settings, + context); } else { pool = std::make_shared( - max_streams, - sum_marks, - min_marks_for_concurrent_read, std::move(parts_with_range), storage_snapshot, prewhere_info, @@ -456,22 +449,28 @@ Pipe ReadFromMergeTree::readFromPool( reader_settings, required_columns, virt_column_names, - context, - false); + pool_settings, + context); } - auto * logger = &Poco::Logger::get(data.getLogName() + " (SelectExecutor)"); - LOG_DEBUG(logger, "Reading approx. {} rows with {} streams", total_rows, max_streams); + LOG_DEBUG(log, "Reading approx. {} rows with {} streams", total_rows, pool_settings.threads); - for (size_t i = 0; i < max_streams; ++i) + /// The reason why we change this setting is because MergeTreeReadPool takes the full task + /// ignoring min_marks_to_read setting in case of remote disk (see MergeTreeReadPool::getTask). + /// In this case, we won't limit the number of rows to read based on adaptive granularity settings. + auto block_size_copy = block_size; + block_size_copy.min_marks_to_read = pool_settings.min_marks_for_concurrent_read; + + Pipes pipes; + for (size_t i = 0; i < pool_settings.threads; ++i) { - auto algorithm = std::make_unique( - i, pool, min_marks_for_concurrent_read, max_block_size, - settings.preferred_block_size_bytes, settings.preferred_max_column_in_block_size_bytes, - data, storage_snapshot, use_uncompressed_cache, - prewhere_info, actions_settings, reader_settings, virt_column_names); + auto algorithm = std::make_unique(i); - auto source = std::make_shared(std::move(algorithm)); + auto processor = std::make_unique( + pool, std::move(algorithm), data, prewhere_info, + actions_settings, block_size_copy, reader_settings, virt_column_names); + + auto source = std::make_shared(std::move(processor)); if (i == 0) source->addTotalRowsApprox(total_rows); @@ -485,17 +484,65 @@ Pipe ReadFromMergeTree::readFromPool( return pipe; } -template -ProcessorPtr ReadFromMergeTree::createSource( - const RangesInDataPart & part, - const Names & required_columns, - bool use_uncompressed_cache, - bool has_limit_below_one_block, - MergeTreeInOrderReadPoolParallelReplicasPtr pool) +Pipe ReadFromMergeTree::readInOrder( + RangesInDataParts parts_with_ranges, + Names required_columns, + PoolSettings pool_settings, + ReadType read_type, + UInt64 limit) { - auto total_rows = part.getRowsCount(); - if (query_info.limit > 0 && query_info.limit < total_rows) - total_rows = query_info.limit; + /// For reading in order it makes sense to read only + /// one range per task to reduce number of read rows. + bool has_limit_below_one_block = read_type != ReadType::Default && limit && limit < block_size.max_block_size_rows; + MergeTreeReadPoolPtr pool; + + if (is_parallel_reading_from_replicas) + { + const auto & client_info = context->getClientInfo(); + ParallelReadingExtension extension + { + .all_callback = all_ranges_callback.value(), + .callback = read_task_callback.value(), + .count_participating_replicas = client_info.count_participating_replicas, + .number_of_current_replica = client_info.number_of_current_replica, + .columns_to_read = required_columns, + }; + + pool_settings.min_marks_for_concurrent_read = static_cast( + pool_settings.min_marks_for_concurrent_read * context->getSettingsRef().parallel_replicas_single_task_marks_count_multiplier); + + CoordinationMode mode = read_type == ReadType::InOrder + ? CoordinationMode::WithOrder + : CoordinationMode::ReverseOrder; + + pool = std::make_shared( + std::move(extension), + mode, + parts_with_ranges, + storage_snapshot, + prewhere_info, + actions_settings, + reader_settings, + required_columns, + virt_column_names, + pool_settings, + context); + } + else + { + pool = std::make_shared( + has_limit_below_one_block, + read_type, + parts_with_ranges, + storage_snapshot, + prewhere_info, + actions_settings, + reader_settings, + required_columns, + virt_column_names, + pool_settings, + context); + } /// Actually it means that parallel reading from replicas enabled /// and we have to collaborate with initiator. @@ -504,37 +551,34 @@ ProcessorPtr ReadFromMergeTree::createSource( /// because we don't know actual amount of read rows in case when limit is set. bool set_rows_approx = !is_parallel_reading_from_replicas && !reader_settings.read_in_order; - auto algorithm = std::make_unique( - data, storage_snapshot, part.data_part, part.alter_conversions, max_block_size, preferred_block_size_bytes, - preferred_max_column_in_block_size_bytes, required_columns, part.ranges, use_uncompressed_cache, prewhere_info, - actions_settings, reader_settings, pool, virt_column_names, part.part_index_in_query, has_limit_below_one_block); - - auto source = std::make_shared(std::move(algorithm)); - - if (set_rows_approx) - source->addTotalRowsApprox(total_rows); - - return source; -} - -Pipe ReadFromMergeTree::readInOrder( - RangesInDataParts parts_with_range, - Names required_columns, - ReadType read_type, - bool use_uncompressed_cache, - UInt64 limit, - MergeTreeInOrderReadPoolParallelReplicasPtr pool) -{ Pipes pipes; - /// For reading in order it makes sense to read only - /// one range per task to reduce number of read rows. - bool has_limit_below_one_block = read_type != ReadType::Default && limit && limit < max_block_size; - - for (const auto & part : parts_with_range) + for (size_t i = 0; i < parts_with_ranges.size(); ++i) { - auto source = read_type == ReadType::InReverseOrder - ? createSource(part, required_columns, use_uncompressed_cache, has_limit_below_one_block, pool) - : createSource(part, required_columns, use_uncompressed_cache, has_limit_below_one_block, pool); + const auto & part_with_ranges = parts_with_ranges[i]; + + UInt64 total_rows = part_with_ranges.getRowsCount(); + if (query_info.limit > 0 && query_info.limit < total_rows) + total_rows = query_info.limit; + + LOG_TRACE(log, "Reading {} ranges in{}order from part {}, approx. {} rows starting from {}", + part_with_ranges.ranges.size(), + read_type == ReadType::InReverseOrder ? " reverse " : " ", + part_with_ranges.data_part->name, total_rows, + part_with_ranges.data_part->index_granularity.getMarkStartingRow(part_with_ranges.ranges.front().begin)); + + MergeTreeSelectAlgorithmPtr algorithm; + if (read_type == ReadType::InReverseOrder) + algorithm = std::make_unique(i); + else + algorithm = std::make_unique(i); + + auto processor = std::make_unique( + pool, std::move(algorithm), data, prewhere_info, + actions_settings, block_size, reader_settings, virt_column_names); + + auto source = std::make_shared(std::move(processor)); + if (set_rows_approx) + source->addTotalRowsApprox(total_rows); pipes.emplace_back(std::move(source)); } @@ -553,16 +597,33 @@ Pipe ReadFromMergeTree::readInOrder( } Pipe ReadFromMergeTree::read( - RangesInDataParts parts_with_range, Names required_columns, ReadType read_type, - size_t max_streams, size_t min_marks_for_concurrent_read, bool use_uncompressed_cache) + RangesInDataParts parts_with_range, + Names required_columns, + ReadType read_type, + size_t max_streams, + size_t min_marks_for_concurrent_read, + bool use_uncompressed_cache) { + const auto & settings = context->getSettingsRef(); + size_t sum_marks = parts_with_range.getMarksCountAllParts(); + + PoolSettings pool_settings + { + .threads = max_streams, + .sum_marks = sum_marks, + .min_marks_for_concurrent_read = min_marks_for_concurrent_read, + .preferred_block_size_bytes = settings.preferred_block_size_bytes, + .use_uncompressed_cache = use_uncompressed_cache, + .use_const_size_tasks_for_remote_reading = settings.merge_tree_use_const_size_tasks_for_remote_reading, + }; + if (read_type == ReadType::ParallelReplicas) - return readFromPoolParallelReplicas(parts_with_range, required_columns, max_streams, min_marks_for_concurrent_read, use_uncompressed_cache); + return readFromPoolParallelReplicas(std::move(parts_with_range), std::move(required_columns), std::move(pool_settings)); if (read_type == ReadType::Default && max_streams > 1) - return readFromPool(parts_with_range, required_columns, max_streams, min_marks_for_concurrent_read, use_uncompressed_cache); + return readFromPool(std::move(parts_with_range), std::move(required_columns), std::move(pool_settings)); - auto pipe = readInOrder(parts_with_range, required_columns, read_type, use_uncompressed_cache, /*limit */0, /*pool*/nullptr); + auto pipe = readInOrder(parts_with_range, required_columns, pool_settings, read_type, /*limit=*/ 0); /// Use ConcatProcessor to concat sources together. /// It is needed to read in parts order (and so in PK order) if single thread is used. @@ -585,7 +646,6 @@ struct PartRangesReadInfo size_t index_granularity_bytes = 0; size_t max_marks_to_use_cache = 0; size_t min_marks_for_concurrent_read = 0; - bool use_uncompressed_cache = false; PartRangesReadInfo( @@ -663,8 +723,12 @@ Pipe ReadFromMergeTree::spreadMarkRangesAmongStreams(RangesInDataParts && parts_ auto read_type = is_parallel_reading_from_replicas ? ReadType::ParallelReplicas : ReadType::Default; - return read(std::move(parts_with_ranges), column_names, read_type, - num_streams, info.min_marks_for_concurrent_read, info.use_uncompressed_cache); + return read(std::move(parts_with_ranges), + column_names, + read_type, + num_streams, + info.min_marks_for_concurrent_read, + info.use_uncompressed_cache); } static ActionsDAGPtr createProjection(const Block & header) @@ -715,7 +779,8 @@ Pipe ReadFromMergeTree::spreadMarkRangesAmongStreamsWithOrder( /// Let's split ranges to avoid reading much data. auto split_ranges - = [rows_granularity = data_settings->index_granularity, my_max_block_size = max_block_size](const auto & ranges, int direction) + = [rows_granularity = data_settings->index_granularity, my_max_block_size = block_size.max_block_size_rows] + (const auto & ranges, int direction) { MarkRanges new_ranges; const size_t max_marks_in_range = (my_max_block_size + rows_granularity - 1) / rows_granularity; @@ -762,109 +827,93 @@ Pipe ReadFromMergeTree::spreadMarkRangesAmongStreamsWithOrder( const size_t min_marks_per_stream = (info.sum_marks - 1) / num_streams + 1; bool need_preliminary_merge = (parts_with_ranges.size() > settings.read_in_order_two_level_merge_threshold); - std::vector splitted_parts_and_ranges; - splitted_parts_and_ranges.reserve(num_streams); + const auto read_type = input_order_info->direction == 1 ? ReadType::InOrder : ReadType::InReverseOrder; - const auto read_type = input_order_info->direction == 1 - ? ReadFromMergeTree::ReadType::InOrder - : ReadFromMergeTree::ReadType::InReverseOrder; - - MergeTreeInOrderReadPoolParallelReplicasPtr pool; - - if (is_parallel_reading_from_replicas) + PoolSettings pool_settings { - const auto & client_info = context->getClientInfo(); - auto extension = ParallelReadingExtension - { - .all_callback = all_ranges_callback.value(), - .callback = read_task_callback.value(), - .count_participating_replicas = client_info.count_participating_replicas, - .number_of_current_replica = client_info.number_of_current_replica, - .columns_to_read = column_names - }; - - auto min_marks_for_concurrent_read = info.min_marks_for_concurrent_read; - min_marks_for_concurrent_read = static_cast(min_marks_for_concurrent_read * settings.parallel_replicas_single_task_marks_count_multiplier); - - pool = std::make_shared( - parts_with_ranges, - extension, - read_type == ReadFromMergeTree::ReadType::InOrder ? CoordinationMode::WithOrder : CoordinationMode::ReverseOrder, - min_marks_for_concurrent_read); - } - - - for (size_t i = 0; i < num_streams && !parts_with_ranges.empty(); ++i) - { - size_t need_marks = min_marks_per_stream; - RangesInDataParts new_parts; - - /// Loop over parts. - /// We will iteratively take part or some subrange of a part from the back - /// and assign a stream to read from it. - while (need_marks > 0 && !parts_with_ranges.empty()) - { - RangesInDataPart part = parts_with_ranges.back(); - parts_with_ranges.pop_back(); - size_t & marks_in_part = info.sum_marks_in_parts.back(); - - /// We will not take too few rows from a part. - if (marks_in_part >= info.min_marks_for_concurrent_read && need_marks < info.min_marks_for_concurrent_read) - need_marks = info.min_marks_for_concurrent_read; - - /// Do not leave too few rows in the part. - if (marks_in_part > need_marks && marks_in_part - need_marks < info.min_marks_for_concurrent_read) - need_marks = marks_in_part; - - MarkRanges ranges_to_get_from_part; - - /// We take full part if it contains enough marks or - /// if we know limit and part contains less than 'limit' rows. - bool take_full_part = marks_in_part <= need_marks || (input_order_info->limit && input_order_info->limit < part.getRowsCount()); - - /// We take the whole part if it is small enough. - if (take_full_part) - { - ranges_to_get_from_part = part.ranges; - - need_marks -= marks_in_part; - info.sum_marks_in_parts.pop_back(); - } - else - { - /// Loop through ranges in part. Take enough ranges to cover "need_marks". - while (need_marks > 0) - { - if (part.ranges.empty()) - throw Exception(ErrorCodes::LOGICAL_ERROR, "Unexpected end of ranges while spreading marks among streams"); - - MarkRange & range = part.ranges.front(); - - const size_t marks_in_range = range.end - range.begin; - const size_t marks_to_get_from_range = std::min(marks_in_range, need_marks); - - ranges_to_get_from_part.emplace_back(range.begin, range.begin + marks_to_get_from_range); - range.begin += marks_to_get_from_range; - marks_in_part -= marks_to_get_from_range; - need_marks -= marks_to_get_from_range; - if (range.begin == range.end) - part.ranges.pop_front(); - } - parts_with_ranges.emplace_back(part); - } - - ranges_to_get_from_part = split_ranges(ranges_to_get_from_part, input_order_info->direction); - new_parts.emplace_back(part.data_part, part.alter_conversions, part.part_index_in_query, std::move(ranges_to_get_from_part)); - } - - splitted_parts_and_ranges.emplace_back(std::move(new_parts)); - } + .min_marks_for_concurrent_read = info.min_marks_for_concurrent_read, + .preferred_block_size_bytes = settings.preferred_block_size_bytes, + .use_uncompressed_cache = info.use_uncompressed_cache, + }; Pipes pipes; - for (auto & item : splitted_parts_and_ranges) + if (is_parallel_reading_from_replicas) { - pipes.emplace_back(readInOrder(std::move(item), column_names, read_type, - info.use_uncompressed_cache, input_order_info->limit, pool)); + pipes.emplace_back(readInOrder(std::move(parts_with_ranges), column_names, pool_settings, read_type, input_order_info->limit)); + } + else + { + std::vector splitted_parts_and_ranges; + splitted_parts_and_ranges.reserve(num_streams); + + for (size_t i = 0; i < num_streams && !parts_with_ranges.empty(); ++i) + { + size_t need_marks = min_marks_per_stream; + RangesInDataParts new_parts; + + /// Loop over parts. + /// We will iteratively take part or some subrange of a part from the back + /// and assign a stream to read from it. + while (need_marks > 0 && !parts_with_ranges.empty()) + { + RangesInDataPart part = parts_with_ranges.back(); + parts_with_ranges.pop_back(); + size_t & marks_in_part = info.sum_marks_in_parts.back(); + + /// We will not take too few rows from a part. + if (marks_in_part >= info.min_marks_for_concurrent_read && need_marks < info.min_marks_for_concurrent_read) + need_marks = info.min_marks_for_concurrent_read; + + /// Do not leave too few rows in the part. + if (marks_in_part > need_marks && marks_in_part - need_marks < info.min_marks_for_concurrent_read) + need_marks = marks_in_part; + + MarkRanges ranges_to_get_from_part; + + /// We take full part if it contains enough marks or + /// if we know limit and part contains less than 'limit' rows. + bool take_full_part = marks_in_part <= need_marks || (input_order_info->limit && input_order_info->limit < part.getRowsCount()); + + /// We take the whole part if it is small enough. + if (take_full_part) + { + ranges_to_get_from_part = part.ranges; + + need_marks -= marks_in_part; + info.sum_marks_in_parts.pop_back(); + } + else + { + /// Loop through ranges in part. Take enough ranges to cover "need_marks". + while (need_marks > 0) + { + if (part.ranges.empty()) + throw Exception(ErrorCodes::LOGICAL_ERROR, "Unexpected end of ranges while spreading marks among streams"); + + MarkRange & range = part.ranges.front(); + + const size_t marks_in_range = range.end - range.begin; + const size_t marks_to_get_from_range = std::min(marks_in_range, need_marks); + + ranges_to_get_from_part.emplace_back(range.begin, range.begin + marks_to_get_from_range); + range.begin += marks_to_get_from_range; + marks_in_part -= marks_to_get_from_range; + need_marks -= marks_to_get_from_range; + if (range.begin == range.end) + part.ranges.pop_front(); + } + parts_with_ranges.emplace_back(part); + } + + ranges_to_get_from_part = split_ranges(ranges_to_get_from_part, input_order_info->direction); + new_parts.emplace_back(part.data_part, part.alter_conversions, part.part_index_in_query, std::move(ranges_to_get_from_part)); + } + + splitted_parts_and_ranges.emplace_back(std::move(new_parts)); + } + + for (auto && item : splitted_parts_and_ranges) + pipes.emplace_back(readInOrder(std::move(item), column_names, pool_settings, read_type, input_order_info->limit)); } Block pipe_header; @@ -898,7 +947,7 @@ Pipe ReadFromMergeTree::spreadMarkRangesAmongStreamsWithOrder( if (pipe.numOutputPorts() > 1) { auto transform = std::make_shared( - pipe.getHeader(), pipe.numOutputPorts(), sort_description, max_block_size, /*max_block_size_bytes=*/0, SortingQueueStrategy::Batch); + pipe.getHeader(), pipe.numOutputPorts(), sort_description, block_size.max_block_size_rows, /*max_block_size_bytes=*/0, SortingQueueStrategy::Batch); pipe.addTransform(std::move(transform)); } @@ -931,7 +980,7 @@ static void addMergingFinal( const SortDescription & sort_description, MergeTreeData::MergingParams merging_params, Names partition_key_columns, - size_t max_block_size) + size_t max_block_size_rows) { const auto & header = pipe.getHeader(); size_t num_outputs = pipe.numOutputPorts(); @@ -944,31 +993,31 @@ static void addMergingFinal( { case MergeTreeData::MergingParams::Ordinary: return std::make_shared(header, num_outputs, - sort_description, max_block_size, /*max_block_size_bytes=*/0, SortingQueueStrategy::Batch); + sort_description, max_block_size_rows, /*max_block_size_bytes=*/0, SortingQueueStrategy::Batch); case MergeTreeData::MergingParams::Collapsing: return std::make_shared(header, num_outputs, - sort_description, merging_params.sign_column, true, max_block_size, /*max_block_size_bytes=*/0); + sort_description, merging_params.sign_column, true, max_block_size_rows, /*max_block_size_bytes=*/0); case MergeTreeData::MergingParams::Summing: return std::make_shared(header, num_outputs, - sort_description, merging_params.columns_to_sum, partition_key_columns, max_block_size, /*max_block_size_bytes=*/0); + sort_description, merging_params.columns_to_sum, partition_key_columns, max_block_size_rows, /*max_block_size_bytes=*/0); case MergeTreeData::MergingParams::Aggregating: return std::make_shared(header, num_outputs, - sort_description, max_block_size, /*max_block_size_bytes=*/0); + sort_description, max_block_size_rows, /*max_block_size_bytes=*/0); case MergeTreeData::MergingParams::Replacing: return std::make_shared(header, num_outputs, - sort_description, merging_params.is_deleted_column, merging_params.version_column, max_block_size, /*max_block_size_bytes=*/0, /*out_row_sources_buf_*/ nullptr, /*use_average_block_sizes*/ false, /*cleanup*/ !merging_params.is_deleted_column.empty()); + sort_description, merging_params.is_deleted_column, merging_params.version_column, max_block_size_rows, /*max_block_size_bytes=*/0, /*out_row_sources_buf_*/ nullptr, /*use_average_block_sizes*/ false, /*cleanup*/ !merging_params.is_deleted_column.empty()); case MergeTreeData::MergingParams::VersionedCollapsing: return std::make_shared(header, num_outputs, - sort_description, merging_params.sign_column, max_block_size, /*max_block_size_bytes=*/0); + sort_description, merging_params.sign_column, max_block_size_rows, /*max_block_size_bytes=*/0); case MergeTreeData::MergingParams::Graphite: return std::make_shared(header, num_outputs, - sort_description, max_block_size, /*max_block_size_bytes=*/0, merging_params.graphite_params, now); + sort_description, max_block_size_rows, /*max_block_size_bytes=*/0, merging_params.graphite_params, now); } UNREACHABLE(); @@ -1064,11 +1113,12 @@ Pipe ReadFromMergeTree::spreadMarkRangesAmongStreamsFinal( return this->read( std::move(parts), column_names, - ReadFromMergeTree::ReadType::InOrder, + ReadType::InOrder, 1 /* num_streams */, 0 /* min_marks_for_concurrent_read */, info.use_uncompressed_cache); }; + pipes = buildPipesForReadingByPKRanges( metadata_for_reading->getPrimaryKey(), sorting_expr, @@ -1080,7 +1130,7 @@ Pipe ReadFromMergeTree::spreadMarkRangesAmongStreamsFinal( else { pipes.emplace_back(read( - std::move(new_parts), column_names, ReadFromMergeTree::ReadType::InOrder, num_streams, 0, info.use_uncompressed_cache)); + std::move(new_parts), column_names, ReadType::InOrder, num_streams, 0, info.use_uncompressed_cache)); pipes.back().addSimpleTransform([sorting_expr](const Block & header) { return std::make_shared(header, sorting_expr); }); @@ -1121,7 +1171,7 @@ Pipe ReadFromMergeTree::spreadMarkRangesAmongStreamsFinal( sort_description, data.merging_params, partition_key_columns, - max_block_size); + block_size.max_block_size_rows); partition_pipes.emplace_back(Pipe::unitePipes(std::move(pipes))); } @@ -1141,7 +1191,7 @@ Pipe ReadFromMergeTree::spreadMarkRangesAmongStreamsFinal( if (sum_marks_in_lonely_parts < num_streams_for_lonely_parts * min_marks_for_concurrent_read && lonely_parts.size() < num_streams_for_lonely_parts) num_streams_for_lonely_parts = std::max((sum_marks_in_lonely_parts + min_marks_for_concurrent_read - 1) / min_marks_for_concurrent_read, lonely_parts.size()); - auto pipe = read(std::move(lonely_parts), column_names, ReadFromMergeTree::ReadType::Default, + auto pipe = read(std::move(lonely_parts), column_names, ReadType::Default, num_streams_for_lonely_parts, min_marks_for_concurrent_read, info.use_uncompressed_cache); /// Drop temporary columns, added by 'sorting_key_expr' @@ -1603,11 +1653,13 @@ void ReadFromMergeTree::updatePrewhereInfo(const PrewhereInfoPtr & prewhere_info { query_info.prewhere_info = prewhere_info_value; prewhere_info = prewhere_info_value; - output_stream = DataStream{.header = IMergeTreeSelectAlgorithm::transformHeader( + + output_stream = DataStream{.header = MergeTreeSelectProcessor::transformHeader( storage_snapshot->getSampleBlockForColumns(real_column_names), prewhere_info_value, data.getPartitionValueType(), virt_column_names)}; + updateSortDescriptionForOutputStream( *output_stream, storage_snapshot->getMetadataForQuery()->getSortingKeyColumns(), diff --git a/src/Processors/QueryPlan/ReadFromMergeTree.h b/src/Processors/QueryPlan/ReadFromMergeTree.h index cb2a3a8ddf9..b6e0d925238 100644 --- a/src/Processors/QueryPlan/ReadFromMergeTree.h +++ b/src/Processors/QueryPlan/ReadFromMergeTree.h @@ -87,25 +87,7 @@ public: }; using IndexStats = std::vector; - - enum class ReadType - { - /// By default, read will use MergeTreeReadPool and return pipe with num_streams outputs. - /// If num_streams == 1, will read without pool, in order specified in parts. - Default, - /// Read in sorting key order. - /// Returned pipe will have the number of ports equals to parts.size(). - /// Parameter num_streams_ is ignored in this case. - /// User should add MergingSorted itself if needed. - InOrder, - /// The same as InOrder, but in reverse order. - /// For every part, read ranges and granules from end to begin. Also add ReverseTransform. - InReverseOrder, - /// A special type of reading where every replica - /// talks to a remote coordinator (which is located on the initiator node) - /// and who spreads marks and parts across them. - ParallelReplicas, - }; + using ReadType = MergeTreeReadType; struct AnalysisResult { @@ -113,7 +95,7 @@ public: MergeTreeDataSelectSamplingData sampling; IndexStats index_stats; Names column_names_to_read; - ReadFromMergeTree::ReadType read_type = ReadFromMergeTree::ReadType::Default; + ReadType read_type = ReadType::Default; UInt64 total_parts = 0; UInt64 parts_before_pk = 0; UInt64 selected_parts = 0; @@ -222,7 +204,7 @@ public: const MergeTreeData::DataPartsVector & getParts() const { return prepared_parts; } const MergeTreeData & getMergeTreeData() const { return data; } - size_t getMaxBlockSize() const { return max_block_size; } + size_t getMaxBlockSize() const { return block_size.max_block_size_rows; } size_t getNumStreams() const { return requested_num_streams; } bool isParallelReadingEnabled() const { return read_task_callback != std::nullopt; } @@ -270,12 +252,10 @@ private: StorageMetadataPtr metadata_for_reading; ContextPtr context; + const MergeTreeReadTask::BlockSizeParams block_size; - const size_t max_block_size; size_t requested_num_streams; size_t output_streams_limit = 0; - const size_t preferred_block_size_bytes; - const size_t preferred_max_column_in_block_size_bytes; const bool sample_factor_column_queried; /// Used for aggregation optimisation (see DB::QueryPlanOptimizations::tryAggregateEachPartitionIndependently). @@ -291,16 +271,14 @@ private: UInt64 selected_rows = 0; UInt64 selected_marks = 0; + using PoolSettings = MergeTreeReadPoolBase::PoolSettings; + Pipe read(RangesInDataParts parts_with_range, Names required_columns, ReadType read_type, size_t max_streams, size_t min_marks_for_concurrent_read, bool use_uncompressed_cache); - Pipe readFromPool(RangesInDataParts parts_with_ranges, Names required_columns, size_t max_streams, size_t min_marks_for_concurrent_read, bool use_uncompressed_cache); - Pipe readFromPoolParallelReplicas(RangesInDataParts parts_with_ranges, Names required_columns, size_t max_streams, size_t min_marks_for_concurrent_read, bool use_uncompressed_cache); - Pipe readInOrder(RangesInDataParts parts_with_range, Names required_columns, ReadType read_type, bool use_uncompressed_cache, UInt64 limit, MergeTreeInOrderReadPoolParallelReplicasPtr pool); + Pipe readFromPool(RangesInDataParts parts_with_range, Names required_columns, PoolSettings pool_settings); + Pipe readFromPoolParallelReplicas(RangesInDataParts parts_with_range, Names required_columns, PoolSettings pool_settings); + Pipe readInOrder(RangesInDataParts parts_with_ranges, Names required_columns, PoolSettings pool_settings, ReadType read_type, UInt64 limit); - template - ProcessorPtr createSource(const RangesInDataPart & part, const Names & required_columns, bool use_uncompressed_cache, bool has_limit_below_one_block, MergeTreeInOrderReadPoolParallelReplicasPtr pool); - - Pipe spreadMarkRanges( - RangesInDataParts && parts_with_ranges, size_t num_streams, AnalysisResult & result, ActionsDAGPtr & result_projection); + Pipe spreadMarkRanges(RangesInDataParts && parts_with_ranges, size_t num_streams, AnalysisResult & result, ActionsDAGPtr & result_projection); Pipe groupStreamsByPartition(AnalysisResult & result, ActionsDAGPtr & result_projection); diff --git a/src/Storages/MergeTree/IMergeTreeReadPool.h b/src/Storages/MergeTree/IMergeTreeReadPool.h index efdfca51c0a..31aa61d5032 100644 --- a/src/Storages/MergeTree/IMergeTreeReadPool.h +++ b/src/Storages/MergeTree/IMergeTreeReadPool.h @@ -3,24 +3,27 @@ #include #include #include -#include +#include namespace DB { -struct MergeTreeReadTask; -using MergeTreeReadTaskPtr = std::unique_ptr; - +/// The interface that determines how tasks for reading (MergeTreeReadTask) +/// are distributed among data parts with ranges. class IMergeTreeReadPool : private boost::noncopyable { public: virtual ~IMergeTreeReadPool() = default; - + virtual String getName() const = 0; virtual Block getHeader() const = 0; - virtual MergeTreeReadTaskPtr getTask(size_t thread) = 0; + /// Returns true if tasks are returned in the same order as the order of ranges passed to pool + virtual bool preservesOrderOfRanges() const = 0; + /// task_idx is an implentation defined identifier that helps + /// to get required task. E.g. it may be number of thread, index of + virtual MergeTreeReadTaskPtr getTask(size_t task_idx, MergeTreeReadTask * previous_task) = 0; virtual void profileFeedback(ReadBufferFromFileBase::ProfileInfo info) = 0; }; diff --git a/src/Storages/MergeTree/IMergeTreeReader.h b/src/Storages/MergeTree/IMergeTreeReader.h index dafe7b2d4ba..a7e60254217 100644 --- a/src/Storages/MergeTree/IMergeTreeReader.h +++ b/src/Storages/MergeTree/IMergeTreeReader.h @@ -3,7 +3,6 @@ #include #include #include -#include #include #include diff --git a/src/Storages/MergeTree/MergeTreeBaseSelectProcessor.cpp b/src/Storages/MergeTree/MergeTreeBaseSelectProcessor.cpp deleted file mode 100644 index d7836ac01b0..00000000000 --- a/src/Storages/MergeTree/MergeTreeBaseSelectProcessor.cpp +++ /dev/null @@ -1,696 +0,0 @@ -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include - -namespace ProfileEvents -{ - extern const Event WaitPrefetchTaskMicroseconds; -}; - -namespace DB -{ - -namespace ErrorCodes -{ - extern const int ILLEGAL_TYPE_OF_COLUMN_FOR_FILTER; - extern const int LOGICAL_ERROR; - extern const int QUERY_WAS_CANCELLED; -} - -static void injectNonConstVirtualColumns( - size_t rows, - Block & block, - const Names & virtual_columns); - -static void injectPartConstVirtualColumns( - size_t rows, - Block & block, - MergeTreeReadTask * task, - const DataTypePtr & partition_value_type, - const Names & virtual_columns); - - -IMergeTreeSelectAlgorithm::IMergeTreeSelectAlgorithm( - Block header, - const MergeTreeData & storage_, - const StorageSnapshotPtr & storage_snapshot_, - const PrewhereInfoPtr & prewhere_info_, - const ExpressionActionsSettings & actions_settings_, - UInt64 max_block_size_rows_, - UInt64 preferred_block_size_bytes_, - UInt64 preferred_max_column_in_block_size_bytes_, - const MergeTreeReaderSettings & reader_settings_, - bool use_uncompressed_cache_, - const Names & virt_column_names_) - : storage(storage_) - , storage_snapshot(storage_snapshot_) - , prewhere_info(prewhere_info_) - , actions_settings(actions_settings_) - , prewhere_actions(getPrewhereActions(prewhere_info, actions_settings, reader_settings_.enable_multiple_prewhere_read_steps)) - , max_block_size_rows(max_block_size_rows_) - , preferred_block_size_bytes(preferred_block_size_bytes_) - , preferred_max_column_in_block_size_bytes(preferred_max_column_in_block_size_bytes_) - , reader_settings(reader_settings_) - , use_uncompressed_cache(use_uncompressed_cache_) - , virt_column_names(virt_column_names_) - , partition_value_type(storage.getPartitionValueType()) - , owned_uncompressed_cache(use_uncompressed_cache ? storage.getContext()->getUncompressedCache() : nullptr) - , owned_mark_cache(storage.getContext()->getMarkCache()) -{ - header_without_const_virtual_columns = applyPrewhereActions(std::move(header), prewhere_info); - size_t non_const_columns_offset = header_without_const_virtual_columns.columns(); - injectNonConstVirtualColumns(0, header_without_const_virtual_columns, virt_column_names); - - for (size_t col_num = non_const_columns_offset; col_num < header_without_const_virtual_columns.columns(); ++col_num) - non_const_virtual_column_names.emplace_back(header_without_const_virtual_columns.getByPosition(col_num).name); - - result_header = header_without_const_virtual_columns; - injectPartConstVirtualColumns(0, result_header, nullptr, partition_value_type, virt_column_names); - - if (!prewhere_actions.steps.empty()) - LOG_TRACE(log, "PREWHERE condition was split into {} steps: {}", prewhere_actions.steps.size(), prewhere_actions.dumpConditions()); - - if (prewhere_info) - LOG_TEST(log, "Original PREWHERE DAG:\n{}\nPREWHERE actions:\n{}", - (prewhere_info->prewhere_actions ? prewhere_info->prewhere_actions->dumpDAG(): std::string("")), - (!prewhere_actions.steps.empty() ? prewhere_actions.dump() : std::string(""))); -} - -bool tryBuildPrewhereSteps(PrewhereInfoPtr prewhere_info, const ExpressionActionsSettings & actions_settings, PrewhereExprInfo & prewhere); - -PrewhereExprInfo IMergeTreeSelectAlgorithm::getPrewhereActions(PrewhereInfoPtr prewhere_info, const ExpressionActionsSettings & actions_settings, bool enable_multiple_prewhere_read_steps) -{ - PrewhereExprInfo prewhere_actions; - if (prewhere_info) - { - if (prewhere_info->row_level_filter) - { - PrewhereExprStep row_level_filter_step - { - .type = PrewhereExprStep::Filter, - .actions = std::make_shared(prewhere_info->row_level_filter, actions_settings), - .filter_column_name = prewhere_info->row_level_column_name, - .remove_filter_column = true, - .need_filter = true, - .perform_alter_conversions = true, - }; - - prewhere_actions.steps.emplace_back(std::make_shared(std::move(row_level_filter_step))); - } - - if (!enable_multiple_prewhere_read_steps || - !tryBuildPrewhereSteps(prewhere_info, actions_settings, prewhere_actions)) - { - PrewhereExprStep prewhere_step - { - .type = PrewhereExprStep::Filter, - .actions = std::make_shared(prewhere_info->prewhere_actions, actions_settings), - .filter_column_name = prewhere_info->prewhere_column_name, - .remove_filter_column = prewhere_info->remove_prewhere_column, - .need_filter = prewhere_info->need_filter, - .perform_alter_conversions = true, - }; - - prewhere_actions.steps.emplace_back(std::make_shared(std::move(prewhere_step))); - } - } - - return prewhere_actions; -} - - -bool IMergeTreeSelectAlgorithm::getNewTask() -{ - if (getNewTaskImpl()) - { - finalizeNewTask(); - return true; - } - return false; -} - - -ChunkAndProgress IMergeTreeSelectAlgorithm::read() -{ - while (!is_cancelled) - { - try - { - if ((!task || task->isFinished()) && !getNewTask()) - break; - } - catch (const Exception & e) - { - /// See MergeTreeBaseSelectProcessor::getTaskFromBuffer() - if (e.code() == ErrorCodes::QUERY_WAS_CANCELLED) - break; - throw; - } - - auto res = readFromPart(); - - if (res.row_count) - { - injectVirtualColumns(res.block, res.row_count, task.get(), partition_value_type, virt_column_names); - - /// Reorder the columns according to result_header - Columns ordered_columns; - ordered_columns.reserve(result_header.columns()); - for (size_t i = 0; i < result_header.columns(); ++i) - { - auto name = result_header.getByPosition(i).name; - ordered_columns.push_back(res.block.getByName(name).column); - } - - return ChunkAndProgress{ - .chunk = Chunk(ordered_columns, res.row_count), - .num_read_rows = res.num_read_rows, - .num_read_bytes = res.num_read_bytes, - .is_finished = false}; - } - else - { - return {Chunk(), res.num_read_rows, res.num_read_bytes, false}; - } - } - - return {Chunk(), 0, 0, true}; -} - -void IMergeTreeSelectAlgorithm::initializeMergeTreeReadersForCurrentTask( - const IMergeTreeReader::ValueSizeMap & value_size_map, - const ReadBufferFromFileBase::ProfileCallback & profile_callback) -{ - if (!task) - throw Exception(ErrorCodes::LOGICAL_ERROR, "There is no task"); - - if (task->reader.valid()) - { - ProfileEventTimeIncrement watch(ProfileEvents::WaitPrefetchTaskMicroseconds); - reader = task->reader.get(); - } - else - { - reader = task->data_part->getReader( - task->task_columns.columns, storage_snapshot, task->mark_ranges, - owned_uncompressed_cache.get(), owned_mark_cache.get(), - task->alter_conversions, reader_settings, value_size_map, profile_callback); - } - - if (!task->pre_reader_for_step.empty()) - { - ProfileEventTimeIncrement watch(ProfileEvents::WaitPrefetchTaskMicroseconds); - pre_reader_for_step.clear(); - for (auto & pre_reader : task->pre_reader_for_step) - pre_reader_for_step.push_back(pre_reader.get()); - } - else - { - initializeMergeTreePreReadersForPart( - task->data_part, task->alter_conversions, - task->task_columns, task->mark_ranges, - value_size_map, profile_callback); - } -} - -void IMergeTreeSelectAlgorithm::initializeMergeTreeReadersForPart( - const MergeTreeData::DataPartPtr & data_part, - const AlterConversionsPtr & alter_conversions, - const MergeTreeReadTaskColumns & task_columns, - const MarkRanges & mark_ranges, - const IMergeTreeReader::ValueSizeMap & value_size_map, - const ReadBufferFromFileBase::ProfileCallback & profile_callback) -{ - reader = data_part->getReader( - task_columns.columns, storage_snapshot, mark_ranges, - owned_uncompressed_cache.get(), owned_mark_cache.get(), - alter_conversions, reader_settings, value_size_map, profile_callback); - - initializeMergeTreePreReadersForPart( - data_part, alter_conversions, task_columns, - mark_ranges, value_size_map, profile_callback); -} - -void IMergeTreeSelectAlgorithm::initializeMergeTreePreReadersForPart( - const MergeTreeData::DataPartPtr & data_part, - const AlterConversionsPtr & alter_conversions, - const MergeTreeReadTaskColumns & task_columns, - const MarkRanges & mark_ranges, - const IMergeTreeReader::ValueSizeMap & value_size_map, - const ReadBufferFromFileBase::ProfileCallback & profile_callback) -{ - pre_reader_for_step.clear(); - - /// Add lightweight delete filtering step - if (reader_settings.apply_deleted_mask && data_part->hasLightweightDelete()) - { - pre_reader_for_step.push_back( - data_part->getReader( - {LightweightDeleteDescription::FILTER_COLUMN}, storage_snapshot, - mark_ranges, owned_uncompressed_cache.get(), owned_mark_cache.get(), - alter_conversions, reader_settings, value_size_map, profile_callback)); - } - - for (const auto & pre_columns_per_step : task_columns.pre_columns) - { - pre_reader_for_step.push_back( - data_part->getReader( - pre_columns_per_step, storage_snapshot, mark_ranges, - owned_uncompressed_cache.get(), owned_mark_cache.get(), - alter_conversions, reader_settings, value_size_map, profile_callback)); - } -} - -void IMergeTreeSelectAlgorithm::initializeRangeReaders(MergeTreeReadTask & current_task) -{ - return initializeRangeReadersImpl( - current_task.range_reader, current_task.pre_range_readers, prewhere_actions, - reader.get(), current_task.data_part->hasLightweightDelete(), reader_settings, - pre_reader_for_step, lightweight_delete_filter_step, non_const_virtual_column_names); -} - -void IMergeTreeSelectAlgorithm::initializeRangeReadersImpl( - MergeTreeRangeReader & range_reader, - std::deque & pre_range_readers, - const PrewhereExprInfo & prewhere_actions, - IMergeTreeReader * reader, - bool has_lightweight_delete, - const MergeTreeReaderSettings & reader_settings, - const std::vector> & pre_reader_for_step, - const PrewhereExprStep & lightweight_delete_filter_step, - const Names & non_const_virtual_column_names) -{ - MergeTreeRangeReader * prev_reader = nullptr; - bool last_reader = false; - size_t pre_readers_shift = 0; - - /// Add filtering step with lightweight delete mask - if (reader_settings.apply_deleted_mask && has_lightweight_delete) - { - MergeTreeRangeReader pre_range_reader(pre_reader_for_step[0].get(), prev_reader, &lightweight_delete_filter_step, last_reader, non_const_virtual_column_names); - pre_range_readers.push_back(std::move(pre_range_reader)); - prev_reader = &pre_range_readers.back(); - pre_readers_shift++; - } - - if (prewhere_actions.steps.size() + pre_readers_shift != pre_reader_for_step.size()) - { - throw Exception( - ErrorCodes::LOGICAL_ERROR, - "PREWHERE steps count mismatch, actions: {}, readers: {}", - prewhere_actions.steps.size(), pre_reader_for_step.size()); - } - - for (size_t i = 0; i < prewhere_actions.steps.size(); ++i) - { - last_reader = reader->getColumns().empty() && (i + 1 == prewhere_actions.steps.size()); - - MergeTreeRangeReader current_reader( - pre_reader_for_step[i + pre_readers_shift].get(), - prev_reader, prewhere_actions.steps[i].get(), - last_reader, non_const_virtual_column_names); - - pre_range_readers.push_back(std::move(current_reader)); - prev_reader = &pre_range_readers.back(); - } - - if (!last_reader) - { - range_reader = MergeTreeRangeReader(reader, prev_reader, nullptr, true, non_const_virtual_column_names); - } - else - { - /// If all columns are read by pre_range_readers than move last pre_range_reader into range_reader - range_reader = std::move(pre_range_readers.back()); - pre_range_readers.pop_back(); - } -} - -static UInt64 estimateNumRows(const MergeTreeReadTask & current_task, UInt64 current_preferred_block_size_bytes, - UInt64 current_max_block_size_rows, UInt64 current_preferred_max_column_in_block_size_bytes, double min_filtration_ratio, size_t min_marks_to_read) -{ - const MergeTreeRangeReader & current_reader = current_task.range_reader; - - if (!current_task.size_predictor) - return static_cast(current_max_block_size_rows); - - /// Calculates number of rows will be read using preferred_block_size_bytes. - /// Can't be less than avg_index_granularity. - size_t rows_to_read = current_task.size_predictor->estimateNumRows(current_preferred_block_size_bytes); - if (!rows_to_read) - return rows_to_read; - auto total_row_in_current_granule = current_reader.numRowsInCurrentGranule(); - rows_to_read = std::max(total_row_in_current_granule, rows_to_read); - - if (current_preferred_max_column_in_block_size_bytes) - { - /// Calculates number of rows will be read using preferred_max_column_in_block_size_bytes. - auto rows_to_read_for_max_size_column - = current_task.size_predictor->estimateNumRowsForMaxSizeColumn(current_preferred_max_column_in_block_size_bytes); - double filtration_ratio = std::max(min_filtration_ratio, 1.0 - current_task.size_predictor->filtered_rows_ratio); - auto rows_to_read_for_max_size_column_with_filtration - = static_cast(rows_to_read_for_max_size_column / filtration_ratio); - - /// If preferred_max_column_in_block_size_bytes is used, number of rows to read can be less than current_index_granularity. - rows_to_read = std::min(rows_to_read, rows_to_read_for_max_size_column_with_filtration); - } - - auto unread_rows_in_current_granule = current_reader.numPendingRowsInCurrentGranule(); - if (unread_rows_in_current_granule >= rows_to_read) - return rows_to_read; - - const MergeTreeIndexGranularity & index_granularity = current_task.data_part->index_granularity; - - return index_granularity.countMarksForRows(current_reader.currentMark(), rows_to_read, current_reader.numReadRowsInCurrentGranule(), min_marks_to_read); -} - - -IMergeTreeSelectAlgorithm::BlockAndProgress IMergeTreeSelectAlgorithm::readFromPartImpl() -{ - if (task->size_predictor) - task->size_predictor->startBlock(); - - const UInt64 current_max_block_size_rows = max_block_size_rows; - const UInt64 current_preferred_block_size_bytes = preferred_block_size_bytes; - const UInt64 current_preferred_max_column_in_block_size_bytes = preferred_max_column_in_block_size_bytes; - const double min_filtration_ratio = 0.00001; - - UInt64 recommended_rows = estimateNumRows(*task, current_preferred_block_size_bytes, - current_max_block_size_rows, current_preferred_max_column_in_block_size_bytes, min_filtration_ratio, min_marks_to_read); - UInt64 rows_to_read = std::max(static_cast(1), std::min(current_max_block_size_rows, recommended_rows)); - - auto read_result = task->range_reader.read(rows_to_read, task->mark_ranges); - - /// All rows were filtered. Repeat. - if (read_result.num_rows == 0) - read_result.columns.clear(); - - const auto & sample_block = task->range_reader.getSampleBlock(); - if (read_result.num_rows != 0 && sample_block.columns() != read_result.columns.size()) - throw Exception(ErrorCodes::LOGICAL_ERROR, "Inconsistent number of columns got from MergeTreeRangeReader. " - "Have {} in sample block and {} columns in list", - toString(sample_block.columns()), toString(read_result.columns.size())); - - /// TODO: check columns have the same types as in header. - - UInt64 num_filtered_rows = read_result.numReadRows() - read_result.num_rows; - - size_t num_read_rows = read_result.numReadRows(); - size_t num_read_bytes = read_result.numBytesRead(); - - if (task->size_predictor) - { - task->size_predictor->updateFilteredRowsRation(read_result.numReadRows(), num_filtered_rows); - - if (!read_result.columns.empty()) - task->size_predictor->update(sample_block, read_result.columns, read_result.num_rows); - } - - Block block; - if (read_result.num_rows != 0) - block = sample_block.cloneWithColumns(read_result.columns); - - BlockAndProgress res = { - .block = std::move(block), - .row_count = read_result.num_rows, - .num_read_rows = num_read_rows, - .num_read_bytes = num_read_bytes }; - - return res; -} - - -IMergeTreeSelectAlgorithm::BlockAndProgress IMergeTreeSelectAlgorithm::readFromPart() -{ - if (!task->range_reader.isInitialized()) - initializeRangeReaders(*task); - - return readFromPartImpl(); -} - - -namespace -{ - struct VirtualColumnsInserter - { - explicit VirtualColumnsInserter(Block & block_) : block(block_) {} - - bool columnExists(const String & name) const { return block.has(name); } - - void insertUInt8Column(const ColumnPtr & column, const String & name) - { - block.insert({column, std::make_shared(), name}); - } - - void insertUInt64Column(const ColumnPtr & column, const String & name) - { - block.insert({column, std::make_shared(), name}); - } - - void insertUUIDColumn(const ColumnPtr & column, const String & name) - { - block.insert({column, std::make_shared(), name}); - } - - void insertLowCardinalityColumn(const ColumnPtr & column, const String & name) - { - block.insert({column, std::make_shared(std::make_shared()), name}); - } - - void insertPartitionValueColumn( - size_t rows, const Row & partition_value, const DataTypePtr & partition_value_type, const String & name) - { - ColumnPtr column; - if (rows) - column = partition_value_type->createColumnConst(rows, Tuple(partition_value.begin(), partition_value.end())) - ->convertToFullColumnIfConst(); - else - column = partition_value_type->createColumn(); - - block.insert({column, partition_value_type, name}); - } - - Block & block; - }; -} - -/// Adds virtual columns that are not const for all rows -static void injectNonConstVirtualColumns( - size_t rows, - Block & block, - const Names & virtual_columns) -{ - VirtualColumnsInserter inserter(block); - for (const auto & virtual_column_name : virtual_columns) - { - if (virtual_column_name == "_part_offset") - { - if (!rows) - { - inserter.insertUInt64Column(DataTypeUInt64().createColumn(), virtual_column_name); - } - else - { - if (!inserter.columnExists(virtual_column_name)) - throw Exception(ErrorCodes::LOGICAL_ERROR, - "Column {} must have been filled part reader", - virtual_column_name); - } - } - - if (virtual_column_name == LightweightDeleteDescription::FILTER_COLUMN.name) - { - /// If _row_exists column isn't present in the part then fill it here with 1s - ColumnPtr column; - if (rows) - column = LightweightDeleteDescription::FILTER_COLUMN.type->createColumnConst(rows, 1)->convertToFullColumnIfConst(); - else - column = LightweightDeleteDescription::FILTER_COLUMN.type->createColumn(); - - inserter.insertUInt8Column(column, virtual_column_name); - } - } -} - -/// Adds virtual columns that are const for the whole part -static void injectPartConstVirtualColumns( - size_t rows, - Block & block, - MergeTreeReadTask * task, - const DataTypePtr & partition_value_type, - const Names & virtual_columns) -{ - VirtualColumnsInserter inserter(block); - /// add virtual columns - /// Except _sample_factor, which is added from the outside. - if (!virtual_columns.empty()) - { - if (unlikely(rows && !task)) - throw Exception(ErrorCodes::LOGICAL_ERROR, "Cannot insert virtual columns to non-empty chunk without specified task."); - - const IMergeTreeDataPart * part = nullptr; - if (rows) - { - part = task->data_part.get(); - if (part->isProjectionPart()) - part = part->getParentPart(); - } - for (const auto & virtual_column_name : virtual_columns) - { - if (virtual_column_name == "_part") - { - ColumnPtr column; - if (rows) - column = DataTypeLowCardinality{std::make_shared()} - .createColumnConst(rows, part->name) - ->convertToFullColumnIfConst(); - else - column = DataTypeLowCardinality{std::make_shared()}.createColumn(); - - inserter.insertLowCardinalityColumn(column, virtual_column_name); - } - else if (virtual_column_name == "_part_index") - { - ColumnPtr column; - if (rows) - column = DataTypeUInt64().createColumnConst(rows, task->part_index_in_query)->convertToFullColumnIfConst(); - else - column = DataTypeUInt64().createColumn(); - - inserter.insertUInt64Column(column, virtual_column_name); - } - else if (virtual_column_name == "_part_uuid") - { - ColumnPtr column; - if (rows) - column = DataTypeUUID().createColumnConst(rows, part->uuid)->convertToFullColumnIfConst(); - else - column = DataTypeUUID().createColumn(); - - inserter.insertUUIDColumn(column, virtual_column_name); - } - else if (virtual_column_name == "_partition_id") - { - ColumnPtr column; - if (rows) - column = DataTypeLowCardinality{std::make_shared()} - .createColumnConst(rows, part->info.partition_id) - ->convertToFullColumnIfConst(); - else - column = DataTypeLowCardinality{std::make_shared()}.createColumn(); - - inserter.insertLowCardinalityColumn(column, virtual_column_name); - } - else if (virtual_column_name == "_partition_value") - { - if (rows) - inserter.insertPartitionValueColumn(rows, part->partition.value, partition_value_type, virtual_column_name); - else - inserter.insertPartitionValueColumn(rows, {}, partition_value_type, virtual_column_name); - } - } - } -} - -void IMergeTreeSelectAlgorithm::injectVirtualColumns( - Block & block, size_t row_count, MergeTreeReadTask * task, const DataTypePtr & partition_value_type, const Names & virtual_columns) -{ - /// First add non-const columns that are filled by the range reader and then const columns that we will fill ourselves. - /// Note that the order is important: virtual columns filled by the range reader must go first - injectNonConstVirtualColumns(row_count, block, virtual_columns); - injectPartConstVirtualColumns(row_count, block, task, partition_value_type, virtual_columns); -} - -Block IMergeTreeSelectAlgorithm::applyPrewhereActions(Block block, const PrewhereInfoPtr & prewhere_info) -{ - if (prewhere_info) - { - if (prewhere_info->row_level_filter) - { - block = prewhere_info->row_level_filter->updateHeader(std::move(block)); - auto & row_level_column = block.getByName(prewhere_info->row_level_column_name); - if (!row_level_column.type->canBeUsedInBooleanContext()) - { - throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_COLUMN_FOR_FILTER, "Invalid type for filter in PREWHERE: {}", - row_level_column.type->getName()); - } - - block.erase(prewhere_info->row_level_column_name); - } - - if (prewhere_info->prewhere_actions) - { - block = prewhere_info->prewhere_actions->updateHeader(std::move(block)); - - auto & prewhere_column = block.getByName(prewhere_info->prewhere_column_name); - if (!prewhere_column.type->canBeUsedInBooleanContext()) - { - throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_COLUMN_FOR_FILTER, "Invalid type for filter in PREWHERE: {}", - prewhere_column.type->getName()); - } - - if (prewhere_info->remove_prewhere_column) - { - block.erase(prewhere_info->prewhere_column_name); - } - else if (prewhere_info->need_filter) - { - WhichDataType which(removeNullable(recursiveRemoveLowCardinality(prewhere_column.type))); - - if (which.isNativeInt() || which.isNativeUInt()) - prewhere_column.column = prewhere_column.type->createColumnConst(block.rows(), 1u)->convertToFullColumnIfConst(); - else if (which.isFloat()) - prewhere_column.column = prewhere_column.type->createColumnConst(block.rows(), 1.0f)->convertToFullColumnIfConst(); - else - throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_COLUMN_FOR_FILTER, - "Illegal type {} of column for filter", - prewhere_column.type->getName()); - } - } - } - - return block; -} - -Block IMergeTreeSelectAlgorithm::transformHeader( - Block block, const PrewhereInfoPtr & prewhere_info, const DataTypePtr & partition_value_type, const Names & virtual_columns) -{ - auto transformed = applyPrewhereActions(std::move(block), prewhere_info); - injectVirtualColumns(transformed, 0, nullptr, partition_value_type, virtual_columns); - return transformed; -} - -std::unique_ptr IMergeTreeSelectAlgorithm::getSizePredictor( - const MergeTreeData::DataPartPtr & data_part, - const MergeTreeReadTaskColumns & task_columns, - const Block & sample_block) -{ - const auto & required_column_names = task_columns.columns.getNames(); - NameSet complete_column_names(required_column_names.begin(), required_column_names.end()); - for (const auto & pre_columns_per_step : task_columns.pre_columns) - { - const auto & required_pre_column_names = pre_columns_per_step.getNames(); - complete_column_names.insert(required_pre_column_names.begin(), required_pre_column_names.end()); - } - - return std::make_unique( - data_part, Names(complete_column_names.begin(), complete_column_names.end()), sample_block); -} - - -IMergeTreeSelectAlgorithm::~IMergeTreeSelectAlgorithm() = default; - -} diff --git a/src/Storages/MergeTree/MergeTreeBaseSelectProcessor.h b/src/Storages/MergeTree/MergeTreeBaseSelectProcessor.h deleted file mode 100644 index 42043c03e85..00000000000 --- a/src/Storages/MergeTree/MergeTreeBaseSelectProcessor.h +++ /dev/null @@ -1,217 +0,0 @@ -#pragma once -#include -#include -#include -#include -#include -#include - - -namespace DB -{ - -class IMergeTreeReader; -class UncompressedCache; -class MarkCache; -struct PrewhereExprInfo; - -struct ChunkAndProgress -{ - Chunk chunk; - size_t num_read_rows = 0; - size_t num_read_bytes = 0; - /// Explicitly indicate that we have read all data. - /// This is needed to occasionally return empty chunk to indicate the progress while the rows are filtered out in PREWHERE. - bool is_finished = false; -}; - -struct ParallelReadingExtension -{ - MergeTreeAllRangesCallback all_callback; - MergeTreeReadTaskCallback callback; - size_t count_participating_replicas{0}; - size_t number_of_current_replica{0}; - /// This is needed to estimate the number of bytes - /// between a pair of marks to perform one request - /// over the network for a 1Gb of data. - Names columns_to_read; -}; - -/// Base class for MergeTreeThreadSelectAlgorithm and MergeTreeSelectAlgorithm -class IMergeTreeSelectAlgorithm -{ -public: - IMergeTreeSelectAlgorithm( - Block header, - const MergeTreeData & storage_, - const StorageSnapshotPtr & storage_snapshot_, - const PrewhereInfoPtr & prewhere_info_, - const ExpressionActionsSettings & actions_settings, - UInt64 max_block_size_rows_, - UInt64 preferred_block_size_bytes_, - UInt64 preferred_max_column_in_block_size_bytes_, - const MergeTreeReaderSettings & reader_settings_, - bool use_uncompressed_cache_, - const Names & virt_column_names_ = {}); - - virtual ~IMergeTreeSelectAlgorithm(); - - static Block transformHeader( - Block block, const PrewhereInfoPtr & prewhere_info, const DataTypePtr & partition_value_type, const Names & virtual_columns); - - static std::unique_ptr getSizePredictor( - const MergeTreeData::DataPartPtr & data_part, - const MergeTreeReadTaskColumns & task_columns, - const Block & sample_block); - - Block getHeader() const { return result_header; } - - ChunkAndProgress read(); - - void cancel() { is_cancelled = true; } - - const MergeTreeReaderSettings & getSettings() const { return reader_settings; } - - virtual std::string getName() const = 0; - - static PrewhereExprInfo getPrewhereActions(PrewhereInfoPtr prewhere_info, const ExpressionActionsSettings & actions_settings, bool enable_multiple_prewhere_read_steps); - -protected: - /// This struct allow to return block with no columns but with non-zero number of rows similar to Chunk - struct BlockAndProgress - { - Block block; - size_t row_count = 0; - size_t num_read_rows = 0; - size_t num_read_bytes = 0; - }; - - /// Creates new this->task and return a flag whether it was successful or not - virtual bool getNewTaskImpl() = 0; - /// Creates new readers for a task it is needed. These methods are separate, because - /// in case of parallel reading from replicas the whole task could be denied by a coodinator - /// or it could modified somehow. - virtual void finalizeNewTask() = 0; - - size_t estimateMaxBatchSizeForHugeRanges(); - - /// Closes readers and unlock part locks - virtual void finish() = 0; - - virtual BlockAndProgress readFromPart(); - - BlockAndProgress readFromPartImpl(); - - /// Used for filling header with no rows as well as block with data - static void - injectVirtualColumns(Block & block, size_t row_count, MergeTreeReadTask * task, const DataTypePtr & partition_value_type, const Names & virtual_columns); - -protected: - static void initializeRangeReadersImpl( - MergeTreeRangeReader & range_reader, - std::deque & pre_range_readers, - const PrewhereExprInfo & prewhere_actions, - IMergeTreeReader * reader, - bool has_lightweight_delete, - const MergeTreeReaderSettings & reader_settings, - const std::vector> & pre_reader_for_step, - const PrewhereExprStep & lightweight_delete_filter_step, - const Names & non_const_virtual_column_names); - - /// Sets up data readers for each step of prewhere and where - void initializeMergeTreeReadersForCurrentTask( - const IMergeTreeReader::ValueSizeMap & value_size_map, - const ReadBufferFromFileBase::ProfileCallback & profile_callback); - - void initializeMergeTreeReadersForPart( - const MergeTreeData::DataPartPtr & data_part, - const AlterConversionsPtr & alter_conversions, - const MergeTreeReadTaskColumns & task_columns, - const MarkRanges & mark_ranges, - const IMergeTreeReader::ValueSizeMap & value_size_map, - const ReadBufferFromFileBase::ProfileCallback & profile_callback); - - /// Sets up range readers corresponding to data readers - void initializeRangeReaders(MergeTreeReadTask & task); - - const MergeTreeData & storage; - StorageSnapshotPtr storage_snapshot; - - /// This step is added when the part has lightweight delete mask - const PrewhereExprStep lightweight_delete_filter_step - { - .type = PrewhereExprStep::Filter, - .actions = nullptr, - .filter_column_name = LightweightDeleteDescription::FILTER_COLUMN.name, - .remove_filter_column = true, - .need_filter = true, - .perform_alter_conversions = true, - }; - - PrewhereInfoPtr prewhere_info; - ExpressionActionsSettings actions_settings; - PrewhereExprInfo prewhere_actions; - - UInt64 max_block_size_rows; - UInt64 preferred_block_size_bytes; - UInt64 preferred_max_column_in_block_size_bytes; - - MergeTreeReaderSettings reader_settings; - - bool use_uncompressed_cache; - - Names virt_column_names; - - /// These columns will be filled by the merge tree range reader - Names non_const_virtual_column_names; - - DataTypePtr partition_value_type; - - /// This header is used for chunks from readFromPart(). - Block header_without_const_virtual_columns; - /// A result of getHeader(). A chunk which this header is returned from read(). - Block result_header; - - UncompressedCachePtr owned_uncompressed_cache; - MarkCachePtr owned_mark_cache; - - using MergeTreeReaderPtr = std::unique_ptr; - MergeTreeReaderPtr reader; - std::vector pre_reader_for_step; - - MergeTreeReadTaskPtr task; - - /// This setting is used in base algorithm only to additionally limit the number of granules to read. - /// It is changed in ctor of MergeTreeThreadSelectAlgorithm. - /// - /// The reason why we have it here is because MergeTreeReadPool takes the full task - /// ignoring min_marks_to_read setting in case of remote disk (see MergeTreeReadPool::getTask). - /// In this case, we won't limit the number of rows to read based on adaptive granularity settings. - /// - /// Big reading tasks are better for remote disk and prefetches. - /// So, for now it's easier to limit max_rows_to_read. - /// Somebody need to refactor this later. - size_t min_marks_to_read = 0; - -private: - Poco::Logger * log = &Poco::Logger::get("MergeTreeBaseSelectProcessor"); - - std::atomic is_cancelled{false}; - - bool getNewTask(); - - /// Initialize pre readers. - void initializeMergeTreePreReadersForPart( - const MergeTreeData::DataPartPtr & data_part, - const AlterConversionsPtr & alter_conversions, - const MergeTreeReadTaskColumns & task_columns, - const MarkRanges & mark_ranges, - const IMergeTreeReader::ValueSizeMap & value_size_map, - const ReadBufferFromFileBase::ProfileCallback & profile_callback); - - static Block applyPrewhereActions(Block block, const PrewhereInfoPtr & prewhere_info); -}; - -using MergeTreeSelectAlgorithmPtr = std::unique_ptr; - -} diff --git a/src/Storages/MergeTree/MergeTreeBlockReadUtils.cpp b/src/Storages/MergeTree/MergeTreeBlockReadUtils.cpp index 48779aa0df7..f5f0fa6f726 100644 --- a/src/Storages/MergeTree/MergeTreeBlockReadUtils.cpp +++ b/src/Storages/MergeTree/MergeTreeBlockReadUtils.cpp @@ -6,7 +6,7 @@ #include #include #include -#include +#include #include #include #include @@ -16,6 +16,7 @@ namespace DB { + namespace ErrorCodes { extern const int LOGICAL_ERROR; @@ -136,43 +137,6 @@ NameSet injectRequiredColumns( return injected_columns; } - -MergeTreeReadTask::MergeTreeReadTask( - const DataPartPtr & data_part_, - const AlterConversionsPtr & alter_conversions_, - const MarkRanges & mark_ranges_, - size_t part_index_in_query_, - const NameSet & column_name_set_, - const MergeTreeReadTaskColumns & task_columns_, - MergeTreeBlockSizePredictorPtr size_predictor_, - Priority priority_, - std::future reader_, - std::vector> && pre_reader_for_step_) - : data_part{data_part_} - , alter_conversions{alter_conversions_} - , mark_ranges{mark_ranges_} - , part_index_in_query{part_index_in_query_} - , column_name_set{column_name_set_} - , task_columns{task_columns_} - , size_predictor{size_predictor_} - , reader(std::move(reader_)) - , pre_reader_for_step(std::move(pre_reader_for_step_)) - , priority(priority_) -{ -} - -MergeTreeReadTask::~MergeTreeReadTask() -{ - if (reader.valid()) - reader.wait(); - - for (const auto & pre_reader : pre_reader_for_step) - { - if (pre_reader.valid()) - pre_reader.wait(); - } -} - MergeTreeBlockSizePredictor::MergeTreeBlockSizePredictor( const DataPartPtr & data_part_, const Names & columns, const Block & sample_block) : data_part(data_part_) @@ -195,9 +159,8 @@ void MergeTreeBlockSizePredictor::initialize(const Block & sample_block, const C for (size_t pos = 0; pos < num_columns; ++pos) { const auto & column_with_type_and_name = sample_block.getByPosition(pos); - const String & column_name = column_with_type_and_name.name; - const ColumnPtr & column_data = from_update ? columns[pos] - : column_with_type_and_name.column; + const auto & column_name = column_with_type_and_name.name; + const auto & column_data = from_update ? columns[pos] : column_with_type_and_name.column; if (!from_update && !names_set.contains(column_name)) continue; @@ -246,7 +209,6 @@ void MergeTreeBlockSizePredictor::startBlock() info.size_bytes = 0; } - /// TODO: add last_read_row_in_part parameter to take into account gaps between adjacent ranges void MergeTreeBlockSizePredictor::update(const Block & sample_block, const Columns & columns, size_t num_rows, double decay) { @@ -296,7 +258,7 @@ void MergeTreeBlockSizePredictor::update(const Block & sample_block, const Colum } -MergeTreeReadTaskColumns getReadTaskColumns( +MergeTreeReadTask::Columns getReadTaskColumns( const IMergeTreeDataPartInfoForReader & data_part_info_for_reader, const StorageSnapshotPtr & storage_snapshot, const Names & required_columns, @@ -317,7 +279,7 @@ MergeTreeReadTaskColumns getReadTaskColumns( injectRequiredColumns( data_part_info_for_reader, storage_snapshot, with_subcolumns, column_to_read_after_prewhere); - MergeTreeReadTaskColumns result; + MergeTreeReadTask::Columns result; auto options = GetColumnsOptions(GetColumnsOptions::All) .withExtendedObjects() .withSystemColumns(); @@ -365,7 +327,7 @@ MergeTreeReadTaskColumns getReadTaskColumns( if (prewhere_info) { - auto prewhere_actions = IMergeTreeSelectAlgorithm::getPrewhereActions( + auto prewhere_actions = MergeTreeSelectProcessor::getPrewhereActions( prewhere_info, actions_settings, reader_settings.enable_multiple_prewhere_read_steps); @@ -387,16 +349,4 @@ MergeTreeReadTaskColumns getReadTaskColumns( return result; } - -std::string MergeTreeReadTaskColumns::dump() const -{ - WriteBufferFromOwnString s; - for (size_t i = 0; i < pre_columns.size(); ++i) - { - s << "STEP " << i << ": " << pre_columns[i].toString() << "\n"; - } - s << "COLUMNS: " << columns.toString() << "\n"; - return s.str(); -} - } diff --git a/src/Storages/MergeTree/MergeTreeBlockReadUtils.h b/src/Storages/MergeTree/MergeTreeBlockReadUtils.h index 5df3e600a1a..9417d47814a 100644 --- a/src/Storages/MergeTree/MergeTreeBlockReadUtils.h +++ b/src/Storages/MergeTree/MergeTreeBlockReadUtils.h @@ -2,29 +2,15 @@ #include #include -#include -#include -#include -#include -#include +#include namespace DB { -class MergeTreeData; -struct MergeTreeReadTask; struct MergeTreeReaderSettings; -struct MergeTreeBlockSizePredictor; class IMergeTreeDataPartInfoForReader; -using MergeTreeReadTaskPtr = std::unique_ptr; -using MergeTreeBlockSizePredictorPtr = std::shared_ptr; - -class IMergeTreeDataPart; -using DataPartPtr = std::shared_ptr; - - /** If some of the requested columns are not in the part, * then find out which columns may need to be read further, * so that you can calculate the DEFAULT expression for these columns. @@ -36,64 +22,7 @@ NameSet injectRequiredColumns( bool with_subcolumns, Names & columns); -struct MergeTreeReadTaskColumns -{ - /// column names to read during WHERE - NamesAndTypesList columns; - /// column names to read during each PREWHERE step - std::vector pre_columns; - - std::string dump() const; -}; - -/// A batch of work for MergeTreeThreadSelectProcessor -struct MergeTreeReadTask -{ - /// Data part which should be read while performing this task - DataPartPtr data_part; - /// Alter converversionss that should be applied on-fly for part. - AlterConversionsPtr alter_conversions; - /// Ranges to read from `data_part`. - MarkRanges mark_ranges; - /// for virtual `part_index` virtual column - size_t part_index_in_query; - /// used to determine whether column should be filtered during PREWHERE or WHERE - const NameSet & column_name_set; - /// column names to read during PREWHERE and WHERE - const MergeTreeReadTaskColumns & task_columns; - /// Used to satistfy preferred_block_size_bytes limitation - MergeTreeBlockSizePredictorPtr size_predictor; - /// Used to save current range processing status - MergeTreeRangeReader range_reader; - /// Range readers for multiple filtering steps: row level security, PREWHERE etc. - /// NOTE: we take references to elements and push_back new elements, that's why it is a deque but not a vector - std::deque pre_range_readers; - - using MergeTreeReaderPtr = std::unique_ptr; - std::future reader; - std::vector> pre_reader_for_step; - - Priority priority; - - bool isFinished() const { return mark_ranges.empty() && range_reader.isCurrentRangeFinished(); } - - MergeTreeReadTask( - const DataPartPtr & data_part_, - const AlterConversionsPtr & alter_conversions_, - const MarkRanges & mark_ranges_, - size_t part_index_in_query_, - const NameSet & column_name_set_, - const MergeTreeReadTaskColumns & task_columns_, - MergeTreeBlockSizePredictorPtr size_predictor_, - Priority priority_ = {}, - std::future reader_ = {}, - std::vector> && pre_reader_for_step_ = {}); - - ~MergeTreeReadTask(); -}; - - -MergeTreeReadTaskColumns getReadTaskColumns( +MergeTreeReadTask::Columns getReadTaskColumns( const IMergeTreeDataPartInfoForReader & data_part_info_for_reader, const StorageSnapshotPtr & storage_snapshot, const Names & required_columns, @@ -119,7 +48,6 @@ struct MergeTreeBlockSizePredictor return block_size_bytes; } - /// Predicts what number of rows should be read to exhaust byte quota per column inline size_t estimateNumRowsForMaxSizeColumn(size_t bytes_quota) const { @@ -153,7 +81,6 @@ struct MergeTreeBlockSizePredictor static double calculateDecay() { return 1. - std::pow(TARGET_WEIGHT, 1. / NUM_UPDATES_TO_TARGET_WEIGHT); } protected: - DataPartPtr data_part; struct ColumnInfo diff --git a/src/Storages/MergeTree/MergeTreeData.cpp b/src/Storages/MergeTree/MergeTreeData.cpp index f7138e73c77..f9f0fc704ad 100644 --- a/src/Storages/MergeTree/MergeTreeData.cpp +++ b/src/Storages/MergeTree/MergeTreeData.cpp @@ -68,7 +68,7 @@ #include #include #include -#include +#include #include #include #include diff --git a/src/Storages/MergeTree/MergeTreeInOrderSelectProcessor.cpp b/src/Storages/MergeTree/MergeTreeInOrderSelectProcessor.cpp deleted file mode 100644 index 2676adfac2d..00000000000 --- a/src/Storages/MergeTree/MergeTreeInOrderSelectProcessor.cpp +++ /dev/null @@ -1,76 +0,0 @@ -#include -#include "Storages/MergeTree/RangesInDataPart.h" -#include - -namespace DB -{ - -namespace ErrorCodes -{ - extern const int MEMORY_LIMIT_EXCEEDED; -} - -bool MergeTreeInOrderSelectAlgorithm::getNewTaskImpl() -try -{ - if (all_mark_ranges.empty()) - return false; - - if (!reader) - initializeReaders(); - - MarkRanges mark_ranges_for_task; - - if (!pool) - { - /// If we need to read few rows, set one range per task to reduce number of read data. - if (has_limit_below_one_block) - { - mark_ranges_for_task = MarkRanges{}; - mark_ranges_for_task.emplace_front(std::move(all_mark_ranges.front())); - all_mark_ranges.pop_front(); - } - else - { - mark_ranges_for_task = std::move(all_mark_ranges); - all_mark_ranges.clear(); - } - } - else - { - auto description = RangesInDataPartDescription{ - .info = data_part->info, - /// We just ignore all the distribution done before - /// Everything will be done on coordinator side - .ranges = {}, - }; - - mark_ranges_for_task = pool->getNewTask(description); - - if (mark_ranges_for_task.empty()) - return false; - } - - auto size_predictor = (preferred_block_size_bytes == 0) ? nullptr - : getSizePredictor(data_part, task_columns, sample_block); - - task = std::make_unique( - data_part, - alter_conversions, - mark_ranges_for_task, - part_index_in_query, - column_name_set, - task_columns, - std::move(size_predictor)); - - return true; -} -catch (...) -{ - /// Suspicion of the broken part. A part is added to the queue for verification. - if (getCurrentExceptionCode() != ErrorCodes::MEMORY_LIMIT_EXCEEDED) - storage.reportBrokenPart(data_part); - throw; -} - -} diff --git a/src/Storages/MergeTree/MergeTreeInOrderSelectProcessor.h b/src/Storages/MergeTree/MergeTreeInOrderSelectProcessor.h deleted file mode 100644 index da332c64232..00000000000 --- a/src/Storages/MergeTree/MergeTreeInOrderSelectProcessor.h +++ /dev/null @@ -1,33 +0,0 @@ -#pragma once -#include -#include - -namespace DB -{ - - -/// Used to read data from single part with select query in order of primary key. -/// Cares about PREWHERE, virtual columns, indexes etc. -/// To read data from multiple parts, Storage (MergeTree) creates multiple such objects. -class MergeTreeInOrderSelectAlgorithm final : public MergeTreeSelectAlgorithm -{ -public: - template - explicit MergeTreeInOrderSelectAlgorithm(Args &&... args) - : MergeTreeSelectAlgorithm{std::forward(args)...} - { - LOG_TRACE(log, "Reading {} ranges in order from part {}, approx. {} rows starting from {}", - all_mark_ranges.size(), data_part->name, total_rows, - data_part->index_granularity.getMarkStartingRow(all_mark_ranges.front().begin)); - } - - String getName() const override { return "MergeTreeInOrder"; } - -private: - bool getNewTaskImpl() override; - void finalizeNewTask() override {} - - Poco::Logger * log = &Poco::Logger::get("MergeTreeInOrderSelectProcessor"); -}; - -} diff --git a/src/Storages/MergeTree/MergeTreePrefetchedReadPool.cpp b/src/Storages/MergeTree/MergeTreePrefetchedReadPool.cpp index 901801d8187..8e0681bc5e6 100644 --- a/src/Storages/MergeTree/MergeTreePrefetchedReadPool.cpp +++ b/src/Storages/MergeTree/MergeTreePrefetchedReadPool.cpp @@ -5,7 +5,6 @@ #include #include #include -#include #include #include #include @@ -14,10 +13,10 @@ #include #include - namespace ProfileEvents { extern const Event MergeTreePrefetchedReadPoolInit; + extern const Event WaitPrefetchTaskMicroseconds; } namespace DB @@ -29,78 +28,79 @@ namespace ErrorCodes extern const int BAD_ARGUMENTS; } +bool MergeTreePrefetchedReadPool::TaskHolder::operator<(const TaskHolder & other) const +{ + chassert(task->priority >= 0); + chassert(other.task->priority >= 0); + /// With default std::priority_queue, top() returns largest element. + /// So closest to 0 will be on top with this comparator. + return task->priority > other.task->priority; /// Less is better. +} + +void MergeTreePrefetchedReadPool::ReadersFuture::wait() +{ + main.wait(); + for (auto & reader_future : prewhere) + reader_future.wait(); +} + +MergeTreeReadTask::Readers MergeTreePrefetchedReadPool::ReadersFuture::get() +{ + MergeTreeReadTask::Readers readers; + ProfileEventTimeIncrement watch(ProfileEvents::WaitPrefetchTaskMicroseconds); + + readers.main = main.get(); + for (auto & reader_future : prewhere) + readers.prewhere.push_back(reader_future.get()); + return readers; +} + MergeTreePrefetchedReadPool::MergeTreePrefetchedReadPool( - size_t threads, - size_t sum_marks_, - size_t min_marks_for_concurrent_read_, RangesInDataParts && parts_, const StorageSnapshotPtr & storage_snapshot_, const PrewhereInfoPtr & prewhere_info_, const ExpressionActionsSettings & actions_settings_, + const MergeTreeReaderSettings & reader_settings_, const Names & column_names_, const Names & virtual_column_names_, - size_t preferred_block_size_bytes_, - const MergeTreeReaderSettings & reader_settings_, - ContextPtr context_, - bool use_uncompressed_cache_, - bool is_remote_read_, - const MergeTreeSettings & storage_settings_) - : WithContext(context_) - , log(&Poco::Logger::get("MergeTreePrefetchedReadPool(" + (parts_.empty() ? "" : parts_.front().data_part->storage.getStorageID().getNameForLogs()) + ")")) - , header(storage_snapshot_->getSampleBlockForColumns(column_names_)) - , mark_cache(context_->getGlobalContext()->getMarkCache().get()) - , uncompressed_cache(use_uncompressed_cache_ ? context_->getGlobalContext()->getUncompressedCache().get() : nullptr) - , profile_callback([this](ReadBufferFromFileBase::ProfileInfo info_) { profileFeedback(info_); }) - , index_granularity_bytes(storage_settings_.index_granularity_bytes) - , fixed_index_granularity(storage_settings_.index_granularity) - , storage_snapshot(storage_snapshot_) - , column_names(column_names_) - , virtual_column_names(virtual_column_names_) - , prewhere_info(prewhere_info_) - , actions_settings(actions_settings_) - , reader_settings(reader_settings_) - , is_remote_read(is_remote_read_) + const PoolSettings & settings_, + const ContextPtr & context_) + : MergeTreeReadPoolBase( + std::move(parts_), + storage_snapshot_, + prewhere_info_, + actions_settings_, + reader_settings_, + column_names_, + virtual_column_names_, + settings_, + context_) + , WithContext(context_) , prefetch_threadpool(getContext()->getPrefetchThreadpool()) + , log(&Poco::Logger::get("MergeTreePrefetchedReadPool(" + (parts_.empty() ? "" : parts_.front().data_part->storage.getStorageID().getNameForLogs()) + ")")) { /// Tasks creation might also create a lost of readers - check they do not /// do any time consuming operations in ctor. ProfileEventTimeIncrement watch(ProfileEvents::MergeTreePrefetchedReadPoolInit); - parts_infos = getPartsInfos(parts_, preferred_block_size_bytes_); - threads_tasks = createThreadsTasks(threads, sum_marks_, min_marks_for_concurrent_read_); + fillPerPartStatistics(); + fillPerThreadTasks(pool_settings.threads, pool_settings.sum_marks); } -struct MergeTreePrefetchedReadPool::PartInfo +MergeTreePrefetchedReadPool::~MergeTreePrefetchedReadPool() { - MergeTreeData::DataPartPtr data_part; - AlterConversionsPtr alter_conversions; - size_t part_index_in_query; - size_t sum_marks = 0; - MarkRanges ranges; + for (auto & [_, tasks] : per_thread_tasks) + { + for (auto & task : tasks) + { + if (task->readers_future.valid()) + task->readers_future.wait(); + } + } +} - NameSet column_name_set; - MergeTreeReadTaskColumns task_columns; - MergeTreeBlockSizePredictorPtr size_predictor; - - size_t approx_size_of_mark = 0; - size_t prefetch_step_marks = 0; - - size_t estimated_memory_usage_for_single_prefetch = 0; - size_t required_readers_num = 0; -}; - -std::future MergeTreePrefetchedReadPool::createPrefetchedReader( - const IMergeTreeDataPart & data_part, - const NamesAndTypesList & columns, - const AlterConversionsPtr & alter_conversions, - const MarkRanges & required_ranges, - Priority priority) const +std::future MergeTreePrefetchedReadPool::createPrefetchedReader(MergeTreeReaderPtr reader, Priority priority) { - auto reader = data_part.getReader( - columns, storage_snapshot, required_ranges, - uncompressed_cache, mark_cache, alter_conversions, reader_settings, - IMergeTreeReader::ValueSizeMap{}, profile_callback); - /// In order to make a prefetch we need to wait for marks to be loaded. But we just created /// a reader (which starts loading marks in its constructor), then if we do prefetch right /// after creating a reader, it will be very inefficient. We can do prefetch for all parts @@ -116,58 +116,42 @@ std::future MergeTreePrefetchedReadPool::createPrefetchedRea my_reader->prefetchBeginOfRange(priority); return std::move(my_reader); }; + return scheduleFromThreadPool(std::move(task), prefetch_threadpool, "ReadPrepare", priority); } -void MergeTreePrefetchedReadPool::createPrefetchedReaderForTask(MergeTreeReadTask & task) const +void MergeTreePrefetchedReadPool::createPrefetchedReadersForTask(ThreadTask & task) { - if (task.reader.valid()) + if (task.readers_future.valid()) throw Exception(ErrorCodes::LOGICAL_ERROR, "Task already has a reader"); - task.reader = createPrefetchedReader(*task.data_part, task.task_columns.columns, task.alter_conversions, task.mark_ranges, task.priority); + auto extras = getExtras(); + auto readers = MergeTreeReadTask::createReaders(task.read_info, extras, task.ranges); - if (reader_settings.apply_deleted_mask && task.data_part->hasLightweightDelete()) - { - auto pre_reader = createPrefetchedReader(*task.data_part, {LightweightDeleteDescription::FILTER_COLUMN}, task.alter_conversions, task.mark_ranges, task.priority); - task.pre_reader_for_step.push_back(std::move(pre_reader)); - } - - for (const auto & pre_columns_per_step : task.task_columns.pre_columns) - { - auto pre_reader = createPrefetchedReader(*task.data_part, pre_columns_per_step, task.alter_conversions, task.mark_ranges, task.priority); - task.pre_reader_for_step.push_back(std::move(pre_reader)); - } + task.readers_future.main = createPrefetchedReader(std::move(readers.main), task.priority); + for (auto && reader : readers.prewhere) + task.readers_future.prewhere.push_back(createPrefetchedReader(std::move(reader), task.priority)); } -bool MergeTreePrefetchedReadPool::TaskHolder::operator <(const TaskHolder & other) const -{ - chassert(task->priority >= 0); - chassert(other.task->priority >= 0); - return task->priority > other.task->priority; /// Less is better. - /// With default std::priority_queue, top() returns largest element. - /// So closest to 0 will be on top with this comparator. -} - -void MergeTreePrefetchedReadPool::startPrefetches() const +void MergeTreePrefetchedReadPool::startPrefetches() { if (prefetch_queue.empty()) return; - [[maybe_unused]] TaskHolder prev(nullptr, 0); + [[maybe_unused]] TaskHolder prev; [[maybe_unused]] const Priority highest_priority{reader_settings.read_settings.priority.value + 1}; assert(prefetch_queue.top().task->priority == highest_priority); + while (!prefetch_queue.empty()) { - const auto & top = prefetch_queue.top(); - createPrefetchedReaderForTask(*top.task); + auto & top = prefetch_queue.top(); + createPrefetchedReadersForTask(*top.task); #ifndef NDEBUG if (prev.task) { assert(top.task->priority >= highest_priority); if (prev.thread_id == top.thread_id) - { assert(prev.task->priority < top.task->priority); - } } prev = top; #endif @@ -175,11 +159,11 @@ void MergeTreePrefetchedReadPool::startPrefetches() const } } -MergeTreeReadTaskPtr MergeTreePrefetchedReadPool::getTask(size_t thread) +MergeTreeReadTaskPtr MergeTreePrefetchedReadPool::getTask(size_t task_idx, MergeTreeReadTask * previous_task) { std::lock_guard lock(mutex); - if (threads_tasks.empty()) + if (per_thread_tasks.empty()) return nullptr; if (!started_prefetches) @@ -188,112 +172,129 @@ MergeTreeReadTaskPtr MergeTreePrefetchedReadPool::getTask(size_t thread) startPrefetches(); } - auto it = threads_tasks.find(thread); - if (it == threads_tasks.end()) - { - ThreadsTasks::iterator non_prefetched_tasks_to_steal = threads_tasks.end(); - ThreadsTasks::iterator prefetched_tasks_to_steal = threads_tasks.end(); - int64_t best_prefetched_task_priority = -1; - - /// There is no point stealing in order (like in MergeTreeReadPool, where tasks can be stolen - /// only from the next thread). Even if we steal task from the next thread, which reads from - /// the same part as we just read, it might seem that we can reuse our own reader, do some - /// seek avoiding and it will have a good result as we avoided seek (new request). But it is - /// not so, because this next task will most likely have its own reader a prefetch already on - /// the fly. (Not to mention that in fact we cannot reuse our own reader if initially we did - /// not accounted this range into range request to object storage). - for (auto thread_tasks_it = threads_tasks.begin(); thread_tasks_it != threads_tasks.end(); ++thread_tasks_it) - { - /// Prefer to steal tasks which have an initialized reader (with prefetched data). Thus we avoid - /// losing a prefetch by creating our own reader (or resusing our own reader if the part - /// is the same as last read by this thread). - auto & thread_tasks = thread_tasks_it->second; - auto task_it = std::find_if( - thread_tasks.begin(), thread_tasks.end(), - [](const auto & task) { return task->reader.valid(); }); - - if (task_it == thread_tasks.end()) - { - /// The follow back to non-prefetched task should lie on the thread which - /// has more tasks than others. - if (non_prefetched_tasks_to_steal == threads_tasks.end() - || non_prefetched_tasks_to_steal->second.size() < thread_tasks.size()) - non_prefetched_tasks_to_steal = thread_tasks_it; - } - /// Try to steal task with the best (lowest) priority (because it will be executed faster). - else if (prefetched_tasks_to_steal == threads_tasks.end() - || (*task_it)->priority < best_prefetched_task_priority) - { - best_prefetched_task_priority = (*task_it)->priority; - chassert(best_prefetched_task_priority >= 0); - prefetched_tasks_to_steal = thread_tasks_it; - } - } - - if (prefetched_tasks_to_steal != threads_tasks.end()) - { - auto & thread_tasks = prefetched_tasks_to_steal->second; - assert(!thread_tasks.empty()); - - auto task_it = std::find_if( - thread_tasks.begin(), thread_tasks.end(), - [](const auto & task) { return task->reader.valid(); }); - assert(task_it != thread_tasks.end()); - - auto task = std::move(*task_it); - thread_tasks.erase(task_it); - - if (thread_tasks.empty()) - threads_tasks.erase(prefetched_tasks_to_steal); - - return task; - } - - /// TODO: it also makes sense to first try to steal from the next thread if it has ranges - /// from the same part as current thread last read - to reuse the reader. - - if (non_prefetched_tasks_to_steal != threads_tasks.end()) - { - auto & thread_tasks = non_prefetched_tasks_to_steal->second; - assert(!thread_tasks.empty()); - - /// Get second half of the tasks. - const size_t total_tasks = thread_tasks.size(); - const size_t half = total_tasks / 2; - auto half_it = thread_tasks.begin() + half; - assert(half_it != thread_tasks.end()); - - /// Give them to current thread, as current thread's tasks list is empty. - auto & current_thread_tasks = threads_tasks[thread]; - current_thread_tasks.insert( - current_thread_tasks.end(), make_move_iterator(half_it), make_move_iterator(thread_tasks.end())); - - /// Erase them from the thread from which we steal. - thread_tasks.resize(half); - if (thread_tasks.empty()) - threads_tasks.erase(non_prefetched_tasks_to_steal); - - auto task = std::move(current_thread_tasks.front()); - current_thread_tasks.erase(current_thread_tasks.begin()); - if (current_thread_tasks.empty()) - threads_tasks.erase(thread); - - return task; - } - - return nullptr; - } + auto it = per_thread_tasks.find(task_idx); + if (it == per_thread_tasks.end()) + return stealTask(task_idx, previous_task); auto & thread_tasks = it->second; assert(!thread_tasks.empty()); - auto task = std::move(thread_tasks.front()); + auto thread_task = std::move(thread_tasks.front()); thread_tasks.pop_front(); if (thread_tasks.empty()) - threads_tasks.erase(it); + per_thread_tasks.erase(it); - return task; + return createTask(*thread_task, previous_task); +} + +MergeTreeReadTaskPtr MergeTreePrefetchedReadPool::stealTask(size_t thread, MergeTreeReadTask * previous_task) +{ + auto non_prefetched_tasks_to_steal = per_thread_tasks.end(); + auto prefetched_tasks_to_steal = per_thread_tasks.end(); + int64_t best_prefetched_task_priority = -1; + + /// There is no point stealing in order (like in MergeTreeReadPool, where tasks can be stolen + /// only from the next thread). Even if we steal task from the next thread, which reads from + /// the same part as we just read, it might seem that we can reuse our own reader, do some + /// seek avoiding and it will have a good result as we avoided seek (new request). But it is + /// not so, because this next task will most likely have its own reader a prefetch already on + /// the fly. (Not to mention that in fact we cannot reuse our own reader if initially we did + /// not accounted this range into range request to object storage). + for (auto thread_tasks_it = per_thread_tasks.begin(); thread_tasks_it != per_thread_tasks.end(); ++thread_tasks_it) + { + /// Prefer to steal tasks which have an initialized reader (with prefetched data). Thus we avoid + /// losing a prefetch by creating our own reader (or resusing our own reader if the part + /// is the same as last read by this thread). + auto & thread_tasks = thread_tasks_it->second; + + auto task_it = std::find_if( + thread_tasks.begin(), thread_tasks.end(), + [](const auto & task) { return task->readers_future.valid(); }); + + if (task_it == thread_tasks.end()) + { + /// The follow back to non-prefetched task should lie on the thread which + /// has more tasks than others. + if (non_prefetched_tasks_to_steal == per_thread_tasks.end() + || non_prefetched_tasks_to_steal->second.size() < thread_tasks.size()) + non_prefetched_tasks_to_steal = thread_tasks_it; + } + /// Try to steal task with the best (lowest) priority (because it will be executed faster). + else if (prefetched_tasks_to_steal == per_thread_tasks.end() + || (*task_it)->priority < best_prefetched_task_priority) + { + best_prefetched_task_priority = (*task_it)->priority; + chassert(best_prefetched_task_priority >= 0); + prefetched_tasks_to_steal = thread_tasks_it; + } + } + + if (prefetched_tasks_to_steal != per_thread_tasks.end()) + { + auto & thread_tasks = prefetched_tasks_to_steal->second; + assert(!thread_tasks.empty()); + + auto task_it = std::find_if( + thread_tasks.begin(), thread_tasks.end(), + [](const auto & task) { return task->readers_future.valid(); }); + + assert(task_it != thread_tasks.end()); + auto thread_task = std::move(*task_it); + thread_tasks.erase(task_it); + + if (thread_tasks.empty()) + per_thread_tasks.erase(prefetched_tasks_to_steal); + + return createTask(*thread_task, previous_task); + } + + /// TODO: it also makes sense to first try to steal from the next thread if it has ranges + /// from the same part as current thread last read - to reuse the reader. + if (non_prefetched_tasks_to_steal != per_thread_tasks.end()) + { + auto & thread_tasks = non_prefetched_tasks_to_steal->second; + assert(!thread_tasks.empty()); + + /// Get second half of the tasks. + const size_t total_tasks = thread_tasks.size(); + const size_t half = total_tasks / 2; + auto half_it = thread_tasks.begin() + half; + assert(half_it != thread_tasks.end()); + + /// Give them to current thread, as current thread's tasks list is empty. + auto & current_thread_tasks = per_thread_tasks[thread]; + current_thread_tasks.insert( + current_thread_tasks.end(), make_move_iterator(half_it), make_move_iterator(thread_tasks.end())); + + /// Erase them from the thread from which we steal. + thread_tasks.resize(half); + if (thread_tasks.empty()) + per_thread_tasks.erase(non_prefetched_tasks_to_steal); + + auto thread_task = std::move(current_thread_tasks.front()); + current_thread_tasks.erase(current_thread_tasks.begin()); + if (current_thread_tasks.empty()) + per_thread_tasks.erase(thread); + + return createTask(*thread_task, previous_task); + } + + return nullptr; +} + +MergeTreeReadTaskPtr MergeTreePrefetchedReadPool::createTask(ThreadTask & task, MergeTreeReadTask * previous_task) +{ + if (task.readers_future.valid()) + { + auto size_predictor = task.read_info->shared_size_predictor + ? std::make_unique(*task.read_info->shared_size_predictor) + : nullptr; + + return std::make_unique(task.read_info, task.readers_future.get(), task.ranges, std::move(size_predictor)); + } + + return MergeTreeReadPoolBase::createTask(task.read_info, task.ranges, previous_task); } size_t getApproximateSizeOfGranule(const IMergeTreeDataPart & part, const Names & columns_to_read) @@ -304,154 +305,111 @@ size_t getApproximateSizeOfGranule(const IMergeTreeDataPart & part, const Names return columns_size.data_compressed / part.getMarksCount(); } -MergeTreePrefetchedReadPool::PartsInfos MergeTreePrefetchedReadPool::getPartsInfos( - const RangesInDataParts & parts, size_t preferred_block_size_bytes) const +void MergeTreePrefetchedReadPool::fillPerPartStatistics() { - PartsInfos result; - Block sample_block = storage_snapshot->metadata->getSampleBlock(); + per_part_statistics.clear(); + per_part_statistics.reserve(parts_ranges.size()); const auto & settings = getContext()->getSettingsRef(); - const bool predict_block_size_bytes = preferred_block_size_bytes > 0; - for (const auto & part : parts) + for (size_t i = 0; i < parts_ranges.size(); ++i) { - auto part_info = std::make_unique(); - - part_info->data_part = part.data_part; - part_info->alter_conversions = part.alter_conversions; - part_info->part_index_in_query = part.part_index_in_query; - part_info->ranges = part.ranges; - std::sort(part_info->ranges.begin(), part_info->ranges.end()); - - LoadedMergeTreeDataPartInfoForReader part_reader_info(part.data_part, part_info->alter_conversions); + auto & part_stat = per_part_statistics.emplace_back(); + const auto & read_info = *per_part_infos[i]; /// Sum up total size of all mark ranges in a data part. - for (const auto & range : part.ranges) - part_info->sum_marks += range.end - range.begin; + for (const auto & range : parts_ranges[i].ranges) + part_stat.sum_marks += range.end - range.begin; const auto & columns = settings.merge_tree_determine_task_size_by_prewhere_columns && prewhere_info ? prewhere_info->prewhere_actions->getRequiredColumnsNames() : column_names; - part_info->approx_size_of_mark = getApproximateSizeOfGranule(*part_info->data_part, columns); - const auto task_columns = getReadTaskColumns( - part_reader_info, - storage_snapshot, - column_names, - virtual_column_names, - prewhere_info, - actions_settings, - reader_settings, - /* with_subcolumns */ true); + part_stat.approx_size_of_mark = getApproximateSizeOfGranule(*read_info.data_part, columns); - part_info->size_predictor = !predict_block_size_bytes - ? nullptr - : IMergeTreeSelectAlgorithm::getSizePredictor(part.data_part, task_columns, sample_block); - - /// Will be used to distinguish between PREWHERE and WHERE columns when applying filter. - const auto & required_column_names = task_columns.columns.getNames(); - part_info->column_name_set = {required_column_names.begin(), required_column_names.end()}; - part_info->task_columns = task_columns; + auto update_stat_for_column = [&](const auto & column_name) + { + size_t column_size = read_info.data_part->getColumnSize(column_name).data_compressed; + part_stat.estimated_memory_usage_for_single_prefetch += std::min(column_size, settings.prefetch_buffer_size); + ++part_stat.required_readers_num; + }; /// adjustBufferSize(), which is done in MergeTreeReaderStream and MergeTreeReaderCompact, /// lowers buffer size if file size (or required read range) is less. So we know that the /// settings.prefetch_buffer_size will be lowered there, therefore we account it here as well. /// But here we make a more approximate lowering (because we do not have loaded marks yet), /// while in adjustBufferSize it will be presize. - for (const auto & col : task_columns.columns) - { - const auto col_size = part.data_part->getColumnSize(col.name).data_compressed; - part_info->estimated_memory_usage_for_single_prefetch += std::min(col_size, settings.prefetch_buffer_size); - ++part_info->required_readers_num; - } - if (reader_settings.apply_deleted_mask && part.data_part->hasLightweightDelete()) - { - const auto col_size = part.data_part->getColumnSize( - LightweightDeleteDescription::FILTER_COLUMN.name).data_compressed; - part_info->estimated_memory_usage_for_single_prefetch += std::min(col_size, settings.prefetch_buffer_size); - ++part_info->required_readers_num; - } - if (prewhere_info) - { - for (const auto & cols : task_columns.pre_columns) - { - for (const auto & col : cols) - { - const size_t col_size = part.data_part->getColumnSize(col.name).data_compressed; - part_info->estimated_memory_usage_for_single_prefetch += std::min(col_size, settings.prefetch_buffer_size); - ++part_info->required_readers_num; - } - } - } + for (const auto & column : read_info.task_columns.columns) + update_stat_for_column(column.name); - result.push_back(std::move(part_info)); + if (reader_settings.apply_deleted_mask && read_info.data_part->hasLightweightDelete()) + update_stat_for_column(LightweightDeleteDescription::FILTER_COLUMN.name); + + for (const auto & pre_columns : read_info.task_columns.pre_columns) + for (const auto & column : pre_columns) + update_stat_for_column(column.name); } - - return result; } -MergeTreePrefetchedReadPool::ThreadsTasks MergeTreePrefetchedReadPool::createThreadsTasks( - size_t threads, size_t sum_marks, size_t /* min_marks_for_concurrent_read */) const +void MergeTreePrefetchedReadPool::fillPerThreadTasks(size_t threads, size_t sum_marks) { - if (parts_infos.empty()) - return {}; + if (per_part_infos.empty()) + return; const auto & context = getContext(); const auto & settings = context->getSettingsRef(); size_t total_size_approx = 0; - for (const auto & part : parts_infos) - { - total_size_approx += part->sum_marks * part->approx_size_of_mark; - } + for (const auto & part : per_part_statistics) + total_size_approx += part.sum_marks * part.approx_size_of_mark; size_t min_prefetch_step_marks = 0; - - for (const auto & part : parts_infos) + for (size_t i = 0; i < per_part_infos.size(); ++i) { + auto & part_stat = per_part_statistics[i]; + if (settings.filesystem_prefetch_step_marks) { - part->prefetch_step_marks = settings.filesystem_prefetch_step_marks; + part_stat.prefetch_step_marks = settings.filesystem_prefetch_step_marks; } - else if (settings.filesystem_prefetch_step_bytes && part->approx_size_of_mark) + else if (settings.filesystem_prefetch_step_bytes && part_stat.approx_size_of_mark) { - part->prefetch_step_marks = std::max( - 1, static_cast(std::round(static_cast(settings.filesystem_prefetch_step_bytes) / part->approx_size_of_mark))); + part_stat.prefetch_step_marks = std::max( + 1, static_cast(std::round(static_cast(settings.filesystem_prefetch_step_bytes) / part_stat.approx_size_of_mark))); } /// This limit is important to avoid spikes of slow aws getObject requests when parallelizing within one file. /// (The default is taken from here https://docs.aws.amazon.com/whitepapers/latest/s3-optimizing-performance-best-practices/use-byte-range-fetches.html). - if (part->approx_size_of_mark + if (part_stat.approx_size_of_mark && settings.filesystem_prefetch_min_bytes_for_single_read_task - && part->approx_size_of_mark < settings.filesystem_prefetch_min_bytes_for_single_read_task) + && part_stat.approx_size_of_mark < settings.filesystem_prefetch_min_bytes_for_single_read_task) { const size_t min_prefetch_step_marks_by_total_cols = static_cast( - std::ceil(static_cast(settings.filesystem_prefetch_min_bytes_for_single_read_task) / part->approx_size_of_mark)); + std::ceil(static_cast(settings.filesystem_prefetch_min_bytes_for_single_read_task) / part_stat.approx_size_of_mark)); + /// At least one task to start working on it right now and another one to prefetch in the meantime. const size_t new_min_prefetch_step_marks = std::min(min_prefetch_step_marks_by_total_cols, sum_marks / threads / 2); if (min_prefetch_step_marks < new_min_prefetch_step_marks) { LOG_DEBUG(log, "Increasing min prefetch step from {} to {}", min_prefetch_step_marks, new_min_prefetch_step_marks); - min_prefetch_step_marks = new_min_prefetch_step_marks; } } - if (part->prefetch_step_marks < min_prefetch_step_marks) + if (part_stat.prefetch_step_marks < min_prefetch_step_marks) { - LOG_DEBUG(log, "Increasing prefetch step from {} to {}", part->prefetch_step_marks, min_prefetch_step_marks); - - part->prefetch_step_marks = min_prefetch_step_marks; + LOG_DEBUG(log, "Increasing prefetch step from {} to {}", part_stat.prefetch_step_marks, min_prefetch_step_marks); + part_stat.prefetch_step_marks = min_prefetch_step_marks; } LOG_DEBUG( log, "Part: {}, sum_marks: {}, approx mark size: {}, prefetch_step_bytes: {}, prefetch_step_marks: {}, (ranges: {})", - part->data_part->name, - part->sum_marks, - part->approx_size_of_mark, + parts_ranges[i].data_part->name, + part_stat.sum_marks, + part_stat.approx_size_of_mark, settings.filesystem_prefetch_step_bytes, - part->prefetch_step_marks, - toString(part->ranges)); + part_stat.prefetch_step_marks, + toString(parts_ranges[i].ranges)); } const size_t min_marks_per_thread = (sum_marks - 1) / threads + 1; @@ -469,13 +427,24 @@ MergeTreePrefetchedReadPool::ThreadsTasks MergeTreePrefetchedReadPool::createThr size_t allowed_memory_usage = settings.filesystem_prefetch_max_memory_usage; if (!allowed_memory_usage) throw Exception(ErrorCodes::BAD_ARGUMENTS, "Setting `filesystem_prefetch_max_memory_usage` must be non-zero"); + std::optional allowed_prefetches_num = settings.filesystem_prefetches_limit ? std::optional(settings.filesystem_prefetches_limit) : std::nullopt; - ThreadsTasks result_threads_tasks; + per_thread_tasks.clear(); size_t total_tasks = 0; - for (size_t i = 0, part_idx = 0; i < threads && part_idx < parts_infos.size(); ++i) + + /// Make a copy to modify ranges. + std::vector per_part_ranges; + per_part_ranges.reserve(parts_ranges.size()); + for (const auto & part_with_ranges : parts_ranges) + { + auto & part_ranges = per_part_ranges.emplace_back(part_with_ranges.ranges); + std::sort(part_ranges.begin(), part_ranges.end()); + } + + for (size_t i = 0, part_idx = 0; i < threads && part_idx < per_part_infos.size(); ++i) { int64_t need_marks = min_marks_per_thread; @@ -486,119 +455,102 @@ MergeTreePrefetchedReadPool::ThreadsTasks MergeTreePrefetchedReadPool::createThr /// reads from pool which are from reader. Priority priority{reader_settings.read_settings.priority.value + 1}; - while (need_marks > 0 && part_idx < parts_infos.size()) + while (need_marks > 0 && part_idx < per_part_infos.size()) { - auto & part = *parts_infos[part_idx]; - size_t & marks_in_part = part.sum_marks; + auto & part_stat = per_part_statistics[part_idx]; + auto & part_ranges = per_part_ranges[part_idx]; - if (marks_in_part == 0) + if (part_stat.sum_marks == 0) { ++part_idx; continue; } MarkRanges ranges_to_get_from_part; - size_t marks_to_get_from_part = std::min(need_marks, marks_in_part); + size_t marks_to_get_from_part = std::min(need_marks, part_stat.sum_marks); /// Split by prefetch step even if !allow_prefetch below. Because it will allow /// to make a better distribution of tasks which did not fill into memory limit /// or prefetches limit through tasks stealing. - if (part.prefetch_step_marks) + if (part_stat.prefetch_step_marks) { - marks_to_get_from_part = std::min(marks_to_get_from_part, part.prefetch_step_marks); + marks_to_get_from_part = std::min(marks_to_get_from_part, part_stat.prefetch_step_marks); } - if (marks_in_part == marks_to_get_from_part) + if (part_stat.sum_marks == marks_to_get_from_part) { - ranges_to_get_from_part = part.ranges; + ranges_to_get_from_part = part_ranges; } else { - if (part.sum_marks < marks_to_get_from_part) + if (part_stat.sum_marks < marks_to_get_from_part) { throw Exception( ErrorCodes::LOGICAL_ERROR, "Requested {} marks from part {}, but part has only {} marks", - marks_to_get_from_part, part.data_part->name, part.sum_marks); + marks_to_get_from_part, per_part_infos[part_idx]->data_part->name, part_stat.sum_marks); } - size_t get_marks_num = marks_to_get_from_part; - while (get_marks_num > 0) + size_t num_marks_to_get = marks_to_get_from_part; + while (num_marks_to_get > 0) { - MarkRange & range = part.ranges.front(); + MarkRange & range = part_ranges.front(); const size_t marks_in_range = range.end - range.begin; - const size_t marks_to_get_from_range = std::min(marks_in_range, get_marks_num); - get_marks_num -= marks_to_get_from_range; + const size_t marks_to_get_from_range = std::min(marks_in_range, num_marks_to_get); + num_marks_to_get -= marks_to_get_from_range; ranges_to_get_from_part.emplace_back(range.begin, range.begin + marks_to_get_from_range); range.begin += marks_to_get_from_range; if (range.begin == range.end) { - part.ranges.pop_front(); + part_ranges.pop_front(); } - else if (!get_marks_num && part.prefetch_step_marks && range.end - range.begin < part.prefetch_step_marks) + else if (!num_marks_to_get && part_stat.prefetch_step_marks && range.end - range.begin < part_stat.prefetch_step_marks) { - /// We already have `get_marks_num` marks, but current mark range has + /// We already have `num_marks_to_get` marks, but current mark range has /// less than `prefetch_step_marks` marks, then add them too. ranges_to_get_from_part.emplace_back(range.begin, range.end); marks_to_get_from_part += range.end - range.begin; - part.ranges.pop_front(); + part_ranges.pop_front(); } } } need_marks -= marks_to_get_from_part; sum_marks -= marks_to_get_from_part; - marks_in_part -= marks_to_get_from_part; - - auto curr_task_size_predictor = !part.size_predictor ? nullptr - : std::make_unique(*part.size_predictor); /// make a copy - - auto read_task = std::make_unique( - part.data_part, - part.alter_conversions, - ranges_to_get_from_part, - part.part_index_in_query, - part.column_name_set, - part.task_columns, - std::move(curr_task_size_predictor)); - - read_task->priority = priority; + part_stat.sum_marks -= marks_to_get_from_part; bool allow_prefetch = false; if (allowed_memory_usage - && (allowed_prefetches_num.has_value() == false || allowed_prefetches_num.value() > 0)) + && (!allowed_prefetches_num.has_value() || allowed_prefetches_num.value() > 0)) { - allow_prefetch = part.estimated_memory_usage_for_single_prefetch <= allowed_memory_usage - && (allowed_prefetches_num.has_value() == false - || part.required_readers_num <= allowed_prefetches_num.value()); + allow_prefetch = part_stat.estimated_memory_usage_for_single_prefetch <= allowed_memory_usage + && (!allowed_prefetches_num.has_value() || part_stat.required_readers_num <= allowed_prefetches_num.value()); if (allow_prefetch) { - allowed_memory_usage -= part.estimated_memory_usage_for_single_prefetch; + allowed_memory_usage -= part_stat.estimated_memory_usage_for_single_prefetch; if (allowed_prefetches_num.has_value()) - *allowed_prefetches_num -= part.required_readers_num; + *allowed_prefetches_num -= part_stat.required_readers_num; } } + auto thread_task = std::make_unique(per_part_infos[part_idx], ranges_to_get_from_part, priority); if (allow_prefetch) - { - prefetch_queue.emplace(TaskHolder(read_task.get(), i)); - } - ++priority.value; + prefetch_queue.emplace(TaskHolder{thread_task.get(), i}); - result_threads_tasks[i].push_back(std::move(read_task)); + per_thread_tasks[i].push_back(std::move(thread_task)); + + ++priority.value; ++total_tasks; } } - LOG_TEST(log, "Result tasks {} for {} threads: {}", total_tasks, threads, dumpTasks(result_threads_tasks)); - - return result_threads_tasks; + LOG_TEST(log, "Result tasks {} for {} threads: {}", total_tasks, threads, dumpTasks(per_thread_tasks)); } -std::string MergeTreePrefetchedReadPool::dumpTasks(const ThreadsTasks & tasks) +std::string MergeTreePrefetchedReadPool::dumpTasks(const TasksPerThread & tasks) { WriteBufferFromOwnString result; for (const auto & [thread_id, thread_tasks] : tasks) @@ -611,9 +563,9 @@ std::string MergeTreePrefetchedReadPool::dumpTasks(const ThreadsTasks & tasks) { result << '\t'; result << ++no << ": "; - result << "reader: " << task->reader.valid() << ", "; - result << "part: " << task->data_part->name << ", "; - result << "ranges: " << toString(task->mark_ranges); + result << "reader future: " << task->readers_future.valid() << ", "; + result << "part: " << task->read_info->data_part->name << ", "; + result << "ranges: " << toString(task->ranges); } } } diff --git a/src/Storages/MergeTree/MergeTreePrefetchedReadPool.h b/src/Storages/MergeTree/MergeTreePrefetchedReadPool.h index ae97b32c7f9..7cee16c6ff2 100644 --- a/src/Storages/MergeTree/MergeTreePrefetchedReadPool.h +++ b/src/Storages/MergeTree/MergeTreePrefetchedReadPool.h @@ -1,9 +1,6 @@ #pragma once - +#include #include -#include -#include -#include #include #include #include @@ -16,99 +13,100 @@ using MergeTreeReaderPtr = std::unique_ptr; /// A class which is responsible for creating read tasks /// which are later taken by readers via getTask method. /// Does prefetching for the read tasks it creates. -class MergeTreePrefetchedReadPool : public IMergeTreeReadPool, private WithContext +class MergeTreePrefetchedReadPool : public MergeTreeReadPoolBase, private WithContext { public: MergeTreePrefetchedReadPool( - size_t threads, - size_t sum_marks_, - size_t min_marks_for_concurrent_read_, RangesInDataParts && parts_, const StorageSnapshotPtr & storage_snapshot_, const PrewhereInfoPtr & prewhere_info_, const ExpressionActionsSettings & actions_settings_, + const MergeTreeReaderSettings & reader_settings_, const Names & column_names_, const Names & virtual_column_names_, - size_t preferred_block_size_bytes_, - const MergeTreeReaderSettings & reader_settings_, - ContextPtr context_, - bool use_uncompressed_cache_, - bool is_remote_read_, - const MergeTreeSettings & storage_settings_); + const PoolSettings & settings_, + const ContextPtr & context_); - MergeTreeReadTaskPtr getTask(size_t thread) override; + ~MergeTreePrefetchedReadPool() override; + + String getName() const override { return "PrefetchedReadPool"; } + bool preservesOrderOfRanges() const override { return false; } + MergeTreeReadTaskPtr getTask(size_t task_idx, MergeTreeReadTask * previous_task) override; void profileFeedback(ReadBufferFromFileBase::ProfileInfo) override {} - Block getHeader() const override { return header; } - static bool checkReadMethodAllowed(LocalFSReadMethod method); static bool checkReadMethodAllowed(RemoteFSReadMethod method); private: - struct PartInfo; - using PartInfoPtr = std::shared_ptr; - using PartsInfos = std::vector; - using MergeTreeReadTaskPtr = std::unique_ptr; - using ThreadTasks = std::deque; - using ThreadsTasks = std::map; + struct PartStatistic + { + size_t sum_marks = 0; - std::future createPrefetchedReader( - const IMergeTreeDataPart & data_part, - const NamesAndTypesList & columns, - const AlterConversionsPtr & alter_conversions, - const MarkRanges & required_ranges, - Priority priority) const; + size_t approx_size_of_mark = 0; + size_t prefetch_step_marks = 0; - void createPrefetchedReaderForTask(MergeTreeReadTask & task) const; + size_t estimated_memory_usage_for_single_prefetch = 0; + size_t required_readers_num = 0; + }; - size_t getApproxSizeOfGranule(const IMergeTreeDataPart & part) const; + struct ReadersFuture + { + void wait(); + MergeTreeReadTask::Readers get(); + bool valid() const { return main.valid(); } - PartsInfos getPartsInfos(const RangesInDataParts & parts, size_t preferred_block_size_bytes) const; + std::future main; + std::vector> prewhere; + }; - ThreadsTasks createThreadsTasks( - size_t threads, - size_t sum_marks, - size_t min_marks_for_concurrent_read) const; + struct ThreadTask + { + using InfoPtr = MergeTreeReadTask::InfoPtr; - void startPrefetches() const; + ThreadTask(InfoPtr read_info_, MarkRanges ranges_, Priority priority_) + : read_info(std::move(read_info_)), ranges(std::move(ranges_)), priority(priority_) + { + } - static std::string dumpTasks(const ThreadsTasks & tasks); - - Poco::Logger * log; - - Block header; - MarkCache * mark_cache; - UncompressedCache * uncompressed_cache; - ReadBufferFromFileBase::ProfileCallback profile_callback; - size_t index_granularity_bytes; - size_t fixed_index_granularity; - - StorageSnapshotPtr storage_snapshot; - const Names column_names; - const Names virtual_column_names; - PrewhereInfoPtr prewhere_info; - const ExpressionActionsSettings actions_settings; - const MergeTreeReaderSettings reader_settings; - RangesInDataParts parts_ranges; - - [[ maybe_unused ]] const bool is_remote_read; - ThreadPool & prefetch_threadpool; - - PartsInfos parts_infos; - - ThreadsTasks threads_tasks; - std::mutex mutex; + InfoPtr read_info; + MarkRanges ranges; + Priority priority; + ReadersFuture readers_future; + }; struct TaskHolder { - explicit TaskHolder(MergeTreeReadTask * task_, size_t thread_id_) : task(task_), thread_id(thread_id_) {} - MergeTreeReadTask * task; - size_t thread_id; - bool operator <(const TaskHolder & other) const; + ThreadTask * task = nullptr; + size_t thread_id = 0; + bool operator<(const TaskHolder & other) const; }; - mutable std::priority_queue prefetch_queue; /// the smallest on top + + using ThreadTaskPtr = std::unique_ptr; + using ThreadTasks = std::deque; + using TasksPerThread = std::map; + using PartStatistics = std::vector; + + void fillPerPartStatistics(); + void fillPerThreadTasks(size_t threads, size_t sum_marks); + + void startPrefetches(); + void createPrefetchedReadersForTask(ThreadTask & task); + std::future createPrefetchedReader(MergeTreeReaderPtr reader, Priority priority); + + MergeTreeReadTaskPtr stealTask(size_t thread, MergeTreeReadTask * previous_task); + MergeTreeReadTaskPtr createTask(ThreadTask & thread_task, MergeTreeReadTask * previous_task); + + static std::string dumpTasks(const TasksPerThread & tasks); + + mutable std::mutex mutex; + ThreadPool & prefetch_threadpool; + + PartStatistics per_part_statistics; + TasksPerThread per_thread_tasks; + std::priority_queue prefetch_queue; /// the smallest on top bool started_prefetches = false; + Poco::Logger * log; /// A struct which allows to track max number of tasks which were in the /// threadpool simultaneously (similar to CurrentMetrics, but the result diff --git a/src/Storages/MergeTree/MergeTreeReadPool.cpp b/src/Storages/MergeTree/MergeTreeReadPool.cpp index 896769d9355..8ed7a9d8707 100644 --- a/src/Storages/MergeTree/MergeTreeReadPool.cpp +++ b/src/Storages/MergeTree/MergeTreeReadPool.cpp @@ -1,5 +1,6 @@ +#include "Storages/MergeTree/MergeTreeBlockReadUtils.h" +#include "Storages/MergeTree/MergeTreeReadTask.h" #include -#include #include #include #include @@ -32,9 +33,6 @@ size_t getApproxSizeOfPart(const IMergeTreeDataPart & part, const Names & column } MergeTreeReadPool::MergeTreeReadPool( - size_t threads_, - size_t sum_marks_, - size_t min_marks_for_concurrent_read_, RangesInDataParts && parts_, const StorageSnapshotPtr & storage_snapshot_, const PrewhereInfoPtr & prewhere_info_, @@ -42,29 +40,22 @@ MergeTreeReadPool::MergeTreeReadPool( const MergeTreeReaderSettings & reader_settings_, const Names & column_names_, const Names & virtual_column_names_, - ContextPtr context_, - bool do_not_steal_tasks_) - : storage_snapshot(storage_snapshot_) - , column_names(column_names_) - , virtual_column_names(virtual_column_names_) - , min_marks_for_concurrent_read(min_marks_for_concurrent_read_) - , prewhere_info(prewhere_info_) - , actions_settings(actions_settings_) - , reader_settings(reader_settings_) - , parts_ranges(std::move(parts_)) - , predict_block_size_bytes(context_->getSettingsRef().preferred_block_size_bytes > 0) - , do_not_steal_tasks(do_not_steal_tasks_) - , merge_tree_use_const_size_tasks_for_remote_reading(context_->getSettingsRef().merge_tree_use_const_size_tasks_for_remote_reading) + const PoolSettings & settings_, + const ContextPtr & context_) + : MergeTreeReadPoolBase( + std::move(parts_), + storage_snapshot_, + prewhere_info_, + actions_settings_, + reader_settings_, + column_names_, + virtual_column_names_, + settings_, + context_) + , min_marks_for_concurrent_read(pool_settings.min_marks_for_concurrent_read) , backoff_settings{context_->getSettingsRef()} - , backoff_state{threads_} + , backoff_state{pool_settings.threads} { - /// parts don't contain duplicate MergeTreeDataPart's. - const auto per_part_sum_marks = fillPerPartInfo( - parts_ranges, storage_snapshot, is_part_on_remote_disk, - predict_block_size_bytes, - column_names, virtual_column_names, prewhere_info, - actions_settings, reader_settings, per_part_params); - if (std::ranges::count(is_part_on_remote_disk, true)) { const auto & settings = context_->getSettingsRef(); @@ -76,6 +67,7 @@ MergeTreeReadPool::MergeTreeReadPool( const auto & columns = settings.merge_tree_determine_task_size_by_prewhere_columns && prewhere_info ? prewhere_info->prewhere_actions->getRequiredColumnsNames() : column_names_; + total_compressed_bytes += getApproxSizeOfPart(*part.data_part, columns); total_marks += part.getMarksCount(); } @@ -85,118 +77,60 @@ MergeTreeReadPool::MergeTreeReadPool( const auto min_bytes_per_task = settings.merge_tree_min_bytes_per_task_for_remote_reading; const auto avg_mark_bytes = std::max(total_compressed_bytes / total_marks, 1); /// We're taking min here because number of tasks shouldn't be too low - it will make task stealing impossible. - const auto heuristic_min_marks = std::min(total_marks / threads_, min_bytes_per_task / avg_mark_bytes); + const auto heuristic_min_marks = std::min(total_marks / pool_settings.threads, min_bytes_per_task / avg_mark_bytes); + if (heuristic_min_marks > min_marks_for_concurrent_read) - { min_marks_for_concurrent_read = heuristic_min_marks; - } } } - fillPerThreadInfo(threads_, sum_marks_, per_part_sum_marks, parts_ranges); + fillPerThreadInfo(pool_settings.threads, pool_settings.sum_marks); } -std::vector MergeTreeReadPool::fillPerPartInfo( - const RangesInDataParts & parts, - const StorageSnapshotPtr & storage_snapshot, - std::vector & is_part_on_remote_disk, - bool & predict_block_size_bytes, - const Names & column_names, - const Names & virtual_column_names, - const PrewhereInfoPtr & prewhere_info, - const ExpressionActionsSettings & actions_settings, - const MergeTreeReaderSettings & reader_settings, - std::vector & per_part_params) -{ - std::vector per_part_sum_marks; - Block sample_block = storage_snapshot->metadata->getSampleBlock(); - is_part_on_remote_disk.resize(parts.size()); - - for (const auto i : collections::range(0, parts.size())) - { - const auto & part = parts[i]; -#ifndef NDEBUG - assertSortedAndNonIntersecting(part.ranges); -#endif - - bool part_on_remote_disk = part.data_part->isStoredOnRemoteDisk(); - is_part_on_remote_disk[i] = part_on_remote_disk; - - /// Read marks for every data part. - size_t sum_marks = 0; - for (const auto & range : part.ranges) - sum_marks += range.end - range.begin; - - per_part_sum_marks.push_back(sum_marks); - - auto & per_part = per_part_params.emplace_back(); - per_part.data_part = part; - - LoadedMergeTreeDataPartInfoForReader part_info(part.data_part, part.alter_conversions); - auto task_columns = getReadTaskColumns( - part_info, storage_snapshot, column_names, virtual_column_names, - prewhere_info, actions_settings, - reader_settings, /*with_subcolumns=*/ true); - - auto size_predictor = !predict_block_size_bytes ? nullptr - : IMergeTreeSelectAlgorithm::getSizePredictor(part.data_part, task_columns, sample_block); - - per_part.size_predictor = std::move(size_predictor); - - /// will be used to distinguish between PREWHERE and WHERE columns when applying filter - const auto & required_column_names = task_columns.columns.getNames(); - per_part.column_name_set = {required_column_names.begin(), required_column_names.end()}; - per_part.task_columns = std::move(task_columns); - } - - return per_part_sum_marks; -} - -MergeTreeReadTaskPtr MergeTreeReadPool::getTask(size_t thread) +MergeTreeReadTaskPtr MergeTreeReadPool::getTask(size_t task_idx, MergeTreeReadTask * previous_task) { const std::lock_guard lock{mutex}; /// If number of threads was lowered due to backoff, then will assign work only for maximum 'backoff_state.current_threads' threads. - if (thread >= backoff_state.current_threads) + if (task_idx >= backoff_state.current_threads) return nullptr; if (remaining_thread_tasks.empty()) return nullptr; - const auto tasks_remaining_for_this_thread = !threads_tasks[thread].sum_marks_in_parts.empty(); - if (!tasks_remaining_for_this_thread && do_not_steal_tasks) + const auto tasks_remaining_for_this_thread = !threads_tasks[task_idx].sum_marks_in_parts.empty(); + if (!tasks_remaining_for_this_thread && pool_settings.do_not_steal_tasks) return nullptr; /// Steal task if nothing to do and it's not prohibited - auto thread_idx = thread; + auto thread_idx = task_idx; if (!tasks_remaining_for_this_thread) { auto it = remaining_thread_tasks.lower_bound(backoff_state.current_threads); // Grab the entire tasks of a thread which is killed by backoff if (it != remaining_thread_tasks.end()) { - threads_tasks[thread] = std::move(threads_tasks[*it]); + threads_tasks[task_idx] = std::move(threads_tasks[*it]); remaining_thread_tasks.erase(it); - remaining_thread_tasks.insert(thread); + remaining_thread_tasks.insert(task_idx); } else // Try steal tasks from the next thread { - it = remaining_thread_tasks.upper_bound(thread); + it = remaining_thread_tasks.upper_bound(task_idx); if (it == remaining_thread_tasks.end()) it = remaining_thread_tasks.begin(); thread_idx = *it; } } + auto & thread_tasks = threads_tasks[thread_idx]; - auto & thread_task = thread_tasks.parts_and_ranges.back(); - const auto part_idx = thread_task.part_idx; - auto & part = per_part_params[part_idx].data_part; + const auto part_idx = thread_task.part_idx; auto & marks_in_part = thread_tasks.sum_marks_in_parts.back(); size_t need_marks; - if (is_part_on_remote_disk[part_idx] && !merge_tree_use_const_size_tasks_for_remote_reading) + if (is_part_on_remote_disk[part_idx] && !pool_settings.use_const_size_tasks_for_remote_reading) need_marks = marks_in_part; else /// Get whole part to read if it is small enough. need_marks = std::min(marks_in_part, min_marks_for_concurrent_read); @@ -239,28 +173,12 @@ MergeTreeReadTaskPtr MergeTreeReadPool::getTask(size_t thread) } } - const auto & per_part = per_part_params[part_idx]; - auto curr_task_size_predictor = !per_part.size_predictor ? nullptr - : std::make_unique(*per_part.size_predictor); /// make a copy - - return std::make_unique( - part.data_part, - part.alter_conversions, - ranges_to_get_from_part, - part.part_index_in_query, - per_part.column_name_set, - per_part.task_columns, - std::move(curr_task_size_predictor)); -} - -Block MergeTreeReadPool::getHeader() const -{ - return storage_snapshot->getSampleBlockForColumns(column_names); + return createTask(per_part_infos[part_idx], std::move(ranges_to_get_from_part), previous_task); } void MergeTreeReadPool::profileFeedback(ReadBufferFromFileBase::ProfileInfo info) { - if (backoff_settings.min_read_latency_ms == 0 || do_not_steal_tasks) + if (backoff_settings.min_read_latency_ms == 0 || pool_settings.do_not_steal_tasks) return; if (info.nanoseconds < backoff_settings.min_read_latency_ms * 1000000) @@ -297,13 +215,10 @@ void MergeTreeReadPool::profileFeedback(ReadBufferFromFileBase::ProfileInfo info LOG_DEBUG(log, "Will lower number of threads to {}", backoff_state.current_threads); } - -void MergeTreeReadPool::fillPerThreadInfo( - size_t threads, size_t sum_marks, std::vector per_part_sum_marks, - const RangesInDataParts & parts) +void MergeTreeReadPool::fillPerThreadInfo(size_t threads, size_t sum_marks) { threads_tasks.resize(threads); - if (parts.empty()) + if (parts_ranges.empty()) return; struct PartInfo @@ -316,17 +231,19 @@ void MergeTreeReadPool::fillPerThreadInfo( using PartsInfo = std::vector; std::queue parts_queue; + auto per_part_sum_marks = getPerPartSumMarks(); + { /// Group parts by disk name. /// We try minimize the number of threads concurrently read from the same disk. /// It improves the performance for JBOD architecture. std::map> parts_per_disk; - for (size_t i = 0; i < parts.size(); ++i) + for (size_t i = 0; i < parts_ranges.size(); ++i) { - PartInfo part_info{parts[i], per_part_sum_marks[i], i}; - if (parts[i].data_part->isStoredOnDisk()) - parts_per_disk[parts[i].data_part->getDataPartStorage().getDiskName()].push_back(std::move(part_info)); + PartInfo part_info{parts_ranges[i], per_part_sum_marks[i], i}; + if (parts_ranges[i].data_part->isStoredOnDisk()) + parts_per_disk[parts_ranges[i].data_part->getDataPartStorage().getDiskName()].push_back(std::move(part_info)); else parts_per_disk[""].push_back(std::move(part_info)); } @@ -346,7 +263,7 @@ void MergeTreeReadPool::fillPerThreadInfo( while (need_marks > 0 && !parts_queue.empty()) { auto & current_parts = parts_queue.front(); - RangesInDataPart & part = current_parts.back().part; + auto & part_with_ranges = current_parts.back().part; size_t & marks_in_part = current_parts.back().sum_marks; const auto part_idx = current_parts.back().part_idx; @@ -366,7 +283,7 @@ void MergeTreeReadPool::fillPerThreadInfo( /// Get whole part to read if it is small enough. if (marks_in_part <= need_marks) { - ranges_to_get_from_part = part.ranges; + ranges_to_get_from_part = part_with_ranges.ranges; marks_in_ranges = marks_in_part; need_marks -= marks_in_part; @@ -379,10 +296,10 @@ void MergeTreeReadPool::fillPerThreadInfo( /// Loop through part ranges. while (need_marks > 0) { - if (part.ranges.empty()) + if (part_with_ranges.ranges.empty()) throw Exception(ErrorCodes::LOGICAL_ERROR, "Unexpected end of ranges while spreading marks among threads"); - MarkRange & range = part.ranges.front(); + MarkRange & range = part_with_ranges.ranges.front(); const size_t marks_in_range = range.end - range.begin; const size_t marks_to_get_from_range = std::min(marks_in_range, need_marks); @@ -392,11 +309,11 @@ void MergeTreeReadPool::fillPerThreadInfo( marks_in_part -= marks_to_get_from_range; need_marks -= marks_to_get_from_range; if (range.begin == range.end) - part.ranges.pop_front(); + part_with_ranges.ranges.pop_front(); } } - threads_tasks[i].parts_and_ranges.push_back({ part_idx, ranges_to_get_from_part }); + threads_tasks[i].parts_and_ranges.push_back({part_idx, ranges_to_get_from_part}); threads_tasks[i].sum_marks_in_parts.push_back(marks_in_ranges); if (marks_in_ranges != 0) remaining_thread_tasks.insert(i); @@ -415,152 +332,4 @@ void MergeTreeReadPool::fillPerThreadInfo( } } - -MergeTreeReadPoolParallelReplicas::~MergeTreeReadPoolParallelReplicas() = default; - - -Block MergeTreeReadPoolParallelReplicas::getHeader() const -{ - return storage_snapshot->getSampleBlockForColumns(extension.columns_to_read); -} - -MergeTreeReadTaskPtr MergeTreeReadPoolParallelReplicas::getTask(size_t thread) -{ - /// This parameter is needed only to satisfy the interface - UNUSED(thread); - - std::lock_guard lock(mutex); - - if (no_more_tasks_available) - return nullptr; - - if (buffered_ranges.empty()) - { - auto result = extension.callback(ParallelReadRequest( - CoordinationMode::Default, - extension.number_of_current_replica, - min_marks_for_concurrent_read * threads, - /// For Default coordination mode we don't need to pass part names. - RangesInDataPartsDescription{})); - - if (!result || result->finish) - { - no_more_tasks_available = true; - return nullptr; - } - - buffered_ranges = std::move(result->description); - } - - if (buffered_ranges.empty()) - throw Exception(ErrorCodes::LOGICAL_ERROR, "No tasks to read. This is a bug"); - - auto & current_task = buffered_ranges.front(); - - RangesInDataPart part; - size_t part_idx = 0; - for (size_t index = 0; index < per_part_params.size(); ++index) - { - auto & other_part = per_part_params[index]; - if (other_part.data_part.data_part->info == current_task.info) - { - part = other_part.data_part; - part_idx = index; - break; - } - } - - MarkRanges ranges_to_read; - size_t current_sum_marks = 0; - while (current_sum_marks < min_marks_for_concurrent_read && !current_task.ranges.empty()) - { - auto diff = min_marks_for_concurrent_read - current_sum_marks; - auto range = current_task.ranges.front(); - if (range.getNumberOfMarks() > diff) - { - auto new_range = range; - new_range.end = range.begin + diff; - range.begin += diff; - - current_task.ranges.front() = range; - ranges_to_read.push_back(new_range); - current_sum_marks += new_range.getNumberOfMarks(); - continue; - } - - ranges_to_read.push_back(range); - current_sum_marks += range.getNumberOfMarks(); - current_task.ranges.pop_front(); - } - - if (current_task.ranges.empty()) - buffered_ranges.pop_front(); - - const auto & per_part = per_part_params[part_idx]; - - auto curr_task_size_predictor - = !per_part.size_predictor ? nullptr : std::make_unique(*per_part.size_predictor); /// make a copy - - return std::make_unique( - part.data_part, - part.alter_conversions, - ranges_to_read, - part.part_index_in_query, - per_part.column_name_set, - per_part.task_columns, - std::move(curr_task_size_predictor)); -} - - -MarkRanges MergeTreeInOrderReadPoolParallelReplicas::getNewTask(RangesInDataPartDescription description) -{ - std::lock_guard lock(mutex); - - auto get_from_buffer = [&]() -> std::optional - { - for (auto & desc : buffered_tasks) - { - if (desc.info == description.info && !desc.ranges.empty()) - { - auto result = std::move(desc.ranges); - desc.ranges = MarkRanges{}; - return result; - } - } - return std::nullopt; - }; - - if (auto result = get_from_buffer(); result) - return result.value(); - - if (no_more_tasks) - return {}; - - auto response = extension.callback(ParallelReadRequest( - mode, - extension.number_of_current_replica, - min_marks_for_concurrent_read * request.size(), - request - )); - - if (!response || response->description.empty() || response->finish) - { - no_more_tasks = true; - return {}; - } - - /// Fill the buffer - for (size_t i = 0; i < request.size(); ++i) - { - auto & new_ranges = response->description[i].ranges; - auto & old_ranges = buffered_tasks[i].ranges; - std::move(new_ranges.begin(), new_ranges.end(), std::back_inserter(old_ranges)); - } - - if (auto result = get_from_buffer(); result) - return result.value(); - - return {}; -} - } diff --git a/src/Storages/MergeTree/MergeTreeReadPool.h b/src/Storages/MergeTree/MergeTreeReadPool.h index 68d5438cb3d..3a1af947cae 100644 --- a/src/Storages/MergeTree/MergeTreeReadPool.h +++ b/src/Storages/MergeTree/MergeTreeReadPool.h @@ -1,38 +1,30 @@ #pragma once - +#include #include -#include -#include #include #include #include -#include #include #include -#include - #include namespace DB { -/** Provides read tasks for MergeTreeThreadSelectProcessor`s in fine-grained batches, allowing for more +/** Provides read tasks for MergeTreeThreadSelectAlgorithm in fine-grained batches, allowing for more * uniform distribution of work amongst multiple threads. All parts and their ranges are divided into `threads` * workloads with at most `sum_marks / threads` marks. Then, threads are performing reads from these workloads * in "sequential" manner, requesting work in small batches. As soon as some thread has exhausted * it's workload, it either is signaled that no more work is available (`do_not_steal_tasks == false`) or * continues taking small batches from other threads' workloads (`do_not_steal_tasks == true`). */ -class MergeTreeReadPool : public IMergeTreeReadPool +class MergeTreeReadPool : public MergeTreeReadPoolBase { public: struct BackoffSettings; MergeTreeReadPool( - size_t threads_, - size_t sum_marks_, - size_t min_marks_for_concurrent_read_, RangesInDataParts && parts_, const StorageSnapshotPtr & storage_snapshot_, const PrewhereInfoPtr & prewhere_info_, @@ -40,12 +32,14 @@ public: const MergeTreeReaderSettings & reader_settings_, const Names & column_names_, const Names & virtual_column_names_, - ContextPtr context_, - bool do_not_steal_tasks_ = false); + const PoolSettings & settings_, + const ContextPtr & context_); ~MergeTreeReadPool() override = default; - MergeTreeReadTaskPtr getTask(size_t thread) override; + String getName() const override { return "ReadPool"; } + bool preservesOrderOfRanges() const override { return false; } + MergeTreeReadTaskPtr getTask(size_t task_idx, MergeTreeReadTask * previous_task) override; /** Each worker could call this method and pass information about read performance. * If read performance is too low, pool could decide to lower number of threads: do not assign more tasks to several threads. @@ -53,8 +47,6 @@ public: */ void profileFeedback(ReadBufferFromFileBase::ProfileInfo info) override; - Block getHeader() const override; - /** Pull could dynamically lower (backoff) number of threads, if read operation are too slow. * Settings for that backoff. */ @@ -82,50 +74,12 @@ public: BackoffSettings() : min_read_latency_ms(0) {} }; - struct PerPartParams - { - MergeTreeReadTaskColumns task_columns; - NameSet column_name_set; - MergeTreeBlockSizePredictorPtr size_predictor; - RangesInDataPart data_part; - }; - - static std::vector fillPerPartInfo( - const RangesInDataParts & parts, - const StorageSnapshotPtr & storage_snapshot, - std::vector & is_part_on_remote_disk, - bool & predict_block_size_bytes, - const Names & column_names, - const Names & virtual_column_names, - const PrewhereInfoPtr & prewhere_info, - const ExpressionActionsSettings & actions_settings_, - const MergeTreeReaderSettings & reader_settings_, - std::vector & per_part_params); - private: - void fillPerThreadInfo( - size_t threads, size_t sum_marks, std::vector per_part_sum_marks, - const RangesInDataParts & parts); - - /// Initialized in constructor - StorageSnapshotPtr storage_snapshot; - const Names column_names; - const Names virtual_column_names; - size_t min_marks_for_concurrent_read{0}; - PrewhereInfoPtr prewhere_info; - ExpressionActionsSettings actions_settings; - MergeTreeReaderSettings reader_settings; - RangesInDataParts parts_ranges; - bool predict_block_size_bytes; - bool do_not_steal_tasks; - bool merge_tree_use_const_size_tasks_for_remote_reading = false; - - std::vector per_part_params; - std::vector is_part_on_remote_disk; - - BackoffSettings backoff_settings; + void fillPerThreadInfo(size_t threads, size_t sum_marks); mutable std::mutex mutex; + size_t min_marks_for_concurrent_read = 0; + /// State to track numbers of slow reads. struct BackoffState { @@ -135,16 +89,10 @@ private: explicit BackoffState(size_t threads) : current_threads(threads) {} }; + + const BackoffSettings backoff_settings; BackoffState backoff_state; - struct Part - { - MergeTreeData::DataPartPtr data_part; - size_t part_index_in_query; - }; - - std::vector parts_with_idx; - struct ThreadTask { struct PartIndexAndRange @@ -159,123 +107,8 @@ private: std::vector threads_tasks; std::set remaining_thread_tasks; + Poco::Logger * log = &Poco::Logger::get("MergeTreeReadPool"); - }; -class MergeTreeReadPoolParallelReplicas : public IMergeTreeReadPool -{ -public: - MergeTreeReadPoolParallelReplicas( - StorageSnapshotPtr storage_snapshot_, - size_t threads_, - ParallelReadingExtension extension_, - const RangesInDataParts & parts_, - const PrewhereInfoPtr & prewhere_info_, - const ExpressionActionsSettings & actions_settings_, - const MergeTreeReaderSettings & reader_settings_, - const Names & column_names_, - const Names & virtual_column_names_, - size_t min_marks_for_concurrent_read_) - : extension(extension_) - , threads(threads_) - , prewhere_info(prewhere_info_) - , actions_settings(actions_settings_) - , reader_settings(reader_settings_) - , storage_snapshot(storage_snapshot_) - , min_marks_for_concurrent_read(min_marks_for_concurrent_read_) - , column_names(column_names_) - , virtual_column_names(virtual_column_names_) - , parts_ranges(std::move(parts_)) - { - MergeTreeReadPool::fillPerPartInfo( - parts_ranges, storage_snapshot, is_part_on_remote_disk, - predict_block_size_bytes, column_names, virtual_column_names, prewhere_info, - actions_settings, reader_settings, per_part_params); - - extension.all_callback(InitialAllRangesAnnouncement( - CoordinationMode::Default, - parts_ranges.getDescriptions(), - extension.number_of_current_replica - )); - } - - ~MergeTreeReadPoolParallelReplicas() override; - - Block getHeader() const override; - - MergeTreeReadTaskPtr getTask(size_t thread) override; - - void profileFeedback(ReadBufferFromFileBase::ProfileInfo) override {} - -private: - ParallelReadingExtension extension; - - RangesInDataPartsDescription buffered_ranges; - size_t threads; - bool no_more_tasks_available{false}; - Poco::Logger * log = &Poco::Logger::get("MergeTreeReadPoolParallelReplicas"); - - std::mutex mutex; - - PrewhereInfoPtr prewhere_info; - ExpressionActionsSettings actions_settings; - MergeTreeReaderSettings reader_settings; - StorageSnapshotPtr storage_snapshot; - size_t min_marks_for_concurrent_read; - const Names column_names; - const Names virtual_column_names; - RangesInDataParts parts_ranges; - - bool predict_block_size_bytes = false; - std::vector is_part_on_remote_disk; - std::vector per_part_params; -}; - -using MergeTreeReadPoolParallelReplicasPtr = std::shared_ptr; - - -class MergeTreeInOrderReadPoolParallelReplicas : private boost::noncopyable -{ -public: - MergeTreeInOrderReadPoolParallelReplicas( - RangesInDataParts parts_, - ParallelReadingExtension extension_, - CoordinationMode mode_, - size_t min_marks_for_concurrent_read_) - : parts_ranges(parts_) - , extension(extension_) - , mode(mode_) - , min_marks_for_concurrent_read(min_marks_for_concurrent_read_) - { - for (const auto & part : parts_ranges) - request.push_back({part.data_part->info, MarkRanges{}}); - - for (const auto & part : parts_ranges) - buffered_tasks.push_back({part.data_part->info, MarkRanges{}}); - - extension.all_callback(InitialAllRangesAnnouncement( - mode, - parts_ranges.getDescriptions(), - extension.number_of_current_replica - )); - } - - MarkRanges getNewTask(RangesInDataPartDescription description); - - - RangesInDataParts parts_ranges; - ParallelReadingExtension extension; - CoordinationMode mode; - size_t min_marks_for_concurrent_read{0}; - - bool no_more_tasks{false}; - RangesInDataPartsDescription request; - RangesInDataPartsDescription buffered_tasks; - - std::mutex mutex; -}; - -using MergeTreeInOrderReadPoolParallelReplicasPtr = std::shared_ptr; - } diff --git a/src/Storages/MergeTree/MergeTreeReadPoolBase.cpp b/src/Storages/MergeTree/MergeTreeReadPoolBase.cpp new file mode 100644 index 00000000000..446baccd961 --- /dev/null +++ b/src/Storages/MergeTree/MergeTreeReadPoolBase.cpp @@ -0,0 +1,149 @@ +#include +#include +#include + +namespace DB +{ + +MergeTreeReadPoolBase::MergeTreeReadPoolBase( + RangesInDataParts && parts_, + const StorageSnapshotPtr & storage_snapshot_, + const PrewhereInfoPtr & prewhere_info_, + const ExpressionActionsSettings & actions_settings_, + const MergeTreeReaderSettings & reader_settings_, + const Names & column_names_, + const Names & virtual_column_names_, + const PoolSettings & pool_settings_, + const ContextPtr & context_) + : parts_ranges(std::move(parts_)) + , storage_snapshot(storage_snapshot_) + , prewhere_info(prewhere_info_) + , actions_settings(actions_settings_) + , reader_settings(reader_settings_) + , column_names(column_names_) + , virtual_column_names(virtual_column_names_) + , pool_settings(pool_settings_) + , owned_mark_cache(context_->getGlobalContext()->getMarkCache()) + , owned_uncompressed_cache(pool_settings_.use_uncompressed_cache ? context_->getGlobalContext()->getUncompressedCache() : nullptr) + , header(storage_snapshot->getSampleBlockForColumns(column_names)) + , profile_callback([this](ReadBufferFromFileBase::ProfileInfo info_) { profileFeedback(info_); }) +{ + fillPerPartInfos(); +} + +void MergeTreeReadPoolBase::fillPerPartInfos() +{ + per_part_infos.reserve(parts_ranges.size()); + is_part_on_remote_disk.reserve(parts_ranges.size()); + + auto sample_block = storage_snapshot->metadata->getSampleBlock(); + + for (const auto & part_with_ranges : parts_ranges) + { +#ifndef NDEBUG + assertSortedAndNonIntersecting(part_with_ranges.ranges); +#endif + + MergeTreeReadTask::Info read_task_info; + + read_task_info.data_part = part_with_ranges.data_part; + read_task_info.part_index_in_query = part_with_ranges.part_index_in_query; + read_task_info.alter_conversions = part_with_ranges.alter_conversions; + + LoadedMergeTreeDataPartInfoForReader part_info(part_with_ranges.data_part, part_with_ranges.alter_conversions); + + read_task_info.task_columns = getReadTaskColumns( + part_info, storage_snapshot, column_names, virtual_column_names, + prewhere_info, actions_settings, + reader_settings, /*with_subcolumns=*/ true); + + if (pool_settings.preferred_block_size_bytes > 0) + { + const auto & result_column_names = read_task_info.task_columns.columns.getNames(); + NameSet all_column_names(result_column_names.begin(), result_column_names.end()); + + for (const auto & pre_columns_per_step : read_task_info.task_columns.pre_columns) + { + const auto & pre_column_names = pre_columns_per_step.getNames(); + all_column_names.insert(pre_column_names.begin(), pre_column_names.end()); + } + + read_task_info.shared_size_predictor = std::make_unique( + read_task_info.data_part, + Names(all_column_names.begin(), all_column_names.end()), + sample_block); + } + + is_part_on_remote_disk.push_back(part_with_ranges.data_part->isStoredOnRemoteDisk()); + per_part_infos.push_back(std::make_shared(std::move(read_task_info))); + } +} + +std::vector MergeTreeReadPoolBase::getPerPartSumMarks() const +{ + std::vector per_part_sum_marks; + per_part_sum_marks.reserve(parts_ranges.size()); + + for (const auto & part_with_ranges : parts_ranges) + { + size_t sum_marks = 0; + for (const auto & range : part_with_ranges.ranges) + sum_marks += range.end - range.begin; + + per_part_sum_marks.push_back(sum_marks); + } + + return per_part_sum_marks; +} + +MergeTreeReadTaskPtr MergeTreeReadPoolBase::createTask( + MergeTreeReadTask::InfoPtr read_info, + MarkRanges ranges, + MergeTreeReadTask * previous_task) const +{ + auto task_size_predictor = read_info->shared_size_predictor + ? std::make_unique(*read_info->shared_size_predictor) + : nullptr; /// make a copy + + auto get_part_name = [](const auto & task_info) -> const String & + { + return task_info.data_part->isProjectionPart() ? task_info.data_part->getParentPart()->name : task_info.data_part->name; + }; + + auto extras = getExtras(); + MergeTreeReadTask::Readers task_readers; + + if (!previous_task) + { + task_readers = MergeTreeReadTask::createReaders(read_info, extras, ranges); + } + else if (get_part_name(previous_task->getInfo()) != get_part_name(*read_info)) + { + extras.value_size_map = previous_task->getMainReader().getAvgValueSizeHints(); + task_readers = MergeTreeReadTask::createReaders(read_info, extras, ranges); + } + else + { + task_readers = previous_task->releaseReaders(); + } + + return std::make_unique( + read_info, + std::move(task_readers), + std::move(ranges), + std::move(task_size_predictor)); +} + +MergeTreeReadTask::Extras MergeTreeReadPoolBase::getExtras() const +{ + return + { + .uncompressed_cache = owned_uncompressed_cache.get(), + .mark_cache = owned_mark_cache.get(), + .reader_settings = reader_settings, + .storage_snapshot = storage_snapshot, + .profile_callback = profile_callback, + }; +} + +} diff --git a/src/Storages/MergeTree/MergeTreeReadPoolBase.h b/src/Storages/MergeTree/MergeTreeReadPoolBase.h new file mode 100644 index 00000000000..0081063cd37 --- /dev/null +++ b/src/Storages/MergeTree/MergeTreeReadPoolBase.h @@ -0,0 +1,67 @@ +#pragma once +#include +#include +#include + +namespace DB +{ + +class MergeTreeReadPoolBase : public IMergeTreeReadPool +{ +public: + struct PoolSettings + { + size_t threads = 0; + size_t sum_marks = 0; + size_t min_marks_for_concurrent_read = 0; + size_t preferred_block_size_bytes = 0; + + bool use_uncompressed_cache = false; + bool do_not_steal_tasks = false; + bool use_const_size_tasks_for_remote_reading = false; + }; + + MergeTreeReadPoolBase( + RangesInDataParts && parts_, + const StorageSnapshotPtr & storage_snapshot_, + const PrewhereInfoPtr & prewhere_info_, + const ExpressionActionsSettings & actions_settings_, + const MergeTreeReaderSettings & reader_settings_, + const Names & column_names_, + const Names & virtual_column_names_, + const PoolSettings & settings_, + const ContextPtr & context_); + + Block getHeader() const override { return header; } + +protected: + /// Initialized in constructor + const RangesInDataParts parts_ranges; + const StorageSnapshotPtr storage_snapshot; + const PrewhereInfoPtr prewhere_info; + const ExpressionActionsSettings actions_settings; + const MergeTreeReaderSettings reader_settings; + const Names column_names; + const Names virtual_column_names; + const PoolSettings pool_settings; + const MarkCachePtr owned_mark_cache; + const UncompressedCachePtr owned_uncompressed_cache; + const Block header; + + void fillPerPartInfos(); + std::vector getPerPartSumMarks() const; + + MergeTreeReadTaskPtr createTask( + MergeTreeReadTask::InfoPtr read_info, + MarkRanges ranges, + MergeTreeReadTask * previous_task) const; + + MergeTreeReadTask::Extras getExtras() const; + + std::vector per_part_infos; + std::vector is_part_on_remote_disk; + + ReadBufferFromFileBase::ProfileCallback profile_callback; +}; + +} diff --git a/src/Storages/MergeTree/MergeTreeReadPoolInOrder.cpp b/src/Storages/MergeTree/MergeTreeReadPoolInOrder.cpp new file mode 100644 index 00000000000..37f009beb87 --- /dev/null +++ b/src/Storages/MergeTree/MergeTreeReadPoolInOrder.cpp @@ -0,0 +1,74 @@ +#include + +namespace DB +{ + +namespace ErrorCodes +{ + extern const int LOGICAL_ERROR; +} + +MergeTreeReadPoolInOrder::MergeTreeReadPoolInOrder( + bool has_limit_below_one_block_, + MergeTreeReadType read_type_, + RangesInDataParts parts_, + const StorageSnapshotPtr & storage_snapshot_, + const PrewhereInfoPtr & prewhere_info_, + const ExpressionActionsSettings & actions_settings_, + const MergeTreeReaderSettings & reader_settings_, + const Names & column_names_, + const Names & virtual_column_names_, + const PoolSettings & settings_, + const ContextPtr & context_) + : MergeTreeReadPoolBase( + std::move(parts_), + storage_snapshot_, + prewhere_info_, + actions_settings_, + reader_settings_, + column_names_, + virtual_column_names_, + settings_, + context_) + , has_limit_below_one_block(has_limit_below_one_block_) + , read_type(read_type_) +{ + per_part_mark_ranges.reserve(parts_ranges.size()); + for (const auto & part_with_ranges : parts_ranges) + per_part_mark_ranges.push_back(part_with_ranges.ranges); +} + +MergeTreeReadTaskPtr MergeTreeReadPoolInOrder::getTask(size_t task_idx, MergeTreeReadTask * previous_task) +{ + if (task_idx >= per_part_infos.size()) + throw Exception(ErrorCodes::LOGICAL_ERROR, + "Requested task with idx {}, but there are only {} parts", + task_idx, per_part_infos.size()); + + auto & all_mark_ranges = per_part_mark_ranges[task_idx]; + if (all_mark_ranges.empty()) + return nullptr; + + MarkRanges mark_ranges_for_task; + if (read_type == MergeTreeReadType::InReverseOrder) + { + /// Read ranges from right to left. + mark_ranges_for_task.emplace_back(std::move(all_mark_ranges.back())); + all_mark_ranges.pop_back(); + } + else if (has_limit_below_one_block) + { + /// If we need to read few rows, set one range per task to reduce number of read data. + mark_ranges_for_task.emplace_back(std::move(all_mark_ranges.front())); + all_mark_ranges.pop_front(); + } + else + { + mark_ranges_for_task = std::move(all_mark_ranges); + all_mark_ranges.clear(); + } + + return createTask(per_part_infos[task_idx], std::move(mark_ranges_for_task), previous_task); +} + +} diff --git a/src/Storages/MergeTree/MergeTreeReadPoolInOrder.h b/src/Storages/MergeTree/MergeTreeReadPoolInOrder.h new file mode 100644 index 00000000000..d9cc1ba4984 --- /dev/null +++ b/src/Storages/MergeTree/MergeTreeReadPoolInOrder.h @@ -0,0 +1,35 @@ +#pragma once +#include + +namespace DB +{ + +class MergeTreeReadPoolInOrder : public MergeTreeReadPoolBase +{ +public: + MergeTreeReadPoolInOrder( + bool has_limit_below_one_block_, + MergeTreeReadType read_type_, + RangesInDataParts parts_, + const StorageSnapshotPtr & storage_snapshot_, + const PrewhereInfoPtr & prewhere_info_, + const ExpressionActionsSettings & actions_settings_, + const MergeTreeReaderSettings & reader_settings_, + const Names & column_names_, + const Names & virtual_column_names_, + const PoolSettings & settings_, + const ContextPtr & context_); + + String getName() const override { return "ReadPoolInOrder"; } + bool preservesOrderOfRanges() const override { return true; } + MergeTreeReadTaskPtr getTask(size_t task_idx, MergeTreeReadTask * previous_task) override; + void profileFeedback(ReadBufferFromFileBase::ProfileInfo) override {} + +private: + const bool has_limit_below_one_block; + const MergeTreeReadType read_type; + + std::vector per_part_mark_ranges; +}; + +} diff --git a/src/Storages/MergeTree/MergeTreeReadPoolParallelReplicas.cpp b/src/Storages/MergeTree/MergeTreeReadPoolParallelReplicas.cpp new file mode 100644 index 00000000000..e61ddf0d122 --- /dev/null +++ b/src/Storages/MergeTree/MergeTreeReadPoolParallelReplicas.cpp @@ -0,0 +1,110 @@ +#include + +namespace DB +{ + +namespace ErrorCodes +{ + extern const int LOGICAL_ERROR; +} + +MergeTreeReadPoolParallelReplicas::MergeTreeReadPoolParallelReplicas( + ParallelReadingExtension extension_, + RangesInDataParts && parts_, + const StorageSnapshotPtr & storage_snapshot_, + const PrewhereInfoPtr & prewhere_info_, + const ExpressionActionsSettings & actions_settings_, + const MergeTreeReaderSettings & reader_settings_, + const Names & column_names_, + const Names & virtual_column_names_, + const PoolSettings & settings_, + const ContextPtr & context_) + : MergeTreeReadPoolBase( + std::move(parts_), + storage_snapshot_, + prewhere_info_, + actions_settings_, + reader_settings_, + column_names_, + virtual_column_names_, + settings_, + context_) + , extension(std::move(extension_)) +{ + extension.all_callback(InitialAllRangesAnnouncement( + CoordinationMode::Default, + parts_ranges.getDescriptions(), + extension.number_of_current_replica + )); +} + +MergeTreeReadTaskPtr MergeTreeReadPoolParallelReplicas::getTask(size_t /*task_idx*/, MergeTreeReadTask * previous_task) +{ + std::lock_guard lock(mutex); + + if (no_more_tasks_available) + return nullptr; + + if (buffered_ranges.empty()) + { + auto result = extension.callback(ParallelReadRequest( + CoordinationMode::Default, + extension.number_of_current_replica, + pool_settings.min_marks_for_concurrent_read * pool_settings.threads, + /// For Default coordination mode we don't need to pass part names. + RangesInDataPartsDescription{})); + + if (!result || result->finish) + { + no_more_tasks_available = true; + return nullptr; + } + + buffered_ranges = std::move(result->description); + } + + if (buffered_ranges.empty()) + throw Exception(ErrorCodes::LOGICAL_ERROR, "No tasks to read. This is a bug"); + + auto & current_task = buffered_ranges.front(); + + size_t part_idx = 0; + for (size_t index = 0; index < per_part_infos.size(); ++index) + { + if (per_part_infos[index]->data_part->info == current_task.info) + { + part_idx = index; + break; + } + } + + MarkRanges ranges_to_read; + size_t current_sum_marks = 0; + while (current_sum_marks < pool_settings.min_marks_for_concurrent_read && !current_task.ranges.empty()) + { + auto diff = pool_settings.min_marks_for_concurrent_read - current_sum_marks; + auto range = current_task.ranges.front(); + if (range.getNumberOfMarks() > diff) + { + auto new_range = range; + new_range.end = range.begin + diff; + range.begin += diff; + + current_task.ranges.front() = range; + ranges_to_read.push_back(new_range); + current_sum_marks += new_range.getNumberOfMarks(); + continue; + } + + ranges_to_read.push_back(range); + current_sum_marks += range.getNumberOfMarks(); + current_task.ranges.pop_front(); + } + + if (current_task.ranges.empty()) + buffered_ranges.pop_front(); + + return createTask(per_part_infos[part_idx], std::move(ranges_to_read), previous_task); +} + +} diff --git a/src/Storages/MergeTree/MergeTreeReadPoolParallelReplicas.h b/src/Storages/MergeTree/MergeTreeReadPoolParallelReplicas.h new file mode 100644 index 00000000000..08020565ec4 --- /dev/null +++ b/src/Storages/MergeTree/MergeTreeReadPoolParallelReplicas.h @@ -0,0 +1,39 @@ +#pragma once +#include +#include + +namespace DB +{ + +class MergeTreeReadPoolParallelReplicas : public MergeTreeReadPoolBase +{ +public: + MergeTreeReadPoolParallelReplicas( + ParallelReadingExtension extension_, + RangesInDataParts && parts_, + const StorageSnapshotPtr & storage_snapshot_, + const PrewhereInfoPtr & prewhere_info_, + const ExpressionActionsSettings & actions_settings_, + const MergeTreeReaderSettings & reader_settings_, + const Names & column_names_, + const Names & virtual_column_names_, + const PoolSettings & settings_, + const ContextPtr & context_); + + ~MergeTreeReadPoolParallelReplicas() override = default; + + String getName() const override { return "ReadPoolParallelReplicas"; } + bool preservesOrderOfRanges() const override { return false; } + void profileFeedback(ReadBufferFromFileBase::ProfileInfo) override {} + MergeTreeReadTaskPtr getTask(size_t task_idx, MergeTreeReadTask * previous_task) override; + +private: + mutable std::mutex mutex; + + const ParallelReadingExtension extension; + RangesInDataPartsDescription buffered_ranges; + bool no_more_tasks_available{false}; + Poco::Logger * log = &Poco::Logger::get("MergeTreeReadPoolParallelReplicas"); +}; + +} diff --git a/src/Storages/MergeTree/MergeTreeReadPoolParallelReplicasInOrder.cpp b/src/Storages/MergeTree/MergeTreeReadPoolParallelReplicasInOrder.cpp new file mode 100644 index 00000000000..a822a517933 --- /dev/null +++ b/src/Storages/MergeTree/MergeTreeReadPoolParallelReplicasInOrder.cpp @@ -0,0 +1,106 @@ +#include + +namespace DB +{ + +namespace ErrorCodes +{ + extern const int LOGICAL_ERROR; +} + +MergeTreeReadPoolParallelReplicasInOrder::MergeTreeReadPoolParallelReplicasInOrder( + ParallelReadingExtension extension_, + CoordinationMode mode_, + RangesInDataParts parts_, + const StorageSnapshotPtr & storage_snapshot_, + const PrewhereInfoPtr & prewhere_info_, + const ExpressionActionsSettings & actions_settings_, + const MergeTreeReaderSettings & reader_settings_, + const Names & column_names_, + const Names & virtual_column_names_, + const PoolSettings & settings_, + const ContextPtr & context_) + : MergeTreeReadPoolBase( + std::move(parts_), + storage_snapshot_, + prewhere_info_, + actions_settings_, + reader_settings_, + column_names_, + virtual_column_names_, + settings_, + context_) + , extension(std::move(extension_)) + , mode(mode_) +{ + for (const auto & part : parts_ranges) + request.push_back({part.data_part->info, MarkRanges{}}); + + for (const auto & part : parts_ranges) + buffered_tasks.push_back({part.data_part->info, MarkRanges{}}); + + extension.all_callback(InitialAllRangesAnnouncement( + mode, + parts_ranges.getDescriptions(), + extension.number_of_current_replica + )); +} + +MergeTreeReadTaskPtr MergeTreeReadPoolParallelReplicasInOrder::getTask(size_t task_idx, MergeTreeReadTask * previous_task) +{ + std::lock_guard lock(mutex); + + if (task_idx >= per_part_infos.size()) + throw Exception(ErrorCodes::LOGICAL_ERROR, + "Requested task with idx {}, but there are only {} parts", + task_idx, per_part_infos.size()); + + const auto & part_info = per_part_infos[task_idx]->data_part->info; + auto get_from_buffer = [&]() -> std::optional + { + for (auto & desc : buffered_tasks) + { + if (desc.info == part_info && !desc.ranges.empty()) + { + auto result = std::move(desc.ranges); + desc.ranges = MarkRanges{}; + return result; + } + } + return std::nullopt; + }; + + if (auto result = get_from_buffer(); result) + return createTask(per_part_infos[task_idx], std::move(*result), previous_task); + + if (no_more_tasks) + return nullptr; + + auto response = extension.callback(ParallelReadRequest( + mode, + extension.number_of_current_replica, + pool_settings.min_marks_for_concurrent_read * request.size(), + request + )); + + if (!response || response->description.empty() || response->finish) + { + no_more_tasks = true; + return nullptr; + } + + /// Fill the buffer + for (size_t i = 0; i < request.size(); ++i) + { + auto & new_ranges = response->description[i].ranges; + auto & old_ranges = buffered_tasks[i].ranges; + std::move(new_ranges.begin(), new_ranges.end(), std::back_inserter(old_ranges)); + } + + if (auto result = get_from_buffer(); result) + return createTask(per_part_infos[task_idx], std::move(*result), previous_task); + + return nullptr; +} + +} diff --git a/src/Storages/MergeTree/MergeTreeReadPoolParallelReplicasInOrder.h b/src/Storages/MergeTree/MergeTreeReadPoolParallelReplicasInOrder.h new file mode 100644 index 00000000000..3e5f8f5dfba --- /dev/null +++ b/src/Storages/MergeTree/MergeTreeReadPoolParallelReplicasInOrder.h @@ -0,0 +1,40 @@ +#pragma once +#include +#include + +namespace DB +{ + +class MergeTreeReadPoolParallelReplicasInOrder : public MergeTreeReadPoolBase +{ +public: + MergeTreeReadPoolParallelReplicasInOrder( + ParallelReadingExtension extension_, + CoordinationMode mode_, + RangesInDataParts parts_, + const StorageSnapshotPtr & storage_snapshot_, + const PrewhereInfoPtr & prewhere_info_, + const ExpressionActionsSettings & actions_settings_, + const MergeTreeReaderSettings & reader_settings_, + const Names & column_names_, + const Names & virtual_column_names_, + const PoolSettings & settings_, + const ContextPtr & context_); + + String getName() const override { return "ReadPoolParallelReplicasInOrder"; } + bool preservesOrderOfRanges() const override { return true; } + void profileFeedback(ReadBufferFromFileBase::ProfileInfo) override {} + MergeTreeReadTaskPtr getTask(size_t task_idx, MergeTreeReadTask * previous_task) override; + +private: + const ParallelReadingExtension extension; + const CoordinationMode mode; + + bool no_more_tasks{false}; + RangesInDataPartsDescription request; + RangesInDataPartsDescription buffered_tasks; + + mutable std::mutex mutex; +}; + +}; diff --git a/src/Storages/MergeTree/MergeTreeReadTask.cpp b/src/Storages/MergeTree/MergeTreeReadTask.cpp new file mode 100644 index 00000000000..9593a1fb1f3 --- /dev/null +++ b/src/Storages/MergeTree/MergeTreeReadTask.cpp @@ -0,0 +1,193 @@ +#include +#include + +namespace DB +{ + +namespace ErrorCodes +{ + extern const int LOGICAL_ERROR; +} + +String MergeTreeReadTask::Columns::dump() const +{ + WriteBufferFromOwnString s; + for (size_t i = 0; i < pre_columns.size(); ++i) + { + s << "STEP " << i << ": " << pre_columns[i].toString() << "\n"; + } + s << "COLUMNS: " << columns.toString() << "\n"; + return s.str(); +} + +MergeTreeReadTask::MergeTreeReadTask( + InfoPtr info_, + Readers readers_, + MarkRanges mark_ranges_, + MergeTreeBlockSizePredictorPtr size_predictor_) + : info(std::move(info_)) + , readers(std::move(readers_)) + , mark_ranges(std::move(mark_ranges_)) + , size_predictor(std::move(size_predictor_)) +{ +} + +MergeTreeReadTask::Readers MergeTreeReadTask::createReaders( + const InfoPtr & read_info, const Extras & extras, const MarkRanges & ranges) +{ + Readers new_readers; + + auto create_reader = [&](const NamesAndTypesList & columns_to_read) + { + return read_info->data_part->getReader( + columns_to_read, extras.storage_snapshot, ranges, + extras.uncompressed_cache, extras.mark_cache, + read_info->alter_conversions, extras.reader_settings, extras.value_size_map, extras.profile_callback); + }; + + new_readers.main = create_reader(read_info->task_columns.columns); + + /// Add lightweight delete filtering step + if (extras.reader_settings.apply_deleted_mask && read_info->data_part->hasLightweightDelete()) + new_readers.prewhere.push_back(create_reader({LightweightDeleteDescription::FILTER_COLUMN})); + + for (const auto & pre_columns_per_step : read_info->task_columns.pre_columns) + new_readers.prewhere.push_back(create_reader(pre_columns_per_step)); + + return new_readers; +} + +MergeTreeReadTask::RangeReaders MergeTreeReadTask::createRangeReaders( + const Readers & task_readers, + const PrewhereExprInfo & prewhere_actions, + const Names & non_const_virtual_column_names) +{ + MergeTreeReadTask::RangeReaders new_range_readers; + if (prewhere_actions.steps.size() != task_readers.prewhere.size()) + throw Exception( + ErrorCodes::LOGICAL_ERROR, + "PREWHERE steps count mismatch, actions: {}, readers: {}", + prewhere_actions.steps.size(), task_readers.prewhere.size()); + + MergeTreeRangeReader * prev_reader = nullptr; + bool last_reader = false; + + for (size_t i = 0; i < prewhere_actions.steps.size(); ++i) + { + last_reader = task_readers.main->getColumns().empty() && (i + 1 == prewhere_actions.steps.size()); + + MergeTreeRangeReader current_reader( + task_readers.prewhere[i].get(), + prev_reader, prewhere_actions.steps[i].get(), + last_reader, non_const_virtual_column_names); + + new_range_readers.prewhere.push_back(std::move(current_reader)); + prev_reader = &new_range_readers.prewhere.back(); + } + + if (!last_reader) + { + new_range_readers.main = MergeTreeRangeReader(task_readers.main.get(), prev_reader, nullptr, true, non_const_virtual_column_names); + } + else + { + /// If all columns are read by prewhere range readers than move last prewhere range reader to main. + new_range_readers.main = std::move(new_range_readers.prewhere.back()); + new_range_readers.prewhere.pop_back(); + } + + return new_range_readers; +} + +void MergeTreeReadTask::initializeRangeReaders( + const PrewhereExprInfo & prewhere_actions, + const Names & non_const_virtual_column_names) +{ + if (range_readers.main.isInitialized()) + throw Exception(ErrorCodes::LOGICAL_ERROR, "Range reader is already initialized"); + + range_readers = createRangeReaders(readers, prewhere_actions, non_const_virtual_column_names); +} + +UInt64 MergeTreeReadTask::estimateNumRows(const BlockSizeParams & params) const +{ + if (!size_predictor) + return static_cast(params.max_block_size_rows); + + /// Calculates number of rows will be read using preferred_block_size_bytes. + /// Can't be less than avg_index_granularity. + size_t rows_to_read = size_predictor->estimateNumRows(params.preferred_block_size_bytes); + if (!rows_to_read) + return rows_to_read; + + auto total_row_in_current_granule = range_readers.main.numRowsInCurrentGranule(); + rows_to_read = std::max(total_row_in_current_granule, rows_to_read); + + if (params.preferred_max_column_in_block_size_bytes) + { + /// Calculates number of rows will be read using preferred_max_column_in_block_size_bytes. + auto rows_to_read_for_max_size_column = size_predictor->estimateNumRowsForMaxSizeColumn(params.preferred_max_column_in_block_size_bytes); + + double filtration_ratio = std::max(params.min_filtration_ratio, 1.0 - size_predictor->filtered_rows_ratio); + auto rows_to_read_for_max_size_column_with_filtration + = static_cast(rows_to_read_for_max_size_column / filtration_ratio); + + /// If preferred_max_column_in_block_size_bytes is used, number of rows to read can be less than current_index_granularity. + rows_to_read = std::min(rows_to_read, rows_to_read_for_max_size_column_with_filtration); + } + + auto unread_rows_in_current_granule = range_readers.main.numPendingRowsInCurrentGranule(); + if (unread_rows_in_current_granule >= rows_to_read) + return rows_to_read; + + const auto & index_granularity = info->data_part->index_granularity; + return index_granularity.countMarksForRows(range_readers.main.currentMark(), rows_to_read, range_readers.main.numReadRowsInCurrentGranule(), params.min_marks_to_read); +} + +MergeTreeReadTask::BlockAndProgress MergeTreeReadTask::read(const BlockSizeParams & params) +{ + if (size_predictor) + size_predictor->startBlock(); + + UInt64 recommended_rows = estimateNumRows(params); + UInt64 rows_to_read = std::max(static_cast(1), std::min(params.max_block_size_rows, recommended_rows)); + + auto read_result = range_readers.main.read(rows_to_read, mark_ranges); + + /// All rows were filtered. Repeat. + if (read_result.num_rows == 0) + read_result.columns.clear(); + + const auto & sample_block = range_readers.main.getSampleBlock(); + if (read_result.num_rows != 0 && sample_block.columns() != read_result.columns.size()) + throw Exception(ErrorCodes::LOGICAL_ERROR, "Inconsistent number of columns got from MergeTreeRangeReader. " + "Have {} in sample block and {} columns in list", + toString(sample_block.columns()), toString(read_result.columns.size())); + + /// TODO: check columns have the same types as in header. + UInt64 num_filtered_rows = read_result.numReadRows() - read_result.num_rows; + + size_t num_read_rows = read_result.numReadRows(); + size_t num_read_bytes = read_result.numBytesRead(); + + if (size_predictor) + { + size_predictor->updateFilteredRowsRation(read_result.numReadRows(), num_filtered_rows); + if (!read_result.columns.empty()) + size_predictor->update(sample_block, read_result.columns, read_result.num_rows); + } + + Block block; + if (read_result.num_rows != 0) + block = sample_block.cloneWithColumns(read_result.columns); + + BlockAndProgress res = { + .block = std::move(block), + .row_count = read_result.num_rows, + .num_read_rows = num_read_rows, + .num_read_bytes = num_read_bytes }; + + return res; +} + +} diff --git a/src/Storages/MergeTree/MergeTreeReadTask.h b/src/Storages/MergeTree/MergeTreeReadTask.h new file mode 100644 index 00000000000..8bb02a2a869 --- /dev/null +++ b/src/Storages/MergeTree/MergeTreeReadTask.h @@ -0,0 +1,156 @@ +#pragma once + +#include +#include +#include +#include +#include +#include +#include + +namespace DB +{ + +class UncompressedCache; +class MarkCache; + +struct MergeTreeBlockSizePredictor; +using MergeTreeBlockSizePredictorPtr = std::shared_ptr; + +class IMergeTreeDataPart; +using DataPartPtr = std::shared_ptr; +using MergeTreeReaderPtr = std::unique_ptr; + +enum class MergeTreeReadType +{ + /// By default, read will use MergeTreeReadPool and return pipe with num_streams outputs. + /// If num_streams == 1, will read without pool, in order specified in parts. + Default, + /// Read in sorting key order. + /// Returned pipe will have the number of ports equals to parts.size(). + /// Parameter num_streams_ is ignored in this case. + /// User should add MergingSorted itself if needed. + InOrder, + /// The same as InOrder, but in reverse order. + /// For every part, read ranges and granules from end to begin. Also add ReverseTransform. + InReverseOrder, + /// A special type of reading where every replica + /// talks to a remote coordinator (which is located on the initiator node) + /// and who spreads marks and parts across them. + ParallelReplicas, +}; + +/// A batch of work for MergeTreeSelectProcessor +struct MergeTreeReadTask : private boost::noncopyable +{ +public: + struct Columns + { + /// Column names to read during WHERE + NamesAndTypesList columns; + /// Column names to read during each PREWHERE step + std::vector pre_columns; + + String dump() const; + }; + + struct Info + { + /// Data part which should be read while performing this task + DataPartPtr data_part; + /// For virtual `part_index` virtual column + size_t part_index_in_query; + /// Alter converversionss that should be applied on-fly for part. + AlterConversionsPtr alter_conversions; + /// Column names to read during PREWHERE and WHERE + Columns task_columns; + /// Shared initialied size predictor. It is copied for each new task. + MergeTreeBlockSizePredictorPtr shared_size_predictor; + }; + + using InfoPtr = std::shared_ptr; + + /// Extra params that required for creation of reader. + struct Extras + { + UncompressedCache * uncompressed_cache = nullptr; + MarkCache * mark_cache = nullptr; + MergeTreeReaderSettings reader_settings; + StorageSnapshotPtr storage_snapshot; + IMergeTreeReader::ValueSizeMap value_size_map; + ReadBufferFromFileBase::ProfileCallback profile_callback; + }; + + struct Readers + { + MergeTreeReaderPtr main; + std::vector prewhere; + }; + + struct RangeReaders + { + /// Used to save current range processing status + MergeTreeRangeReader main; + + /// Range readers for multiple filtering steps: row level security, PREWHERE etc. + /// NOTE: we take references to elements and push_back new elements, that's why it is a deque but not a vector + std::deque prewhere; + }; + + struct BlockSizeParams + { + UInt64 max_block_size_rows = DEFAULT_BLOCK_SIZE; + UInt64 preferred_block_size_bytes = 1000000; + UInt64 preferred_max_column_in_block_size_bytes = 0; + UInt64 min_marks_to_read = 0; + double min_filtration_ratio = 0.00001; + }; + + /// The result of reading from task. + struct BlockAndProgress + { + Block block; + size_t row_count = 0; + size_t num_read_rows = 0; + size_t num_read_bytes = 0; + }; + + MergeTreeReadTask(InfoPtr info_, Readers readers_, MarkRanges mark_ranges_, MergeTreeBlockSizePredictorPtr size_predictor_); + + void initializeRangeReaders(const PrewhereExprInfo & prewhere_actions, const Names & non_const_virtual_column_names); + + BlockAndProgress read(const BlockSizeParams & params); + bool isFinished() const { return mark_ranges.empty() && range_readers.main.isCurrentRangeFinished(); } + + const Info & getInfo() const { return *info; } + const MergeTreeRangeReader & getMainRangeReader() const { return range_readers.main; } + const IMergeTreeReader & getMainReader() const { return *readers.main; } + + Readers releaseReaders() { return std::move(readers); } + + static Readers createReaders(const InfoPtr & read_info, const Extras & extras, const MarkRanges & ranges); + static RangeReaders createRangeReaders(const Readers & readers, const PrewhereExprInfo & prewhere_actions, const Names & non_const_virtual_column_names); + +private: + UInt64 estimateNumRows(const BlockSizeParams & params) const; + + /// Shared information required for reading. + InfoPtr info; + + /// Readers for data_part of this task. + /// May be reused and released to the next task. + Readers readers; + + /// Range readers to read mark_ranges from data_part + RangeReaders range_readers; + + /// Ranges to read from data_part + MarkRanges mark_ranges; + + /// Used to satistfy preferred_block_size_bytes limitation + MergeTreeBlockSizePredictorPtr size_predictor; +}; + +using MergeTreeReadTaskPtr = std::unique_ptr; + +} diff --git a/src/Storages/MergeTree/MergeTreeReverseSelectProcessor.cpp b/src/Storages/MergeTree/MergeTreeReverseSelectProcessor.cpp deleted file mode 100644 index bb515bcd10f..00000000000 --- a/src/Storages/MergeTree/MergeTreeReverseSelectProcessor.cpp +++ /dev/null @@ -1,134 +0,0 @@ -#include -#include -#include "Storages/MergeTree/MergeTreeBaseSelectProcessor.h" - -namespace DB -{ - -namespace ErrorCodes -{ - extern const int MEMORY_LIMIT_EXCEEDED; -} - -bool MergeTreeReverseSelectAlgorithm::getNewTaskImpl() -try -{ - if (pool) - return getNewTaskParallelReplicas(); - else - return getNewTaskOrdinaryReading(); -} -catch (...) -{ - /// Suspicion of the broken part. A part is added to the queue for verification. - if (getCurrentExceptionCode() != ErrorCodes::MEMORY_LIMIT_EXCEEDED) - storage.reportBrokenPart(data_part); - throw; -} - - -bool MergeTreeReverseSelectAlgorithm::getNewTaskOrdinaryReading() -{ - if (chunks.empty() && all_mark_ranges.empty()) - return false; - - /// We have some blocks to return in buffer. - /// Return true to continue reading, but actually don't create a task. - if (all_mark_ranges.empty()) - return true; - - if (!reader) - initializeReaders(); - - /// Read ranges from right to left. - MarkRanges mark_ranges_for_task{std::move(all_mark_ranges.back())}; - all_mark_ranges.pop_back(); - - auto size_predictor = (preferred_block_size_bytes == 0) ? nullptr - : getSizePredictor(data_part, task_columns, sample_block); - - task = std::make_unique( - data_part, - alter_conversions, - mark_ranges_for_task, - part_index_in_query, - column_name_set, - task_columns, - std::move(size_predictor)); - - return true; - -} - -bool MergeTreeReverseSelectAlgorithm::getNewTaskParallelReplicas() -{ - if (chunks.empty() && no_more_tasks) - return false; - - /// We have some blocks to return in buffer. - /// Return true to continue reading, but actually don't create a task. - if (no_more_tasks) - return true; - - if (!reader) - initializeReaders(); - - auto description = RangesInDataPartDescription{ - .info = data_part->info, - /// We just ignore all the distribution done before - /// Everything will be done on coordinator side - .ranges = {}, - }; - - auto mark_ranges_for_task = pool->getNewTask(description); - if (mark_ranges_for_task.empty()) - { - /// If we have chunks in buffer - return true to continue reading from them - return !chunks.empty(); - } - - auto size_predictor = (preferred_block_size_bytes == 0) ? nullptr - : getSizePredictor(data_part, task_columns, sample_block); - - task = std::make_unique( - data_part, - alter_conversions, - mark_ranges_for_task, - part_index_in_query, - column_name_set, - task_columns, - std::move(size_predictor)); - - return true; -} - -MergeTreeReverseSelectAlgorithm::BlockAndProgress MergeTreeReverseSelectAlgorithm::readFromPart() -{ - BlockAndProgress res; - - if (!chunks.empty()) - { - res = std::move(chunks.back()); - chunks.pop_back(); - return res; - } - - if (!task->range_reader.isInitialized()) - initializeRangeReaders(*task); - - while (!task->isFinished()) - { - auto chunk = readFromPartImpl(); - chunks.push_back(std::move(chunk)); - } - - if (chunks.empty()) - return {}; - - res = std::move(chunks.back()); - chunks.pop_back(); - - return res; -} - -} diff --git a/src/Storages/MergeTree/MergeTreeReverseSelectProcessor.h b/src/Storages/MergeTree/MergeTreeReverseSelectProcessor.h deleted file mode 100644 index 618f2855695..00000000000 --- a/src/Storages/MergeTree/MergeTreeReverseSelectProcessor.h +++ /dev/null @@ -1,44 +0,0 @@ -#pragma once -#include -#include - - -namespace DB -{ - -/// Used to read data from single part with select query -/// in reverse order of primary key. -/// Cares about PREWHERE, virtual columns, indexes etc. -/// To read data from multiple parts, Storage (MergeTree) creates multiple such objects. -class MergeTreeReverseSelectAlgorithm final : public MergeTreeSelectAlgorithm -{ -public: - template - explicit MergeTreeReverseSelectAlgorithm(Args &&... args) - : MergeTreeSelectAlgorithm{std::forward(args)...} - { - LOG_TRACE(log, "Reading {} ranges in reverse order from part {}, approx. {} rows starting from {}", - all_mark_ranges.size(), data_part->name, total_rows, - data_part->index_granularity.getMarkStartingRow(all_mark_ranges.front().begin)); - } - - String getName() const override { return "MergeTreeReverse"; } - -private: - bool getNewTaskImpl() override; - void finalizeNewTask() override {} - - bool getNewTaskParallelReplicas(); - bool getNewTaskOrdinaryReading(); - - BlockAndProgress readFromPart() override; - - std::vector chunks; - - /// Used for parallel replicas - bool no_more_tasks{false}; - - Poco::Logger * log = &Poco::Logger::get("MergeTreeReverseSelectProcessor"); -}; - -} diff --git a/src/Storages/MergeTree/MergeTreeSelectAlgorithms.cpp b/src/Storages/MergeTree/MergeTreeSelectAlgorithms.cpp new file mode 100644 index 00000000000..8bc4377cffb --- /dev/null +++ b/src/Storages/MergeTree/MergeTreeSelectAlgorithms.cpp @@ -0,0 +1,87 @@ +#include +#include + +namespace DB +{ + +namespace ErrorCodes +{ + extern const int LOGICAL_ERROR; +} + +MergeTreeThreadSelectAlgorithm::TaskResult MergeTreeThreadSelectAlgorithm::getNewTask(IMergeTreeReadPool & pool, MergeTreeReadTask * previous_task) +{ + TaskResult res; + res.first = pool.getTask(thread_idx, previous_task); + res.second = !!res.first; + return res; +} + +MergeTreeReadTask::BlockAndProgress MergeTreeThreadSelectAlgorithm::readFromTask(MergeTreeReadTask * task, const MergeTreeReadTask::BlockSizeParams & params) +{ + if (!task) + return {}; + + return task->read(params); +} + +IMergeTreeSelectAlgorithm::TaskResult MergeTreeInOrderSelectAlgorithm::getNewTask(IMergeTreeReadPool & pool, MergeTreeReadTask * previous_task) +{ + if (!pool.preservesOrderOfRanges()) + throw Exception(ErrorCodes::LOGICAL_ERROR, + "MergeTreeInOrderSelectAlgorithm requires read pool that preserves order of ranges, got: {}", pool.getName()); + + TaskResult res; + res.first = pool.getTask(part_idx, previous_task); + res.second = !!res.first; + return res; +} + +MergeTreeReadTask::BlockAndProgress MergeTreeInOrderSelectAlgorithm::readFromTask(MergeTreeReadTask * task, const BlockSizeParams & params) +{ + if (!task) + return {}; + + return task->read(params); +} + +IMergeTreeSelectAlgorithm::TaskResult MergeTreeInReverseOrderSelectAlgorithm::getNewTask(IMergeTreeReadPool & pool, MergeTreeReadTask * previous_task) +{ + if (!pool.preservesOrderOfRanges()) + throw Exception(ErrorCodes::LOGICAL_ERROR, + "MergeTreeInReverseOrderSelectAlgorithm requires read pool that preserves order of ranges, got: {}", pool.getName()); + + TaskResult res; + res.first = pool.getTask(part_idx, previous_task); + /// We may have some chunks to return in buffer. + /// Set continue_reading to true but actually don't create a new task. + res.second = !!res.first || !chunks.empty(); + return res; +} + +MergeTreeReadTask::BlockAndProgress MergeTreeInReverseOrderSelectAlgorithm::readFromTask(MergeTreeReadTask * task, const BlockSizeParams & params) +{ + MergeTreeReadTask::BlockAndProgress res; + + if (!chunks.empty()) + { + res = std::move(chunks.back()); + chunks.pop_back(); + return res; + } + + if (!task) + return {}; + + while (!task->isFinished()) + chunks.push_back(task->read(params)); + + if (chunks.empty()) + return {}; + + res = std::move(chunks.back()); + chunks.pop_back(); + return res; +} + +} diff --git a/src/Storages/MergeTree/MergeTreeSelectAlgorithms.h b/src/Storages/MergeTree/MergeTreeSelectAlgorithms.h new file mode 100644 index 00000000000..a6254a90687 --- /dev/null +++ b/src/Storages/MergeTree/MergeTreeSelectAlgorithms.h @@ -0,0 +1,64 @@ +#pragma once + +#include +#include + +namespace DB +{ + +class IMergeTreeReadPool; + +class IMergeTreeSelectAlgorithm : private boost::noncopyable +{ +public: + /// The pair of {task, continue_reading}. + using TaskResult = std::pair; + using BlockSizeParams = MergeTreeReadTask::BlockSizeParams; + + virtual ~IMergeTreeSelectAlgorithm() = default; + + virtual String getName() const = 0; + virtual TaskResult getNewTask(IMergeTreeReadPool & pool, MergeTreeReadTask * previous_task) = 0; + virtual MergeTreeReadTask::BlockAndProgress readFromTask(MergeTreeReadTask * task, const BlockSizeParams & params) = 0; +}; + +using MergeTreeSelectAlgorithmPtr = std::unique_ptr; + +class MergeTreeThreadSelectAlgorithm : public IMergeTreeSelectAlgorithm +{ +public: + explicit MergeTreeThreadSelectAlgorithm(size_t thread_idx_) : thread_idx(thread_idx_) {} + String getName() const override { return "Thread"; } + TaskResult getNewTask(IMergeTreeReadPool & pool, MergeTreeReadTask * previous_task) override; + MergeTreeReadTask::BlockAndProgress readFromTask(MergeTreeReadTask * task, const BlockSizeParams & params) override; + +private: + const size_t thread_idx; +}; + +class MergeTreeInOrderSelectAlgorithm : public IMergeTreeSelectAlgorithm +{ +public: + explicit MergeTreeInOrderSelectAlgorithm(size_t part_idx_) : part_idx(part_idx_) {} + String getName() const override { return "InOrder"; } + TaskResult getNewTask(IMergeTreeReadPool & pool, MergeTreeReadTask * previous_task) override; + MergeTreeReadTask::BlockAndProgress readFromTask(MergeTreeReadTask * task, const BlockSizeParams & params) override; + +private: + const size_t part_idx; +}; + +class MergeTreeInReverseOrderSelectAlgorithm : public IMergeTreeSelectAlgorithm +{ +public: + explicit MergeTreeInReverseOrderSelectAlgorithm(size_t part_idx_) : part_idx(part_idx_) {} + String getName() const override { return "InReverseOrder"; } + TaskResult getNewTask(IMergeTreeReadPool & pool, MergeTreeReadTask * previous_task) override; + MergeTreeReadTask::BlockAndProgress readFromTask(MergeTreeReadTask * task, const BlockSizeParams & params) override; + +private: + const size_t part_idx; + std::vector chunks; +}; + +} diff --git a/src/Storages/MergeTree/MergeTreeSelectProcessor.cpp b/src/Storages/MergeTree/MergeTreeSelectProcessor.cpp index e5a45ad5554..975fad1ab6b 100644 --- a/src/Storages/MergeTree/MergeTreeSelectProcessor.cpp +++ b/src/Storages/MergeTree/MergeTreeSelectProcessor.cpp @@ -1,85 +1,435 @@ #include -#include -#include -#include -#include - +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include namespace DB { -MergeTreeSelectAlgorithm::MergeTreeSelectAlgorithm( +namespace ErrorCodes +{ + extern const int ILLEGAL_TYPE_OF_COLUMN_FOR_FILTER; + extern const int LOGICAL_ERROR; + extern const int QUERY_WAS_CANCELLED; +} + +static void injectNonConstVirtualColumns( + size_t rows, + Block & block, + const Names & virtual_columns); + +static void injectPartConstVirtualColumns( + size_t rows, + Block & block, + MergeTreeReadTask * task, + const DataTypePtr & partition_value_type, + const Names & virtual_columns); + +MergeTreeSelectProcessor::MergeTreeSelectProcessor( + MergeTreeReadPoolPtr pool_, + MergeTreeSelectAlgorithmPtr algorithm_, const MergeTreeData & storage_, - const StorageSnapshotPtr & storage_snapshot_, - const MergeTreeData::DataPartPtr & owned_data_part_, - const AlterConversionsPtr & alter_conversions_, - UInt64 max_block_size_rows_, - size_t preferred_block_size_bytes_, - size_t preferred_max_column_in_block_size_bytes_, - Names required_columns_, - MarkRanges mark_ranges_, - bool use_uncompressed_cache_, const PrewhereInfoPtr & prewhere_info_, const ExpressionActionsSettings & actions_settings_, + const MergeTreeReadTask::BlockSizeParams & block_size_params_, const MergeTreeReaderSettings & reader_settings_, - MergeTreeInOrderReadPoolParallelReplicasPtr pool_, - const Names & virt_column_names_, - size_t part_index_in_query_, - bool has_limit_below_one_block_) - : IMergeTreeSelectAlgorithm{ - storage_snapshot_->getSampleBlockForColumns(required_columns_), - storage_, storage_snapshot_, prewhere_info_, actions_settings_, max_block_size_rows_, - preferred_block_size_bytes_, preferred_max_column_in_block_size_bytes_, - reader_settings_, use_uncompressed_cache_, virt_column_names_}, - required_columns{std::move(required_columns_)}, - data_part{owned_data_part_}, - alter_conversions(alter_conversions_), - sample_block(storage_snapshot_->metadata->getSampleBlock()), - all_mark_ranges(std::move(mark_ranges_)), - part_index_in_query(part_index_in_query_), - has_limit_below_one_block(has_limit_below_one_block_), - pool(pool_), - total_rows(data_part->index_granularity.getRowsCountInRanges(all_mark_ranges)) + const Names & virt_column_names_) + : pool(std::move(pool_)) + , algorithm(std::move(algorithm_)) + , prewhere_info(prewhere_info_) + , actions_settings(actions_settings_) + , prewhere_actions(getPrewhereActions(prewhere_info, actions_settings, reader_settings_.enable_multiple_prewhere_read_steps)) + , reader_settings(reader_settings_) + , block_size_params(block_size_params_) + , virt_column_names(virt_column_names_) + , partition_value_type(storage_.getPartitionValueType()) { - ordered_names = header_without_const_virtual_columns.getNames(); + if (reader_settings.apply_deleted_mask) + { + PrewhereExprStep step + { + .type = PrewhereExprStep::Filter, + .actions = nullptr, + .filter_column_name = LightweightDeleteDescription::FILTER_COLUMN.name, + .remove_filter_column = true, + .need_filter = true, + .perform_alter_conversions = true, + }; + + lightweight_delete_filter_step = std::make_shared(std::move(step)); + } + + header_without_const_virtual_columns = applyPrewhereActions(pool->getHeader(), prewhere_info); + size_t non_const_columns_offset = header_without_const_virtual_columns.columns(); + injectNonConstVirtualColumns(0, header_without_const_virtual_columns, virt_column_names); + + for (size_t col_num = non_const_columns_offset; col_num < header_without_const_virtual_columns.columns(); ++col_num) + non_const_virtual_column_names.emplace_back(header_without_const_virtual_columns.getByPosition(col_num).name); + + result_header = header_without_const_virtual_columns; + injectPartConstVirtualColumns(0, result_header, nullptr, partition_value_type, virt_column_names); + + if (!prewhere_actions.steps.empty()) + LOG_TRACE(log, "PREWHERE condition was split into {} steps: {}", prewhere_actions.steps.size(), prewhere_actions.dumpConditions()); + + if (prewhere_info) + LOG_TEST(log, "Original PREWHERE DAG:\n{}\nPREWHERE actions:\n{}", + (prewhere_info->prewhere_actions ? prewhere_info->prewhere_actions->dumpDAG(): std::string("")), + (!prewhere_actions.steps.empty() ? prewhere_actions.dump() : std::string(""))); } -void MergeTreeSelectAlgorithm::initializeReaders() +String MergeTreeSelectProcessor::getName() const { - LoadedMergeTreeDataPartInfoForReader part_info(data_part, alter_conversions); - - task_columns = getReadTaskColumns( - part_info, storage_snapshot, - required_columns, virt_column_names, - prewhere_info, - actions_settings, reader_settings, /*with_subcolumns=*/ true); - - /// Will be used to distinguish between PREWHERE and WHERE columns when applying filter - const auto & column_names = task_columns.columns.getNames(); - column_name_set = NameSet{column_names.begin(), column_names.end()}; - - if (use_uncompressed_cache) - owned_uncompressed_cache = storage.getContext()->getUncompressedCache(); - - owned_mark_cache = storage.getContext()->getMarkCache(); - - initializeMergeTreeReadersForPart( - data_part, alter_conversions, task_columns, - all_mark_ranges, /*value_size_map=*/ {}, /*profile_callback=*/ {}); + return fmt::format("MergeTreeSelect(pool: {}, algorithm: {})", pool->getName(), algorithm->getName()); } +bool tryBuildPrewhereSteps(PrewhereInfoPtr prewhere_info, const ExpressionActionsSettings & actions_settings, PrewhereExprInfo & prewhere); -void MergeTreeSelectAlgorithm::finish() +PrewhereExprInfo MergeTreeSelectProcessor::getPrewhereActions(PrewhereInfoPtr prewhere_info, const ExpressionActionsSettings & actions_settings, bool enable_multiple_prewhere_read_steps) { - /** Close the files (before destroying the object). - * When many sources are created, but simultaneously reading only a few of them, - * buffers don't waste memory. - */ - reader.reset(); - pre_reader_for_step.clear(); - data_part.reset(); + PrewhereExprInfo prewhere_actions; + if (prewhere_info) + { + if (prewhere_info->row_level_filter) + { + PrewhereExprStep row_level_filter_step + { + .type = PrewhereExprStep::Filter, + .actions = std::make_shared(prewhere_info->row_level_filter, actions_settings), + .filter_column_name = prewhere_info->row_level_column_name, + .remove_filter_column = true, + .need_filter = true, + .perform_alter_conversions = true, + }; + + prewhere_actions.steps.emplace_back(std::make_shared(std::move(row_level_filter_step))); + } + + if (!enable_multiple_prewhere_read_steps || + !tryBuildPrewhereSteps(prewhere_info, actions_settings, prewhere_actions)) + { + PrewhereExprStep prewhere_step + { + .type = PrewhereExprStep::Filter, + .actions = std::make_shared(prewhere_info->prewhere_actions, actions_settings), + .filter_column_name = prewhere_info->prewhere_column_name, + .remove_filter_column = prewhere_info->remove_prewhere_column, + .need_filter = prewhere_info->need_filter, + .perform_alter_conversions = true, + }; + + prewhere_actions.steps.emplace_back(std::make_shared(std::move(prewhere_step))); + } + } + + return prewhere_actions; } -MergeTreeSelectAlgorithm::~MergeTreeSelectAlgorithm() = default; +ChunkAndProgress MergeTreeSelectProcessor::read() +{ + while (!is_cancelled) + { + try + { + bool continue_reading = true; + if (!task || task->isFinished()) + std::tie(task, continue_reading) = algorithm->getNewTask(*pool, task.get()); + + if (!continue_reading) + break; + } + catch (const Exception & e) + { + if (e.code() == ErrorCodes::QUERY_WAS_CANCELLED) + break; + throw; + } + + if (task && !task->getMainRangeReader().isInitialized()) + initializeRangeReaders(); + + auto res = algorithm->readFromTask(task.get(), block_size_params); + + if (res.row_count) + { + injectVirtualColumns(res.block, res.row_count, task.get(), partition_value_type, virt_column_names); + + /// Reorder the columns according to result_header + Columns ordered_columns; + ordered_columns.reserve(result_header.columns()); + for (size_t i = 0; i < result_header.columns(); ++i) + { + auto name = result_header.getByPosition(i).name; + ordered_columns.push_back(res.block.getByName(name).column); + } + + return ChunkAndProgress{ + .chunk = Chunk(ordered_columns, res.row_count), + .num_read_rows = res.num_read_rows, + .num_read_bytes = res.num_read_bytes, + .is_finished = false}; + } + else + { + return {Chunk(), res.num_read_rows, res.num_read_bytes, false}; + } + } + + return {Chunk(), 0, 0, true}; +} + +void MergeTreeSelectProcessor::initializeRangeReaders() +{ + PrewhereExprInfo all_prewhere_actions; + if (lightweight_delete_filter_step && task->getInfo().data_part->hasLightweightDelete()) + all_prewhere_actions.steps.push_back(lightweight_delete_filter_step); + + for (const auto & step : prewhere_actions.steps) + all_prewhere_actions.steps.push_back(step); + + task->initializeRangeReaders(all_prewhere_actions, non_const_virtual_column_names); +} + + +namespace +{ + struct VirtualColumnsInserter + { + explicit VirtualColumnsInserter(Block & block_) : block(block_) {} + + bool columnExists(const String & name) const { return block.has(name); } + + void insertUInt8Column(const ColumnPtr & column, const String & name) + { + block.insert({column, std::make_shared(), name}); + } + + void insertUInt64Column(const ColumnPtr & column, const String & name) + { + block.insert({column, std::make_shared(), name}); + } + + void insertUUIDColumn(const ColumnPtr & column, const String & name) + { + block.insert({column, std::make_shared(), name}); + } + + void insertLowCardinalityColumn(const ColumnPtr & column, const String & name) + { + block.insert({column, std::make_shared(std::make_shared()), name}); + } + + void insertPartitionValueColumn( + size_t rows, const Row & partition_value, const DataTypePtr & partition_value_type, const String & name) + { + ColumnPtr column; + if (rows) + column = partition_value_type->createColumnConst(rows, Tuple(partition_value.begin(), partition_value.end())) + ->convertToFullColumnIfConst(); + else + column = partition_value_type->createColumn(); + + block.insert({column, partition_value_type, name}); + } + + Block & block; + }; +} + +/// Adds virtual columns that are not const for all rows +static void injectNonConstVirtualColumns( + size_t rows, + Block & block, + const Names & virtual_columns) +{ + VirtualColumnsInserter inserter(block); + for (const auto & virtual_column_name : virtual_columns) + { + if (virtual_column_name == "_part_offset") + { + if (!rows) + { + inserter.insertUInt64Column(DataTypeUInt64().createColumn(), virtual_column_name); + } + else + { + if (!inserter.columnExists(virtual_column_name)) + throw Exception(ErrorCodes::LOGICAL_ERROR, + "Column {} must have been filled part reader", + virtual_column_name); + } + } + + if (virtual_column_name == LightweightDeleteDescription::FILTER_COLUMN.name) + { + /// If _row_exists column isn't present in the part then fill it here with 1s + ColumnPtr column; + if (rows) + column = LightweightDeleteDescription::FILTER_COLUMN.type->createColumnConst(rows, 1)->convertToFullColumnIfConst(); + else + column = LightweightDeleteDescription::FILTER_COLUMN.type->createColumn(); + + inserter.insertUInt8Column(column, virtual_column_name); + } + } +} + +/// Adds virtual columns that are const for the whole part +static void injectPartConstVirtualColumns( + size_t rows, + Block & block, + MergeTreeReadTask * task, + const DataTypePtr & partition_value_type, + const Names & virtual_columns) +{ + VirtualColumnsInserter inserter(block); + /// add virtual columns + /// Except _sample_factor, which is added from the outside. + if (!virtual_columns.empty()) + { + if (unlikely(rows && !task)) + throw Exception(ErrorCodes::LOGICAL_ERROR, "Cannot insert virtual columns to non-empty chunk without specified task."); + + const IMergeTreeDataPart * part = nullptr; + + if (rows) + { + part = task->getInfo().data_part.get(); + if (part->isProjectionPart()) + part = part->getParentPart(); + } + + for (const auto & virtual_column_name : virtual_columns) + { + if (virtual_column_name == "_part") + { + ColumnPtr column; + if (rows) + column = DataTypeLowCardinality{std::make_shared()} + .createColumnConst(rows, part->name) + ->convertToFullColumnIfConst(); + else + column = DataTypeLowCardinality{std::make_shared()}.createColumn(); + + inserter.insertLowCardinalityColumn(column, virtual_column_name); + } + else if (virtual_column_name == "_part_index") + { + ColumnPtr column; + if (rows) + column = DataTypeUInt64().createColumnConst(rows, task->getInfo().part_index_in_query)->convertToFullColumnIfConst(); + else + column = DataTypeUInt64().createColumn(); + + inserter.insertUInt64Column(column, virtual_column_name); + } + else if (virtual_column_name == "_part_uuid") + { + ColumnPtr column; + if (rows) + column = DataTypeUUID().createColumnConst(rows, part->uuid)->convertToFullColumnIfConst(); + else + column = DataTypeUUID().createColumn(); + + inserter.insertUUIDColumn(column, virtual_column_name); + } + else if (virtual_column_name == "_partition_id") + { + ColumnPtr column; + if (rows) + column = DataTypeLowCardinality{std::make_shared()} + .createColumnConst(rows, part->info.partition_id) + ->convertToFullColumnIfConst(); + else + column = DataTypeLowCardinality{std::make_shared()}.createColumn(); + + inserter.insertLowCardinalityColumn(column, virtual_column_name); + } + else if (virtual_column_name == "_partition_value") + { + if (rows) + inserter.insertPartitionValueColumn(rows, part->partition.value, partition_value_type, virtual_column_name); + else + inserter.insertPartitionValueColumn(rows, {}, partition_value_type, virtual_column_name); + } + } + } +} + +void MergeTreeSelectProcessor::injectVirtualColumns( + Block & block, size_t row_count, MergeTreeReadTask * task, const DataTypePtr & partition_value_type, const Names & virtual_columns) +{ + /// First add non-const columns that are filled by the range reader and then const columns that we will fill ourselves. + /// Note that the order is important: virtual columns filled by the range reader must go first + injectNonConstVirtualColumns(row_count, block, virtual_columns); + injectPartConstVirtualColumns(row_count, block, task, partition_value_type, virtual_columns); +} + +Block MergeTreeSelectProcessor::applyPrewhereActions(Block block, const PrewhereInfoPtr & prewhere_info) +{ + if (prewhere_info) + { + if (prewhere_info->row_level_filter) + { + block = prewhere_info->row_level_filter->updateHeader(std::move(block)); + auto & row_level_column = block.getByName(prewhere_info->row_level_column_name); + if (!row_level_column.type->canBeUsedInBooleanContext()) + { + throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_COLUMN_FOR_FILTER, "Invalid type for filter in PREWHERE: {}", + row_level_column.type->getName()); + } + + block.erase(prewhere_info->row_level_column_name); + } + + if (prewhere_info->prewhere_actions) + { + block = prewhere_info->prewhere_actions->updateHeader(std::move(block)); + + auto & prewhere_column = block.getByName(prewhere_info->prewhere_column_name); + if (!prewhere_column.type->canBeUsedInBooleanContext()) + { + throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_COLUMN_FOR_FILTER, "Invalid type for filter in PREWHERE: {}", + prewhere_column.type->getName()); + } + + if (prewhere_info->remove_prewhere_column) + { + block.erase(prewhere_info->prewhere_column_name); + } + else if (prewhere_info->need_filter) + { + WhichDataType which(removeNullable(recursiveRemoveLowCardinality(prewhere_column.type))); + + if (which.isNativeInt() || which.isNativeUInt()) + prewhere_column.column = prewhere_column.type->createColumnConst(block.rows(), 1u)->convertToFullColumnIfConst(); + else if (which.isFloat()) + prewhere_column.column = prewhere_column.type->createColumnConst(block.rows(), 1.0f)->convertToFullColumnIfConst(); + else + throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_COLUMN_FOR_FILTER, + "Illegal type {} of column for filter", + prewhere_column.type->getName()); + } + } + } + + return block; +} + +Block MergeTreeSelectProcessor::transformHeader( + Block block, const PrewhereInfoPtr & prewhere_info, const DataTypePtr & partition_value_type, const Names & virtual_columns) +{ + auto transformed = applyPrewhereActions(std::move(block), prewhere_info); + injectVirtualColumns(transformed, 0, nullptr, partition_value_type, virtual_columns); + return transformed; +} } diff --git a/src/Storages/MergeTree/MergeTreeSelectProcessor.h b/src/Storages/MergeTree/MergeTreeSelectProcessor.h index 5f4f49bf075..a5178cda55d 100644 --- a/src/Storages/MergeTree/MergeTreeSelectProcessor.h +++ b/src/Storages/MergeTree/MergeTreeSelectProcessor.h @@ -1,78 +1,118 @@ #pragma once -#include #include -#include -#include -#include -#include +#include +#include +#include +#include +#include namespace DB { +struct PrewhereExprInfo; -/// Used to read data from single part with select query -/// Cares about PREWHERE, virtual columns, indexes etc. -/// To read data from multiple parts, Storage (MergeTree) creates multiple such objects. -class MergeTreeSelectAlgorithm : public IMergeTreeSelectAlgorithm +struct ChunkAndProgress { -public: - MergeTreeSelectAlgorithm( - const MergeTreeData & storage, - const StorageSnapshotPtr & storage_snapshot_, - const MergeTreeData::DataPartPtr & owned_data_part_, - const AlterConversionsPtr & alter_conversions_, - UInt64 max_block_size_rows, - size_t preferred_block_size_bytes, - size_t preferred_max_column_in_block_size_bytes, - Names required_columns_, - MarkRanges mark_ranges, - bool use_uncompressed_cache, - const PrewhereInfoPtr & prewhere_info, - const ExpressionActionsSettings & actions_settings_, - const MergeTreeReaderSettings & reader_settings, - MergeTreeInOrderReadPoolParallelReplicasPtr pool_, - const Names & virt_column_names = {}, - size_t part_index_in_query_ = 0, - bool has_limit_below_one_block_ = false); - - ~MergeTreeSelectAlgorithm() override; - -protected: - /// Defer initialization from constructor, because it may be heavy - /// and it's better to do it lazily in `getNewTaskImpl`, which is executing in parallel. - void initializeReaders(); - void finish() final; - - /// Used by Task - Names required_columns; - /// Names from header. Used in order to order columns in read blocks. - Names ordered_names; - NameSet column_name_set; - - MergeTreeReadTaskColumns task_columns; - - /// Data part will not be removed if the pointer owns it - MergeTreeData::DataPartPtr data_part; - - /// Alter converversionss that should be applied on-fly for part. - AlterConversionsPtr alter_conversions; - - /// Cache getSampleBlock call, which might be heavy. - Block sample_block; - - /// Mark ranges we should read (in ascending order) - MarkRanges all_mark_ranges; - /// Value of _part_index virtual column (used only in SelectExecutor) - size_t part_index_in_query = 0; - /// If true, every task will be created only with one range. - /// It reduces amount of read data for queries with small LIMIT. - bool has_limit_below_one_block = false; - - /// Pool for reading in order - MergeTreeInOrderReadPoolParallelReplicasPtr pool; - - size_t total_rows = 0; + Chunk chunk; + size_t num_read_rows = 0; + size_t num_read_bytes = 0; + /// Explicitly indicate that we have read all data. + /// This is needed to occasionally return empty chunk to indicate the progress while the rows are filtered out in PREWHERE. + bool is_finished = false; }; +struct ParallelReadingExtension +{ + MergeTreeAllRangesCallback all_callback; + MergeTreeReadTaskCallback callback; + size_t count_participating_replicas{0}; + size_t number_of_current_replica{0}; + /// This is needed to estimate the number of bytes + /// between a pair of marks to perform one request + /// over the network for a 1Gb of data. + Names columns_to_read; +}; + +/// Base class for MergeTreeThreadSelectAlgorithm and MergeTreeSelectAlgorithm +class MergeTreeSelectProcessor : private boost::noncopyable +{ +public: + MergeTreeSelectProcessor( + MergeTreeReadPoolPtr pool_, + MergeTreeSelectAlgorithmPtr algorithm_, + const MergeTreeData & storage_, + const PrewhereInfoPtr & prewhere_info_, + const ExpressionActionsSettings & actions_settings_, + const MergeTreeReadTask::BlockSizeParams & block_size_params_, + const MergeTreeReaderSettings & reader_settings_, + const Names & virt_column_names_); + + String getName() const; + + static Block transformHeader( + Block block, + const PrewhereInfoPtr & prewhere_info, + const DataTypePtr & partition_value_type, + const Names & virtual_columns); + + Block getHeader() const { return result_header; } + + ChunkAndProgress read(); + + void cancel() { is_cancelled = true; } + + const MergeTreeReaderSettings & getSettings() const { return reader_settings; } + + static PrewhereExprInfo getPrewhereActions( + PrewhereInfoPtr prewhere_info, + const ExpressionActionsSettings & actions_settings, + bool enable_multiple_prewhere_read_steps); + +private: + /// This struct allow to return block with no columns but with non-zero number of rows similar to Chunk + struct BlockAndProgress + { + Block block; + size_t row_count = 0; + size_t num_read_rows = 0; + size_t num_read_bytes = 0; + }; + + /// Used for filling header with no rows as well as block with data + static void injectVirtualColumns(Block & block, size_t row_count, MergeTreeReadTask * task, const DataTypePtr & partition_value_type, const Names & virtual_columns); + static Block applyPrewhereActions(Block block, const PrewhereInfoPtr & prewhere_info); + + /// Sets up range readers corresponding to data readers + void initializeRangeReaders(); + + const MergeTreeReadPoolPtr pool; + const MergeTreeSelectAlgorithmPtr algorithm; + + const PrewhereInfoPtr prewhere_info; + const ExpressionActionsSettings actions_settings; + const PrewhereExprInfo prewhere_actions; + + const MergeTreeReaderSettings reader_settings; + const MergeTreeReadTask::BlockSizeParams block_size_params; + const Names virt_column_names; + const DataTypePtr partition_value_type; + + /// Current task to read from. + MergeTreeReadTaskPtr task; + /// This step is added when the part has lightweight delete mask + PrewhereExprStepPtr lightweight_delete_filter_step; + /// These columns will be filled by the merge tree range reader + Names non_const_virtual_column_names; + /// This header is used for chunks from readFromPart(). + Block header_without_const_virtual_columns; + /// A result of getHeader(). A chunk which this header is returned from read(). + Block result_header; + + Poco::Logger * log = &Poco::Logger::get("MergeTreeSelectProcessor"); + std::atomic is_cancelled{false}; +}; + +using MergeTreeSelectProcessorPtr = std::unique_ptr; + } diff --git a/src/Storages/MergeTree/MergeTreeSource.cpp b/src/Storages/MergeTree/MergeTreeSource.cpp index 69fbdd5a64d..a450505f7a8 100644 --- a/src/Storages/MergeTree/MergeTreeSource.cpp +++ b/src/Storages/MergeTree/MergeTreeSource.cpp @@ -1,5 +1,5 @@ #include -#include +#include #include #include #include @@ -24,7 +24,7 @@ struct MergeTreeSource::AsyncReadingState /// which can be called from background thread. /// Invariant: /// * background thread changes status InProgress -> IsFinished - /// * (status == InProgress) => (MergeTreeBaseSelectProcessor is alive) + /// * (status == InProgress) => (MergeTreeSelectProcessor is alive) void setResult(ChunkAndProgress chunk_) { @@ -118,7 +118,7 @@ struct MergeTreeSource::AsyncReadingState /// (executing thread) (bg pool thread) /// Control::finish() /// stage = Stage::IsFinished; - /// ~MergeTreeBaseSelectProcessor() + /// ~MergeTreeSelectProcessor() /// ~AsyncReadingState() /// control->stage != Stage::InProgress /// ~EventFD() @@ -133,12 +133,12 @@ private: }; #endif -MergeTreeSource::MergeTreeSource(MergeTreeSelectAlgorithmPtr algorithm_) - : ISource(algorithm_->getHeader()) - , algorithm(std::move(algorithm_)) +MergeTreeSource::MergeTreeSource(MergeTreeSelectProcessorPtr processor_) + : ISource(processor_->getHeader()) + , processor(std::move(processor_)) { #if defined(OS_LINUX) - if (algorithm->getSettings().use_asynchronous_read_from_pool) + if (processor->getSettings().use_asynchronous_read_from_pool) async_reading_state = std::make_unique(); #endif } @@ -147,12 +147,12 @@ MergeTreeSource::~MergeTreeSource() = default; std::string MergeTreeSource::getName() const { - return algorithm->getName(); + return processor->getName(); } void MergeTreeSource::onCancel() { - algorithm->cancel(); + processor->cancel(); } ISource::Status MergeTreeSource::prepare() @@ -184,7 +184,7 @@ Chunk MergeTreeSource::processReadResult(ChunkAndProgress chunk) finished = chunk.is_finished; /// We can return a chunk with no rows even if are not finished. - /// This allows to report progress when all the rows are filtered out inside MergeTreeBaseSelectProcessor by PREWHERE logic. + /// This allows to report progress when all the rows are filtered out inside MergeTreeSelectProcessor by PREWHERE logic. return std::move(chunk.chunk); } @@ -200,7 +200,7 @@ std::optional MergeTreeSource::tryGenerate() chassert(async_reading_state->getStage() == AsyncReadingState::Stage::NotStarted); /// It is important to store control into job. - /// Otherwise, race between job and ~MergeTreeBaseSelectProcessor is possible. + /// Otherwise, race between job and ~MergeTreeSelectProcessor is possible. auto job = [this, control = async_reading_state->start()]() mutable { auto holder = std::move(control); @@ -208,7 +208,7 @@ std::optional MergeTreeSource::tryGenerate() try { OpenTelemetry::SpanHolder span{"MergeTreeSource::tryGenerate()"}; - holder->setResult(algorithm->read()); + holder->setResult(processor->read()); } catch (...) { @@ -223,7 +223,7 @@ std::optional MergeTreeSource::tryGenerate() #endif OpenTelemetry::SpanHolder span{"MergeTreeSource::tryGenerate()"}; - return processReadResult(algorithm->read()); + return processReadResult(processor->read()); } #if defined(OS_LINUX) diff --git a/src/Storages/MergeTree/MergeTreeSource.h b/src/Storages/MergeTree/MergeTreeSource.h index 463faad0fab..655f0ee6ebe 100644 --- a/src/Storages/MergeTree/MergeTreeSource.h +++ b/src/Storages/MergeTree/MergeTreeSource.h @@ -4,15 +4,15 @@ namespace DB { -class IMergeTreeSelectAlgorithm; -using MergeTreeSelectAlgorithmPtr = std::unique_ptr; +class MergeTreeSelectProcessor; +using MergeTreeSelectProcessorPtr = std::unique_ptr; struct ChunkAndProgress; class MergeTreeSource final : public ISource { public: - explicit MergeTreeSource(MergeTreeSelectAlgorithmPtr algorithm_); + explicit MergeTreeSource(MergeTreeSelectProcessorPtr processor_); ~MergeTreeSource() override; std::string getName() const override; @@ -29,7 +29,7 @@ protected: void onCancel() override; private: - MergeTreeSelectAlgorithmPtr algorithm; + MergeTreeSelectProcessorPtr processor; #if defined(OS_LINUX) struct AsyncReadingState; diff --git a/src/Storages/MergeTree/MergeTreeThreadSelectProcessor.cpp b/src/Storages/MergeTree/MergeTreeThreadSelectProcessor.cpp deleted file mode 100644 index 892ae9ead87..00000000000 --- a/src/Storages/MergeTree/MergeTreeThreadSelectProcessor.cpp +++ /dev/null @@ -1,73 +0,0 @@ -#include -#include -#include -#include - - -namespace DB -{ - -MergeTreeThreadSelectAlgorithm::MergeTreeThreadSelectAlgorithm( - size_t thread_, - IMergeTreeReadPoolPtr pool_, - size_t min_marks_for_concurrent_read_, - size_t max_block_size_rows_, - size_t preferred_block_size_bytes_, - size_t preferred_max_column_in_block_size_bytes_, - const MergeTreeData & storage_, - const StorageSnapshotPtr & storage_snapshot_, - bool use_uncompressed_cache_, - const PrewhereInfoPtr & prewhere_info_, - const ExpressionActionsSettings & actions_settings_, - const MergeTreeReaderSettings & reader_settings_, - const Names & virt_column_names_) - : IMergeTreeSelectAlgorithm{ - pool_->getHeader(), storage_, storage_snapshot_, prewhere_info_, actions_settings_, max_block_size_rows_, - preferred_block_size_bytes_, preferred_max_column_in_block_size_bytes_, - reader_settings_, use_uncompressed_cache_, virt_column_names_}, - thread{thread_}, - pool{std::move(pool_)} -{ - min_marks_to_read = min_marks_for_concurrent_read_; -} - -/// Requests read task from MergeTreeReadPool and signals whether it got one -bool MergeTreeThreadSelectAlgorithm::getNewTaskImpl() -{ - task = pool->getTask(thread); - return static_cast(task); -} - - -void MergeTreeThreadSelectAlgorithm::finalizeNewTask() -{ - const std::string part_name = task->data_part->isProjectionPart() ? task->data_part->getParentPart()->name : task->data_part->name; - - /// Allows pool to reduce number of threads in case of too slow reads. - auto profile_callback = [this](ReadBufferFromFileBase::ProfileInfo info_) { pool->profileFeedback(info_); }; - IMergeTreeReader::ValueSizeMap value_size_map; - - if (reader && part_name != last_read_part_name) - { - value_size_map = reader->getAvgValueSizeHints(); - } - - /// task->reader.valid() means there is a prefetched reader in this test, use it. - const bool init_new_readers = !reader || task->reader.valid() || part_name != last_read_part_name; - if (init_new_readers) - initializeMergeTreeReadersForCurrentTask(value_size_map, profile_callback); - - last_read_part_name = part_name; -} - - -void MergeTreeThreadSelectAlgorithm::finish() -{ - reader.reset(); - pre_reader_for_step.clear(); -} - - -MergeTreeThreadSelectAlgorithm::~MergeTreeThreadSelectAlgorithm() = default; - -} diff --git a/src/Storages/MergeTree/MergeTreeThreadSelectProcessor.h b/src/Storages/MergeTree/MergeTreeThreadSelectProcessor.h deleted file mode 100644 index 4d9c9c92daf..00000000000 --- a/src/Storages/MergeTree/MergeTreeThreadSelectProcessor.h +++ /dev/null @@ -1,54 +0,0 @@ -#pragma once -#include - - -namespace DB -{ - -class IMergeTreeReadPool; -using IMergeTreeReadPoolPtr = std::shared_ptr; - -/** Used in conjunction with MergeTreeReadPool, asking it for more work to do and performing whatever reads it is asked - * to perform. - */ -class MergeTreeThreadSelectAlgorithm final : public IMergeTreeSelectAlgorithm -{ -public: - MergeTreeThreadSelectAlgorithm( - size_t thread_, - IMergeTreeReadPoolPtr pool_, - size_t min_marks_for_concurrent_read, - size_t max_block_size_, - size_t preferred_block_size_bytes_, - size_t preferred_max_column_in_block_size_bytes_, - const MergeTreeData & storage_, - const StorageSnapshotPtr & storage_snapshot_, - bool use_uncompressed_cache_, - const PrewhereInfoPtr & prewhere_info_, - const ExpressionActionsSettings & actions_settings_, - const MergeTreeReaderSettings & reader_settings_, - const Names & virt_column_names_); - - String getName() const override { return "MergeTreeThread"; } - - ~MergeTreeThreadSelectAlgorithm() override; - -protected: - /// Requests read task from MergeTreeReadPool and signals whether it got one - bool getNewTaskImpl() override; - - void finalizeNewTask() override; - - void finish() override; - -private: - /// "thread" index (there are N threads and each thread is assigned index in interval [0..N-1]) - size_t thread; - - IMergeTreeReadPoolPtr pool; - - /// Last part read in this thread - std::string last_read_part_name; -}; - -} diff --git a/tests/queries/0_stateless/01861_explain_pipeline.reference b/tests/queries/0_stateless/01861_explain_pipeline.reference index 427b3eaefc0..d2370755a84 100644 --- a/tests/queries/0_stateless/01861_explain_pipeline.reference +++ b/tests/queries/0_stateless/01861_explain_pipeline.reference @@ -4,7 +4,7 @@ ExpressionTransform ExpressionTransform ReplacingSorted 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 0 0 1 1 2 2 @@ -20,9 +20,9 @@ ExpressionTransform × 2 FilterSortedStreamByRange Description: filter values in [(5), +inf) ExpressionTransform - MergeTreeInOrder 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) 0 → 1 ReplacingSorted 2 → 1 FilterSortedStreamByRange × 2 Description: filter values in [-inf, (5)) ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 diff --git a/tests/queries/0_stateless/02149_read_in_order_fixed_prefix.reference b/tests/queries/0_stateless/02149_read_in_order_fixed_prefix.reference index d1662cdeb81..f2a4ef1f634 100644 --- a/tests/queries/0_stateless/02149_read_in_order_fixed_prefix.reference +++ b/tests/queries/0_stateless/02149_read_in_order_fixed_prefix.reference @@ -14,7 +14,7 @@ ExpressionTransform (Expression) ExpressionTransform × 2 (ReadFromMergeTree) - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 2020-10-01 9 2020-10-01 9 2020-10-01 9 @@ -32,9 +32,9 @@ ExpressionTransform ExpressionTransform × 2 (ReadFromMergeTree) ReverseTransform - MergeTreeReverse 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InReverseOrder) 0 → 1 ReverseTransform - MergeTreeReverse 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InReverseOrder) 0 → 1 2020-10-01 9 2020-10-01 9 2020-10-01 9 @@ -51,7 +51,7 @@ ExpressionTransform (Expression) ExpressionTransform × 2 (ReadFromMergeTree) - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 2020-10-11 0 2020-10-11 0 2020-10-11 0 @@ -65,7 +65,7 @@ ExpressionTransform (Expression) ExpressionTransform (ReadFromMergeTree) - MergeTreeInOrder 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) 0 → 1 (Expression) ExpressionTransform (Limit) @@ -77,7 +77,7 @@ ExpressionTransform ExpressionTransform (ReadFromMergeTree) ExpressionTransform - MergeTreeInOrder 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) 0 → 1 2020-10-11 0 0 2020-10-11 0 10 2020-10-11 0 20 @@ -93,7 +93,7 @@ ExpressionTransform (Expression) ExpressionTransform (ReadFromMergeTree) - MergeTreeInOrder 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) 0 → 1 (Expression) ExpressionTransform (Limit) @@ -107,7 +107,7 @@ ExpressionTransform ExpressionTransform (ReadFromMergeTree) ExpressionTransform - MergeTreeInOrder 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) 0 → 1 2020-10-12 0 2020-10-12 1 2020-10-12 2 @@ -129,7 +129,7 @@ ExpressionTransform ExpressionTransform (ReadFromMergeTree) ReverseTransform - MergeTreeReverse 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InReverseOrder) 0 → 1 (Expression) ExpressionTransform (Limit) @@ -142,7 +142,7 @@ ExpressionTransform (ReadFromMergeTree) ExpressionTransform ReverseTransform - MergeTreeReverse 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InReverseOrder) 0 → 1 2020-10-12 99999 2020-10-12 99998 2020-10-12 99997 @@ -169,7 +169,7 @@ ExpressionTransform (Expression) ExpressionTransform (ReadFromMergeTree) - MergeTreeInOrder 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) 0 → 1 2020-10-10 00:00:00 0.01 2020-10-10 00:00:00 0.01 2020-10-10 00:00:00 0.01 @@ -183,7 +183,7 @@ ExpressionTransform (Expression) ExpressionTransform (ReadFromMergeTree) - MergeTreeInOrder 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) 0 → 1 2020-10-10 00:00:00 0.01 2020-10-10 00:00:00 0.01 2020-10-10 00:00:00 0.01 diff --git a/tests/queries/0_stateless/02233_optimize_aggregation_in_order_prefix.reference b/tests/queries/0_stateless/02233_optimize_aggregation_in_order_prefix.reference index f98effbec67..d6c1c5d3fc4 100644 --- a/tests/queries/0_stateless/02233_optimize_aggregation_in_order_prefix.reference +++ b/tests/queries/0_stateless/02233_optimize_aggregation_in_order_prefix.reference @@ -19,7 +19,7 @@ ExpressionTransform × 2 (Expression) ExpressionTransform (ReadFromMergeTree) - MergeTreeInOrder 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) 0 → 1 explain pipeline select parent_key, child_key, count() from data_02233 group by parent_key, child_key with totals order by parent_key, child_key settings max_threads=1, optimize_aggregation_in_order=0, read_in_order_two_level_merge_threshold=1; (Expression) ExpressionTransform × 2 @@ -36,7 +36,7 @@ ExpressionTransform × 2 (Expression) ExpressionTransform (ReadFromMergeTree) - MergeTreeInOrder 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) 0 → 1 select parent_key, child_key, count() from data_02233 group by parent_key, child_key with totals order by parent_key, child_key settings max_threads=1, optimize_aggregation_in_order=1; 0 0 4 0 1 3 diff --git a/tests/queries/0_stateless/02317_distinct_in_order_optimization_explain.reference b/tests/queries/0_stateless/02317_distinct_in_order_optimization_explain.reference index 21ce47cc685..da07e94cead 100644 --- a/tests/queries/0_stateless/02317_distinct_in_order_optimization_explain.reference +++ b/tests/queries/0_stateless/02317_distinct_in_order_optimization_explain.reference @@ -38,19 +38,19 @@ DistinctTransform DistinctTransform -- Check reading in order for distinct -- disabled, distinct columns match sorting key -MergeTreeThread +algorithm: Thread -- enabled, distinct columns match sorting key -MergeTreeInOrder +algorithm: InOrder -- enabled, distinct columns form prefix of sorting key -MergeTreeInOrder +algorithm: InOrder -- enabled, distinct columns DON't form prefix of sorting key -MergeTreeThread +algorithm: Thread -- enabled, distinct columns contains constant columns, non-const columns form prefix of sorting key -MergeTreeInOrder +algorithm: InOrder -- enabled, distinct columns contains constant columns, non-const columns match prefix of sorting key -MergeTreeInOrder +algorithm: InOrder -- enabled, only part of distinct columns form prefix of sorting key -MergeTreeThread +algorithm: Thread === disable new analyzer === -- enabled, check that sorting properties are propagated from ReadFromMergeTree till preliminary distinct Sorting (Stream): a ASC, b ASC diff --git a/tests/queries/0_stateless/02317_distinct_in_order_optimization_explain.sh b/tests/queries/0_stateless/02317_distinct_in_order_optimization_explain.sh index db8fa7c1600..07c7bc4af56 100755 --- a/tests/queries/0_stateless/02317_distinct_in_order_optimization_explain.sh +++ b/tests/queries/0_stateless/02317_distinct_in_order_optimization_explain.sh @@ -12,8 +12,8 @@ GREP_DISTINCT="grep 'DistinctSortedChunkTransform\|DistinctSortedStreamTransform TRIM_LEADING_SPACES="sed -e 's/^[ \t]*//'" REMOVE_NON_LETTERS="sed 's/[^a-zA-Z]//g'" FIND_DISTINCT="$GREP_DISTINCT | $TRIM_LEADING_SPACES | $REMOVE_NON_LETTERS" -FIND_READING_IN_ORDER="grep 'MergeTreeInOrder' | $TRIM_LEADING_SPACES | $REMOVE_NON_LETTERS" -FIND_READING_DEFAULT="grep 'MergeTreeThread' | $TRIM_LEADING_SPACES | $REMOVE_NON_LETTERS" +FIND_READING_IN_ORDER="grep -o 'algorithm: InOrder' | $TRIM_LEADING_SPACES" +FIND_READING_DEFAULT="grep -o 'algorithm: Thread' | $TRIM_LEADING_SPACES" FIND_SORTING_PROPERTIES="grep 'Sorting (Stream)' | $TRIM_LEADING_SPACES" $CLICKHOUSE_CLIENT -q "drop table if exists distinct_in_order_explain sync" diff --git a/tests/queries/0_stateless/02340_parts_refcnt_mergetree.reference b/tests/queries/0_stateless/02340_parts_refcnt_mergetree.reference index e225ce389cb..ae4fafae829 100644 --- a/tests/queries/0_stateless/02340_parts_refcnt_mergetree.reference +++ b/tests/queries/0_stateless/02340_parts_refcnt_mergetree.reference @@ -1,2 +1,2 @@ -data_02340 1_2_2_0 5 -data_02340_rep 1_0_0_0 5 +data_02340 1_2_2_0 6 +data_02340_rep 1_0_0_0 6 diff --git a/tests/queries/0_stateless/02343_aggregation_pipeline.reference b/tests/queries/0_stateless/02343_aggregation_pipeline.reference index ca838fdf4e0..acb0368c3cc 100644 --- a/tests/queries/0_stateless/02343_aggregation_pipeline.reference +++ b/tests/queries/0_stateless/02343_aggregation_pipeline.reference @@ -100,7 +100,7 @@ ExpressionTransform × 16 (Expression) ExpressionTransform (ReadFromMergeTree) - MergeTreeInOrder 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) 0 → 1 (ReadFromRemote) explain pipeline SELECT k1, k3, sum(value) v FROM remote('127.0.0.{1,2}', currentDatabase(), proj_agg_02343) GROUP BY k1, k3 SETTINGS distributed_aggregation_memory_efficient = 1; (Expression) @@ -117,7 +117,7 @@ ExpressionTransform × 16 (Expression) ExpressionTransform (ReadFromMergeTree) - MergeTreeInOrder 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) 0 → 1 (ReadFromRemote) -- { echoOn } @@ -144,6 +144,6 @@ ExpressionTransform × 2 (Expression) ExpressionTransform × 2 (ReadFromMergeTree) - MergeTreeThread × 2 0 → 1 + MergeTreeSelect(pool: ReadPool, algorithm: Thread) × 2 0 → 1 (ReadFromRemote) (ReadFromRemote) diff --git a/tests/queries/0_stateless/02343_aggregation_pipeline.sql b/tests/queries/0_stateless/02343_aggregation_pipeline.sql index b018cf21f91..ee82c8d7c47 100644 --- a/tests/queries/0_stateless/02343_aggregation_pipeline.sql +++ b/tests/queries/0_stateless/02343_aggregation_pipeline.sql @@ -5,6 +5,8 @@ set max_threads = 16; set prefer_localhost_replica = 1; set optimize_aggregation_in_order = 0; set max_block_size = 65505; +set allow_prefetched_read_pool_for_remote_filesystem = 0; +set allow_prefetched_read_pool_for_local_filesystem = 0; -- { echoOn } diff --git a/tests/queries/0_stateless/02404_memory_bound_merging.reference b/tests/queries/0_stateless/02404_memory_bound_merging.reference index 41a3b6bf8ec..0409c48f846 100644 --- a/tests/queries/0_stateless/02404_memory_bound_merging.reference +++ b/tests/queries/0_stateless/02404_memory_bound_merging.reference @@ -16,7 +16,7 @@ ExpressionTransform × 4 (Expression) ExpressionTransform × 4 (ReadFromMergeTree) - MergeTreeInOrder × 4 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 4 0 → 1 (ReadFromRemote) select a from remote(test_cluster_two_shards, currentDatabase(), t) group by a order by a limit 5 offset 100500; 100500 @@ -47,7 +47,7 @@ ExpressionTransform × 4 (Expression) ExpressionTransform × 4 (ReadFromMergeTree) - MergeTreeInOrder × 4 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 4 0 → 1 (ReadFromRemote) (ReadFromRemote) select a from remote(test_cluster_two_shards, currentDatabase(), dist_t) group by a order by a limit 5 offset 100500; @@ -84,7 +84,7 @@ ExpressionTransform (Expression) ExpressionTransform × 4 (ReadFromMergeTree) - MergeTreeInOrder × 4 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 4 0 → 1 (ReadFromRemote) select a, count() from dist_t_different_dbs group by a order by a limit 5 offset 500; 500 2000 diff --git a/tests/queries/0_stateless/02404_memory_bound_merging.sql b/tests/queries/0_stateless/02404_memory_bound_merging.sql index 23c29ca8d93..a9577aa4222 100644 --- a/tests/queries/0_stateless/02404_memory_bound_merging.sql +++ b/tests/queries/0_stateless/02404_memory_bound_merging.sql @@ -57,7 +57,9 @@ select a, count() from dist_t_different_dbs group by a, b order by a limit 5 off -- { echoOff } -- -create table pr_t(a UInt64, b UInt64) engine=MergeTree order by a; +drop table if exists pr_t; + +create table pr_t(a UInt64, b UInt64) engine=MergeTree order by a settings storage_policy = 's3'; insert into pr_t select number % 1000, number % 1000 from numbers_mt(1e6); set allow_experimental_parallel_reading_from_replicas = 1; diff --git a/tests/queries/0_stateless/02493_max_streams_for_merge_tree_reading.reference b/tests/queries/0_stateless/02493_max_streams_for_merge_tree_reading.reference index dfa09193761..ad541387510 100644 --- a/tests/queries/0_stateless/02493_max_streams_for_merge_tree_reading.reference +++ b/tests/queries/0_stateless/02493_max_streams_for_merge_tree_reading.reference @@ -3,33 +3,33 @@ -- The number of output streams is limited by max_streams_for_merge_tree_reading select sum(x) from t settings max_threads=32, max_streams_for_merge_tree_reading=16, allow_asynchronous_read_from_io_pool_for_merge_tree=0; 49999995000000 -select * from (explain pipeline select sum(x) from t settings max_threads=32, max_streams_for_merge_tree_reading=16, allow_asynchronous_read_from_io_pool_for_merge_tree=0) where explain like '%Resize%' or explain like '%MergeTreeThread%'; +select * from (explain pipeline select sum(x) from t settings max_threads=32, max_streams_for_merge_tree_reading=16, allow_asynchronous_read_from_io_pool_for_merge_tree=0) where explain like '%Resize%' or explain like '%MergeTreeSelect%'; Resize 16 → 32 StrictResize 16 → 16 - MergeTreeThread × 16 0 → 1 + MergeTreeSelect(pool: ReadPool, algorithm: Thread) × 16 0 → 1 -- Without asynchronous_read, max_streams_for_merge_tree_reading limits max_streams * max_streams_to_max_threads_ratio select sum(x) from t settings max_threads=4, max_streams_for_merge_tree_reading=16, allow_asynchronous_read_from_io_pool_for_merge_tree=0, max_streams_to_max_threads_ratio=8; 49999995000000 -select * from (explain pipeline select sum(x) from t settings max_threads=4, max_streams_for_merge_tree_reading=16, allow_asynchronous_read_from_io_pool_for_merge_tree=0, max_streams_to_max_threads_ratio=8) where explain like '%Resize%' or explain like '%MergeTreeThread%'; +select * from (explain pipeline select sum(x) from t settings max_threads=4, max_streams_for_merge_tree_reading=16, allow_asynchronous_read_from_io_pool_for_merge_tree=0, max_streams_to_max_threads_ratio=8) where explain like '%Resize%' or explain like '%MergeTreeSelect%'; Resize 16 → 4 StrictResize 16 → 16 - MergeTreeThread × 16 0 → 1 + MergeTreeSelect(pool: ReadPool, algorithm: Thread) × 16 0 → 1 -- With asynchronous_read, read in max_streams_for_merge_tree_reading async streams and resize to max_threads select sum(x) from t settings max_threads=4, max_streams_for_merge_tree_reading=16, allow_asynchronous_read_from_io_pool_for_merge_tree=1; 49999995000000 -select * from (explain pipeline select sum(x) from t settings max_threads=4, max_streams_for_merge_tree_reading=16, allow_asynchronous_read_from_io_pool_for_merge_tree=1) where explain like '%Resize%' or explain like '%MergeTreeThread%'; +select * from (explain pipeline select sum(x) from t settings max_threads=4, max_streams_for_merge_tree_reading=16, allow_asynchronous_read_from_io_pool_for_merge_tree=1) where explain like '%Resize%' or explain like '%MergeTreeSelect%'; Resize 4 → 4 StrictResize 4 → 4 Resize 16 → 4 - MergeTreeThread × 16 0 → 1 + MergeTreeSelect(pool: ReadPool, algorithm: Thread) × 16 0 → 1 -- With asynchronous_read, read using max_streams * max_streams_to_max_threads_ratio async streams, resize to max_streams_for_merge_tree_reading outp[ut streams, resize to max_threads after aggregation select sum(x) from t settings max_threads=4, max_streams_for_merge_tree_reading=16, allow_asynchronous_read_from_io_pool_for_merge_tree=1, max_streams_to_max_threads_ratio=8; 49999995000000 -select * from (explain pipeline select sum(x) from t settings max_threads=4, max_streams_for_merge_tree_reading=16, allow_asynchronous_read_from_io_pool_for_merge_tree=1, max_streams_to_max_threads_ratio=8) where explain like '%Resize%' or explain like '%MergeTreeThread%'; +select * from (explain pipeline select sum(x) from t settings max_threads=4, max_streams_for_merge_tree_reading=16, allow_asynchronous_read_from_io_pool_for_merge_tree=1, max_streams_to_max_threads_ratio=8) where explain like '%Resize%' or explain like '%MergeTreeSelect%'; Resize 16 → 4 StrictResize 16 → 16 Resize 32 → 16 - MergeTreeThread × 32 0 → 1 + MergeTreeSelect(pool: ReadPool, algorithm: Thread) × 32 0 → 1 -- For read-in-order, disable everything set query_plan_remove_redundant_sorting=0; -- to keep reading in order select sum(x) from (select x from t order by x) settings max_threads=4, max_streams_for_merge_tree_reading=16, allow_asynchronous_read_from_io_pool_for_merge_tree=1, optimize_read_in_order=1, query_plan_read_in_order=1; diff --git a/tests/queries/0_stateless/02493_max_streams_for_merge_tree_reading.sql b/tests/queries/0_stateless/02493_max_streams_for_merge_tree_reading.sql index c8643b5c758..f2e81273f12 100644 --- a/tests/queries/0_stateless/02493_max_streams_for_merge_tree_reading.sql +++ b/tests/queries/0_stateless/02493_max_streams_for_merge_tree_reading.sql @@ -4,23 +4,26 @@ drop table if exists t; create table t (x UInt64) engine = MergeTree order by x; insert into t select number from numbers_mt(10000000) settings max_insert_threads=8; +set allow_prefetched_read_pool_for_remote_filesystem = 0; +set allow_prefetched_read_pool_for_local_filesystem = 0; + -- { echo } -- The number of output streams is limited by max_streams_for_merge_tree_reading select sum(x) from t settings max_threads=32, max_streams_for_merge_tree_reading=16, allow_asynchronous_read_from_io_pool_for_merge_tree=0; -select * from (explain pipeline select sum(x) from t settings max_threads=32, max_streams_for_merge_tree_reading=16, allow_asynchronous_read_from_io_pool_for_merge_tree=0) where explain like '%Resize%' or explain like '%MergeTreeThread%'; +select * from (explain pipeline select sum(x) from t settings max_threads=32, max_streams_for_merge_tree_reading=16, allow_asynchronous_read_from_io_pool_for_merge_tree=0) where explain like '%Resize%' or explain like '%MergeTreeSelect%'; -- Without asynchronous_read, max_streams_for_merge_tree_reading limits max_streams * max_streams_to_max_threads_ratio select sum(x) from t settings max_threads=4, max_streams_for_merge_tree_reading=16, allow_asynchronous_read_from_io_pool_for_merge_tree=0, max_streams_to_max_threads_ratio=8; -select * from (explain pipeline select sum(x) from t settings max_threads=4, max_streams_for_merge_tree_reading=16, allow_asynchronous_read_from_io_pool_for_merge_tree=0, max_streams_to_max_threads_ratio=8) where explain like '%Resize%' or explain like '%MergeTreeThread%'; +select * from (explain pipeline select sum(x) from t settings max_threads=4, max_streams_for_merge_tree_reading=16, allow_asynchronous_read_from_io_pool_for_merge_tree=0, max_streams_to_max_threads_ratio=8) where explain like '%Resize%' or explain like '%MergeTreeSelect%'; -- With asynchronous_read, read in max_streams_for_merge_tree_reading async streams and resize to max_threads select sum(x) from t settings max_threads=4, max_streams_for_merge_tree_reading=16, allow_asynchronous_read_from_io_pool_for_merge_tree=1; -select * from (explain pipeline select sum(x) from t settings max_threads=4, max_streams_for_merge_tree_reading=16, allow_asynchronous_read_from_io_pool_for_merge_tree=1) where explain like '%Resize%' or explain like '%MergeTreeThread%'; +select * from (explain pipeline select sum(x) from t settings max_threads=4, max_streams_for_merge_tree_reading=16, allow_asynchronous_read_from_io_pool_for_merge_tree=1) where explain like '%Resize%' or explain like '%MergeTreeSelect%'; -- With asynchronous_read, read using max_streams * max_streams_to_max_threads_ratio async streams, resize to max_streams_for_merge_tree_reading outp[ut streams, resize to max_threads after aggregation select sum(x) from t settings max_threads=4, max_streams_for_merge_tree_reading=16, allow_asynchronous_read_from_io_pool_for_merge_tree=1, max_streams_to_max_threads_ratio=8; -select * from (explain pipeline select sum(x) from t settings max_threads=4, max_streams_for_merge_tree_reading=16, allow_asynchronous_read_from_io_pool_for_merge_tree=1, max_streams_to_max_threads_ratio=8) where explain like '%Resize%' or explain like '%MergeTreeThread%'; +select * from (explain pipeline select sum(x) from t settings max_threads=4, max_streams_for_merge_tree_reading=16, allow_asynchronous_read_from_io_pool_for_merge_tree=1, max_streams_to_max_threads_ratio=8) where explain like '%Resize%' or explain like '%MergeTreeSelect%'; -- For read-in-order, disable everything set query_plan_remove_redundant_sorting=0; -- to keep reading in order diff --git a/tests/queries/0_stateless/02521_aggregation_by_partitions.reference b/tests/queries/0_stateless/02521_aggregation_by_partitions.reference index 188a6889002..67a131ff853 100644 --- a/tests/queries/0_stateless/02521_aggregation_by_partitions.reference +++ b/tests/queries/0_stateless/02521_aggregation_by_partitions.reference @@ -7,13 +7,13 @@ ExpressionTransform × 16 ExpressionTransform × 4 (ReadFromMergeTree) Resize 3 → 1 - MergeTreeThread × 3 0 → 1 + MergeTreeSelect(pool: ReadPool, algorithm: Thread) × 3 0 → 1 Resize 3 → 1 - MergeTreeThread × 3 0 → 1 + MergeTreeSelect(pool: ReadPool, algorithm: Thread) × 3 0 → 1 Resize 3 → 1 - MergeTreeThread × 3 0 → 1 + MergeTreeSelect(pool: ReadPool, algorithm: Thread) × 3 0 → 1 Resize 3 → 1 - MergeTreeThread × 3 0 → 1 + MergeTreeSelect(pool: ReadPool, algorithm: Thread) × 3 0 → 1 1000000 (Expression) ExpressionTransform × 16 @@ -24,21 +24,21 @@ ExpressionTransform × 16 ExpressionTransform × 8 (ReadFromMergeTree) Resize 2 → 1 - MergeTreeThread × 2 0 → 1 + MergeTreeSelect(pool: ReadPool, algorithm: Thread) × 2 0 → 1 Resize 2 → 1 - MergeTreeThread × 2 0 → 1 + MergeTreeSelect(pool: ReadPool, algorithm: Thread) × 2 0 → 1 Resize 2 → 1 - MergeTreeThread × 2 0 → 1 + MergeTreeSelect(pool: ReadPool, algorithm: Thread) × 2 0 → 1 Resize 2 → 1 - MergeTreeThread × 2 0 → 1 + MergeTreeSelect(pool: ReadPool, algorithm: Thread) × 2 0 → 1 Resize 2 → 1 - MergeTreeThread × 2 0 → 1 + MergeTreeSelect(pool: ReadPool, algorithm: Thread) × 2 0 → 1 Resize 2 → 1 - MergeTreeThread × 2 0 → 1 + MergeTreeSelect(pool: ReadPool, algorithm: Thread) × 2 0 → 1 Resize 2 → 1 - MergeTreeThread × 2 0 → 1 + MergeTreeSelect(pool: ReadPool, algorithm: Thread) × 2 0 → 1 Resize 2 → 1 - MergeTreeThread × 2 0 → 1 + MergeTreeSelect(pool: ReadPool, algorithm: Thread) × 2 0 → 1 1000000 (Expression) ExpressionTransform × 16 @@ -49,37 +49,37 @@ ExpressionTransform × 16 ExpressionTransform × 16 (ReadFromMergeTree) Concat 2 → 1 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 Concat 2 → 1 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 Concat 2 → 1 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 Concat 2 → 1 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 Concat 2 → 1 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 Concat 2 → 1 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 Concat 2 → 1 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 Concat 2 → 1 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 Concat 2 → 1 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 Concat 2 → 1 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 Concat 2 → 1 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 Concat 2 → 1 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 Concat 2 → 1 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 Concat 2 → 1 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 Concat 2 → 1 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 Concat 2 → 1 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 1000000 1000000 (Expression) @@ -94,16 +94,16 @@ ExpressionTransform × 16 ExpressionTransform × 4 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 1000000 (Expression) ExpressionTransform × 16 @@ -117,28 +117,28 @@ ExpressionTransform × 16 ExpressionTransform × 8 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 1000000 (Expression) ExpressionTransform × 16 @@ -151,52 +151,52 @@ ExpressionTransform × 16 ExpressionTransform × 16 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 1000000 Skip merging: 1 Skip merging: 1 diff --git a/tests/queries/0_stateless/02521_aggregation_by_partitions.sql b/tests/queries/0_stateless/02521_aggregation_by_partitions.sql index 73d58bb6d6c..767529c9783 100644 --- a/tests/queries/0_stateless/02521_aggregation_by_partitions.sql +++ b/tests/queries/0_stateless/02521_aggregation_by_partitions.sql @@ -5,6 +5,9 @@ set allow_aggregate_partitions_independently = 1; set force_aggregate_partitions_independently = 1; set optimize_use_projections = 0; +set allow_prefetched_read_pool_for_remote_filesystem = 0; +set allow_prefetched_read_pool_for_local_filesystem = 0; + create table t1(a UInt32) engine=MergeTree order by tuple() partition by a % 4 settings index_granularity = 8192, index_granularity_bytes = 10485760; system stop merges t1; diff --git a/tests/queries/0_stateless/02554_fix_grouping_sets_predicate_push_down.reference b/tests/queries/0_stateless/02554_fix_grouping_sets_predicate_push_down.reference index 440f668c614..794fe5944cd 100644 --- a/tests/queries/0_stateless/02554_fix_grouping_sets_predicate_push_down.reference +++ b/tests/queries/0_stateless/02554_fix_grouping_sets_predicate_push_down.reference @@ -40,7 +40,7 @@ ExpressionTransform × 2 (Expression) ExpressionTransform (ReadFromMergeTree) - MergeTreeInOrder 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) 0 → 1 ---Result--- 2023-01-05 all @@ -59,4 +59,4 @@ ExpressionTransform × 2 (Filter) FilterTransform (ReadFromMergeTree) - MergeTreeInOrder 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) 0 → 1 diff --git a/tests/queries/0_stateless/02582_async_reading_with_small_limit.reference b/tests/queries/0_stateless/02582_async_reading_with_small_limit.reference index a7f994d4b13..770e0046c27 100644 --- a/tests/queries/0_stateless/02582_async_reading_with_small_limit.reference +++ b/tests/queries/0_stateless/02582_async_reading_with_small_limit.reference @@ -4,4 +4,4 @@ ExpressionTransform Limit (ReadFromMergeTree) Concat 3 → 1 - MergeTreeInOrder × 3 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 3 0 → 1 diff --git a/tests/queries/0_stateless/02582_async_reading_with_small_limit.sql b/tests/queries/0_stateless/02582_async_reading_with_small_limit.sql index 6ea6f880712..65b51014258 100644 --- a/tests/queries/0_stateless/02582_async_reading_with_small_limit.sql +++ b/tests/queries/0_stateless/02582_async_reading_with_small_limit.sql @@ -1,3 +1,5 @@ +drop table if exists t; + create table t(a UInt64) engine=MergeTree order by tuple(); system stop merges t; diff --git a/tests/queries/0_stateless/02676_distinct_reading_in_order_analyzer.reference b/tests/queries/0_stateless/02676_distinct_reading_in_order_analyzer.reference index 016202cfb66..bb9cfc5efca 100644 --- a/tests/queries/0_stateless/02676_distinct_reading_in_order_analyzer.reference +++ b/tests/queries/0_stateless/02676_distinct_reading_in_order_analyzer.reference @@ -1 +1 @@ -MergeTreeInOrder +MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) 0 → 1 diff --git a/tests/queries/0_stateless/02676_distinct_reading_in_order_analyzer.sql b/tests/queries/0_stateless/02676_distinct_reading_in_order_analyzer.sql index f00c1322e1d..a280a384c40 100644 --- a/tests/queries/0_stateless/02676_distinct_reading_in_order_analyzer.sql +++ b/tests/queries/0_stateless/02676_distinct_reading_in_order_analyzer.sql @@ -5,4 +5,4 @@ set allow_experimental_analyzer=1; create table t (a UInt64, b UInt64) engine=MergeTree() order by (a); insert into t select number % 2, number from numbers(10); -select splitByChar(' ', trimBoth(explain))[1] from (explain pipeline select distinct a from t) where explain like '%MergeTreeInOrder%'; +select trimBoth(explain) from (explain pipeline select distinct a from t) where explain like '%InOrder%'; diff --git a/tests/queries/0_stateless/02780_final_streams_data_skipping_index.reference b/tests/queries/0_stateless/02780_final_streams_data_skipping_index.reference index 5242c625325..85a66dfb34c 100644 --- a/tests/queries/0_stateless/02780_final_streams_data_skipping_index.reference +++ b/tests/queries/0_stateless/02780_final_streams_data_skipping_index.reference @@ -12,12 +12,12 @@ ExpressionTransform × 2 FilterSortedStreamByRange × 2 Description: filter values in [(999424), +inf) ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 AggregatingSortedTransform FilterSortedStreamByRange Description: filter values in [-inf, (999424)) ExpressionTransform - MergeTreeInOrder 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) 0 → 1 EXPLAIN PIPELINE SELECT * FROM data FINAL WHERE v1 >= now() - INTERVAL 180 DAY SETTINGS max_threads=2, max_final_threads=2, force_data_skipping_indices='v1_index', use_skip_indexes_if_final=0 FORMAT LineAsString; @@ -31,9 +31,9 @@ ExpressionTransform × 2 FilterSortedStreamByRange × 2 Description: filter values in [(999424), +inf) ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 AggregatingSortedTransform FilterSortedStreamByRange Description: filter values in [-inf, (999424)) ExpressionTransform - MergeTreeInOrder 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) 0 → 1 diff --git a/tests/queries/0_stateless/02798_explain_settings_not_applied_bug.reference b/tests/queries/0_stateless/02798_explain_settings_not_applied_bug.reference index 6fc36a0ba01..ffb61b54f9a 100644 --- a/tests/queries/0_stateless/02798_explain_settings_not_applied_bug.reference +++ b/tests/queries/0_stateless/02798_explain_settings_not_applied_bug.reference @@ -1,11 +1,11 @@ - explain + explain - (Expression) - ExpressionTransform - (Aggregating) - FinalizeAggregatedTransform - AggregatingInOrderTransform - (Expression) - ExpressionTransform - (ReadFromMergeTree) - MergeTreeInOrder 0 → 1 + (Expression) + ExpressionTransform + (Aggregating) + FinalizeAggregatedTransform + AggregatingInOrderTransform + (Expression) + ExpressionTransform + (ReadFromMergeTree) + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) 0 → 1 From 7bc5627912312871bfc2a7c5c55b8954f7f9e22f Mon Sep 17 00:00:00 2001 From: Anton Popov Date: Tue, 29 Aug 2023 14:21:58 +0000 Subject: [PATCH 086/433] fix typos --- src/Storages/MergeTree/IMergeTreeReadPool.h | 2 +- src/Storages/MergeTree/MergeTreeReadTask.h | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/src/Storages/MergeTree/IMergeTreeReadPool.h b/src/Storages/MergeTree/IMergeTreeReadPool.h index 31aa61d5032..904933a8410 100644 --- a/src/Storages/MergeTree/IMergeTreeReadPool.h +++ b/src/Storages/MergeTree/IMergeTreeReadPool.h @@ -21,7 +21,7 @@ public: /// Returns true if tasks are returned in the same order as the order of ranges passed to pool virtual bool preservesOrderOfRanges() const = 0; - /// task_idx is an implentation defined identifier that helps + /// task_idx is an implementation defined identifier that helps /// to get required task. E.g. it may be number of thread, index of virtual MergeTreeReadTaskPtr getTask(size_t task_idx, MergeTreeReadTask * previous_task) = 0; virtual void profileFeedback(ReadBufferFromFileBase::ProfileInfo info) = 0; diff --git a/src/Storages/MergeTree/MergeTreeReadTask.h b/src/Storages/MergeTree/MergeTreeReadTask.h index 8bb02a2a869..8d2f0657fd1 100644 --- a/src/Storages/MergeTree/MergeTreeReadTask.h +++ b/src/Storages/MergeTree/MergeTreeReadTask.h @@ -64,7 +64,7 @@ public: AlterConversionsPtr alter_conversions; /// Column names to read during PREWHERE and WHERE Columns task_columns; - /// Shared initialied size predictor. It is copied for each new task. + /// Shared initialized size predictor. It is copied for each new task. MergeTreeBlockSizePredictorPtr shared_size_predictor; }; From 81975807461d0d4f93aabbb432d5370e8e8e7e6c Mon Sep 17 00:00:00 2001 From: Anton Popov Date: Tue, 29 Aug 2023 17:23:52 +0000 Subject: [PATCH 087/433] fix tests and better interface --- .../MergeTree/MergeTreePrefetchedReadPool.cpp | 62 +++++++++---------- .../MergeTree/MergeTreePrefetchedReadPool.h | 26 +++++--- ...1_mergetree_read_in_order_spread.reference | 6 +- .../02404_memory_bound_merging.sql | 2 +- 4 files changed, 50 insertions(+), 46 deletions(-) diff --git a/src/Storages/MergeTree/MergeTreePrefetchedReadPool.cpp b/src/Storages/MergeTree/MergeTreePrefetchedReadPool.cpp index 8e0681bc5e6..d97b674df37 100644 --- a/src/Storages/MergeTree/MergeTreePrefetchedReadPool.cpp +++ b/src/Storages/MergeTree/MergeTreePrefetchedReadPool.cpp @@ -37,22 +37,34 @@ bool MergeTreePrefetchedReadPool::TaskHolder::operator<(const TaskHolder & other return task->priority > other.task->priority; /// Less is better. } -void MergeTreePrefetchedReadPool::ReadersFuture::wait() +MergeTreePrefetchedReadPool::PrefetechedReaders::PrefetechedReaders( + MergeTreeReadTask::Readers readers_, + Priority priority_, + MergeTreePrefetchedReadPool & pool_) + : is_valid(true) + , readers(std::move(readers_)) { - main.wait(); - for (auto & reader_future : prewhere) - reader_future.wait(); + prefetch_futures.push_back(pool_.createPrefetchedFuture(readers.main.get(), priority_)); + + for (const auto & reader : readers.prewhere) + prefetch_futures.push_back(pool_.createPrefetchedFuture(reader.get(), priority_)); } -MergeTreeReadTask::Readers MergeTreePrefetchedReadPool::ReadersFuture::get() +void MergeTreePrefetchedReadPool::PrefetechedReaders::wait() { - MergeTreeReadTask::Readers readers; ProfileEventTimeIncrement watch(ProfileEvents::WaitPrefetchTaskMicroseconds); + for (auto & prefetch_future : prefetch_futures) + prefetch_future.wait(); +} - readers.main = main.get(); - for (auto & reader_future : prewhere) - readers.prewhere.push_back(reader_future.get()); - return readers; +MergeTreeReadTask::Readers MergeTreePrefetchedReadPool::PrefetechedReaders::get() +{ + ProfileEventTimeIncrement watch(ProfileEvents::WaitPrefetchTaskMicroseconds); + for (auto & prefetch_future : prefetch_futures) + prefetch_future.get(); + + is_valid = false; + return std::move(readers); } MergeTreePrefetchedReadPool::MergeTreePrefetchedReadPool( @@ -87,37 +99,22 @@ MergeTreePrefetchedReadPool::MergeTreePrefetchedReadPool( fillPerThreadTasks(pool_settings.threads, pool_settings.sum_marks); } -MergeTreePrefetchedReadPool::~MergeTreePrefetchedReadPool() -{ - for (auto & [_, tasks] : per_thread_tasks) - { - for (auto & task : tasks) - { - if (task->readers_future.valid()) - task->readers_future.wait(); - } - } -} - -std::future MergeTreePrefetchedReadPool::createPrefetchedReader(MergeTreeReaderPtr reader, Priority priority) +std::future MergeTreePrefetchedReadPool::createPrefetchedFuture(IMergeTreeReader * reader, Priority priority) { /// In order to make a prefetch we need to wait for marks to be loaded. But we just created /// a reader (which starts loading marks in its constructor), then if we do prefetch right /// after creating a reader, it will be very inefficient. We can do prefetch for all parts /// only inside this MergeTreePrefetchedReadPool, where read tasks are created and distributed, /// and we cannot block either, therefore make prefetch inside the pool and put the future - /// into the read task (MergeTreeReadTask). When a thread calls getTask(), it will wait for - /// it (if not yet ready) after getting the task. - auto task = [=, my_reader = std::move(reader), context = getContext()]() mutable -> MergeTreeReaderPtr && + /// into the thread task. When a thread calls getTask(), it will wait for it is not ready yet. + auto task = [=, context = getContext()]() mutable { /// For async read metrics in system.query_log. PrefetchIncrement watch(context->getAsyncReadCounters()); - - my_reader->prefetchBeginOfRange(priority); - return std::move(my_reader); + reader->prefetchBeginOfRange(priority); }; - return scheduleFromThreadPool(std::move(task), prefetch_threadpool, "ReadPrepare", priority); + return scheduleFromThreadPool(std::move(task), prefetch_threadpool, "ReadPrepare", priority); } void MergeTreePrefetchedReadPool::createPrefetchedReadersForTask(ThreadTask & task) @@ -127,10 +124,7 @@ void MergeTreePrefetchedReadPool::createPrefetchedReadersForTask(ThreadTask & ta auto extras = getExtras(); auto readers = MergeTreeReadTask::createReaders(task.read_info, extras, task.ranges); - - task.readers_future.main = createPrefetchedReader(std::move(readers.main), task.priority); - for (auto && reader : readers.prewhere) - task.readers_future.prewhere.push_back(createPrefetchedReader(std::move(reader), task.priority)); + task.readers_future = PrefetechedReaders(std::move(readers), task.priority, *this); } void MergeTreePrefetchedReadPool::startPrefetches() diff --git a/src/Storages/MergeTree/MergeTreePrefetchedReadPool.h b/src/Storages/MergeTree/MergeTreePrefetchedReadPool.h index 7cee16c6ff2..f530879f86a 100644 --- a/src/Storages/MergeTree/MergeTreePrefetchedReadPool.h +++ b/src/Storages/MergeTree/MergeTreePrefetchedReadPool.h @@ -27,8 +27,6 @@ public: const PoolSettings & settings_, const ContextPtr & context_); - ~MergeTreePrefetchedReadPool() override; - String getName() const override { return "PrefetchedReadPool"; } bool preservesOrderOfRanges() const override { return false; } MergeTreeReadTaskPtr getTask(size_t task_idx, MergeTreeReadTask * previous_task) override; @@ -50,14 +48,20 @@ private: size_t required_readers_num = 0; }; - struct ReadersFuture + class PrefetechedReaders { + public: + PrefetechedReaders() = default; + PrefetechedReaders(MergeTreeReadTask::Readers readers_, Priority priority_, MergeTreePrefetchedReadPool & pool_); + void wait(); MergeTreeReadTask::Readers get(); - bool valid() const { return main.valid(); } + bool valid() const { return is_valid; } - std::future main; - std::vector> prewhere; + private: + bool is_valid = false; + MergeTreeReadTask::Readers readers; + std::vector> prefetch_futures; }; struct ThreadTask @@ -69,10 +73,16 @@ private: { } + ~ThreadTask() + { + if (readers_future.valid()) + readers_future.wait(); + } + InfoPtr read_info; MarkRanges ranges; Priority priority; - ReadersFuture readers_future; + PrefetechedReaders readers_future; }; struct TaskHolder @@ -92,7 +102,7 @@ private: void startPrefetches(); void createPrefetchedReadersForTask(ThreadTask & task); - std::future createPrefetchedReader(MergeTreeReaderPtr reader, Priority priority); + std::future createPrefetchedFuture(IMergeTreeReader * reader, Priority priority); MergeTreeReadTaskPtr stealTask(size_t thread, MergeTreeReadTask * previous_task); MergeTreeReadTaskPtr createTask(ThreadTask & thread_task, MergeTreeReadTask * previous_task); diff --git a/tests/queries/0_stateless/01551_mergetree_read_in_order_spread.reference b/tests/queries/0_stateless/01551_mergetree_read_in_order_spread.reference index 1d76d9bd631..443f6d3ae93 100644 --- a/tests/queries/0_stateless/01551_mergetree_read_in_order_spread.reference +++ b/tests/queries/0_stateless/01551_mergetree_read_in_order_spread.reference @@ -9,9 +9,9 @@ ExpressionTransform × 3 ExpressionTransform × 3 (ReadFromMergeTree) ExpressionTransform × 4 - MergeTreeInOrder 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) 0 → 1 MergingSortedTransform 2 → 1 ExpressionTransform × 2 - MergeTreeInOrder × 2 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) × 2 0 → 1 ExpressionTransform - MergeTreeInOrder 0 → 1 + MergeTreeSelect(pool: ReadPoolInOrder, algorithm: InOrder) 0 → 1 diff --git a/tests/queries/0_stateless/02404_memory_bound_merging.sql b/tests/queries/0_stateless/02404_memory_bound_merging.sql index a9577aa4222..5e017e79309 100644 --- a/tests/queries/0_stateless/02404_memory_bound_merging.sql +++ b/tests/queries/0_stateless/02404_memory_bound_merging.sql @@ -59,7 +59,7 @@ select a, count() from dist_t_different_dbs group by a, b order by a limit 5 off drop table if exists pr_t; -create table pr_t(a UInt64, b UInt64) engine=MergeTree order by a settings storage_policy = 's3'; +create table pr_t(a UInt64, b UInt64) engine=MergeTree order by a; insert into pr_t select number % 1000, number % 1000 from numbers_mt(1e6); set allow_experimental_parallel_reading_from_replicas = 1; From 11c94246c774217bbf3b64b8e6166e22b8f0b3ab Mon Sep 17 00:00:00 2001 From: Duc Canh Le Date: Wed, 30 Aug 2023 05:05:57 +0000 Subject: [PATCH 088/433] do not implicitly read pk + version and compute sort expression for lonely parts Signed-off-by: Duc Canh Le --- .../QueryPlan/ReadFromMergeTree.cpp | 115 ++++++---- ...ross_partitions_final_all_lonely.reference | 21 -- ...rge_across_partitions_final_all_lonely.sql | 32 --- ...oss_partitions_final_with_lonely.reference | 202 ++++++++++++++++++ ...ge_across_partitions_final_with_lonely.sql | 49 +++++ 5 files changed, 321 insertions(+), 98 deletions(-) delete mode 100644 tests/queries/0_stateless/02868_no_merge_across_partitions_final_all_lonely.reference delete mode 100644 tests/queries/0_stateless/02868_no_merge_across_partitions_final_all_lonely.sql create mode 100644 tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.reference create mode 100644 tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.sql diff --git a/src/Processors/QueryPlan/ReadFromMergeTree.cpp b/src/Processors/QueryPlan/ReadFromMergeTree.cpp index 02b3d7a07fa..e868cb3845a 100644 --- a/src/Processors/QueryPlan/ReadFromMergeTree.cpp +++ b/src/Processors/QueryPlan/ReadFromMergeTree.cpp @@ -982,7 +982,7 @@ Pipe ReadFromMergeTree::spreadMarkRangesAmongStreamsFinal( RangesInDataParts && parts_with_ranges, size_t num_streams, const Names & origin_column_names, const Names & column_names, ActionsDAGPtr & out_projection) { const auto & settings = context->getSettingsRef(); - const auto data_settings = data.getSettings(); + const auto & data_settings = data.getSettings(); PartRangesReadInfo info(parts_with_ranges, settings, *data_settings); @@ -1017,7 +1017,8 @@ Pipe ReadFromMergeTree::spreadMarkRangesAmongStreamsFinal( parts_to_merge_ranges.push_back(parts_with_ranges.end()); } - Pipes partition_pipes; + Pipes merging_pipes; + Pipes no_merging_pipes; /// If do_not_merge_across_partitions_select_final is true and num_streams > 1 /// we will store lonely parts with level > 0 to use parallel select on them. @@ -1028,20 +1029,21 @@ Pipe ReadFromMergeTree::spreadMarkRangesAmongStreamsFinal( for (size_t range_index = 0; range_index < parts_to_merge_ranges.size() - 1; ++range_index) { + /// If do_not_merge_across_partitions_select_final is true and there is only one part in partition + /// with level > 0 then we won't post-process this part, and if num_streams > 1 we + /// can use parallel select on such parts. + bool no_merging_final = settings.do_not_merge_across_partitions_select_final && + std::distance(parts_to_merge_ranges[range_index], parts_to_merge_ranges[range_index + 1]) == 1 && + parts_to_merge_ranges[range_index]->data_part->info.level > 0 && + data.merging_params.is_deleted_column.empty(); Pipes pipes; { RangesInDataParts new_parts; - /// If do_not_merge_across_partitions_select_final is true and there is only one part in partition - /// with level > 0 then we won't postprocess this part and if num_streams > 1 we - /// can use parallel select on such parts. We save such parts in one vector and then use - /// MergeTreeReadPool and MergeTreeThreadSelectProcessor for parallel select. - if (num_streams > 1 && settings.do_not_merge_across_partitions_select_final && - std::distance(parts_to_merge_ranges[range_index], parts_to_merge_ranges[range_index + 1]) == 1 && - parts_to_merge_ranges[range_index]->data_part->info.level > 0 - && data.merging_params.is_deleted_column.empty()) + if (no_merging_final) { - sum_marks_in_lonely_parts += parts_to_merge_ranges[range_index]->getMarksCount(); + if (num_streams > 1) + sum_marks_in_lonely_parts += parts_to_merge_ranges[range_index]->getMarksCount(); lonely_parts.push_back(std::move(*parts_to_merge_ranges[range_index])); continue; } @@ -1091,16 +1093,6 @@ Pipe ReadFromMergeTree::spreadMarkRangesAmongStreamsFinal( out_projection = createProjection(pipes.front().getHeader()); } - /// If do_not_merge_across_partitions_select_final is true and there is only one part in partition - /// with level > 0 then we won't postprocess this part - if (settings.do_not_merge_across_partitions_select_final && - std::distance(parts_to_merge_ranges[range_index], parts_to_merge_ranges[range_index + 1]) == 1 && - parts_to_merge_ranges[range_index]->data_part->info.level > 0 && - data.merging_params.is_deleted_column.empty()) - { - partition_pipes.emplace_back(Pipe::unitePipes(std::move(pipes))); - continue; - } Names sort_columns = metadata_for_reading->getSortingKeyColumns(); SortDescription sort_description; @@ -1123,39 +1115,72 @@ Pipe ReadFromMergeTree::spreadMarkRangesAmongStreamsFinal( partition_key_columns, max_block_size); - partition_pipes.emplace_back(Pipe::unitePipes(std::move(pipes))); + merging_pipes.emplace_back(Pipe::unitePipes(std::move(pipes))); } if (!lonely_parts.empty()) { - size_t num_streams_for_lonely_parts = num_streams * lonely_parts.size(); + Pipe pipe; + if (num_streams > 1) + { + size_t num_streams_for_lonely_parts = num_streams * lonely_parts.size(); - const size_t min_marks_for_concurrent_read = MergeTreeDataSelectExecutor::minMarksForConcurrentRead( - settings.merge_tree_min_rows_for_concurrent_read, - settings.merge_tree_min_bytes_for_concurrent_read, - data_settings->index_granularity, - info.index_granularity_bytes, - sum_marks_in_lonely_parts); + const size_t min_marks_for_concurrent_read = MergeTreeDataSelectExecutor::minMarksForConcurrentRead( + settings.merge_tree_min_rows_for_concurrent_read, + settings.merge_tree_min_bytes_for_concurrent_read, + data_settings->index_granularity, + info.index_granularity_bytes, + sum_marks_in_lonely_parts); - /// Reduce the number of num_streams_for_lonely_parts if the data is small. - if (sum_marks_in_lonely_parts < num_streams_for_lonely_parts * min_marks_for_concurrent_read && lonely_parts.size() < num_streams_for_lonely_parts) - num_streams_for_lonely_parts = std::max((sum_marks_in_lonely_parts + min_marks_for_concurrent_read - 1) / min_marks_for_concurrent_read, lonely_parts.size()); + /// Reduce the number of num_streams_for_lonely_parts if the data is small. + if (sum_marks_in_lonely_parts < num_streams_for_lonely_parts * min_marks_for_concurrent_read + && lonely_parts.size() < num_streams_for_lonely_parts) + num_streams_for_lonely_parts = std::max( + (sum_marks_in_lonely_parts + min_marks_for_concurrent_read - 1) / min_marks_for_concurrent_read, + lonely_parts.size()); - auto pipe = read(std::move(lonely_parts), partition_pipes.empty() ? origin_column_names : column_names, ReadFromMergeTree::ReadType::Default, - num_streams_for_lonely_parts, min_marks_for_concurrent_read, info.use_uncompressed_cache); - - /// Drop temporary columns, added by 'sorting_key_expr' - if (!out_projection) - out_projection = createProjection(pipe.getHeader()); - - if (!partition_pipes.empty()) - pipe.addSimpleTransform([sorting_expr](const Block & header) - { return std::make_shared(header, sorting_expr); }); - - partition_pipes.emplace_back(std::move(pipe)); + pipe = read( + std::move(lonely_parts), + origin_column_names, + ReadFromMergeTree::ReadType::Default, + num_streams_for_lonely_parts, + min_marks_for_concurrent_read, + info.use_uncompressed_cache); + } + else + { + pipe = read( + std::move(lonely_parts), + origin_column_names, + ReadFromMergeTree::ReadType::InOrder, + num_streams, + 0, + info.use_uncompressed_cache); + } + no_merging_pipes.emplace_back(std::move(pipe)); } - return Pipe::unitePipes(std::move(partition_pipes)); + if (!merging_pipes.empty() && !no_merging_pipes.empty()) + { + out_projection = nullptr; /// We do projection here + Pipes pipes; + pipes.resize(2); + pipes[0] = Pipe::unitePipes(std::move(merging_pipes)); + pipes[1] = Pipe::unitePipes(std::move(no_merging_pipes)); + auto conversion_action = ActionsDAG::makeConvertingActions( + pipes[0].getHeader().getColumnsWithTypeAndName(), + pipes[1].getHeader().getColumnsWithTypeAndName(), + ActionsDAG::MatchColumnsMode::Name); + pipes[0].addSimpleTransform( + [conversion_action](const Block & header) + { + auto converting_expr = std::make_shared(conversion_action); + return std::make_shared(header, converting_expr); + }); + return Pipe::unitePipes(std::move(pipes)); + } + else + return merging_pipes.empty() ? Pipe::unitePipes(std::move(no_merging_pipes)) : Pipe::unitePipes(std::move(merging_pipes)); } MergeTreeDataSelectAnalysisResultPtr ReadFromMergeTree::selectRangesToRead( diff --git a/tests/queries/0_stateless/02868_no_merge_across_partitions_final_all_lonely.reference b/tests/queries/0_stateless/02868_no_merge_across_partitions_final_all_lonely.reference deleted file mode 100644 index 9f4677b0f16..00000000000 --- a/tests/queries/0_stateless/02868_no_merge_across_partitions_final_all_lonely.reference +++ /dev/null @@ -1,21 +0,0 @@ -(Expression) -ExpressionTransform × 16 -Header: max(val) UInt64: max(val) UInt64 UInt64(size = 0) - count() UInt64: count() UInt64 UInt64(size = 0) - (Aggregating) - Resize 4 → 16 - Header × 16 : max(val) UInt64: max(val) UInt64 UInt64(size = 0) - count() UInt64: count() UInt64 UInt64(size = 0) - AggregatingTransform × 4 - Header: max(val) UInt64: max(val) UInt64 UInt64(size = 0) - count() UInt64: count() UInt64 UInt64(size = 0) - StrictResize 4 → 4 - Header × 4 : val UInt64: val UInt64 UInt64(size = 0) - (Expression) - ExpressionTransform × 4 - Header: val UInt64: val UInt64 UInt64(size = 0) - (ReadFromMergeTree) - ExpressionTransform × 4 - Header: val UInt64: val UInt64 UInt64(size = 0) - MergeTreeThread × 4 0 → 1 - Header: val UInt64: val UInt64 UInt64(size = 0) diff --git a/tests/queries/0_stateless/02868_no_merge_across_partitions_final_all_lonely.sql b/tests/queries/0_stateless/02868_no_merge_across_partitions_final_all_lonely.sql deleted file mode 100644 index d2c40bb002a..00000000000 --- a/tests/queries/0_stateless/02868_no_merge_across_partitions_final_all_lonely.sql +++ /dev/null @@ -1,32 +0,0 @@ -DROP TABLE IF EXISTS all_lonely; - -CREATE TABLE all_lonely -( - `id` UInt64, - `dt` Date, - `val` UInt64, - `version` UInt64 -) -ENGINE = ReplacingMergeTree(version) -PARTITION BY dt -ORDER BY (id); - -INSERT INTO all_lonely SELECT number, '2022-10-28', number*10, 0 FROM numbers(10000); -INSERT INTO all_lonely SELECT number+500000, '2022-10-28', number*10, 1 FROM numbers(10000); -OPTIMIZE TABLE all_lonely PARTITION '2022-10-28' FINAL; - - -INSERT INTO all_lonely SELECT number, '2022-10-29', number*10, 0 FROM numbers(10000); -INSERT INTO all_lonely SELECT number+500000, '2022-10-29', number*10, 1 FROM numbers(10000); -OPTIMIZE TABLE all_lonely PARTITION '2022-10-29' FINAL; - -INSERT INTO all_lonely SELECT number, '2022-10-30', number*10, 0 FROM numbers(10000); -INSERT INTO all_lonely SELECT number+500000, '2022-10-30', number*10, 1 FROM numbers(10000); -OPTIMIZE TABLE all_lonely PARTITION '2022-10-30' FINAL; - - -INSERT INTO all_lonely SELECT number, '2022-10-31', number*10, 0 FROM numbers(10000); -INSERT INTO all_lonely SELECT number+500000, '2022-10-31', number*10, 1 FROM numbers(10000); -OPTIMIZE TABLE all_lonely PARTITION '2022-10-31' FINAL; - -EXPLAIN PIPELINE header=1 SELECT max(val), count(*) FROM all_lonely FINAL SETTINGS do_not_merge_across_partitions_select_final = 1, max_threads = 16; diff --git a/tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.reference b/tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.reference new file mode 100644 index 00000000000..a4b37bc423b --- /dev/null +++ b/tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.reference @@ -0,0 +1,202 @@ +(Expression) +ExpressionTransform × 16 +Header: max(val) UInt64: max(val) UInt64 UInt64(size = 0) + count() UInt64: count() UInt64 UInt64(size = 0) + (Aggregating) + Resize 6 → 16 + Header × 16 : max(val) UInt64: max(val) UInt64 UInt64(size = 0) + count() UInt64: count() UInt64 UInt64(size = 0) + AggregatingTransform × 6 + Header: max(val) UInt64: max(val) UInt64 UInt64(size = 0) + count() UInt64: count() UInt64 UInt64(size = 0) + StrictResize 6 → 6 + Header × 6 : val UInt64: val UInt64 UInt64(size = 0) + (Expression) + ExpressionTransform × 6 + Header: val UInt64: val UInt64 UInt64(size = 0) + (ReadFromMergeTree) + MergeTreeThread × 4 0 → 1 + Header: val UInt64: val UInt64 UInt64(size = 0) + ExpressionTransform × 2 + Header: val UInt64: val UInt64 UInt64(size = 0) + ReplacingSorted 2 → 1 + Header: id UInt64: id UInt64 UInt64(size = 0) + val UInt64: val UInt64 UInt64(size = 0) + version UInt64: version UInt64 UInt64(size = 0) + FilterSortedStreamByRange × 2 + Header: id UInt64: id UInt64 UInt64(size = 0) + val UInt64: val UInt64 UInt64(size = 0) + version UInt64: version UInt64 UInt64(size = 0) + Description: filter values in [(50000), +inf) + ExpressionTransform × 2 + Header: id UInt64: id UInt64 UInt64(size = 0) + val UInt64: val UInt64 UInt64(size = 0) + version UInt64: version UInt64 UInt64(size = 0) + MergeTreeInOrder × 2 0 → 1 + Header: id UInt64: id UInt64 UInt64(size = 0) + val UInt64: val UInt64 UInt64(size = 0) + version UInt64: version UInt64 UInt64(size = 0) + ReplacingSorted 2 → 1 + Header: id UInt64: id UInt64 UInt64(size = 0) + val UInt64: val UInt64 UInt64(size = 0) + version UInt64: version UInt64 UInt64(size = 0) + FilterSortedStreamByRange × 2 + Header: id UInt64: id UInt64 UInt64(size = 0) + val UInt64: val UInt64 UInt64(size = 0) + version UInt64: version UInt64 UInt64(size = 0) + Description: filter values in [-inf, (50000)) + ExpressionTransform × 2 + Header: id UInt64: id UInt64 UInt64(size = 0) + val UInt64: val UInt64 UInt64(size = 0) + version UInt64: version UInt64 UInt64(size = 0) + MergeTreeInOrder × 2 0 → 1 + Header: id UInt64: id UInt64 UInt64(size = 0) + val UInt64: val UInt64 UInt64(size = 0) + version UInt64: version UInt64 UInt64(size = 0) +(Expression) +ExpressionTransform × 16 +Header: max(val) UInt64: max(val) UInt64 UInt64(size = 0) + count() UInt64: count() UInt64 UInt64(size = 0) + (Aggregating) + Resize 4 → 16 + Header × 16 : max(val) UInt64: max(val) UInt64 UInt64(size = 0) + count() UInt64: count() UInt64 UInt64(size = 0) + AggregatingTransform × 4 + Header: max(val) UInt64: max(val) UInt64 UInt64(size = 0) + count() UInt64: count() UInt64 UInt64(size = 0) + StrictResize 4 → 4 + Header × 4 : val UInt64: val UInt64 UInt64(size = 0) + (Expression) + ExpressionTransform × 4 + Header: val UInt64: val UInt64 UInt64(size = 0) + (Filter) + FilterTransform × 4 + Header: val UInt64: val UInt64 UInt64(size = 0) + (ReadFromMergeTree) + MergeTreeThread × 4 0 → 1 + Header: dt Date: dt Date UInt16(size = 0) + val UInt64: val UInt64 UInt64(size = 0) +(Expression) +ExpressionTransform +Header: max(val) UInt64: max(val) UInt64 UInt64(size = 0) + count() UInt64: count() UInt64 UInt64(size = 0) + (Aggregating) + Resize 4 → 1 + Header: max(val) UInt64: max(val) UInt64 UInt64(size = 0) + count() UInt64: count() UInt64 UInt64(size = 0) + AggregatingTransform × 4 + Header: max(val) UInt64: max(val) UInt64 UInt64(size = 0) + count() UInt64: count() UInt64 UInt64(size = 0) + StrictResize 4 → 4 + Header × 4 : val UInt64: val UInt64 UInt64(size = 0) + (Expression) + ExpressionTransform × 4 + Header: val UInt64: val UInt64 UInt64(size = 0) + (Filter) + FilterTransform × 4 + Header: val UInt64: val UInt64 UInt64(size = 0) + (ReadFromMergeTree) + MergeTreeInOrder × 4 0 → 1 + Header: dt Date: dt Date UInt16(size = 0) + val UInt64: val UInt64 UInt64(size = 0) +(Expression) +ExpressionTransform × 16 +Header: max(val) UInt64: max(val) UInt64 UInt64(size = 0) + count() UInt64: count() UInt64 UInt64(size = 0) + (Aggregating) + Resize 6 → 16 + Header × 16 : max(val_0) UInt64: max(val_0) UInt64 UInt64(size = 0) + count() UInt64: count() UInt64 UInt64(size = 0) + AggregatingTransform × 6 + Header: max(val_0) UInt64: max(val_0) UInt64 UInt64(size = 0) + count() UInt64: count() UInt64 UInt64(size = 0) + StrictResize 6 → 6 + Header × 6 : val_0 UInt64: val_0 UInt64 UInt64(size = 0) + (Expression) + ExpressionTransform × 6 + Header: val_0 UInt64: val_0 UInt64 UInt64(size = 0) + (ReadFromMergeTree) + MergeTreeThread × 4 0 → 1 + Header: val UInt64: val UInt64 UInt64(size = 0) + ExpressionTransform × 2 + Header: val UInt64: val UInt64 UInt64(size = 0) + ReplacingSorted 2 → 1 + Header: id UInt64: id UInt64 UInt64(size = 0) + val UInt64: val UInt64 UInt64(size = 0) + version UInt64: version UInt64 UInt64(size = 0) + FilterSortedStreamByRange × 2 + Header: id UInt64: id UInt64 UInt64(size = 0) + val UInt64: val UInt64 UInt64(size = 0) + version UInt64: version UInt64 UInt64(size = 0) + Description: filter values in [(50000), +inf) + ExpressionTransform × 2 + Header: id UInt64: id UInt64 UInt64(size = 0) + val UInt64: val UInt64 UInt64(size = 0) + version UInt64: version UInt64 UInt64(size = 0) + MergeTreeInOrder × 2 0 → 1 + Header: id UInt64: id UInt64 UInt64(size = 0) + val UInt64: val UInt64 UInt64(size = 0) + version UInt64: version UInt64 UInt64(size = 0) + ReplacingSorted 2 → 1 + Header: id UInt64: id UInt64 UInt64(size = 0) + val UInt64: val UInt64 UInt64(size = 0) + version UInt64: version UInt64 UInt64(size = 0) + FilterSortedStreamByRange × 2 + Header: id UInt64: id UInt64 UInt64(size = 0) + val UInt64: val UInt64 UInt64(size = 0) + version UInt64: version UInt64 UInt64(size = 0) + Description: filter values in [-inf, (50000)) + ExpressionTransform × 2 + Header: id UInt64: id UInt64 UInt64(size = 0) + val UInt64: val UInt64 UInt64(size = 0) + version UInt64: version UInt64 UInt64(size = 0) + MergeTreeInOrder × 2 0 → 1 + Header: id UInt64: id UInt64 UInt64(size = 0) + val UInt64: val UInt64 UInt64(size = 0) + version UInt64: version UInt64 UInt64(size = 0) +(Expression) +ExpressionTransform × 16 +Header: max(val) UInt64: max(val) UInt64 UInt64(size = 0) + count() UInt64: count() UInt64 UInt64(size = 0) + (Aggregating) + Resize 4 → 16 + Header × 16 : max(val_0) UInt64: max(val_0) UInt64 UInt64(size = 0) + count() UInt64: count() UInt64 UInt64(size = 0) + AggregatingTransform × 4 + Header: max(val_0) UInt64: max(val_0) UInt64 UInt64(size = 0) + count() UInt64: count() UInt64 UInt64(size = 0) + StrictResize 4 → 4 + Header × 4 : val_0 UInt64: val_0 UInt64 UInt64(size = 0) + (Expression) + ExpressionTransform × 4 + Header: val_0 UInt64: val_0 UInt64 UInt64(size = 0) + (Filter) + FilterTransform × 4 + Header: val_0 UInt64: val_0 UInt64 UInt64(size = 0) + (ReadFromMergeTree) + MergeTreeThread × 4 0 → 1 + Header: val UInt64: val UInt64 UInt64(size = 0) + dt Date: dt Date UInt16(size = 0) +(Expression) +ExpressionTransform +Header: max(val) UInt64: max(val) UInt64 UInt64(size = 0) + count() UInt64: count() UInt64 UInt64(size = 0) + (Aggregating) + Resize 4 → 1 + Header: max(val_0) UInt64: max(val_0) UInt64 UInt64(size = 0) + count() UInt64: count() UInt64 UInt64(size = 0) + AggregatingTransform × 4 + Header: max(val_0) UInt64: max(val_0) UInt64 UInt64(size = 0) + count() UInt64: count() UInt64 UInt64(size = 0) + StrictResize 4 → 4 + Header × 4 : val_0 UInt64: val_0 UInt64 UInt64(size = 0) + (Expression) + ExpressionTransform × 4 + Header: val_0 UInt64: val_0 UInt64 UInt64(size = 0) + (Filter) + FilterTransform × 4 + Header: val_0 UInt64: val_0 UInt64 UInt64(size = 0) + (ReadFromMergeTree) + MergeTreeInOrder × 4 0 → 1 + Header: val UInt64: val UInt64 UInt64(size = 0) + dt Date: dt Date UInt16(size = 0) diff --git a/tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.sql b/tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.sql new file mode 100644 index 00000000000..526c9204d96 --- /dev/null +++ b/tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.sql @@ -0,0 +1,49 @@ +DROP TABLE IF EXISTS with_lonely; + +CREATE TABLE with_lonely +( + `id` UInt64, + `dt` Date, + `val` UInt64, + `version` UInt64 +) +ENGINE = ReplacingMergeTree(version) +PARTITION BY dt +ORDER BY (id); + +INSERT INTO with_lonely SELECT number, '2022-10-28', number*10, 0 FROM numbers(10000); +INSERT INTO with_lonely SELECT number+500000, '2022-10-28', number*10, 1 FROM numbers(10000); +OPTIMIZE TABLE with_lonely PARTITION '2022-10-28' FINAL; + + +INSERT INTO with_lonely SELECT number, '2022-10-29', number*10, 0 FROM numbers(10000); +INSERT INTO with_lonely SELECT number+500000, '2022-10-29', number*10, 1 FROM numbers(10000); +OPTIMIZE TABLE with_lonely PARTITION '2022-10-29' FINAL; + +INSERT INTO with_lonely SELECT number, '2022-10-30', number*10, 0 FROM numbers(10000); +INSERT INTO with_lonely SELECT number+500000, '2022-10-30', number*10, 1 FROM numbers(10000); +OPTIMIZE TABLE with_lonely PARTITION '2022-10-30' FINAL; + + +INSERT INTO with_lonely SELECT number, '2022-10-31', number*10, 0 FROM numbers(10000); +INSERT INTO with_lonely SELECT number+500000, '2022-10-31', number*10, 1 FROM numbers(10000); +OPTIMIZE TABLE with_lonely PARTITION '2022-10-31' FINAL; + +SYSTEM STOP MERGES with_lonely; +INSERT INTO with_lonely SELECT number, '2022-11-01', number*10, 0 FROM numbers(1000); +INSERT INTO with_lonely SELECT number+50000, '2022-11-01', number*10, 1 FROM numbers(1000); +INSERT INTO with_lonely SELECT number+60000, '2022-11-01', number*10, 2 FROM numbers(1000); + +SET do_not_merge_across_partitions_select_final = 1, max_threads = 16; + +-- mix lonely parts and non-lonely parts +EXPLAIN PIPELINE header=1 SELECT max(val), count(*) FROM with_lonely FINAL SETTINGS allow_experimental_analyzer = 0; +-- only lonely parts +EXPLAIN PIPELINE header=1 SELECT max(val), count(*) FROM with_lonely FINAL WHERE dt < '2022-11-01' SETTINGS allow_experimental_analyzer = 0; +-- only lonely parts but max_thread = 1, so reading lonely parts with in-order +EXPLAIN PIPELINE header=1 SELECT max(val), count(*) FROM with_lonely FINAL WHERE dt < '2022-11-01' SETTINGS max_threads = 1, allow_experimental_analyzer = 0; + + +EXPLAIN PIPELINE header=1 SELECT max(val), count(*) FROM with_lonely FINAL SETTINGS allow_experimental_analyzer = 1; +EXPLAIN PIPELINE header=1 SELECT max(val), count(*) FROM with_lonely FINAL WHERE dt < '2022-11-01' SETTINGS allow_experimental_analyzer = 1; +EXPLAIN PIPELINE header=1 SELECT max(val), count(*) FROM with_lonely FINAL WHERE dt < '2022-11-01' SETTINGS max_threads = 1, allow_experimental_analyzer = 1; From 7618b2d112c1de743c7fc1f778991e01114aa72b Mon Sep 17 00:00:00 2001 From: vdimir Date: Wed, 30 Aug 2023 12:05:04 +0200 Subject: [PATCH 089/433] Add partial_result_update_duration_ms to settings.md --- docs/en/operations/settings/settings.md | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/docs/en/operations/settings/settings.md b/docs/en/operations/settings/settings.md index 4d4ff5fc235..3163ffc5ac3 100644 --- a/docs/en/operations/settings/settings.md +++ b/docs/en/operations/settings/settings.md @@ -4643,3 +4643,11 @@ SELECT toFloat64('1.7091'), toFloat64('1.5008753E7') SETTINGS precise_float_pars │ 1.7091 │ 15008753 │ └─────────────────────┴──────────────────────────┘ ``` + +## partial_result_update_duration_ms + +Interval (in milliseconds) for sending updates with partial data about the result table to the client (in interactive mode) during query execution. Setting to 0 disables partial results. Only supported for single-threaded GROUP BY without key, ORDER BY, LIMIT and OFFSET. + +## max_rows_in_partial_result + +Maximum rows to show in the partial result after every real-time update while the query runs (use partial result limit + OFFSET as a value in case of OFFSET in the query). From b7438e99107e84969d82a2cfe850083e381fbfb0 Mon Sep 17 00:00:00 2001 From: Anton Popov Date: Wed, 30 Aug 2023 15:53:27 +0000 Subject: [PATCH 090/433] fix test --- tests/queries/1_stateful/00177_memory_bound_merging.sh | 5 +---- 1 file changed, 1 insertion(+), 4 deletions(-) diff --git a/tests/queries/1_stateful/00177_memory_bound_merging.sh b/tests/queries/1_stateful/00177_memory_bound_merging.sh index 2c531b064db..ce889b338d6 100755 --- a/tests/queries/1_stateful/00177_memory_bound_merging.sh +++ b/tests/queries/1_stateful/00177_memory_bound_merging.sh @@ -8,9 +8,6 @@ CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) check_replicas_read_in_order() { - # to check this we actually look for at least one log message from MergeTreeInOrderSelectProcessor. - # hopefully logger's names are a bit more stable than log messages itself - # # NOTE: lack of "current_database = '$CLICKHOUSE_DATABASE'" filter is made on purpose $CLICKHOUSE_CLIENT -nq " SYSTEM FLUSH LOGS; @@ -18,7 +15,7 @@ check_replicas_read_in_order() { SELECT COUNT() > 0 FROM system.text_log WHERE query_id IN (SELECT query_id FROM system.query_log WHERE query_id != '$1' AND initial_query_id = '$1' AND event_date >= yesterday()) - AND event_date >= yesterday() AND logger_name = 'MergeTreeInOrderSelectProcessor'" + AND event_date >= yesterday() AND message ILIKE '%Reading%ranges in order%'" } # replicas should use reading in order following initiator's decision to execute aggregation in order. From f4dd6be8a861e13b63e7de4cc8013d4c60a2bffe Mon Sep 17 00:00:00 2001 From: slvrtrn Date: Wed, 30 Aug 2023 18:59:39 +0200 Subject: [PATCH 091/433] WIP prepared statements --- .../PreparedStatementsTest.java | 155 +++++++++++++ src/Core/MySQL/PacketsProtocolBinary.cpp | 60 +++-- src/Core/MySQL/PacketsProtocolBinary.h | 4 +- src/Core/MySQL/PacketsProtocolText.cpp | 33 ++- src/Core/MySQL/PacketsProtocolText.h | 209 ++++++++++-------- .../Formats/Impl/MySQLOutputFormat.cpp | 2 +- tests/integration/requirements.txt | 30 +++ .../prepared_statements.reference | 1 + tests/integration/test_mysql_protocol/test.py | 122 ++++++++-- 9 files changed, 478 insertions(+), 138 deletions(-) create mode 100644 docker/test/integration/mysql_java_client/PreparedStatementsTest.java create mode 100644 tests/integration/requirements.txt create mode 100644 tests/integration/test_mysql_protocol/prepared_statements.reference diff --git a/docker/test/integration/mysql_java_client/PreparedStatementsTest.java b/docker/test/integration/mysql_java_client/PreparedStatementsTest.java new file mode 100644 index 00000000000..0534f2b761d --- /dev/null +++ b/docker/test/integration/mysql_java_client/PreparedStatementsTest.java @@ -0,0 +1,155 @@ +import com.mysql.cj.MysqlType; + +import java.sql.*; + +public class PreparedStatementsTest { + public static void main(String[] args) { + int i = 0; + String host = "127.0.0.1"; + String port = "9004"; + String user = "default"; + String password = ""; + String database = "default"; + while (i < args.length) { + switch (args[i]) { + case "--host": + host = args[++i]; + break; + case "--port": + port = args[++i]; + break; + case "--user": + user = args[++i]; + break; + case "--password": + password = args[++i]; + break; + case "--database": + database = args[++i]; + break; + default: + i++; + break; + } + } + + try { + Class.forName("com.mysql.cj.jdbc.Driver"); + /* + * useServerPrepStmts allows us to send COM_STMT_PREPARE and COM_STMT_EXECUTE to + * test the binary protocol + */ + String jdbcUrl = String.format("jdbc:mysql://%s:%s/%s?useSSL=false&useServerPrepStmts=true", + host, port, database); + Connection conn = DriverManager.getConnection(jdbcUrl, user, password); + testSimpleDataTypes(conn); + testStringTypes(conn); + testDecimalTypes(conn); + testMiscTypes(conn); + // testDateTypes(conn); + conn.close(); + } catch (Exception e) { + e.printStackTrace(); + System.exit(1); + } + } + + private static void testSimpleDataTypes(Connection conn) throws SQLException { + System.out.println("### testSimpleDataTypes"); + ResultSet rs = conn.prepareStatement("SELECT * FROM ps_simple_data_types").executeQuery(); + int rowNum = 1; + while (rs.next()) { + System.out.printf("Row #%d\n", rowNum++); + System.out.printf("Type: %s, value: %d\n", getMysqlType(rs, "i8"), rs.getInt("i8")); + System.out.printf("Type: %s, value: %d\n", getMysqlType(rs, "i16"), rs.getInt("i16")); + System.out.printf("Type: %s, value: %d\n", getMysqlType(rs, "i32"), rs.getInt("i32")); + System.out.printf("Type: %s, value: %d\n", getMysqlType(rs, "i64"), rs.getLong("i64")); + System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "i128"), rs.getString("i128")); + System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "i256"), rs.getString("i256")); + System.out.printf("Type: %s, value: %d\n", getMysqlType(rs, "ui8"), rs.getInt("ui8")); + System.out.printf("Type: %s, value: %d\n", getMysqlType(rs, "ui16"), rs.getInt("ui16")); + System.out.printf("Type: %s, value: %d\n", getMysqlType(rs, "ui32"), rs.getLong("ui32")); + System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "ui64"), rs.getString("ui64")); + System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "ui128"), rs.getString("ui128")); + System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "ui256"), rs.getString("ui256")); + System.out.printf("Type: %s, value: %f\n", getMysqlType(rs, "f32"), rs.getFloat("f32")); + System.out.printf("Type: %s, value: %f\n", getMysqlType(rs, "f64"), rs.getFloat("f64")); + System.out.printf("Type: %s, value: %b\n", getMysqlType(rs, "b"), rs.getBoolean("b")); + } + System.out.println(); + } + + private static void testStringTypes(Connection conn) throws SQLException { + System.out.println("### testStringTypes"); + ResultSet rs = conn.prepareStatement("SELECT * FROM ps_string_types").executeQuery(); + int rowNum = 1; + while (rs.next()) { + System.out.printf("Row #%d\n", rowNum++); + System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "s"), rs.getString("s")); + System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "sn"), rs.getString("sn")); + System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "lc"), rs.getString("lc")); + System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "nlc"), rs.getString("nlc")); + } + System.out.println(); + } + + private static void testDecimalTypes(Connection conn) throws SQLException { + System.out.println("### testDecimalTypes"); + ResultSet rs = conn.prepareStatement("SELECT * FROM ps_decimal_types").executeQuery(); + int rowNum = 1; + while (rs.next()) { + System.out.printf("Row #%d\n", rowNum++); + System.out.printf("Type: %s, value: %s\n", + getMysqlType(rs, "d32"), + rs.getBigDecimal("d32").toPlainString()); + System.out.printf("Type: %s, value: %s\n", + getMysqlType(rs, "d64"), + rs.getBigDecimal("d64").toPlainString()); + System.out.printf("Type: %s, value: %s\n", + getMysqlType(rs, "d128_native"), + rs.getBigDecimal("d128_native").toPlainString()); + System.out.printf("Type: %s, value: %s\n", + getMysqlType(rs, "d128_text"), + rs.getString("d128_text")); + System.out.printf("Type: %s, value: %s\n", + getMysqlType(rs, "d256"), + rs.getString("d256")); + } + System.out.println(); + } + + private static void testDateTypes(Connection conn) throws SQLException { + System.out.println("### testDateTypes"); + ResultSet rs = conn.prepareStatement("SELECT * FROM ps_date_types").executeQuery(); + int rowNum = 1; + while (rs.next()) { + System.out.printf("Row #%d\n", rowNum++); + System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "d"), rs.getDate("d")); + System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "d32"), rs.getDate("d32")); + System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "dt"), rs.getTimestamp("dt")); + System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "dt64"), rs.getTimestamp("dt64")); + } + System.out.println(); + } + + private static void testMiscTypes(Connection conn) throws SQLException { + System.out.println("### testMiscTypes"); + ResultSet rs = conn.prepareStatement("SELECT d FROM ps_misc_types").executeQuery(); + int rowNum = 1; + while (rs.next()) { + System.out.printf("Row #%d\n", rowNum++); + System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "a"), rs.getString("a")); + System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "u"), rs.getString("u")); + System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "t"), rs.getString("t")); + System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "m"), rs.getString("m")); + } + System.out.println(); + } + + private static MysqlType getMysqlType(ResultSet rs, String columnLabel) throws SQLException { + ResultSetMetaData meta = rs.getMetaData(); + return MysqlType.getByJdbcType(meta.getColumnType(rs.findColumn(columnLabel))); + + } + +} diff --git a/src/Core/MySQL/PacketsProtocolBinary.cpp b/src/Core/MySQL/PacketsProtocolBinary.cpp index 287dda269e6..2b5dce78ab3 100644 --- a/src/Core/MySQL/PacketsProtocolBinary.cpp +++ b/src/Core/MySQL/PacketsProtocolBinary.cpp @@ -3,11 +3,14 @@ #include #include #include +#include #include +#include "Common/logger_useful.h" #include "Columns/ColumnLowCardinality.h" #include "Columns/ColumnVector.h" #include "DataTypes/DataTypeLowCardinality.h" #include "DataTypes/DataTypeNullable.h" +#include "DataTypes/DataTypesNumber.h" #include "Formats/FormatSettings.h" #include "IO/WriteBufferFromString.h" #include "base/types.h" @@ -26,24 +29,37 @@ namespace MySQLProtocol { /// See https://dev.mysql.com/doc/dev/mysql-server/8.1.0/page_protocol_binary_resultset.html#sect_protocol_binary_resultset_row payload_size = 1 + null_bitmap_size; - // LOG_TRACE(&Poco::Logger::get("ResultSetRow"), "Null bitmap size: {}", null_bitmap_size); + LOG_TRACE(&Poco::Logger::get("ResultSetRow"), "Null bitmap size: {}", null_bitmap_size); FormatSettings format_settings; for (size_t i = 0; i < columns.size(); ++i) { ColumnPtr col = columns[i]; - + LOG_TRACE(&Poco::Logger::get("col->isNullAt"), "isNullAt: {}, {}", row_num, col->isNullAt(row_num)); if (col->isNullAt(row_num)) { - null_bitmap[i / 8] |= 1 << i % 8; + size_t byte = (i + 2) / 8; + int bit = 1 << ((i + 2) % 8); + null_bitmap[byte] |= bit; + continue; // NULLs are stored in the null bitmap only } - TypeIndex type_index = removeNullable(removeLowCardinality(data_types[i]))->getTypeId(); + DataTypePtr data_type = removeLowCardinality(removeNullable((data_types[i]))); + TypeIndex type_index = data_type->getTypeId(); switch (type_index) { case TypeIndex::Int8: - case TypeIndex::UInt8: payload_size += 1; break; + case TypeIndex::UInt8: + if (data_type->getName() == "Bool") + { + payload_size += 2; // BIT MySQL type is string in binary + } + else + { + payload_size += 1; + } + break; case TypeIndex::Int16: case TypeIndex::UInt16: payload_size += 2; @@ -91,6 +107,7 @@ namespace MySQLProtocol break; } default: + LOG_TRACE(&Poco::Logger::get("Type default"), "{} is {}", col->getName(), data_type->getName()); WriteBufferFromOwnString ostr; serializations[i]->serializeText(*columns[i], row_num, ostr, format_settings); payload_size += getLengthEncodedStringSize(ostr.str()); @@ -109,64 +126,69 @@ namespace MySQLProtocol ColumnPtr col = columns[i]; if (col->isNullAt(row_num)) { - continue; // NULLs are stored in the null bitmap only + continue; } - TypeIndex type_index = removeNullable(removeLowCardinality(data_types[i]))->getTypeId(); + DataTypePtr data_type = removeLowCardinality(removeNullable((data_types[i]))); + TypeIndex type_index = data_type->getTypeId(); switch (type_index) { case TypeIndex::UInt8: { - UInt64 value = col->get64(row_num); + UInt8 value = assert_cast &>(*col).getData()[row_num]; + if (data_type->getName() == "Bool") + { + buffer.write(static_cast(1)); + } buffer.write(reinterpret_cast(&value), 1); break; } case TypeIndex::UInt16: { - UInt64 value = col->get64(row_num); + UInt16 value = assert_cast &>(*col).getData()[row_num]; buffer.write(reinterpret_cast(&value), 2); break; } case TypeIndex::UInt32: { - UInt64 value = col->get64(row_num); + UInt32 value = assert_cast &>(*col).getData()[row_num]; buffer.write(reinterpret_cast(&value), 4); break; } case TypeIndex::UInt64: { - UInt64 value = col->get64(row_num); + UInt64 value = assert_cast &>(*col).getData()[row_num]; buffer.write(reinterpret_cast(&value), 8); break; } case TypeIndex::Int8: { - UInt64 value = col->get64(row_num); + Int8 value = assert_cast &>(*col).getData()[row_num]; buffer.write(reinterpret_cast(&value), 1); break; } case TypeIndex::Int16: { - UInt64 value = col->get64(row_num); + Int16 value = assert_cast &>(*col).getData()[row_num]; buffer.write(reinterpret_cast(&value), 2); break; } case TypeIndex::Int32: { - UInt64 value = col->get64(row_num); + Int32 value = assert_cast &>(*col).getData()[row_num]; buffer.write(reinterpret_cast(&value), 4); break; } case TypeIndex::Int64: { - UInt64 value = col->get64(row_num); + Int64 value = assert_cast &>(*col).getData()[row_num]; buffer.write(reinterpret_cast(&value), 8); break; } case TypeIndex::Float32: { - Float32 value = col->getFloat32(row_num); + Float32 value = assert_cast &>(*col).getData()[row_num]; buffer.write(reinterpret_cast(&value), 4); break; } case TypeIndex::Float64: { - Float64 value = col->getFloat64(row_num); + Float64 value = assert_cast &>(*col).getData()[row_num]; buffer.write(reinterpret_cast(&value), 8); break; } case TypeIndex::Date: { - UInt64 value = col->get64(row_num); + UInt64 value = assert_cast &>(*col).getData()[row_num]; if (value != 0) { Poco::DateTime dt = Poco::DateTime(Poco::Timestamp(value * 1000 * 1000)); @@ -185,7 +207,7 @@ namespace MySQLProtocol break; } case TypeIndex::DateTime: { - UInt64 value = col->get64(row_num); + UInt64 value = assert_cast &>(*col).getData()[row_num]; if (value != 0) { Poco::DateTime dt = Poco::DateTime(Poco::Timestamp(value * 1000 * 1000)); diff --git a/src/Core/MySQL/PacketsProtocolBinary.h b/src/Core/MySQL/PacketsProtocolBinary.h index 69936e527c1..1a201ce3a4a 100644 --- a/src/Core/MySQL/PacketsProtocolBinary.h +++ b/src/Core/MySQL/PacketsProtocolBinary.h @@ -28,8 +28,8 @@ namespace MySQLProtocol std::vector serialized = std::vector(columns.size()); - size_t null_bitmap_size = (columns.size() + 7) / 8; - std::vector null_bitmap = std::vector(null_bitmap_size, 0); + size_t null_bitmap_size = (columns.size() + 7 + 2) / 8; + std::vector null_bitmap = std::vector(null_bitmap_size, static_cast(0)); size_t payload_size = 0; diff --git a/src/Core/MySQL/PacketsProtocolText.cpp b/src/Core/MySQL/PacketsProtocolText.cpp index 9c5bf6b6e05..bb5412bb5a5 100644 --- a/src/Core/MySQL/PacketsProtocolText.cpp +++ b/src/Core/MySQL/PacketsProtocolText.cpp @@ -2,7 +2,9 @@ #include #include #include +#include "Common/assert_cast.h" #include "Core/MySQL/IMySQLWritePacket.h" +#include "DataTypes/DataTypesDecimal.h" namespace DB { @@ -142,17 +144,25 @@ namespace MySQLProtocol } } - ColumnDefinition getColumnDefinition(const String & column_name, const TypeIndex type_index) + ColumnDefinition getColumnDefinition(const String & column_name, const DataTypePtr & data_type) { ColumnType column_type; CharacterSet charset = CharacterSet::binary; int flags = 0; uint8_t decimals = 0; + TypeIndex type_index = data_type->getTypeId(); switch (type_index) { case TypeIndex::UInt8: - column_type = ColumnType::MYSQL_TYPE_TINY; - flags = ColumnDefinitionFlags::BINARY_FLAG | ColumnDefinitionFlags::UNSIGNED_FLAG; + if (data_type->getName() == "Bool") + { + column_type = ColumnType::MYSQL_TYPE_BIT; + } + else + { + column_type = ColumnType::MYSQL_TYPE_TINY; + flags = ColumnDefinitionFlags::BINARY_FLAG | ColumnDefinitionFlags::UNSIGNED_FLAG; + } break; case TypeIndex::UInt16: column_type = ColumnType::MYSQL_TYPE_SHORT; @@ -202,11 +212,24 @@ namespace MySQLProtocol break; case TypeIndex::Decimal32: case TypeIndex::Decimal64: - case TypeIndex::Decimal128: - /// MySQL Decimal has max 65 precision and 30 scale. Thus, Decimal256 is reported as a string column_type = ColumnType::MYSQL_TYPE_DECIMAL; flags = ColumnDefinitionFlags::BINARY_FLAG; break; + case TypeIndex::Decimal128: { + // MySQL Decimal has max 65 precision and 30 scale + const auto & type = assert_cast(*data_type); + if (type.getPrecision() > 65 || type.getScale() > 30) + { + column_type = ColumnType::MYSQL_TYPE_STRING; + charset = CharacterSet::utf8_general_ci; + } + else + { + column_type = ColumnType::MYSQL_TYPE_DECIMAL; + flags = ColumnDefinitionFlags::BINARY_FLAG; + } + break; + } default: column_type = ColumnType::MYSQL_TYPE_STRING; charset = CharacterSet::utf8_general_ci; diff --git a/src/Core/MySQL/PacketsProtocolText.h b/src/Core/MySQL/PacketsProtocolText.h index aeeedc4dbf8..d9ce913494b 100644 --- a/src/Core/MySQL/PacketsProtocolText.h +++ b/src/Core/MySQL/PacketsProtocolText.h @@ -12,122 +12,135 @@ namespace DB namespace MySQLProtocol { -namespace ProtocolText -{ + namespace ProtocolText + { -enum CharacterSet -{ - utf8_general_ci = 33, - binary = 63 -}; + enum CharacterSet + { + utf8_general_ci = 33, + binary = 63 + }; -// https://dev.mysql.com/doc/dev/mysql-server/latest/group__group__cs__column__definition__flags.html -enum ColumnDefinitionFlags -{ - UNSIGNED_FLAG = 32, - BINARY_FLAG = 128 -}; + // https://dev.mysql.com/doc/dev/mysql-server/latest/group__group__cs__column__definition__flags.html + enum ColumnDefinitionFlags + { + UNSIGNED_FLAG = 32, + BINARY_FLAG = 128 + }; -enum ColumnType -{ - MYSQL_TYPE_DECIMAL = 0x00, - MYSQL_TYPE_TINY = 0x01, - MYSQL_TYPE_SHORT = 0x02, - MYSQL_TYPE_LONG = 0x03, - MYSQL_TYPE_FLOAT = 0x04, - MYSQL_TYPE_DOUBLE = 0x05, - MYSQL_TYPE_NULL = 0x06, - MYSQL_TYPE_TIMESTAMP = 0x07, - MYSQL_TYPE_LONGLONG = 0x08, - MYSQL_TYPE_INT24 = 0x09, - MYSQL_TYPE_DATE = 0x0a, - MYSQL_TYPE_TIME = 0x0b, - MYSQL_TYPE_DATETIME = 0x0c, - MYSQL_TYPE_YEAR = 0x0d, - MYSQL_TYPE_NEWDATE = 0x0e, - MYSQL_TYPE_VARCHAR = 0x0f, - MYSQL_TYPE_BIT = 0x10, - MYSQL_TYPE_TIMESTAMP2 = 0x11, - MYSQL_TYPE_DATETIME2 = 0x12, - MYSQL_TYPE_TIME2 = 0x13, - MYSQL_TYPE_JSON = 0xf5, - MYSQL_TYPE_NEWDECIMAL = 0xf6, - MYSQL_TYPE_ENUM = 0xf7, - MYSQL_TYPE_SET = 0xf8, - MYSQL_TYPE_TINY_BLOB = 0xf9, - MYSQL_TYPE_MEDIUM_BLOB = 0xfa, - MYSQL_TYPE_LONG_BLOB = 0xfb, - MYSQL_TYPE_BLOB = 0xfc, - MYSQL_TYPE_VAR_STRING = 0xfd, - MYSQL_TYPE_STRING = 0xfe, - MYSQL_TYPE_GEOMETRY = 0xff -}; + enum ColumnType + { + MYSQL_TYPE_DECIMAL = 0x00, + MYSQL_TYPE_TINY = 0x01, + MYSQL_TYPE_SHORT = 0x02, + MYSQL_TYPE_LONG = 0x03, + MYSQL_TYPE_FLOAT = 0x04, + MYSQL_TYPE_DOUBLE = 0x05, + MYSQL_TYPE_NULL = 0x06, + MYSQL_TYPE_TIMESTAMP = 0x07, + MYSQL_TYPE_LONGLONG = 0x08, + MYSQL_TYPE_INT24 = 0x09, + MYSQL_TYPE_DATE = 0x0a, + MYSQL_TYPE_TIME = 0x0b, + MYSQL_TYPE_DATETIME = 0x0c, + MYSQL_TYPE_YEAR = 0x0d, + MYSQL_TYPE_NEWDATE = 0x0e, + MYSQL_TYPE_VARCHAR = 0x0f, + MYSQL_TYPE_BIT = 0x10, + MYSQL_TYPE_TIMESTAMP2 = 0x11, + MYSQL_TYPE_DATETIME2 = 0x12, + MYSQL_TYPE_TIME2 = 0x13, + MYSQL_TYPE_JSON = 0xf5, + MYSQL_TYPE_NEWDECIMAL = 0xf6, + MYSQL_TYPE_ENUM = 0xf7, + MYSQL_TYPE_SET = 0xf8, + MYSQL_TYPE_TINY_BLOB = 0xf9, + MYSQL_TYPE_MEDIUM_BLOB = 0xfa, + MYSQL_TYPE_LONG_BLOB = 0xfb, + MYSQL_TYPE_BLOB = 0xfc, + MYSQL_TYPE_VAR_STRING = 0xfd, + MYSQL_TYPE_STRING = 0xfe, + MYSQL_TYPE_GEOMETRY = 0xff + }; -class ResultSetRow : public IMySQLWritePacket -{ -protected: - const Columns & columns; - int row_num; - size_t payload_size = 0; - std::vector serialized; + class ResultSetRow : public IMySQLWritePacket + { + protected: + const Columns & columns; + int row_num; + size_t payload_size = 0; + std::vector serialized; - size_t getPayloadSize() const override; + size_t getPayloadSize() const override; - void writePayloadImpl(WriteBuffer & buffer) const override; + void writePayloadImpl(WriteBuffer & buffer) const override; -public: - ResultSetRow(const Serializations & serializations, const Columns & columns_, int row_num_); -}; + public: + ResultSetRow(const Serializations & serializations, const Columns & columns_, int row_num_); + }; -class ComFieldList : public LimitedReadPacket -{ -public: - String table, field_wildcard; + class ComFieldList : public LimitedReadPacket + { + public: + String table, field_wildcard; - void readPayloadImpl(ReadBuffer & payload) override; -}; + void readPayloadImpl(ReadBuffer & payload) override; + }; -class ColumnDefinition : public IMySQLWritePacket, public IMySQLReadPacket -{ -public: - String schema; - String table; - String org_table; - String name; - String org_name; - size_t next_length = 0x0c; - uint16_t character_set; - uint32_t column_length; - ColumnType column_type; - uint16_t flags; - uint8_t decimals = 0x00; - /// https://dev.mysql.com/doc/internals/en/com-query-response.html#column-definition - /// There are extra fields in the packet for column defaults - bool is_comm_field_list_response = false; + class ColumnDefinition : public IMySQLWritePacket, public IMySQLReadPacket + { + public: + String schema; + String table; + String org_table; + String name; + String org_name; + size_t next_length = 0x0c; + uint16_t character_set; + uint32_t column_length; + ColumnType column_type; + uint16_t flags; + uint8_t decimals = 0x00; + /// https://dev.mysql.com/doc/internals/en/com-query-response.html#column-definition + /// There are extra fields in the packet for column defaults + bool is_comm_field_list_response = false; -protected: - size_t getPayloadSize() const override; + protected: + size_t getPayloadSize() const override; - void readPayloadImpl(ReadBuffer & payload) override; + void readPayloadImpl(ReadBuffer & payload) override; - void writePayloadImpl(WriteBuffer & buffer) const override; + void writePayloadImpl(WriteBuffer & buffer) const override; -public: - ColumnDefinition(); + public: + ColumnDefinition(); - ColumnDefinition( - String schema_, String table_, String org_table_, String name_, String org_name_, uint16_t character_set_, uint32_t column_length_, - ColumnType column_type_, uint16_t flags_, uint8_t decimals_, bool with_defaults_ = false); + ColumnDefinition( + String schema_, + String table_, + String org_table_, + String name_, + String org_name_, + uint16_t character_set_, + uint32_t column_length_, + ColumnType column_type_, + uint16_t flags_, + uint8_t decimals_, + bool with_defaults_ = false); - /// Should be used when column metadata (original name, table, original table, database) is unknown. - ColumnDefinition( - String name_, uint16_t character_set_, uint32_t column_length_, ColumnType column_type_, uint16_t flags_, uint8_t decimals_); + /// Should be used when column metadata (original name, table, original table, database) is unknown. + ColumnDefinition( + String name_, + uint16_t character_set_, + uint32_t column_length_, + ColumnType column_type_, + uint16_t flags_, + uint8_t decimals_); + }; -}; + ColumnDefinition getColumnDefinition(const String & column_name, const DataTypePtr & data_type); -ColumnDefinition getColumnDefinition(const String & column_name, const TypeIndex index); - -} + } } diff --git a/src/Processors/Formats/Impl/MySQLOutputFormat.cpp b/src/Processors/Formats/Impl/MySQLOutputFormat.cpp index 3dafe560281..d25b68507fc 100644 --- a/src/Processors/Formats/Impl/MySQLOutputFormat.cpp +++ b/src/Processors/Formats/Impl/MySQLOutputFormat.cpp @@ -53,7 +53,7 @@ void MySQLOutputFormat::writePrefix() for (size_t i = 0; i < header.columns(); ++i) { const auto & column_name = header.getColumnsWithTypeAndName()[i].name; - packet_endpoint->sendPacket(getColumnDefinition(column_name, data_types[i]->getTypeId())); + packet_endpoint->sendPacket(getColumnDefinition(column_name, data_types[i])); } if (!(client_capabilities & Capability::CLIENT_DEPRECATE_EOF) && !use_binary_result_set) diff --git a/tests/integration/requirements.txt b/tests/integration/requirements.txt new file mode 100644 index 00000000000..11efc43eded --- /dev/null +++ b/tests/integration/requirements.txt @@ -0,0 +1,30 @@ +PyMySQL +aerospike +avro +cassandra-driver +confluent-kafka +dicttoxml +docker +docker-compose +grpcio +grpcio-tools +kafka-python +kazoo +minio +lz4 +protobuf +psycopg2-binary +pymongo +pytz +pytest +pytest-timeout +redis +tzlocal==2.1 +urllib3 +requests-kerberos +dict2xml +hypothesis +pyhdfs +pika +meilisearch +nats-py diff --git a/tests/integration/test_mysql_protocol/prepared_statements.reference b/tests/integration/test_mysql_protocol/prepared_statements.reference new file mode 100644 index 00000000000..257cc5642cb --- /dev/null +++ b/tests/integration/test_mysql_protocol/prepared_statements.reference @@ -0,0 +1 @@ +foo diff --git a/tests/integration/test_mysql_protocol/test.py b/tests/integration/test_mysql_protocol/test.py index a5c82e1a2c6..c0bc178c5bb 100644 --- a/tests/integration/test_mysql_protocol/test.py +++ b/tests/integration/test_mysql_protocol/test.py @@ -10,7 +10,6 @@ import logging import docker import pymysql.connections import pytest -from docker.models.containers import Container from helpers.cluster import ClickHouseCluster, get_docker_compose_path, run_and_check SCRIPT_DIR = os.path.dirname(os.path.realpath(__file__)) @@ -150,7 +149,6 @@ def java_container(): def test_mysql_client(started_cluster): - # type: (ClickHouseCluster) -> None code, (stdout, stderr) = started_cluster.mysql_client_container.exec_run( """ mysql --protocol tcp -h {host} -P {port} default -u user_with_double_sha1 --password=abacaba @@ -651,7 +649,6 @@ def test_python_client(started_cluster): def test_golang_client(started_cluster, golang_container): - # type: (str, Container) -> None with open(os.path.join(SCRIPT_DIR, "golang.reference"), "rb") as fp: reference = fp.read() @@ -687,7 +684,6 @@ def test_golang_client(started_cluster, golang_container): def test_php_client(started_cluster, php_container): - # type: (str, Container) -> None code, (stdout, stderr) = php_container.exec_run( "php -f test.php {host} {port} default 123".format( host=started_cluster.get_instance_ip("node"), port=server_port @@ -764,7 +760,6 @@ def test_mysqljs_client(started_cluster, nodejs_container): def test_java_client(started_cluster, java_container): - # type: (str, Container) -> None with open(os.path.join(SCRIPT_DIR, "java.reference")) as fp: reference = fp.read() @@ -810,6 +805,107 @@ def test_java_client(started_cluster, java_container): assert stdout.decode() == reference +def test_prepared_statements(started_cluster, java_container): + with open(os.path.join(SCRIPT_DIR, "prepared_statements.reference")) as fp: + reference = fp.read() + + node.query(""" + CREATE TABLE ps_simple_data_types + ( + i8 Int8, + i16 Int16, + i32 Int32, + i64 Int64, + i128 Int128, + i256 Int256, + ui8 UInt8, + ui16 UInt16, + ui32 UInt32, + ui64 UInt64, + ui128 UInt128, + ui256 UInt256, + f32 Float32, + f64 Float64, + b Boolean + ) ENGINE MergeTree ORDER BY i8; + """, settings={"password": "123"}) + node.query(""" + INSERT INTO ps_simple_data_types + VALUES (127, 32767, 2147483647, 9223372036854775807, 170141183460469231731687303715884105727, + 57896044618658097711785492504343953926634992332820282019728792003956564819967, + 255, 65535, 4294967295, 18446744073709551615, 340282366920938463463374607431768211455, + 115792089237316195423570985008687907853269984665640564039457584007913129639935, + 1.234, 3.35245141223232, FALSE), + (-128, -32768, -2147483648, -9223372036854775808, -170141183460469231731687303715884105728, + -57896044618658097711785492504343953926634992332820282019728792003956564819968, + 120, 1234, 51234, 421342, 15324355, 41345135123432, + -0.7968956, -0.113259, TRUE); + """, settings={"password": "123"}) + + node.query(""" + CREATE TABLE ps_string_types + ( + s String, + sn Nullable(String), + lc LowCardinality(String), + nlc LowCardinality(Nullable(String)) + ) ENGINE MergeTree ORDER BY s; + """, settings={"password": "123"}) + node.query(""" + INSERT INTO ps_string_types + VALUES ('foo', 'bar', 'qaz', 'qux'), + ('42', NULL, 'test', NULL); + """, settings={"password": "123"}) + + node.query(""" + CREATE TABLE ps_decimal_types + ( + d32 Decimal(9, 2), + d64 Decimal(18, 3), + d128_native Decimal(30, 10), + d128_text Decimal(38, 31), + d256 Decimal(76, 20) + ) ENGINE MergeTree ORDER BY d32; + """, settings={"password": "123"}) + node.query(""" + INSERT INTO ps_decimal_types + VALUES (1234567.89, + 123456789123456.789, + 12345678912345678912.1234567891, + 1234567.8912345678912345678911234567891, + 12345678912345678912345678911234567891234567891234567891.12345678911234567891), + (-1.55, 6.03, 5, -1224124.23423, -54342.3); + """, settings={"password": "123"}) + + node.query(""" + CREATE TABLE ps_misc_types + ( + + a Array(String), + u UUID, + t Tuple(Int32, String), + m Map(String, Int32) + ) ENGINE MergeTree ORDER BY u; + """, settings={"password": "123"}) + node.query(""" + SELECT * + FROM ps_misc_types; + INSERT INTO ps_misc_types + VALUES (['foo', 'bar'], '5da5038d-788f-48c6-b510-babb41c538d3', (42, 'qaz'), {'qux': 144, 'text': 255}), + ([], '9a0ccc06-2578-4861-8534-631c9d40f3f7', (0, ''), {}); + """, settings={"password": "123"}) + + code, (stdout, stderr) = java_container.exec_run( + "java PreparedStatementsTest --host {host} --port {port} --user user_with_double_sha1 --password abacaba --database " + "default".format( + host=started_cluster.get_instance_ip("node"), port=server_port + ), + demux=True, + ) + assert code == 0 + assert stdout.decode() == reference + + def test_types(started_cluster): client = pymysql.connections.Connection( host=started_cluster.get_instance_ip("node"), @@ -844,14 +940,14 @@ def test_types(started_cluster): result = cursor.fetchall()[0] expected = [ - ("Int8_column", -(2**7)), - ("UInt8_column", 2**8 - 1), - ("Int16_column", -(2**15)), - ("UInt16_column", 2**16 - 1), - ("Int32_column", -(2**31)), - ("UInt32_column", 2**32 - 1), - ("Int64_column", -(2**63)), - ("UInt64_column", 2**64 - 1), + ("Int8_column", -(2 ** 7)), + ("UInt8_column", 2 ** 8 - 1), + ("Int16_column", -(2 ** 15)), + ("UInt16_column", 2 ** 16 - 1), + ("Int32_column", -(2 ** 31)), + ("UInt32_column", 2 ** 32 - 1), + ("Int64_column", -(2 ** 63)), + ("UInt64_column", 2 ** 64 - 1), ("String_column", "тест"), ("FixedString_column", "тест"), ("Float32_column", 1.5), From 06afe0c2aa43828d830bca354cf99c3ea65b4534 Mon Sep 17 00:00:00 2001 From: Duc Canh Le Date: Thu, 31 Aug 2023 06:27:06 +0000 Subject: [PATCH 092/433] more stable stateless test + add perf. test --- tests/performance/final_with_lonely_parts.xml | 77 +++++++ ...oss_partitions_final_with_lonely.reference | 208 +----------------- ...rge_across_partitions_final_with_lonely.sh | 60 +++++ ...ge_across_partitions_final_with_lonely.sql | 49 ----- 4 files changed, 143 insertions(+), 251 deletions(-) create mode 100644 tests/performance/final_with_lonely_parts.xml create mode 100755 tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.sh delete mode 100644 tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.sql diff --git a/tests/performance/final_with_lonely_parts.xml b/tests/performance/final_with_lonely_parts.xml new file mode 100644 index 00000000000..075f639956b --- /dev/null +++ b/tests/performance/final_with_lonely_parts.xml @@ -0,0 +1,77 @@ + + + 1 + 1 + + + + CREATE TABLE with_lonely + ( + `id` UInt64, + `dt` Date, + `val` UInt64, + `version` UInt64 + ) + ENGINE = ReplacingMergeTree(version) + PARTITION BY dt + ORDER BY (id); + + + + INSERT INTO with_lonely SELECT number, '2022-10-28', number*10, 0 FROM numbers(3000000); + + + INSERT INTO with_lonely SELECT number+500000, '2022-10-28', number*10, 1 FROM + numbers(3000000); + + OPTIMIZE TABLE with_lonely PARTITION '2022-10-28' FINAL; + + + INSERT INTO with_lonely SELECT number, '2022-10-29', number*10, 0 FROM numbers(3000000); + + + INSERT INTO with_lonely SELECT number+500000, '2022-10-29', number*10, 1 FROM + numbers(3000000); + + OPTIMIZE TABLE with_lonely PARTITION '2022-10-29' FINAL; + + + INSERT INTO with_lonely SELECT number, '2022-10-30', number*10, 0 FROM numbers(3000000); + + + INSERT INTO with_lonely SELECT number+500000, '2022-10-30', number*10, 1 FROM + numbers(3000000); + + OPTIMIZE TABLE with_lonely PARTITION '2022-10-30' FINAL; + + + INSERT INTO with_lonely SELECT number, '2022-10-31', number*10, 0 FROM numbers(3000000); + + + INSERT INTO with_lonely SELECT number+500000, '2022-10-31', number*10, 1 FROM + numbers(3000000); + + OPTIMIZE TABLE with_lonely PARTITION '2022-10-31' FINAL; + + + INSERT INTO with_lonely SELECT number, '2022-11-01', number*10, 0 FROM numbers(500000); + + + INSERT INTO with_lonely SELECT number+50000, '2022-11-01', number*10, 1 FROM + numbers(500000); + + + INSERT INTO with_lonely SELECT number+60000, '2022-11-01', number*10, 2 FROM + numbers(500000); + + + INSERT INTO with_lonely SELECT number+70000, '2022-11-01', number*10, 3 FROM + numbers(500000); + + + + SELECT max(val), count(*) FROM with_lonely FINAL; + SELECT max(val), count(*) FROM with_lonely FINAL WHERE dt != '2022-11-01'; + + DROP TABLE IF EXISTS with_lonely; + diff --git a/tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.reference b/tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.reference index a4b37bc423b..72837206fac 100644 --- a/tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.reference +++ b/tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.reference @@ -1,202 +1,6 @@ -(Expression) -ExpressionTransform × 16 -Header: max(val) UInt64: max(val) UInt64 UInt64(size = 0) - count() UInt64: count() UInt64 UInt64(size = 0) - (Aggregating) - Resize 6 → 16 - Header × 16 : max(val) UInt64: max(val) UInt64 UInt64(size = 0) - count() UInt64: count() UInt64 UInt64(size = 0) - AggregatingTransform × 6 - Header: max(val) UInt64: max(val) UInt64 UInt64(size = 0) - count() UInt64: count() UInt64 UInt64(size = 0) - StrictResize 6 → 6 - Header × 6 : val UInt64: val UInt64 UInt64(size = 0) - (Expression) - ExpressionTransform × 6 - Header: val UInt64: val UInt64 UInt64(size = 0) - (ReadFromMergeTree) - MergeTreeThread × 4 0 → 1 - Header: val UInt64: val UInt64 UInt64(size = 0) - ExpressionTransform × 2 - Header: val UInt64: val UInt64 UInt64(size = 0) - ReplacingSorted 2 → 1 - Header: id UInt64: id UInt64 UInt64(size = 0) - val UInt64: val UInt64 UInt64(size = 0) - version UInt64: version UInt64 UInt64(size = 0) - FilterSortedStreamByRange × 2 - Header: id UInt64: id UInt64 UInt64(size = 0) - val UInt64: val UInt64 UInt64(size = 0) - version UInt64: version UInt64 UInt64(size = 0) - Description: filter values in [(50000), +inf) - ExpressionTransform × 2 - Header: id UInt64: id UInt64 UInt64(size = 0) - val UInt64: val UInt64 UInt64(size = 0) - version UInt64: version UInt64 UInt64(size = 0) - MergeTreeInOrder × 2 0 → 1 - Header: id UInt64: id UInt64 UInt64(size = 0) - val UInt64: val UInt64 UInt64(size = 0) - version UInt64: version UInt64 UInt64(size = 0) - ReplacingSorted 2 → 1 - Header: id UInt64: id UInt64 UInt64(size = 0) - val UInt64: val UInt64 UInt64(size = 0) - version UInt64: version UInt64 UInt64(size = 0) - FilterSortedStreamByRange × 2 - Header: id UInt64: id UInt64 UInt64(size = 0) - val UInt64: val UInt64 UInt64(size = 0) - version UInt64: version UInt64 UInt64(size = 0) - Description: filter values in [-inf, (50000)) - ExpressionTransform × 2 - Header: id UInt64: id UInt64 UInt64(size = 0) - val UInt64: val UInt64 UInt64(size = 0) - version UInt64: version UInt64 UInt64(size = 0) - MergeTreeInOrder × 2 0 → 1 - Header: id UInt64: id UInt64 UInt64(size = 0) - val UInt64: val UInt64 UInt64(size = 0) - version UInt64: version UInt64 UInt64(size = 0) -(Expression) -ExpressionTransform × 16 -Header: max(val) UInt64: max(val) UInt64 UInt64(size = 0) - count() UInt64: count() UInt64 UInt64(size = 0) - (Aggregating) - Resize 4 → 16 - Header × 16 : max(val) UInt64: max(val) UInt64 UInt64(size = 0) - count() UInt64: count() UInt64 UInt64(size = 0) - AggregatingTransform × 4 - Header: max(val) UInt64: max(val) UInt64 UInt64(size = 0) - count() UInt64: count() UInt64 UInt64(size = 0) - StrictResize 4 → 4 - Header × 4 : val UInt64: val UInt64 UInt64(size = 0) - (Expression) - ExpressionTransform × 4 - Header: val UInt64: val UInt64 UInt64(size = 0) - (Filter) - FilterTransform × 4 - Header: val UInt64: val UInt64 UInt64(size = 0) - (ReadFromMergeTree) - MergeTreeThread × 4 0 → 1 - Header: dt Date: dt Date UInt16(size = 0) - val UInt64: val UInt64 UInt64(size = 0) -(Expression) -ExpressionTransform -Header: max(val) UInt64: max(val) UInt64 UInt64(size = 0) - count() UInt64: count() UInt64 UInt64(size = 0) - (Aggregating) - Resize 4 → 1 - Header: max(val) UInt64: max(val) UInt64 UInt64(size = 0) - count() UInt64: count() UInt64 UInt64(size = 0) - AggregatingTransform × 4 - Header: max(val) UInt64: max(val) UInt64 UInt64(size = 0) - count() UInt64: count() UInt64 UInt64(size = 0) - StrictResize 4 → 4 - Header × 4 : val UInt64: val UInt64 UInt64(size = 0) - (Expression) - ExpressionTransform × 4 - Header: val UInt64: val UInt64 UInt64(size = 0) - (Filter) - FilterTransform × 4 - Header: val UInt64: val UInt64 UInt64(size = 0) - (ReadFromMergeTree) - MergeTreeInOrder × 4 0 → 1 - Header: dt Date: dt Date UInt16(size = 0) - val UInt64: val UInt64 UInt64(size = 0) -(Expression) -ExpressionTransform × 16 -Header: max(val) UInt64: max(val) UInt64 UInt64(size = 0) - count() UInt64: count() UInt64 UInt64(size = 0) - (Aggregating) - Resize 6 → 16 - Header × 16 : max(val_0) UInt64: max(val_0) UInt64 UInt64(size = 0) - count() UInt64: count() UInt64 UInt64(size = 0) - AggregatingTransform × 6 - Header: max(val_0) UInt64: max(val_0) UInt64 UInt64(size = 0) - count() UInt64: count() UInt64 UInt64(size = 0) - StrictResize 6 → 6 - Header × 6 : val_0 UInt64: val_0 UInt64 UInt64(size = 0) - (Expression) - ExpressionTransform × 6 - Header: val_0 UInt64: val_0 UInt64 UInt64(size = 0) - (ReadFromMergeTree) - MergeTreeThread × 4 0 → 1 - Header: val UInt64: val UInt64 UInt64(size = 0) - ExpressionTransform × 2 - Header: val UInt64: val UInt64 UInt64(size = 0) - ReplacingSorted 2 → 1 - Header: id UInt64: id UInt64 UInt64(size = 0) - val UInt64: val UInt64 UInt64(size = 0) - version UInt64: version UInt64 UInt64(size = 0) - FilterSortedStreamByRange × 2 - Header: id UInt64: id UInt64 UInt64(size = 0) - val UInt64: val UInt64 UInt64(size = 0) - version UInt64: version UInt64 UInt64(size = 0) - Description: filter values in [(50000), +inf) - ExpressionTransform × 2 - Header: id UInt64: id UInt64 UInt64(size = 0) - val UInt64: val UInt64 UInt64(size = 0) - version UInt64: version UInt64 UInt64(size = 0) - MergeTreeInOrder × 2 0 → 1 - Header: id UInt64: id UInt64 UInt64(size = 0) - val UInt64: val UInt64 UInt64(size = 0) - version UInt64: version UInt64 UInt64(size = 0) - ReplacingSorted 2 → 1 - Header: id UInt64: id UInt64 UInt64(size = 0) - val UInt64: val UInt64 UInt64(size = 0) - version UInt64: version UInt64 UInt64(size = 0) - FilterSortedStreamByRange × 2 - Header: id UInt64: id UInt64 UInt64(size = 0) - val UInt64: val UInt64 UInt64(size = 0) - version UInt64: version UInt64 UInt64(size = 0) - Description: filter values in [-inf, (50000)) - ExpressionTransform × 2 - Header: id UInt64: id UInt64 UInt64(size = 0) - val UInt64: val UInt64 UInt64(size = 0) - version UInt64: version UInt64 UInt64(size = 0) - MergeTreeInOrder × 2 0 → 1 - Header: id UInt64: id UInt64 UInt64(size = 0) - val UInt64: val UInt64 UInt64(size = 0) - version UInt64: version UInt64 UInt64(size = 0) -(Expression) -ExpressionTransform × 16 -Header: max(val) UInt64: max(val) UInt64 UInt64(size = 0) - count() UInt64: count() UInt64 UInt64(size = 0) - (Aggregating) - Resize 4 → 16 - Header × 16 : max(val_0) UInt64: max(val_0) UInt64 UInt64(size = 0) - count() UInt64: count() UInt64 UInt64(size = 0) - AggregatingTransform × 4 - Header: max(val_0) UInt64: max(val_0) UInt64 UInt64(size = 0) - count() UInt64: count() UInt64 UInt64(size = 0) - StrictResize 4 → 4 - Header × 4 : val_0 UInt64: val_0 UInt64 UInt64(size = 0) - (Expression) - ExpressionTransform × 4 - Header: val_0 UInt64: val_0 UInt64 UInt64(size = 0) - (Filter) - FilterTransform × 4 - Header: val_0 UInt64: val_0 UInt64 UInt64(size = 0) - (ReadFromMergeTree) - MergeTreeThread × 4 0 → 1 - Header: val UInt64: val UInt64 UInt64(size = 0) - dt Date: dt Date UInt16(size = 0) -(Expression) -ExpressionTransform -Header: max(val) UInt64: max(val) UInt64 UInt64(size = 0) - count() UInt64: count() UInt64 UInt64(size = 0) - (Aggregating) - Resize 4 → 1 - Header: max(val_0) UInt64: max(val_0) UInt64 UInt64(size = 0) - count() UInt64: count() UInt64 UInt64(size = 0) - AggregatingTransform × 4 - Header: max(val_0) UInt64: max(val_0) UInt64 UInt64(size = 0) - count() UInt64: count() UInt64 UInt64(size = 0) - StrictResize 4 → 4 - Header × 4 : val_0 UInt64: val_0 UInt64 UInt64(size = 0) - (Expression) - ExpressionTransform × 4 - Header: val_0 UInt64: val_0 UInt64 UInt64(size = 0) - (Filter) - FilterTransform × 4 - Header: val_0 UInt64: val_0 UInt64 UInt64(size = 0) - (ReadFromMergeTree) - MergeTreeInOrder × 4 0 → 1 - Header: val UInt64: val UInt64 UInt64(size = 0) - dt Date: dt Date UInt16(size = 0) +90 110 +90 80 +90 80 +90 110 +90 80 +90 80 diff --git a/tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.sh b/tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.sh new file mode 100755 index 00000000000..d01e4f94350 --- /dev/null +++ b/tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.sh @@ -0,0 +1,60 @@ +#!/usr/bin/env bash + +set -e + +CUR_DIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) +# shellcheck source=../shell_config.sh +. "$CUR_DIR"/../shell_config.sh + +${CLICKHOUSE_CLIENT} -nm -q """ +DROP TABLE IF EXISTS with_lonely; + +CREATE TABLE with_lonely +( + id UInt64, + dt Date, + val UInt64, + version UInt64 +) +ENGINE = ReplacingMergeTree(version) +PARTITION BY dt +ORDER BY (id); +""" + +create_optimize_partition() { + ${CLICKHOUSE_CLIENT} -nm -q """ + INSERT INTO with_lonely SELECT number, '$1', number*10, 0 FROM numbers(10); + INSERT INTO with_lonely SELECT number+500000, '$1', number*10, 1 FROM numbers(10); + """ + cnt=$(${CLICKHOUSE_CLIENT} -q "SELECT count() FROM system.parts WHERE database = currentDatabase() AND table = 'with_lonely' AND partition = '$1' AND active") + while [ "$cnt" -ne "1" ]; do + ${CLICKHOUSE_CLIENT} -q "OPTIMIZE TABLE with_lonely PARTITION '$1' FINAL;" + cnt=$(${CLICKHOUSE_CLIENT} -q "SELECT count() FROM system.parts WHERE database = currentDatabase() AND table = 'with_lonely' AND partition = '$1' AND active") + done +} + +create_optimize_partition "2022-10-28" +create_optimize_partition "2022-10-29" +create_optimize_partition "2022-10-30" +create_optimize_partition "2022-10-31" + +${CLICKHOUSE_CLIENT} -nm -q """ +SYSTEM STOP MERGES with_lonely; + +INSERT INTO with_lonely SELECT number, '2022-11-01', number*10, 0 FROM numbers(10); +INSERT INTO with_lonely SELECT number+50000, '2022-11-01', number*10, 1 FROM numbers(10); +INSERT INTO with_lonely SELECT number+60000, '2022-11-01', number*10, 2 FROM numbers(10); +""" + +CLICKHOUSE_CLIENT="${CLICKHOUSE_CLIENT} --do_not_merge_across_partitions_select_final 1" + +# mix lonely parts and non-lonely parts +${CLICKHOUSE_CLIENT} --allow_experimental_analyzer 1 --max_bytes_to_read 2000 -q "SELECT max(val), count(*) FROM with_lonely FINAL;" +# only lonely parts +${CLICKHOUSE_CLIENT} --allow_experimental_analyzer 1 --max_bytes_to_read 1000 -q "SELECT max(val), count(*) FROM with_lonely FINAL WHERE dt < '2022-11-01';" +# only lonely parts but max_thread = 1, so reading lonely parts with in-order +${CLICKHOUSE_CLIENT} --allow_experimental_analyzer 1 --max_threads 1 --max_bytes_to_read 1000 -q "SELECT max(val), count(*) FROM with_lonely FINAL WHERE dt < '2022-11-01';" + +${CLICKHOUSE_CLIENT} --allow_experimental_analyzer 0 --max_bytes_to_read 2000 -q "SELECT max(val), count(*) FROM with_lonely FINAL;" +${CLICKHOUSE_CLIENT} --allow_experimental_analyzer 0 --max_bytes_to_read 1000 -q "SELECT max(val), count(*) FROM with_lonely FINAL WHERE dt < '2022-11-01';" +${CLICKHOUSE_CLIENT} --allow_experimental_analyzer 0 --max_threads 1 --max_bytes_to_read 1000 -q "SELECT max(val), count(*) FROM with_lonely FINAL WHERE dt < '2022-11-01';" diff --git a/tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.sql b/tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.sql deleted file mode 100644 index 526c9204d96..00000000000 --- a/tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.sql +++ /dev/null @@ -1,49 +0,0 @@ -DROP TABLE IF EXISTS with_lonely; - -CREATE TABLE with_lonely -( - `id` UInt64, - `dt` Date, - `val` UInt64, - `version` UInt64 -) -ENGINE = ReplacingMergeTree(version) -PARTITION BY dt -ORDER BY (id); - -INSERT INTO with_lonely SELECT number, '2022-10-28', number*10, 0 FROM numbers(10000); -INSERT INTO with_lonely SELECT number+500000, '2022-10-28', number*10, 1 FROM numbers(10000); -OPTIMIZE TABLE with_lonely PARTITION '2022-10-28' FINAL; - - -INSERT INTO with_lonely SELECT number, '2022-10-29', number*10, 0 FROM numbers(10000); -INSERT INTO with_lonely SELECT number+500000, '2022-10-29', number*10, 1 FROM numbers(10000); -OPTIMIZE TABLE with_lonely PARTITION '2022-10-29' FINAL; - -INSERT INTO with_lonely SELECT number, '2022-10-30', number*10, 0 FROM numbers(10000); -INSERT INTO with_lonely SELECT number+500000, '2022-10-30', number*10, 1 FROM numbers(10000); -OPTIMIZE TABLE with_lonely PARTITION '2022-10-30' FINAL; - - -INSERT INTO with_lonely SELECT number, '2022-10-31', number*10, 0 FROM numbers(10000); -INSERT INTO with_lonely SELECT number+500000, '2022-10-31', number*10, 1 FROM numbers(10000); -OPTIMIZE TABLE with_lonely PARTITION '2022-10-31' FINAL; - -SYSTEM STOP MERGES with_lonely; -INSERT INTO with_lonely SELECT number, '2022-11-01', number*10, 0 FROM numbers(1000); -INSERT INTO with_lonely SELECT number+50000, '2022-11-01', number*10, 1 FROM numbers(1000); -INSERT INTO with_lonely SELECT number+60000, '2022-11-01', number*10, 2 FROM numbers(1000); - -SET do_not_merge_across_partitions_select_final = 1, max_threads = 16; - --- mix lonely parts and non-lonely parts -EXPLAIN PIPELINE header=1 SELECT max(val), count(*) FROM with_lonely FINAL SETTINGS allow_experimental_analyzer = 0; --- only lonely parts -EXPLAIN PIPELINE header=1 SELECT max(val), count(*) FROM with_lonely FINAL WHERE dt < '2022-11-01' SETTINGS allow_experimental_analyzer = 0; --- only lonely parts but max_thread = 1, so reading lonely parts with in-order -EXPLAIN PIPELINE header=1 SELECT max(val), count(*) FROM with_lonely FINAL WHERE dt < '2022-11-01' SETTINGS max_threads = 1, allow_experimental_analyzer = 0; - - -EXPLAIN PIPELINE header=1 SELECT max(val), count(*) FROM with_lonely FINAL SETTINGS allow_experimental_analyzer = 1; -EXPLAIN PIPELINE header=1 SELECT max(val), count(*) FROM with_lonely FINAL WHERE dt < '2022-11-01' SETTINGS allow_experimental_analyzer = 1; -EXPLAIN PIPELINE header=1 SELECT max(val), count(*) FROM with_lonely FINAL WHERE dt < '2022-11-01' SETTINGS max_threads = 1, allow_experimental_analyzer = 1; From 8483d1c701c56db00296ea5ad712e35f527c0579 Mon Sep 17 00:00:00 2001 From: slvrtrn Date: Thu, 31 Aug 2023 13:26:27 +0200 Subject: [PATCH 093/433] Binary protocol tests --- .../integration/mysql_java_client/Dockerfile | 13 +-- .../PreparedStatementsTest.java | 36 ++++---- .../integration/mysql_java_client/Test.java | 3 +- src/Core/MySQL/PacketsProtocolBinary.cpp | 84 ++++++++++++------ src/Core/MySQL/PacketsProtocolText.cpp | 1 + .../Serializations/SerializationDate32.cpp | 14 +-- .../prepared_statements.reference | 82 ++++++++++++++++- .../prepared_statements_test.sql | 86 ++++++++++++++++++ tests/integration/test_mysql_protocol/test.py | 88 +------------------ 9 files changed, 262 insertions(+), 145 deletions(-) create mode 100644 tests/integration/test_mysql_protocol/prepared_statements_test.sql diff --git a/docker/test/integration/mysql_java_client/Dockerfile b/docker/test/integration/mysql_java_client/Dockerfile index 0abf50cd493..75e3c9de138 100644 --- a/docker/test/integration/mysql_java_client/Dockerfile +++ b/docker/test/integration/mysql_java_client/Dockerfile @@ -1,10 +1,10 @@ # docker build -t clickhouse/mysql-java-client . # MySQL Java client docker container -FROM ubuntu:18.04 +FROM ubuntu:22.04 RUN apt-get update && \ - apt-get install -y software-properties-common build-essential openjdk-8-jdk libmysql-java curl + apt-get install -y software-properties-common build-essential openjdk-8-jdk curl RUN rm -rf \ /var/lib/apt/lists/* \ @@ -12,10 +12,11 @@ RUN rm -rf \ /tmp/* \ RUN apt-get clean -ARG ver=5.1.46 -RUN curl -L -o /mysql-connector-java-${ver}.jar https://repo1.maven.org/maven2/mysql/mysql-connector-java/${ver}/mysql-connector-java-${ver}.jar -ENV CLASSPATH=$CLASSPATH:/mysql-connector-java-${ver}.jar +ARG ver=8.1.0 +RUN curl -L -o /mysql-connector-j-${ver}.jar https://repo1.maven.org/maven2/com/mysql/mysql-connector-j/${ver}/mysql-connector-j-${ver}.jar +ENV CLASSPATH=$CLASSPATH:/mysql-connector-j-${ver}.jar WORKDIR /jdbc COPY Test.java Test.java -RUN javac Test.java +COPY PreparedStatementsTest.java PreparedStatementsTest.java +RUN javac Test.java PreparedStatementsTest.java diff --git a/docker/test/integration/mysql_java_client/PreparedStatementsTest.java b/docker/test/integration/mysql_java_client/PreparedStatementsTest.java index 0534f2b761d..cb479e13099 100644 --- a/docker/test/integration/mysql_java_client/PreparedStatementsTest.java +++ b/docker/test/integration/mysql_java_client/PreparedStatementsTest.java @@ -33,20 +33,19 @@ public class PreparedStatementsTest { } } + // useServerPrepStmts allows us to send COM_STMT_PREPARE and COM_STMT_EXECUTE + // to test the binary protocol implementation + String jdbcUrl = String.format("jdbc:mysql://%s:%s/%s?useSSL=false&useServerPrepStmts=true", + host, port, database); + try { Class.forName("com.mysql.cj.jdbc.Driver"); - /* - * useServerPrepStmts allows us to send COM_STMT_PREPARE and COM_STMT_EXECUTE to - * test the binary protocol - */ - String jdbcUrl = String.format("jdbc:mysql://%s:%s/%s?useSSL=false&useServerPrepStmts=true", - host, port, database); Connection conn = DriverManager.getConnection(jdbcUrl, user, password); testSimpleDataTypes(conn); testStringTypes(conn); testDecimalTypes(conn); testMiscTypes(conn); - // testDateTypes(conn); + testDateTypes(conn); conn.close(); } catch (Exception e) { e.printStackTrace(); @@ -99,20 +98,15 @@ public class PreparedStatementsTest { int rowNum = 1; while (rs.next()) { System.out.printf("Row #%d\n", rowNum++); - System.out.printf("Type: %s, value: %s\n", - getMysqlType(rs, "d32"), + System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "d32"), rs.getBigDecimal("d32").toPlainString()); - System.out.printf("Type: %s, value: %s\n", - getMysqlType(rs, "d64"), + System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "d64"), rs.getBigDecimal("d64").toPlainString()); - System.out.printf("Type: %s, value: %s\n", - getMysqlType(rs, "d128_native"), + System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "d128_native"), rs.getBigDecimal("d128_native").toPlainString()); - System.out.printf("Type: %s, value: %s\n", - getMysqlType(rs, "d128_text"), + System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "d128_text"), rs.getString("d128_text")); - System.out.printf("Type: %s, value: %s\n", - getMysqlType(rs, "d256"), + System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "d256"), rs.getString("d256")); } System.out.println(); @@ -127,14 +121,16 @@ public class PreparedStatementsTest { System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "d"), rs.getDate("d")); System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "d32"), rs.getDate("d32")); System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "dt"), rs.getTimestamp("dt")); - System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "dt64"), rs.getTimestamp("dt64")); + System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "dt64_3"), rs.getString("dt64_3")); + System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "dt64_6"), rs.getString("dt64_6")); + System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "dt64_9"), rs.getString("dt64_9")); } System.out.println(); } private static void testMiscTypes(Connection conn) throws SQLException { System.out.println("### testMiscTypes"); - ResultSet rs = conn.prepareStatement("SELECT d FROM ps_misc_types").executeQuery(); + ResultSet rs = conn.prepareStatement("SELECT * FROM ps_misc_types").executeQuery(); int rowNum = 1; while (rs.next()) { System.out.printf("Row #%d\n", rowNum++); @@ -149,7 +145,5 @@ public class PreparedStatementsTest { private static MysqlType getMysqlType(ResultSet rs, String columnLabel) throws SQLException { ResultSetMetaData meta = rs.getMetaData(); return MysqlType.getByJdbcType(meta.getColumnType(rs.findColumn(columnLabel))); - } - } diff --git a/docker/test/integration/mysql_java_client/Test.java b/docker/test/integration/mysql_java_client/Test.java index 2e256d5dc44..752441718a9 100644 --- a/docker/test/integration/mysql_java_client/Test.java +++ b/docker/test/integration/mysql_java_client/Test.java @@ -46,6 +46,7 @@ class JavaConnectorTest { Connection conn = null; Statement stmt = null; try { + Class.forName("com.mysql.cj.jdbc.Driver"); conn = DriverManager.getConnection(jdbcUrl, user, password); stmt = conn.createStatement(); stmt.executeUpdate(CREATE_TABLE_SQL); @@ -69,7 +70,7 @@ class JavaConnectorTest { stmt.close(); conn.close(); - } catch (SQLException e) { + } catch (Exception e) { e.printStackTrace(); System.exit(1); } diff --git a/src/Core/MySQL/PacketsProtocolBinary.cpp b/src/Core/MySQL/PacketsProtocolBinary.cpp index 2b5dce78ab3..ebb256cb462 100644 --- a/src/Core/MySQL/PacketsProtocolBinary.cpp +++ b/src/Core/MySQL/PacketsProtocolBinary.cpp @@ -2,17 +2,18 @@ #include #include #include -#include -#include -#include -#include "Common/logger_useful.h" +#include "Common/LocalDate.h" +#include "Common/LocalDateTime.h" #include "Columns/ColumnLowCardinality.h" #include "Columns/ColumnVector.h" +#include "Columns/ColumnsDateTime.h" #include "DataTypes/DataTypeLowCardinality.h" #include "DataTypes/DataTypeNullable.h" #include "DataTypes/DataTypesNumber.h" #include "Formats/FormatSettings.h" #include "IO/WriteBufferFromString.h" +#include "base/DayNum.h" +#include "base/Decimal.h" #include "base/types.h" namespace DB @@ -29,12 +30,10 @@ namespace MySQLProtocol { /// See https://dev.mysql.com/doc/dev/mysql-server/8.1.0/page_protocol_binary_resultset.html#sect_protocol_binary_resultset_row payload_size = 1 + null_bitmap_size; - LOG_TRACE(&Poco::Logger::get("ResultSetRow"), "Null bitmap size: {}", null_bitmap_size); FormatSettings format_settings; for (size_t i = 0; i < columns.size(); ++i) { ColumnPtr col = columns[i]; - LOG_TRACE(&Poco::Logger::get("col->isNullAt"), "isNullAt: {}, {}", row_num, col->isNullAt(row_num)); if (col->isNullAt(row_num)) { size_t byte = (i + 2) / 8; @@ -75,7 +74,19 @@ namespace MySQLProtocol payload_size += 8; break; case TypeIndex::Date: { - UInt64 value = col->get64(row_num); + UInt16 value = assert_cast &>(*col).getData()[row_num]; + if (value == 0) + { + payload_size += 1; // length only, no other fields + } + else + { + payload_size += 5; + } + break; + } + case TypeIndex::Date32: { + Int32 value = assert_cast &>(*col).getData()[row_num]; if (value == 0) { payload_size += 1; // length only, no other fields @@ -87,15 +98,15 @@ namespace MySQLProtocol break; } case TypeIndex::DateTime: { - UInt64 value = col->get64(row_num); + UInt32 value = assert_cast &>(*col).getData()[row_num]; if (value == 0) { payload_size += 1; // length only, no other fields } else { - Poco::DateTime dt = Poco::DateTime(Poco::Timestamp(value * 1000 * 1000)); - if (dt.second() == 0 && dt.minute() == 0 && dt.hour() == 0) + LocalDateTime ldt = LocalDateTime(value, DateLUT::instance(getDateTimeTimezone(*data_type))); + if (ldt.second() == 0 && ldt.minute() == 0 && ldt.hour() == 0) { payload_size += 5; } @@ -107,7 +118,6 @@ namespace MySQLProtocol break; } default: - LOG_TRACE(&Poco::Logger::get("Type default"), "{} is {}", col->getName(), data_type->getName()); WriteBufferFromOwnString ostr; serializations[i]->serializeText(*columns[i], row_num, ostr, format_settings); payload_size += getLengthEncodedStringSize(ostr.str()); @@ -188,14 +198,33 @@ namespace MySQLProtocol break; } case TypeIndex::Date: { - UInt64 value = assert_cast &>(*col).getData()[row_num]; + UInt16 value = assert_cast &>(*col).getData()[row_num]; if (value != 0) { - Poco::DateTime dt = Poco::DateTime(Poco::Timestamp(value * 1000 * 1000)); + LocalDate ld = LocalDate(DayNum(value)); buffer.write(static_cast(4)); // bytes_following - int year = dt.year(); - int month = dt.month(); - int day = dt.day(); + auto year = ld.year(); + auto month = ld.month(); + auto day = ld.day(); + buffer.write(reinterpret_cast(&year), 2); + buffer.write(reinterpret_cast(&month), 1); + buffer.write(reinterpret_cast(&day), 1); + } + else + { + buffer.write(static_cast(0)); + } + break; + } + case TypeIndex::Date32: { + Int32 value = assert_cast &>(*col).getData()[row_num]; + if (value != 0) + { + LocalDate ld = LocalDate(ExtendedDayNum(value)); + buffer.write(static_cast(4)); // bytes_following + auto year = ld.year(); + auto month = ld.month(); + auto day = ld.day(); buffer.write(reinterpret_cast(&year), 2); buffer.write(reinterpret_cast(&month), 1); buffer.write(reinterpret_cast(&day), 1); @@ -207,24 +236,24 @@ namespace MySQLProtocol break; } case TypeIndex::DateTime: { - UInt64 value = assert_cast &>(*col).getData()[row_num]; + UInt32 value = assert_cast &>(*col).getData()[row_num]; if (value != 0) { - Poco::DateTime dt = Poco::DateTime(Poco::Timestamp(value * 1000 * 1000)); - bool is_date_time = !(dt.hour() == 0 && dt.minute() == 0 && dt.second() == 0); - size_t bytes_following = is_date_time ? 7 : 4; + LocalDateTime ldt = LocalDateTime(value, DateLUT::instance(getDateTimeTimezone(*data_type))); + int year = ldt.year(); + int month = ldt.month(); + int day = ldt.day(); + int hour = ldt.hour(); + int minute = ldt.minute(); + int second = ldt.second(); + bool has_time = !(hour == 0 && minute == 0 && second == 0); + size_t bytes_following = has_time ? 7 : 4; buffer.write(reinterpret_cast(&bytes_following), 1); - int year = dt.year(); - int month = dt.month(); - int day = dt.day(); buffer.write(reinterpret_cast(&year), 2); buffer.write(reinterpret_cast(&month), 1); buffer.write(reinterpret_cast(&day), 1); - if (is_date_time) + if (has_time) { - int hour = dt.hourAMPM(); - int minute = dt.minute(); - int second = dt.second(); buffer.write(reinterpret_cast(&hour), 1); buffer.write(reinterpret_cast(&minute), 1); buffer.write(reinterpret_cast(&second), 1); @@ -236,6 +265,7 @@ namespace MySQLProtocol } break; } + default: writeLengthEncodedString(serialized[i], buffer); break; diff --git a/src/Core/MySQL/PacketsProtocolText.cpp b/src/Core/MySQL/PacketsProtocolText.cpp index bb5412bb5a5..712b203bc2a 100644 --- a/src/Core/MySQL/PacketsProtocolText.cpp +++ b/src/Core/MySQL/PacketsProtocolText.cpp @@ -203,6 +203,7 @@ namespace MySQLProtocol decimals = 31; break; case TypeIndex::Date: + case TypeIndex::Date32: column_type = ColumnType::MYSQL_TYPE_DATE; flags = ColumnDefinitionFlags::BINARY_FLAG; break; diff --git a/src/DataTypes/Serializations/SerializationDate32.cpp b/src/DataTypes/Serializations/SerializationDate32.cpp index 851710de839..68213a14afc 100644 --- a/src/DataTypes/Serializations/SerializationDate32.cpp +++ b/src/DataTypes/Serializations/SerializationDate32.cpp @@ -28,12 +28,14 @@ void SerializationDate32::deserializeTextEscaped(IColumn & column, ReadBuffer & assert_cast(column).getData().push_back(x); } -void SerializationDate32::serializeTextEscaped(const IColumn & column, size_t row_num, WriteBuffer & ostr, const FormatSettings & settings) const +void SerializationDate32::serializeTextEscaped( + const IColumn & column, size_t row_num, WriteBuffer & ostr, const FormatSettings & settings) const { serializeText(column, row_num, ostr, settings); } -void SerializationDate32::serializeTextQuoted(const IColumn & column, size_t row_num, WriteBuffer & ostr, const FormatSettings & settings) const +void SerializationDate32::serializeTextQuoted( + const IColumn & column, size_t row_num, WriteBuffer & ostr, const FormatSettings & settings) const { writeChar('\'', ostr); serializeText(column, row_num, ostr, settings); @@ -46,10 +48,11 @@ void SerializationDate32::deserializeTextQuoted(IColumn & column, ReadBuffer & i assertChar('\'', istr); readDateText(x, istr, time_zone); assertChar('\'', istr); - assert_cast(column).getData().push_back(x); /// It's important to do this at the end - for exception safety. + assert_cast(column).getData().push_back(x); /// It's important to do this at the end - for exception safety. } -void SerializationDate32::serializeTextJSON(const IColumn & column, size_t row_num, WriteBuffer & ostr, const FormatSettings & settings) const +void SerializationDate32::serializeTextJSON( + const IColumn & column, size_t row_num, WriteBuffer & ostr, const FormatSettings & settings) const { writeChar('"', ostr); serializeText(column, row_num, ostr, settings); @@ -65,7 +68,8 @@ void SerializationDate32::deserializeTextJSON(IColumn & column, ReadBuffer & ist assert_cast(column).getData().push_back(x); } -void SerializationDate32::serializeTextCSV(const IColumn & column, size_t row_num, WriteBuffer & ostr, const FormatSettings & settings) const +void SerializationDate32::serializeTextCSV( + const IColumn & column, size_t row_num, WriteBuffer & ostr, const FormatSettings & settings) const { writeChar('"', ostr); serializeText(column, row_num, ostr, settings); diff --git a/tests/integration/test_mysql_protocol/prepared_statements.reference b/tests/integration/test_mysql_protocol/prepared_statements.reference index 257cc5642cb..3e23620c07e 100644 --- a/tests/integration/test_mysql_protocol/prepared_statements.reference +++ b/tests/integration/test_mysql_protocol/prepared_statements.reference @@ -1 +1,81 @@ -foo +### testSimpleDataTypes +Row #1 +Type: TINYINT, value: -128 +Type: SMALLINT, value: -32768 +Type: INT, value: -2147483648 +Type: BIGINT, value: -9223372036854775808 +Type: CHAR, value: -170141183460469231731687303715884105728 +Type: CHAR, value: -57896044618658097711785492504343953926634992332820282019728792003956564819968 +Type: TINYINT, value: 120 +Type: SMALLINT, value: 1234 +Type: INT, value: 51234 +Type: BIGINT, value: 421342 +Type: CHAR, value: 15324355 +Type: CHAR, value: 41345135123432 +Type: FLOAT, value: -0.796896 +Type: DOUBLE, value: -0.113259 +Type: BIT, value: true +Row #2 +Type: TINYINT, value: 127 +Type: SMALLINT, value: 32767 +Type: INT, value: 2147483647 +Type: BIGINT, value: 9223372036854775807 +Type: CHAR, value: 170141183460469231731687303715884105727 +Type: CHAR, value: 57896044618658097711785492504343953926634992332820282019728792003956564819967 +Type: TINYINT, value: 255 +Type: SMALLINT, value: 65535 +Type: INT, value: 4294967295 +Type: BIGINT, value: 18446744073709551615 +Type: CHAR, value: 340282366920938463463374607431768211455 +Type: CHAR, value: 115792089237316195423570985008687907853269984665640564039457584007913129639935 +Type: FLOAT, value: 1.234000 +Type: DOUBLE, value: 3.352451 +Type: BIT, value: false + +### testStringTypes +Row #1 +Type: CHAR, value: 42 +Type: CHAR, value: null +Type: CHAR, value: test +Type: CHAR, value: null +Row #2 +Type: CHAR, value: foo +Type: CHAR, value: bar +Type: CHAR, value: qaz +Type: CHAR, value: qux + +### testDecimalTypes +Row #1 +Type: DECIMAL, value: -1.55 +Type: DECIMAL, value: 6.03 +Type: DECIMAL, value: 5 +Type: CHAR, value: -1224124.23423 +Type: CHAR, value: -54342.3 +Row #2 +Type: DECIMAL, value: 1234567.89 +Type: DECIMAL, value: 123456789123456.789 +Type: DECIMAL, value: 12345678912345678912.1234567891 +Type: CHAR, value: 1234567.8912345678912345678911234567891 +Type: CHAR, value: 12345678912345678912345678911234567891234567891234567891.12345678911234567891 + +### testMiscTypes +Row #1 +Type: CHAR, value: [] +Type: CHAR, value: 9a0ccc06-2578-4861-8534-631c9d40f3f7 +Type: CHAR, value: (0,'') +Type: CHAR, value: {} +Row #2 +Type: CHAR, value: ['foo','bar'] +Type: CHAR, value: 5da5038d-788f-48c6-b510-babb41c538d3 +Type: CHAR, value: (42,'qaz') +Type: CHAR, value: {'qux':144,'text':255} + +### testDateTypes +Row #1 +Type: DATE, value: 2149-06-06 +Type: DATE, value: 2178-04-16 +Type: TIMESTAMP, value: 2106-02-07 06:28:15.0 +Type: CHAR, value: 2106-02-07 06:28:15.123 +Type: CHAR, value: 2106-02-07 06:28:15.123456 +Type: CHAR, value: 2106-02-07 06:28:15.123456789 + diff --git a/tests/integration/test_mysql_protocol/prepared_statements_test.sql b/tests/integration/test_mysql_protocol/prepared_statements_test.sql new file mode 100644 index 00000000000..4fb53a14a61 --- /dev/null +++ b/tests/integration/test_mysql_protocol/prepared_statements_test.sql @@ -0,0 +1,86 @@ +CREATE TABLE ps_simple_data_types +( + i8 Int8, + i16 Int16, + i32 Int32, + i64 Int64, + i128 Int128, + i256 Int256, + ui8 UInt8, + ui16 UInt16, + ui32 UInt32, + ui64 UInt64, + ui128 UInt128, + ui256 UInt256, + f32 Float32, + f64 Float64, + b Boolean +) ENGINE MergeTree ORDER BY i8; + +INSERT INTO ps_simple_data_types +VALUES (127, 32767, 2147483647, 9223372036854775807, 170141183460469231731687303715884105727, + 57896044618658097711785492504343953926634992332820282019728792003956564819967, + 255, 65535, 4294967295, 18446744073709551615, 340282366920938463463374607431768211455, + 115792089237316195423570985008687907853269984665640564039457584007913129639935, + 1.234, 3.35245141223232, FALSE), + (-128, -32768, -2147483648, -9223372036854775808, -170141183460469231731687303715884105728, + -57896044618658097711785492504343953926634992332820282019728792003956564819968, + 120, 1234, 51234, 421342, 15324355, 41345135123432, + -0.7968956, -0.113259, TRUE); + +CREATE TABLE ps_string_types +( + s String, + sn Nullable(String), + lc LowCardinality(String), + nlc LowCardinality(Nullable(String)) +) ENGINE MergeTree ORDER BY s; + +INSERT INTO ps_string_types +VALUES ('foo', 'bar', 'qaz', 'qux'), + ('42', NULL, 'test', NULL); + +CREATE TABLE ps_decimal_types +( + d32 Decimal(9, 2), + d64 Decimal(18, 3), + d128_native Decimal(30, 10), + d128_text Decimal(38, 31), + d256 Decimal(76, 20) +) ENGINE MergeTree ORDER BY d32; + +INSERT INTO ps_decimal_types +VALUES (1234567.89, + 123456789123456.789, + 12345678912345678912.1234567891, + 1234567.8912345678912345678911234567891, + 12345678912345678912345678911234567891234567891234567891.12345678911234567891), + (-1.55, 6.03, 5, -1224124.23423, -54342.3); + +CREATE TABLE ps_misc_types +( + a Array(String), + u UUID, + t Tuple(Int32, String), + m Map(String, Int32) +) ENGINE MergeTree ORDER BY u; + +INSERT INTO ps_misc_types +VALUES (['foo', 'bar'], '5da5038d-788f-48c6-b510-babb41c538d3', (42, 'qaz'), {'qux': 144, 'text': 255}), + ([], '9a0ccc06-2578-4861-8534-631c9d40f3f7', (0, ''), {}); + +CREATE TABLE ps_date_types +( + d Date, + d32 Date32, + dt DateTime, + dt64_3 DateTime64(3, 'UTC'), + dt64_6 DateTime64(6, 'UTC'), + dt64_9 DateTime64(9, 'UTC') +) ENGINE MergeTree ORDER BY d; + +INSERT INTO ps_date_types +VALUES ('2149-06-06', '2178-04-16', '2106-02-07 06:28:15', + '2106-02-07 06:28:15.123', + '2106-02-07 06:28:15.123456', + '2106-02-07 06:28:15.123456789'); diff --git a/tests/integration/test_mysql_protocol/test.py b/tests/integration/test_mysql_protocol/test.py index c0bc178c5bb..19a52da1158 100644 --- a/tests/integration/test_mysql_protocol/test.py +++ b/tests/integration/test_mysql_protocol/test.py @@ -809,91 +809,11 @@ def test_prepared_statements(started_cluster, java_container): with open(os.path.join(SCRIPT_DIR, "prepared_statements.reference")) as fp: reference = fp.read() - node.query(""" - CREATE TABLE ps_simple_data_types - ( - i8 Int8, - i16 Int16, - i32 Int32, - i64 Int64, - i128 Int128, - i256 Int256, - ui8 UInt8, - ui16 UInt16, - ui32 UInt32, - ui64 UInt64, - ui128 UInt128, - ui256 UInt256, - f32 Float32, - f64 Float64, - b Boolean - ) ENGINE MergeTree ORDER BY i8; - """, settings={"password": "123"}) - node.query(""" - INSERT INTO ps_simple_data_types - VALUES (127, 32767, 2147483647, 9223372036854775807, 170141183460469231731687303715884105727, - 57896044618658097711785492504343953926634992332820282019728792003956564819967, - 255, 65535, 4294967295, 18446744073709551615, 340282366920938463463374607431768211455, - 115792089237316195423570985008687907853269984665640564039457584007913129639935, - 1.234, 3.35245141223232, FALSE), - (-128, -32768, -2147483648, -9223372036854775808, -170141183460469231731687303715884105728, - -57896044618658097711785492504343953926634992332820282019728792003956564819968, - 120, 1234, 51234, 421342, 15324355, 41345135123432, - -0.7968956, -0.113259, TRUE); - """, settings={"password": "123"}) + with open(os.path.join(SCRIPT_DIR, "prepared_statements_test.sql")) as sql: + statements = list(filter(lambda s: s != '', map(lambda s: s.strip().replace('\n', ' '), sql.read().split(';')))) - node.query(""" - CREATE TABLE ps_string_types - ( - s String, - sn Nullable(String), - lc LowCardinality(String), - nlc LowCardinality(Nullable(String)) - ) ENGINE MergeTree ORDER BY s; - """, settings={"password": "123"}) - node.query(""" - INSERT INTO ps_string_types - VALUES ('foo', 'bar', 'qaz', 'qux'), - ('42', NULL, 'test', NULL); - """, settings={"password": "123"}) - - node.query(""" - CREATE TABLE ps_decimal_types - ( - d32 Decimal(9, 2), - d64 Decimal(18, 3), - d128_native Decimal(30, 10), - d128_text Decimal(38, 31), - d256 Decimal(76, 20) - ) ENGINE MergeTree ORDER BY d32; - """, settings={"password": "123"}) - node.query(""" - INSERT INTO ps_decimal_types - VALUES (1234567.89, - 123456789123456.789, - 12345678912345678912.1234567891, - 1234567.8912345678912345678911234567891, - 12345678912345678912345678911234567891234567891234567891.12345678911234567891), - (-1.55, 6.03, 5, -1224124.23423, -54342.3); - """, settings={"password": "123"}) - - node.query(""" - CREATE TABLE ps_misc_types - ( - - a Array(String), - u UUID, - t Tuple(Int32, String), - m Map(String, Int32) - ) ENGINE MergeTree ORDER BY u; - """, settings={"password": "123"}) - node.query(""" - SELECT * - FROM ps_misc_types; - INSERT INTO ps_misc_types - VALUES (['foo', 'bar'], '5da5038d-788f-48c6-b510-babb41c538d3', (42, 'qaz'), {'qux': 144, 'text': 255}), - ([], '9a0ccc06-2578-4861-8534-631c9d40f3f7', (0, ''), {}); - """, settings={"password": "123"}) + for statement in statements: + node.query(statement, settings={"password": "123"}) code, (stdout, stderr) = java_container.exec_run( "java PreparedStatementsTest --host {host} --port {port} --user user_with_double_sha1 --password abacaba --database " From 77111a229d6058f3b4010dca7b371e7641fd8b60 Mon Sep 17 00:00:00 2001 From: Anton Popov Date: Thu, 31 Aug 2023 12:59:21 +0000 Subject: [PATCH 094/433] fix clang-tidy --- src/Storages/MergeTree/MergeTreePrefetchedReadPool.cpp | 2 +- src/Storages/MergeTree/MergeTreeReadPoolInOrder.cpp | 1 - 2 files changed, 1 insertion(+), 2 deletions(-) diff --git a/src/Storages/MergeTree/MergeTreePrefetchedReadPool.cpp b/src/Storages/MergeTree/MergeTreePrefetchedReadPool.cpp index d97b674df37..7bd038fe5cf 100644 --- a/src/Storages/MergeTree/MergeTreePrefetchedReadPool.cpp +++ b/src/Storages/MergeTree/MergeTreePrefetchedReadPool.cpp @@ -138,7 +138,7 @@ void MergeTreePrefetchedReadPool::startPrefetches() while (!prefetch_queue.empty()) { - auto & top = prefetch_queue.top(); + const auto & top = prefetch_queue.top(); createPrefetchedReadersForTask(*top.task); #ifndef NDEBUG if (prev.task) diff --git a/src/Storages/MergeTree/MergeTreeReadPoolInOrder.cpp b/src/Storages/MergeTree/MergeTreeReadPoolInOrder.cpp index 37f009beb87..1b621ad5055 100644 --- a/src/Storages/MergeTree/MergeTreeReadPoolInOrder.cpp +++ b/src/Storages/MergeTree/MergeTreeReadPoolInOrder.cpp @@ -65,7 +65,6 @@ MergeTreeReadTaskPtr MergeTreeReadPoolInOrder::getTask(size_t task_idx, MergeTre else { mark_ranges_for_task = std::move(all_mark_ranges); - all_mark_ranges.clear(); } return createTask(per_part_infos[task_idx], std::move(mark_ranges_for_task), previous_task); From 7739dd7157b1cd04d05746514aa9ce83f1b437a0 Mon Sep 17 00:00:00 2001 From: Antonio Andelic Date: Thu, 31 Aug 2023 14:52:42 +0000 Subject: [PATCH 095/433] Fix integration tests --- programs/server/Server.cpp | 2 +- src/Common/ZooKeeper/ZooKeeper.cpp | 4 ++-- src/Coordination/CoordinationSettings.cpp | 3 +++ tests/integration/helpers/cluster.py | 2 +- 4 files changed, 7 insertions(+), 4 deletions(-) diff --git a/programs/server/Server.cpp b/programs/server/Server.cpp index 6f94e483e20..6d49fde1691 100644 --- a/programs/server/Server.cpp +++ b/programs/server/Server.cpp @@ -1425,7 +1425,7 @@ try const auto interserver_listen_hosts = getInterserverListenHosts(config()); const auto listen_try = getListenTry(config()); - if (config().has("keeper_server")) + if (config().has("keeper_server.server_id")) { #if USE_NURAFT //// If we don't have configured connection probably someone trying to use clickhouse-server instead diff --git a/src/Common/ZooKeeper/ZooKeeper.cpp b/src/Common/ZooKeeper/ZooKeeper.cpp index 8b719265869..d099b09827c 100644 --- a/src/Common/ZooKeeper/ZooKeeper.cpp +++ b/src/Common/ZooKeeper/ZooKeeper.cpp @@ -1427,7 +1427,7 @@ void validateZooKeeperConfig(const Poco::Util::AbstractConfiguration & config) bool hasZooKeeperConfig(const Poco::Util::AbstractConfiguration & config) { - return config.has("zookeeper") || config.has("keeper") || (config.has("keeper_server") && config.getBool("keeper_server.use_cluster", true)); + return config.has("zookeeper") || config.has("keeper") || (config.has("keeper_server.raft_configuration") && config.getBool("keeper_server.use_cluster", true)); } String getZooKeeperConfigName(const Poco::Util::AbstractConfiguration & config) @@ -1438,7 +1438,7 @@ String getZooKeeperConfigName(const Poco::Util::AbstractConfiguration & config) if (config.has("keeper")) return "keeper"; - if (config.has("keeper_server") && config.getBool("keeper_server.use_cluster", true)) + if (config.has("keeper_server.raft_configuration") && config.getBool("keeper_server.use_cluster", true)) return "keeper_server"; throw DB::Exception(DB::ErrorCodes::NO_ELEMENTS_IN_CONFIG, "There is no Zookeeper configuration in server config"); diff --git a/src/Coordination/CoordinationSettings.cpp b/src/Coordination/CoordinationSettings.cpp index 61e08f9fde3..aadedf19204 100644 --- a/src/Coordination/CoordinationSettings.cpp +++ b/src/Coordination/CoordinationSettings.cpp @@ -152,6 +152,9 @@ void KeeperConfigurationAndSettings::dump(WriteBufferFromOwnString & buf) const writeText("raft_limits_reconnect_limit=", buf); write_int(static_cast(coordination_settings->raft_limits_reconnect_limit)); + + writeText("async_replication=", buf); + write_bool(coordination_settings->async_replication); } KeeperConfigurationAndSettingsPtr diff --git a/tests/integration/helpers/cluster.py b/tests/integration/helpers/cluster.py index c48bd4303c0..708614ed48c 100644 --- a/tests/integration/helpers/cluster.py +++ b/tests/integration/helpers/cluster.py @@ -4260,7 +4260,7 @@ class ClickHouseInstance: if len(self.custom_dictionaries_paths): write_embedded_config("0_common_enable_dictionaries.xml", self.config_d_dir) - # write_embedded_config("0_common_enable_keeper_async_replication.xml", self.config_d_dir) + write_embedded_config("0_common_enable_keeper_async_replication.xml", self.config_d_dir) logging.debug("Generate and write macros file") macros = self.macros.copy() From 3999994eb4d973ef56af829dac0fba1352d54478 Mon Sep 17 00:00:00 2001 From: robot-clickhouse Date: Thu, 31 Aug 2023 15:07:26 +0000 Subject: [PATCH 096/433] Automatic style fix --- tests/integration/helpers/cluster.py | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/tests/integration/helpers/cluster.py b/tests/integration/helpers/cluster.py index 708614ed48c..1e92bdaa788 100644 --- a/tests/integration/helpers/cluster.py +++ b/tests/integration/helpers/cluster.py @@ -4260,7 +4260,9 @@ class ClickHouseInstance: if len(self.custom_dictionaries_paths): write_embedded_config("0_common_enable_dictionaries.xml", self.config_d_dir) - write_embedded_config("0_common_enable_keeper_async_replication.xml", self.config_d_dir) + write_embedded_config( + "0_common_enable_keeper_async_replication.xml", self.config_d_dir + ) logging.debug("Generate and write macros file") macros = self.macros.copy() From 5f1a75ceae6960598a4f440aa62a04b85bfacf56 Mon Sep 17 00:00:00 2001 From: Antonio Andelic Date: Fri, 1 Sep 2023 07:39:36 +0000 Subject: [PATCH 097/433] Fix tests --- docker/test/upgrade/run.sh | 12 +++++++++++ src/Coordination/Changelog.cpp | 2 +- src/Coordination/CoordinationSettings.cpp | 21 ++++++++++++------- tests/integration/helpers/cluster.py | 7 ++++--- .../test_keeper_four_word_command/test.py | 4 +++- 5 files changed, 33 insertions(+), 13 deletions(-) diff --git a/docker/test/upgrade/run.sh b/docker/test/upgrade/run.sh index 0b30ab9dbf7..886efe3ff33 100644 --- a/docker/test/upgrade/run.sh +++ b/docker/test/upgrade/run.sh @@ -60,6 +60,12 @@ install_packages previous_release_package_folder # available for dump via clickhouse-local configure +# async_replication setting doesn't exist on some older versions +sudo cat /etc/clickhouse-server/config.d/keeper_port.xml \ + | sed "/1<\/async_replication>/d" \ + > /etc/clickhouse-server/config.d/keeper_port.xml.tmp +sudo mv /etc/clickhouse-server/config.d/keeper_port.xml.tmp /etc/clickhouse-server/config.d/keeper_port.xml + # it contains some new settings, but we can safely remove it rm /etc/clickhouse-server/config.d/merge_tree.xml rm /etc/clickhouse-server/config.d/enable_wait_for_shutdown_replicated_tables.xml @@ -82,6 +88,12 @@ sudo cat /etc/clickhouse-server/config.d/keeper_port.xml \ > /etc/clickhouse-server/config.d/keeper_port.xml.tmp sudo mv /etc/clickhouse-server/config.d/keeper_port.xml.tmp /etc/clickhouse-server/config.d/keeper_port.xml +# async_replication setting doesn't exist on some older versions +sudo cat /etc/clickhouse-server/config.d/keeper_port.xml \ + | sed "/1<\/async_replication>/d" \ + > /etc/clickhouse-server/config.d/keeper_port.xml.tmp +sudo mv /etc/clickhouse-server/config.d/keeper_port.xml.tmp /etc/clickhouse-server/config.d/keeper_port.xml + # But we still need default disk because some tables loaded only into it sudo cat /etc/clickhouse-server/config.d/s3_storage_policy_by_default.xml \ | sed "s|
s3
|
s3
default|" \ diff --git a/src/Coordination/Changelog.cpp b/src/Coordination/Changelog.cpp index 2af68173588..7030775904c 100644 --- a/src/Coordination/Changelog.cpp +++ b/src/Coordination/Changelog.cpp @@ -1021,7 +1021,7 @@ void Changelog::writeThread() const auto flush_logs = [&](const auto & flush) { - LOG_TRACE(log, "Flushing {} logs", pending_appends); + LOG_TEST(log, "Flushing {} logs", pending_appends); { std::lock_guard writer_lock(writer_mutex); diff --git a/src/Coordination/CoordinationSettings.cpp b/src/Coordination/CoordinationSettings.cpp index aadedf19204..8688f6f5a40 100644 --- a/src/Coordination/CoordinationSettings.cpp +++ b/src/Coordination/CoordinationSettings.cpp @@ -22,16 +22,19 @@ void CoordinationSettings::loadFromConfig(const String & config_elem, const Poco Poco::Util::AbstractConfiguration::Keys config_keys; config.keys(config_elem, config_keys); - try + for (const String & key : config_keys) { - for (const String & key : config_keys) + try + { set(key, config.getString(config_elem + "." + key)); - } - catch (Exception & e) - { - if (e.code() == ErrorCodes::UNKNOWN_SETTING) - e.addMessage("in Coordination settings config"); - throw; + } + catch (Exception & e) + { + if (e.code() == ErrorCodes::UNKNOWN_SETTING) + LOG_WARNING(&Poco::Logger::get("CoordinationSettings"), "Found unknown coordination setting in config: '{}'", key); + else + throw; + } } } @@ -134,6 +137,8 @@ void KeeperConfigurationAndSettings::dump(WriteBufferFromOwnString & buf) const write_int(coordination_settings->max_requests_batch_size); writeText("max_requests_batch_bytes_size=", buf); write_int(coordination_settings->max_requests_batch_bytes_size); + writeText("max_flush_batch_size=", buf); + write_int(coordination_settings->max_flush_batch_size); writeText("max_request_queue_size=", buf); write_int(coordination_settings->max_request_queue_size); writeText("max_requests_quick_batch_size=", buf); diff --git a/tests/integration/helpers/cluster.py b/tests/integration/helpers/cluster.py index 1e92bdaa788..7ee86d74d35 100644 --- a/tests/integration/helpers/cluster.py +++ b/tests/integration/helpers/cluster.py @@ -4260,9 +4260,10 @@ class ClickHouseInstance: if len(self.custom_dictionaries_paths): write_embedded_config("0_common_enable_dictionaries.xml", self.config_d_dir) - write_embedded_config( - "0_common_enable_keeper_async_replication.xml", self.config_d_dir - ) + if self.tag == "latest": + write_embedded_config( + "0_common_enable_keeper_async_replication.xml", self.config_d_dir + ) logging.debug("Generate and write macros file") macros = self.macros.copy() diff --git a/tests/integration/test_keeper_four_word_command/test.py b/tests/integration/test_keeper_four_word_command/test.py index 1d5bc6a6541..0de7de8c3be 100644 --- a/tests/integration/test_keeper_four_word_command/test.py +++ b/tests/integration/test_keeper_four_word_command/test.py @@ -279,8 +279,9 @@ def test_cmd_conf(started_cluster): assert result["stale_log_gap"] == "10000" assert result["fresh_log_gap"] == "200" - assert result["max_requests_batch_size"] == "100" + assert result["max_requests_batch_size"] == "1000" assert result["max_requests_batch_bytes_size"] == "102400" + assert result["max_flush_batch_size"] == "1000" assert result["max_request_queue_size"] == "100000" assert result["max_requests_quick_batch_size"] == "100" assert result["quorum_reads"] == "false" @@ -290,6 +291,7 @@ def test_cmd_conf(started_cluster): assert result["compress_snapshots_with_zstd_format"] == "true" assert result["configuration_change_tries_count"] == "20" + assert result["async_replication"] == "true" finally: close_keeper_socket(client) From 0c9653dac5ebeb239b2e4098d995e5d2608d1c66 Mon Sep 17 00:00:00 2001 From: yariks5s Date: Fri, 1 Sep 2023 14:01:06 +0000 Subject: [PATCH 098/433] implemented for formatDateTime and toString --- src/Functions/FunctionsConversion.h | 81 ++++++++++++------- src/Functions/formatDateTime.cpp | 38 ++++++--- .../02863_non_const_timezone_check.reference | 20 +++++ .../02863_non_const_timezone_check.sql | 19 +++++ 4 files changed, 120 insertions(+), 38 deletions(-) create mode 100644 tests/queries/0_stateless/02863_non_const_timezone_check.reference create mode 100644 tests/queries/0_stateless/02863_non_const_timezone_check.sql diff --git a/src/Functions/FunctionsConversion.h b/src/Functions/FunctionsConversion.h index 102d4339554..8a0b82107c3 100644 --- a/src/Functions/FunctionsConversion.h +++ b/src/Functions/FunctionsConversion.h @@ -203,18 +203,15 @@ struct ConvertImpl } } - if constexpr (std::is_same_v && std::is_same_v) + if constexpr (std::is_same_v && std::is_same_v) { - static_assert(std::is_same_v, "UInt128 and UUID types must be same"); - if constexpr (std::endian::native == std::endian::little) - { - vec_to[i].items[1] = vec_from[i].toUnderType().items[0]; - vec_to[i].items[0] = vec_from[i].toUnderType().items[1]; - } - else - { - vec_to[i] = vec_from[i].toUnderType(); - } + static_assert( + std::is_same_v, + "UInt128 and UUID types must be same"); + + vec_to[i].items[1] = vec_from[i].toUnderType().items[0]; + vec_to[i].items[0] = vec_from[i].toUnderType().items[1]; + continue; } @@ -886,20 +883,32 @@ struct ConvertImplconvertToFullColumnIfConst(); + + const auto & col_with_type_and_name = columnGetNested(datetime_arg); + ColumnUInt8::MutablePtr null_map = copyNullMap(datetime_arg.column); - const auto & col_with_type_and_name = columnGetNested(arguments[0]); const auto & type = static_cast(*col_with_type_and_name.type); const DateLUTImpl * time_zone = nullptr; + const ColumnConst * time_zone_column = nullptr; + + if (arguments.size() > 1) + time_zone_column = checkAndGetColumnConst(arguments[1].column.get()); + else if (arguments.size() == 1) + time_zone = &DateLUT::instance(); if constexpr (std::is_same_v || std::is_same_v) time_zone = &DateLUT::instance(); /// For argument of Date or DateTime type, second argument with time zone could be specified. if constexpr (std::is_same_v || std::is_same_v) { - auto non_null_args = createBlockWithNestedColumns(arguments); - time_zone = &extractTimeZoneFromFunctionArguments(non_null_args, 1, 0); + if (time_zone_column) + { + auto non_null_args = createBlockWithNestedColumns(arguments); + time_zone = &extractTimeZoneFromFunctionArguments(non_null_args, 1, 0); + } } if (const auto col_from = checkAndGetColumn(col_with_type_and_name.column.get())) @@ -930,6 +939,13 @@ struct ConvertImpl 1) + { + if (!arguments[1].column.get()->getDataAt(i).toString().empty()) + time_zone = &DateLUT::instance(arguments[1].column.get()->getDataAt(i).toString()); + else + throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Provided time zone must be non-empty and be a valid time zone"); + } bool is_ok = FormatImpl::template execute(vec_from[i], write_buffer, &type, time_zone); null_map->getData()[i] |= !is_ok; writeChar(0, write_buffer); @@ -940,6 +956,13 @@ struct ConvertImpl 1) + { + if (!arguments[1].column.get()->getDataAt(i).toString().empty()) + time_zone = &DateLUT::instance(arguments[1].column.get()->getDataAt(i).toString()); + else + throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Provided time zone must be non-empty and be a valid time zone"); + } FormatImpl::template execute(vec_from[i], write_buffer, &type, time_zone); writeChar(0, write_buffer); offsets_to[i] = write_buffer.count(); @@ -1844,20 +1867,20 @@ public: mandatory_args.push_back({"scale", &isNativeInteger, &isColumnConst, "const Integer"}); } - // toString(DateTime or DateTime64, [timezone: String (const/non-const)]) + // toString(DateTime or DateTime64, [timezone: String]) if ((std::is_same_v && !arguments.empty() && (isDateTime64(arguments[0].type) || isDateTime(arguments[0].type))) - // toUnixTimestamp(value[, timezone : String(const/non-const)]) + // toUnixTimestamp(value[, timezone : String]) || std::is_same_v - // toDate(value[, timezone : String(const/non-const)]) + // toDate(value[, timezone : String]) || std::is_same_v // TODO: shall we allow timestamp argument for toDate? DateTime knows nothing about timezones and this argument is ignored below. - // toDate32(value[, timezone : String(const/non-const)]) + // toDate32(value[, timezone : String]) || std::is_same_v - // toDateTime(value[, timezone: String(const/non-const)]) + // toDateTime(value[, timezone: String]) || std::is_same_v - // toDateTime64(value, scale : Integer[, timezone: String(const/non-const)]) + // toDateTime64(value, scale : Integer[, timezone: String]) || std::is_same_v) - { - optional_args.push_back({"timezone", &isString, nullptr, "const String/String"}); + { + optional_args.push_back({"timezone", &isString, nullptr, "String"}); } validateFunctionArgumentTypes(*this, arguments, mandatory_args, optional_args); @@ -1895,13 +1918,13 @@ public: if (to_datetime64 || scale != 0) /// toDateTime('xxxx-xx-xx xx:xx:xx', 0) return DateTime return std::make_shared(scale, - extractTimeZoneNameFromFunctionArguments(arguments, timezone_arg_position, 0, true)); + extractTimeZoneNameFromFunctionArguments(arguments, timezone_arg_position, 0, false)); - return std::make_shared(extractTimeZoneNameFromFunctionArguments(arguments, timezone_arg_position, 0, true)); + return std::make_shared(extractTimeZoneNameFromFunctionArguments(arguments, timezone_arg_position, 0, false)); } if constexpr (std::is_same_v) - return std::make_shared(extractTimeZoneNameFromFunctionArguments(arguments, timezone_arg_position, 0, true)); + return std::make_shared(extractTimeZoneNameFromFunctionArguments(arguments, timezone_arg_position, 0, false)); else if constexpr (std::is_same_v) throw Exception(ErrorCodes::LOGICAL_ERROR, "Unexpected branch in code of conversion function: it is a bug."); else @@ -2142,7 +2165,7 @@ public: bool useDefaultImplementationForConstants() const override { return true; } bool canBeExecutedOnDefaultArguments() const override { return false; } - ColumnNumbers getArgumentsThatAreAlwaysConstant() const override { return {1}; } + ColumnNumbers getArgumentsThatAreAlwaysConstant() const override { return {}; } DataTypePtr getReturnTypeImpl(const ColumnsWithTypeAndName & arguments) const override { @@ -2814,9 +2837,9 @@ protected: bool useDefaultImplementationForNulls() const override { return false; } /// CAST(Nothing, T) -> T bool useDefaultImplementationForNothing() const override { return false; } - bool useDefaultImplementationForConstants() const override { return true; } + bool useDefaultImplementationForConstants() const override { return false; } bool useDefaultImplementationForLowCardinalityColumns() const override { return false; } - ColumnNumbers getArgumentsThatAreAlwaysConstant() const override { return {1}; } + // ColumnNumbers getArgumentsThatAreAlwaysConstant() const override { return {1}; } private: WrapperType wrapper_function; diff --git a/src/Functions/formatDateTime.cpp b/src/Functions/formatDateTime.cpp index 50772866648..7c4b9ea3d36 100644 --- a/src/Functions/formatDateTime.cpp +++ b/src/Functions/formatDateTime.cpp @@ -746,7 +746,7 @@ public: bool isSuitableForShortCircuitArgumentsExecution(const DataTypesWithConstInfo & /*arguments*/) const override { return false; } - ColumnNumbers getArgumentsThatAreAlwaysConstant() const override { return {1, 2}; } + ColumnNumbers getArgumentsThatAreAlwaysConstant() const override { return {1}; } bool isVariadic() const override { return true; } size_t getNumberOfArguments() const override { return 0; } @@ -855,11 +855,18 @@ public: template ColumnPtr executeType(const ColumnsWithTypeAndName & arguments, const DataTypePtr &) const { - auto * times = checkAndGetColumn(arguments[0].column.get()); + auto non_const_datetime = arguments[0].column->convertToFullColumnIfConst(); + auto * times = checkAndGetColumn(non_const_datetime.get()); if (!times) return nullptr; const ColumnConst * format_column = checkAndGetColumnConst(arguments[1].column.get()); + const ColumnConst * time_zone_column = nullptr; + const DateLUTImpl * time_zone = nullptr; + if (arguments.size() == 2) + time_zone = &DateLUT::instance(); + else if (arguments.size() > 2) + time_zone_column = checkAndGetColumnConst(arguments[2].column.get()); if (!format_column) throw Exception(ErrorCodes::ILLEGAL_COLUMN, "Illegal column {} of second ('format') argument of function {}. Must be constant string.", @@ -893,15 +900,19 @@ public: String out_template; size_t out_template_size = parseFormat(format, instructions, scale, mysql_with_only_fixed_length_formatters, out_template); - const DateLUTImpl * time_zone_tmp = nullptr; if (castType(arguments[0].type.get(), [&]([[maybe_unused]] const auto & type) { return true; })) - time_zone_tmp = &extractTimeZoneFromFunctionArguments(arguments, 2, 0); + { + if (time_zone_column) + time_zone = &extractTimeZoneFromFunctionArguments(arguments, 2, 0); + } else if (std::is_same_v || std::is_same_v) - time_zone_tmp = &extractTimeZoneFromFunctionArguments(arguments, 2, 0); + { + if (time_zone_column) + time_zone = &extractTimeZoneFromFunctionArguments(arguments, 2, 0); + } else - time_zone_tmp = &DateLUT::instance(); + time_zone = &DateLUT::instance(); - const DateLUTImpl & time_zone = *time_zone_tmp; const auto & vec = times->getData(); auto col_res = ColumnString::create(); @@ -941,6 +952,13 @@ public: auto * pos = begin; for (size_t i = 0; i < vec.size(); ++i) { + if (!time_zone_column && arguments.size() > 2) + { + if (!arguments[2].column.get()->getDataAt(i).toString().empty()) + time_zone = &DateLUT::instance(arguments[2].column.get()->getDataAt(i).toString()); + else + throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Provided time zone must be non-empty and be a valid time zone"); + } if constexpr (std::is_same_v) { auto c = DecimalUtils::split(vec[i], scale); @@ -954,12 +972,14 @@ public: } for (auto & instruction : instructions) - instruction.perform(pos, static_cast(c.whole), c.fractional, scale, time_zone); + { + instruction.perform(pos, static_cast(c.whole), c.fractional, scale, *time_zone); + } } else { for (auto & instruction : instructions) - instruction.perform(pos, static_cast(vec[i]), 0, 0, time_zone); + instruction.perform(pos, static_cast(vec[i]), 0, 0, *time_zone); } *pos++ = '\0'; diff --git a/tests/queries/0_stateless/02863_non_const_timezone_check.reference b/tests/queries/0_stateless/02863_non_const_timezone_check.reference new file mode 100644 index 00000000000..024b0ff35a4 --- /dev/null +++ b/tests/queries/0_stateless/02863_non_const_timezone_check.reference @@ -0,0 +1,20 @@ +173000 + 083000 + 173000 + 083000 + 173000 + 173000 + 083000 + 173000 + 083000 + 173000 + 2023-08-25 17:30:00 + 2023-08-25 08:30:00 + 2023-08-25 17:30:00 + 2023-08-25 08:30:00 + 2023-08-25 17:30:00 + 2023-08-25 17:30:00 + 2023-08-25 08:30:00 + 2023-08-25 17:30:00 + 2023-08-25 08:30:00 + 2023-08-25 17:30:00 \ No newline at end of file diff --git a/tests/queries/0_stateless/02863_non_const_timezone_check.sql b/tests/queries/0_stateless/02863_non_const_timezone_check.sql new file mode 100644 index 00000000000..cc54cea9ca0 --- /dev/null +++ b/tests/queries/0_stateless/02863_non_const_timezone_check.sql @@ -0,0 +1,19 @@ +DROP TABLE IF EXISTS Dates; + +CREATE TABLE Dates (date DateTime('UTC')) ENGINE = MergeTree() ORDER BY date; + +INSERT INTO Dates VALUES ('2023-08-25 15:30:00'); + +SELECT formatDateTime((SELECT date FROM Dates), '%H%i%S', number % 2 ? 'America/Los_Angeles' : 'Europe/Amsterdam') FROM numbers(5); + +SELECT formatDateTime((SELECT materialize(date) FROM Dates), '%H%i%S', number % 2 ? 'America/Los_Angeles' : 'Europe/Amsterdam') FROM numbers(5); + +SELECT formatDateTime((SELECT materialize(date) FROM Dates), '%H%i%S', number % 2 ? '' : 'Europe/Amsterdam') FROM numbers(5); -- {serverError 43} + +SELECT toString((SELECT date FROM Dates), number % 2 ? 'America/Los_Angeles' : 'Europe/Amsterdam') FROM numbers(5); + +SELECT toString((SELECT materialize(date) FROM Dates), number % 2 ? 'America/Los_Angeles' : 'Europe/Amsterdam') FROM numbers(5); + +SELECT toString((SELECT materialize(date) FROM Dates), number % 2 ? 'America/Los_Angeles' : '') FROM numbers(5); -- {serverError 43} + +DROP TABLE Dates; From c9f1ff03db74e1462adce7fd28b287c0dcb45a79 Mon Sep 17 00:00:00 2001 From: yariks5s Date: Fri, 1 Sep 2023 14:14:00 +0000 Subject: [PATCH 099/433] Minor fixes --- src/Functions/FunctionsConversion.h | 4 +- .../extractTimeZoneFromFunctionArguments.cpp | 4 +- .../02863_non_const_timezone_check.reference | 38 +++++++++---------- 3 files changed, 23 insertions(+), 23 deletions(-) diff --git a/src/Functions/FunctionsConversion.h b/src/Functions/FunctionsConversion.h index 8a0b82107c3..bb730507233 100644 --- a/src/Functions/FunctionsConversion.h +++ b/src/Functions/FunctionsConversion.h @@ -2165,7 +2165,7 @@ public: bool useDefaultImplementationForConstants() const override { return true; } bool canBeExecutedOnDefaultArguments() const override { return false; } - ColumnNumbers getArgumentsThatAreAlwaysConstant() const override { return {}; } + ColumnNumbers getArgumentsThatAreAlwaysConstant() const override { return {1}; } DataTypePtr getReturnTypeImpl(const ColumnsWithTypeAndName & arguments) const override { @@ -2839,7 +2839,7 @@ protected: bool useDefaultImplementationForNothing() const override { return false; } bool useDefaultImplementationForConstants() const override { return false; } bool useDefaultImplementationForLowCardinalityColumns() const override { return false; } - // ColumnNumbers getArgumentsThatAreAlwaysConstant() const override { return {1}; } + ColumnNumbers getArgumentsThatAreAlwaysConstant() const override { return {1}; } private: WrapperType wrapper_function; diff --git a/src/Functions/extractTimeZoneFromFunctionArguments.cpp b/src/Functions/extractTimeZoneFromFunctionArguments.cpp index 38ce0467aee..7168c68c9c9 100644 --- a/src/Functions/extractTimeZoneFromFunctionArguments.cpp +++ b/src/Functions/extractTimeZoneFromFunctionArguments.cpp @@ -30,11 +30,11 @@ std::string extractTimeZoneNameFromColumn(const IColumn * column, const String & } -std::string extractTimeZoneNameFromFunctionArguments(const ColumnsWithTypeAndName & arguments, size_t time_zone_arg_num, size_t datetime_arg_num, [[maybe_unused]] bool allow_nonconst_timezone_arguments) +std::string extractTimeZoneNameFromFunctionArguments(const ColumnsWithTypeAndName & arguments, size_t time_zone_arg_num, size_t datetime_arg_num, bool allow_nonconst_timezone_arguments) { /// Explicit time zone may be passed in last argument. if ((arguments.size() == time_zone_arg_num + 1) - && (arguments[time_zone_arg_num].column)) + && (!allow_nonconst_timezone_arguments || arguments[time_zone_arg_num].column)) { return extractTimeZoneNameFromColumn(arguments[time_zone_arg_num].column.get(), arguments[time_zone_arg_num].name); } diff --git a/tests/queries/0_stateless/02863_non_const_timezone_check.reference b/tests/queries/0_stateless/02863_non_const_timezone_check.reference index 024b0ff35a4..7efacab94d3 100644 --- a/tests/queries/0_stateless/02863_non_const_timezone_check.reference +++ b/tests/queries/0_stateless/02863_non_const_timezone_check.reference @@ -1,20 +1,20 @@ 173000 - 083000 - 173000 - 083000 - 173000 - 173000 - 083000 - 173000 - 083000 - 173000 - 2023-08-25 17:30:00 - 2023-08-25 08:30:00 - 2023-08-25 17:30:00 - 2023-08-25 08:30:00 - 2023-08-25 17:30:00 - 2023-08-25 17:30:00 - 2023-08-25 08:30:00 - 2023-08-25 17:30:00 - 2023-08-25 08:30:00 - 2023-08-25 17:30:00 \ No newline at end of file +083000 +173000 +083000 +173000 +173000 +083000 +173000 +083000 +173000 +2023-08-25 17:30:00 +2023-08-25 08:30:00 +2023-08-25 17:30:00 +2023-08-25 08:30:00 +2023-08-25 17:30:00 +2023-08-25 17:30:00 +2023-08-25 08:30:00 +2023-08-25 17:30:00 +2023-08-25 08:30:00 +2023-08-25 17:30:00 From 1e01aa917a3e9001055c645a496b280a9f6ee5cd Mon Sep 17 00:00:00 2001 From: yariks5s Date: Fri, 1 Sep 2023 14:18:29 +0000 Subject: [PATCH 100/433] Updated FunctionsConversion.h --- src/Functions/FunctionsConversion.h | 19 +++++++++++-------- 1 file changed, 11 insertions(+), 8 deletions(-) diff --git a/src/Functions/FunctionsConversion.h b/src/Functions/FunctionsConversion.h index bb730507233..4dc64d16f09 100644 --- a/src/Functions/FunctionsConversion.h +++ b/src/Functions/FunctionsConversion.h @@ -203,15 +203,18 @@ struct ConvertImpl } } - if constexpr (std::is_same_v && std::is_same_v) + if constexpr (std::is_same_v && std::is_same_v) { - static_assert( - std::is_same_v, - "UInt128 and UUID types must be same"); - - vec_to[i].items[1] = vec_from[i].toUnderType().items[0]; - vec_to[i].items[0] = vec_from[i].toUnderType().items[1]; - + static_assert(std::is_same_v, "UInt128 and UUID types must be same"); + if constexpr (std::endian::native == std::endian::little) + { + vec_to[i].items[1] = vec_from[i].toUnderType().items[0]; + vec_to[i].items[0] = vec_from[i].toUnderType().items[1]; + } + else + { + vec_to[i] = vec_from[i].toUnderType(); + } continue; } From 278cc494a249b8979ebd70c11aebbb65138790bb Mon Sep 17 00:00:00 2001 From: Yarik Briukhovetskyi <114298166+yariks5s@users.noreply.github.com> Date: Fri, 1 Sep 2023 17:40:04 +0200 Subject: [PATCH 101/433] Updated FunctionsConversion.h --- src/Functions/FunctionsConversion.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/Functions/FunctionsConversion.h b/src/Functions/FunctionsConversion.h index 4dc64d16f09..1e047e9b6ff 100644 --- a/src/Functions/FunctionsConversion.h +++ b/src/Functions/FunctionsConversion.h @@ -2840,7 +2840,7 @@ protected: bool useDefaultImplementationForNulls() const override { return false; } /// CAST(Nothing, T) -> T bool useDefaultImplementationForNothing() const override { return false; } - bool useDefaultImplementationForConstants() const override { return false; } + bool useDefaultImplementationForConstants() const override { return true; } bool useDefaultImplementationForLowCardinalityColumns() const override { return false; } ColumnNumbers getArgumentsThatAreAlwaysConstant() const override { return {1}; } From ef4dcf909fa0e8b447fb881e273021b256e85ae2 Mon Sep 17 00:00:00 2001 From: serxa Date: Fri, 1 Sep 2023 16:23:15 +0000 Subject: [PATCH 102/433] fix build --- src/Disks/ObjectStorages/DiskObjectStorage.cpp | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/Disks/ObjectStorages/DiskObjectStorage.cpp b/src/Disks/ObjectStorages/DiskObjectStorage.cpp index b755a4e0b2a..8d7e9c60e31 100644 --- a/src/Disks/ObjectStorages/DiskObjectStorage.cpp +++ b/src/Disks/ObjectStorages/DiskObjectStorage.cpp @@ -66,8 +66,8 @@ DiskObjectStorage::DiskObjectStorage( , metadata_storage(std::move(metadata_storage_)) , object_storage(std::move(object_storage_)) , send_metadata(config.getBool(config_prefix + ".send_metadata", false)) - , read_resource_name(config.getString(config_prefix + ".read_resource", "") - , write_resource_name(config.getString(config_prefix + ".write_resource", "") + , read_resource_name(config.getString(config_prefix + ".read_resource", "")) + , write_resource_name(config.getString(config_prefix + ".write_resource", "")) , metadata_helper(std::make_unique(this, ReadSettings{})) {} From 14e2032d929f325ad2188032df50a2ab199d69a8 Mon Sep 17 00:00:00 2001 From: serxa Date: Fri, 1 Sep 2023 16:28:03 +0000 Subject: [PATCH 103/433] add documentation --- .../mergetree-family/mergetree.md | 4 + docs/en/operations/workload-scheduling.md | 149 ++++++++++++++++++ 2 files changed, 153 insertions(+) create mode 100644 docs/en/operations/workload-scheduling.md diff --git a/docs/en/engines/table-engines/mergetree-family/mergetree.md b/docs/en/engines/table-engines/mergetree-family/mergetree.md index 4f506126682..5d0f9f5d6c6 100644 --- a/docs/en/engines/table-engines/mergetree-family/mergetree.md +++ b/docs/en/engines/table-engines/mergetree-family/mergetree.md @@ -1139,6 +1139,8 @@ Optional parameters: - `s3_max_put_burst` — Max number of requests that can be issued simultaneously before hitting request per second limit. By default (`0` value) equals to `s3_max_put_rps`. - `s3_max_get_rps` — Maximum GET requests per second rate before throttling. Default value is `0` (unlimited). - `s3_max_get_burst` — Max number of requests that can be issued simultaneously before hitting request per second limit. By default (`0` value) equals to `s3_max_get_rps`. +- `read_resource` — Resource name to be used for [scheduling](/docs/en/operations/workload-scheduling.md) of read requests to this disk. Default value is empty string (IO scheduling is not enabled for this disk). +- `write_resource` — Resource name to be used for [scheduling](/docs/en/operations/workload-scheduling.md) of write requests to this disk. Default value is empty string (IO scheduling is not enabled for this disk). ### Configuring the cache @@ -1251,6 +1253,8 @@ Other parameters: * `cache_enabled` - Allows to cache mark and index files on local FS. Default value is `true`. * `cache_path` - Path on local FS where to store cached mark and index files. Default value is `/var/lib/clickhouse/disks//cache/`. * `skip_access_check` - If true, disk access checks will not be performed on disk start-up. Default value is `false`. +* `read_resource` — Resource name to be used for [scheduling](/docs/en/operations/workload-scheduling.md) of read requests to this disk. Default value is empty string (IO scheduling is not enabled for this disk). +* `write_resource` — Resource name to be used for [scheduling](/docs/en/operations/workload-scheduling.md) of write requests to this disk. Default value is empty string (IO scheduling is not enabled for this disk). Examples of working configurations can be found in integration tests directory (see e.g. [test_merge_tree_azure_blob_storage](https://github.com/ClickHouse/ClickHouse/blob/master/tests/integration/test_merge_tree_azure_blob_storage/configs/config.d/storage_conf.xml) or [test_azure_blob_storage_zero_copy_replication](https://github.com/ClickHouse/ClickHouse/blob/master/tests/integration/test_azure_blob_storage_zero_copy_replication/configs/config.d/storage_conf.xml)). diff --git a/docs/en/operations/workload-scheduling.md b/docs/en/operations/workload-scheduling.md new file mode 100644 index 00000000000..1761f99b469 --- /dev/null +++ b/docs/en/operations/workload-scheduling.md @@ -0,0 +1,149 @@ +--- +slug: /en/operations/workload-scheduling +sidebar_position: 69 +sidebar_label: "Workload scheduling" +title: "Workload scheduling" +--- + +When ClickHouse execute multiple queries simultaneously, they may be using shared resources (e.g. disks). Scheduling constraints and policies can be applied to regulate how resources are utilized and shared between different workloads. For every resource a scheduling hierarchy can be configured. Hierarchy root represents a resource, while leafs are queues, holding requests that exceed resource capacity. + +:::note +Currently only remote disk IO can be scheduled using described method. For CPU scheduling see settings about thread pools and [`concurrent_threads_soft_limit_num`](server-configuration-parameters/settings.md#concurrent_threads_soft_limit_num). For flexible memory limits see [Memory overcommit](settings/memory-overcommit.md) +::: + +## Disk configuration + +To enable IO scheduling for a specific disk, you have to specify `read_resource` and/or `write_resource` in storage configuration. It says ClickHouse what resource should be used for every read and write requests with given disk. Read and write resource can refer to the same resource name, which is useful for local SSDs or HDDs. Multiple different disks also can refer to the same resource, which is useful for remote disks: if you want to be able to allow fair division of network bandwidth between e.g. "production" and "development" workloads. + +Example: +```xml + + + ... + + + s3 + https://clickhouse-public-datasets.s3.amazonaws.com/my-bucket/root-path/ + your_access_key_id + your_secret_access_key + network_read + network_write + + + + + +
+ s3 +
+ + + + + +``` + +## Workload markup {#workload_markup} + +Queries can be marked with setting `workload` to distinguish different workloads. If `workload` is not set, than value "default" is used. Note that you are able to specify the other value using settings profiles. Setting constraints can be used to make `workload` constant if you want all queries from the user to be marked with fixed value of `workload` setting. + +Let's consider an example of a system with two different workloads: "production" and "development". + +```sql +SELECT count() FROM my_table WHERE value = 42 SETTINGS workload = 'production' +SELECT count() FROM my_table WHERE value = 13 SETTINGS workload = 'development' +``` + +## Resource scheduling hierarchy + +From the standpoint of scheduling subsystem a resource represents a hierarchy of scheduling nodes. + +```mermaid +graph TD + subgraph network_read + nr_root(("/")) + -->|100 concurrent requests| nr_fair("fair") + -->|75% bandwidth| nr_prod["prod"] + nr_fair + -->|25% bandwidth| nr_dev["dev"] + end + + subgraph network_write + nw_root(("/")) + -->|100 concurrent requests| nw_fair("fair") + -->|75% bandwidth| nw_prod["prod"] + nw_fair + -->|25% bandwidth| nw_dev["dev"] + end +``` + +**Possible node types:** +* inflight_limit (constraint) - blocks if either number of concurrent in-flight requests exceeds `max_requests`, or their total cost exceeds `max_cost`; must have a single child. +* fair (policy) - selects the next request to serve from one of its children nodes according to max-min fairness; children nodes can specify `weight` (default is 1). +* priority (policy) - selects the next request to serve from one of its children nodes according to static priorities (lower value means higher priority); children nodes can specify `priority` (default is 0). +* fifo (queue) - leaf of the hierarchy capable of holding requests that exceed resource capacity. + +The following example shows how to define IO scheduling hierarchies shown in the picture: + +```xml + + + + + inflight_limit + 100 + + + fair + + + fifo + 3 + + + fifo + + + + + inflight_limit + 100 + + + fair + + + fifo + 3 + + + fifo + + + + +``` + +## Workload classifiers + +Workload classifiers are used to define mapping from `workload` specified by a query into leaf-queues that should be used for specific resources. At the moment, workload classification is simple: only static mapping is available. + +Example: +```xml + + + + /fair/prod + /fair/prod + + + /fair/dev + /fair/dev + + + /fair/dev + /fair/dev + + + +``` From 1d270fca2817cf50d3c2c3a8136b76b386b5f698 Mon Sep 17 00:00:00 2001 From: Yarik Briukhovetskyi <114298166+yariks5s@users.noreply.github.com> Date: Fri, 1 Sep 2023 18:30:39 +0200 Subject: [PATCH 104/433] Updated FunctionsConversion.h --- src/Functions/FunctionsConversion.h | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) diff --git a/src/Functions/FunctionsConversion.h b/src/Functions/FunctionsConversion.h index 1e047e9b6ff..34430130fba 100644 --- a/src/Functions/FunctionsConversion.h +++ b/src/Functions/FunctionsConversion.h @@ -1945,7 +1945,14 @@ public: } bool useDefaultImplementationForConstants() const override { return true; } - + ColumnNumbers getArgumentsThatAreAlwaysConstant() const override + { + if constexpr (std::is_same_v) + return {}; + else if constexpr (std::is_same_v) + return {2}; + return {1}; + } bool canBeExecutedOnDefaultArguments() const override { return false; } ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override From 096b44e1e5dcf46e081b9d8a123462c222c8fd0b Mon Sep 17 00:00:00 2001 From: pufit Date: Fri, 1 Sep 2023 23:34:47 -0400 Subject: [PATCH 105/433] Implementing `reconfig` command. Parser improvements --- programs/keeper-client/Commands.cpp | 93 ++++++++++++++++++------- programs/keeper-client/Commands.h | 17 +++++ programs/keeper-client/KeeperClient.cpp | 63 ++++++++++------- programs/keeper-client/KeeperClient.h | 4 +- programs/keeper-client/Parser.cpp | 33 ++++----- src/Common/ZooKeeper/ZooKeeper.cpp | 39 +++++++++++ src/Common/ZooKeeper/ZooKeeper.h | 13 ++++ 7 files changed, 189 insertions(+), 73 deletions(-) diff --git a/programs/keeper-client/Commands.cpp b/programs/keeper-client/Commands.cpp index c822a631798..e265d7f4858 100644 --- a/programs/keeper-client/Commands.cpp +++ b/programs/keeper-client/Commands.cpp @@ -9,11 +9,11 @@ namespace DB bool LSCommand::parse(IParser::Pos & pos, std::shared_ptr & node, Expected & expected) const { - String arg; - if (!parseKeeperPath(pos, expected, arg)) + String path; + if (!parseKeeperPath(pos, expected, path)) return true; - node->args.push_back(std::move(arg)); + node->args.push_back(std::move(path)); return true; } @@ -42,11 +42,11 @@ void LSCommand::execute(const ASTKeeperQuery * query, KeeperClient * client) con bool CDCommand::parse(IParser::Pos & pos, std::shared_ptr & node, Expected & expected) const { - String arg; - if (!parseKeeperPath(pos, expected, arg)) + String path; + if (!parseKeeperPath(pos, expected, path)) return true; - node->args.push_back(std::move(arg)); + node->args.push_back(std::move(path)); return true; } @@ -64,11 +64,12 @@ void CDCommand::execute(const ASTKeeperQuery * query, KeeperClient * client) con bool SetCommand::parse(IParser::Pos & pos, std::shared_ptr & node, Expected & expected) const { - String arg; - if (!parseKeeperPath(pos, expected, arg)) + String path; + if (!parseKeeperPath(pos, expected, path)) return false; - node->args.push_back(std::move(arg)); + node->args.push_back(std::move(path)); + String arg; if (!parseKeeperArg(pos, expected, arg)) return false; node->args.push_back(std::move(arg)); @@ -93,11 +94,12 @@ void SetCommand::execute(const ASTKeeperQuery * query, KeeperClient * client) co bool CreateCommand::parse(IParser::Pos & pos, std::shared_ptr & node, Expected & expected) const { - String arg; - if (!parseKeeperPath(pos, expected, arg)) + String path; + if (!parseKeeperPath(pos, expected, path)) return false; - node->args.push_back(std::move(arg)); + node->args.push_back(std::move(path)); + String arg; if (!parseKeeperArg(pos, expected, arg)) return false; node->args.push_back(std::move(arg)); @@ -143,10 +145,10 @@ void TouchCommand::execute(const ASTKeeperQuery * query, KeeperClient * client) bool GetCommand::parse(IParser::Pos & pos, std::shared_ptr & node, Expected & expected) const { - String arg; - if (!parseKeeperPath(pos, expected, arg)) + String path; + if (!parseKeeperPath(pos, expected, path)) return false; - node->args.push_back(std::move(arg)); + node->args.push_back(std::move(path)); return true; } @@ -158,11 +160,11 @@ void GetCommand::execute(const ASTKeeperQuery * query, KeeperClient * client) co bool GetStatCommand::parse(IParser::Pos & pos, std::shared_ptr & node, Expected & expected) const { - String arg; - if (!parseKeeperPath(pos, expected, arg)) + String path; + if (!parseKeeperPath(pos, expected, path)) return true; - node->args.push_back(std::move(arg)); + node->args.push_back(std::move(path)); return true; } @@ -325,10 +327,10 @@ void FindBigFamily::execute(const ASTKeeperQuery * query, KeeperClient * client) bool RMCommand::parse(IParser::Pos & pos, std::shared_ptr & node, Expected & expected) const { - String arg; - if (!parseKeeperPath(pos, expected, arg)) + String path; + if (!parseKeeperPath(pos, expected, path)) return false; - node->args.push_back(std::move(arg)); + node->args.push_back(std::move(path)); return true; } @@ -340,10 +342,10 @@ void RMCommand::execute(const ASTKeeperQuery * query, KeeperClient * client) con bool RMRCommand::parse(IParser::Pos & pos, std::shared_ptr & node, Expected & expected) const { - String arg; - if (!parseKeeperPath(pos, expected, arg)) + String path; + if (!parseKeeperPath(pos, expected, path)) return false; - node->args.push_back(std::move(arg)); + node->args.push_back(std::move(path)); return true; } @@ -355,6 +357,49 @@ void RMRCommand::execute(const ASTKeeperQuery * query, KeeperClient * client) co [client, path]{ client->zookeeper->removeRecursive(path); }); } +bool ReconfigCommand::parse(IParser::Pos & pos, std::shared_ptr & node, DB::Expected & expected) const +{ + ReconfigCommand::Operation operation; + if (ParserKeyword{"ADD"}.ignore(pos, expected)) + operation = ReconfigCommand::Operation::ADD; + else if (ParserKeyword{"REMOVE"}.ignore(pos, expected)) + operation = ReconfigCommand::Operation::REMOVE; + else if (ParserKeyword{"SET"}.ignore(pos, expected)) + operation = ReconfigCommand::Operation::SET; + else + return false; + + node->args.push_back(operation); + ParserToken{TokenType::Whitespace}.ignore(pos); + + String arg; + if (!parseKeeperArg(pos, expected, arg)) + return false; + node->args.push_back(std::move(arg)); + + return true; +} + +void ReconfigCommand::execute(const DB::ASTKeeperQuery * query, DB::KeeperClient * client) const +{ + String joining; + String leaving; + String new_members; + + auto operation = query->args[0].get(); + if (operation == static_cast(ReconfigCommand::Operation::ADD)) + joining = query->args[1].safeGet(); + else if (operation == static_cast(ReconfigCommand::Operation::REMOVE)) + leaving = query->args[1].safeGet(); + else if (operation == static_cast(ReconfigCommand::Operation::SET)) + new_members = query->args[1].safeGet(); + else + UNREACHABLE(); + + auto response = client->zookeeper->reconfig(joining, leaving, new_members); + std::cout << response.value << '\n'; +} + bool HelpCommand::parse(IParser::Pos & /* pos */, std::shared_ptr & /* node */, Expected & /* expected */) const { return true; diff --git a/programs/keeper-client/Commands.h b/programs/keeper-client/Commands.h index 7d64fef1738..c2f7bd9b896 100644 --- a/programs/keeper-client/Commands.h +++ b/programs/keeper-client/Commands.h @@ -177,6 +177,23 @@ class RMRCommand : public IKeeperClientCommand String getHelpMessage() const override { return "{} -- Recursively deletes path. Confirmation required"; } }; +class ReconfigCommand : public IKeeperClientCommand +{ + enum class Operation : Int64 { + ADD = 0, + REMOVE = 1, + SET = 2, + }; + + String getName() const override { return "reconfig"; } + + bool parse(IParser::Pos & pos, std::shared_ptr & node, Expected & expected) const override; + + void execute(const ASTKeeperQuery * query, KeeperClient * client) const override; + + String getHelpMessage() const override { return "{} \"\" [version] -- Reconfigures a ZooKeeper cluster. See https://clickhouse.com/docs/en/guides/sre/keeper/clickhouse-keeper#reconfiguration"; } +}; + class HelpCommand : public IKeeperClientCommand { String getName() const override { return "help"; } diff --git a/programs/keeper-client/KeeperClient.cpp b/programs/keeper-client/KeeperClient.cpp index 893be4d25d1..021baf75f64 100644 --- a/programs/keeper-client/KeeperClient.cpp +++ b/programs/keeper-client/KeeperClient.cpp @@ -83,8 +83,11 @@ std::vector KeeperClient::getCompletions(const String & prefix) const void KeeperClient::askConfirmation(const String & prompt, std::function && callback) { + if (!ask_confirmation) + return callback(); + std::cout << prompt << " Continue?\n"; - need_confirmation = true; + waiting_confirmation = true; confirmation_callback = callback; } @@ -184,6 +187,7 @@ void KeeperClient::initialize(Poco::Util::Application & /* self */) std::make_shared(), std::make_shared(), std::make_shared(), + std::make_shared(), std::make_shared(), std::make_shared(), }); @@ -216,17 +220,6 @@ void KeeperClient::initialize(Poco::Util::Application & /* self */) EventNotifier::init(); } -void KeeperClient::executeQuery(const String & query) -{ - std::vector queries; - boost::algorithm::split(queries, query, boost::is_any_of(";")); - - for (const auto & query_text : queries) - { - if (!query_text.empty()) - processQueryText(query_text); - } -} bool KeeperClient::processQueryText(const String & text) { @@ -235,29 +228,44 @@ bool KeeperClient::processQueryText(const String & text) try { - if (need_confirmation) + if (waiting_confirmation) { - need_confirmation = false; + waiting_confirmation = false; if (text.size() == 1 && (text == "y" || text == "Y")) confirmation_callback(); return true; } KeeperParser parser; - String message; const char * begin = text.data(); - ASTPtr res = tryParseQuery(parser, begin, begin + text.size(), message, true, "", false, 0, 0, false); + const char * end = begin + text.size(); - if (!res) + while (begin < end) { - std::cerr << message << "\n"; - return true; + String message; + ASTPtr res = tryParseQuery( + parser, + begin, + end, + /* out_error_message = */ message, + /* hilite = */ true, + /* description = */ "", + /* allow_multi_statements = */ true, + /* max_query_size = */ 0, + /* max_parser_depth = */ 0, + /* skip_insignificant = */ false); + + if (!res) + { + std::cerr << message << "\n"; + return true; + } + + auto * query = res->as(); + + auto command = KeeperClient::commands.find(query->command); + command->second->execute(query, this); } - - auto * query = res->as(); - - auto command = KeeperClient::commands.find(query->command); - command->second->execute(query, this); } catch (Coordination::Exception & err) { @@ -286,7 +294,7 @@ void KeeperClient::runInteractive() while (true) { String prompt; - if (need_confirmation) + if (waiting_confirmation) prompt = "[y/n] "; else prompt = cwd.string() + " :) "; @@ -320,7 +328,10 @@ int KeeperClient::main(const std::vector & /* args */) zookeeper = std::make_unique(zk_args); if (config().has("query")) - executeQuery(config().getString("query")); + { + ask_confirmation = false; + processQueryText(config().getString("query")); + } else runInteractive(); diff --git a/programs/keeper-client/KeeperClient.h b/programs/keeper-client/KeeperClient.h index e7fa5cdc5fe..4a20a45de6e 100644 --- a/programs/keeper-client/KeeperClient.h +++ b/programs/keeper-client/KeeperClient.h @@ -50,7 +50,6 @@ public: protected: void runInteractive(); bool processQueryText(const String & text); - void executeQuery(const String & query); void loadCommands(std::vector && new_commands); @@ -61,7 +60,8 @@ protected: zkutil::ZooKeeperArgs zk_args; - bool need_confirmation = false; + bool ask_confirmation = true; + bool waiting_confirmation = false; std::vector registered_commands_and_four_letter_words; }; diff --git a/programs/keeper-client/Parser.cpp b/programs/keeper-client/Parser.cpp index 3420ccb2219..228ce93b78d 100644 --- a/programs/keeper-client/Parser.cpp +++ b/programs/keeper-client/Parser.cpp @@ -7,43 +7,34 @@ namespace DB bool parseKeeperArg(IParser::Pos & pos, Expected & expected, String & result) { - expected.add(pos, getTokenName(TokenType::BareWord)); - - if (pos->type == TokenType::BareWord) + if (pos->type == TokenType::QuotedIdentifier || pos->type == TokenType::StringLiteral) { - result = String(pos->begin, pos->end); - ++pos; + if (!parseIdentifierOrStringLiteral(pos, expected, result)) + return false; + ParserToken{TokenType::Whitespace}.ignore(pos); - return true; } - bool status = parseIdentifierOrStringLiteral(pos, expected, result); - ParserToken{TokenType::Whitespace}.ignore(pos); - return status; -} - -bool parseKeeperPath(IParser::Pos & pos, Expected & expected, String & path) -{ - expected.add(pos, "path"); - - if (pos->type == TokenType::QuotedIdentifier || pos->type == TokenType::StringLiteral) - return parseIdentifierOrStringLiteral(pos, expected, path); - - String result; - while (pos->type != TokenType::Whitespace && pos->type != TokenType::EndOfStream) + while (pos->type != TokenType::Whitespace && pos->type != TokenType::EndOfStream && pos->type != TokenType::Semicolon) { result.append(pos->begin, pos->end); ++pos; } + ParserToken{TokenType::Whitespace}.ignore(pos); if (result.empty()) return false; - path = result; return true; } +bool parseKeeperPath(IParser::Pos & pos, Expected & expected, String & path) +{ + expected.add(pos, "path"); + return parseKeeperArg(pos, expected, path); +} + bool KeeperParser::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) { auto query = std::make_shared(); diff --git a/src/Common/ZooKeeper/ZooKeeper.cpp b/src/Common/ZooKeeper/ZooKeeper.cpp index 8b719265869..adc78035abb 100644 --- a/src/Common/ZooKeeper/ZooKeeper.cpp +++ b/src/Common/ZooKeeper/ZooKeeper.cpp @@ -877,6 +877,24 @@ void ZooKeeper::handleEphemeralNodeExistence(const std::string & path, const std } } +Coordination::ReconfigResponse ZooKeeper::reconfig( + const std::string & joining, + const std::string & leaving, + const std::string & new_members, + int32_t version) +{ + auto future_result = asyncReconfig(joining, leaving, new_members, version); + + if (future_result.wait_for(std::chrono::milliseconds(args.operation_timeout_ms)) != std::future_status::ready) + { + impl->finalize(fmt::format("Operation timeout on {}", Coordination::OpNum::Reconfig)); + throw KeeperException(Coordination::Error::ZOPERATIONTIMEOUT); + } + + return future_result.get(); +} + + ZooKeeperPtr ZooKeeper::startNewSession() const { return std::make_shared(args, zk_log); @@ -1226,6 +1244,27 @@ std::future ZooKeeper::asyncSync(const std::string & return future; } +std::future ZooKeeper::asyncReconfig( + const std::string & joining, + const std::string & leaving, + const std::string & new_members, + int32_t version) +{ + auto promise = std::make_shared>(); + auto future = promise->get_future(); + + auto callback = [promise](const Coordination::ReconfigResponse & response) mutable + { + if (response.error != Coordination::Error::ZOK) + promise->set_exception(std::make_exception_ptr(KeeperException(response.error))); + else + promise->set_value(response); + }; + + impl->reconfig(joining, leaving, new_members, version, std::move(callback)); + return future; +} + void ZooKeeper::finalize(const String & reason) { impl->finalize(reason); diff --git a/src/Common/ZooKeeper/ZooKeeper.h b/src/Common/ZooKeeper/ZooKeeper.h index 1126cc189b5..4b598147301 100644 --- a/src/Common/ZooKeeper/ZooKeeper.h +++ b/src/Common/ZooKeeper/ZooKeeper.h @@ -449,6 +449,12 @@ public: /// disappear automatically after 3x session_timeout. void handleEphemeralNodeExistence(const std::string & path, const std::string & fast_delete_if_equal_value); + Coordination::ReconfigResponse reconfig( + const std::string & joining, + const std::string & leaving, + const std::string & new_members, + int32_t version = -1); + /// Async interface (a small subset of operations is implemented). /// /// Usage: @@ -529,6 +535,13 @@ public: const std::string & path, Coordination::ListRequestType list_request_type = Coordination::ListRequestType::ALL); + using FutureReconfig = std::future; + FutureReconfig asyncReconfig( + const std::string & joining, + const std::string & leaving, + const std::string & new_members, + int32_t version = -1); + void finalize(const String & reason); void setZooKeeperLog(std::shared_ptr zk_log_); From adcab639a47edc19e2c336aa21c2a00ce4b8e24f Mon Sep 17 00:00:00 2001 From: serxa Date: Sun, 3 Sep 2023 21:48:33 +0000 Subject: [PATCH 106/433] add `system.scheduler` table for introspection --- src/IO/IResourceManager.h | 5 +- src/IO/ISchedulerConstraint.h | 3 + src/IO/ISchedulerNode.h | 11 +- src/IO/ISchedulerQueue.h | 6 + src/IO/Resource/DynamicResourceManager.cpp | 30 ++++- src/IO/Resource/DynamicResourceManager.h | 3 +- src/IO/Resource/FairPolicy.h | 24 ++++ src/IO/Resource/FifoQueue.h | 18 +++ src/IO/Resource/PriorityPolicy.h | 9 ++ src/IO/Resource/SemaphoreConstraint.h | 29 ++++- src/IO/Resource/StaticResourceManager.h | 5 + src/IO/ResourceBudget.h | 5 + src/IO/SchedulerRoot.h | 8 ++ .../System/StorageSystemScheduler.cpp | 103 ++++++++++++++++++ src/Storages/System/StorageSystemScheduler.h | 22 ++++ src/Storages/System/attachSystemTables.cpp | 2 + 16 files changed, 276 insertions(+), 7 deletions(-) create mode 100644 src/Storages/System/StorageSystemScheduler.cpp create mode 100644 src/Storages/System/StorageSystemScheduler.h diff --git a/src/IO/IResourceManager.h b/src/IO/IResourceManager.h index ff372698ed5..7a078c565d8 100644 --- a/src/IO/IResourceManager.h +++ b/src/IO/IResourceManager.h @@ -7,7 +7,7 @@ #include #include -#include +#include namespace DB { @@ -46,6 +46,9 @@ public: /// Obtain a classifier instance required to get access to resources. /// Note that it holds resource configuration, so should be destructed when query is done. virtual ClassifierPtr acquire(const String & classifier_name) = 0; + + /// For introspection, see `system.scheduler` table + virtual void forEachNode(std::function visitor) = 0; }; using ResourceManagerPtr = std::shared_ptr; diff --git a/src/IO/ISchedulerConstraint.h b/src/IO/ISchedulerConstraint.h index 47f6905e265..16a421f1240 100644 --- a/src/IO/ISchedulerConstraint.h +++ b/src/IO/ISchedulerConstraint.h @@ -44,6 +44,9 @@ public: } } + /// For introspection of current state (true = satisfied, false = violated) + virtual bool isSatisfied(); + protected: // Reference to nearest parent that is also derived from ISchedulerConstraint. // Request can traverse through multiple constraints while being dequeue from hierarchy, diff --git a/src/IO/ISchedulerNode.h b/src/IO/ISchedulerNode.h index 1c33c033744..943d41d95d2 100644 --- a/src/IO/ISchedulerNode.h +++ b/src/IO/ISchedulerNode.h @@ -3,6 +3,7 @@ #include #include #include +#include "base/types.h" #include #include @@ -176,7 +177,10 @@ public: /// Returns true iff node is active virtual bool isActive() = 0; - /// Returns the first request to be executed as the first component of resuting pair. + /// Returns number of active children + virtual size_t activeChildren() = 0; + + /// Returns the first request to be executed as the first component of resulting pair. /// The second pair component is `true` iff node is still active after dequeueing. virtual std::pair dequeueRequest() = 0; @@ -215,6 +219,11 @@ public: String basename; SchedulerNodeInfo info; ISchedulerNode * parent = nullptr; + + /// Introspection + std::atomic dequeued_requests{0}; + std::atomic dequeued_cost{0}; + std::atomic busy_periods{0}; }; using SchedulerNodePtr = std::shared_ptr; diff --git a/src/IO/ISchedulerQueue.h b/src/IO/ISchedulerQueue.h index fc2f3943d26..2e190a529e3 100644 --- a/src/IO/ISchedulerQueue.h +++ b/src/IO/ISchedulerQueue.h @@ -50,6 +50,12 @@ public: /// Should be called outside of scheduling subsystem, implementation must be thread-safe. virtual void enqueueRequest(ResourceRequest * request) = 0; + /// For introspection + ResourceCost getBudget() const + { + return budget.get(); + } + private: // Allows multiple consumers to synchronize with common "debit/credit" balance. // 1) (positive) to avoid wasting of allocated but not used resource (e.g in case of a failure); diff --git a/src/IO/Resource/DynamicResourceManager.cpp b/src/IO/Resource/DynamicResourceManager.cpp index df0de6575f4..964f4f662e6 100644 --- a/src/IO/Resource/DynamicResourceManager.cpp +++ b/src/IO/Resource/DynamicResourceManager.cpp @@ -9,6 +9,7 @@ #include #include +#include namespace DB { @@ -217,13 +218,36 @@ void DynamicResourceManager::updateConfiguration(const Poco::Util::AbstractConfi ClassifierPtr DynamicResourceManager::acquire(const String & classifier_name) { // Acquire a reference to the current state - StatePtr state_; + StatePtr state_ref; { std::lock_guard lock{mutex}; - state_ = state; + state_ref = state; } - return std::make_shared(state_, classifier_name); + return std::make_shared(state_ref, classifier_name); +} + +void DynamicResourceManager::forEachNode(std::function visitor) +{ + // Acquire a reference to the current state + StatePtr state_ref; + { + std::lock_guard lock{mutex}; + state_ref = state; + } + + std::promise promise; + auto future = promise.get_future(); + scheduler.event_queue->enqueue([state_ref, visitor, &promise] + { + for (auto & [name, resource] : state_ref->resources) + for (auto & [path, node] : resource->nodes) + visitor(name, path, node.type, node.ptr); + promise.set_value(); + }); + + // Block until execution is done in the scheduler thread + future.get(); } void registerDynamicResourceManager(ResourceManagerFactory & factory) diff --git a/src/IO/Resource/DynamicResourceManager.h b/src/IO/Resource/DynamicResourceManager.h index aa1147f1fb2..10d784503ca 100644 --- a/src/IO/Resource/DynamicResourceManager.h +++ b/src/IO/Resource/DynamicResourceManager.h @@ -19,7 +19,7 @@ namespace DB * `ClassifierPtr` is acquired and held. * * Manager can update configuration after initialization. During update, new version of resources are also - * attached to scheduler, so multiple version can coexist for a short perid. This will violate constraints + * attached to scheduler, so multiple version can coexist for a short period. This will violate constraints * (e.g. in-fly-limit), because different version have independent nodes to impose constraints, the same * violation will apply to fairness. Old version exists as long as there is at least one classifier * instance referencing it. Classifiers are typically attached to queries and will be destructed with them. @@ -30,6 +30,7 @@ public: DynamicResourceManager(); void updateConfiguration(const Poco::Util::AbstractConfiguration & config) override; ClassifierPtr acquire(const String & classifier_name) override; + void forEachNode(std::function visitor) override; private: /// Holds everything required to work with one specific configuration diff --git a/src/IO/Resource/FairPolicy.h b/src/IO/Resource/FairPolicy.h index 9c0c78f057c..7e632efc8fd 100644 --- a/src/IO/Resource/FairPolicy.h +++ b/src/IO/Resource/FairPolicy.h @@ -6,6 +6,7 @@ #include #include +#include #include #include @@ -176,8 +177,11 @@ public: max_vruntime = 0; } system_vruntime = max_vruntime; + busy_periods++; } + dequeued_requests++; + dequeued_cost += request->cost; return {request, heap_size > 0}; } @@ -186,12 +190,32 @@ public: return heap_size > 0; } + size_t activeChildren() override + { + return heap_size; + } + void activateChild(ISchedulerNode * child) override { // Find this child; this is O(1), thanks to inactive index we hold in `parent.idx` activateChildImpl(child->info.parent.idx); } + // For introspection + double getSystemVRuntime() const + { + return system_vruntime; + } + + std::optional getChildVRuntime(ISchedulerNode * child) const + { + for (const auto & item : items) { + if (child == item.child) + return item.vruntime; + } + return std::nullopt; + } + private: void activateChildImpl(size_t inactive_idx) { diff --git a/src/IO/Resource/FifoQueue.h b/src/IO/Resource/FifoQueue.h index f3ff15ad461..e0689e3c8da 100644 --- a/src/IO/Resource/FifoQueue.h +++ b/src/IO/Resource/FifoQueue.h @@ -39,6 +39,7 @@ public: { std::unique_lock lock(mutex); request->enqueue_ns = clock_gettime_ns(); + queue_cost += request->cost; bool was_empty = requests.empty(); requests.push_back(request); if (was_empty) @@ -52,6 +53,11 @@ public: return {nullptr, false}; ResourceRequest * result = requests.front(); requests.pop_front(); + if (requests.empty()) + busy_periods++; + queue_cost -= result->cost; + dequeued_requests++; + dequeued_cost += result->cost; return {result, !requests.empty()}; } @@ -61,6 +67,11 @@ public: return !requests.empty(); } + size_t activeChildren() override + { + return 0; + } + void activateChild(ISchedulerNode *) override { assert(false); // queue cannot have children @@ -83,8 +94,15 @@ public: return nullptr; } + std::pair getQueueLengthAndCost() + { + std::unique_lock lock(mutex); + return {requests.size(), queue_cost}; + } + private: std::mutex mutex; + Int64 queue_cost = 0; std::deque requests; }; diff --git a/src/IO/Resource/PriorityPolicy.h b/src/IO/Resource/PriorityPolicy.h index 3c091dcc85a..0a06a4fdc09 100644 --- a/src/IO/Resource/PriorityPolicy.h +++ b/src/IO/Resource/PriorityPolicy.h @@ -113,8 +113,12 @@ public: { std::pop_heap(items.begin(), items.end()); items.pop_back(); + if (items.empty()) + busy_periods++; } + dequeued_requests++; + dequeued_cost += request->cost; return {request, !items.empty()}; } @@ -123,6 +127,11 @@ public: return !items.empty(); } + size_t activeChildren() override + { + return items.size(); + } + void activateChild(ISchedulerNode * child) override { bool activate_parent = items.empty(); diff --git a/src/IO/Resource/SemaphoreConstraint.h b/src/IO/Resource/SemaphoreConstraint.h index 237e63eaddb..26437d3da87 100644 --- a/src/IO/Resource/SemaphoreConstraint.h +++ b/src/IO/Resource/SemaphoreConstraint.h @@ -78,7 +78,10 @@ public: requests++; cost += request->cost; child_active = child_now_active; - + if (!active()) + busy_periods++; + dequeued_requests++; + dequeued_cost += request->cost; return {request, active()}; } @@ -113,6 +116,30 @@ public: return active(); } + size_t activeChildren() override + { + std::unique_lock lock(mutex); + return child_active ? 1 : 0; + } + + bool isSatisfied() override + { + std::unique_lock lock(mutex); + return satisfied(); + } + + std::pair getInflights() + { + std::unique_lock lock(mutex); + return {requests, cost}; + } + + std::pair getLimits() + { + std::unique_lock lock(mutex); + return {max_requests, max_cost}; + } + private: bool satisfied() const { diff --git a/src/IO/Resource/StaticResourceManager.h b/src/IO/Resource/StaticResourceManager.h index 066dbf4ebf8..b8712b22b73 100644 --- a/src/IO/Resource/StaticResourceManager.h +++ b/src/IO/Resource/StaticResourceManager.h @@ -22,6 +22,11 @@ public: ClassifierPtr acquire(const String & classifier_name) override; + void forEachNode(std::function visitor) override + { + UNUSED(visitor); + } + private: struct Resource { diff --git a/src/IO/ResourceBudget.h b/src/IO/ResourceBudget.h index 7f67f9cfc10..0adad45ba91 100644 --- a/src/IO/ResourceBudget.h +++ b/src/IO/ResourceBudget.h @@ -48,6 +48,11 @@ public: available.fetch_add(estimated_cost - real_cost); } + ResourceCost get() const + { + return available.load(); + } + private: std::atomic available = 0; // requested - consumed }; diff --git a/src/IO/SchedulerRoot.h b/src/IO/SchedulerRoot.h index f9af2099b8c..75a1b3af72a 100644 --- a/src/IO/SchedulerRoot.h +++ b/src/IO/SchedulerRoot.h @@ -156,6 +156,8 @@ public: else current = current->next; // Just move round-robin pointer + dequeued_requests++; + dequeued_cost += request->cost; return {request, current != nullptr}; } @@ -164,6 +166,11 @@ public: return current != nullptr; } + size_t activeChildren() override + { + return 0; + } + void activateChild(ISchedulerNode * child) override { activate(TResource::get(child->info)); @@ -205,6 +212,7 @@ private: value->next = nullptr; value->prev = nullptr; current = nullptr; + busy_periods++; return; } else // Just move current to next to avoid invalidation diff --git a/src/Storages/System/StorageSystemScheduler.cpp b/src/Storages/System/StorageSystemScheduler.cpp new file mode 100644 index 00000000000..bb73619b972 --- /dev/null +++ b/src/Storages/System/StorageSystemScheduler.cpp @@ -0,0 +1,103 @@ +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include "IO/ResourceRequest.h" + + +namespace DB +{ + +NamesAndTypesList StorageSystemScheduler::getNamesAndTypes() +{ + NamesAndTypesList names_and_types{ + {"resource", std::make_shared()}, + {"path", std::make_shared()}, + {"type", std::make_shared()}, + {"weight", std::make_shared()}, + {"priority", std::make_shared()}, + {"is_active", std::make_shared()}, + {"active_children", std::make_shared()}, + {"dequeued_requests", std::make_shared()}, + {"dequeued_cost", std::make_shared()}, + {"busy_periods", std::make_shared()}, + {"vruntime", std::make_shared(std::make_shared())}, + {"system_vruntime", std::make_shared(std::make_shared())}, + {"queue_length", std::make_shared(std::make_shared())}, + {"queue_cost", std::make_shared(std::make_shared())}, + {"budget", std::make_shared(std::make_shared())}, + {"is_satisfied", std::make_shared(std::make_shared())}, + {"inflight_requests", std::make_shared(std::make_shared())}, + {"inflight_cost", std::make_shared(std::make_shared())}, + {"max_requests", std::make_shared(std::make_shared())}, + {"max_cost", std::make_shared(std::make_shared())}, + }; + return names_and_types; +} + + +void StorageSystemScheduler::fillData(MutableColumns & res_columns, ContextPtr context, const SelectQueryInfo &) const +{ + context->getResourceManager()->forEachNode([&] (const String & resource, const String & path, const String & type, const SchedulerNodePtr & node) { + size_t i = 0; + res_columns[i++]->insert(resource); + res_columns[i++]->insert(path); + res_columns[i++]->insert(type); + res_columns[i++]->insert(node->info.weight); + res_columns[i++]->insert(node->info.priority.value); + res_columns[i++]->insert(node->isActive()); + res_columns[i++]->insert(node->activeChildren()); + res_columns[i++]->insert(node->dequeued_requests.load()); + res_columns[i++]->insert(node->dequeued_cost.load()); + res_columns[i++]->insert(node->busy_periods.load()); + + Field vruntime; + Field system_vruntime; + Field queue_length; + Field queue_cost; + Field budget; + Field is_satisfied; + Field inflight_requests; + Field inflight_cost; + Field max_requests; + Field max_cost; + + if (auto parent = dynamic_cast(node->parent)) + vruntime = parent->getChildVRuntime(node.get()); + if (auto ptr = dynamic_cast(node.get())) + system_vruntime = ptr->getSystemVRuntime(); + if (auto ptr = dynamic_cast(node.get())) + std::tie(queue_length, queue_cost) = ptr->getQueueLengthAndCost(); + if (auto ptr = dynamic_cast(node.get())) + budget = ptr->getBudget(); + if (auto ptr = dynamic_cast(node.get())) + is_satisfied = ptr->isSatisfied(); + if (auto ptr = dynamic_cast(node.get())) + { + std::tie(inflight_requests, inflight_cost) = ptr->getInflights(); + std::tie(max_requests, max_cost) = ptr->getLimits(); + } + + res_columns[i++]->insert(vruntime); + res_columns[i++]->insert(system_vruntime); + res_columns[i++]->insert(queue_length); + res_columns[i++]->insert(queue_cost); + res_columns[i++]->insert(budget); + res_columns[i++]->insert(is_satisfied); + res_columns[i++]->insert(inflight_requests); + res_columns[i++]->insert(inflight_cost); + res_columns[i++]->insert(max_requests); + res_columns[i++]->insert(max_cost); + }); +} + +} diff --git a/src/Storages/System/StorageSystemScheduler.h b/src/Storages/System/StorageSystemScheduler.h new file mode 100644 index 00000000000..31d14862209 --- /dev/null +++ b/src/Storages/System/StorageSystemScheduler.h @@ -0,0 +1,22 @@ +#pragma once + +#include + + +namespace DB +{ +class Context; + +/// Implements `system.scheduler` table, which allows you to get information about scheduling nodes. +class StorageSystemScheduler final : public IStorageSystemOneBlock +{ +public: + std::string getName() const override { return "SystemScheduler"; } + static NamesAndTypesList getNamesAndTypes(); + +protected: + using IStorageSystemOneBlock::IStorageSystemOneBlock; + void fillData(MutableColumns & res_columns, ContextPtr context, const SelectQueryInfo &) const override; +}; + +} diff --git a/src/Storages/System/attachSystemTables.cpp b/src/Storages/System/attachSystemTables.cpp index b770f784c01..e4e19ce2e06 100644 --- a/src/Storages/System/attachSystemTables.cpp +++ b/src/Storages/System/attachSystemTables.cpp @@ -83,6 +83,7 @@ #include #include #include +#include #if USE_RDKAFKA #include @@ -148,6 +149,7 @@ void attachSystemTablesLocal(ContextPtr context, IDatabase & system_database) attach(context, system_database, "backups"); attach(context, system_database, "schema_inference_cache"); attach(context, system_database, "dropped_tables"); + attach(context, system_database, "scheduler"); #if USE_RDKAFKA attach(context, system_database, "kafka_consumers"); #endif From 8fdd118e0c338f5a34d410cdcac874de7b3bc199 Mon Sep 17 00:00:00 2001 From: serxa Date: Sun, 3 Sep 2023 22:40:00 +0000 Subject: [PATCH 107/433] fix build --- src/IO/ISchedulerConstraint.h | 4 ++-- src/Storages/System/StorageSystemScheduler.cpp | 17 ++++++++++------- 2 files changed, 12 insertions(+), 9 deletions(-) diff --git a/src/IO/ISchedulerConstraint.h b/src/IO/ISchedulerConstraint.h index 16a421f1240..05bed7c3df8 100644 --- a/src/IO/ISchedulerConstraint.h +++ b/src/IO/ISchedulerConstraint.h @@ -21,7 +21,7 @@ namespace DB class ISchedulerConstraint : public ISchedulerNode { public: - ISchedulerConstraint(EventQueue * event_queue_, const Poco::Util::AbstractConfiguration & config = emptyConfig(), const String & config_prefix = {}) + explicit ISchedulerConstraint(EventQueue * event_queue_, const Poco::Util::AbstractConfiguration & config = emptyConfig(), const String & config_prefix = {}) : ISchedulerNode(event_queue_, config, config_prefix) {} @@ -45,7 +45,7 @@ public: } /// For introspection of current state (true = satisfied, false = violated) - virtual bool isSatisfied(); + virtual bool isSatisfied() = 0; protected: // Reference to nearest parent that is also derived from ISchedulerConstraint. diff --git a/src/Storages/System/StorageSystemScheduler.cpp b/src/Storages/System/StorageSystemScheduler.cpp index bb73619b972..2c3b726433e 100644 --- a/src/Storages/System/StorageSystemScheduler.cpp +++ b/src/Storages/System/StorageSystemScheduler.cpp @@ -71,17 +71,20 @@ void StorageSystemScheduler::fillData(MutableColumns & res_columns, ContextPtr c Field max_requests; Field max_cost; - if (auto parent = dynamic_cast(node->parent)) - vruntime = parent->getChildVRuntime(node.get()); - if (auto ptr = dynamic_cast(node.get())) + if (auto * parent = dynamic_cast(node->parent)) + { + if (auto value = parent->getChildVRuntime(node.get())) + vruntime = *value; + } + if (auto * ptr = dynamic_cast(node.get())) system_vruntime = ptr->getSystemVRuntime(); - if (auto ptr = dynamic_cast(node.get())) + if (auto * ptr = dynamic_cast(node.get())) std::tie(queue_length, queue_cost) = ptr->getQueueLengthAndCost(); - if (auto ptr = dynamic_cast(node.get())) + if (auto * ptr = dynamic_cast(node.get())) budget = ptr->getBudget(); - if (auto ptr = dynamic_cast(node.get())) + if (auto * ptr = dynamic_cast(node.get())) is_satisfied = ptr->isSatisfied(); - if (auto ptr = dynamic_cast(node.get())) + if (auto * ptr = dynamic_cast(node.get())) { std::tie(inflight_requests, inflight_cost) = ptr->getInflights(); std::tie(max_requests, max_cost) = ptr->getLimits(); From 964412ee64dcf7f20c4f674215ba88747e58e08d Mon Sep 17 00:00:00 2001 From: Duc Canh Le Date: Mon, 4 Sep 2023 02:47:26 +0000 Subject: [PATCH 108/433] no random settings in stateless test Signed-off-by: Duc Canh Le --- ...erge_across_partitions_final_with_lonely.reference | 3 --- ...68_no_merge_across_partitions_final_with_lonely.sh | 11 ++++------- 2 files changed, 4 insertions(+), 10 deletions(-) diff --git a/tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.reference b/tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.reference index 72837206fac..126459c2db0 100644 --- a/tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.reference +++ b/tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.reference @@ -1,6 +1,3 @@ 90 110 90 80 90 80 -90 110 -90 80 -90 80 diff --git a/tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.sh b/tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.sh index d01e4f94350..4bc29ce4233 100755 --- a/tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.sh +++ b/tests/queries/0_stateless/02868_no_merge_across_partitions_final_with_lonely.sh @@ -1,4 +1,5 @@ #!/usr/bin/env bash +# Tags: no-random-settings, no-random-merge-tree-settings set -e @@ -49,12 +50,8 @@ INSERT INTO with_lonely SELECT number+60000, '2022-11-01', number*10, 2 FROM num CLICKHOUSE_CLIENT="${CLICKHOUSE_CLIENT} --do_not_merge_across_partitions_select_final 1" # mix lonely parts and non-lonely parts -${CLICKHOUSE_CLIENT} --allow_experimental_analyzer 1 --max_bytes_to_read 2000 -q "SELECT max(val), count(*) FROM with_lonely FINAL;" +${CLICKHOUSE_CLIENT} --max_bytes_to_read 2000 -q "SELECT max(val), count(*) FROM with_lonely FINAL;" # only lonely parts -${CLICKHOUSE_CLIENT} --allow_experimental_analyzer 1 --max_bytes_to_read 1000 -q "SELECT max(val), count(*) FROM with_lonely FINAL WHERE dt < '2022-11-01';" +${CLICKHOUSE_CLIENT} --max_bytes_to_read 1000 -q "SELECT max(val), count(*) FROM with_lonely FINAL WHERE dt < '2022-11-01';" # only lonely parts but max_thread = 1, so reading lonely parts with in-order -${CLICKHOUSE_CLIENT} --allow_experimental_analyzer 1 --max_threads 1 --max_bytes_to_read 1000 -q "SELECT max(val), count(*) FROM with_lonely FINAL WHERE dt < '2022-11-01';" - -${CLICKHOUSE_CLIENT} --allow_experimental_analyzer 0 --max_bytes_to_read 2000 -q "SELECT max(val), count(*) FROM with_lonely FINAL;" -${CLICKHOUSE_CLIENT} --allow_experimental_analyzer 0 --max_bytes_to_read 1000 -q "SELECT max(val), count(*) FROM with_lonely FINAL WHERE dt < '2022-11-01';" -${CLICKHOUSE_CLIENT} --allow_experimental_analyzer 0 --max_threads 1 --max_bytes_to_read 1000 -q "SELECT max(val), count(*) FROM with_lonely FINAL WHERE dt < '2022-11-01';" +${CLICKHOUSE_CLIENT} --max_threads 1 --max_bytes_to_read 1000 -q "SELECT max(val), count(*) FROM with_lonely FINAL WHERE dt < '2022-11-01';" From 5db1961129bc60969b44728481049321c3358c68 Mon Sep 17 00:00:00 2001 From: Ziy1-Tan Date: Wed, 31 May 2023 00:17:03 +0800 Subject: [PATCH 109/433] Allow specifying `--query` multiple times in the command line --- docs/en/interfaces/cli.md | 4 ++-- docs/ru/interfaces/cli.md | 2 +- docs/zh/interfaces/cli.md | 2 +- src/Client/ClientBase.cpp | 20 ++++++++-------- src/Client/ClientBase.h | 1 + ...771_specify_query_multiple_times.reference | 13 ++++++++++ .../02771_specify_query_multiple_times.sh | 24 +++++++++++++++++++ 7 files changed, 52 insertions(+), 14 deletions(-) create mode 100644 tests/queries/0_stateless/02771_specify_query_multiple_times.reference create mode 100755 tests/queries/0_stateless/02771_specify_query_multiple_times.sh diff --git a/docs/en/interfaces/cli.md b/docs/en/interfaces/cli.md index 36afb94433a..175ff858d09 100644 --- a/docs/en/interfaces/cli.md +++ b/docs/en/interfaces/cli.md @@ -30,7 +30,7 @@ It may lack support for new features. ## Usage {#cli_usage} -The client can be used in interactive and non-interactive (batch) mode. +The client can be used in interactive and non-interactive (batch) mode. ### Gather your connection details @@ -177,7 +177,7 @@ You can pass parameters to `clickhouse-client` (all parameters have a default va - `--user, -u` – The username. Default value: default. - `--password` – The password. Default value: empty string. - `--ask-password` - Prompt the user to enter a password. -- `--query, -q` – The query to process when using non-interactive mode. Cannot be used simultaneously with `--queries-file`. +- `--query, -q` – The query to process when using non-interactive mode. Specifying `--query` multiple times are allowed(`--query "SELECT 1;" --query "SELECT 2;"...`). Cannot be used simultaneously with `--queries-file`. - `--queries-file` – file path with queries to execute. Cannot be used simultaneously with `--query`. - `--multiquery, -n` – If specified, multiple queries separated by semicolons can be listed after the `--query` option. For convenience, it is also possible to omit `--query` and pass the queries directly after `--multiquery`. - `--multiline, -m` – If specified, allow multiline queries (do not send the query on Enter). diff --git a/docs/ru/interfaces/cli.md b/docs/ru/interfaces/cli.md index aa6ae3629e8..47ab6474fc0 100644 --- a/docs/ru/interfaces/cli.md +++ b/docs/ru/interfaces/cli.md @@ -128,7 +128,7 @@ $ clickhouse-client --param_tbl="numbers" --param_db="system" --param_col="numbe - `--port` — порт для подключения, по умолчанию — 9000. Обратите внимание: для HTTP-интерфейса и нативного интерфейса используются разные порты. - `--user, -u` — имя пользователя, по умолчанию — ‘default’. - `--password` — пароль, по умолчанию — пустая строка. -- `--query, -q` — запрос для выполнения, при использовании в неинтерактивном режиме. +- `--query, -q` — запрос для выполнения, при использовании в неинтерактивном режиме. Допускается указание `--query` несколько раз (`--query "SELECT 1;" --query "SELECT 2;"...`). - `--queries-file` - путь к файлу с запросами для выполнения. Необходимо указать только одну из опций: `query` или `queries-file`. - `--database, -d` — выбрать текущую БД. Без указания значение берется из настроек сервера (по умолчанию — БД ‘default’). - `--multiline, -m` — если указано — разрешить многострочные запросы, не отправлять запрос по нажатию Enter. diff --git a/docs/zh/interfaces/cli.md b/docs/zh/interfaces/cli.md index 80d13154a76..a6b4d10dd2f 100644 --- a/docs/zh/interfaces/cli.md +++ b/docs/zh/interfaces/cli.md @@ -116,7 +116,7 @@ $ clickhouse-client --param_tuple_in_tuple="(10, ('dt', 10))" -q "SELECT * FROM - `--port` – 连接的端口，默认值：9000。注意HTTP接口以及TCP原生接口使用的是不同端口。 - `--user, -u` – 用户名。 默认值：`default`。 - `--password` – 密码。 默认值：空字符串。 -- `--query, -q` – 使用非交互模式查询。 +- `--query, -q` – 使用非交互模式查询。 允许多次指定 `--query`（`--query "SELECT 1;" --query "SELECT 2;"...`）。 - `--database, -d` – 默认当前操作的数据库. 默认值：服务端默认的配置（默认是`default`）。 - `--multiline, -m` – 如果指定，允许多行语句查询（Enter仅代表换行，不代表查询语句完结）。 - `--multiquery, -n` – 如果指定, 允许处理用`;`号分隔的多个查询，只在非交互模式下生效。 diff --git a/src/Client/ClientBase.cpp b/src/Client/ClientBase.cpp index 19244134617..46d6e882d2f 100644 --- a/src/Client/ClientBase.cpp +++ b/src/Client/ClientBase.cpp @@ -2484,23 +2484,23 @@ void ClientBase::runNonInteractive() return; } - String text; - if (config().has("query")) + if (!queries.empty()) { - text += config().getRawString("query"); /// Poco configuration should not process substitutions in form of ${...} inside query. + for (const auto & query : queries) + { + bool result = query_fuzzer_runs ? processWithFuzzing(query) : processQueryText(query); + if (!result) + return; + } } else { /// If 'query' parameter is not set, read a query from stdin. /// The query is read entirely into memory (streaming is disabled). ReadBufferFromFileDescriptor in(STDIN_FILENO); + String text; readStringUntilEOF(text, in); } - - if (query_fuzzer_runs) - processWithFuzzing(text); - else - processQueryText(text); } @@ -2700,7 +2700,7 @@ void ClientBase::init(int argc, char ** argv) ("config-file,C", po::value(), "config-file path") - ("query,q", po::value(), "query") + ("query,q", po::value>(), R"(query; specifying --query multiple times are allowed(--query "SELECT 1;" --query "SELECT 2;"...))") ("queries-file", po::value>()->multitoken(), "file path with queries to execute; multiple files can be specified (--queries-file file1 file2...)") ("multiquery,n", "If specified, multiple queries separated by semicolons can be listed after --query. For convenience, it is also possible to omit --query and pass the queries directly after --multiquery.") @@ -2789,7 +2789,7 @@ void ClientBase::init(int argc, char ** argv) if (options.count("time")) print_time_to_stderr = true; if (options.count("query")) - config().setString("query", options["query"].as()); + queries = options["query"].as>(); if (options.count("query_id")) config().setString("query_id", options["query_id"].as()); if (options.count("database")) diff --git a/src/Client/ClientBase.h b/src/Client/ClientBase.h index d877905302d..9a2a78a2b38 100644 --- a/src/Client/ClientBase.h +++ b/src/Client/ClientBase.h @@ -202,6 +202,7 @@ protected: std::optional suggest; bool load_suggestions = false; + std::vector queries; /// If not empty, queries will be read from these strings std::vector queries_files; /// If not empty, queries will be read from these files std::vector interleave_queries_files; /// If not empty, run queries from these files before processing every file from 'queries_files'. std::vector cmd_options; diff --git a/tests/queries/0_stateless/02771_specify_query_multiple_times.reference b/tests/queries/0_stateless/02771_specify_query_multiple_times.reference new file mode 100644 index 00000000000..18a0b6bde4d --- /dev/null +++ b/tests/queries/0_stateless/02771_specify_query_multiple_times.reference @@ -0,0 +1,13 @@ +101 +202 +303 +404 +404 +404 +505 +Syntax error +Syntax error +Empty query +Empty query +BAD_ARGUMENTS +Bad arguments diff --git a/tests/queries/0_stateless/02771_specify_query_multiple_times.sh b/tests/queries/0_stateless/02771_specify_query_multiple_times.sh new file mode 100755 index 00000000000..e477b0deef9 --- /dev/null +++ b/tests/queries/0_stateless/02771_specify_query_multiple_times.sh @@ -0,0 +1,24 @@ +#!/usr/bin/env bash + +CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) +# shellcheck source=../shell_config.sh +. "$CURDIR"/../shell_config.sh + +$CLICKHOUSE_LOCAL --query "SELECT 101" +$CLICKHOUSE_LOCAL --query "SELECT 202" --query "SELECT 303" +$CLICKHOUSE_LOCAL --query "SELECT 404" --query "SELECT 404" --query "SELECT 404" +$CLICKHOUSE_LOCAL --query --query 2>&1 +$CLICKHOUSE_LOCAL --query "" --query "" 2>&1 +$CLICKHOUSE_LOCAL --query "SELECT 505;"--query + +# Abort if any invalid SQL +$CLICKHOUSE_LOCAL --query "SELECT 606; S" --query "SELECT 606" 2>&1 | grep -o 'Syntax error' +$CLICKHOUSE_LOCAL --query "SELECT 707" --query "SELECT 808; S" 2>&1 | grep -o '201' +$CLICKHOUSE_LOCAL --query "SELECT 909" --query "SELECT 909; S" 2>&1 | grep -o 'Syntax error' +$CLICKHOUSE_LOCAL --query "; SELECT 111;" --query "SELECT 111;" 2>&1 | grep -o 'Empty query' +$CLICKHOUSE_LOCAL --query "SELECT 222;" --query "; SELECT 222;" 2>&1 | grep -o '201' +$CLICKHOUSE_LOCAL --query "SELECT 333;" --query "; SELECT 333;" 2>&1 | grep -o 'Empty query' + + +$CLICKHOUSE_LOCAL --query --query "SELECT 444;" 2>&1 | grep -o 'BAD_ARGUMENTS' +$CLICKHOUSE_LOCAL --query "SELECT 555;" --query 2>&1 | grep -o 'Bad arguments' From fe7d636b935da44b706e3a7fb2fbcf524481f337 Mon Sep 17 00:00:00 2001 From: Ziy1-Tan Date: Wed, 31 May 2023 09:40:31 +0800 Subject: [PATCH 110/433] Improve docs Signed-off-by: Ziy1-Tan --- docs/en/interfaces/cli.md | 2 +- src/Client/ClientBase.cpp | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/docs/en/interfaces/cli.md b/docs/en/interfaces/cli.md index 175ff858d09..ab04e3352bd 100644 --- a/docs/en/interfaces/cli.md +++ b/docs/en/interfaces/cli.md @@ -177,7 +177,7 @@ You can pass parameters to `clickhouse-client` (all parameters have a default va - `--user, -u` – The username. Default value: default. - `--password` – The password. Default value: empty string. - `--ask-password` - Prompt the user to enter a password. -- `--query, -q` – The query to process when using non-interactive mode. Specifying `--query` multiple times are allowed(`--query "SELECT 1;" --query "SELECT 2;"...`). Cannot be used simultaneously with `--queries-file`. +- `--query, -q` – The query to process when using non-interactive mode. Specifying `--query` multiple times is allowed (`--query "SELECT 1;" --query "SELECT 2;"...`). Cannot be used simultaneously with `--queries-file`. - `--queries-file` – file path with queries to execute. Cannot be used simultaneously with `--query`. - `--multiquery, -n` – If specified, multiple queries separated by semicolons can be listed after the `--query` option. For convenience, it is also possible to omit `--query` and pass the queries directly after `--multiquery`. - `--multiline, -m` – If specified, allow multiline queries (do not send the query on Enter). diff --git a/src/Client/ClientBase.cpp b/src/Client/ClientBase.cpp index 46d6e882d2f..934394601f2 100644 --- a/src/Client/ClientBase.cpp +++ b/src/Client/ClientBase.cpp @@ -2700,7 +2700,7 @@ void ClientBase::init(int argc, char ** argv) ("config-file,C", po::value(), "config-file path") - ("query,q", po::value>(), R"(query; specifying --query multiple times are allowed(--query "SELECT 1;" --query "SELECT 2;"...))") + ("query,q", po::value>(), R"(query; can be specified multiple times (--query "SELECT 1;" --query "SELECT 2;"...))") ("queries-file", po::value>()->multitoken(), "file path with queries to execute; multiple files can be specified (--queries-file file1 file2...)") ("multiquery,n", "If specified, multiple queries separated by semicolons can be listed after --query. For convenience, it is also possible to omit --query and pass the queries directly after --multiquery.") From 44994fe51be89f4f2d831ca887581d30a4996382 Mon Sep 17 00:00:00 2001 From: Robert Schulze Date: Mon, 4 Sep 2023 11:40:56 +0000 Subject: [PATCH 111/433] Incorporate review feedback --- docs/en/interfaces/cli.md | 4 +-- .../operations/utilities/clickhouse-local.md | 4 +-- src/Client/ClientBase.cpp | 25 +++++++++++++------ src/Client/ClientBase.h | 2 +- .../02771_multiple_query_arguments.reference | 17 +++++++++++++ .../02771_multiple_query_arguments.sh | 21 ++++++++++++++++ ...771_specify_query_multiple_times.reference | 13 ---------- .../02771_specify_query_multiple_times.sh | 24 ------------------ 8 files changed, 60 insertions(+), 50 deletions(-) create mode 100644 tests/queries/0_stateless/02771_multiple_query_arguments.reference create mode 100755 tests/queries/0_stateless/02771_multiple_query_arguments.sh delete mode 100644 tests/queries/0_stateless/02771_specify_query_multiple_times.reference delete mode 100755 tests/queries/0_stateless/02771_specify_query_multiple_times.sh diff --git a/docs/en/interfaces/cli.md b/docs/en/interfaces/cli.md index ab04e3352bd..f018f3a248e 100644 --- a/docs/en/interfaces/cli.md +++ b/docs/en/interfaces/cli.md @@ -177,8 +177,8 @@ You can pass parameters to `clickhouse-client` (all parameters have a default va - `--user, -u` – The username. Default value: default. - `--password` – The password. Default value: empty string. - `--ask-password` - Prompt the user to enter a password. -- `--query, -q` – The query to process when using non-interactive mode. Specifying `--query` multiple times is allowed (`--query "SELECT 1;" --query "SELECT 2;"...`). Cannot be used simultaneously with `--queries-file`. -- `--queries-file` – file path with queries to execute. Cannot be used simultaneously with `--query`. +- `--query, -q` – The query to process when using non-interactive mode. `--query` can be specified multiple times, e.g. `--query "SELECT 1" --query "SELECT 2"`. Cannot be used simultaneously with `--queries-file`. +- `--queries-file` – file path with queries to execute. `--queries-file` can be specified multiple times, e.g. `--query queries1.sql --query queries2.sql`. Cannot be used simultaneously with `--query`. - `--multiquery, -n` – If specified, multiple queries separated by semicolons can be listed after the `--query` option. For convenience, it is also possible to omit `--query` and pass the queries directly after `--multiquery`. - `--multiline, -m` – If specified, allow multiline queries (do not send the query on Enter). - `--database, -d` – Select the current default database. Default value: the current database from the server settings (‘default’ by default). diff --git a/docs/en/operations/utilities/clickhouse-local.md b/docs/en/operations/utilities/clickhouse-local.md index 737c2b81dee..c863282efc1 100644 --- a/docs/en/operations/utilities/clickhouse-local.md +++ b/docs/en/operations/utilities/clickhouse-local.md @@ -202,8 +202,8 @@ Arguments: - `-S`, `--structure` — table structure for input data. - `--input-format` — input format, `TSV` by default. - `-f`, `--file` — path to data, `stdin` by default. -- `-q`, `--query` — queries to execute with `;` as delimiter. Cannot be used simultaneously with `--queries-file`. -- `--queries-file` - file path with queries to execute. Cannot be used simultaneously with `--query`. +- `-q`, `--query` — queries to execute with `;` as delimiter. `--query` can be specified multiple times, e.g. `--query "SELECT 1" --query "SELECT 2"`. Cannot be used simultaneously with `--queries-file`. +- `--queries-file` - file path with queries to execute. `--queries-file` can be specified multiple times, e.g. `--query queries1.sql --query queries2.sql`. Cannot be used simultaneously with `--query`. - `--multiquery, -n` – If specified, multiple queries separated by semicolons can be listed after the `--query` option. For convenience, it is also possible to omit `--query` and pass the queries directly after `--multiquery`. - `-N`, `--table` — table name where to put output data, `table` by default. - `--format`, `--output-format` — output format, `TSV` by default. diff --git a/src/Client/ClientBase.cpp b/src/Client/ClientBase.cpp index 934394601f2..64fd37cc90d 100644 --- a/src/Client/ClientBase.cpp +++ b/src/Client/ClientBase.cpp @@ -2488,9 +2488,16 @@ void ClientBase::runNonInteractive() { for (const auto & query : queries) { - bool result = query_fuzzer_runs ? processWithFuzzing(query) : processQueryText(query); - if (!result) - return; + if (query_fuzzer_runs) + { + if (!processWithFuzzing(query)) + return; + } + else + { + if (!processQueryText(query)) + return; + } } } else @@ -2500,6 +2507,10 @@ void ClientBase::runNonInteractive() ReadBufferFromFileDescriptor in(STDIN_FILENO); String text; readStringUntilEOF(text, in); + if (query_fuzzer_runs) + processWithFuzzing(text); + else + processQueryText(text); } } @@ -2700,9 +2711,8 @@ void ClientBase::init(int argc, char ** argv) ("config-file,C", po::value(), "config-file path") - ("query,q", po::value>(), R"(query; can be specified multiple times (--query "SELECT 1;" --query "SELECT 2;"...))") - ("queries-file", po::value>()->multitoken(), - "file path with queries to execute; multiple files can be specified (--queries-file file1 file2...)") + ("query,q", po::value>(), R"(query; can be specified multiple times (--query "SELECT 1" --query "SELECT 2"...))") + ("queries-file", po::value>()->multitoken(), "file path with queries to execute; multiple files can be specified (--queries-file file1 file2...)") ("multiquery,n", "If specified, multiple queries separated by semicolons can be listed after --query. For convenience, it is also possible to omit --query and pass the queries directly after --multiquery.") ("multiline,m", "If specified, allow multiline queries (do not send the query on Enter)") ("database,d", po::value(), "database") @@ -2723,8 +2733,7 @@ void ClientBase::init(int argc, char ** argv) ("log-level", po::value(), "log level") ("server_logs_file", po::value(), "put server logs into specified file") - ("suggestion_limit", po::value()->default_value(10000), - "Suggestion limit for how many databases, tables and columns to fetch.") + ("suggestion_limit", po::value()->default_value(10000), "Suggestion limit for how many databases, tables and columns to fetch.") ("format,f", po::value(), "default output format") ("vertical,E", "vertical output format, same as --format=Vertical or FORMAT Vertical or \\G at end of command") diff --git a/src/Client/ClientBase.h b/src/Client/ClientBase.h index 9a2a78a2b38..eabc79b7432 100644 --- a/src/Client/ClientBase.h +++ b/src/Client/ClientBase.h @@ -202,7 +202,7 @@ protected: std::optional suggest; bool load_suggestions = false; - std::vector queries; /// If not empty, queries will be read from these strings + std::vector queries; /// Queries passed via '--query' std::vector queries_files; /// If not empty, queries will be read from these files std::vector interleave_queries_files; /// If not empty, run queries from these files before processing every file from 'queries_files'. std::vector cmd_options; diff --git a/tests/queries/0_stateless/02771_multiple_query_arguments.reference b/tests/queries/0_stateless/02771_multiple_query_arguments.reference new file mode 100644 index 00000000000..5cad23947c8 --- /dev/null +++ b/tests/queries/0_stateless/02771_multiple_query_arguments.reference @@ -0,0 +1,17 @@ +101 +101 +202 +202 +Multi-statements are not allowed +Empty query +Bad arguments +Syntax error +101 +101 +202 +202 +303 +303 +303 +Bad arguments +Syntax error diff --git a/tests/queries/0_stateless/02771_multiple_query_arguments.sh b/tests/queries/0_stateless/02771_multiple_query_arguments.sh new file mode 100755 index 00000000000..435e0a33315 --- /dev/null +++ b/tests/queries/0_stateless/02771_multiple_query_arguments.sh @@ -0,0 +1,21 @@ +#!/usr/bin/env bash + +CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) +# shellcheck source=../shell_config.sh +. "$CURDIR"/../shell_config.sh + +# clickhouse-client +$CLICKHOUSE_CLIENT --query "SELECT 101" --query "SELECT 101" +$CLICKHOUSE_CLIENT --query "SELECT 202;" --query "SELECT 202;" +$CLICKHOUSE_CLIENT --query "SELECT 303" --query "SELECT 303; SELECT 303" 2>&1 | grep -o 'Multi-statements are not allowed' +$CLICKHOUSE_CLIENT --query "" --query "" 2>&1 | grep -o 'Empty query' +$CLICKHOUSE_CLIENT --query "SELECT 303" --query 2>&1 | grep -o 'Bad arguments' +$CLICKHOUSE_CLIENT --query "SELECT 303" --query "SELE" 2>&1 | grep -o 'Syntax error' + +# clickhouse-local +$CLICKHOUSE_LOCAL --query "SELECT 101" --query "SELECT 101" +$CLICKHOUSE_LOCAL --query "SELECT 202;" --query "SELECT 202;" +$CLICKHOUSE_LOCAL --query "SELECT 303" --query "SELECT 303; SELECT 303" 2>&1 # behaves differently than clickhouse-client, TODO make it consistent +$CLICKHOUSE_LOCAL --query "" --query "" 2>&1 # behaves equally different than clickhouse-client TODO +$CLICKHOUSE_LOCAL --query "SELECT 303" --query 2>&1 | grep -o 'Bad arguments' +$CLICKHOUSE_LOCAL --query "SELECT 303" --query "SELE" 2>&1 | grep -o 'Syntax error' diff --git a/tests/queries/0_stateless/02771_specify_query_multiple_times.reference b/tests/queries/0_stateless/02771_specify_query_multiple_times.reference deleted file mode 100644 index 18a0b6bde4d..00000000000 --- a/tests/queries/0_stateless/02771_specify_query_multiple_times.reference +++ /dev/null @@ -1,13 +0,0 @@ -101 -202 -303 -404 -404 -404 -505 -Syntax error -Syntax error -Empty query -Empty query -BAD_ARGUMENTS -Bad arguments diff --git a/tests/queries/0_stateless/02771_specify_query_multiple_times.sh b/tests/queries/0_stateless/02771_specify_query_multiple_times.sh deleted file mode 100755 index e477b0deef9..00000000000 --- a/tests/queries/0_stateless/02771_specify_query_multiple_times.sh +++ /dev/null @@ -1,24 +0,0 @@ -#!/usr/bin/env bash - -CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) -# shellcheck source=../shell_config.sh -. "$CURDIR"/../shell_config.sh - -$CLICKHOUSE_LOCAL --query "SELECT 101" -$CLICKHOUSE_LOCAL --query "SELECT 202" --query "SELECT 303" -$CLICKHOUSE_LOCAL --query "SELECT 404" --query "SELECT 404" --query "SELECT 404" -$CLICKHOUSE_LOCAL --query --query 2>&1 -$CLICKHOUSE_LOCAL --query "" --query "" 2>&1 -$CLICKHOUSE_LOCAL --query "SELECT 505;"--query - -# Abort if any invalid SQL -$CLICKHOUSE_LOCAL --query "SELECT 606; S" --query "SELECT 606" 2>&1 | grep -o 'Syntax error' -$CLICKHOUSE_LOCAL --query "SELECT 707" --query "SELECT 808; S" 2>&1 | grep -o '201' -$CLICKHOUSE_LOCAL --query "SELECT 909" --query "SELECT 909; S" 2>&1 | grep -o 'Syntax error' -$CLICKHOUSE_LOCAL --query "; SELECT 111;" --query "SELECT 111;" 2>&1 | grep -o 'Empty query' -$CLICKHOUSE_LOCAL --query "SELECT 222;" --query "; SELECT 222;" 2>&1 | grep -o '201' -$CLICKHOUSE_LOCAL --query "SELECT 333;" --query "; SELECT 333;" 2>&1 | grep -o 'Empty query' - - -$CLICKHOUSE_LOCAL --query --query "SELECT 444;" 2>&1 | grep -o 'BAD_ARGUMENTS' -$CLICKHOUSE_LOCAL --query "SELECT 555;" --query 2>&1 | grep -o 'Bad arguments' From b5233efc20a1ad18ae347cf90b320f0b9cf8a0d1 Mon Sep 17 00:00:00 2001 From: serxa Date: Mon, 4 Sep 2023 12:35:17 +0000 Subject: [PATCH 112/433] fix system table and scheduler config reloading --- src/IO/ISchedulerNode.h | 14 ++++++++++++-- src/IO/Resource/FairPolicy.h | 2 ++ src/IO/Resource/FifoQueue.h | 2 ++ src/IO/Resource/PriorityPolicy.h | 2 ++ src/IO/Resource/SemaphoreConstraint.h | 2 ++ src/IO/SchedulerRoot.h | 2 ++ src/Storages/System/StorageSystemScheduler.cpp | 2 +- 7 files changed, 23 insertions(+), 3 deletions(-) diff --git a/src/IO/ISchedulerNode.h b/src/IO/ISchedulerNode.h index 943d41d95d2..f7fc9257128 100644 --- a/src/IO/ISchedulerNode.h +++ b/src/IO/ISchedulerNode.h @@ -69,6 +69,13 @@ struct SchedulerNodeInfo { priority.value = value; } + + // To check if configuration update required + bool equals(const SchedulerNodeInfo & o) const + { + // `parent` data is not compared intentionally (it is not part of configuration settings) + return weight == o.weight && priority == o.priority; + } }; /* @@ -158,8 +165,11 @@ public: virtual ~ISchedulerNode() {} - // Checks if two nodes configuration is equal - virtual bool equals(ISchedulerNode * other) = 0; + /// Checks if two nodes configuration is equal + virtual bool equals(ISchedulerNode * other) + { + return info.equals(other->info); + } /// Attach new child virtual void attachChild(const std::shared_ptr & child) = 0; diff --git a/src/IO/Resource/FairPolicy.h b/src/IO/Resource/FairPolicy.h index 7e632efc8fd..3c7ccfcff89 100644 --- a/src/IO/Resource/FairPolicy.h +++ b/src/IO/Resource/FairPolicy.h @@ -51,6 +51,8 @@ public: bool equals(ISchedulerNode * other) override { + if (!ISchedulerNode::equals(other)) + return false; if (auto * o = dynamic_cast(other)) return true; return false; diff --git a/src/IO/Resource/FifoQueue.h b/src/IO/Resource/FifoQueue.h index e0689e3c8da..e95e817719a 100644 --- a/src/IO/Resource/FifoQueue.h +++ b/src/IO/Resource/FifoQueue.h @@ -30,6 +30,8 @@ public: bool equals(ISchedulerNode * other) override { + if (!ISchedulerNode::equals(other)) + return false; if (auto * o = dynamic_cast(other)) return true; return false; diff --git a/src/IO/Resource/PriorityPolicy.h b/src/IO/Resource/PriorityPolicy.h index 0a06a4fdc09..86d4fadb9dc 100644 --- a/src/IO/Resource/PriorityPolicy.h +++ b/src/IO/Resource/PriorityPolicy.h @@ -42,6 +42,8 @@ public: bool equals(ISchedulerNode * other) override { + if (!ISchedulerNode::equals(other)) + return false; if (auto * o = dynamic_cast(other)) return true; return false; diff --git a/src/IO/Resource/SemaphoreConstraint.h b/src/IO/Resource/SemaphoreConstraint.h index 26437d3da87..9160d0431de 100644 --- a/src/IO/Resource/SemaphoreConstraint.h +++ b/src/IO/Resource/SemaphoreConstraint.h @@ -27,6 +27,8 @@ public: bool equals(ISchedulerNode * other) override { + if (!ISchedulerNode::equals(other)) + return false; if (auto * o = dynamic_cast(other)) return max_requests == o->max_requests && max_cost == o->max_cost; return false; diff --git a/src/IO/SchedulerRoot.h b/src/IO/SchedulerRoot.h index 75a1b3af72a..748632615bc 100644 --- a/src/IO/SchedulerRoot.h +++ b/src/IO/SchedulerRoot.h @@ -97,6 +97,8 @@ public: bool equals(ISchedulerNode * other) override { + if (!ISchedulerNode::equals(other)) + return false; if (auto * o = dynamic_cast(other)) return true; return false; diff --git a/src/Storages/System/StorageSystemScheduler.cpp b/src/Storages/System/StorageSystemScheduler.cpp index 2c3b726433e..101914a34ae 100644 --- a/src/Storages/System/StorageSystemScheduler.cpp +++ b/src/Storages/System/StorageSystemScheduler.cpp @@ -26,7 +26,7 @@ NamesAndTypesList StorageSystemScheduler::getNamesAndTypes() {"weight", std::make_shared()}, {"priority", std::make_shared()}, {"is_active", std::make_shared()}, - {"active_children", std::make_shared()}, + {"active_children", std::make_shared()}, {"dequeued_requests", std::make_shared()}, {"dequeued_cost", std::make_shared()}, {"busy_periods", std::make_shared()}, From 033970716e405331d04a30cf6ef1f3cad1a28274 Mon Sep 17 00:00:00 2001 From: yariks5s Date: Mon, 4 Sep 2023 13:30:24 +0000 Subject: [PATCH 113/433] fixed tests --- src/Functions/FunctionsConversion.h | 5 ++++- src/Functions/formatDateTime.cpp | 4 ++-- 2 files changed, 6 insertions(+), 3 deletions(-) diff --git a/src/Functions/FunctionsConversion.h b/src/Functions/FunctionsConversion.h index 34430130fba..f02f92efef9 100644 --- a/src/Functions/FunctionsConversion.h +++ b/src/Functions/FunctionsConversion.h @@ -900,7 +900,10 @@ struct ConvertImpl 1) time_zone_column = checkAndGetColumnConst(arguments[1].column.get()); else if (arguments.size() == 1) - time_zone = &DateLUT::instance(); + { + auto non_null_args = createBlockWithNestedColumns(arguments); + time_zone = &extractTimeZoneFromFunctionArguments(non_null_args, 1, 0); + } if constexpr (std::is_same_v || std::is_same_v) time_zone = &DateLUT::instance(); diff --git a/src/Functions/formatDateTime.cpp b/src/Functions/formatDateTime.cpp index 7c4b9ea3d36..0514cc27851 100644 --- a/src/Functions/formatDateTime.cpp +++ b/src/Functions/formatDateTime.cpp @@ -864,7 +864,7 @@ public: const ColumnConst * time_zone_column = nullptr; const DateLUTImpl * time_zone = nullptr; if (arguments.size() == 2) - time_zone = &DateLUT::instance(); + time_zone = &extractTimeZoneFromFunctionArguments(arguments, 2, 0); else if (arguments.size() > 2) time_zone_column = checkAndGetColumnConst(arguments[2].column.get()); if (!format_column) @@ -953,7 +953,7 @@ public: for (size_t i = 0; i < vec.size(); ++i) { if (!time_zone_column && arguments.size() > 2) - { + { if (!arguments[2].column.get()->getDataAt(i).toString().empty()) time_zone = &DateLUT::instance(arguments[2].column.get()->getDataAt(i).toString()); else From 3d597a7d8edf902e30a924e91218e5f40365d7e5 Mon Sep 17 00:00:00 2001 From: Christian Clauss Date: Mon, 4 Sep 2023 15:43:45 +0200 Subject: [PATCH 114/433] Fix typo --- tests/fuzz/README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tests/fuzz/README.md b/tests/fuzz/README.md index 440ae1aac1f..d9d89fbfe96 100644 --- a/tests/fuzz/README.md +++ b/tests/fuzz/README.md @@ -1,4 +1,4 @@ -The list of funtions generated via following query +The list of functions generated via following query ``` clickhouse-client -q "select concat('\"', name, '\"') from system.functions union all select concat('\"', alias_to, '\"') from system.functions where alias_to != '' " > functions.dict From aa8e9c3cbc5822983fba3a54737be5ff8a87a601 Mon Sep 17 00:00:00 2001 From: serxa Date: Mon, 4 Sep 2023 14:18:54 +0000 Subject: [PATCH 115/433] fix style --- src/IO/Resource/FairPolicy.h | 3 ++- src/Storages/System/StorageSystemScheduler.cpp | 3 ++- 2 files changed, 4 insertions(+), 2 deletions(-) diff --git a/src/IO/Resource/FairPolicy.h b/src/IO/Resource/FairPolicy.h index 3c7ccfcff89..57b26344658 100644 --- a/src/IO/Resource/FairPolicy.h +++ b/src/IO/Resource/FairPolicy.h @@ -211,7 +211,8 @@ public: std::optional getChildVRuntime(ISchedulerNode * child) const { - for (const auto & item : items) { + for (const auto & item : items) + { if (child == item.child) return item.vruntime; } diff --git a/src/Storages/System/StorageSystemScheduler.cpp b/src/Storages/System/StorageSystemScheduler.cpp index 101914a34ae..ca0a138559e 100644 --- a/src/Storages/System/StorageSystemScheduler.cpp +++ b/src/Storages/System/StorageSystemScheduler.cpp @@ -47,7 +47,8 @@ NamesAndTypesList StorageSystemScheduler::getNamesAndTypes() void StorageSystemScheduler::fillData(MutableColumns & res_columns, ContextPtr context, const SelectQueryInfo &) const { - context->getResourceManager()->forEachNode([&] (const String & resource, const String & path, const String & type, const SchedulerNodePtr & node) { + context->getResourceManager()->forEachNode([&] (const String & resource, const String & path, const String & type, const SchedulerNodePtr & node) + { size_t i = 0; res_columns[i++]->insert(resource); res_columns[i++]->insert(path); From 4af150db5f717a37dcd6113665a33a5513e1a42c Mon Sep 17 00:00:00 2001 From: yariks5s Date: Mon, 4 Sep 2023 14:22:03 +0000 Subject: [PATCH 116/433] fixed suggestions --- src/Functions/formatDateTime.cpp | 25 +++++++++++++------------ 1 file changed, 13 insertions(+), 12 deletions(-) diff --git a/src/Functions/formatDateTime.cpp b/src/Functions/formatDateTime.cpp index 0514cc27851..39ae7fe735c 100644 --- a/src/Functions/formatDateTime.cpp +++ b/src/Functions/formatDateTime.cpp @@ -860,19 +860,20 @@ public: if (!times) return nullptr; - const ColumnConst * format_column = checkAndGetColumnConst(arguments[1].column.get()); - const ColumnConst * time_zone_column = nullptr; - const DateLUTImpl * time_zone = nullptr; - if (arguments.size() == 2) - time_zone = &extractTimeZoneFromFunctionArguments(arguments, 2, 0); - else if (arguments.size() > 2) - time_zone_column = checkAndGetColumnConst(arguments[2].column.get()); - if (!format_column) + String format; + if (const auto * format_column = checkAndGetColumnConst(arguments[1].column.get())) + format = format_column->getValue(); + else throw Exception(ErrorCodes::ILLEGAL_COLUMN, "Illegal column {} of second ('format') argument of function {}. Must be constant string.", arguments[1].column->getName(), getName()); - String format = format_column->getValue(); + const ColumnConst * const_time_zone_column = nullptr; + const DateLUTImpl * time_zone = nullptr; + if (arguments.size() == 2) + time_zone = &extractTimeZoneFromFunctionArguments(arguments, 2, 0); + else if (arguments.size() > 2) + const_time_zone_column = checkAndGetColumnConst(arguments[2].column.get()); UInt32 scale [[maybe_unused]] = 0; if constexpr (std::is_same_v) @@ -902,12 +903,12 @@ public: if (castType(arguments[0].type.get(), [&]([[maybe_unused]] const auto & type) { return true; })) { - if (time_zone_column) + if (const_time_zone_column) time_zone = &extractTimeZoneFromFunctionArguments(arguments, 2, 0); } else if (std::is_same_v || std::is_same_v) { - if (time_zone_column) + if (const_time_zone_column) time_zone = &extractTimeZoneFromFunctionArguments(arguments, 2, 0); } else @@ -952,7 +953,7 @@ public: auto * pos = begin; for (size_t i = 0; i < vec.size(); ++i) { - if (!time_zone_column && arguments.size() > 2) + if (!const_time_zone_column && arguments.size() > 2) { if (!arguments[2].column.get()->getDataAt(i).toString().empty()) time_zone = &DateLUT::instance(arguments[2].column.get()->getDataAt(i).toString()); From 11fea581bcf1c2816983a76c362f51b1fbd37f2c Mon Sep 17 00:00:00 2001 From: Robert Schulze Date: Mon, 4 Sep 2023 14:48:38 +0000 Subject: [PATCH 117/433] Fix tests --- programs/client/Client.cpp | 6 +++--- programs/local/LocalServer.cpp | 10 +++++----- src/Client/ClientBase.cpp | 6 +++--- .../02751_multiquery_with_argument.reference | 3 --- .../0_stateless/02751_multiquery_with_argument.sh | 5 +---- 5 files changed, 12 insertions(+), 18 deletions(-) diff --git a/programs/client/Client.cpp b/programs/client/Client.cpp index 929e59ed852..64823f9ec7f 100644 --- a/programs/client/Client.cpp +++ b/programs/client/Client.cpp @@ -1189,7 +1189,7 @@ void Client::processOptions(const OptionsDescription & options_description, void Client::processConfig() { - if (config().has("query") && config().has("queries-file")) + if (!queries.empty() && config().has("queries-file")) throw Exception(ErrorCodes::BAD_ARGUMENTS, "Options '--query' and '--queries-file' cannot be specified at the same time"); /// Batch mode is enabled if one of the following is true: @@ -1200,9 +1200,9 @@ void Client::processConfig() /// - --queries-file command line option is present. /// The value of the option is used as file with query (or of multiple queries) to execute. - delayed_interactive = config().has("interactive") && (config().has("query") || config().has("queries-file")); + delayed_interactive = config().has("interactive") && (!queries.empty() || config().has("queries-file")); if (stdin_is_a_tty - && (delayed_interactive || (!config().has("query") && queries_files.empty()))) + && (delayed_interactive || (queries.empty() && queries_files.empty()))) { is_interactive = true; } diff --git a/programs/local/LocalServer.cpp b/programs/local/LocalServer.cpp index 5c522b678ef..cacfe4dc162 100644 --- a/programs/local/LocalServer.cpp +++ b/programs/local/LocalServer.cpp @@ -319,7 +319,7 @@ static bool checkIfStdinIsRegularFile() std::string LocalServer::getInitialCreateTableQuery() { - if (!config().has("table-structure") && !config().has("table-file") && !config().has("table-data-format") && (!checkIfStdinIsRegularFile() || !config().has("query"))) + if (!config().has("table-structure") && !config().has("table-file") && !config().has("table-data-format") && (!checkIfStdinIsRegularFile() || queries.empty())) return {}; auto table_name = backQuoteIfNeed(config().getString("table-name", "table")); @@ -461,7 +461,7 @@ try if (first_time) { - if (queries_files.empty() && !config().has("query")) + if (queries_files.empty() && queries.empty()) { std::cerr << "\033[31m" << "ClickHouse compiled in fuzzing mode." << "\033[0m" << std::endl; std::cerr << "\033[31m" << "You have to provide a query with --query or --queries-file option." << "\033[0m" << std::endl; @@ -473,7 +473,7 @@ try #else is_interactive = stdin_is_a_tty && (config().hasOption("interactive") - || (!config().has("query") && !config().has("table-structure") && queries_files.empty() && !config().has("table-file"))); + || (queries.empty() && !config().has("table-structure") && queries_files.empty() && !config().has("table-file"))); #endif if (!is_interactive) { @@ -569,10 +569,10 @@ void LocalServer::updateLoggerLevel(const String & logs_level) void LocalServer::processConfig() { - if (config().has("query") && config().has("queries-file")) + if (!queries.empty() && config().has("queries-file")) throw Exception(ErrorCodes::BAD_ARGUMENTS, "Options '--query' and '--queries-file' cannot be specified at the same time"); - delayed_interactive = config().has("interactive") && (config().has("query") || config().has("queries-file")); + delayed_interactive = config().has("interactive") && (!queries.empty() || config().has("queries-file")); if (is_interactive && !delayed_interactive) { if (config().has("multiquery")) diff --git a/src/Client/ClientBase.cpp b/src/Client/ClientBase.cpp index 64fd37cc90d..7d34e65356e 100644 --- a/src/Client/ClientBase.cpp +++ b/src/Client/ClientBase.cpp @@ -2680,8 +2680,8 @@ void ClientBase::init(int argc, char ** argv) stderr_is_a_tty = isatty(STDERR_FILENO); terminal_width = getTerminalWidth(); - Arguments common_arguments{""}; /// 0th argument is ignored. std::vector external_tables_arguments; + Arguments common_arguments = {""}; /// 0th argument is ignored. std::vector hosts_and_ports_arguments; readArguments(argc, argv, common_arguments, external_tables_arguments, hosts_and_ports_arguments); @@ -2699,7 +2699,6 @@ void ClientBase::init(int argc, char ** argv) } - po::variables_map options; OptionsDescription options_description; options_description.main_description.emplace(createOptionsDescription("Main options", terminal_width)); @@ -2711,7 +2710,7 @@ void ClientBase::init(int argc, char ** argv) ("config-file,C", po::value(), "config-file path") - ("query,q", po::value>(), R"(query; can be specified multiple times (--query "SELECT 1" --query "SELECT 2"...))") + ("query,q", po::value>()->multitoken(), R"(query; can be specified multiple times (--query "SELECT 1" --query "SELECT 2"...))") ("queries-file", po::value>()->multitoken(), "file path with queries to execute; multiple files can be specified (--queries-file file1 file2...)") ("multiquery,n", "If specified, multiple queries separated by semicolons can be listed after --query. For convenience, it is also possible to omit --query and pass the queries directly after --multiquery.") ("multiline,m", "If specified, allow multiline queries (do not send the query on Enter)") @@ -2771,6 +2770,7 @@ void ClientBase::init(int argc, char ** argv) std::transform(external_options.begin(), external_options.end(), std::back_inserter(cmd_options), getter); } + po::variables_map options; parseAndCheckOptions(options_description, options, common_arguments); po::notify(options); diff --git a/tests/queries/0_stateless/02751_multiquery_with_argument.reference b/tests/queries/0_stateless/02751_multiquery_with_argument.reference index df9771b6bd3..843fffb476c 100644 --- a/tests/queries/0_stateless/02751_multiquery_with_argument.reference +++ b/tests/queries/0_stateless/02751_multiquery_with_argument.reference @@ -18,9 +18,6 @@ Bad arguments Bad arguments Bad arguments BAD_ARGUMENTS -Bad arguments BAD_ARGUMENTS Bad arguments Bad arguments -Bad arguments -Bad arguments diff --git a/tests/queries/0_stateless/02751_multiquery_with_argument.sh b/tests/queries/0_stateless/02751_multiquery_with_argument.sh index ce53ede3331..d742cc0ad90 100755 --- a/tests/queries/0_stateless/02751_multiquery_with_argument.sh +++ b/tests/queries/0_stateless/02751_multiquery_with_argument.sh @@ -30,9 +30,6 @@ $CLICKHOUSE_LOCAL -n --multiquery "SELECT 307; SELECT 308;" 2>&1 | grep -o 'Bad $CLICKHOUSE_LOCAL --multiquery "SELECT 309; SELECT 310;" --multiquery 2>&1 | grep -o 'Bad arguments' $CLICKHOUSE_LOCAL --multiquery "SELECT 311;" --multiquery "SELECT 312;" 2>&1 | grep -o 'Bad arguments' $CLICKHOUSE_LOCAL --multiquery "SELECT 313;" -n "SELECT 314;" 2>&1 | grep -o 'BAD_ARGUMENTS' -$CLICKHOUSE_LOCAL --multiquery "SELECT 315;" --query "SELECT 316;" 2>&1 | grep -o 'Bad arguments' $CLICKHOUSE_LOCAL -n "SELECT 320" --query "SELECT 317;" 2>&1 | grep -o 'BAD_ARGUMENTS' -$CLICKHOUSE_LOCAL --query --multiquery --multiquery "SELECT 318;" 2>&1 | grep -o 'Bad arguments' -$CLICKHOUSE_LOCAL --query --multiquery "SELECT 319;" 2>&1 | grep -o 'Bad arguments' $CLICKHOUSE_LOCAL --query -n "SELECT 400;" 2>&1 | grep -o 'Bad arguments' -$CLICKHOUSE_LOCAL --query -n --multiquery "SELECT 401;" 2>&1 | grep -o 'Bad arguments' \ No newline at end of file +$CLICKHOUSE_LOCAL --query -n --multiquery "SELECT 401;" 2>&1 | grep -o 'Bad arguments' From 4f4640ec90a450a39dd41dd0f798299b88d5983f Mon Sep 17 00:00:00 2001 From: slvrtrn Date: Mon, 4 Sep 2023 20:49:29 +0200 Subject: [PATCH 118/433] Add DateTime64 support --- .../PreparedStatementsTest.java | 108 +++++--- src/Core/MySQL/PacketsProtocolBinary.cpp | 248 ++++++++++++------ src/Core/MySQL/PacketsProtocolText.cpp | 4 + .../prepared_statements.reference | 157 ++++++----- .../prepared_statements_test.sql | 33 ++- 5 files changed, 364 insertions(+), 186 deletions(-) diff --git a/docker/test/integration/mysql_java_client/PreparedStatementsTest.java b/docker/test/integration/mysql_java_client/PreparedStatementsTest.java index cb479e13099..9efb3330fd5 100644 --- a/docker/test/integration/mysql_java_client/PreparedStatementsTest.java +++ b/docker/test/integration/mysql_java_client/PreparedStatementsTest.java @@ -46,6 +46,8 @@ public class PreparedStatementsTest { testDecimalTypes(conn); testMiscTypes(conn); testDateTypes(conn); + testUnusualDateTime64Scales(conn); + testDateTimeTimezones(conn); conn.close(); } catch (Exception e) { e.printStackTrace(); @@ -59,21 +61,21 @@ public class PreparedStatementsTest { int rowNum = 1; while (rs.next()) { System.out.printf("Row #%d\n", rowNum++); - System.out.printf("Type: %s, value: %d\n", getMysqlType(rs, "i8"), rs.getInt("i8")); - System.out.printf("Type: %s, value: %d\n", getMysqlType(rs, "i16"), rs.getInt("i16")); - System.out.printf("Type: %s, value: %d\n", getMysqlType(rs, "i32"), rs.getInt("i32")); - System.out.printf("Type: %s, value: %d\n", getMysqlType(rs, "i64"), rs.getLong("i64")); - System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "i128"), rs.getString("i128")); - System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "i256"), rs.getString("i256")); - System.out.printf("Type: %s, value: %d\n", getMysqlType(rs, "ui8"), rs.getInt("ui8")); - System.out.printf("Type: %s, value: %d\n", getMysqlType(rs, "ui16"), rs.getInt("ui16")); - System.out.printf("Type: %s, value: %d\n", getMysqlType(rs, "ui32"), rs.getLong("ui32")); - System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "ui64"), rs.getString("ui64")); - System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "ui128"), rs.getString("ui128")); - System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "ui256"), rs.getString("ui256")); - System.out.printf("Type: %s, value: %f\n", getMysqlType(rs, "f32"), rs.getFloat("f32")); - System.out.printf("Type: %s, value: %f\n", getMysqlType(rs, "f64"), rs.getFloat("f64")); - System.out.printf("Type: %s, value: %b\n", getMysqlType(rs, "b"), rs.getBoolean("b")); + System.out.printf("%s, value: %d\n", getMysqlType(rs, "i8"), rs.getInt("i8")); + System.out.printf("%s, value: %d\n", getMysqlType(rs, "i16"), rs.getInt("i16")); + System.out.printf("%s, value: %d\n", getMysqlType(rs, "i32"), rs.getInt("i32")); + System.out.printf("%s, value: %d\n", getMysqlType(rs, "i64"), rs.getLong("i64")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "i128"), rs.getString("i128")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "i256"), rs.getString("i256")); + System.out.printf("%s, value: %d\n", getMysqlType(rs, "ui8"), rs.getInt("ui8")); + System.out.printf("%s, value: %d\n", getMysqlType(rs, "ui16"), rs.getInt("ui16")); + System.out.printf("%s, value: %d\n", getMysqlType(rs, "ui32"), rs.getLong("ui32")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "ui64"), rs.getString("ui64")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "ui128"), rs.getString("ui128")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "ui256"), rs.getString("ui256")); + System.out.printf("%s, value: %f\n", getMysqlType(rs, "f32"), rs.getFloat("f32")); + System.out.printf("%s, value: %f\n", getMysqlType(rs, "f64"), rs.getFloat("f64")); + System.out.printf("%s, value: %b\n", getMysqlType(rs, "b"), rs.getBoolean("b")); } System.out.println(); } @@ -84,10 +86,10 @@ public class PreparedStatementsTest { int rowNum = 1; while (rs.next()) { System.out.printf("Row #%d\n", rowNum++); - System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "s"), rs.getString("s")); - System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "sn"), rs.getString("sn")); - System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "lc"), rs.getString("lc")); - System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "nlc"), rs.getString("nlc")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "s"), rs.getString("s")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "sn"), rs.getString("sn")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "lc"), rs.getString("lc")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "nlc"), rs.getString("nlc")); } System.out.println(); } @@ -98,16 +100,12 @@ public class PreparedStatementsTest { int rowNum = 1; while (rs.next()) { System.out.printf("Row #%d\n", rowNum++); - System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "d32"), - rs.getBigDecimal("d32").toPlainString()); - System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "d64"), - rs.getBigDecimal("d64").toPlainString()); - System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "d128_native"), + System.out.printf("%s, value: %s\n", getMysqlType(rs, "d32"), rs.getBigDecimal("d32").toPlainString()); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "d64"), rs.getBigDecimal("d64").toPlainString()); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "d128_native"), rs.getBigDecimal("d128_native").toPlainString()); - System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "d128_text"), - rs.getString("d128_text")); - System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "d256"), - rs.getString("d256")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "d128_text"), rs.getString("d128_text")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "d256"), rs.getString("d256")); } System.out.println(); } @@ -118,12 +116,41 @@ public class PreparedStatementsTest { int rowNum = 1; while (rs.next()) { System.out.printf("Row #%d\n", rowNum++); - System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "d"), rs.getDate("d")); - System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "d32"), rs.getDate("d32")); - System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "dt"), rs.getTimestamp("dt")); - System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "dt64_3"), rs.getString("dt64_3")); - System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "dt64_6"), rs.getString("dt64_6")); - System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "dt64_9"), rs.getString("dt64_9")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "d"), rs.getDate("d")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "d32"), rs.getDate("d32")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "dt"), rs.getTimestamp("dt")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "dt64_3"), rs.getTimestamp("dt64_3")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "dt64_6"), rs.getTimestamp("dt64_6")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "dt64_9"), rs.getTimestamp("dt64_9")); + } + System.out.println(); + } + + private static void testUnusualDateTime64Scales(Connection conn) throws SQLException { + System.out.println("### testUnusualDateTime64Scales"); + ResultSet rs = conn.prepareStatement("SELECT * FROM ps_unusual_datetime64_scales").executeQuery(); + int rowNum = 1; + while (rs.next()) { + System.out.printf("Row #%d\n", rowNum++); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "dt64_0"), rs.getTimestamp("dt64_0")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "dt64_1"), rs.getTimestamp("dt64_1")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "dt64_2"), rs.getTimestamp("dt64_2")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "dt64_4"), rs.getTimestamp("dt64_4")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "dt64_5"), rs.getTimestamp("dt64_5")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "dt64_7"), rs.getTimestamp("dt64_7")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "dt64_8"), rs.getTimestamp("dt64_8")); + } + System.out.println(); + } + + private static void testDateTimeTimezones(Connection conn) throws SQLException { + System.out.println("### testDateTimeTimezones"); + ResultSet rs = conn.prepareStatement("SELECT * FROM ps_datetime_timezones").executeQuery(); + int rowNum = 1; + while (rs.next()) { + System.out.printf("Row #%d\n", rowNum++); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "dt"), rs.getTimestamp("dt")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "dt64_3"), rs.getTimestamp("dt64_3")); } System.out.println(); } @@ -134,16 +161,17 @@ public class PreparedStatementsTest { int rowNum = 1; while (rs.next()) { System.out.printf("Row #%d\n", rowNum++); - System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "a"), rs.getString("a")); - System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "u"), rs.getString("u")); - System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "t"), rs.getString("t")); - System.out.printf("Type: %s, value: %s\n", getMysqlType(rs, "m"), rs.getString("m")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "a"), rs.getString("a")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "u"), rs.getString("u")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "t"), rs.getString("t")); + System.out.printf("%s, value: %s\n", getMysqlType(rs, "m"), rs.getString("m")); } System.out.println(); } - private static MysqlType getMysqlType(ResultSet rs, String columnLabel) throws SQLException { + private static String getMysqlType(ResultSet rs, String columnLabel) throws SQLException { ResultSetMetaData meta = rs.getMetaData(); - return MysqlType.getByJdbcType(meta.getColumnType(rs.findColumn(columnLabel))); + return String.format("%s type is %s", columnLabel, + MysqlType.getByJdbcType(meta.getColumnType(rs.findColumn(columnLabel)))); } } diff --git a/src/Core/MySQL/PacketsProtocolBinary.cpp b/src/Core/MySQL/PacketsProtocolBinary.cpp index ebb256cb462..ca206870a76 100644 --- a/src/Core/MySQL/PacketsProtocolBinary.cpp +++ b/src/Core/MySQL/PacketsProtocolBinary.cpp @@ -7,6 +7,8 @@ #include "Columns/ColumnLowCardinality.h" #include "Columns/ColumnVector.h" #include "Columns/ColumnsDateTime.h" +#include "Core/DecimalFunctions.h" +#include "DataTypes/DataTypeDateTime64.h" #include "DataTypes/DataTypeLowCardinality.h" #include "DataTypes/DataTypeNullable.h" #include "DataTypes/DataTypesNumber.h" @@ -74,46 +76,66 @@ namespace MySQLProtocol payload_size += 8; break; case TypeIndex::Date: { - UInt16 value = assert_cast &>(*col).getData()[row_num]; - if (value == 0) - { - payload_size += 1; // length only, no other fields - } - else - { - payload_size += 5; - } + payload_size += 5; break; } case TypeIndex::Date32: { - Int32 value = assert_cast &>(*col).getData()[row_num]; - if (value == 0) - { - payload_size += 1; // length only, no other fields - } - else - { - payload_size += 5; - } + payload_size += 5; break; } case TypeIndex::DateTime: { UInt32 value = assert_cast &>(*col).getData()[row_num]; - if (value == 0) + LocalDateTime ldt = LocalDateTime(value, DateLUT::instance(getDateTimeTimezone(*data_type))); + + bool has_time = !(ldt.hour() == 0 && ldt.minute() == 0 && ldt.second() == 0); + if (has_time) { - payload_size += 1; // length only, no other fields + payload_size += 8; } else { - LocalDateTime ldt = LocalDateTime(value, DateLUT::instance(getDateTimeTimezone(*data_type))); - if (ldt.second() == 0 && ldt.minute() == 0 && ldt.hour() == 0) - { - payload_size += 5; - } - else - { - payload_size += 8; - } + payload_size += 5; + } + break; + } + case TypeIndex::DateTime64: { + const auto * date_time_type = typeid_cast(data_type.get()); + UInt32 scale = date_time_type->getScale(); + + static constexpr UInt32 MaxScale = DecimalUtils::max_precision; + scale = scale > MaxScale ? MaxScale : scale; + + const auto dt64 = assert_cast(*col).getData()[row_num]; + auto components = DecimalUtils::split(dt64, scale); + + using T = typename DateTime64::NativeType; + if (dt64.value < 0 && components.fractional) + { + components.fractional + = DecimalUtils::scaleMultiplier(scale) + (components.whole ? T(-1) : T(1)) * components.fractional; + --components.whole; + } + if (scale > 6) + { + // MySQL Timestamp has max scale of 6 + components.fractional /= static_cast(pow(10, scale - 6)); + } + + LocalDateTime ldt = LocalDateTime(components.whole, DateLUT::instance(getDateTimeTimezone(*data_type))); + + bool has_microseconds = components.fractional != 0; + bool has_time = !(ldt.hour() == 0 && ldt.minute() == 0 && ldt.second() == 0); + if (has_microseconds) + { + payload_size += 12; + } + else if (has_time) + { + payload_size += 8; + } + else + { + payload_size += 5; } break; } @@ -199,73 +221,133 @@ namespace MySQLProtocol } case TypeIndex::Date: { UInt16 value = assert_cast &>(*col).getData()[row_num]; - if (value != 0) - { - LocalDate ld = LocalDate(DayNum(value)); - buffer.write(static_cast(4)); // bytes_following - auto year = ld.year(); - auto month = ld.month(); - auto day = ld.day(); - buffer.write(reinterpret_cast(&year), 2); - buffer.write(reinterpret_cast(&month), 1); - buffer.write(reinterpret_cast(&day), 1); - } - else - { - buffer.write(static_cast(0)); - } + LocalDate ld = LocalDate(DayNum(value)); + buffer.write(static_cast(4)); // bytes_following + auto year = ld.year(); + auto month = ld.month(); + auto day = ld.day(); + buffer.write(reinterpret_cast(&year), 2); + buffer.write(reinterpret_cast(&month), 1); + buffer.write(reinterpret_cast(&day), 1); break; } case TypeIndex::Date32: { Int32 value = assert_cast &>(*col).getData()[row_num]; - if (value != 0) - { - LocalDate ld = LocalDate(ExtendedDayNum(value)); - buffer.write(static_cast(4)); // bytes_following - auto year = ld.year(); - auto month = ld.month(); - auto day = ld.day(); - buffer.write(reinterpret_cast(&year), 2); - buffer.write(reinterpret_cast(&month), 1); - buffer.write(reinterpret_cast(&day), 1); - } - else - { - buffer.write(static_cast(0)); - } + LocalDate ld = LocalDate(ExtendedDayNum(value)); + buffer.write(static_cast(4)); // bytes_following + auto year = ld.year(); + auto month = ld.month(); + auto day = ld.day(); + buffer.write(reinterpret_cast(&year), 2); + buffer.write(reinterpret_cast(&month), 1); + buffer.write(reinterpret_cast(&day), 1); break; } case TypeIndex::DateTime: { UInt32 value = assert_cast &>(*col).getData()[row_num]; - if (value != 0) + String timezone = getDateTimeTimezone(*data_type); + LocalDateTime ldt = LocalDateTime(value, DateLUT::instance(timezone)); + int year = ldt.year(); + int month = ldt.month(); + int day = ldt.day(); + int hour = ldt.hour(); + int minute = ldt.minute(); + int second = ldt.second(); + bool has_time = !(hour == 0 && minute == 0 && second == 0); + size_t bytes_following = has_time ? 7 : 4; + buffer.write(reinterpret_cast(&bytes_following), 1); + buffer.write(reinterpret_cast(&year), 2); + buffer.write(reinterpret_cast(&month), 1); + buffer.write(reinterpret_cast(&day), 1); + if (has_time) { - LocalDateTime ldt = LocalDateTime(value, DateLUT::instance(getDateTimeTimezone(*data_type))); - int year = ldt.year(); - int month = ldt.month(); - int day = ldt.day(); - int hour = ldt.hour(); - int minute = ldt.minute(); - int second = ldt.second(); - bool has_time = !(hour == 0 && minute == 0 && second == 0); - size_t bytes_following = has_time ? 7 : 4; - buffer.write(reinterpret_cast(&bytes_following), 1); - buffer.write(reinterpret_cast(&year), 2); - buffer.write(reinterpret_cast(&month), 1); - buffer.write(reinterpret_cast(&day), 1); - if (has_time) - { - buffer.write(reinterpret_cast(&hour), 1); - buffer.write(reinterpret_cast(&minute), 1); - buffer.write(reinterpret_cast(&second), 1); - } - } - else - { - buffer.write(static_cast(0)); + buffer.write(reinterpret_cast(&hour), 1); + buffer.write(reinterpret_cast(&minute), 1); + buffer.write(reinterpret_cast(&second), 1); } break; } + case TypeIndex::DateTime64: { + const auto * date_time_type = typeid_cast(data_type.get()); + UInt32 scale = date_time_type->getScale(); + static constexpr UInt32 MaxScale = DecimalUtils::max_precision; + scale = scale > MaxScale ? MaxScale : scale; + + const auto dt64 = assert_cast(*col).getData()[row_num]; + auto components = DecimalUtils::split(dt64, scale); + + using T = typename DateTime64::NativeType; + if (dt64.value < 0 && components.fractional) + { + components.fractional + = DecimalUtils::scaleMultiplier(scale) + (components.whole ? T(-1) : T(1)) * components.fractional; + --components.whole; + } + + if (components.fractional != 0) + { + if (scale > 6) + { + // MySQL Timestamp has max scale of 6 + components.fractional /= static_cast(pow(10, scale - 6)); + } + else + { + // fractional == 1 is a different microsecond value depending on the scale + // Scale 1 = 100 000 + // Scale 2 = 010 000 + // Scale 3 = 001 000 + // Scale 4 = 000 100 + // Scale 5 = 000 010 + // Scale 6 = 000 001 + components.fractional *= static_cast(pow(10, 6 - scale)); + } + } + + String timezone = getDateTimeTimezone(*data_type); + std::cout << "Timezone is " << timezone << std::endl; + LocalDateTime ldt = LocalDateTime(components.whole, DateLUT::instance(timezone)); + auto year = ldt.year(); + auto month = ldt.month(); + auto day = ldt.day(); + auto hour = ldt.hour(); + auto minute = ldt.minute(); + auto second = ldt.second(); + + bool has_time = !(hour == 0 && minute == 0 && second == 0); + bool has_microseconds = components.fractional != 0; + + if (has_microseconds) + { + buffer.write(static_cast(11)); // bytes_following + buffer.write(reinterpret_cast(&year), 2); + buffer.write(reinterpret_cast(&month), 1); + buffer.write(reinterpret_cast(&day), 1); + buffer.write(reinterpret_cast(&hour), 1); + buffer.write(reinterpret_cast(&minute), 1); + buffer.write(reinterpret_cast(&second), 1); + buffer.write(reinterpret_cast(&components.fractional), 4); + } + else if (has_time) + { + buffer.write(static_cast(7)); // bytes_following + buffer.write(reinterpret_cast(&year), 2); + buffer.write(reinterpret_cast(&month), 1); + buffer.write(reinterpret_cast(&day), 1); + buffer.write(reinterpret_cast(&hour), 1); + buffer.write(reinterpret_cast(&minute), 1); + buffer.write(reinterpret_cast(&second), 1); + } + else + { + buffer.write(static_cast(4)); // bytes_following + buffer.write(reinterpret_cast(&year), 2); + buffer.write(reinterpret_cast(&month), 1); + buffer.write(reinterpret_cast(&day), 1); + } + break; + } default: writeLengthEncodedString(serialized[i], buffer); break; diff --git a/src/Core/MySQL/PacketsProtocolText.cpp b/src/Core/MySQL/PacketsProtocolText.cpp index 712b203bc2a..9f4de31e44d 100644 --- a/src/Core/MySQL/PacketsProtocolText.cpp +++ b/src/Core/MySQL/PacketsProtocolText.cpp @@ -211,6 +211,10 @@ namespace MySQLProtocol column_type = ColumnType::MYSQL_TYPE_DATETIME; flags = ColumnDefinitionFlags::BINARY_FLAG; break; + case TypeIndex::DateTime64: + column_type = ColumnType::MYSQL_TYPE_DATETIME; + flags = ColumnDefinitionFlags::BINARY_FLAG; + break; case TypeIndex::Decimal32: case TypeIndex::Decimal64: column_type = ColumnType::MYSQL_TYPE_DECIMAL; diff --git a/tests/integration/test_mysql_protocol/prepared_statements.reference b/tests/integration/test_mysql_protocol/prepared_statements.reference index 3e23620c07e..402e64348ae 100644 --- a/tests/integration/test_mysql_protocol/prepared_statements.reference +++ b/tests/integration/test_mysql_protocol/prepared_statements.reference @@ -1,81 +1,114 @@ ### testSimpleDataTypes Row #1 -Type: TINYINT, value: -128 -Type: SMALLINT, value: -32768 -Type: INT, value: -2147483648 -Type: BIGINT, value: -9223372036854775808 -Type: CHAR, value: -170141183460469231731687303715884105728 -Type: CHAR, value: -57896044618658097711785492504343953926634992332820282019728792003956564819968 -Type: TINYINT, value: 120 -Type: SMALLINT, value: 1234 -Type: INT, value: 51234 -Type: BIGINT, value: 421342 -Type: CHAR, value: 15324355 -Type: CHAR, value: 41345135123432 -Type: FLOAT, value: -0.796896 -Type: DOUBLE, value: -0.113259 -Type: BIT, value: true +i8 type is TINYINT, value: -128 +i16 type is SMALLINT, value: -32768 +i32 type is INT, value: -2147483648 +i64 type is BIGINT, value: -9223372036854775808 +i128 type is CHAR, value: -170141183460469231731687303715884105728 +i256 type is CHAR, value: -57896044618658097711785492504343953926634992332820282019728792003956564819968 +ui8 type is TINYINT, value: 120 +ui16 type is SMALLINT, value: 1234 +ui32 type is INT, value: 51234 +ui64 type is BIGINT, value: 421342 +ui128 type is CHAR, value: 15324355 +ui256 type is CHAR, value: 41345135123432 +f32 type is FLOAT, value: -0.796896 +f64 type is DOUBLE, value: -0.113259 +b type is BIT, value: true Row #2 -Type: TINYINT, value: 127 -Type: SMALLINT, value: 32767 -Type: INT, value: 2147483647 -Type: BIGINT, value: 9223372036854775807 -Type: CHAR, value: 170141183460469231731687303715884105727 -Type: CHAR, value: 57896044618658097711785492504343953926634992332820282019728792003956564819967 -Type: TINYINT, value: 255 -Type: SMALLINT, value: 65535 -Type: INT, value: 4294967295 -Type: BIGINT, value: 18446744073709551615 -Type: CHAR, value: 340282366920938463463374607431768211455 -Type: CHAR, value: 115792089237316195423570985008687907853269984665640564039457584007913129639935 -Type: FLOAT, value: 1.234000 -Type: DOUBLE, value: 3.352451 -Type: BIT, value: false +i8 type is TINYINT, value: 127 +i16 type is SMALLINT, value: 32767 +i32 type is INT, value: 2147483647 +i64 type is BIGINT, value: 9223372036854775807 +i128 type is CHAR, value: 170141183460469231731687303715884105727 +i256 type is CHAR, value: 57896044618658097711785492504343953926634992332820282019728792003956564819967 +ui8 type is TINYINT, value: 255 +ui16 type is SMALLINT, value: 65535 +ui32 type is INT, value: 4294967295 +ui64 type is BIGINT, value: 18446744073709551615 +ui128 type is CHAR, value: 340282366920938463463374607431768211455 +ui256 type is CHAR, value: 115792089237316195423570985008687907853269984665640564039457584007913129639935 +f32 type is FLOAT, value: 1.234000 +f64 type is DOUBLE, value: 3.352451 +b type is BIT, value: false ### testStringTypes Row #1 -Type: CHAR, value: 42 -Type: CHAR, value: null -Type: CHAR, value: test -Type: CHAR, value: null +s type is CHAR, value: 42 +sn type is CHAR, value: null +lc type is CHAR, value: test +nlc type is CHAR, value: null Row #2 -Type: CHAR, value: foo -Type: CHAR, value: bar -Type: CHAR, value: qaz -Type: CHAR, value: qux +s type is CHAR, value: foo +sn type is CHAR, value: bar +lc type is CHAR, value: qaz +nlc type is CHAR, value: qux ### testDecimalTypes Row #1 -Type: DECIMAL, value: -1.55 -Type: DECIMAL, value: 6.03 -Type: DECIMAL, value: 5 -Type: CHAR, value: -1224124.23423 -Type: CHAR, value: -54342.3 +d32 type is DECIMAL, value: -1.55 +d64 type is DECIMAL, value: 6.03 +d128_native type is DECIMAL, value: 5 +d128_text type is CHAR, value: -1224124.23423 +d256 type is CHAR, value: -54342.3 Row #2 -Type: DECIMAL, value: 1234567.89 -Type: DECIMAL, value: 123456789123456.789 -Type: DECIMAL, value: 12345678912345678912.1234567891 -Type: CHAR, value: 1234567.8912345678912345678911234567891 -Type: CHAR, value: 12345678912345678912345678911234567891234567891234567891.12345678911234567891 +d32 type is DECIMAL, value: 1234567.89 +d64 type is DECIMAL, value: 123456789123456.789 +d128_native type is DECIMAL, value: 12345678912345678912.1234567891 +d128_text type is CHAR, value: 1234567.8912345678912345678911234567891 +d256 type is CHAR, value: 12345678912345678912345678911234567891234567891234567891.12345678911234567891 ### testMiscTypes Row #1 -Type: CHAR, value: [] -Type: CHAR, value: 9a0ccc06-2578-4861-8534-631c9d40f3f7 -Type: CHAR, value: (0,'') -Type: CHAR, value: {} +a type is CHAR, value: [] +u type is CHAR, value: 9a0ccc06-2578-4861-8534-631c9d40f3f7 +t type is CHAR, value: (0,'') +m type is CHAR, value: {} Row #2 -Type: CHAR, value: ['foo','bar'] -Type: CHAR, value: 5da5038d-788f-48c6-b510-babb41c538d3 -Type: CHAR, value: (42,'qaz') -Type: CHAR, value: {'qux':144,'text':255} +a type is CHAR, value: ['foo','bar'] +u type is CHAR, value: 5da5038d-788f-48c6-b510-babb41c538d3 +t type is CHAR, value: (42,'qaz') +m type is CHAR, value: {'qux':144,'text':255} ### testDateTypes Row #1 -Type: DATE, value: 2149-06-06 -Type: DATE, value: 2178-04-16 -Type: TIMESTAMP, value: 2106-02-07 06:28:15.0 -Type: CHAR, value: 2106-02-07 06:28:15.123 -Type: CHAR, value: 2106-02-07 06:28:15.123456 -Type: CHAR, value: 2106-02-07 06:28:15.123456789 +d type is DATE, value: 1970-01-01 +d32 type is DATE, value: 1900-01-01 +dt type is TIMESTAMP, value: 1970-01-01 01:00:00.0 +dt64_3 type is TIMESTAMP, value: 1900-01-01 00:00:00.0 +dt64_6 type is TIMESTAMP, value: 1900-01-01 00:00:00.000001 +dt64_9 type is TIMESTAMP, value: 1900-01-01 00:00:00.0 +Row #2 +d type is DATE, value: 2149-06-06 +d32 type is DATE, value: 2178-04-16 +dt type is TIMESTAMP, value: 2106-02-07 06:28:15.0 +dt64_3 type is TIMESTAMP, value: 2106-02-07 06:28:15.123 +dt64_6 type is TIMESTAMP, value: 2106-02-07 06:28:15.123456 +dt64_9 type is TIMESTAMP, value: 2106-02-07 06:28:15.123456 + +### testUnusualDateTime64Scales +Row #1 +dt64_0 type is TIMESTAMP, value: 2022-04-13 03:17:45.0 +dt64_1 type is TIMESTAMP, value: 2022-04-13 03:17:45.1 +dt64_2 type is TIMESTAMP, value: 2022-04-13 03:17:45.12 +dt64_4 type is TIMESTAMP, value: 2022-04-13 03:17:45.1234 +dt64_5 type is TIMESTAMP, value: 2022-04-13 03:17:45.12345 +dt64_7 type is TIMESTAMP, value: 2022-04-13 03:17:45.123456 +dt64_8 type is TIMESTAMP, value: 2022-04-13 03:17:45.123456 +Row #2 +dt64_0 type is TIMESTAMP, value: 2022-04-13 03:17:45.0 +dt64_1 type is TIMESTAMP, value: 2022-04-13 03:17:45.1 +dt64_2 type is TIMESTAMP, value: 2022-04-13 03:17:45.01 +dt64_4 type is TIMESTAMP, value: 2022-04-13 03:17:45.0001 +dt64_5 type is TIMESTAMP, value: 2022-04-13 03:17:45.00001 +dt64_7 type is TIMESTAMP, value: 2022-04-13 03:17:45.0 +dt64_8 type is TIMESTAMP, value: 2022-04-13 03:17:45.0 + +### testDateTimeTimezones +Row #1 +dt type is TIMESTAMP, value: 1970-01-01 01:00:00.0 +dt64_3 type is TIMESTAMP, value: 1969-12-31 16:00:00.0 +Row #2 +dt type is TIMESTAMP, value: 2022-09-04 20:31:05.0 +dt64_3 type is TIMESTAMP, value: 2022-09-04 20:31:05.022 diff --git a/tests/integration/test_mysql_protocol/prepared_statements_test.sql b/tests/integration/test_mysql_protocol/prepared_statements_test.sql index 4fb53a14a61..a32ce9db1d5 100644 --- a/tests/integration/test_mysql_protocol/prepared_statements_test.sql +++ b/tests/integration/test_mysql_protocol/prepared_statements_test.sql @@ -83,4 +83,35 @@ INSERT INTO ps_date_types VALUES ('2149-06-06', '2178-04-16', '2106-02-07 06:28:15', '2106-02-07 06:28:15.123', '2106-02-07 06:28:15.123456', - '2106-02-07 06:28:15.123456789'); + '2106-02-07 06:28:15.123456789'), + ('1970-01-01', '1900-01-01', '1970-01-01 00:00:00', + '1900-01-01 00:00:00.001', + '1900-01-01 00:00:00.000001', + '1900-01-01 00:00:00.000000001');; + +CREATE TABLE ps_unusual_datetime64_scales +( + dt64_0 DateTime64(0, 'UTC'), + dt64_1 DateTime64(1, 'UTC'), + dt64_2 DateTime64(2, 'UTC'), + dt64_4 DateTime64(4, 'UTC'), + dt64_5 DateTime64(5, 'UTC'), + dt64_7 DateTime64(7, 'UTC'), + dt64_8 DateTime64(8, 'UTC') +) ENGINE MergeTree ORDER BY dt64_0; + +INSERT INTO ps_unusual_datetime64_scales +VALUES ('2022-04-13 03:17:45', + '2022-04-13 03:17:45.1', + '2022-04-13 03:17:45.12', + '2022-04-13 03:17:45.1234', + '2022-04-13 03:17:45.12345', + '2022-04-13 03:17:45.1234567', + '2022-04-13 03:17:45.12345678'), + ('2022-04-13 03:17:45', + '2022-04-13 03:17:45.1', + '2022-04-13 03:17:45.01', + '2022-04-13 03:17:45.0001', + '2022-04-13 03:17:45.00001', + '2022-04-13 03:17:45.0000001', + '2022-04-13 03:17:45.00000001'); From 2245a8df9dd1f66ab2bc1cb35be19f0b3ca97a94 Mon Sep 17 00:00:00 2001 From: slvrtrn Date: Mon, 4 Sep 2023 20:58:24 +0200 Subject: [PATCH 119/433] Revert format changes --- src/Core/MySQL/PacketsGeneric.cpp | 499 ++++++++++++------------- src/Core/MySQL/PacketsProtocolText.cpp | 452 +++++++++++----------- src/DataTypes/DataTypesDecimal.cpp | 18 +- 3 files changed, 478 insertions(+), 491 deletions(-) diff --git a/src/Core/MySQL/PacketsGeneric.cpp b/src/Core/MySQL/PacketsGeneric.cpp index 88183890399..23f197bcfe8 100644 --- a/src/Core/MySQL/PacketsGeneric.cpp +++ b/src/Core/MySQL/PacketsGeneric.cpp @@ -8,264 +8,253 @@ namespace DB namespace MySQLProtocol { - namespace Generic +namespace Generic +{ + +static const size_t MYSQL_ERRMSG_SIZE = 512; + +void SSLRequest::readPayloadImpl(ReadBuffer & buf) +{ + buf.readStrict(reinterpret_cast(&capability_flags), 4); + buf.readStrict(reinterpret_cast(&max_packet_size), 4); + buf.readStrict(reinterpret_cast(&character_set), 1); +} + +OKPacket::OKPacket(uint32_t capabilities_) + : header(0x00), capabilities(capabilities_), affected_rows(0x00), last_insert_id(0x00), status_flags(0x00) +{ +} + +OKPacket::OKPacket( + uint8_t header_, uint32_t capabilities_, uint64_t affected_rows_, uint32_t status_flags_, int16_t warnings_, + String session_state_changes_, String info_) + : header(header_), capabilities(capabilities_), affected_rows(affected_rows_), last_insert_id(0), warnings(warnings_), + status_flags(status_flags_), session_state_changes(std::move(session_state_changes_)), info(std::move(info_)) +{ +} + +size_t OKPacket::getPayloadSize() const +{ + size_t result = 2 + getLengthEncodedNumberSize(affected_rows); + + if (capabilities & CLIENT_PROTOCOL_41) { - - static const size_t MYSQL_ERRMSG_SIZE = 512; - - void SSLRequest::readPayloadImpl(ReadBuffer & buf) - { - buf.readStrict(reinterpret_cast(&capability_flags), 4); - buf.readStrict(reinterpret_cast(&max_packet_size), 4); - buf.readStrict(reinterpret_cast(&character_set), 1); - } - - OKPacket::OKPacket(uint32_t capabilities_) - : header(0x00), capabilities(capabilities_), affected_rows(0x00), last_insert_id(0x00), status_flags(0x00) - { - } - - OKPacket::OKPacket( - uint8_t header_, - uint32_t capabilities_, - uint64_t affected_rows_, - uint32_t status_flags_, - int16_t warnings_, - String session_state_changes_, - String info_) - : header(header_) - , capabilities(capabilities_) - , affected_rows(affected_rows_) - , last_insert_id(0) - , warnings(warnings_) - , status_flags(status_flags_) - , session_state_changes(std::move(session_state_changes_)) - , info(std::move(info_)) - { - } - - size_t OKPacket::getPayloadSize() const - { - size_t result = 2 + getLengthEncodedNumberSize(affected_rows); - - if (capabilities & CLIENT_PROTOCOL_41) - { - result += 4; - } - else if (capabilities & CLIENT_TRANSACTIONS) - { - result += 2; - } - - if (capabilities & CLIENT_SESSION_TRACK) - { - result += getLengthEncodedStringSize(info); - if (status_flags & SERVER_SESSION_STATE_CHANGED) - result += getLengthEncodedStringSize(session_state_changes); - } - else - { - result += info.size(); - } - - return result; - } - - void OKPacket::readPayloadImpl(ReadBuffer & payload) - - { - payload.readStrict(reinterpret_cast(&header), 1); - affected_rows = readLengthEncodedNumber(payload); - last_insert_id = readLengthEncodedNumber(payload); - - if (capabilities & CLIENT_PROTOCOL_41) - { - payload.readStrict(reinterpret_cast(&status_flags), 2); - payload.readStrict(reinterpret_cast(&warnings), 2); - } - else if (capabilities & CLIENT_TRANSACTIONS) - { - payload.readStrict(reinterpret_cast(&status_flags), 2); - } - - if (capabilities & CLIENT_SESSION_TRACK) - { - readLengthEncodedString(info, payload); - if (status_flags & SERVER_SESSION_STATE_CHANGED) - { - readLengthEncodedString(session_state_changes, payload); - } - } - else - { - readString(info, payload); - } - } - - void OKPacket::writePayloadImpl(WriteBuffer & buffer) const - - { - buffer.write(header); - writeLengthEncodedNumber(affected_rows, buffer); - writeLengthEncodedNumber(last_insert_id, buffer); /// last insert-id - - if (capabilities & CLIENT_PROTOCOL_41) - { - buffer.write(reinterpret_cast(&status_flags), 2); - buffer.write(reinterpret_cast(&warnings), 2); - } - else if (capabilities & CLIENT_TRANSACTIONS) - { - buffer.write(reinterpret_cast(&status_flags), 2); - } - - if (capabilities & CLIENT_SESSION_TRACK) - { - writeLengthEncodedString(info, buffer); - if (status_flags & SERVER_SESSION_STATE_CHANGED) - writeLengthEncodedString(session_state_changes, buffer); - } - else - { - writeString(info, buffer); - } - } - - EOFPacket::EOFPacket() : warnings(0x00), status_flags(0x00) - { - } - - EOFPacket::EOFPacket(int warnings_, int status_flags_) : warnings(warnings_), status_flags(status_flags_) - { - } - - size_t EOFPacket::getPayloadSize() const - { - return 5; - } - - void EOFPacket::readPayloadImpl(ReadBuffer & payload) - { - payload.readStrict(reinterpret_cast(&header), 1); - assert(header == 0xfe); - payload.readStrict(reinterpret_cast(&warnings), 2); - payload.readStrict(reinterpret_cast(&status_flags), 2); - } - - void EOFPacket::writePayloadImpl(WriteBuffer & buffer) const - { - buffer.write(header); // EOF header - buffer.write(reinterpret_cast(&warnings), 2); - buffer.write(reinterpret_cast(&status_flags), 2); - } - - void AuthSwitchPacket::readPayloadImpl(ReadBuffer & payload) - { - payload.readStrict(reinterpret_cast(&header), 1); - assert(header == 0xfe); - readStringUntilEOF(plugin_name, payload); - } - - ERRPacket::ERRPacket() : error_code(0x00) - { - } - - ERRPacket::ERRPacket(int error_code_, String sql_state_, String error_message_) - : error_code(error_code_), sql_state(std::move(sql_state_)), error_message(std::move(error_message_)) - { - } - - size_t ERRPacket::getPayloadSize() const - { - return 4 + sql_state.length() + std::min(error_message.length(), MYSQL_ERRMSG_SIZE); - } - - void ERRPacket::readPayloadImpl(ReadBuffer & payload) - { - payload.readStrict(reinterpret_cast(&header), 1); - assert(header == 0xff); - - payload.readStrict(reinterpret_cast(&error_code), 2); - - /// SQL State [optional: # + 5bytes string] - UInt8 sharp = static_cast(*payload.position()); - if (sharp == 0x23) - { - payload.ignore(1); - sql_state.resize(5); - payload.readStrict(reinterpret_cast(sql_state.data()), 5); - } - readString(error_message, payload); - } - - void ERRPacket::writePayloadImpl(WriteBuffer & buffer) const - { - buffer.write(header); - buffer.write(reinterpret_cast(&error_code), 2); - buffer.write('#'); - buffer.write(sql_state.data(), sql_state.length()); - buffer.write(error_message.data(), std::min(error_message.length(), MYSQL_ERRMSG_SIZE)); - } - - ResponsePacket::ResponsePacket(UInt32 server_capability_flags_) : ok(OKPacket(server_capability_flags_)) - { - } - - ResponsePacket::ResponsePacket(UInt32 server_capability_flags_, bool is_handshake_) - : ok(OKPacket(server_capability_flags_)), is_handshake(is_handshake_) - { - } - - void ResponsePacket::readPayloadImpl(ReadBuffer & payload) - { - UInt16 header = static_cast(*payload.position()); - switch (header) - { - case PACKET_OK: - packetType = PACKET_OK; - ok.readPayloadWithUnpacked(payload); - break; - case PACKET_ERR: - packetType = PACKET_ERR; - err.readPayloadWithUnpacked(payload); - break; - case PACKET_EOF: - if (is_handshake) - { - packetType = PACKET_AUTH_SWITCH; - auth_switch.readPayloadWithUnpacked(payload); - } - else - { - packetType = PACKET_EOF; - eof.readPayloadWithUnpacked(payload); - } - break; - case PACKET_LOCALINFILE: - packetType = PACKET_LOCALINFILE; - break; - default: - packetType = PACKET_OK; - column_length = readLengthEncodedNumber(payload); - } - } - - LengthEncodedNumber::LengthEncodedNumber(uint64_t value_) : value(value_) - { - } - - size_t LengthEncodedNumber::getPayloadSize() const - { - return getLengthEncodedNumberSize(value); - } - - void LengthEncodedNumber::writePayloadImpl(WriteBuffer & buffer) const - { - writeLengthEncodedNumber(value, buffer); - } - + result += 4; + } + else if (capabilities & CLIENT_TRANSACTIONS) + { + result += 2; } + if (capabilities & CLIENT_SESSION_TRACK) + { + result += getLengthEncodedStringSize(info); + if (status_flags & SERVER_SESSION_STATE_CHANGED) + result += getLengthEncodedStringSize(session_state_changes); + } + else + { + result += info.size(); + } + + return result; +} + +void OKPacket::readPayloadImpl(ReadBuffer & payload) + +{ + payload.readStrict(reinterpret_cast(&header), 1); + affected_rows = readLengthEncodedNumber(payload); + last_insert_id = readLengthEncodedNumber(payload); + + if (capabilities & CLIENT_PROTOCOL_41) + { + payload.readStrict(reinterpret_cast(&status_flags), 2); + payload.readStrict(reinterpret_cast(&warnings), 2); + } + else if (capabilities & CLIENT_TRANSACTIONS) + { + payload.readStrict(reinterpret_cast(&status_flags), 2); + } + + if (capabilities & CLIENT_SESSION_TRACK) + { + readLengthEncodedString(info, payload); + if (status_flags & SERVER_SESSION_STATE_CHANGED) + { + readLengthEncodedString(session_state_changes, payload); + } + } + else + { + readString(info, payload); + } +} + +void OKPacket::writePayloadImpl(WriteBuffer & buffer) const + +{ + buffer.write(header); + writeLengthEncodedNumber(affected_rows, buffer); + writeLengthEncodedNumber(last_insert_id, buffer); /// last insert-id + + if (capabilities & CLIENT_PROTOCOL_41) + { + buffer.write(reinterpret_cast(&status_flags), 2); + buffer.write(reinterpret_cast(&warnings), 2); + } + else if (capabilities & CLIENT_TRANSACTIONS) + { + buffer.write(reinterpret_cast(&status_flags), 2); + } + + if (capabilities & CLIENT_SESSION_TRACK) + { + writeLengthEncodedString(info, buffer); + if (status_flags & SERVER_SESSION_STATE_CHANGED) + writeLengthEncodedString(session_state_changes, buffer); + } + else + { + writeString(info, buffer); + } +} + +EOFPacket::EOFPacket() : warnings(0x00), status_flags(0x00) +{ +} + +EOFPacket::EOFPacket(int warnings_, int status_flags_) : warnings(warnings_), status_flags(status_flags_) +{ +} + +size_t EOFPacket::getPayloadSize() const +{ + return 5; +} + +void EOFPacket::readPayloadImpl(ReadBuffer & payload) +{ + payload.readStrict(reinterpret_cast(&header), 1); + assert(header == 0xfe); + payload.readStrict(reinterpret_cast(&warnings), 2); + payload.readStrict(reinterpret_cast(&status_flags), 2); +} + +void EOFPacket::writePayloadImpl(WriteBuffer & buffer) const +{ + buffer.write(header); // EOF header + buffer.write(reinterpret_cast(&warnings), 2); + buffer.write(reinterpret_cast(&status_flags), 2); +} + +void AuthSwitchPacket::readPayloadImpl(ReadBuffer & payload) +{ + payload.readStrict(reinterpret_cast(&header), 1); + assert(header == 0xfe); + readStringUntilEOF(plugin_name, payload); +} + +ERRPacket::ERRPacket() : error_code(0x00) +{ +} + +ERRPacket::ERRPacket(int error_code_, String sql_state_, String error_message_) + : error_code(error_code_), sql_state(std::move(sql_state_)), error_message(std::move(error_message_)) +{ +} + +size_t ERRPacket::getPayloadSize() const +{ + return 4 + sql_state.length() + std::min(error_message.length(), MYSQL_ERRMSG_SIZE); +} + +void ERRPacket::readPayloadImpl(ReadBuffer & payload) +{ + payload.readStrict(reinterpret_cast(&header), 1); + assert(header == 0xff); + + payload.readStrict(reinterpret_cast(&error_code), 2); + + /// SQL State [optional: # + 5bytes string] + UInt8 sharp = static_cast(*payload.position()); + if (sharp == 0x23) + { + payload.ignore(1); + sql_state.resize(5); + payload.readStrict(reinterpret_cast(sql_state.data()), 5); + } + readString(error_message, payload); +} + +void ERRPacket::writePayloadImpl(WriteBuffer & buffer) const +{ + buffer.write(header); + buffer.write(reinterpret_cast(&error_code), 2); + buffer.write('#'); + buffer.write(sql_state.data(), sql_state.length()); + buffer.write(error_message.data(), std::min(error_message.length(), MYSQL_ERRMSG_SIZE)); +} + +ResponsePacket::ResponsePacket(UInt32 server_capability_flags_) : ok(OKPacket(server_capability_flags_)) +{ +} + +ResponsePacket::ResponsePacket(UInt32 server_capability_flags_, bool is_handshake_) + : ok(OKPacket(server_capability_flags_)), is_handshake(is_handshake_) +{ +} + +void ResponsePacket::readPayloadImpl(ReadBuffer & payload) +{ + UInt16 header = static_cast(*payload.position()); + switch (header) + { + case PACKET_OK: + packetType = PACKET_OK; + ok.readPayloadWithUnpacked(payload); + break; + case PACKET_ERR: + packetType = PACKET_ERR; + err.readPayloadWithUnpacked(payload); + break; + case PACKET_EOF: + if (is_handshake) + { + packetType = PACKET_AUTH_SWITCH; + auth_switch.readPayloadWithUnpacked(payload); + } + else + { + packetType = PACKET_EOF; + eof.readPayloadWithUnpacked(payload); + } + break; + case PACKET_LOCALINFILE: + packetType = PACKET_LOCALINFILE; + break; + default: + packetType = PACKET_OK; + column_length = readLengthEncodedNumber(payload); + } +} + +LengthEncodedNumber::LengthEncodedNumber(uint64_t value_) : value(value_) +{ +} + +size_t LengthEncodedNumber::getPayloadSize() const +{ + return getLengthEncodedNumberSize(value); +} + +void LengthEncodedNumber::writePayloadImpl(WriteBuffer & buffer) const +{ + writeLengthEncodedNumber(value, buffer); +} + +} + } } diff --git a/src/Core/MySQL/PacketsProtocolText.cpp b/src/Core/MySQL/PacketsProtocolText.cpp index 9f4de31e44d..f01f2ff9b61 100644 --- a/src/Core/MySQL/PacketsProtocolText.cpp +++ b/src/Core/MySQL/PacketsProtocolText.cpp @@ -12,238 +12,238 @@ namespace DB namespace MySQLProtocol { - namespace ProtocolText +namespace ProtocolText +{ + +ResultSetRow::ResultSetRow(const Serializations & serializations, const Columns & columns_, int row_num_) + : columns(columns_), row_num(row_num_) +{ + for (size_t i = 0; i < columns.size(); ++i) { - - ResultSetRow::ResultSetRow(const Serializations & serializations, const Columns & columns_, int row_num_) - : columns(columns_), row_num(row_num_) + if (columns[i]->isNullAt(row_num)) { - for (size_t i = 0; i < columns.size(); ++i) - { - if (columns[i]->isNullAt(row_num)) - { - payload_size += 1; - serialized.emplace_back("\xfb"); - } - else - { - WriteBufferFromOwnString ostr; - serializations[i]->serializeText(*columns[i], row_num, ostr, FormatSettings()); - payload_size += getLengthEncodedStringSize(ostr.str()); - serialized.push_back(std::move(ostr.str())); - } - } + payload_size += 1; + serialized.emplace_back("\xfb"); } - - size_t ResultSetRow::getPayloadSize() const + else { - return payload_size; + WriteBufferFromOwnString ostr; + serializations[i]->serializeText(*columns[i], row_num, ostr, FormatSettings()); + payload_size += getLengthEncodedStringSize(ostr.str()); + serialized.push_back(std::move(ostr.str())); } - - void ResultSetRow::writePayloadImpl(WriteBuffer & buffer) const - { - for (size_t i = 0; i < columns.size(); ++i) - { - if (columns[i]->isNullAt(row_num)) - buffer.write(serialized[i].data(), 1); - else - writeLengthEncodedString(serialized[i], buffer); - } - } - - void ComFieldList::readPayloadImpl(ReadBuffer & payload) - { - // Command byte has been already read from payload. - readNullTerminated(table, payload); - readStringUntilEOF(field_wildcard, payload); - } - - ColumnDefinition::ColumnDefinition() : character_set(0x00), column_length(0), column_type(MYSQL_TYPE_DECIMAL), flags(0x00) - { - } - - ColumnDefinition::ColumnDefinition( - String schema_, - String table_, - String org_table_, - String name_, - String org_name_, - uint16_t character_set_, - uint32_t column_length_, - ColumnType column_type_, - uint16_t flags_, - uint8_t decimals_, - bool with_defaults_) - : schema(std::move(schema_)) - , table(std::move(table_)) - , org_table(std::move(org_table_)) - , name(std::move(name_)) - , org_name(std::move(org_name_)) - , character_set(character_set_) - , column_length(column_length_) - , column_type(column_type_) - , flags(flags_) - , decimals(decimals_) - , is_comm_field_list_response(with_defaults_) - { - } - - ColumnDefinition::ColumnDefinition( - String name_, uint16_t character_set_, uint32_t column_length_, ColumnType column_type_, uint16_t flags_, uint8_t decimals_) - : ColumnDefinition("", "", "", std::move(name_), "", character_set_, column_length_, column_type_, flags_, decimals_) - { - } - - size_t ColumnDefinition::getPayloadSize() const - { - return 12 + getLengthEncodedStringSize("def") + getLengthEncodedStringSize(schema) + getLengthEncodedStringSize(table) - + getLengthEncodedStringSize(org_table) + getLengthEncodedStringSize(name) + getLengthEncodedStringSize(org_name) - + getLengthEncodedNumberSize(next_length) + is_comm_field_list_response; - } - - void ColumnDefinition::readPayloadImpl(ReadBuffer & payload) - { - String def; - readLengthEncodedString(def, payload); - assert(def == "def"); - readLengthEncodedString(schema, payload); - readLengthEncodedString(table, payload); - readLengthEncodedString(org_table, payload); - readLengthEncodedString(name, payload); - readLengthEncodedString(org_name, payload); - next_length = readLengthEncodedNumber(payload); - payload.readStrict(reinterpret_cast(&character_set), 2); - payload.readStrict(reinterpret_cast(&column_length), 4); - payload.readStrict(reinterpret_cast(&column_type), 1); - payload.readStrict(reinterpret_cast(&flags), 2); - payload.readStrict(reinterpret_cast(&decimals), 1); - payload.ignore(2); - } - - void ColumnDefinition::writePayloadImpl(WriteBuffer & buffer) const - { - writeLengthEncodedString(std::string("def"), buffer); /// always "def" - writeLengthEncodedString(schema, buffer); - writeLengthEncodedString(table, buffer); - writeLengthEncodedString(org_table, buffer); - writeLengthEncodedString(name, buffer); - writeLengthEncodedString(org_name, buffer); - writeLengthEncodedNumber(next_length, buffer); - buffer.write(reinterpret_cast(&character_set), 2); - buffer.write(reinterpret_cast(&column_length), 4); - buffer.write(reinterpret_cast(&column_type), 1); - buffer.write(reinterpret_cast(&flags), 2); - buffer.write(reinterpret_cast(&decimals), 1); - writeChar(0x0, 2, buffer); - if (is_comm_field_list_response) - { - /// We should write length encoded int with string size - /// followed by string with some "default values" (possibly it's column defaults). - /// But we just send NULL for simplicity. - writeChar(0xfb, buffer); - } - } - - ColumnDefinition getColumnDefinition(const String & column_name, const DataTypePtr & data_type) - { - ColumnType column_type; - CharacterSet charset = CharacterSet::binary; - int flags = 0; - uint8_t decimals = 0; - TypeIndex type_index = data_type->getTypeId(); - switch (type_index) - { - case TypeIndex::UInt8: - if (data_type->getName() == "Bool") - { - column_type = ColumnType::MYSQL_TYPE_BIT; - } - else - { - column_type = ColumnType::MYSQL_TYPE_TINY; - flags = ColumnDefinitionFlags::BINARY_FLAG | ColumnDefinitionFlags::UNSIGNED_FLAG; - } - break; - case TypeIndex::UInt16: - column_type = ColumnType::MYSQL_TYPE_SHORT; - flags = ColumnDefinitionFlags::BINARY_FLAG | ColumnDefinitionFlags::UNSIGNED_FLAG; - break; - case TypeIndex::UInt32: - column_type = ColumnType::MYSQL_TYPE_LONG; - flags = ColumnDefinitionFlags::BINARY_FLAG | ColumnDefinitionFlags::UNSIGNED_FLAG; - break; - case TypeIndex::UInt64: - column_type = ColumnType::MYSQL_TYPE_LONGLONG; - flags = ColumnDefinitionFlags::BINARY_FLAG | ColumnDefinitionFlags::UNSIGNED_FLAG; - break; - case TypeIndex::Int8: - column_type = ColumnType::MYSQL_TYPE_TINY; - flags = ColumnDefinitionFlags::BINARY_FLAG; - break; - case TypeIndex::Int16: - column_type = ColumnType::MYSQL_TYPE_SHORT; - flags = ColumnDefinitionFlags::BINARY_FLAG; - break; - case TypeIndex::Int32: - column_type = ColumnType::MYSQL_TYPE_LONG; - flags = ColumnDefinitionFlags::BINARY_FLAG; - break; - case TypeIndex::Int64: - column_type = ColumnType::MYSQL_TYPE_LONGLONG; - flags = ColumnDefinitionFlags::BINARY_FLAG; - break; - case TypeIndex::Float32: - column_type = ColumnType::MYSQL_TYPE_FLOAT; - flags = ColumnDefinitionFlags::BINARY_FLAG; - decimals = 31; - break; - case TypeIndex::Float64: - column_type = ColumnType::MYSQL_TYPE_DOUBLE; - flags = ColumnDefinitionFlags::BINARY_FLAG; - decimals = 31; - break; - case TypeIndex::Date: - case TypeIndex::Date32: - column_type = ColumnType::MYSQL_TYPE_DATE; - flags = ColumnDefinitionFlags::BINARY_FLAG; - break; - case TypeIndex::DateTime: - column_type = ColumnType::MYSQL_TYPE_DATETIME; - flags = ColumnDefinitionFlags::BINARY_FLAG; - break; - case TypeIndex::DateTime64: - column_type = ColumnType::MYSQL_TYPE_DATETIME; - flags = ColumnDefinitionFlags::BINARY_FLAG; - break; - case TypeIndex::Decimal32: - case TypeIndex::Decimal64: - column_type = ColumnType::MYSQL_TYPE_DECIMAL; - flags = ColumnDefinitionFlags::BINARY_FLAG; - break; - case TypeIndex::Decimal128: { - // MySQL Decimal has max 65 precision and 30 scale - const auto & type = assert_cast(*data_type); - if (type.getPrecision() > 65 || type.getScale() > 30) - { - column_type = ColumnType::MYSQL_TYPE_STRING; - charset = CharacterSet::utf8_general_ci; - } - else - { - column_type = ColumnType::MYSQL_TYPE_DECIMAL; - flags = ColumnDefinitionFlags::BINARY_FLAG; - } - break; - } - default: - column_type = ColumnType::MYSQL_TYPE_STRING; - charset = CharacterSet::utf8_general_ci; - break; - } - return ColumnDefinition(column_name, charset, 0, column_type, flags, decimals); - } - } +} + +size_t ResultSetRow::getPayloadSize() const +{ + return payload_size; +} + +void ResultSetRow::writePayloadImpl(WriteBuffer & buffer) const +{ + for (size_t i = 0; i < columns.size(); ++i) + { + if (columns[i]->isNullAt(row_num)) + buffer.write(serialized[i].data(), 1); + else + writeLengthEncodedString(serialized[i], buffer); + } +} + +void ComFieldList::readPayloadImpl(ReadBuffer & payload) +{ + // Command byte has been already read from payload. + readNullTerminated(table, payload); + readStringUntilEOF(field_wildcard, payload); +} + +ColumnDefinition::ColumnDefinition() : character_set(0x00), column_length(0), column_type(MYSQL_TYPE_DECIMAL), flags(0x00) +{ +} + +ColumnDefinition::ColumnDefinition( + String schema_, + String table_, + String org_table_, + String name_, + String org_name_, + uint16_t character_set_, + uint32_t column_length_, + ColumnType column_type_, + uint16_t flags_, + uint8_t decimals_, + bool with_defaults_) + : schema(std::move(schema_)) + , table(std::move(table_)) + , org_table(std::move(org_table_)) + , name(std::move(name_)) + , org_name(std::move(org_name_)) + , character_set(character_set_) + , column_length(column_length_) + , column_type(column_type_) + , flags(flags_) + , decimals(decimals_) + , is_comm_field_list_response(with_defaults_) +{ +} + +ColumnDefinition::ColumnDefinition( + String name_, uint16_t character_set_, uint32_t column_length_, ColumnType column_type_, uint16_t flags_, uint8_t decimals_) + : ColumnDefinition("", "", "", std::move(name_), "", character_set_, column_length_, column_type_, flags_, decimals_) +{ +} + +size_t ColumnDefinition::getPayloadSize() const +{ + return 12 + getLengthEncodedStringSize("def") + getLengthEncodedStringSize(schema) + getLengthEncodedStringSize(table) + + getLengthEncodedStringSize(org_table) + getLengthEncodedStringSize(name) + getLengthEncodedStringSize(org_name) + + getLengthEncodedNumberSize(next_length) + is_comm_field_list_response; +} + +void ColumnDefinition::readPayloadImpl(ReadBuffer & payload) +{ + String def; + readLengthEncodedString(def, payload); + assert(def == "def"); + readLengthEncodedString(schema, payload); + readLengthEncodedString(table, payload); + readLengthEncodedString(org_table, payload); + readLengthEncodedString(name, payload); + readLengthEncodedString(org_name, payload); + next_length = readLengthEncodedNumber(payload); + payload.readStrict(reinterpret_cast(&character_set), 2); + payload.readStrict(reinterpret_cast(&column_length), 4); + payload.readStrict(reinterpret_cast(&column_type), 1); + payload.readStrict(reinterpret_cast(&flags), 2); + payload.readStrict(reinterpret_cast(&decimals), 1); + payload.ignore(2); +} + +void ColumnDefinition::writePayloadImpl(WriteBuffer & buffer) const +{ + writeLengthEncodedString(std::string("def"), buffer); /// always "def" + writeLengthEncodedString(schema, buffer); + writeLengthEncodedString(table, buffer); + writeLengthEncodedString(org_table, buffer); + writeLengthEncodedString(name, buffer); + writeLengthEncodedString(org_name, buffer); + writeLengthEncodedNumber(next_length, buffer); + buffer.write(reinterpret_cast(&character_set), 2); + buffer.write(reinterpret_cast(&column_length), 4); + buffer.write(reinterpret_cast(&column_type), 1); + buffer.write(reinterpret_cast(&flags), 2); + buffer.write(reinterpret_cast(&decimals), 1); + writeChar(0x0, 2, buffer); + if (is_comm_field_list_response) + { + /// We should write length encoded int with string size + /// followed by string with some "default values" (possibly it's column defaults). + /// But we just send NULL for simplicity. + writeChar(0xfb, buffer); + } +} + +ColumnDefinition getColumnDefinition(const String & column_name, const DataTypePtr & data_type) +{ + ColumnType column_type; + CharacterSet charset = CharacterSet::binary; + int flags = 0; + uint8_t decimals = 0; + TypeIndex type_index = data_type->getTypeId(); + switch (type_index) + { + case TypeIndex::UInt8: + if (data_type->getName() == "Bool") + { + column_type = ColumnType::MYSQL_TYPE_BIT; + } + else + { + column_type = ColumnType::MYSQL_TYPE_TINY; + flags = ColumnDefinitionFlags::BINARY_FLAG | ColumnDefinitionFlags::UNSIGNED_FLAG; + } + break; + case TypeIndex::UInt16: + column_type = ColumnType::MYSQL_TYPE_SHORT; + flags = ColumnDefinitionFlags::BINARY_FLAG | ColumnDefinitionFlags::UNSIGNED_FLAG; + break; + case TypeIndex::UInt32: + column_type = ColumnType::MYSQL_TYPE_LONG; + flags = ColumnDefinitionFlags::BINARY_FLAG | ColumnDefinitionFlags::UNSIGNED_FLAG; + break; + case TypeIndex::UInt64: + column_type = ColumnType::MYSQL_TYPE_LONGLONG; + flags = ColumnDefinitionFlags::BINARY_FLAG | ColumnDefinitionFlags::UNSIGNED_FLAG; + break; + case TypeIndex::Int8: + column_type = ColumnType::MYSQL_TYPE_TINY; + flags = ColumnDefinitionFlags::BINARY_FLAG; + break; + case TypeIndex::Int16: + column_type = ColumnType::MYSQL_TYPE_SHORT; + flags = ColumnDefinitionFlags::BINARY_FLAG; + break; + case TypeIndex::Int32: + column_type = ColumnType::MYSQL_TYPE_LONG; + flags = ColumnDefinitionFlags::BINARY_FLAG; + break; + case TypeIndex::Int64: + column_type = ColumnType::MYSQL_TYPE_LONGLONG; + flags = ColumnDefinitionFlags::BINARY_FLAG; + break; + case TypeIndex::Float32: + column_type = ColumnType::MYSQL_TYPE_FLOAT; + flags = ColumnDefinitionFlags::BINARY_FLAG; + decimals = 31; + break; + case TypeIndex::Float64: + column_type = ColumnType::MYSQL_TYPE_DOUBLE; + flags = ColumnDefinitionFlags::BINARY_FLAG; + decimals = 31; + break; + case TypeIndex::Date: + case TypeIndex::Date32: + column_type = ColumnType::MYSQL_TYPE_DATE; + flags = ColumnDefinitionFlags::BINARY_FLAG; + break; + case TypeIndex::DateTime: + column_type = ColumnType::MYSQL_TYPE_DATETIME; + flags = ColumnDefinitionFlags::BINARY_FLAG; + break; + case TypeIndex::DateTime64: + column_type = ColumnType::MYSQL_TYPE_DATETIME; + flags = ColumnDefinitionFlags::BINARY_FLAG; + break; + case TypeIndex::Decimal32: + case TypeIndex::Decimal64: + column_type = ColumnType::MYSQL_TYPE_DECIMAL; + flags = ColumnDefinitionFlags::BINARY_FLAG; + break; + case TypeIndex::Decimal128: { + // MySQL Decimal has max 65 precision and 30 scale + const auto & type = assert_cast(*data_type); + if (type.getPrecision() > 65 || type.getScale() > 30) + { + column_type = ColumnType::MYSQL_TYPE_STRING; + charset = CharacterSet::utf8_general_ci; + } + else + { + column_type = ColumnType::MYSQL_TYPE_DECIMAL; + flags = ColumnDefinitionFlags::BINARY_FLAG; + } + break; + } + default: + column_type = ColumnType::MYSQL_TYPE_STRING; + charset = CharacterSet::utf8_general_ci; + break; + } + return ColumnDefinition(column_name, charset, 0, column_type, flags, decimals); +} + +} } diff --git a/src/DataTypes/DataTypesDecimal.cpp b/src/DataTypes/DataTypesDecimal.cpp index 2af216529e5..41373d1f5e7 100644 --- a/src/DataTypes/DataTypesDecimal.cpp +++ b/src/DataTypes/DataTypesDecimal.cpp @@ -81,14 +81,14 @@ SerializationPtr DataTypeDecimal::doGetDefaultSerialization() const static DataTypePtr create(const ASTPtr & arguments) { if (!arguments || arguments->children.size() != 2) - throw Exception( - ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH, "Decimal data type family must have exactly two arguments: precision and scale"); + throw Exception(ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH, + "Decimal data type family must have exactly two arguments: precision and scale"); const auto * precision = arguments->children[0]->as(); const auto * scale = arguments->children[1]->as(); - if (!precision || precision->value.getType() != Field::Types::UInt64 || !scale - || !(scale->value.getType() == Field::Types::Int64 || scale->value.getType() == Field::Types::UInt64)) + if (!precision || precision->value.getType() != Field::Types::UInt64 || + !scale || !(scale->value.getType() == Field::Types::Int64 || scale->value.getType() == Field::Types::UInt64)) throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Decimal data type family must have two numbers as its arguments"); UInt64 precision_value = precision->value.get(); @@ -101,15 +101,13 @@ template static DataTypePtr createExact(const ASTPtr & arguments) { if (!arguments || arguments->children.size() != 1) - throw Exception( - ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH, - "Decimal32 | Decimal64 | Decimal128 | Decimal256 data type family must have exactly one arguments: scale"); + throw Exception(ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH, + "Decimal32 | Decimal64 | Decimal128 | Decimal256 data type family must have exactly one arguments: scale"); const auto * scale_arg = arguments->children[0]->as(); if (!scale_arg || !(scale_arg->value.getType() == Field::Types::Int64 || scale_arg->value.getType() == Field::Types::UInt64)) - throw Exception( - ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, - "Decimal32 | Decimal64 | Decimal128 | Decimal256 data type family must have a one number as its argument"); + throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, + "Decimal32 | Decimal64 | Decimal128 | Decimal256 data type family must have a one number as its argument"); UInt64 precision = DecimalUtils::max_precision; UInt64 scale = scale_arg->value.get(); From bb0eff9669d03ddc1be52f613ccbe35f6de484d2 Mon Sep 17 00:00:00 2001 From: slvrtrn Date: Mon, 4 Sep 2023 21:15:26 +0200 Subject: [PATCH 120/433] Revert format changes --- src/Core/MySQL/PacketsGeneric.cpp | 8 +- src/Core/MySQL/PacketsProtocolText.cpp | 42 ++-- src/Core/MySQL/PacketsProtocolText.h | 224 +++++++++--------- src/DataTypes/DataTypesDecimal.cpp | 2 +- .../Serializations/SerializationDate32.cpp | 14 +- src/Formats/FormatSettings.h | 8 +- .../Formats/Impl/MySQLOutputFormat.cpp | 8 +- src/Server/MySQLHandler.cpp | 92 +++---- src/Server/MySQLHandler.h | 22 +- 9 files changed, 191 insertions(+), 229 deletions(-) diff --git a/src/Core/MySQL/PacketsGeneric.cpp b/src/Core/MySQL/PacketsGeneric.cpp index 23f197bcfe8..af80797d5c1 100644 --- a/src/Core/MySQL/PacketsGeneric.cpp +++ b/src/Core/MySQL/PacketsGeneric.cpp @@ -29,7 +29,7 @@ OKPacket::OKPacket( uint8_t header_, uint32_t capabilities_, uint64_t affected_rows_, uint32_t status_flags_, int16_t warnings_, String session_state_changes_, String info_) : header(header_), capabilities(capabilities_), affected_rows(affected_rows_), last_insert_id(0), warnings(warnings_), - status_flags(status_flags_), session_state_changes(std::move(session_state_changes_)), info(std::move(info_)) + status_flags(status_flags_), session_state_changes(std::move(session_state_changes_)), info(std::move(info_)) { } @@ -124,7 +124,8 @@ EOFPacket::EOFPacket() : warnings(0x00), status_flags(0x00) { } -EOFPacket::EOFPacket(int warnings_, int status_flags_) : warnings(warnings_), status_flags(status_flags_) +EOFPacket::EOFPacket(int warnings_, int status_flags_) + : warnings(warnings_), status_flags(status_flags_) { } @@ -196,7 +197,8 @@ void ERRPacket::writePayloadImpl(WriteBuffer & buffer) const buffer.write(error_message.data(), std::min(error_message.length(), MYSQL_ERRMSG_SIZE)); } -ResponsePacket::ResponsePacket(UInt32 server_capability_flags_) : ok(OKPacket(server_capability_flags_)) +ResponsePacket::ResponsePacket(UInt32 server_capability_flags_) + : ok(OKPacket(server_capability_flags_)) { } diff --git a/src/Core/MySQL/PacketsProtocolText.cpp b/src/Core/MySQL/PacketsProtocolText.cpp index f01f2ff9b61..ebfff1c2370 100644 --- a/src/Core/MySQL/PacketsProtocolText.cpp +++ b/src/Core/MySQL/PacketsProtocolText.cpp @@ -58,33 +58,17 @@ void ComFieldList::readPayloadImpl(ReadBuffer & payload) readStringUntilEOF(field_wildcard, payload); } -ColumnDefinition::ColumnDefinition() : character_set(0x00), column_length(0), column_type(MYSQL_TYPE_DECIMAL), flags(0x00) +ColumnDefinition::ColumnDefinition() + : character_set(0x00), column_length(0), column_type(MYSQL_TYPE_DECIMAL), flags(0x00) { } ColumnDefinition::ColumnDefinition( - String schema_, - String table_, - String org_table_, - String name_, - String org_name_, - uint16_t character_set_, - uint32_t column_length_, - ColumnType column_type_, - uint16_t flags_, - uint8_t decimals_, - bool with_defaults_) - : schema(std::move(schema_)) - , table(std::move(table_)) - , org_table(std::move(org_table_)) - , name(std::move(name_)) - , org_name(std::move(org_name_)) - , character_set(character_set_) - , column_length(column_length_) - , column_type(column_type_) - , flags(flags_) - , decimals(decimals_) - , is_comm_field_list_response(with_defaults_) + String schema_, String table_, String org_table_, String name_, String org_name_, uint16_t character_set_, uint32_t column_length_, + ColumnType column_type_, uint16_t flags_, uint8_t decimals_, bool with_defaults_) + : schema(std::move(schema_)), table(std::move(table_)), org_table(std::move(org_table_)), name(std::move(name_)), + org_name(std::move(org_name_)), character_set(character_set_), column_length(column_length_), column_type(column_type_), + flags(flags_), decimals(decimals_), is_comm_field_list_response(with_defaults_) { } @@ -96,9 +80,15 @@ ColumnDefinition::ColumnDefinition( size_t ColumnDefinition::getPayloadSize() const { - return 12 + getLengthEncodedStringSize("def") + getLengthEncodedStringSize(schema) + getLengthEncodedStringSize(table) - + getLengthEncodedStringSize(org_table) + getLengthEncodedStringSize(name) + getLengthEncodedStringSize(org_name) - + getLengthEncodedNumberSize(next_length) + is_comm_field_list_response; + return 12 + + getLengthEncodedStringSize("def") + + getLengthEncodedStringSize(schema) + + getLengthEncodedStringSize(table) + + getLengthEncodedStringSize(org_table) + + getLengthEncodedStringSize(name) + + getLengthEncodedStringSize(org_name) + + getLengthEncodedNumberSize(next_length) + + is_comm_field_list_response; } void ColumnDefinition::readPayloadImpl(ReadBuffer & payload) diff --git a/src/Core/MySQL/PacketsProtocolText.h b/src/Core/MySQL/PacketsProtocolText.h index d9ce913494b..1de9aabe9bb 100644 --- a/src/Core/MySQL/PacketsProtocolText.h +++ b/src/Core/MySQL/PacketsProtocolText.h @@ -12,135 +12,135 @@ namespace DB namespace MySQLProtocol { - namespace ProtocolText - { +namespace ProtocolText +{ - enum CharacterSet - { - utf8_general_ci = 33, - binary = 63 - }; +enum CharacterSet +{ + utf8_general_ci = 33, + binary = 63 +}; - // https://dev.mysql.com/doc/dev/mysql-server/latest/group__group__cs__column__definition__flags.html - enum ColumnDefinitionFlags - { - UNSIGNED_FLAG = 32, - BINARY_FLAG = 128 - }; +// https://dev.mysql.com/doc/dev/mysql-server/latest/group__group__cs__column__definition__flags.html +enum ColumnDefinitionFlags +{ + UNSIGNED_FLAG = 32, + BINARY_FLAG = 128 +}; - enum ColumnType - { - MYSQL_TYPE_DECIMAL = 0x00, - MYSQL_TYPE_TINY = 0x01, - MYSQL_TYPE_SHORT = 0x02, - MYSQL_TYPE_LONG = 0x03, - MYSQL_TYPE_FLOAT = 0x04, - MYSQL_TYPE_DOUBLE = 0x05, - MYSQL_TYPE_NULL = 0x06, - MYSQL_TYPE_TIMESTAMP = 0x07, - MYSQL_TYPE_LONGLONG = 0x08, - MYSQL_TYPE_INT24 = 0x09, - MYSQL_TYPE_DATE = 0x0a, - MYSQL_TYPE_TIME = 0x0b, - MYSQL_TYPE_DATETIME = 0x0c, - MYSQL_TYPE_YEAR = 0x0d, - MYSQL_TYPE_NEWDATE = 0x0e, - MYSQL_TYPE_VARCHAR = 0x0f, - MYSQL_TYPE_BIT = 0x10, - MYSQL_TYPE_TIMESTAMP2 = 0x11, - MYSQL_TYPE_DATETIME2 = 0x12, - MYSQL_TYPE_TIME2 = 0x13, - MYSQL_TYPE_JSON = 0xf5, - MYSQL_TYPE_NEWDECIMAL = 0xf6, - MYSQL_TYPE_ENUM = 0xf7, - MYSQL_TYPE_SET = 0xf8, - MYSQL_TYPE_TINY_BLOB = 0xf9, - MYSQL_TYPE_MEDIUM_BLOB = 0xfa, - MYSQL_TYPE_LONG_BLOB = 0xfb, - MYSQL_TYPE_BLOB = 0xfc, - MYSQL_TYPE_VAR_STRING = 0xfd, - MYSQL_TYPE_STRING = 0xfe, - MYSQL_TYPE_GEOMETRY = 0xff - }; +enum ColumnType +{ + MYSQL_TYPE_DECIMAL = 0x00, + MYSQL_TYPE_TINY = 0x01, + MYSQL_TYPE_SHORT = 0x02, + MYSQL_TYPE_LONG = 0x03, + MYSQL_TYPE_FLOAT = 0x04, + MYSQL_TYPE_DOUBLE = 0x05, + MYSQL_TYPE_NULL = 0x06, + MYSQL_TYPE_TIMESTAMP = 0x07, + MYSQL_TYPE_LONGLONG = 0x08, + MYSQL_TYPE_INT24 = 0x09, + MYSQL_TYPE_DATE = 0x0a, + MYSQL_TYPE_TIME = 0x0b, + MYSQL_TYPE_DATETIME = 0x0c, + MYSQL_TYPE_YEAR = 0x0d, + MYSQL_TYPE_NEWDATE = 0x0e, + MYSQL_TYPE_VARCHAR = 0x0f, + MYSQL_TYPE_BIT = 0x10, + MYSQL_TYPE_TIMESTAMP2 = 0x11, + MYSQL_TYPE_DATETIME2 = 0x12, + MYSQL_TYPE_TIME2 = 0x13, + MYSQL_TYPE_JSON = 0xf5, + MYSQL_TYPE_NEWDECIMAL = 0xf6, + MYSQL_TYPE_ENUM = 0xf7, + MYSQL_TYPE_SET = 0xf8, + MYSQL_TYPE_TINY_BLOB = 0xf9, + MYSQL_TYPE_MEDIUM_BLOB = 0xfa, + MYSQL_TYPE_LONG_BLOB = 0xfb, + MYSQL_TYPE_BLOB = 0xfc, + MYSQL_TYPE_VAR_STRING = 0xfd, + MYSQL_TYPE_STRING = 0xfe, + MYSQL_TYPE_GEOMETRY = 0xff +}; - class ResultSetRow : public IMySQLWritePacket - { - protected: - const Columns & columns; - int row_num; - size_t payload_size = 0; - std::vector serialized; +class ResultSetRow : public IMySQLWritePacket +{ +protected: + const Columns & columns; + int row_num; + size_t payload_size = 0; + std::vector serialized; - size_t getPayloadSize() const override; + size_t getPayloadSize() const override; - void writePayloadImpl(WriteBuffer & buffer) const override; + void writePayloadImpl(WriteBuffer & buffer) const override; - public: - ResultSetRow(const Serializations & serializations, const Columns & columns_, int row_num_); - }; +public: + ResultSetRow(const Serializations & serializations, const Columns & columns_, int row_num_); +}; - class ComFieldList : public LimitedReadPacket - { - public: - String table, field_wildcard; +class ComFieldList : public LimitedReadPacket +{ +public: + String table, field_wildcard; - void readPayloadImpl(ReadBuffer & payload) override; - }; + void readPayloadImpl(ReadBuffer & payload) override; +}; - class ColumnDefinition : public IMySQLWritePacket, public IMySQLReadPacket - { - public: - String schema; - String table; - String org_table; - String name; - String org_name; - size_t next_length = 0x0c; - uint16_t character_set; - uint32_t column_length; - ColumnType column_type; - uint16_t flags; - uint8_t decimals = 0x00; - /// https://dev.mysql.com/doc/internals/en/com-query-response.html#column-definition - /// There are extra fields in the packet for column defaults - bool is_comm_field_list_response = false; +class ColumnDefinition : public IMySQLWritePacket, public IMySQLReadPacket +{ +public: + String schema; + String table; + String org_table; + String name; + String org_name; + size_t next_length = 0x0c; + uint16_t character_set; + uint32_t column_length; + ColumnType column_type; + uint16_t flags; + uint8_t decimals = 0x00; + /// https://dev.mysql.com/doc/internals/en/com-query-response.html#column-definition + /// There are extra fields in the packet for column defaults + bool is_comm_field_list_response = false; - protected: - size_t getPayloadSize() const override; +protected: + size_t getPayloadSize() const override; - void readPayloadImpl(ReadBuffer & payload) override; + void readPayloadImpl(ReadBuffer & payload) override; - void writePayloadImpl(WriteBuffer & buffer) const override; + void writePayloadImpl(WriteBuffer & buffer) const override; - public: - ColumnDefinition(); +public: + ColumnDefinition(); - ColumnDefinition( - String schema_, - String table_, - String org_table_, - String name_, - String org_name_, - uint16_t character_set_, - uint32_t column_length_, - ColumnType column_type_, - uint16_t flags_, - uint8_t decimals_, - bool with_defaults_ = false); + ColumnDefinition( + String schema_, + String table_, + String org_table_, + String name_, + String org_name_, + uint16_t character_set_, + uint32_t column_length_, + ColumnType column_type_, + uint16_t flags_, + uint8_t decimals_, + bool with_defaults_ = false); - /// Should be used when column metadata (original name, table, original table, database) is unknown. - ColumnDefinition( - String name_, - uint16_t character_set_, - uint32_t column_length_, - ColumnType column_type_, - uint16_t flags_, - uint8_t decimals_); - }; + /// Should be used when column metadata (original name, table, original table, database) is unknown. + ColumnDefinition( + String name_, + uint16_t character_set_, + uint32_t column_length_, + ColumnType column_type_, + uint16_t flags_, + uint8_t decimals_); +}; - ColumnDefinition getColumnDefinition(const String & column_name, const DataTypePtr & data_type); - - } +ColumnDefinition getColumnDefinition(const String & column_name, const DataTypePtr & data_type); + +} } diff --git a/src/DataTypes/DataTypesDecimal.cpp b/src/DataTypes/DataTypesDecimal.cpp index 41373d1f5e7..e830363d9f1 100644 --- a/src/DataTypes/DataTypesDecimal.cpp +++ b/src/DataTypes/DataTypesDecimal.cpp @@ -1,13 +1,13 @@ #include #include +#include #include #include #include #include #include #include -#include #include diff --git a/src/DataTypes/Serializations/SerializationDate32.cpp b/src/DataTypes/Serializations/SerializationDate32.cpp index 68213a14afc..851710de839 100644 --- a/src/DataTypes/Serializations/SerializationDate32.cpp +++ b/src/DataTypes/Serializations/SerializationDate32.cpp @@ -28,14 +28,12 @@ void SerializationDate32::deserializeTextEscaped(IColumn & column, ReadBuffer & assert_cast(column).getData().push_back(x); } -void SerializationDate32::serializeTextEscaped( - const IColumn & column, size_t row_num, WriteBuffer & ostr, const FormatSettings & settings) const +void SerializationDate32::serializeTextEscaped(const IColumn & column, size_t row_num, WriteBuffer & ostr, const FormatSettings & settings) const { serializeText(column, row_num, ostr, settings); } -void SerializationDate32::serializeTextQuoted( - const IColumn & column, size_t row_num, WriteBuffer & ostr, const FormatSettings & settings) const +void SerializationDate32::serializeTextQuoted(const IColumn & column, size_t row_num, WriteBuffer & ostr, const FormatSettings & settings) const { writeChar('\'', ostr); serializeText(column, row_num, ostr, settings); @@ -48,11 +46,10 @@ void SerializationDate32::deserializeTextQuoted(IColumn & column, ReadBuffer & i assertChar('\'', istr); readDateText(x, istr, time_zone); assertChar('\'', istr); - assert_cast(column).getData().push_back(x); /// It's important to do this at the end - for exception safety. + assert_cast(column).getData().push_back(x); /// It's important to do this at the end - for exception safety. } -void SerializationDate32::serializeTextJSON( - const IColumn & column, size_t row_num, WriteBuffer & ostr, const FormatSettings & settings) const +void SerializationDate32::serializeTextJSON(const IColumn & column, size_t row_num, WriteBuffer & ostr, const FormatSettings & settings) const { writeChar('"', ostr); serializeText(column, row_num, ostr, settings); @@ -68,8 +65,7 @@ void SerializationDate32::deserializeTextJSON(IColumn & column, ReadBuffer & ist assert_cast(column).getData().push_back(x); } -void SerializationDate32::serializeTextCSV( - const IColumn & column, size_t row_num, WriteBuffer & ostr, const FormatSettings & settings) const +void SerializationDate32::serializeTextCSV(const IColumn & column, size_t row_num, WriteBuffer & ostr, const FormatSettings & settings) const { writeChar('"', ostr); serializeText(column, row_num, ostr, settings); diff --git a/src/Formats/FormatSettings.h b/src/Formats/FormatSettings.h index 986d79a0b7b..2f1d64e7bc6 100644 --- a/src/Formats/FormatSettings.h +++ b/src/Formats/FormatSettings.h @@ -1,7 +1,7 @@ #pragma once -#include #include +#include #include #include @@ -48,9 +48,9 @@ struct FormatSettings enum class DateTimeInputFormat { - Basic, /// Default format for fast parsing: YYYY-MM-DD hh:mm:ss (ISO-8601 without fractional part and timezone) or NNNNNNNNNN unix timestamp. - BestEffort, /// Use sophisticated rules to parse whatever possible. - BestEffortUS /// Use sophisticated rules to parse American style: mm/dd/yyyy + Basic, /// Default format for fast parsing: YYYY-MM-DD hh:mm:ss (ISO-8601 without fractional part and timezone) or NNNNNNNNNN unix timestamp. + BestEffort, /// Use sophisticated rules to parse whatever possible. + BestEffortUS /// Use sophisticated rules to parse American style: mm/dd/yyyy }; DateTimeInputFormat date_time_input_format = DateTimeInputFormat::Basic; diff --git a/src/Processors/Formats/Impl/MySQLOutputFormat.cpp b/src/Processors/Formats/Impl/MySQLOutputFormat.cpp index d25b68507fc..eaae8c6555c 100644 --- a/src/Processors/Formats/Impl/MySQLOutputFormat.cpp +++ b/src/Processors/Formats/Impl/MySQLOutputFormat.cpp @@ -17,7 +17,8 @@ using namespace MySQLProtocol::ProtocolText; using namespace MySQLProtocol::ProtocolBinary; MySQLOutputFormat::MySQLOutputFormat(WriteBuffer & out_, const Block & header_, const FormatSettings & settings_) - : IOutputFormat(header_, out_), client_capabilities(settings_.mysql_wire.client_capabilities) + : IOutputFormat(header_, out_) + , client_capabilities(settings_.mysql_wire.client_capabilities) { /// MySQlWire is a special format that is usually used as output format for MySQL protocol connections. /// In this case we have a correct `sequence_id` stored in `settings_.mysql_wire`. @@ -137,8 +138,9 @@ void registerOutputFormatMySQLWire(FormatFactory & factory) { factory.registerOutputFormat( "MySQLWire", - [](WriteBuffer & buf, const Block & sample, const FormatSettings & settings) - { return std::make_shared(buf, sample, settings); }); + [](WriteBuffer & buf, + const Block & sample, + const FormatSettings & settings) { return std::make_shared(buf, sample, settings); }); } } diff --git a/src/Server/MySQLHandler.cpp b/src/Server/MySQLHandler.cpp index 3715dfea9f7..79b645eafd2 100644 --- a/src/Server/MySQLHandler.cpp +++ b/src/Server/MySQLHandler.cpp @@ -67,7 +67,10 @@ static String killConnectionIdReplacementQuery(const String & query); static String selectLimitReplacementQuery(const String & query); MySQLHandler::MySQLHandler( - IServer & server_, TCPServer & tcp_server_, const Poco::Net::StreamSocket & socket_, bool ssl_enabled, uint32_t connection_id_) + IServer & server_, + TCPServer & tcp_server_, + const Poco::Net::StreamSocket & socket_, + bool ssl_enabled, uint32_t connection_id_) : Poco::Net::TCPServerConnection(socket_) , server(server_) , tcp_server(tcp_server_) @@ -75,8 +78,7 @@ MySQLHandler::MySQLHandler( , connection_id(connection_id_) , auth_plugin(new MySQLProtocol::Authentication::Native41()) { - server_capabilities = CLIENT_PROTOCOL_41 | CLIENT_SECURE_CONNECTION | CLIENT_PLUGIN_AUTH | CLIENT_PLUGIN_AUTH_LENENC_CLIENT_DATA - | CLIENT_CONNECT_WITH_DB | CLIENT_DEPRECATE_EOF; + server_capabilities = CLIENT_PROTOCOL_41 | CLIENT_SECURE_CONNECTION | CLIENT_PLUGIN_AUTH | CLIENT_PLUGIN_AUTH_LENENC_CLIENT_DATA | CLIENT_CONNECT_WITH_DB | CLIENT_DEPRECATE_EOF; if (ssl_enabled) server_capabilities |= CLIENT_SSL; @@ -102,13 +104,8 @@ void MySQLHandler::run() try { - Handshake handshake( - server_capabilities, - connection_id, - VERSION_STRING + String("-") + VERSION_NAME, - auth_plugin->getName(), - auth_plugin->getAuthPluginData(), - CharacterSet::utf8_general_ci); + Handshake handshake(server_capabilities, connection_id, VERSION_STRING + String("-") + VERSION_NAME, + auth_plugin->getName(), auth_plugin->getAuthPluginData(), CharacterSet::utf8_general_ci); packet_endpoint->sendPacket(handshake, true); LOG_TRACE(log, "Sent handshake"); @@ -118,10 +115,8 @@ void MySQLHandler::run() client_capabilities = handshake_response.capability_flags; max_packet_size = handshake_response.max_packet_size ? handshake_response.max_packet_size : MAX_PACKET_LENGTH; - LOG_TRACE( - log, - "Capabilities: {}, max_packet_size: {}, character_set: {}, user: {}, auth_response length: {}, database: {}, auth_plugin_name: " - "{}", + LOG_TRACE(log, + "Capabilities: {}, max_packet_size: {}, character_set: {}, user: {}, auth_response length: {}, database: {}, auth_plugin_name: {}", handshake_response.capability_flags, handshake_response.max_packet_size, static_cast(handshake_response.character_set), @@ -165,8 +160,8 @@ void MySQLHandler::run() // For commands which are executed without MemoryTracker. LimitReadBuffer limited_payload(payload, 10000, /* trow_exception */ true, /* exact_limit */ {}, "too long MySQL packet."); - LOG_DEBUG( - log, "Received command: {}. Connection id: {}.", static_cast(static_cast(command)), connection_id); + LOG_DEBUG(log, "Received command: {}. Connection id: {}.", + static_cast(static_cast(command)), connection_id); if (!tcp_server.isOpen()) return; @@ -232,15 +227,13 @@ void MySQLHandler::finishHandshake(MySQLProtocol::ConnectionPhase::HandshakeResp size_t pos = 0; /// Reads at least count and at most packet_size bytes. - auto read_bytes = [this, &buf, &pos, &packet_size](size_t count) -> void - { + auto read_bytes = [this, &buf, &pos, &packet_size](size_t count) -> void { while (pos < count) { int ret = socket().receiveBytes(buf + pos, static_cast(packet_size - pos)); if (ret == 0) { - throw Exception( - ErrorCodes::CANNOT_READ_ALL_DATA, "Cannot read all data. Bytes read: {}. Bytes expected: 3", std::to_string(pos)); + throw Exception(ErrorCodes::CANNOT_READ_ALL_DATA, "Cannot read all data. Bytes read: {}. Bytes expected: 3", std::to_string(pos)); } pos += ret; } @@ -279,8 +272,7 @@ void MySQLHandler::authenticate(const String & user_name, const String & auth_pl authPluginSSL(); } - std::optional auth_response - = auth_plugin_name == auth_plugin->getName() ? std::make_optional(initial_auth_response) : std::nullopt; + std::optional auth_response = auth_plugin_name == auth_plugin->getName() ? std::make_optional(initial_auth_response) : std::nullopt; auth_plugin->authenticate(user_name, *session, auth_response, packet_endpoint, secure_connection, socket().peerAddress()); } catch (const Exception & exc) @@ -312,17 +304,8 @@ void MySQLHandler::comFieldList(ReadBuffer & payload) for (const NameAndTypePair & column : metadata_snapshot->getColumns().getAll()) { ColumnDefinition column_definition( - database, - packet.table, - packet.table, - column.name, - column.name, - CharacterSet::binary, - 100, - ColumnType::MYSQL_TYPE_STRING, - 0, - 0, - true); + database, packet.table, packet.table, column.name, column.name, CharacterSet::binary, 100, ColumnType::MYSQL_TYPE_STRING, 0, 0, true + ); packet_endpoint->sendPacket(column_definition); } packet_endpoint->sendPacket(OKPacket(0xfe, client_capabilities, 0, 0, 0), true); @@ -367,16 +350,15 @@ void MySQLHandler::comQuery(ReadBuffer & payload, bool use_binary_protocol_resul query_context->setCurrentQueryId(fmt::format("mysql:{}:{}", connection_id, toString(UUIDHelpers::generateV4()))); CurrentThread::QueryScope query_scope{query_context}; - std::atomic affected_rows{0}; + std::atomic affected_rows {0}; auto prev = query_context->getProgressCallback(); - query_context->setProgressCallback( - [&, my_prev = prev](const Progress & progress) - { - if (my_prev) - my_prev(progress); + query_context->setProgressCallback([&, my_prev = prev](const Progress & progress) + { + if (my_prev) + my_prev(progress); - affected_rows += progress.written_rows; - }); + affected_rows += progress.written_rows; + }); FormatSettings format_settings; format_settings.mysql_wire.client_capabilities = client_capabilities; @@ -475,17 +457,13 @@ void MySQLHandler::comStmtClose(ReadBuffer & payload) void MySQLHandler::authPluginSSL() { - throw Exception( - ErrorCodes::SUPPORT_IS_DISABLED, - "ClickHouse was built without SSL support. Try specifying password using double SHA1 in users.xml."); + throw Exception(ErrorCodes::SUPPORT_IS_DISABLED, + "ClickHouse was built without SSL support. Try specifying password using double SHA1 in users.xml."); } void MySQLHandler::finishHandshakeSSL( - [[maybe_unused]] size_t packet_size, - [[maybe_unused]] char * buf, - [[maybe_unused]] size_t pos, - [[maybe_unused]] std::function read_bytes, - [[maybe_unused]] MySQLProtocol::ConnectionPhase::HandshakeResponse & packet) + [[maybe_unused]] size_t packet_size, [[maybe_unused]] char * buf, [[maybe_unused]] size_t pos, + [[maybe_unused]] std::function read_bytes, [[maybe_unused]] MySQLProtocol::ConnectionPhase::HandshakeResponse & packet) { throw Exception(ErrorCodes::SUPPORT_IS_DISABLED, "Client requested SSL, while it is disabled."); } @@ -499,9 +477,10 @@ MySQLHandlerSSL::MySQLHandlerSSL( uint32_t connection_id_, RSA & public_key_, RSA & private_key_) - : MySQLHandler(server_, tcp_server_, socket_, ssl_enabled, connection_id_), public_key(public_key_), private_key(private_key_) -{ -} + : MySQLHandler(server_, tcp_server_, socket_, ssl_enabled, connection_id_) + , public_key(public_key_) + , private_key(private_key_) +{} void MySQLHandlerSSL::authPluginSSL() { @@ -509,10 +488,7 @@ void MySQLHandlerSSL::authPluginSSL() } void MySQLHandlerSSL::finishHandshakeSSL( - size_t packet_size, - char * buf, - size_t pos, - std::function read_bytes, + size_t packet_size, char *buf, size_t pos, std::function read_bytes, MySQLProtocol::ConnectionPhase::HandshakeResponse & packet) { read_bytes(packet_size); /// Reading rest SSLRequest. @@ -542,8 +518,8 @@ static bool isFederatedServerSetupSetCommand(const String & query) "|(^(SET AUTOCOMMIT(.*)))" "|(^(SET sql_mode(.*)))" "|(^(SET @@(.*)))" - "|(^(SET SESSION TRANSACTION ISOLATION LEVEL(.*)))", - std::regex::icase}; + "|(^(SET SESSION TRANSACTION ISOLATION LEVEL(.*)))" + , std::regex::icase}; return 1 == std::regex_match(query, expr); } diff --git a/src/Server/MySQLHandler.h b/src/Server/MySQLHandler.h index a412b647ae2..240057e8ba7 100644 --- a/src/Server/MySQLHandler.h +++ b/src/Server/MySQLHandler.h @@ -19,7 +19,7 @@ namespace CurrentMetrics { -extern const Metric MySQLConnection; + extern const Metric MySQLConnection; } namespace DB @@ -32,7 +32,11 @@ class MySQLHandler : public Poco::Net::TCPServerConnection { public: MySQLHandler( - IServer & server_, TCPServer & tcp_server_, const Poco::Net::StreamSocket & socket_, bool ssl_enabled, uint32_t connection_id_); + IServer & server_, + TCPServer & tcp_server_, + const Poco::Net::StreamSocket & socket_, + bool ssl_enabled, + uint32_t connection_id_); void run() final; @@ -59,12 +63,7 @@ protected: void comStmtClose(ReadBuffer & payload); virtual void authPluginSSL(); - virtual void finishHandshakeSSL( - size_t packet_size, - char * buf, - size_t pos, - std::function read_bytes, - MySQLProtocol::ConnectionPhase::HandshakeResponse & packet); + virtual void finishHandshakeSSL(size_t packet_size, char * buf, size_t pos, std::function read_bytes, MySQLProtocol::ConnectionPhase::HandshakeResponse & packet); IServer & server; TCPServer & tcp_server; @@ -110,11 +109,8 @@ private: void authPluginSSL() override; void finishHandshakeSSL( - size_t packet_size, - char * buf, - size_t pos, - std::function read_bytes, - MySQLProtocol::ConnectionPhase::HandshakeResponse & packet) override; + size_t packet_size, char * buf, size_t pos, + std::function read_bytes, MySQLProtocol::ConnectionPhase::HandshakeResponse & packet) override; RSA & public_key; RSA & private_key; From 6e0a25436893fc0d7d401b08baf4da2d2c6ca7df Mon Sep 17 00:00:00 2001 From: slvrtrn Date: Mon, 4 Sep 2023 21:50:06 +0200 Subject: [PATCH 121/433] Update tests, update imports, remove requirements --- .../Formats/Impl/MySQLOutputFormat.cpp | 3 +- tests/integration/requirements.txt | 30 ------------------- .../prepared_statements.reference | 4 +-- .../prepared_statements_test.sql | 12 +++++++- 4 files changed, 14 insertions(+), 35 deletions(-) delete mode 100644 tests/integration/requirements.txt diff --git a/src/Processors/Formats/Impl/MySQLOutputFormat.cpp b/src/Processors/Formats/Impl/MySQLOutputFormat.cpp index eaae8c6555c..2f056daf8e4 100644 --- a/src/Processors/Formats/Impl/MySQLOutputFormat.cpp +++ b/src/Processors/Formats/Impl/MySQLOutputFormat.cpp @@ -1,3 +1,4 @@ +#include #include #include #include @@ -5,8 +6,6 @@ #include #include #include -#include -#include "Common/logger_useful.h" namespace DB { diff --git a/tests/integration/requirements.txt b/tests/integration/requirements.txt deleted file mode 100644 index 11efc43eded..00000000000 --- a/tests/integration/requirements.txt +++ /dev/null @@ -1,30 +0,0 @@ -PyMySQL -aerospike -avro -cassandra-driver -confluent-kafka -dicttoxml -docker -docker-compose -grpcio -grpcio-tools -kafka-python -kazoo -minio -lz4 -protobuf -psycopg2-binary -pymongo -pytz -pytest -pytest-timeout -redis -tzlocal==2.1 -urllib3 -requests-kerberos -dict2xml -hypothesis -pyhdfs -pika -meilisearch -nats-py diff --git a/tests/integration/test_mysql_protocol/prepared_statements.reference b/tests/integration/test_mysql_protocol/prepared_statements.reference index 402e64348ae..6ea4b218f56 100644 --- a/tests/integration/test_mysql_protocol/prepared_statements.reference +++ b/tests/integration/test_mysql_protocol/prepared_statements.reference @@ -74,8 +74,8 @@ m type is CHAR, value: {'qux':144,'text':255} Row #1 d type is DATE, value: 1970-01-01 d32 type is DATE, value: 1900-01-01 -dt type is TIMESTAMP, value: 1970-01-01 01:00:00.0 -dt64_3 type is TIMESTAMP, value: 1900-01-01 00:00:00.0 +dt type is TIMESTAMP, value: 1970-01-01 00:00:00.0 +dt64_3 type is TIMESTAMP, value: 1900-01-01 00:00:00.001 dt64_6 type is TIMESTAMP, value: 1900-01-01 00:00:00.000001 dt64_9 type is TIMESTAMP, value: 1900-01-01 00:00:00.0 Row #2 diff --git a/tests/integration/test_mysql_protocol/prepared_statements_test.sql b/tests/integration/test_mysql_protocol/prepared_statements_test.sql index a32ce9db1d5..0302fe5aca7 100644 --- a/tests/integration/test_mysql_protocol/prepared_statements_test.sql +++ b/tests/integration/test_mysql_protocol/prepared_statements_test.sql @@ -73,7 +73,7 @@ CREATE TABLE ps_date_types ( d Date, d32 Date32, - dt DateTime, + dt DateTime('UTC'), dt64_3 DateTime64(3, 'UTC'), dt64_6 DateTime64(6, 'UTC'), dt64_9 DateTime64(9, 'UTC') @@ -115,3 +115,13 @@ VALUES ('2022-04-13 03:17:45', '2022-04-13 03:17:45.00001', '2022-04-13 03:17:45.0000001', '2022-04-13 03:17:45.00000001'); + +CREATE TABLE ps_datetime_timezones +( + dt DateTime('Europe/Amsterdam'), + dt64_3 DateTime64(3, 'Asia/Shanghai') +) ENGINE MergeTree ORDER BY dt; + +INSERT INTO ps_datetime_timezones +VALUES ('2022-09-04 20:31:05', '2022-09-04 20:31:05.022'), + ('1970-01-01 00:00:00', '1969-12-31 16:00:00'); From 319a51c0b70579b73dff64b40580acd939a80ab0 Mon Sep 17 00:00:00 2001 From: slvrtrn Date: Mon, 4 Sep 2023 22:05:51 +0200 Subject: [PATCH 122/433] Address style checks --- src/Core/MySQL/PacketsPreparedStatements.cpp | 6 ----- src/Core/MySQL/PacketsPreparedStatements.h | 5 +--- src/Core/MySQL/PacketsProtocolBinary.cpp | 1 - tests/integration/test_mysql_protocol/test.py | 23 +++++++++++-------- 4 files changed, 15 insertions(+), 20 deletions(-) diff --git a/src/Core/MySQL/PacketsPreparedStatements.cpp b/src/Core/MySQL/PacketsPreparedStatements.cpp index eddded585d5..dcff294ebad 100644 --- a/src/Core/MySQL/PacketsPreparedStatements.cpp +++ b/src/Core/MySQL/PacketsPreparedStatements.cpp @@ -1,5 +1,4 @@ #include -#include #include #include @@ -25,11 +24,6 @@ namespace MySQLProtocol buffer.write(0x0); // RESULTSET_METADATA_NONE } - void PrepareStatementResponseOK::readPayloadImpl([[maybe_unused]] ReadBuffer & payload) - { - throw Exception(ErrorCodes::NOT_IMPLEMENTED, "PrepareStatementResponseOK::readPayloadImpl is not implemented"); - } - PrepareStatementResponseOK::PrepareStatementResponseOK( uint32_t statement_id_, uint16_t num_columns_, uint16_t num_params_, uint16_t warnings_count_) : statement_id(statement_id_), num_columns(num_columns_), num_params(num_params_), warnings_count(warnings_count_) diff --git a/src/Core/MySQL/PacketsPreparedStatements.h b/src/Core/MySQL/PacketsPreparedStatements.h index 7f180d4d142..5aca459fc3a 100644 --- a/src/Core/MySQL/PacketsPreparedStatements.h +++ b/src/Core/MySQL/PacketsPreparedStatements.h @@ -1,6 +1,5 @@ #pragma once -#include #include namespace DB @@ -10,7 +9,7 @@ namespace MySQLProtocol namespace PreparedStatements { // https://dev.mysql.com/doc/dev/mysql-server/latest/page_protocol_com_stmt_prepare.html#sect_protocol_com_stmt_prepare_response_ok - class PrepareStatementResponseOK : public IMySQLWritePacket, public IMySQLReadPacket + class PrepareStatementResponseOK : public IMySQLWritePacket { public: uint8_t status = 0x00; @@ -23,8 +22,6 @@ namespace MySQLProtocol protected: size_t getPayloadSize() const override; - void readPayloadImpl(ReadBuffer & payload) override; - void writePayloadImpl(WriteBuffer & buffer) const override; public: diff --git a/src/Core/MySQL/PacketsProtocolBinary.cpp b/src/Core/MySQL/PacketsProtocolBinary.cpp index ca206870a76..9e5e2993595 100644 --- a/src/Core/MySQL/PacketsProtocolBinary.cpp +++ b/src/Core/MySQL/PacketsProtocolBinary.cpp @@ -306,7 +306,6 @@ namespace MySQLProtocol } String timezone = getDateTimeTimezone(*data_type); - std::cout << "Timezone is " << timezone << std::endl; LocalDateTime ldt = LocalDateTime(components.whole, DateLUT::instance(timezone)); auto year = ldt.year(); auto month = ldt.month(); diff --git a/tests/integration/test_mysql_protocol/test.py b/tests/integration/test_mysql_protocol/test.py index 19a52da1158..0f8c1300cdc 100644 --- a/tests/integration/test_mysql_protocol/test.py +++ b/tests/integration/test_mysql_protocol/test.py @@ -810,7 +810,12 @@ def test_prepared_statements(started_cluster, java_container): reference = fp.read() with open(os.path.join(SCRIPT_DIR, "prepared_statements_test.sql")) as sql: - statements = list(filter(lambda s: s != '', map(lambda s: s.strip().replace('\n', ' '), sql.read().split(';')))) + statements = list( + filter( + lambda s: s != "", + map(lambda s: s.strip().replace("\n", " "), sql.read().split(";")), + ) + ) for statement in statements: node.query(statement, settings={"password": "123"}) @@ -860,14 +865,14 @@ def test_types(started_cluster): result = cursor.fetchall()[0] expected = [ - ("Int8_column", -(2 ** 7)), - ("UInt8_column", 2 ** 8 - 1), - ("Int16_column", -(2 ** 15)), - ("UInt16_column", 2 ** 16 - 1), - ("Int32_column", -(2 ** 31)), - ("UInt32_column", 2 ** 32 - 1), - ("Int64_column", -(2 ** 63)), - ("UInt64_column", 2 ** 64 - 1), + ("Int8_column", -(2**7)), + ("UInt8_column", 2**8 - 1), + ("Int16_column", -(2**15)), + ("UInt16_column", 2**16 - 1), + ("Int32_column", -(2**31)), + ("UInt32_column", 2**32 - 1), + ("Int64_column", -(2**63)), + ("UInt64_column", 2**64 - 1), ("String_column", "тест"), ("FixedString_column", "тест"), ("Float32_column", 1.5), From 9fa4850630223d9c6dd02c5b1204efb3903d338f Mon Sep 17 00:00:00 2001 From: slvrtrn Date: Mon, 4 Sep 2023 22:36:29 +0200 Subject: [PATCH 123/433] Try to resolve style check --- src/Interpreters/InterpreterShowColumnsQuery.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/Interpreters/InterpreterShowColumnsQuery.cpp b/src/Interpreters/InterpreterShowColumnsQuery.cpp index 922f9887a82..6c55ef0e6e9 100644 --- a/src/Interpreters/InterpreterShowColumnsQuery.cpp +++ b/src/Interpreters/InterpreterShowColumnsQuery.cpp @@ -37,7 +37,7 @@ String InterpreterShowColumnsQuery::getRewrittenQuery() SELECT name AS field, type AS type, - if(startsWith(type, 'Nullable'), 'YES', 'NO') AS null, + if(startsWith(type, 'Nullable'), 'YES', 'NO') AS `null`, trim(concatWithSeparator(' ', if (is_in_primary_key, 'PRI', ''), if (is_in_sorting_key, 'SOR', ''))) AS key, if (default_kind IN ('ALIAS', 'DEFAULT', 'MATERIALIZED'), default_expression, NULL) AS default, '' AS extra )"; From eddf914421a964bb3ebb590f75e08ca905a74178 Mon Sep 17 00:00:00 2001 From: slvrtrn Date: Mon, 4 Sep 2023 23:00:24 +0200 Subject: [PATCH 124/433] Clear prepared statements when it's too many --- src/Server/MySQLHandler.cpp | 2 ++ 1 file changed, 2 insertions(+) diff --git a/src/Server/MySQLHandler.cpp b/src/Server/MySQLHandler.cpp index 79b645eafd2..219e9c15ac2 100644 --- a/src/Server/MySQLHandler.cpp +++ b/src/Server/MySQLHandler.cpp @@ -389,6 +389,8 @@ void MySQLHandler::comStmtPrepare(DB::ReadBuffer & payload) if (prepared_statements_map.size() > 10000) /// Shouldn't happen in reality as COM_STMT_CLOSE cleans up the elements { LOG_ERROR(log, "Too many prepared statements"); + current_prepared_statement_id = 0; + prepared_statements_map.clear(); packet_endpoint->sendPacket(ERRPacket(), true); return; } From 8aeccd078bc32050ba5c173a299a6568387c4f65 Mon Sep 17 00:00:00 2001 From: Igor Nikonov Date: Mon, 4 Sep 2023 21:16:25 +0000 Subject: [PATCH 125/433] Try to catch timeout --- tests/ci/fast_test_check.py | 22 ++++++++++++++++------ 1 file changed, 16 insertions(+), 6 deletions(-) diff --git a/tests/ci/fast_test_check.py b/tests/ci/fast_test_check.py index 460e17acd37..57646dec4c5 100644 --- a/tests/ci/fast_test_check.py +++ b/tests/ci/fast_test_check.py @@ -32,6 +32,8 @@ from stopwatch import Stopwatch from tee_popen import TeePopen from upload_result_helper import upload_results from version_helper import get_version_from_repo +from report import TestResult +from subprocess import TimeoutExpired NAME = "Fast test" @@ -151,12 +153,17 @@ def main(): os.makedirs(logs_path) run_log_path = os.path.join(logs_path, "run.log") - with TeePopen(run_cmd, run_log_path, timeout=90 * 60) as process: - retcode = process.wait() - if retcode == 0: - logging.info("Run successfully") - else: - logging.info("Run failed") + expired_timeout = None + with TeePopen(run_cmd, run_log_path, timeout= 65) as process: + try: + retcode = process.wait() + if retcode == 0: + logging.info("Run successfully") + else: + logging.info("Run failed") + except TimeoutExpired as timeout_ex: + logging.info(f"Timeout expired for process execution: {run_cmd}") + expired_timeout = timeout_ex.timeout subprocess.check_call(f"sudo chown -R ubuntu:ubuntu {temp_path}", shell=True) @@ -188,6 +195,9 @@ def main(): else: state, description, test_results, additional_logs = process_results(output_path) + if (expired_timeout is not None): + test_results.append(TestResult(f"Timeout expired for process execution: {run_cmd}", "FAIL", expired_timeout)) + ch_helper = ClickHouseHelper() s3_path_prefix = os.path.join( get_release_or_pr(pr_info, get_version_from_repo())[0], From 0f5c49f6b5e74c98aa89ecacb1b176d719f1113e Mon Sep 17 00:00:00 2001 From: robot-clickhouse Date: Mon, 4 Sep 2023 21:23:58 +0000 Subject: [PATCH 126/433] Automatic style fix --- tests/ci/fast_test_check.py | 12 +++++++++--- 1 file changed, 9 insertions(+), 3 deletions(-) diff --git a/tests/ci/fast_test_check.py b/tests/ci/fast_test_check.py index 57646dec4c5..522c0f9f290 100644 --- a/tests/ci/fast_test_check.py +++ b/tests/ci/fast_test_check.py @@ -154,7 +154,7 @@ def main(): run_log_path = os.path.join(logs_path, "run.log") expired_timeout = None - with TeePopen(run_cmd, run_log_path, timeout= 65) as process: + with TeePopen(run_cmd, run_log_path, timeout=65) as process: try: retcode = process.wait() if retcode == 0: @@ -195,8 +195,14 @@ def main(): else: state, description, test_results, additional_logs = process_results(output_path) - if (expired_timeout is not None): - test_results.append(TestResult(f"Timeout expired for process execution: {run_cmd}", "FAIL", expired_timeout)) + if expired_timeout is not None: + test_results.append( + TestResult( + f"Timeout expired for process execution: {run_cmd}", + "FAIL", + expired_timeout, + ) + ) ch_helper = ClickHouseHelper() s3_path_prefix = os.path.join( From 767cac44dafda95d32800b5c5747db6585dea0af Mon Sep 17 00:00:00 2001 From: yariks5s Date: Tue, 5 Sep 2023 08:13:12 +0000 Subject: [PATCH 127/433] minor requests --- src/Functions/FunctionsConversion.h | 4 ++-- src/Functions/formatDateTime.cpp | 2 +- tests/queries/0_stateless/02863_non_const_timezone_check.sql | 4 ++-- 3 files changed, 5 insertions(+), 5 deletions(-) diff --git a/src/Functions/FunctionsConversion.h b/src/Functions/FunctionsConversion.h index f02f92efef9..4d4fe5f2d3b 100644 --- a/src/Functions/FunctionsConversion.h +++ b/src/Functions/FunctionsConversion.h @@ -950,7 +950,7 @@ struct ConvertImplgetDataAt(i).toString().empty()) time_zone = &DateLUT::instance(arguments[1].column.get()->getDataAt(i).toString()); else - throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Provided time zone must be non-empty and be a valid time zone"); + throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Provided time zone must be non-empty"); } bool is_ok = FormatImpl::template execute(vec_from[i], write_buffer, &type, time_zone); null_map->getData()[i] |= !is_ok; @@ -967,7 +967,7 @@ struct ConvertImplgetDataAt(i).toString().empty()) time_zone = &DateLUT::instance(arguments[1].column.get()->getDataAt(i).toString()); else - throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Provided time zone must be non-empty and be a valid time zone"); + throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Provided time zone must be non-empty"); } FormatImpl::template execute(vec_from[i], write_buffer, &type, time_zone); writeChar(0, write_buffer); diff --git a/src/Functions/formatDateTime.cpp b/src/Functions/formatDateTime.cpp index 39ae7fe735c..ac3013424fb 100644 --- a/src/Functions/formatDateTime.cpp +++ b/src/Functions/formatDateTime.cpp @@ -958,7 +958,7 @@ public: if (!arguments[2].column.get()->getDataAt(i).toString().empty()) time_zone = &DateLUT::instance(arguments[2].column.get()->getDataAt(i).toString()); else - throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Provided time zone must be non-empty and be a valid time zone"); + throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Provided time zone must be non-empty"); } if constexpr (std::is_same_v) { diff --git a/tests/queries/0_stateless/02863_non_const_timezone_check.sql b/tests/queries/0_stateless/02863_non_const_timezone_check.sql index cc54cea9ca0..4cb5457ae80 100644 --- a/tests/queries/0_stateless/02863_non_const_timezone_check.sql +++ b/tests/queries/0_stateless/02863_non_const_timezone_check.sql @@ -8,12 +8,12 @@ SELECT formatDateTime((SELECT date FROM Dates), '%H%i%S', number % 2 ? 'America/ SELECT formatDateTime((SELECT materialize(date) FROM Dates), '%H%i%S', number % 2 ? 'America/Los_Angeles' : 'Europe/Amsterdam') FROM numbers(5); -SELECT formatDateTime((SELECT materialize(date) FROM Dates), '%H%i%S', number % 2 ? '' : 'Europe/Amsterdam') FROM numbers(5); -- {serverError 43} +SELECT formatDateTime((SELECT materialize(date) FROM Dates), '%H%i%S', number % 2 ? '' : 'Europe/Amsterdam') FROM numbers(5); -- {serverError ILLEGAL_TYPE_OF_ARGUMENT} SELECT toString((SELECT date FROM Dates), number % 2 ? 'America/Los_Angeles' : 'Europe/Amsterdam') FROM numbers(5); SELECT toString((SELECT materialize(date) FROM Dates), number % 2 ? 'America/Los_Angeles' : 'Europe/Amsterdam') FROM numbers(5); -SELECT toString((SELECT materialize(date) FROM Dates), number % 2 ? 'America/Los_Angeles' : '') FROM numbers(5); -- {serverError 43} +SELECT toString((SELECT materialize(date) FROM Dates), number % 2 ? 'America/Los_Angeles' : '') FROM numbers(5); -- {serverError ILLEGAL_TYPE_OF_ARGUMENT} DROP TABLE Dates; From 3159560c6cc955e2e4b03e82101a1d12186ea450 Mon Sep 17 00:00:00 2001 From: Salvatore Mesoraca Date: Tue, 5 Sep 2023 18:24:53 +0200 Subject: [PATCH 128/433] Fix filesystem cache test A missing % in the query prevented it from producing any result at all. --- .../02241_filesystem_cache_on_write_operations.reference | 2 ++ .../0_stateless/02241_filesystem_cache_on_write_operations.sh | 2 +- 2 files changed, 3 insertions(+), 1 deletion(-) diff --git a/tests/queries/0_stateless/02241_filesystem_cache_on_write_operations.reference b/tests/queries/0_stateless/02241_filesystem_cache_on_write_operations.reference index c6099b96d31..157837983f7 100644 --- a/tests/queries/0_stateless/02241_filesystem_cache_on_write_operations.reference +++ b/tests/queries/0_stateless/02241_filesystem_cache_on_write_operations.reference @@ -26,6 +26,7 @@ state: DOWNLOADED 24 84045 32 167243 41 250541 +INSERT INTO test_02241 SELECT number, toString(number) FROM numbers(5000000) 0 5010500 18816 Using storage policy: local_cache @@ -56,5 +57,6 @@ state: DOWNLOADED 24 84045 32 167243 41 250541 +INSERT INTO test_02241 SELECT number, toString(number) FROM numbers(5000000) 0 5010500 18816 diff --git a/tests/queries/0_stateless/02241_filesystem_cache_on_write_operations.sh b/tests/queries/0_stateless/02241_filesystem_cache_on_write_operations.sh index 3a22ddfd7f5..96f61cf61e8 100755 --- a/tests/queries/0_stateless/02241_filesystem_cache_on_write_operations.sh +++ b/tests/queries/0_stateless/02241_filesystem_cache_on_write_operations.sh @@ -113,7 +113,7 @@ for STORAGE_POLICY in 's3_cache' 'local_cache'; do FROM system.query_log WHERE - query LIKE 'SELECT number, toString(number) FROM numbers(5000000)%' + query LIKE '%SELECT number, toString(number) FROM numbers(5000000)%' AND type = 'QueryFinish' AND current_database = currentDatabase() ORDER BY From 17cb6a8e3f0fa8c8f3c65834e968282aef2a343c Mon Sep 17 00:00:00 2001 From: Robert Schulze Date: Tue, 5 Sep 2023 18:34:33 +0000 Subject: [PATCH 129/433] SHOW COLUMNS: update docs, refresh expected results, add comment about bug --- docs/en/sql-reference/statements/show.md | 2 +- .../InterpreterShowColumnsQuery.cpp | 3 + .../0_stateless/02706_show_columns.reference | 68 +++++++++---------- 3 files changed, 38 insertions(+), 35 deletions(-) diff --git a/docs/en/sql-reference/statements/show.md b/docs/en/sql-reference/statements/show.md index 1c399d2072b..994c2bb24c7 100644 --- a/docs/en/sql-reference/statements/show.md +++ b/docs/en/sql-reference/statements/show.md @@ -208,7 +208,7 @@ The optional keyword `FULL` causes the output to include the collation, comment The statement produces a result table with the following structure: - field - The name of the column (String) - type - The column data type (String) -- null - If the column data type is Nullable (UInt8) +- null - `YES` if the column data type is Nullable, `NO` otherwise (UInt8) - key - `PRI` if the column is part of the primary key, `SOR` if the column is part of the sorting key, empty otherwise (String) - default - Default expression of the column if it is of type `ALIAS`, `DEFAULT`, or `MATERIALIZED`, otherwise `NULL`. (Nullable(String)) - extra - Additional information, currently unused (String) diff --git a/src/Interpreters/InterpreterShowColumnsQuery.cpp b/src/Interpreters/InterpreterShowColumnsQuery.cpp index 6c55ef0e6e9..ff1bab062d8 100644 --- a/src/Interpreters/InterpreterShowColumnsQuery.cpp +++ b/src/Interpreters/InterpreterShowColumnsQuery.cpp @@ -42,6 +42,9 @@ SELECT if (default_kind IN ('ALIAS', 'DEFAULT', 'MATERIALIZED'), default_expression, NULL) AS default, '' AS extra )"; + // Known issue: Field 'null' is wrong for types like 'LowCardinality(Nullable(String))'. Can't simply replace 'startsWith' by + // `hasSubsequence` as that would return `true` for non-nullable types such as `Tuple(Nullable(String), String)`... + // TODO Interpret query.extended. It is supposed to show internal/virtual columns. Need to fetch virtual column names, see // IStorage::getVirtuals(). We can't easily do that via SQL. diff --git a/tests/queries/0_stateless/02706_show_columns.reference b/tests/queries/0_stateless/02706_show_columns.reference index 9d7e8485e19..29e5329d63d 100644 --- a/tests/queries/0_stateless/02706_show_columns.reference +++ b/tests/queries/0_stateless/02706_show_columns.reference @@ -1,47 +1,47 @@ --- Aliases of SHOW COLUMNS -int32 Nullable(Int32) 1 \N -str String 0 SOR \N -uint64 UInt64 0 PRI SOR \N -int32 Nullable(Int32) 1 \N -str String 0 SOR \N -uint64 UInt64 0 PRI SOR \N +int32 Nullable(Int32) YES \N +str String NO SOR \N +uint64 UInt64 NO PRI SOR \N +int32 Nullable(Int32) YES \N +str String NO SOR \N +uint64 UInt64 NO PRI SOR \N --- EXTENDED -int32 Nullable(Int32) 1 \N -str String 0 SOR \N -uint64 UInt64 0 PRI SOR \N +int32 Nullable(Int32) YES \N +str String NO SOR \N +uint64 UInt64 NO PRI SOR \N --- FULL -int32 Nullable(Int32) 1 \N \N example comment -str String 0 SOR \N \N -uint64 UInt64 0 PRI SOR \N \N +int32 Nullable(Int32) YES \N \N example comment +str String NO SOR \N \N +uint64 UInt64 NO PRI SOR \N \N --- LIKE -int32 Nullable(Int32) 1 \N -uint64 UInt64 0 PRI SOR \N +int32 Nullable(Int32) YES \N +uint64 UInt64 NO PRI SOR \N --- NOT LIKE -str String 0 SOR \N +str String NO SOR \N --- ILIKE -int32 Nullable(Int32) 1 \N -uint64 UInt64 0 PRI SOR \N +int32 Nullable(Int32) YES \N +uint64 UInt64 NO PRI SOR \N --- NOT ILIKE -str String 0 SOR \N +str String NO SOR \N --- WHERE -int32 Nullable(Int32) 1 \N -uint64 UInt64 0 PRI SOR \N +int32 Nullable(Int32) YES \N +uint64 UInt64 NO PRI SOR \N --- LIMIT -int32 Nullable(Int32) 1 \N +int32 Nullable(Int32) YES \N --- Check with weird table names -c String 0 PRI SOR \N -c String 0 PRI SOR \N -c String 0 PRI SOR \N -c String 0 PRI SOR \N +c String NO PRI SOR \N +c String NO PRI SOR \N +c String NO PRI SOR \N +c String NO PRI SOR \N --- Original table -int32 Nullable(Int32) 1 \N -str String 0 SOR \N -uint64 UInt64 0 PRI SOR \N +int32 Nullable(Int32) YES \N +str String NO SOR \N +uint64 UInt64 NO PRI SOR \N --- Equally named table in other database -int32 Int32 0 \N -str String 0 \N -uint64 UInt64 0 PRI SOR \N +int32 Int32 NO \N +str String NO \N +uint64 UInt64 NO PRI SOR \N --- Short form -int32 Int32 0 \N -str String 0 \N -uint64 UInt64 0 PRI SOR \N +int32 Int32 NO \N +str String NO \N +uint64 UInt64 NO PRI SOR \N From 82b1c09150fb8dd521abb333446ffeef1b463efa Mon Sep 17 00:00:00 2001 From: Vitaly Baranov Date: Tue, 5 Sep 2023 20:46:13 +0200 Subject: [PATCH 130/433] Simplify coordination of UUIDs. --- src/Backups/RestoreCoordinationLocal.cpp | 5 ++--- src/Backups/RestoreCoordinationRemote.cpp | 13 ++++--------- src/Interpreters/InterpreterCreateQuery.cpp | 7 +++++++ src/Parsers/ASTCreateQuery.cpp | 5 ++++- src/Parsers/ASTCreateQuery.h | 2 +- 5 files changed, 18 insertions(+), 14 deletions(-) diff --git a/src/Backups/RestoreCoordinationLocal.cpp b/src/Backups/RestoreCoordinationLocal.cpp index 6e9a05e06c2..1bd2f8e2ed1 100644 --- a/src/Backups/RestoreCoordinationLocal.cpp +++ b/src/Backups/RestoreCoordinationLocal.cpp @@ -73,8 +73,8 @@ void RestoreCoordinationLocal::generateUUIDForTable(ASTCreateQuery & create_quer return; } - create_query.setUUID({}); - auto new_uuids = create_query.generateRandomUUID(); + auto new_uuids = create_query.generateRandomUUID(/* always_generate_new_uuid= */ true); + String new_query_str = serializeAST(create_query); { @@ -82,7 +82,6 @@ void RestoreCoordinationLocal::generateUUIDForTable(ASTCreateQuery & create_quer if (find_in_map()) return; create_query_uuids[query_str] = new_uuids; - create_query_uuids[new_query_str] = new_uuids; } } diff --git a/src/Backups/RestoreCoordinationRemote.cpp b/src/Backups/RestoreCoordinationRemote.cpp index 077921f4d32..28cdfd2c519 100644 --- a/src/Backups/RestoreCoordinationRemote.cpp +++ b/src/Backups/RestoreCoordinationRemote.cpp @@ -237,10 +237,7 @@ bool RestoreCoordinationRemote::acquireReplicatedSQLObjects(const String & loade void RestoreCoordinationRemote::generateUUIDForTable(ASTCreateQuery & create_query) { String query_str = serializeAST(create_query); - - create_query.setUUID({}); - String new_uuids_str = create_query.generateRandomUUID().toString(); - String new_query_str = serializeAST(create_query); + String new_uuids_str = create_query.generateRandomUUID(/* always_generate_new_uuid= */ true).toString(); auto holder = with_retries.createRetriesControlHolder("generateUUIDForTable"); holder.retries_ctl.retryLoop( @@ -249,25 +246,23 @@ void RestoreCoordinationRemote::generateUUIDForTable(ASTCreateQuery & create_que with_retries.renewZooKeeper(zk); String path = zookeeper_path + "/table_uuids/" + escapeForFileName(query_str); - String path2 = zookeeper_path + "/table_uuids/" + escapeForFileName(new_query_str); Coordination::Requests ops; ops.emplace_back(zkutil::makeCreateRequest(path, new_uuids_str, zkutil::CreateMode::Persistent)); - ops.emplace_back(zkutil::makeCreateRequest(path2, new_uuids_str, zkutil::CreateMode::Persistent)); Coordination::Responses responses; - Coordination::Error res = zk->tryMulti(ops, responses); + Coordination::Error res = zk->tryCreate(path, new_uuids_str, zkutil::CreateMode::Persistent); if (res == Coordination::Error::ZOK) return; - if ((res == Coordination::Error::ZNODEEXISTS) && (responses[0]->error == Coordination::Error::ZNODEEXISTS)) + if (res == Coordination::Error::ZNODEEXISTS) { create_query.setUUID(ASTCreateQuery::UUIDs::fromString(zk->get(path))); return; } - zkutil::KeeperMultiException::check(res, ops, responses); + zkutil::KeeperException::fromPath(res, path); }); } diff --git a/src/Interpreters/InterpreterCreateQuery.cpp b/src/Interpreters/InterpreterCreateQuery.cpp index 6da99a79f91..1cd34c2a0f6 100644 --- a/src/Interpreters/InterpreterCreateQuery.cpp +++ b/src/Interpreters/InterpreterCreateQuery.cpp @@ -1024,8 +1024,15 @@ void InterpreterCreateQuery::assertOrSetUUID(ASTCreateQuery & create, const Data bool is_on_cluster = getContext()->getClientInfo().query_kind == ClientInfo::QueryKind::SECONDARY_QUERY; bool has_uuid = create.uuid != UUIDHelpers::Nil || create.to_inner_uuid != UUIDHelpers::Nil; if (has_uuid && !is_on_cluster && !internal) + { + /// We don't show the following error message either + /// 1) if it's a secondary query (an initiator of a CREATE TABLE ON CLUSTER query + /// doesn't know the exact database engines on replicas and generates an UUID, and then the replicas are free to ignore that UUID); or + /// 2) if it's an internal query (for example RESTORE uses internal queries to create tables and it generates an UUID + /// before creating a table to be possibly ignored if the database engine doesn't need it). throw Exception(ErrorCodes::INCORRECT_QUERY, "{} UUID specified, but engine of database {} is not Atomic", kind, create.getDatabase()); + } /// The database doesn't support UUID so we'll ignore it. The UUID could be set here because of either /// a) the initiator of `ON CLUSTER` query generated it to ensure the same UUIDs are used on different hosts; or diff --git a/src/Parsers/ASTCreateQuery.cpp b/src/Parsers/ASTCreateQuery.cpp index bb0e4008491..1562586bd93 100644 --- a/src/Parsers/ASTCreateQuery.cpp +++ b/src/Parsers/ASTCreateQuery.cpp @@ -484,8 +484,11 @@ ASTCreateQuery::UUIDs ASTCreateQuery::UUIDs::fromString(const String & str) return res; } -ASTCreateQuery::UUIDs ASTCreateQuery::generateRandomUUID() +ASTCreateQuery::UUIDs ASTCreateQuery::generateRandomUUID(bool always_generate_new_uuid) { + if (always_generate_new_uuid) + setUUID({}); + if (uuid == UUIDHelpers::Nil) uuid = UUIDHelpers::generateV4(); diff --git a/src/Parsers/ASTCreateQuery.h b/src/Parsers/ASTCreateQuery.h index 25d910913a4..28f5e05802b 100644 --- a/src/Parsers/ASTCreateQuery.h +++ b/src/Parsers/ASTCreateQuery.h @@ -155,7 +155,7 @@ public: String toString() const; static UUIDs fromString(const String & str); }; - UUIDs generateRandomUUID(); + UUIDs generateRandomUUID(bool always_generate_new_uuid = false); void setUUID(const UUIDs & uuids); protected: From 767f2472c2cee3e788222c1ff14b20ec1b1f6845 Mon Sep 17 00:00:00 2001 From: Joe Lynch Date: Fri, 1 Sep 2023 14:04:01 +0200 Subject: [PATCH 131/433] Add elapsed_ns to HTTP headers X-ClickHouse-Progress and X-ClickHouse-Summary The TCP interface progress has this field. This is not a super accurate measure of server side query time, but trying to measure from the client is even worse. --- docs/en/interfaces/http.md | 18 ++++---- docs/ru/interfaces/http.md | 16 +++---- docs/zh/interfaces/http.md | 18 ++++---- src/IO/Progress.cpp | 8 +++- src/IO/Progress.h | 2 + .../WriteBufferFromHTTPServerResponse.cpp | 2 + ...0416_pocopatch_progress_in_http_headers.sh | 6 +-- .../0_stateless/02136_scalar_progress.sh | 2 +- .../02373_progress_contain_result.sh | 2 +- .../02423_insert_summary_behaviour.sh | 12 +++--- .../02457_insert_select_progress_http.sh | 2 +- .../02869_http_headers_elapsed_ns.reference | 2 + .../02869_http_headers_elapsed_ns.sh | 43 +++++++++++++++++++ 13 files changed, 94 insertions(+), 39 deletions(-) create mode 100644 tests/queries/0_stateless/02869_http_headers_elapsed_ns.reference create mode 100755 tests/queries/0_stateless/02869_http_headers_elapsed_ns.sh diff --git a/docs/en/interfaces/http.md b/docs/en/interfaces/http.md index 37821f0fee1..f2091dd0900 100644 --- a/docs/en/interfaces/http.md +++ b/docs/en/interfaces/http.md @@ -56,7 +56,7 @@ Connection: Close Content-Type: text/tab-separated-values; charset=UTF-8 X-ClickHouse-Server-Display-Name: clickhouse.ru-central1.internal X-ClickHouse-Query-Id: 5abe861c-239c-467f-b955-8a201abb8b7f -X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","peak_memory_usage":"0"} +X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","elapsed_ns":"662334,"peak_memory_usage":"0"} 1 ``` @@ -286,9 +286,9 @@ Similarly, you can use ClickHouse sessions in the HTTP protocol. To do this, you You can receive information about the progress of a query in `X-ClickHouse-Progress` response headers. To do this, enable [send_progress_in_http_headers](../operations/settings/settings.md#settings-send_progress_in_http_headers). Example of the header sequence: ``` text -X-ClickHouse-Progress: {"read_rows":"2752512","read_bytes":"240570816","total_rows_to_read":"8880128","peak_memory_usage":"4371480"} -X-ClickHouse-Progress: {"read_rows":"5439488","read_bytes":"482285394","total_rows_to_read":"8880128","peak_memory_usage":"13621616"} -X-ClickHouse-Progress: {"read_rows":"8783786","read_bytes":"819092887","total_rows_to_read":"8880128","peak_memory_usage":"23155600"} +X-ClickHouse-Progress: {"read_rows":"2752512","read_bytes":"240570816","total_rows_to_read":"8880128","elapsed_ns":"662334","peak_memory_usage":"4371480"} +X-ClickHouse-Progress: {"read_rows":"5439488","read_bytes":"482285394","total_rows_to_read":"8880128","elapsed_ns":"992334","peak_memory_usage":"13621616"} +X-ClickHouse-Progress: {"read_rows":"8783786","read_bytes":"819092887","total_rows_to_read":"8880128","elapsed_ns":"1232334","peak_memory_usage":"23155600"} ``` Possible header fields: @@ -416,7 +416,7 @@ $ curl -v 'http://localhost:8123/predefined_query' < X-ClickHouse-Format: Template < X-ClickHouse-Timezone: Asia/Shanghai < Keep-Alive: timeout=3 -< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","peak_memory_usage":"0"} +< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","elapsed_ns":"662334","peak_memory_usage":"0"} < # HELP "Query" "Number of executing queries" # TYPE "Query" counter @@ -581,7 +581,7 @@ $ curl -vv -H 'XXX:xxx' 'http://localhost:8123/hi' < Content-Type: text/html; charset=UTF-8 < Transfer-Encoding: chunked < Keep-Alive: timeout=3 -< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","peak_memory_usage":"0"} +< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","elapsed_ns":"662334","peak_memory_usage":"0"} < * Connection #0 to host localhost left intact Say Hi!% @@ -621,7 +621,7 @@ $ curl -v -H 'XXX:xxx' 'http://localhost:8123/get_config_static_handler' < Content-Type: text/plain; charset=UTF-8 < Transfer-Encoding: chunked < Keep-Alive: timeout=3 -< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","peak_memory_usage":"0"} +< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","elapsed_ns":"662334","peak_memory_usage":"0"} < * Connection #0 to host localhost left intact
% @@ -673,7 +673,7 @@ $ curl -vv -H 'XXX:xxx' 'http://localhost:8123/get_absolute_path_static_handler' < Content-Type: text/html; charset=UTF-8 < Transfer-Encoding: chunked < Keep-Alive: timeout=3 -< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","peak_memory_usage":"0"} +< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","elapsed_ns":"662334","peak_memory_usage":"0"} < Absolute Path File * Connection #0 to host localhost left intact @@ -692,7 +692,7 @@ $ curl -vv -H 'XXX:xxx' 'http://localhost:8123/get_relative_path_static_handler' < Content-Type: text/html; charset=UTF-8 < Transfer-Encoding: chunked < Keep-Alive: timeout=3 -< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","peak_memory_usage":"0"} +< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","elapsed_ns":"662334","peak_memory_usage":"0"} < Relative Path File * Connection #0 to host localhost left intact diff --git a/docs/ru/interfaces/http.md b/docs/ru/interfaces/http.md index ca7fbdc219b..ae3f100a6e8 100644 --- a/docs/ru/interfaces/http.md +++ b/docs/ru/interfaces/http.md @@ -50,7 +50,7 @@ Connection: Close Content-Type: text/tab-separated-values; charset=UTF-8 X-ClickHouse-Server-Display-Name: clickhouse.ru-central1.internal X-ClickHouse-Query-Id: 5abe861c-239c-467f-b955-8a201abb8b7f -X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","peak_memory_usage":"0"} +X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","elapsed_ns":"662334","peak_memory_usage":"0"} 1 ``` @@ -267,9 +267,9 @@ $ echo 'SELECT number FROM system.numbers LIMIT 10' | curl 'http://localhost:812 Прогресс выполнения запроса можно отслеживать с помощью заголовков ответа `X-ClickHouse-Progress`. Для этого включите [send_progress_in_http_headers](../operations/settings/settings.md#settings-send_progress_in_http_headers). Пример последовательности заголовков: ``` text -X-ClickHouse-Progress: {"read_rows":"2752512","read_bytes":"240570816","total_rows_to_read":"8880128","peak_memory_usage":"4371480"} -X-ClickHouse-Progress: {"read_rows":"5439488","read_bytes":"482285394","total_rows_to_read":"8880128","peak_memory_usage":"13621616"} -X-ClickHouse-Progress: {"read_rows":"8783786","read_bytes":"819092887","total_rows_to_read":"8880128","peak_memory_usage":"23155600"} +X-ClickHouse-Progress: {"read_rows":"2752512","read_bytes":"240570816","total_rows_to_read":"8880128","elapsed_ns":"662334","peak_memory_usage":"4371480"} +X-ClickHouse-Progress: {"read_rows":"5439488","read_bytes":"482285394","total_rows_to_read":"8880128","elapsed_ns":"992334","peak_memory_usage":"13621616"} +X-ClickHouse-Progress: {"read_rows":"8783786","read_bytes":"819092887","total_rows_to_read":"8880128","elapsed_ns":"1232334","peak_memory_usage":"23155600"} ``` Возможные поля заголовка: @@ -530,7 +530,7 @@ $ curl -vv -H 'XXX:xxx' 'http://localhost:8123/hi' < Content-Type: text/html; charset=UTF-8 < Transfer-Encoding: chunked < Keep-Alive: timeout=3 -< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","peak_memory_usage":"0"} +< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","elapsed_ns":"662334","peak_memory_usage":"0"} < * Connection #0 to host localhost left intact Say Hi!% @@ -570,7 +570,7 @@ $ curl -v -H 'XXX:xxx' 'http://localhost:8123/get_config_static_handler' < Content-Type: text/plain; charset=UTF-8 < Transfer-Encoding: chunked < Keep-Alive: timeout=3 -< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","peak_memory_usage":"0"} +< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","elapsed_ns":"662334","peak_memory_usage":"0"} < * Connection #0 to host localhost left intact
% @@ -622,7 +622,7 @@ $ curl -vv -H 'XXX:xxx' 'http://localhost:8123/get_absolute_path_static_handler' < Content-Type: text/html; charset=UTF-8 < Transfer-Encoding: chunked < Keep-Alive: timeout=3 -< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","peak_memory_usage":"0"} +< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","elapsed_ns":"662334","peak_memory_usage":"0"} < Absolute Path File * Connection #0 to host localhost left intact @@ -641,7 +641,7 @@ $ curl -vv -H 'XXX:xxx' 'http://localhost:8123/get_relative_path_static_handler' < Content-Type: text/html; charset=UTF-8 < Transfer-Encoding: chunked < Keep-Alive: timeout=3 -< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","peak_memory_usage":"0"} +< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","elapsed_ns":"662334","peak_memory_usage":"0"} < Relative Path File * Connection #0 to host localhost left intact diff --git a/docs/zh/interfaces/http.md b/docs/zh/interfaces/http.md index f84768beccc..b1501cbb5d6 100644 --- a/docs/zh/interfaces/http.md +++ b/docs/zh/interfaces/http.md @@ -53,7 +53,7 @@ Connection: Close Content-Type: text/tab-separated-values; charset=UTF-8 X-ClickHouse-Server-Display-Name: clickhouse.ru-central1.internal X-ClickHouse-Query-Id: 5abe861c-239c-467f-b955-8a201abb8b7f -X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","peak_memory_usage":"0"} +X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","elapsed_ns":"662334","peak_memory_usage":"0"} 1 ``` @@ -262,9 +262,9 @@ $ echo 'SELECT number FROM system.numbers LIMIT 10' | curl 'http://localhost:812 您可以在`X-ClickHouse-Progress`响应头中收到查询进度的信息。为此，启用[Http Header携带进度](../operations/settings/settings.md#settings-send_progress_in_http_headers)。示例： ``` text -X-ClickHouse-Progress: {"read_rows":"2752512","read_bytes":"240570816","total_rows_to_read":"8880128","peak_memory_usage":"4371480"} -X-ClickHouse-Progress: {"read_rows":"5439488","read_bytes":"482285394","total_rows_to_read":"8880128","peak_memory_usage":"13621616"} -X-ClickHouse-Progress: {"read_rows":"8783786","read_bytes":"819092887","total_rows_to_read":"8880128","peak_memory_usage":"23155600"} +X-ClickHouse-Progress: {"read_rows":"2752512","read_bytes":"240570816","total_rows_to_read":"8880128","elapsed_ns":"662334","peak_memory_usage":"4371480"} +X-ClickHouse-Progress: {"read_rows":"5439488","read_bytes":"482285394","total_rows_to_read":"8880128","elapsed_ns":"992334","peak_memory_usage":"13621616"} +X-ClickHouse-Progress: {"read_rows":"8783786","read_bytes":"819092887","total_rows_to_read":"8880128","elapsed_ns":"1232334","peak_memory_usage":"23155600"} ``` 显示字段信息: @@ -363,7 +363,7 @@ $ curl -v 'http://localhost:8123/predefined_query' < X-ClickHouse-Format: Template < X-ClickHouse-Timezone: Asia/Shanghai < Keep-Alive: timeout=3 -< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","peak_memory_usage":"0"} +< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","elapsed_ns":"662334","peak_memory_usage":"0"} < # HELP "Query" "Number of executing queries" # TYPE "Query" counter @@ -521,7 +521,7 @@ $ curl -vv -H 'XXX:xxx' 'http://localhost:8123/hi' < Content-Type: text/html; charset=UTF-8 < Transfer-Encoding: chunked < Keep-Alive: timeout=3 -< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","peak_memory_usage":"0"} +< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","elapsed_ns":"662334","peak_memory_usage":"0"} < * Connection #0 to host localhost left intact Say Hi!% @@ -561,7 +561,7 @@ $ curl -v -H 'XXX:xxx' 'http://localhost:8123/get_config_static_handler' < Content-Type: text/plain; charset=UTF-8 < Transfer-Encoding: chunked < Keep-Alive: timeout=3 -< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","peak_memory_usage":"0"} +< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","elapsed_ns":"662334","peak_memory_usage":"0"} < * Connection #0 to host localhost left intact
% @@ -613,7 +613,7 @@ $ curl -vv -H 'XXX:xxx' 'http://localhost:8123/get_absolute_path_static_handler' < Content-Type: text/html; charset=UTF-8 < Transfer-Encoding: chunked < Keep-Alive: timeout=3 -< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","peak_memory_usage":"0"} +< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","elapsed_ns":"662334","peak_memory_usage":"0"} < Absolute Path File * Connection #0 to host localhost left intact @@ -632,7 +632,7 @@ $ curl -vv -H 'XXX:xxx' 'http://localhost:8123/get_relative_path_static_handler' < Content-Type: text/html; charset=UTF-8 < Transfer-Encoding: chunked < Keep-Alive: timeout=3 -< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","peak_memory_usage":"0"} +< X-ClickHouse-Summary: {"read_rows":"0","read_bytes":"0","written_rows":"0","written_bytes":"0","total_rows_to_read":"0","elapsed_ns":"662334","peak_memory_usage":"0"} < Relative Path File * Connection #0 to host localhost left intact diff --git a/src/IO/Progress.cpp b/src/IO/Progress.cpp index 620d2f0f762..56fb7fe317e 100644 --- a/src/IO/Progress.cpp +++ b/src/IO/Progress.cpp @@ -90,7 +90,8 @@ void ProgressValues::writeJSON(WriteBuffer & out, bool add_braces) const writeText(result_rows, out); writeCString("\",\"result_bytes\":\"", out); writeText(result_bytes, out); - writeCString("\"", out); + writeCString("\",\"elapsed_ns\":\"", out); + writeText(elapsed_ns, out); if (add_braces) writeCString("}", out); } @@ -239,4 +240,9 @@ void Progress::writeJSON(WriteBuffer & out, bool add_braces) const getValues().writeJSON(out, add_braces); } +void Progress::incrementElapsedNs(UInt64 elapsed_ns_) +{ + elapsed_ns.fetch_add(elapsed_ns_, std::memory_order_relaxed); +} + } diff --git a/src/IO/Progress.h b/src/IO/Progress.h index 936ed5ea342..288598cae78 100644 --- a/src/IO/Progress.h +++ b/src/IO/Progress.h @@ -124,6 +124,8 @@ struct Progress /// Each value separately is changed atomically (but not whole object). bool incrementPiecewiseAtomically(const Progress & rhs); + void incrementElapsedNs(UInt64 elapsed_ns_); + void reset(); ProgressValues getValues() const; diff --git a/src/Server/HTTP/WriteBufferFromHTTPServerResponse.cpp b/src/Server/HTTP/WriteBufferFromHTTPServerResponse.cpp index 9da24cfd9cb..62f492fa0a9 100644 --- a/src/Server/HTTP/WriteBufferFromHTTPServerResponse.cpp +++ b/src/Server/HTTP/WriteBufferFromHTTPServerResponse.cpp @@ -48,6 +48,7 @@ void WriteBufferFromHTTPServerResponse::writeHeaderProgressImpl(const char * hea void WriteBufferFromHTTPServerResponse::writeHeaderSummary() { + accumulated_progress.incrementElapsedNs(progress_watch.elapsed()); writeHeaderProgressImpl("X-ClickHouse-Summary: "); } @@ -164,6 +165,7 @@ void WriteBufferFromHTTPServerResponse::onProgress(const Progress & progress, In peak_memory_usage = peak_memory_usage_; if (send_progress && progress_watch.elapsed() >= send_progress_interval_ms * 1000000) { + accumulated_progress.incrementElapsedNs(progress_watch.elapsed()); progress_watch.restart(); /// Send all common headers before our special progress headers. diff --git a/tests/queries/0_stateless/00416_pocopatch_progress_in_http_headers.sh b/tests/queries/0_stateless/00416_pocopatch_progress_in_http_headers.sh index e77aa1bf250..6ec88fae351 100755 --- a/tests/queries/0_stateless/00416_pocopatch_progress_in_http_headers.sh +++ b/tests/queries/0_stateless/00416_pocopatch_progress_in_http_headers.sh @@ -10,7 +10,7 @@ result="" lines_expected=4 counter=0 while [ $counter -lt $RETRIES ] && [ "$(echo "$result" | wc -l)" != "$lines_expected" ]; do - result=$(${CLICKHOUSE_CURL} -vsS "${CLICKHOUSE_URL}&max_block_size=5&send_progress_in_http_headers=1&http_headers_progress_interval_ms=0" -d 'SELECT max(number) FROM numbers(10)' 2>&1 | grep -E 'Content-Encoding|X-ClickHouse-Progress|^[0-9]' | sed 's/,\"peak_mem[^}]*//') + result=$(${CLICKHOUSE_CURL} -vsS "${CLICKHOUSE_URL}&max_block_size=5&send_progress_in_http_headers=1&http_headers_progress_interval_ms=0" -d 'SELECT max(number) FROM numbers(10)' 2>&1 | grep -E 'Content-Encoding|X-ClickHouse-Progress|^[0-9]' | sed 's/,\"elapsed_ns[^}]*//') let counter=counter+1 done echo "$result" @@ -19,7 +19,7 @@ result="" lines_expected=12 counter=0 while [ $counter -lt $RETRIES ] && [ "$(echo "$result" | wc -l)" != "$lines_expected" ]; do - result=$(${CLICKHOUSE_CURL} -vsS "${CLICKHOUSE_URL}&max_block_size=1&send_progress_in_http_headers=1&http_headers_progress_interval_ms=0&output_format_parallel_formatting=0" -d 'SELECT number FROM numbers(10)' 2>&1 | grep -E 'Content-Encoding|X-ClickHouse-Progress|^[0-9]'| sed 's/,\"peak_mem[^}]*//') + result=$(${CLICKHOUSE_CURL} -vsS "${CLICKHOUSE_URL}&max_block_size=1&send_progress_in_http_headers=1&http_headers_progress_interval_ms=0&output_format_parallel_formatting=0" -d 'SELECT number FROM numbers(10)' 2>&1 | grep -E 'Content-Encoding|X-ClickHouse-Progress|^[0-9]'| sed 's/,\"elapsed_ns[^}]*//') let counter=counter+1 done echo "$result" @@ -46,7 +46,7 @@ ${CLICKHOUSE_CURL} -vsS "${CLICKHOUSE_URL}" -H 'Accept-Encoding: gzip' -d 'CREAT result="" counter=0 while [ $counter -lt $RETRIES ] && [ -z "$result" ]; do - result=$(${CLICKHOUSE_CURL} -vsS "${CLICKHOUSE_URL}&max_block_size=1&http_headers_progress_interval_ms=0&send_progress_in_http_headers=1" -d 'INSERT INTO insert_number_query (record) SELECT number FROM system.numbers LIMIT 10' 2>&1 | grep -E 'Content-Encoding|X-ClickHouse-Summary|^[0-9]' | sed 's/,\"peak_mem[^}]*//') + result=$(${CLICKHOUSE_CURL} -vsS "${CLICKHOUSE_URL}&max_block_size=1&http_headers_progress_interval_ms=0&send_progress_in_http_headers=1" -d 'INSERT INTO insert_number_query (record) SELECT number FROM system.numbers LIMIT 10' 2>&1 | grep -E 'Content-Encoding|X-ClickHouse-Summary|^[0-9]' | sed 's/,\"elapsed_ns[^}]*//') let counter=counter+1 done echo "$result" diff --git a/tests/queries/0_stateless/02136_scalar_progress.sh b/tests/queries/0_stateless/02136_scalar_progress.sh index 517fe19ecd4..f2016b04f82 100755 --- a/tests/queries/0_stateless/02136_scalar_progress.sh +++ b/tests/queries/0_stateless/02136_scalar_progress.sh @@ -4,4 +4,4 @@ CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) # shellcheck source=../shell_config.sh . "$CURDIR"/../shell_config.sh -$CLICKHOUSE_CURL -sS "${CLICKHOUSE_URL}&wait_end_of_query=1&send_progress_in_http_headers=1&http_headers_progress_interval_ms=0" -d "SELECT (SELECT max(number), count(number) FROM numbers(100000) settings max_block_size=65505);" -v 2>&1 | grep -E "X-ClickHouse-Summary|X-ClickHouse-Progress" | sed 's/,\"peak_mem[^}]*//' +$CLICKHOUSE_CURL -sS "${CLICKHOUSE_URL}&wait_end_of_query=1&send_progress_in_http_headers=1&http_headers_progress_interval_ms=0" -d "SELECT (SELECT max(number), count(number) FROM numbers(100000) settings max_block_size=65505);" -v 2>&1 | grep -E "X-ClickHouse-Summary|X-ClickHouse-Progress" | sed 's/,\"elapsed_ns[^}]*//' diff --git a/tests/queries/0_stateless/02373_progress_contain_result.sh b/tests/queries/0_stateless/02373_progress_contain_result.sh index c84af0ee269..c87a5ec7615 100755 --- a/tests/queries/0_stateless/02373_progress_contain_result.sh +++ b/tests/queries/0_stateless/02373_progress_contain_result.sh @@ -6,4 +6,4 @@ CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) echo 'SELECT 1 FROM numbers(100)' | ${CLICKHOUSE_CURL_COMMAND} -v "${CLICKHOUSE_URL}&wait_end_of_query=1&send_progress_in_http_headers=0" --data-binary @- 2>&1 | - grep 'X-ClickHouse-Summary' | sed 's/,\"peak_mem[^}]*//' + grep 'X-ClickHouse-Summary' | sed 's/,\"elapsed_ns[^}]*//' diff --git a/tests/queries/0_stateless/02423_insert_summary_behaviour.sh b/tests/queries/0_stateless/02423_insert_summary_behaviour.sh index a0b3c519806..b184d9ccf47 100755 --- a/tests/queries/0_stateless/02423_insert_summary_behaviour.sh +++ b/tests/queries/0_stateless/02423_insert_summary_behaviour.sh @@ -11,11 +11,11 @@ $CLICKHOUSE_CLIENT -q "CREATE MATERIALIZED VIEW floats_to_target TO target_1 AS $CLICKHOUSE_CLIENT -q "CREATE MATERIALIZED VIEW floats_to_target_2 TO target_2 AS SELECT * FROM floats, numbers(2) n" echo "No materialized views" -${CLICKHOUSE_CURL} "${CLICKHOUSE_URL}&wait_end_of_query=1&query=INSERT+INTO+target_1" -d "VALUES(1.0)" -v 2>&1 | grep 'X-ClickHouse-Summary' | sed 's/,\"peak_mem[^}]*//' -$CLICKHOUSE_LOCAL -q "SELECT number::Float64 AS v FROM numbers(10)" --format Native | ${CLICKHOUSE_CURL} -sS "${CLICKHOUSE_URL}&wait_end_of_query=1&query=INSERT+INTO+target_1+FORMAT+Native" --data-binary @- -v 2>&1 | grep 'X-ClickHouse-Summary' | sed 's/,\"peak_mem[^}]*//' -$CLICKHOUSE_LOCAL -q "SELECT number::Float64 AS v FROM numbers(10)" --format RowBinary | ${CLICKHOUSE_CURL} -sS "${CLICKHOUSE_URL}&wait_end_of_query=1&query=INSERT+INTO+target_1+FORMAT+RowBinary" --data-binary @- -v 2>&1 | grep 'X-ClickHouse-Summary' | sed 's/,\"peak_mem[^}]*//' +${CLICKHOUSE_CURL} "${CLICKHOUSE_URL}&wait_end_of_query=1&query=INSERT+INTO+target_1" -d "VALUES(1.0)" -v 2>&1 | grep 'X-ClickHouse-Summary' | sed 's/,\"elapsed_ns[^}]*//' +$CLICKHOUSE_LOCAL -q "SELECT number::Float64 AS v FROM numbers(10)" --format Native | ${CLICKHOUSE_CURL} -sS "${CLICKHOUSE_URL}&wait_end_of_query=1&query=INSERT+INTO+target_1+FORMAT+Native" --data-binary @- -v 2>&1 | grep 'X-ClickHouse-Summary' | sed 's/,\"elapsed_ns[^}]*//' +$CLICKHOUSE_LOCAL -q "SELECT number::Float64 AS v FROM numbers(10)" --format RowBinary | ${CLICKHOUSE_CURL} -sS "${CLICKHOUSE_URL}&wait_end_of_query=1&query=INSERT+INTO+target_1+FORMAT+RowBinary" --data-binary @- -v 2>&1 | grep 'X-ClickHouse-Summary' | sed 's/,\"elapsed_ns[^}]*//' echo "With materialized views" -${CLICKHOUSE_CURL} "${CLICKHOUSE_URL}&wait_end_of_query=1&query=INSERT+INTO+floats" -d "VALUES(1.0)" -v 2>&1 | grep 'X-ClickHouse-Summary' | sed 's/,\"peak_mem[^}]*//' -$CLICKHOUSE_LOCAL -q "SELECT number::Float64 AS v FROM numbers(10)" --format Native | ${CLICKHOUSE_CURL} -sS "${CLICKHOUSE_URL}&wait_end_of_query=1&query=INSERT+INTO+floats+FORMAT+Native" --data-binary @- -v 2>&1 | grep 'X-ClickHouse-Summary' | sed 's/,\"peak_mem[^}]*//' -$CLICKHOUSE_LOCAL -q "SELECT number::Float64 AS v FROM numbers(10)" --format RowBinary | ${CLICKHOUSE_CURL} -sS "${CLICKHOUSE_URL}&wait_end_of_query=1&query=INSERT+INTO+floats+FORMAT+RowBinary" --data-binary @- -v 2>&1 | grep 'X-ClickHouse-Summary' | sed 's/,\"peak_mem[^}]*//' +${CLICKHOUSE_CURL} "${CLICKHOUSE_URL}&wait_end_of_query=1&query=INSERT+INTO+floats" -d "VALUES(1.0)" -v 2>&1 | grep 'X-ClickHouse-Summary' | sed 's/,\"elapsed_ns[^}]*//' +$CLICKHOUSE_LOCAL -q "SELECT number::Float64 AS v FROM numbers(10)" --format Native | ${CLICKHOUSE_CURL} -sS "${CLICKHOUSE_URL}&wait_end_of_query=1&query=INSERT+INTO+floats+FORMAT+Native" --data-binary @- -v 2>&1 | grep 'X-ClickHouse-Summary' | sed 's/,\"elapsed_ns[^}]*//' +$CLICKHOUSE_LOCAL -q "SELECT number::Float64 AS v FROM numbers(10)" --format RowBinary | ${CLICKHOUSE_CURL} -sS "${CLICKHOUSE_URL}&wait_end_of_query=1&query=INSERT+INTO+floats+FORMAT+RowBinary" --data-binary @- -v 2>&1 | grep 'X-ClickHouse-Summary' | sed 's/,\"elapsed_ns[^}]*//' diff --git a/tests/queries/0_stateless/02457_insert_select_progress_http.sh b/tests/queries/0_stateless/02457_insert_select_progress_http.sh index 7f7fc67ae00..ae62ee4b77e 100755 --- a/tests/queries/0_stateless/02457_insert_select_progress_http.sh +++ b/tests/queries/0_stateless/02457_insert_select_progress_http.sh @@ -5,5 +5,5 @@ CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) . "$CURDIR"/../shell_config.sh ${CLICKHOUSE_CURL} -sS "${CLICKHOUSE_URL}&send_progress_in_http_headers=1&http_headers_progress_interval_ms=0" -d @- <<< "insert into function null('_ Int') select * from numbers(5) settings max_block_size=1" -v |& { - grep -F -e X-ClickHouse-Progress: -e X-ClickHouse-Summary: | sed 's/,\"peak_mem[^}]*//' + grep -F -e X-ClickHouse-Progress: -e X-ClickHouse-Summary: | sed 's/,\"elapsed_ns[^}]*//' } diff --git a/tests/queries/0_stateless/02869_http_headers_elapsed_ns.reference b/tests/queries/0_stateless/02869_http_headers_elapsed_ns.reference new file mode 100644 index 00000000000..f89715e69fd --- /dev/null +++ b/tests/queries/0_stateless/02869_http_headers_elapsed_ns.reference @@ -0,0 +1,2 @@ +elapsed_ns in progress are all non zero +elapsed_ns in summary is not zero diff --git a/tests/queries/0_stateless/02869_http_headers_elapsed_ns.sh b/tests/queries/0_stateless/02869_http_headers_elapsed_ns.sh new file mode 100755 index 00000000000..107d8a10345 --- /dev/null +++ b/tests/queries/0_stateless/02869_http_headers_elapsed_ns.sh @@ -0,0 +1,43 @@ +#!/usr/bin/env bash + +CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) +# shellcheck source=../shell_config.sh +. "$CURDIR"/../shell_config.sh + + +CURL_OUTPUT=$(echo 'SELECT number FROM numbers(10)' | \ + ${CLICKHOUSE_CURL} -vsS "${CLICKHOUSE_URL}&max_block_size=1&send_progress_in_http_headers=1&http_headers_progress_interval_ms=0&output_format_parallel_formatting=0" --data-binary @- 2>&1) + +ELAPSED_NS_PROGRESS=$(echo "${CURL_OUTPUT}" | \ + grep 'X-ClickHouse-Progress' | \ + awk '{print $3}' | \ + sed -E 's/.*"elapsed_ns":"?([^,"]*)"?.*/\1/' + ) + +ELAPSED_NS_SUMMARY=$(echo "${CURL_OUTPUT}" | \ + grep 'X-ClickHouse-Summary' | \ + awk '{print $3}' | \ + sed -E 's/.*"elapsed_ns":"?([^,"]*)"?.*/\1/' + ) + + +ALL_ARE_NON_ZERO=1 +while read -r line; do + if [ "$line" -eq 0 ]; then + ALL_ARE_NON_ZERO=0 + break + fi +done <<< "$ELAPSED_NS_PROGRESS" + +if [ "$ALL_ARE_NON_ZERO" -eq 1 ] && [ "$(echo "$ELAPSED_NS_SUMMARY" | wc -l)" -gt 0 ]; then + echo "elapsed_ns in progress are all non zero" +else + echo "elapsed_ns in progress are all zero!" +fi + +if [ "$ELAPSED_NS_SUMMARY" -ne 0 ]; +then + echo "elapsed_ns in summary is not zero" +else + echo "elapsed_ns in summary is zero!" +fi From 91f637b24a0b713561039fff8b980c43e344f441 Mon Sep 17 00:00:00 2001 From: Joe Lynch Date: Mon, 4 Sep 2023 17:56:14 +0200 Subject: [PATCH 132/433] fix python test --- .../test_postgresql_replica_database_engine_1/test.py | 1 + 1 file changed, 1 insertion(+) diff --git a/tests/integration/test_postgresql_replica_database_engine_1/test.py b/tests/integration/test_postgresql_replica_database_engine_1/test.py index 57ee920d49c..c118080a572 100644 --- a/tests/integration/test_postgresql_replica_database_engine_1/test.py +++ b/tests/integration/test_postgresql_replica_database_engine_1/test.py @@ -414,6 +414,7 @@ def test_many_concurrent_queries(started_cluster): port=started_cluster.postgres_port, database=True, ) + cursor = conn.cursor() pg_manager.create_and_fill_postgres_tables(NUM_TABLES, numbers=10000) def attack(thread_id): From d69541786974f454ae86b7bfd68edb718ba45108 Mon Sep 17 00:00:00 2001 From: HarryLeeIBM Date: Tue, 5 Sep 2023 13:38:02 -0700 Subject: [PATCH 133/433] Fix precise float parsing issue on s390x --- src/IO/readFloatText.h | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) diff --git a/src/IO/readFloatText.h b/src/IO/readFloatText.h index da4719b8dcb..feab9589c2e 100644 --- a/src/IO/readFloatText.h +++ b/src/IO/readFloatText.h @@ -238,8 +238,15 @@ ReturnType readFloatTextPreciseImpl(T & x, ReadBuffer & buf) ++num_copied_chars; } - auto res = fast_float::from_chars(tmp_buf, tmp_buf + num_copied_chars, x); - + fast_float::from_chars_result res; + if constexpr (std::endian::native == std::endian::little) + res = fast_float::from_chars(tmp_buf, tmp_buf + num_copied_chars, x); + else + { + Float64 x64 = 0.0; + res = fast_float::from_chars(tmp_buf, tmp_buf + num_copied_chars, x64); + x = static_cast(x64); + } if (unlikely(res.ec != std::errc())) { if constexpr (throw_exception) From d36edea9e2da0c5e42a0b269fd6de576b4419378 Mon Sep 17 00:00:00 2001 From: Robert Schulze Date: Tue, 5 Sep 2023 16:51:54 +0000 Subject: [PATCH 134/433] Consolidate GCD codec tests (follow up to #53149) --- .../sql-reference/statements/create/table.md | 8 ++-- src/Compression/CompressionCodecDelta.cpp | 1 - src/Compression/CompressionCodecGCD.cpp | 17 +++---- ...cimal_with_default_precision_and_scale.sql | 2 +- .../02868_gcd_codec_test_data.reference | 1 - .../0_stateless/02868_gcd_codec_test_data.sql | 15 ------- ...2870_gcd_codec_test_single_codec.reference | 0 .../02870_gcd_codec_test_single_codec.sql | 2 - ...es.reference => 02872_gcd_codec.reference} | 4 ++ ...upported_types.sql => 02872_gcd_codec.sql} | 45 +++++++++++++++---- ..._gcd_codec_test_only_zero_values.reference | 3 -- .../02873_gcd_codec_test_only_zero_values.sql | 3 -- 12 files changed, 51 insertions(+), 50 deletions(-) delete mode 100644 tests/queries/0_stateless/02868_gcd_codec_test_data.reference delete mode 100644 tests/queries/0_stateless/02868_gcd_codec_test_data.sql delete mode 100644 tests/queries/0_stateless/02870_gcd_codec_test_single_codec.reference delete mode 100644 tests/queries/0_stateless/02870_gcd_codec_test_single_codec.sql rename tests/queries/0_stateless/{02872_gcd_codec_test_supported_types.reference => 02872_gcd_codec.reference} (99%) rename tests/queries/0_stateless/{02872_gcd_codec_test_supported_types.sql => 02872_gcd_codec.sql} (75%) delete mode 100644 tests/queries/0_stateless/02873_gcd_codec_test_only_zero_values.reference delete mode 100644 tests/queries/0_stateless/02873_gcd_codec_test_only_zero_values.sql diff --git a/docs/en/sql-reference/statements/create/table.md b/docs/en/sql-reference/statements/create/table.md index a50874cdb95..2a0b70fb19b 100644 --- a/docs/en/sql-reference/statements/create/table.md +++ b/docs/en/sql-reference/statements/create/table.md @@ -391,19 +391,19 @@ DEFLATE_QPL is not available in ClickHouse Cloud. ### Specialized Codecs -These codecs are designed to make compression more effective by using specific features of data. Some of these codecs do not compress data themself. Instead, they prepare the data for a common purpose codec, which compresses it better than without this preparation. +These codecs are designed to make compression more effective by exploiting specific features of the data. Some of these codecs do not compress data themself, they instead preprocess the data such that a second compression stage using a general-purpose codec can achieve a higher data compression rate. #### Delta -`Delta(delta_bytes)` — Compression approach in which raw values are replaced by the difference of two neighboring values, except for the first value that stays unchanged. Up to `delta_bytes` are used for storing delta values, so `delta_bytes` is the maximum size of raw values. Possible `delta_bytes` values: 1, 2, 4, 8. The default value for `delta_bytes` is `sizeof(type)` if equal to 1, 2, 4, or 8. In all other cases, it’s 1. Delta is a data preparation codec, i.e. cannot be used stand-alone. +`Delta(delta_bytes)` — Compression approach in which raw values are replaced by the difference of two neighboring values, except for the first value that stays unchanged. Up to `delta_bytes` are used for storing delta values, so `delta_bytes` is the maximum size of raw values. Possible `delta_bytes` values: 1, 2, 4, 8. The default value for `delta_bytes` is `sizeof(type)` if equal to 1, 2, 4, or 8. In all other cases, it’s 1. Delta is a data preparation codec, i.e. it cannot be used stand-alone. #### DoubleDelta -`DoubleDelta(bytes_size)` — Calculates delta of deltas and writes it in compact binary form. Possible `bytes_size` values: 1, 2, 4, 8, the default value is `sizeof(type)` if equal to 1, 2, 4, or 8. In all other cases, it’s 1. Optimal compression rates are achieved for monotonic sequences with a constant stride, such as time series data. Can be used with any fixed-width type. Implements the algorithm used in Gorilla TSDB, extending it to support 64-bit types. Uses 1 extra bit for 32-bit deltas: 5-bit prefixes instead of 4-bit prefixes. For additional information, see Compressing Time Stamps in [Gorilla: A Fast, Scalable, In-Memory Time Series Database](http://www.vldb.org/pvldb/vol8/p1816-teller.pdf). DoubleDelta is a data preparation codec, i.e. cannot be used stand-alone. +`DoubleDelta(bytes_size)` — Calculates delta of deltas and writes it in compact binary form. Possible `bytes_size` values: 1, 2, 4, 8, the default value is `sizeof(type)` if equal to 1, 2, 4, or 8. In all other cases, it’s 1. Optimal compression rates are achieved for monotonic sequences with a constant stride, such as time series data. Can be used with any fixed-width type. Implements the algorithm used in Gorilla TSDB, extending it to support 64-bit types. Uses 1 extra bit for 32-bit deltas: 5-bit prefixes instead of 4-bit prefixes. For additional information, see Compressing Time Stamps in [Gorilla: A Fast, Scalable, In-Memory Time Series Database](http://www.vldb.org/pvldb/vol8/p1816-teller.pdf). DoubleDelta is a data preparation codec, i.e. it cannot be used stand-alone. #### GCD -`GCD()` - - Calculates the greatest common denominator (GCD) of the values in the column, then divides each value by the GCD. Can be used with integer, decimal and date/time columns. A viable use case are timestamps or monetary values with high precision. GCD is a data preparation codec, i.e. cannot be used stand-alone. +`GCD()` - - Calculates the greatest common denominator (GCD) of the values in the column, then divides each value by the GCD. Can be used with integer, decimal and date/time columns. A viable use case are values which change (increase or decrease) in multiples of the GCD, e.g. 24, 28, 16, 24, 8, 24 (GCD = 4). GCD is a data preparation codec, i.e. it cannot be used stand-alone. #### Gorilla diff --git a/src/Compression/CompressionCodecDelta.cpp b/src/Compression/CompressionCodecDelta.cpp index 51472aedaf0..26beedb411f 100644 --- a/src/Compression/CompressionCodecDelta.cpp +++ b/src/Compression/CompressionCodecDelta.cpp @@ -5,7 +5,6 @@ #include #include #include -#include namespace DB diff --git a/src/Compression/CompressionCodecGCD.cpp b/src/Compression/CompressionCodecGCD.cpp index 7c6e4d86f6c..cb71f252ff5 100644 --- a/src/Compression/CompressionCodecGCD.cpp +++ b/src/Compression/CompressionCodecGCD.cpp @@ -3,14 +3,8 @@ #include #include #include -#include -#include -#include #include "Common/Exception.h" #include "DataTypes/IDataType.h" -#include "base/Decimal_fwd.h" -#include "base/types.h" -#include "config.h" #include #include @@ -84,7 +78,7 @@ void compressDataForType(const char * source, UInt32 source_size, char * dest) const char * const source_end = source + source_size; - T gcd_divider{}; + T gcd_divider = 0; const auto * cur_source = source; while (gcd_divider != T(1) && cur_source < source_end) { @@ -100,7 +94,7 @@ void compressDataForType(const char * source, UInt32 source_size, char * dest) if constexpr (sizeof(T) <= 8) { - /// libdivide support only UInt32 and UInt64. + /// libdivide supports only UInt32 and UInt64. using LibdivideT = std::conditional_t; libdivide::divider divider(static_cast(gcd_divider)); cur_source = source; @@ -126,8 +120,6 @@ void compressDataForType(const char * source, UInt32 source_size, char * dest) template void decompressDataForType(const char * source, UInt32 source_size, char * dest, UInt32 output_size) { - const char * const output_end = dest + output_size; - if (source_size % sizeof(T) != 0) throw Exception(ErrorCodes::CANNOT_DECOMPRESS, "Cannot GCD decompress, data size {} is not aligned to {}", source_size, sizeof(T)); @@ -135,11 +127,14 @@ void decompressDataForType(const char * source, UInt32 source_size, char * dest, throw Exception(ErrorCodes::CANNOT_DECOMPRESS, "Cannot GCD decompress, data size {} is less than {}", source_size, sizeof(T)); const char * const source_end = source + source_size; + const char * const dest_end = dest + output_size; + const T gcd_multiplier = unalignedLoad(source); source += sizeof(T); + while (source < source_end) { - if (dest + sizeof(T) > output_end) [[unlikely]] + if (dest + sizeof(T) > dest_end) [[unlikely]] throw Exception(ErrorCodes::CANNOT_DECOMPRESS, "Cannot decompress the data"); unalignedStore(dest, unalignedLoad(source) * gcd_multiplier); diff --git a/tests/queries/0_stateless/00700_decimal_with_default_precision_and_scale.sql b/tests/queries/0_stateless/00700_decimal_with_default_precision_and_scale.sql index 4d6048ed7ae..5132b593bcc 100644 --- a/tests/queries/0_stateless/00700_decimal_with_default_precision_and_scale.sql +++ b/tests/queries/0_stateless/00700_decimal_with_default_precision_and_scale.sql @@ -12,7 +12,7 @@ ORDER BY (d2, d3); INSERT INTO decimal (d1, d2, d3) VALUES (4.2, 4.2, 4.2); -SELECT type FROM system.columns WHERE table = 'decimal' AND database = currentDatabase(); +SELECT type FROM system.columns WHERE table = 'decimal' AND database = currentDatabase() ORDER BY type; SELECT toTypeName(d2), toTypeName(d3) FROM decimal LIMIT 1; diff --git a/tests/queries/0_stateless/02868_gcd_codec_test_data.reference b/tests/queries/0_stateless/02868_gcd_codec_test_data.reference deleted file mode 100644 index 573541ac970..00000000000 --- a/tests/queries/0_stateless/02868_gcd_codec_test_data.reference +++ /dev/null @@ -1 +0,0 @@ -0 diff --git a/tests/queries/0_stateless/02868_gcd_codec_test_data.sql b/tests/queries/0_stateless/02868_gcd_codec_test_data.sql deleted file mode 100644 index 1039ebcecb8..00000000000 --- a/tests/queries/0_stateless/02868_gcd_codec_test_data.sql +++ /dev/null @@ -1,15 +0,0 @@ -DROP TABLE IF EXISTS table_none; -CREATE TABLE table_none (id UInt64, ui UInt256 CODEC(LZ4)) ENGINE = Memory; -INSERT INTO table_none SELECT * FROM generateRandom() LIMIT 50; - -DROP TABLE IF EXISTS table_gcd_codec; -CREATE TABLE table_gcd_codec (id UInt64, ui UInt256 CODEC(GCD, LZ4)) ENGINE = Memory; -INSERT INTO table_gcd_codec SELECT * FROM table_none; - -SELECT COUNT(*) -FROM ( - SELECT table_none.id, table_none.ui AS ui1, table_gcd_codec.id, table_gcd_codec.ui AS ui2 - FROM table_none - JOIN table_gcd_codec ON table_none.id = table_gcd_codec.id -) -WHERE ui1 != ui2; diff --git a/tests/queries/0_stateless/02870_gcd_codec_test_single_codec.reference b/tests/queries/0_stateless/02870_gcd_codec_test_single_codec.reference deleted file mode 100644 index e69de29bb2d..00000000000 diff --git a/tests/queries/0_stateless/02870_gcd_codec_test_single_codec.sql b/tests/queries/0_stateless/02870_gcd_codec_test_single_codec.sql deleted file mode 100644 index d101884e379..00000000000 --- a/tests/queries/0_stateless/02870_gcd_codec_test_single_codec.sql +++ /dev/null @@ -1,2 +0,0 @@ -DROP TABLE IF EXISTS table_gcd_codec; -CREATE TABLE table_gcd_codec (str UInt64 CODEC(GCD)) ENGINE = Memory; -- { serverError 36 } diff --git a/tests/queries/0_stateless/02872_gcd_codec_test_supported_types.reference b/tests/queries/0_stateless/02872_gcd_codec.reference similarity index 99% rename from tests/queries/0_stateless/02872_gcd_codec_test_supported_types.reference rename to tests/queries/0_stateless/02872_gcd_codec.reference index 2eb7ed1e475..1dd1b67e047 100644 --- a/tests/queries/0_stateless/02872_gcd_codec_test_supported_types.reference +++ b/tests/queries/0_stateless/02872_gcd_codec.reference @@ -1,4 +1,5 @@ 0 +0 1 2 3 @@ -998,3 +999,6 @@ 1970-01-01 02:00:47.000 1970-01-01 02:00:48.000 1970-01-01 02:00:49.000 +0 +0 +0 diff --git a/tests/queries/0_stateless/02872_gcd_codec_test_supported_types.sql b/tests/queries/0_stateless/02872_gcd_codec.sql similarity index 75% rename from tests/queries/0_stateless/02872_gcd_codec_test_supported_types.sql rename to tests/queries/0_stateless/02872_gcd_codec.sql index 3b6d27a7683..245a1211052 100644 --- a/tests/queries/0_stateless/02872_gcd_codec_test_supported_types.sql +++ b/tests/queries/0_stateless/02872_gcd_codec.sql @@ -1,4 +1,28 @@ --- Int +-- GCD codec can't be used stand-alone +CREATE TEMPORARY TABLE table_gcd_codec (n UInt64 CODEC(GCD)) ENGINE = Memory; -- { serverError BAD_ARGUMENTS } + +-- GCD codec rejects non-integer/decimal/datetime types +CREATE TEMPORARY TABLE table_gcd_codec (str String CODEC(GCD, LZ4)) ENGINE = Memory; -- { serverError BAD_ARGUMENTS } + +-- Basic random-based correctness test +CREATE TEMPORARY TABLE table_lz4 (id UInt64, ui UInt256 CODEC(LZ4)) ENGINE = Memory; +INSERT INTO table_lz4 SELECT * FROM generateRandom() LIMIT 50; + +CREATE TEMPORARY TABLE table_gcd (id UInt64, ui UInt256 CODEC(GCD, LZ4)) ENGINE = Memory; +INSERT INTO table_gcd SELECT * FROM table_lz4; + +SELECT COUNT(*) +FROM ( + SELECT table_lz4.id, table_lz4.ui AS ui1, table_gcd.id, table_gcd.ui AS ui2 + FROM table_lz4 JOIN table_gcd + ON table_lz4.id = table_gcd.id +) +WHERE ui1 != ui2; + +------------------------------------------------------------------------------------------- +-- Compression/decompression works for all data types supported by GCD codec + +-- Int* CREATE TEMPORARY TABLE table_gcd_codec_uint8 (n UInt8 CODEC(GCD, LZ4)) ENGINE = Memory; CREATE TEMPORARY TABLE table_gcd_codec_uint16 (n UInt16 CODEC(GCD, LZ4)) ENGINE = Memory; CREATE TEMPORARY TABLE table_gcd_codec_uint32 (n UInt32 CODEC(GCD, LZ4)) ENGINE = Memory; @@ -20,8 +44,7 @@ SELECT * FROM table_gcd_codec_uint64; SELECT * FROM table_gcd_codec_uint128; SELECT * FROM table_gcd_codec_uint256; - --- UInt +-- UInt* CREATE TEMPORARY TABLE table_gcd_codec_int8 (n Int8 CODEC(GCD, LZ4)) ENGINE = Memory; CREATE TEMPORARY TABLE table_gcd_codec_int16 (n Int16 CODEC(GCD, LZ4)) ENGINE = Memory; CREATE TEMPORARY TABLE table_gcd_codec_int32 (n Int32 CODEC(GCD, LZ4)) ENGINE = Memory; @@ -43,8 +66,7 @@ SELECT * FROM table_gcd_codec_int64; SELECT * FROM table_gcd_codec_int128; SELECT * FROM table_gcd_codec_int256; - --- Decimal +-- Decimal* CREATE TEMPORARY TABLE table_gcd_codec_decimal32 (n Decimal32(1) CODEC(GCD, LZ4)) ENGINE = Memory; CREATE TEMPORARY TABLE table_gcd_codec_decimal64 (n Decimal64(1) CODEC(GCD, LZ4)) ENGINE = Memory; CREATE TEMPORARY TABLE table_gcd_codec_decimal128 (n Decimal128(1) CODEC(GCD, LZ4)) ENGINE = Memory; @@ -60,8 +82,7 @@ SELECT * FROM table_gcd_codec_decimal64; SELECT * FROM table_gcd_codec_decimal128; SELECT * FROM table_gcd_codec_decimal256; - --- Date +-- Date[32] CREATE TEMPORARY TABLE table_gcd_codec_date (n Date CODEC(GCD, LZ4)) ENGINE = Memory; CREATE TEMPORARY TABLE table_gcd_codec_date32 (n Date32 CODEC(GCD, LZ4)) ENGINE = Memory; @@ -71,8 +92,7 @@ INSERT INTO table_gcd_codec_date32 SELECT number FROM system.numbers LIMIT 50; SELECT * FROM table_gcd_codec_date; SELECT * FROM table_gcd_codec_date32; - --- DateTime +-- DateTimeTime[64] CREATE TEMPORARY TABLE table_gcd_codec_datetime (n DateTime('Asia/Istanbul') CODEC(GCD, LZ4)) ENGINE = Memory; CREATE TEMPORARY TABLE table_gcd_codec_datetime64 (n DateTime64(3, 'Asia/Istanbul') CODEC(GCD, LZ4)) ENGINE = Memory; @@ -81,3 +101,10 @@ INSERT INTO table_gcd_codec_datetime64 SELECT number FROM system.numbers LIMIT 5 SELECT * FROM table_gcd_codec_datetime; SELECT * FROM table_gcd_codec_datetime64; + + +-- A column with all 0 values can be compressed/decompressed + +CREATE TEMPORARY TABLE table_gcd_codec_only_zero_values (n UInt8 CODEC(GCD, LZ4)) ENGINE = Memory; +INSERT INTO table_gcd_codec_only_zero_values VALUES (0), (0), (0); +SELECT * FROM table_gcd_codec_only_zero_values; diff --git a/tests/queries/0_stateless/02873_gcd_codec_test_only_zero_values.reference b/tests/queries/0_stateless/02873_gcd_codec_test_only_zero_values.reference deleted file mode 100644 index bb0b1cf658d..00000000000 --- a/tests/queries/0_stateless/02873_gcd_codec_test_only_zero_values.reference +++ /dev/null @@ -1,3 +0,0 @@ -0 -0 -0 diff --git a/tests/queries/0_stateless/02873_gcd_codec_test_only_zero_values.sql b/tests/queries/0_stateless/02873_gcd_codec_test_only_zero_values.sql deleted file mode 100644 index 7200e0c5d84..00000000000 --- a/tests/queries/0_stateless/02873_gcd_codec_test_only_zero_values.sql +++ /dev/null @@ -1,3 +0,0 @@ -CREATE TABLE table_gcd_codec_only_zero_values (n UInt8 CODEC(GCD, LZ4)) ENGINE = Memory; -INSERT INTO table_gcd_codec_only_zero_values VALUES (0), (0), (0); -SELECT * FROM table_gcd_codec_only_zero_values; From 7b236c8c38318f47f3431a70e73f348a1f984874 Mon Sep 17 00:00:00 2001 From: Alexey Gerasimchuck Date: Tue, 5 Sep 2023 22:21:18 +0000 Subject: [PATCH 135/433] fixed dereference problem --- src/Interpreters/Context.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/Interpreters/Context.cpp b/src/Interpreters/Context.cpp index c3601ee9d90..306db8ff491 100644 --- a/src/Interpreters/Context.cpp +++ b/src/Interpreters/Context.cpp @@ -982,7 +982,7 @@ void Context::setTemporaryStorageInCache(const String & cache_disk_name, size_t auto file_cache = FileCacheFactory::instance().getByName(disk_ptr->getCacheName()).cache; if (!file_cache) - throw Exception(ErrorCodes::NO_ELEMENTS_IN_CONFIG, "Cache '{}' is not found", file_cache->getBasePath()); + throw Exception(ErrorCodes::NO_ELEMENTS_IN_CONFIG, "Cache '{}' is not found", disk_ptr->getCacheName()); LOG_DEBUG(shared->log, "Using file cache ({}) for temporary files", file_cache->getBasePath()); From b41a4d5c7a1df2c5b03d155f5475bb8cecb24fae Mon Sep 17 00:00:00 2001 From: Robert Schulze Date: Wed, 6 Sep 2023 08:43:11 +0000 Subject: [PATCH 136/433] Fix test --- .../00700_decimal_with_default_precision_and_scale.reference | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/tests/queries/0_stateless/00700_decimal_with_default_precision_and_scale.reference b/tests/queries/0_stateless/00700_decimal_with_default_precision_and_scale.reference index 79e219b89fd..e4db0ebf5ef 100644 --- a/tests/queries/0_stateless/00700_decimal_with_default_precision_and_scale.reference +++ b/tests/queries/0_stateless/00700_decimal_with_default_precision_and_scale.reference @@ -1,4 +1,4 @@ -Decimal(9, 8) -Decimal(18, 0) Decimal(10, 0) +Decimal(18, 0) +Decimal(9, 8) Decimal(18, 0) Decimal(10, 0) From 401496aa1f97fa1723dd7a2fe4dee5acd201478d Mon Sep 17 00:00:00 2001 From: Anton Popov Date: Tue, 5 Sep 2023 13:03:15 +0200 Subject: [PATCH 137/433] Apply suggestions from code review Co-authored-by: Nikita Mikhaylov --- src/Processors/QueryPlan/ReadFromMergeTree.cpp | 1 + src/Storages/MergeTree/IMergeTreeReadPool.h | 2 +- 2 files changed, 2 insertions(+), 1 deletion(-) diff --git a/src/Processors/QueryPlan/ReadFromMergeTree.cpp b/src/Processors/QueryPlan/ReadFromMergeTree.cpp index 8c7a4d86e5b..8deab0ae6c8 100644 --- a/src/Processors/QueryPlan/ReadFromMergeTree.cpp +++ b/src/Processors/QueryPlan/ReadFromMergeTree.cpp @@ -837,6 +837,7 @@ Pipe ReadFromMergeTree::spreadMarkRangesAmongStreamsWithOrder( }; Pipes pipes; + /// For parallel replicas the split will be performed on the initiator side. if (is_parallel_reading_from_replicas) { pipes.emplace_back(readInOrder(std::move(parts_with_ranges), column_names, pool_settings, read_type, input_order_info->limit)); diff --git a/src/Storages/MergeTree/IMergeTreeReadPool.h b/src/Storages/MergeTree/IMergeTreeReadPool.h index 904933a8410..944d933e167 100644 --- a/src/Storages/MergeTree/IMergeTreeReadPool.h +++ b/src/Storages/MergeTree/IMergeTreeReadPool.h @@ -22,7 +22,7 @@ public: virtual bool preservesOrderOfRanges() const = 0; /// task_idx is an implementation defined identifier that helps - /// to get required task. E.g. it may be number of thread, index of + /// to get required task. E.g. it may be number of thread in case of Default reading type or an index of a part in case of InOrder/InReverseOrder reading type. virtual MergeTreeReadTaskPtr getTask(size_t task_idx, MergeTreeReadTask * previous_task) = 0; virtual void profileFeedback(ReadBufferFromFileBase::ProfileInfo info) = 0; }; From 3ef6d0da39488cd28a9014e6c37c3740e18b216e Mon Sep 17 00:00:00 2001 From: Igor Nikonov Date: Wed, 6 Sep 2023 11:02:17 +0000 Subject: [PATCH 138/433] Set status and description --- tests/ci/fast_test_check.py | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/tests/ci/fast_test_check.py b/tests/ci/fast_test_check.py index 522c0f9f290..38e2eca3420 100644 --- a/tests/ci/fast_test_check.py +++ b/tests/ci/fast_test_check.py @@ -26,13 +26,12 @@ from docker_pull_helper import get_image_with_version from env_helper import S3_BUILDS_BUCKET, TEMP_PATH from get_robot_token import get_best_robot_token from pr_info import FORCE_TESTS_LABEL, PRInfo -from report import TestResults, read_test_results +from report import TestResult, TestResults, read_test_results from s3_helper import S3Helper from stopwatch import Stopwatch from tee_popen import TeePopen from upload_result_helper import upload_results from version_helper import get_version_from_repo -from report import TestResult from subprocess import TimeoutExpired NAME = "Fast test" @@ -203,6 +202,8 @@ def main(): expired_timeout, ) ) + state = "failure" + description = test_results[-1].name ch_helper = ClickHouseHelper() s3_path_prefix = os.path.join( From a22e219be96ec2277278e6565447a3ad0cd62db9 Mon Sep 17 00:00:00 2001 From: Igor Nikonov Date: Wed, 6 Sep 2023 11:51:43 +0000 Subject: [PATCH 139/433] Fix --- tests/ci/fast_test_check.py | 20 ++++++++++---------- 1 file changed, 10 insertions(+), 10 deletions(-) diff --git a/tests/ci/fast_test_check.py b/tests/ci/fast_test_check.py index 38e2eca3420..ce10c9dce51 100644 --- a/tests/ci/fast_test_check.py +++ b/tests/ci/fast_test_check.py @@ -152,17 +152,17 @@ def main(): os.makedirs(logs_path) run_log_path = os.path.join(logs_path, "run.log") - expired_timeout = None - with TeePopen(run_cmd, run_log_path, timeout=65) as process: - try: + timeout = 65 + timeout_expired = False + with TeePopen(run_cmd, run_log_path, timeout=timeout) as process: retcode = process.wait() - if retcode == 0: + if process.timeout_exceeded: + logging.info(f"Timeout expired for process execution: {run_cmd}") + timeout_expired = True + elif retcode == 0: logging.info("Run successfully") else: logging.info("Run failed") - except TimeoutExpired as timeout_ex: - logging.info(f"Timeout expired for process execution: {run_cmd}") - expired_timeout = timeout_ex.timeout subprocess.check_call(f"sudo chown -R ubuntu:ubuntu {temp_path}", shell=True) @@ -194,15 +194,15 @@ def main(): else: state, description, test_results, additional_logs = process_results(output_path) - if expired_timeout is not None: + if timeout_expired is not None: test_results.append( TestResult( f"Timeout expired for process execution: {run_cmd}", "FAIL", - expired_timeout, + timeout, ) ) - state = "failure" + state = "timeout" description = test_results[-1].name ch_helper = ClickHouseHelper() From f08d445a0fc083b088e6e830c4f7ca4036503fe7 Mon Sep 17 00:00:00 2001 From: robot-clickhouse Date: Wed, 6 Sep 2023 11:59:51 +0000 Subject: [PATCH 140/433] Automatic style fix --- tests/ci/fast_test_check.py | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) diff --git a/tests/ci/fast_test_check.py b/tests/ci/fast_test_check.py index ce10c9dce51..4af1ca0812a 100644 --- a/tests/ci/fast_test_check.py +++ b/tests/ci/fast_test_check.py @@ -155,14 +155,14 @@ def main(): timeout = 65 timeout_expired = False with TeePopen(run_cmd, run_log_path, timeout=timeout) as process: - retcode = process.wait() - if process.timeout_exceeded: - logging.info(f"Timeout expired for process execution: {run_cmd}") - timeout_expired = True - elif retcode == 0: - logging.info("Run successfully") - else: - logging.info("Run failed") + retcode = process.wait() + if process.timeout_exceeded: + logging.info(f"Timeout expired for process execution: {run_cmd}") + timeout_expired = True + elif retcode == 0: + logging.info("Run successfully") + else: + logging.info("Run failed") subprocess.check_call(f"sudo chown -R ubuntu:ubuntu {temp_path}", shell=True) From f6f2e731a7ee557c0c718352797346544cc7ae83 Mon Sep 17 00:00:00 2001 From: Igor Nikonov Date: Wed, 6 Sep 2023 12:44:24 +0000 Subject: [PATCH 141/433] Fix state --- tests/ci/fast_test_check.py | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/tests/ci/fast_test_check.py b/tests/ci/fast_test_check.py index 4af1ca0812a..afb3a93a617 100644 --- a/tests/ci/fast_test_check.py +++ b/tests/ci/fast_test_check.py @@ -32,7 +32,6 @@ from stopwatch import Stopwatch from tee_popen import TeePopen from upload_result_helper import upload_results from version_helper import get_version_from_repo -from subprocess import TimeoutExpired NAME = "Fast test" @@ -202,7 +201,7 @@ def main(): timeout, ) ) - state = "timeout" + state = "failure" description = test_results[-1].name ch_helper = ClickHouseHelper() From f98007a4ed67954c076b00d282a7a39961719c65 Mon Sep 17 00:00:00 2001 From: Mikhail Koviazin Date: Thu, 31 Aug 2023 12:28:39 +0000 Subject: [PATCH 142/433] Replace Turbo-Base64 with aklomp/base64 Closes #31643 --- .gitmodules | 4 +- contrib/CMakeLists.txt | 2 +- contrib/aklomp-base64 | 1 + contrib/aklomp-base64-cmake/.gitignore | 1 + contrib/aklomp-base64-cmake/CMakeLists.txt | 84 ++++ contrib/aklomp-base64-cmake/README.md | 5 + contrib/aklomp-base64-cmake/config.h.in | 8 + contrib/aklomp-base64-cmake/tables.h | 393 ++++++++++++++++++ contrib/base64 | 1 - contrib/base64-cmake/CMakeLists.txt | 60 --- docker/test/fasttest/run.sh | 2 +- docs/en/operations/system-tables/licenses.md | 4 +- docs/ru/operations/system-tables/licenses.md | 2 +- docs/zh/operations/system-tables/licenses.md | 2 +- src/Functions/FunctionBase64Conversion.h | 24 +- src/Functions/base64Decode.cpp | 1 - src/Functions/base64Encode.cpp | 1 - .../00732_base64_functions.reference | 5 +- .../0_stateless/00732_base64_functions.sql | 4 +- utils/check-style/check-include | 2 +- 20 files changed, 517 insertions(+), 89 deletions(-) create mode 160000 contrib/aklomp-base64 create mode 100644 contrib/aklomp-base64-cmake/.gitignore create mode 100644 contrib/aklomp-base64-cmake/CMakeLists.txt create mode 100644 contrib/aklomp-base64-cmake/README.md create mode 100644 contrib/aklomp-base64-cmake/config.h.in create mode 100644 contrib/aklomp-base64-cmake/tables.h delete mode 160000 contrib/base64 delete mode 100644 contrib/base64-cmake/CMakeLists.txt diff --git a/.gitmodules b/.gitmodules index 4ebccaec4e9..63f0221c2f1 100644 --- a/.gitmodules +++ b/.gitmodules @@ -41,8 +41,8 @@ path = contrib/boost url = https://github.com/ClickHouse/boost [submodule "contrib/base64"] - path = contrib/base64 - url = https://github.com/ClickHouse/Turbo-Base64 + path = contrib/aklomp-base64 + url = https://github.com/aklomp/base64.git [submodule "contrib/arrow"] path = contrib/arrow url = https://github.com/ClickHouse/arrow diff --git a/contrib/CMakeLists.txt b/contrib/CMakeLists.txt index 7b98cf43b09..f9c3fddff40 100644 --- a/contrib/CMakeLists.txt +++ b/contrib/CMakeLists.txt @@ -135,7 +135,7 @@ add_contrib (aws-cmake aws-cmake ) -add_contrib (base64-cmake base64) +add_contrib (aklomp-base64-cmake aklomp-base64) add_contrib (simdjson-cmake simdjson) add_contrib (rapidjson-cmake rapidjson) add_contrib (fastops-cmake fastops) diff --git a/contrib/aklomp-base64 b/contrib/aklomp-base64 new file mode 160000 index 00000000000..e77bd70bdd8 --- /dev/null +++ b/contrib/aklomp-base64 @@ -0,0 +1 @@ +Subproject commit e77bd70bdd860c52c561568cffb251d88bba064c diff --git a/contrib/aklomp-base64-cmake/.gitignore b/contrib/aklomp-base64-cmake/.gitignore new file mode 100644 index 00000000000..0e56cf2f8c1 --- /dev/null +++ b/contrib/aklomp-base64-cmake/.gitignore @@ -0,0 +1 @@ +config.h diff --git a/contrib/aklomp-base64-cmake/CMakeLists.txt b/contrib/aklomp-base64-cmake/CMakeLists.txt new file mode 100644 index 00000000000..74ac1c8e6c3 --- /dev/null +++ b/contrib/aklomp-base64-cmake/CMakeLists.txt @@ -0,0 +1,84 @@ +if(ARCH_AMD64 OR ARCH_AARCH64 OR ARCH_PPC64LE OR ARCH_S390X) + option (ENABLE_BASE64 "Enable base64" ${ENABLE_LIBRARIES}) +elseif(ENABLE_BASE64) + message (${RECONFIGURE_MESSAGE_LEVEL} "base64 library is only supported on x86_64 and aarch64") +endif() + +if (NOT ENABLE_BASE64) + message(STATUS "Not using base64") + return() +endif() + +SET(LIBRARY_DIR "${ClickHouse_SOURCE_DIR}/contrib/aklomp-base64") + +set(HAVE_NEON32 0) +set(HAVE_NEON64 0) +if (ARCH_AARCH64) + set(HAVE_NEON64 1) +endif () + +macro(cast_to_bool var instruction) + if (HAVE_${var}) + set(base64_${var} 1) + set(base64_${var}_opt ${instruction}) + else() + set(base64_${var} 0) + endif() +endmacro() + +cast_to_bool(SSSE3 "-mssse3") +cast_to_bool(SSE41 "-msse4.1") +cast_to_bool(SSE42 "-msse4.2") +cast_to_bool(AVX "-mavx") +cast_to_bool(AVX2 "-mavx2") + +set(HAVE_FAST_UNALIGNED_ACCESS 0) +if(HAVE_SSSE3 OR HAVE_SSE41 OR HAVE_SSE42 OR HAVE_AVX OR HAVE_AVX2 OR ARCH_AARCH64) + set(HAVE_FAST_UNALIGNED_ACCESS 1) +endif () + +configure_file(config.h.in ${CMAKE_CURRENT_BINARY_DIR}/config.h) + +add_library(_base64 + # Library files + "${LIBRARY_DIR}/lib/lib.c" + "${LIBRARY_DIR}/lib/codec_choose.c" + + "${LIBRARY_DIR}/lib/tables/tables.c" + + # Codec implementations + "${LIBRARY_DIR}/lib/arch/generic/codec.c" + + "${LIBRARY_DIR}/lib/arch/ssse3/codec.c" + "${LIBRARY_DIR}/lib/arch/sse41/codec.c" + "${LIBRARY_DIR}/lib/arch/sse42/codec.c" + "${LIBRARY_DIR}/lib/arch/avx/codec.c" + "${LIBRARY_DIR}/lib/arch/avx2/codec.c" + "${LIBRARY_DIR}/lib/arch/avx512/codec.c" + + "${LIBRARY_DIR}/lib/arch/neon32/codec.c" + "${LIBRARY_DIR}/lib/arch/neon64/codec.c" + "${LIBRARY_DIR}/lib/codecs.h" + "${CMAKE_CURRENT_BINARY_DIR}/config.h" + "${CMAKE_CURRENT_SOURCE_DIR}/tables.h") + +if(HAVE_AVX) + set_source_files_properties(${LIBRARY_DIR}/lib/arch/avx/codec.c PROPERTIES COMPILE_FLAGS -mavx) +endif() +if(HAVE_AVX2) + set_source_files_properties(${LIBRARY_DIR}/lib/arch/avx2/codec.c PROPERTIES COMPILE_FLAGS -mavx2) +endif() +if(HAVE_SSE41) + set_source_files_properties(${LIBRARY_DIR}/lib/arch/sse41/codec.c PROPERTIES COMPILE_FLAGS -msse4.1) +endif() +if(HAVE_SSE42) + set_source_files_properties(${LIBRARY_DIR}/lib/arch/sse42/codec.c PROPERTIES COMPILE_FLAGS -msse4.2) +endif() +if(HAVE_SSSE3) + set_source_files_properties(${LIBRARY_DIR}/lib/arch/ssse3/codec.c PROPERTIES COMPILE_FLAGS -mssse3) +endif() + +target_include_directories(_base64 SYSTEM PUBLIC ${LIBRARY_DIR}/include) +target_include_directories(_base64 PRIVATE ${CMAKE_CURRENT_SOURCE_DIR} ${CMAKE_CURRENT_BINARY_DIR}) + +add_library(ch_contrib::base64 ALIAS _base64) diff --git a/contrib/aklomp-base64-cmake/README.md b/contrib/aklomp-base64-cmake/README.md new file mode 100644 index 00000000000..24d2fc23ba8 --- /dev/null +++ b/contrib/aklomp-base64-cmake/README.md @@ -0,0 +1,5 @@ +`tables.h` file was generated by the base64 library with it's makefile: +``` +cc -std=c99 -O3 -Wall -Wextra -pedantic -o lib/tables/table_generator lib/tables/table_generator.c +./lib/tables/table_generator > lib/tables.h +``` diff --git a/contrib/aklomp-base64-cmake/config.h.in b/contrib/aklomp-base64-cmake/config.h.in new file mode 100644 index 00000000000..951bf4f88b1 --- /dev/null +++ b/contrib/aklomp-base64-cmake/config.h.in @@ -0,0 +1,8 @@ +#cmakedefine01 HAVE_NEON32 +#cmakedefine01 HAVE_NEON64 +#cmakedefine01 HAVE_SSSE3 +#cmakedefine01 HAVE_SSE41 +#cmakedefine01 HAVE_SSE42 +#cmakedefine01 HAVE_AVX +#cmakedefine01 HAVE_AVX2 +#cmakedefine01 HAVE_FAST_UNALIGNED_ACCESS diff --git a/contrib/aklomp-base64-cmake/tables.h b/contrib/aklomp-base64-cmake/tables.h new file mode 100644 index 00000000000..f5d951fa79c --- /dev/null +++ b/contrib/aklomp-base64-cmake/tables.h @@ -0,0 +1,393 @@ +#include +#define CHAR62 '+' +#define CHAR63 '/' +#define CHARPAD '=' + + +#if BASE64_LITTLE_ENDIAN + + +/* SPECIAL DECODE TABLES FOR LITTLE ENDIAN (INTEL) CPUS */ + +const uint32_t base64_table_dec_32bit_d0[256] = { +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0x000000f8, 0xffffffff, 0xffffffff, 0xffffffff, 0x000000fc, +0x000000d0, 0x000000d4, 0x000000d8, 0x000000dc, 0x000000e0, 0x000000e4, +0x000000e8, 0x000000ec, 0x000000f0, 0x000000f4, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0x00000000, +0x00000004, 0x00000008, 0x0000000c, 0x00000010, 0x00000014, 0x00000018, +0x0000001c, 0x00000020, 0x00000024, 0x00000028, 0x0000002c, 0x00000030, +0x00000034, 0x00000038, 0x0000003c, 0x00000040, 0x00000044, 0x00000048, +0x0000004c, 0x00000050, 0x00000054, 0x00000058, 0x0000005c, 0x00000060, +0x00000064, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0x00000068, 0x0000006c, 0x00000070, 0x00000074, 0x00000078, +0x0000007c, 0x00000080, 0x00000084, 0x00000088, 0x0000008c, 0x00000090, +0x00000094, 0x00000098, 0x0000009c, 0x000000a0, 0x000000a4, 0x000000a8, +0x000000ac, 0x000000b0, 0x000000b4, 0x000000b8, 0x000000bc, 0x000000c0, +0x000000c4, 0x000000c8, 0x000000cc, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff +}; + + +const uint32_t base64_table_dec_32bit_d1[256] = { +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0x0000e003, 0xffffffff, 0xffffffff, 0xffffffff, 0x0000f003, +0x00004003, 0x00005003, 0x00006003, 0x00007003, 0x00008003, 0x00009003, +0x0000a003, 0x0000b003, 0x0000c003, 0x0000d003, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0x00000000, +0x00001000, 0x00002000, 0x00003000, 0x00004000, 0x00005000, 0x00006000, +0x00007000, 0x00008000, 0x00009000, 0x0000a000, 0x0000b000, 0x0000c000, +0x0000d000, 0x0000e000, 0x0000f000, 0x00000001, 0x00001001, 0x00002001, +0x00003001, 0x00004001, 0x00005001, 0x00006001, 0x00007001, 0x00008001, +0x00009001, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0x0000a001, 0x0000b001, 0x0000c001, 0x0000d001, 0x0000e001, +0x0000f001, 0x00000002, 0x00001002, 0x00002002, 0x00003002, 0x00004002, +0x00005002, 0x00006002, 0x00007002, 0x00008002, 0x00009002, 0x0000a002, +0x0000b002, 0x0000c002, 0x0000d002, 0x0000e002, 0x0000f002, 0x00000003, +0x00001003, 0x00002003, 0x00003003, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff +}; + + +const uint32_t base64_table_dec_32bit_d2[256] = { +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0x00800f00, 0xffffffff, 0xffffffff, 0xffffffff, 0x00c00f00, +0x00000d00, 0x00400d00, 0x00800d00, 0x00c00d00, 0x00000e00, 0x00400e00, +0x00800e00, 0x00c00e00, 0x00000f00, 0x00400f00, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0x00000000, +0x00400000, 0x00800000, 0x00c00000, 0x00000100, 0x00400100, 0x00800100, +0x00c00100, 0x00000200, 0x00400200, 0x00800200, 0x00c00200, 0x00000300, +0x00400300, 0x00800300, 0x00c00300, 0x00000400, 0x00400400, 0x00800400, +0x00c00400, 0x00000500, 0x00400500, 0x00800500, 0x00c00500, 0x00000600, +0x00400600, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0x00800600, 0x00c00600, 0x00000700, 0x00400700, 0x00800700, +0x00c00700, 0x00000800, 0x00400800, 0x00800800, 0x00c00800, 0x00000900, +0x00400900, 0x00800900, 0x00c00900, 0x00000a00, 0x00400a00, 0x00800a00, +0x00c00a00, 0x00000b00, 0x00400b00, 0x00800b00, 0x00c00b00, 0x00000c00, +0x00400c00, 0x00800c00, 0x00c00c00, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff +}; + + +const uint32_t base64_table_dec_32bit_d3[256] = { +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0x003e0000, 0xffffffff, 0xffffffff, 0xffffffff, 0x003f0000, +0x00340000, 0x00350000, 0x00360000, 0x00370000, 0x00380000, 0x00390000, +0x003a0000, 0x003b0000, 0x003c0000, 0x003d0000, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0x00000000, +0x00010000, 0x00020000, 0x00030000, 0x00040000, 0x00050000, 0x00060000, +0x00070000, 0x00080000, 0x00090000, 0x000a0000, 0x000b0000, 0x000c0000, +0x000d0000, 0x000e0000, 0x000f0000, 0x00100000, 0x00110000, 0x00120000, +0x00130000, 0x00140000, 0x00150000, 0x00160000, 0x00170000, 0x00180000, +0x00190000, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0x001a0000, 0x001b0000, 0x001c0000, 0x001d0000, 0x001e0000, +0x001f0000, 0x00200000, 0x00210000, 0x00220000, 0x00230000, 0x00240000, +0x00250000, 0x00260000, 0x00270000, 0x00280000, 0x00290000, 0x002a0000, +0x002b0000, 0x002c0000, 0x002d0000, 0x002e0000, 0x002f0000, 0x00300000, +0x00310000, 0x00320000, 0x00330000, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff +}; + + +#else + + +/* SPECIAL DECODE TABLES FOR BIG ENDIAN (IBM/MOTOROLA/SUN) CPUS */ + +const uint32_t base64_table_dec_32bit_d0[256] = { +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xf8000000, 0xffffffff, 0xffffffff, 0xffffffff, 0xfc000000, +0xd0000000, 0xd4000000, 0xd8000000, 0xdc000000, 0xe0000000, 0xe4000000, +0xe8000000, 0xec000000, 0xf0000000, 0xf4000000, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0x00000000, +0x04000000, 0x08000000, 0x0c000000, 0x10000000, 0x14000000, 0x18000000, +0x1c000000, 0x20000000, 0x24000000, 0x28000000, 0x2c000000, 0x30000000, +0x34000000, 0x38000000, 0x3c000000, 0x40000000, 0x44000000, 0x48000000, +0x4c000000, 0x50000000, 0x54000000, 0x58000000, 0x5c000000, 0x60000000, +0x64000000, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0x68000000, 0x6c000000, 0x70000000, 0x74000000, 0x78000000, +0x7c000000, 0x80000000, 0x84000000, 0x88000000, 0x8c000000, 0x90000000, +0x94000000, 0x98000000, 0x9c000000, 0xa0000000, 0xa4000000, 0xa8000000, +0xac000000, 0xb0000000, 0xb4000000, 0xb8000000, 0xbc000000, 0xc0000000, +0xc4000000, 0xc8000000, 0xcc000000, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff +}; + + +const uint32_t base64_table_dec_32bit_d1[256] = { +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0x03e00000, 0xffffffff, 0xffffffff, 0xffffffff, 0x03f00000, +0x03400000, 0x03500000, 0x03600000, 0x03700000, 0x03800000, 0x03900000, +0x03a00000, 0x03b00000, 0x03c00000, 0x03d00000, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0x00000000, +0x00100000, 0x00200000, 0x00300000, 0x00400000, 0x00500000, 0x00600000, +0x00700000, 0x00800000, 0x00900000, 0x00a00000, 0x00b00000, 0x00c00000, +0x00d00000, 0x00e00000, 0x00f00000, 0x01000000, 0x01100000, 0x01200000, +0x01300000, 0x01400000, 0x01500000, 0x01600000, 0x01700000, 0x01800000, +0x01900000, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0x01a00000, 0x01b00000, 0x01c00000, 0x01d00000, 0x01e00000, +0x01f00000, 0x02000000, 0x02100000, 0x02200000, 0x02300000, 0x02400000, +0x02500000, 0x02600000, 0x02700000, 0x02800000, 0x02900000, 0x02a00000, +0x02b00000, 0x02c00000, 0x02d00000, 0x02e00000, 0x02f00000, 0x03000000, +0x03100000, 0x03200000, 0x03300000, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff +}; + + +const uint32_t base64_table_dec_32bit_d2[256] = { +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0x000f8000, 0xffffffff, 0xffffffff, 0xffffffff, 0x000fc000, +0x000d0000, 0x000d4000, 0x000d8000, 0x000dc000, 0x000e0000, 0x000e4000, +0x000e8000, 0x000ec000, 0x000f0000, 0x000f4000, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0x00000000, +0x00004000, 0x00008000, 0x0000c000, 0x00010000, 0x00014000, 0x00018000, +0x0001c000, 0x00020000, 0x00024000, 0x00028000, 0x0002c000, 0x00030000, +0x00034000, 0x00038000, 0x0003c000, 0x00040000, 0x00044000, 0x00048000, +0x0004c000, 0x00050000, 0x00054000, 0x00058000, 0x0005c000, 0x00060000, +0x00064000, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0x00068000, 0x0006c000, 0x00070000, 0x00074000, 0x00078000, +0x0007c000, 0x00080000, 0x00084000, 0x00088000, 0x0008c000, 0x00090000, +0x00094000, 0x00098000, 0x0009c000, 0x000a0000, 0x000a4000, 0x000a8000, +0x000ac000, 0x000b0000, 0x000b4000, 0x000b8000, 0x000bc000, 0x000c0000, +0x000c4000, 0x000c8000, 0x000cc000, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff +}; + + +const uint32_t base64_table_dec_32bit_d3[256] = { +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0x00003e00, 0xffffffff, 0xffffffff, 0xffffffff, 0x00003f00, +0x00003400, 0x00003500, 0x00003600, 0x00003700, 0x00003800, 0x00003900, +0x00003a00, 0x00003b00, 0x00003c00, 0x00003d00, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0x00000000, +0x00000100, 0x00000200, 0x00000300, 0x00000400, 0x00000500, 0x00000600, +0x00000700, 0x00000800, 0x00000900, 0x00000a00, 0x00000b00, 0x00000c00, +0x00000d00, 0x00000e00, 0x00000f00, 0x00001000, 0x00001100, 0x00001200, +0x00001300, 0x00001400, 0x00001500, 0x00001600, 0x00001700, 0x00001800, +0x00001900, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0x00001a00, 0x00001b00, 0x00001c00, 0x00001d00, 0x00001e00, +0x00001f00, 0x00002000, 0x00002100, 0x00002200, 0x00002300, 0x00002400, +0x00002500, 0x00002600, 0x00002700, 0x00002800, 0x00002900, 0x00002a00, +0x00002b00, 0x00002c00, 0x00002d00, 0x00002e00, 0x00002f00, 0x00003000, +0x00003100, 0x00003200, 0x00003300, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, +0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff +}; + + +#endif diff --git a/contrib/base64 b/contrib/base64 deleted file mode 160000 index 8628e258090..00000000000 --- a/contrib/base64 +++ /dev/null @@ -1 +0,0 @@ -Subproject commit 8628e258090f9eb76d90ac3c91e1ab4690e9aa11 diff --git a/contrib/base64-cmake/CMakeLists.txt b/contrib/base64-cmake/CMakeLists.txt deleted file mode 100644 index 333e0a96a0b..00000000000 --- a/contrib/base64-cmake/CMakeLists.txt +++ /dev/null @@ -1,60 +0,0 @@ -if(ARCH_AMD64 OR ARCH_AARCH64 OR ARCH_PPC64LE OR ARCH_S390X) - option (ENABLE_BASE64 "Enable base64" ${ENABLE_LIBRARIES}) -elseif(ENABLE_BASE64) - message (${RECONFIGURE_MESSAGE_LEVEL} "base64 library is only supported on x86_64 and aarch64") -endif() - -if (NOT ENABLE_BASE64) - message(STATUS "Not using base64") - return() -endif() - -SET(LIBRARY_DIR "${ClickHouse_SOURCE_DIR}/contrib/base64") - -add_library(_base64_scalar OBJECT "${LIBRARY_DIR}/turbob64c.c" "${LIBRARY_DIR}/turbob64d.c") -add_library(_base64_ssse3 OBJECT "${LIBRARY_DIR}/turbob64sse.c") # This file also contains code for ARM NEON - -if (ARCH_AMD64) - add_library(_base64_avx OBJECT "${LIBRARY_DIR}/turbob64sse.c") # This is not a mistake. One file is compiled twice. - add_library(_base64_avx2 OBJECT "${LIBRARY_DIR}/turbob64avx2.c") -endif () - -target_compile_options(_base64_scalar PRIVATE -falign-loops) - -if (ARCH_AMD64) - target_compile_options(_base64_ssse3 PRIVATE -mno-avx -mno-avx2 -mssse3 -falign-loops) - target_compile_options(_base64_avx PRIVATE -falign-loops -mavx) - target_compile_options(_base64_avx2 PRIVATE -falign-loops -mavx2) -else () - if (ARCH_PPC64LE) - target_compile_options(_base64_ssse3 PRIVATE -D__SSSE3__ -falign-loops) - else() - target_compile_options(_base64_ssse3 PRIVATE -falign-loops) - endif() -endif () - -if (ARCH_AMD64) - add_library(_base64 - $ - $ - $ - $) -else () - add_library(_base64 - $ - $) -endif () - -target_include_directories(_base64 SYSTEM PUBLIC ${LIBRARY_DIR}) - -if (XCODE OR XCODE_VERSION) - # https://gitlab.kitware.com/cmake/cmake/issues/17457 - # Some native build systems may not like targets that have only object files, so consider adding at least one real source file - # This applies to Xcode. - if (NOT EXISTS "${CMAKE_CURRENT_BINARY_DIR}/dummy.c") - file(WRITE "${CMAKE_CURRENT_BINARY_DIR}/dummy.c" "") - endif () - target_sources(_base64 PRIVATE "${CMAKE_CURRENT_BINARY_DIR}/dummy.c") -endif () - -add_library(ch_contrib::base64 ALIAS _base64) diff --git a/docker/test/fasttest/run.sh b/docker/test/fasttest/run.sh index 477bbdb6919..296a132d3e3 100755 --- a/docker/test/fasttest/run.sh +++ b/docker/test/fasttest/run.sh @@ -120,7 +120,7 @@ function clone_submodules contrib/libxml2 contrib/libunwind contrib/fmtlib - contrib/base64 + contrib/aklomp-base64 contrib/cctz contrib/libcpuid contrib/libdivide diff --git a/docs/en/operations/system-tables/licenses.md b/docs/en/operations/system-tables/licenses.md index 0f09d559d8b..7a444105489 100644 --- a/docs/en/operations/system-tables/licenses.md +++ b/docs/en/operations/system-tables/licenses.md @@ -1,7 +1,7 @@ --- slug: /en/operations/system-tables/licenses --- -# licenses +# licenses Contains licenses of third-party libraries that are located in the [contrib](https://github.com/ClickHouse/ClickHouse/tree/master/contrib) directory of ClickHouse sources. @@ -27,7 +27,7 @@ SELECT library_name, license_type, license_path FROM system.licenses LIMIT 15 │ aws-c-event-stream │ Apache │ /contrib/aws-c-event-stream/LICENSE │ │ aws-checksums │ Apache │ /contrib/aws-checksums/LICENSE │ │ aws │ Apache │ /contrib/aws/LICENSE.txt │ -│ base64 │ BSD 2-clause │ /contrib/base64/LICENSE │ +│ base64 │ BSD 2-clause │ /contrib/aklomp-base64/LICENSE │ │ boost │ Boost │ /contrib/boost/LICENSE_1_0.txt │ │ brotli │ MIT │ /contrib/brotli/LICENSE │ │ capnproto │ MIT │ /contrib/capnproto/LICENSE │ diff --git a/docs/ru/operations/system-tables/licenses.md b/docs/ru/operations/system-tables/licenses.md index 13407a27c94..41cbe25d0f7 100644 --- a/docs/ru/operations/system-tables/licenses.md +++ b/docs/ru/operations/system-tables/licenses.md @@ -27,7 +27,7 @@ SELECT library_name, license_type, license_path FROM system.licenses LIMIT 15 │ aws-c-event-stream │ Apache │ /contrib/aws-c-event-stream/LICENSE │ │ aws-checksums │ Apache │ /contrib/aws-checksums/LICENSE │ │ aws │ Apache │ /contrib/aws/LICENSE.txt │ -│ base64 │ BSD 2-clause │ /contrib/base64/LICENSE │ +│ base64 │ BSD 2-clause │ /contrib/aklomp-base64/LICENSE │ │ boost │ Boost │ /contrib/boost/LICENSE_1_0.txt │ │ brotli │ MIT │ /contrib/brotli/LICENSE │ │ capnproto │ MIT │ /contrib/capnproto/LICENSE │ diff --git a/docs/zh/operations/system-tables/licenses.md b/docs/zh/operations/system-tables/licenses.md index dc09e65264d..9711617bdfd 100644 --- a/docs/zh/operations/system-tables/licenses.md +++ b/docs/zh/operations/system-tables/licenses.md @@ -27,7 +27,7 @@ SELECT library_name, license_type, license_path FROM system.licenses LIMIT 15 │ aws-c-event-stream │ Apache │ /contrib/aws-c-event-stream/LICENSE │ │ aws-checksums │ Apache │ /contrib/aws-checksums/LICENSE │ │ aws │ Apache │ /contrib/aws/LICENSE.txt │ -│ base64 │ BSD 2-clause │ /contrib/base64/LICENSE │ +│ base64 │ BSD 2-clause │ /contrib/aklomp-base64/LICENSE │ │ boost │ Boost │ /contrib/boost/LICENSE_1_0.txt │ │ brotli │ MIT │ /contrib/brotli/LICENSE │ │ capnproto │ MIT │ /contrib/capnproto/LICENSE │ diff --git a/src/Functions/FunctionBase64Conversion.h b/src/Functions/FunctionBase64Conversion.h index ef340a33149..7bd4bfd777b 100644 --- a/src/Functions/FunctionBase64Conversion.h +++ b/src/Functions/FunctionBase64Conversion.h @@ -1,4 +1,5 @@ #pragma once +#include #include "config.h" #if USE_BASE64 @@ -8,7 +9,7 @@ # include # include # include -# include +# include # include # include @@ -27,11 +28,9 @@ namespace Detail { inline size_t base64Decode(const std::span src, UInt8 * dst) { -# if defined(__aarch64__) - return tb64sdec(reinterpret_cast(src.data()), src.size(), reinterpret_cast(dst)); -# else - return _tb64d(reinterpret_cast(src.data()), src.size(), reinterpret_cast(dst)); -# endif + size_t outlen = 0; + base64_decode(reinterpret_cast(src.data()), src.size(), reinterpret_cast(dst), &outlen, 0); + return outlen; } } @@ -46,15 +45,9 @@ struct Base64Encode static size_t performCoding(const std::span src, UInt8 * dst) { - /* - * Some bug in sse arm64 implementation? - * `base64Encode(repeat('a', 46))` returns wrong padding character - */ -# if defined(__aarch64__) - return tb64senc(reinterpret_cast(src.data()), src.size(), reinterpret_cast(dst)); -# else - return _tb64e(reinterpret_cast(src.data()), src.size(), reinterpret_cast(dst)); -# endif + size_t outlen = 0; + base64_encode(reinterpret_cast(src.data()), src.size(), reinterpret_cast(dst), &outlen, 0); + return outlen; } }; @@ -225,6 +218,7 @@ private: return dst_column; } }; + } #endif diff --git a/src/Functions/base64Decode.cpp b/src/Functions/base64Decode.cpp index 4060aafe1a3..842ceeb3a87 100644 --- a/src/Functions/base64Decode.cpp +++ b/src/Functions/base64Decode.cpp @@ -7,7 +7,6 @@ namespace DB { REGISTER_FUNCTION(Base64Decode) { - tb64ini(0, 0); factory.registerFunction>(); /// MysQL compatibility alias. diff --git a/src/Functions/base64Encode.cpp b/src/Functions/base64Encode.cpp index 773db7e09d9..38b45790a3a 100644 --- a/src/Functions/base64Encode.cpp +++ b/src/Functions/base64Encode.cpp @@ -7,7 +7,6 @@ namespace DB { REGISTER_FUNCTION(Base64Encode) { - tb64ini(0, 0); factory.registerFunction>(); /// MysQL compatibility alias. diff --git a/tests/queries/0_stateless/00732_base64_functions.reference b/tests/queries/0_stateless/00732_base64_functions.reference index 5dc1ba03b89..007648f0cd4 100644 --- a/tests/queries/0_stateless/00732_base64_functions.reference +++ b/tests/queries/0_stateless/00732_base64_functions.reference @@ -13,6 +13,9 @@ foob fooba foobar 1 1 - +This used to be an error +fooba +This also used to be an error +fooba Zm9v foo diff --git a/tests/queries/0_stateless/00732_base64_functions.sql b/tests/queries/0_stateless/00732_base64_functions.sql index adba0cdebbd..2c771c842ac 100644 --- a/tests/queries/0_stateless/00732_base64_functions.sql +++ b/tests/queries/0_stateless/00732_base64_functions.sql @@ -7,13 +7,15 @@ SELECT base64Encode(val) FROM (select arrayJoin(['', 'f', 'fo', 'foo', 'foob', ' SELECT base64Decode(val) FROM (select arrayJoin(['', 'Zg==', 'Zm8=', 'Zm9v', 'Zm9vYg==', 'Zm9vYmE=', 'Zm9vYmFy']) val); SELECT base64Decode(base64Encode('foo')) = 'foo', base64Encode(base64Decode('Zm9v')) == 'Zm9v'; +SELECT 'This used to be an error'; SELECT tryBase64Decode('Zm9vYmF=Zm9v'); SELECT base64Encode(val, 'excess argument') FROM (select arrayJoin(['', 'f', 'fo', 'foo', 'foob', 'fooba', 'foobar']) val); -- { serverError 42 } SELECT base64Decode(val, 'excess argument') FROM (select arrayJoin(['', 'Zg==', 'Zm8=', 'Zm9v', 'Zm9vYg==', 'Zm9vYmE=', 'Zm9vYmFy']) val); -- { serverError 42 } SELECT tryBase64Decode('Zm9vYmF=Zm9v', 'excess argument'); -- { serverError 42 } -SELECT base64Decode('Zm9vYmF=Zm9v'); -- { serverError 117 } +SELECT 'This also used to be an error'; +SELECT base64Decode('Zm9vYmF=Zm9v'); select base64Encode(toFixedString('foo', 3)); select base64Decode(toFixedString('Zm9v', 4)); diff --git a/utils/check-style/check-include b/utils/check-style/check-include index 3c0c6103958..efa5121157d 100755 --- a/utils/check-style/check-include +++ b/utils/check-style/check-include @@ -51,7 +51,7 @@ inc="-I. \ -I./base/daemon \ -I./base/consistent-hashing \ -I./contrib/libhdfs3/include \ --I./contrib/base64/include \ +-I./contrib/aklomp-base64/include \ -I./contrib/protobuf/src \ -I./contrib/cppkafka/include \ -I./contrib/librdkafka-cmake/include \ From f2f24610d002a98d535072a57458956488e9c54b Mon Sep 17 00:00:00 2001 From: Robert Schulze Date: Wed, 6 Sep 2023 15:04:18 +0200 Subject: [PATCH 143/433] Improve English grammar by 8.3% --- docs/en/sql-reference/statements/create/table.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/docs/en/sql-reference/statements/create/table.md b/docs/en/sql-reference/statements/create/table.md index 2a0b70fb19b..5e94a5fdc6f 100644 --- a/docs/en/sql-reference/statements/create/table.md +++ b/docs/en/sql-reference/statements/create/table.md @@ -403,7 +403,7 @@ These codecs are designed to make compression more effective by exploiting speci #### GCD -`GCD()` - - Calculates the greatest common denominator (GCD) of the values in the column, then divides each value by the GCD. Can be used with integer, decimal and date/time columns. A viable use case are values which change (increase or decrease) in multiples of the GCD, e.g. 24, 28, 16, 24, 8, 24 (GCD = 4). GCD is a data preparation codec, i.e. it cannot be used stand-alone. +`GCD()` - - Calculates the greatest common denominator (GCD) of the values in the column, then divides each value by the GCD. Can be used with integer, decimal and date/time columns. The codec is well suited for columns with values that change (increase or decrease) in multiples of the GCD, e.g. 24, 28, 16, 24, 8, 24 (GCD = 4). GCD is a data preparation codec, i.e. it cannot be used stand-alone. #### Gorilla From e858f0b880688c9115c38a1a04078cbe5b9fbd2e Mon Sep 17 00:00:00 2001 From: Igor Nikonov Date: Wed, 6 Sep 2023 13:05:56 +0000 Subject: [PATCH 144/433] Fix description length --- tests/ci/fast_test_check.py | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/tests/ci/fast_test_check.py b/tests/ci/fast_test_check.py index bead2f2f61f..e1548bf801e 100644 --- a/tests/ci/fast_test_check.py +++ b/tests/ci/fast_test_check.py @@ -185,6 +185,11 @@ def main(): state, description, test_results, additional_logs = process_results(output_path) if timeout_expired is not None: + name = f"Timeout expired for process execution: {run_cmd}" + if len(name) > 140: + name = name[:136] + name += " ..." + test_results.append( TestResult( f"Timeout expired for process execution: {run_cmd}", From 0c802df85582a7e985e46230b06ee16deeb8d934 Mon Sep 17 00:00:00 2001 From: yariks5s Date: Wed, 6 Sep 2023 13:33:11 +0000 Subject: [PATCH 145/433] small improvement --- src/Functions/FunctionsConversion.h | 34 ++++++++++++++--------------- 1 file changed, 17 insertions(+), 17 deletions(-) diff --git a/src/Functions/FunctionsConversion.h b/src/Functions/FunctionsConversion.h index ed84ca56749..d1e51ee240b 100644 --- a/src/Functions/FunctionsConversion.h +++ b/src/Functions/FunctionsConversion.h @@ -884,35 +884,32 @@ struct ConvertImplconvertToFullColumnIfConst(); - - const auto & col_with_type_and_name = columnGetNested(datetime_arg); - ColumnUInt8::MutablePtr null_map = copyNullMap(datetime_arg.column); - - const auto & type = static_cast(*col_with_type_and_name.type); const DateLUTImpl * time_zone = nullptr; const ColumnConst * time_zone_column = nullptr; - if (arguments.size() > 1) - time_zone_column = checkAndGetColumnConst(arguments[1].column.get()); - else if (arguments.size() == 1) + if (arguments.size() == 1) { auto non_null_args = createBlockWithNestedColumns(arguments); time_zone = &extractTimeZoneFromFunctionArguments(non_null_args, 1, 0); } - - if constexpr (std::is_same_v || std::is_same_v) - time_zone = &DateLUT::instance(); - /// For argument of Date or DateTime type, second argument with time zone could be specified. - if constexpr (std::is_same_v || std::is_same_v) + else /// When we have a column for timezone { - if (time_zone_column) + datetime_arg.column = datetime_arg.column->convertToFullColumnIfConst(); + + if constexpr (std::is_same_v || std::is_same_v) + time_zone = &DateLUT::instance(); + /// For argument of Date or DateTime type, second argument with time zone could be specified. + if constexpr (std::is_same_v || std::is_same_v) { - auto non_null_args = createBlockWithNestedColumns(arguments); - time_zone = &extractTimeZoneFromFunctionArguments(non_null_args, 1, 0); + if ((time_zone_column = checkAndGetColumnConst(arguments[1].column.get()))) + { + auto non_null_args = createBlockWithNestedColumns(arguments); + time_zone = &extractTimeZoneFromFunctionArguments(non_null_args, 1, 0); + } } } + const auto & col_with_type_and_name = columnGetNested(datetime_arg); if (const auto col_from = checkAndGetColumn(col_with_type_and_name.column.get())) { @@ -937,6 +934,9 @@ struct ConvertImpl write_buffer(data_to); + const auto & type = static_cast(*col_with_type_and_name.type); + + ColumnUInt8::MutablePtr null_map = copyNullMap(datetime_arg.column); if (null_map) { From babbecc646acf48b4176ef07013728100948a494 Mon Sep 17 00:00:00 2001 From: Igor Nikonov Date: Wed, 6 Sep 2023 15:00:32 +0000 Subject: [PATCH 146/433] Fix --- tests/ci/fast_test_check.py | 11 +++-------- 1 file changed, 3 insertions(+), 8 deletions(-) diff --git a/tests/ci/fast_test_check.py b/tests/ci/fast_test_check.py index e1548bf801e..97e4f2a8713 100644 --- a/tests/ci/fast_test_check.py +++ b/tests/ci/fast_test_check.py @@ -185,18 +185,13 @@ def main(): state, description, test_results, additional_logs = process_results(output_path) if timeout_expired is not None: - name = f"Timeout expired for process execution: {run_cmd}" + name = f"Timeout expired for process execution" if len(name) > 140: name = name[:136] name += " ..." - test_results.append( - TestResult( - f"Timeout expired for process execution: {run_cmd}", - "FAIL", - timeout, - ) - ) + test_results.append(TestResult(name, "FAIL", timeout)) + state = "failure" description = test_results[-1].name From e5a3a9319fc00814cb87fc842e645d244ae84472 Mon Sep 17 00:00:00 2001 From: Mikhail Koviazin Date: Wed, 6 Sep 2023 15:17:52 +0000 Subject: [PATCH 147/433] Review fixes * Remove generated `tables.h` and use the ones from upstream instead * Add comment about compiling AVX512 codec --- contrib/aklomp-base64-cmake/CMakeLists.txt | 10 +- contrib/aklomp-base64-cmake/README.md | 5 - contrib/aklomp-base64-cmake/tables.h | 393 --------------------- 3 files changed, 8 insertions(+), 400 deletions(-) delete mode 100644 contrib/aklomp-base64-cmake/README.md delete mode 100644 contrib/aklomp-base64-cmake/tables.h diff --git a/contrib/aklomp-base64-cmake/CMakeLists.txt b/contrib/aklomp-base64-cmake/CMakeLists.txt index 74ac1c8e6c3..17fa3dcec02 100644 --- a/contrib/aklomp-base64-cmake/CMakeLists.txt +++ b/contrib/aklomp-base64-cmake/CMakeLists.txt @@ -54,13 +54,19 @@ add_library(_base64 "${LIBRARY_DIR}/lib/arch/sse42/codec.c" "${LIBRARY_DIR}/lib/arch/avx/codec.c" "${LIBRARY_DIR}/lib/arch/avx2/codec.c" + # Even though AVX512 is not supported, we still need to compile this file + # because the run-time codec choose requires the functions from this file + # to be present. "${LIBRARY_DIR}/lib/arch/avx512/codec.c" + # Tables + "${LIBRARY_DIR}/lib/tables/table_dec_32bit.h" + "${LIBRARY_DIR}/lib/tables/table_enc_12bit.h" + "${LIBRARY_DIR}/lib/arch/neon32/codec.c" "${LIBRARY_DIR}/lib/arch/neon64/codec.c" "${LIBRARY_DIR}/lib/codecs.h" - "${CMAKE_CURRENT_BINARY_DIR}/config.h" - "${CMAKE_CURRENT_SOURCE_DIR}/tables.h") + "${CMAKE_CURRENT_BINARY_DIR}/config.h") if(HAVE_AVX) set_source_files_properties(${LIBRARY_DIR}/lib/arch/avx/codec.c PROPERTIES COMPILE_FLAGS -mavx) diff --git a/contrib/aklomp-base64-cmake/README.md b/contrib/aklomp-base64-cmake/README.md deleted file mode 100644 index 24d2fc23ba8..00000000000 --- a/contrib/aklomp-base64-cmake/README.md +++ /dev/null @@ -1,5 +0,0 @@ -`tables.h` file was generated by the base64 library with it's makefile: -``` -cc -std=c99 -O3 -Wall -Wextra -pedantic -o lib/tables/table_generator lib/tables/table_generator.c -./lib/tables/table_generator > lib/tables.h -``` diff --git a/contrib/aklomp-base64-cmake/tables.h b/contrib/aklomp-base64-cmake/tables.h deleted file mode 100644 index f5d951fa79c..00000000000 --- a/contrib/aklomp-base64-cmake/tables.h +++ /dev/null @@ -1,393 +0,0 @@ -#include -#define CHAR62 '+' -#define CHAR63 '/' -#define CHARPAD '=' - - -#if BASE64_LITTLE_ENDIAN - - -/* SPECIAL DECODE TABLES FOR LITTLE ENDIAN (INTEL) CPUS */ - -const uint32_t base64_table_dec_32bit_d0[256] = { -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0x000000f8, 0xffffffff, 0xffffffff, 0xffffffff, 0x000000fc, -0x000000d0, 0x000000d4, 0x000000d8, 0x000000dc, 0x000000e0, 0x000000e4, -0x000000e8, 0x000000ec, 0x000000f0, 0x000000f4, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0x00000000, -0x00000004, 0x00000008, 0x0000000c, 0x00000010, 0x00000014, 0x00000018, -0x0000001c, 0x00000020, 0x00000024, 0x00000028, 0x0000002c, 0x00000030, -0x00000034, 0x00000038, 0x0000003c, 0x00000040, 0x00000044, 0x00000048, -0x0000004c, 0x00000050, 0x00000054, 0x00000058, 0x0000005c, 0x00000060, -0x00000064, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0x00000068, 0x0000006c, 0x00000070, 0x00000074, 0x00000078, -0x0000007c, 0x00000080, 0x00000084, 0x00000088, 0x0000008c, 0x00000090, -0x00000094, 0x00000098, 0x0000009c, 0x000000a0, 0x000000a4, 0x000000a8, -0x000000ac, 0x000000b0, 0x000000b4, 0x000000b8, 0x000000bc, 0x000000c0, -0x000000c4, 0x000000c8, 0x000000cc, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff -}; - - -const uint32_t base64_table_dec_32bit_d1[256] = { -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0x0000e003, 0xffffffff, 0xffffffff, 0xffffffff, 0x0000f003, -0x00004003, 0x00005003, 0x00006003, 0x00007003, 0x00008003, 0x00009003, -0x0000a003, 0x0000b003, 0x0000c003, 0x0000d003, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0x00000000, -0x00001000, 0x00002000, 0x00003000, 0x00004000, 0x00005000, 0x00006000, -0x00007000, 0x00008000, 0x00009000, 0x0000a000, 0x0000b000, 0x0000c000, -0x0000d000, 0x0000e000, 0x0000f000, 0x00000001, 0x00001001, 0x00002001, -0x00003001, 0x00004001, 0x00005001, 0x00006001, 0x00007001, 0x00008001, -0x00009001, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0x0000a001, 0x0000b001, 0x0000c001, 0x0000d001, 0x0000e001, -0x0000f001, 0x00000002, 0x00001002, 0x00002002, 0x00003002, 0x00004002, -0x00005002, 0x00006002, 0x00007002, 0x00008002, 0x00009002, 0x0000a002, -0x0000b002, 0x0000c002, 0x0000d002, 0x0000e002, 0x0000f002, 0x00000003, -0x00001003, 0x00002003, 0x00003003, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff -}; - - -const uint32_t base64_table_dec_32bit_d2[256] = { -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0x00800f00, 0xffffffff, 0xffffffff, 0xffffffff, 0x00c00f00, -0x00000d00, 0x00400d00, 0x00800d00, 0x00c00d00, 0x00000e00, 0x00400e00, -0x00800e00, 0x00c00e00, 0x00000f00, 0x00400f00, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0x00000000, -0x00400000, 0x00800000, 0x00c00000, 0x00000100, 0x00400100, 0x00800100, -0x00c00100, 0x00000200, 0x00400200, 0x00800200, 0x00c00200, 0x00000300, -0x00400300, 0x00800300, 0x00c00300, 0x00000400, 0x00400400, 0x00800400, -0x00c00400, 0x00000500, 0x00400500, 0x00800500, 0x00c00500, 0x00000600, -0x00400600, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0x00800600, 0x00c00600, 0x00000700, 0x00400700, 0x00800700, -0x00c00700, 0x00000800, 0x00400800, 0x00800800, 0x00c00800, 0x00000900, -0x00400900, 0x00800900, 0x00c00900, 0x00000a00, 0x00400a00, 0x00800a00, -0x00c00a00, 0x00000b00, 0x00400b00, 0x00800b00, 0x00c00b00, 0x00000c00, -0x00400c00, 0x00800c00, 0x00c00c00, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff -}; - - -const uint32_t base64_table_dec_32bit_d3[256] = { -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0x003e0000, 0xffffffff, 0xffffffff, 0xffffffff, 0x003f0000, -0x00340000, 0x00350000, 0x00360000, 0x00370000, 0x00380000, 0x00390000, -0x003a0000, 0x003b0000, 0x003c0000, 0x003d0000, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0x00000000, -0x00010000, 0x00020000, 0x00030000, 0x00040000, 0x00050000, 0x00060000, -0x00070000, 0x00080000, 0x00090000, 0x000a0000, 0x000b0000, 0x000c0000, -0x000d0000, 0x000e0000, 0x000f0000, 0x00100000, 0x00110000, 0x00120000, -0x00130000, 0x00140000, 0x00150000, 0x00160000, 0x00170000, 0x00180000, -0x00190000, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0x001a0000, 0x001b0000, 0x001c0000, 0x001d0000, 0x001e0000, -0x001f0000, 0x00200000, 0x00210000, 0x00220000, 0x00230000, 0x00240000, -0x00250000, 0x00260000, 0x00270000, 0x00280000, 0x00290000, 0x002a0000, -0x002b0000, 0x002c0000, 0x002d0000, 0x002e0000, 0x002f0000, 0x00300000, -0x00310000, 0x00320000, 0x00330000, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff -}; - - -#else - - -/* SPECIAL DECODE TABLES FOR BIG ENDIAN (IBM/MOTOROLA/SUN) CPUS */ - -const uint32_t base64_table_dec_32bit_d0[256] = { -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xf8000000, 0xffffffff, 0xffffffff, 0xffffffff, 0xfc000000, -0xd0000000, 0xd4000000, 0xd8000000, 0xdc000000, 0xe0000000, 0xe4000000, -0xe8000000, 0xec000000, 0xf0000000, 0xf4000000, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0x00000000, -0x04000000, 0x08000000, 0x0c000000, 0x10000000, 0x14000000, 0x18000000, -0x1c000000, 0x20000000, 0x24000000, 0x28000000, 0x2c000000, 0x30000000, -0x34000000, 0x38000000, 0x3c000000, 0x40000000, 0x44000000, 0x48000000, -0x4c000000, 0x50000000, 0x54000000, 0x58000000, 0x5c000000, 0x60000000, -0x64000000, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0x68000000, 0x6c000000, 0x70000000, 0x74000000, 0x78000000, -0x7c000000, 0x80000000, 0x84000000, 0x88000000, 0x8c000000, 0x90000000, -0x94000000, 0x98000000, 0x9c000000, 0xa0000000, 0xa4000000, 0xa8000000, -0xac000000, 0xb0000000, 0xb4000000, 0xb8000000, 0xbc000000, 0xc0000000, -0xc4000000, 0xc8000000, 0xcc000000, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff -}; - - -const uint32_t base64_table_dec_32bit_d1[256] = { -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0x03e00000, 0xffffffff, 0xffffffff, 0xffffffff, 0x03f00000, -0x03400000, 0x03500000, 0x03600000, 0x03700000, 0x03800000, 0x03900000, -0x03a00000, 0x03b00000, 0x03c00000, 0x03d00000, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0x00000000, -0x00100000, 0x00200000, 0x00300000, 0x00400000, 0x00500000, 0x00600000, -0x00700000, 0x00800000, 0x00900000, 0x00a00000, 0x00b00000, 0x00c00000, -0x00d00000, 0x00e00000, 0x00f00000, 0x01000000, 0x01100000, 0x01200000, -0x01300000, 0x01400000, 0x01500000, 0x01600000, 0x01700000, 0x01800000, -0x01900000, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0x01a00000, 0x01b00000, 0x01c00000, 0x01d00000, 0x01e00000, -0x01f00000, 0x02000000, 0x02100000, 0x02200000, 0x02300000, 0x02400000, -0x02500000, 0x02600000, 0x02700000, 0x02800000, 0x02900000, 0x02a00000, -0x02b00000, 0x02c00000, 0x02d00000, 0x02e00000, 0x02f00000, 0x03000000, -0x03100000, 0x03200000, 0x03300000, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff -}; - - -const uint32_t base64_table_dec_32bit_d2[256] = { -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0x000f8000, 0xffffffff, 0xffffffff, 0xffffffff, 0x000fc000, -0x000d0000, 0x000d4000, 0x000d8000, 0x000dc000, 0x000e0000, 0x000e4000, -0x000e8000, 0x000ec000, 0x000f0000, 0x000f4000, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0x00000000, -0x00004000, 0x00008000, 0x0000c000, 0x00010000, 0x00014000, 0x00018000, -0x0001c000, 0x00020000, 0x00024000, 0x00028000, 0x0002c000, 0x00030000, -0x00034000, 0x00038000, 0x0003c000, 0x00040000, 0x00044000, 0x00048000, -0x0004c000, 0x00050000, 0x00054000, 0x00058000, 0x0005c000, 0x00060000, -0x00064000, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0x00068000, 0x0006c000, 0x00070000, 0x00074000, 0x00078000, -0x0007c000, 0x00080000, 0x00084000, 0x00088000, 0x0008c000, 0x00090000, -0x00094000, 0x00098000, 0x0009c000, 0x000a0000, 0x000a4000, 0x000a8000, -0x000ac000, 0x000b0000, 0x000b4000, 0x000b8000, 0x000bc000, 0x000c0000, -0x000c4000, 0x000c8000, 0x000cc000, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff -}; - - -const uint32_t base64_table_dec_32bit_d3[256] = { -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0x00003e00, 0xffffffff, 0xffffffff, 0xffffffff, 0x00003f00, -0x00003400, 0x00003500, 0x00003600, 0x00003700, 0x00003800, 0x00003900, -0x00003a00, 0x00003b00, 0x00003c00, 0x00003d00, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0x00000000, -0x00000100, 0x00000200, 0x00000300, 0x00000400, 0x00000500, 0x00000600, -0x00000700, 0x00000800, 0x00000900, 0x00000a00, 0x00000b00, 0x00000c00, -0x00000d00, 0x00000e00, 0x00000f00, 0x00001000, 0x00001100, 0x00001200, -0x00001300, 0x00001400, 0x00001500, 0x00001600, 0x00001700, 0x00001800, -0x00001900, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0x00001a00, 0x00001b00, 0x00001c00, 0x00001d00, 0x00001e00, -0x00001f00, 0x00002000, 0x00002100, 0x00002200, 0x00002300, 0x00002400, -0x00002500, 0x00002600, 0x00002700, 0x00002800, 0x00002900, 0x00002a00, -0x00002b00, 0x00002c00, 0x00002d00, 0x00002e00, 0x00002f00, 0x00003000, -0x00003100, 0x00003200, 0x00003300, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, -0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff -}; - - -#endif From 4b5784f7d021ca03b1759de944d777f780a6538d Mon Sep 17 00:00:00 2001 From: Kruglov Pavel <48961922+Avogar@users.noreply.github.com> Date: Wed, 6 Sep 2023 18:05:20 +0200 Subject: [PATCH 148/433] Fix test --- .../02845_group_by_constant_keys.sql | 25 +++++++++---------- 1 file changed, 12 insertions(+), 13 deletions(-) diff --git a/tests/queries/0_stateless/02845_group_by_constant_keys.sql b/tests/queries/0_stateless/02845_group_by_constant_keys.sql index e16fd3fd1d2..053ad3ecdc4 100644 --- a/tests/queries/0_stateless/02845_group_by_constant_keys.sql +++ b/tests/queries/0_stateless/02845_group_by_constant_keys.sql @@ -10,20 +10,19 @@ insert into test select * from numbers(10); insert into test select * from numbers(10); insert into test select * from numbers(10); -select count() from test group by _file settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=0, compile_aggregate_expressions=0; -select count() from test group by _file settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=1, compile_aggregate_expressions=0; -select count() from test group by _file settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=0, compile_aggregate_expressions=1; -select count() from test group by _file settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=1, compile_aggregate_expressions=1; +select count() from test group by _file order by _file settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=0, compile_aggregate_expressions=0; +select count() from test group by _file order by _file settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=1, compile_aggregate_expressions=0; +select count() from test group by _file order by _file settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=0, compile_aggregate_expressions=1; +select count() from test group by _file order by _file settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=1, compile_aggregate_expressions=1; -select count(), _file from test group by _file settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=0, compile_aggregate_expressions=0; -select count(), _file from test group by _file settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=1, compile_aggregate_expressions=0; -select count(), _file from test group by _file settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=0, compile_aggregate_expressions=1; -select count(), _file from test group by _file settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=1, compile_aggregate_expressions=1; +select count(), _file from test group by _file order by _file settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=0, compile_aggregate_expressions=0; +select count(), _file from test group by _file order by _file settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=1, compile_aggregate_expressions=0; +select count(), _file from test group by _file order by _file settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=0, compile_aggregate_expressions=1; +select count(), _file from test group by _file order by _file settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=1, compile_aggregate_expressions=1; -select count() from test group by _file, _path settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=0, compile_aggregate_expressions=0; -select count() from test group by _file, _path settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=1, compile_aggregate_expressions=0; -select count() from test group by _file, _path settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=0, compile_aggregate_expressions=1; -select count() from test group by _file, _path settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=1, compile_aggregate_expressions=1; +select count() from test group by _file, _path order by _file, _path settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=0, compile_aggregate_expressions=0; +select count() from test group by _file, _path order by _file, _path settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=1, compile_aggregate_expressions=0; +select count() from test group by _file, _path order by _file, _path settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=0, compile_aggregate_expressions=1; +select count() from test group by _file, _path order by _file, _path settings optimize_group_by_constant_keys=1, enable_software_prefetch_in_aggregation=1, compile_aggregate_expressions=1; drop table test; - From b10639594abf7b3bf28a27b8974e221b3700d256 Mon Sep 17 00:00:00 2001 From: Kruglov Pavel <48961922+Avogar@users.noreply.github.com> Date: Wed, 6 Sep 2023 18:05:43 +0200 Subject: [PATCH 149/433] Update test reference --- .../02845_group_by_constant_keys.reference | 24 +++++++++---------- 1 file changed, 12 insertions(+), 12 deletions(-) diff --git a/tests/queries/0_stateless/02845_group_by_constant_keys.reference b/tests/queries/0_stateless/02845_group_by_constant_keys.reference index 67cbdf0c025..5473729c960 100644 --- a/tests/queries/0_stateless/02845_group_by_constant_keys.reference +++ b/tests/queries/0_stateless/02845_group_by_constant_keys.reference @@ -14,18 +14,18 @@ 10 10 10 -10 data.Parquet -10 data.2.Parquet -10 data.1.Parquet -10 data.Parquet -10 data.2.Parquet -10 data.1.Parquet -10 data.Parquet -10 data.2.Parquet -10 data.1.Parquet -10 data.Parquet -10 data.2.Parquet -10 data.1.Parquet +10 data.1.JSON +10 data.2.JSON +10 data.JSON +10 data.1.JSON +10 data.2.JSON +10 data.JSON +10 data.1.JSON +10 data.2.JSON +10 data.JSON +10 data.1.JSON +10 data.2.JSON +10 data.JSON 10 10 10 From 692f4d1f6718c52030a9e63511147580e945d24a Mon Sep 17 00:00:00 2001 From: Kruglov Pavel <48961922+Avogar@users.noreply.github.com> Date: Wed, 6 Sep 2023 18:09:56 +0200 Subject: [PATCH 150/433] Fix test --- .../0_stateless/02842_capn_proto_outfile_without_schema.sh | 1 + 1 file changed, 1 insertion(+) diff --git a/tests/queries/0_stateless/02842_capn_proto_outfile_without_schema.sh b/tests/queries/0_stateless/02842_capn_proto_outfile_without_schema.sh index fde08484026..ae57cdd7543 100755 --- a/tests/queries/0_stateless/02842_capn_proto_outfile_without_schema.sh +++ b/tests/queries/0_stateless/02842_capn_proto_outfile_without_schema.sh @@ -1,4 +1,5 @@ #!/usr/bin/env bash +# Tags: no-fasttest CUR_DIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) # shellcheck source=../shell_config.sh From bcb058f999f9826dfcb9beb5435354b95abacc71 Mon Sep 17 00:00:00 2001 From: johanngan Date: Mon, 12 Jun 2023 15:41:55 -0500 Subject: [PATCH 151/433] Add case insensitive and dot-all modes to RegExpTree dictionary The new per-dictionary settings control regex match semantics around case sensitivity and the '.' wildcard with newlines. They must be set at the dictionary level since they're applied to regex engines at pattern-compile-time. - regexp_dict_flag_case_insensitive: case insensitive matching - regexp_dict_flag_dotall: '.' matches all characters including newlines They correspond to HS_FLAG_CASELESS and HS_FLAG_DOTALL in Vectorscan and case_sensitive and dot_nl in RE2. These are the most useful options compatible with the internal behavior of RegExpTreeDictionary around splitting up simple and complex patterns between Vectorscan and RE2. The alternative is to use (?i) and/or (?s) for all patterns. However, (?s) isn't handled properly by OptimizedRegularExpression::analyze(). And while (?i) is, it still causes the dictionary to treat the pattern as "complex" for sequential scanning with RE2 rather than multi-matching with Vectorscan, even though Vectorscan supports case insensitive literal matching. Setting dictionary-wide flags is both more convenient, and circumvents these problems. --- docs/en/sql-reference/dictionaries/index.md | 6 ++ src/Core/Settings.h | 2 + src/Dictionaries/RegExpTreeDictionary.cpp | 28 +++++++++- src/Dictionaries/RegExpTreeDictionary.h | 9 ++- ...04_regexp_dictionary_yaml_source.reference | 12 ++++ .../02504_regexp_dictionary_yaml_source.sh | 56 +++++++++++++++++++ 6 files changed, 108 insertions(+), 5 deletions(-) diff --git a/docs/en/sql-reference/dictionaries/index.md b/docs/en/sql-reference/dictionaries/index.md index dd8031461e0..80f728a33df 100644 --- a/docs/en/sql-reference/dictionaries/index.md +++ b/docs/en/sql-reference/dictionaries/index.md @@ -2361,6 +2361,12 @@ Result: └────────────────────────────────────────┴───────────────────────────────────────────────────────────────────────────────────────┘ ``` +#### Matching Modes + +Pattern matching behavior can be modified with certain dictionary settings: +- `regexp_dict_flag_case_insensitive`: Use case-insensitive matching (defaults to `false`). Can be overridden in individual expressions with `(?i)` and `(?-i)`. +- `regexp_dict_flag_dotall`: Allow '.' to match newline characters (defaults to `false`). + ### Use Regular Expression Tree Dictionary in ClickHouse Cloud Above used `YAMLRegExpTree` source works in ClickHouse Open Source but not in ClickHouse Cloud. To use regexp tree dictionaries in ClickHouse could, first create a regexp tree dictionary from a YAML file locally in ClickHouse Open Source, then dump this dictionary into a CSV file using the `dictionary` table function and the [INTO OUTFILE](../statements/select/into-outfile.md) clause. diff --git a/src/Core/Settings.h b/src/Core/Settings.h index 620cc8fd67f..b3a44a7a8ee 100644 --- a/src/Core/Settings.h +++ b/src/Core/Settings.h @@ -1053,6 +1053,8 @@ class IColumn; \ M(Bool, format_display_secrets_in_show_and_select, false, "Do not hide secrets in SHOW and SELECT queries.", IMPORTANT) \ M(Bool, regexp_dict_allow_hyperscan, true, "Allow regexp_tree dictionary using Hyperscan library.", 0) \ + M(Bool, regexp_dict_flag_case_insensitive, false, "Use case-insensitive matching for a regexp_tree dictionary. Can be overridden in individual expressions with (?i) and (?-i).", 0) \ + M(Bool, regexp_dict_flag_dotall, false, "Allow '.' to match newline characters for a regexp_tree dictionary.", 0) \ \ M(Bool, dictionary_use_async_executor, false, "Execute a pipeline for reading from a dictionary with several threads. It's supported only by DIRECT dictionary with CLICKHOUSE source.", 0) \ M(Bool, precise_float_parsing, false, "Prefer more precise (but slower) float parsing algorithm", 0) \ diff --git a/src/Dictionaries/RegExpTreeDictionary.cpp b/src/Dictionaries/RegExpTreeDictionary.cpp index e9ba4c6268e..29ef71b3ce0 100644 --- a/src/Dictionaries/RegExpTreeDictionary.cpp +++ b/src/Dictionaries/RegExpTreeDictionary.cpp @@ -206,6 +206,8 @@ void RegExpTreeDictionary::initRegexNodes(Block & block) re2_st::RE2::Options regexp_options; regexp_options.set_log_errors(false); + regexp_options.set_case_sensitive(!flag_case_insensitive); + regexp_options.set_dot_nl(flag_dotall); RegexTreeNodePtr node = std::make_shared(id, parent_id, regex, regexp_options); int num_captures = std::min(node->searcher.NumberOfCapturingGroups() + 1, 10); @@ -330,11 +332,20 @@ void RegExpTreeDictionary::loadData() std::vector flags; std::vector lengths; + // Notes: + // - Always set HS_FLAG_SINGLEMATCH because we only care about whether a pattern matches at least once + // - HS_FLAG_CASELESS is supported by hs_compile_lit_multi, so we should set it if flag_case_insensitive is set. + // - HS_FLAG_DOTALL is not supported by hs_compile_lit_multi, but the '.' wildcard can't appear in any of the simple regexps + // anyway, so even if flag_dotall is set, we only need to configure the RE2 searcher, and don't need to set any Hyperscan flags. + unsigned int flag_bits = HS_FLAG_SINGLEMATCH; + if (flag_case_insensitive) + flag_bits |= HS_FLAG_CASELESS; + for (const std::string & simple_regexp : simple_regexps) { patterns.push_back(simple_regexp.data()); lengths.push_back(simple_regexp.size()); - flags.push_back(HS_FLAG_SINGLEMATCH); + flags.push_back(flag_bits); } hs_database_t * db = nullptr; @@ -380,12 +391,16 @@ RegExpTreeDictionary::RegExpTreeDictionary( const DictionaryStructure & structure_, DictionarySourcePtr source_ptr_, Configuration configuration_, - bool use_vectorscan_) + bool use_vectorscan_, + bool flag_case_insensitive_, + bool flag_dotall_) : IDictionary(id_), structure(structure_), source_ptr(source_ptr_), configuration(configuration_), use_vectorscan(use_vectorscan_), + flag_case_insensitive(flag_case_insensitive_), + flag_dotall(flag_dotall_), logger(&Poco::Logger::get("RegExpTreeDictionary")) { if (auto * ch_source = typeid_cast(source_ptr.get())) @@ -859,7 +874,14 @@ void registerDictionaryRegExpTree(DictionaryFactory & factory) auto context = copyContextAndApplySettingsFromDictionaryConfig(global_context, config, config_prefix); - return std::make_unique(dict_id, dict_struct, std::move(source_ptr), configuration, context->getSettings().regexp_dict_allow_hyperscan); + return std::make_unique( + dict_id, + dict_struct, + std::move(source_ptr), + configuration, + context->getSettings().regexp_dict_allow_hyperscan, + context->getSettings().regexp_dict_flag_case_insensitive, + context->getSettings().regexp_dict_flag_dotall); }; factory.registerLayout("regexp_tree", create_layout, true); diff --git a/src/Dictionaries/RegExpTreeDictionary.h b/src/Dictionaries/RegExpTreeDictionary.h index 7f1d0ee1e88..62008bb5aae 100644 --- a/src/Dictionaries/RegExpTreeDictionary.h +++ b/src/Dictionaries/RegExpTreeDictionary.h @@ -49,7 +49,9 @@ public: const DictionaryStructure & structure_, DictionarySourcePtr source_ptr_, Configuration configuration_, - bool use_vectorscan_); + bool use_vectorscan_, + bool flag_case_insensitive_, + bool flag_dotall_); std::string getTypeName() const override { return name; } @@ -85,7 +87,8 @@ public: std::shared_ptr clone() const override { - return std::make_shared(getDictionaryID(), structure, source_ptr->clone(), configuration, use_vectorscan); + return std::make_shared( + getDictionaryID(), structure, source_ptr->clone(), configuration, use_vectorscan, flag_case_insensitive, flag_dotall); } ColumnUInt8::Ptr hasKeys(const Columns &, const DataTypes &) const override @@ -189,6 +192,8 @@ private: using RegexTreeNodePtr = std::shared_ptr; bool use_vectorscan; + bool flag_case_insensitive; + bool flag_dotall; std::vector simple_regexps; std::vector regexp_ids; diff --git a/tests/queries/0_stateless/02504_regexp_dictionary_yaml_source.reference b/tests/queries/0_stateless/02504_regexp_dictionary_yaml_source.reference index 79871e3716c..de5f14ee1ff 100644 --- a/tests/queries/0_stateless/02504_regexp_dictionary_yaml_source.reference +++ b/tests/queries/0_stateless/02504_regexp_dictionary_yaml_source.reference @@ -27,3 +27,15 @@ GitHub Documentation ['GitHub Documentation','GitHub'] Documentation ['Documentation'] +['foo','(?i)foo','(?-i)foo'] +['(?i)foo'] +['hello.*world','(?i)hello.*world','(?-i)hello.*world'] +[] +['(?i)hello.*world'] +[] +['foo','(?i)foo','(?-i)foo'] +['foo','(?i)foo'] +['hello.*world','(?i)hello.*world','(?-i)hello.*world'] +['hello.*world','(?i)hello.*world','(?-i)hello.*world'] +['hello.*world','(?i)hello.*world'] +['hello.*world','(?i)hello.*world'] diff --git a/tests/queries/0_stateless/02504_regexp_dictionary_yaml_source.sh b/tests/queries/0_stateless/02504_regexp_dictionary_yaml_source.sh index 5e8985406ae..34ea8b9bfbe 100755 --- a/tests/queries/0_stateless/02504_regexp_dictionary_yaml_source.sh +++ b/tests/queries/0_stateless/02504_regexp_dictionary_yaml_source.sh @@ -239,10 +239,66 @@ select dictGet('regexp_dict3', 'tag', '/docs'); select dictGetAll('regexp_dict3', 'tag', '/docs'); " +# Test case-insensitive and dot-all match modes +cat > "$yaml" < Date: Wed, 6 Sep 2023 16:39:45 +0000 Subject: [PATCH 152/433] Fix flaky test test_storage_azure_blob_storage/test.py::test_schema_inference_cache --- .../test_storage_azure_blob_storage/test.py | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-) diff --git a/tests/integration/test_storage_azure_blob_storage/test.py b/tests/integration/test_storage_azure_blob_storage/test.py index 4cde1f9d597..734e9bb899c 100644 --- a/tests/integration/test_storage_azure_blob_storage/test.py +++ b/tests/integration/test_storage_azure_blob_storage/test.py @@ -35,7 +35,7 @@ def cluster(): cluster.shutdown() -def azure_query(node, query, expect_error="false", try_num=10, settings={}): +def azure_query(node, query, expect_error="false", try_num=10, settings={}, query_on_retry = None): for i in range(try_num): try: if expect_error == "true": @@ -62,6 +62,8 @@ def azure_query(node, query, expect_error="false", try_num=10, settings={}): break if not retry or i == try_num - 1: raise Exception(ex) + if query_on_retry is not None: + node.query(query_on_retry) continue @@ -761,8 +763,10 @@ def run_describe_query(instance, file, connection_string): azure_query(instance, query) -def run_count_query(instance, file, connection_string): +def run_count_query(instance, file, connection_string, drop_cache_on_retry = False): query = f"select count() from azureBlobStorage('{connection_string}', 'cont', '{file}', auto, auto, 'x UInt64')" + if drop_cache_on_retry: + return azure_query(node=instance,query=query,query_on_retry="system drop schema cache for azure") return azure_query(instance, query) @@ -872,7 +876,6 @@ def test_schema_inference_cache(cluster): files = "test_cache{0,1,2,3}.jsonl" run_describe_query(node, files, connection_string) - print(node.query("select * from system.schema_inference_cache format Pretty")) check_cache_hits(node, files) node.query(f"system drop schema cache for azure") @@ -957,7 +960,7 @@ def test_schema_inference_cache(cluster): node.query(f"system drop schema cache for azure") check_cache(node, []) - res = run_count_query(node, "test_cache{0,1}.csv", connection_string) + res = run_count_query(node, "test_cache{0,1}.csv", connection_string, True) assert int(res) == 300 check_cache_misses(node, "test_cache{0,1}.csv", 2) From 4368c00052717b9ad6c69bb6e39a4036621bc0e9 Mon Sep 17 00:00:00 2001 From: Alexander Sapin Date: Wed, 6 Sep 2023 18:43:56 +0200 Subject: [PATCH 153/433] Fix logical error in vertical merge + replacing merge tree + optimize cleanup --- .../Algorithms/ReplacingSortedAlgorithm.cpp | 24 +++++++++++++------ .../Algorithms/ReplacingSortedAlgorithm.h | 4 +++- .../Merges/ReplacingSortedTransform.h | 6 +++-- src/Storages/MergeTree/MergeTask.cpp | 14 ++++++----- src/Storages/MergeTree/MergeTask.h | 1 + ...ge_tree_with_block_number_column.reference | 13 ++++++++++ ...ng_merge_tree_with_block_number_column.sql | 23 ++++++++++++++++++ 7 files changed, 69 insertions(+), 16 deletions(-) create mode 100644 tests/queries/0_stateless/02861_replacing_merge_tree_with_block_number_column.reference create mode 100644 tests/queries/0_stateless/02861_replacing_merge_tree_with_block_number_column.sql diff --git a/src/Processors/Merges/Algorithms/ReplacingSortedAlgorithm.cpp b/src/Processors/Merges/Algorithms/ReplacingSortedAlgorithm.cpp index db770de858c..4d2443b1e46 100644 --- a/src/Processors/Merges/Algorithms/ReplacingSortedAlgorithm.cpp +++ b/src/Processors/Merges/Algorithms/ReplacingSortedAlgorithm.cpp @@ -21,10 +21,14 @@ ReplacingSortedAlgorithm::ReplacingSortedAlgorithm( size_t max_block_size_bytes, WriteBuffer * out_row_sources_buf_, bool use_average_block_sizes, - bool cleanup_) + bool cleanup_, + size_t * cleanedup_rows_count_) : IMergingAlgorithmWithSharedChunks(header_, num_inputs, std::move(description_), out_row_sources_buf_, max_row_refs) - , merged_data(header_.cloneEmptyColumns(), use_average_block_sizes, max_block_size_rows, max_block_size_bytes), cleanup(cleanup_) + , merged_data(header_.cloneEmptyColumns(), use_average_block_sizes, max_block_size_rows, max_block_size_bytes) + , cleanup(cleanup_) + , cleanedup_rows_count(cleanedup_rows_count_) { + if (!is_deleted_column.empty()) is_deleted_column_number = header_.getPositionByName(is_deleted_column); if (!version_column.empty()) @@ -74,10 +78,13 @@ IMergingAlgorithm::Status ReplacingSortedAlgorithm::merge() /// Write the data for the previous primary key. if (!selected_row.empty()) { - if (is_deleted_column_number!=-1) + if (is_deleted_column_number != -1) { - if (!(cleanup && assert_cast(*(*selected_row.all_columns)[is_deleted_column_number]).getData()[selected_row.row_num])) + uint8_t value = assert_cast(*(*selected_row.all_columns)[is_deleted_column_number]).getData()[selected_row.row_num]; + if (!cleanup || !value) insertRow(); + else if (cleanedup_rows_count != nullptr) + *cleanedup_rows_count += current_row_sources.size(); } else insertRow(); @@ -91,7 +98,7 @@ IMergingAlgorithm::Status ReplacingSortedAlgorithm::merge() if (out_row_sources_buf) current_row_sources.emplace_back(current.impl->order, true); - if ((is_deleted_column_number!=-1)) + if (is_deleted_column_number != -1) { const UInt8 is_deleted = assert_cast(*current->all_columns[is_deleted_column_number]).getData()[current->getRow()]; if ((is_deleted != 1) && (is_deleted != 0)) @@ -129,10 +136,13 @@ IMergingAlgorithm::Status ReplacingSortedAlgorithm::merge() /// We will write the data for the last primary key. if (!selected_row.empty()) { - if (is_deleted_column_number!=-1) + if (is_deleted_column_number != -1) { - if (!(cleanup && assert_cast(*(*selected_row.all_columns)[is_deleted_column_number]).getData()[selected_row.row_num])) + uint8_t value = assert_cast(*(*selected_row.all_columns)[is_deleted_column_number]).getData()[selected_row.row_num]; + if (!cleanup || !value) insertRow(); + else if (cleanedup_rows_count != nullptr) + *cleanedup_rows_count += current_row_sources.size(); } else insertRow(); diff --git a/src/Processors/Merges/Algorithms/ReplacingSortedAlgorithm.h b/src/Processors/Merges/Algorithms/ReplacingSortedAlgorithm.h index 4d8de55b032..6ee138aca88 100644 --- a/src/Processors/Merges/Algorithms/ReplacingSortedAlgorithm.h +++ b/src/Processors/Merges/Algorithms/ReplacingSortedAlgorithm.h @@ -27,7 +27,8 @@ public: size_t max_block_size_bytes, WriteBuffer * out_row_sources_buf_ = nullptr, bool use_average_block_sizes = false, - bool cleanup = false); + bool cleanup = false, + size_t * cleanedup_rows_count = nullptr); Status merge() override; @@ -37,6 +38,7 @@ private: ssize_t is_deleted_column_number = -1; ssize_t version_column_number = -1; bool cleanup = false; + size_t * cleanedup_rows_count = nullptr; using RowRef = detail::RowRefWithOwnedChunk; static constexpr size_t max_row_refs = 2; /// last, current. diff --git a/src/Processors/Merges/ReplacingSortedTransform.h b/src/Processors/Merges/ReplacingSortedTransform.h index 9cd2f29a862..7e293db1aa8 100644 --- a/src/Processors/Merges/ReplacingSortedTransform.h +++ b/src/Processors/Merges/ReplacingSortedTransform.h @@ -19,7 +19,8 @@ public: size_t max_block_size_bytes, WriteBuffer * out_row_sources_buf_ = nullptr, bool use_average_block_sizes = false, - bool cleanup = false) + bool cleanup = false, + size_t * cleanedup_rows_count = nullptr) : IMergingTransform( num_inputs, header, header, /*have_all_inputs_=*/ true, /*limit_hint_=*/ 0, /*always_read_till_end_=*/ false, header, @@ -31,7 +32,8 @@ public: max_block_size_bytes, out_row_sources_buf_, use_average_block_sizes, - cleanup) + cleanup, + cleanedup_rows_count) { } diff --git a/src/Storages/MergeTree/MergeTask.cpp b/src/Storages/MergeTree/MergeTask.cpp index 8f39c31eae0..df607d36402 100644 --- a/src/Storages/MergeTree/MergeTask.cpp +++ b/src/Storages/MergeTree/MergeTask.cpp @@ -487,6 +487,7 @@ bool MergeTask::VerticalMergeStage::prepareVerticalMergeForAllColumns() const size_t sum_input_rows_exact = global_ctx->merge_list_element_ptr->rows_read; size_t input_rows_filtered = *global_ctx->input_rows_filtered; + size_t cleanedup_rows_count = global_ctx->cleanedup_rows_count; global_ctx->merge_list_element_ptr->columns_written = global_ctx->merging_column_names.size(); global_ctx->merge_list_element_ptr->progress.store(ctx->column_sizes->keyColumnsWeight(), std::memory_order_relaxed); @@ -499,12 +500,13 @@ bool MergeTask::VerticalMergeStage::prepareVerticalMergeForAllColumns() const /// In special case, when there is only one source part, and no rows were skipped, we may have /// skipped writing rows_sources file. Otherwise rows_sources_count must be equal to the total /// number of input rows. - if ((rows_sources_count > 0 || global_ctx->future_part->parts.size() > 1) && sum_input_rows_exact != rows_sources_count + input_rows_filtered) + if ((rows_sources_count > 0 || global_ctx->future_part->parts.size() > 1) + && sum_input_rows_exact != rows_sources_count + input_rows_filtered + cleanedup_rows_count) throw Exception( - ErrorCodes::LOGICAL_ERROR, - "Number of rows in source parts ({}) excluding filtered rows ({}) differs from number " - "of bytes written to rows_sources file ({}). It is a bug.", - sum_input_rows_exact, input_rows_filtered, rows_sources_count); + ErrorCodes::LOGICAL_ERROR, + "Number of rows in source parts ({}) excluding filtered rows ({}) and cleaned up rows ({}) differs from number " + "of bytes written to rows_sources file ({}). It is a bug.", + sum_input_rows_exact, input_rows_filtered, cleanedup_rows_count, rows_sources_count); ctx->rows_sources_read_buf = std::make_unique(ctx->tmp_disk->readFile(fileName(ctx->rows_sources_file->path()))); @@ -975,7 +977,7 @@ void MergeTask::ExecuteAndFinalizeHorizontalPart::createMergedStream() merged_transform = std::make_shared( header, pipes.size(), sort_description, ctx->merging_params.is_deleted_column, ctx->merging_params.version_column, merge_block_size_rows, merge_block_size_bytes, ctx->rows_sources_write_buf.get(), ctx->blocks_are_granules_size, - (data_settings->clean_deleted_rows != CleanDeletedRows::Never) || global_ctx->cleanup); + (data_settings->clean_deleted_rows != CleanDeletedRows::Never) || global_ctx->cleanup, &global_ctx->cleanedup_rows_count); break; case MergeTreeData::MergingParams::Graphite: diff --git a/src/Storages/MergeTree/MergeTask.h b/src/Storages/MergeTree/MergeTask.h index 9b777345c1d..402d3c26e49 100644 --- a/src/Storages/MergeTree/MergeTask.h +++ b/src/Storages/MergeTree/MergeTask.h @@ -145,6 +145,7 @@ private: bool deduplicate{false}; Names deduplicate_by_columns{}; bool cleanup{false}; + size_t cleanedup_rows_count{0}; NamesAndTypesList gathering_columns{}; NamesAndTypesList merging_columns{}; diff --git a/tests/queries/0_stateless/02861_replacing_merge_tree_with_block_number_column.reference b/tests/queries/0_stateless/02861_replacing_merge_tree_with_block_number_column.reference new file mode 100644 index 00000000000..9c9caa22139 --- /dev/null +++ b/tests/queries/0_stateless/02861_replacing_merge_tree_with_block_number_column.reference @@ -0,0 +1,13 @@ +== Only last version remains after OPTIMIZE W/ CLEANUP == +d1 5 0 +d2 1 0 +d3 1 0 +d4 1 0 +d5 1 0 +d6 3 0 +== OPTIMIZE W/ CLEANUP (remove d6) == +d1 5 0 +d2 1 0 +d3 1 0 +d4 1 0 +d5 1 0 diff --git a/tests/queries/0_stateless/02861_replacing_merge_tree_with_block_number_column.sql b/tests/queries/0_stateless/02861_replacing_merge_tree_with_block_number_column.sql new file mode 100644 index 00000000000..7b78e2900e7 --- /dev/null +++ b/tests/queries/0_stateless/02861_replacing_merge_tree_with_block_number_column.sql @@ -0,0 +1,23 @@ +DROP TABLE IF EXISTS test; +CREATE TABLE test (uid String, version UInt32, is_deleted UInt8) ENGINE = ReplacingMergeTree(version, is_deleted) Order by (uid) SETTINGS vertical_merge_algorithm_min_rows_to_activate = 1, + vertical_merge_algorithm_min_columns_to_activate = 0, + min_rows_for_wide_part = 1, + min_bytes_for_wide_part = 1; + +-- Expect d6 to be version=3 is_deleted=false +INSERT INTO test (*) VALUES ('d1', 1, 0), ('d1', 2, 1), ('d1', 3, 0), ('d1', 4, 1), ('d1', 5, 0), ('d2', 1, 0), ('d3', 1, 0), ('d4', 1, 0), ('d5', 1, 0), ('d6', 1, 0), ('d6', 3, 0); +-- Insert previous version of 'd6' but only v=3 is_deleted=false will remain +INSERT INTO test (*) VALUES ('d1', 1, 0), ('d1', 2, 1), ('d1', 3, 0), ('d1', 4, 1), ('d1', 5, 0), ('d2', 1, 0), ('d3', 1, 0), ('d4', 1, 0), ('d5', 1, 0), ('d6', 1, 0), ('d6', 2, 1); +SELECT '== Only last version remains after OPTIMIZE W/ CLEANUP =='; +OPTIMIZE TABLE test FINAL CLEANUP; +select * from test order by uid; + +-- insert d6 v=3 is_deleted=true (timestamp more recent so this version should be the one take into acount) +INSERT INTO test (*) VALUES ('d1', 1, 0), ('d1', 2, 1), ('d1', 3, 0), ('d1', 4, 1), ('d1', 5, 0), ('d2', 1, 0), ('d3', 1, 0), ('d4', 1, 0), ('d5', 1, 0), ('d6', 1, 0), ('d6', 3, 1); + +SELECT '== OPTIMIZE W/ CLEANUP (remove d6) =='; +OPTIMIZE TABLE test FINAL CLEANUP; +-- No d6 anymore +select * from test order by uid; + +DROP TABLE IF EXISTS test; From ce673d6beab8203b95b50feea8862038ab2bee51 Mon Sep 17 00:00:00 2001 From: Arthur Passos Date: Wed, 6 Sep 2023 13:59:39 -0300 Subject: [PATCH 154/433] arrow-parquet account for monotonically increasing offsets across multiple batches --- .../Formats/Impl/ArrowColumnToCHColumn.cpp | 17 ++++++++++++++++- 1 file changed, 16 insertions(+), 1 deletion(-) diff --git a/src/Processors/Formats/Impl/ArrowColumnToCHColumn.cpp b/src/Processors/Formats/Impl/ArrowColumnToCHColumn.cpp index 76f39b07a05..d6ae39b1542 100644 --- a/src/Processors/Formats/Impl/ArrowColumnToCHColumn.cpp +++ b/src/Processors/Formats/Impl/ArrowColumnToCHColumn.cpp @@ -447,10 +447,25 @@ static ColumnPtr readOffsetsFromArrowListColumn(std::shared_ptr 0. + * If they are not monotonically increasing, it'll always be 0. + * Therefore, we subtract the previous offset from the current offset to get the corresponding CH "offset". + * + * This workaround should not be confused with the one above. That workaround deals with multiple chunks and internal representation + * of offsets, not batches. + * */ + auto previous_offset = arrow_offsets.Value(0); + for (int64_t i = 1; i < arrow_offsets.length(); ++i) { auto offset = arrow_offsets.Value(i); - offsets_data.emplace_back(start_offset + offset); + offsets_data.emplace_back(start_offset + offset - previous_offset); } } return offsets_column; From c3fb0ef5d4211c5c09e61776fd6dc0eb503bbb8c Mon Sep 17 00:00:00 2001 From: Kruglov Pavel <48961922+Avogar@users.noreply.github.com> Date: Wed, 6 Sep 2023 19:42:14 +0200 Subject: [PATCH 155/433] Fix style --- tests/integration/test_storage_azure_blob_storage/test.py | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/tests/integration/test_storage_azure_blob_storage/test.py b/tests/integration/test_storage_azure_blob_storage/test.py index 734e9bb899c..b58f0272b7f 100644 --- a/tests/integration/test_storage_azure_blob_storage/test.py +++ b/tests/integration/test_storage_azure_blob_storage/test.py @@ -766,7 +766,12 @@ def run_describe_query(instance, file, connection_string): def run_count_query(instance, file, connection_string, drop_cache_on_retry = False): query = f"select count() from azureBlobStorage('{connection_string}', 'cont', '{file}', auto, auto, 'x UInt64')" if drop_cache_on_retry: - return azure_query(node=instance,query=query,query_on_retry="system drop schema cache for azure") + return azure_query( + node=instance, + query=query, + query_on_retry="system drop schema cache for azure" + ) + return azure_query(instance, query) From ab101eaff8ff36aa31c50bf3c20e2e4e23fe571a Mon Sep 17 00:00:00 2001 From: avogar Date: Wed, 6 Sep 2023 18:21:24 +0000 Subject: [PATCH 156/433] Fix possible error 'URI contains invalid characters' in s3 table function --- src/Storages/StorageS3.cpp | 8 ++++++-- src/TableFunctions/TableFunctionS3.cpp | 9 +++++---- ...esigned_url_and_url_with_special_characters.reference | 0 ..._s3_presigned_url_and_url_with_special_characters.sql | 4 ++++ 4 files changed, 15 insertions(+), 6 deletions(-) create mode 100644 tests/queries/0_stateless/02873_s3_presigned_url_and_url_with_special_characters.reference create mode 100644 tests/queries/0_stateless/02873_s3_presigned_url_and_url_with_special_characters.sql diff --git a/src/Storages/StorageS3.cpp b/src/Storages/StorageS3.cpp index 24b2279bfdc..5078d21141b 100644 --- a/src/Storages/StorageS3.cpp +++ b/src/Storages/StorageS3.cpp @@ -1246,11 +1246,13 @@ void StorageS3::processNamedCollectionResult(StorageS3::Configuration & configur validateNamedCollection(collection, required_configuration_keys, optional_configuration_keys); auto filename = collection.getOrDefault("filename", ""); + String url; if (!filename.empty()) - configuration.url = S3::URI(std::filesystem::path(collection.get("url")) / filename); + url = std::filesystem::path(collection.get("url")) / filename; else - configuration.url = S3::URI(collection.get("url")); + url = collection.get("url"); + configuration.url = S3::URI(url); configuration.auth_settings.access_key_id = collection.getOrDefault("access_key_id", ""); configuration.auth_settings.secret_access_key = collection.getOrDefault("secret_access_key", ""); configuration.auth_settings.use_environment_credentials = collection.getOrDefault("use_environment_credentials", 1); @@ -1258,6 +1260,8 @@ void StorageS3::processNamedCollectionResult(StorageS3::Configuration & configur configuration.auth_settings.expiration_window_seconds = collection.getOrDefault("expiration_window_seconds", S3::DEFAULT_EXPIRATION_WINDOW_SECONDS); configuration.format = collection.getOrDefault("format", configuration.format); + if (configuration.format == "auto") + configuration.format = FormatFactory::instance().getFormatFromFileName(Poco::URI(url).getPath(), true); configuration.compression_method = collection.getOrDefault("compression_method", collection.getOrDefault("compression", "auto")); configuration.structure = collection.getOrDefault("structure", "auto"); diff --git a/src/TableFunctions/TableFunctionS3.cpp b/src/TableFunctions/TableFunctionS3.cpp index 0ae3a19dc7f..94ce29f1116 100644 --- a/src/TableFunctions/TableFunctionS3.cpp +++ b/src/TableFunctions/TableFunctionS3.cpp @@ -152,7 +152,8 @@ void TableFunctionS3::parseArgumentsImpl(ASTs & args, const ContextPtr & context } /// This argument is always the first - configuration.url = S3::URI(checkAndGetLiteralArgument(args[0], "url")); + String url = checkAndGetLiteralArgument(args[0], "url"); + configuration.url = S3::URI(url); if (args_to_idx.contains("format")) { @@ -176,12 +177,12 @@ void TableFunctionS3::parseArgumentsImpl(ASTs & args, const ContextPtr & context configuration.auth_settings.secret_access_key = checkAndGetLiteralArgument(args[args_to_idx["secret_access_key"]], "secret_access_key"); configuration.auth_settings.no_sign_request = no_sign_request; + + if (configuration.format == "auto") + configuration.format = FormatFactory::instance().getFormatFromFileName(Poco::URI(url).getPath(), true); } configuration.keys = {configuration.url.key}; - - if (configuration.format == "auto") - configuration.format = FormatFactory::instance().getFormatFromFileName(Poco::URI(configuration.url.uri.getPath()).getPath(), true); } void TableFunctionS3::parseArguments(const ASTPtr & ast_function, ContextPtr context) diff --git a/tests/queries/0_stateless/02873_s3_presigned_url_and_url_with_special_characters.reference b/tests/queries/0_stateless/02873_s3_presigned_url_and_url_with_special_characters.reference new file mode 100644 index 00000000000..e69de29bb2d diff --git a/tests/queries/0_stateless/02873_s3_presigned_url_and_url_with_special_characters.sql b/tests/queries/0_stateless/02873_s3_presigned_url_and_url_with_special_characters.sql new file mode 100644 index 00000000000..e6954fd9cbf --- /dev/null +++ b/tests/queries/0_stateless/02873_s3_presigned_url_and_url_with_special_characters.sql @@ -0,0 +1,4 @@ +select * from s3('https://datasets-documentation.s3.eu-west-3.amazonaws.com/MyPrefix/BU%20-%20UNIT%20-%201/*.parquet'); -- { serverError CANNOT_EXTRACT_TABLE_STRUCTURE } + +select * from s3('https://datasets-documentation.s3.eu-west-3.amazonaws.com/MyPrefix/*.parquet?some_tocken=ABCD'); -- { serverError CANNOT_EXTRACT_TABLE_STRUCTURE } + From 664efaf3fa3bcd7029dcc71558e5bdbc06fcff7f Mon Sep 17 00:00:00 2001 From: serxa Date: Wed, 6 Sep 2023 18:35:40 +0000 Subject: [PATCH 157/433] add test --- tests/integration/test_scheduler/__init__.py | 0 .../test_scheduler/configs/scheduler.xml | 62 ++++++++++++ tests/integration/test_scheduler/test.py | 97 +++++++++++++++++++ 3 files changed, 159 insertions(+) create mode 100644 tests/integration/test_scheduler/__init__.py create mode 100644 tests/integration/test_scheduler/configs/scheduler.xml create mode 100644 tests/integration/test_scheduler/test.py diff --git a/tests/integration/test_scheduler/__init__.py b/tests/integration/test_scheduler/__init__.py new file mode 100644 index 00000000000..e69de29bb2d diff --git a/tests/integration/test_scheduler/configs/scheduler.xml b/tests/integration/test_scheduler/configs/scheduler.xml new file mode 100644 index 00000000000..fe01cb8d662 --- /dev/null +++ b/tests/integration/test_scheduler/configs/scheduler.xml @@ -0,0 +1,62 @@ + + + + + s3 + http://minio1:9001/root/data/ + minio + minio123 + 33554432 + 10 + 10 + network_read + network_write + + + + + +
+ s3 +
+ + + + + + + inflight_limit1000000 + priority + fifo0 + fair1 + fifo9 + fifo1 + + + inflight_limit1000000 + priority + fifo0 + fair1 + fifo9 + fifo1 + + + + + /prio/admin + /prio/admin + + + /prio/fair/prod + /prio/fair/prod + + + /prio/fair/dev + /prio/fair/dev + + + /prio/fair/dev + /prio/fair/dev + + + diff --git a/tests/integration/test_scheduler/test.py b/tests/integration/test_scheduler/test.py new file mode 100644 index 00000000000..7900e5b649e --- /dev/null +++ b/tests/integration/test_scheduler/test.py @@ -0,0 +1,97 @@ +# pylint: disable=unused-argument +# pylint: disable=redefined-outer-name +# pylint: disable=line-too-long + +import time +import threading +import pytest + +from helpers.cluster import ClickHouseCluster + +cluster = ClickHouseCluster(__file__) + +node = cluster.add_instance( + "node", + stay_alive=True, + main_configs=["configs/scheduler.xml"], + with_minio=True, +) + + +@pytest.fixture(scope="module", autouse=True) +def start_cluster(): + try: + cluster.start() + yield + finally: + cluster.shutdown() + + +def test_s3_disk(): + node.query( + f""" + drop table if exists data; + create table data (key UInt64 CODEC(NONE)) engine=MergeTree() order by tuple() settings min_bytes_for_wide_part=1e9, storage_policy='s3'; + """ + ) + + def write_query(workload): + try: + node.query(f"insert into data select * from numbers(1e5) settings workload='{workload}'") + except QueryRuntimeException: + pass + thread1 = threading.Thread(target=write_query, args=["development"]) + thread2 = threading.Thread(target=write_query, args=["production"]) + thread3 = threading.Thread(target=write_query, args=["admin"]) + + thread1.start() + thread2.start() + thread3.start() + + thread3.join() + thread2.join() + thread1.join() + + assert ( + node.query(f"select dequeued_requests>0 from system.scheduler where resource='network_write' and path='/prio/admin'") + == "1\n" + ) + assert ( + node.query(f"select dequeued_requests>0 from system.scheduler where resource='network_write' and path='/prio/fair/dev'") + == "1\n" + ) + assert ( + node.query(f"select dequeued_requests>0 from system.scheduler where resource='network_write' and path='/prio/fair/prod'") + == "1\n" + ) + + def read_query(workload): + try: + node.query(f"select sum(key*key) from data settings workload='{workload}'") + except QueryRuntimeException: + pass + thread1 = threading.Thread(target=read_query, args=["development"]) + thread2 = threading.Thread(target=read_query, args=["production"]) + thread3 = threading.Thread(target=read_query, args=["admin"]) + + thread1.start() + thread2.start() + thread3.start() + + thread3.join() + thread2.join() + thread1.join() + + assert ( + node.query(f"select dequeued_requests>0 from system.scheduler where resource='network_read' and path='/prio/admin'") + == "1\n" + ) + assert ( + node.query(f"select dequeued_requests>0 from system.scheduler where resource='network_read' and path='/prio/fair/dev'") + == "1\n" + ) + assert ( + node.query(f"select dequeued_requests>0 from system.scheduler where resource='network_read' and path='/prio/fair/prod'") + == "1\n" + ) + From cd7c57f5d47ab46f3120faf115554deb9d486af8 Mon Sep 17 00:00:00 2001 From: robot-clickhouse Date: Wed, 6 Sep 2023 19:00:05 +0000 Subject: [PATCH 158/433] Automatic style fix --- tests/integration/test_scheduler/test.py | 31 ++++++++++++++++++------ 1 file changed, 23 insertions(+), 8 deletions(-) diff --git a/tests/integration/test_scheduler/test.py b/tests/integration/test_scheduler/test.py index 7900e5b649e..e6def99c076 100644 --- a/tests/integration/test_scheduler/test.py +++ b/tests/integration/test_scheduler/test.py @@ -37,9 +37,12 @@ def test_s3_disk(): def write_query(workload): try: - node.query(f"insert into data select * from numbers(1e5) settings workload='{workload}'") + node.query( + f"insert into data select * from numbers(1e5) settings workload='{workload}'" + ) except QueryRuntimeException: pass + thread1 = threading.Thread(target=write_query, args=["development"]) thread2 = threading.Thread(target=write_query, args=["production"]) thread3 = threading.Thread(target=write_query, args=["admin"]) @@ -53,15 +56,21 @@ def test_s3_disk(): thread1.join() assert ( - node.query(f"select dequeued_requests>0 from system.scheduler where resource='network_write' and path='/prio/admin'") + node.query( + f"select dequeued_requests>0 from system.scheduler where resource='network_write' and path='/prio/admin'" + ) == "1\n" ) assert ( - node.query(f"select dequeued_requests>0 from system.scheduler where resource='network_write' and path='/prio/fair/dev'") + node.query( + f"select dequeued_requests>0 from system.scheduler where resource='network_write' and path='/prio/fair/dev'" + ) == "1\n" ) assert ( - node.query(f"select dequeued_requests>0 from system.scheduler where resource='network_write' and path='/prio/fair/prod'") + node.query( + f"select dequeued_requests>0 from system.scheduler where resource='network_write' and path='/prio/fair/prod'" + ) == "1\n" ) @@ -70,6 +79,7 @@ def test_s3_disk(): node.query(f"select sum(key*key) from data settings workload='{workload}'") except QueryRuntimeException: pass + thread1 = threading.Thread(target=read_query, args=["development"]) thread2 = threading.Thread(target=read_query, args=["production"]) thread3 = threading.Thread(target=read_query, args=["admin"]) @@ -83,15 +93,20 @@ def test_s3_disk(): thread1.join() assert ( - node.query(f"select dequeued_requests>0 from system.scheduler where resource='network_read' and path='/prio/admin'") + node.query( + f"select dequeued_requests>0 from system.scheduler where resource='network_read' and path='/prio/admin'" + ) == "1\n" ) assert ( - node.query(f"select dequeued_requests>0 from system.scheduler where resource='network_read' and path='/prio/fair/dev'") + node.query( + f"select dequeued_requests>0 from system.scheduler where resource='network_read' and path='/prio/fair/dev'" + ) == "1\n" ) assert ( - node.query(f"select dequeued_requests>0 from system.scheduler where resource='network_read' and path='/prio/fair/prod'") + node.query( + f"select dequeued_requests>0 from system.scheduler where resource='network_read' and path='/prio/fair/prod'" + ) == "1\n" ) - From e11e5fbe2e8e50b298a05f6c8c8c8d2e9b0ffbef Mon Sep 17 00:00:00 2001 From: Kruglov Pavel <48961922+Avogar@users.noreply.github.com> Date: Wed, 6 Sep 2023 21:10:30 +0200 Subject: [PATCH 159/433] Fix fasttest --- .../02873_s3_presigned_url_and_url_with_special_characters.sql | 2 ++ 1 file changed, 2 insertions(+) diff --git a/tests/queries/0_stateless/02873_s3_presigned_url_and_url_with_special_characters.sql b/tests/queries/0_stateless/02873_s3_presigned_url_and_url_with_special_characters.sql index e6954fd9cbf..da76a5cb88f 100644 --- a/tests/queries/0_stateless/02873_s3_presigned_url_and_url_with_special_characters.sql +++ b/tests/queries/0_stateless/02873_s3_presigned_url_and_url_with_special_characters.sql @@ -1,3 +1,5 @@ +-- Tags: no-fasttest + select * from s3('https://datasets-documentation.s3.eu-west-3.amazonaws.com/MyPrefix/BU%20-%20UNIT%20-%201/*.parquet'); -- { serverError CANNOT_EXTRACT_TABLE_STRUCTURE } select * from s3('https://datasets-documentation.s3.eu-west-3.amazonaws.com/MyPrefix/*.parquet?some_tocken=ABCD'); -- { serverError CANNOT_EXTRACT_TABLE_STRUCTURE } From 77697211cd2cf565fd248db9842360bde40d5e7f Mon Sep 17 00:00:00 2001 From: serxa Date: Wed, 6 Sep 2023 19:30:39 +0000 Subject: [PATCH 160/433] add docs for `system.scheduler` + small fix --- docs/en/operations/system-tables/scheduler.md | 64 +++++++++++++++++++ docs/en/operations/workload-scheduling.md | 10 ++- src/IO/ISchedulerNode.h | 2 +- .../System/StorageSystemScheduler.cpp | 2 +- 4 files changed, 73 insertions(+), 5 deletions(-) create mode 100644 docs/en/operations/system-tables/scheduler.md diff --git a/docs/en/operations/system-tables/scheduler.md b/docs/en/operations/system-tables/scheduler.md new file mode 100644 index 00000000000..168bfe708c4 --- /dev/null +++ b/docs/en/operations/system-tables/scheduler.md @@ -0,0 +1,64 @@ +--- +slug: /en/operations/system-tables/scheduler +--- +# scheduler + +Contains information and status for [scheduling nodes](/docs/en/operations/workload-scheduling.md/#hierarchy) residing on the local server. +This table can be used for monitoring. The table contains a row for every scheduling node. + +Example: + +``` sql +SELECT * +FROM system.scheduler +WHERE resource = 'network_read' AND path = '/prio/fair/prod' +FORMAT Vertical +``` + +``` text +Row 1: +────── +resource: network_read +path: /prio/fair/prod +type: fifo +weight: 5 +priority: 0 +is_active: 0 +active_children: 0 +dequeued_requests: 67 +dequeued_cost: 4692272 +busy_periods: 63 +vruntime: 938454.1999999989 +system_vruntime: ᴺᵁᴸᴸ +queue_length: 0 +queue_cost: 0 +budget: -60524 +is_satisfied: ᴺᵁᴸᴸ +inflight_requests: ᴺᵁᴸᴸ +inflight_cost: ᴺᵁᴸᴸ +max_requests: ᴺᵁᴸᴸ +max_cost: ᴺᵁᴸᴸ +``` + +Columns: + +- `resource` (`String`) - Resource name +- `path` (`String`) - Path to a scheduling node within this resource scheduling hierarchy +- `type` (`String`) - Type of a scheduling node. +- `weight` (`Float64`) - Weight of a node, used by a parent node of `fair`` type. +- `priority` (`Int64`) - Priority of a node, used by a parent node of 'priority' type (Lower value means higher priority). +- `is_active` (`UInt8`) - Whether this node is currently active - has resource requests to be dequeued and constraints satisfied. +- `active_children` (`UInt64`) - The number of children in active state. +- `dequeued_requests` (`UInt64`) - The total number of resource requests dequeued from this node. +- `dequeued_cost` (`UInt64`) - The sum of costs (e.g. size in bytes) of all requests dequeued from this node. +- `busy_periods` (`UInt64`) - The total number of deactivations of this node. +- `vruntime` (`Nullable(Float64)`) - For children of `fair` nodes only. Virtual runtime of a node used by SFQ algorithm to select the next child to process in a max-min fair manner. +- `system_vruntime` (`Nullable(Float64)`) - For `fair` nodes only. Virtual runtime showing `vruntime` of the last processed resource request. Used during child activation as the new value of `vruntime`. +- `queue_length` (`Nullable(UInt64)`) - For `fifo` nodes only. Current number of resource requests residing in the queue. +- `queue_cost` (`Nullable(UInt64)`) - For `fifo` nodes only. Sum of costs (e.g. size in bytes) of all requests residing in the queue. +- `budget` (`Nullable(Int64)`) - For `fifo` nodes only. The number of available "cost units" for new resource requests. Can appear in case of discrepancy of estimated and real costs of resource requests (e.g. after read/write failure) +- `is_satisfied` (`Nullable(UInt8)`) - For constraint nodes only (e.g. `inflight_limit`). Equals `1` if all the constraint of this node are satisfied. +- `inflight_requests` (`Nullable(Int64)`) - For `inflight_limit` nodes only. The number of resource requests dequeued from this node, that are currently in consumption state. +- `inflight_cost` (`Nullable(Int64)`) - For `inflight_limit` nodes only. The sum of costs (e.g. bytes) of all resource requests dequeued from this node, that are currently in consumption state. +- `max_requests` (`Nullable(Int64)`) - For `inflight_limit` nodes only. Upper limit for `inflight_requests` leading to constraint violation. +- `max_cost` (`Nullable(Int64)`) - For `inflight_limit` nodes only. Upper limit for `inflight_cost` leading to constraint violation. diff --git a/docs/en/operations/workload-scheduling.md b/docs/en/operations/workload-scheduling.md index 1761f99b469..c6529d5b81f 100644 --- a/docs/en/operations/workload-scheduling.md +++ b/docs/en/operations/workload-scheduling.md @@ -11,7 +11,7 @@ When ClickHouse execute multiple queries simultaneously, they may be using share Currently only remote disk IO can be scheduled using described method. For CPU scheduling see settings about thread pools and [`concurrent_threads_soft_limit_num`](server-configuration-parameters/settings.md#concurrent_threads_soft_limit_num). For flexible memory limits see [Memory overcommit](settings/memory-overcommit.md) ::: -## Disk configuration +## Disk configuration {#disk-config} To enable IO scheduling for a specific disk, you have to specify `read_resource` and/or `write_resource` in storage configuration. It says ClickHouse what resource should be used for every read and write requests with given disk. Read and write resource can refer to the same resource name, which is useful for local SSDs or HDDs. Multiple different disks also can refer to the same resource, which is useful for remote disks: if you want to be able to allow fair division of network bandwidth between e.g. "production" and "development" workloads. @@ -54,7 +54,7 @@ SELECT count() FROM my_table WHERE value = 42 SETTINGS workload = 'production' SELECT count() FROM my_table WHERE value = 13 SETTINGS workload = 'development' ``` -## Resource scheduling hierarchy +## Resource scheduling hierarchy {#hierarchy} From the standpoint of scheduling subsystem a resource represents a hierarchy of scheduling nodes. @@ -124,7 +124,7 @@ The following example shows how to define IO scheduling hierarchies shown in the ``` -## Workload classifiers +## Workload classifiers {#classifiers} Workload classifiers are used to define mapping from `workload` specified by a query into leaf-queues that should be used for specific resources. At the moment, workload classification is simple: only static mapping is available. @@ -147,3 +147,7 @@ Example: ``` + + +## See also + - [system.scheduler](/docs/en/operations/system-tables/scheduler.md) diff --git a/src/IO/ISchedulerNode.h b/src/IO/ISchedulerNode.h index f7fc9257128..644cc708bca 100644 --- a/src/IO/ISchedulerNode.h +++ b/src/IO/ISchedulerNode.h @@ -232,7 +232,7 @@ public: /// Introspection std::atomic dequeued_requests{0}; - std::atomic dequeued_cost{0}; + std::atomic dequeued_cost{0}; std::atomic busy_periods{0}; }; diff --git a/src/Storages/System/StorageSystemScheduler.cpp b/src/Storages/System/StorageSystemScheduler.cpp index ca0a138559e..9f4307fca3a 100644 --- a/src/Storages/System/StorageSystemScheduler.cpp +++ b/src/Storages/System/StorageSystemScheduler.cpp @@ -28,7 +28,7 @@ NamesAndTypesList StorageSystemScheduler::getNamesAndTypes() {"is_active", std::make_shared()}, {"active_children", std::make_shared()}, {"dequeued_requests", std::make_shared()}, - {"dequeued_cost", std::make_shared()}, + {"dequeued_cost", std::make_shared()}, {"busy_periods", std::make_shared()}, {"vruntime", std::make_shared(std::make_shared())}, {"system_vruntime", std::make_shared(std::make_shared())}, From d372776de62239ba6919e355e5d0baebf359f347 Mon Sep 17 00:00:00 2001 From: Robert Schulze Date: Wed, 6 Sep 2023 19:46:52 +0000 Subject: [PATCH 161/433] Fix more tests --- ...d_optimize_skip_select_on_unused_shards.reference | 12 ++++++------ ...p_select_on_unused_shards_with_prewhere.reference | 12 ++++++------ 2 files changed, 12 insertions(+), 12 deletions(-) diff --git a/tests/queries/0_stateless/00754_distributed_optimize_skip_select_on_unused_shards.reference b/tests/queries/0_stateless/00754_distributed_optimize_skip_select_on_unused_shards.reference index aa509893230..add8c239ade 100644 --- a/tests/queries/0_stateless/00754_distributed_optimize_skip_select_on_unused_shards.reference +++ b/tests/queries/0_stateless/00754_distributed_optimize_skip_select_on_unused_shards.reference @@ -1,7 +1,7 @@ OK OK 1 -FAIL +OK 0 4 2 @@ -9,8 +9,8 @@ FAIL 1 1 4 -FAIL -FAIL -FAIL -FAIL -FAIL +OK +OK +OK +OK +OK diff --git a/tests/queries/0_stateless/00754_distributed_optimize_skip_select_on_unused_shards_with_prewhere.reference b/tests/queries/0_stateless/00754_distributed_optimize_skip_select_on_unused_shards_with_prewhere.reference index 611f0fd2585..4c66ccfd2a2 100644 --- a/tests/queries/0_stateless/00754_distributed_optimize_skip_select_on_unused_shards_with_prewhere.reference +++ b/tests/queries/0_stateless/00754_distributed_optimize_skip_select_on_unused_shards_with_prewhere.reference @@ -1,15 +1,15 @@ OK OK 1 -FAIL +OK 0 1 4 4 2 4 -FAIL -FAIL -FAIL -FAIL -FAIL +OK +OK +OK +OK +OK From a8027be612986b060b8bd285ac503c9d2c1e08d6 Mon Sep 17 00:00:00 2001 From: Arthur Passos Date: Wed, 6 Sep 2023 17:25:27 -0300 Subject: [PATCH 162/433] fix --- .../Formats/Impl/ArrowColumnToCHColumn.cpp | 24 +++++-------------- 1 file changed, 6 insertions(+), 18 deletions(-) diff --git a/src/Processors/Formats/Impl/ArrowColumnToCHColumn.cpp b/src/Processors/Formats/Impl/ArrowColumnToCHColumn.cpp index d6ae39b1542..fc522cd7569 100644 --- a/src/Processors/Formats/Impl/ArrowColumnToCHColumn.cpp +++ b/src/Processors/Formats/Impl/ArrowColumnToCHColumn.cpp @@ -427,26 +427,12 @@ static ColumnPtr readOffsetsFromArrowListColumn(std::shared_ptr &>(*offsets_column).getData(); offsets_data.reserve(arrow_column->length()); - uint64_t start_offset = 0u; - for (int chunk_i = 0, num_chunks = arrow_column->num_chunks(); chunk_i < num_chunks; ++chunk_i) { arrow::ListArray & list_chunk = dynamic_cast(*(arrow_column->chunk(chunk_i))); auto arrow_offsets_array = list_chunk.offsets(); auto & arrow_offsets = dynamic_cast(*arrow_offsets_array); - /* - * It seems like arrow::ListArray::values() (nested column data) might or might not be shared across chunks. - * When it is shared, the offsets will be monotonically increasing. Otherwise, the offsets will be zero based. - * In order to account for both cases, the starting offset is updated whenever a zero-based offset is found. - * More info can be found in: https://lists.apache.org/thread/rrwfb9zo2dc58dhd9rblf20xd7wmy7jm and - * https://github.com/ClickHouse/ClickHouse/pull/43297 - * */ - if (list_chunk.offset() == 0) - { - start_offset = offsets_data.back(); - } - /* * CH uses element size as "offsets", while arrow uses actual offsets as offsets. * That's why CH usually starts reading offsets with i=1 and i=0 is ignored. @@ -457,15 +443,17 @@ static ColumnPtr readOffsetsFromArrowListColumn(std::shared_ptr Date: Wed, 6 Sep 2023 17:26:34 -0300 Subject: [PATCH 163/433] docs --- src/Processors/Formats/Impl/ArrowColumnToCHColumn.cpp | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/src/Processors/Formats/Impl/ArrowColumnToCHColumn.cpp b/src/Processors/Formats/Impl/ArrowColumnToCHColumn.cpp index fc522cd7569..2a9892b7219 100644 --- a/src/Processors/Formats/Impl/ArrowColumnToCHColumn.cpp +++ b/src/Processors/Formats/Impl/ArrowColumnToCHColumn.cpp @@ -444,7 +444,8 @@ static ColumnPtr readOffsetsFromArrowListColumn(std::shared_ptr Date: Wed, 6 Sep 2023 22:27:41 +0000 Subject: [PATCH 164/433] Disable prefer_localhost_replica in tests --- .../test.py | 19 +++++++++++-------- 1 file changed, 11 insertions(+), 8 deletions(-) diff --git a/tests/integration/test_parallel_replicas_over_distributed/test.py b/tests/integration/test_parallel_replicas_over_distributed/test.py index 6525901d366..c73d52b6326 100644 --- a/tests/integration/test_parallel_replicas_over_distributed/test.py +++ b/tests/integration/test_parallel_replicas_over_distributed/test.py @@ -107,21 +107,24 @@ def create_tables(cluster, table_name): pytest.param("test_single_shard_multiple_replicas", 3, 0), pytest.param("test_single_shard_multiple_replicas", 4, 0), pytest.param("test_single_shard_multiple_replicas", 10, 0), + # prefer_localhost_replica=1 - pytest.param("test_single_shard_multiple_replicas", 2, 1), - pytest.param("test_single_shard_multiple_replicas", 3, 1), - pytest.param("test_single_shard_multiple_replicas", 4, 1), - pytest.param("test_single_shard_multiple_replicas", 10, 1), + # pytest.param("test_single_shard_multiple_replicas", 2, 1), + # pytest.param("test_single_shard_multiple_replicas", 3, 1), + # pytest.param("test_single_shard_multiple_replicas", 4, 1), + # pytest.param("test_single_shard_multiple_replicas", 10, 1), + # prefer_localhost_replica=0 pytest.param("test_multiple_shards_multiple_replicas", 2, 0), pytest.param("test_multiple_shards_multiple_replicas", 3, 0), pytest.param("test_multiple_shards_multiple_replicas", 4, 0), pytest.param("test_multiple_shards_multiple_replicas", 10, 0), + # prefer_localhost_replica=1 - pytest.param("test_multiple_shards_multiple_replicas", 2, 1), - pytest.param("test_multiple_shards_multiple_replicas", 3, 1), - pytest.param("test_multiple_shards_multiple_replicas", 4, 1), - pytest.param("test_multiple_shards_multiple_replicas", 10, 1), + # pytest.param("test_multiple_shards_multiple_replicas", 2, 1), + # pytest.param("test_multiple_shards_multiple_replicas", 3, 1), + # pytest.param("test_multiple_shards_multiple_replicas", 4, 1), + # pytest.param("test_multiple_shards_multiple_replicas", 10, 1), ], ) def test_parallel_replicas_over_distributed( From 6404768aa4b691ba546f7e984abd39e08a1e311f Mon Sep 17 00:00:00 2001 From: robot-clickhouse Date: Wed, 6 Sep 2023 22:45:17 +0000 Subject: [PATCH 165/433] Automatic style fix --- .../test_parallel_replicas_over_distributed/test.py | 3 --- 1 file changed, 3 deletions(-) diff --git a/tests/integration/test_parallel_replicas_over_distributed/test.py b/tests/integration/test_parallel_replicas_over_distributed/test.py index c73d52b6326..1df2126a4ff 100644 --- a/tests/integration/test_parallel_replicas_over_distributed/test.py +++ b/tests/integration/test_parallel_replicas_over_distributed/test.py @@ -107,19 +107,16 @@ def create_tables(cluster, table_name): pytest.param("test_single_shard_multiple_replicas", 3, 0), pytest.param("test_single_shard_multiple_replicas", 4, 0), pytest.param("test_single_shard_multiple_replicas", 10, 0), - # prefer_localhost_replica=1 # pytest.param("test_single_shard_multiple_replicas", 2, 1), # pytest.param("test_single_shard_multiple_replicas", 3, 1), # pytest.param("test_single_shard_multiple_replicas", 4, 1), # pytest.param("test_single_shard_multiple_replicas", 10, 1), - # prefer_localhost_replica=0 pytest.param("test_multiple_shards_multiple_replicas", 2, 0), pytest.param("test_multiple_shards_multiple_replicas", 3, 0), pytest.param("test_multiple_shards_multiple_replicas", 4, 0), pytest.param("test_multiple_shards_multiple_replicas", 10, 0), - # prefer_localhost_replica=1 # pytest.param("test_multiple_shards_multiple_replicas", 2, 1), # pytest.param("test_multiple_shards_multiple_replicas", 3, 1), From 7ab56a13b8cfb0298df35ce3017ce032843ce6ad Mon Sep 17 00:00:00 2001 From: Igor Nikonov Date: Wed, 6 Sep 2023 22:49:28 +0000 Subject: [PATCH 166/433] Polishing --- tests/ci/fast_test_check.py | 12 ++++-------- 1 file changed, 4 insertions(+), 8 deletions(-) diff --git a/tests/ci/fast_test_check.py b/tests/ci/fast_test_check.py index 97e4f2a8713..96eb144da2c 100644 --- a/tests/ci/fast_test_check.py +++ b/tests/ci/fast_test_check.py @@ -21,6 +21,7 @@ from commit_status_helper import ( get_commit, post_commit_status, update_mergeable_check, + format_description, ) from docker_pull_helper import get_image_with_version from env_helper import S3_BUILDS_BUCKET, TEMP_PATH @@ -185,15 +186,10 @@ def main(): state, description, test_results, additional_logs = process_results(output_path) if timeout_expired is not None: - name = f"Timeout expired for process execution" - if len(name) > 140: - name = name[:136] - name += " ..." - - test_results.append(TestResult(name, "FAIL", timeout)) - + test_result_name = "Timeout for fast test check is expired" + test_results.append(TestResult(test_result_name, "FAIL", timeout)) state = "failure" - description = test_results[-1].name + description = format_description(test_result_name) ch_helper = ClickHouseHelper() s3_path_prefix = os.path.join( From ef2350cfc7e8b6727cbe7e08f3787ee0669bf736 Mon Sep 17 00:00:00 2001 From: pufit Date: Wed, 6 Sep 2023 20:36:39 -0400 Subject: [PATCH 167/433] Implementing `exists`, `sync` commands. Better tests --- .../utilities/clickhouse-keeper-client.md | 5 + programs/keeper-client/Commands.cpp | 30 ++++ programs/keeper-client/Commands.h | 27 ++- programs/keeper-client/KeeperClient.cpp | 37 +++- programs/keeper-client/KeeperClient.h | 3 + tests/integration/helpers/keeper_utils.py | 164 +++++++++++++++++- tests/integration/test_keeper_client/test.py | 147 +++++++--------- .../test_keeper_reconfig_add/test.py | 68 ++++---- .../test_keeper_reconfig_remove/test.py | 77 ++++---- .../test_keeper_reconfig_remove_many/test.py | 71 +++----- .../test.py | 51 +++--- .../__init__.py | 0 .../configs/keeper1.xml | 35 ---- .../configs/keeper2.xml | 35 ---- .../configs/keeper3.xml | 35 ---- .../configs/keeper4.xml | 21 --- .../test.py | 43 ----- 17 files changed, 430 insertions(+), 419 deletions(-) delete mode 100644 tests/integration/test_keeper_reconfig_replace_leader_in_one_command/__init__.py delete mode 100644 tests/integration/test_keeper_reconfig_replace_leader_in_one_command/configs/keeper1.xml delete mode 100644 tests/integration/test_keeper_reconfig_replace_leader_in_one_command/configs/keeper2.xml delete mode 100644 tests/integration/test_keeper_reconfig_replace_leader_in_one_command/configs/keeper3.xml delete mode 100644 tests/integration/test_keeper_reconfig_replace_leader_in_one_command/configs/keeper4.xml delete mode 100644 tests/integration/test_keeper_reconfig_replace_leader_in_one_command/test.py diff --git a/docs/en/operations/utilities/clickhouse-keeper-client.md b/docs/en/operations/utilities/clickhouse-keeper-client.md index 2140d22b620..9f24ba9587b 100644 --- a/docs/en/operations/utilities/clickhouse-keeper-client.md +++ b/docs/en/operations/utilities/clickhouse-keeper-client.md @@ -16,6 +16,8 @@ A client application to interact with clickhouse-keeper by its native protocol. - `--session-timeout=TIMEOUT` — Set session timeout in seconds. Default value: 10s. - `--operation-timeout=TIMEOUT` — Set operation timeout in seconds. Default value: 10s. - `--history-file=FILE_PATH` — Set path of history file. Default value: `~/.keeper-client-history`. +- `--log-level=LEVEL` — Set log level. Default value: `information`. +- `--no-confirmation` — If set, will not require a confirmation on several commands. Default value `false` for interactive and `true` for query - `--help` — Shows the help message. ## Example {#clickhouse-keeper-client-example} @@ -44,6 +46,7 @@ keeper foo bar - `ls [path]` -- Lists the nodes for the given path (default: cwd) - `cd [path]` -- Change the working path (default `.`) +- `exists ` -- Returns `1` if node exists, `0` otherwise - `set [version]` -- Updates the node's value. Only update if version matches (default: -1) - `create [mode]` -- Creates new node with the set value - `touch ` -- Creates new node with an empty string as value. Doesn't throw an exception if the node already exists @@ -56,3 +59,5 @@ keeper foo bar - `find_super_nodes [path]` -- Finds nodes with number of children larger than some threshold for the given path (default `.`) - `delete_stale_backups` -- Deletes ClickHouse nodes used for backups that are now inactive - `find_big_family [path] [n]` -- Returns the top n nodes with the biggest family in the subtree (default path = `.` and n = 10) +- `sync ` -- Synchronizes node between processes and leader +- `reconfig "" [version]` -- Reconfigures Keeper cluster. See https://clickhouse.com/docs/en/guides/sre/keeper/clickhouse-keeper#reconfiguration diff --git a/programs/keeper-client/Commands.cpp b/programs/keeper-client/Commands.cpp index e265d7f4858..65bc719f1be 100644 --- a/programs/keeper-client/Commands.cpp +++ b/programs/keeper-client/Commands.cpp @@ -158,6 +158,21 @@ void GetCommand::execute(const ASTKeeperQuery * query, KeeperClient * client) co std::cout << client->zookeeper->get(client->getAbsolutePath(query->args[0].safeGet())) << "\n"; } +bool ExistsCommand::parse(IParser::Pos & pos, std::shared_ptr & node, DB::Expected & expected) const +{ + String path; + if (!parseKeeperPath(pos, expected, path)) + return false; + node->args.push_back(std::move(path)); + + return true; +} + +void ExistsCommand::execute(const DB::ASTKeeperQuery * query, DB::KeeperClient * client) const +{ + std::cout << client->zookeeper->exists(client->getAbsolutePath(query->args[0].safeGet())) << "\n"; +} + bool GetStatCommand::parse(IParser::Pos & pos, std::shared_ptr & node, Expected & expected) const { String path; @@ -400,6 +415,21 @@ void ReconfigCommand::execute(const DB::ASTKeeperQuery * query, DB::KeeperClient std::cout << response.value << '\n'; } +bool SyncCommand::parse(IParser::Pos & pos, std::shared_ptr & node, DB::Expected & expected) const +{ + String path; + if (!parseKeeperPath(pos, expected, path)) + return false; + node->args.push_back(std::move(path)); + + return true; +} + +void SyncCommand::execute(const DB::ASTKeeperQuery * query, DB::KeeperClient * client) const +{ + std::cout << client->zookeeper->sync(client->getAbsolutePath(query->args[0].safeGet())) << "\n"; +} + bool HelpCommand::parse(IParser::Pos & /* pos */, std::shared_ptr & /* node */, Expected & /* expected */) const { return true; diff --git a/programs/keeper-client/Commands.h b/programs/keeper-client/Commands.h index c2f7bd9b896..19d5ee2a516 100644 --- a/programs/keeper-client/Commands.h +++ b/programs/keeper-client/Commands.h @@ -101,6 +101,17 @@ class GetCommand : public IKeeperClientCommand String getHelpMessage() const override { return "{} -- Returns the node's value"; } }; +class ExistsCommand : public IKeeperClientCommand +{ + String getName() const override { return "exists"; } + + bool parse(IParser::Pos & pos, std::shared_ptr & node, Expected & expected) const override; + + void execute(const ASTKeeperQuery * query, KeeperClient * client) const override; + + String getHelpMessage() const override { return "{} -- Returns `1` if node exists, `0` otherwise"; } +}; + class GetStatCommand : public IKeeperClientCommand { String getName() const override { return "get_stat"; } @@ -179,7 +190,8 @@ class RMRCommand : public IKeeperClientCommand class ReconfigCommand : public IKeeperClientCommand { - enum class Operation : Int64 { + enum class Operation : Int64 + { ADD = 0, REMOVE = 1, SET = 2, @@ -191,7 +203,18 @@ class ReconfigCommand : public IKeeperClientCommand void execute(const ASTKeeperQuery * query, KeeperClient * client) const override; - String getHelpMessage() const override { return "{} \"\" [version] -- Reconfigures a ZooKeeper cluster. See https://clickhouse.com/docs/en/guides/sre/keeper/clickhouse-keeper#reconfiguration"; } + String getHelpMessage() const override { return "{} \"\" [version] -- Reconfigures Keeper cluster. See https://clickhouse.com/docs/en/guides/sre/keeper/clickhouse-keeper#reconfiguration"; } +}; + +class SyncCommand: public IKeeperClientCommand +{ + String getName() const override { return "sync"; } + + bool parse(IParser::Pos & pos, std::shared_ptr & node, Expected & expected) const override; + + void execute(const ASTKeeperQuery * query, KeeperClient * client) const override; + + String getHelpMessage() const override { return "{} -- Synchronizes node between processes and leader"; } }; class HelpCommand : public IKeeperClientCommand diff --git a/programs/keeper-client/KeeperClient.cpp b/programs/keeper-client/KeeperClient.cpp index afaf94448d5..890b937e384 100644 --- a/programs/keeper-client/KeeperClient.cpp +++ b/programs/keeper-client/KeeperClient.cpp @@ -173,6 +173,14 @@ void KeeperClient::defineOptions(Poco::Util::OptionSet & options) Poco::Util::Option("log-level", "", "set log level") .argument("") .binding("log-level")); + + options.addOption( + Poco::Util::Option("no-confirmation", "", "if set, will not require a confirmation on several commands. default false for interactive and true for query") + .binding("no-confirmation")); + + options.addOption( + Poco::Util::Option("tests-mode", "", "run keeper-client in a special mode for tests. all commands output are separated by special symbols. default false") + .binding("tests-mode")); } void KeeperClient::initialize(Poco::Util::Application & /* self */) @@ -187,6 +195,7 @@ void KeeperClient::initialize(Poco::Util::Application & /* self */) std::make_shared(), std::make_shared(), std::make_shared(), + std::make_shared(), std::make_shared(), std::make_shared(), std::make_shared(), @@ -194,6 +203,7 @@ void KeeperClient::initialize(Poco::Util::Application & /* self */) std::make_shared(), std::make_shared(), std::make_shared(), + std::make_shared(), std::make_shared(), std::make_shared(), }); @@ -233,7 +243,6 @@ void KeeperClient::initialize(Poco::Util::Application & /* self */) EventNotifier::init(); } - bool KeeperClient::processQueryText(const String & text) { if (exit_strings.find(text) != exit_strings.end()) @@ -287,7 +296,7 @@ bool KeeperClient::processQueryText(const String & text) return true; } -void KeeperClient::runInteractive() +void KeeperClient::runInteractiveReplxx() { LineReader::Patterns query_extenders = {"\\"}; @@ -321,6 +330,26 @@ void KeeperClient::runInteractive() } } +void KeeperClient::runInteractiveInputStream() +{ + for (String input; std::getline(std::cin, input);) + { + if (!processQueryText(input)) + break; + + std::cout << "\a\a\a\a" << std::endl; + std::cerr << std::flush; + } +} + +void KeeperClient::runInteractive() +{ + if (config().hasOption("tests-mode")) + runInteractiveInputStream(); + else + runInteractiveReplxx(); +} + int KeeperClient::main(const std::vector & /* args */) { if (config().hasOption("help")) @@ -370,9 +399,11 @@ int KeeperClient::main(const std::vector & /* args */) zk_args.operation_timeout_ms = config().getInt("operation-timeout", 10) * 1000; zookeeper = std::make_unique(zk_args); + if (config().has("no-confirmation") || config().has("query")) + ask_confirmation = false; + if (config().has("query")) { - ask_confirmation = false; processQueryText(config().getString("query")); } else diff --git a/programs/keeper-client/KeeperClient.h b/programs/keeper-client/KeeperClient.h index 4a20a45de6e..0d3db3c2f02 100644 --- a/programs/keeper-client/KeeperClient.h +++ b/programs/keeper-client/KeeperClient.h @@ -49,6 +49,9 @@ public: protected: void runInteractive(); + void runInteractiveReplxx(); + void runInteractiveInputStream(); + bool processQueryText(const String & text); void loadCommands(std::vector && new_commands); diff --git a/tests/integration/helpers/keeper_utils.py b/tests/integration/helpers/keeper_utils.py index 93ea3fa74b7..a9ee3075750 100644 --- a/tests/integration/helpers/keeper_utils.py +++ b/tests/integration/helpers/keeper_utils.py @@ -1,6 +1,164 @@ +import io +import subprocess import socket import time +import typing as tp +import contextlib +import select from kazoo.client import KazooClient +from helpers.cluster import ClickHouseCluster, ClickHouseInstance +from helpers.client import CommandRequest + + +def execute_keeper_client_query(cluster: ClickHouseCluster, node: ClickHouseInstance, query: str) -> str: + request = CommandRequest( + [ + cluster.server_bin_path, + "keeper-client", + "--host", + str(cluster.get_instance_ip(node.name)), + "--port", + str(cluster.zookeeper_port), + "-q", + query, + ], + stdin="", + ) + + return request.get_answer() + + +class KeeperException(Exception): + pass + + +class KeeperClient(object): + + SEPARATOR = b'\a\a\a\a\n' + + def __init__(self, bin_path: str, host: str, port: int): + self.bin_path = bin_path + self.host = host + self.port = port + + self.proc = subprocess.Popen( + [ + bin_path, + 'keeper-client', + '--host', + host, + '--port', + str(port), + '--log-level', + 'error', + '--tests-mode', + '--no-confirmation', + ], + stdin=subprocess.PIPE, + stdout=subprocess.PIPE, + stderr=subprocess.PIPE, + ) + + self.poller = select.epoll() + self.poller.register(self.proc.stdout) + self.poller.register(self.proc.stderr) + + self._fd_nums = { + self.proc.stdout.fileno(): self.proc.stdout, + self.proc.stderr.fileno(): self.proc.stderr, + } + + self.stopped = False + + def execute_query(self, query: str, timeout: float = 10.) -> str: + output = io.BytesIO() + + self.proc.stdin.write(query.encode() + b'\n') + self.proc.stdin.flush() + + events = self.poller.poll(timeout) + + for fd_num, event in events: + if event & (select.EPOLLIN | select.EPOLLPRI): + file = self._fd_nums[fd_num] + + if file == self.proc.stdout: + while True: + chunk = file.readline() + if chunk.endswith(self.SEPARATOR): + break + + output.write(chunk) + + elif file == self.proc.stderr: + assert self.proc.stdout.readline() == self.SEPARATOR + raise KeeperException(self.proc.stderr.readline().strip().decode()) + + else: + raise ValueError(f'Failed to read from pipe. Flag {event}') + + data = output.getvalue().strip().decode() + return data + + def cd(self, path: str, timeout: float = 10.): + self.execute_query(f'cd {path}', timeout) + + def ls(self, path: str, timeout: float = 10.) -> list[str]: + return self.execute_query(f'ls {path}', timeout).split(' ') + + def create(self, path: str, value: str, timeout: float = 10.): + self.execute_query(f'create {path} {value}', timeout) + + def get(self, path: str, timeout: float = 10.) -> str: + return self.execute_query(f'get {path}', timeout) + + def exists(self, path: str, timeout: float = 10.) -> bool: + return bool(int(self.execute_query(f'exists {path}', timeout))) + + def stop(self): + if not self.stopped: + self.stopped = True + self.proc.communicate(b'exit\n', timeout=10.) + + def sync(self, path: str, timeout: float = 10.): + self.execute_query(f'sync {path}', timeout) + + def touch(self, path: str, timeout: float = 10.): + self.execute_query(f'touch {path}', timeout) + + def find_big_family(self, path: str, n: int = 10, timeout: float = 10.) -> str: + return self.execute_query(f'find_big_family {path} {n}', timeout) + + def find_super_nodes(self, threshold: int, timeout: float = 10.) -> str: + return self.execute_query(f'find_super_nodes {threshold}', timeout) + + def delete_stale_backups(self, timeout: float = 10.) -> str: + return self.execute_query('delete_stale_backups', timeout) + + def reconfig(self, joining: tp.Optional[str], leaving: tp.Optional[str], new_members: tp.Optional[str], timeout: float = 10.) -> str: + if bool(joining) + bool(leaving) + bool(new_members) != 1: + raise ValueError('Exactly one of joining, leaving or new_members must be specified') + + if joining is not None: + operation = 'add' + elif leaving is not None: + operation = 'remove' + elif new_members is not None: + operation = 'set' + else: + raise ValueError('At least one of joining, leaving or new_members must be specified') + + return self.execute_query(f'reconfig {operation} {joining or leaving or new_members}', timeout) + + @classmethod + @contextlib.contextmanager + def from_cluster(cls, cluster: ClickHouseCluster, keeper_node: str, port: tp.Optional[int] = None) -> 'KeeperClient': + client = cls(cluster.server_bin_path, cluster.get_instance_ip(keeper_node), port or cluster.zookeeper_port) + + try: + yield client + finally: + client.stop() def get_keeper_socket(cluster, node, port=9181): @@ -70,14 +228,14 @@ def get_fake_zk(cluster, node, timeout: float = 30.0) -> KazooClient: return _fake -def get_config_str(zk: KazooClient) -> str: +def get_config_str(zk: KeeperClient) -> str: """ Return decoded contents of /keeper/config node """ - return zk.get("/keeper/config")[0].decode("utf-8") + return zk.get("/keeper/config") -def wait_configs_equal(left_config: str, right_zk: KazooClient, timeout: float = 30.0): +def wait_configs_equal(left_config: str, right_zk: KeeperClient, timeout: float = 30.0): """ Check whether get /keeper/config result in left_config is equal to get /keeper/config on right_zk ZK connection. diff --git a/tests/integration/test_keeper_client/test.py b/tests/integration/test_keeper_client/test.py index 8f7056a5afd..c82917372b8 100644 --- a/tests/integration/test_keeper_client/test.py +++ b/tests/integration/test_keeper_client/test.py @@ -1,7 +1,7 @@ import pytest -from helpers.client import CommandRequest from helpers.cluster import ClickHouseCluster from helpers.test_tools import TSV +from helpers.keeper_utils import KeeperClient cluster = ClickHouseCluster(__file__) @@ -24,39 +24,28 @@ def started_cluster(): cluster.shutdown() -def keeper_query(query: str): - return CommandRequest( - [ - cluster.server_bin_path, - "keeper-client", - "--host", - str(cluster.get_instance_ip("zoo1")), - "--port", - str(cluster.zookeeper_port), - "-q", - query, - ], - stdin="", - ) +@pytest.fixture(scope="function") +def client(started_cluster): + with KeeperClient.from_cluster(cluster, "zoo1") as keeper_client: + yield keeper_client -def test_big_family(): - command = keeper_query( - "touch test_big_family;" - "touch test_big_family/1;" - "touch test_big_family/1/1;" - "touch test_big_family/1/2;" - "touch test_big_family/1/3;" - "touch test_big_family/1/4;" - "touch test_big_family/1/5;" - "touch test_big_family/2;" - "touch test_big_family/2/1;" - "touch test_big_family/2/2;" - "touch test_big_family/2/3;" - "find_big_family test_big_family;" - ) +def test_big_family(client: KeeperClient): + client.touch("/test_big_family") + client.touch("/test_big_family/1") + client.touch("/test_big_family/1/1") + client.touch("/test_big_family/1/2") + client.touch("/test_big_family/1/3") + client.touch("/test_big_family/1/4") + client.touch("/test_big_family/1/5") + client.touch("/test_big_family/2") + client.touch("/test_big_family/2/1") + client.touch("/test_big_family/2/2") + client.touch("/test_big_family/2/3") - assert command.get_answer() == TSV( + response = client.find_big_family("/test_big_family") + + assert response == TSV( [ ["/test_big_family/1", "5"], ["/test_big_family/2", "3"], @@ -71,34 +60,33 @@ def test_big_family(): ] ) - command = keeper_query("find_big_family test_big_family 1;") + response = client.find_big_family("/test_big_family", 1) - assert command.get_answer() == TSV( + assert response == TSV( [ ["/test_big_family/1", "5"], ] ) -def test_find_super_nodes(): - command = keeper_query( - "touch test_find_super_nodes;" - "touch test_find_super_nodes/1;" - "touch test_find_super_nodes/1/1;" - "touch test_find_super_nodes/1/2;" - "touch test_find_super_nodes/1/3;" - "touch test_find_super_nodes/1/4;" - "touch test_find_super_nodes/1/5;" - "touch test_find_super_nodes/2;" - "touch test_find_super_nodes/2/1;" - "touch test_find_super_nodes/2/2;" - "touch test_find_super_nodes/2/3;" - "touch test_find_super_nodes/2/4;" - "cd test_find_super_nodes;" - "find_super_nodes 4;" - ) +def test_find_super_nodes(client: KeeperClient): + client.touch("/test_find_super_nodes") + client.touch("/test_find_super_nodes/1") + client.touch("/test_find_super_nodes/1/1") + client.touch("/test_find_super_nodes/1/2") + client.touch("/test_find_super_nodes/1/3") + client.touch("/test_find_super_nodes/1/4") + client.touch("/test_find_super_nodes/1/5") + client.touch("/test_find_super_nodes/2") + client.touch("/test_find_super_nodes/2/1") + client.touch("/test_find_super_nodes/2/2") + client.touch("/test_find_super_nodes/2/3") + client.touch("/test_find_super_nodes/2/4") - assert command.get_answer() == TSV( + client.cd("/test_find_super_nodes") + + response = client.find_super_nodes(4) + assert response == TSV( [ ["/test_find_super_nodes/1", "5"], ["/test_find_super_nodes/2", "4"], @@ -106,41 +94,38 @@ def test_find_super_nodes(): ) -def test_delete_stale_backups(): - command = keeper_query( - "touch /clickhouse;" - "touch /clickhouse/backups;" - "touch /clickhouse/backups/1;" - "touch /clickhouse/backups/1/stage;" - "touch /clickhouse/backups/1/stage/alive123;" - "touch /clickhouse/backups/2;" - "touch /clickhouse/backups/2/stage;" - "touch /clickhouse/backups/2/stage/dead123;" - "delete_stale_backups;" - "y;" - "ls clickhouse/backups;" - ) +def test_delete_stale_backups(client: KeeperClient): + client.touch("/clickhouse") + client.touch("/clickhouse/backups") + client.touch("/clickhouse/backups/1") + client.touch("/clickhouse/backups/1/stage") + client.touch("/clickhouse/backups/1/stage/alive123") + client.touch("/clickhouse/backups/2") + client.touch("/clickhouse/backups/2/stage") + client.touch("/clickhouse/backups/2/stage/dead123") - assert command.get_answer() == ( - "You are going to delete all inactive backups in /clickhouse/backups. Continue?\n" + response = client.delete_stale_backups() + + assert response == ( 'Found backup "/clickhouse/backups/1", checking if it\'s active\n' 'Backup "/clickhouse/backups/1" is active, not going to delete\n' 'Found backup "/clickhouse/backups/2", checking if it\'s active\n' - 'Backup "/clickhouse/backups/2" is not active, deleting it\n' - "1\n" + 'Backup "/clickhouse/backups/2" is not active, deleting it' ) - -def test_base_commands(): - command = keeper_query( - "create test_create_zk_node1 testvalue1;" - "create test_create_zk_node_2 testvalue2;" - "get test_create_zk_node1;" - ) - - assert command.get_answer() == "testvalue1\n" + assert client.ls("/clickhouse/backups") == ["1"] -def test_four_letter_word_commands(): - command = keeper_query("ruok") - assert command.get_answer() == "imok\n" +def test_base_commands(client: KeeperClient): + client.create("/test_create_zk_node1", "testvalue1") + client.create("/test_create_zk_node_2", "testvalue2") + assert client.get("/test_create_zk_node1") == "testvalue1" + + client.create("/123", "1=2") + client.create("/123/321", "'foo;bar'") + assert client.get("/123") == "1=2" + assert client.get("/123/321") == "foo;bar" + + +def test_four_letter_word_commands(client: KeeperClient): + assert client.execute_query("ruok") == "imok" diff --git a/tests/integration/test_keeper_reconfig_add/test.py b/tests/integration/test_keeper_reconfig_add/test.py index 2c2da7403a1..6e05766c271 100644 --- a/tests/integration/test_keeper_reconfig_add/test.py +++ b/tests/integration/test_keeper_reconfig_add/test.py @@ -1,11 +1,10 @@ #!/usr/bin/env python3 import pytest -from helpers.cluster import ClickHouseCluster +from helpers.cluster import ClickHouseCluster, ClickHouseInstance import helpers.keeper_utils as ku import os -from kazoo.client import KazooClient -from kazoo.exceptions import BadArgumentsException +import typing as tp cluster = ClickHouseCluster(__file__) CONFIG_DIR = os.path.join(os.path.dirname(os.path.realpath(__file__)), "configs") @@ -19,11 +18,7 @@ part_of_cluster = "now this node is the part of cluster" zk1, zk2, zk3 = None, None, None -def get_fake_zk(node): - return ku.get_fake_zk(cluster, node) - - -@pytest.fixture(scope="module") +@pytest.fixture(scope="module", autouse=True) def started_cluster(): try: cluster.start() @@ -43,21 +38,26 @@ def started_cluster(): yield cluster finally: + conn: tp.Optional[ku.KeeperClient] for conn in [zk1, zk2, zk3]: - if conn: + if conn is not None: conn.stop() - conn.close() cluster.shutdown() -def test_reconfig_add(started_cluster): +def create_client(node: ClickHouseInstance): + return ku.KeeperClient(cluster.server_bin_path, cluster.get_instance_ip(node.name), 9181) + + +def test_reconfig_add(): """ Add a node to another node. Then add another node to two. """ + global zk1, zk2, zk3 + zk1 = create_client(node1) - zk1 = get_fake_zk(node1) - config = ku.get_config_str(zk1) + config = zk1.get("/keeper/config") print("Initial config", config) assert len(config.split("\n")) == 1 @@ -65,24 +65,22 @@ def test_reconfig_add(started_cluster): assert "node2" not in config assert "node3" not in config - with pytest.raises(BadArgumentsException): + with pytest.raises(ku.KeeperException): # duplicate id with different endpoint zk1.reconfig(joining="server.1=localhost:1337", leaving=None, new_members=None) - with pytest.raises(BadArgumentsException): + with pytest.raises(ku.KeeperException): # duplicate endpoint zk1.reconfig(joining="server.8=node1:9234", leaving=None, new_members=None) for i in range(100): - zk1.create(f"/test_three_{i}", b"somedata") + zk1.create(f"/test_three_{i}", "somedata") node2.start_clickhouse() - config, _ = zk1.reconfig( + config = zk1.reconfig( joining="server.2=node2:9234", leaving=None, new_members=None ) ku.wait_until_connected(cluster, node2) - - config = config.decode("utf-8") print("After adding 2", config) assert len(config.split("\n")) == 2 @@ -90,12 +88,12 @@ def test_reconfig_add(started_cluster): assert "node2" in config assert "node3" not in config - zk2 = get_fake_zk(node2) + zk2 = create_client(node2) ku.wait_configs_equal(config, zk2) for i in range(100): - assert zk2.exists(f"/test_three_{i}") is not None - zk2.create(f"/test_three_{100 + i}", b"somedata") + assert zk2.exists(f"/test_three_{i}") + zk2.create(f"/test_three_{100 + i}", "somedata") # Why not both? # One node will process add_srv request, other will pull out updated config, apply @@ -107,23 +105,21 @@ def test_reconfig_add(started_cluster): assert node2.contains_in_log(part_of_cluster) zk1.stop() - zk1.close() - zk1 = get_fake_zk(node1) + zk1 = create_client(node1) zk1.sync("/test_three_0") for i in range(200): - assert zk1.exists(f"/test_three_{i}") is not None + assert zk1.exists(f"/test_three_{i}") for i in range(100): - zk2.create(f"/test_four_{i}", b"somedata") + zk2.create(f"/test_four_{i}", "somedata") node3.start_clickhouse() - config, _ = zk2.reconfig( + config = zk2.reconfig( joining="server.3=node3:9234", leaving=None, new_members=None ) ku.wait_until_connected(cluster, node3) - config = config.decode("utf-8") print("After adding 3", config) assert len(config.split("\n")) == 3 @@ -131,25 +127,23 @@ def test_reconfig_add(started_cluster): assert "node2" in config assert "node3" in config - zk3 = get_fake_zk(node3) + zk3 = create_client(node3) ku.wait_configs_equal(config, zk3) for i in range(100): - assert zk3.exists(f"/test_four_{i}") is not None - zk3.create(f"/test_four_{100 + i}", b"somedata") + assert zk3.exists(f"/test_four_{i}") + zk3.create(f"/test_four_{100 + i}", "somedata") zk1.stop() - zk1.close() - zk1 = get_fake_zk(node1) + zk1 = create_client(node1) zk1.sync("/test_four_0") zk2.stop() - zk2.close() - zk2 = get_fake_zk(node2) + zk2 = create_client(node2) zk2.sync("/test_four_0") for i in range(200): - assert zk1.exists(f"/test_four_{i}") is not None - assert zk2.exists(f"/test_four_{i}") is not None + assert zk1.exists(f"/test_four_{i}") + assert zk2.exists(f"/test_four_{i}") assert node3.contains_in_log(part_of_cluster) diff --git a/tests/integration/test_keeper_reconfig_remove/test.py b/tests/integration/test_keeper_reconfig_remove/test.py index fb0a9472df3..daab94c59c4 100644 --- a/tests/integration/test_keeper_reconfig_remove/test.py +++ b/tests/integration/test_keeper_reconfig_remove/test.py @@ -1,11 +1,11 @@ #!/usr/bin/env python3 +import subprocess import pytest -from helpers.cluster import ClickHouseCluster +from helpers.cluster import ClickHouseCluster, ClickHouseInstance import helpers.keeper_utils as ku import os -from kazoo.client import KazooClient -from kazoo.exceptions import BadVersionException, BadArgumentsException +import typing as tp cluster = ClickHouseCluster(__file__) CONFIG_DIR = os.path.join(os.path.dirname(os.path.realpath(__file__)), "configs") @@ -23,16 +23,16 @@ def started_cluster(): cluster.start() yield cluster finally: + conn: tp.Optional[ku.KeeperClient] for conn in [zk1, zk2, zk3]: if conn: conn.stop() - conn.close() cluster.shutdown() -def get_fake_zk(node): - return ku.get_fake_zk(cluster, node) +def create_client(node: ClickHouseInstance): + return ku.KeeperClient(cluster.server_bin_path, cluster.get_instance_ip(node.name), 9181) def test_reconfig_remove_followers_from_3(started_cluster): @@ -42,9 +42,9 @@ def test_reconfig_remove_followers_from_3(started_cluster): Check that remaining node is in standalone mode. """ - zk1 = get_fake_zk(node1) - config, _ = zk1.get("/keeper/config") - config = config.decode("utf-8") + global zk1, zk2, zk3 + zk1 = create_client(node1) + config = zk1.get("/keeper/config") print("Initial config", config) assert len(config.split("\n")) == 3 @@ -52,36 +52,33 @@ def test_reconfig_remove_followers_from_3(started_cluster): assert "node2" in config assert "node3" in config - with pytest.raises(BadVersionException): - zk1.reconfig(joining=None, leaving="1", new_members=None, from_config=20) - with pytest.raises(BadArgumentsException): + with pytest.raises(ValueError): zk1.reconfig(joining=None, leaving=None, new_members=None) - with pytest.raises(BadArgumentsException): + with pytest.raises(ku.KeeperException): # bulk reconfiguration is not supported zk1.reconfig(joining=None, leaving=None, new_members="3") - with pytest.raises(BadArgumentsException): + with pytest.raises(ValueError): zk1.reconfig(joining="1", leaving="1", new_members="3") - with pytest.raises(BadArgumentsException): + with pytest.raises(ku.KeeperException): # at least one node must be left zk1.reconfig(joining=None, leaving="1,2,3", new_members=None) for i in range(100): - zk1.create(f"/test_two_{i}", b"somedata") + zk1.create(f"/test_two_{i}", "somedata") - zk2 = get_fake_zk(node2) + zk2 = create_client(node2) zk2.sync("/test_two_0") ku.wait_configs_equal(config, zk2) - zk3 = get_fake_zk(node3) + zk3 = create_client(node3) zk3.sync("/test_two_0") ku.wait_configs_equal(config, zk3) for i in range(100): - assert zk2.exists(f"test_two_{i}") is not None - assert zk3.exists(f"test_two_{i}") is not None + assert zk2.exists(f"test_two_{i}") + assert zk3.exists(f"test_two_{i}") - config, _ = zk1.reconfig(joining=None, leaving="3", new_members=None) - config = config.decode("utf-8") + config = zk1.reconfig(joining=None, leaving="3", new_members=None) print("After removing 3", config) assert len(config.split("\n")) == 2 @@ -90,35 +87,26 @@ def test_reconfig_remove_followers_from_3(started_cluster): assert "node3" not in config zk2.stop() - zk2.close() - zk2 = get_fake_zk(node2) + zk2 = create_client(node2) ku.wait_configs_equal(config, zk2) for i in range(100): - assert zk2.exists(f"test_two_{i}") is not None - zk2.create(f"/test_two_{100 + i}", b"otherdata") + assert zk2.exists(f"test_two_{i}") + zk2.create(f"/test_two_{100 + i}", "otherdata") zk1.stop() - zk1.close() - zk1 = get_fake_zk(node1) + zk1 = create_client(node1) zk1.sync("/test_two_0") for i in range(200): - assert zk1.exists(f"test_two_{i}") is not None - - with pytest.raises(Exception): - zk3.stop() - zk3.close() - zk3 = get_fake_zk(node3) - zk3.sync("/test_two_0") + assert zk1.exists(f"test_two_{i}") assert node3.contains_in_log(log_msg_removed) for i in range(100): - zk2.create(f"/test_two_{200 + i}", b"otherdata") + zk2.create(f"/test_two_{200 + i}", "otherdata") - config, _ = zk1.reconfig(joining=None, leaving="2", new_members=None) - config = config.decode("utf-8") + config = zk1.reconfig(joining=None, leaving="2", new_members=None) print("After removing 2", config) assert len(config.split("\n")) == 1 @@ -127,19 +115,12 @@ def test_reconfig_remove_followers_from_3(started_cluster): assert "node3" not in config zk1.stop() - zk1.close() - zk1 = get_fake_zk(node1) + zk1 = create_client(node1) zk1.sync("/test_two_0") for i in range(300): - assert zk1.exists(f"test_two_{i}") is not None - - with pytest.raises(Exception): - zk2.stop() - zk2.close() - zk2 = get_fake_zk(node2) - zk2.sync("/test_two_0") + assert zk1.exists(f"test_two_{i}") assert not node1.contains_in_log(log_msg_removed) assert node2.contains_in_log(log_msg_removed) - assert "Mode: standalone" in zk1.command(b"stat") + assert "Mode: standalone" in zk1.execute_query('stat') diff --git a/tests/integration/test_keeper_reconfig_remove_many/test.py b/tests/integration/test_keeper_reconfig_remove_many/test.py index ec0d8b95eff..e99327e1098 100644 --- a/tests/integration/test_keeper_reconfig_remove_many/test.py +++ b/tests/integration/test_keeper_reconfig_remove_many/test.py @@ -1,11 +1,11 @@ #!/usr/bin/env python3 import pytest -from helpers.cluster import ClickHouseCluster +from helpers.cluster import ClickHouseCluster, ClickHouseInstance import helpers.keeper_utils as ku import os -from kazoo.client import KazooClient, KazooState -from kazoo.exceptions import BadVersionException, BadArgumentsException +import typing as tp + cluster = ClickHouseCluster(__file__) CONFIG_DIR = os.path.join(os.path.dirname(os.path.realpath(__file__)), "configs") @@ -26,49 +26,49 @@ def started_cluster(): cluster.start() yield cluster finally: + conn: tp.Optional[ku.KeeperClient] for conn in [zk1, zk2, zk3, zk4, zk5]: if conn: conn.stop() - conn.close() cluster.shutdown() -def get_fake_zk(node): - return ku.get_fake_zk(cluster, node) +def create_client(node: ClickHouseInstance): + return ku.KeeperClient(cluster.server_bin_path, cluster.get_instance_ip(node.name), 9181) def test_reconfig_remove_2_and_leader(started_cluster): """ Remove 2 followers from a cluster of 5. Remove leader from 3 nodes. """ + global zk1, zk2, zk3, zk4, zk5 - zk1 = get_fake_zk(node1) + zk1 = create_client(node1) config = ku.get_config_str(zk1) print("Initial config", config) assert len(config.split("\n")) == 5 for i in range(100): - zk1.create(f"/test_two_{i}", b"somedata") + zk1.create(f"/test_two_{i}", "somedata") - zk4 = get_fake_zk(node4) + zk4 = create_client(node4) zk4.sync("/test_two_0") ku.wait_configs_equal(config, zk4) - zk5 = get_fake_zk(node5) + zk5 = create_client(node5) zk5.sync("/test_two_0") ku.wait_configs_equal(config, zk5) for i in range(100): - assert zk4.exists(f"test_two_{i}") is not None - assert zk5.exists(f"test_two_{i}") is not None + assert zk4.exists(f"test_two_{i}") + assert zk5.exists(f"test_two_{i}") - zk4.create(f"/test_two_{100 + i}", b"otherdata") + zk4.create(f"/test_two_{100 + i}", "otherdata") - zk2 = get_fake_zk(node2) - config, _ = zk2.reconfig(joining=None, leaving="4,5", new_members=None) - config = config.decode("utf-8") + zk2 = create_client(node2) + config = zk2.reconfig(joining=None, leaving="4,5", new_members=None) print("After removing 4,5", config) assert len(config.split("\n")) == 3 @@ -79,27 +79,14 @@ def test_reconfig_remove_2_and_leader(started_cluster): assert "node5" not in config zk1.stop() - zk1.close() - zk1 = get_fake_zk(node1) + zk1 = create_client(node1) zk1.sync("/test_two_0") ku.wait_configs_equal(config, zk1) for i in range(200): - assert zk1.exists(f"test_two_{i}") is not None - assert zk2.exists(f"test_two_{i}") is not None - - with pytest.raises(Exception): - zk4.stop() - zk4.close() - zk4 = get_fake_zk(node4) - zk4.sync("/test_two_0") - - with pytest.raises(Exception): - zk5.stop() - zk5.close() - zk5 = get_fake_zk(node5) - zk5.sync("/test_two_0") + assert zk1.exists(f"test_two_{i}") + assert zk2.exists(f"test_two_{i}") assert not node1.contains_in_log(log_msg_removed) assert not node2.contains_in_log(log_msg_removed) @@ -110,11 +97,10 @@ def test_reconfig_remove_2_and_leader(started_cluster): assert ku.is_leader(cluster, node1) for i in range(100): - zk1.create(f"/test_leader_{i}", b"somedata") + zk1.create(f"/test_leader_{i}", "somedata") # when a leader gets a remove request, it must yield leadership - config, _ = zk1.reconfig(joining=None, leaving="1", new_members=None) - config = config.decode("utf-8") + config = zk1.reconfig(joining=None, leaving="1", new_members=None) print("After removing 1 (leader)", config) assert len(config.split("\n")) == 2 @@ -125,24 +111,17 @@ def test_reconfig_remove_2_and_leader(started_cluster): assert "node5" not in config zk2.stop() - zk2.close() - zk2 = get_fake_zk(node2) + zk2 = create_client(node2) zk2.sync("/test_leader_0") ku.wait_configs_equal(config, zk2) - zk3 = get_fake_zk(node3) + zk3 = create_client(node3) zk3.sync("/test_leader_0") ku.wait_configs_equal(config, zk3) for i in range(100): - assert zk2.exists(f"test_leader_{i}") is not None - assert zk3.exists(f"test_leader_{i}") is not None - - with pytest.raises(Exception): - zk1.stop() - zk1.close() - zk1 = get_fake_zk(node1) - zk1.sync("/test_leader_0") + assert zk2.exists(f"test_leader_{i}") + assert zk3.exists(f"test_leader_{i}") assert node1.contains_in_log(log_msg_removed) assert not node2.contains_in_log(log_msg_removed) diff --git a/tests/integration/test_keeper_reconfig_replace_leader/test.py b/tests/integration/test_keeper_reconfig_replace_leader/test.py index ef1d5394b67..bbb5b66eaa5 100644 --- a/tests/integration/test_keeper_reconfig_replace_leader/test.py +++ b/tests/integration/test_keeper_reconfig_replace_leader/test.py @@ -1,11 +1,10 @@ #!/usr/bin/env python3 import pytest -from helpers.cluster import ClickHouseCluster +from helpers.cluster import ClickHouseCluster, ClickHouseInstance from os.path import join, dirname, realpath -import time import helpers.keeper_utils as ku -from kazoo.client import KazooClient, KazooState +import typing as tp cluster = ClickHouseCluster(__file__) CONFIG_DIR = join(dirname(realpath(__file__)), "configs") @@ -31,24 +30,26 @@ def started_cluster(): yield cluster finally: + conn: tp.Optional[ku.KeeperClient] for conn in [zk1, zk2, zk3, zk4]: if conn: conn.stop() - conn.close() cluster.shutdown() -def get_fake_zk(node): - return ku.get_fake_zk(cluster, node) +def create_client(node: ClickHouseInstance): + return ku.KeeperClient( + cluster.server_bin_path, cluster.get_instance_ip(node.name), 9181 + ) def test_reconfig_replace_leader(started_cluster): """ Remove leader from a cluster of 3 and add a new node via two commands. """ - - zk1 = get_fake_zk(node1) + global zk1, zk2, zk3, zk4 + zk1 = create_client(node1) config = ku.get_config_str(zk1) assert len(config.split("\n")) == 3 @@ -58,23 +59,22 @@ def test_reconfig_replace_leader(started_cluster): assert "node4" not in config for i in range(100): - zk1.create(f"/test_four_{i}", b"somedata") + zk1.create(f"/test_four_{i}", "somedata") - zk2 = get_fake_zk(node2) + zk2 = create_client(node2) zk2.sync("/test_four_0") ku.wait_configs_equal(config, zk2) - zk3 = get_fake_zk(node3) + zk3 = create_client(node3) zk3.sync("/test_four_0") ku.wait_configs_equal(config, zk3) for i in range(100): - assert zk2.exists(f"/test_four_{i}") is not None - assert zk3.exists(f"/test_four_{i}") is not None + assert zk2.exists(f"/test_four_{i}") + assert zk3.exists(f"/test_four_{i}") assert ku.is_leader(cluster, node1) - config, _ = zk2.reconfig(joining=None, leaving="1", new_members=None) - config = config.decode("utf-8") + config = zk2.reconfig(joining=None, leaving="1", new_members=None) print("After removing 1 (leader)", config) assert len(config.split("\n")) == 2 @@ -85,17 +85,10 @@ def test_reconfig_replace_leader(started_cluster): ku.wait_configs_equal(config, zk2) - with pytest.raises(Exception): - zk1.stop() - zk1.close() - zk1 = get_fake_zk(node1) - zk1.sync("/test_four_0") - node4.start_clickhouse() - config, _ = zk2.reconfig( + config = zk2.reconfig( joining="server.4=node4:9234", leaving=None, new_members=None ) - config = config.decode("utf-8") ku.wait_until_connected(cluster, node4) print("After adding 4", config) @@ -105,22 +98,20 @@ def test_reconfig_replace_leader(started_cluster): assert "node3" in config assert "node4" in config - zk4 = get_fake_zk(node4) + zk4 = create_client(node4) ku.wait_configs_equal(config, zk4) for i in range(100): - assert zk4.exists(f"test_four_{i}") is not None - zk4.create(f"/test_four_{100 + i}", b"somedata") + assert zk4.exists(f"test_four_{i}") + zk4.create(f"/test_four_{100 + i}", "somedata") zk2.stop() - zk2.close() - zk2 = get_fake_zk(node2) + zk2 = create_client(node2) zk2.sync("/test_four_0") ku.wait_configs_equal(config, zk2) zk3.stop() - zk3.close() - zk3 = get_fake_zk(node3) + zk3 = create_client(node3) zk3.sync("/test_four_0") ku.wait_configs_equal(config, zk3) diff --git a/tests/integration/test_keeper_reconfig_replace_leader_in_one_command/__init__.py b/tests/integration/test_keeper_reconfig_replace_leader_in_one_command/__init__.py deleted file mode 100644 index e69de29bb2d..00000000000 diff --git a/tests/integration/test_keeper_reconfig_replace_leader_in_one_command/configs/keeper1.xml b/tests/integration/test_keeper_reconfig_replace_leader_in_one_command/configs/keeper1.xml deleted file mode 100644 index 71f3403aca3..00000000000 --- a/tests/integration/test_keeper_reconfig_replace_leader_in_one_command/configs/keeper1.xml +++ /dev/null @@ -1,35 +0,0 @@ - - - 9181 - 1 - /var/lib/clickhouse/coordination/log - /var/lib/clickhouse/coordination/snapshots - true - - - 5000 - 10000 - trace - - - - - 1 - node1 - 9234 - - - 2 - node2 - 9234 - true - - - 3 - node3 - 9234 - true - - - - diff --git a/tests/integration/test_keeper_reconfig_replace_leader_in_one_command/configs/keeper2.xml b/tests/integration/test_keeper_reconfig_replace_leader_in_one_command/configs/keeper2.xml deleted file mode 100644 index faefb4d1102..00000000000 --- a/tests/integration/test_keeper_reconfig_replace_leader_in_one_command/configs/keeper2.xml +++ /dev/null @@ -1,35 +0,0 @@ - - - 9181 - 2 - /var/lib/clickhouse/coordination/log - /var/lib/clickhouse/coordination/snapshots - true - - - 5000 - 10000 - trace - - - - - 1 - node1 - 9234 - - - 2 - node2 - 9234 - true - - - 3 - node3 - 9234 - true - - - - diff --git a/tests/integration/test_keeper_reconfig_replace_leader_in_one_command/configs/keeper3.xml b/tests/integration/test_keeper_reconfig_replace_leader_in_one_command/configs/keeper3.xml deleted file mode 100644 index 80a9caa92c2..00000000000 --- a/tests/integration/test_keeper_reconfig_replace_leader_in_one_command/configs/keeper3.xml +++ /dev/null @@ -1,35 +0,0 @@ - - - 9181 - 3 - /var/lib/clickhouse/coordination/log - /var/lib/clickhouse/coordination/snapshots - true - - - 5000 - 10000 - trace - - - - - 1 - node1 - 9234 - - - 2 - node2 - 9234 - true - - - 3 - node3 - 9234 - true - - - - diff --git a/tests/integration/test_keeper_reconfig_replace_leader_in_one_command/configs/keeper4.xml b/tests/integration/test_keeper_reconfig_replace_leader_in_one_command/configs/keeper4.xml deleted file mode 100644 index 9fd88fe5d63..00000000000 --- a/tests/integration/test_keeper_reconfig_replace_leader_in_one_command/configs/keeper4.xml +++ /dev/null @@ -1,21 +0,0 @@ - - - 9181 - 4 - /var/lib/clickhouse/coordination/log - /var/lib/clickhouse/coordination/snapshots - true - - - 5000 - 10000 - trace - - - - 2 node2 9234 - 3 node3 9234 - 4 node4 9234 - - - diff --git a/tests/integration/test_keeper_reconfig_replace_leader_in_one_command/test.py b/tests/integration/test_keeper_reconfig_replace_leader_in_one_command/test.py deleted file mode 100644 index b099d0513e1..00000000000 --- a/tests/integration/test_keeper_reconfig_replace_leader_in_one_command/test.py +++ /dev/null @@ -1,43 +0,0 @@ -#!/usr/bin/env python3 - -import pytest -from helpers.cluster import ClickHouseCluster -from os.path import join, dirname, realpath -import time -import helpers.keeper_utils as ku -from kazoo.client import KazooClient, KazooState - -cluster = ClickHouseCluster(__file__) -CONFIG_DIR = join(dirname(realpath(__file__)), "configs") - -node1 = cluster.add_instance("node1", main_configs=["configs/keeper1.xml"]) -node2 = cluster.add_instance("node2", main_configs=["configs/keeper2.xml"]) -node3 = cluster.add_instance("node3", main_configs=["configs/keeper3.xml"]) -node4 = cluster.add_instance("node4", stay_alive=True) -zk1, zk2, zk3, zk4 = None, None, None, None - - -@pytest.fixture(scope="module") -def started_cluster(): - try: - cluster.start() - - node4.stop_clickhouse() - node4.copy_file_to_container( - join(CONFIG_DIR, "keeper4.xml"), - "/etc/clickhouse-server/config.d/keeper.xml", - ) - - yield cluster - - finally: - for conn in [zk1, zk2, zk3, zk4]: - if conn: - conn.stop() - conn.close() - - cluster.shutdown() - - -def get_fake_zk(node): - return ku.get_fake_zk(cluster, node) From cc7892691940132fa0c5ac65bab3ccdb08e98024 Mon Sep 17 00:00:00 2001 From: robot-clickhouse Date: Thu, 7 Sep 2023 00:44:54 +0000 Subject: [PATCH 168/433] Automatic style fix --- tests/integration/helpers/keeper_utils.py | 105 +++++++++++------- .../test_keeper_reconfig_add/test.py | 12 +- .../test_keeper_reconfig_remove/test.py | 6 +- .../test_keeper_reconfig_remove_many/test.py | 4 +- .../test.py | 4 +- 5 files changed, 75 insertions(+), 56 deletions(-) diff --git a/tests/integration/helpers/keeper_utils.py b/tests/integration/helpers/keeper_utils.py index a9ee3075750..767416a3dbc 100644 --- a/tests/integration/helpers/keeper_utils.py +++ b/tests/integration/helpers/keeper_utils.py @@ -10,7 +10,9 @@ from helpers.cluster import ClickHouseCluster, ClickHouseInstance from helpers.client import CommandRequest -def execute_keeper_client_query(cluster: ClickHouseCluster, node: ClickHouseInstance, query: str) -> str: +def execute_keeper_client_query( + cluster: ClickHouseCluster, node: ClickHouseInstance, query: str +) -> str: request = CommandRequest( [ cluster.server_bin_path, @@ -33,8 +35,7 @@ class KeeperException(Exception): class KeeperClient(object): - - SEPARATOR = b'\a\a\a\a\n' + SEPARATOR = b"\a\a\a\a\n" def __init__(self, bin_path: str, host: str, port: int): self.bin_path = bin_path @@ -44,15 +45,15 @@ class KeeperClient(object): self.proc = subprocess.Popen( [ bin_path, - 'keeper-client', - '--host', + "keeper-client", + "--host", host, - '--port', + "--port", str(port), - '--log-level', - 'error', - '--tests-mode', - '--no-confirmation', + "--log-level", + "error", + "--tests-mode", + "--no-confirmation", ], stdin=subprocess.PIPE, stdout=subprocess.PIPE, @@ -70,10 +71,10 @@ class KeeperClient(object): self.stopped = False - def execute_query(self, query: str, timeout: float = 10.) -> str: + def execute_query(self, query: str, timeout: float = 10.0) -> str: output = io.BytesIO() - self.proc.stdin.write(query.encode() + b'\n') + self.proc.stdin.write(query.encode() + b"\n") self.proc.stdin.flush() events = self.poller.poll(timeout) @@ -95,65 +96,83 @@ class KeeperClient(object): raise KeeperException(self.proc.stderr.readline().strip().decode()) else: - raise ValueError(f'Failed to read from pipe. Flag {event}') + raise ValueError(f"Failed to read from pipe. Flag {event}") data = output.getvalue().strip().decode() return data - def cd(self, path: str, timeout: float = 10.): - self.execute_query(f'cd {path}', timeout) + def cd(self, path: str, timeout: float = 10.0): + self.execute_query(f"cd {path}", timeout) - def ls(self, path: str, timeout: float = 10.) -> list[str]: - return self.execute_query(f'ls {path}', timeout).split(' ') + def ls(self, path: str, timeout: float = 10.0) -> list[str]: + return self.execute_query(f"ls {path}", timeout).split(" ") - def create(self, path: str, value: str, timeout: float = 10.): - self.execute_query(f'create {path} {value}', timeout) + def create(self, path: str, value: str, timeout: float = 10.0): + self.execute_query(f"create {path} {value}", timeout) - def get(self, path: str, timeout: float = 10.) -> str: - return self.execute_query(f'get {path}', timeout) + def get(self, path: str, timeout: float = 10.0) -> str: + return self.execute_query(f"get {path}", timeout) - def exists(self, path: str, timeout: float = 10.) -> bool: - return bool(int(self.execute_query(f'exists {path}', timeout))) + def exists(self, path: str, timeout: float = 10.0) -> bool: + return bool(int(self.execute_query(f"exists {path}", timeout))) def stop(self): if not self.stopped: self.stopped = True - self.proc.communicate(b'exit\n', timeout=10.) + self.proc.communicate(b"exit\n", timeout=10.0) - def sync(self, path: str, timeout: float = 10.): - self.execute_query(f'sync {path}', timeout) + def sync(self, path: str, timeout: float = 10.0): + self.execute_query(f"sync {path}", timeout) - def touch(self, path: str, timeout: float = 10.): - self.execute_query(f'touch {path}', timeout) + def touch(self, path: str, timeout: float = 10.0): + self.execute_query(f"touch {path}", timeout) - def find_big_family(self, path: str, n: int = 10, timeout: float = 10.) -> str: - return self.execute_query(f'find_big_family {path} {n}', timeout) + def find_big_family(self, path: str, n: int = 10, timeout: float = 10.0) -> str: + return self.execute_query(f"find_big_family {path} {n}", timeout) - def find_super_nodes(self, threshold: int, timeout: float = 10.) -> str: - return self.execute_query(f'find_super_nodes {threshold}', timeout) + def find_super_nodes(self, threshold: int, timeout: float = 10.0) -> str: + return self.execute_query(f"find_super_nodes {threshold}", timeout) - def delete_stale_backups(self, timeout: float = 10.) -> str: - return self.execute_query('delete_stale_backups', timeout) + def delete_stale_backups(self, timeout: float = 10.0) -> str: + return self.execute_query("delete_stale_backups", timeout) - def reconfig(self, joining: tp.Optional[str], leaving: tp.Optional[str], new_members: tp.Optional[str], timeout: float = 10.) -> str: + def reconfig( + self, + joining: tp.Optional[str], + leaving: tp.Optional[str], + new_members: tp.Optional[str], + timeout: float = 10.0, + ) -> str: if bool(joining) + bool(leaving) + bool(new_members) != 1: - raise ValueError('Exactly one of joining, leaving or new_members must be specified') + raise ValueError( + "Exactly one of joining, leaving or new_members must be specified" + ) if joining is not None: - operation = 'add' + operation = "add" elif leaving is not None: - operation = 'remove' + operation = "remove" elif new_members is not None: - operation = 'set' + operation = "set" else: - raise ValueError('At least one of joining, leaving or new_members must be specified') + raise ValueError( + "At least one of joining, leaving or new_members must be specified" + ) - return self.execute_query(f'reconfig {operation} {joining or leaving or new_members}', timeout) + return self.execute_query( + f"reconfig {operation} {joining or leaving or new_members}", timeout + ) @classmethod @contextlib.contextmanager - def from_cluster(cls, cluster: ClickHouseCluster, keeper_node: str, port: tp.Optional[int] = None) -> 'KeeperClient': - client = cls(cluster.server_bin_path, cluster.get_instance_ip(keeper_node), port or cluster.zookeeper_port) + def from_cluster( + cls, cluster: ClickHouseCluster, keeper_node: str, port: tp.Optional[int] = None + ) -> "KeeperClient": + client = cls( + cluster.server_bin_path, + cluster.get_instance_ip(keeper_node), + port or cluster.zookeeper_port, + ) try: yield client diff --git a/tests/integration/test_keeper_reconfig_add/test.py b/tests/integration/test_keeper_reconfig_add/test.py index 6e05766c271..724bfdef492 100644 --- a/tests/integration/test_keeper_reconfig_add/test.py +++ b/tests/integration/test_keeper_reconfig_add/test.py @@ -47,7 +47,9 @@ def started_cluster(): def create_client(node: ClickHouseInstance): - return ku.KeeperClient(cluster.server_bin_path, cluster.get_instance_ip(node.name), 9181) + return ku.KeeperClient( + cluster.server_bin_path, cluster.get_instance_ip(node.name), 9181 + ) def test_reconfig_add(): @@ -77,9 +79,7 @@ def test_reconfig_add(): zk1.create(f"/test_three_{i}", "somedata") node2.start_clickhouse() - config = zk1.reconfig( - joining="server.2=node2:9234", leaving=None, new_members=None - ) + config = zk1.reconfig(joining="server.2=node2:9234", leaving=None, new_members=None) ku.wait_until_connected(cluster, node2) print("After adding 2", config) @@ -115,9 +115,7 @@ def test_reconfig_add(): zk2.create(f"/test_four_{i}", "somedata") node3.start_clickhouse() - config = zk2.reconfig( - joining="server.3=node3:9234", leaving=None, new_members=None - ) + config = zk2.reconfig(joining="server.3=node3:9234", leaving=None, new_members=None) ku.wait_until_connected(cluster, node3) print("After adding 3", config) diff --git a/tests/integration/test_keeper_reconfig_remove/test.py b/tests/integration/test_keeper_reconfig_remove/test.py index daab94c59c4..d23d771edba 100644 --- a/tests/integration/test_keeper_reconfig_remove/test.py +++ b/tests/integration/test_keeper_reconfig_remove/test.py @@ -32,7 +32,9 @@ def started_cluster(): def create_client(node: ClickHouseInstance): - return ku.KeeperClient(cluster.server_bin_path, cluster.get_instance_ip(node.name), 9181) + return ku.KeeperClient( + cluster.server_bin_path, cluster.get_instance_ip(node.name), 9181 + ) def test_reconfig_remove_followers_from_3(started_cluster): @@ -123,4 +125,4 @@ def test_reconfig_remove_followers_from_3(started_cluster): assert not node1.contains_in_log(log_msg_removed) assert node2.contains_in_log(log_msg_removed) - assert "Mode: standalone" in zk1.execute_query('stat') + assert "Mode: standalone" in zk1.execute_query("stat") diff --git a/tests/integration/test_keeper_reconfig_remove_many/test.py b/tests/integration/test_keeper_reconfig_remove_many/test.py index e99327e1098..0f9d2fce374 100644 --- a/tests/integration/test_keeper_reconfig_remove_many/test.py +++ b/tests/integration/test_keeper_reconfig_remove_many/test.py @@ -35,7 +35,9 @@ def started_cluster(): def create_client(node: ClickHouseInstance): - return ku.KeeperClient(cluster.server_bin_path, cluster.get_instance_ip(node.name), 9181) + return ku.KeeperClient( + cluster.server_bin_path, cluster.get_instance_ip(node.name), 9181 + ) def test_reconfig_remove_2_and_leader(started_cluster): diff --git a/tests/integration/test_keeper_reconfig_replace_leader/test.py b/tests/integration/test_keeper_reconfig_replace_leader/test.py index bbb5b66eaa5..76a8eb092e2 100644 --- a/tests/integration/test_keeper_reconfig_replace_leader/test.py +++ b/tests/integration/test_keeper_reconfig_replace_leader/test.py @@ -86,9 +86,7 @@ def test_reconfig_replace_leader(started_cluster): ku.wait_configs_equal(config, zk2) node4.start_clickhouse() - config = zk2.reconfig( - joining="server.4=node4:9234", leaving=None, new_members=None - ) + config = zk2.reconfig(joining="server.4=node4:9234", leaving=None, new_members=None) ku.wait_until_connected(cluster, node4) print("After adding 4", config) From c2f4dc0f14f55cf9785da86cb2bf07c556e22d1e Mon Sep 17 00:00:00 2001 From: Alexey Gerasimchuck Date: Thu, 7 Sep 2023 01:32:39 +0000 Subject: [PATCH 169/433] Implemented peak_threads_usage --- docs/en/operations/system-tables/query_log.md | 3 +- docs/ru/operations/system-tables/query_log.md | 3 +- src/Common/ConcurrencyControl.cpp | 2 +- src/Common/ThreadStatus.h | 16 +++- src/Interpreters/ProcessList.cpp | 4 +- src/Interpreters/ProcessList.h | 1 + src/Interpreters/QueryLog.cpp | 3 + src/Interpreters/QueryLog.h | 1 + src/Interpreters/ThreadStatusExt.cpp | 22 ++++- src/Interpreters/executeQuery.cpp | 1 + .../MergeTree/MergePlainMergeTreeTask.cpp | 1 - .../02871_peak_threads_usage.reference | 14 +++ .../0_stateless/02871_peak_threads_usage.sh | 85 +++++++++++++++++++ 13 files changed, 146 insertions(+), 10 deletions(-) create mode 100644 tests/queries/0_stateless/02871_peak_threads_usage.reference create mode 100755 tests/queries/0_stateless/02871_peak_threads_usage.sh diff --git a/docs/en/operations/system-tables/query_log.md b/docs/en/operations/system-tables/query_log.md index 7143520835f..ced97166702 100644 --- a/docs/en/operations/system-tables/query_log.md +++ b/docs/en/operations/system-tables/query_log.md @@ -101,7 +101,8 @@ Columns: - `ProfileEvents` ([Map(String, UInt64)](../../sql-reference/data-types/map.md)) — ProfileEvents that measure different metrics. The description of them could be found in the table [system.events](../../operations/system-tables/events.md#system_tables-events) - `Settings` ([Map(String, String)](../../sql-reference/data-types/map.md)) — Settings that were changed when the client ran the query. To enable logging changes to settings, set the `log_query_settings` parameter to 1. - `log_comment` ([String](../../sql-reference/data-types/string.md)) — Log comment. It can be set to arbitrary string no longer than [max_query_size](../../operations/settings/settings.md#settings-max_query_size). An empty string if it is not defined. -- `thread_ids` ([Array(UInt64)](../../sql-reference/data-types/array.md)) — Thread ids that are participating in query execution. +- `thread_ids` ([Array(UInt64)](../../sql-reference/data-types/array.md)) — Thread ids that are participating in query execution. These threads may not have run simultaneously. +- `peak_threads_usage` ([UInt64)](../../sql-reference/data-types/int-uint.md)) — Maximum count of simultaneous threads executing the query. - `used_aggregate_functions` ([Array(String)](../../sql-reference/data-types/array.md)) — Canonical names of `aggregate functions`, which were used during query execution. - `used_aggregate_function_combinators` ([Array(String)](../../sql-reference/data-types/array.md)) — Canonical names of `aggregate functions combinators`, which were used during query execution. - `used_database_engines` ([Array(String)](../../sql-reference/data-types/array.md)) — Canonical names of `database engines`, which were used during query execution. diff --git a/docs/ru/operations/system-tables/query_log.md b/docs/ru/operations/system-tables/query_log.md index 29793188d3d..8d79aa0eef7 100644 --- a/docs/ru/operations/system-tables/query_log.md +++ b/docs/ru/operations/system-tables/query_log.md @@ -99,7 +99,8 @@ ClickHouse не удаляет данные из таблица автомати - `ProfileEvents` ([Map(String, UInt64)](../../sql-reference/data-types/array.md)) — счетчики для изменения различных метрик. Описание метрик можно получить из таблицы [system.events](#system_tables-events)(#system_tables-events - `Settings` ([Map(String, String)](../../sql-reference/data-types/array.md)) — имена настроек, которые меняются, когда клиент выполняет запрос. Чтобы разрешить логирование изменений настроек, установите параметр `log_query_settings` равным 1. - `log_comment` ([String](../../sql-reference/data-types/string.md)) — комментарий к записи в логе. Представляет собой произвольную строку, длина которой должна быть не больше, чем [max_query_size](../../operations/settings/settings.md#settings-max_query_size). Если нет комментария, то пустая строка. -- `thread_ids` ([Array(UInt64)](../../sql-reference/data-types/array.md)) — идентификаторы потоков, участвующих в обработке запросов. +- `thread_ids` ([Array(UInt64)](../../sql-reference/data-types/array.md)) — идентификаторы потоков, участвующих в обработке запросов, эти потоки не обязательно выполняются одновременно. +- `peak_threads_usage` ([UInt64)](../../sql-reference/data-types/int-uint.md)) — максимальное количество одновременно работавших потоков, участвоваших в обработке запроса. - `used_aggregate_functions` ([Array(String)](../../sql-reference/data-types/array.md)) — канонические имена `агрегатных функций`, использованных при выполнении запроса. - `used_aggregate_function_combinators` ([Array(String)](../../sql-reference/data-types/array.md)) — канонические имена `комбинаторов агрегатных функций`, использованных при выполнении запроса. - `used_database_engines` ([Array(String)](../../sql-reference/data-types/array.md)) — канонические имена `движков баз данных`, использованных при выполнении запроса. diff --git a/src/Common/ConcurrencyControl.cpp b/src/Common/ConcurrencyControl.cpp index de46f0e90ca..ebc1c3f9f42 100644 --- a/src/Common/ConcurrencyControl.cpp +++ b/src/Common/ConcurrencyControl.cpp @@ -57,7 +57,7 @@ ConcurrencyControl::Allocation::Allocation(ConcurrencyControl & parent_, SlotCou *waiter = this; } -// Grant single slot to allocation, returns true iff more slot(s) are required +// Grant single slot to allocation, returns true if more slot(s) are required bool ConcurrencyControl::Allocation::grant() { std::unique_lock lock{mutex}; diff --git a/src/Common/ThreadStatus.h b/src/Common/ThreadStatus.h index 7c8dbdb68bd..02e1f50ff2d 100644 --- a/src/Common/ThreadStatus.h +++ b/src/Common/ThreadStatus.h @@ -107,15 +107,25 @@ public: static ThreadGroupPtr createForBackgroundProcess(ContextPtr storage_context); std::vector getInvolvedThreadIds() const; - void linkThread(UInt64 thread_it); + size_t getPeakThreadsUsage() const; + + void linkThread(UInt64 thread_id); + void unlinkThread(); private: mutable std::mutex mutex; /// Set up at creation, no race when reading - SharedData shared_data; + SharedData shared_data TSA_GUARDED_BY(mutex); + /// Set of all thread ids which has been attached to the group - std::unordered_set thread_ids; + std::unordered_set thread_ids TSA_GUARDED_BY(mutex); + + /// Count of simultaneously working threads + UInt64 active_thread_count TSA_GUARDED_BY(mutex) = 0; + + /// Peak threads count in the group + size_t peak_threads_usage TSA_GUARDED_BY(mutex) = 0; }; /** diff --git a/src/Interpreters/ProcessList.cpp b/src/Interpreters/ProcessList.cpp index c299572ef41..8572470abc1 100644 --- a/src/Interpreters/ProcessList.cpp +++ b/src/Interpreters/ProcessList.cpp @@ -590,8 +590,10 @@ QueryStatusInfo QueryStatus::getInfo(bool get_thread_list, bool get_profile_even res.peak_memory_usage = thread_group->memory_tracker.getPeak(); if (get_thread_list) + { res.thread_ids = thread_group->getInvolvedThreadIds(); - + res.peak_threads_usage = thread_group->getPeakThreadsUsage(); + } if (get_profile_events) res.profile_counters = std::make_shared(thread_group->performance_counters.getPartiallyAtomicSnapshot()); } diff --git a/src/Interpreters/ProcessList.h b/src/Interpreters/ProcessList.h index 2eea49e1267..75a0eaa34bc 100644 --- a/src/Interpreters/ProcessList.h +++ b/src/Interpreters/ProcessList.h @@ -67,6 +67,7 @@ struct QueryStatusInfo /// Optional fields, filled by query std::vector thread_ids; + size_t peak_threads_usage; std::shared_ptr profile_counters; std::shared_ptr query_settings; std::string current_database; diff --git a/src/Interpreters/QueryLog.cpp b/src/Interpreters/QueryLog.cpp index df21e82305a..57f3968fba1 100644 --- a/src/Interpreters/QueryLog.cpp +++ b/src/Interpreters/QueryLog.cpp @@ -118,6 +118,7 @@ NamesAndTypesList QueryLogElement::getNamesAndTypes() {"log_comment", std::make_shared()}, {"thread_ids", std::make_shared(std::make_shared())}, + {"peak_threads_usage", std::make_shared()}, {"ProfileEvents", std::make_shared(low_cardinality_string, std::make_shared())}, {"Settings", std::make_shared(low_cardinality_string, low_cardinality_string)}, @@ -230,6 +231,8 @@ void QueryLogElement::appendToBlock(MutableColumns & columns) const columns[i++]->insert(threads_array); } + columns[i++]->insert(peak_threads_usage); + if (profile_counters) { auto * column = columns[i++].get(); diff --git a/src/Interpreters/QueryLog.h b/src/Interpreters/QueryLog.h index 5bc80280eac..733be5840e5 100644 --- a/src/Interpreters/QueryLog.h +++ b/src/Interpreters/QueryLog.h @@ -91,6 +91,7 @@ struct QueryLogElement String log_comment; std::vector thread_ids; + UInt64 peak_threads_usage{}; std::shared_ptr profile_counters; std::shared_ptr async_read_counters; std::shared_ptr query_settings; diff --git a/src/Interpreters/ThreadStatusExt.cpp b/src/Interpreters/ThreadStatusExt.cpp index 6ee8ec987db..e0b94e8534a 100644 --- a/src/Interpreters/ThreadStatusExt.cpp +++ b/src/Interpreters/ThreadStatusExt.cpp @@ -61,10 +61,26 @@ std::vector ThreadGroup::getInvolvedThreadIds() const return res; } -void ThreadGroup::linkThread(UInt64 thread_it) +size_t ThreadGroup::getPeakThreadsUsage() const { std::lock_guard lock(mutex); - thread_ids.insert(thread_it); + return peak_threads_usage; +} + + +void ThreadGroup::linkThread(UInt64 thread_id) +{ + std::lock_guard lock(mutex); + thread_ids.insert(thread_id); + + ++active_thread_count; + peak_threads_usage = std::max(peak_threads_usage, active_thread_count); +} + +void ThreadGroup::unlinkThread() +{ + std::lock_guard lock(mutex); + --active_thread_count; } ThreadGroupPtr ThreadGroup::createForQuery(ContextPtr query_context_, std::function fatal_error_callback_) @@ -243,6 +259,8 @@ void ThreadStatus::detachFromGroup() /// Extract MemoryTracker out from query and user context memory_tracker.setParent(&total_memory_tracker); + thread_group->unlinkThread(); + thread_group.reset(); query_id_from_query_context.clear(); diff --git a/src/Interpreters/executeQuery.cpp b/src/Interpreters/executeQuery.cpp index bac6807b682..1bfeeaa8ad4 100644 --- a/src/Interpreters/executeQuery.cpp +++ b/src/Interpreters/executeQuery.cpp @@ -241,6 +241,7 @@ addStatusInfoToQueryLogElement(QueryLogElement & element, const QueryStatusInfo element.memory_usage = info.peak_memory_usage > 0 ? info.peak_memory_usage : 0; element.thread_ids = info.thread_ids; + element.peak_threads_usage = info.peak_threads_usage; element.profile_counters = info.profile_counters; /// We need to refresh the access info since dependent views might have added extra information, either during diff --git a/src/Storages/MergeTree/MergePlainMergeTreeTask.cpp b/src/Storages/MergeTree/MergePlainMergeTreeTask.cpp index 3f5753a0c95..c218acce903 100644 --- a/src/Storages/MergeTree/MergePlainMergeTreeTask.cpp +++ b/src/Storages/MergeTree/MergePlainMergeTreeTask.cpp @@ -40,7 +40,6 @@ bool MergePlainMergeTreeTask::executeStep() if (merge_list_entry) { switcher.emplace((*merge_list_entry)->thread_group); - } switch (state) diff --git a/tests/queries/0_stateless/02871_peak_threads_usage.reference b/tests/queries/0_stateless/02871_peak_threads_usage.reference new file mode 100644 index 00000000000..d772a2c4b4e --- /dev/null +++ b/tests/queries/0_stateless/02871_peak_threads_usage.reference @@ -0,0 +1,14 @@ +1 2 1 1 +2 2 1 1 +3 2 1 1 +4 2 1 1 +5 4 1 1 +6 6 1 1 +7 2 1 1 +8 2 1 1 +9 2 1 1 +10 6 1 1 +11 6 1 1 +12 6 1 1 +13 2 1 1 +14 2 1 1 diff --git a/tests/queries/0_stateless/02871_peak_threads_usage.sh b/tests/queries/0_stateless/02871_peak_threads_usage.sh new file mode 100755 index 00000000000..dfb3e665020 --- /dev/null +++ b/tests/queries/0_stateless/02871_peak_threads_usage.sh @@ -0,0 +1,85 @@ +#!/usr/bin/env bash +# Tags: no-parallel +# Tag no-parallel: Avoid using threads in other parallel queries. + +CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) +# shellcheck source=../shell_config.sh +. "$CURDIR"/../shell_config.sh + +QUERY_OPTIONS=( + "--log_query_threads=1" + "--log_queries_min_type=QUERY_FINISH" + "--log_queries=1" + "--format=Null" + "--use_concurrency_control=0" +) + +UNIQUE_QUERY_ID="02871_1_$$" + +# TCPHandler and QueryPullPipeEx threads are always part of the query thread group, but those threads are not within the max_threads limit. +${CLICKHOUSE_CLIENT} --query_id="${UNIQUE_QUERY_ID}_1" --query='SELECT 1' "${QUERY_OPTIONS[@]}" +${CLICKHOUSE_CLIENT} --query_id="${UNIQUE_QUERY_ID}_2" --query='SELECT 1 SETTINGS max_threads = 1' "${QUERY_OPTIONS[@]}" +${CLICKHOUSE_CLIENT} --query_id="${UNIQUE_QUERY_ID}_3" --query='SELECT 1 SETTINGS max_threads = 8' "${QUERY_OPTIONS[@]}" +${CLICKHOUSE_CLIENT} --query_id="${UNIQUE_QUERY_ID}_4" --query='SELECT * FROM numbers_mt(500000) SETTINGS max_threads = 1' "${QUERY_OPTIONS[@]}" +${CLICKHOUSE_CLIENT} --query_id="${UNIQUE_QUERY_ID}_5" --query='SELECT * FROM numbers_mt(500000) SETTINGS max_threads = 2' "${QUERY_OPTIONS[@]}" +${CLICKHOUSE_CLIENT} --query_id="${UNIQUE_QUERY_ID}_6" --query='SELECT * FROM numbers_mt(500000) SETTINGS max_threads = 4' "${QUERY_OPTIONS[@]}" +${CLICKHOUSE_CLIENT} --query_id="${UNIQUE_QUERY_ID}_7" --query='SELECT * FROM numbers_mt(5000), numbers(5000) SETTINGS max_threads = 1, joined_subquery_requires_alias=0' "${QUERY_OPTIONS[@]}" +${CLICKHOUSE_CLIENT} --query_id="${UNIQUE_QUERY_ID}_8" --query='SELECT * FROM numbers_mt(5000), numbers(5000) SETTINGS max_threads = 4, joined_subquery_requires_alias=0' "${QUERY_OPTIONS[@]}" + +${CLICKHOUSE_CLIENT} --query_id="${UNIQUE_QUERY_ID}_9" -mn --query=""" +SELECT count() FROM + (SELECT number FROM numbers_mt(1,100000) + UNION ALL SELECT number FROM numbers_mt(10000, 200000) + UNION ALL SELECT number FROM numbers_mt(30000, 40000) + UNION ALL SELECT number FROM numbers_mt(30000, 40000) + UNION ALL SELECT number FROM numbers_mt(300000, 400000) + UNION ALL SELECT number FROM numbers_mt(300000, 400000) + UNION ALL SELECT number FROM numbers_mt(300000, 4000000) + UNION ALL SELECT number FROM numbers_mt(300000, 4000000) + ) SETTINGS max_threads = 1""" "${QUERY_OPTIONS[@]}" + +${CLICKHOUSE_CLIENT} --query_id="${UNIQUE_QUERY_ID}_10" -mn --query=""" +SELECT count() FROM + (SELECT number FROM numbers_mt(1,100000) + UNION ALL SELECT number FROM numbers_mt(10000, 2000) + UNION ALL SELECT number FROM numbers_mt(30000, 40000) + UNION ALL SELECT number FROM numbers_mt(30000, 40) + UNION ALL SELECT number FROM numbers_mt(300000, 400) + UNION ALL SELECT number FROM numbers_mt(300000, 4000) + UNION ALL SELECT number FROM numbers_mt(300000, 40000) + UNION ALL SELECT number FROM numbers_mt(300000, 4000000) + ) SETTINGS max_threads = 4""" "${QUERY_OPTIONS[@]}" + +${CLICKHOUSE_CLIENT} --query_id="${UNIQUE_QUERY_ID}_11" -mn --query=""" +SELECT count() FROM + (SELECT number FROM numbers_mt(1,100000) + UNION ALL SELECT number FROM numbers_mt(1, 1) + UNION ALL SELECT number FROM numbers_mt(1, 1) + UNION ALL SELECT number FROM numbers_mt(1, 1) + UNION ALL SELECT number FROM numbers_mt(1, 1) + UNION ALL SELECT number FROM numbers_mt(1, 1) + UNION ALL SELECT number FROM numbers_mt(1, 1) + UNION ALL SELECT number FROM numbers_mt(1, 4000000) + ) SETTINGS max_threads = 4""" "${QUERY_OPTIONS[@]}" + +${CLICKHOUSE_CLIENT} --query_id="${UNIQUE_QUERY_ID}_12" -mn --query=""" +SELECT sum(number) FROM numbers_mt(100000) +GROUP BY number % 2 +WITH TOTALS ORDER BY number % 2 +SETTINGS max_threads = 4""" "${QUERY_OPTIONS[@]}" + +${CLICKHOUSE_CLIENT} --query_id="${UNIQUE_QUERY_ID}_13" -mn --query="SELECT * FROM numbers(100000) SETTINGS max_threads = 1" "${QUERY_OPTIONS[@]}" + +${CLICKHOUSE_CLIENT} --query_id="${UNIQUE_QUERY_ID}_14" -mn --query="SELECT * FROM numbers(100000) SETTINGS max_threads = 4" "${QUERY_OPTIONS[@]}" + +${CLICKHOUSE_CLIENT} -q "SYSTEM FLUSH LOGS" +for i in {1..14} +do + ${CLICKHOUSE_CLIENT} -mn --query=""" + SELECT '${i}', + peak_threads_usage, + (select count() from system.query_thread_log WHERE system.query_thread_log.query_id = '${UNIQUE_QUERY_ID}_${i}' AND current_database = currentDatabase()) = length(thread_ids), + length(thread_ids) >= peak_threads_usage + FROM system.query_log + WHERE type = 'QueryFinish' AND query_id = '${UNIQUE_QUERY_ID}_${i}' AND current_database = currentDatabase()" +done From fef80f176ea3f88e6d8219b4448c581a0acda578 Mon Sep 17 00:00:00 2001 From: pufit Date: Wed, 6 Sep 2023 21:56:14 -0400 Subject: [PATCH 170/433] Fix typo --- docs/en/operations/utilities/clickhouse-keeper-client.md | 2 +- programs/keeper-client/Commands.h | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/docs/en/operations/utilities/clickhouse-keeper-client.md b/docs/en/operations/utilities/clickhouse-keeper-client.md index 9f24ba9587b..42fe4fd196e 100644 --- a/docs/en/operations/utilities/clickhouse-keeper-client.md +++ b/docs/en/operations/utilities/clickhouse-keeper-client.md @@ -60,4 +60,4 @@ keeper foo bar - `delete_stale_backups` -- Deletes ClickHouse nodes used for backups that are now inactive - `find_big_family [path] [n]` -- Returns the top n nodes with the biggest family in the subtree (default path = `.` and n = 10) - `sync ` -- Synchronizes node between processes and leader -- `reconfig "" [version]` -- Reconfigures Keeper cluster. See https://clickhouse.com/docs/en/guides/sre/keeper/clickhouse-keeper#reconfiguration +- `reconfig "" [version]` -- Reconfigure Keeper cluster. See https://clickhouse.com/docs/en/guides/sre/keeper/clickhouse-keeper#reconfiguration diff --git a/programs/keeper-client/Commands.h b/programs/keeper-client/Commands.h index 19d5ee2a516..aed98bfa0fb 100644 --- a/programs/keeper-client/Commands.h +++ b/programs/keeper-client/Commands.h @@ -203,7 +203,7 @@ class ReconfigCommand : public IKeeperClientCommand void execute(const ASTKeeperQuery * query, KeeperClient * client) const override; - String getHelpMessage() const override { return "{} \"\" [version] -- Reconfigures Keeper cluster. See https://clickhouse.com/docs/en/guides/sre/keeper/clickhouse-keeper#reconfiguration"; } + String getHelpMessage() const override { return "{} \"\" [version] -- Reconfigure Keeper cluster. See https://clickhouse.com/docs/en/guides/sre/keeper/clickhouse-keeper#reconfiguration"; } }; class SyncCommand: public IKeeperClientCommand From 4b764f7b6564c1723dd0547e377fd0c1c7dbddbd Mon Sep 17 00:00:00 2001 From: Nikolay Degterinsky Date: Thu, 7 Sep 2023 02:10:14 +0000 Subject: [PATCH 171/433] Fix segfault in arrayExists AST optimization --- .../RewriteArrayExistsFunctionVisitor.cpp | 19 +++++++++++++++++++ .../RewriteArrayExistsFunctionVisitor.h | 2 +- .../02834_array_exists_segfault.reference | 0 .../02834_array_exists_segfault.sql | 2 ++ 4 files changed, 22 insertions(+), 1 deletion(-) create mode 100644 tests/queries/0_stateless/02834_array_exists_segfault.reference create mode 100644 tests/queries/0_stateless/02834_array_exists_segfault.sql diff --git a/src/Interpreters/RewriteArrayExistsFunctionVisitor.cpp b/src/Interpreters/RewriteArrayExistsFunctionVisitor.cpp index 8d1a37647ad..eb392cd90a8 100644 --- a/src/Interpreters/RewriteArrayExistsFunctionVisitor.cpp +++ b/src/Interpreters/RewriteArrayExistsFunctionVisitor.cpp @@ -2,6 +2,7 @@ #include #include #include +#include namespace DB { @@ -76,4 +77,22 @@ void RewriteArrayExistsFunctionMatcher::visit(const ASTFunction & func, ASTPtr & } } +bool RewriteArrayExistsFunctionMatcher::needChildVisit(const ASTPtr & ast, const ASTPtr & child) +{ + /// SELECT ... JOIN ... ON arrayExists(x -> x = 1, arr) + /// A JOIN on expression is invalid, ignore it to avoid segmentation fault + + if (auto * join = ast->as()) + { + if (auto * func = child->as()) + { + if (func->name == "arrayExists") + return false; + } + } + + return true; +} + + } diff --git a/src/Interpreters/RewriteArrayExistsFunctionVisitor.h b/src/Interpreters/RewriteArrayExistsFunctionVisitor.h index 763dd2d4df2..ea9363ccee9 100644 --- a/src/Interpreters/RewriteArrayExistsFunctionVisitor.h +++ b/src/Interpreters/RewriteArrayExistsFunctionVisitor.h @@ -18,7 +18,7 @@ public: static void visit(ASTPtr & ast, Data &); static void visit(const ASTFunction &, ASTPtr & ast, Data &); - static bool needChildVisit(const ASTPtr &, const ASTPtr &) { return true; } + static bool needChildVisit(const ASTPtr & ast, const ASTPtr & child); }; using RewriteArrayExistsFunctionVisitor = InDepthNodeVisitor; diff --git a/tests/queries/0_stateless/02834_array_exists_segfault.reference b/tests/queries/0_stateless/02834_array_exists_segfault.reference new file mode 100644 index 00000000000..e69de29bb2d diff --git a/tests/queries/0_stateless/02834_array_exists_segfault.sql b/tests/queries/0_stateless/02834_array_exists_segfault.sql new file mode 100644 index 00000000000..e691cf36505 --- /dev/null +++ b/tests/queries/0_stateless/02834_array_exists_segfault.sql @@ -0,0 +1,2 @@ +CREATE TABLE 02834_t (id UInt64, arr Array(UInt64)) ENGINE = MergeTree ORDER BY id; +WITH subquery AS (SELECT []) SELECT t.* FROM 02834_t AS t JOIN subquery ON arrayExists(x -> x = 1, t.arr); -- { serverError INVALID_JOIN_ON_EXPRESSION } From 9918a85e78eab868590f3be12f89172ae12b4cfd Mon Sep 17 00:00:00 2001 From: Nikolay Degterinsky Date: Thu, 7 Sep 2023 02:35:21 +0000 Subject: [PATCH 172/433] Better --- .../RewriteArrayExistsFunctionVisitor.cpp | 36 +++++++++++++------ .../RewriteArrayExistsFunctionVisitor.h | 2 +- .../02834_array_exists_segfault.sql | 2 ++ 3 files changed, 28 insertions(+), 12 deletions(-) diff --git a/src/Interpreters/RewriteArrayExistsFunctionVisitor.cpp b/src/Interpreters/RewriteArrayExistsFunctionVisitor.cpp index eb392cd90a8..adee5679bdf 100644 --- a/src/Interpreters/RewriteArrayExistsFunctionVisitor.cpp +++ b/src/Interpreters/RewriteArrayExistsFunctionVisitor.cpp @@ -15,6 +15,28 @@ void RewriteArrayExistsFunctionMatcher::visit(ASTPtr & ast, Data & data) visit(*func, ast, data); } + else if (auto * join = ast->as()) + { + if (join->using_expression_list) + { + auto it = std::find(join->children.begin(), join->children.end(), join->using_expression_list); + + visit(join->using_expression_list, data); + + if (it && *it != join->using_expression_list) + *it = join->using_expression_list; + } + + if (join->on_expression) + { + auto it = std::find(join->children.begin(), join->children.end(), join->on_expression); + + visit(join->on_expression, data); + + if (it && *it != join->on_expression) + *it = join->on_expression; + } + } } void RewriteArrayExistsFunctionMatcher::visit(const ASTFunction & func, ASTPtr & ast, Data &) @@ -77,19 +99,11 @@ void RewriteArrayExistsFunctionMatcher::visit(const ASTFunction & func, ASTPtr & } } -bool RewriteArrayExistsFunctionMatcher::needChildVisit(const ASTPtr & ast, const ASTPtr & child) +bool RewriteArrayExistsFunctionMatcher::needChildVisit(const ASTPtr & ast, const ASTPtr &) { - /// SELECT ... JOIN ... ON arrayExists(x -> x = 1, arr) - /// A JOIN on expression is invalid, ignore it to avoid segmentation fault - + /// Children of ASTTableJoin are handled separately in visit() function if (auto * join = ast->as()) - { - if (auto * func = child->as()) - { - if (func->name == "arrayExists") - return false; - } - } + return false; return true; } diff --git a/src/Interpreters/RewriteArrayExistsFunctionVisitor.h b/src/Interpreters/RewriteArrayExistsFunctionVisitor.h index ea9363ccee9..22c28fd005b 100644 --- a/src/Interpreters/RewriteArrayExistsFunctionVisitor.h +++ b/src/Interpreters/RewriteArrayExistsFunctionVisitor.h @@ -18,7 +18,7 @@ public: static void visit(ASTPtr & ast, Data &); static void visit(const ASTFunction &, ASTPtr & ast, Data &); - static bool needChildVisit(const ASTPtr & ast, const ASTPtr & child); + static bool needChildVisit(const ASTPtr & ast, const ASTPtr &); }; using RewriteArrayExistsFunctionVisitor = InDepthNodeVisitor; diff --git a/tests/queries/0_stateless/02834_array_exists_segfault.sql b/tests/queries/0_stateless/02834_array_exists_segfault.sql index e691cf36505..3cf457610fa 100644 --- a/tests/queries/0_stateless/02834_array_exists_segfault.sql +++ b/tests/queries/0_stateless/02834_array_exists_segfault.sql @@ -1,2 +1,4 @@ +DROP TABLE IF EXISTS 02834_t; CREATE TABLE 02834_t (id UInt64, arr Array(UInt64)) ENGINE = MergeTree ORDER BY id; WITH subquery AS (SELECT []) SELECT t.* FROM 02834_t AS t JOIN subquery ON arrayExists(x -> x = 1, t.arr); -- { serverError INVALID_JOIN_ON_EXPRESSION } +DROP TABLE 02834_t; From 1c8216b5c65b06a3c664cd57db790156f4e424f2 Mon Sep 17 00:00:00 2001 From: Alexey Gerasimchuck Date: Thu, 7 Sep 2023 03:27:04 +0000 Subject: [PATCH 173/433] added assertion --- src/Interpreters/ThreadStatusExt.cpp | 1 + 1 file changed, 1 insertion(+) diff --git a/src/Interpreters/ThreadStatusExt.cpp b/src/Interpreters/ThreadStatusExt.cpp index e0b94e8534a..efb8c6792bc 100644 --- a/src/Interpreters/ThreadStatusExt.cpp +++ b/src/Interpreters/ThreadStatusExt.cpp @@ -80,6 +80,7 @@ void ThreadGroup::linkThread(UInt64 thread_id) void ThreadGroup::unlinkThread() { std::lock_guard lock(mutex); + chassert(active_thread_count > 0); --active_thread_count; } From d95c29227f8768316b0c2f56e290ac03ce7ad706 Mon Sep 17 00:00:00 2001 From: Nikolay Degterinsky Date: Thu, 7 Sep 2023 03:46:27 +0000 Subject: [PATCH 174/433] Add STD alias to stddevPop function for MySQL compatibility --- .../aggregate-functions/reference/stddevpop.md | 6 +++++- .../aggregate-functions/reference/stddevsamp.md | 4 +++- .../sql-reference/aggregate-functions/reference/varpop.md | 4 +++- .../sql-reference/aggregate-functions/reference/varsamp.md | 4 +++- src/AggregateFunctions/AggregateFunctionSecondMoment.cpp | 1 + 5 files changed, 15 insertions(+), 4 deletions(-) diff --git a/docs/en/sql-reference/aggregate-functions/reference/stddevpop.md b/docs/en/sql-reference/aggregate-functions/reference/stddevpop.md index 7ad7e37e5c2..0ddbf700bd3 100644 --- a/docs/en/sql-reference/aggregate-functions/reference/stddevpop.md +++ b/docs/en/sql-reference/aggregate-functions/reference/stddevpop.md @@ -7,6 +7,10 @@ sidebar_position: 30 The result is equal to the square root of [varPop](../../../sql-reference/aggregate-functions/reference/varpop.md). -:::note +Alias: +- `STD` +- `STDDEV_POP` + +:::note This function uses a numerically unstable algorithm. If you need [numerical stability](https://en.wikipedia.org/wiki/Numerical_stability) in calculations, use the `stddevPopStable` function. It works slower but provides a lower computational error. ::: \ No newline at end of file diff --git a/docs/en/sql-reference/aggregate-functions/reference/stddevsamp.md b/docs/en/sql-reference/aggregate-functions/reference/stddevsamp.md index 068725c4991..01484c2b02e 100644 --- a/docs/en/sql-reference/aggregate-functions/reference/stddevsamp.md +++ b/docs/en/sql-reference/aggregate-functions/reference/stddevsamp.md @@ -7,6 +7,8 @@ sidebar_position: 31 The result is equal to the square root of [varSamp](../../../sql-reference/aggregate-functions/reference/varsamp.md). -:::note +Alias: `STDDEV_SAMP`. + +:::note This function uses a numerically unstable algorithm. If you need [numerical stability](https://en.wikipedia.org/wiki/Numerical_stability) in calculations, use the `stddevSampStable` function. It works slower but provides a lower computational error. ::: \ No newline at end of file diff --git a/docs/en/sql-reference/aggregate-functions/reference/varpop.md b/docs/en/sql-reference/aggregate-functions/reference/varpop.md index 0a665c83e74..751688b0830 100644 --- a/docs/en/sql-reference/aggregate-functions/reference/varpop.md +++ b/docs/en/sql-reference/aggregate-functions/reference/varpop.md @@ -9,6 +9,8 @@ Calculates the amount `Σ((x - x̅)^2) / n`, where `n` is the sample size and `x In other words, dispersion for a set of values. Returns `Float64`. -:::note +Alias: `VAR_POP`. + +:::note This function uses a numerically unstable algorithm. If you need [numerical stability](https://en.wikipedia.org/wiki/Numerical_stability) in calculations, use the `varPopStable` function. It works slower but provides a lower computational error. ::: \ No newline at end of file diff --git a/docs/en/sql-reference/aggregate-functions/reference/varsamp.md b/docs/en/sql-reference/aggregate-functions/reference/varsamp.md index 76639d2d7a0..9b2b94936ec 100644 --- a/docs/en/sql-reference/aggregate-functions/reference/varsamp.md +++ b/docs/en/sql-reference/aggregate-functions/reference/varsamp.md @@ -11,6 +11,8 @@ It represents an unbiased estimate of the variance of a random variable if passe Returns `Float64`. When `n <= 1`, returns `+∞`. -:::note +Alias: `VAR_SAMP`. + +:::note This function uses a numerically unstable algorithm. If you need [numerical stability](https://en.wikipedia.org/wiki/Numerical_stability) in calculations, use the `varSampStable` function. It works slower but provides a lower computational error. ::: diff --git a/src/AggregateFunctions/AggregateFunctionSecondMoment.cpp b/src/AggregateFunctions/AggregateFunctionSecondMoment.cpp index 123baac3e37..80fbe2511d9 100644 --- a/src/AggregateFunctions/AggregateFunctionSecondMoment.cpp +++ b/src/AggregateFunctions/AggregateFunctionSecondMoment.cpp @@ -19,6 +19,7 @@ void registerAggregateFunctionsStatisticsSecondMoment(AggregateFunctionFactory & factory.registerAlias("VAR_POP", "varPop", AggregateFunctionFactory::CaseInsensitive); factory.registerAlias("STDDEV_SAMP", "stddevSamp", AggregateFunctionFactory::CaseInsensitive); factory.registerAlias("STDDEV_POP", "stddevPop", AggregateFunctionFactory::CaseInsensitive); + factory.registerAlias("STD", "stddevPop", AggregateFunctionFactory::CaseInsensitive); } } From f1e4ca4da5565f8be79d3aca08ac5fc65c57c58e Mon Sep 17 00:00:00 2001 From: Alexey Gerasimchuck Date: Thu, 7 Sep 2023 04:56:13 +0000 Subject: [PATCH 175/433] changed data_type --- src/Common/ThreadStatus.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/Common/ThreadStatus.h b/src/Common/ThreadStatus.h index 02e1f50ff2d..970bb5cd6e6 100644 --- a/src/Common/ThreadStatus.h +++ b/src/Common/ThreadStatus.h @@ -122,7 +122,7 @@ private: std::unordered_set thread_ids TSA_GUARDED_BY(mutex); /// Count of simultaneously working threads - UInt64 active_thread_count TSA_GUARDED_BY(mutex) = 0; + size_t active_thread_count TSA_GUARDED_BY(mutex) = 0; /// Peak threads count in the group size_t peak_threads_usage TSA_GUARDED_BY(mutex) = 0; From 2c85c28113671ee7cceb334626e5b35c001f45a1 Mon Sep 17 00:00:00 2001 From: Igor Nikonov Date: Thu, 7 Sep 2023 07:14:40 +0000 Subject: [PATCH 176/433] Cleanup --- .../test_parallel_replicas_over_distributed/test.py | 12 ------------ 1 file changed, 12 deletions(-) diff --git a/tests/integration/test_parallel_replicas_over_distributed/test.py b/tests/integration/test_parallel_replicas_over_distributed/test.py index 1df2126a4ff..50db95c90b2 100644 --- a/tests/integration/test_parallel_replicas_over_distributed/test.py +++ b/tests/integration/test_parallel_replicas_over_distributed/test.py @@ -102,26 +102,14 @@ def create_tables(cluster, table_name): @pytest.mark.parametrize( "cluster,max_parallel_replicas,prefer_localhost_replica", [ - # prefer_localhost_replica=0 pytest.param("test_single_shard_multiple_replicas", 2, 0), pytest.param("test_single_shard_multiple_replicas", 3, 0), pytest.param("test_single_shard_multiple_replicas", 4, 0), pytest.param("test_single_shard_multiple_replicas", 10, 0), - # prefer_localhost_replica=1 - # pytest.param("test_single_shard_multiple_replicas", 2, 1), - # pytest.param("test_single_shard_multiple_replicas", 3, 1), - # pytest.param("test_single_shard_multiple_replicas", 4, 1), - # pytest.param("test_single_shard_multiple_replicas", 10, 1), - # prefer_localhost_replica=0 pytest.param("test_multiple_shards_multiple_replicas", 2, 0), pytest.param("test_multiple_shards_multiple_replicas", 3, 0), pytest.param("test_multiple_shards_multiple_replicas", 4, 0), pytest.param("test_multiple_shards_multiple_replicas", 10, 0), - # prefer_localhost_replica=1 - # pytest.param("test_multiple_shards_multiple_replicas", 2, 1), - # pytest.param("test_multiple_shards_multiple_replicas", 3, 1), - # pytest.param("test_multiple_shards_multiple_replicas", 4, 1), - # pytest.param("test_multiple_shards_multiple_replicas", 10, 1), ], ) def test_parallel_replicas_over_distributed( From 14f77a00d4ba6d44c951f4b698298ca66409fc01 Mon Sep 17 00:00:00 2001 From: Nikolay Degterinsky Date: Thu, 7 Sep 2023 07:48:08 +0000 Subject: [PATCH 177/433] Add test --- tests/queries/0_stateless/02833_std_alias.reference | 2 ++ tests/queries/0_stateless/02833_std_alias.sql | 8 ++++++++ 2 files changed, 10 insertions(+) create mode 100644 tests/queries/0_stateless/02833_std_alias.reference create mode 100644 tests/queries/0_stateless/02833_std_alias.sql diff --git a/tests/queries/0_stateless/02833_std_alias.reference b/tests/queries/0_stateless/02833_std_alias.reference new file mode 100644 index 00000000000..d8e5e1556a7 --- /dev/null +++ b/tests/queries/0_stateless/02833_std_alias.reference @@ -0,0 +1,2 @@ +6.408619196051518 5.848925577403085 +6.408619196051518 5.848925577403085 diff --git a/tests/queries/0_stateless/02833_std_alias.sql b/tests/queries/0_stateless/02833_std_alias.sql new file mode 100644 index 00000000000..256990f3f3a --- /dev/null +++ b/tests/queries/0_stateless/02833_std_alias.sql @@ -0,0 +1,8 @@ +DROP TABLE IF EXISTS series; +CREATE TABLE series(i UInt32, x Float64, y Float64) ENGINE = Memory; +INSERT INTO series(i, x, y) VALUES (1, 5.6,-4.4),(2, -9.6,3),(3, -1.3,-4),(4, 5.3,9.7),(5, 4.4,0.037),(6, -8.6,-7.8),(7, 5.1,9.3),(8, 7.9,-3.6),(9, -8.2,0.62),(10, -3,7.3); + +SELECT std(x), std(y) FROM series; +SELECT stddevPop(x), stddevPop(y) FROM series; + +DROP TABLE series; From c795b9c32f9c6eba1d0fd36fc4681691eab796a6 Mon Sep 17 00:00:00 2001 From: Antonio Andelic Date: Thu, 7 Sep 2023 07:54:06 +0000 Subject: [PATCH 178/433] Check for overflow before addition in anova function --- src/AggregateFunctions/Moments.h | 4 ++++ .../0_stateless/02874_analysis_of_variance_overflow.reference | 0 .../0_stateless/02874_analysis_of_variance_overflow.sql | 1 + 3 files changed, 5 insertions(+) create mode 100644 tests/queries/0_stateless/02874_analysis_of_variance_overflow.reference create mode 100644 tests/queries/0_stateless/02874_analysis_of_variance_overflow.sql diff --git a/src/AggregateFunctions/Moments.h b/src/AggregateFunctions/Moments.h index a34f960794c..099bc11d0c8 100644 --- a/src/AggregateFunctions/Moments.h +++ b/src/AggregateFunctions/Moments.h @@ -459,6 +459,10 @@ struct AnalysisOfVarianceMoments void add(T value, size_t group) { + if (group == std::numeric_limits::max()) + throw Exception(ErrorCodes::BAD_ARGUMENTS, "Too many groups for analysis of variance (should be no more than {}, got {})", + MAX_GROUPS_NUMBER, group); + resizeIfNeeded(group + 1); xs1[group] += value; xs2[group] += value * value; diff --git a/tests/queries/0_stateless/02874_analysis_of_variance_overflow.reference b/tests/queries/0_stateless/02874_analysis_of_variance_overflow.reference new file mode 100644 index 00000000000..e69de29bb2d diff --git a/tests/queries/0_stateless/02874_analysis_of_variance_overflow.sql b/tests/queries/0_stateless/02874_analysis_of_variance_overflow.sql new file mode 100644 index 00000000000..67fb4d28acb --- /dev/null +++ b/tests/queries/0_stateless/02874_analysis_of_variance_overflow.sql @@ -0,0 +1 @@ +SELECT analysisOfVariance(1, 18446744073709551615); -- { serverError BAD_ARGUMENTS } From b067213c5a5f460ac80e95875c34a80e4ef1ac50 Mon Sep 17 00:00:00 2001 From: Mikhail Koviazin Date: Thu, 7 Sep 2023 08:08:02 +0000 Subject: [PATCH 179/433] aklomp-base64: fix the submodule name --- .gitmodules | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/.gitmodules b/.gitmodules index 63f0221c2f1..caed96b0320 100644 --- a/.gitmodules +++ b/.gitmodules @@ -40,9 +40,6 @@ [submodule "contrib/boost"] path = contrib/boost url = https://github.com/ClickHouse/boost -[submodule "contrib/base64"] - path = contrib/aklomp-base64 - url = https://github.com/aklomp/base64.git [submodule "contrib/arrow"] path = contrib/arrow url = https://github.com/ClickHouse/arrow @@ -348,3 +345,6 @@ [submodule "contrib/robin-map"] path = contrib/robin-map url = https://github.com/Tessil/robin-map.git +[submodule "contrib/aklomp-base64"] + path = contrib/aklomp-base64 + url = https://github.com/aklomp/base64.git From fd76bc6412e97b72940110cbbfd65a79bdea23a8 Mon Sep 17 00:00:00 2001 From: Han Fei Date: Thu, 7 Sep 2023 10:12:39 +0200 Subject: [PATCH 180/433] try to fix build for freeBSD --- contrib/libunwind | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/contrib/libunwind b/contrib/libunwind index 2a82f19272f..f12a21bdd60 160000 --- a/contrib/libunwind +++ b/contrib/libunwind @@ -1 +1 @@ -Subproject commit 2a82f19272f37d345001f6568ee32655aeb2e54d +Subproject commit f12a21bdd601232465791a32ed054bb1dded8e02 From d11a9b4175e540192a781f73dced9e4ea9cdfa19 Mon Sep 17 00:00:00 2001 From: serxa Date: Thu, 7 Sep 2023 08:51:35 +0000 Subject: [PATCH 181/433] update spell dict --- utils/check-style/aspell-ignore/en/aspell-dict.txt | 12 ++++++++---- 1 file changed, 8 insertions(+), 4 deletions(-) diff --git a/utils/check-style/aspell-ignore/en/aspell-dict.txt b/utils/check-style/aspell-ignore/en/aspell-dict.txt index 2c29fd9369e..231110c4f3b 100644 --- a/utils/check-style/aspell-ignore/en/aspell-dict.txt +++ b/utils/check-style/aspell-ignore/en/aspell-dict.txt @@ -1329,10 +1329,10 @@ ddl deallocation deallocations debian +decodeHTMLComponent decodeURLComponent decodeURLFormComponent decodeXMLComponent -decodeHTMLComponent decompressor decrypt decrypted @@ -1346,17 +1346,18 @@ defaultRoles defaultValueOfArgumentType defaultValueOfTypeName delim -deltaLake -deltaSum -deltaSumTimestamp deltalake +deltaLake deltasum +deltaSum deltasumtimestamp +deltaSumTimestamp demangle denormalize denormalized denormalizing denormals +dequeued deserialization deserialized deserializing @@ -1441,6 +1442,7 @@ farmFingerprint farmHash fastops fcoverage +fifo filesystem filesystemAvailable filesystemCapacity @@ -1607,6 +1609,7 @@ incrementing indexHint indexOf infi +inflight initcap initcapUTF initialQueryID @@ -2536,6 +2539,7 @@ visitParamExtractRaw visitParamExtractString visitParamExtractUInt visitParamHas +vruntime wchc wchs webpage From c36a3fbd3900379f5c1efa80807e01a4fa1329fe Mon Sep 17 00:00:00 2001 From: zvonand Date: Thu, 7 Sep 2023 12:00:25 +0300 Subject: [PATCH 182/433] renamed setting --- docs/en/operations/settings/settings.md | 6 ++--- docs/en/sql-reference/table-functions/file.md | 4 ++-- docs/en/sql-reference/table-functions/hdfs.md | 4 ++-- docs/ru/operations/settings/settings.md | 6 ++--- docs/ru/sql-reference/table-functions/file.md | 2 +- docs/ru/sql-reference/table-functions/hdfs.md | 4 ++-- src/Core/Settings.h | 2 +- src/Storages/HDFS/StorageHDFS.cpp | 20 ++++++++-------- src/Storages/StorageFile.cpp | 24 +++++++++---------- ...02771_multidirectory_globs_storage_file.sh | 2 +- 10 files changed, 37 insertions(+), 37 deletions(-) diff --git a/docs/en/operations/settings/settings.md b/docs/en/operations/settings/settings.md index ac53b46da7e..2fac021fe2d 100644 --- a/docs/en/operations/settings/settings.md +++ b/docs/en/operations/settings/settings.md @@ -4660,7 +4660,7 @@ The default value is `false`. true ``` -## ignore_eacces_multidirectory_globs {#ignore_eacces_multidirectory_globs} +## ignore_access_denied_multidirectory_globs {#ignore_access_denied_multidirectory_globs} Allows to ignore 'permission denied' errors when using multi-directory `{}` globs for [File](../../sql-reference/table-functions/file.md#globs_in_path) and [HDFS](../../sql-reference/table-functions/hdfs.md) storages. This setting is only applicable when multi directory `{}` glob is used. @@ -4689,12 +4689,12 @@ It happens because a multi-directory glob requires a recursive search in _all_ a If this setting is on, all inaccessible directories will be silently skipped, even if they are explicitly specified inside `{}`. ```sql -SELECT _path, _file FROM file('my_directory/{data1/f1,data2/f2}.csv', CSV) SETTINGS ignore_eacces_multidirectory_globs = 0; +SELECT _path, _file FROM file('my_directory/{data1/f1,data2/f2}.csv', CSV) SETTINGS ignore_access_denied_multidirectory_globs = 0; Code: 1001. DB::Exception: std::__1::__fs::filesystem::filesystem_error: filesystem error: in directory_iterator::directory_iterator(...): Permission denied ``` ```sql -SELECT _path, _file FROM file('my_directory/{data1/f1,data2/f2}.csv', CSV) SETTINGS ignore_eacces_multidirectory_globs = 1; +SELECT _path, _file FROM file('my_directory/{data1/f1,data2/f2}.csv', CSV) SETTINGS ignore_access_denied_multidirectory_globs = 1; ┌─_path───────────────────┬─_file───────┐ │ │ │ diff --git a/docs/en/sql-reference/table-functions/file.md b/docs/en/sql-reference/table-functions/file.md index f12ac732c22..06571fd724b 100644 --- a/docs/en/sql-reference/table-functions/file.md +++ b/docs/en/sql-reference/table-functions/file.md @@ -141,7 +141,7 @@ Multiple path components can have globs. For being processed file must exist and - `*` — Substitutes any number of any characters except `/` including empty string. - `?` — Substitutes any single character. -- `{some_string,another_string,yet_another_one}` — Substitutes any of strings `'some_string', 'another_string', 'yet_another_one'`. In case at least one of strings contains `/`, `'permission denied'` errors may be ignored using [ignore_eacces_multidirectory_globs](/docs/en/operations/settings/settings.md#ignore_eacces_multidirectory_globs) setting for file & HDFS. +- `{some_string,another_string,yet_another_one}` — Substitutes any of strings `'some_string', 'another_string', 'yet_another_one'`. In case at least one of strings contains `/`, `'permission denied'` errors may be ignored using [ignore_access_denied_multidirectory_globs](/docs/en/operations/settings/settings.md#ignore_access_denied_multidirectory_globs) setting for file & HDFS. - `{N..M}` — Substitutes any number in range from N to M including both borders. - `**` - Fetches all files inside the folder recursively. @@ -210,7 +210,7 @@ SELECT count(*) FROM file('big_dir/**/file002', 'CSV', 'name String, value UInt3 - [engine_file_allow_create_multiple_files](/docs/en/operations/settings/settings.md#engine_file_allow_create_multiple_files) - allows to create a new file on each insert if format has suffix. Disabled by default. - [engine_file_skip_empty_files](/docs/en/operations/settings/settings.md#engine_file_skip_empty_files) - allows to skip empty files while reading. Disabled by default. - [storage_file_read_method](/docs/en/operations/settings/settings.md#engine-file-emptyif-not-exists) - method of reading data from storage file, one of: read, pread, mmap (only for clickhouse-local). Default value: `pread` for clickhouse-server, `mmap` for clickhouse-local. -- [ignore_eacces_multidirectory_globs](/docs/en/operations/settings/settings.md#ignore_eacces_multidirectory_globs) - allows to ignore permission denied errors for multi-directory globs. +- [ignore_access_denied_multidirectory_globs](/docs/en/operations/settings/settings.md#ignore_access_denied_multidirectory_globs) - allows to ignore permission denied errors for multi-directory globs. **See Also** diff --git a/docs/en/sql-reference/table-functions/hdfs.md b/docs/en/sql-reference/table-functions/hdfs.md index 0e4ee632b71..b50356202f7 100644 --- a/docs/en/sql-reference/table-functions/hdfs.md +++ b/docs/en/sql-reference/table-functions/hdfs.md @@ -45,7 +45,7 @@ Multiple path components can have globs. For being processed file should exists - `*` — Substitutes any number of any characters except `/` including empty string. - `?` — Substitutes any single character. -- `{some_string,another_string,yet_another_one}` — Substitutes any of strings `'some_string', 'another_string', 'yet_another_one'`. In case at least one of strings contains `/`, `'permission denied'` errors may be ignored using [ignore_eacces_multidirectory_globs](/docs/en/operations/settings/settings.md#ignore_eacces_multidirectory_globs) setting. +- `{some_string,another_string,yet_another_one}` — Substitutes any of strings `'some_string', 'another_string', 'yet_another_one'`. In case at least one of strings contains `/`, `'permission denied'` errors may be ignored using [ignore_access_denied_multidirectory_globs](/docs/en/operations/settings/settings.md#ignore_access_denied_multidirectory_globs) setting. - `{N..M}` — Substitutes any number in range from N to M including both borders. Constructions with `{}` are similar to the [remote table function](../../sql-reference/table-functions/remote.md)). @@ -102,7 +102,7 @@ FROM hdfs('hdfs://hdfs1:9000/big_dir/file{0..9}{0..9}{0..9}', 'CSV', 'name Strin - [hdfs_truncate_on_insert](/docs/en/operations/settings/settings.md#hdfs-truncate-on-insert) - allows to truncate file before insert into it. Disabled by default. - [hdfs_create_multiple_files](/docs/en/operations/settings/settings.md#hdfs_allow_create_multiple_files) - allows to create a new file on each insert if format has suffix. Disabled by default. - [hdfs_skip_empty_files](/docs/en/operations/settings/settings.md#hdfs_skip_empty_files) - allows to skip empty files while reading. Disabled by default. -- [ignore_eacces_multidirectory_globs](/docs/en/operations/settings/settings.md#ignore_eacces_multidirectory_globs) - allows to ignore permission denied errors for multi-directory globs. +- [ignore_access_denied_multidirectory_globs](/docs/en/operations/settings/settings.md#ignore_access_denied_multidirectory_globs) - allows to ignore permission denied errors for multi-directory globs. **See Also** diff --git a/docs/ru/operations/settings/settings.md b/docs/ru/operations/settings/settings.md index 3f0bc4ee2ec..73ffb9d6a12 100644 --- a/docs/ru/operations/settings/settings.md +++ b/docs/ru/operations/settings/settings.md @@ -4207,7 +4207,7 @@ SELECT toFloat64('1.7091'), toFloat64('1.5008753E7') SETTINGS precise_float_pars └─────────────────────┴──────────────────────────┘ ``` -## ignore_eacces_multidirectory_globs {#ignore_eacces_multidirectory_globs} +## ignore_access_denied_multidirectory_globs {#ignore_access_denied_multidirectory_globs} Позволяет игнорировать ошибку 'permission denied', возникающую при использовании шаблона `{}`, содержащего `/` внутри себя. Работает для [File](../../sql-reference/table-functions/file.md#globs_in_path) и [HDFS](../../sql-reference/table-functions/hdfs.md). @@ -4237,12 +4237,12 @@ my_directory/ Если данная настройка имеет значение 1, то недоступные директории будут тихо пропущены, даже если они явно указаны внутри `{}`. ```sql -SELECT _path, _file FROM file('my_directory/{data1/f1,data2/f2}.csv', CSV) SETTINGS ignore_eacces_multidirectory_globs = 0; +SELECT _path, _file FROM file('my_directory/{data1/f1,data2/f2}.csv', CSV) SETTINGS ignore_access_denied_multidirectory_globs = 0; Code: 1001. DB::Exception: std::__1::__fs::filesystem::filesystem_error: filesystem error: in directory_iterator::directory_iterator(...): Permission denied ``` ```sql -SELECT _path, _file FROM file('my_directory/{data1/f1,data2/f2}.csv', CSV) SETTINGS ignore_eacces_multidirectory_globs = 1; +SELECT _path, _file FROM file('my_directory/{data1/f1,data2/f2}.csv', CSV) SETTINGS ignore_access_denied_multidirectory_globs = 1; ┌─_path───────────────────┬─_file───────┐ │ │ │ diff --git a/docs/ru/sql-reference/table-functions/file.md b/docs/ru/sql-reference/table-functions/file.md index 32d98bd19ee..4c12b7cca66 100644 --- a/docs/ru/sql-reference/table-functions/file.md +++ b/docs/ru/sql-reference/table-functions/file.md @@ -79,7 +79,7 @@ SELECT * FROM file('test.csv', 'CSV', 'column1 UInt32, column2 UInt32, column3 U - `*` — заменяет любое количество любых символов кроме `/`, включая отсутствие символов. - `?` — заменяет ровно один любой символ. -- `{some_string,another_string,yet_another_one}` — заменяет любую из строк `'some_string', 'another_string', 'yet_another_one'`. В случае, если в какой-либо из строк содержится `/`, то ошибки доступа (permission denied) к существующим, но недоступным директориям/файлам могут быть проигнорированы при помощи настройки [ignore_eacces_multidirectory_globs](/docs/ru/operations/settings/settings.md#ignore_eacces_multidirectory_globs). +- `{some_string,another_string,yet_another_one}` — заменяет любую из строк `'some_string', 'another_string', 'yet_another_one'`. В случае, если в какой-либо из строк содержится `/`, то ошибки доступа (permission denied) к существующим, но недоступным директориям/файлам могут быть проигнорированы при помощи настройки [ignore_access_denied_multidirectory_globs](/docs/ru/operations/settings/settings.md#ignore_access_denied_multidirectory_globs). - `{N..M}` — заменяет любое число в интервале от `N` до `M` включительно (может содержать ведущие нули). Конструкция с `{}` аналогична табличной функции [remote](remote.md). diff --git a/docs/ru/sql-reference/table-functions/hdfs.md b/docs/ru/sql-reference/table-functions/hdfs.md index d74248bd93c..98cb1fbfd8a 100644 --- a/docs/ru/sql-reference/table-functions/hdfs.md +++ b/docs/ru/sql-reference/table-functions/hdfs.md @@ -43,7 +43,7 @@ LIMIT 2 - `*` — Заменяет любое количество любых символов кроме `/`, включая отсутствие символов. - `?` — Заменяет ровно один любой символ. -- `{some_string,another_string,yet_another_one}` — Заменяет любую из строк `'some_string', 'another_string', 'yet_another_one'`. В случае, если в какой-либо из строк содержится `/`, то ошибки доступа (permission denied) к существующим, но недоступным директориям/файлам могут быть проигнорированы при помощи настройки [ignore_eacces_multidirectory_globs](/docs/ru/operations/settings/settings.md#ignore_eacces_multidirectory_globs). +- `{some_string,another_string,yet_another_one}` — Заменяет любую из строк `'some_string', 'another_string', 'yet_another_one'`. В случае, если в какой-либо из строк содержится `/`, то ошибки доступа (permission denied) к существующим, но недоступным директориям/файлам могут быть проигнорированы при помощи настройки [ignore_access_denied_multidirectory_globs](/docs/ru/operations/settings/settings.md#ignore_access_denied_multidirectory_globs). - `{N..M}` — Заменяет любое число в интервале от `N` до `M` включительно (может содержать ведущие нули). Конструкция с `{}` аналогична табличной функции [remote](remote.md). @@ -61,5 +61,5 @@ LIMIT 2 **Смотрите также** - [Виртуальные столбцы](index.md#table_engines-virtual_columns) -- Параметр [ignore_eacces_multidirectory_globs](/docs/ru/operations/settings/settings.md#ignore_eacces_multidirectory_globs) +- Параметр [ignore_access_denied_multidirectory_globs](/docs/ru/operations/settings/settings.md#ignore_access_denied_multidirectory_globs) diff --git a/src/Core/Settings.h b/src/Core/Settings.h index 4706f8914f4..441498d50f8 100644 --- a/src/Core/Settings.h +++ b/src/Core/Settings.h @@ -797,7 +797,7 @@ class IColumn; M(Timezone, session_timezone, "", "This setting can be removed in the future due to potential caveats. It is experimental and is not suitable for production usage. The default timezone for current session or query. The server default timezone if empty.", 0) \ M(Bool, allow_create_index_without_type, false, "Allow CREATE INDEX query without TYPE. Query will be ignored. Made for SQL compatibility tests.", 0)\ M(Bool, create_index_ignore_unique, false, "Ignore UNIQUE keyword in CREATE UNIQUE INDEX. Made for SQL compatibility tests.", 0) \ - M(Bool, ignore_eacces_multidirectory_globs, false, "Ignore access denied errors when processing multi-directory globs for file & HDFS.", 0)\ + M(Bool, ignore_access_denied_multidirectory_globs, false, "Ignore access denied errors when processing multi-directory globs for file & HDFS.", 0)\ // End of COMMON_SETTINGS // Please add settings related to formats into the FORMAT_FACTORY_SETTINGS and move obsolete settings to OBSOLETE_SETTINGS. diff --git a/src/Storages/HDFS/StorageHDFS.cpp b/src/Storages/HDFS/StorageHDFS.cpp index 09c205930dc..fc904c83413 100644 --- a/src/Storages/HDFS/StorageHDFS.cpp +++ b/src/Storages/HDFS/StorageHDFS.cpp @@ -74,7 +74,7 @@ namespace std::vector LSWithRegexpMatching(const String & path_for_ls, const HDFSFSPtr & fs, const String & for_match, - bool ignore_eacces_multidirectory_globs); + bool ignore_access_denied_multidirectory_globs); /* * When `{...}` has any `/`s, it must be processed in a different way: @@ -91,7 +91,7 @@ namespace re2::RE2 & matcher, const size_t max_depth, const size_t next_slash_after_glob_pos, - bool ignore_eacces_multidirectory_globs) + bool ignore_access_denied_multidirectory_globs) { /// We don't need to go all the way in every directory if max_depth is reached /// as it is upper limit of depth by simply counting `/`s in curly braces @@ -104,8 +104,8 @@ namespace { // ignore file not found (as in LSWithRegexpMatching) // keep throw other exception, libhdfs3 doesn't have function to get exception type, so use errno. - // ignore permission denied if ignore_eacces_multidirectory_globs is true - if (!(ignore_eacces_multidirectory_globs && errno == EACCES)) + // ignore permission denied if ignore_access_denied_multidirectory_globs is true + if (!(ignore_access_denied_multidirectory_globs && errno == EACCES)) throw Exception( ErrorCodes::ACCESS_DENIED, "Cannot list directory {}: {}", path_for_ls, String(hdfsGetLastError())); } @@ -134,7 +134,7 @@ namespace { std::vector result_part = LSWithRegexpMatching( fs::path(full_path) / "" , fs, suffix_with_globs.substr(next_slash_after_glob_pos), - ignore_eacces_multidirectory_globs); + ignore_access_denied_multidirectory_globs); std::move(result_part.begin(), result_part.end(), std::back_inserter(result)); } } @@ -142,7 +142,7 @@ namespace { std::vector result_part = LSWithFoldedRegexpMatching( fs::path(full_path), fs, processed_suffix + dir_or_file_name, suffix_with_globs, - matcher, max_depth - 1, next_slash_after_glob_pos, ignore_eacces_multidirectory_globs); + matcher, max_depth - 1, next_slash_after_glob_pos, ignore_access_denied_multidirectory_globs); std::move(result_part.begin(), result_part.end(), std::back_inserter(result)); } } @@ -156,7 +156,7 @@ namespace const String & path_for_ls, const HDFSFSPtr & fs, const String & for_match, - bool ignore_eacces_multidirectory_globs) + bool ignore_access_denied_multidirectory_globs) { const size_t first_glob_pos = for_match.find_first_of("*?{"); const bool has_glob = first_glob_pos != std::string::npos; @@ -199,7 +199,7 @@ namespace if (slashes_in_glob) { return LSWithFoldedRegexpMatching(fs::path(prefix_without_globs), fs, "", suffix_with_globs, matcher, - slashes_in_glob, next_slash_after_glob_pos, ignore_eacces_multidirectory_globs); + slashes_in_glob, next_slash_after_glob_pos, ignore_access_denied_multidirectory_globs); } HDFSFileInfo ls; @@ -233,7 +233,7 @@ namespace if (re2::RE2::FullMatch(file_name, matcher)) { std::vector result_part = LSWithRegexpMatching(fs::path(full_path) / "", fs, - suffix_with_globs.substr(next_slash_after_glob_pos), ignore_eacces_multidirectory_globs); + suffix_with_globs.substr(next_slash_after_glob_pos), ignore_access_denied_multidirectory_globs); /// Recursion depth is limited by pattern. '*' works only for depth = 1, for depth = 2 pattern path is '*/*'. So we do not need additional check. std::move(result_part.begin(), result_part.end(), std::back_inserter(result)); } @@ -261,7 +261,7 @@ namespace HDFSBuilderWrapper builder = createHDFSBuilder(uri_without_path + "/", context->getGlobalContext()->getConfigRef()); HDFSFSPtr fs = createHDFSFS(builder.get()); - auto res = LSWithRegexpMatching("/", fs, path_from_uri, context->getSettingsRef().ignore_eacces_multidirectory_globs); + auto res = LSWithRegexpMatching("/", fs, path_from_uri, context->getSettingsRef().ignore_access_denied_multidirectory_globs); return res; } } diff --git a/src/Storages/StorageFile.cpp b/src/Storages/StorageFile.cpp index 84f1479881a..ee27757f1c2 100644 --- a/src/Storages/StorageFile.cpp +++ b/src/Storages/StorageFile.cpp @@ -104,7 +104,7 @@ void listFilesWithRegexpMatchingImpl( const std::string & path_for_ls, const std::string & for_match, size_t & total_bytes_to_read, - bool ignore_eacces_multidirectory_globs, + bool ignore_access_denied_multidirectory_globs, std::vector & result, bool recursive = false); @@ -123,14 +123,14 @@ void listFilesWithFoldedRegexpMatchingImpl(const std::string & path_for_ls, size_t & total_bytes_to_read, const size_t max_depth, const size_t next_slash_after_glob_pos, - bool ignore_eacces_multidirectory_globs, + bool ignore_access_denied_multidirectory_globs, std::vector & result) { if (!max_depth) return; const fs::directory_iterator end; - fs::directory_iterator it = ignore_eacces_multidirectory_globs + fs::directory_iterator it = ignore_access_denied_multidirectory_globs ? fs::directory_iterator(path_for_ls, fs::directory_options::skip_permission_denied) : fs::directory_iterator(path_for_ls); for (; it != end; ++it) @@ -150,7 +150,7 @@ void listFilesWithFoldedRegexpMatchingImpl(const std::string & path_for_ls, { listFilesWithRegexpMatchingImpl(fs::path(full_path) / "" , suffix_with_globs.substr(next_slash_after_glob_pos), - total_bytes_to_read, ignore_eacces_multidirectory_globs, result); + total_bytes_to_read, ignore_access_denied_multidirectory_globs, result); } } else if (it->is_directory()) @@ -158,7 +158,7 @@ void listFilesWithFoldedRegexpMatchingImpl(const std::string & path_for_ls, listFilesWithFoldedRegexpMatchingImpl(fs::path(full_path), processed_suffix + dir_or_file_name, suffix_with_globs, matcher, total_bytes_to_read, max_depth - 1, next_slash_after_glob_pos, - ignore_eacces_multidirectory_globs, result); + ignore_access_denied_multidirectory_globs, result); } } @@ -171,7 +171,7 @@ void listFilesWithRegexpMatchingImpl( const std::string & path_for_ls, const std::string & for_match, size_t & total_bytes_to_read, - bool ignore_eacces_multidirectory_globs, + bool ignore_access_denied_multidirectory_globs, std::vector & result, bool recursive) { @@ -231,7 +231,7 @@ void listFilesWithRegexpMatchingImpl( { listFilesWithFoldedRegexpMatchingImpl(fs::path(prefix_without_globs), "", suffix_with_globs, matcher, total_bytes_to_read, slashes_in_glob, next_slash_after_glob_pos, - ignore_eacces_multidirectory_globs, result); + ignore_access_denied_multidirectory_globs, result); return; } @@ -257,12 +257,12 @@ void listFilesWithRegexpMatchingImpl( { listFilesWithRegexpMatchingImpl(fs::path(full_path).append(it->path().string()) / "", looking_for_directory ? suffix_with_globs.substr(next_slash_after_glob_pos) : current_glob, - total_bytes_to_read, ignore_eacces_multidirectory_globs, result, recursive); + total_bytes_to_read, ignore_access_denied_multidirectory_globs, result, recursive); } else if (looking_for_directory && re2::RE2::FullMatch(file_name, matcher)) /// Recursion depth is limited by pattern. '*' works only for depth = 1, for depth = 2 pattern path is '*/*'. So we do not need additional check. listFilesWithRegexpMatchingImpl(fs::path(full_path) / "", suffix_with_globs.substr(next_slash_after_glob_pos), - total_bytes_to_read, ignore_eacces_multidirectory_globs, result); + total_bytes_to_read, ignore_access_denied_multidirectory_globs, result); } } } @@ -271,10 +271,10 @@ std::vector listFilesWithRegexpMatching( const std::string & path_for_ls, const std::string & for_match, size_t & total_bytes_to_read, - bool ignore_eacces_multidirectory_globs) + bool ignore_access_denied_multidirectory_globs) { std::vector result; - listFilesWithRegexpMatchingImpl(path_for_ls, for_match, total_bytes_to_read, ignore_eacces_multidirectory_globs, result); + listFilesWithRegexpMatchingImpl(path_for_ls, for_match, total_bytes_to_read, ignore_access_denied_multidirectory_globs, result); return result; } @@ -439,7 +439,7 @@ Strings StorageFile::getPathsList(const String & table_path, const String & user else { /// We list only non-directory files. - paths = listFilesWithRegexpMatching("/", path, total_bytes_to_read, context->getSettingsRef().ignore_eacces_multidirectory_globs); + paths = listFilesWithRegexpMatching("/", path, total_bytes_to_read, context->getSettingsRef().ignore_access_denied_multidirectory_globs); can_be_directory = false; } diff --git a/tests/queries/0_stateless/02771_multidirectory_globs_storage_file.sh b/tests/queries/0_stateless/02771_multidirectory_globs_storage_file.sh index 984ae47d222..b0918d4641d 100755 --- a/tests/queries/0_stateless/02771_multidirectory_globs_storage_file.sh +++ b/tests/queries/0_stateless/02771_multidirectory_globs_storage_file.sh @@ -33,6 +33,6 @@ ${CLICKHOUSE_CLIENT} --query "SELECT *, _file FROM file('${user_files_path}/${CL # Add a directory to test against permission_denied rm -rf ${user_files_path:?}/${CLICKHOUSE_TEST_UNIQUE_NAME:?}/dir_inaccessible/ && mkdir -p ${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/dir_inaccessible/ && chmod 000 ${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/dir_inaccessible/ -${CLICKHOUSE_CLIENT} --query "SELECT *, _file FROM file('${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/dir{?/subdir?1/da,2/subdir2?/da}ta1.csv', CSV) SETTINGS ignore_eacces_multidirectory_globs = 1;" +${CLICKHOUSE_CLIENT} --query "SELECT *, _file FROM file('${user_files_path}/${CLICKHOUSE_TEST_UNIQUE_NAME}/dir{?/subdir?1/da,2/subdir2?/da}ta1.csv', CSV) SETTINGS ignore_access_denied_multidirectory_globs = 1;" rm -rf ${user_files_path:?}/${CLICKHOUSE_TEST_UNIQUE_NAME:?} From 05446d837702cfedb5b794060a1475424cce386b Mon Sep 17 00:00:00 2001 From: Alexander Sapin Date: Thu, 7 Sep 2023 11:46:28 +0200 Subject: [PATCH 183/433] Rename test --- ...eference => 02861_replacing_merge_tree_with_cleanup.reference} | 0 ...ber_column.sql => 02861_replacing_merge_tree_with_cleanup.sql} | 0 2 files changed, 0 insertions(+), 0 deletions(-) rename tests/queries/0_stateless/{02861_replacing_merge_tree_with_block_number_column.reference => 02861_replacing_merge_tree_with_cleanup.reference} (100%) rename tests/queries/0_stateless/{02861_replacing_merge_tree_with_block_number_column.sql => 02861_replacing_merge_tree_with_cleanup.sql} (100%) diff --git a/tests/queries/0_stateless/02861_replacing_merge_tree_with_block_number_column.reference b/tests/queries/0_stateless/02861_replacing_merge_tree_with_cleanup.reference similarity index 100% rename from tests/queries/0_stateless/02861_replacing_merge_tree_with_block_number_column.reference rename to tests/queries/0_stateless/02861_replacing_merge_tree_with_cleanup.reference diff --git a/tests/queries/0_stateless/02861_replacing_merge_tree_with_block_number_column.sql b/tests/queries/0_stateless/02861_replacing_merge_tree_with_cleanup.sql similarity index 100% rename from tests/queries/0_stateless/02861_replacing_merge_tree_with_block_number_column.sql rename to tests/queries/0_stateless/02861_replacing_merge_tree_with_cleanup.sql From 25c06c9764317e8d5eeb3555a03bcc7229088d47 Mon Sep 17 00:00:00 2001 From: Fern Date: Thu, 7 Sep 2023 10:47:42 +0100 Subject: [PATCH 184/433] Update sampling-query-profiler.md Adding a note to clarify that the sample query provided, in CH Cloud, might return empty information as it cycles through nodes of the cluster. You need to add clusterAllReplicas. --- .../optimizing-performance/sampling-query-profiler.md | 2 ++ 1 file changed, 2 insertions(+) diff --git a/docs/en/operations/optimizing-performance/sampling-query-profiler.md b/docs/en/operations/optimizing-performance/sampling-query-profiler.md index f5d0e5d6aed..9988bfc44bc 100644 --- a/docs/en/operations/optimizing-performance/sampling-query-profiler.md +++ b/docs/en/operations/optimizing-performance/sampling-query-profiler.md @@ -11,6 +11,8 @@ ClickHouse runs sampling profiler that allows analyzing query execution. Using p Query profiler is automatically enabled in ClickHouse Cloud and you can run a sample query as follows +:::note If you are running the following query in ClickHouse Cloud, make sure to change `FROM system.trace_log` to `FROM clusterAllReplicas(default, system.trace_log)` to select from all nodes of the cluster ::: + ``` sql SELECT count(), From cd3c121a620718581d2169e722aafc4524cc03e3 Mon Sep 17 00:00:00 2001 From: Robert Schulze Date: Thu, 7 Sep 2023 09:50:49 +0000 Subject: [PATCH 185/433] Various fixups --- contrib/aklomp-base64-cmake/CMakeLists.txt | 20 +++---- docs/en/operations/system-tables/licenses.md | 13 +---- docs/ru/operations/system-tables/licenses.md | 13 +---- docs/zh/operations/system-tables/licenses.md | 14 +---- src/Functions/FunctionBase64Conversion.h | 57 +++++++------------ .../00732_base64_functions.reference | 13 ++++- .../0_stateless/00732_base64_functions.sql | 20 ++++--- .../0_stateless/01092_base64.reference | 1 - tests/queries/0_stateless/01092_base64.sql | 4 -- 9 files changed, 54 insertions(+), 101 deletions(-) delete mode 100644 tests/queries/0_stateless/01092_base64.reference delete mode 100644 tests/queries/0_stateless/01092_base64.sql diff --git a/contrib/aklomp-base64-cmake/CMakeLists.txt b/contrib/aklomp-base64-cmake/CMakeLists.txt index 17fa3dcec02..4607d1cc9b9 100644 --- a/contrib/aklomp-base64-cmake/CMakeLists.txt +++ b/contrib/aklomp-base64-cmake/CMakeLists.txt @@ -1,8 +1,4 @@ -if(ARCH_AMD64 OR ARCH_AARCH64 OR ARCH_PPC64LE OR ARCH_S390X) - option (ENABLE_BASE64 "Enable base64" ${ENABLE_LIBRARIES}) -elseif(ENABLE_BASE64) - message (${RECONFIGURE_MESSAGE_LEVEL} "base64 library is only supported on x86_64 and aarch64") -endif() +option (ENABLE_BASE64 "Enable base64" ${ENABLE_LIBRARIES}) if (NOT ENABLE_BASE64) message(STATUS "Not using base64") @@ -13,7 +9,7 @@ SET(LIBRARY_DIR "${ClickHouse_SOURCE_DIR}/contrib/aklomp-base64") set(HAVE_NEON32 0) set(HAVE_NEON64 0) -if (ARCH_AARCH64) +if (ARCH_AARCH64 AND NOT NO_ARMV81_OR_HIGHER) set(HAVE_NEON64 1) endif () @@ -33,7 +29,7 @@ cast_to_bool(AVX "-mavx") cast_to_bool(AVX2 "-mavx2") set(HAVE_FAST_UNALIGNED_ACCESS 0) -if(HAVE_SSSE3 OR HAVE_SSE41 OR HAVE_SSE42 OR HAVE_AVX OR HAVE_AVX2 OR ARCH_AARCH64) +if(HAVE_SSSE3 OR HAVE_SSE41 OR HAVE_SSE42 OR HAVE_AVX OR HAVE_AVX2 OR HAVE_NEON64) set(HAVE_FAST_UNALIGNED_ACCESS 1) endif () @@ -48,23 +44,21 @@ add_library(_base64 # Codec implementations "${LIBRARY_DIR}/lib/arch/generic/codec.c" - + "${LIBRARY_DIR}/lib/arch/neon32/codec.c" + "${LIBRARY_DIR}/lib/arch/neon64/codec.c" "${LIBRARY_DIR}/lib/arch/ssse3/codec.c" "${LIBRARY_DIR}/lib/arch/sse41/codec.c" "${LIBRARY_DIR}/lib/arch/sse42/codec.c" + # ClickHouse does not support AVX* by default. We still need to compile these files + # because the run-time codec detection requires functions from these file to be present. "${LIBRARY_DIR}/lib/arch/avx/codec.c" "${LIBRARY_DIR}/lib/arch/avx2/codec.c" - # Even though AVX512 is not supported, we still need to compile this file - # because the run-time codec choose requires the functions from this file - # to be present. "${LIBRARY_DIR}/lib/arch/avx512/codec.c" # Tables "${LIBRARY_DIR}/lib/tables/table_dec_32bit.h" "${LIBRARY_DIR}/lib/tables/table_enc_12bit.h" - "${LIBRARY_DIR}/lib/arch/neon32/codec.c" - "${LIBRARY_DIR}/lib/arch/neon64/codec.c" "${LIBRARY_DIR}/lib/codecs.h" "${CMAKE_CURRENT_BINARY_DIR}/config.h") diff --git a/docs/en/operations/system-tables/licenses.md b/docs/en/operations/system-tables/licenses.md index 7a444105489..c436c0ade3b 100644 --- a/docs/en/operations/system-tables/licenses.md +++ b/docs/en/operations/system-tables/licenses.md @@ -20,21 +20,10 @@ SELECT library_name, license_type, license_path FROM system.licenses LIMIT 15 ``` text ┌─library_name───────┬─license_type─┬─license_path────────────────────────┐ -│ FastMemcpy │ MIT │ /contrib/FastMemcpy/LICENSE │ -│ arrow │ Apache │ /contrib/arrow/LICENSE.txt │ -│ avro │ Apache │ /contrib/avro/LICENSE.txt │ │ aws-c-common │ Apache │ /contrib/aws-c-common/LICENSE │ -│ aws-c-event-stream │ Apache │ /contrib/aws-c-event-stream/LICENSE │ -│ aws-checksums │ Apache │ /contrib/aws-checksums/LICENSE │ -│ aws │ Apache │ /contrib/aws/LICENSE.txt │ │ base64 │ BSD 2-clause │ /contrib/aklomp-base64/LICENSE │ -│ boost │ Boost │ /contrib/boost/LICENSE_1_0.txt │ │ brotli │ MIT │ /contrib/brotli/LICENSE │ -│ capnproto │ MIT │ /contrib/capnproto/LICENSE │ -│ cassandra │ Apache │ /contrib/cassandra/LICENSE.txt │ -│ cctz │ Apache │ /contrib/cctz/LICENSE.txt │ -│ cityhash102 │ MIT │ /contrib/cityhash102/COPYING │ -│ cppkafka │ BSD 2-clause │ /contrib/cppkafka/LICENSE │ +│ [...] │ [...] │ [...] │ └────────────────────┴──────────────┴─────────────────────────────────────┘ ``` diff --git a/docs/ru/operations/system-tables/licenses.md b/docs/ru/operations/system-tables/licenses.md index 41cbe25d0f7..e35b5d5e3be 100644 --- a/docs/ru/operations/system-tables/licenses.md +++ b/docs/ru/operations/system-tables/licenses.md @@ -20,21 +20,10 @@ SELECT library_name, license_type, license_path FROM system.licenses LIMIT 15 ``` text ┌─library_name───────┬─license_type─┬─license_path────────────────────────┐ -│ FastMemcpy │ MIT │ /contrib/FastMemcpy/LICENSE │ -│ arrow │ Apache │ /contrib/arrow/LICENSE.txt │ -│ avro │ Apache │ /contrib/avro/LICENSE.txt │ │ aws-c-common │ Apache │ /contrib/aws-c-common/LICENSE │ -│ aws-c-event-stream │ Apache │ /contrib/aws-c-event-stream/LICENSE │ -│ aws-checksums │ Apache │ /contrib/aws-checksums/LICENSE │ -│ aws │ Apache │ /contrib/aws/LICENSE.txt │ │ base64 │ BSD 2-clause │ /contrib/aklomp-base64/LICENSE │ -│ boost │ Boost │ /contrib/boost/LICENSE_1_0.txt │ │ brotli │ MIT │ /contrib/brotli/LICENSE │ -│ capnproto │ MIT │ /contrib/capnproto/LICENSE │ -│ cassandra │ Apache │ /contrib/cassandra/LICENSE.txt │ -│ cctz │ Apache │ /contrib/cctz/LICENSE.txt │ -│ cityhash102 │ MIT │ /contrib/cityhash102/COPYING │ -│ cppkafka │ BSD 2-clause │ /contrib/cppkafka/LICENSE │ +│ [...] │ [...] │ [...] │ └────────────────────┴──────────────┴─────────────────────────────────────┘ ``` diff --git a/docs/zh/operations/system-tables/licenses.md b/docs/zh/operations/system-tables/licenses.md index 9711617bdfd..38260491dc0 100644 --- a/docs/zh/operations/system-tables/licenses.md +++ b/docs/zh/operations/system-tables/licenses.md @@ -20,21 +20,9 @@ SELECT library_name, license_type, license_path FROM system.licenses LIMIT 15 ``` text ┌─library_name───────┬─license_type─┬─license_path────────────────────────┐ -│ FastMemcpy │ MIT │ /contrib/FastMemcpy/LICENSE │ -│ arrow │ Apache │ /contrib/arrow/LICENSE.txt │ -│ avro │ Apache │ /contrib/avro/LICENSE.txt │ │ aws-c-common │ Apache │ /contrib/aws-c-common/LICENSE │ -│ aws-c-event-stream │ Apache │ /contrib/aws-c-event-stream/LICENSE │ -│ aws-checksums │ Apache │ /contrib/aws-checksums/LICENSE │ -│ aws │ Apache │ /contrib/aws/LICENSE.txt │ │ base64 │ BSD 2-clause │ /contrib/aklomp-base64/LICENSE │ -│ boost │ Boost │ /contrib/boost/LICENSE_1_0.txt │ │ brotli │ MIT │ /contrib/brotli/LICENSE │ -│ capnproto │ MIT │ /contrib/capnproto/LICENSE │ -│ cassandra │ Apache │ /contrib/cassandra/LICENSE.txt │ -│ cctz │ Apache │ /contrib/cctz/LICENSE.txt │ -│ cityhash102 │ MIT │ /contrib/cityhash102/COPYING │ -│ cppkafka │ BSD 2-clause │ /contrib/cppkafka/LICENSE │ +│ [...] │ [...] │ [...] │ └────────────────────┴──────────────┴─────────────────────────────────────┘ - ``` diff --git a/src/Functions/FunctionBase64Conversion.h b/src/Functions/FunctionBase64Conversion.h index 7bd4bfd777b..15dcbaf8352 100644 --- a/src/Functions/FunctionBase64Conversion.h +++ b/src/Functions/FunctionBase64Conversion.h @@ -1,5 +1,4 @@ #pragma once -#include #include "config.h" #if USE_BASE64 @@ -12,6 +11,7 @@ # include # include +# include # include namespace DB @@ -24,26 +24,16 @@ namespace ErrorCodes extern const int INCORRECT_DATA; } -namespace Detail -{ - inline size_t base64Decode(const std::span src, UInt8 * dst) - { - size_t outlen = 0; - base64_decode(reinterpret_cast(src.data()), src.size(), reinterpret_cast(dst), &outlen, 0); - return outlen; - } -} - struct Base64Encode { static constexpr auto name = "base64Encode"; - static size_t getBufferSize(const size_t string_length, const size_t string_count) + static size_t getBufferSize(size_t string_length, size_t string_count) { return ((string_length - string_count) / 3 + string_count) * 4 + string_count; } - static size_t performCoding(const std::span src, UInt8 * dst) + static size_t perform(const std::span src, UInt8 * dst) { size_t outlen = 0; base64_encode(reinterpret_cast(src.data()), src.size(), reinterpret_cast(dst), &outlen, 0); @@ -55,15 +45,17 @@ struct Base64Decode { static constexpr auto name = "base64Decode"; - static size_t getBufferSize(const size_t string_length, const size_t string_count) + static size_t getBufferSize(size_t string_length, size_t string_count) { return ((string_length - string_count) / 4 + string_count) * 3 + string_count; } - static size_t performCoding(const std::span src, UInt8 * dst) + static size_t perform(const std::span src, UInt8 * dst) { - const auto outlen = Detail::base64Decode(src, dst); - if (src.size() > 0 && !outlen) + size_t outlen = 0; + int rc = base64_decode(reinterpret_cast(src.data()), src.size(), reinterpret_cast(dst), &outlen, 0); + + if (rc != 1) throw Exception( ErrorCodes::INCORRECT_DATA, "Failed to {} input '{}'", @@ -78,17 +70,16 @@ struct TryBase64Decode { static constexpr auto name = "tryBase64Decode"; - static size_t getBufferSize(const size_t string_length, const size_t string_count) + static size_t getBufferSize(size_t string_length, size_t string_count) { return Base64Decode::getBufferSize(string_length, string_count); } - static size_t performCoding(const std::span src, UInt8 * dst) + static size_t perform(const std::span src, UInt8 * dst) { - if (src.empty()) - return 0; + size_t outlen = 0; + base64_decode(reinterpret_cast(src.data()), src.size(), reinterpret_cast(dst), &outlen, 0); - const auto outlen = Detail::base64Decode(src, dst); // during decoding character array can be partially polluted // if fail, revert back and clean if (!outlen) @@ -112,20 +103,16 @@ public: DataTypePtr getReturnTypeImpl(const ColumnsWithTypeAndName & arguments) const override { - if (arguments.size() != 1) - throw Exception(ErrorCodes::BAD_ARGUMENTS, "Wrong number of arguments for function {}: 1 expected.", getName()); + FunctionArgumentDescriptors mandatory_arguments{ + {"value", &isStringOrFixedString, nullptr, "String or FixedString"} + }; - if (!WhichDataType(arguments[0].type).isStringOrFixedString()) - throw Exception( - ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, - "Illegal type {} of 1st argument of function {}. Must be FixedString or String.", - arguments[0].type->getName(), - getName()); + validateFunctionArgumentTypes(*this, arguments, mandatory_arguments); return std::make_shared(); } - ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr &, const size_t input_rows_count) const override + ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override { const auto & input_column = arguments[0].column; if (const auto * src_column_as_fixed_string = checkAndGetColumn(*input_column)) @@ -141,7 +128,7 @@ public: } private: - static ColumnPtr execute(const ColumnString & src_column, const size_t src_row_count) + static ColumnPtr execute(const ColumnString & src_column, size_t src_row_count) { auto dst_column = ColumnString::create(); auto & dst_chars = dst_column->getChars(); @@ -162,7 +149,7 @@ private: for (size_t row = 0; row < src_row_count; ++row) { const size_t src_length = src_offsets[row] - src_offset_prev - 1; - const auto outlen = Func::performCoding({src, src_length}, dst_pos); + const auto outlen = Func::perform({src, src_length}, dst_pos); /// Base64 library is using AVX-512 with some shuffle operations. /// Memory sanitizer don't understand if there was uninitialized memory in SIMD register but it was not used in the result of shuffle. @@ -181,7 +168,7 @@ private: return dst_column; } - static ColumnPtr execute(const ColumnFixedString & src_column, const size_t src_row_count) + static ColumnPtr execute(const ColumnFixedString & src_column, size_t src_row_count) { auto dst_column = ColumnString::create(); auto & dst_chars = dst_column->getChars(); @@ -200,7 +187,7 @@ private: for (size_t row = 0; row < src_row_count; ++row) { - const auto outlen = Func::performCoding({src, src_n}, dst_pos); + const auto outlen = Func::perform({src, src_n}, dst_pos); /// Base64 library is using AVX-512 with some shuffle operations. /// Memory sanitizer don't understand if there was uninitialized memory in SIMD register but it was not used in the result of shuffle. diff --git a/tests/queries/0_stateless/00732_base64_functions.reference b/tests/queries/0_stateless/00732_base64_functions.reference index 007648f0cd4..19486ff95de 100644 --- a/tests/queries/0_stateless/00732_base64_functions.reference +++ b/tests/queries/0_stateless/00732_base64_functions.reference @@ -12,10 +12,17 @@ foo foob fooba foobar -1 1 -This used to be an error + +f +fo +foo +foob fooba -This also used to be an error +foobar +1 1 +1 1 fooba Zm9v foo +foo +TEcgT3B0aW11cw== diff --git a/tests/queries/0_stateless/00732_base64_functions.sql b/tests/queries/0_stateless/00732_base64_functions.sql index 2c771c842ac..4272cca19ee 100644 --- a/tests/queries/0_stateless/00732_base64_functions.sql +++ b/tests/queries/0_stateless/00732_base64_functions.sql @@ -5,17 +5,21 @@ SET send_logs_level = 'fatal'; SELECT base64Encode(val) FROM (select arrayJoin(['', 'f', 'fo', 'foo', 'foob', 'fooba', 'foobar']) val); SELECT base64Decode(val) FROM (select arrayJoin(['', 'Zg==', 'Zm8=', 'Zm9v', 'Zm9vYg==', 'Zm9vYmE=', 'Zm9vYmFy']) val); +SELECT tryBase64Decode(val) FROM (select arrayJoin(['', 'Zg==', 'Zm8=', 'Zm9v', 'Zm9vYg==', 'Zm9vYmE=', 'Zm9vYmFy']) val); + SELECT base64Decode(base64Encode('foo')) = 'foo', base64Encode(base64Decode('Zm9v')) == 'Zm9v'; +SELECT tryBase64Decode(base64Encode('foo')) = 'foo', base64Encode(tryBase64Decode('Zm9v')) == 'Zm9v'; -SELECT 'This used to be an error'; +SELECT base64Encode('foo', 'excess argument'); -- { serverError NUMBER_OF_ARGUMENTS_DOESNT_MATCH } +SELECT base64Decode('foo', 'excess argument'); -- { serverError NUMBER_OF_ARGUMENTS_DOESNT_MATCH } +SELECT tryBase64Decode('foo', 'excess argument'); -- { serverError NUMBER_OF_ARGUMENTS_DOESNT_MATCH } + +SELECT base64Decode('Zm9vYmF=Zm9v'); -- { serverError INCORRECT_DATA } SELECT tryBase64Decode('Zm9vYmF=Zm9v'); -SELECT base64Encode(val, 'excess argument') FROM (select arrayJoin(['', 'f', 'fo', 'foo', 'foob', 'fooba', 'foobar']) val); -- { serverError 42 } -SELECT base64Decode(val, 'excess argument') FROM (select arrayJoin(['', 'Zg==', 'Zm8=', 'Zm9v', 'Zm9vYg==', 'Zm9vYmE=', 'Zm9vYmFy']) val); -- { serverError 42 } -SELECT tryBase64Decode('Zm9vYmF=Zm9v', 'excess argument'); -- { serverError 42 } - -SELECT 'This also used to be an error'; -SELECT base64Decode('Zm9vYmF=Zm9v'); - select base64Encode(toFixedString('foo', 3)); select base64Decode(toFixedString('Zm9v', 4)); +select tryBase64Decode(toFixedString('Zm9v', 4)); + +-- This query reproduces a bug in TurboBase64 library (which we no longer use) +select distinct base64Encode(materialize('LG Optimus')) from numbers(100); diff --git a/tests/queries/0_stateless/01092_base64.reference b/tests/queries/0_stateless/01092_base64.reference deleted file mode 100644 index 4c41ac31946..00000000000 --- a/tests/queries/0_stateless/01092_base64.reference +++ /dev/null @@ -1 +0,0 @@ -TEcgT3B0aW11cw== diff --git a/tests/queries/0_stateless/01092_base64.sql b/tests/queries/0_stateless/01092_base64.sql deleted file mode 100644 index f50cf49d270..00000000000 --- a/tests/queries/0_stateless/01092_base64.sql +++ /dev/null @@ -1,4 +0,0 @@ --- Tags: no-fasttest - --- This query reproduces a bug in TurboBase64 library. -select distinct base64Encode(materialize('LG Optimus')) from numbers(100); From fc105aa442f6ece56dfaeb5d6541c793bde412f1 Mon Sep 17 00:00:00 2001 From: Robert Schulze Date: Thu, 7 Sep 2023 10:59:19 +0000 Subject: [PATCH 186/433] Fix another test --- tests/queries/0_stateless/01339_client_unrecognized_option.sh | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tests/queries/0_stateless/01339_client_unrecognized_option.sh b/tests/queries/0_stateless/01339_client_unrecognized_option.sh index b3488d010c2..0cfb705185e 100755 --- a/tests/queries/0_stateless/01339_client_unrecognized_option.sh +++ b/tests/queries/0_stateless/01339_client_unrecognized_option.sh @@ -11,7 +11,7 @@ $CLICKHOUSE_CLIENT -xyzgarbage 2>&1 | grep -q "UNRECOGNIZED_ARGUMENTS" && echo ' $CLICKHOUSE_CLIENT --xyzgarbage 2>&1 | grep -q "UNRECOGNIZED_ARGUMENTS" && echo 'OK' || echo 'FAIL' -cat /etc/passwd | sed 's/:/\t/g' | $CLICKHOUSE_CLIENT --query="SELECT shell, count() AS c FROM passwd GROUP BY shell ORDER BY c DESC" --external --file=- --name=passwd --structure='login String, unused String, uid UInt16, gid UInt16, comment String, home String, shell String' xyzgarbage 2>&1 | grep -q "BAD_ARGUMENTS" && echo 'OK' || echo 'FAIL' +cat /etc/passwd | sed 's/:/\t/g' | $CLICKHOUSE_CLIENT --query="SELECT shell, count() AS c FROM passwd GROUP BY shell ORDER BY c DESC" --external --file=- --name=passwd --structure='login String, unused String, uid UInt16, gid UInt16, comment String, home String, shell String' xyzgarbage 2>&1 | grep -q "SYNTAX_ERROR" && echo 'OK' || echo 'FAIL' cat /etc/passwd | sed 's/:/\t/g' | $CLICKHOUSE_CLIENT --query="SELECT shell, count() AS c FROM passwd GROUP BY shell ORDER BY c DESC" --external -xyzgarbage --file=- --name=passwd --structure='login String, unused String, uid UInt16, gid UInt16, comment String, home String, shell String' 2>&1 | grep -q "Bad arguments" && echo 'OK' || echo 'FAIL' From c6ef811b82e4c1ccc30c224ab285255399a7bfc7 Mon Sep 17 00:00:00 2001 From: avogar Date: Thu, 7 Sep 2023 11:05:06 +0000 Subject: [PATCH 187/433] Fix tests --- src/Storages/StorageS3.cpp | 8 ++------ src/TableFunctions/TableFunctionS3.cpp | 5 +++++ 2 files changed, 7 insertions(+), 6 deletions(-) diff --git a/src/Storages/StorageS3.cpp b/src/Storages/StorageS3.cpp index 5078d21141b..24b2279bfdc 100644 --- a/src/Storages/StorageS3.cpp +++ b/src/Storages/StorageS3.cpp @@ -1246,13 +1246,11 @@ void StorageS3::processNamedCollectionResult(StorageS3::Configuration & configur validateNamedCollection(collection, required_configuration_keys, optional_configuration_keys); auto filename = collection.getOrDefault("filename", ""); - String url; if (!filename.empty()) - url = std::filesystem::path(collection.get("url")) / filename; + configuration.url = S3::URI(std::filesystem::path(collection.get("url")) / filename); else - url = collection.get("url"); + configuration.url = S3::URI(collection.get("url")); - configuration.url = S3::URI(url); configuration.auth_settings.access_key_id = collection.getOrDefault("access_key_id", ""); configuration.auth_settings.secret_access_key = collection.getOrDefault("secret_access_key", ""); configuration.auth_settings.use_environment_credentials = collection.getOrDefault("use_environment_credentials", 1); @@ -1260,8 +1258,6 @@ void StorageS3::processNamedCollectionResult(StorageS3::Configuration & configur configuration.auth_settings.expiration_window_seconds = collection.getOrDefault("expiration_window_seconds", S3::DEFAULT_EXPIRATION_WINDOW_SECONDS); configuration.format = collection.getOrDefault("format", configuration.format); - if (configuration.format == "auto") - configuration.format = FormatFactory::instance().getFormatFromFileName(Poco::URI(url).getPath(), true); configuration.compression_method = collection.getOrDefault("compression_method", collection.getOrDefault("compression", "auto")); configuration.structure = collection.getOrDefault("structure", "auto"); diff --git a/src/TableFunctions/TableFunctionS3.cpp b/src/TableFunctions/TableFunctionS3.cpp index 94ce29f1116..2e9527cff6c 100644 --- a/src/TableFunctions/TableFunctionS3.cpp +++ b/src/TableFunctions/TableFunctionS3.cpp @@ -58,6 +58,11 @@ void TableFunctionS3::parseArgumentsImpl(ASTs & args, const ContextPtr & context if (auto named_collection = tryGetNamedCollectionWithOverrides(args, context)) { StorageS3::processNamedCollectionResult(configuration, *named_collection); + if (configuration.format == "auto") + { + String file_path = named_collection->getOrDefault("filename", Poco::URI(named_collection->get("url")).getPath()); + configuration.format = FormatFactory::instance().getFormatFromFileName(file_path, true); + } } else { From b06412ae60fb50f8bfbf6635e6b575d49f11d96e Mon Sep 17 00:00:00 2001 From: Nikolay Degterinsky Date: Thu, 7 Sep 2023 11:15:49 +0000 Subject: [PATCH 188/433] Add addDate function for compatibility with MySQL --- .../functions/date-time-functions.md | 39 ++++++++ src/Functions/addDate.cpp | 88 +++++++++++++++++++ ...new_functions_must_be_documented.reference | 1 + .../02834_add_date_function.reference | 4 + .../0_stateless/02834_add_date_function.sql | 9 ++ 5 files changed, 141 insertions(+) create mode 100644 src/Functions/addDate.cpp create mode 100644 tests/queries/0_stateless/02834_add_date_function.reference create mode 100644 tests/queries/0_stateless/02834_add_date_function.sql diff --git a/docs/en/sql-reference/functions/date-time-functions.md b/docs/en/sql-reference/functions/date-time-functions.md index 87d84425029..3940aef8fbf 100644 --- a/docs/en/sql-reference/functions/date-time-functions.md +++ b/docs/en/sql-reference/functions/date-time-functions.md @@ -987,6 +987,10 @@ Result: └───────────────────────────────────────────────┘ ``` +**See Also** + +- [addDate](#addDate) + ## date\_sub Subtracts the time interval or date interval from the provided date or date with time. @@ -1134,6 +1138,41 @@ Result: └──────────────────────────────────────────────────────────────┘ ``` +## addDate + +Adds the time interval or date interval to the provided date or date with time. + +**Syntax** + +``` sql +addDate(date, interval) +``` + +**Arguments** + +- `date` — The date or date with time to which `value` is added. [Date](../../sql-reference/data-types/date.md) or [DateTime](../../sql-reference/data-types/datetime.md). +- `interval` — Interval to add. [Interval](../../sql-reference/data-types/special-data-types/interval.md). + +**Returned value** + +Date or date with time obtained by adding `value`, expressed in `unit`, to `date`. + +Type: [Date](../../sql-reference/data-types/date.md) or [DateTime](../../sql-reference/data-types/datetime.md). + +**Example** + +```sql +SELECT addDate(toDate('2018-01-01'), INTERVAL 3 YEAR); +``` + +Result: + +```text +┌─addDate(toDate('2018-01-01'), toIntervalYear(3))─┐ +│ 2021-01-01 │ +└──────────────────────────────────────────────────┘ +``` + ## now Returns the current date and time at the moment of query analysis. The function is a constant expression. diff --git a/src/Functions/addDate.cpp b/src/Functions/addDate.cpp new file mode 100644 index 00000000000..a34ded341ac --- /dev/null +++ b/src/Functions/addDate.cpp @@ -0,0 +1,88 @@ +#include +#include + +namespace DB +{ +namespace ErrorCodes +{ + extern const int ILLEGAL_TYPE_OF_ARGUMENT; +} + +namespace +{ + +class FunctionAddDate : public IFunction +{ +public: + static constexpr auto name = "addDate"; + + explicit FunctionAddDate(ContextPtr context_) : context(context_) {} + + static FunctionPtr create(ContextPtr context) { return std::make_shared(context); } + + String getName() const override { return name; } + + bool isSuitableForShortCircuitArgumentsExecution(const DataTypesWithConstInfo & /*arguments*/) const override { return false; } + size_t getNumberOfArguments() const override { return 2; } + + DataTypePtr getReturnTypeImpl(const ColumnsWithTypeAndName & arguments) const override + { + if (!isDateOrDate32(arguments[0].type) && !isDateTime(arguments[0].type) && !isDateTime64(arguments[0].type)) + throw Exception( + ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, + "Illegal type {} of 1st argument of function {}. Should be a date or a date with time", + arguments[0].type->getName(), + getName()); + + if (!isInterval(arguments[1].type)) + throw Exception( + ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, + "Illegal type {} of 1st argument of function {}. Should be an interval", + arguments[0].type->getName(), + getName()); + + auto plus = FunctionFactory::instance().get("plus", context); + auto plus_build = plus->build(arguments); + + return plus_build->getResultType(); + } + + bool useDefaultImplementationForConstants() const override { return true; } + ColumnNumbers getArgumentsThatAreAlwaysConstant() const override { return {0, 2}; } + + ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override + { + if (!isDateOrDate32(arguments[0].type) && !isDateTime(arguments[0].type) && !isDateTime64(arguments[0].type)) + throw Exception( + ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, + "Illegal type {} of 1st argument of function {}. Should be a date or a date with time", + arguments[0].type->getName(), + getName()); + + if (!isInterval(arguments[1].type)) + throw Exception( + ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, + "Illegal type {} of 1st argument of function {}. Should be an interval", + arguments[0].type->getName(), + getName()); + + auto plus = FunctionFactory::instance().get("plus", context); + auto plus_build = plus->build(arguments); + + auto res_type = plus_build->getResultType(); + return plus_build->execute(arguments, res_type, input_rows_count); + } + +private: + ContextPtr context; +}; + +} + + +REGISTER_FUNCTION(AddInterval) +{ + factory.registerFunction(); +} + +} diff --git a/tests/queries/0_stateless/02415_all_new_functions_must_be_documented.reference b/tests/queries/0_stateless/02415_all_new_functions_must_be_documented.reference index 61a2e4e9f02..1a7ffdb1fa1 100644 --- a/tests/queries/0_stateless/02415_all_new_functions_must_be_documented.reference +++ b/tests/queries/0_stateless/02415_all_new_functions_must_be_documented.reference @@ -68,6 +68,7 @@ accurateCastOrDefault accurateCastOrNull acos acosh +addDate addDays addHours addMicroseconds diff --git a/tests/queries/0_stateless/02834_add_date_function.reference b/tests/queries/0_stateless/02834_add_date_function.reference new file mode 100644 index 00000000000..8912c8c8f63 --- /dev/null +++ b/tests/queries/0_stateless/02834_add_date_function.reference @@ -0,0 +1,4 @@ +2022-05-07 00:05:00 +2022-05-07 00:05:00.000 +2022-05-07 00:05:00 +2022-05-07 00:05:00.000 diff --git a/tests/queries/0_stateless/02834_add_date_function.sql b/tests/queries/0_stateless/02834_add_date_function.sql new file mode 100644 index 00000000000..1bf4cb0976e --- /dev/null +++ b/tests/queries/0_stateless/02834_add_date_function.sql @@ -0,0 +1,9 @@ +SELECT addDate('2022-05-07'::Date, INTERVAL 5 MINUTE); +SELECT addDate('2022-05-07'::Date32, INTERVAL 5 MINUTE); +SELECT addDate('2022-05-07'::DateTime, INTERVAL 5 MINUTE); +SELECT addDate('2022-05-07'::DateTime64, INTERVAL 5 MINUTE); + +SELECT addDate('2022-05-07'::Date); -- { serverError NUMBER_OF_ARGUMENTS_DOESNT_MATCH } +SELECT addDate('2022-05-07'::Date, INTERVAL 5 MINUTE, 5); -- { serverError NUMBER_OF_ARGUMENTS_DOESNT_MATCH } +SELECT addDate('2022-05-07'::Date, 10); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT addDate('1234', INTERVAL 5 MINUTE); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } From 9a3ea9d6dc7ce9c7f1217e0f15ade11e2953f44a Mon Sep 17 00:00:00 2001 From: Nikolay Degterinsky Date: Thu, 7 Sep 2023 11:31:09 +0000 Subject: [PATCH 189/433] Fix style --- utils/check-style/aspell-ignore/en/aspell-dict.txt | 1 + 1 file changed, 1 insertion(+) diff --git a/utils/check-style/aspell-ignore/en/aspell-dict.txt b/utils/check-style/aspell-ignore/en/aspell-dict.txt index a314815e2c4..aa130681f50 100644 --- a/utils/check-style/aspell-ignore/en/aspell-dict.txt +++ b/utils/check-style/aspell-ignore/en/aspell-dict.txt @@ -982,6 +982,7 @@ acos acosh activecube activerecord +addDate addDays addHours addMinutes From e821cd4511214e2ff66e89cf7cb7ddc6a27c45f4 Mon Sep 17 00:00:00 2001 From: Nikolay Degterinsky Date: Thu, 7 Sep 2023 12:01:07 +0000 Subject: [PATCH 190/433] Fix formatting of column declaration --- src/Parsers/ASTColumnDeclaration.cpp | 4 ---- 1 file changed, 4 deletions(-) diff --git a/src/Parsers/ASTColumnDeclaration.cpp b/src/Parsers/ASTColumnDeclaration.cpp index 12d000d5e9f..e9b490a1be3 100644 --- a/src/Parsers/ASTColumnDeclaration.cpp +++ b/src/Parsers/ASTColumnDeclaration.cpp @@ -77,10 +77,6 @@ void ASTColumnDeclaration::formatImpl(const FormatSettings & settings, FormatSta << (*null_modifier ? "" : "NOT ") << "NULL" << (settings.hilite ? hilite_none : ""); } - if (primary_key_specifier) - settings.ostr << ' ' << (settings.hilite ? hilite_keyword : "") - << "PRIMARY KEY" << (settings.hilite ? hilite_none : ""); - if (default_expression) { settings.ostr << ' ' << (settings.hilite ? hilite_keyword : "") << default_specifier << (settings.hilite ? hilite_none : ""); From 7d86d8b559739cb3d12d2dbd1ba213155cdfe7d8 Mon Sep 17 00:00:00 2001 From: Arthur Passos Date: Thu, 7 Sep 2023 09:22:54 -0300 Subject: [PATCH 191/433] trigger ci From 5afca8a903c318c4f01830827e5fed5c202344a9 Mon Sep 17 00:00:00 2001 From: yariks5s Date: Thu, 7 Sep 2023 12:41:01 +0000 Subject: [PATCH 192/433] optimized toString() --- src/Functions/FunctionsConversion.h | 234 ++++++++++++++++++---------- 1 file changed, 155 insertions(+), 79 deletions(-) diff --git a/src/Functions/FunctionsConversion.h b/src/Functions/FunctionsConversion.h index d1e51ee240b..98d28cf03a8 100644 --- a/src/Functions/FunctionsConversion.h +++ b/src/Functions/FunctionsConversion.h @@ -58,6 +58,7 @@ #include #include #include +#include "DataTypes/IDataType.h" #include @@ -883,104 +884,179 @@ struct ConvertImpl) { - auto non_null_args = createBlockWithNestedColumns(arguments); - time_zone = &extractTimeZoneFromFunctionArguments(non_null_args, 1, 0); + auto datetime_arg = arguments[0]; + + const DateLUTImpl * time_zone = nullptr; + const ColumnConst * time_zone_column = nullptr; + + if (arguments.size() == 1) + { + auto non_null_args = createBlockWithNestedColumns(arguments); + time_zone = &extractTimeZoneFromFunctionArguments(non_null_args, 1, 0); + } + else /// When we have a column for timezone + { + datetime_arg.column = datetime_arg.column->convertToFullColumnIfConst(); + + if constexpr (std::is_same_v || std::is_same_v) + time_zone = &DateLUT::instance(); + /// For argument of Date or DateTime type, second argument with time zone could be specified. + if constexpr (std::is_same_v || std::is_same_v) + { + if ((time_zone_column = checkAndGetColumnConst(arguments[1].column.get()))) + { + auto non_null_args = createBlockWithNestedColumns(arguments); + time_zone = &extractTimeZoneFromFunctionArguments(non_null_args, 1, 0); + } + } + } + const auto & col_with_type_and_name = columnGetNested(datetime_arg); + + if (const auto col_from = checkAndGetColumn(col_with_type_and_name.column.get())) + { + auto col_to = ColumnString::create(); + + const typename ColVecType::Container & vec_from = col_from->getData(); + ColumnString::Chars & data_to = col_to->getChars(); + ColumnString::Offsets & offsets_to = col_to->getOffsets(); + size_t size = vec_from.size(); + + if constexpr (std::is_same_v) + data_to.resize(size * (strlen("YYYY-MM-DD") + 1)); + else if constexpr (std::is_same_v) + data_to.resize(size * (strlen("YYYY-MM-DD") + 1)); + else if constexpr (std::is_same_v) + data_to.resize(size * (strlen("YYYY-MM-DD hh:mm:ss") + 1)); + else if constexpr (std::is_same_v) + data_to.resize(size * (strlen("YYYY-MM-DD hh:mm:ss.") + col_from->getScale() + 1)); + else + data_to.resize(size * 3); /// Arbitrary + + offsets_to.resize(size); + + WriteBufferFromVector write_buffer(data_to); + const auto & type = static_cast(*col_with_type_and_name.type); + + ColumnUInt8::MutablePtr null_map = copyNullMap(datetime_arg.column); + + if (null_map) + { + for (size_t i = 0; i < size; ++i) + { + if (!time_zone_column && arguments.size() > 1) + { + if (!arguments[1].column.get()->getDataAt(i).toString().empty()) + time_zone = &DateLUT::instance(arguments[1].column.get()->getDataAt(i).toString()); + else + throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Provided time zone must be non-empty"); + } + bool is_ok = FormatImpl::template execute(vec_from[i], write_buffer, &type, time_zone); + null_map->getData()[i] |= !is_ok; + writeChar(0, write_buffer); + offsets_to[i] = write_buffer.count(); + } + } + else + { + for (size_t i = 0; i < size; ++i) + { + if (!time_zone_column && arguments.size() > 1) + { + if (!arguments[1].column.get()->getDataAt(i).toString().empty()) + time_zone = &DateLUT::instance(arguments[1].column.get()->getDataAt(i).toString()); + else + throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Provided time zone must be non-empty"); + } + FormatImpl::template execute(vec_from[i], write_buffer, &type, time_zone); + writeChar(0, write_buffer); + offsets_to[i] = write_buffer.count(); + } + } + + write_buffer.finalize(); + + if (null_map) + return ColumnNullable::create(std::move(col_to), std::move(null_map)); + return col_to; + } + else + throw Exception(ErrorCodes::ILLEGAL_COLUMN, "Illegal column {} of first argument of function {}", + arguments[0].column->getName(), Name::name); } - else /// When we have a column for timezone + else { - datetime_arg.column = datetime_arg.column->convertToFullColumnIfConst(); + ColumnUInt8::MutablePtr null_map = copyNullMap(arguments[0].column); + + const auto & col_with_type_and_name = columnGetNested(arguments[0]); + const auto & type = static_cast(*col_with_type_and_name.type); + + const DateLUTImpl * time_zone = nullptr; if constexpr (std::is_same_v || std::is_same_v) time_zone = &DateLUT::instance(); /// For argument of Date or DateTime type, second argument with time zone could be specified. if constexpr (std::is_same_v || std::is_same_v) { - if ((time_zone_column = checkAndGetColumnConst(arguments[1].column.get()))) - { - auto non_null_args = createBlockWithNestedColumns(arguments); - time_zone = &extractTimeZoneFromFunctionArguments(non_null_args, 1, 0); - } + auto non_null_args = createBlockWithNestedColumns(arguments); + time_zone = &extractTimeZoneFromFunctionArguments(non_null_args, 1, 0); } - } - const auto & col_with_type_and_name = columnGetNested(datetime_arg); - if (const auto col_from = checkAndGetColumn(col_with_type_and_name.column.get())) - { - auto col_to = ColumnString::create(); - - const typename ColVecType::Container & vec_from = col_from->getData(); - ColumnString::Chars & data_to = col_to->getChars(); - ColumnString::Offsets & offsets_to = col_to->getOffsets(); - size_t size = vec_from.size(); - - if constexpr (std::is_same_v) - data_to.resize(size * (strlen("YYYY-MM-DD") + 1)); - else if constexpr (std::is_same_v) - data_to.resize(size * (strlen("YYYY-MM-DD") + 1)); - else if constexpr (std::is_same_v) - data_to.resize(size * (strlen("YYYY-MM-DD hh:mm:ss") + 1)); - else if constexpr (std::is_same_v) - data_to.resize(size * (strlen("YYYY-MM-DD hh:mm:ss.") + col_from->getScale() + 1)); - else - data_to.resize(size * 3); /// Arbitrary - - offsets_to.resize(size); - - WriteBufferFromVector write_buffer(data_to); - const auto & type = static_cast(*col_with_type_and_name.type); - - ColumnUInt8::MutablePtr null_map = copyNullMap(datetime_arg.column); - - if (null_map) + if (const auto col_from = checkAndGetColumn(col_with_type_and_name.column.get())) { - for (size_t i = 0; i < size; ++i) + auto col_to = ColumnString::create(); + + const typename ColVecType::Container & vec_from = col_from->getData(); + ColumnString::Chars & data_to = col_to->getChars(); + ColumnString::Offsets & offsets_to = col_to->getOffsets(); + size_t size = vec_from.size(); + + if constexpr (std::is_same_v) + data_to.resize(size * (strlen("YYYY-MM-DD") + 1)); + else if constexpr (std::is_same_v) + data_to.resize(size * (strlen("YYYY-MM-DD") + 1)); + else if constexpr (std::is_same_v) + data_to.resize(size * (strlen("YYYY-MM-DD hh:mm:ss") + 1)); + else if constexpr (std::is_same_v) + data_to.resize(size * (strlen("YYYY-MM-DD hh:mm:ss.") + col_from->getScale() + 1)); + else + data_to.resize(size * 3); /// Arbitrary + + offsets_to.resize(size); + + WriteBufferFromVector write_buffer(data_to); + + if (null_map) { - if (!time_zone_column && arguments.size() > 1) + for (size_t i = 0; i < size; ++i) { - if (!arguments[1].column.get()->getDataAt(i).toString().empty()) - time_zone = &DateLUT::instance(arguments[1].column.get()->getDataAt(i).toString()); - else - throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Provided time zone must be non-empty"); + bool is_ok = FormatImpl::template execute(vec_from[i], write_buffer, &type, time_zone); + null_map->getData()[i] |= !is_ok; + writeChar(0, write_buffer); + offsets_to[i] = write_buffer.count(); } - bool is_ok = FormatImpl::template execute(vec_from[i], write_buffer, &type, time_zone); - null_map->getData()[i] |= !is_ok; - writeChar(0, write_buffer); - offsets_to[i] = write_buffer.count(); } + else + { + for (size_t i = 0; i < size; ++i) + { + FormatImpl::template execute(vec_from[i], write_buffer, &type, time_zone); + writeChar(0, write_buffer); + offsets_to[i] = write_buffer.count(); + } + } + + write_buffer.finalize(); + + if (null_map) + return ColumnNullable::create(std::move(col_to), std::move(null_map)); + return col_to; } else - { - for (size_t i = 0; i < size; ++i) - { - if (!time_zone_column && arguments.size() > 1) - { - if (!arguments[1].column.get()->getDataAt(i).toString().empty()) - time_zone = &DateLUT::instance(arguments[1].column.get()->getDataAt(i).toString()); - else - throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Provided time zone must be non-empty"); - } - FormatImpl::template execute(vec_from[i], write_buffer, &type, time_zone); - writeChar(0, write_buffer); - offsets_to[i] = write_buffer.count(); - } - } - - write_buffer.finalize(); - - if (null_map) - return ColumnNullable::create(std::move(col_to), std::move(null_map)); - return col_to; + throw Exception(ErrorCodes::ILLEGAL_COLUMN, "Illegal column {} of first argument of function {}", + arguments[0].column->getName(), Name::name); } - else - throw Exception(ErrorCodes::ILLEGAL_COLUMN, "Illegal column {} of first argument of function {}", - arguments[0].column->getName(), Name::name); } }; From 75dc6d9708ed6768ff220b55e98643b6c3e2f37c Mon Sep 17 00:00:00 2001 From: Alex Cheng Date: Thu, 7 Sep 2023 20:41:32 +0800 Subject: [PATCH 193/433] small fine-tune for using ColumnNullable pointer --- src/Processors/Transforms/MergeJoinTransform.cpp | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/src/Processors/Transforms/MergeJoinTransform.cpp b/src/Processors/Transforms/MergeJoinTransform.cpp index 46abd5c891d..5ee8d677dec 100644 --- a/src/Processors/Transforms/MergeJoinTransform.cpp +++ b/src/Processors/Transforms/MergeJoinTransform.cpp @@ -52,12 +52,12 @@ int nullableCompareAt(const IColumn & left_column, const IColumn & right_column, if (left_nullable && right_nullable) { - int res = left_column.compareAt(lhs_pos, rhs_pos, right_column, null_direction_hint); + int res = left_nullable->compareAt(lhs_pos, rhs_pos, right_column, null_direction_hint); if (res) return res; /// NULL != NULL case - if (left_column.isNullAt(lhs_pos)) + if (left_nullable->isNullAt(lhs_pos)) return null_direction_hint; return 0; @@ -68,7 +68,7 @@ int nullableCompareAt(const IColumn & left_column, const IColumn & right_column, { if (const auto * left_nullable = checkAndGetColumn(left_column)) { - if (left_column.isNullAt(lhs_pos)) + if (left_nullable->isNullAt(lhs_pos)) return null_direction_hint; return left_nullable->getNestedColumn().compareAt(lhs_pos, rhs_pos, right_column, null_direction_hint); } @@ -78,7 +78,7 @@ int nullableCompareAt(const IColumn & left_column, const IColumn & right_column, { if (const auto * right_nullable = checkAndGetColumn(right_column)) { - if (right_column.isNullAt(rhs_pos)) + if (right_nullable->isNullAt(rhs_pos)) return -null_direction_hint; return left_column.compareAt(lhs_pos, rhs_pos, right_nullable->getNestedColumn(), null_direction_hint); } From a0317e461a1f227895de22933b040cc4cd45f28c Mon Sep 17 00:00:00 2001 From: Igor Nikonov Date: Thu, 7 Sep 2023 12:41:36 +0000 Subject: [PATCH 194/433] Better description --- tests/ci/fast_test_check.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/tests/ci/fast_test_check.py b/tests/ci/fast_test_check.py index 96eb144da2c..052aafb3395 100644 --- a/tests/ci/fast_test_check.py +++ b/tests/ci/fast_test_check.py @@ -145,8 +145,8 @@ def main(): logs_path.mkdir(parents=True, exist_ok=True) run_log_path = logs_path / "run.log" - timeout = 65 timeout_expired = False + timeout = 90 * 60 with TeePopen(run_cmd, run_log_path, timeout=timeout) as process: retcode = process.wait() if process.timeout_exceeded: @@ -186,7 +186,7 @@ def main(): state, description, test_results, additional_logs = process_results(output_path) if timeout_expired is not None: - test_result_name = "Timeout for fast test check is expired" + test_result_name = "Check timeout expired" test_results.append(TestResult(test_result_name, "FAIL", timeout)) state = "failure" description = format_description(test_result_name) From c354fcaff7e5a53e2c14265aac595d6ad9f065dd Mon Sep 17 00:00:00 2001 From: Igor Nikonov Date: Thu, 7 Sep 2023 12:56:22 +0000 Subject: [PATCH 195/433] Timeout report for stress_check --- tests/ci/fast_test_check.py | 4 ++-- tests/ci/stress_check.py | 20 ++++++++++++++++---- 2 files changed, 18 insertions(+), 6 deletions(-) diff --git a/tests/ci/fast_test_check.py b/tests/ci/fast_test_check.py index 052aafb3395..3d6d43a2d54 100644 --- a/tests/ci/fast_test_check.py +++ b/tests/ci/fast_test_check.py @@ -150,7 +150,7 @@ def main(): with TeePopen(run_cmd, run_log_path, timeout=timeout) as process: retcode = process.wait() if process.timeout_exceeded: - logging.info(f"Timeout expired for process execution: {run_cmd}") + logging.info(f"Timeout expired for command: {run_cmd}") timeout_expired = True elif retcode == 0: logging.info("Run successfully") @@ -185,7 +185,7 @@ def main(): else: state, description, test_results, additional_logs = process_results(output_path) - if timeout_expired is not None: + if timeout_expired: test_result_name = "Check timeout expired" test_results.append(TestResult(test_result_name, "FAIL", timeout)) state = "failure" diff --git a/tests/ci/stress_check.py b/tests/ci/stress_check.py index 7bac4e1d511..933bdc2ba60 100644 --- a/tests/ci/stress_check.py +++ b/tests/ci/stress_check.py @@ -16,12 +16,12 @@ from clickhouse_helper import ( ClickHouseHelper, prepare_tests_results_for_clickhouse, ) -from commit_status_helper import RerunHelper, get_commit, post_commit_status +from commit_status_helper import RerunHelper, get_commit, post_commit_status, format_description from docker_pull_helper import DockerImage, get_image_with_version from env_helper import TEMP_PATH, REPO_COPY, REPORTS_PATH from get_robot_token import get_best_robot_token from pr_info import PRInfo -from report import TestResults, read_test_results +from report import TestResult, TestResults, read_test_results from s3_helper import S3Helper from stopwatch import Stopwatch from tee_popen import TeePopen @@ -169,9 +169,14 @@ def run_stress_test(docker_image_name): ) logging.info("Going to run stress test: %s", run_command) - with TeePopen(run_command, run_log_path, timeout=60 * 150) as process: + timeout_expired = False + timeout = 60 * 150 + with TeePopen(run_command, run_log_path, timeout=timeout) as process: retcode = process.wait() - if retcode == 0: + if process.timeout_exceeded: + logging.info(f"Timeout expired for command: {run_command}") + timeout_expired = True + elif retcode == 0: logging.info("Run successfully") else: logging.info("Run failed") @@ -183,6 +188,13 @@ def run_stress_test(docker_image_name): state, description, test_results, additional_logs = process_results( result_path, server_log_path, run_log_path ) + + if timeout_expired: + test_result_name = "Check timeout expired" + test_results.append(TestResult(test_result_name, "FAIL", timeout)) + state = "failure" + description = format_description(test_result_name) + ch_helper = ClickHouseHelper() report_url = upload_results( From 27455a5882d95b474dbae8311c75d62449ec1990 Mon Sep 17 00:00:00 2001 From: Igor Nikonov Date: Thu, 7 Sep 2023 12:57:35 +0000 Subject: [PATCH 196/433] Test timeout to fail --- tests/ci/stress_check.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tests/ci/stress_check.py b/tests/ci/stress_check.py index 933bdc2ba60..9f832192886 100644 --- a/tests/ci/stress_check.py +++ b/tests/ci/stress_check.py @@ -170,7 +170,7 @@ def run_stress_test(docker_image_name): logging.info("Going to run stress test: %s", run_command) timeout_expired = False - timeout = 60 * 150 + timeout = 60 * 1 with TeePopen(run_command, run_log_path, timeout=timeout) as process: retcode = process.wait() if process.timeout_exceeded: From a5b508d6c5d6a87c6bd77c6af4b09c16514270bc Mon Sep 17 00:00:00 2001 From: Joe Lynch Date: Thu, 7 Sep 2023 14:59:33 +0200 Subject: [PATCH 197/433] Fix json in header --- docker/test/stateless/Dockerfile | 1 + src/IO/Progress.cpp | 1 + .../0_stateless/02869_http_headers_elapsed_ns.sh | 12 ++++++------ 3 files changed, 8 insertions(+), 6 deletions(-) diff --git a/docker/test/stateless/Dockerfile b/docker/test/stateless/Dockerfile index 35a6e9c365b..d96531e84c2 100644 --- a/docker/test/stateless/Dockerfile +++ b/docker/test/stateless/Dockerfile @@ -40,6 +40,7 @@ RUN apt-get update -y \ cargo \ zstd \ file \ + jq \ pv \ zip \ p7zip-full \ diff --git a/src/IO/Progress.cpp b/src/IO/Progress.cpp index 56fb7fe317e..1dcb206496f 100644 --- a/src/IO/Progress.cpp +++ b/src/IO/Progress.cpp @@ -92,6 +92,7 @@ void ProgressValues::writeJSON(WriteBuffer & out, bool add_braces) const writeText(result_bytes, out); writeCString("\",\"elapsed_ns\":\"", out); writeText(elapsed_ns, out); + writeCString("\"", out); if (add_braces) writeCString("}", out); } diff --git a/tests/queries/0_stateless/02869_http_headers_elapsed_ns.sh b/tests/queries/0_stateless/02869_http_headers_elapsed_ns.sh index 107d8a10345..df6302bb82a 100755 --- a/tests/queries/0_stateless/02869_http_headers_elapsed_ns.sh +++ b/tests/queries/0_stateless/02869_http_headers_elapsed_ns.sh @@ -8,17 +8,17 @@ CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) CURL_OUTPUT=$(echo 'SELECT number FROM numbers(10)' | \ ${CLICKHOUSE_CURL} -vsS "${CLICKHOUSE_URL}&max_block_size=1&send_progress_in_http_headers=1&http_headers_progress_interval_ms=0&output_format_parallel_formatting=0" --data-binary @- 2>&1) -ELAPSED_NS_PROGRESS=$(echo "${CURL_OUTPUT}" | \ +ELAPSED_NS_PROGRESS="$(echo "${CURL_OUTPUT}" | \ grep 'X-ClickHouse-Progress' | \ awk '{print $3}' | \ - sed -E 's/.*"elapsed_ns":"?([^,"]*)"?.*/\1/' - ) + jq -cM '.elapsed_ns | tonumber' + )" -ELAPSED_NS_SUMMARY=$(echo "${CURL_OUTPUT}" | \ +ELAPSED_NS_SUMMARY="$(echo "${CURL_OUTPUT}" | \ grep 'X-ClickHouse-Summary' | \ awk '{print $3}' | \ - sed -E 's/.*"elapsed_ns":"?([^,"]*)"?.*/\1/' - ) + jq -cM '.elapsed_ns | tonumber' + )" ALL_ARE_NON_ZERO=1 From e761532d7ad94bd5ce351b696d1d82accaa2fee4 Mon Sep 17 00:00:00 2001 From: Robert Schulze Date: Thu, 7 Sep 2023 15:00:06 +0200 Subject: [PATCH 198/433] Fix style --- src/Functions/FunctionBase64Conversion.h | 2 -- 1 file changed, 2 deletions(-) diff --git a/src/Functions/FunctionBase64Conversion.h b/src/Functions/FunctionBase64Conversion.h index 15dcbaf8352..f52dec0eaf7 100644 --- a/src/Functions/FunctionBase64Conversion.h +++ b/src/Functions/FunctionBase64Conversion.h @@ -18,9 +18,7 @@ namespace DB { namespace ErrorCodes { - extern const int BAD_ARGUMENTS; extern const int ILLEGAL_COLUMN; - extern const int ILLEGAL_TYPE_OF_ARGUMENT; extern const int INCORRECT_DATA; } From c680c951d3d371342a37d4607b1876238d3382dc Mon Sep 17 00:00:00 2001 From: robot-clickhouse Date: Thu, 7 Sep 2023 13:12:53 +0000 Subject: [PATCH 199/433] Automatic style fix --- tests/ci/stress_check.py | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/tests/ci/stress_check.py b/tests/ci/stress_check.py index 9f832192886..9e607f7da82 100644 --- a/tests/ci/stress_check.py +++ b/tests/ci/stress_check.py @@ -16,7 +16,12 @@ from clickhouse_helper import ( ClickHouseHelper, prepare_tests_results_for_clickhouse, ) -from commit_status_helper import RerunHelper, get_commit, post_commit_status, format_description +from commit_status_helper import ( + RerunHelper, + get_commit, + post_commit_status, + format_description, +) from docker_pull_helper import DockerImage, get_image_with_version from env_helper import TEMP_PATH, REPO_COPY, REPORTS_PATH from get_robot_token import get_best_robot_token From c9635ddacd6e06191b6798b91e00c051b8b645f2 Mon Sep 17 00:00:00 2001 From: Sergei Trifonov Date: Thu, 7 Sep 2023 15:48:49 +0200 Subject: [PATCH 200/433] Update src/IO/ISchedulerNode.h Co-authored-by: Kseniia Sumarokova <54203879+kssenii@users.noreply.github.com> --- src/IO/ISchedulerNode.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/IO/ISchedulerNode.h b/src/IO/ISchedulerNode.h index 644cc708bca..5cf1ae94216 100644 --- a/src/IO/ISchedulerNode.h +++ b/src/IO/ISchedulerNode.h @@ -3,7 +3,7 @@ #include #include #include -#include "base/types.h" +#include #include #include From 958f102daae2a8e33d59529918c7126c1b08afa2 Mon Sep 17 00:00:00 2001 From: Sergei Trifonov Date: Thu, 7 Sep 2023 15:58:31 +0200 Subject: [PATCH 201/433] Update src/IO/IResourceManager.h Co-authored-by: Kseniia Sumarokova <54203879+kssenii@users.noreply.github.com> --- src/IO/IResourceManager.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/src/IO/IResourceManager.h b/src/IO/IResourceManager.h index 7a078c565d8..b4270cd1935 100644 --- a/src/IO/IResourceManager.h +++ b/src/IO/IResourceManager.h @@ -48,7 +48,8 @@ public: virtual ClassifierPtr acquire(const String & classifier_name) = 0; /// For introspection, see `system.scheduler` table - virtual void forEachNode(std::function visitor) = 0; + using VisitorFunc = std::function; + void forEachNode(VisitorFunc visitor) override; }; using ResourceManagerPtr = std::shared_ptr; From a94ae9a7abcf7379145d93fc9465264c2aaae5f4 Mon Sep 17 00:00:00 2001 From: serxa Date: Thu, 7 Sep 2023 14:02:58 +0000 Subject: [PATCH 202/433] review fixes --- docs/en/operations/workload-scheduling.md | 14 +++++++------- src/IO/Resource/ClassifiersConfig.cpp | 4 ++-- src/IO/Resource/ClassifiersConfig.h | 4 ++-- src/IO/Resource/DynamicResourceManager.cpp | 2 +- src/IO/Resource/DynamicResourceManager.h | 2 +- src/IO/Resource/SemaphoreConstraint.h | 2 +- src/IO/Resource/StaticResourceManager.h | 2 +- .../tests/gtest_resource_manager_hierarchical.cpp | 8 ++++---- .../tests/gtest_resource_manager_static.cpp | 8 ++++---- .../test_scheduler/configs/scheduler.xml | 4 ++-- 10 files changed, 25 insertions(+), 25 deletions(-) diff --git a/docs/en/operations/workload-scheduling.md b/docs/en/operations/workload-scheduling.md index c6529d5b81f..6ed6ced41b4 100644 --- a/docs/en/operations/workload-scheduling.md +++ b/docs/en/operations/workload-scheduling.md @@ -78,10 +78,10 @@ graph TD ``` **Possible node types:** -* inflight_limit (constraint) - blocks if either number of concurrent in-flight requests exceeds `max_requests`, or their total cost exceeds `max_cost`; must have a single child. -* fair (policy) - selects the next request to serve from one of its children nodes according to max-min fairness; children nodes can specify `weight` (default is 1). -* priority (policy) - selects the next request to serve from one of its children nodes according to static priorities (lower value means higher priority); children nodes can specify `priority` (default is 0). -* fifo (queue) - leaf of the hierarchy capable of holding requests that exceed resource capacity. +* `inflight_limit` (constraint) - blocks if either number of concurrent in-flight requests exceeds `max_requests`, or their total cost exceeds `max_cost`; must have a single child. +* `fair` (policy) - selects the next request to serve from one of its children nodes according to max-min fairness; children nodes can specify `weight` (default is 1). +* `priority` (policy) - selects the next request to serve from one of its children nodes according to static priorities (lower value means higher priority); children nodes can specify `priority` (default is 0). +* `fifo` (queue) - leaf of the hierarchy capable of holding requests that exceed resource capacity. The following example shows how to define IO scheduling hierarchies shown in the picture: @@ -124,14 +124,14 @@ The following example shows how to define IO scheduling hierarchies shown in the ``` -## Workload classifiers {#classifiers} +## Workload classifiers {#workload_classifiers} Workload classifiers are used to define mapping from `workload` specified by a query into leaf-queues that should be used for specific resources. At the moment, workload classification is simple: only static mapping is available. Example: ```xml - + /fair/prod /fair/prod @@ -144,7 +144,7 @@ Example: /fair/dev /fair/dev - + ``` diff --git a/src/IO/Resource/ClassifiersConfig.cpp b/src/IO/Resource/ClassifiersConfig.cpp index fcd4655e2e4..7dc4d517138 100644 --- a/src/IO/Resource/ClassifiersConfig.cpp +++ b/src/IO/Resource/ClassifiersConfig.cpp @@ -21,7 +21,7 @@ ClassifierDescription::ClassifierDescription(const Poco::Util::AbstractConfigura ClassifiersConfig::ClassifiersConfig(const Poco::Util::AbstractConfiguration & config) { Poco::Util::AbstractConfiguration::Keys keys; - const String config_prefix = "classifiers"; + const String config_prefix = "workload_classifiers"; config.keys(config_prefix, keys); for (const auto & key : keys) classifiers.emplace(std::piecewise_construct, @@ -34,7 +34,7 @@ const ClassifierDescription & ClassifiersConfig::get(const String & classifier_n if (auto it = classifiers.find(classifier_name); it != classifiers.end()) return it->second; else - throw Exception(ErrorCodes::RESOURCE_NOT_FOUND, "Unknown classifier '{}' to access resources", classifier_name); + throw Exception(ErrorCodes::RESOURCE_NOT_FOUND, "Unknown workload classifier '{}' to access resources", classifier_name); } } diff --git a/src/IO/Resource/ClassifiersConfig.h b/src/IO/Resource/ClassifiersConfig.h index 96e2bd0f0b9..186c49943ad 100644 --- a/src/IO/Resource/ClassifiersConfig.h +++ b/src/IO/Resource/ClassifiersConfig.h @@ -15,14 +15,14 @@ struct ClassifierDescription : std::unordered_map /* * Loads a config with the following format: - * + * * * /path/to/queue * /path/to/another/queue * * ... * ... - * + * */ class ClassifiersConfig { diff --git a/src/IO/Resource/DynamicResourceManager.cpp b/src/IO/Resource/DynamicResourceManager.cpp index 964f4f662e6..b9803d8079d 100644 --- a/src/IO/Resource/DynamicResourceManager.cpp +++ b/src/IO/Resource/DynamicResourceManager.cpp @@ -227,7 +227,7 @@ ClassifierPtr DynamicResourceManager::acquire(const String & classifier_name) return std::make_shared(state_ref, classifier_name); } -void DynamicResourceManager::forEachNode(std::function visitor) +void DynamicResourceManager::forEachNode(IResourceManager::VisitorFunc visitor) { // Acquire a reference to the current state StatePtr state_ref; diff --git a/src/IO/Resource/DynamicResourceManager.h b/src/IO/Resource/DynamicResourceManager.h index 10d784503ca..3372d40a285 100644 --- a/src/IO/Resource/DynamicResourceManager.h +++ b/src/IO/Resource/DynamicResourceManager.h @@ -30,7 +30,7 @@ public: DynamicResourceManager(); void updateConfiguration(const Poco::Util::AbstractConfiguration & config) override; ClassifierPtr acquire(const String & classifier_name) override; - void forEachNode(std::function visitor) override; + void forEachNode(VisitorFunc visitor) override; private: /// Holds everything required to work with one specific configuration diff --git a/src/IO/Resource/SemaphoreConstraint.h b/src/IO/Resource/SemaphoreConstraint.h index 9160d0431de..9c6ce43d6ea 100644 --- a/src/IO/Resource/SemaphoreConstraint.h +++ b/src/IO/Resource/SemaphoreConstraint.h @@ -121,7 +121,7 @@ public: size_t activeChildren() override { std::unique_lock lock(mutex); - return child_active ? 1 : 0; + return child_active; } bool isSatisfied() override diff --git a/src/IO/Resource/StaticResourceManager.h b/src/IO/Resource/StaticResourceManager.h index b8712b22b73..5ec6a35750b 100644 --- a/src/IO/Resource/StaticResourceManager.h +++ b/src/IO/Resource/StaticResourceManager.h @@ -22,7 +22,7 @@ public: ClassifierPtr acquire(const String & classifier_name) override; - void forEachNode(std::function visitor) override + void forEachNode(VisitorFunc visitor) override { UNUSED(visitor); } diff --git a/src/IO/Resource/tests/gtest_resource_manager_hierarchical.cpp b/src/IO/Resource/tests/gtest_resource_manager_hierarchical.cpp index 43773559f03..949a1ee0264 100644 --- a/src/IO/Resource/tests/gtest_resource_manager_hierarchical.cpp +++ b/src/IO/Resource/tests/gtest_resource_manager_hierarchical.cpp @@ -24,10 +24,10 @@ TEST(IOResourceDynamicResourceManager, Smoke) fifo3 - + /fair/A /fair/B - + )CONFIG"); @@ -71,11 +71,11 @@ TEST(IOResourceDynamicResourceManager, Fairness) fifo - + /fair/A /fair/B /fair/leader - + )CONFIG"); diff --git a/src/IO/Resource/tests/gtest_resource_manager_static.cpp b/src/IO/Resource/tests/gtest_resource_manager_static.cpp index 976eac41a49..9c5e86e9ffc 100644 --- a/src/IO/Resource/tests/gtest_resource_manager_static.cpp +++ b/src/IO/Resource/tests/gtest_resource_manager_static.cpp @@ -24,10 +24,10 @@ TEST(IOResourceStaticResourceManager, Smoke) 1 - + /prio/A /prio/B - + )CONFIG"); @@ -70,13 +70,13 @@ TEST(IOResourceStaticResourceManager, Prioritization) - + /prio/A /prio/B /prio/C /prio/D /prio/leader - + )CONFIG"); diff --git a/tests/integration/test_scheduler/configs/scheduler.xml b/tests/integration/test_scheduler/configs/scheduler.xml index fe01cb8d662..523ba1a5a98 100644 --- a/tests/integration/test_scheduler/configs/scheduler.xml +++ b/tests/integration/test_scheduler/configs/scheduler.xml @@ -41,7 +41,7 @@ fifo1 - + /prio/admin /prio/admin @@ -58,5 +58,5 @@ /prio/fair/dev /prio/fair/dev - + From fae0e13baa18629eeb40baaf94f8cc7c0844bbab Mon Sep 17 00:00:00 2001 From: Antonio Andelic Date: Thu, 7 Sep 2023 14:34:36 +0000 Subject: [PATCH 203/433] Properly remove added interserver endpoint --- src/Storages/StorageReplicatedMergeTree.cpp | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git a/src/Storages/StorageReplicatedMergeTree.cpp b/src/Storages/StorageReplicatedMergeTree.cpp index e6edb99e301..e52b1159b40 100644 --- a/src/Storages/StorageReplicatedMergeTree.cpp +++ b/src/Storages/StorageReplicatedMergeTree.cpp @@ -4890,6 +4890,16 @@ void StorageReplicatedMergeTree::startupImpl(bool from_attach_thread) if (from_attach_thread) { restarting_thread.shutdown(/* part_of_full_shutdown */false); + + auto data_parts_exchange_ptr = std::atomic_exchange(&data_parts_exchange_endpoint, InterserverIOEndpointPtr{}); + if (data_parts_exchange_ptr) + { + getContext()->getInterserverIOHandler().removeEndpointIfExists(data_parts_exchange_ptr->getId(getEndpointName())); + /// Ask all parts exchange handlers to finish asap. New ones will fail to start + data_parts_exchange_ptr->blocker.cancelForever(); + /// Wait for all of them + std::lock_guard lock(data_parts_exchange_ptr->rwlock); + } } else { From fd0da1f27855fc47284430fd6c8f63a00cf533cb Mon Sep 17 00:00:00 2001 From: Suzy Wang Date: Thu, 7 Sep 2023 11:00:12 -0300 Subject: [PATCH 204/433] hashtable order fix for s390x --- src/Common/HashTable/Hash.h | 25 +++++++++++++++++++ src/Common/HashTable/StringHashTable.h | 22 ++++++++++++++++ .../integer_hash_tables_and_hashes.cpp | 5 ++++ src/Interpreters/AggregationCommon.h | 5 +++- src/Interpreters/Aggregator.h | 5 +++- 5 files changed, 60 insertions(+), 2 deletions(-) diff --git a/src/Common/HashTable/Hash.h b/src/Common/HashTable/Hash.h index 49ab875297c..870e6eaa863 100644 --- a/src/Common/HashTable/Hash.h +++ b/src/Common/HashTable/Hash.h @@ -393,6 +393,18 @@ struct UInt128HashCRC32 } }; +#elif defined(__s390x__) && __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ + +struct UInt128HashCRC32 +{ + size_t operator()(UInt128 x) const + { + UInt64 crc = -1ULL; + crc = s390x_crc32(crc, x.items[UInt128::_impl::little(0)]); + crc = s390x_crc32(crc, x.items[UInt128::_impl::little(1)]); + return crc; + } +}; #else /// On other platforms we do not use CRC32. NOTE This can be confusing. @@ -451,6 +463,19 @@ struct UInt256HashCRC32 } }; +#elif defined(__s390x__) && __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ +struct UInt256HashCRC32 +{ + size_t operator()(UInt256 x) const + { + UInt64 crc = -1ULL; + crc = s390x_crc32(crc, x.items[UInt256::_impl::little(0)]); + crc = s390x_crc32(crc, x.items[UInt256::_impl::little(1)]); + crc = s390x_crc32(crc, x.items[UInt256::_impl::little(2)]); + crc = s390x_crc32(crc, x.items[UInt256::_impl::little(3)]); + return crc; + } +}; #else /// We do not need to use CRC32 on other platforms. NOTE This can be confusing. diff --git a/src/Common/HashTable/StringHashTable.h b/src/Common/HashTable/StringHashTable.h index a9d078f6cf3..f91e89dc134 100644 --- a/src/Common/HashTable/StringHashTable.h +++ b/src/Common/HashTable/StringHashTable.h @@ -71,6 +71,28 @@ struct StringHashTableHash res = _mm_crc32_u64(res, key.c); return res; } +#elif defined(__s390x__) && __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ + size_t ALWAYS_INLINE operator()(StringKey8 key) const + { + size_t res = -1ULL; + res = s390x_crc32(res, key); + return res; + } + size_t ALWAYS_INLINE operator()(StringKey16 key) const + { + size_t res = -1ULL; + res = s390x_crc32(res, key.items[UInt128::_impl::little(0)]); + res = s390x_crc32(res, key.items[UInt128::_impl::little(1)]); + return res; + } + size_t ALWAYS_INLINE operator()(StringKey24 key) const + { + size_t res = -1ULL; + res = s390x_crc32(res, key.a); + res = s390x_crc32(res, key.b); + res = s390x_crc32(res, key.c); + return res; + } #else size_t ALWAYS_INLINE operator()(StringKey8 key) const { diff --git a/src/Common/benchmarks/integer_hash_tables_and_hashes.cpp b/src/Common/benchmarks/integer_hash_tables_and_hashes.cpp index 2a4e366f1c5..40747f9f4db 100644 --- a/src/Common/benchmarks/integer_hash_tables_and_hashes.cpp +++ b/src/Common/benchmarks/integer_hash_tables_and_hashes.cpp @@ -188,12 +188,17 @@ namespace Hashes #include #endif + #if defined(__s390x__) && __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ + #include + #endif struct CRC32Hash { size_t operator()(Key x) const { #ifdef __SSE4_2__ return _mm_crc32_u64(-1ULL, x); + #elif defined(__s390x__) && __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ + return s390x_crc32(-1ULL, x); #else /// On other platforms we do not have CRC32. NOTE This can be confusing. return intHash64(x); diff --git a/src/Interpreters/AggregationCommon.h b/src/Interpreters/AggregationCommon.h index 77ca2c49b31..722675f62ed 100644 --- a/src/Interpreters/AggregationCommon.h +++ b/src/Interpreters/AggregationCommon.h @@ -90,7 +90,10 @@ void fillFixedBatch(size_t keys_size, const ColumnRawPtrs & key_columns, const S /// Note: here we violate strict aliasing. /// It should be ok as log as we do not reffer to any value from `out` before filling. const char * source = static_cast(column)->getRawDataBegin(); - T * dest = reinterpret_cast(reinterpret_cast(out.data()) + offset); + size_t offset_to = offset; + if constexpr (std::endian::native == std::endian::big) + offset_to = sizeof(Key) - sizeof(T) - offset; + T * dest = reinterpret_cast(reinterpret_cast(out.data()) + offset_to); fillFixedBatch(num_rows, reinterpret_cast(source), dest); offset += sizeof(T); } diff --git a/src/Interpreters/Aggregator.h b/src/Interpreters/Aggregator.h index 4f2c86606c5..48a5cbb8aff 100644 --- a/src/Interpreters/Aggregator.h +++ b/src/Interpreters/Aggregator.h @@ -512,7 +512,10 @@ struct AggregationMethodKeysFixed else { size_t size = key_sizes[i]; - observed_column->insertData(reinterpret_cast(&key) + pos, size); + size_t offset_to = pos; + if constexpr (std::endian::native == std::endian::big) + offset_to = sizeof(Key) - size - pos; + observed_column->insertData(reinterpret_cast(&key) + offset_to, size); pos += size; } } From 7abc9c896d8ac053036fc03fc7391d8a65b6e255 Mon Sep 17 00:00:00 2001 From: Kruglov Pavel <48961922+Avogar@users.noreply.github.com> Date: Thu, 7 Sep 2023 18:06:46 +0200 Subject: [PATCH 205/433] Fix style --- tests/integration/test_storage_azure_blob_storage/test.py | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) diff --git a/tests/integration/test_storage_azure_blob_storage/test.py b/tests/integration/test_storage_azure_blob_storage/test.py index b58f0272b7f..e386f200f7f 100644 --- a/tests/integration/test_storage_azure_blob_storage/test.py +++ b/tests/integration/test_storage_azure_blob_storage/test.py @@ -35,7 +35,9 @@ def cluster(): cluster.shutdown() -def azure_query(node, query, expect_error="false", try_num=10, settings={}, query_on_retry = None): +def azure_query( + node, query, expect_error="false", try_num=10, settings={}, query_on_retry = None +): for i in range(try_num): try: if expect_error == "true": @@ -763,13 +765,13 @@ def run_describe_query(instance, file, connection_string): azure_query(instance, query) -def run_count_query(instance, file, connection_string, drop_cache_on_retry = False): +def run_count_query(instance, file, connection_string, drop_cache_on_retry=False): query = f"select count() from azureBlobStorage('{connection_string}', 'cont', '{file}', auto, auto, 'x UInt64')" if drop_cache_on_retry: return azure_query( node=instance, query=query, - query_on_retry="system drop schema cache for azure" + query_on_retry="system drop schema cache for azure", ) return azure_query(instance, query) From 2d6503dd7f242c16255de036a6c89f91bf332679 Mon Sep 17 00:00:00 2001 From: Kseniia Sumarokova <54203879+kssenii@users.noreply.github.com> Date: Thu, 7 Sep 2023 18:10:27 +0200 Subject: [PATCH 206/433] Update README.md --- tests/fuzz/README.md | 1 - 1 file changed, 1 deletion(-) diff --git a/tests/fuzz/README.md b/tests/fuzz/README.md index d9d89fbfe96..5b864babde3 100644 --- a/tests/fuzz/README.md +++ b/tests/fuzz/README.md @@ -10,5 +10,4 @@ The list of datatypes generated via following query: clickhouse-client -q "select concat('\"', name, '\"') from system.data_type_families union all select concat('\"', alias_to, '\"') from system.data_type_families where alias_to != '' " > datatypes.dict ``` - Then merge all dictionaries into one (all.dict) From 8c2b45aab37fa5bdc48475b0ea1bfeda9686e81c Mon Sep 17 00:00:00 2001 From: Kruglov Pavel <48961922+Avogar@users.noreply.github.com> Date: Thu, 7 Sep 2023 18:42:04 +0200 Subject: [PATCH 207/433] Fix style --- tests/integration/test_storage_azure_blob_storage/test.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tests/integration/test_storage_azure_blob_storage/test.py b/tests/integration/test_storage_azure_blob_storage/test.py index e386f200f7f..3ec0506c525 100644 --- a/tests/integration/test_storage_azure_blob_storage/test.py +++ b/tests/integration/test_storage_azure_blob_storage/test.py @@ -36,7 +36,7 @@ def cluster(): def azure_query( - node, query, expect_error="false", try_num=10, settings={}, query_on_retry = None + node, query, expect_error="false", try_num=10, settings={}, query_on_retry=None ): for i in range(try_num): try: From 67b27643edf8b4738f0b3e1247c00eb6808bb030 Mon Sep 17 00:00:00 2001 From: Anton Popov Date: Thu, 7 Sep 2023 16:56:13 +0000 Subject: [PATCH 208/433] fix creash in case of exception in prefetched pool --- src/Storages/MergeTree/MergeTreePrefetchedReadPool.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/Storages/MergeTree/MergeTreePrefetchedReadPool.cpp b/src/Storages/MergeTree/MergeTreePrefetchedReadPool.cpp index 7bd038fe5cf..465dcd6a7ab 100644 --- a/src/Storages/MergeTree/MergeTreePrefetchedReadPool.cpp +++ b/src/Storages/MergeTree/MergeTreePrefetchedReadPool.cpp @@ -59,11 +59,11 @@ void MergeTreePrefetchedReadPool::PrefetechedReaders::wait() MergeTreeReadTask::Readers MergeTreePrefetchedReadPool::PrefetechedReaders::get() { + SCOPE_EXIT({ is_valid = false; }); ProfileEventTimeIncrement watch(ProfileEvents::WaitPrefetchTaskMicroseconds); for (auto & prefetch_future : prefetch_futures) prefetch_future.get(); - is_valid = false; return std::move(readers); } From 46e1d05a89fd5b78660765cc8be0f3699fba18ba Mon Sep 17 00:00:00 2001 From: Robert Schulze Date: Thu, 7 Sep 2023 19:18:13 +0000 Subject: [PATCH 209/433] Fix another test --- .../test_backward_compatibility/test_functions.py | 5 +++++ tests/queries/0_stateless/00732_base64_functions.reference | 1 + tests/queries/0_stateless/00732_base64_functions.sql | 3 +++ 3 files changed, 9 insertions(+) diff --git a/tests/integration/test_backward_compatibility/test_functions.py b/tests/integration/test_backward_compatibility/test_functions.py index c86c3ba0ab2..607ac74d725 100644 --- a/tests/integration/test_backward_compatibility/test_functions.py +++ b/tests/integration/test_backward_compatibility/test_functions.py @@ -148,6 +148,11 @@ def test_string_functions(start_cluster): # 22.8 Backward Incompatible Change: Extended range of Date32 "toDate32OrZero", "toDate32OrDefault", + # 23.9 changed the base64-handling library from Turbo base64 to aklomp-base64. They differ in the way they deal with base64 values + # that are not properly padded by '=', for example below test value v='foo'. (Depending on the specification/context, padding is + # mandatory or optional). The former lib produces a value based on implicit padding, the latter lib throws an error. + "FROM_BASE64", + "base64Decode", ] functions = filter(lambda x: x not in excludes, functions) diff --git a/tests/queries/0_stateless/00732_base64_functions.reference b/tests/queries/0_stateless/00732_base64_functions.reference index 19486ff95de..f97c19427e7 100644 --- a/tests/queries/0_stateless/00732_base64_functions.reference +++ b/tests/queries/0_stateless/00732_base64_functions.reference @@ -22,6 +22,7 @@ foobar 1 1 1 1 fooba +~ Zm9v foo foo diff --git a/tests/queries/0_stateless/00732_base64_functions.sql b/tests/queries/0_stateless/00732_base64_functions.sql index 4272cca19ee..99268004003 100644 --- a/tests/queries/0_stateless/00732_base64_functions.sql +++ b/tests/queries/0_stateless/00732_base64_functions.sql @@ -17,6 +17,9 @@ SELECT tryBase64Decode('foo', 'excess argument'); -- { serverError NUMBER_OF_ARG SELECT base64Decode('Zm9vYmF=Zm9v'); -- { serverError INCORRECT_DATA } SELECT tryBase64Decode('Zm9vYmF=Zm9v'); +SELECT base64Decode('foo'); -- { serverError INCORRECT_DATA } +SELECT tryBase64Decode('foo'); + select base64Encode(toFixedString('foo', 3)); select base64Decode(toFixedString('Zm9v', 4)); select tryBase64Decode(toFixedString('Zm9v', 4)); From 59844b0a9d5e00bdd8774fa423ff4cb01a6372ea Mon Sep 17 00:00:00 2001 From: Julia Kartseva Date: Mon, 4 Sep 2023 19:50:46 -0700 Subject: [PATCH 210/433] Add SHOW FUNCTIONS support to client Accepts CH client queries in the form of SHOW FUNCTIONS [LIKE | ILIKE '< pattern>'] Retrieves all columns from `functions` table in system database. Fixes #52757 --- docs/en/sql-reference/statements/show.md | 13 ++++++ src/Interpreters/InterpreterFactory.cpp | 6 +++ .../InterpreterShowFunctionsQuery.cpp | 46 +++++++++++++++++++ .../InterpreterShowFunctionsQuery.h | 27 +++++++++++ src/Parsers/ASTShowFunctionsQuery.cpp | 25 ++++++++++ src/Parsers/ASTShowFunctionsQuery.h | 23 ++++++++++ src/Parsers/ParserQueryWithOutput.cpp | 3 ++ src/Parsers/ParserShowFunctionsQuery.cpp | 35 ++++++++++++++ src/Parsers/ParserShowFunctionsQuery.h | 19 ++++++++ .../0_stateless/00419_show_sql_queries.sh | 1 + .../02875_show_functions.reference | 0 .../0_stateless/02875_show_functions.sh | 14 ++++++ 12 files changed, 212 insertions(+) create mode 100644 src/Interpreters/InterpreterShowFunctionsQuery.cpp create mode 100644 src/Interpreters/InterpreterShowFunctionsQuery.h create mode 100644 src/Parsers/ASTShowFunctionsQuery.cpp create mode 100644 src/Parsers/ASTShowFunctionsQuery.h create mode 100644 src/Parsers/ParserShowFunctionsQuery.cpp create mode 100644 src/Parsers/ParserShowFunctionsQuery.h create mode 100644 tests/queries/0_stateless/02875_show_functions.reference create mode 100755 tests/queries/0_stateless/02875_show_functions.sh diff --git a/docs/en/sql-reference/statements/show.md b/docs/en/sql-reference/statements/show.md index 1c399d2072b..e94718394de 100644 --- a/docs/en/sql-reference/statements/show.md +++ b/docs/en/sql-reference/statements/show.md @@ -638,3 +638,16 @@ Outputs the content of the [system.table_engines](../../operations/system-tables **See Also** - [system.table_engines](../../operations/system-tables/table_engines.md) table + +## SHOW FUNCTIONS + +``` sql +SHOW FUNCTIONS [LIKE | ILIKE ''] +``` + +Outputs the content of the [system.functions](../../operations/system-tables/functions.md) table. + +If either `LIKE` or `ILIKE` clause is specified, the query returns a list of system functions whose names match the provided ``. + +**See Also** +- [system.functions](../../operations/system-tables/functions.md) table diff --git a/src/Interpreters/InterpreterFactory.cpp b/src/Interpreters/InterpreterFactory.cpp index 27c94119750..de3a3d68d39 100644 --- a/src/Interpreters/InterpreterFactory.cpp +++ b/src/Interpreters/InterpreterFactory.cpp @@ -19,6 +19,7 @@ #include #include #include +#include #include #include #include @@ -80,6 +81,7 @@ #include #include #include +#include #include #include #include @@ -203,6 +205,10 @@ std::unique_ptr InterpreterFactory::get(ASTPtr & query, ContextMut { return std::make_unique(query, context); } + else if (query->as()) + { + return std::make_unique(query, context); + } else if (query->as()) { return std::make_unique(query, context); diff --git a/src/Interpreters/InterpreterShowFunctionsQuery.cpp b/src/Interpreters/InterpreterShowFunctionsQuery.cpp new file mode 100644 index 00000000000..efadb929451 --- /dev/null +++ b/src/Interpreters/InterpreterShowFunctionsQuery.cpp @@ -0,0 +1,46 @@ +#include + +#include +#include +#include +#include + +namespace DB +{ + +InterpreterShowFunctionsQuery::InterpreterShowFunctionsQuery(const ASTPtr & query_ptr_, ContextMutablePtr context_) + : WithMutableContext(context_), query_ptr(query_ptr_) +{ +} + +BlockIO InterpreterShowFunctionsQuery::execute() +{ + return executeQuery(getRewrittenQuery(), getContext(), true); +} + +String InterpreterShowFunctionsQuery::getRewrittenQuery() +{ + constexpr const char * functions_table = "functions"; + + const auto & query = query_ptr->as(); + + DatabasePtr systemDb = DatabaseCatalog::instance().getSystemDatabase(); + + String rewritten_query = fmt::format( + R"( +SELECT * +FROM {}.{})", + systemDb->getDatabaseName(), + functions_table); + + if (!query.like.empty()) + { + rewritten_query += " WHERE name "; + rewritten_query += query.case_insensitive_like ? "ILIKE " : "LIKE "; + rewritten_query += fmt::format("'{}'", query.like); + } + + return rewritten_query; +} + +} diff --git a/src/Interpreters/InterpreterShowFunctionsQuery.h b/src/Interpreters/InterpreterShowFunctionsQuery.h new file mode 100644 index 00000000000..f15e1ae67b2 --- /dev/null +++ b/src/Interpreters/InterpreterShowFunctionsQuery.h @@ -0,0 +1,27 @@ +#pragma once + +#include +#include + +namespace DB +{ + +class Context; + +class InterpreterShowFunctionsQuery : public IInterpreter, WithMutableContext +{ +public: + InterpreterShowFunctionsQuery(const ASTPtr & query_ptr_, ContextMutablePtr context_); + + BlockIO execute() override; + + bool ignoreQuota() const override { return true; } + bool ignoreLimits() const override { return true; } + +private: + ASTPtr query_ptr; + + String getRewrittenQuery(); +}; + +} diff --git a/src/Parsers/ASTShowFunctionsQuery.cpp b/src/Parsers/ASTShowFunctionsQuery.cpp new file mode 100644 index 00000000000..9253dcf5cb2 --- /dev/null +++ b/src/Parsers/ASTShowFunctionsQuery.cpp @@ -0,0 +1,25 @@ +#include + +#include + +namespace DB +{ + +ASTPtr ASTShowFunctionsQuery::clone() const +{ + auto res = std::make_shared(*this); + res->children.clear(); + cloneOutputOptions(*res); + return res; +} + +void ASTShowFunctionsQuery::formatQueryImpl(const FormatSettings & settings, FormatState &, FormatStateStacked) const +{ + settings.ostr << (settings.hilite ? hilite_keyword : "") << "SHOW FUNCTIONS" << (settings.hilite ? hilite_none : ""); + + if (!like.empty()) + settings.ostr << (settings.hilite ? hilite_keyword : "") << (case_insensitive_like ? " ILIKE " : " LIKE ") + << (settings.hilite ? hilite_none : "") << DB::quote << like; +} + +} diff --git a/src/Parsers/ASTShowFunctionsQuery.h b/src/Parsers/ASTShowFunctionsQuery.h new file mode 100644 index 00000000000..6993f939888 --- /dev/null +++ b/src/Parsers/ASTShowFunctionsQuery.h @@ -0,0 +1,23 @@ +#pragma once + +#include +#include + +namespace DB +{ + +class ASTShowFunctionsQuery : public ASTQueryWithOutput +{ +public: + bool case_insensitive_like = false; + String like; + + String getID(char) const override { return "ShowFunctions"; } + ASTPtr clone() const override; + QueryKind getQueryKind() const override { return QueryKind::Show; } + +protected: + void formatQueryImpl(const FormatSettings & settings, FormatState &, FormatStateStacked) const override; +}; + +} diff --git a/src/Parsers/ParserQueryWithOutput.cpp b/src/Parsers/ParserQueryWithOutput.cpp index a2391495071..9a71bc222b5 100644 --- a/src/Parsers/ParserQueryWithOutput.cpp +++ b/src/Parsers/ParserQueryWithOutput.cpp @@ -18,6 +18,7 @@ #include #include #include +#include #include #include #include @@ -40,6 +41,7 @@ bool ParserQueryWithOutput::parseImpl(Pos & pos, ASTPtr & node, Expected & expec ParserShowTablesQuery show_tables_p; ParserShowColumnsQuery show_columns_p; ParserShowEnginesQuery show_engine_p; + ParserShowFunctionsQuery show_functions_p; ParserShowIndexesQuery show_indexes_p; ParserSelectWithUnionQuery select_p; ParserTablePropertiesQuery table_p; @@ -71,6 +73,7 @@ bool ParserQueryWithOutput::parseImpl(Pos & pos, ASTPtr & node, Expected & expec || show_tables_p.parse(pos, query, expected) || show_columns_p.parse(pos, query, expected) || show_engine_p.parse(pos, query, expected) + || show_functions_p.parse(pos, query, expected) || show_indexes_p.parse(pos, query, expected) || table_p.parse(pos, query, expected) || describe_cache_p.parse(pos, query, expected) diff --git a/src/Parsers/ParserShowFunctionsQuery.cpp b/src/Parsers/ParserShowFunctionsQuery.cpp new file mode 100644 index 00000000000..524d936c4f1 --- /dev/null +++ b/src/Parsers/ParserShowFunctionsQuery.cpp @@ -0,0 +1,35 @@ +#include + +#include +#include +#include +#include + +namespace DB +{ + +bool ParserShowFunctionsQuery::parseImpl(Pos & pos, ASTPtr & node, Expected & expected) +{ + ASTPtr like; + + auto query = std::make_shared(); + if (!ParserKeyword("SHOW FUNCTIONS").ignore(pos, expected)) + return false; + + if (bool insensitive = ParserKeyword("ILIKE").ignore(pos, expected); insensitive || ParserKeyword("LIKE").ignore(pos, expected)) + { + if (insensitive) + query->case_insensitive_like = true; + + if (!ParserStringLiteral().parse(pos, like, expected)) + return false; + } + + if (like) + query->like = like->as().value.safeGet(); + node = query; + + return true; +} + +} diff --git a/src/Parsers/ParserShowFunctionsQuery.h b/src/Parsers/ParserShowFunctionsQuery.h new file mode 100644 index 00000000000..25241d180db --- /dev/null +++ b/src/Parsers/ParserShowFunctionsQuery.h @@ -0,0 +1,19 @@ +#pragma once + +#include + +namespace DB +{ + +/** Parses queries of the form + * SHOW FUNCTIONS [LIKE | ILIKE ''] + */ +class ParserShowFunctionsQuery : public IParserBase +{ +protected: + const char * getName() const override { return "SHOW FUNCTIONS query"; } + + bool parseImpl(Pos & pos, ASTPtr & node, Expected & expected) override; +}; + +} diff --git a/tests/queries/0_stateless/00419_show_sql_queries.sh b/tests/queries/0_stateless/00419_show_sql_queries.sh index 607703b385a..99252eeb1ba 100755 --- a/tests/queries/0_stateless/00419_show_sql_queries.sh +++ b/tests/queries/0_stateless/00419_show_sql_queries.sh @@ -8,3 +8,4 @@ $CLICKHOUSE_CLIENT -q "SHOW PROCESSLIST" &>/dev/null $CLICKHOUSE_CLIENT -q "SHOW DATABASES" &>/dev/null $CLICKHOUSE_CLIENT -q "SHOW TABLES" &>/dev/null $CLICKHOUSE_CLIENT -q "SHOW ENGINES" &>/dev/null +$CLICKHOUSE_CLIENT -q "SHOW FUNCTIONS" &>/dev/null diff --git a/tests/queries/0_stateless/02875_show_functions.reference b/tests/queries/0_stateless/02875_show_functions.reference new file mode 100644 index 00000000000..e69de29bb2d diff --git a/tests/queries/0_stateless/02875_show_functions.sh b/tests/queries/0_stateless/02875_show_functions.sh new file mode 100755 index 00000000000..6f8da63ca9e --- /dev/null +++ b/tests/queries/0_stateless/02875_show_functions.sh @@ -0,0 +1,14 @@ +#!/usr/bin/env bash + +CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) +# shellcheck source=../shell_config.sh +. "$CURDIR"/../shell_config.sh + +diff -q <($CLICKHOUSE_CLIENT -q "SELECT * from system.functions") \ + <($CLICKHOUSE_CLIENT -q "SHOW FUNCTIONS") + +diff -q <($CLICKHOUSE_CLIENT -q "SELECT * FROM system.functions WHERE name ILIKE 'quantile%'") \ + <($CLICKHOUSE_CLIENT -q "SHOW FUNCTIONS ILIKE 'quantile%'") + +diff -q <($CLICKHOUSE_CLIENT -q "SELECT * FROM system.functions WHERE name LIKE 'median%'") \ + <($CLICKHOUSE_CLIENT -q "SHOW FUNCTIONS LIKE 'median%'") From 411a97ff36f9855e4c3f3e85403a9a140ee2ca33 Mon Sep 17 00:00:00 2001 From: Sema Checherinda Date: Thu, 7 Sep 2023 21:15:30 +0000 Subject: [PATCH 211/433] reproduce a bug in removeSharedRecursive --- .../__init__.py | 0 .../configs/config.d/storage_conf.xml | 43 +++ .../configs/config.d/users.xml | 7 + .../test.py | 334 ++++++++++++++++++ 4 files changed, 384 insertions(+) create mode 100644 tests/integration/test_replicated_zero_copy_projection_mutation/__init__.py create mode 100644 tests/integration/test_replicated_zero_copy_projection_mutation/configs/config.d/storage_conf.xml create mode 100644 tests/integration/test_replicated_zero_copy_projection_mutation/configs/config.d/users.xml create mode 100644 tests/integration/test_replicated_zero_copy_projection_mutation/test.py diff --git a/tests/integration/test_replicated_zero_copy_projection_mutation/__init__.py b/tests/integration/test_replicated_zero_copy_projection_mutation/__init__.py new file mode 100644 index 00000000000..e69de29bb2d diff --git a/tests/integration/test_replicated_zero_copy_projection_mutation/configs/config.d/storage_conf.xml b/tests/integration/test_replicated_zero_copy_projection_mutation/configs/config.d/storage_conf.xml new file mode 100644 index 00000000000..08c5ccdde88 --- /dev/null +++ b/tests/integration/test_replicated_zero_copy_projection_mutation/configs/config.d/storage_conf.xml @@ -0,0 +1,43 @@ + + + + + + s3 + http://minio1:9001/root/data/ + minio + minio123 + true + + + s3 + http://resolver:8081/root/data/ + minio + minio123 + true + 0 + + + + + +
+ s3 +
+ + + + +
+ broken_s3 +
+ + + + + + + 1 + + + diff --git a/tests/integration/test_replicated_zero_copy_projection_mutation/configs/config.d/users.xml b/tests/integration/test_replicated_zero_copy_projection_mutation/configs/config.d/users.xml new file mode 100644 index 00000000000..246de9ecb96 --- /dev/null +++ b/tests/integration/test_replicated_zero_copy_projection_mutation/configs/config.d/users.xml @@ -0,0 +1,7 @@ + + + + 1 + + + diff --git a/tests/integration/test_replicated_zero_copy_projection_mutation/test.py b/tests/integration/test_replicated_zero_copy_projection_mutation/test.py new file mode 100644 index 00000000000..10c1ce3f89a --- /dev/null +++ b/tests/integration/test_replicated_zero_copy_projection_mutation/test.py @@ -0,0 +1,334 @@ +import logging +import time +from contextlib import contextmanager +import pathlib + +import pytest + +from helpers.mock_servers import start_s3_mock +from helpers.cluster import ClickHouseCluster +from helpers.test_tools import assert_eq_with_retry + + +def args_to_dict(**kwargs): + return kwargs + + +@pytest.fixture(scope="module") +def cluster(): + try: + cluster = ClickHouseCluster(__file__) + + kwargs = args_to_dict( + main_configs=[ + "configs/config.d/storage_conf.xml", + ], + user_configs=[ + "configs/config.d/users.xml", + ], + with_minio=True, + with_zookeeper=True, + stay_alive=True, + ) + + cluster.add_instance("node1", **kwargs) + cluster.add_instance("node2", **kwargs) + + cluster.start() + yield cluster + + finally: + cluster.shutdown() + + +@pytest.fixture(scope="module") +def all_cluster_nodes(cluster): + yield cluster.instances.values() + + +@pytest.fixture(scope="module") +def first_cluster_node(cluster): + yield cluster.instances["node1"] + + +@pytest.fixture(scope="module") +def second_cluster_node(cluster): + yield cluster.instances["node2"] + + +@pytest.fixture(scope="module") +def init_broken_s3(cluster): + yield start_s3_mock(cluster, "broken_s3", "8081") + + +@pytest.fixture(scope="function") +def broken_s3(init_broken_s3): + init_broken_s3.reset() + yield init_broken_s3 + + +def list_objects(cluster, path="data/", hint="list_objects"): + minio = cluster.minio_client + objects = list(minio.list_objects(cluster.minio_bucket, path, recursive=True)) + names = [x.object_name for x in objects] + names.sort() + logging.info(f"{hint} ({len(objects)}): {names}") + return names + + +def wait_for_delete_s3_objects(cluster, expected, timeout=30): + while timeout > 0: + if len(list_objects(cluster, "data/")) == expected: + return + timeout -= 1 + time.sleep(1) + final_listing = list_objects(cluster, "data/") + assert len(final_listing) == expected, ",".join(final_listing) + + +def remove_all_s3_objects(cluster): + minio = cluster.minio_client + for obj in list_objects(cluster, "data/"): + minio.remove_object(cluster.minio_bucket, obj) + + +@pytest.fixture(autouse=True, scope="function") +def clear_minio(cluster): + try: + # CH do some writes to the S3 at start. For example, file data/clickhouse_access_check_{server_uuid}. + # Set the timeout there as 10 sec in order to resolve the race with that file exists. + wait_for_delete_s3_objects(cluster, 0, timeout=10) + except: + # Remove extra objects to prevent tests cascade failing + remove_all_s3_objects(cluster) + + yield + + +@contextmanager +def drop_table_guard(nodes, table): + for node in nodes: + node.query(f"DROP TABLE IF EXISTS {table} SYNC") + try: + yield + finally: + for node in nodes: + node.query(f"DROP TABLE IF EXISTS {table} SYNC") + + +def test_all_projection_files_are_dropped_when_part_is_dropped(cluster, first_cluster_node): + node = first_cluster_node + + with drop_table_guard([node], "test_all_projection_files_are_dropped"): + node.query( + """ + CREATE TABLE test_all_projection_files_are_dropped(a UInt32, b UInt32) + ENGINE MergeTree() + ORDER BY a + SETTINGS storage_policy='s3', old_parts_lifetime=0 + """ + ) + + objects_empty_table = list_objects(cluster) + + node.query( + "ALTER TABLE test_all_projection_files_are_dropped ADD projection b_order (SELECT a, b ORDER BY b)") + node.query( + "ALTER TABLE test_all_projection_files_are_dropped MATERIALIZE projection b_order") + + node.query( + """ + INSERT INTO test_all_projection_files_are_dropped + VALUES (1, 105), (5, 101), (3, 103), (4, 102), (2, 104) + """ + ) + + node.query( + "ALTER TABLE test_all_projection_files_are_dropped DROP PARTITION ID 'all'" + ) + + objects_at_the_end = list_objects(cluster) + assert objects_at_the_end == objects_empty_table + + +def test_hardlinks_preserved_when_projection_dropped(cluster, all_cluster_nodes, first_cluster_node, second_cluster_node): + with drop_table_guard(all_cluster_nodes, "test_hardlinks_preserved_when_projection_dropped"): + create_query = ( + """ + CREATE TABLE test_hardlinks_preserved_when_projection_dropped + ( + a UInt32, + b UInt32, + c UInt32, + PROJECTION projection_order_by_b + ( + SELECT a, b ORDER BY b + ) + ) + ENGINE ReplicatedMergeTree('/clickhouse/tables/test_projection', '{instance}') + ORDER BY a + """ + ) + + first_node_settings = ( + """ + SETTINGS + storage_policy='s3', + old_parts_lifetime=0 + """ + ) + + # big old_parts_lifetime value makes second node to hold outdated part for us, we make it as broken_on_start + second_node_settings = ( + """ + SETTINGS + storage_policy='s3', + old_parts_lifetime=10000 + """ + ) + + first_cluster_node.query(create_query + first_node_settings) + second_cluster_node.query(create_query + second_node_settings) + + first_cluster_node.query("SYSTEM FLUSH LOGS") + table_uuid = first_cluster_node.query( + """ + SELECT uuid FROM system.tables + WHERE name = 'test_hardlinks_preserved_when_projection_dropped' + """ + ).strip() + + first_cluster_node.query( + """ + INSERT INTO test_hardlinks_preserved_when_projection_dropped + VALUES (1, 105, 1), (5, 101, 1), (3, 103, 1), (4, 102, 1), (2, 104, 1) + """ + ) + + # second_cluster_node will fetch the mutated part when it is ready on first_cluster_node + second_cluster_node.query("SYSTEM STOP MERGES") + + first_cluster_node.query( + """ + ALTER TABLE test_hardlinks_preserved_when_projection_dropped + UPDATE c = 2 where c = 1 + """, + settings={"mutations_sync": "1"} + ) + + assert_eq_with_retry(first_cluster_node, "SELECT COUNT() FROM system.replication_queue", "0") + + # the mutated part is ready on first_cluster_node, second replica just fetches it + second_cluster_node.query("SYSTEM START MERGES") + + data = first_cluster_node.query( + """ + SELECT * FROM system.parts + WHERE name = 'all_0_0_0' + AND table = 'test_hardlinks_preserved_when_projection_dropped' + AND not active + FORMAT Vertical + """ + ) + + # fist node removed outdated part + assert_eq_with_retry( + first_cluster_node, + """ + SELECT removal_state FROM system.parts + WHERE name = 'all_0_0_0' AND table = 'test_hardlinks_preserved_when_projection_dropped' AND not active + """, + "", + retry_count=300, + sleep_time=1 + ) + + hardlinks = first_cluster_node.query( + f""" + SELECT value + FROM system.zookeeper + WHERE + path like '/clickhouse/zero_copy/zero_copy_s3/{table_uuid}' AND name = 'all_0_0_0' + """, + settings={"allow_unrestricted_reads_from_keeper": "1"} + ).strip().split() + assert len(hardlinks) > 0, ",".join(hardlinks) + assert any(["proj/" in x for x in hardlinks]), ",".join(hardlinks) + + logging.info("hardlinks has been planted") + + part_path_on_second_node = second_cluster_node.query( + """ + SELECT path FROM system.parts + WHERE + name = 'all_0_0_0' AND table = 'test_hardlinks_preserved_when_projection_dropped' + """ + ).strip() + + # that corrupts outdatated part all_0_0_0 + script = ( + f"INDEX_FILE={part_path_on_second_node}/primary.cidx" + """ + cp $INDEX_FILE $INDEX_FILE.backup + echo "unexpected data in metadata file" | cat > $INDEX_FILE + """ + ) + second_cluster_node.exec_in_container(["bash", "-c", script]) + + # corrupted outdatated part all_0_0_0 is detached as broken_on_start + second_cluster_node.restart_clickhouse() + + second_cluster_node.query("SYSTEM WAIT LOADING PARTS test_hardlinks_preserved_when_projection_dropped") + + second_cluster_node.query("SYSTEM FLUSH LOGS") + + broken_parts = second_cluster_node.query( + """ + SELECT name, reason, path FROM system.detached_parts + WHERE + table = 'test_hardlinks_preserved_when_projection_dropped' + """ + ).strip().split('\n') + + assert len(broken_parts) == 1, broken_parts + broken_part_name, reason, broken_part_path_on_second_node = broken_parts[0].split('\t') + assert "broken-on-start" == reason + + script = ( + f"INDEX_FILE={broken_part_path_on_second_node}/primary.cidx" + """ + mv $INDEX_FILE.backup $INDEX_FILE + """ + ) + second_cluster_node.exec_in_container(["bash", "-c", script]) + + second_cluster_node.query( + f""" + ALTER TABLE test_hardlinks_preserved_when_projection_dropped + DROP DETACHED PART '{broken_part_name}' + """, + settings={"allow_drop_detached": "1"} + ) + + res = first_cluster_node.query( + """ + CHECK TABLE test_hardlinks_preserved_when_projection_dropped + """ + ).strip() + + # corrupted + assert res == "0" + + data = first_cluster_node.query( + """ + SELECT a, b FROM test_hardlinks_preserved_when_projection_dropped + WHERE b > 104 + ORDER BY b + """ + ).split("\n") + + assert len(data) == 5, data + + + + From 44e432deabdbb1eda096004bb71ea93750f62599 Mon Sep 17 00:00:00 2001 From: Sema Checherinda Date: Thu, 7 Sep 2023 23:47:35 +0200 Subject: [PATCH 212/433] style fix --- .../configs/config.d/storage_conf.xml | 15 --- .../test.py | 114 +++++++++--------- 2 files changed, 59 insertions(+), 70 deletions(-) diff --git a/tests/integration/test_replicated_zero_copy_projection_mutation/configs/config.d/storage_conf.xml b/tests/integration/test_replicated_zero_copy_projection_mutation/configs/config.d/storage_conf.xml index 08c5ccdde88..44d043b944f 100644 --- a/tests/integration/test_replicated_zero_copy_projection_mutation/configs/config.d/storage_conf.xml +++ b/tests/integration/test_replicated_zero_copy_projection_mutation/configs/config.d/storage_conf.xml @@ -9,14 +9,6 @@ minio123 true - - s3 - http://resolver:8081/root/data/ - minio - minio123 - true - 0 - @@ -26,13 +18,6 @@ - - -
- broken_s3 -
- - diff --git a/tests/integration/test_replicated_zero_copy_projection_mutation/test.py b/tests/integration/test_replicated_zero_copy_projection_mutation/test.py index 10c1ce3f89a..3f594158a46 100644 --- a/tests/integration/test_replicated_zero_copy_projection_mutation/test.py +++ b/tests/integration/test_replicated_zero_copy_projection_mutation/test.py @@ -116,7 +116,9 @@ def drop_table_guard(nodes, table): node.query(f"DROP TABLE IF EXISTS {table} SYNC") -def test_all_projection_files_are_dropped_when_part_is_dropped(cluster, first_cluster_node): +def test_all_projection_files_are_dropped_when_part_is_dropped( + cluster, first_cluster_node +): node = first_cluster_node with drop_table_guard([node], "test_all_projection_files_are_dropped"): @@ -132,9 +134,11 @@ def test_all_projection_files_are_dropped_when_part_is_dropped(cluster, first_cl objects_empty_table = list_objects(cluster) node.query( - "ALTER TABLE test_all_projection_files_are_dropped ADD projection b_order (SELECT a, b ORDER BY b)") + "ALTER TABLE test_all_projection_files_are_dropped ADD projection b_order (SELECT a, b ORDER BY b)" + ) node.query( - "ALTER TABLE test_all_projection_files_are_dropped MATERIALIZE projection b_order") + "ALTER TABLE test_all_projection_files_are_dropped MATERIALIZE projection b_order" + ) node.query( """ @@ -151,10 +155,13 @@ def test_all_projection_files_are_dropped_when_part_is_dropped(cluster, first_cl assert objects_at_the_end == objects_empty_table -def test_hardlinks_preserved_when_projection_dropped(cluster, all_cluster_nodes, first_cluster_node, second_cluster_node): - with drop_table_guard(all_cluster_nodes, "test_hardlinks_preserved_when_projection_dropped"): - create_query = ( - """ +def test_hardlinks_preserved_when_projection_dropped( + cluster, all_cluster_nodes, first_cluster_node, second_cluster_node +): + with drop_table_guard( + all_cluster_nodes, "test_hardlinks_preserved_when_projection_dropped" + ): + create_query = """ CREATE TABLE test_hardlinks_preserved_when_projection_dropped ( a UInt32, @@ -167,25 +174,20 @@ def test_hardlinks_preserved_when_projection_dropped(cluster, all_cluster_nodes, ) ENGINE ReplicatedMergeTree('/clickhouse/tables/test_projection', '{instance}') ORDER BY a - """ - ) + """ - first_node_settings = ( - """ + first_node_settings = """ SETTINGS storage_policy='s3', old_parts_lifetime=0 - """ - ) + """ # big old_parts_lifetime value makes second node to hold outdated part for us, we make it as broken_on_start - second_node_settings = ( - """ + second_node_settings = """ SETTINGS storage_policy='s3', old_parts_lifetime=10000 - """ - ) + """ first_cluster_node.query(create_query + first_node_settings) second_cluster_node.query(create_query + second_node_settings) @@ -213,45 +215,43 @@ def test_hardlinks_preserved_when_projection_dropped(cluster, all_cluster_nodes, ALTER TABLE test_hardlinks_preserved_when_projection_dropped UPDATE c = 2 where c = 1 """, - settings={"mutations_sync": "1"} + settings={"mutations_sync": "1"}, ) - assert_eq_with_retry(first_cluster_node, "SELECT COUNT() FROM system.replication_queue", "0") + assert_eq_with_retry( + first_cluster_node, "SELECT COUNT() FROM system.replication_queue", "0" + ) # the mutated part is ready on first_cluster_node, second replica just fetches it second_cluster_node.query("SYSTEM START MERGES") - data = first_cluster_node.query( - """ - SELECT * FROM system.parts - WHERE name = 'all_0_0_0' - AND table = 'test_hardlinks_preserved_when_projection_dropped' - AND not active - FORMAT Vertical - """ - ) - # fist node removed outdated part assert_eq_with_retry( first_cluster_node, """ SELECT removal_state FROM system.parts - WHERE name = 'all_0_0_0' AND table = 'test_hardlinks_preserved_when_projection_dropped' AND not active + WHERE name = 'all_0_0_0' + AND table = 'test_hardlinks_preserved_when_projection_dropped' + AND not active """, "", retry_count=300, - sleep_time=1 + sleep_time=1, ) - hardlinks = first_cluster_node.query( - f""" - SELECT value - FROM system.zookeeper - WHERE - path like '/clickhouse/zero_copy/zero_copy_s3/{table_uuid}' AND name = 'all_0_0_0' - """, - settings={"allow_unrestricted_reads_from_keeper": "1"} - ).strip().split() + hardlinks = ( + first_cluster_node.query( + f""" + SELECT value + FROM system.zookeeper + WHERE + path like '/clickhouse/zero_copy/zero_copy_s3/{table_uuid}' AND name = 'all_0_0_0' + """, + settings={"allow_unrestricted_reads_from_keeper": "1"}, + ) + .strip() + .split() + ) assert len(hardlinks) > 0, ",".join(hardlinks) assert any(["proj/" in x for x in hardlinks]), ",".join(hardlinks) @@ -278,20 +278,28 @@ def test_hardlinks_preserved_when_projection_dropped(cluster, all_cluster_nodes, # corrupted outdatated part all_0_0_0 is detached as broken_on_start second_cluster_node.restart_clickhouse() - second_cluster_node.query("SYSTEM WAIT LOADING PARTS test_hardlinks_preserved_when_projection_dropped") + second_cluster_node.query( + "SYSTEM WAIT LOADING PARTS test_hardlinks_preserved_when_projection_dropped" + ) second_cluster_node.query("SYSTEM FLUSH LOGS") - broken_parts = second_cluster_node.query( - """ - SELECT name, reason, path FROM system.detached_parts - WHERE - table = 'test_hardlinks_preserved_when_projection_dropped' - """ - ).strip().split('\n') - + broken_parts = ( + second_cluster_node.query( + """ + SELECT name, reason, path FROM system.detached_parts + WHERE + table = 'test_hardlinks_preserved_when_projection_dropped' + """ + ) + .strip() + .split("\n") + ) assert len(broken_parts) == 1, broken_parts - broken_part_name, reason, broken_part_path_on_second_node = broken_parts[0].split('\t') + # style checker black asked to do this. It is crazy + broken_part_name, reason, broken_part_path_on_second_node = broken_parts[ + 0 + ].split("\t") assert "broken-on-start" == reason script = ( @@ -307,7 +315,7 @@ def test_hardlinks_preserved_when_projection_dropped(cluster, all_cluster_nodes, ALTER TABLE test_hardlinks_preserved_when_projection_dropped DROP DETACHED PART '{broken_part_name}' """, - settings={"allow_drop_detached": "1"} + settings={"allow_drop_detached": "1"}, ) res = first_cluster_node.query( @@ -328,7 +336,3 @@ def test_hardlinks_preserved_when_projection_dropped(cluster, all_cluster_nodes, ).split("\n") assert len(data) == 5, data - - - - From 0b396ee7b4ac9c2df1fea61a471a3f3d71b57073 Mon Sep 17 00:00:00 2001 From: Leonardo Maciel Date: Thu, 7 Sep 2023 22:27:22 -0300 Subject: [PATCH 213/433] Fixed detectLanguage mixed example --- docs/en/sql-reference/functions/nlp-functions.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/docs/en/sql-reference/functions/nlp-functions.md b/docs/en/sql-reference/functions/nlp-functions.md index bb127a939f3..3e0458d226d 100644 --- a/docs/en/sql-reference/functions/nlp-functions.md +++ b/docs/en/sql-reference/functions/nlp-functions.md @@ -204,7 +204,7 @@ Other possible results: Query: ```sql -SELECT detectLanguageMixed('Je pense que je ne parviendrai jamais à parler français comme un natif. Where there’s a will, there’s a way.'); +SELECT detectLanguage('Je pense que je ne parviendrai jamais à parler français comme un natif. Where there’s a will, there’s a way.'); ``` Result: From 926310728f109a474c17751b5ca0a88c1e25a1ed Mon Sep 17 00:00:00 2001 From: pufit Date: Thu, 7 Sep 2023 22:48:04 -0400 Subject: [PATCH 214/433] Fix issues --- programs/keeper-client/Commands.cpp | 22 ++++++++++++++-------- programs/keeper-client/Commands.h | 2 +- 2 files changed, 15 insertions(+), 9 deletions(-) diff --git a/programs/keeper-client/Commands.cpp b/programs/keeper-client/Commands.cpp index 65bc719f1be..757f90fadcf 100644 --- a/programs/keeper-client/Commands.cpp +++ b/programs/keeper-client/Commands.cpp @@ -402,14 +402,20 @@ void ReconfigCommand::execute(const DB::ASTKeeperQuery * query, DB::KeeperClient String new_members; auto operation = query->args[0].get(); - if (operation == static_cast(ReconfigCommand::Operation::ADD)) - joining = query->args[1].safeGet(); - else if (operation == static_cast(ReconfigCommand::Operation::REMOVE)) - leaving = query->args[1].safeGet(); - else if (operation == static_cast(ReconfigCommand::Operation::SET)) - new_members = query->args[1].safeGet(); - else - UNREACHABLE(); + switch (operation) + { + case static_cast(ReconfigCommand::Operation::ADD): + joining = query->args[1].safeGet(); + break; + case static_cast(ReconfigCommand::Operation::REMOVE): + leaving = query->args[1].safeGet(); + break; + case static_cast(ReconfigCommand::Operation::SET): + new_members = query->args[1].safeGet(); + break; + default: + UNREACHABLE(); + } auto response = client->zookeeper->reconfig(joining, leaving, new_members); std::cout << response.value << '\n'; diff --git a/programs/keeper-client/Commands.h b/programs/keeper-client/Commands.h index aed98bfa0fb..f9d4292584b 100644 --- a/programs/keeper-client/Commands.h +++ b/programs/keeper-client/Commands.h @@ -190,7 +190,7 @@ class RMRCommand : public IKeeperClientCommand class ReconfigCommand : public IKeeperClientCommand { - enum class Operation : Int64 + enum class Operation : UInt8 { ADD = 0, REMOVE = 1, From f40d3a25cff76a22d5095015e6849542e2119aa0 Mon Sep 17 00:00:00 2001 From: Nikolay Degterinsky <43110995+evillique@users.noreply.github.com> Date: Fri, 8 Sep 2023 05:22:21 +0200 Subject: [PATCH 215/433] Fix build --- src/Interpreters/RewriteArrayExistsFunctionVisitor.cpp | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/Interpreters/RewriteArrayExistsFunctionVisitor.cpp b/src/Interpreters/RewriteArrayExistsFunctionVisitor.cpp index adee5679bdf..64e03767c49 100644 --- a/src/Interpreters/RewriteArrayExistsFunctionVisitor.cpp +++ b/src/Interpreters/RewriteArrayExistsFunctionVisitor.cpp @@ -19,7 +19,7 @@ void RewriteArrayExistsFunctionMatcher::visit(ASTPtr & ast, Data & data) { if (join->using_expression_list) { - auto it = std::find(join->children.begin(), join->children.end(), join->using_expression_list); + auto * it = std::find(join->children.begin(), join->children.end(), join->using_expression_list); visit(join->using_expression_list, data); @@ -29,7 +29,7 @@ void RewriteArrayExistsFunctionMatcher::visit(ASTPtr & ast, Data & data) if (join->on_expression) { - auto it = std::find(join->children.begin(), join->children.end(), join->on_expression); + auto * it = std::find(join->children.begin(), join->children.end(), join->on_expression); visit(join->on_expression, data); From 5862c4ec93a9945f4d7a607ba9b917f9e863b11b Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?J=C3=A1nos=20Benjamin=20Antal?= Date: Fri, 8 Sep 2023 08:42:56 +0200 Subject: [PATCH 216/433] Use not the absolute latest docker image tag (#54369) * Add params for select queries in clickhouse_helper.py * Find the latest image tags from master * Get full git history * Do not checkout submodules * Automatic style fix * Fix style errors * Make logging pretty * Revert accidentally committed changes * Eliminate pformat * Update all workflows * Remove flag with default value * Improve exception handling * Utilizing the loop condition in while loop * Make the logic robust over temporary errors * Prettify logs * Automatic style fix * Add accidentally removed line * Revert "Make the logic robust over temporary errors" This reverts commit 91e6b5d7bac04ef0129a76711d00a88ef9bd2416. * Do not fail in case the tag cannot be queried from ClickHouse * Support partial finding of tags --------- Co-authored-by: robot-clickhouse --- .github/workflows/backport_branches.yml | 1 + .github/workflows/docs_check.yml | 1 + .github/workflows/master.yml | 1 + .github/workflows/nightly.yml | 1 + .github/workflows/pull_request.yml | 1 + .github/workflows/release_branches.yml | 1 + tests/ci/clickhouse_helper.py | 20 ++++-- tests/ci/docker_images_check.py | 22 +------ tests/ci/docker_images_helper.py | 30 +++++++++ tests/ci/docker_manifests_merge.py | 84 ++++++++++++++++++++++++- tests/ci/docker_test.py | 3 +- 11 files changed, 137 insertions(+), 28 deletions(-) create mode 100644 tests/ci/docker_images_helper.py diff --git a/.github/workflows/backport_branches.yml b/.github/workflows/backport_branches.yml index d69168b01ee..a9503136b1a 100644 --- a/.github/workflows/backport_branches.yml +++ b/.github/workflows/backport_branches.yml @@ -76,6 +76,7 @@ jobs: uses: ClickHouse/checkout@v1 with: clear-repository: true + fetch-depth: 0 # to find ancestor merge commits necessary for finding proper docker tags - name: Download changed aarch64 images uses: actions/download-artifact@v3 with: diff --git a/.github/workflows/docs_check.yml b/.github/workflows/docs_check.yml index a0d0e49b95b..a6be21fb14a 100644 --- a/.github/workflows/docs_check.yml +++ b/.github/workflows/docs_check.yml @@ -73,6 +73,7 @@ jobs: uses: ClickHouse/checkout@v1 with: clear-repository: true + fetch-depth: 0 # to find ancestor merge commits necessary for finding proper docker tags - name: Download changed aarch64 images uses: actions/download-artifact@v3 with: diff --git a/.github/workflows/master.yml b/.github/workflows/master.yml index 93ef4aa9da3..0a21fb93beb 100644 --- a/.github/workflows/master.yml +++ b/.github/workflows/master.yml @@ -60,6 +60,7 @@ jobs: uses: ClickHouse/checkout@v1 with: clear-repository: true + fetch-depth: 0 # to find ancestor merge commits necessary for finding proper docker tags - name: Download changed aarch64 images uses: actions/download-artifact@v3 with: diff --git a/.github/workflows/nightly.yml b/.github/workflows/nightly.yml index 9de0444bd83..8162dc37223 100644 --- a/.github/workflows/nightly.yml +++ b/.github/workflows/nightly.yml @@ -53,6 +53,7 @@ jobs: uses: ClickHouse/checkout@v1 with: clear-repository: true + fetch-depth: 0 # to find ancestor merge commits necessary for finding proper docker tags - name: Download changed aarch64 images uses: actions/download-artifact@v3 with: diff --git a/.github/workflows/pull_request.yml b/.github/workflows/pull_request.yml index ce135846dd5..3845ebdcac7 100644 --- a/.github/workflows/pull_request.yml +++ b/.github/workflows/pull_request.yml @@ -94,6 +94,7 @@ jobs: uses: ClickHouse/checkout@v1 with: clear-repository: true + fetch-depth: 0 # to find ancestor merge commits necessary for finding proper docker tags - name: Download changed aarch64 images uses: actions/download-artifact@v3 with: diff --git a/.github/workflows/release_branches.yml b/.github/workflows/release_branches.yml index 21284815583..6d999d1bee7 100644 --- a/.github/workflows/release_branches.yml +++ b/.github/workflows/release_branches.yml @@ -52,6 +52,7 @@ jobs: uses: ClickHouse/checkout@v1 with: clear-repository: true + fetch-depth: 0 # to find ancestor merge commits necessary for finding proper docker tags - name: Download changed aarch64 images uses: actions/download-artifact@v3 with: diff --git a/tests/ci/clickhouse_helper.py b/tests/ci/clickhouse_helper.py index c2304ca7ad3..0d05b811a4f 100644 --- a/tests/ci/clickhouse_helper.py +++ b/tests/ci/clickhouse_helper.py @@ -13,6 +13,10 @@ from pr_info import PRInfo from report import TestResults +class CHException(Exception): + pass + + class InsertException(Exception): pass @@ -131,12 +135,16 @@ class ClickHouseHelper: if not safe: raise - def _select_and_get_json_each_row(self, db, query): + def _select_and_get_json_each_row(self, db, query, query_params): params = { "database": db, "query": query, "default_format": "JSONEachRow", } + if query_params is not None: + for name, value in query_params.items(): + params[f"param_{name}"] = str(value) + for i in range(5): response = None try: @@ -144,15 +152,15 @@ class ClickHouseHelper: response.raise_for_status() return response.text except Exception as ex: - logging.warning("Cannot insert with exception %s", str(ex)) + logging.warning("Select query failed with exception %s", str(ex)) if response: - logging.warning("Reponse text %s", response.text) + logging.warning("Response text %s", response.text) time.sleep(0.1 * i) - raise Exception("Cannot fetch data from clickhouse") + raise CHException("Cannot fetch data from clickhouse") - def select_json_each_row(self, db, query): - text = self._select_and_get_json_each_row(db, query) + def select_json_each_row(self, db, query, query_params=None): + text = self._select_and_get_json_each_row(db, query, query_params) result = [] for line in text.split("\n"): if line: diff --git a/tests/ci/docker_images_check.py b/tests/ci/docker_images_check.py index 16a58a90dcf..ab20e7fb9cf 100644 --- a/tests/ci/docker_images_check.py +++ b/tests/ci/docker_images_check.py @@ -9,7 +9,7 @@ import subprocess import time import sys from pathlib import Path -from typing import Any, Dict, List, Optional, Set, Tuple, Union +from typing import Any, List, Optional, Set, Tuple, Union from github import Github @@ -23,13 +23,12 @@ from s3_helper import S3Helper from stopwatch import Stopwatch from tee_popen import TeePopen from upload_result_helper import upload_results +from docker_images_helper import ImagesDict, IMAGES_FILE_PATH, get_images_dict NAME = "Push to Dockerhub" TEMP_PATH = os.path.join(RUNNER_TEMP, "docker_images_check") -ImagesDict = Dict[str, dict] - class DockerImage: def __init__( @@ -78,21 +77,6 @@ class DockerImage: return f"DockerImage(path={self.path},repo={self.repo},parent={self.parent})" -def get_images_dict(repo_path: str, image_file_path: str) -> ImagesDict: - """Return images suppose to build on the current architecture host""" - images_dict = {} - path_to_images_file = os.path.join(repo_path, image_file_path) - if os.path.exists(path_to_images_file): - with open(path_to_images_file, "rb") as dict_file: - images_dict = json.load(dict_file) - else: - logging.info( - "Image file %s doesn't exist in repo %s", image_file_path, repo_path - ) - - return images_dict - - def get_changed_docker_images( pr_info: PRInfo, images_dict: ImagesDict ) -> Set[DockerImage]: @@ -410,7 +394,7 @@ def main(): shutil.rmtree(TEMP_PATH) os.makedirs(TEMP_PATH) - images_dict = get_images_dict(GITHUB_WORKSPACE, "docker/images.json") + images_dict = get_images_dict(GITHUB_WORKSPACE, IMAGES_FILE_PATH) pr_info = PRInfo() if args.all: diff --git a/tests/ci/docker_images_helper.py b/tests/ci/docker_images_helper.py new file mode 100644 index 00000000000..6aed6e82ce3 --- /dev/null +++ b/tests/ci/docker_images_helper.py @@ -0,0 +1,30 @@ +#!/usr/bin/env python3 + +import json +import logging +import os +from typing import Dict, List + +IMAGES_FILE_PATH = "docker/images.json" + +ImagesDict = Dict[str, dict] + + +def get_images_dict(repo_path: str, images_file_path: str) -> ImagesDict: + """Return images suppose to build on the current architecture host""" + images_dict = {} + path_to_images_file = os.path.join(repo_path, images_file_path) + if os.path.exists(path_to_images_file): + with open(path_to_images_file, "rb") as dict_file: + images_dict = json.load(dict_file) + else: + logging.info( + "Image file %s doesn't exist in repo %s", images_file_path, repo_path + ) + + return images_dict + + +def get_image_names(repo_path: str, images_file_path: str) -> List[str]: + images_dict = get_images_dict(repo_path, images_file_path) + return [info["name"] for (_, info) in images_dict.items()] diff --git a/tests/ci/docker_manifests_merge.py b/tests/ci/docker_manifests_merge.py index d89708b9277..56923c5e2d9 100644 --- a/tests/ci/docker_manifests_merge.py +++ b/tests/ci/docker_manifests_merge.py @@ -9,10 +9,16 @@ import subprocess from typing import List, Dict, Tuple from github import Github -from clickhouse_helper import ClickHouseHelper, prepare_tests_results_for_clickhouse +from clickhouse_helper import ( + ClickHouseHelper, + prepare_tests_results_for_clickhouse, + CHException, +) from commit_status_helper import format_description, get_commit, post_commit_status -from env_helper import RUNNER_TEMP +from docker_images_helper import IMAGES_FILE_PATH, get_image_names +from env_helper import RUNNER_TEMP, GITHUB_WORKSPACE from get_robot_token import get_best_robot_token, get_parameter_from_ssm +from git_helper import Runner from pr_info import PRInfo from report import TestResults, TestResult from s3_helper import S3Helper @@ -167,6 +173,74 @@ def create_manifest(image: str, tags: List[str], push: bool) -> Tuple[str, str]: return manifest, "OK" +def enrich_images(changed_images: Dict[str, str]) -> None: + all_image_names = get_image_names(GITHUB_WORKSPACE, IMAGES_FILE_PATH) + + images_to_find_tags_for = [ + image for image in all_image_names if image not in changed_images + ] + images_to_find_tags_for.sort() + + logging.info( + "Trying to find versions for images:\n %s", "\n ".join(images_to_find_tags_for) + ) + + COMMIT_SHA_BATCH_SIZE = 100 + MAX_COMMIT_BATCHES_TO_CHECK = 10 + # Gets the sha of the last COMMIT_SHA_BATCH_SIZE commits after skipping some commits (see below) + LAST_N_ANCESTOR_SHA_COMMAND = f"git log --format=format:'%H' --max-count={COMMIT_SHA_BATCH_SIZE} --skip={{}} --merges" + git_runner = Runner() + + GET_COMMIT_SHAS_QUERY = """ + WITH {commit_shas:Array(String)} AS commit_shas, + {images:Array(String)} AS images + SELECT + substring(test_name, 1, position(test_name, ':') -1) AS image_name, + argMax(commit_sha, check_start_time) AS commit_sha + FROM checks + WHERE + check_name == 'Push multi-arch images to Dockerhub' + AND position(test_name, checks.commit_sha) + AND checks.commit_sha IN commit_shas + AND image_name IN images + GROUP BY image_name + """ + + batch_count = 0 + ch_helper = ClickHouseHelper() + + while ( + batch_count <= MAX_COMMIT_BATCHES_TO_CHECK and len(images_to_find_tags_for) != 0 + ): + commit_shas = git_runner( + LAST_N_ANCESTOR_SHA_COMMAND.format(batch_count * COMMIT_SHA_BATCH_SIZE) + ).split("\n") + + result = ch_helper.select_json_each_row( + "default", + GET_COMMIT_SHAS_QUERY, + {"commit_shas": commit_shas, "images": images_to_find_tags_for}, + ) + result.sort(key=lambda x: x["image_name"]) + + logging.info( + "Found images for commits %s..%s:\n %s", + commit_shas[0], + commit_shas[-1], + "\n ".join(f"{im['image_name']}:{im['commit_sha']}" for im in result), + ) + + for row in result: + image_name = row["image_name"] + commit_sha = row["commit_sha"] + # As we only get the SHAs of merge commits from master, the PR number will be always 0 + tag = f"0-{commit_sha}" + changed_images[image_name] = tag + images_to_find_tags_for.remove(image_name) + + batch_count += 1 + + def main(): logging.basicConfig(level=logging.INFO) stopwatch = Stopwatch() @@ -198,6 +272,12 @@ def main(): if test_result != "OK": status = "failure" + try: + # changed_images now contains all the images that are changed in this PR. Let's find the latest tag for the images that are not changed. + enrich_images(changed_images) + except CHException as ex: + logging.warning("Couldn't get proper tags for not changed images: %s", ex) + with open( os.path.join(args.path, "changed_images.json"), "w", encoding="utf-8" ) as ci: diff --git a/tests/ci/docker_test.py b/tests/ci/docker_test.py index 14db348ef36..ac43069a679 100644 --- a/tests/ci/docker_test.py +++ b/tests/ci/docker_test.py @@ -9,6 +9,7 @@ from env_helper import GITHUB_RUN_URL from pr_info import PRInfo from report import TestResult import docker_images_check as di +from docker_images_helper import get_images_dict from version_helper import get_version_from_string import docker_server as ds @@ -31,7 +32,7 @@ class TestDockerImageCheck(unittest.TestCase): images = sorted( list( di.get_changed_docker_images( - pr_info, di.get_images_dict("/", self.docker_images_path) + pr_info, get_images_dict("/", self.docker_images_path) ) ) ) From 5c2fd38f7367b8112a4f086fd140e0a7869857d8 Mon Sep 17 00:00:00 2001 From: chen768959 <934103231@qq.com> Date: Fri, 8 Sep 2023 15:05:04 +0800 Subject: [PATCH 217/433] extract checkUInt64ToIn64Conversion and convertUInt64ToInt64IfPossible to separate functions --- src/DataTypes/FieldToDataType.cpp | 94 +++++++++++-------------------- src/DataTypes/FieldToDataType.h | 10 ++++ 2 files changed, 44 insertions(+), 60 deletions(-) diff --git a/src/DataTypes/FieldToDataType.cpp b/src/DataTypes/FieldToDataType.cpp index e25a2f546f2..837aae6753a 100644 --- a/src/DataTypes/FieldToDataType.cpp +++ b/src/DataTypes/FieldToDataType.cpp @@ -136,39 +136,17 @@ DataTypePtr FieldToDataType::operator() (const Array & x) const DataTypes element_types; element_types.reserve(x.size()); - auto checkIfConversionSigned = [&](bool& has_signed_int, bool& has_uint64, bool& uint64_could_opposite, - const DataTypePtr & type, const Field & elem) - { - if (uint64_could_opposite) - { - has_signed_int |= WhichDataType(type).isNativeInt(); - - if (type->getTypeId() == TypeIndex::UInt64) - { - has_uint64 = true; - uint64_could_opposite &= (elem.template get() <= std::numeric_limits::max()); - } - } - }; - bool has_signed_int = false; - bool has_uint64 = false; - bool uint64_could_opposite = true; + bool uint64_convert_possible = true; for (const Field & elem : x) { DataTypePtr type = applyVisitor(*this, elem); element_types.emplace_back(type); - checkIfConversionSigned(has_signed_int, has_uint64, uint64_could_opposite, type, elem); + checkUInt64ToIn64Conversion(has_signed_int, uint64_convert_possible, type, elem); } - // Convert the UInt64 type to Int64 in order to cover other signed_integer types - // and obtain the least super type of all ints. - if (has_signed_int && has_uint64 && uint64_could_opposite) - { - for (auto & type : element_types) - if (type->getTypeId() == TypeIndex::UInt64) - type = std::make_shared(); - } + if (has_signed_int && uint64_convert_possible) + convertUInt64ToInt64IfPossible(element_types); return std::make_shared(getLeastSupertype(element_types)); } @@ -196,53 +174,27 @@ DataTypePtr FieldToDataType::operator() (const Map & map) const key_types.reserve(map.size()); value_types.reserve(map.size()); - auto checkIfConversionSigned = [&](bool& has_signed_int, bool& has_uint64, bool& uint64_could_opposite, - const DataTypePtr & type, const Field & elem) - { - if (uint64_could_opposite) - { - has_signed_int |= WhichDataType(type).isNativeInt(); - - if (type->getTypeId() == TypeIndex::UInt64) - { - has_uint64 = true; - uint64_could_opposite &= (elem.template get() <= std::numeric_limits::max()); - } - } - }; - - auto updateUInt64Types = [](DataTypes types) - { - for (auto& type : types) - if (type->getTypeId() == TypeIndex::UInt64) - type = std::make_shared(); - }; - bool k_has_signed_int = false; - bool k_has_uint64 = false; - bool k_uint64_could_opposite = true; + bool k_uint64_convert_possible = true; bool v_has_signed_int = false; - bool v_has_uint64 = false; - bool v_uint64_could_opposite = true; + bool v_uint64_convert_possible = true; for (const auto & elem : map) { const auto & tuple = elem.safeGet(); assert(tuple.size() == 2); DataTypePtr k_type = applyVisitor(*this, tuple[0]); key_types.push_back(k_type); - checkIfConversionSigned(k_has_signed_int, k_has_uint64, k_uint64_could_opposite, k_type, tuple[0]); + checkUInt64ToIn64Conversion(k_has_signed_int, k_uint64_convert_possible, k_type, tuple[0]); DataTypePtr v_type = applyVisitor(*this, tuple[1]); value_types.push_back(v_type); - checkIfConversionSigned(v_has_signed_int, v_has_uint64, v_uint64_could_opposite, v_type, tuple[1]); + checkUInt64ToIn64Conversion(v_has_signed_int, v_uint64_convert_possible, v_type, tuple[1]); } - // Convert the UInt64 type to Int64 in order to cover other signed_integer types - // and obtain the least super type of all ints. - if (k_has_signed_int && k_has_uint64 && k_uint64_could_opposite) - updateUInt64Types(key_types); + if (k_has_signed_int && k_uint64_convert_possible) + convertUInt64ToInt64IfPossible(key_types); - if (v_has_signed_int && v_has_uint64 && v_uint64_could_opposite) - updateUInt64Types(value_types); + if (v_has_signed_int && v_uint64_convert_possible) + convertUInt64ToInt64IfPossible(value_types); return std::make_shared( getLeastSupertype(key_types), @@ -275,6 +227,28 @@ DataTypePtr FieldToDataType::operator()(const bool &) const return DataTypeFactory::instance().get("Bool"); } +template +void FieldToDataType::checkUInt64ToIn64Conversion(bool & has_signed_int, bool & uint64_convert_possible, const DataTypePtr & type, const Field & elem) const +{ + if (uint64_convert_possible) + { + bool is_native_int = WhichDataType(type).isNativeInt(); + + if (is_native_int) + has_signed_int |= is_native_int; + else if (type->getTypeId() == TypeIndex::UInt64) + uint64_convert_possible &= (elem.template get() <= std::numeric_limits::max()); + } +} + +template +void FieldToDataType::convertUInt64ToInt64IfPossible(DataTypes & data_types) const +{ + for (auto& type : data_types) + if (type->getTypeId() == TypeIndex::UInt64) + type = std::make_shared(); +} + template class FieldToDataType; template class FieldToDataType; template class FieldToDataType; diff --git a/src/DataTypes/FieldToDataType.h b/src/DataTypes/FieldToDataType.h index 8febadc1a0d..d1a3f11e8de 100644 --- a/src/DataTypes/FieldToDataType.h +++ b/src/DataTypes/FieldToDataType.h @@ -45,6 +45,16 @@ public: DataTypePtr operator() (const UInt256 & x) const; DataTypePtr operator() (const Int256 & x) const; DataTypePtr operator() (const bool & x) const; + +private: + // The conditions for converting UInt64 to Int64 are: + // 1. The existence of Int. + // 2. The existence of UInt64, and the UInt64 value must be <= Int64.max. + void checkUInt64ToIn64Conversion(bool& has_signed_int, bool& uint64_convert_possible, const DataTypePtr & type, const Field & elem) const; + + // Convert the UInt64 type to Int64 in order to cover other signed_integer types + // and obtain the least super type of all ints. + void convertUInt64ToInt64IfPossible(DataTypes & data_types) const; }; } From 80aea109b2352f13f84a60336dac02a199a9fcd7 Mon Sep 17 00:00:00 2001 From: Andrey Zvonov <32552679+zvonand@users.noreply.github.com> Date: Fri, 8 Sep 2023 12:49:15 +0300 Subject: [PATCH 218/433] Update aspell-dict.txt --- utils/check-style/aspell-ignore/en/aspell-dict.txt | 1 - 1 file changed, 1 deletion(-) diff --git a/utils/check-style/aspell-ignore/en/aspell-dict.txt b/utils/check-style/aspell-ignore/en/aspell-dict.txt index caf504e232d..465e7344174 100644 --- a/utils/check-style/aspell-ignore/en/aspell-dict.txt +++ b/utils/check-style/aspell-ignore/en/aspell-dict.txt @@ -1392,7 +1392,6 @@ dragonbox dropoff dumpColumnStructure durations -eacces ecto embeddings emptyArray From 5e0caa669e0c8e9dd6494f45289c90f75f4e419a Mon Sep 17 00:00:00 2001 From: vdimir Date: Fri, 8 Sep 2023 09:56:53 +0000 Subject: [PATCH 219/433] Move sucessull checks in commit status under a spoiler --- tests/ci/commit_status_helper.py | 50 ++++++++++++++++++++++++-------- 1 file changed, 38 insertions(+), 12 deletions(-) diff --git a/tests/ci/commit_status_helper.py b/tests/ci/commit_status_helper.py index ac1498ae745..4d6037c76a8 100644 --- a/tests/ci/commit_status_helper.py +++ b/tests/ci/commit_status_helper.py @@ -201,12 +201,10 @@ def generate_status_comment(pr_info: PRInfo, statuses: CommitStatuses) -> str: comment_body = ( f"\n" - f"This is an automated comment for commit {pr_info.sha} with " - f"description of existing statuses. It's updated for the latest CI running\n" - f"The full report is available [here]({report_url})\n" - f"{worst_state}\n\n" - "\n" - "" + f"*This is an automated comment for commit {pr_info.sha} with " + f"description of existing statuses. It's updated for the latest CI running*\n" + f"📄 [Click here]({report_url}) to open a full report in a separate page\n" + f"{worst_state}\n\n" ) # group checks by the name to get the worst one per each grouped_statuses = {} # type: Dict[CheckDescription, CommitStatuses] @@ -230,17 +228,45 @@ def generate_status_comment(pr_info: PRInfo, statuses: CommitStatuses) -> str: else: grouped_statuses[cd] = [status] - table_rows = [] # type: List[str] + table_header = ( + "
Check nameDescriptionStatus
\n" + "\n" + "\n" + ) + table_footer = "\n
Check nameDescriptionStatus
\n" + + details_header = "
Successfull checks\n" + details_footer = "
\n" + + visible_table_rows = [] # type: List[str] + hidden_table_rows = [] # type: List[str] for desc, gs in grouped_statuses.items(): - table_rows.append( + state = get_worst_state(gs) + table_row = ( f"{desc.name}{desc.description}" - f"{beauty_state(get_worst_state(gs))}\n" + f"{beauty_state(state)}\n" ) + if state == SUCCESS: + hidden_table_rows.append(table_row) + else: + visible_table_rows.append(table_row) - table_rows.sort() + visible_table_rows.sort() + hidden_table_rows.sort() - comment_footer = "" - return "".join([comment_body, *table_rows, comment_footer]) + return "".join( + [ + comment_body, + table_header, + *visible_table_rows, + table_footer, + details_header, + table_header, + *hidden_table_rows, + table_footer, + details_footer, + ] + ) def get_worst_state(statuses: CommitStatuses) -> str: From 89a7f9057949370573a369848c4a71460a23f959 Mon Sep 17 00:00:00 2001 From: Robert Schulze Date: Fri, 8 Sep 2023 10:25:28 +0000 Subject: [PATCH 220/433] Simplify and enable AVX-512 --- contrib/aklomp-base64-cmake/CMakeLists.txt | 61 +++++++--------------- contrib/aklomp-base64-cmake/config.h.in | 7 +-- 2 files changed, 23 insertions(+), 45 deletions(-) diff --git a/contrib/aklomp-base64-cmake/CMakeLists.txt b/contrib/aklomp-base64-cmake/CMakeLists.txt index 4607d1cc9b9..9accee5f534 100644 --- a/contrib/aklomp-base64-cmake/CMakeLists.txt +++ b/contrib/aklomp-base64-cmake/CMakeLists.txt @@ -7,30 +7,18 @@ endif() SET(LIBRARY_DIR "${ClickHouse_SOURCE_DIR}/contrib/aklomp-base64") -set(HAVE_NEON32 0) -set(HAVE_NEON64 0) -if (ARCH_AARCH64 AND NOT NO_ARMV81_OR_HIGHER) +# These defines enable/disable SIMD codecs in base64's runtime codec dispatch. +# We don't want to limit ourselves --> enable all. +set(HAVE_SSSE3 1) +set(HAVE_SSE41 1) +set(HAVE_SSE42 1) +set(HAVE_AVX 1) +set(HAVE_AVX2 1) +set(HAVE_AVX512 1) +# (no runtime dispatch on ARM) +if (NOT NO_ARMV81_OR_HIGHER) set(HAVE_NEON64 1) -endif () - -macro(cast_to_bool var instruction) - if (HAVE_${var}) - set(base64_${var} 1) - set(base64_${var}_opt ${instruction}) - else() - set(base64_${var} 0) - endif() -endmacro() - -cast_to_bool(SSSE3 "-mssse3") -cast_to_bool(SSE41 "-msse4.1") -cast_to_bool(SSE42 "-msse4.2") -cast_to_bool(AVX "-mavx") -cast_to_bool(AVX2 "-mavx2") - -set(HAVE_FAST_UNALIGNED_ACCESS 0) -if(HAVE_SSSE3 OR HAVE_SSE41 OR HAVE_SSE42 OR HAVE_AVX OR HAVE_AVX2 OR HAVE_NEON64) - set(HAVE_FAST_UNALIGNED_ACCESS 1) + set(HAVE_NEON32 0) endif () configure_file(config.h.in ${CMAKE_CURRENT_BINARY_DIR}/config.h) @@ -44,16 +32,14 @@ add_library(_base64 # Codec implementations "${LIBRARY_DIR}/lib/arch/generic/codec.c" - "${LIBRARY_DIR}/lib/arch/neon32/codec.c" - "${LIBRARY_DIR}/lib/arch/neon64/codec.c" "${LIBRARY_DIR}/lib/arch/ssse3/codec.c" "${LIBRARY_DIR}/lib/arch/sse41/codec.c" "${LIBRARY_DIR}/lib/arch/sse42/codec.c" - # ClickHouse does not support AVX* by default. We still need to compile these files - # because the run-time codec detection requires functions from these file to be present. "${LIBRARY_DIR}/lib/arch/avx/codec.c" "${LIBRARY_DIR}/lib/arch/avx2/codec.c" "${LIBRARY_DIR}/lib/arch/avx512/codec.c" + "${LIBRARY_DIR}/lib/arch/neon32/codec.c" + "${LIBRARY_DIR}/lib/arch/neon64/codec.c" # Tables "${LIBRARY_DIR}/lib/tables/table_dec_32bit.h" @@ -62,21 +48,12 @@ add_library(_base64 "${LIBRARY_DIR}/lib/codecs.h" "${CMAKE_CURRENT_BINARY_DIR}/config.h") -if(HAVE_AVX) - set_source_files_properties(${LIBRARY_DIR}/lib/arch/avx/codec.c PROPERTIES COMPILE_FLAGS -mavx) -endif() -if(HAVE_AVX2) - set_source_files_properties(${LIBRARY_DIR}/lib/arch/avx2/codec.c PROPERTIES COMPILE_FLAGS -mavx2) -endif() -if(HAVE_SSE41) - set_source_files_properties(${LIBRARY_DIR}/lib/arch/sse41/codec.c PROPERTIES COMPILE_FLAGS -msse4.1) -endif() -if(HAVE_SSE42) - set_source_files_properties(${LIBRARY_DIR}/lib/arch/sse42/codec.c PROPERTIES COMPILE_FLAGS -msse4.2) -endif() -if(HAVE_SSSE3) - set_source_files_properties(${LIBRARY_DIR}/lib/arch/ssse3/codec.c PROPERTIES COMPILE_FLAGS -mssse3) -endif() +set_source_files_properties(${LIBRARY_DIR}/lib/arch/ssse3/codec.c PROPERTIES COMPILE_FLAGS "-mssse3") +set_source_files_properties(${LIBRARY_DIR}/lib/arch/sse41/codec.c PROPERTIES COMPILE_FLAGS "-msse4.1") +set_source_files_properties(${LIBRARY_DIR}/lib/arch/sse42/codec.c PROPERTIES COMPILE_FLAGS "-msse4.2") +set_source_files_properties(${LIBRARY_DIR}/lib/arch/avx/codec.c PROPERTIES COMPILE_FLAGS "-mavx") +set_source_files_properties(${LIBRARY_DIR}/lib/arch/avx2/codec.c PROPERTIES COMPILE_FLAGS "-mavx2") +set_source_files_properties(${LIBRARY_DIR}/lib/arch/avx512/codec.c PROPERTIES COMPILE_FLAGS "-mavx512vl -mavx512vbmi") target_include_directories(_base64 SYSTEM PUBLIC ${LIBRARY_DIR}/include) target_include_directories(_base64 PRIVATE ${CMAKE_CURRENT_SOURCE_DIR} ${CMAKE_CURRENT_BINARY_DIR}) diff --git a/contrib/aklomp-base64-cmake/config.h.in b/contrib/aklomp-base64-cmake/config.h.in index 951bf4f88b1..4dc84632b30 100644 --- a/contrib/aklomp-base64-cmake/config.h.in +++ b/contrib/aklomp-base64-cmake/config.h.in @@ -1,8 +1,9 @@ -#cmakedefine01 HAVE_NEON32 -#cmakedefine01 HAVE_NEON64 #cmakedefine01 HAVE_SSSE3 #cmakedefine01 HAVE_SSE41 #cmakedefine01 HAVE_SSE42 #cmakedefine01 HAVE_AVX #cmakedefine01 HAVE_AVX2 -#cmakedefine01 HAVE_FAST_UNALIGNED_ACCESS +#cmakedefine01 HAVE_AVX512 + +#cmakedefine01 HAVE_NEON32 +#cmakedefine01 HAVE_NEON64 From 66479b69660e999c0f0a58a7aa976e1714869fbb Mon Sep 17 00:00:00 2001 From: yariks5s Date: Fri, 8 Sep 2023 10:54:25 +0000 Subject: [PATCH 221/433] code cleanup and performance improvement --- src/Functions/FunctionsConversion.h | 23 ++--------------------- 1 file changed, 2 insertions(+), 21 deletions(-) diff --git a/src/Functions/FunctionsConversion.h b/src/Functions/FunctionsConversion.h index 79bac2d17e8..6cdf961c450 100644 --- a/src/Functions/FunctionsConversion.h +++ b/src/Functions/FunctionsConversion.h @@ -990,20 +990,11 @@ struct ConvertImpl(*col_with_type_and_name.type); const DateLUTImpl * time_zone = nullptr; - if constexpr (std::is_same_v || std::is_same_v) - time_zone = &DateLUT::instance(); - /// For argument of Date or DateTime type, second argument with time zone could be specified. - if constexpr (std::is_same_v || std::is_same_v) - { - auto non_null_args = createBlockWithNestedColumns(arguments); - time_zone = &extractTimeZoneFromFunctionArguments(non_null_args, 1, 0); - } - if (const auto col_from = checkAndGetColumn(col_with_type_and_name.column.get())) { auto col_to = ColumnString::create(); @@ -1013,17 +1004,7 @@ struct ConvertImplgetOffsets(); size_t size = vec_from.size(); - if constexpr (std::is_same_v) - data_to.resize(size * (strlen("YYYY-MM-DD") + 1)); - else if constexpr (std::is_same_v) - data_to.resize(size * (strlen("YYYY-MM-DD") + 1)); - else if constexpr (std::is_same_v) - data_to.resize(size * (strlen("YYYY-MM-DD hh:mm:ss") + 1)); - else if constexpr (std::is_same_v) - data_to.resize(size * (strlen("YYYY-MM-DD hh:mm:ss.") + col_from->getScale() + 1)); - else - data_to.resize(size * 3); /// Arbitrary - + data_to.resize(size * 3); offsets_to.resize(size); WriteBufferFromVector write_buffer(data_to); From 2643e483106b571c4d4ea66adf9cecf5014736f5 Mon Sep 17 00:00:00 2001 From: Vitaly Baranov Date: Fri, 8 Sep 2023 13:23:59 +0200 Subject: [PATCH 222/433] Remove unused code --- src/Backups/RestoreCoordinationRemote.cpp | 5 ----- 1 file changed, 5 deletions(-) diff --git a/src/Backups/RestoreCoordinationRemote.cpp b/src/Backups/RestoreCoordinationRemote.cpp index 28cdfd2c519..c71466ad8f4 100644 --- a/src/Backups/RestoreCoordinationRemote.cpp +++ b/src/Backups/RestoreCoordinationRemote.cpp @@ -246,11 +246,6 @@ void RestoreCoordinationRemote::generateUUIDForTable(ASTCreateQuery & create_que with_retries.renewZooKeeper(zk); String path = zookeeper_path + "/table_uuids/" + escapeForFileName(query_str); - - Coordination::Requests ops; - ops.emplace_back(zkutil::makeCreateRequest(path, new_uuids_str, zkutil::CreateMode::Persistent)); - - Coordination::Responses responses; Coordination::Error res = zk->tryCreate(path, new_uuids_str, zkutil::CreateMode::Persistent); if (res == Coordination::Error::ZOK) From 4a1e2959fa2b10f86592663fac7ba15891b73ea3 Mon Sep 17 00:00:00 2001 From: Antonio Andelic Date: Fri, 8 Sep 2023 12:45:01 +0000 Subject: [PATCH 223/433] Apply PR comments --- src/Coordination/Changelog.cpp | 12 +++++++++++- src/Coordination/CoordinationSettings.cpp | 19 ++++++++----------- src/Coordination/KeeperDispatcher.cpp | 17 ++++++++++++----- src/Coordination/KeeperDispatcher.h | 2 +- src/Coordination/KeeperStateMachine.cpp | 2 +- src/Coordination/KeeperStateManager.cpp | 13 ++++++------- 6 files changed, 39 insertions(+), 26 deletions(-) diff --git a/src/Coordination/Changelog.cpp b/src/Coordination/Changelog.cpp index 7030775904c..3c2004a1b75 100644 --- a/src/Coordination/Changelog.cpp +++ b/src/Coordination/Changelog.cpp @@ -1048,18 +1048,26 @@ void Changelog::writeThread() LOG_WARNING(log, "Changelog is shut down"); }; + /// NuRaft writes a batch of request by first calling multiple store requests, i.e. AppendLog + /// finished by a flush request + /// We assume that after some number of appends, we always get flush request while (true) { if (try_batch_flush) { try_batch_flush = false; + /// we have Flush request stored in write operation + /// but we try to get new append operations + /// if there are none, we apply the currently set Flush + chassert(std::holds_alternative(write_operation)); if (!write_operations.tryPop(write_operation)) { chassert(batch_append_ok); const auto & flush = std::get(write_operation); flush_logs(flush); notify_append_completion(); - continue; + if (!write_operations.pop(write_operation)) + break; } } else if (!write_operations.pop(write_operation)) @@ -1092,10 +1100,12 @@ void Changelog::writeThread() try_batch_flush = true; continue; } + /// we need to flush because we have maximum allowed pending records flush_logs(flush); } else { + std::lock_guard lock{durable_idx_mutex}; *flush.failed = true; } notify_append_completion(); diff --git a/src/Coordination/CoordinationSettings.cpp b/src/Coordination/CoordinationSettings.cpp index 8688f6f5a40..1f27823182a 100644 --- a/src/Coordination/CoordinationSettings.cpp +++ b/src/Coordination/CoordinationSettings.cpp @@ -22,19 +22,16 @@ void CoordinationSettings::loadFromConfig(const String & config_elem, const Poco Poco::Util::AbstractConfiguration::Keys config_keys; config.keys(config_elem, config_keys); - for (const String & key : config_keys) + try { - try - { + for (const String & key : config_keys) set(key, config.getString(config_elem + "." + key)); - } - catch (Exception & e) - { - if (e.code() == ErrorCodes::UNKNOWN_SETTING) - LOG_WARNING(&Poco::Logger::get("CoordinationSettings"), "Found unknown coordination setting in config: '{}'", key); - else - throw; - } + } + catch (Exception & e) + { + if (e.code() == ErrorCodes::UNKNOWN_SETTING) + e.addMessage("in Coordination settings config"); + throw; } } diff --git a/src/Coordination/KeeperDispatcher.cpp b/src/Coordination/KeeperDispatcher.cpp index ecd81661d8f..d93bdb21db0 100644 --- a/src/Coordination/KeeperDispatcher.cpp +++ b/src/Coordination/KeeperDispatcher.cpp @@ -149,7 +149,8 @@ void KeeperDispatcher::requestThread() }; /// Waiting until previous append will be successful, or batch is big enough - while (!shutdown_called && !has_reconfig_request && !prev_result_done() && current_batch.size() <= max_batch_size + while (!shutdown_called && !has_reconfig_request && + !prev_result_done() && current_batch.size() <= max_batch_size && current_batch_bytes_size < max_batch_bytes_size) { try_get_request(); @@ -190,18 +191,24 @@ void KeeperDispatcher::requestThread() if (prev_result) result_buf = forceWaitAndProcessResult(prev_result, current_batch); + /// In case of older version or disabled async replication, result buf will be set to value of `commit` function + /// which always returns nullptr + /// in that case we don't have to do manual wait because are already sure that the batch was committed when we get + /// the result back + /// otherwise, we need to manually wait until the batch is committed if (result_buf) { nuraft::buffer_serializer bs(result_buf); auto log_idx = bs.get_u64(); + /// we will wake up this thread on each commit so we need to run it in loop until the last request of batch is committed while (true) { - auto current_last_committed_idx = last_committed_log_idx.load(std::memory_order_relaxed); + auto current_last_committed_idx = our_last_committed_log_idx.load(std::memory_order_relaxed); if (current_last_committed_idx >= log_idx) break; - last_committed_log_idx.wait(current_last_committed_idx); + our_last_committed_log_idx.wait(current_last_committed_idx); } } } @@ -397,8 +404,8 @@ void KeeperDispatcher::initialize(const Poco::Util::AbstractConfiguration & conf } } - last_committed_log_idx.store(log_idx, std::memory_order_relaxed); - last_committed_log_idx.notify_all(); + our_last_committed_log_idx.store(log_idx, std::memory_order_relaxed); + our_last_committed_log_idx.notify_all(); }); try diff --git a/src/Coordination/KeeperDispatcher.h b/src/Coordination/KeeperDispatcher.h index 31a8f80f252..f11102291f0 100644 --- a/src/Coordination/KeeperDispatcher.h +++ b/src/Coordination/KeeperDispatcher.h @@ -107,7 +107,7 @@ private: public: std::mutex read_request_queue_mutex; - std::atomic last_committed_log_idx = 0; + std::atomic our_last_committed_log_idx = 0; /// queue of read requests that can be processed after a request with specific session ID and XID is committed std::unordered_map> read_request_queue; diff --git a/src/Coordination/KeeperStateMachine.cpp b/src/Coordination/KeeperStateMachine.cpp index 5a8cf88ea73..e753695fd46 100644 --- a/src/Coordination/KeeperStateMachine.cpp +++ b/src/Coordination/KeeperStateMachine.cpp @@ -167,7 +167,7 @@ nuraft::ptr KeeperStateMachine::pre_commit(uint64_t log_idx, nur request_for_session->zxid = log_idx; preprocess(*request_for_session); - auto result = nuraft::buffer::alloc(8); + auto result = nuraft::buffer::alloc(sizeof(log_idx)); nuraft::buffer_serializer ss(result); ss.put_u64(log_idx); return result; diff --git a/src/Coordination/KeeperStateManager.cpp b/src/Coordination/KeeperStateManager.cpp index f52d0749b10..879a42f6258 100644 --- a/src/Coordination/KeeperStateManager.cpp +++ b/src/Coordination/KeeperStateManager.cpp @@ -250,15 +250,14 @@ KeeperStateManager::KeeperStateManager( , log_store(nuraft::cs_new( LogFileSettings { - .force_sync = coordination_settings->force_sync, - .compress_logs = coordination_settings->compress_logs, - .rotate_interval = coordination_settings->rotate_log_storage_interval, - .max_size = coordination_settings->max_log_file_size, - .overallocate_size = coordination_settings->log_file_overallocate_size - }, + .force_sync = coordination_settings->force_sync, + .compress_logs = coordination_settings->compress_logs, + .rotate_interval = coordination_settings->rotate_log_storage_interval, + .max_size = coordination_settings->max_log_file_size, + .overallocate_size = coordination_settings->log_file_overallocate_size}, FlushSettings { - .max_flush_batch_size = coordination_settings->max_flush_batch_size, + .max_flush_batch_size = coordination_settings->max_flush_batch_size, }, keeper_context_)) , server_state_file_name(server_state_file_name_) From 3c9d9ce39ef1f95c18cb2ac467bceef438572f58 Mon Sep 17 00:00:00 2001 From: Robert Schulze Date: Fri, 8 Sep 2023 13:10:35 +0000 Subject: [PATCH 224/433] Fix build --- contrib/aklomp-base64-cmake/CMakeLists.txt | 81 ++++++++++++---------- 1 file changed, 44 insertions(+), 37 deletions(-) diff --git a/contrib/aklomp-base64-cmake/CMakeLists.txt b/contrib/aklomp-base64-cmake/CMakeLists.txt index 9accee5f534..4b988fad860 100644 --- a/contrib/aklomp-base64-cmake/CMakeLists.txt +++ b/contrib/aklomp-base64-cmake/CMakeLists.txt @@ -7,53 +7,60 @@ endif() SET(LIBRARY_DIR "${ClickHouse_SOURCE_DIR}/contrib/aklomp-base64") -# These defines enable/disable SIMD codecs in base64's runtime codec dispatch. -# We don't want to limit ourselves --> enable all. -set(HAVE_SSSE3 1) -set(HAVE_SSE41 1) -set(HAVE_SSE42 1) -set(HAVE_AVX 1) -set(HAVE_AVX2 1) -set(HAVE_AVX512 1) -# (no runtime dispatch on ARM) -if (NOT NO_ARMV81_OR_HIGHER) - set(HAVE_NEON64 1) - set(HAVE_NEON32 0) +if (ARCH_AMD64) + # These defines enable/disable SIMD codecs in base64's runtime codec dispatch. + # We don't want to limit ourselves --> enable all. + set(HAVE_SSSE3 1) + set(HAVE_SSE41 1) + set(HAVE_SSE42 1) + set(HAVE_AVX 1) + set(HAVE_AVX2 1) + set(HAVE_AVX512 1) +endif () + +if (ARCH_AARCH64) + # The choice of HAVE_NEON* depends on the target machine because base64 provides + # no runtime dispatch on ARM. NEON is only mandatory with the normal build profile. + if(NOT NO_ARMV81_OR_HIGHER) + set(HAVE_NEON64 1) + set(HAVE_NEON32 0) + endif () endif () configure_file(config.h.in ${CMAKE_CURRENT_BINARY_DIR}/config.h) add_library(_base64 - # Library files - "${LIBRARY_DIR}/lib/lib.c" - "${LIBRARY_DIR}/lib/codec_choose.c" + "${LIBRARY_DIR}/lib/lib.c" + "${LIBRARY_DIR}/lib/codec_choose.c" - "${LIBRARY_DIR}/lib/tables/tables.c" + "${LIBRARY_DIR}/lib/tables/tables.c" + "${LIBRARY_DIR}/lib/tables/table_dec_32bit.h" + "${LIBRARY_DIR}/lib/tables/table_enc_12bit.h" - # Codec implementations - "${LIBRARY_DIR}/lib/arch/generic/codec.c" - "${LIBRARY_DIR}/lib/arch/ssse3/codec.c" - "${LIBRARY_DIR}/lib/arch/sse41/codec.c" - "${LIBRARY_DIR}/lib/arch/sse42/codec.c" - "${LIBRARY_DIR}/lib/arch/avx/codec.c" - "${LIBRARY_DIR}/lib/arch/avx2/codec.c" - "${LIBRARY_DIR}/lib/arch/avx512/codec.c" - "${LIBRARY_DIR}/lib/arch/neon32/codec.c" - "${LIBRARY_DIR}/lib/arch/neon64/codec.c" + "${LIBRARY_DIR}/lib/codecs.h" - # Tables - "${LIBRARY_DIR}/lib/tables/table_dec_32bit.h" - "${LIBRARY_DIR}/lib/tables/table_enc_12bit.h" + "${CMAKE_CURRENT_BINARY_DIR}/config.h" - "${LIBRARY_DIR}/lib/codecs.h" - "${CMAKE_CURRENT_BINARY_DIR}/config.h") + "${LIBRARY_DIR}/lib/arch/generic/codec.c" + "${LIBRARY_DIR}/lib/arch/ssse3/codec.c" + "${LIBRARY_DIR}/lib/arch/sse41/codec.c" + "${LIBRARY_DIR}/lib/arch/sse42/codec.c" + "${LIBRARY_DIR}/lib/arch/avx/codec.c" + "${LIBRARY_DIR}/lib/arch/avx2/codec.c" + "${LIBRARY_DIR}/lib/arch/avx512/codec.c" -set_source_files_properties(${LIBRARY_DIR}/lib/arch/ssse3/codec.c PROPERTIES COMPILE_FLAGS "-mssse3") -set_source_files_properties(${LIBRARY_DIR}/lib/arch/sse41/codec.c PROPERTIES COMPILE_FLAGS "-msse4.1") -set_source_files_properties(${LIBRARY_DIR}/lib/arch/sse42/codec.c PROPERTIES COMPILE_FLAGS "-msse4.2") -set_source_files_properties(${LIBRARY_DIR}/lib/arch/avx/codec.c PROPERTIES COMPILE_FLAGS "-mavx") -set_source_files_properties(${LIBRARY_DIR}/lib/arch/avx2/codec.c PROPERTIES COMPILE_FLAGS "-mavx2") -set_source_files_properties(${LIBRARY_DIR}/lib/arch/avx512/codec.c PROPERTIES COMPILE_FLAGS "-mavx512vl -mavx512vbmi") + "${LIBRARY_DIR}/lib/arch/neon32/codec.c" + "${LIBRARY_DIR}/lib/arch/neon64/codec.c" +) + +if (ARCH_AMD64) + set_source_files_properties(${LIBRARY_DIR}/lib/arch/ssse3/codec.c PROPERTIES COMPILE_FLAGS "-mssse3") + set_source_files_properties(${LIBRARY_DIR}/lib/arch/sse41/codec.c PROPERTIES COMPILE_FLAGS "-msse4.1") + set_source_files_properties(${LIBRARY_DIR}/lib/arch/sse42/codec.c PROPERTIES COMPILE_FLAGS "-msse4.2") + set_source_files_properties(${LIBRARY_DIR}/lib/arch/avx/codec.c PROPERTIES COMPILE_FLAGS "-mavx") + set_source_files_properties(${LIBRARY_DIR}/lib/arch/avx2/codec.c PROPERTIES COMPILE_FLAGS "-mavx2") + set_source_files_properties(${LIBRARY_DIR}/lib/arch/avx512/codec.c PROPERTIES COMPILE_FLAGS "-mavx512vl -mavx512vbmi") +endif() target_include_directories(_base64 SYSTEM PUBLIC ${LIBRARY_DIR}/include) target_include_directories(_base64 PRIVATE ${CMAKE_CURRENT_SOURCE_DIR} ${CMAKE_CURRENT_BINARY_DIR}) From 12ce57369bd8cd80043257ffa87d519f6add4279 Mon Sep 17 00:00:00 2001 From: Yarik Briukhovetskyi <114298166+yariks5s@users.noreply.github.com> Date: Fri, 8 Sep 2023 15:36:33 +0200 Subject: [PATCH 225/433] Update FunctionsConversion.h --- src/Functions/FunctionsConversion.h | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-) diff --git a/src/Functions/FunctionsConversion.h b/src/Functions/FunctionsConversion.h index 6cdf961c450..b7e21b81b21 100644 --- a/src/Functions/FunctionsConversion.h +++ b/src/Functions/FunctionsConversion.h @@ -993,8 +993,6 @@ struct ConvertImpl(*col_with_type_and_name.type); - const DateLUTImpl * time_zone = nullptr; - if (const auto col_from = checkAndGetColumn(col_with_type_and_name.column.get())) { auto col_to = ColumnString::create(); @@ -1013,7 +1011,8 @@ struct ConvertImpl::template execute(vec_from[i], write_buffer, &type, time_zone); + bool is_ok = FormatImpl::template execute(vec_from[i], write_buffer, &type, nullptr); + /// We don't use timezones in this branch null_map->getData()[i] |= !is_ok; writeChar(0, write_buffer); offsets_to[i] = write_buffer.count(); @@ -1023,7 +1022,7 @@ struct ConvertImpl::template execute(vec_from[i], write_buffer, &type, time_zone); + FormatImpl::template execute(vec_from[i], write_buffer, &type, nullptr); writeChar(0, write_buffer); offsets_to[i] = write_buffer.count(); } From 4cfd46a0cc86709610e283b5cc9781bb7b79c170 Mon Sep 17 00:00:00 2001 From: Andrey Zvonov <32552679+zvonand@users.noreply.github.com> Date: Fri, 8 Sep 2023 17:29:09 +0300 Subject: [PATCH 226/433] Update settings.md --- docs/en/operations/settings/settings.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/docs/en/operations/settings/settings.md b/docs/en/operations/settings/settings.md index 2fac021fe2d..9a9c4a57e29 100644 --- a/docs/en/operations/settings/settings.md +++ b/docs/en/operations/settings/settings.md @@ -4663,7 +4663,7 @@ The default value is `false`. ## ignore_access_denied_multidirectory_globs {#ignore_access_denied_multidirectory_globs} Allows to ignore 'permission denied' errors when using multi-directory `{}` globs for [File](../../sql-reference/table-functions/file.md#globs_in_path) and [HDFS](../../sql-reference/table-functions/hdfs.md) storages. -This setting is only applicable when multi directory `{}` glob is used. +This setting is only applicable to multi directory `{}` globs. Possible values: `0`, `1`. From 4cc0b7c7f3a9f32cbf27b23829faed5d66a08736 Mon Sep 17 00:00:00 2001 From: Han Fei Date: Fri, 8 Sep 2023 16:34:02 +0200 Subject: [PATCH 227/433] update --- contrib/libunwind | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/contrib/libunwind b/contrib/libunwind index f12a21bdd60..30cc1d3fd36 160000 --- a/contrib/libunwind +++ b/contrib/libunwind @@ -1 +1 @@ -Subproject commit f12a21bdd601232465791a32ed054bb1dded8e02 +Subproject commit 30cc1d3fd3655a5cfa0ab112fe320fb9fc0a8344 From ee608d7c815a466aa2e85fb505aab70ff2a6581e Mon Sep 17 00:00:00 2001 From: Denny Crane Date: Fri, 8 Sep 2023 12:32:31 -0300 Subject: [PATCH 228/433] Update http.md --- docs/en/interfaces/http.md | 23 +++++++++++++++++++++++ 1 file changed, 23 insertions(+) diff --git a/docs/en/interfaces/http.md b/docs/en/interfaces/http.md index 37821f0fee1..01db04b9fe0 100644 --- a/docs/en/interfaces/http.md +++ b/docs/en/interfaces/http.md @@ -18,6 +18,8 @@ $ curl 'http://localhost:8123/' Ok. ``` +Also see: [HTTP response codes caveats](#http_response_codes_caveats). + Sometimes, `curl` command is not available on user operating systems. On Ubuntu or Debian, run `sudo apt install curl`. Please refer this [documentation](https://curl.se/download.html) to install it before running the examples. Web UI can be accessed here: `http://localhost:8123/play`. @@ -323,6 +325,27 @@ $ curl -sS 'http://localhost:8123/?max_result_bytes=4000000&buffer_size=3000000& Use buffering to avoid situations where a query processing error occurred after the response code and HTTP headers were sent to the client. In this situation, an error message is written at the end of the response body, and on the client-side, the error can only be detected at the parsing stage. +## HTTP response codes caveats {#http_response_codes_caveats} + +Because of limitation of HTTP protocol, HTTP 200 response code does not guarantee that a query was successful. + +Here is an example: + +``` +curl -v -Ss "http://localhost:8123/?max_block_size=1&query=select+sleepEachRow(0.001),throwIf(number=2)from+numbers(5)" +* Trying 127.0.0.1:8123... +... +< HTTP/1.1 200 OK +... +Code: 395. DB::Exception: Value passed to 'throwIf' function is non-zero: while executing 'FUNCTION throwIf(equals(number, 2) :: 1) -> throwIf(equals(number, 2)) +``` + +The reason for this behavior is the nature of the HTTP protocol. The HTTP header is sent first with an HTTP code of 200, followed by the HTTP body, and then the error is injected into the body as plain text. +This behavior is independent of the format used, whether it's `Native`, `TSV`, or `JSON`; the error message will always be in the middle of the response stream. +You can mitigate this problem by enabling `wait_end_of_query=1` ([Response Buffering](#response-buffering)). In this case, the sending of the HTTP header is delayed until the entire query is resolved. +However, this does not completely solve the problem because the result must still fit within the `http_response_buffer_size`, and other settings like `send_progress_in_http_headers` can interfere with the delay of the header. +The only way to catch all errors is to analyze the HTTP body before parsing it using the required format. + ### Queries with Parameters {#cli-queries-with-parameters} You can create a query with parameters and pass values for them from the corresponding HTTP request parameters. For more information, see [Queries with Parameters for CLI](../interfaces/cli.md#cli-queries-with-parameters). From b0b8255c6f53c753532abc5a676eeaee9dfdbaf7 Mon Sep 17 00:00:00 2001 From: vdimir Date: Fri, 8 Sep 2023 15:36:32 +0000 Subject: [PATCH 229/433] Set PR status into title --- tests/ci/commit_status_helper.py | 73 ++++++++++++++++++++------------ 1 file changed, 45 insertions(+), 28 deletions(-) diff --git a/tests/ci/commit_status_helper.py b/tests/ci/commit_status_helper.py index 4d6037c76a8..b9c0392ecc4 100644 --- a/tests/ci/commit_status_helper.py +++ b/tests/ci/commit_status_helper.py @@ -4,6 +4,7 @@ import csv import os import time from typing import Dict, List, Optional, Union +from collections import defaultdict import logging from github import Github @@ -128,6 +129,28 @@ def post_commit_status( logging.error("Failed to update the status comment, continue anyway") +STATUS_ICON_MAP = defaultdict( + str, + { + ERROR: "🔴", + FAILURE: "🔴", + PENDING: "🟡", + SUCCESS: "🟢", + }, +) + + +def update_pr_title_icon(pr: Github.PullRequest, status: str) -> None: + new_title = pr.title + new_status_icon = STATUS_ICON_MAP[status] + if new_title and new_title[-1] != new_status_icon: + if new_title[-1] in set(STATUS_ICON_MAP.values()): + new_title = new_title[:-1] + " " + new_status_icon + else: + new_title = new_title + " " + new_status_icon + pr.edit(title=new_title) + + def set_status_comment(commit: Commit, pr_info: PRInfo) -> None: """It adds or updates the comment status to all Pull Requests but for release one, so the method does nothing for simple pushes and pull requests with @@ -167,6 +190,8 @@ def set_status_comment(commit: Commit, pr_info: PRInfo) -> None: comment = ic break + update_pr_title_icon(pr, get_worst_state(statuses)) + if comment is None: pr.create_issue_comment(comment_body) return @@ -180,15 +205,6 @@ def set_status_comment(commit: Commit, pr_info: PRInfo) -> None: def generate_status_comment(pr_info: PRInfo, statuses: CommitStatuses) -> str: """The method generates the comment body, as well it updates the CI report""" - def beauty_state(state: str) -> str: - if state == SUCCESS: - return f"🟢 {state}" - if state == PENDING: - return f"🟡 {state}" - if state in [ERROR, FAILURE]: - return f"🔴 {state}" - return state - report_url = create_ci_report(pr_info, statuses) worst_state = get_worst_state(statuses) if not worst_state: @@ -196,14 +212,14 @@ def generate_status_comment(pr_info: PRInfo, statuses: CommitStatuses) -> str: # the function should not be used on empty statuses worst_state = "The commit doesn't have the statuses yet" else: - worst_state = f"The overall status of the commit is {beauty_state(worst_state)}" + worst_state = f"The overall status of the commit is {STATUS_ICON_MAP[worst_state]} {worst_state}" comment_body = ( f"\n" f"*This is an automated comment for commit {pr_info.sha} with " - f"description of existing statuses. It's updated for the latest CI running*\n" - f"📄 [Click here]({report_url}) to open a full report in a separate page\n" + f"description of existing statuses. It's updated for the latest CI running*\n\n" + f"[📄 Click here]({report_url}) to open a full report in a separate page\n" f"{worst_state}\n\n" ) # group checks by the name to get the worst one per each @@ -244,29 +260,30 @@ def generate_status_comment(pr_info: PRInfo, statuses: CommitStatuses) -> str: state = get_worst_state(gs) table_row = ( f"{desc.name}{desc.description}" - f"{beauty_state(state)}\n" + f"{STATUS_ICON_MAP[worst_state]} {worst_state}\n" ) if state == SUCCESS: hidden_table_rows.append(table_row) else: visible_table_rows.append(table_row) - visible_table_rows.sort() - hidden_table_rows.sort() + result = [comment_body] - return "".join( - [ - comment_body, - table_header, - *visible_table_rows, - table_footer, - details_header, - table_header, - *hidden_table_rows, - table_footer, - details_footer, - ] - ) + if visible_table_rows: + visible_table_rows.sort() + result.append(table_header) + result.extend(visible_table_rows) + result.append(table_footer) + + if hidden_table_rows: + hidden_table_rows.sort() + result.append(details_header) + result.append(table_header) + result.extend(hidden_table_rows) + result.append(table_footer) + result.append(details_footer) + + return "".join(result) def get_worst_state(statuses: CommitStatuses) -> str: From 465d84aba7412082123d475df3fb83d927062f15 Mon Sep 17 00:00:00 2001 From: vdimir Date: Fri, 8 Sep 2023 15:39:07 +0000 Subject: [PATCH 230/433] Set raw_logs in create_ci_report --- tests/ci/commit_status_helper.py | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) diff --git a/tests/ci/commit_status_helper.py b/tests/ci/commit_status_helper.py index b9c0392ecc4..52ee80d66f0 100644 --- a/tests/ci/commit_status_helper.py +++ b/tests/ci/commit_status_helper.py @@ -298,7 +298,14 @@ def create_ci_report(pr_info: PRInfo, statuses: CommitStatuses) -> str: log_urls = None if status.target_url is not None: log_urls = [status.target_url] - test_results.append(TestResult(status.context, status.state, log_urls=log_urls)) + raw_logs = None + if status.description: + raw_logs = status.description + test_results.append( + TestResult( + status.context, status.state, log_urls=log_urls, raw_logs=raw_logs + ) + ) return upload_results( S3Helper(), pr_info.number, pr_info.sha, test_results, [], CI_STATUS_NAME ) From a46d1974ac751df34fbf047ac795b0e4ebf495fe Mon Sep 17 00:00:00 2001 From: vdimir Date: Fri, 8 Sep 2023 15:49:13 +0000 Subject: [PATCH 231/433] fix --- tests/ci/commit_status_helper.py | 19 ++++++++----------- 1 file changed, 8 insertions(+), 11 deletions(-) diff --git a/tests/ci/commit_status_helper.py b/tests/ci/commit_status_helper.py index 52ee80d66f0..207a04f06ad 100644 --- a/tests/ci/commit_status_helper.py +++ b/tests/ci/commit_status_helper.py @@ -12,6 +12,7 @@ from github.GithubObject import _NotSetType, NotSet as NotSet from github.Commit import Commit from github.CommitStatus import CommitStatus from github.IssueComment import IssueComment +from github.PullRequest import PullRequest from github.Repository import Repository from ci_config import CI_CONFIG, REQUIRED_CHECKS, CHECK_DESCRIPTIONS, CheckDescription @@ -140,9 +141,11 @@ STATUS_ICON_MAP = defaultdict( ) -def update_pr_title_icon(pr: Github.PullRequest, status: str) -> None: - new_title = pr.title +def update_pr_title_icon(pr: PullRequest, status: str) -> None: new_status_icon = STATUS_ICON_MAP[status] + if not new_status_icon: + return + new_title = pr.title if new_title and new_title[-1] != new_status_icon: if new_title[-1] in set(STATUS_ICON_MAP.values()): new_title = new_title[:-1] + " " + new_status_icon @@ -207,20 +210,14 @@ def generate_status_comment(pr_info: PRInfo, statuses: CommitStatuses) -> str: report_url = create_ci_report(pr_info, statuses) worst_state = get_worst_state(statuses) - if not worst_state: - # Theoretically possible, although - # the function should not be used on empty statuses - worst_state = "The commit doesn't have the statuses yet" - else: - worst_state = f"The overall status of the commit is {STATUS_ICON_MAP[worst_state]} {worst_state}" comment_body = ( f"\n" f"*This is an automated comment for commit {pr_info.sha} with " f"description of existing statuses. It's updated for the latest CI running*\n\n" - f"[📄 Click here]({report_url}) to open a full report in a separate page\n" - f"{worst_state}\n\n" + f"[{STATUS_ICON_MAP[worst_state]} Click here]({report_url}) to open a full report in a separate page\n" + f"\n" ) # group checks by the name to get the worst one per each grouped_statuses = {} # type: Dict[CheckDescription, CommitStatuses] @@ -260,7 +257,7 @@ def generate_status_comment(pr_info: PRInfo, statuses: CommitStatuses) -> str: state = get_worst_state(gs) table_row = ( f"{desc.name}{desc.description}" - f"{STATUS_ICON_MAP[worst_state]} {worst_state}\n" + f"{STATUS_ICON_MAP[state]} {state}\n" ) if state == SUCCESS: hidden_table_rows.append(table_row) From 6a51d9fd7d2a373741f79ac7f9100225d79ef8bd Mon Sep 17 00:00:00 2001 From: vdimir Date: Fri, 8 Sep 2023 16:01:13 +0000 Subject: [PATCH 232/433] do not merge this commit --- src/Storages/StorageReplicatedMergeTree.cpp | 3 +-- tests/ci/commit_status_helper.py | 2 +- 2 files changed, 2 insertions(+), 3 deletions(-) diff --git a/src/Storages/StorageReplicatedMergeTree.cpp b/src/Storages/StorageReplicatedMergeTree.cpp index e6edb99e301..840a77868d6 100644 --- a/src/Storages/StorageReplicatedMergeTree.cpp +++ b/src/Storages/StorageReplicatedMergeTree.cpp @@ -7656,8 +7656,7 @@ void StorageReplicatedMergeTree::replacePartitionFrom( entry_replace.columns_version = -1; } - if (replace) - { + if (replace) { /// Cancel concurrent inserts in range clearLockedBlockNumbersInPartition(*zookeeper, drop_range.partition_id, drop_range.max_block, drop_range.max_block); /// Remove deduplication block_ids of replacing parts diff --git a/tests/ci/commit_status_helper.py b/tests/ci/commit_status_helper.py index 207a04f06ad..9788e629cb8 100644 --- a/tests/ci/commit_status_helper.py +++ b/tests/ci/commit_status_helper.py @@ -248,7 +248,7 @@ def generate_status_comment(pr_info: PRInfo, statuses: CommitStatuses) -> str: ) table_footer = "\n\n" - details_header = "
Successfull checks\n" + details_header = "
Successful checks\n" details_footer = "
\n" visible_table_rows = [] # type: List[str] From 2d178e9bfdc229329a2e95e8c0e8aa2693c01bca Mon Sep 17 00:00:00 2001 From: Denny Crane Date: Fri, 8 Sep 2023 15:58:24 -0300 Subject: [PATCH 233/433] Update type-conversion-functions.md --- .../functions/type-conversion-functions.md | 23 +++++++++++++++---- 1 file changed, 18 insertions(+), 5 deletions(-) diff --git a/docs/en/sql-reference/functions/type-conversion-functions.md b/docs/en/sql-reference/functions/type-conversion-functions.md index 36f40b37238..4d2e6ccd9b5 100644 --- a/docs/en/sql-reference/functions/type-conversion-functions.md +++ b/docs/en/sql-reference/functions/type-conversion-functions.md @@ -892,16 +892,29 @@ Query: ``` sql SELECT - now() AS now_local, - toString(now(), 'Asia/Yekaterinburg') AS now_yekat; + time_zone, + now() AS ts, + toString(ts, time_zone) +FROM system.time_zones +WHERE time_zone LIKE 'Europe%' +LIMIT 10 ``` Result: ```response -┌───────────now_local─┬─now_yekat───────────┐ -│ 2016-06-15 00:11:21 │ 2016-06-15 02:11:21 │ -└─────────────────────┴─────────────────────┘ +┌─time_zone─────────┬──────────────────ts─┬─toString(now(), time_zone)─┐ +│ Europe/Amsterdam │ 2023-09-08 18:51:22 │ 2023-09-08 20:51:22 │ +│ Europe/Andorra │ 2023-09-08 18:51:22 │ 2023-09-08 20:51:22 │ +│ Europe/Astrakhan │ 2023-09-08 18:51:22 │ 2023-09-08 22:51:22 │ +│ Europe/Athens │ 2023-09-08 18:51:22 │ 2023-09-08 21:51:22 │ +│ Europe/Belfast │ 2023-09-08 18:51:22 │ 2023-09-08 19:51:22 │ +│ Europe/Belgrade │ 2023-09-08 18:51:22 │ 2023-09-08 20:51:22 │ +│ Europe/Berlin │ 2023-09-08 18:51:22 │ 2023-09-08 20:51:22 │ +│ Europe/Bratislava │ 2023-09-08 18:51:22 │ 2023-09-08 20:51:22 │ +│ Europe/Brussels │ 2023-09-08 18:51:22 │ 2023-09-08 20:51:22 │ +│ Europe/Bucharest │ 2023-09-08 18:51:22 │ 2023-09-08 21:51:22 │ +└───────────────────┴─────────────────────┴────────────────────────────┘ ``` Also see the `toUnixTimestamp` function. From a0ef063b2d8fa839b176ccffa268ec8d9350e083 Mon Sep 17 00:00:00 2001 From: Denny Crane Date: Fri, 8 Sep 2023 16:04:35 -0300 Subject: [PATCH 234/433] Update date-time-functions.md --- .../functions/date-time-functions.md | 27 +++++++++++++++++++ 1 file changed, 27 insertions(+) diff --git a/docs/en/sql-reference/functions/date-time-functions.md b/docs/en/sql-reference/functions/date-time-functions.md index 6bcc9d654c5..03cd75b21ab 100644 --- a/docs/en/sql-reference/functions/date-time-functions.md +++ b/docs/en/sql-reference/functions/date-time-functions.md @@ -1494,6 +1494,33 @@ Result: └─────────────────────────────────────────────────────────────────────┘ ``` +Additionally, the `formatDateTime` function can take a third String argument containing the name of the time zone. Example: `Asia/Istanbul`. In this case, the time is formatted according to the specified time zone. + +**Example** + +```sql +SELECT + time_zone, + now() AS ts, + formatDateTime(ts, '%T', time_zone) AS str_time_timezone +FROM system.time_zones +WHERE time_zone LIKE 'Europe%' +LIMIT 10 + +┌─time_zone─────────┬──────────────────ts─┬─str_time_timezone─┐ +│ Europe/Amsterdam │ 2023-09-08 19:03:14 │ 21:03:14 │ +│ Europe/Andorra │ 2023-09-08 19:03:14 │ 21:03:14 │ +│ Europe/Astrakhan │ 2023-09-08 19:03:14 │ 23:03:14 │ +│ Europe/Athens │ 2023-09-08 19:03:14 │ 22:03:14 │ +│ Europe/Belfast │ 2023-09-08 19:03:14 │ 20:03:14 │ +│ Europe/Belgrade │ 2023-09-08 19:03:14 │ 21:03:14 │ +│ Europe/Berlin │ 2023-09-08 19:03:14 │ 21:03:14 │ +│ Europe/Bratislava │ 2023-09-08 19:03:14 │ 21:03:14 │ +│ Europe/Brussels │ 2023-09-08 19:03:14 │ 21:03:14 │ +│ Europe/Bucharest │ 2023-09-08 19:03:14 │ 22:03:14 │ +└───────────────────┴─────────────────────┴───────────────────┘ +``` + **See Also** - [formatDateTimeInJodaSyntax](##formatDateTimeInJodaSyntax) From 3f58e2f1342b4c61bb9c4100d463d94344fdf768 Mon Sep 17 00:00:00 2001 From: Denny Crane Date: Fri, 8 Sep 2023 16:10:16 -0300 Subject: [PATCH 235/433] Update date-time-functions.md --- docs/en/sql-reference/functions/date-time-functions.md | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/docs/en/sql-reference/functions/date-time-functions.md b/docs/en/sql-reference/functions/date-time-functions.md index 03cd75b21ab..02dea43c9f9 100644 --- a/docs/en/sql-reference/functions/date-time-functions.md +++ b/docs/en/sql-reference/functions/date-time-functions.md @@ -237,6 +237,11 @@ type_samoa: DateTime('US/Samoa') int32samoa: 1546300800 ``` +**See Also** + +- [formatDateTime](#date_time_functions-formatDateTime) - supports non-constant timezone. +- [toString](type-conversion-functions.md#tostring) - supports non-constant timezone. + ## timeZoneOf Returns the timezone name of [DateTime](../../sql-reference/data-types/datetime.md) or [DateTime64](../../sql-reference/data-types/datetime64.md) data types. From 24928217b9f52bce1e95f92554dc1cf1fb53525f Mon Sep 17 00:00:00 2001 From: Denny Crane Date: Fri, 8 Sep 2023 16:14:00 -0300 Subject: [PATCH 236/433] Update date-time-functions.md --- .../functions/date-time-functions.md | 28 +++++++++---------- 1 file changed, 14 insertions(+), 14 deletions(-) diff --git a/docs/en/sql-reference/functions/date-time-functions.md b/docs/en/sql-reference/functions/date-time-functions.md index 02dea43c9f9..cc8469e82ef 100644 --- a/docs/en/sql-reference/functions/date-time-functions.md +++ b/docs/en/sql-reference/functions/date-time-functions.md @@ -1505,25 +1505,25 @@ Additionally, the `formatDateTime` function can take a third String argument con ```sql SELECT - time_zone, now() AS ts, - formatDateTime(ts, '%T', time_zone) AS str_time_timezone + time_zone, + formatDateTime(ts, '%T', time_zone) AS str_tz_time FROM system.time_zones WHERE time_zone LIKE 'Europe%' LIMIT 10 -┌─time_zone─────────┬──────────────────ts─┬─str_time_timezone─┐ -│ Europe/Amsterdam │ 2023-09-08 19:03:14 │ 21:03:14 │ -│ Europe/Andorra │ 2023-09-08 19:03:14 │ 21:03:14 │ -│ Europe/Astrakhan │ 2023-09-08 19:03:14 │ 23:03:14 │ -│ Europe/Athens │ 2023-09-08 19:03:14 │ 22:03:14 │ -│ Europe/Belfast │ 2023-09-08 19:03:14 │ 20:03:14 │ -│ Europe/Belgrade │ 2023-09-08 19:03:14 │ 21:03:14 │ -│ Europe/Berlin │ 2023-09-08 19:03:14 │ 21:03:14 │ -│ Europe/Bratislava │ 2023-09-08 19:03:14 │ 21:03:14 │ -│ Europe/Brussels │ 2023-09-08 19:03:14 │ 21:03:14 │ -│ Europe/Bucharest │ 2023-09-08 19:03:14 │ 22:03:14 │ -└───────────────────┴─────────────────────┴───────────────────┘ +┌──────────────────ts─┬─time_zone─────────┬─str_tz_time─┐ +│ 2023-09-08 19:13:40 │ Europe/Amsterdam │ 21:13:40 │ +│ 2023-09-08 19:13:40 │ Europe/Andorra │ 21:13:40 │ +│ 2023-09-08 19:13:40 │ Europe/Astrakhan │ 23:13:40 │ +│ 2023-09-08 19:13:40 │ Europe/Athens │ 22:13:40 │ +│ 2023-09-08 19:13:40 │ Europe/Belfast │ 20:13:40 │ +│ 2023-09-08 19:13:40 │ Europe/Belgrade │ 21:13:40 │ +│ 2023-09-08 19:13:40 │ Europe/Berlin │ 21:13:40 │ +│ 2023-09-08 19:13:40 │ Europe/Bratislava │ 21:13:40 │ +│ 2023-09-08 19:13:40 │ Europe/Brussels │ 21:13:40 │ +│ 2023-09-08 19:13:40 │ Europe/Bucharest │ 22:13:40 │ +└─────────────────────┴───────────────────┴─────────────┘ ``` **See Also** From f5f741c118525eeb9d31666d6c53f72eb46f8622 Mon Sep 17 00:00:00 2001 From: Denny Crane Date: Fri, 8 Sep 2023 16:15:33 -0300 Subject: [PATCH 237/433] Update type-conversion-functions.md --- .../functions/type-conversion-functions.md | 28 +++++++++---------- 1 file changed, 14 insertions(+), 14 deletions(-) diff --git a/docs/en/sql-reference/functions/type-conversion-functions.md b/docs/en/sql-reference/functions/type-conversion-functions.md index 4d2e6ccd9b5..0df72b5818c 100644 --- a/docs/en/sql-reference/functions/type-conversion-functions.md +++ b/docs/en/sql-reference/functions/type-conversion-functions.md @@ -892,9 +892,9 @@ Query: ``` sql SELECT - time_zone, now() AS ts, - toString(ts, time_zone) + time_zone, + toString(ts, time_zone) AS str_tz_datetime FROM system.time_zones WHERE time_zone LIKE 'Europe%' LIMIT 10 @@ -903,18 +903,18 @@ LIMIT 10 Result: ```response -┌─time_zone─────────┬──────────────────ts─┬─toString(now(), time_zone)─┐ -│ Europe/Amsterdam │ 2023-09-08 18:51:22 │ 2023-09-08 20:51:22 │ -│ Europe/Andorra │ 2023-09-08 18:51:22 │ 2023-09-08 20:51:22 │ -│ Europe/Astrakhan │ 2023-09-08 18:51:22 │ 2023-09-08 22:51:22 │ -│ Europe/Athens │ 2023-09-08 18:51:22 │ 2023-09-08 21:51:22 │ -│ Europe/Belfast │ 2023-09-08 18:51:22 │ 2023-09-08 19:51:22 │ -│ Europe/Belgrade │ 2023-09-08 18:51:22 │ 2023-09-08 20:51:22 │ -│ Europe/Berlin │ 2023-09-08 18:51:22 │ 2023-09-08 20:51:22 │ -│ Europe/Bratislava │ 2023-09-08 18:51:22 │ 2023-09-08 20:51:22 │ -│ Europe/Brussels │ 2023-09-08 18:51:22 │ 2023-09-08 20:51:22 │ -│ Europe/Bucharest │ 2023-09-08 18:51:22 │ 2023-09-08 21:51:22 │ -└───────────────────┴─────────────────────┴────────────────────────────┘ +┌──────────────────ts─┬─time_zone─────────┬─str_tz_datetime─────┐ +│ 2023-09-08 19:14:59 │ Europe/Amsterdam │ 2023-09-08 21:14:59 │ +│ 2023-09-08 19:14:59 │ Europe/Andorra │ 2023-09-08 21:14:59 │ +│ 2023-09-08 19:14:59 │ Europe/Astrakhan │ 2023-09-08 23:14:59 │ +│ 2023-09-08 19:14:59 │ Europe/Athens │ 2023-09-08 22:14:59 │ +│ 2023-09-08 19:14:59 │ Europe/Belfast │ 2023-09-08 20:14:59 │ +│ 2023-09-08 19:14:59 │ Europe/Belgrade │ 2023-09-08 21:14:59 │ +│ 2023-09-08 19:14:59 │ Europe/Berlin │ 2023-09-08 21:14:59 │ +│ 2023-09-08 19:14:59 │ Europe/Bratislava │ 2023-09-08 21:14:59 │ +│ 2023-09-08 19:14:59 │ Europe/Brussels │ 2023-09-08 21:14:59 │ +│ 2023-09-08 19:14:59 │ Europe/Bucharest │ 2023-09-08 22:14:59 │ +└─────────────────────┴───────────────────┴─────────────────────┘ ``` Also see the `toUnixTimestamp` function. From 4df83b09615e824eb514c6f60b9a0d4f72de66a9 Mon Sep 17 00:00:00 2001 From: Alexey Milovidov Date: Fri, 8 Sep 2023 22:36:09 +0200 Subject: [PATCH 238/433] Remove useless line --- tests/config/install.sh | 1 - 1 file changed, 1 deletion(-) diff --git a/tests/config/install.sh b/tests/config/install.sh index e4d17597fbf..b65f2cc8dd1 100755 --- a/tests/config/install.sh +++ b/tests/config/install.sh @@ -21,7 +21,6 @@ ln -sf $SRC_PATH/config.d/text_log.xml $DEST_SERVER_PATH/config.d/ ln -sf $SRC_PATH/config.d/custom_settings_prefixes.xml $DEST_SERVER_PATH/config.d/ ln -sf $SRC_PATH/config.d/enable_access_control_improvements.xml $DEST_SERVER_PATH/config.d/ ln -sf $SRC_PATH/config.d/macros.xml $DEST_SERVER_PATH/config.d/ -ln -sf $SRC_PATH/config.d/disks.xml $DEST_SERVER_PATH/config.d/ ln -sf $SRC_PATH/config.d/secure_ports.xml $DEST_SERVER_PATH/config.d/ ln -sf $SRC_PATH/config.d/clusters.xml $DEST_SERVER_PATH/config.d/ ln -sf $SRC_PATH/config.d/graphite.xml $DEST_SERVER_PATH/config.d/ From da69a9fda02145347424abf36e1c1be61f20da01 Mon Sep 17 00:00:00 2001 From: Dmitry Novik Date: Fri, 8 Sep 2023 21:41:42 +0000 Subject: [PATCH 239/433] Add a log message on replicated table drop --- src/Storages/StorageReplicatedMergeTree.cpp | 10 ++++++++++ .../0_stateless/01165_lost_part_empty_partition.sql | 2 ++ 2 files changed, 12 insertions(+) diff --git a/src/Storages/StorageReplicatedMergeTree.cpp b/src/Storages/StorageReplicatedMergeTree.cpp index e52b1159b40..aac85b4afc0 100644 --- a/src/Storages/StorageReplicatedMergeTree.cpp +++ b/src/Storages/StorageReplicatedMergeTree.cpp @@ -981,6 +981,16 @@ void StorageReplicatedMergeTree::drop() { /// Session could expire, get it again zookeeper = getZooKeeperIfTableShutDown(); + + auto lost_part_count_path = fs::path(zookeeper_path) / "lost_part_count"; + Coordination::Stat lost_part_count_stat; + String lost_part_count_str; + if (zookeeper->tryGet(lost_part_count_path, lost_part_count_str, &lost_part_count_stat)) + { + UInt64 lost_part_count = lost_part_count_str.empty() ? 0 : parse(lost_part_count_str); + if (lost_part_count > 0) + LOG_DEBUG(log, "Dropping table with non-zero lost_part_count equal to {}", lost_part_count); + } dropReplica(zookeeper, zookeeper_path, replica_name, log, getSettings(), &has_metadata_in_zookeeper); } } diff --git a/tests/queries/0_stateless/01165_lost_part_empty_partition.sql b/tests/queries/0_stateless/01165_lost_part_empty_partition.sql index 924798b0050..8486a6ac2e6 100644 --- a/tests/queries/0_stateless/01165_lost_part_empty_partition.sql +++ b/tests/queries/0_stateless/01165_lost_part_empty_partition.sql @@ -12,6 +12,7 @@ drop table rmt1; system sync replica rmt2; select lost_part_count from system.replicas where database = currentDatabase() and table = 'rmt2'; drop table rmt2; +select count() from system.text_log where logger_name like '%' || currentDatabase() || '%' and message ilike '%table with non-zero lost_part_count equal to%'; create table rmt1 (d DateTime, n int) engine=ReplicatedMergeTree('/test/01165/{database}/rmt', '1') order by n partition by tuple(); @@ -24,6 +25,7 @@ drop table rmt1; system sync replica rmt2; select lost_part_count from system.replicas where database = currentDatabase() and table = 'rmt2'; drop table rmt2; +select count() from system.text_log where logger_name like '%' || currentDatabase() || '%' and message ilike '%table with non-zero lost_part_count equal to%'; create table rmt1 (n UInt8, m Int32, d Date, t DateTime) engine=ReplicatedMergeTree('/test/01165/{database}/rmt', '1') order by n partition by (n, m, d, t); From 7a396139dfcff29c5e9d461fbc283d69d02cb784 Mon Sep 17 00:00:00 2001 From: Igor Nikonov Date: Fri, 8 Sep 2023 21:53:38 +0000 Subject: [PATCH 240/433] Cleanup: unnecessary SelectQueryInfo usage around distributed --- .../ClusterProxy/executeQuery.cpp | 13 +++++++------ src/Interpreters/ClusterProxy/executeQuery.h | 5 ++++- src/Storages/StorageDistributed.cpp | 19 +++++++++++-------- src/Storages/StorageDistributed.h | 11 ++++++++--- src/Storages/StorageMergeTree.cpp | 2 +- src/Storages/StorageReplicatedMergeTree.cpp | 2 +- 6 files changed, 32 insertions(+), 20 deletions(-) diff --git a/src/Interpreters/ClusterProxy/executeQuery.cpp b/src/Interpreters/ClusterProxy/executeQuery.cpp index 9dab803c78d..a8a365255d9 100644 --- a/src/Interpreters/ClusterProxy/executeQuery.cpp +++ b/src/Interpreters/ClusterProxy/executeQuery.cpp @@ -38,7 +38,7 @@ ContextMutablePtr updateSettingsForCluster(bool interserver_mode, ContextPtr context, const Settings & settings, const StorageID & main_table, - const SelectQueryInfo * query_info, + const ASTPtr & additional_filter_ast, Poco::Logger * log) { Settings new_settings = settings; @@ -115,11 +115,11 @@ ContextMutablePtr updateSettingsForCluster(bool interserver_mode, /// /// Here we don't try to analyze setting again. In case if query_info->additional_filter_ast is not empty, some filter was applied. /// It's just easier to add this filter for a source table. - if (query_info && query_info->additional_filter_ast) + if (additional_filter_ast) { Tuple tuple; tuple.push_back(main_table.getShortName()); - tuple.push_back(queryToString(query_info->additional_filter_ast)); + tuple.push_back(queryToString(additional_filter_ast)); new_settings.additional_table_filters.value.push_back(std::move(tuple)); } @@ -174,7 +174,8 @@ void executeQuery( std::vector plans; SelectStreamFactory::Shards remote_shards; - auto new_context = updateSettingsForCluster(!query_info.getCluster()->getSecret().empty(), context, settings, main_table, &query_info, log); + auto new_context = updateSettingsForCluster(!not_optimized_cluster->getSecret().empty(), context, settings, + main_table, query_info.additional_filter_ast, log); new_context->increaseDistributedDepth(); size_t shards = query_info.getCluster()->getShardCount(); @@ -269,7 +270,7 @@ void executeQueryWithParallelReplicas( SelectStreamFactory & stream_factory, const ASTPtr & query_ast, ContextPtr context, - const SelectQueryInfo & query_info, + std::shared_ptr storage_limits, const ClusterPtr & not_optimized_cluster) { const auto & settings = context->getSettingsRef(); @@ -333,7 +334,7 @@ void executeQueryWithParallelReplicas( std::move(scalars), std::move(external_tables), &Poco::Logger::get("ReadFromParallelRemoteReplicasStep"), - query_info.storage_limits); + std::move(storage_limits)); query_plan.addStep(std::move(read_from_remote)); } diff --git a/src/Interpreters/ClusterProxy/executeQuery.h b/src/Interpreters/ClusterProxy/executeQuery.h index 511914e99e4..03f6e558a92 100644 --- a/src/Interpreters/ClusterProxy/executeQuery.h +++ b/src/Interpreters/ClusterProxy/executeQuery.h @@ -20,6 +20,9 @@ using ExpressionActionsPtr = std::shared_ptr; struct StorageID; +struct StorageLimits; +using StorageLimitsList = std::list; + namespace ClusterProxy { @@ -66,7 +69,7 @@ void executeQueryWithParallelReplicas( SelectStreamFactory & stream_factory, const ASTPtr & query_ast, ContextPtr context, - const SelectQueryInfo & query_info, + std::shared_ptr storage_limits, const ClusterPtr & not_optimized_cluster); } diff --git a/src/Storages/StorageDistributed.cpp b/src/Storages/StorageDistributed.cpp index a4d3c566967..7c556b1198a 100644 --- a/src/Storages/StorageDistributed.cpp +++ b/src/Storages/StorageDistributed.cpp @@ -75,6 +75,7 @@ #include #include #include +#include #include #include @@ -437,7 +438,9 @@ QueryProcessingStage::Enum StorageDistributed::getQueryProcessingStage( { /// Always calculate optimized cluster here, to avoid conditions during read() /// (Anyway it will be calculated in the read()) - ClusterPtr optimized_cluster = getOptimizedCluster(local_context, storage_snapshot, query_info); + const auto & select = query_info.query->as(); + auto syntax_analyzer_result = query_info.syntax_analyzer_result; + ClusterPtr optimized_cluster = getOptimizedCluster(local_context, storage_snapshot, select, syntax_analyzer_result); if (optimized_cluster) { LOG_DEBUG(log, "Skipping irrelevant shards - the query will be sent to the following shards of the cluster (shard numbers): {}", @@ -1303,8 +1306,8 @@ ClusterPtr StorageDistributed::getCluster() const return owned_cluster ? owned_cluster : getContext()->getCluster(cluster_name); } -ClusterPtr StorageDistributed::getOptimizedCluster( - ContextPtr local_context, const StorageSnapshotPtr & storage_snapshot, const SelectQueryInfo & query_info) const +ClusterPtr StorageDistributed::getOptimizedCluster(ContextPtr local_context, const StorageSnapshotPtr & storage_snapshot, + const ASTSelectQuery & select, const TreeRewriterResultPtr & syntax_analyzer_result) const { ClusterPtr cluster = getCluster(); const Settings & settings = local_context->getSettingsRef(); @@ -1313,7 +1316,7 @@ ClusterPtr StorageDistributed::getOptimizedCluster( if (has_sharding_key && sharding_key_is_usable) { - ClusterPtr optimized = skipUnusedShards(cluster, query_info, storage_snapshot, local_context); + ClusterPtr optimized = skipUnusedShards(cluster, select, syntax_analyzer_result, storage_snapshot, local_context); if (optimized) return optimized; } @@ -1362,16 +1365,16 @@ IColumn::Selector StorageDistributed::createSelector(const ClusterPtr cluster, c /// using constraints from "PREWHERE" and "WHERE" conditions, otherwise returns `nullptr` ClusterPtr StorageDistributed::skipUnusedShards( ClusterPtr cluster, - const SelectQueryInfo & query_info, + const ASTSelectQuery & select, + const TreeRewriterResultPtr & syntax_analyzer_result, const StorageSnapshotPtr & storage_snapshot, ContextPtr local_context) const { - const auto & select = query_info.query->as(); if (!select.prewhere() && !select.where()) return nullptr; /// FIXME: support analyzer - if (!query_info.syntax_analyzer_result) + if (!syntax_analyzer_result) return nullptr; ASTPtr condition_ast; @@ -1380,7 +1383,7 @@ ClusterPtr StorageDistributed::skipUnusedShards( { ASTPtr select_without_join_ptr = select.clone(); ASTSelectQuery select_without_join = select_without_join_ptr->as(); - TreeRewriterResult analyzer_result_without_join = *query_info.syntax_analyzer_result; + TreeRewriterResult analyzer_result_without_join = *syntax_analyzer_result; removeJoin(select_without_join, analyzer_result_without_join, local_context); if (!select_without_join.prewhere() && !select_without_join.where()) diff --git a/src/Storages/StorageDistributed.h b/src/Storages/StorageDistributed.h index 064254f65d4..3f28378eaa8 100644 --- a/src/Storages/StorageDistributed.h +++ b/src/Storages/StorageDistributed.h @@ -28,6 +28,9 @@ using DiskPtr = std::shared_ptr; class ExpressionActions; using ExpressionActionsPtr = std::shared_ptr; +struct TreeRewriterResult; +using TreeRewriterResultPtr = std::shared_ptr; + /** A distributed table that resides on multiple servers. * Uses data from the specified database and tables on each server. * @@ -182,10 +185,12 @@ private: /// Apply the following settings: /// - optimize_skip_unused_shards /// - force_optimize_skip_unused_shards - ClusterPtr getOptimizedCluster(ContextPtr, const StorageSnapshotPtr & storage_snapshot, const SelectQueryInfo & query_info) const; + ClusterPtr getOptimizedCluster(ContextPtr local_context, const StorageSnapshotPtr & storage_snapshot, + const ASTSelectQuery & select, const TreeRewriterResultPtr & syntax_analyzer_result) const; - ClusterPtr skipUnusedShards( - ClusterPtr cluster, const SelectQueryInfo & query_info, const StorageSnapshotPtr & storage_snapshot, ContextPtr context) const; + ClusterPtr skipUnusedShards( + ClusterPtr cluster, const ASTSelectQuery & select, const TreeRewriterResultPtr & syntax_analyzer_result, + const StorageSnapshotPtr & storage_snapshot, ContextPtr context) const; /// This method returns optimal query processing stage. /// diff --git a/src/Storages/StorageMergeTree.cpp b/src/Storages/StorageMergeTree.cpp index 719c75fd5e5..ac2c7baf857 100644 --- a/src/Storages/StorageMergeTree.cpp +++ b/src/Storages/StorageMergeTree.cpp @@ -239,7 +239,7 @@ void StorageMergeTree::read( ClusterProxy::executeQueryWithParallelReplicas( query_plan, getStorageID(), /*remove_table_function_ptr*/ nullptr, select_stream_factory, modified_query_ast, - local_context, query_info, cluster); + local_context, query_info.storage_limits, cluster); } else { diff --git a/src/Storages/StorageReplicatedMergeTree.cpp b/src/Storages/StorageReplicatedMergeTree.cpp index e52b1159b40..f9538f60760 100644 --- a/src/Storages/StorageReplicatedMergeTree.cpp +++ b/src/Storages/StorageReplicatedMergeTree.cpp @@ -5190,7 +5190,7 @@ void StorageReplicatedMergeTree::readParallelReplicasImpl( query_plan, getStorageID(), /* table_func_ptr= */ nullptr, select_stream_factory, modified_query_ast, - local_context, query_info, parallel_replicas_cluster); + local_context, query_info.storage_limits, parallel_replicas_cluster); } void StorageReplicatedMergeTree::readLocalImpl( From 0728df37e628cdfdeea0833e4cf0d4fca24eef3e Mon Sep 17 00:00:00 2001 From: Alexey Milovidov Date: Sat, 9 Sep 2023 01:44:14 +0200 Subject: [PATCH 241/433] Add `instance_type` column to CI Logs and the `checks` table --- docker/test/base/setup_export_logs.sh | 8 ++++---- tests/ci/build_check.py | 5 ++++- tests/ci/clickhouse_helper.py | 16 ++++++++++++---- tests/ci/performance_comparison_check.py | 4 +++- 4 files changed, 23 insertions(+), 10 deletions(-) diff --git a/docker/test/base/setup_export_logs.sh b/docker/test/base/setup_export_logs.sh index f610ccba0f7..0d6c0e368db 100755 --- a/docker/test/base/setup_export_logs.sh +++ b/docker/test/base/setup_export_logs.sh @@ -15,8 +15,8 @@ CLICKHOUSE_CI_LOGS_USER=${CLICKHOUSE_CI_LOGS_USER:-ci} # Pre-configured destination cluster, where to export the data CLICKHOUSE_CI_LOGS_CLUSTER=${CLICKHOUSE_CI_LOGS_CLUSTER:-system_logs_export} -EXTRA_COLUMNS=${EXTRA_COLUMNS:-"pull_request_number UInt32, commit_sha String, check_start_time DateTime, check_name LowCardinality(String), instance_type LowCardinality(String), "} -EXTRA_COLUMNS_EXPRESSION=${EXTRA_COLUMNS_EXPRESSION:-"0 AS pull_request_number, '' AS commit_sha, now() AS check_start_time, '' AS check_name, '' AS instance_type"} +EXTRA_COLUMNS=${EXTRA_COLUMNS:-"pull_request_number UInt32, commit_sha String, check_start_time DateTime, check_name LowCardinality(String), instance_type LowCardinality(String), instance_id String, "} +EXTRA_COLUMNS_EXPRESSION=${EXTRA_COLUMNS_EXPRESSION:-"0 AS pull_request_number, '' AS commit_sha, now() AS check_start_time, '' AS check_name, '' AS instance_type, '' AS instance_id"} EXTRA_ORDER_BY_COLUMNS=${EXTRA_ORDER_BY_COLUMNS:-"check_name, "} function __set_connection_args @@ -125,9 +125,9 @@ function setup_logs_replication echo 'Create %_log tables' clickhouse-client --query "SHOW TABLES FROM system LIKE '%\\_log'" | while read -r table do - # Calculate hash of its structure: + # Calculate hash of its structure. Note: 1 is the version of extra columns - increment it if extra columns are changed: hash=$(clickhouse-client --query " - SELECT sipHash64(groupArray((name, type))) + SELECT sipHash64(1, groupArray((name, type))) FROM (SELECT name, type FROM system.columns WHERE database = 'system' AND table = '$table' ORDER BY position) diff --git a/tests/ci/build_check.py b/tests/ci/build_check.py index 85b2c42100d..498fd836436 100644 --- a/tests/ci/build_check.py +++ b/tests/ci/build_check.py @@ -34,6 +34,7 @@ from clickhouse_helper import ( CiLogsCredentials, prepare_tests_results_for_clickhouse, get_instance_type, + get_instance_id, ) from stopwatch import Stopwatch @@ -364,6 +365,7 @@ def main(): ci_logs_credentials = CiLogsCredentials(Path("/dev/null")) if ci_logs_credentials.host: instance_type = get_instance_type() + instance_id = get_instance_id() query = f"""INSERT INTO build_time_trace ( pull_request_number, @@ -371,6 +373,7 @@ def main(): check_start_time, check_name, instance_type, + instance_id, file, library, time, @@ -386,7 +389,7 @@ def main(): avgMs, args_name ) -SELECT {pr_info.number}, '{pr_info.sha}', '{stopwatch.start_time_str}', '{build_name}', '{instance_type}', * +SELECT {pr_info.number}, '{pr_info.sha}', '{stopwatch.start_time_str}', '{build_name}', '{instance_type}', '{instance_id}', * FROM input(' file String, library String, diff --git a/tests/ci/clickhouse_helper.py b/tests/ci/clickhouse_helper.py index 0d05b811a4f..ff0e2677759 100644 --- a/tests/ci/clickhouse_helper.py +++ b/tests/ci/clickhouse_helper.py @@ -168,9 +168,8 @@ class ClickHouseHelper: return result -# Obtain the machine type from IMDS: -def get_instance_type(): - url = "http://169.254.169.254/latest/meta-data/instance-type" +def _query_imds(path): + url = f"http://169.254.169.254/{path}" for i in range(5): try: response = requests.get(url, timeout=1) @@ -184,6 +183,14 @@ def get_instance_type(): continue return "" +# Obtain the machine type from IMDS: +def get_instance_type(): + return _query_imds("latest/meta-data/instance-type") + +# Obtain the instance id from IMDS: +def get_instance_id(): + return _query_imds("latest/meta-data/instance-id") + def prepare_tests_results_for_clickhouse( pr_info: PRInfo, @@ -222,6 +229,7 @@ def prepare_tests_results_for_clickhouse( head_repo=head_repo, task_url=pr_info.task_url, instance_type=get_instance_type(), + instance_id=get_instance_id(), ) # Always publish a total record for all checks. For checks with individual @@ -288,7 +296,7 @@ class CiLogsCredentials: extra_columns = ( f"{pr_info.number} AS pull_request_number, '{pr_info.sha}' AS commit_sha, " f"toDateTime('{check_start_time}', 'UTC') AS check_start_time, '{check_name}' AS check_name, " - f"'{get_instance_type()}' AS instance_type" + f"'{get_instance_type()}' AS instance_type, '{get_instance_id()}' AS instance_id, " ) return ( f'-e EXTRA_COLUMNS_EXPRESSION="{extra_columns}" ' diff --git a/tests/ci/performance_comparison_check.py b/tests/ci/performance_comparison_check.py index 9cc96c4dbff..909fd32d2b7 100644 --- a/tests/ci/performance_comparison_check.py +++ b/tests/ci/performance_comparison_check.py @@ -20,7 +20,7 @@ from get_robot_token import get_best_robot_token, get_parameter_from_ssm from pr_info import PRInfo from s3_helper import S3Helper from tee_popen import TeePopen -from clickhouse_helper import get_instance_type +from clickhouse_helper import get_instance_type, get_instance_id from stopwatch import Stopwatch IMAGE_NAME = "clickhouse/performance-comparison" @@ -38,11 +38,13 @@ def get_run_command( image, ): instance_type = get_instance_type() + instance_id = get_instance_id() envs = [ f"-e CHECK_START_TIME='{check_start_time}'", f"-e CHECK_NAME='{check_name}'", f"-e INSTANCE_TYPE='{instance_type}'", + f"-e INSTANCE_ID='{instance_id}'", f"-e PR_TO_TEST={pr_to_test}", f"-e SHA_TO_TEST={sha_to_test}", ] From 03a755732aed456166018e94278bc47d5b2fdf1e Mon Sep 17 00:00:00 2001 From: Alexey Milovidov Date: Sat, 9 Sep 2023 03:10:23 +0300 Subject: [PATCH 242/433] Revert "Add settings for real-time updates during query execution" --- docs/en/operations/settings/settings.md | 8 - src/Client/ClientBase.cpp | 36 +- src/Client/ClientBase.h | 15 - src/Core/Settings.h | 3 - src/Interpreters/Aggregator.cpp | 23 -- src/Interpreters/Aggregator.h | 2 - src/Processors/Chunk.cpp | 3 +- .../Executors/CompletedPipelineExecutor.cpp | 4 +- src/Processors/Executors/ExecutingGraph.cpp | 1 + .../Executors/ExecutionThreadContext.h | 13 +- src/Processors/Executors/ExecutorTasks.cpp | 15 +- src/Processors/Executors/ExecutorTasks.h | 2 +- src/Processors/Executors/PipelineExecutor.cpp | 5 +- src/Processors/Executors/PipelineExecutor.h | 5 +- .../PullingAsyncPipelineExecutor.cpp | 7 +- .../Executors/PullingAsyncPipelineExecutor.h | 2 +- .../Executors/PullingPipelineExecutor.cpp | 2 +- .../PushingAsyncPipelineExecutor.cpp | 2 +- .../Executors/PushingPipelineExecutor.cpp | 2 +- src/Processors/Formats/IOutputFormat.cpp | 110 ++---- src/Processors/Formats/IOutputFormat.h | 13 +- .../Formats/Impl/PrettyBlockOutputFormat.cpp | 34 +- .../Formats/Impl/PrettyBlockOutputFormat.h | 6 - .../Impl/PrettyCompactBlockOutputFormat.cpp | 3 +- src/Processors/Formats/LazyOutputFormat.h | 5 +- src/Processors/IProcessor.cpp | 5 - src/Processors/IProcessor.h | 23 -- src/Processors/LimitTransform.cpp | 9 +- src/Processors/LimitTransform.h | 4 - .../QueryPlan/BuildQueryPipelineSettings.cpp | 7 +- .../QueryPlan/BuildQueryPipelineSettings.h | 3 - src/Processors/QueryPlan/QueryPlan.cpp | 5 - src/Processors/QueryPlan/SortingStep.h | 2 - .../AggregatingPartialResultTransform.cpp | 47 --- .../AggregatingPartialResultTransform.h | 29 -- .../Transforms/AggregatingTransform.cpp | 12 - .../Transforms/AggregatingTransform.h | 21 -- .../Transforms/ExpressionTransform.cpp | 6 - .../Transforms/ExpressionTransform.h | 5 - .../LimitPartialResultTransform.cpp | 42 --- .../Transforms/LimitPartialResultTransform.h | 36 -- .../Transforms/LimitsCheckingTransform.cpp | 1 - .../Transforms/LimitsCheckingTransform.h | 2 - .../MergeSortingPartialResultTransform.cpp | 48 --- .../MergeSortingPartialResultTransform.h | 28 -- .../Transforms/MergeSortingTransform.cpp | 12 - .../Transforms/MergeSortingTransform.h | 11 - .../Transforms/PartialResultTransform.cpp | 80 ----- .../Transforms/PartialResultTransform.h | 57 ---- src/QueryPipeline/Pipe.cpp | 204 +++--------- src/QueryPipeline/Pipe.h | 25 +- src/QueryPipeline/QueryPipeline.cpp | 50 +-- src/QueryPipeline/QueryPipeline.h | 7 +- src/QueryPipeline/QueryPipelineBuilder.cpp | 21 +- src/QueryPipeline/QueryPipelineBuilder.h | 6 - src/Server/TCPHandler.cpp | 3 +- .../0_stateless/02010_lc_native.python | 291 +++++++++++++--- .../02210_processors_profile_log.reference | 1 - .../02458_insert_select_progress_tcp.python | 244 ++++++++++++-- .../02750_settings_alias_tcp_protocol.python | 212 +++++++++++- ...rting_result_during_query_execution.python | 93 ------ ...ng_result_during_query_execution.reference | 38 --- ...l_sorting_result_during_query_execution.sh | 8 - ...ating_result_during_query_execution.python | 125 ------- ...ng_result_during_query_execution.reference | 88 ----- ...gregating_result_during_query_execution.sh | 8 - .../queries/0_stateless/helpers/tcp_client.py | 313 ------------------ 67 files changed, 774 insertions(+), 1779 deletions(-) delete mode 100644 src/Processors/Transforms/AggregatingPartialResultTransform.cpp delete mode 100644 src/Processors/Transforms/AggregatingPartialResultTransform.h delete mode 100644 src/Processors/Transforms/LimitPartialResultTransform.cpp delete mode 100644 src/Processors/Transforms/LimitPartialResultTransform.h delete mode 100644 src/Processors/Transforms/MergeSortingPartialResultTransform.cpp delete mode 100644 src/Processors/Transforms/MergeSortingPartialResultTransform.h delete mode 100644 src/Processors/Transforms/PartialResultTransform.cpp delete mode 100644 src/Processors/Transforms/PartialResultTransform.h delete mode 100755 tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python delete mode 100644 tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.reference delete mode 100755 tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.sh delete mode 100644 tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python delete mode 100644 tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.reference delete mode 100755 tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.sh delete mode 100644 tests/queries/0_stateless/helpers/tcp_client.py diff --git a/docs/en/operations/settings/settings.md b/docs/en/operations/settings/settings.md index a1989ab3d13..66b8c923d5c 100644 --- a/docs/en/operations/settings/settings.md +++ b/docs/en/operations/settings/settings.md @@ -4644,14 +4644,6 @@ SELECT toFloat64('1.7091'), toFloat64('1.5008753E7') SETTINGS precise_float_pars └─────────────────────┴──────────────────────────┘ ``` -## partial_result_update_duration_ms - -Interval (in milliseconds) for sending updates with partial data about the result table to the client (in interactive mode) during query execution. Setting to 0 disables partial results. Only supported for single-threaded GROUP BY without key, ORDER BY, LIMIT and OFFSET. - -## max_rows_in_partial_result - -Maximum rows to show in the partial result after every real-time update while the query runs (use partial result limit + OFFSET as a value in case of OFFSET in the query). - ## validate_tcp_client_information {#validate-tcp-client-information} Determines whether validation of client information enabled when query packet is received from a client using a TCP connection. diff --git a/src/Client/ClientBase.cpp b/src/Client/ClientBase.cpp index cc60095884e..859665c64b3 100644 --- a/src/Client/ClientBase.cpp +++ b/src/Client/ClientBase.cpp @@ -441,20 +441,7 @@ void ClientBase::onData(Block & block, ASTPtr parsed_query) if (!block) return; - if (block.rows() == 0 && partial_result_mode == PartialResultMode::Active) - { - partial_result_mode = PartialResultMode::Inactive; - if (is_interactive) - { - progress_indication.clearProgressOutput(*tty_buf); - std::cout << "Full result:" << std::endl; - progress_indication.writeProgress(*tty_buf); - } - } - - if (partial_result_mode == PartialResultMode::Inactive) - processed_rows += block.rows(); - + processed_rows += block.rows(); /// Even if all blocks are empty, we still need to initialize the output stream to write empty resultset. initOutputFormat(block, parsed_query); @@ -464,20 +451,13 @@ void ClientBase::onData(Block & block, ASTPtr parsed_query) if (block.rows() == 0 || (query_fuzzer_runs != 0 && processed_rows >= 100)) return; - if (!is_interactive && partial_result_mode == PartialResultMode::Active) - return; - /// If results are written INTO OUTFILE, we can avoid clearing progress to avoid flicker. if (need_render_progress && tty_buf && (!select_into_file || select_into_file_and_stdout)) progress_indication.clearProgressOutput(*tty_buf); try { - if (partial_result_mode == PartialResultMode::Active) - output_format->writePartialResult(materializeBlock(block)); - else - output_format->write(materializeBlock(block)); - + output_format->write(materializeBlock(block)); written_first_block = true; } catch (const Exception &) @@ -541,9 +521,6 @@ void ClientBase::onProfileInfo(const ProfileInfo & profile_info) void ClientBase::initOutputFormat(const Block & block, ASTPtr parsed_query) try { - if (partial_result_mode == PartialResultMode::NotInit) - partial_result_mode = PartialResultMode::Active; - if (!output_format) { /// Ignore all results when fuzzing as they can be huge. @@ -954,14 +931,6 @@ void ClientBase::processOrdinaryQuery(const String & query_to_execute, ASTPtr pa const auto & settings = global_context->getSettingsRef(); const Int32 signals_before_stop = settings.partial_result_on_first_cancel ? 2 : 1; - bool has_partial_result_setting = settings.partial_result_update_duration_ms.totalMilliseconds() > 0; - - if (has_partial_result_setting) - { - partial_result_mode = PartialResultMode::NotInit; - if (is_interactive) - std::cout << "Partial result:" << std::endl; - } int retries_left = 10; while (retries_left) @@ -1767,7 +1736,6 @@ void ClientBase::processParsedSingleQuery(const String & full_query, const Strin } processed_rows = 0; - partial_result_mode = PartialResultMode::Inactive; written_first_block = false; progress_indication.resetProgress(); profile_events.watch.restart(); diff --git a/src/Client/ClientBase.h b/src/Client/ClientBase.h index e2394259f85..eabc79b7432 100644 --- a/src/Client/ClientBase.h +++ b/src/Client/ClientBase.h @@ -272,21 +272,6 @@ protected: size_t processed_rows = 0; /// How many rows have been read or written. bool print_num_processed_rows = false; /// Whether to print the number of processed rows at - enum class PartialResultMode: UInt8 - { - /// Query doesn't show partial result before the first block with 0 rows. - /// The first block with 0 rows initializes the output table format using its header. - NotInit, - - /// Query shows partial result after the first and before the second block with 0 rows. - /// The second block with 0 rows indicates that that receiving blocks with partial result has been completed and next blocks will be with the full result. - Active, - - /// Query doesn't show partial result at all. - Inactive, - }; - PartialResultMode partial_result_mode = PartialResultMode::Inactive; - bool print_stack_trace = false; /// The last exception that was received from the server. Is used for the /// return code in batch mode. diff --git a/src/Core/Settings.h b/src/Core/Settings.h index 719def2a649..620cc8fd67f 100644 --- a/src/Core/Settings.h +++ b/src/Core/Settings.h @@ -309,9 +309,6 @@ class IColumn; \ M(Bool, partial_result_on_first_cancel, false, "Allows query to return a partial result after cancel.", 0) \ \ - M(Milliseconds, partial_result_update_duration_ms, 0, "Interval (in milliseconds) for sending updates with partial data about the result table to the client (in interactive mode) during query execution. Setting to 0 disables partial results. Only supported for single-threaded GROUP BY without key, ORDER BY, LIMIT and OFFSET.", 0) \ - M(UInt64, max_rows_in_partial_result, 10, "Maximum rows to show in the partial result after every real-time update while the query runs (use partial result limit + OFFSET as a value in case of OFFSET in the query).", 0) \ - \ M(Bool, ignore_on_cluster_for_replicated_udf_queries, false, "Ignore ON CLUSTER clause for replicated UDF management queries.", 0) \ M(Bool, ignore_on_cluster_for_replicated_access_entities_queries, false, "Ignore ON CLUSTER clause for replicated access entities management queries.", 0) \ /** Settings for testing hedged requests */ \ diff --git a/src/Interpreters/Aggregator.cpp b/src/Interpreters/Aggregator.cpp index 23ee097ebff..9249c3ce4ce 100644 --- a/src/Interpreters/Aggregator.cpp +++ b/src/Interpreters/Aggregator.cpp @@ -2272,29 +2272,6 @@ Block Aggregator::prepareBlockAndFillWithoutKey(AggregatedDataVariants & data_va return block; } -Block Aggregator::prepareBlockAndFillWithoutKeySnapshot(AggregatedDataVariants & data_variants) const -{ - size_t rows = 1; - bool final = true; - - auto && out_cols - = prepareOutputBlockColumns(params, aggregate_functions, getHeader(final), data_variants.aggregates_pools, final, rows); - auto && [key_columns, raw_key_columns, aggregate_columns, final_aggregate_columns, aggregate_columns_data] = out_cols; - - AggregatedDataWithoutKey & data = data_variants.without_key; - - /// Always single-thread. It's safe to pass current arena from 'aggregates_pool'. - for (size_t insert_i = 0; insert_i < params.aggregates_size; ++insert_i) - aggregate_functions[insert_i]->insertResultInto( - data + offsets_of_aggregate_states[insert_i], - *final_aggregate_columns[insert_i], - data_variants.aggregates_pool); - - Block block = finalizeBlock(params, getHeader(final), std::move(out_cols), final, rows); - - return block; -} - template Aggregator::ConvertToBlockRes Aggregator::prepareBlockAndFillSingleLevel(AggregatedDataVariants & data_variants, bool final) const diff --git a/src/Interpreters/Aggregator.h b/src/Interpreters/Aggregator.h index 6bfaa76f9b3..4f2c86606c5 100644 --- a/src/Interpreters/Aggregator.h +++ b/src/Interpreters/Aggregator.h @@ -1210,7 +1210,6 @@ private: friend class ConvertingAggregatedToChunksSource; friend class ConvertingAggregatedToChunksWithMergingSource; friend class AggregatingInOrderTransform; - friend class AggregatingPartialResultTransform; /// Data structure of source blocks. Block header; @@ -1392,7 +1391,6 @@ private: std::atomic * is_cancelled = nullptr) const; Block prepareBlockAndFillWithoutKey(AggregatedDataVariants & data_variants, bool final, bool is_overflows) const; - Block prepareBlockAndFillWithoutKeySnapshot(AggregatedDataVariants & data_variants) const; BlocksList prepareBlocksAndFillTwoLevel(AggregatedDataVariants & data_variants, bool final, ThreadPool * thread_pool) const; template diff --git a/src/Processors/Chunk.cpp b/src/Processors/Chunk.cpp index cd442085eca..3839a8963b2 100644 --- a/src/Processors/Chunk.cpp +++ b/src/Processors/Chunk.cpp @@ -14,8 +14,7 @@ namespace ErrorCodes extern const int POSITION_OUT_OF_BOUND; } -Chunk::Chunk(DB::Columns columns_, UInt64 num_rows_) - : columns(std::move(columns_)), num_rows(num_rows_) +Chunk::Chunk(DB::Columns columns_, UInt64 num_rows_) : columns(std::move(columns_)), num_rows(num_rows_) { checkNumRowsIsConsistent(); } diff --git a/src/Processors/Executors/CompletedPipelineExecutor.cpp b/src/Processors/Executors/CompletedPipelineExecutor.cpp index c30586e194e..598a51bf0c7 100644 --- a/src/Processors/Executors/CompletedPipelineExecutor.cpp +++ b/src/Processors/Executors/CompletedPipelineExecutor.cpp @@ -75,7 +75,7 @@ void CompletedPipelineExecutor::execute() if (interactive_timeout_ms) { data = std::make_unique(); - data->executor = std::make_shared(pipeline.processors, pipeline.process_list_element, pipeline.partial_result_duration_ms); + data->executor = std::make_shared(pipeline.processors, pipeline.process_list_element); data->executor->setReadProgressCallback(pipeline.getReadProgressCallback()); /// Avoid passing this to lambda, copy ptr to data instead. @@ -105,7 +105,7 @@ void CompletedPipelineExecutor::execute() } else { - PipelineExecutor executor(pipeline.processors, pipeline.process_list_element, pipeline.partial_result_duration_ms); + PipelineExecutor executor(pipeline.processors, pipeline.process_list_element); executor.setReadProgressCallback(pipeline.getReadProgressCallback()); executor.execute(pipeline.getNumThreads(), pipeline.getConcurrencyControl()); } diff --git a/src/Processors/Executors/ExecutingGraph.cpp b/src/Processors/Executors/ExecutingGraph.cpp index 6a946b4a4b9..27f6a454b24 100644 --- a/src/Processors/Executors/ExecutingGraph.cpp +++ b/src/Processors/Executors/ExecutingGraph.cpp @@ -260,6 +260,7 @@ bool ExecutingGraph::updateNode(uint64_t pid, Queue & queue, Queue & async_queue { pid = updated_processors.top(); updated_processors.pop(); + /// In this method we have ownership on node. auto & node = *nodes[pid]; diff --git a/src/Processors/Executors/ExecutionThreadContext.h b/src/Processors/Executors/ExecutionThreadContext.h index 85788a70771..eb048f8ab09 100644 --- a/src/Processors/Executors/ExecutionThreadContext.h +++ b/src/Processors/Executors/ExecutionThreadContext.h @@ -30,12 +30,6 @@ private: /// Callback for read progress. ReadProgressCallback * read_progress_callback = nullptr; - /// Timer that stops optimization of running local tasks instead of queuing them. - /// It provides local progress for each IProcessor task, allowing the partial result of the request to be always sended to the user. - Stopwatch watch; - /// Time period that limits the maximum allowed duration for optimizing the scheduling of local tasks within the executor - const UInt64 partial_result_duration_ms; - public: #ifndef NDEBUG /// Time for different processing stages. @@ -68,13 +62,8 @@ public: void setException(std::exception_ptr exception_) { exception = exception_; } void rethrowExceptionIfHas(); - bool needWatchRestartForPartialResultProgress() { return partial_result_duration_ms != 0 && partial_result_duration_ms < watch.elapsedMilliseconds(); } - void restartWatch() { watch.restart(); } - - explicit ExecutionThreadContext(size_t thread_number_, bool profile_processors_, bool trace_processors_, ReadProgressCallback * callback, UInt64 partial_result_duration_ms_) + explicit ExecutionThreadContext(size_t thread_number_, bool profile_processors_, bool trace_processors_, ReadProgressCallback * callback) : read_progress_callback(callback) - , watch(CLOCK_MONOTONIC) - , partial_result_duration_ms(partial_result_duration_ms_) , thread_number(thread_number_) , profile_processors(profile_processors_) , trace_processors(trace_processors_) diff --git a/src/Processors/Executors/ExecutorTasks.cpp b/src/Processors/Executors/ExecutorTasks.cpp index 08920592391..e61d225a968 100644 --- a/src/Processors/Executors/ExecutorTasks.cpp +++ b/src/Processors/Executors/ExecutorTasks.cpp @@ -108,15 +108,8 @@ void ExecutorTasks::pushTasks(Queue & queue, Queue & async_queue, ExecutionThrea { context.setTask(nullptr); - /// If sending partial results is allowed and local tasks scheduling optimization is repeated longer than the limit - /// or new task need to send partial result later, skip optimization for this iteration. - /// Otherwise take local task from queue if has one. - if ((!queue.empty() && queue.front()->processor->isPartialResultProcessor()) - || context.needWatchRestartForPartialResultProgress()) - { - context.restartWatch(); - } - else if (!queue.empty() && !context.hasAsyncTasks()) + /// Take local task from queue if has one. + if (!queue.empty() && !context.hasAsyncTasks()) { context.setTask(queue.front()); queue.pop(); @@ -146,7 +139,7 @@ void ExecutorTasks::pushTasks(Queue & queue, Queue & async_queue, ExecutionThrea } } -void ExecutorTasks::init(size_t num_threads_, size_t use_threads_, bool profile_processors, bool trace_processors, ReadProgressCallback * callback, UInt64 partial_result_duration_ms) +void ExecutorTasks::init(size_t num_threads_, size_t use_threads_, bool profile_processors, bool trace_processors, ReadProgressCallback * callback) { num_threads = num_threads_; use_threads = use_threads_; @@ -158,7 +151,7 @@ void ExecutorTasks::init(size_t num_threads_, size_t use_threads_, bool profile_ executor_contexts.reserve(num_threads); for (size_t i = 0; i < num_threads; ++i) - executor_contexts.emplace_back(std::make_unique(i, profile_processors, trace_processors, callback, partial_result_duration_ms)); + executor_contexts.emplace_back(std::make_unique(i, profile_processors, trace_processors, callback)); } } diff --git a/src/Processors/Executors/ExecutorTasks.h b/src/Processors/Executors/ExecutorTasks.h index ab6d5e91411..d35f8de94d1 100644 --- a/src/Processors/Executors/ExecutorTasks.h +++ b/src/Processors/Executors/ExecutorTasks.h @@ -58,7 +58,7 @@ public: void tryGetTask(ExecutionThreadContext & context); void pushTasks(Queue & queue, Queue & async_queue, ExecutionThreadContext & context); - void init(size_t num_threads_, size_t use_threads_, bool profile_processors, bool trace_processors, ReadProgressCallback * callback, UInt64 partial_result_duration_ms); + void init(size_t num_threads_, size_t use_threads_, bool profile_processors, bool trace_processors, ReadProgressCallback * callback); void fill(Queue & queue); void upscale(size_t use_threads_); diff --git a/src/Processors/Executors/PipelineExecutor.cpp b/src/Processors/Executors/PipelineExecutor.cpp index 77779e2cec2..37af391fba3 100644 --- a/src/Processors/Executors/PipelineExecutor.cpp +++ b/src/Processors/Executors/PipelineExecutor.cpp @@ -33,9 +33,8 @@ namespace ErrorCodes } -PipelineExecutor::PipelineExecutor(std::shared_ptr & processors, QueryStatusPtr elem, UInt64 partial_result_duration_ms_) +PipelineExecutor::PipelineExecutor(std::shared_ptr & processors, QueryStatusPtr elem) : process_list_element(std::move(elem)) - , partial_result_duration_ms(partial_result_duration_ms_) { if (process_list_element) { @@ -329,7 +328,7 @@ void PipelineExecutor::initializeExecution(size_t num_threads, bool concurrency_ Queue queue; graph->initializeExecution(queue); - tasks.init(num_threads, use_threads, profile_processors, trace_processors, read_progress_callback.get(), partial_result_duration_ms); + tasks.init(num_threads, use_threads, profile_processors, trace_processors, read_progress_callback.get()); tasks.fill(queue); if (num_threads > 1) diff --git a/src/Processors/Executors/PipelineExecutor.h b/src/Processors/Executors/PipelineExecutor.h index 6cb0e6c4ac1..dee12dad282 100644 --- a/src/Processors/Executors/PipelineExecutor.h +++ b/src/Processors/Executors/PipelineExecutor.h @@ -33,7 +33,7 @@ public: /// During pipeline execution new processors can appear. They will be added to existing set. /// /// Explicit graph representation is built in constructor. Throws if graph is not correct. - explicit PipelineExecutor(std::shared_ptr & processors, QueryStatusPtr elem, UInt64 partial_result_duration_ms_ = 0); + explicit PipelineExecutor(std::shared_ptr & processors, QueryStatusPtr elem); ~PipelineExecutor(); /// Execute pipeline in multiple threads. Must be called once. @@ -90,9 +90,6 @@ private: ReadProgressCallbackPtr read_progress_callback; - /// Duration between sending partial result through the pipeline - const UInt64 partial_result_duration_ms; - using Queue = std::queue; void initializeExecution(size_t num_threads, bool concurrency_control); /// Initialize executor contexts and task_queue. diff --git a/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp b/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp index 95a2022bf93..345bec395b2 100644 --- a/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp +++ b/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp @@ -41,13 +41,12 @@ struct PullingAsyncPipelineExecutor::Data } }; -PullingAsyncPipelineExecutor::PullingAsyncPipelineExecutor(QueryPipeline & pipeline_, bool has_partial_result_setting) : pipeline(pipeline_) +PullingAsyncPipelineExecutor::PullingAsyncPipelineExecutor(QueryPipeline & pipeline_) : pipeline(pipeline_) { if (!pipeline.pulling()) throw Exception(ErrorCodes::LOGICAL_ERROR, "Pipeline for PullingAsyncPipelineExecutor must be pulling"); - lazy_format = std::make_shared(pipeline.output->getHeader(), /*is_partial_result_protocol_active*/ has_partial_result_setting); - + lazy_format = std::make_shared(pipeline.output->getHeader()); pipeline.complete(lazy_format); } @@ -104,7 +103,7 @@ bool PullingAsyncPipelineExecutor::pull(Chunk & chunk, uint64_t milliseconds) if (!data) { data = std::make_unique(); - data->executor = std::make_shared(pipeline.processors, pipeline.process_list_element, pipeline.partial_result_duration_ms); + data->executor = std::make_shared(pipeline.processors, pipeline.process_list_element); data->executor->setReadProgressCallback(pipeline.getReadProgressCallback()); data->lazy_format = lazy_format.get(); diff --git a/src/Processors/Executors/PullingAsyncPipelineExecutor.h b/src/Processors/Executors/PullingAsyncPipelineExecutor.h index 202ecbf281b..361bcc0155c 100644 --- a/src/Processors/Executors/PullingAsyncPipelineExecutor.h +++ b/src/Processors/Executors/PullingAsyncPipelineExecutor.h @@ -21,7 +21,7 @@ struct ProfileInfo; class PullingAsyncPipelineExecutor { public: - explicit PullingAsyncPipelineExecutor(QueryPipeline & pipeline_, bool has_partial_result_setting = false); + explicit PullingAsyncPipelineExecutor(QueryPipeline & pipeline_); ~PullingAsyncPipelineExecutor(); /// Get structure of returned block or chunk. diff --git a/src/Processors/Executors/PullingPipelineExecutor.cpp b/src/Processors/Executors/PullingPipelineExecutor.cpp index f79f15c19bf..cbf73c5cb07 100644 --- a/src/Processors/Executors/PullingPipelineExecutor.cpp +++ b/src/Processors/Executors/PullingPipelineExecutor.cpp @@ -44,7 +44,7 @@ bool PullingPipelineExecutor::pull(Chunk & chunk) { if (!executor) { - executor = std::make_shared(pipeline.processors, pipeline.process_list_element, pipeline.partial_result_duration_ms); + executor = std::make_shared(pipeline.processors, pipeline.process_list_element); executor->setReadProgressCallback(pipeline.getReadProgressCallback()); } diff --git a/src/Processors/Executors/PushingAsyncPipelineExecutor.cpp b/src/Processors/Executors/PushingAsyncPipelineExecutor.cpp index f3ed24e7e96..a816ab9ca7f 100644 --- a/src/Processors/Executors/PushingAsyncPipelineExecutor.cpp +++ b/src/Processors/Executors/PushingAsyncPipelineExecutor.cpp @@ -167,7 +167,7 @@ void PushingAsyncPipelineExecutor::start() started = true; data = std::make_unique(); - data->executor = std::make_shared(pipeline.processors, pipeline.process_list_element, pipeline.partial_result_duration_ms); + data->executor = std::make_shared(pipeline.processors, pipeline.process_list_element); data->executor->setReadProgressCallback(pipeline.getReadProgressCallback()); data->source = pushing_source.get(); diff --git a/src/Processors/Executors/PushingPipelineExecutor.cpp b/src/Processors/Executors/PushingPipelineExecutor.cpp index f2b018792c7..696932932df 100644 --- a/src/Processors/Executors/PushingPipelineExecutor.cpp +++ b/src/Processors/Executors/PushingPipelineExecutor.cpp @@ -87,7 +87,7 @@ void PushingPipelineExecutor::start() return; started = true; - executor = std::make_shared(pipeline.processors, pipeline.process_list_element, pipeline.partial_result_duration_ms); + executor = std::make_shared(pipeline.processors, pipeline.process_list_element); executor->setReadProgressCallback(pipeline.getReadProgressCallback()); if (!executor->executeStep(&input_wait_flag)) diff --git a/src/Processors/Formats/IOutputFormat.cpp b/src/Processors/Formats/IOutputFormat.cpp index e691e32a7bc..88a6fb1e92f 100644 --- a/src/Processors/Formats/IOutputFormat.cpp +++ b/src/Processors/Formats/IOutputFormat.cpp @@ -1,89 +1,40 @@ #include #include -#include namespace DB { -IOutputFormat::IOutputFormat(const Block & header_, WriteBuffer & out_, bool is_partial_result_protocol_active_) - : IProcessor({header_, header_, header_, header_}, {}) - , out(out_) - , is_partial_result_protocol_active(is_partial_result_protocol_active_) +IOutputFormat::IOutputFormat(const Block & header_, WriteBuffer & out_) + : IProcessor({header_, header_, header_}, {}), out(out_) { } -void IOutputFormat::setCurrentChunk(InputPort & input, PortKind kind) -{ - current_chunk = input.pull(true); - current_block_kind = kind; - has_input = true; -} - -IOutputFormat::Status IOutputFormat::prepareMainAndPartialResult() -{ - bool need_data = false; - for (auto kind : {Main, PartialResult}) - { - auto & input = getPort(kind); - - if (input.isFinished()) - continue; - - if (kind == PartialResult && main_input_activated) - { - input.close(); - continue; - } - - input.setNeeded(); - need_data = true; - - if (!input.hasData()) - continue; - - setCurrentChunk(input, kind); - return Status::Ready; - } - - if (need_data) - return Status::NeedData; - - return Status::Finished; -} - -IOutputFormat::Status IOutputFormat::prepareTotalsAndExtremes() -{ - for (auto kind : {Totals, Extremes}) - { - auto & input = getPort(kind); - - if (!input.isConnected() || input.isFinished()) - continue; - - input.setNeeded(); - if (!input.hasData()) - return Status::NeedData; - - setCurrentChunk(input, kind); - return Status::Ready; - } - - return Status::Finished; -} - IOutputFormat::Status IOutputFormat::prepare() { if (has_input) return Status::Ready; - auto status = prepareMainAndPartialResult(); - if (status != Status::Finished) - return status; + for (auto kind : {Main, Totals, Extremes}) + { + auto & input = getPort(kind); - status = prepareTotalsAndExtremes(); - if (status != Status::Finished) - return status; + if (kind != Main && !input.isConnected()) + continue; + + if (input.isFinished()) + continue; + + input.setNeeded(); + + if (!input.hasData()) + return Status::NeedData; + + current_chunk = input.pull(true); + current_block_kind = kind; + has_input = true; + return Status::Ready; + } finished = true; @@ -132,18 +83,8 @@ void IOutputFormat::work() case Main: result_rows += current_chunk.getNumRows(); result_bytes += current_chunk.allocatedBytes(); - if (is_partial_result_protocol_active && !main_input_activated && current_chunk.hasRows()) - { - /// Sending an empty block signals to the client that partial results are terminated, - /// and only data from the main pipeline will be forwarded. - consume(Chunk(current_chunk.cloneEmptyColumns(), 0)); - main_input_activated = true; - } consume(std::move(current_chunk)); break; - case PartialResult: - consumePartialResult(std::move(current_chunk)); - break; case Totals: writeSuffixIfNeeded(); if (auto totals = prepareTotals(std::move(current_chunk))) @@ -178,15 +119,6 @@ void IOutputFormat::write(const Block & block) flush(); } -void IOutputFormat::writePartialResult(const Block & block) -{ - writePrefixIfNeeded(); - consumePartialResult(Chunk(block.getColumns(), block.rows())); - - if (auto_flush) - flush(); -} - void IOutputFormat::finalize() { if (finalized) diff --git a/src/Processors/Formats/IOutputFormat.h b/src/Processors/Formats/IOutputFormat.h index 470d24e9a22..58700a978ff 100644 --- a/src/Processors/Formats/IOutputFormat.h +++ b/src/Processors/Formats/IOutputFormat.h @@ -23,9 +23,9 @@ class WriteBuffer; class IOutputFormat : public IProcessor { public: - enum PortKind { Main = 0, Totals = 1, Extremes = 2, PartialResult = 3 }; + enum PortKind { Main = 0, Totals = 1, Extremes = 2 }; - IOutputFormat(const Block & header_, WriteBuffer & out_, bool is_partial_result_protocol_active_ = false); + IOutputFormat(const Block & header_, WriteBuffer & out_); Status prepare() override; void work() override; @@ -54,7 +54,6 @@ public: /// TODO: separate formats and processors. void write(const Block & block); - void writePartialResult(const Block & block); void finalize(); @@ -119,7 +118,6 @@ protected: virtual void consume(Chunk) = 0; virtual void consumeTotals(Chunk) {} virtual void consumeExtremes(Chunk) {} - virtual void consumePartialResult(Chunk) {} virtual void finalizeImpl() {} virtual void finalizeBuffers() {} virtual void writePrefix() {} @@ -168,7 +166,6 @@ protected: Chunk current_chunk; PortKind current_block_kind = PortKind::Main; - bool main_input_activated = false; bool has_input = false; bool finished = false; bool finalized = false; @@ -183,15 +180,9 @@ protected: Statistics statistics; private: - void setCurrentChunk(InputPort & input, PortKind kind); - IOutputFormat::Status prepareMainAndPartialResult(); - IOutputFormat::Status prepareTotalsAndExtremes(); - size_t rows_read_before = 0; bool are_totals_written = false; - bool is_partial_result_protocol_active = false; - /// Counters for consumed chunks. Are used for QueryLog. size_t result_rows = 0; size_t result_bytes = 0; diff --git a/src/Processors/Formats/Impl/PrettyBlockOutputFormat.cpp b/src/Processors/Formats/Impl/PrettyBlockOutputFormat.cpp index 6fa891297f6..14648e68f94 100644 --- a/src/Processors/Formats/Impl/PrettyBlockOutputFormat.cpp +++ b/src/Processors/Formats/Impl/PrettyBlockOutputFormat.cpp @@ -134,8 +134,7 @@ void PrettyBlockOutputFormat::write(Chunk chunk, PortKind port_kind) { if (total_rows >= format_settings.pretty.max_rows) { - if (port_kind != PortKind::PartialResult) - total_rows += chunk.getNumRows(); + total_rows += chunk.getNumRows(); return; } if (mono_block) @@ -316,8 +315,7 @@ void PrettyBlockOutputFormat::writeChunk(const Chunk & chunk, PortKind port_kind } writeString(bottom_separator_s, out); - if (port_kind != PortKind::PartialResult) - total_rows += num_rows; + total_rows += num_rows; } @@ -390,34 +388,6 @@ void PrettyBlockOutputFormat::consumeExtremes(Chunk chunk) write(std::move(chunk), PortKind::Extremes); } -void PrettyBlockOutputFormat::clearLastLines(size_t lines_number) -{ - /// http://en.wikipedia.org/wiki/ANSI_escape_code - #define MOVE_TO_PREV_LINE "\033[A" - #define CLEAR_TO_END_OF_LINE "\033[K" - - static const char * clear_prev_line = MOVE_TO_PREV_LINE \ - CLEAR_TO_END_OF_LINE; - - /// Move cursor to the beginning of line - writeCString("\r", out); - - for (size_t line = 0; line < lines_number; ++line) - { - writeCString(clear_prev_line, out); - } -} - -void PrettyBlockOutputFormat::consumePartialResult(Chunk chunk) -{ - if (prev_partial_block_rows > 0) - /// number of rows + header line + footer line - clearLastLines(prev_partial_block_rows + 2); - - prev_partial_block_rows = chunk.getNumRows(); - write(std::move(chunk), PortKind::PartialResult); -} - void PrettyBlockOutputFormat::writeMonoChunkIfNeeded() { diff --git a/src/Processors/Formats/Impl/PrettyBlockOutputFormat.h b/src/Processors/Formats/Impl/PrettyBlockOutputFormat.h index 92466dce3ff..dfb23ac63f9 100644 --- a/src/Processors/Formats/Impl/PrettyBlockOutputFormat.h +++ b/src/Processors/Formats/Impl/PrettyBlockOutputFormat.h @@ -28,12 +28,7 @@ protected: void consumeTotals(Chunk) override; void consumeExtremes(Chunk) override; - void clearLastLines(size_t lines_number); - void consumePartialResult(Chunk) override; - size_t total_rows = 0; - size_t prev_partial_block_rows = 0; - size_t row_number_width = 7; // "10000. " const FormatSettings format_settings; @@ -60,7 +55,6 @@ protected: void resetFormatterImpl() override { total_rows = 0; - prev_partial_block_rows = 0; } private: diff --git a/src/Processors/Formats/Impl/PrettyCompactBlockOutputFormat.cpp b/src/Processors/Formats/Impl/PrettyCompactBlockOutputFormat.cpp index 3a04d86b1ad..2ba9ec725e2 100644 --- a/src/Processors/Formats/Impl/PrettyCompactBlockOutputFormat.cpp +++ b/src/Processors/Formats/Impl/PrettyCompactBlockOutputFormat.cpp @@ -194,8 +194,7 @@ void PrettyCompactBlockOutputFormat::writeChunk(const Chunk & chunk, PortKind po writeBottom(max_widths); - if (port_kind != PortKind::PartialResult) - total_rows += num_rows; + total_rows += num_rows; } diff --git a/src/Processors/Formats/LazyOutputFormat.h b/src/Processors/Formats/LazyOutputFormat.h index bbcfdbb7193..9cf609ed2d7 100644 --- a/src/Processors/Formats/LazyOutputFormat.h +++ b/src/Processors/Formats/LazyOutputFormat.h @@ -14,8 +14,8 @@ class LazyOutputFormat : public IOutputFormat { public: - explicit LazyOutputFormat(const Block & header, bool is_partial_result_protocol_active = false) - : IOutputFormat(header, out, is_partial_result_protocol_active), queue(2) {} + explicit LazyOutputFormat(const Block & header) + : IOutputFormat(header, out), queue(2) {} String getName() const override { return "LazyOutputFormat"; } @@ -49,7 +49,6 @@ protected: void consumeTotals(Chunk chunk) override { totals = std::move(chunk); } void consumeExtremes(Chunk chunk) override { extremes = std::move(chunk); } - void consumePartialResult(Chunk chunk) override { consume(std::move(chunk)); } private: diff --git a/src/Processors/IProcessor.cpp b/src/Processors/IProcessor.cpp index 2f294a32531..8b160153733 100644 --- a/src/Processors/IProcessor.cpp +++ b/src/Processors/IProcessor.cpp @@ -40,10 +40,5 @@ std::string IProcessor::statusToName(Status status) UNREACHABLE(); } -ProcessorPtr IProcessor::getPartialResultProcessorPtr(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) -{ - return current_processor->getPartialResultProcessor(current_processor, partial_result_limit, partial_result_duration_ms); -} - } diff --git a/src/Processors/IProcessor.h b/src/Processors/IProcessor.h index 51a0bb1c121..c6bef186877 100644 --- a/src/Processors/IProcessor.h +++ b/src/Processors/IProcessor.h @@ -164,8 +164,6 @@ public: static std::string statusToName(Status status); - static ProcessorPtr getPartialResultProcessorPtr(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms); - /** Method 'prepare' is responsible for all cheap ("instantaneous": O(1) of data volume, no wait) calculations. * * It may access input and output ports, @@ -237,22 +235,6 @@ public: throw Exception(ErrorCodes::NOT_IMPLEMENTED, "Method 'expandPipeline' is not implemented for {} processor", getName()); } - enum class PartialResultStatus - { - /// Processor currently doesn't support work with the partial result pipeline. - NotSupported, - - /// Processor can be skipped in the partial result pipeline. - SkipSupported, - - /// Processor creates a light-weight copy of itself in the partial result pipeline. - /// The copy can create snapshots of the original processor or transform small blocks of data in the same way as the original processor - FullSupported, - }; - - virtual bool isPartialResultProcessor() const { return false; } - virtual PartialResultStatus getPartialResultProcessorSupportStatus() const { return PartialResultStatus::NotSupported; } - /// In case if query was cancelled executor will wait till all processors finish their jobs. /// Generally, there is no reason to check this flag. However, it may be reasonable for long operations (e.g. i/o). bool isCancelled() const { return is_cancelled.load(std::memory_order_acquire); } @@ -387,11 +369,6 @@ public: protected: virtual void onCancel() {} - virtual ProcessorPtr getPartialResultProcessor(const ProcessorPtr & /*current_processor*/, UInt64 /*partial_result_limit*/, UInt64 /*partial_result_duration_ms*/) - { - throw Exception(ErrorCodes::NOT_IMPLEMENTED, "Method 'getPartialResultProcessor' is not implemented for {} processor", getName()); - } - private: /// For: /// - elapsed_us diff --git a/src/Processors/LimitTransform.cpp b/src/Processors/LimitTransform.cpp index b2bf3c28eee..5e24062d67a 100644 --- a/src/Processors/LimitTransform.cpp +++ b/src/Processors/LimitTransform.cpp @@ -1,5 +1,5 @@ #include -#include + namespace DB { @@ -180,6 +180,7 @@ LimitTransform::Status LimitTransform::preparePair(PortsData & data) return Status::NeedData; data.current_chunk = input.pull(true); + auto rows = data.current_chunk.getNumRows(); if (rows_before_limit_at_least && !data.input_port_has_counter) @@ -366,11 +367,5 @@ bool LimitTransform::sortColumnsEqualAt(const ColumnRawPtrs & current_chunk_sort return true; } -ProcessorPtr LimitTransform::getPartialResultProcessor(const ProcessorPtr & /*current_processor*/, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) -{ - const auto & header = inputs.front().getHeader(); - return std::make_shared(header, partial_result_limit, partial_result_duration_ms, limit, offset); -} - } diff --git a/src/Processors/LimitTransform.h b/src/Processors/LimitTransform.h index eac5f9e8d6d..33ff968985f 100644 --- a/src/Processors/LimitTransform.h +++ b/src/Processors/LimitTransform.h @@ -55,8 +55,6 @@ private: ColumnRawPtrs extractSortColumns(const Columns & columns) const; bool sortColumnsEqualAt(const ColumnRawPtrs & current_chunk_sort_columns, UInt64 current_chunk_row_num) const; - ProcessorPtr getPartialResultProcessor(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) override; - public: LimitTransform( const Block & header_, UInt64 limit_, UInt64 offset_, size_t num_streams = 1, @@ -75,8 +73,6 @@ public: void setRowsBeforeLimitCounter(RowsBeforeLimitCounterPtr counter) override { rows_before_limit_at_least.swap(counter); } void setInputPortHasCounter(size_t pos) { ports_data[pos].input_port_has_counter = true; } - - PartialResultStatus getPartialResultProcessorSupportStatus() const override { return PartialResultStatus::FullSupported; } }; } diff --git a/src/Processors/QueryPlan/BuildQueryPipelineSettings.cpp b/src/Processors/QueryPlan/BuildQueryPipelineSettings.cpp index 60ac30389a1..fb3ed7f80fc 100644 --- a/src/Processors/QueryPlan/BuildQueryPipelineSettings.cpp +++ b/src/Processors/QueryPlan/BuildQueryPipelineSettings.cpp @@ -9,12 +9,7 @@ namespace DB BuildQueryPipelineSettings BuildQueryPipelineSettings::fromContext(ContextPtr from) { BuildQueryPipelineSettings settings; - - const auto & context_settings = from->getSettingsRef(); - settings.partial_result_limit = context_settings.max_rows_in_partial_result; - settings.partial_result_duration_ms = context_settings.partial_result_update_duration_ms.totalMilliseconds(); - - settings.actions_settings = ExpressionActionsSettings::fromSettings(context_settings, CompileExpressions::yes); + settings.actions_settings = ExpressionActionsSettings::fromSettings(from->getSettingsRef(), CompileExpressions::yes); settings.process_list_element = from->getProcessListElement(); settings.progress_callback = from->getProgressCallback(); return settings; diff --git a/src/Processors/QueryPlan/BuildQueryPipelineSettings.h b/src/Processors/QueryPlan/BuildQueryPipelineSettings.h index 0410bf925d1..3b5e4e06953 100644 --- a/src/Processors/QueryPlan/BuildQueryPipelineSettings.h +++ b/src/Processors/QueryPlan/BuildQueryPipelineSettings.h @@ -19,9 +19,6 @@ struct BuildQueryPipelineSettings QueryStatusPtr process_list_element; ProgressCallback progress_callback = nullptr; - UInt64 partial_result_limit = 0; - UInt64 partial_result_duration_ms = 0; - const ExpressionActionsSettings & getActionsSettings() const { return actions_settings; } static BuildQueryPipelineSettings fromContext(ContextPtr from); }; diff --git a/src/Processors/QueryPlan/QueryPlan.cpp b/src/Processors/QueryPlan/QueryPlan.cpp index 5d38bfb42c4..8054209c1c3 100644 --- a/src/Processors/QueryPlan/QueryPlan.cpp +++ b/src/Processors/QueryPlan/QueryPlan.cpp @@ -168,8 +168,6 @@ QueryPipelineBuilderPtr QueryPlan::buildQueryPipeline( QueryPipelineBuilderPtr last_pipeline; - bool has_partial_result_setting = build_pipeline_settings.partial_result_duration_ms > 0; - std::stack stack; stack.push(Frame{.node = root}); @@ -196,9 +194,6 @@ QueryPipelineBuilderPtr QueryPlan::buildQueryPipeline( } else stack.push(Frame{.node = frame.node->children[next_child]}); - - if (has_partial_result_setting && last_pipeline && !last_pipeline->isPartialResultActive()) - last_pipeline->activatePartialResult(build_pipeline_settings.partial_result_limit, build_pipeline_settings.partial_result_duration_ms); } last_pipeline->setProgressCallback(build_pipeline_settings.progress_callback); diff --git a/src/Processors/QueryPlan/SortingStep.h b/src/Processors/QueryPlan/SortingStep.h index a72cab05754..371a24ac6f2 100644 --- a/src/Processors/QueryPlan/SortingStep.h +++ b/src/Processors/QueryPlan/SortingStep.h @@ -27,8 +27,6 @@ public: size_t max_bytes_before_external_sort = 0; TemporaryDataOnDiskScopePtr tmp_data = nullptr; size_t min_free_disk_space = 0; - UInt64 partial_result_limit = 0; - UInt64 partial_result_duration_ms = 0; explicit Settings(const Context & context); explicit Settings(size_t max_block_size_); diff --git a/src/Processors/Transforms/AggregatingPartialResultTransform.cpp b/src/Processors/Transforms/AggregatingPartialResultTransform.cpp deleted file mode 100644 index cf8ce72e096..00000000000 --- a/src/Processors/Transforms/AggregatingPartialResultTransform.cpp +++ /dev/null @@ -1,47 +0,0 @@ -#include - -namespace DB -{ - -AggregatingPartialResultTransform::AggregatingPartialResultTransform( - const Block & input_header, const Block & output_header, AggregatingTransformPtr aggregating_transform_, - UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_) - : PartialResultTransform(input_header, output_header, partial_result_limit_, partial_result_duration_ms_) - , aggregating_transform(std::move(aggregating_transform_)) - , transform_aggregator(input_header, aggregating_transform->params->params) - {} - -void AggregatingPartialResultTransform::transformPartialResult(Chunk & chunk) -{ - auto & params = aggregating_transform->params->params; - - bool no_more_keys = false; - AggregatedDataVariants variants; - ColumnRawPtrs key_columns(params.keys_size); - Aggregator::AggregateColumns aggregate_columns(params.aggregates_size); - - const UInt64 num_rows = chunk.getNumRows(); - transform_aggregator.executeOnBlock(chunk.detachColumns(), 0, num_rows, variants, key_columns, aggregate_columns, no_more_keys); - - auto transformed_block = transform_aggregator.convertToBlocks(variants, /*final*/ true, /*max_threads*/ 1).front(); - - chunk = convertToChunk(transformed_block); -} - -PartialResultTransform::ShaphotResult AggregatingPartialResultTransform::getRealProcessorSnapshot() -{ - std::lock_guard lock(aggregating_transform->snapshot_mutex); - if (aggregating_transform->is_generate_initialized) - return {{}, SnaphotStatus::Stopped}; - - if (aggregating_transform->variants.empty()) - return {{}, SnaphotStatus::NotReady}; - - auto & snapshot_aggregator = aggregating_transform->params->aggregator; - auto & snapshot_variants = aggregating_transform->many_data->variants; - auto block = snapshot_aggregator.prepareBlockAndFillWithoutKeySnapshot(*snapshot_variants.at(0)); - - return {convertToChunk(block), SnaphotStatus::Ready}; -} - -} diff --git a/src/Processors/Transforms/AggregatingPartialResultTransform.h b/src/Processors/Transforms/AggregatingPartialResultTransform.h deleted file mode 100644 index f7bac3a5394..00000000000 --- a/src/Processors/Transforms/AggregatingPartialResultTransform.h +++ /dev/null @@ -1,29 +0,0 @@ -#pragma once - -#include -#include -#include - -namespace DB -{ - -class AggregatingPartialResultTransform : public PartialResultTransform -{ -public: - using AggregatingTransformPtr = std::shared_ptr; - - AggregatingPartialResultTransform( - const Block & input_header, const Block & output_header, AggregatingTransformPtr aggregating_transform_, - UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_); - - String getName() const override { return "AggregatingPartialResultTransform"; } - - void transformPartialResult(Chunk & chunk) override; - ShaphotResult getRealProcessorSnapshot() override; - -private: - AggregatingTransformPtr aggregating_transform; - Aggregator transform_aggregator; -}; - -} diff --git a/src/Processors/Transforms/AggregatingTransform.cpp b/src/Processors/Transforms/AggregatingTransform.cpp index b4d2785bed2..4bd000797a6 100644 --- a/src/Processors/Transforms/AggregatingTransform.cpp +++ b/src/Processors/Transforms/AggregatingTransform.cpp @@ -1,4 +1,3 @@ -#include #include #include @@ -658,8 +657,6 @@ void AggregatingTransform::consume(Chunk chunk) src_rows += num_rows; src_bytes += chunk.bytes(); - std::lock_guard lock(snapshot_mutex); - if (params->params.only_merge) { auto block = getInputs().front().getHeader().cloneWithColumns(chunk.detachColumns()); @@ -679,7 +676,6 @@ void AggregatingTransform::initGenerate() if (is_generate_initialized) return; - std::lock_guard lock(snapshot_mutex); is_generate_initialized = true; /// If there was no data, and we aggregate without keys, and we must return single row with the result of empty aggregation. @@ -810,12 +806,4 @@ void AggregatingTransform::initGenerate() } } -ProcessorPtr AggregatingTransform::getPartialResultProcessor(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) -{ - const auto & input_header = inputs.front().getHeader(); - const auto & output_header = outputs.front().getHeader(); - auto aggregating_processor = std::dynamic_pointer_cast(current_processor); - return std::make_shared(input_header, output_header, std::move(aggregating_processor), partial_result_limit, partial_result_duration_ms); -} - } diff --git a/src/Processors/Transforms/AggregatingTransform.h b/src/Processors/Transforms/AggregatingTransform.h index 791cd12326f..38baa4d0394 100644 --- a/src/Processors/Transforms/AggregatingTransform.h +++ b/src/Processors/Transforms/AggregatingTransform.h @@ -170,23 +170,9 @@ public: void work() override; Processors expandPipeline() override; - PartialResultStatus getPartialResultProcessorSupportStatus() const override - { - /// Currently AggregatingPartialResultTransform support only single-thread aggregation without key. - - /// TODO: check that insert results from aggregator.prepareBlockAndFillWithoutKey return values without - /// changing of the aggregator state when aggregation with keys will be supported in AggregatingPartialResultTransform. - bool is_partial_result_supported = params->params.keys_size == 0 /// Aggregation without key. - && many_data->variants.size() == 1; /// Use only one stream for aggregation. - - return is_partial_result_supported ? PartialResultStatus::FullSupported : PartialResultStatus::NotSupported; - } - protected: void consume(Chunk chunk); - ProcessorPtr getPartialResultProcessor(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) override; - private: /// To read the data that was flushed into the temporary data file. Processors processors; @@ -226,13 +212,6 @@ private: bool is_consume_started = false; - friend class AggregatingPartialResultTransform; - /// The mutex protects variables that are used for creating a snapshot of the current processor. - /// The current implementation of AggregatingPartialResultTransform uses the 'is_generate_initialized' variable to check - /// whether the processor has started sending data through the main pipeline, and the corresponding partial result processor should stop creating snapshots. - /// Additionally, the mutex protects the 'params->aggregator' and 'many_data->variants' variables, which are used to get data from them for a snapshot. - std::mutex snapshot_mutex; - void initGenerate(); }; diff --git a/src/Processors/Transforms/ExpressionTransform.cpp b/src/Processors/Transforms/ExpressionTransform.cpp index 78dace56e4e..0d3341b000c 100644 --- a/src/Processors/Transforms/ExpressionTransform.cpp +++ b/src/Processors/Transforms/ExpressionTransform.cpp @@ -25,12 +25,6 @@ void ExpressionTransform::transform(Chunk & chunk) chunk.setColumns(block.getColumns(), num_rows); } -ProcessorPtr ExpressionTransform::getPartialResultProcessor(const ProcessorPtr & /*current_processor*/, UInt64 /*partial_result_limit*/, UInt64 /*partial_result_duration_ms*/) -{ - const auto & header = getInputPort().getHeader(); - return std::make_shared(header, expression); -} - ConvertingTransform::ConvertingTransform(const Block & header_, ExpressionActionsPtr expression_) : ExceptionKeepingTransform(header_, ExpressionTransform::transformHeader(header_, expression_->getActionsDAG())) , expression(std::move(expression_)) diff --git a/src/Processors/Transforms/ExpressionTransform.h b/src/Processors/Transforms/ExpressionTransform.h index 8250f25f0f8..791c7d7ba73 100644 --- a/src/Processors/Transforms/ExpressionTransform.h +++ b/src/Processors/Transforms/ExpressionTransform.h @@ -26,15 +26,10 @@ public: static Block transformHeader(Block header, const ActionsDAG & expression); - PartialResultStatus getPartialResultProcessorSupportStatus() const override { return PartialResultStatus::FullSupported; } - protected: void transform(Chunk & chunk) override; - ProcessorPtr getPartialResultProcessor(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) override; - private: - ExpressionActionsPtr expression; }; diff --git a/src/Processors/Transforms/LimitPartialResultTransform.cpp b/src/Processors/Transforms/LimitPartialResultTransform.cpp deleted file mode 100644 index c9eaa9dc7dd..00000000000 --- a/src/Processors/Transforms/LimitPartialResultTransform.cpp +++ /dev/null @@ -1,42 +0,0 @@ -#include -#include - -namespace DB -{ - -LimitPartialResultTransform::LimitPartialResultTransform( - const Block & header, - UInt64 partial_result_limit_, - UInt64 partial_result_duration_ms_, - UInt64 limit_, - UInt64 offset_) - : PartialResultTransform(header, partial_result_limit_, partial_result_duration_ms_) - , limit(limit_) - , offset(offset_) - {} - -void LimitPartialResultTransform::transformPartialResult(Chunk & chunk) -{ - UInt64 num_rows = chunk.getNumRows(); - if (num_rows < offset || limit == 0) - { - chunk = {}; - return; - } - - UInt64 length = std::min(limit, num_rows - offset); - - /// Check if some rows should be removed - if (length < num_rows) - { - UInt64 num_columns = chunk.getNumColumns(); - auto columns = chunk.detachColumns(); - - for (UInt64 i = 0; i < num_columns; ++i) - columns[i] = columns[i]->cut(offset, length); - - chunk.setColumns(std::move(columns), length); - } -} - -} diff --git a/src/Processors/Transforms/LimitPartialResultTransform.h b/src/Processors/Transforms/LimitPartialResultTransform.h deleted file mode 100644 index 3a0116b624d..00000000000 --- a/src/Processors/Transforms/LimitPartialResultTransform.h +++ /dev/null @@ -1,36 +0,0 @@ -#pragma once - -#include - -namespace DB -{ - -class LimitTransform; - -/// Currently support only single thread implementation with one input and one output ports -class LimitPartialResultTransform : public PartialResultTransform -{ -public: - using LimitTransformPtr = std::shared_ptr; - - LimitPartialResultTransform( - const Block & header, - UInt64 partial_result_limit_, - UInt64 partial_result_duration_ms_, - UInt64 limit_, - UInt64 offset_); - - String getName() const override { return "LimitPartialResultTransform"; } - - void transformPartialResult(Chunk & chunk) override; - /// LimitsTransform doesn't have a state which can be snapshoted - ShaphotResult getRealProcessorSnapshot() override { return {{}, SnaphotStatus::Stopped}; } - -private: - UInt64 limit; - UInt64 offset; - - LimitTransformPtr limit_transform; -}; - -} diff --git a/src/Processors/Transforms/LimitsCheckingTransform.cpp b/src/Processors/Transforms/LimitsCheckingTransform.cpp index 0557f3f291e..02d2fef808c 100644 --- a/src/Processors/Transforms/LimitsCheckingTransform.cpp +++ b/src/Processors/Transforms/LimitsCheckingTransform.cpp @@ -1,5 +1,4 @@ #include -#include #include namespace DB diff --git a/src/Processors/Transforms/LimitsCheckingTransform.h b/src/Processors/Transforms/LimitsCheckingTransform.h index eabb988dab6..2f96a17c17b 100644 --- a/src/Processors/Transforms/LimitsCheckingTransform.h +++ b/src/Processors/Transforms/LimitsCheckingTransform.h @@ -33,8 +33,6 @@ public: void setQuota(const std::shared_ptr & quota_) { quota = quota_; } - PartialResultStatus getPartialResultProcessorSupportStatus() const override { return PartialResultStatus::SkipSupported; } - protected: void transform(Chunk & chunk) override; diff --git a/src/Processors/Transforms/MergeSortingPartialResultTransform.cpp b/src/Processors/Transforms/MergeSortingPartialResultTransform.cpp deleted file mode 100644 index e4a2af2cdd8..00000000000 --- a/src/Processors/Transforms/MergeSortingPartialResultTransform.cpp +++ /dev/null @@ -1,48 +0,0 @@ -#include - -namespace DB -{ - -MergeSortingPartialResultTransform::MergeSortingPartialResultTransform( - const Block & header, MergeSortingTransformPtr merge_sorting_transform_, - UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_) - : PartialResultTransform(header, partial_result_limit_, partial_result_duration_ms_) - , merge_sorting_transform(std::move(merge_sorting_transform_)) - {} - -PartialResultTransform::ShaphotResult MergeSortingPartialResultTransform::getRealProcessorSnapshot() -{ - std::lock_guard lock(merge_sorting_transform->snapshot_mutex); - if (merge_sorting_transform->generated_prefix) - return {{}, SnaphotStatus::Stopped}; - - if (merge_sorting_transform->chunks.empty()) - return {{}, SnaphotStatus::NotReady}; - - /// Sort all input data - merge_sorting_transform->remerge(); - /// Add a copy of the first `partial_result_limit` rows to a generated_chunk - /// to send it later as a partial result in the next prepare stage of the current processor - auto generated_columns = merge_sorting_transform->chunks[0].cloneEmptyColumns(); - size_t total_rows = 0; - for (const auto & merged_chunk : merge_sorting_transform->chunks) - { - size_t rows = std::min(merged_chunk.getNumRows(), partial_result_limit - total_rows); - if (rows == 0) - break; - - for (size_t position = 0; position < generated_columns.size(); ++position) - { - auto column = merged_chunk.getColumns()[position]; - generated_columns[position]->insertRangeFrom(*column, 0, rows); - } - - total_rows += rows; - } - - auto partial_result = Chunk(std::move(generated_columns), total_rows, merge_sorting_transform->chunks[0].getChunkInfo()); - merge_sorting_transform->enrichChunkWithConstants(partial_result); - return {std::move(partial_result), SnaphotStatus::Ready}; -} - -} diff --git a/src/Processors/Transforms/MergeSortingPartialResultTransform.h b/src/Processors/Transforms/MergeSortingPartialResultTransform.h deleted file mode 100644 index 781aa8e1265..00000000000 --- a/src/Processors/Transforms/MergeSortingPartialResultTransform.h +++ /dev/null @@ -1,28 +0,0 @@ -#pragma once - -#include -#include - -namespace DB -{ - -class MergeSortingPartialResultTransform : public PartialResultTransform -{ -public: - using MergeSortingTransformPtr = std::shared_ptr; - - MergeSortingPartialResultTransform( - const Block & header, MergeSortingTransformPtr merge_sorting_transform_, - UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_); - - String getName() const override { return "MergeSortingPartialResultTransform"; } - - /// MergeSortingTransform always receives chunks in a sorted state, so transformation is not needed - void transformPartialResult(Chunk & /*chunk*/) override {} - ShaphotResult getRealProcessorSnapshot() override; - -private: - MergeSortingTransformPtr merge_sorting_transform; -}; - -} diff --git a/src/Processors/Transforms/MergeSortingTransform.cpp b/src/Processors/Transforms/MergeSortingTransform.cpp index e801e5e16d5..de77711d129 100644 --- a/src/Processors/Transforms/MergeSortingTransform.cpp +++ b/src/Processors/Transforms/MergeSortingTransform.cpp @@ -1,5 +1,4 @@ #include -#include #include #include #include @@ -137,8 +136,6 @@ void MergeSortingTransform::consume(Chunk chunk) /// If there were only const columns in sort description, then there is no need to sort. /// Return the chunk as is. - std::lock_guard lock(snapshot_mutex); - if (description.empty()) { generated_chunk = std::move(chunk); @@ -216,8 +213,6 @@ void MergeSortingTransform::serialize() void MergeSortingTransform::generate() { - std::lock_guard lock(snapshot_mutex); - if (!generated_prefix) { size_t num_tmp_files = tmp_data ? tmp_data->getStreams().size() : 0; @@ -278,11 +273,4 @@ void MergeSortingTransform::remerge() sum_bytes_in_blocks = new_sum_bytes_in_blocks; } -ProcessorPtr MergeSortingTransform::getPartialResultProcessor(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) -{ - const auto & header = inputs.front().getHeader(); - auto merge_sorting_processor = std::dynamic_pointer_cast(current_processor); - return std::make_shared(header, std::move(merge_sorting_processor), partial_result_limit, partial_result_duration_ms); -} - } diff --git a/src/Processors/Transforms/MergeSortingTransform.h b/src/Processors/Transforms/MergeSortingTransform.h index 67f098b4362..e8c180b6903 100644 --- a/src/Processors/Transforms/MergeSortingTransform.h +++ b/src/Processors/Transforms/MergeSortingTransform.h @@ -33,8 +33,6 @@ public: String getName() const override { return "MergeSortingTransform"; } - PartialResultStatus getPartialResultProcessorSupportStatus() const override { return PartialResultStatus::FullSupported; } - protected: void consume(Chunk chunk) override; void serialize() override; @@ -42,8 +40,6 @@ protected: Processors expandPipeline() override; - ProcessorPtr getPartialResultProcessor(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) override; - private: size_t max_bytes_before_remerge; double remerge_lowered_memory_bytes_ratio; @@ -63,13 +59,6 @@ private: void remerge(); ProcessorPtr external_merging_sorted; - - friend class MergeSortingPartialResultTransform; - /// The mutex protects variables that are used for creating a snapshot of the current processor. - /// The current implementation of MergeSortingPartialResultTransform uses the 'generated_prefix' variable to check - /// whether the processor has started sending data through the main pipeline, and the corresponding partial result processor should stop creating snapshots. - /// Additionally, the mutex protects the 'chunks' variable and all variables in the 'remerge' function, which is used to transition 'chunks' to a sorted state. - std::mutex snapshot_mutex; }; } diff --git a/src/Processors/Transforms/PartialResultTransform.cpp b/src/Processors/Transforms/PartialResultTransform.cpp deleted file mode 100644 index 97ff79dee54..00000000000 --- a/src/Processors/Transforms/PartialResultTransform.cpp +++ /dev/null @@ -1,80 +0,0 @@ -#include - -namespace DB -{ - - -PartialResultTransform::PartialResultTransform(const Block & header, UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_) - : PartialResultTransform(header, header, partial_result_limit_, partial_result_duration_ms_) {} - -PartialResultTransform::PartialResultTransform(const Block & input_header, const Block & output_header, UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_) - : IProcessor({input_header}, {output_header}) - , input(inputs.front()) - , output(outputs.front()) - , partial_result_limit(partial_result_limit_) - , partial_result_duration_ms(partial_result_duration_ms_) - , watch(CLOCK_MONOTONIC) - {} - -IProcessor::Status PartialResultTransform::prepare() -{ - if (output.isFinished()) - { - input.close(); - return Status::Finished; - } - - if (finished_getting_snapshots) - { - output.finish(); - return Status::Finished; - } - - if (!output.canPush()) - { - input.setNotNeeded(); - return Status::PortFull; - } - - /// If input data from previous partial result processor is finished then - /// PartialResultTransform ready to create snapshots and send them as a partial result - if (input.isFinished()) - { - if (partial_result.snapshot_status == SnaphotStatus::Ready) - { - partial_result.snapshot_status = SnaphotStatus::NotReady; - output.push(std::move(partial_result.chunk)); - return Status::PortFull; - } - - return Status::Ready; - } - - input.setNeeded(); - if (!input.hasData()) - return Status::NeedData; - - partial_result.chunk = input.pull(); - transformPartialResult(partial_result.chunk); - if (partial_result.chunk.getNumRows() > 0) - { - output.push(std::move(partial_result.chunk)); - return Status::PortFull; - } - - return Status::NeedData; -} - -void PartialResultTransform::work() -{ - if (partial_result_duration_ms < watch.elapsedMilliseconds()) - { - partial_result = getRealProcessorSnapshot(); - if (partial_result.snapshot_status == SnaphotStatus::Stopped) - finished_getting_snapshots = true; - - watch.restart(); - } -} - -} diff --git a/src/Processors/Transforms/PartialResultTransform.h b/src/Processors/Transforms/PartialResultTransform.h deleted file mode 100644 index 4fe87638f38..00000000000 --- a/src/Processors/Transforms/PartialResultTransform.h +++ /dev/null @@ -1,57 +0,0 @@ -#pragma once - -#include - -namespace DB -{ - -/// Processors of this type are used to construct an auxiliary pipeline with processors corresponding to those in the main pipeline. -/// These processors work in two modes: -/// 1) Creating a snapshot of the corresponding processor from the main pipeline once per partial_result_duration_ms (period in milliseconds), and then sending the snapshot through the partial result pipeline. -/// 2) Transforming small blocks of data in the same way as the original processor and sending the transformed data through the partial result pipeline. -/// All processors of this type rely on the invariant that a new block from the previous processor of the partial result pipeline overwrites information about the previous block of the same previous processor. -class PartialResultTransform : public IProcessor -{ -public: - PartialResultTransform(const Block & header, UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_); - PartialResultTransform(const Block & input_header, const Block & output_header, UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_); - - String getName() const override { return "PartialResultTransform"; } - - Status prepare() override; - void work() override; - - bool isPartialResultProcessor() const override { return true; } - -protected: - enum class SnaphotStatus - { - NotReady, // Waiting for data from the previous partial result processor or awaiting a timer before creating the snapshot. - Ready, // Current partial result processor has received a snapshot from the processor in the main pipeline. - Stopped, // The processor from the main pipeline has started sending data, and the pipeline for partial results should use data from the next processors of the main pipeline. - }; - - struct ShaphotResult - { - Chunk chunk; - SnaphotStatus snapshot_status; - }; - - InputPort & input; - OutputPort & output; - - UInt64 partial_result_limit; - UInt64 partial_result_duration_ms; - - ShaphotResult partial_result = {{}, SnaphotStatus::NotReady}; - - bool finished_getting_snapshots = false; - - virtual void transformPartialResult(Chunk & /*chunk*/) = 0; - virtual ShaphotResult getRealProcessorSnapshot() = 0; // { return {{}, SnaphotStatus::Stopped}; } - -private: - Stopwatch watch; -}; - -} diff --git a/src/QueryPipeline/Pipe.cpp b/src/QueryPipeline/Pipe.cpp index 293d152ea65..91ba01c479f 100644 --- a/src/QueryPipeline/Pipe.cpp +++ b/src/QueryPipeline/Pipe.cpp @@ -12,7 +12,6 @@ #include #include #include -#include namespace DB { @@ -168,9 +167,12 @@ Pipe::Pipe(ProcessorPtr source) { checkSource(*source); + if (collected_processors) + collected_processors->emplace_back(source); + output_ports.push_back(&source->getOutputs().front()); header = output_ports.front()->getHeader(); - addProcessor(std::move(source)); + processors->emplace_back(std::move(source)); max_parallel_streams = 1; } @@ -317,16 +319,6 @@ Pipe Pipe::unitePipes(Pipes pipes, Processors * collected_processors, bool allow res.processors->insert(res.processors->end(), pipe.processors->begin(), pipe.processors->end()); res.output_ports.insert(res.output_ports.end(), pipe.output_ports.begin(), pipe.output_ports.end()); - if (res.isPartialResultActive() && pipe.isPartialResultActive()) - { - res.partial_result_ports.insert( - res.partial_result_ports.end(), - pipe.partial_result_ports.begin(), - pipe.partial_result_ports.end()); - } - else - res.dropPartialResult(); - res.max_parallel_streams += pipe.max_parallel_streams; if (pipe.totals_port) @@ -360,11 +352,11 @@ void Pipe::addSource(ProcessorPtr source) else assertBlocksHaveEqualStructure(header, source_header, "Pipes"); - output_ports.push_back(&source->getOutputs().front()); - if (isPartialResultActive()) - partial_result_ports.push_back(nullptr); + if (collected_processors) + collected_processors->emplace_back(source); - addProcessor(std::move(source)); + output_ports.push_back(&source->getOutputs().front()); + processors->emplace_back(std::move(source)); max_parallel_streams = std::max(max_parallel_streams, output_ports.size()); } @@ -382,9 +374,11 @@ void Pipe::addTotalsSource(ProcessorPtr source) assertBlocksHaveEqualStructure(header, source_header, "Pipes"); - totals_port = &source->getOutputs().front(); + if (collected_processors) + collected_processors->emplace_back(source); - addProcessor(std::move(source)); + totals_port = &source->getOutputs().front(); + processors->emplace_back(std::move(source)); } void Pipe::addExtremesSource(ProcessorPtr source) @@ -400,20 +394,11 @@ void Pipe::addExtremesSource(ProcessorPtr source) assertBlocksHaveEqualStructure(header, source_header, "Pipes"); + if (collected_processors) + collected_processors->emplace_back(source); + extremes_port = &source->getOutputs().front(); - - addProcessor(std::move(source)); -} - -void Pipe::activatePartialResult(UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_) -{ - if (is_partial_result_active) - throw Exception(ErrorCodes::LOGICAL_ERROR, "Partial result for Pipe should be initialized only once"); - - is_partial_result_active = true; - partial_result_limit = partial_result_limit_; - partial_result_duration_ms = partial_result_duration_ms_; - partial_result_ports.assign(output_ports.size(), nullptr); + processors->emplace_back(std::move(source)); } static void dropPort(OutputPort *& port, Processors & processors, Processors * collected_processors) @@ -441,15 +426,6 @@ void Pipe::dropExtremes() dropPort(extremes_port, *processors, collected_processors); } -void Pipe::dropPartialResult() -{ - for (auto & port : partial_result_ports) - dropPort(port, *processors, collected_processors); - - is_partial_result_active = false; - partial_result_ports.clear(); -} - void Pipe::addTransform(ProcessorPtr transform) { addTransform(std::move(transform), static_cast(nullptr), static_cast(nullptr)); @@ -480,8 +456,6 @@ void Pipe::addTransform(ProcessorPtr transform, OutputPort * totals, OutputPort if (extremes) extremes_port = extremes; - addPartialResultTransform(transform); - size_t next_output = 0; for (auto & input : inputs) { @@ -532,7 +506,10 @@ void Pipe::addTransform(ProcessorPtr transform, OutputPort * totals, OutputPort if (extremes_port) assertBlocksHaveEqualStructure(header, extremes_port->getHeader(), "Pipes"); - addProcessor(std::move(transform)); + if (collected_processors) + collected_processors->emplace_back(transform); + + processors->emplace_back(std::move(transform)); max_parallel_streams = std::max(max_parallel_streams, output_ports.size()); } @@ -569,8 +546,6 @@ void Pipe::addTransform(ProcessorPtr transform, InputPort * totals, InputPort * extremes_port = nullptr; } - addPartialResultTransform(transform); - bool found_totals = false; bool found_extremes = false; @@ -620,104 +595,14 @@ void Pipe::addTransform(ProcessorPtr transform, InputPort * totals, InputPort * if (extremes_port) assertBlocksHaveEqualStructure(header, extremes_port->getHeader(), "Pipes"); - addProcessor(std::move(transform)); + if (collected_processors) + collected_processors->emplace_back(transform); + + processors->emplace_back(std::move(transform)); max_parallel_streams = std::max(max_parallel_streams, output_ports.size()); } -void Pipe::addPartialResultSimpleTransform(const ProcessorPtr & transform, size_t partial_result_port_id) -{ - if (isPartialResultActive()) - { - auto & partial_result_port = partial_result_ports[partial_result_port_id]; - auto partial_result_status = transform->getPartialResultProcessorSupportStatus(); - - if (partial_result_status == IProcessor::PartialResultStatus::NotSupported) - dropPort(partial_result_port, *processors, collected_processors); - - if (partial_result_status != IProcessor::PartialResultStatus::FullSupported) - return; - - auto partial_result_transform = IProcessor::getPartialResultProcessorPtr(transform, partial_result_limit, partial_result_duration_ms); - - connectPartialResultPort(partial_result_port, partial_result_transform->getInputs().front()); - - partial_result_port = &partial_result_transform->getOutputs().front(); - - addProcessor(std::move(partial_result_transform)); - } -} - -void Pipe::addPartialResultTransform(const ProcessorPtr & transform) -{ - if (isPartialResultActive()) - { - size_t new_outputs_size = transform->getOutputs().size(); - auto partial_result_status = transform->getPartialResultProcessorSupportStatus(); - - if (partial_result_status == IProcessor::PartialResultStatus::SkipSupported && new_outputs_size != partial_result_ports.size()) - throw Exception( - ErrorCodes::LOGICAL_ERROR, - "Cannot skip transform {} in the partial result part of the Pipe because it has {} output ports, but the partial result part expects {} output ports", - transform->getName(), - new_outputs_size, - partial_result_ports.size()); - - if (partial_result_status == IProcessor::PartialResultStatus::NotSupported) - { - for (auto & partial_result_port : partial_result_ports) - dropPort(partial_result_port, *processors, collected_processors); - - partial_result_ports.assign(new_outputs_size, nullptr); - } - - if (partial_result_status != IProcessor::PartialResultStatus::FullSupported) - return; - - auto partial_result_transform = IProcessor::getPartialResultProcessorPtr(transform, partial_result_limit, partial_result_duration_ms); - auto & inputs = partial_result_transform->getInputs(); - - if (inputs.size() != partial_result_ports.size()) - throw Exception( - ErrorCodes::LOGICAL_ERROR, - "Cannot add partial result transform {} to Pipe because it has {} input ports, but {} expected", - partial_result_transform->getName(), - inputs.size(), - partial_result_ports.size()); - - size_t next_port = 0; - for (auto & input : inputs) - { - connectPartialResultPort(partial_result_ports[next_port], input); - ++next_port; - } - - partial_result_ports.assign(new_outputs_size, nullptr); - - next_port = 0; - for (auto & new_partial_result_port : partial_result_transform->getOutputs()) - { - partial_result_ports[next_port] = &new_partial_result_port; - ++next_port; - } - - addProcessor(std::move(partial_result_transform)); - } -} - -void Pipe::connectPartialResultPort(OutputPort * partial_result_port, InputPort & partial_result_transform_port) -{ - if (partial_result_port == nullptr) - { - auto source = std::make_shared(getHeader()); - partial_result_port = &source->getPort(); - - addProcessor(std::move(source)); - } - - connect(*partial_result_port, partial_result_transform_port); -} - void Pipe::addSimpleTransform(const ProcessorGetterWithStreamKind & getter) { if (output_ports.empty()) @@ -725,7 +610,7 @@ void Pipe::addSimpleTransform(const ProcessorGetterWithStreamKind & getter) Block new_header; - auto add_transform = [&](OutputPort *& port, size_t partial_result_port_id, StreamType stream_type) + auto add_transform = [&](OutputPort *& port, StreamType stream_type) { if (!port) return; @@ -761,22 +646,19 @@ void Pipe::addSimpleTransform(const ProcessorGetterWithStreamKind & getter) { connect(*port, transform->getInputs().front()); port = &transform->getOutputs().front(); - if (stream_type == StreamType::Main) - addPartialResultSimpleTransform(transform, partial_result_port_id); - addProcessor(std::move(transform)); + if (collected_processors) + collected_processors->emplace_back(transform); + + processors->emplace_back(std::move(transform)); } }; - size_t partial_result_port_id = 0; for (auto & port : output_ports) - { - add_transform(port, partial_result_port_id, StreamType::Main); - ++partial_result_port_id; - } + add_transform(port, StreamType::Main); - add_transform(totals_port, 0, StreamType::Totals); - add_transform(extremes_port, 0, StreamType::Extremes); + add_transform(totals_port, StreamType::Totals); + add_transform(extremes_port, StreamType::Extremes); header = std::move(new_header); } @@ -797,7 +679,6 @@ void Pipe::addChains(std::vector chains) dropTotals(); dropExtremes(); - dropPartialResult(); size_t max_parallel_streams_for_chains = 0; @@ -816,21 +697,18 @@ void Pipe::addChains(std::vector chains) auto added_processors = Chain::getProcessors(std::move(chains[i])); for (auto & transform : added_processors) - addProcessor(std::move(transform)); + { + if (collected_processors) + collected_processors->emplace_back(transform); + + processors->emplace_back(std::move(transform)); + } } header = std::move(new_header); max_parallel_streams = std::max(max_parallel_streams, max_parallel_streams_for_chains); } -void Pipe::addProcessor(ProcessorPtr processor) -{ - if (collected_processors) - collected_processors->emplace_back(processor); - - processors->emplace_back(std::move(processor)); -} - void Pipe::resize(size_t num_streams, bool force, bool strict) { if (output_ports.empty()) @@ -891,9 +769,6 @@ void Pipe::setSinks(const Pipe::ProcessorGetterWithStreamKind & getter) add_transform(totals_port, StreamType::Totals); add_transform(extremes_port, StreamType::Extremes); - for (auto & port : partial_result_ports) - add_transform(port, StreamType::PartialResult); - output_ports.clear(); header.clear(); } @@ -903,9 +778,6 @@ void Pipe::transform(const Transformer & transformer, bool check_ports) if (output_ports.empty()) throw Exception(ErrorCodes::LOGICAL_ERROR, "Cannot transform empty Pipe"); - /// TODO: Add functionality to work with partial result ports in transformer. - dropPartialResult(); - auto new_processors = transformer(output_ports); /// Create hash table with new processors. diff --git a/src/QueryPipeline/Pipe.h b/src/QueryPipeline/Pipe.h index 70e933bcfd2..09931e38578 100644 --- a/src/QueryPipeline/Pipe.h +++ b/src/QueryPipeline/Pipe.h @@ -48,9 +48,6 @@ public: OutputPort * getOutputPort(size_t pos) const { return output_ports[pos]; } OutputPort * getTotalsPort() const { return totals_port; } OutputPort * getExtremesPort() const { return extremes_port; } - OutputPort * getPartialResultPort(size_t pos) const { return partial_result_ports.empty() ? nullptr : partial_result_ports[pos]; } - - bool isPartialResultActive() { return is_partial_result_active; } /// Add processor to list, add it output ports to output_ports. /// Processor shouldn't have input ports, output ports shouldn't be connected. @@ -61,13 +58,9 @@ public: void addTotalsSource(ProcessorPtr source); void addExtremesSource(ProcessorPtr source); - /// Activate sending partial result during main pipeline execution - void activatePartialResult(UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_); - - /// Drop totals, extremes and partial result (create NullSink for them). + /// Drop totals and extremes (create NullSink for them). void dropTotals(); void dropExtremes(); - void dropPartialResult(); /// Add processor to list. It should have size() input ports with compatible header. /// Output ports should have same headers. @@ -76,16 +69,11 @@ public: void addTransform(ProcessorPtr transform, OutputPort * totals, OutputPort * extremes); void addTransform(ProcessorPtr transform, InputPort * totals, InputPort * extremes); - void addPartialResultTransform(const ProcessorPtr & transform); - void addPartialResultSimpleTransform(const ProcessorPtr & transform, size_t partial_result_port_id); - void connectPartialResultPort(OutputPort * partial_result_port, InputPort & partial_result_transform_port); - enum class StreamType { Main = 0, /// Stream for query data. There may be several streams of this type. Totals, /// Stream for totals. No more than one. Extremes, /// Stream for extremes. No more than one. - PartialResult, /// Stream for partial result data. There may be several streams of this type. }; using ProcessorGetter = std::function; @@ -121,17 +109,10 @@ private: Block header; std::shared_ptr processors; - /// If the variable is true, then each time a processor is added pipe will try - /// to add processor which will send partial result from original processor - bool is_partial_result_active = false; - UInt64 partial_result_limit = 0; - UInt64 partial_result_duration_ms = 0; - - /// Output ports. Totals, extremes and partial results are allowed to be empty. + /// Output ports. Totals and extremes are allowed to be empty. OutputPortRawPtrs output_ports; OutputPort * totals_port = nullptr; OutputPort * extremes_port = nullptr; - OutputPortRawPtrs partial_result_ports; /// It is the max number of processors which can be executed in parallel for each step. /// Usually, it's the same as the number of output ports. @@ -147,8 +128,6 @@ private: static Pipe unitePipes(Pipes pipes, Processors * collected_processors, bool allow_empty_header); void setSinks(const Pipe::ProcessorGetterWithStreamKind & getter); - void addProcessor(ProcessorPtr processor); - friend class QueryPipelineBuilder; friend class QueryPipeline; }; diff --git a/src/QueryPipeline/QueryPipeline.cpp b/src/QueryPipeline/QueryPipeline.cpp index 9a836f68da1..128972b8ff0 100644 --- a/src/QueryPipeline/QueryPipeline.cpp +++ b/src/QueryPipeline/QueryPipeline.cpp @@ -66,8 +66,7 @@ static void checkPulling( Processors & processors, OutputPort * output, OutputPort * totals, - OutputPort * extremes, - OutputPort * partial_result) + OutputPort * extremes) { if (!output || output->isConnected()) throw Exception( @@ -84,15 +83,9 @@ static void checkPulling( ErrorCodes::LOGICAL_ERROR, "Cannot create pulling QueryPipeline because its extremes port is connected"); - if (partial_result && partial_result->isConnected()) - throw Exception( - ErrorCodes::LOGICAL_ERROR, - "Cannot create pulling QueryPipeline because its partial_result port is connected"); - bool found_output = false; bool found_totals = false; bool found_extremes = false; - bool found_partial_result = false; for (const auto & processor : processors) { for (const auto & in : processor->getInputs()) @@ -106,8 +99,6 @@ static void checkPulling( found_totals = true; else if (extremes && &out == extremes) found_extremes = true; - else if (partial_result && &out == partial_result) - found_partial_result = true; else checkOutput(out, processor); } @@ -125,10 +116,6 @@ static void checkPulling( throw Exception( ErrorCodes::LOGICAL_ERROR, "Cannot create pulling QueryPipeline because its extremes port does not belong to any processor"); - if (partial_result && !found_partial_result) - throw Exception( - ErrorCodes::LOGICAL_ERROR, - "Cannot create pulling QueryPipeline because its partial result port does not belong to any processor"); } static void checkCompleted(Processors & processors) @@ -331,20 +318,17 @@ QueryPipeline::QueryPipeline( std::shared_ptr processors_, OutputPort * output_, OutputPort * totals_, - OutputPort * extremes_, - OutputPort * partial_result_) + OutputPort * extremes_) : resources(std::move(resources_)) , processors(std::move(processors_)) , output(output_) , totals(totals_) , extremes(extremes_) - , partial_result(partial_result_) { - checkPulling(*processors, output, totals, extremes, partial_result); + checkPulling(*processors, output, totals, extremes); } QueryPipeline::QueryPipeline(Pipe pipe) - : partial_result_duration_ms(pipe.partial_result_duration_ms) { if (pipe.numOutputPorts() > 0) { @@ -352,10 +336,9 @@ QueryPipeline::QueryPipeline(Pipe pipe) output = pipe.getOutputPort(0); totals = pipe.getTotalsPort(); extremes = pipe.getExtremesPort(); - partial_result = pipe.getPartialResultPort(0); processors = std::move(pipe.processors); - checkPulling(*processors, output, totals, extremes, partial_result); + checkPulling(*processors, output, totals, extremes); } else { @@ -387,7 +370,6 @@ QueryPipeline::QueryPipeline(std::shared_ptr format) auto & format_main = format->getPort(IOutputFormat::PortKind::Main); auto & format_totals = format->getPort(IOutputFormat::PortKind::Totals); auto & format_extremes = format->getPort(IOutputFormat::PortKind::Extremes); - auto & format_partial_result = format->getPort(IOutputFormat::PortKind::PartialResult); if (!totals) { @@ -403,21 +385,12 @@ QueryPipeline::QueryPipeline(std::shared_ptr format) processors->emplace_back(std::move(source)); } - if (!partial_result) - { - auto source = std::make_shared(format_partial_result.getHeader()); - partial_result = &source->getPort(); - processors->emplace_back(std::move(source)); - } - connect(*totals, format_totals); connect(*extremes, format_extremes); - connect(*partial_result, format_partial_result); input = &format_main; totals = nullptr; extremes = nullptr; - partial_result = nullptr; output_format = format.get(); @@ -445,7 +418,6 @@ void QueryPipeline::complete(std::shared_ptr sink) drop(totals, *processors); drop(extremes, *processors); - drop(partial_result, *processors); connect(*output, sink->getPort()); processors->emplace_back(std::move(sink)); @@ -461,7 +433,6 @@ void QueryPipeline::complete(Chain chain) drop(totals, *processors); drop(extremes, *processors); - drop(partial_result, *processors); processors->reserve(processors->size() + chain.getProcessors().size() + 1); for (auto processor : chain.getProcessors()) @@ -487,7 +458,6 @@ void QueryPipeline::complete(Pipe pipe) pipe.resize(1); pipe.dropExtremes(); pipe.dropTotals(); - pipe.dropPartialResult(); connect(*pipe.getOutputPort(0), *input); input = nullptr; @@ -516,13 +486,11 @@ void QueryPipeline::complete(std::shared_ptr format) addMaterializing(output, *processors); addMaterializing(totals, *processors); addMaterializing(extremes, *processors); - addMaterializing(partial_result, *processors); } auto & format_main = format->getPort(IOutputFormat::PortKind::Main); auto & format_totals = format->getPort(IOutputFormat::PortKind::Totals); auto & format_extremes = format->getPort(IOutputFormat::PortKind::Extremes); - auto & format_partial_result = format->getPort(IOutputFormat::PortKind::PartialResult); if (!totals) { @@ -538,22 +506,13 @@ void QueryPipeline::complete(std::shared_ptr format) processors->emplace_back(std::move(source)); } - if (!partial_result) - { - auto source = std::make_shared(format_partial_result.getHeader()); - partial_result = &source->getPort(); - processors->emplace_back(std::move(source)); - } - connect(*output, format_main); connect(*totals, format_totals); connect(*extremes, format_extremes); - connect(*partial_result, format_partial_result); output = nullptr; totals = nullptr; extremes = nullptr; - partial_result = nullptr; initRowsBeforeLimit(format.get()); output_format = format.get(); @@ -725,7 +684,6 @@ void QueryPipeline::convertStructureTo(const ColumnsWithTypeAndName & columns) addExpression(output, actions, *processors); addExpression(totals, actions, *processors); addExpression(extremes, actions, *processors); - addExpression(partial_result, actions, *processors); } std::unique_ptr QueryPipeline::getReadProgressCallback() const diff --git a/src/QueryPipeline/QueryPipeline.h b/src/QueryPipeline/QueryPipeline.h index 20e58bc0f59..f14cf61aac2 100644 --- a/src/QueryPipeline/QueryPipeline.h +++ b/src/QueryPipeline/QueryPipeline.h @@ -75,8 +75,7 @@ public: std::shared_ptr processors_, OutputPort * output_, OutputPort * totals_ = nullptr, - OutputPort * extremes_ = nullptr, - OutputPort * partial_result_ = nullptr); + OutputPort * extremes_ = nullptr); bool initialized() const { return !processors->empty(); } /// When initialized, exactly one of the following is true. @@ -155,7 +154,6 @@ private: OutputPort * output = nullptr; OutputPort * totals = nullptr; OutputPort * extremes = nullptr; - OutputPort * partial_result = nullptr; QueryStatusPtr process_list_element; @@ -164,9 +162,6 @@ private: size_t num_threads = 0; bool concurrency_control = false; - UInt64 partial_result_limit = 0; - UInt64 partial_result_duration_ms = 0; - friend class PushingPipelineExecutor; friend class PullingPipelineExecutor; friend class PushingAsyncPipelineExecutor; diff --git a/src/QueryPipeline/QueryPipelineBuilder.cpp b/src/QueryPipeline/QueryPipelineBuilder.cpp index 90f5ee364f3..f9726339872 100644 --- a/src/QueryPipeline/QueryPipelineBuilder.cpp +++ b/src/QueryPipeline/QueryPipelineBuilder.cpp @@ -110,16 +110,6 @@ void QueryPipelineBuilder::init(QueryPipeline & pipeline) pipe.header = {}; } - if (pipeline.partial_result) - { - /// Set partial result ports only after activation because when activated, it is set to nullptr - pipe.activatePartialResult(pipeline.partial_result_limit, pipeline.partial_result_duration_ms); - pipe.partial_result_ports = {pipeline.partial_result}; - } - - if (!pipeline.partial_result) - pipe.dropPartialResult(); - pipe.totals_port = pipeline.totals; pipe.extremes_port = pipeline.extremes; pipe.max_parallel_streams = pipeline.num_threads; @@ -362,10 +352,6 @@ std::unique_ptr QueryPipelineBuilder::joinPipelinesYShaped left->checkInitializedAndNotCompleted(); right->checkInitializedAndNotCompleted(); - /// TODO: Support joining of partial results from different pipelines. - left->pipe.dropPartialResult(); - right->pipe.dropPartialResult(); - left->pipe.dropExtremes(); right->pipe.dropExtremes(); if (left->getNumStreams() != 1 || right->getNumStreams() != 1) @@ -378,7 +364,6 @@ std::unique_ptr QueryPipelineBuilder::joinPipelinesYShaped auto joining = std::make_shared(join, inputs, out_header, max_block_size); - /// TODO: Support partial results in merge pipelines after joining support above. return mergePipelines(std::move(left), std::move(right), std::move(joining), collected_processors); } @@ -399,10 +384,6 @@ std::unique_ptr QueryPipelineBuilder::joinPipelinesRightLe left->pipe.dropExtremes(); right->pipe.dropExtremes(); - /// TODO: Support joining of partial results from different pipelines. - left->pipe.dropPartialResult(); - right->pipe.dropPartialResult(); - left->pipe.collected_processors = collected_processors; /// Collect the NEW processors for the right pipeline. @@ -653,7 +634,7 @@ PipelineExecutorPtr QueryPipelineBuilder::execute() if (!isCompleted()) throw Exception(ErrorCodes::LOGICAL_ERROR, "Cannot execute pipeline because it is not completed"); - return std::make_shared(pipe.processors, process_list_element, pipe.partial_result_duration_ms); + return std::make_shared(pipe.processors, process_list_element); } Pipe QueryPipelineBuilder::getPipe(QueryPipelineBuilder pipeline, QueryPlanResourceHolder & resources) diff --git a/src/QueryPipeline/QueryPipelineBuilder.h b/src/QueryPipeline/QueryPipelineBuilder.h index 612e7b1652f..5d273df7068 100644 --- a/src/QueryPipeline/QueryPipelineBuilder.h +++ b/src/QueryPipeline/QueryPipelineBuilder.h @@ -85,12 +85,6 @@ public: /// Pipeline will be completed after this transformation. void setSinks(const Pipe::ProcessorGetterWithStreamKind & getter); - /// Activate building separate pipeline for sending partial result. - void activatePartialResult(UInt64 partial_result_limit, UInt64 partial_result_duration_ms) { pipe.activatePartialResult(partial_result_limit, partial_result_duration_ms); } - - /// Check if building of a pipeline for sending partial result active. - bool isPartialResultActive() { return pipe.isPartialResultActive(); } - /// Add totals which returns one chunk with single row with defaults. void addDefaultTotals(); diff --git a/src/Server/TCPHandler.cpp b/src/Server/TCPHandler.cpp index c687a6064b4..96c449b7e17 100644 --- a/src/Server/TCPHandler.cpp +++ b/src/Server/TCPHandler.cpp @@ -885,8 +885,7 @@ void TCPHandler::processOrdinaryQueryWithProcessors() std::unique_lock progress_lock(task_callback_mutex, std::defer_lock); { - bool has_partial_result_setting = query_context->getSettingsRef().partial_result_update_duration_ms.totalMilliseconds() > 0; - PullingAsyncPipelineExecutor executor(pipeline, has_partial_result_setting); + PullingAsyncPipelineExecutor executor(pipeline); CurrentMetrics::Increment query_thread_metric_increment{CurrentMetrics::QueryThread}; Block block; diff --git a/tests/queries/0_stateless/02010_lc_native.python b/tests/queries/0_stateless/02010_lc_native.python index 219fdf04472..6c4220855c8 100755 --- a/tests/queries/0_stateless/02010_lc_native.python +++ b/tests/queries/0_stateless/02010_lc_native.python @@ -1,33 +1,227 @@ #!/usr/bin/env python3 # -*- coding: utf-8 -*- +import socket import os -import sys +import uuid -CURDIR = os.path.dirname(os.path.realpath(__file__)) -sys.path.insert(0, os.path.join(CURDIR, "helpers")) +CLICKHOUSE_HOST = os.environ.get("CLICKHOUSE_HOST", "127.0.0.1") +CLICKHOUSE_PORT = int(os.environ.get("CLICKHOUSE_PORT_TCP", "900000")) +CLICKHOUSE_DATABASE = os.environ.get("CLICKHOUSE_DATABASE", "default") +CLIENT_NAME = "simple native protocol" -from tcp_client import ( - TCPClient, - CLICKHOUSE_DATABASE, - writeVarUInt, - writeStringBinary, - serializeBlockInfo, - assertPacket, -) + +def writeVarUInt(x, ba): + for _ in range(0, 9): + byte = x & 0x7F + if x > 0x7F: + byte |= 0x80 + + ba.append(byte) + + x >>= 7 + if x == 0: + return + + +def writeStringBinary(s, ba): + b = bytes(s, "utf-8") + writeVarUInt(len(s), ba) + ba.extend(b) + + +def readStrict(s, size=1): + res = bytearray() + while size: + cur = s.recv(size) + # if not res: + # raise "Socket is closed" + size -= len(cur) + res.extend(cur) + + return res + + +def readUInt(s, size=1): + res = readStrict(s, size) + val = 0 + for i in range(len(res)): + val += res[i] << (i * 8) + return val + + +def readUInt8(s): + return readUInt(s) + + +def readUInt16(s): + return readUInt(s, 2) + + +def readUInt32(s): + return readUInt(s, 4) + + +def readUInt64(s): + return readUInt(s, 8) + + +def readVarUInt(s): + x = 0 + for i in range(9): + byte = readStrict(s)[0] + x |= (byte & 0x7F) << (7 * i) + + if not byte & 0x80: + return x + + return x + + +def readStringBinary(s): + size = readVarUInt(s) + s = readStrict(s, size) + return s.decode("utf-8") + + +def sendHello(s): + ba = bytearray() + writeVarUInt(0, ba) # Hello + writeStringBinary(CLIENT_NAME, ba) + writeVarUInt(21, ba) + writeVarUInt(9, ba) + writeVarUInt(54449, ba) + writeStringBinary("default", ba) # database + writeStringBinary("default", ba) # user + writeStringBinary("", ba) # pwd + s.sendall(ba) + + +def receiveHello(s): + p_type = readVarUInt(s) + assert p_type == 0 # Hello + server_name = readStringBinary(s) + # print("Server name: ", server_name) + server_version_major = readVarUInt(s) + # print("Major: ", server_version_major) + server_version_minor = readVarUInt(s) + # print("Minor: ", server_version_minor) + server_revision = readVarUInt(s) + # print("Revision: ", server_revision) + server_timezone = readStringBinary(s) + # print("Timezone: ", server_timezone) + server_display_name = readStringBinary(s) + # print("Display name: ", server_display_name) + server_version_patch = readVarUInt(s) + # print("Version patch: ", server_version_patch) + + +def serializeClientInfo(ba, query_id): + writeStringBinary("default", ba) # initial_user + writeStringBinary(query_id, ba) # initial_query_id + writeStringBinary("127.0.0.1:9000", ba) # initial_address + ba.extend([0] * 8) # initial_query_start_time_microseconds + ba.append(1) # TCP + writeStringBinary("os_user", ba) # os_user + writeStringBinary("client_hostname", ba) # client_hostname + writeStringBinary(CLIENT_NAME, ba) # client_name + writeVarUInt(21, ba) + writeVarUInt(9, ba) + writeVarUInt(54449, ba) + writeStringBinary("", ba) # quota_key + writeVarUInt(0, ba) # distributed_depth + writeVarUInt(1, ba) # client_version_patch + ba.append(0) # No telemetry + + +def sendQuery(s, query): + ba = bytearray() + query_id = uuid.uuid4().hex + writeVarUInt(1, ba) # query + writeStringBinary(query_id, ba) + + ba.append(1) # INITIAL_QUERY + + # client info + serializeClientInfo(ba, query_id) + + writeStringBinary("", ba) # No settings + writeStringBinary("", ba) # No interserver secret + writeVarUInt(2, ba) # Stage - Complete + ba.append(0) # No compression + writeStringBinary(query, ba) # query, finally + s.sendall(ba) + + +def serializeBlockInfo(ba): + writeVarUInt(1, ba) # 1 + ba.append(0) # is_overflows + writeVarUInt(2, ba) # 2 + writeVarUInt(0, ba) # 0 + ba.extend([0] * 4) # bucket_num + + +def sendEmptyBlock(s): + ba = bytearray() + writeVarUInt(2, ba) # Data + writeStringBinary("", ba) + serializeBlockInfo(ba) + writeVarUInt(0, ba) # rows + writeVarUInt(0, ba) # columns + s.sendall(ba) + + +def assertPacket(packet, expected): + assert packet == expected, packet + + +def readHeader(s): + packet_type = readVarUInt(s) + if packet_type == 2: # Exception + raise RuntimeError(readException(s)) + assertPacket(packet_type, 1) # Data + + readStringBinary(s) # external table name + # BlockInfo + assertPacket(readVarUInt(s), 1) # 1 + assertPacket(readUInt8(s), 0) # is_overflows + assertPacket(readVarUInt(s), 2) # 2 + assertPacket(readUInt32(s), 4294967295) # bucket_num + assertPacket(readVarUInt(s), 0) # 0 + columns = readVarUInt(s) # rows + rows = readVarUInt(s) # columns + print("Rows {} Columns {}".format(rows, columns)) + for _ in range(columns): + col_name = readStringBinary(s) + type_name = readStringBinary(s) + print("Column {} type {}".format(col_name, type_name)) + + +def readException(s): + code = readUInt32(s) + name = readStringBinary(s) + text = readStringBinary(s) + readStringBinary(s) # trace + assertPacket(readUInt8(s), 0) # has_nested + return "code {}: {}".format(code, text.replace("DB::Exception:", "")) def insertValidLowCardinalityRow(): - with TCPClient() as client: - client.sendQuery( + with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: + s.settimeout(30) + s.connect((CLICKHOUSE_HOST, CLICKHOUSE_PORT)) + sendHello(s) + receiveHello(s) + sendQuery( + s, "insert into {}.tab settings input_format_defaults_for_omitted_fields=0 format TSV".format( CLICKHOUSE_DATABASE ), ) # external tables - client.sendEmptyBlock() - client.readHeader() + sendEmptyBlock(s) + readHeader(s) # Data ba = bytearray() @@ -46,25 +240,31 @@ def insertValidLowCardinalityRow(): writeStringBinary("hello", ba) # key ba.extend([1] + [0] * 7) # num_indexes ba.extend([0] * 8) # UInt64 index (0 for 'hello') - client.send(ba) + s.sendall(ba) # Fin block - client.sendEmptyBlock() + sendEmptyBlock(s) - assertPacket(client.readVarUInt(), 5) # End of stream + assertPacket(readVarUInt(s), 5) # End of stream + s.close() def insertLowCardinalityRowWithIndexOverflow(): - with TCPClient() as client: - client.sendQuery( + with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: + s.settimeout(30) + s.connect((CLICKHOUSE_HOST, CLICKHOUSE_PORT)) + sendHello(s) + receiveHello(s) + sendQuery( + s, "insert into {}.tab settings input_format_defaults_for_omitted_fields=0 format TSV".format( CLICKHOUSE_DATABASE ), ) # external tables - client.sendEmptyBlock() - client.readHeader() + sendEmptyBlock(s) + readHeader(s) # Data ba = bytearray() @@ -83,23 +283,29 @@ def insertLowCardinalityRowWithIndexOverflow(): writeStringBinary("hello", ba) # key ba.extend([1] + [0] * 7) # num_indexes ba.extend([0] * 7 + [1]) # UInt64 index (overflow) - client.send(ba) + s.sendall(ba) - assertPacket(client.readVarUInt(), 2) # Exception - print(client.readException()) + assertPacket(readVarUInt(s), 2) + print(readException(s)) + s.close() def insertLowCardinalityRowWithIncorrectDictType(): - with TCPClient() as client: - client.sendQuery( + with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: + s.settimeout(30) + s.connect((CLICKHOUSE_HOST, CLICKHOUSE_PORT)) + sendHello(s) + receiveHello(s) + sendQuery( + s, "insert into {}.tab settings input_format_defaults_for_omitted_fields=0 format TSV".format( CLICKHOUSE_DATABASE ), ) # external tables - client.sendEmptyBlock() - client.readHeader() + sendEmptyBlock(s) + readHeader(s) # Data ba = bytearray() @@ -118,23 +324,29 @@ def insertLowCardinalityRowWithIncorrectDictType(): writeStringBinary("hello", ba) # key ba.extend([1] + [0] * 7) # num_indexes ba.extend([0] * 8) # UInt64 index (overflow) - client.send(ba) + s.sendall(ba) - assertPacket(client.readVarUInt(), 2) # Exception - print(client.readException()) + assertPacket(readVarUInt(s), 2) + print(readException(s)) + s.close() def insertLowCardinalityRowWithIncorrectAdditionalKeys(): - with TCPClient() as client: - client.sendQuery( + with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: + s.settimeout(30) + s.connect((CLICKHOUSE_HOST, CLICKHOUSE_PORT)) + sendHello(s) + receiveHello(s) + sendQuery( + s, "insert into {}.tab settings input_format_defaults_for_omitted_fields=0 format TSV".format( CLICKHOUSE_DATABASE ), ) # external tables - client.sendEmptyBlock() - client.readHeader() + sendEmptyBlock(s) + readHeader(s) # Data ba = bytearray() @@ -153,10 +365,11 @@ def insertLowCardinalityRowWithIncorrectAdditionalKeys(): writeStringBinary("hello", ba) # key ba.extend([1] + [0] * 7) # num_indexes ba.extend([0] * 8) # UInt64 index (0 for 'hello') - client.send(ba) + s.sendall(ba) - assertPacket(client.readVarUInt(), 2) # Exception - print(client.readException()) + assertPacket(readVarUInt(s), 2) + print(readException(s)) + s.close() def main(): diff --git a/tests/queries/0_stateless/02210_processors_profile_log.reference b/tests/queries/0_stateless/02210_processors_profile_log.reference index f480236111f..41543d0706a 100644 --- a/tests/queries/0_stateless/02210_processors_profile_log.reference +++ b/tests/queries/0_stateless/02210_processors_profile_log.reference @@ -38,5 +38,4 @@ LazyOutputFormat 1 1 1 0 0 LimitsCheckingTransform 1 1 1 1 1 NullSource 1 0 0 0 0 NullSource 1 0 0 0 0 -NullSource 0 0 0 0 0 SourceFromSingleChunk 1 0 0 1 1 diff --git a/tests/queries/0_stateless/02458_insert_select_progress_tcp.python b/tests/queries/0_stateless/02458_insert_select_progress_tcp.python index fdc64a8dba8..92240e109c1 100644 --- a/tests/queries/0_stateless/02458_insert_select_progress_tcp.python +++ b/tests/queries/0_stateless/02458_insert_select_progress_tcp.python @@ -1,30 +1,188 @@ #!/usr/bin/env python3 -import json +import socket import os -import sys +import uuid +import json -CURDIR = os.path.dirname(os.path.realpath(__file__)) -sys.path.insert(0, os.path.join(CURDIR, "helpers")) +CLICKHOUSE_HOST = os.environ.get("CLICKHOUSE_HOST", "127.0.0.1") +CLICKHOUSE_PORT = int(os.environ.get("CLICKHOUSE_PORT_TCP", "900000")) +CLICKHOUSE_DATABASE = os.environ.get("CLICKHOUSE_DATABASE", "default") +CLIENT_NAME = "simple native protocol" -from tcp_client import TCPClient + +def writeVarUInt(x, ba): + for _ in range(0, 9): + byte = x & 0x7F + if x > 0x7F: + byte |= 0x80 + + ba.append(byte) + + x >>= 7 + if x == 0: + return + + +def writeStringBinary(s, ba): + b = bytes(s, "utf-8") + writeVarUInt(len(s), ba) + ba.extend(b) + + +def readStrict(s, size=1): + res = bytearray() + while size: + cur = s.recv(size) + # if not res: + # raise "Socket is closed" + size -= len(cur) + res.extend(cur) + + return res + + +def readUInt(s, size=1): + res = readStrict(s, size) + val = 0 + for i in range(len(res)): + val += res[i] << (i * 8) + return val + + +def readUInt8(s): + return readUInt(s) + + +def readUInt16(s): + return readUInt(s, 2) + + +def readUInt32(s): + return readUInt(s, 4) + + +def readUInt64(s): + return readUInt(s, 8) + + +def readVarUInt(s): + x = 0 + for i in range(9): + byte = readStrict(s)[0] + x |= (byte & 0x7F) << (7 * i) + + if not byte & 0x80: + return x + + return x + + +def readStringBinary(s): + size = readVarUInt(s) + s = readStrict(s, size) + return s.decode("utf-8") + + +def sendHello(s): + ba = bytearray() + writeVarUInt(0, ba) # Hello + writeStringBinary(CLIENT_NAME, ba) + writeVarUInt(21, ba) + writeVarUInt(9, ba) + writeVarUInt(54449, ba) + writeStringBinary(CLICKHOUSE_DATABASE, ba) # database + writeStringBinary("default", ba) # user + writeStringBinary("", ba) # pwd + s.sendall(ba) + + +def receiveHello(s): + p_type = readVarUInt(s) + assert p_type == 0 # Hello + server_name = readStringBinary(s) + # print("Server name: ", server_name) + server_version_major = readVarUInt(s) + # print("Major: ", server_version_major) + server_version_minor = readVarUInt(s) + # print("Minor: ", server_version_minor) + server_revision = readVarUInt(s) + # print("Revision: ", server_revision) + server_timezone = readStringBinary(s) + # print("Timezone: ", server_timezone) + server_display_name = readStringBinary(s) + # print("Display name: ", server_display_name) + server_version_patch = readVarUInt(s) + # print("Version patch: ", server_version_patch) + + +def serializeClientInfo(ba, query_id): + writeStringBinary("default", ba) # initial_user + writeStringBinary(query_id, ba) # initial_query_id + writeStringBinary("127.0.0.1:9000", ba) # initial_address + ba.extend([0] * 8) # initial_query_start_time_microseconds + ba.append(1) # TCP + writeStringBinary("os_user", ba) # os_user + writeStringBinary("client_hostname", ba) # client_hostname + writeStringBinary(CLIENT_NAME, ba) # client_name + writeVarUInt(21, ba) + writeVarUInt(9, ba) + writeVarUInt(54449, ba) + writeStringBinary("", ba) # quota_key + writeVarUInt(0, ba) # distributed_depth + writeVarUInt(1, ba) # client_version_patch + ba.append(0) # No telemetry + + +def sendQuery(s, query): + ba = bytearray() + query_id = uuid.uuid4().hex + writeVarUInt(1, ba) # query + writeStringBinary(query_id, ba) + + ba.append(1) # INITIAL_QUERY + + # client info + serializeClientInfo(ba, query_id) + + writeStringBinary("", ba) # No settings + writeStringBinary("", ba) # No interserver secret + writeVarUInt(2, ba) # Stage - Complete + ba.append(0) # No compression + writeStringBinary(query, ba) # query, finally + s.sendall(ba) + + +def serializeBlockInfo(ba): + writeVarUInt(1, ba) # 1 + ba.append(0) # is_overflows + writeVarUInt(2, ba) # 2 + writeVarUInt(0, ba) # 0 + ba.extend([0] * 4) # bucket_num + + +def sendEmptyBlock(s): + ba = bytearray() + writeVarUInt(2, ba) # Data + writeStringBinary("", ba) + serializeBlockInfo(ba) + writeVarUInt(0, ba) # rows + writeVarUInt(0, ba) # columns + s.sendall(ba) + + +def assertPacket(packet, expected): + assert packet == expected, packet class Progress: - def __init__( - self, - read_rows=0, - read_bytes=0, - total_rows_to_read=0, - written_rows=0, - written_bytes=0, - ): + def __init__(self): # NOTE: this is done in ctor to initialize __dict__ - self.read_rows = read_rows - self.read_bytes = read_bytes - self.total_rows_to_read = total_rows_to_read - self.written_rows = written_rows - self.written_bytes = written_bytes + self.read_rows = 0 + self.read_bytes = 0 + self.total_rows_to_read = 0 + self.written_rows = 0 + self.written_bytes = 0 def __str__(self): return json.dumps(self.__dict__) @@ -37,6 +195,13 @@ class Progress: self.written_bytes += b.written_bytes return self + def readPacket(self, s): + self.read_rows += readVarUInt(s) + self.read_bytes += readVarUInt(s) + self.total_rows_to_read += readVarUInt(s) + self.written_rows += readVarUInt(s) + self.written_bytes += readVarUInt(s) + def __bool__(self): return ( self.read_rows > 0 @@ -47,25 +212,52 @@ class Progress: ) +def readProgress(s): + packet_type = readVarUInt(s) + if packet_type == 2: # Exception + raise RuntimeError(readException(s)) + + if packet_type == 5: # End stream + return None + + assertPacket(packet_type, 3) # Progress + + progress = Progress() + progress.readPacket(s) + return progress + + +def readException(s): + code = readUInt32(s) + name = readStringBinary(s) + text = readStringBinary(s) + readStringBinary(s) # trace + assertPacket(readUInt8(s), 0) # has_nested + return "code {}: {}".format(code, text.replace("DB::Exception:", "")) + + def main(): - with TCPClient() as client: + with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: + s.settimeout(30) + s.connect((CLICKHOUSE_HOST, CLICKHOUSE_PORT)) + sendHello(s) + receiveHello(s) # For 1 second sleep and 1000ms of interactive_delay we definitelly should have non zero progress packet. # NOTE: interactive_delay=0 cannot be used since in this case CompletedPipelineExecutor will not call cancelled callback. - client.sendQuery( + sendQuery( + s, "insert into function null('_ Int') select sleep(1) from numbers(2) settings max_block_size=1, interactive_delay=1000", ) # external tables - client.sendEmptyBlock() + sendEmptyBlock(s) summary_progress = Progress() non_empty_progress_packets = 0 while True: - progress_info = client.readProgress() - if progress_info is None: + progress = readProgress(s) + if progress is None: break - - progress = Progress(*progress_info) summary_progress += progress if progress: non_empty_progress_packets += 1 @@ -76,6 +268,8 @@ def main(): # - 1 or 2 for each SELECT block assert non_empty_progress_packets in (3, 4), f"{non_empty_progress_packets=:}" + s.close() + if __name__ == "__main__": main() diff --git a/tests/queries/0_stateless/02750_settings_alias_tcp_protocol.python b/tests/queries/0_stateless/02750_settings_alias_tcp_protocol.python index 1736807410f..48b27d434ec 100644 --- a/tests/queries/0_stateless/02750_settings_alias_tcp_protocol.python +++ b/tests/queries/0_stateless/02750_settings_alias_tcp_protocol.python @@ -1,23 +1,217 @@ #!/usr/bin/env python3 - +import socket import os -import sys +import uuid +import json -CURDIR = os.path.dirname(os.path.realpath(__file__)) -sys.path.insert(0, os.path.join(CURDIR, "helpers")) +CLICKHOUSE_HOST = os.environ.get("CLICKHOUSE_HOST", "127.0.0.1") +CLICKHOUSE_PORT = int(os.environ.get("CLICKHOUSE_PORT_TCP", "900000")) +CLICKHOUSE_DATABASE = os.environ.get("CLICKHOUSE_DATABASE", "default") +CLIENT_NAME = "simple native protocol" -from tcp_client import TCPClient + +def writeVarUInt(x, ba): + for _ in range(0, 9): + byte = x & 0x7F + if x > 0x7F: + byte |= 0x80 + + ba.append(byte) + + x >>= 7 + if x == 0: + return + + +def writeStringBinary(s, ba): + b = bytes(s, "utf-8") + writeVarUInt(len(s), ba) + ba.extend(b) + + +def readStrict(s, size=1): + res = bytearray() + while size: + cur = s.recv(size) + # if not res: + # raise "Socket is closed" + size -= len(cur) + res.extend(cur) + + return res + + +def readUInt(s, size=1): + res = readStrict(s, size) + val = 0 + for i in range(len(res)): + val += res[i] << (i * 8) + return val + + +def readUInt8(s): + return readUInt(s) + + +def readUInt16(s): + return readUInt(s, 2) + + +def readUInt32(s): + return readUInt(s, 4) + + +def readUInt64(s): + return readUInt(s, 8) + + +def readVarUInt(s): + x = 0 + for i in range(9): + byte = readStrict(s)[0] + x |= (byte & 0x7F) << (7 * i) + + if not byte & 0x80: + return x + + return x + + +def readStringBinary(s): + size = readVarUInt(s) + s = readStrict(s, size) + return s.decode("utf-8") + + +def sendHello(s): + ba = bytearray() + writeVarUInt(0, ba) # Hello + writeStringBinary(CLIENT_NAME, ba) + writeVarUInt(21, ba) + writeVarUInt(9, ba) + writeVarUInt(54449, ba) + writeStringBinary(CLICKHOUSE_DATABASE, ba) # database + writeStringBinary("default", ba) # user + writeStringBinary("", ba) # pwd + s.sendall(ba) + + +def receiveHello(s): + p_type = readVarUInt(s) + assert p_type == 0 # Hello + _server_name = readStringBinary(s) + _server_version_major = readVarUInt(s) + _server_version_minor = readVarUInt(s) + _server_revision = readVarUInt(s) + _server_timezone = readStringBinary(s) + _server_display_name = readStringBinary(s) + _server_version_patch = readVarUInt(s) + + +def serializeClientInfo(ba, query_id): + writeStringBinary("default", ba) # initial_user + writeStringBinary(query_id, ba) # initial_query_id + writeStringBinary("127.0.0.1:9000", ba) # initial_address + ba.extend([0] * 8) # initial_query_start_time_microseconds + ba.append(1) # TCP + writeStringBinary("os_user", ba) # os_user + writeStringBinary("client_hostname", ba) # client_hostname + writeStringBinary(CLIENT_NAME, ba) # client_name + writeVarUInt(21, ba) + writeVarUInt(9, ba) + writeVarUInt(54449, ba) + writeStringBinary("", ba) # quota_key + writeVarUInt(0, ba) # distributed_depth + writeVarUInt(1, ba) # client_version_patch + ba.append(0) # No telemetry + + +def sendQuery(s, query, settings): + ba = bytearray() + query_id = uuid.uuid4().hex + writeVarUInt(1, ba) # query + writeStringBinary(query_id, ba) + + ba.append(1) # INITIAL_QUERY + + # client info + serializeClientInfo(ba, query_id) + + # Settings + for key, value in settings.items(): + writeStringBinary(key, ba) + writeVarUInt(1, ba) # is_important + writeStringBinary(str(value), ba) + writeStringBinary("", ba) # End of settings + + writeStringBinary("", ba) # No interserver secret + writeVarUInt(2, ba) # Stage - Complete + ba.append(0) # No compression + writeStringBinary(query, ba) # query, finally + s.sendall(ba) + + +def serializeBlockInfo(ba): + writeVarUInt(1, ba) # 1 + ba.append(0) # is_overflows + writeVarUInt(2, ba) # 2 + writeVarUInt(0, ba) # 0 + ba.extend([0] * 4) # bucket_num + + +def sendEmptyBlock(s): + ba = bytearray() + writeVarUInt(2, ba) # Data + writeStringBinary("", ba) + serializeBlockInfo(ba) + writeVarUInt(0, ba) # rows + writeVarUInt(0, ba) # columns + s.sendall(ba) + + +def assertPacket(packet, expected): + assert packet == expected, "Got: {}, expected: {}".format(packet, expected) + + +def readResponse(s): + packet_type = readVarUInt(s) + if packet_type == 2: # Exception + raise RuntimeError(readException(s)) + + if packet_type == 1: # Data + return None + if packet_type == 3: # Progress + return None + if packet_type == 5: # End stream + return None + + raise RuntimeError("Unexpected packet: {}".format(packet_type)) + + +def readException(s): + code = readUInt32(s) + _name = readStringBinary(s) + text = readStringBinary(s) + readStringBinary(s) # trace + assertPacket(readUInt8(s), 0) # has_nested + return "code {}: {}".format(code, text.replace("DB::Exception:", "")) def main(): - with TCPClient() as client: - client.sendQuery("select 1", {"replication_alter_partitions_sync": 1}) + with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: + s.settimeout(30) + s.connect((CLICKHOUSE_HOST, CLICKHOUSE_PORT)) + sendHello(s) + receiveHello(s) + sendQuery(s, "select 1", {"replication_alter_partitions_sync": 1}) # external tables - client.sendEmptyBlock() + sendEmptyBlock(s) - while client.readResponse() is not None: + while readResponse(s) is not None: pass + + s.close() print("OK") diff --git a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python deleted file mode 100755 index b3d04d263dc..00000000000 --- a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python +++ /dev/null @@ -1,93 +0,0 @@ -#!/usr/bin/env python3 -# -*- coding: utf-8 -*- - - -import os -import sys - -CURDIR = os.path.dirname(os.path.realpath(__file__)) -sys.path.insert(0, os.path.join(CURDIR, "helpers")) - -from tcp_client import TCPClient - - -def run_query_without_errors(query, support_partial_result): - with TCPClient() as client: - client.sendQuery(query) - - # external tables - client.sendEmptyBlock() - client.readHeader() - - # Partial result - partial_result = client.readDataWithoutProgress()[0] - if support_partial_result: - assert ( - len(partial_result.value) > 0 - ), "Expected at least one block with a non-empty partial result before getting the full result" - - while True: - assert all( - a >= b - for a, b in zip(partial_result.value, partial_result.value[1:]) - ), "Partial result always should be sorted for this test" - - new_partial_result = client.readDataWithoutProgress( - need_print_info=False - )[0] - if len(new_partial_result.value) == 0: - break - - data_size = len(partial_result.value) - assert all( - partial_result.value[i] <= new_partial_result.value[i] - for i in range(data_size) - ), f"New partial result values should always be greater then old one because a new block contains more information about the full data. New result {new_partial_result}. Previous result {partial_result}" - - partial_result = new_partial_result - else: - block_rows = len(partial_result.value) - assert ( - block_rows == 0 - ), f"Expected only empty partial result block before getting the full result, but block has {block_rows} rows" - - # Full result - full_result = client.readDataWithoutProgress()[0] - - data_size = len(partial_result.value) - assert all( - partial_result.value[i] <= full_result.value[i] for i in range(data_size) - ), f"Full result values should always be greater then partial result values. Full result {full_result}. Partial result {partial_result}" - - for result in full_result.value: - print(result) - - -def main(): - # Request with partial result limit less then full limit - run_query_without_errors( - "SELECT number FROM numbers_mt(5e6+1) ORDER BY -number LIMIT 5 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 3", - support_partial_result=True, - ) - - # Request with partial result limit greater then full limit - run_query_without_errors( - "SELECT number FROM numbers_mt(5e6+1) ORDER BY -number LIMIT 3 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 5", - support_partial_result=True, - ) - - # Request with OFFSET - run_query_without_errors( - "SELECT number FROM numbers_mt(5e6+1) ORDER BY -number LIMIT 3 OFFSET 1 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 5", - support_partial_result=True, - ) - - # Request with OFFSET greater then partial result limit (partial result pipeline use blocks with less then OFFSET, so there will be no elements in block after LimitPartialResultTransform) - run_query_without_errors( - "SELECT number FROM numbers_mt(5e6+1) ORDER BY -number LIMIT 3 OFFSET 15 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 5", - support_partial_result=False, - ) - - -if __name__ == "__main__": - main() diff --git a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.reference b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.reference deleted file mode 100644 index 211a193940a..00000000000 --- a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.reference +++ /dev/null @@ -1,38 +0,0 @@ -Rows 0 Columns 1 -Column number type UInt64 -Rows 3 Columns 1 -Column number type UInt64 -Rows 5 Columns 1 -Column number type UInt64 -5000000 -4999999 -4999998 -4999997 -4999996 -Rows 0 Columns 1 -Column number type UInt64 -Rows 3 Columns 1 -Column number type UInt64 -Rows 3 Columns 1 -Column number type UInt64 -5000000 -4999999 -4999998 -Rows 0 Columns 1 -Column number type UInt64 -Rows 3 Columns 1 -Column number type UInt64 -Rows 3 Columns 1 -Column number type UInt64 -4999999 -4999998 -4999997 -Rows 0 Columns 1 -Column number type UInt64 -Rows 0 Columns 1 -Column number type UInt64 -Rows 3 Columns 1 -Column number type UInt64 -4999985 -4999984 -4999983 diff --git a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.sh b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.sh deleted file mode 100755 index 1ed15197dbf..00000000000 --- a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.sh +++ /dev/null @@ -1,8 +0,0 @@ -#!/usr/bin/env bash - -CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) -# shellcheck source=../shell_config.sh -. "$CURDIR"/../shell_config.sh - -# We should have correct env vars from shell_config.sh to run this test -python3 "$CURDIR"/02833_partial_sorting_result_during_query_execution.python diff --git a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python deleted file mode 100644 index 4d869b05580..00000000000 --- a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python +++ /dev/null @@ -1,125 +0,0 @@ -#!/usr/bin/env python3 -# -*- coding: utf-8 -*- - - -import os -import sys - -CURDIR = os.path.dirname(os.path.realpath(__file__)) -sys.path.insert(0, os.path.join(CURDIR, "helpers")) - -from tcp_client import TCPClient - - -def get_keys(results): - return [key for key, _ in results] - - -def check_new_result(new_results, old_results, invariants, rows_limit): - if rows_limit is not None: - assert ( - len(new_results[0].value) <= rows_limit - ), f"Result should have no more then {rows_limit} rows. But it has {len(new_results[0].value)} rows" - - for new_result, old_result in zip(new_results, old_results): - assert ( - new_result.key == old_result.key - ), f"Keys in blocks should be in the same order. Full results keys {get_keys(full_results)}. Partial results keys {get_keys(partial_results)}" - - key = new_result.key - if key in invariants: - new_value = new_result.value - old_value = old_result.value - assert invariants[key]( - old_value, new_value - ), f"Problem with the invariant between new and old result for key: {key}. New value {new_value}. Old value {old_value}" - - -def run_query_without_errors( - query, support_partial_result, invariants=None, rows_limit=None -): - if invariants is None: - invariants = {} - - with TCPClient() as client: - client.sendQuery(query) - - # external tables - client.sendEmptyBlock() - client.readHeader() - - # Partial result - partial_results = client.readDataWithoutProgress() - if support_partial_result: - assert ( - len(partial_results) > 0 and len(partial_results[0].value) > 0 - ), "Expected at least one block with a non-empty partial result before getting the full result" - while True: - new_partial_results = client.readDataWithoutProgress( - need_print_info=False - ) - if len(new_partial_results[0].value) == 0: - break - - check_new_result( - new_partial_results, partial_results, invariants, rows_limit - ) - partial_results = new_partial_results - else: - block_rows = len(partial_results[0].value) - assert ( - block_rows == 0 - ), f"Expected only empty partial result block before getting the full result, but block has {block_rows} rows" - - # Full result - full_results = client.readDataWithoutProgress() - if support_partial_result: - check_new_result(full_results, partial_results, invariants, rows_limit) - - for data in full_results: - if isinstance(data.value[0], int): - print(data.key, data.value) - - -def supported_scenarios_without_key(): - # Simple aggregation query - query = "select median(number), stddevSamp(number), stddevPop(number), max(number), min(number), any(number), count(number), avg(number), sum(number) from numbers_mt(1e7+1) settings max_threads = 1, partial_result_update_duration_ms = 1" - invariants = { - "median(number)": lambda old_value, new_value: old_value <= new_value, - "max(number)": lambda old_value, new_value: old_value <= new_value, - "min(number)": lambda old_value, new_value: old_value >= new_value, - "count(number)": lambda old_value, new_value: old_value <= new_value, - "avg(number)": lambda old_value, new_value: old_value <= new_value, - "sum(number)": lambda old_value, new_value: old_value <= new_value, - } - run_query_without_errors( - query, support_partial_result=True, invariants=invariants, rows_limit=1 - ) - - # Aggregation query with a nested ORDER BY subquery - query = "select median(number), stddevSamp(number), stddevPop(number), max(number), min(number), any(number), count(number), avg(number), sum(number) FROM (SELECT number FROM numbers_mt(1e7) ORDER BY -number LIMIT 3) settings max_threads = 1, partial_result_update_duration_ms=1" - - # Aggregation receives small partial result blocks from ORDER BY which always sends blocks with bigger values - invariants["min(number)"] = lambda old_value, new_value: old_value <= new_value - run_query_without_errors( - query, support_partial_result=True, invariants=invariants, rows_limit=1 - ) - - -def unsupported_scenarios(): - # Currently aggregator for partial result supports only single thread aggregation without key - # Update test when multithreading or aggregation with GROUP BY will be supported for partial result updates - multithread_query = "select sum(number) from numbers_mt(1e7+1) settings max_threads = 2, partial_result_update_duration_ms = 100" - run_query_without_errors(multithread_query, support_partial_result=False) - - group_with_key_query = "select mod2, sum(number) from numbers_mt(1e7+1) group by number % 2 as mod2 settings max_threads = 1, partial_result_update_duration_ms = 100" - run_query_without_errors(group_with_key_query, support_partial_result=False) - - -def main(): - supported_scenarios_without_key() - unsupported_scenarios() - - -if __name__ == "__main__": - main() diff --git a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.reference b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.reference deleted file mode 100644 index a813b18f24f..00000000000 --- a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.reference +++ /dev/null @@ -1,88 +0,0 @@ -Rows 0 Columns 9 -Column median(number) type Float64 -Column stddevSamp(number) type Float64 -Column stddevPop(number) type Float64 -Column max(number) type UInt64 -Column min(number) type UInt64 -Column any(number) type UInt64 -Column count(number) type UInt64 -Column avg(number) type Float64 -Column sum(number) type UInt64 -Rows 1 Columns 9 -Column median(number) type Float64 -Column stddevSamp(number) type Float64 -Column stddevPop(number) type Float64 -Column max(number) type UInt64 -Column min(number) type UInt64 -Column any(number) type UInt64 -Column count(number) type UInt64 -Column avg(number) type Float64 -Column sum(number) type UInt64 -Rows 1 Columns 9 -Column median(number) type Float64 -Column stddevSamp(number) type Float64 -Column stddevPop(number) type Float64 -Column max(number) type UInt64 -Column min(number) type UInt64 -Column any(number) type UInt64 -Column count(number) type UInt64 -Column avg(number) type Float64 -Column sum(number) type UInt64 -max(number) [10000000] -min(number) [0] -any(number) [0] -count(number) [10000001] -sum(number) [50000005000000] -Rows 0 Columns 9 -Column median(number) type Float64 -Column stddevSamp(number) type Float64 -Column stddevPop(number) type Float64 -Column max(number) type UInt64 -Column min(number) type UInt64 -Column any(number) type UInt64 -Column count(number) type UInt64 -Column avg(number) type Float64 -Column sum(number) type UInt64 -Rows 1 Columns 9 -Column median(number) type Float64 -Column stddevSamp(number) type Float64 -Column stddevPop(number) type Float64 -Column max(number) type UInt64 -Column min(number) type UInt64 -Column any(number) type UInt64 -Column count(number) type UInt64 -Column avg(number) type Float64 -Column sum(number) type UInt64 -Rows 1 Columns 9 -Column median(number) type Float64 -Column stddevSamp(number) type Float64 -Column stddevPop(number) type Float64 -Column max(number) type UInt64 -Column min(number) type UInt64 -Column any(number) type UInt64 -Column count(number) type UInt64 -Column avg(number) type Float64 -Column sum(number) type UInt64 -max(number) [9999999] -min(number) [9999997] -any(number) [9999999] -count(number) [3] -sum(number) [29999994] -Rows 0 Columns 1 -Column sum(number) type UInt64 -Rows 0 Columns 1 -Column sum(number) type UInt64 -Rows 1 Columns 1 -Column sum(number) type UInt64 -sum(number) [50000005000000] -Rows 0 Columns 2 -Column mod2 type UInt8 -Column sum(number) type UInt64 -Rows 0 Columns 2 -Column mod2 type UInt8 -Column sum(number) type UInt64 -Rows 2 Columns 2 -Column mod2 type UInt8 -Column sum(number) type UInt64 -mod2 [0, 1] -sum(number) [25000005000000, 25000000000000] diff --git a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.sh b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.sh deleted file mode 100755 index e70a3c53ec4..00000000000 --- a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.sh +++ /dev/null @@ -1,8 +0,0 @@ -#!/usr/bin/env bash - -CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) -# shellcheck source=../shell_config.sh -. "$CURDIR"/../shell_config.sh - -# We should have correct env vars from shell_config.sh to run this test -python3 "$CURDIR"/02834_partial_aggregating_result_during_query_execution.python diff --git a/tests/queries/0_stateless/helpers/tcp_client.py b/tests/queries/0_stateless/helpers/tcp_client.py deleted file mode 100644 index fdc4ab28e04..00000000000 --- a/tests/queries/0_stateless/helpers/tcp_client.py +++ /dev/null @@ -1,313 +0,0 @@ -import socket -import os -import uuid -import struct - -CLICKHOUSE_HOST = os.environ.get("CLICKHOUSE_HOST", "127.0.0.1") -CLICKHOUSE_PORT = int(os.environ.get("CLICKHOUSE_PORT_TCP", "900000")) -CLICKHOUSE_DATABASE = os.environ.get("CLICKHOUSE_DATABASE", "default") -CLIENT_NAME = "simple native protocol" - - -def writeVarUInt(x, ba): - for _ in range(0, 9): - byte = x & 0x7F - if x > 0x7F: - byte |= 0x80 - - ba.append(byte) - - x >>= 7 - if x == 0: - return - - -def writeStringBinary(s, ba): - b = bytes(s, "utf-8") - writeVarUInt(len(s), ba) - ba.extend(b) - - -def serializeClientInfo(ba, query_id): - writeStringBinary("default", ba) # initial_user - writeStringBinary(query_id, ba) # initial_query_id - writeStringBinary("127.0.0.1:9000", ba) # initial_address - ba.extend([0] * 8) # initial_query_start_time_microseconds - ba.append(1) # TCP - writeStringBinary("os_user", ba) # os_user - writeStringBinary("client_hostname", ba) # client_hostname - writeStringBinary(CLIENT_NAME, ba) # client_name - writeVarUInt(21, ba) - writeVarUInt(9, ba) - writeVarUInt(54449, ba) - writeStringBinary("", ba) # quota_key - writeVarUInt(0, ba) # distributed_depth - writeVarUInt(1, ba) # client_version_patch - ba.append(0) # No telemetry - - -def serializeBlockInfo(ba): - writeVarUInt(1, ba) # 1 - ba.append(0) # is_overflows - writeVarUInt(2, ba) # 2 - writeVarUInt(0, ba) # 0 - ba.extend([0] * 4) # bucket_num - - -def assertPacket(packet, expected): - assert packet == expected, "Got: {}, expected: {}".format(packet, expected) - - -class Data(object): - def __init__(self, key, value): - self.key = key - self.value = value - - -class TCPClient(object): - def __init__(self, timeout=30): - self.timeout = timeout - self.socket = None - - def __enter__(self): - self.socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM) - self.socket.settimeout(self.timeout) - self.socket.connect((CLICKHOUSE_HOST, CLICKHOUSE_PORT)) - - self.sendHello() - self.receiveHello() - - return self - - def __exit__(self, exc_type, exc_value, traceback): - if self.socket: - self.socket.close() - - def readStrict(self, size=1): - res = bytearray() - while size: - cur = self.socket.recv(size) - # if not res: - # raise "Socket is closed" - size -= len(cur) - res.extend(cur) - - return res - - def readUInt(self, size=1): - res = self.readStrict(size) - val = 0 - for i in range(len(res)): - val += res[i] << (i * 8) - return val - - def readUInt8(self): - return self.readUInt() - - def readUInt16(self): - return self.readUInt(2) - - def readUInt32(self): - return self.readUInt(4) - - def readUInt64(self): - return self.readUInt(8) - - def readFloat16(self): - return struct.unpack("e", self.readStrict(2)) - - def readFloat32(self): - return struct.unpack("f", self.readStrict(4)) - - def readFloat64(self): - return struct.unpack("d", self.readStrict(8)) - - def readVarUInt(self): - x = 0 - for i in range(9): - byte = self.readStrict()[0] - x |= (byte & 0x7F) << (7 * i) - - if not byte & 0x80: - return x - - return x - - def readStringBinary(self): - size = self.readVarUInt() - s = self.readStrict(size) - return s.decode("utf-8") - - def send(self, byte_array): - self.socket.sendall(byte_array) - - def sendHello(self): - ba = bytearray() - writeVarUInt(0, ba) # Hello - writeStringBinary(CLIENT_NAME, ba) - writeVarUInt(21, ba) - writeVarUInt(9, ba) - writeVarUInt(54449, ba) - writeStringBinary(CLICKHOUSE_DATABASE, ba) # database - writeStringBinary("default", ba) # user - writeStringBinary("", ba) # pwd - self.send(ba) - - def receiveHello(self): - p_type = self.readVarUInt() - assert p_type == 0 # Hello - _server_name = self.readStringBinary() - _server_version_major = self.readVarUInt() - _server_version_minor = self.readVarUInt() - _server_revision = self.readVarUInt() - _server_timezone = self.readStringBinary() - _server_display_name = self.readStringBinary() - _server_version_patch = self.readVarUInt() - - def sendQuery(self, query, settings=None): - if settings == None: - settings = {} # No settings - - ba = bytearray() - query_id = uuid.uuid4().hex - writeVarUInt(1, ba) # query - writeStringBinary(query_id, ba) - - ba.append(1) # INITIAL_QUERY - - # client info - serializeClientInfo(ba, query_id) - - # Settings - for key, value in settings.items(): - writeStringBinary(key, ba) - writeVarUInt(1, ba) # is_important - writeStringBinary(str(value), ba) - writeStringBinary("", ba) # End of settings - - writeStringBinary("", ba) # No interserver secret - writeVarUInt(2, ba) # Stage - Complete - ba.append(0) # No compression - writeStringBinary(query, ba) # query, finally - self.send(ba) - - def sendEmptyBlock(self): - ba = bytearray() - writeVarUInt(2, ba) # Data - writeStringBinary("", ba) - serializeBlockInfo(ba) - writeVarUInt(0, ba) # rows - writeVarUInt(0, ba) # columns - self.send(ba) - - def readException(self): - code = self.readUInt32() - _name = self.readStringBinary() - text = self.readStringBinary() - self.readStringBinary() # trace - assertPacket(self.readUInt8(), 0) # has_nested - return "code {}: {}".format(code, text.replace("DB::Exception:", "")) - - def readPacketType(self): - packet_type = self.readVarUInt() - if packet_type == 2: # Exception - raise RuntimeError(self.readException()) - - return packet_type - - def readResponse(self): - packet_type = self.readPacketType() - if packet_type == 1: # Data - return None - if packet_type == 3: # Progress - return None - if packet_type == 5: # End stream - return None - - raise RuntimeError("Unexpected packet: {}".format(packet_type)) - - def readProgressData(self): - read_rows = self.readVarUInt() - read_bytes = self.readVarUInt() - total_rows_to_read = self.readVarUInt() - written_rows = self.readVarUInt() - written_bytes = self.readVarUInt() - - return read_rows, read_bytes, total_rows_to_read, written_rows, written_bytes - - def readProgress(self): - packet_type = self.readPacketType() - if packet_type == 5: # End stream - return None - assertPacket(packet_type, 3) # Progress - return self.readProgressData() - - def readHeaderInfo(self): - self.readStringBinary() # external table name - # BlockInfo - assertPacket(self.readVarUInt(), 1) # field number 1 - assertPacket(self.readUInt8(), 0) # is_overflows - assertPacket(self.readVarUInt(), 2) # field number 2 - assertPacket(self.readUInt32(), 4294967295) # bucket_num - assertPacket(self.readVarUInt(), 0) # 0 - columns = self.readVarUInt() # rows - rows = self.readVarUInt() # columns - - return columns, rows - - def readHeader(self): - packet_type = self.readPacketType() - assertPacket(packet_type, 1) # Data - - columns, rows = self.readHeaderInfo() - print("Rows {} Columns {}".format(rows, columns)) - for _ in range(columns): - col_name = self.readStringBinary() - type_name = self.readStringBinary() - print("Column {} type {}".format(col_name, type_name)) - - def readRow(self, row_type, rows): - supported_row_types = { - "UInt8": self.readUInt8, - "UInt16": self.readUInt16, - "UInt32": self.readUInt32, - "UInt64": self.readUInt64, - "Float16": self.readFloat16, - "Float32": self.readFloat32, - "Float64": self.readFloat64, - } - if row_type in supported_row_types: - read_type = supported_row_types[row_type] - row = [read_type() for _ in range(rows)] - return row - else: - raise RuntimeError( - "Current python version of tcp client doesn't support the following type of row: {}".format( - row_type - ) - ) - - def readDataWithoutProgress(self, need_print_info=True): - packet_type = self.readPacketType() - while packet_type == 3: # Progress - self.readProgressData() - packet_type = self.readPacketType() - - if packet_type == 5: # End stream - return None - assertPacket(packet_type, 1) # Data - - columns, rows = self.readHeaderInfo() - data = [] - if need_print_info: - print("Rows {} Columns {}".format(rows, columns)) - - for _ in range(columns): - col_name = self.readStringBinary() - type_name = self.readStringBinary() - if need_print_info: - print("Column {} type {}".format(col_name, type_name)) - - data.append(Data(col_name, self.readRow(type_name, rows))) - - return data From 923d7a0bc3c848c05c6772735720620710110e68 Mon Sep 17 00:00:00 2001 From: robot-clickhouse Date: Sat, 9 Sep 2023 00:27:39 +0000 Subject: [PATCH 243/433] Automatic style fix --- tests/ci/clickhouse_helper.py | 2 ++ 1 file changed, 2 insertions(+) diff --git a/tests/ci/clickhouse_helper.py b/tests/ci/clickhouse_helper.py index ff0e2677759..1b2d4c08edb 100644 --- a/tests/ci/clickhouse_helper.py +++ b/tests/ci/clickhouse_helper.py @@ -183,10 +183,12 @@ def _query_imds(path): continue return "" + # Obtain the machine type from IMDS: def get_instance_type(): return _query_imds("latest/meta-data/instance-type") + # Obtain the instance id from IMDS: def get_instance_id(): return _query_imds("latest/meta-data/instance-id") From 71d37d64ab37ee24716868fa83c0421f5168a307 Mon Sep 17 00:00:00 2001 From: Alexey Milovidov Date: Sat, 9 Sep 2023 06:37:02 +0300 Subject: [PATCH 244/433] Revert "Fix issues in accurateCastOrNull" --- src/Functions/FunctionsConversion.h | 60 ++++--------------- .../01556_accurate_cast_or_null.reference | 5 -- .../01556_accurate_cast_or_null.sql | 7 --- ...03_cast_nullable_to_custom_types.reference | 4 +- 4 files changed, 13 insertions(+), 63 deletions(-) diff --git a/src/Functions/FunctionsConversion.h b/src/Functions/FunctionsConversion.h index 79bac2d17e8..98d28cf03a8 100644 --- a/src/Functions/FunctionsConversion.h +++ b/src/Functions/FunctionsConversion.h @@ -88,7 +88,6 @@ namespace ErrorCodes extern const int ILLEGAL_TYPE_OF_ARGUMENT; extern const int NOT_IMPLEMENTED; extern const int CANNOT_INSERT_NULL_IN_ORDINARY_COLUMN; - extern const int CANNOT_PARSE_BOOL; } @@ -1703,19 +1702,7 @@ struct ConvertImplGenericFromString const auto & val = col_from_string->getDataAt(i); ReadBufferFromMemory read_buffer(val.data, val.size); - try - { - serialization_from.deserializeWholeText(column_to, read_buffer, format_settings); - } - catch (const Exception & e) - { - if (e.code() == ErrorCodes::CANNOT_PARSE_BOOL && typeid_cast(&column_to)) - { - column_to.insertDefault(); - continue; - } - throw; - } + serialization_from.deserializeWholeText(column_to, read_buffer, format_settings); if (!read_buffer.eof()) { @@ -4174,21 +4161,15 @@ private: { if constexpr (std::is_same_v) { - ret = [cast_ipv4_ipv6_default_on_conversion_error_value, - input_format_ipv4_default_on_conversion_error_value, - requested_result_is_nullable]( - ColumnsWithTypeAndName & arguments, - const DataTypePtr & result_type, - const ColumnNullable * column_nullable, - size_t) -> ColumnPtr + ret = [cast_ipv4_ipv6_default_on_conversion_error_value, input_format_ipv4_default_on_conversion_error_value, requested_result_is_nullable]( + ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, const ColumnNullable * column_nullable, size_t) + -> ColumnPtr { if (!WhichDataType(result_type).isIPv4()) throw Exception(ErrorCodes::TYPE_MISMATCH, "Wrong result type {}. Expected IPv4", result_type->getName()); const auto * null_map = column_nullable ? &column_nullable->getNullMapData() : nullptr; - if (requested_result_is_nullable) - return convertToIPv4(arguments[0].column, null_map); - else if (cast_ipv4_ipv6_default_on_conversion_error_value || input_format_ipv4_default_on_conversion_error_value) + if (cast_ipv4_ipv6_default_on_conversion_error_value || input_format_ipv4_default_on_conversion_error_value || requested_result_is_nullable) return convertToIPv4(arguments[0].column, null_map); else return convertToIPv4(arguments[0].column, null_map); @@ -4199,22 +4180,16 @@ private: if constexpr (std::is_same_v) { - ret = [cast_ipv4_ipv6_default_on_conversion_error_value, - input_format_ipv6_default_on_conversion_error_value, - requested_result_is_nullable]( - ColumnsWithTypeAndName & arguments, - const DataTypePtr & result_type, - const ColumnNullable * column_nullable, - size_t) -> ColumnPtr + ret = [cast_ipv4_ipv6_default_on_conversion_error_value, input_format_ipv6_default_on_conversion_error_value, requested_result_is_nullable]( + ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, const ColumnNullable * column_nullable, size_t) + -> ColumnPtr { if (!WhichDataType(result_type).isIPv6()) throw Exception( ErrorCodes::TYPE_MISMATCH, "Wrong result type {}. Expected IPv6", result_type->getName()); const auto * null_map = column_nullable ? &column_nullable->getNullMapData() : nullptr; - if (requested_result_is_nullable) - return convertToIPv6(arguments[0].column, null_map); - else if (cast_ipv4_ipv6_default_on_conversion_error_value || input_format_ipv6_default_on_conversion_error_value) + if (cast_ipv4_ipv6_default_on_conversion_error_value || input_format_ipv6_default_on_conversion_error_value || requested_result_is_nullable) return convertToIPv6(arguments[0].column, null_map); else return convertToIPv6(arguments[0].column, null_map); @@ -4225,18 +4200,7 @@ private: if (to_type->getCustomSerialization() && to_type->getCustomName()) { - ret = [requested_result_is_nullable]( - ColumnsWithTypeAndName & arguments, - const DataTypePtr & result_type, - const ColumnNullable * column_nullable, - size_t input_rows_count) -> ColumnPtr - { - auto wrapped_result_type = result_type; - if (requested_result_is_nullable) - wrapped_result_type = makeNullable(result_type); - return ConvertImplGenericFromString::execute( - arguments, wrapped_result_type, column_nullable, input_rows_count); - }; + ret = &ConvertImplGenericFromString::execute; return true; } } @@ -4251,9 +4215,7 @@ private: ErrorCodes::TYPE_MISMATCH, "Wrong result type {}. Expected IPv4", result_type->getName()); const auto * null_map = column_nullable ? &column_nullable->getNullMapData() : nullptr; - if (requested_result_is_nullable) - return convertIPv6ToIPv4(arguments[0].column, null_map); - else if (cast_ipv4_ipv6_default_on_conversion_error_value) + if (cast_ipv4_ipv6_default_on_conversion_error_value || requested_result_is_nullable) return convertIPv6ToIPv4(arguments[0].column, null_map); else return convertIPv6ToIPv4(arguments[0].column, null_map); diff --git a/tests/queries/0_stateless/01556_accurate_cast_or_null.reference b/tests/queries/0_stateless/01556_accurate_cast_or_null.reference index 296b0b553f4..31a9c37421e 100644 --- a/tests/queries/0_stateless/01556_accurate_cast_or_null.reference +++ b/tests/queries/0_stateless/01556_accurate_cast_or_null.reference @@ -42,8 +42,3 @@ 2023-05-30 2149-06-06 1970-01-20 -Tests for issue #38585 -\N -\N -\N -\N diff --git a/tests/queries/0_stateless/01556_accurate_cast_or_null.sql b/tests/queries/0_stateless/01556_accurate_cast_or_null.sql index 5d580a8f32f..f00f6ef837f 100644 --- a/tests/queries/0_stateless/01556_accurate_cast_or_null.sql +++ b/tests/queries/0_stateless/01556_accurate_cast_or_null.sql @@ -49,10 +49,3 @@ SELECT accurateCastOrNull('1xxx', 'Date'); SELECT accurateCastOrNull('2023-05-30', 'Date'); SELECT accurateCastOrNull('2180-01-01', 'Date'); SELECT accurateCastOrNull(19, 'Date'); - -SELECT 'Tests for issue #38585'; -select accurateCastOrNull('test', 'Bool'); -select accurateCast('test', 'Nullable(Bool)'); - -select accurateCastOrNull('test', 'IPv4'); -select accurateCastOrNull('test', 'IPv6'); diff --git a/tests/queries/0_stateless/02303_cast_nullable_to_custom_types.reference b/tests/queries/0_stateless/02303_cast_nullable_to_custom_types.reference index 08ade20ed1d..159ee805f26 100644 --- a/tests/queries/0_stateless/02303_cast_nullable_to_custom_types.reference +++ b/tests/queries/0_stateless/02303_cast_nullable_to_custom_types.reference @@ -17,7 +17,7 @@ true \N 0.0.0.0 \N -\N +0.0.0.0 \N \N \N @@ -25,7 +25,7 @@ true \N :: \N -\N +:: \N \N \N From 154ac14ef60a5426155900bea8d001674ad45dce Mon Sep 17 00:00:00 2001 From: Alexey Milovidov Date: Sat, 9 Sep 2023 05:51:43 +0200 Subject: [PATCH 245/433] Fix error --- tests/ci/clickhouse_helper.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tests/ci/clickhouse_helper.py b/tests/ci/clickhouse_helper.py index ff0e2677759..9460bec189a 100644 --- a/tests/ci/clickhouse_helper.py +++ b/tests/ci/clickhouse_helper.py @@ -296,7 +296,7 @@ class CiLogsCredentials: extra_columns = ( f"{pr_info.number} AS pull_request_number, '{pr_info.sha}' AS commit_sha, " f"toDateTime('{check_start_time}', 'UTC') AS check_start_time, '{check_name}' AS check_name, " - f"'{get_instance_type()}' AS instance_type, '{get_instance_id()}' AS instance_id, " + f"'{get_instance_type()}' AS instance_type, '{get_instance_id()}' AS instance_id" ) return ( f'-e EXTRA_COLUMNS_EXPRESSION="{extra_columns}" ' From b6e9dd96f8f0f2aa74701b5486f9655465433377 Mon Sep 17 00:00:00 2001 From: Igor Nikonov Date: Sat, 9 Sep 2023 06:55:35 +0000 Subject: [PATCH 246/433] Fix build --- src/Interpreters/ClusterProxy/executeQuery.cpp | 2 +- src/Interpreters/ClusterProxy/executeQuery.h | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/src/Interpreters/ClusterProxy/executeQuery.cpp b/src/Interpreters/ClusterProxy/executeQuery.cpp index a8a365255d9..34be2636917 100644 --- a/src/Interpreters/ClusterProxy/executeQuery.cpp +++ b/src/Interpreters/ClusterProxy/executeQuery.cpp @@ -38,7 +38,7 @@ ContextMutablePtr updateSettingsForCluster(bool interserver_mode, ContextPtr context, const Settings & settings, const StorageID & main_table, - const ASTPtr & additional_filter_ast, + ASTPtr additional_filter_ast, Poco::Logger * log) { Settings new_settings = settings; diff --git a/src/Interpreters/ClusterProxy/executeQuery.h b/src/Interpreters/ClusterProxy/executeQuery.h index 03f6e558a92..b663dffb7fa 100644 --- a/src/Interpreters/ClusterProxy/executeQuery.h +++ b/src/Interpreters/ClusterProxy/executeQuery.h @@ -41,7 +41,7 @@ ContextMutablePtr updateSettingsForCluster(bool interserver_mode, ContextPtr context, const Settings & settings, const StorageID & main_table, - const SelectQueryInfo * query_info = nullptr, + ASTPtr additional_filter_ast = nullptr, Poco::Logger * log = nullptr); using AdditionalShardFilterGenerator = std::function; From 5470a5b60fec84c84b22908b706ffc27c3216748 Mon Sep 17 00:00:00 2001 From: Igor Nikonov Date: Sat, 9 Sep 2023 06:58:18 +0000 Subject: [PATCH 247/433] Fix style --- src/Storages/StorageDistributed.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/Storages/StorageDistributed.cpp b/src/Storages/StorageDistributed.cpp index 7c556b1198a..ddc625745b4 100644 --- a/src/Storages/StorageDistributed.cpp +++ b/src/Storages/StorageDistributed.cpp @@ -1316,7 +1316,7 @@ ClusterPtr StorageDistributed::getOptimizedCluster(ContextPtr local_context, con if (has_sharding_key && sharding_key_is_usable) { - ClusterPtr optimized = skipUnusedShards(cluster, select, syntax_analyzer_result, storage_snapshot, local_context); + ClusterPtr optimized = skipUnusedShards(cluster, select, syntax_analyzer_result, storage_snapshot, local_context); if (optimized) return optimized; } From fff2a2d81c126b436b7b49fb1fca749b74409190 Mon Sep 17 00:00:00 2001 From: Igor Nikonov Date: Sat, 9 Sep 2023 08:29:33 +0000 Subject: [PATCH 248/433] Fix style --- src/Storages/StorageDistributed.h | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) diff --git a/src/Storages/StorageDistributed.h b/src/Storages/StorageDistributed.h index 3f28378eaa8..fde414f3123 100644 --- a/src/Storages/StorageDistributed.h +++ b/src/Storages/StorageDistributed.h @@ -188,9 +188,8 @@ private: ClusterPtr getOptimizedCluster(ContextPtr local_context, const StorageSnapshotPtr & storage_snapshot, const ASTSelectQuery & select, const TreeRewriterResultPtr & syntax_analyzer_result) const; - ClusterPtr skipUnusedShards( - ClusterPtr cluster, const ASTSelectQuery & select, const TreeRewriterResultPtr & syntax_analyzer_result, - const StorageSnapshotPtr & storage_snapshot, ContextPtr context) const; + ClusterPtr skipUnusedShards(ClusterPtr cluster, const ASTSelectQuery & select, const TreeRewriterResultPtr & syntax_analyzer_result, + const StorageSnapshotPtr & storage_snapshot, ContextPtr context) const; /// This method returns optimal query processing stage. /// From 1b3da7fc1e1abef16564ea0c3e9b4bfd1808a23b Mon Sep 17 00:00:00 2001 From: serxa Date: Sat, 9 Sep 2023 09:26:18 +0000 Subject: [PATCH 249/433] fix build --- src/IO/IResourceManager.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/IO/IResourceManager.h b/src/IO/IResourceManager.h index b4270cd1935..39fede0e19f 100644 --- a/src/IO/IResourceManager.h +++ b/src/IO/IResourceManager.h @@ -49,7 +49,7 @@ public: /// For introspection, see `system.scheduler` table using VisitorFunc = std::function; - void forEachNode(VisitorFunc visitor) override; + virtual void forEachNode(VisitorFunc visitor) = 0; }; using ResourceManagerPtr = std::shared_ptr; From 9b936c44db62400cca0bf3e3d53dd43898ddd85e Mon Sep 17 00:00:00 2001 From: Nikolai Kochetov Date: Sat, 9 Sep 2023 12:29:39 +0200 Subject: [PATCH 250/433] Revert "Revert "Add settings for real-time updates during query execution"" --- docs/en/operations/settings/settings.md | 8 + src/Client/ClientBase.cpp | 36 +- src/Client/ClientBase.h | 15 + src/Core/Settings.h | 3 + src/Interpreters/Aggregator.cpp | 23 ++ src/Interpreters/Aggregator.h | 2 + src/Processors/Chunk.cpp | 3 +- .../Executors/CompletedPipelineExecutor.cpp | 4 +- src/Processors/Executors/ExecutingGraph.cpp | 1 - .../Executors/ExecutionThreadContext.h | 13 +- src/Processors/Executors/ExecutorTasks.cpp | 15 +- src/Processors/Executors/ExecutorTasks.h | 2 +- src/Processors/Executors/PipelineExecutor.cpp | 5 +- src/Processors/Executors/PipelineExecutor.h | 5 +- .../PullingAsyncPipelineExecutor.cpp | 7 +- .../Executors/PullingAsyncPipelineExecutor.h | 2 +- .../Executors/PullingPipelineExecutor.cpp | 2 +- .../PushingAsyncPipelineExecutor.cpp | 2 +- .../Executors/PushingPipelineExecutor.cpp | 2 +- src/Processors/Formats/IOutputFormat.cpp | 110 ++++-- src/Processors/Formats/IOutputFormat.h | 13 +- .../Formats/Impl/PrettyBlockOutputFormat.cpp | 34 +- .../Formats/Impl/PrettyBlockOutputFormat.h | 6 + .../Impl/PrettyCompactBlockOutputFormat.cpp | 3 +- src/Processors/Formats/LazyOutputFormat.h | 5 +- src/Processors/IProcessor.cpp | 5 + src/Processors/IProcessor.h | 23 ++ src/Processors/LimitTransform.cpp | 9 +- src/Processors/LimitTransform.h | 4 + .../QueryPlan/BuildQueryPipelineSettings.cpp | 7 +- .../QueryPlan/BuildQueryPipelineSettings.h | 3 + src/Processors/QueryPlan/QueryPlan.cpp | 5 + src/Processors/QueryPlan/SortingStep.h | 2 + .../AggregatingPartialResultTransform.cpp | 47 +++ .../AggregatingPartialResultTransform.h | 29 ++ .../Transforms/AggregatingTransform.cpp | 12 + .../Transforms/AggregatingTransform.h | 21 ++ .../Transforms/ExpressionTransform.cpp | 6 + .../Transforms/ExpressionTransform.h | 5 + .../LimitPartialResultTransform.cpp | 42 +++ .../Transforms/LimitPartialResultTransform.h | 36 ++ .../Transforms/LimitsCheckingTransform.cpp | 1 + .../Transforms/LimitsCheckingTransform.h | 2 + .../MergeSortingPartialResultTransform.cpp | 48 +++ .../MergeSortingPartialResultTransform.h | 28 ++ .../Transforms/MergeSortingTransform.cpp | 12 + .../Transforms/MergeSortingTransform.h | 11 + .../Transforms/PartialResultTransform.cpp | 80 +++++ .../Transforms/PartialResultTransform.h | 57 ++++ src/QueryPipeline/Pipe.cpp | 204 +++++++++--- src/QueryPipeline/Pipe.h | 25 +- src/QueryPipeline/QueryPipeline.cpp | 50 ++- src/QueryPipeline/QueryPipeline.h | 7 +- src/QueryPipeline/QueryPipelineBuilder.cpp | 21 +- src/QueryPipeline/QueryPipelineBuilder.h | 6 + src/Server/TCPHandler.cpp | 3 +- .../0_stateless/02010_lc_native.python | 291 +++------------- .../02210_processors_profile_log.reference | 1 + .../02458_insert_select_progress_tcp.python | 244 ++------------ .../02750_settings_alias_tcp_protocol.python | 212 +----------- ...rting_result_during_query_execution.python | 93 ++++++ ...ng_result_during_query_execution.reference | 38 +++ ...l_sorting_result_during_query_execution.sh | 8 + ...ating_result_during_query_execution.python | 125 +++++++ ...ng_result_during_query_execution.reference | 88 +++++ ...gregating_result_during_query_execution.sh | 8 + .../queries/0_stateless/helpers/tcp_client.py | 313 ++++++++++++++++++ 67 files changed, 1779 insertions(+), 774 deletions(-) create mode 100644 src/Processors/Transforms/AggregatingPartialResultTransform.cpp create mode 100644 src/Processors/Transforms/AggregatingPartialResultTransform.h create mode 100644 src/Processors/Transforms/LimitPartialResultTransform.cpp create mode 100644 src/Processors/Transforms/LimitPartialResultTransform.h create mode 100644 src/Processors/Transforms/MergeSortingPartialResultTransform.cpp create mode 100644 src/Processors/Transforms/MergeSortingPartialResultTransform.h create mode 100644 src/Processors/Transforms/PartialResultTransform.cpp create mode 100644 src/Processors/Transforms/PartialResultTransform.h create mode 100755 tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python create mode 100644 tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.reference create mode 100755 tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.sh create mode 100644 tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python create mode 100644 tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.reference create mode 100755 tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.sh create mode 100644 tests/queries/0_stateless/helpers/tcp_client.py diff --git a/docs/en/operations/settings/settings.md b/docs/en/operations/settings/settings.md index 66b8c923d5c..a1989ab3d13 100644 --- a/docs/en/operations/settings/settings.md +++ b/docs/en/operations/settings/settings.md @@ -4644,6 +4644,14 @@ SELECT toFloat64('1.7091'), toFloat64('1.5008753E7') SETTINGS precise_float_pars └─────────────────────┴──────────────────────────┘ ``` +## partial_result_update_duration_ms + +Interval (in milliseconds) for sending updates with partial data about the result table to the client (in interactive mode) during query execution. Setting to 0 disables partial results. Only supported for single-threaded GROUP BY without key, ORDER BY, LIMIT and OFFSET. + +## max_rows_in_partial_result + +Maximum rows to show in the partial result after every real-time update while the query runs (use partial result limit + OFFSET as a value in case of OFFSET in the query). + ## validate_tcp_client_information {#validate-tcp-client-information} Determines whether validation of client information enabled when query packet is received from a client using a TCP connection. diff --git a/src/Client/ClientBase.cpp b/src/Client/ClientBase.cpp index 859665c64b3..cc60095884e 100644 --- a/src/Client/ClientBase.cpp +++ b/src/Client/ClientBase.cpp @@ -441,7 +441,20 @@ void ClientBase::onData(Block & block, ASTPtr parsed_query) if (!block) return; - processed_rows += block.rows(); + if (block.rows() == 0 && partial_result_mode == PartialResultMode::Active) + { + partial_result_mode = PartialResultMode::Inactive; + if (is_interactive) + { + progress_indication.clearProgressOutput(*tty_buf); + std::cout << "Full result:" << std::endl; + progress_indication.writeProgress(*tty_buf); + } + } + + if (partial_result_mode == PartialResultMode::Inactive) + processed_rows += block.rows(); + /// Even if all blocks are empty, we still need to initialize the output stream to write empty resultset. initOutputFormat(block, parsed_query); @@ -451,13 +464,20 @@ void ClientBase::onData(Block & block, ASTPtr parsed_query) if (block.rows() == 0 || (query_fuzzer_runs != 0 && processed_rows >= 100)) return; + if (!is_interactive && partial_result_mode == PartialResultMode::Active) + return; + /// If results are written INTO OUTFILE, we can avoid clearing progress to avoid flicker. if (need_render_progress && tty_buf && (!select_into_file || select_into_file_and_stdout)) progress_indication.clearProgressOutput(*tty_buf); try { - output_format->write(materializeBlock(block)); + if (partial_result_mode == PartialResultMode::Active) + output_format->writePartialResult(materializeBlock(block)); + else + output_format->write(materializeBlock(block)); + written_first_block = true; } catch (const Exception &) @@ -521,6 +541,9 @@ void ClientBase::onProfileInfo(const ProfileInfo & profile_info) void ClientBase::initOutputFormat(const Block & block, ASTPtr parsed_query) try { + if (partial_result_mode == PartialResultMode::NotInit) + partial_result_mode = PartialResultMode::Active; + if (!output_format) { /// Ignore all results when fuzzing as they can be huge. @@ -931,6 +954,14 @@ void ClientBase::processOrdinaryQuery(const String & query_to_execute, ASTPtr pa const auto & settings = global_context->getSettingsRef(); const Int32 signals_before_stop = settings.partial_result_on_first_cancel ? 2 : 1; + bool has_partial_result_setting = settings.partial_result_update_duration_ms.totalMilliseconds() > 0; + + if (has_partial_result_setting) + { + partial_result_mode = PartialResultMode::NotInit; + if (is_interactive) + std::cout << "Partial result:" << std::endl; + } int retries_left = 10; while (retries_left) @@ -1736,6 +1767,7 @@ void ClientBase::processParsedSingleQuery(const String & full_query, const Strin } processed_rows = 0; + partial_result_mode = PartialResultMode::Inactive; written_first_block = false; progress_indication.resetProgress(); profile_events.watch.restart(); diff --git a/src/Client/ClientBase.h b/src/Client/ClientBase.h index eabc79b7432..e2394259f85 100644 --- a/src/Client/ClientBase.h +++ b/src/Client/ClientBase.h @@ -272,6 +272,21 @@ protected: size_t processed_rows = 0; /// How many rows have been read or written. bool print_num_processed_rows = false; /// Whether to print the number of processed rows at + enum class PartialResultMode: UInt8 + { + /// Query doesn't show partial result before the first block with 0 rows. + /// The first block with 0 rows initializes the output table format using its header. + NotInit, + + /// Query shows partial result after the first and before the second block with 0 rows. + /// The second block with 0 rows indicates that that receiving blocks with partial result has been completed and next blocks will be with the full result. + Active, + + /// Query doesn't show partial result at all. + Inactive, + }; + PartialResultMode partial_result_mode = PartialResultMode::Inactive; + bool print_stack_trace = false; /// The last exception that was received from the server. Is used for the /// return code in batch mode. diff --git a/src/Core/Settings.h b/src/Core/Settings.h index 620cc8fd67f..719def2a649 100644 --- a/src/Core/Settings.h +++ b/src/Core/Settings.h @@ -309,6 +309,9 @@ class IColumn; \ M(Bool, partial_result_on_first_cancel, false, "Allows query to return a partial result after cancel.", 0) \ \ + M(Milliseconds, partial_result_update_duration_ms, 0, "Interval (in milliseconds) for sending updates with partial data about the result table to the client (in interactive mode) during query execution. Setting to 0 disables partial results. Only supported for single-threaded GROUP BY without key, ORDER BY, LIMIT and OFFSET.", 0) \ + M(UInt64, max_rows_in_partial_result, 10, "Maximum rows to show in the partial result after every real-time update while the query runs (use partial result limit + OFFSET as a value in case of OFFSET in the query).", 0) \ + \ M(Bool, ignore_on_cluster_for_replicated_udf_queries, false, "Ignore ON CLUSTER clause for replicated UDF management queries.", 0) \ M(Bool, ignore_on_cluster_for_replicated_access_entities_queries, false, "Ignore ON CLUSTER clause for replicated access entities management queries.", 0) \ /** Settings for testing hedged requests */ \ diff --git a/src/Interpreters/Aggregator.cpp b/src/Interpreters/Aggregator.cpp index 9249c3ce4ce..23ee097ebff 100644 --- a/src/Interpreters/Aggregator.cpp +++ b/src/Interpreters/Aggregator.cpp @@ -2272,6 +2272,29 @@ Block Aggregator::prepareBlockAndFillWithoutKey(AggregatedDataVariants & data_va return block; } +Block Aggregator::prepareBlockAndFillWithoutKeySnapshot(AggregatedDataVariants & data_variants) const +{ + size_t rows = 1; + bool final = true; + + auto && out_cols + = prepareOutputBlockColumns(params, aggregate_functions, getHeader(final), data_variants.aggregates_pools, final, rows); + auto && [key_columns, raw_key_columns, aggregate_columns, final_aggregate_columns, aggregate_columns_data] = out_cols; + + AggregatedDataWithoutKey & data = data_variants.without_key; + + /// Always single-thread. It's safe to pass current arena from 'aggregates_pool'. + for (size_t insert_i = 0; insert_i < params.aggregates_size; ++insert_i) + aggregate_functions[insert_i]->insertResultInto( + data + offsets_of_aggregate_states[insert_i], + *final_aggregate_columns[insert_i], + data_variants.aggregates_pool); + + Block block = finalizeBlock(params, getHeader(final), std::move(out_cols), final, rows); + + return block; +} + template Aggregator::ConvertToBlockRes Aggregator::prepareBlockAndFillSingleLevel(AggregatedDataVariants & data_variants, bool final) const diff --git a/src/Interpreters/Aggregator.h b/src/Interpreters/Aggregator.h index 4f2c86606c5..6bfaa76f9b3 100644 --- a/src/Interpreters/Aggregator.h +++ b/src/Interpreters/Aggregator.h @@ -1210,6 +1210,7 @@ private: friend class ConvertingAggregatedToChunksSource; friend class ConvertingAggregatedToChunksWithMergingSource; friend class AggregatingInOrderTransform; + friend class AggregatingPartialResultTransform; /// Data structure of source blocks. Block header; @@ -1391,6 +1392,7 @@ private: std::atomic * is_cancelled = nullptr) const; Block prepareBlockAndFillWithoutKey(AggregatedDataVariants & data_variants, bool final, bool is_overflows) const; + Block prepareBlockAndFillWithoutKeySnapshot(AggregatedDataVariants & data_variants) const; BlocksList prepareBlocksAndFillTwoLevel(AggregatedDataVariants & data_variants, bool final, ThreadPool * thread_pool) const; template diff --git a/src/Processors/Chunk.cpp b/src/Processors/Chunk.cpp index 3839a8963b2..cd442085eca 100644 --- a/src/Processors/Chunk.cpp +++ b/src/Processors/Chunk.cpp @@ -14,7 +14,8 @@ namespace ErrorCodes extern const int POSITION_OUT_OF_BOUND; } -Chunk::Chunk(DB::Columns columns_, UInt64 num_rows_) : columns(std::move(columns_)), num_rows(num_rows_) +Chunk::Chunk(DB::Columns columns_, UInt64 num_rows_) + : columns(std::move(columns_)), num_rows(num_rows_) { checkNumRowsIsConsistent(); } diff --git a/src/Processors/Executors/CompletedPipelineExecutor.cpp b/src/Processors/Executors/CompletedPipelineExecutor.cpp index 598a51bf0c7..c30586e194e 100644 --- a/src/Processors/Executors/CompletedPipelineExecutor.cpp +++ b/src/Processors/Executors/CompletedPipelineExecutor.cpp @@ -75,7 +75,7 @@ void CompletedPipelineExecutor::execute() if (interactive_timeout_ms) { data = std::make_unique(); - data->executor = std::make_shared(pipeline.processors, pipeline.process_list_element); + data->executor = std::make_shared(pipeline.processors, pipeline.process_list_element, pipeline.partial_result_duration_ms); data->executor->setReadProgressCallback(pipeline.getReadProgressCallback()); /// Avoid passing this to lambda, copy ptr to data instead. @@ -105,7 +105,7 @@ void CompletedPipelineExecutor::execute() } else { - PipelineExecutor executor(pipeline.processors, pipeline.process_list_element); + PipelineExecutor executor(pipeline.processors, pipeline.process_list_element, pipeline.partial_result_duration_ms); executor.setReadProgressCallback(pipeline.getReadProgressCallback()); executor.execute(pipeline.getNumThreads(), pipeline.getConcurrencyControl()); } diff --git a/src/Processors/Executors/ExecutingGraph.cpp b/src/Processors/Executors/ExecutingGraph.cpp index 27f6a454b24..6a946b4a4b9 100644 --- a/src/Processors/Executors/ExecutingGraph.cpp +++ b/src/Processors/Executors/ExecutingGraph.cpp @@ -260,7 +260,6 @@ bool ExecutingGraph::updateNode(uint64_t pid, Queue & queue, Queue & async_queue { pid = updated_processors.top(); updated_processors.pop(); - /// In this method we have ownership on node. auto & node = *nodes[pid]; diff --git a/src/Processors/Executors/ExecutionThreadContext.h b/src/Processors/Executors/ExecutionThreadContext.h index eb048f8ab09..85788a70771 100644 --- a/src/Processors/Executors/ExecutionThreadContext.h +++ b/src/Processors/Executors/ExecutionThreadContext.h @@ -30,6 +30,12 @@ private: /// Callback for read progress. ReadProgressCallback * read_progress_callback = nullptr; + /// Timer that stops optimization of running local tasks instead of queuing them. + /// It provides local progress for each IProcessor task, allowing the partial result of the request to be always sended to the user. + Stopwatch watch; + /// Time period that limits the maximum allowed duration for optimizing the scheduling of local tasks within the executor + const UInt64 partial_result_duration_ms; + public: #ifndef NDEBUG /// Time for different processing stages. @@ -62,8 +68,13 @@ public: void setException(std::exception_ptr exception_) { exception = exception_; } void rethrowExceptionIfHas(); - explicit ExecutionThreadContext(size_t thread_number_, bool profile_processors_, bool trace_processors_, ReadProgressCallback * callback) + bool needWatchRestartForPartialResultProgress() { return partial_result_duration_ms != 0 && partial_result_duration_ms < watch.elapsedMilliseconds(); } + void restartWatch() { watch.restart(); } + + explicit ExecutionThreadContext(size_t thread_number_, bool profile_processors_, bool trace_processors_, ReadProgressCallback * callback, UInt64 partial_result_duration_ms_) : read_progress_callback(callback) + , watch(CLOCK_MONOTONIC) + , partial_result_duration_ms(partial_result_duration_ms_) , thread_number(thread_number_) , profile_processors(profile_processors_) , trace_processors(trace_processors_) diff --git a/src/Processors/Executors/ExecutorTasks.cpp b/src/Processors/Executors/ExecutorTasks.cpp index e61d225a968..08920592391 100644 --- a/src/Processors/Executors/ExecutorTasks.cpp +++ b/src/Processors/Executors/ExecutorTasks.cpp @@ -108,8 +108,15 @@ void ExecutorTasks::pushTasks(Queue & queue, Queue & async_queue, ExecutionThrea { context.setTask(nullptr); - /// Take local task from queue if has one. - if (!queue.empty() && !context.hasAsyncTasks()) + /// If sending partial results is allowed and local tasks scheduling optimization is repeated longer than the limit + /// or new task need to send partial result later, skip optimization for this iteration. + /// Otherwise take local task from queue if has one. + if ((!queue.empty() && queue.front()->processor->isPartialResultProcessor()) + || context.needWatchRestartForPartialResultProgress()) + { + context.restartWatch(); + } + else if (!queue.empty() && !context.hasAsyncTasks()) { context.setTask(queue.front()); queue.pop(); @@ -139,7 +146,7 @@ void ExecutorTasks::pushTasks(Queue & queue, Queue & async_queue, ExecutionThrea } } -void ExecutorTasks::init(size_t num_threads_, size_t use_threads_, bool profile_processors, bool trace_processors, ReadProgressCallback * callback) +void ExecutorTasks::init(size_t num_threads_, size_t use_threads_, bool profile_processors, bool trace_processors, ReadProgressCallback * callback, UInt64 partial_result_duration_ms) { num_threads = num_threads_; use_threads = use_threads_; @@ -151,7 +158,7 @@ void ExecutorTasks::init(size_t num_threads_, size_t use_threads_, bool profile_ executor_contexts.reserve(num_threads); for (size_t i = 0; i < num_threads; ++i) - executor_contexts.emplace_back(std::make_unique(i, profile_processors, trace_processors, callback)); + executor_contexts.emplace_back(std::make_unique(i, profile_processors, trace_processors, callback, partial_result_duration_ms)); } } diff --git a/src/Processors/Executors/ExecutorTasks.h b/src/Processors/Executors/ExecutorTasks.h index d35f8de94d1..ab6d5e91411 100644 --- a/src/Processors/Executors/ExecutorTasks.h +++ b/src/Processors/Executors/ExecutorTasks.h @@ -58,7 +58,7 @@ public: void tryGetTask(ExecutionThreadContext & context); void pushTasks(Queue & queue, Queue & async_queue, ExecutionThreadContext & context); - void init(size_t num_threads_, size_t use_threads_, bool profile_processors, bool trace_processors, ReadProgressCallback * callback); + void init(size_t num_threads_, size_t use_threads_, bool profile_processors, bool trace_processors, ReadProgressCallback * callback, UInt64 partial_result_duration_ms); void fill(Queue & queue); void upscale(size_t use_threads_); diff --git a/src/Processors/Executors/PipelineExecutor.cpp b/src/Processors/Executors/PipelineExecutor.cpp index 37af391fba3..77779e2cec2 100644 --- a/src/Processors/Executors/PipelineExecutor.cpp +++ b/src/Processors/Executors/PipelineExecutor.cpp @@ -33,8 +33,9 @@ namespace ErrorCodes } -PipelineExecutor::PipelineExecutor(std::shared_ptr & processors, QueryStatusPtr elem) +PipelineExecutor::PipelineExecutor(std::shared_ptr & processors, QueryStatusPtr elem, UInt64 partial_result_duration_ms_) : process_list_element(std::move(elem)) + , partial_result_duration_ms(partial_result_duration_ms_) { if (process_list_element) { @@ -328,7 +329,7 @@ void PipelineExecutor::initializeExecution(size_t num_threads, bool concurrency_ Queue queue; graph->initializeExecution(queue); - tasks.init(num_threads, use_threads, profile_processors, trace_processors, read_progress_callback.get()); + tasks.init(num_threads, use_threads, profile_processors, trace_processors, read_progress_callback.get(), partial_result_duration_ms); tasks.fill(queue); if (num_threads > 1) diff --git a/src/Processors/Executors/PipelineExecutor.h b/src/Processors/Executors/PipelineExecutor.h index dee12dad282..6cb0e6c4ac1 100644 --- a/src/Processors/Executors/PipelineExecutor.h +++ b/src/Processors/Executors/PipelineExecutor.h @@ -33,7 +33,7 @@ public: /// During pipeline execution new processors can appear. They will be added to existing set. /// /// Explicit graph representation is built in constructor. Throws if graph is not correct. - explicit PipelineExecutor(std::shared_ptr & processors, QueryStatusPtr elem); + explicit PipelineExecutor(std::shared_ptr & processors, QueryStatusPtr elem, UInt64 partial_result_duration_ms_ = 0); ~PipelineExecutor(); /// Execute pipeline in multiple threads. Must be called once. @@ -90,6 +90,9 @@ private: ReadProgressCallbackPtr read_progress_callback; + /// Duration between sending partial result through the pipeline + const UInt64 partial_result_duration_ms; + using Queue = std::queue; void initializeExecution(size_t num_threads, bool concurrency_control); /// Initialize executor contexts and task_queue. diff --git a/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp b/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp index 345bec395b2..95a2022bf93 100644 --- a/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp +++ b/src/Processors/Executors/PullingAsyncPipelineExecutor.cpp @@ -41,12 +41,13 @@ struct PullingAsyncPipelineExecutor::Data } }; -PullingAsyncPipelineExecutor::PullingAsyncPipelineExecutor(QueryPipeline & pipeline_) : pipeline(pipeline_) +PullingAsyncPipelineExecutor::PullingAsyncPipelineExecutor(QueryPipeline & pipeline_, bool has_partial_result_setting) : pipeline(pipeline_) { if (!pipeline.pulling()) throw Exception(ErrorCodes::LOGICAL_ERROR, "Pipeline for PullingAsyncPipelineExecutor must be pulling"); - lazy_format = std::make_shared(pipeline.output->getHeader()); + lazy_format = std::make_shared(pipeline.output->getHeader(), /*is_partial_result_protocol_active*/ has_partial_result_setting); + pipeline.complete(lazy_format); } @@ -103,7 +104,7 @@ bool PullingAsyncPipelineExecutor::pull(Chunk & chunk, uint64_t milliseconds) if (!data) { data = std::make_unique(); - data->executor = std::make_shared(pipeline.processors, pipeline.process_list_element); + data->executor = std::make_shared(pipeline.processors, pipeline.process_list_element, pipeline.partial_result_duration_ms); data->executor->setReadProgressCallback(pipeline.getReadProgressCallback()); data->lazy_format = lazy_format.get(); diff --git a/src/Processors/Executors/PullingAsyncPipelineExecutor.h b/src/Processors/Executors/PullingAsyncPipelineExecutor.h index 361bcc0155c..202ecbf281b 100644 --- a/src/Processors/Executors/PullingAsyncPipelineExecutor.h +++ b/src/Processors/Executors/PullingAsyncPipelineExecutor.h @@ -21,7 +21,7 @@ struct ProfileInfo; class PullingAsyncPipelineExecutor { public: - explicit PullingAsyncPipelineExecutor(QueryPipeline & pipeline_); + explicit PullingAsyncPipelineExecutor(QueryPipeline & pipeline_, bool has_partial_result_setting = false); ~PullingAsyncPipelineExecutor(); /// Get structure of returned block or chunk. diff --git a/src/Processors/Executors/PullingPipelineExecutor.cpp b/src/Processors/Executors/PullingPipelineExecutor.cpp index cbf73c5cb07..f79f15c19bf 100644 --- a/src/Processors/Executors/PullingPipelineExecutor.cpp +++ b/src/Processors/Executors/PullingPipelineExecutor.cpp @@ -44,7 +44,7 @@ bool PullingPipelineExecutor::pull(Chunk & chunk) { if (!executor) { - executor = std::make_shared(pipeline.processors, pipeline.process_list_element); + executor = std::make_shared(pipeline.processors, pipeline.process_list_element, pipeline.partial_result_duration_ms); executor->setReadProgressCallback(pipeline.getReadProgressCallback()); } diff --git a/src/Processors/Executors/PushingAsyncPipelineExecutor.cpp b/src/Processors/Executors/PushingAsyncPipelineExecutor.cpp index a816ab9ca7f..f3ed24e7e96 100644 --- a/src/Processors/Executors/PushingAsyncPipelineExecutor.cpp +++ b/src/Processors/Executors/PushingAsyncPipelineExecutor.cpp @@ -167,7 +167,7 @@ void PushingAsyncPipelineExecutor::start() started = true; data = std::make_unique(); - data->executor = std::make_shared(pipeline.processors, pipeline.process_list_element); + data->executor = std::make_shared(pipeline.processors, pipeline.process_list_element, pipeline.partial_result_duration_ms); data->executor->setReadProgressCallback(pipeline.getReadProgressCallback()); data->source = pushing_source.get(); diff --git a/src/Processors/Executors/PushingPipelineExecutor.cpp b/src/Processors/Executors/PushingPipelineExecutor.cpp index 696932932df..f2b018792c7 100644 --- a/src/Processors/Executors/PushingPipelineExecutor.cpp +++ b/src/Processors/Executors/PushingPipelineExecutor.cpp @@ -87,7 +87,7 @@ void PushingPipelineExecutor::start() return; started = true; - executor = std::make_shared(pipeline.processors, pipeline.process_list_element); + executor = std::make_shared(pipeline.processors, pipeline.process_list_element, pipeline.partial_result_duration_ms); executor->setReadProgressCallback(pipeline.getReadProgressCallback()); if (!executor->executeStep(&input_wait_flag)) diff --git a/src/Processors/Formats/IOutputFormat.cpp b/src/Processors/Formats/IOutputFormat.cpp index 88a6fb1e92f..e691e32a7bc 100644 --- a/src/Processors/Formats/IOutputFormat.cpp +++ b/src/Processors/Formats/IOutputFormat.cpp @@ -1,40 +1,89 @@ #include #include +#include namespace DB { -IOutputFormat::IOutputFormat(const Block & header_, WriteBuffer & out_) - : IProcessor({header_, header_, header_}, {}), out(out_) +IOutputFormat::IOutputFormat(const Block & header_, WriteBuffer & out_, bool is_partial_result_protocol_active_) + : IProcessor({header_, header_, header_, header_}, {}) + , out(out_) + , is_partial_result_protocol_active(is_partial_result_protocol_active_) { } +void IOutputFormat::setCurrentChunk(InputPort & input, PortKind kind) +{ + current_chunk = input.pull(true); + current_block_kind = kind; + has_input = true; +} + +IOutputFormat::Status IOutputFormat::prepareMainAndPartialResult() +{ + bool need_data = false; + for (auto kind : {Main, PartialResult}) + { + auto & input = getPort(kind); + + if (input.isFinished()) + continue; + + if (kind == PartialResult && main_input_activated) + { + input.close(); + continue; + } + + input.setNeeded(); + need_data = true; + + if (!input.hasData()) + continue; + + setCurrentChunk(input, kind); + return Status::Ready; + } + + if (need_data) + return Status::NeedData; + + return Status::Finished; +} + +IOutputFormat::Status IOutputFormat::prepareTotalsAndExtremes() +{ + for (auto kind : {Totals, Extremes}) + { + auto & input = getPort(kind); + + if (!input.isConnected() || input.isFinished()) + continue; + + input.setNeeded(); + if (!input.hasData()) + return Status::NeedData; + + setCurrentChunk(input, kind); + return Status::Ready; + } + + return Status::Finished; +} + IOutputFormat::Status IOutputFormat::prepare() { if (has_input) return Status::Ready; - for (auto kind : {Main, Totals, Extremes}) - { - auto & input = getPort(kind); + auto status = prepareMainAndPartialResult(); + if (status != Status::Finished) + return status; - if (kind != Main && !input.isConnected()) - continue; - - if (input.isFinished()) - continue; - - input.setNeeded(); - - if (!input.hasData()) - return Status::NeedData; - - current_chunk = input.pull(true); - current_block_kind = kind; - has_input = true; - return Status::Ready; - } + status = prepareTotalsAndExtremes(); + if (status != Status::Finished) + return status; finished = true; @@ -83,8 +132,18 @@ void IOutputFormat::work() case Main: result_rows += current_chunk.getNumRows(); result_bytes += current_chunk.allocatedBytes(); + if (is_partial_result_protocol_active && !main_input_activated && current_chunk.hasRows()) + { + /// Sending an empty block signals to the client that partial results are terminated, + /// and only data from the main pipeline will be forwarded. + consume(Chunk(current_chunk.cloneEmptyColumns(), 0)); + main_input_activated = true; + } consume(std::move(current_chunk)); break; + case PartialResult: + consumePartialResult(std::move(current_chunk)); + break; case Totals: writeSuffixIfNeeded(); if (auto totals = prepareTotals(std::move(current_chunk))) @@ -119,6 +178,15 @@ void IOutputFormat::write(const Block & block) flush(); } +void IOutputFormat::writePartialResult(const Block & block) +{ + writePrefixIfNeeded(); + consumePartialResult(Chunk(block.getColumns(), block.rows())); + + if (auto_flush) + flush(); +} + void IOutputFormat::finalize() { if (finalized) diff --git a/src/Processors/Formats/IOutputFormat.h b/src/Processors/Formats/IOutputFormat.h index 58700a978ff..470d24e9a22 100644 --- a/src/Processors/Formats/IOutputFormat.h +++ b/src/Processors/Formats/IOutputFormat.h @@ -23,9 +23,9 @@ class WriteBuffer; class IOutputFormat : public IProcessor { public: - enum PortKind { Main = 0, Totals = 1, Extremes = 2 }; + enum PortKind { Main = 0, Totals = 1, Extremes = 2, PartialResult = 3 }; - IOutputFormat(const Block & header_, WriteBuffer & out_); + IOutputFormat(const Block & header_, WriteBuffer & out_, bool is_partial_result_protocol_active_ = false); Status prepare() override; void work() override; @@ -54,6 +54,7 @@ public: /// TODO: separate formats and processors. void write(const Block & block); + void writePartialResult(const Block & block); void finalize(); @@ -118,6 +119,7 @@ protected: virtual void consume(Chunk) = 0; virtual void consumeTotals(Chunk) {} virtual void consumeExtremes(Chunk) {} + virtual void consumePartialResult(Chunk) {} virtual void finalizeImpl() {} virtual void finalizeBuffers() {} virtual void writePrefix() {} @@ -166,6 +168,7 @@ protected: Chunk current_chunk; PortKind current_block_kind = PortKind::Main; + bool main_input_activated = false; bool has_input = false; bool finished = false; bool finalized = false; @@ -180,9 +183,15 @@ protected: Statistics statistics; private: + void setCurrentChunk(InputPort & input, PortKind kind); + IOutputFormat::Status prepareMainAndPartialResult(); + IOutputFormat::Status prepareTotalsAndExtremes(); + size_t rows_read_before = 0; bool are_totals_written = false; + bool is_partial_result_protocol_active = false; + /// Counters for consumed chunks. Are used for QueryLog. size_t result_rows = 0; size_t result_bytes = 0; diff --git a/src/Processors/Formats/Impl/PrettyBlockOutputFormat.cpp b/src/Processors/Formats/Impl/PrettyBlockOutputFormat.cpp index 14648e68f94..6fa891297f6 100644 --- a/src/Processors/Formats/Impl/PrettyBlockOutputFormat.cpp +++ b/src/Processors/Formats/Impl/PrettyBlockOutputFormat.cpp @@ -134,7 +134,8 @@ void PrettyBlockOutputFormat::write(Chunk chunk, PortKind port_kind) { if (total_rows >= format_settings.pretty.max_rows) { - total_rows += chunk.getNumRows(); + if (port_kind != PortKind::PartialResult) + total_rows += chunk.getNumRows(); return; } if (mono_block) @@ -315,7 +316,8 @@ void PrettyBlockOutputFormat::writeChunk(const Chunk & chunk, PortKind port_kind } writeString(bottom_separator_s, out); - total_rows += num_rows; + if (port_kind != PortKind::PartialResult) + total_rows += num_rows; } @@ -388,6 +390,34 @@ void PrettyBlockOutputFormat::consumeExtremes(Chunk chunk) write(std::move(chunk), PortKind::Extremes); } +void PrettyBlockOutputFormat::clearLastLines(size_t lines_number) +{ + /// http://en.wikipedia.org/wiki/ANSI_escape_code + #define MOVE_TO_PREV_LINE "\033[A" + #define CLEAR_TO_END_OF_LINE "\033[K" + + static const char * clear_prev_line = MOVE_TO_PREV_LINE \ + CLEAR_TO_END_OF_LINE; + + /// Move cursor to the beginning of line + writeCString("\r", out); + + for (size_t line = 0; line < lines_number; ++line) + { + writeCString(clear_prev_line, out); + } +} + +void PrettyBlockOutputFormat::consumePartialResult(Chunk chunk) +{ + if (prev_partial_block_rows > 0) + /// number of rows + header line + footer line + clearLastLines(prev_partial_block_rows + 2); + + prev_partial_block_rows = chunk.getNumRows(); + write(std::move(chunk), PortKind::PartialResult); +} + void PrettyBlockOutputFormat::writeMonoChunkIfNeeded() { diff --git a/src/Processors/Formats/Impl/PrettyBlockOutputFormat.h b/src/Processors/Formats/Impl/PrettyBlockOutputFormat.h index dfb23ac63f9..92466dce3ff 100644 --- a/src/Processors/Formats/Impl/PrettyBlockOutputFormat.h +++ b/src/Processors/Formats/Impl/PrettyBlockOutputFormat.h @@ -28,7 +28,12 @@ protected: void consumeTotals(Chunk) override; void consumeExtremes(Chunk) override; + void clearLastLines(size_t lines_number); + void consumePartialResult(Chunk) override; + size_t total_rows = 0; + size_t prev_partial_block_rows = 0; + size_t row_number_width = 7; // "10000. " const FormatSettings format_settings; @@ -55,6 +60,7 @@ protected: void resetFormatterImpl() override { total_rows = 0; + prev_partial_block_rows = 0; } private: diff --git a/src/Processors/Formats/Impl/PrettyCompactBlockOutputFormat.cpp b/src/Processors/Formats/Impl/PrettyCompactBlockOutputFormat.cpp index 2ba9ec725e2..3a04d86b1ad 100644 --- a/src/Processors/Formats/Impl/PrettyCompactBlockOutputFormat.cpp +++ b/src/Processors/Formats/Impl/PrettyCompactBlockOutputFormat.cpp @@ -194,7 +194,8 @@ void PrettyCompactBlockOutputFormat::writeChunk(const Chunk & chunk, PortKind po writeBottom(max_widths); - total_rows += num_rows; + if (port_kind != PortKind::PartialResult) + total_rows += num_rows; } diff --git a/src/Processors/Formats/LazyOutputFormat.h b/src/Processors/Formats/LazyOutputFormat.h index 9cf609ed2d7..bbcfdbb7193 100644 --- a/src/Processors/Formats/LazyOutputFormat.h +++ b/src/Processors/Formats/LazyOutputFormat.h @@ -14,8 +14,8 @@ class LazyOutputFormat : public IOutputFormat { public: - explicit LazyOutputFormat(const Block & header) - : IOutputFormat(header, out), queue(2) {} + explicit LazyOutputFormat(const Block & header, bool is_partial_result_protocol_active = false) + : IOutputFormat(header, out, is_partial_result_protocol_active), queue(2) {} String getName() const override { return "LazyOutputFormat"; } @@ -49,6 +49,7 @@ protected: void consumeTotals(Chunk chunk) override { totals = std::move(chunk); } void consumeExtremes(Chunk chunk) override { extremes = std::move(chunk); } + void consumePartialResult(Chunk chunk) override { consume(std::move(chunk)); } private: diff --git a/src/Processors/IProcessor.cpp b/src/Processors/IProcessor.cpp index 8b160153733..2f294a32531 100644 --- a/src/Processors/IProcessor.cpp +++ b/src/Processors/IProcessor.cpp @@ -40,5 +40,10 @@ std::string IProcessor::statusToName(Status status) UNREACHABLE(); } +ProcessorPtr IProcessor::getPartialResultProcessorPtr(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) +{ + return current_processor->getPartialResultProcessor(current_processor, partial_result_limit, partial_result_duration_ms); +} + } diff --git a/src/Processors/IProcessor.h b/src/Processors/IProcessor.h index c6bef186877..51a0bb1c121 100644 --- a/src/Processors/IProcessor.h +++ b/src/Processors/IProcessor.h @@ -164,6 +164,8 @@ public: static std::string statusToName(Status status); + static ProcessorPtr getPartialResultProcessorPtr(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms); + /** Method 'prepare' is responsible for all cheap ("instantaneous": O(1) of data volume, no wait) calculations. * * It may access input and output ports, @@ -235,6 +237,22 @@ public: throw Exception(ErrorCodes::NOT_IMPLEMENTED, "Method 'expandPipeline' is not implemented for {} processor", getName()); } + enum class PartialResultStatus + { + /// Processor currently doesn't support work with the partial result pipeline. + NotSupported, + + /// Processor can be skipped in the partial result pipeline. + SkipSupported, + + /// Processor creates a light-weight copy of itself in the partial result pipeline. + /// The copy can create snapshots of the original processor or transform small blocks of data in the same way as the original processor + FullSupported, + }; + + virtual bool isPartialResultProcessor() const { return false; } + virtual PartialResultStatus getPartialResultProcessorSupportStatus() const { return PartialResultStatus::NotSupported; } + /// In case if query was cancelled executor will wait till all processors finish their jobs. /// Generally, there is no reason to check this flag. However, it may be reasonable for long operations (e.g. i/o). bool isCancelled() const { return is_cancelled.load(std::memory_order_acquire); } @@ -369,6 +387,11 @@ public: protected: virtual void onCancel() {} + virtual ProcessorPtr getPartialResultProcessor(const ProcessorPtr & /*current_processor*/, UInt64 /*partial_result_limit*/, UInt64 /*partial_result_duration_ms*/) + { + throw Exception(ErrorCodes::NOT_IMPLEMENTED, "Method 'getPartialResultProcessor' is not implemented for {} processor", getName()); + } + private: /// For: /// - elapsed_us diff --git a/src/Processors/LimitTransform.cpp b/src/Processors/LimitTransform.cpp index 5e24062d67a..b2bf3c28eee 100644 --- a/src/Processors/LimitTransform.cpp +++ b/src/Processors/LimitTransform.cpp @@ -1,5 +1,5 @@ #include - +#include namespace DB { @@ -180,7 +180,6 @@ LimitTransform::Status LimitTransform::preparePair(PortsData & data) return Status::NeedData; data.current_chunk = input.pull(true); - auto rows = data.current_chunk.getNumRows(); if (rows_before_limit_at_least && !data.input_port_has_counter) @@ -367,5 +366,11 @@ bool LimitTransform::sortColumnsEqualAt(const ColumnRawPtrs & current_chunk_sort return true; } +ProcessorPtr LimitTransform::getPartialResultProcessor(const ProcessorPtr & /*current_processor*/, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) +{ + const auto & header = inputs.front().getHeader(); + return std::make_shared(header, partial_result_limit, partial_result_duration_ms, limit, offset); +} + } diff --git a/src/Processors/LimitTransform.h b/src/Processors/LimitTransform.h index 33ff968985f..eac5f9e8d6d 100644 --- a/src/Processors/LimitTransform.h +++ b/src/Processors/LimitTransform.h @@ -55,6 +55,8 @@ private: ColumnRawPtrs extractSortColumns(const Columns & columns) const; bool sortColumnsEqualAt(const ColumnRawPtrs & current_chunk_sort_columns, UInt64 current_chunk_row_num) const; + ProcessorPtr getPartialResultProcessor(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) override; + public: LimitTransform( const Block & header_, UInt64 limit_, UInt64 offset_, size_t num_streams = 1, @@ -73,6 +75,8 @@ public: void setRowsBeforeLimitCounter(RowsBeforeLimitCounterPtr counter) override { rows_before_limit_at_least.swap(counter); } void setInputPortHasCounter(size_t pos) { ports_data[pos].input_port_has_counter = true; } + + PartialResultStatus getPartialResultProcessorSupportStatus() const override { return PartialResultStatus::FullSupported; } }; } diff --git a/src/Processors/QueryPlan/BuildQueryPipelineSettings.cpp b/src/Processors/QueryPlan/BuildQueryPipelineSettings.cpp index fb3ed7f80fc..60ac30389a1 100644 --- a/src/Processors/QueryPlan/BuildQueryPipelineSettings.cpp +++ b/src/Processors/QueryPlan/BuildQueryPipelineSettings.cpp @@ -9,7 +9,12 @@ namespace DB BuildQueryPipelineSettings BuildQueryPipelineSettings::fromContext(ContextPtr from) { BuildQueryPipelineSettings settings; - settings.actions_settings = ExpressionActionsSettings::fromSettings(from->getSettingsRef(), CompileExpressions::yes); + + const auto & context_settings = from->getSettingsRef(); + settings.partial_result_limit = context_settings.max_rows_in_partial_result; + settings.partial_result_duration_ms = context_settings.partial_result_update_duration_ms.totalMilliseconds(); + + settings.actions_settings = ExpressionActionsSettings::fromSettings(context_settings, CompileExpressions::yes); settings.process_list_element = from->getProcessListElement(); settings.progress_callback = from->getProgressCallback(); return settings; diff --git a/src/Processors/QueryPlan/BuildQueryPipelineSettings.h b/src/Processors/QueryPlan/BuildQueryPipelineSettings.h index 3b5e4e06953..0410bf925d1 100644 --- a/src/Processors/QueryPlan/BuildQueryPipelineSettings.h +++ b/src/Processors/QueryPlan/BuildQueryPipelineSettings.h @@ -19,6 +19,9 @@ struct BuildQueryPipelineSettings QueryStatusPtr process_list_element; ProgressCallback progress_callback = nullptr; + UInt64 partial_result_limit = 0; + UInt64 partial_result_duration_ms = 0; + const ExpressionActionsSettings & getActionsSettings() const { return actions_settings; } static BuildQueryPipelineSettings fromContext(ContextPtr from); }; diff --git a/src/Processors/QueryPlan/QueryPlan.cpp b/src/Processors/QueryPlan/QueryPlan.cpp index 8054209c1c3..5d38bfb42c4 100644 --- a/src/Processors/QueryPlan/QueryPlan.cpp +++ b/src/Processors/QueryPlan/QueryPlan.cpp @@ -168,6 +168,8 @@ QueryPipelineBuilderPtr QueryPlan::buildQueryPipeline( QueryPipelineBuilderPtr last_pipeline; + bool has_partial_result_setting = build_pipeline_settings.partial_result_duration_ms > 0; + std::stack stack; stack.push(Frame{.node = root}); @@ -194,6 +196,9 @@ QueryPipelineBuilderPtr QueryPlan::buildQueryPipeline( } else stack.push(Frame{.node = frame.node->children[next_child]}); + + if (has_partial_result_setting && last_pipeline && !last_pipeline->isPartialResultActive()) + last_pipeline->activatePartialResult(build_pipeline_settings.partial_result_limit, build_pipeline_settings.partial_result_duration_ms); } last_pipeline->setProgressCallback(build_pipeline_settings.progress_callback); diff --git a/src/Processors/QueryPlan/SortingStep.h b/src/Processors/QueryPlan/SortingStep.h index 371a24ac6f2..a72cab05754 100644 --- a/src/Processors/QueryPlan/SortingStep.h +++ b/src/Processors/QueryPlan/SortingStep.h @@ -27,6 +27,8 @@ public: size_t max_bytes_before_external_sort = 0; TemporaryDataOnDiskScopePtr tmp_data = nullptr; size_t min_free_disk_space = 0; + UInt64 partial_result_limit = 0; + UInt64 partial_result_duration_ms = 0; explicit Settings(const Context & context); explicit Settings(size_t max_block_size_); diff --git a/src/Processors/Transforms/AggregatingPartialResultTransform.cpp b/src/Processors/Transforms/AggregatingPartialResultTransform.cpp new file mode 100644 index 00000000000..cf8ce72e096 --- /dev/null +++ b/src/Processors/Transforms/AggregatingPartialResultTransform.cpp @@ -0,0 +1,47 @@ +#include + +namespace DB +{ + +AggregatingPartialResultTransform::AggregatingPartialResultTransform( + const Block & input_header, const Block & output_header, AggregatingTransformPtr aggregating_transform_, + UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_) + : PartialResultTransform(input_header, output_header, partial_result_limit_, partial_result_duration_ms_) + , aggregating_transform(std::move(aggregating_transform_)) + , transform_aggregator(input_header, aggregating_transform->params->params) + {} + +void AggregatingPartialResultTransform::transformPartialResult(Chunk & chunk) +{ + auto & params = aggregating_transform->params->params; + + bool no_more_keys = false; + AggregatedDataVariants variants; + ColumnRawPtrs key_columns(params.keys_size); + Aggregator::AggregateColumns aggregate_columns(params.aggregates_size); + + const UInt64 num_rows = chunk.getNumRows(); + transform_aggregator.executeOnBlock(chunk.detachColumns(), 0, num_rows, variants, key_columns, aggregate_columns, no_more_keys); + + auto transformed_block = transform_aggregator.convertToBlocks(variants, /*final*/ true, /*max_threads*/ 1).front(); + + chunk = convertToChunk(transformed_block); +} + +PartialResultTransform::ShaphotResult AggregatingPartialResultTransform::getRealProcessorSnapshot() +{ + std::lock_guard lock(aggregating_transform->snapshot_mutex); + if (aggregating_transform->is_generate_initialized) + return {{}, SnaphotStatus::Stopped}; + + if (aggregating_transform->variants.empty()) + return {{}, SnaphotStatus::NotReady}; + + auto & snapshot_aggregator = aggregating_transform->params->aggregator; + auto & snapshot_variants = aggregating_transform->many_data->variants; + auto block = snapshot_aggregator.prepareBlockAndFillWithoutKeySnapshot(*snapshot_variants.at(0)); + + return {convertToChunk(block), SnaphotStatus::Ready}; +} + +} diff --git a/src/Processors/Transforms/AggregatingPartialResultTransform.h b/src/Processors/Transforms/AggregatingPartialResultTransform.h new file mode 100644 index 00000000000..f7bac3a5394 --- /dev/null +++ b/src/Processors/Transforms/AggregatingPartialResultTransform.h @@ -0,0 +1,29 @@ +#pragma once + +#include +#include +#include + +namespace DB +{ + +class AggregatingPartialResultTransform : public PartialResultTransform +{ +public: + using AggregatingTransformPtr = std::shared_ptr; + + AggregatingPartialResultTransform( + const Block & input_header, const Block & output_header, AggregatingTransformPtr aggregating_transform_, + UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_); + + String getName() const override { return "AggregatingPartialResultTransform"; } + + void transformPartialResult(Chunk & chunk) override; + ShaphotResult getRealProcessorSnapshot() override; + +private: + AggregatingTransformPtr aggregating_transform; + Aggregator transform_aggregator; +}; + +} diff --git a/src/Processors/Transforms/AggregatingTransform.cpp b/src/Processors/Transforms/AggregatingTransform.cpp index 4bd000797a6..b4d2785bed2 100644 --- a/src/Processors/Transforms/AggregatingTransform.cpp +++ b/src/Processors/Transforms/AggregatingTransform.cpp @@ -1,3 +1,4 @@ +#include #include #include @@ -657,6 +658,8 @@ void AggregatingTransform::consume(Chunk chunk) src_rows += num_rows; src_bytes += chunk.bytes(); + std::lock_guard lock(snapshot_mutex); + if (params->params.only_merge) { auto block = getInputs().front().getHeader().cloneWithColumns(chunk.detachColumns()); @@ -676,6 +679,7 @@ void AggregatingTransform::initGenerate() if (is_generate_initialized) return; + std::lock_guard lock(snapshot_mutex); is_generate_initialized = true; /// If there was no data, and we aggregate without keys, and we must return single row with the result of empty aggregation. @@ -806,4 +810,12 @@ void AggregatingTransform::initGenerate() } } +ProcessorPtr AggregatingTransform::getPartialResultProcessor(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) +{ + const auto & input_header = inputs.front().getHeader(); + const auto & output_header = outputs.front().getHeader(); + auto aggregating_processor = std::dynamic_pointer_cast(current_processor); + return std::make_shared(input_header, output_header, std::move(aggregating_processor), partial_result_limit, partial_result_duration_ms); +} + } diff --git a/src/Processors/Transforms/AggregatingTransform.h b/src/Processors/Transforms/AggregatingTransform.h index 38baa4d0394..791cd12326f 100644 --- a/src/Processors/Transforms/AggregatingTransform.h +++ b/src/Processors/Transforms/AggregatingTransform.h @@ -170,9 +170,23 @@ public: void work() override; Processors expandPipeline() override; + PartialResultStatus getPartialResultProcessorSupportStatus() const override + { + /// Currently AggregatingPartialResultTransform support only single-thread aggregation without key. + + /// TODO: check that insert results from aggregator.prepareBlockAndFillWithoutKey return values without + /// changing of the aggregator state when aggregation with keys will be supported in AggregatingPartialResultTransform. + bool is_partial_result_supported = params->params.keys_size == 0 /// Aggregation without key. + && many_data->variants.size() == 1; /// Use only one stream for aggregation. + + return is_partial_result_supported ? PartialResultStatus::FullSupported : PartialResultStatus::NotSupported; + } + protected: void consume(Chunk chunk); + ProcessorPtr getPartialResultProcessor(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) override; + private: /// To read the data that was flushed into the temporary data file. Processors processors; @@ -212,6 +226,13 @@ private: bool is_consume_started = false; + friend class AggregatingPartialResultTransform; + /// The mutex protects variables that are used for creating a snapshot of the current processor. + /// The current implementation of AggregatingPartialResultTransform uses the 'is_generate_initialized' variable to check + /// whether the processor has started sending data through the main pipeline, and the corresponding partial result processor should stop creating snapshots. + /// Additionally, the mutex protects the 'params->aggregator' and 'many_data->variants' variables, which are used to get data from them for a snapshot. + std::mutex snapshot_mutex; + void initGenerate(); }; diff --git a/src/Processors/Transforms/ExpressionTransform.cpp b/src/Processors/Transforms/ExpressionTransform.cpp index 0d3341b000c..78dace56e4e 100644 --- a/src/Processors/Transforms/ExpressionTransform.cpp +++ b/src/Processors/Transforms/ExpressionTransform.cpp @@ -25,6 +25,12 @@ void ExpressionTransform::transform(Chunk & chunk) chunk.setColumns(block.getColumns(), num_rows); } +ProcessorPtr ExpressionTransform::getPartialResultProcessor(const ProcessorPtr & /*current_processor*/, UInt64 /*partial_result_limit*/, UInt64 /*partial_result_duration_ms*/) +{ + const auto & header = getInputPort().getHeader(); + return std::make_shared(header, expression); +} + ConvertingTransform::ConvertingTransform(const Block & header_, ExpressionActionsPtr expression_) : ExceptionKeepingTransform(header_, ExpressionTransform::transformHeader(header_, expression_->getActionsDAG())) , expression(std::move(expression_)) diff --git a/src/Processors/Transforms/ExpressionTransform.h b/src/Processors/Transforms/ExpressionTransform.h index 791c7d7ba73..8250f25f0f8 100644 --- a/src/Processors/Transforms/ExpressionTransform.h +++ b/src/Processors/Transforms/ExpressionTransform.h @@ -26,10 +26,15 @@ public: static Block transformHeader(Block header, const ActionsDAG & expression); + PartialResultStatus getPartialResultProcessorSupportStatus() const override { return PartialResultStatus::FullSupported; } + protected: void transform(Chunk & chunk) override; + ProcessorPtr getPartialResultProcessor(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) override; + private: + ExpressionActionsPtr expression; }; diff --git a/src/Processors/Transforms/LimitPartialResultTransform.cpp b/src/Processors/Transforms/LimitPartialResultTransform.cpp new file mode 100644 index 00000000000..c9eaa9dc7dd --- /dev/null +++ b/src/Processors/Transforms/LimitPartialResultTransform.cpp @@ -0,0 +1,42 @@ +#include +#include + +namespace DB +{ + +LimitPartialResultTransform::LimitPartialResultTransform( + const Block & header, + UInt64 partial_result_limit_, + UInt64 partial_result_duration_ms_, + UInt64 limit_, + UInt64 offset_) + : PartialResultTransform(header, partial_result_limit_, partial_result_duration_ms_) + , limit(limit_) + , offset(offset_) + {} + +void LimitPartialResultTransform::transformPartialResult(Chunk & chunk) +{ + UInt64 num_rows = chunk.getNumRows(); + if (num_rows < offset || limit == 0) + { + chunk = {}; + return; + } + + UInt64 length = std::min(limit, num_rows - offset); + + /// Check if some rows should be removed + if (length < num_rows) + { + UInt64 num_columns = chunk.getNumColumns(); + auto columns = chunk.detachColumns(); + + for (UInt64 i = 0; i < num_columns; ++i) + columns[i] = columns[i]->cut(offset, length); + + chunk.setColumns(std::move(columns), length); + } +} + +} diff --git a/src/Processors/Transforms/LimitPartialResultTransform.h b/src/Processors/Transforms/LimitPartialResultTransform.h new file mode 100644 index 00000000000..3a0116b624d --- /dev/null +++ b/src/Processors/Transforms/LimitPartialResultTransform.h @@ -0,0 +1,36 @@ +#pragma once + +#include + +namespace DB +{ + +class LimitTransform; + +/// Currently support only single thread implementation with one input and one output ports +class LimitPartialResultTransform : public PartialResultTransform +{ +public: + using LimitTransformPtr = std::shared_ptr; + + LimitPartialResultTransform( + const Block & header, + UInt64 partial_result_limit_, + UInt64 partial_result_duration_ms_, + UInt64 limit_, + UInt64 offset_); + + String getName() const override { return "LimitPartialResultTransform"; } + + void transformPartialResult(Chunk & chunk) override; + /// LimitsTransform doesn't have a state which can be snapshoted + ShaphotResult getRealProcessorSnapshot() override { return {{}, SnaphotStatus::Stopped}; } + +private: + UInt64 limit; + UInt64 offset; + + LimitTransformPtr limit_transform; +}; + +} diff --git a/src/Processors/Transforms/LimitsCheckingTransform.cpp b/src/Processors/Transforms/LimitsCheckingTransform.cpp index 02d2fef808c..0557f3f291e 100644 --- a/src/Processors/Transforms/LimitsCheckingTransform.cpp +++ b/src/Processors/Transforms/LimitsCheckingTransform.cpp @@ -1,4 +1,5 @@ #include +#include #include namespace DB diff --git a/src/Processors/Transforms/LimitsCheckingTransform.h b/src/Processors/Transforms/LimitsCheckingTransform.h index 2f96a17c17b..eabb988dab6 100644 --- a/src/Processors/Transforms/LimitsCheckingTransform.h +++ b/src/Processors/Transforms/LimitsCheckingTransform.h @@ -33,6 +33,8 @@ public: void setQuota(const std::shared_ptr & quota_) { quota = quota_; } + PartialResultStatus getPartialResultProcessorSupportStatus() const override { return PartialResultStatus::SkipSupported; } + protected: void transform(Chunk & chunk) override; diff --git a/src/Processors/Transforms/MergeSortingPartialResultTransform.cpp b/src/Processors/Transforms/MergeSortingPartialResultTransform.cpp new file mode 100644 index 00000000000..e4a2af2cdd8 --- /dev/null +++ b/src/Processors/Transforms/MergeSortingPartialResultTransform.cpp @@ -0,0 +1,48 @@ +#include + +namespace DB +{ + +MergeSortingPartialResultTransform::MergeSortingPartialResultTransform( + const Block & header, MergeSortingTransformPtr merge_sorting_transform_, + UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_) + : PartialResultTransform(header, partial_result_limit_, partial_result_duration_ms_) + , merge_sorting_transform(std::move(merge_sorting_transform_)) + {} + +PartialResultTransform::ShaphotResult MergeSortingPartialResultTransform::getRealProcessorSnapshot() +{ + std::lock_guard lock(merge_sorting_transform->snapshot_mutex); + if (merge_sorting_transform->generated_prefix) + return {{}, SnaphotStatus::Stopped}; + + if (merge_sorting_transform->chunks.empty()) + return {{}, SnaphotStatus::NotReady}; + + /// Sort all input data + merge_sorting_transform->remerge(); + /// Add a copy of the first `partial_result_limit` rows to a generated_chunk + /// to send it later as a partial result in the next prepare stage of the current processor + auto generated_columns = merge_sorting_transform->chunks[0].cloneEmptyColumns(); + size_t total_rows = 0; + for (const auto & merged_chunk : merge_sorting_transform->chunks) + { + size_t rows = std::min(merged_chunk.getNumRows(), partial_result_limit - total_rows); + if (rows == 0) + break; + + for (size_t position = 0; position < generated_columns.size(); ++position) + { + auto column = merged_chunk.getColumns()[position]; + generated_columns[position]->insertRangeFrom(*column, 0, rows); + } + + total_rows += rows; + } + + auto partial_result = Chunk(std::move(generated_columns), total_rows, merge_sorting_transform->chunks[0].getChunkInfo()); + merge_sorting_transform->enrichChunkWithConstants(partial_result); + return {std::move(partial_result), SnaphotStatus::Ready}; +} + +} diff --git a/src/Processors/Transforms/MergeSortingPartialResultTransform.h b/src/Processors/Transforms/MergeSortingPartialResultTransform.h new file mode 100644 index 00000000000..781aa8e1265 --- /dev/null +++ b/src/Processors/Transforms/MergeSortingPartialResultTransform.h @@ -0,0 +1,28 @@ +#pragma once + +#include +#include + +namespace DB +{ + +class MergeSortingPartialResultTransform : public PartialResultTransform +{ +public: + using MergeSortingTransformPtr = std::shared_ptr; + + MergeSortingPartialResultTransform( + const Block & header, MergeSortingTransformPtr merge_sorting_transform_, + UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_); + + String getName() const override { return "MergeSortingPartialResultTransform"; } + + /// MergeSortingTransform always receives chunks in a sorted state, so transformation is not needed + void transformPartialResult(Chunk & /*chunk*/) override {} + ShaphotResult getRealProcessorSnapshot() override; + +private: + MergeSortingTransformPtr merge_sorting_transform; +}; + +} diff --git a/src/Processors/Transforms/MergeSortingTransform.cpp b/src/Processors/Transforms/MergeSortingTransform.cpp index de77711d129..e801e5e16d5 100644 --- a/src/Processors/Transforms/MergeSortingTransform.cpp +++ b/src/Processors/Transforms/MergeSortingTransform.cpp @@ -1,4 +1,5 @@ #include +#include #include #include #include @@ -136,6 +137,8 @@ void MergeSortingTransform::consume(Chunk chunk) /// If there were only const columns in sort description, then there is no need to sort. /// Return the chunk as is. + std::lock_guard lock(snapshot_mutex); + if (description.empty()) { generated_chunk = std::move(chunk); @@ -213,6 +216,8 @@ void MergeSortingTransform::serialize() void MergeSortingTransform::generate() { + std::lock_guard lock(snapshot_mutex); + if (!generated_prefix) { size_t num_tmp_files = tmp_data ? tmp_data->getStreams().size() : 0; @@ -273,4 +278,11 @@ void MergeSortingTransform::remerge() sum_bytes_in_blocks = new_sum_bytes_in_blocks; } +ProcessorPtr MergeSortingTransform::getPartialResultProcessor(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) +{ + const auto & header = inputs.front().getHeader(); + auto merge_sorting_processor = std::dynamic_pointer_cast(current_processor); + return std::make_shared(header, std::move(merge_sorting_processor), partial_result_limit, partial_result_duration_ms); +} + } diff --git a/src/Processors/Transforms/MergeSortingTransform.h b/src/Processors/Transforms/MergeSortingTransform.h index e8c180b6903..67f098b4362 100644 --- a/src/Processors/Transforms/MergeSortingTransform.h +++ b/src/Processors/Transforms/MergeSortingTransform.h @@ -33,6 +33,8 @@ public: String getName() const override { return "MergeSortingTransform"; } + PartialResultStatus getPartialResultProcessorSupportStatus() const override { return PartialResultStatus::FullSupported; } + protected: void consume(Chunk chunk) override; void serialize() override; @@ -40,6 +42,8 @@ protected: Processors expandPipeline() override; + ProcessorPtr getPartialResultProcessor(const ProcessorPtr & current_processor, UInt64 partial_result_limit, UInt64 partial_result_duration_ms) override; + private: size_t max_bytes_before_remerge; double remerge_lowered_memory_bytes_ratio; @@ -59,6 +63,13 @@ private: void remerge(); ProcessorPtr external_merging_sorted; + + friend class MergeSortingPartialResultTransform; + /// The mutex protects variables that are used for creating a snapshot of the current processor. + /// The current implementation of MergeSortingPartialResultTransform uses the 'generated_prefix' variable to check + /// whether the processor has started sending data through the main pipeline, and the corresponding partial result processor should stop creating snapshots. + /// Additionally, the mutex protects the 'chunks' variable and all variables in the 'remerge' function, which is used to transition 'chunks' to a sorted state. + std::mutex snapshot_mutex; }; } diff --git a/src/Processors/Transforms/PartialResultTransform.cpp b/src/Processors/Transforms/PartialResultTransform.cpp new file mode 100644 index 00000000000..97ff79dee54 --- /dev/null +++ b/src/Processors/Transforms/PartialResultTransform.cpp @@ -0,0 +1,80 @@ +#include + +namespace DB +{ + + +PartialResultTransform::PartialResultTransform(const Block & header, UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_) + : PartialResultTransform(header, header, partial_result_limit_, partial_result_duration_ms_) {} + +PartialResultTransform::PartialResultTransform(const Block & input_header, const Block & output_header, UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_) + : IProcessor({input_header}, {output_header}) + , input(inputs.front()) + , output(outputs.front()) + , partial_result_limit(partial_result_limit_) + , partial_result_duration_ms(partial_result_duration_ms_) + , watch(CLOCK_MONOTONIC) + {} + +IProcessor::Status PartialResultTransform::prepare() +{ + if (output.isFinished()) + { + input.close(); + return Status::Finished; + } + + if (finished_getting_snapshots) + { + output.finish(); + return Status::Finished; + } + + if (!output.canPush()) + { + input.setNotNeeded(); + return Status::PortFull; + } + + /// If input data from previous partial result processor is finished then + /// PartialResultTransform ready to create snapshots and send them as a partial result + if (input.isFinished()) + { + if (partial_result.snapshot_status == SnaphotStatus::Ready) + { + partial_result.snapshot_status = SnaphotStatus::NotReady; + output.push(std::move(partial_result.chunk)); + return Status::PortFull; + } + + return Status::Ready; + } + + input.setNeeded(); + if (!input.hasData()) + return Status::NeedData; + + partial_result.chunk = input.pull(); + transformPartialResult(partial_result.chunk); + if (partial_result.chunk.getNumRows() > 0) + { + output.push(std::move(partial_result.chunk)); + return Status::PortFull; + } + + return Status::NeedData; +} + +void PartialResultTransform::work() +{ + if (partial_result_duration_ms < watch.elapsedMilliseconds()) + { + partial_result = getRealProcessorSnapshot(); + if (partial_result.snapshot_status == SnaphotStatus::Stopped) + finished_getting_snapshots = true; + + watch.restart(); + } +} + +} diff --git a/src/Processors/Transforms/PartialResultTransform.h b/src/Processors/Transforms/PartialResultTransform.h new file mode 100644 index 00000000000..4fe87638f38 --- /dev/null +++ b/src/Processors/Transforms/PartialResultTransform.h @@ -0,0 +1,57 @@ +#pragma once + +#include + +namespace DB +{ + +/// Processors of this type are used to construct an auxiliary pipeline with processors corresponding to those in the main pipeline. +/// These processors work in two modes: +/// 1) Creating a snapshot of the corresponding processor from the main pipeline once per partial_result_duration_ms (period in milliseconds), and then sending the snapshot through the partial result pipeline. +/// 2) Transforming small blocks of data in the same way as the original processor and sending the transformed data through the partial result pipeline. +/// All processors of this type rely on the invariant that a new block from the previous processor of the partial result pipeline overwrites information about the previous block of the same previous processor. +class PartialResultTransform : public IProcessor +{ +public: + PartialResultTransform(const Block & header, UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_); + PartialResultTransform(const Block & input_header, const Block & output_header, UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_); + + String getName() const override { return "PartialResultTransform"; } + + Status prepare() override; + void work() override; + + bool isPartialResultProcessor() const override { return true; } + +protected: + enum class SnaphotStatus + { + NotReady, // Waiting for data from the previous partial result processor or awaiting a timer before creating the snapshot. + Ready, // Current partial result processor has received a snapshot from the processor in the main pipeline. + Stopped, // The processor from the main pipeline has started sending data, and the pipeline for partial results should use data from the next processors of the main pipeline. + }; + + struct ShaphotResult + { + Chunk chunk; + SnaphotStatus snapshot_status; + }; + + InputPort & input; + OutputPort & output; + + UInt64 partial_result_limit; + UInt64 partial_result_duration_ms; + + ShaphotResult partial_result = {{}, SnaphotStatus::NotReady}; + + bool finished_getting_snapshots = false; + + virtual void transformPartialResult(Chunk & /*chunk*/) = 0; + virtual ShaphotResult getRealProcessorSnapshot() = 0; // { return {{}, SnaphotStatus::Stopped}; } + +private: + Stopwatch watch; +}; + +} diff --git a/src/QueryPipeline/Pipe.cpp b/src/QueryPipeline/Pipe.cpp index 91ba01c479f..293d152ea65 100644 --- a/src/QueryPipeline/Pipe.cpp +++ b/src/QueryPipeline/Pipe.cpp @@ -12,6 +12,7 @@ #include #include #include +#include namespace DB { @@ -167,12 +168,9 @@ Pipe::Pipe(ProcessorPtr source) { checkSource(*source); - if (collected_processors) - collected_processors->emplace_back(source); - output_ports.push_back(&source->getOutputs().front()); header = output_ports.front()->getHeader(); - processors->emplace_back(std::move(source)); + addProcessor(std::move(source)); max_parallel_streams = 1; } @@ -319,6 +317,16 @@ Pipe Pipe::unitePipes(Pipes pipes, Processors * collected_processors, bool allow res.processors->insert(res.processors->end(), pipe.processors->begin(), pipe.processors->end()); res.output_ports.insert(res.output_ports.end(), pipe.output_ports.begin(), pipe.output_ports.end()); + if (res.isPartialResultActive() && pipe.isPartialResultActive()) + { + res.partial_result_ports.insert( + res.partial_result_ports.end(), + pipe.partial_result_ports.begin(), + pipe.partial_result_ports.end()); + } + else + res.dropPartialResult(); + res.max_parallel_streams += pipe.max_parallel_streams; if (pipe.totals_port) @@ -352,11 +360,11 @@ void Pipe::addSource(ProcessorPtr source) else assertBlocksHaveEqualStructure(header, source_header, "Pipes"); - if (collected_processors) - collected_processors->emplace_back(source); - output_ports.push_back(&source->getOutputs().front()); - processors->emplace_back(std::move(source)); + if (isPartialResultActive()) + partial_result_ports.push_back(nullptr); + + addProcessor(std::move(source)); max_parallel_streams = std::max(max_parallel_streams, output_ports.size()); } @@ -374,11 +382,9 @@ void Pipe::addTotalsSource(ProcessorPtr source) assertBlocksHaveEqualStructure(header, source_header, "Pipes"); - if (collected_processors) - collected_processors->emplace_back(source); - totals_port = &source->getOutputs().front(); - processors->emplace_back(std::move(source)); + + addProcessor(std::move(source)); } void Pipe::addExtremesSource(ProcessorPtr source) @@ -394,11 +400,20 @@ void Pipe::addExtremesSource(ProcessorPtr source) assertBlocksHaveEqualStructure(header, source_header, "Pipes"); - if (collected_processors) - collected_processors->emplace_back(source); - extremes_port = &source->getOutputs().front(); - processors->emplace_back(std::move(source)); + + addProcessor(std::move(source)); +} + +void Pipe::activatePartialResult(UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_) +{ + if (is_partial_result_active) + throw Exception(ErrorCodes::LOGICAL_ERROR, "Partial result for Pipe should be initialized only once"); + + is_partial_result_active = true; + partial_result_limit = partial_result_limit_; + partial_result_duration_ms = partial_result_duration_ms_; + partial_result_ports.assign(output_ports.size(), nullptr); } static void dropPort(OutputPort *& port, Processors & processors, Processors * collected_processors) @@ -426,6 +441,15 @@ void Pipe::dropExtremes() dropPort(extremes_port, *processors, collected_processors); } +void Pipe::dropPartialResult() +{ + for (auto & port : partial_result_ports) + dropPort(port, *processors, collected_processors); + + is_partial_result_active = false; + partial_result_ports.clear(); +} + void Pipe::addTransform(ProcessorPtr transform) { addTransform(std::move(transform), static_cast(nullptr), static_cast(nullptr)); @@ -456,6 +480,8 @@ void Pipe::addTransform(ProcessorPtr transform, OutputPort * totals, OutputPort if (extremes) extremes_port = extremes; + addPartialResultTransform(transform); + size_t next_output = 0; for (auto & input : inputs) { @@ -506,10 +532,7 @@ void Pipe::addTransform(ProcessorPtr transform, OutputPort * totals, OutputPort if (extremes_port) assertBlocksHaveEqualStructure(header, extremes_port->getHeader(), "Pipes"); - if (collected_processors) - collected_processors->emplace_back(transform); - - processors->emplace_back(std::move(transform)); + addProcessor(std::move(transform)); max_parallel_streams = std::max(max_parallel_streams, output_ports.size()); } @@ -546,6 +569,8 @@ void Pipe::addTransform(ProcessorPtr transform, InputPort * totals, InputPort * extremes_port = nullptr; } + addPartialResultTransform(transform); + bool found_totals = false; bool found_extremes = false; @@ -595,14 +620,104 @@ void Pipe::addTransform(ProcessorPtr transform, InputPort * totals, InputPort * if (extremes_port) assertBlocksHaveEqualStructure(header, extremes_port->getHeader(), "Pipes"); - if (collected_processors) - collected_processors->emplace_back(transform); - - processors->emplace_back(std::move(transform)); + addProcessor(std::move(transform)); max_parallel_streams = std::max(max_parallel_streams, output_ports.size()); } +void Pipe::addPartialResultSimpleTransform(const ProcessorPtr & transform, size_t partial_result_port_id) +{ + if (isPartialResultActive()) + { + auto & partial_result_port = partial_result_ports[partial_result_port_id]; + auto partial_result_status = transform->getPartialResultProcessorSupportStatus(); + + if (partial_result_status == IProcessor::PartialResultStatus::NotSupported) + dropPort(partial_result_port, *processors, collected_processors); + + if (partial_result_status != IProcessor::PartialResultStatus::FullSupported) + return; + + auto partial_result_transform = IProcessor::getPartialResultProcessorPtr(transform, partial_result_limit, partial_result_duration_ms); + + connectPartialResultPort(partial_result_port, partial_result_transform->getInputs().front()); + + partial_result_port = &partial_result_transform->getOutputs().front(); + + addProcessor(std::move(partial_result_transform)); + } +} + +void Pipe::addPartialResultTransform(const ProcessorPtr & transform) +{ + if (isPartialResultActive()) + { + size_t new_outputs_size = transform->getOutputs().size(); + auto partial_result_status = transform->getPartialResultProcessorSupportStatus(); + + if (partial_result_status == IProcessor::PartialResultStatus::SkipSupported && new_outputs_size != partial_result_ports.size()) + throw Exception( + ErrorCodes::LOGICAL_ERROR, + "Cannot skip transform {} in the partial result part of the Pipe because it has {} output ports, but the partial result part expects {} output ports", + transform->getName(), + new_outputs_size, + partial_result_ports.size()); + + if (partial_result_status == IProcessor::PartialResultStatus::NotSupported) + { + for (auto & partial_result_port : partial_result_ports) + dropPort(partial_result_port, *processors, collected_processors); + + partial_result_ports.assign(new_outputs_size, nullptr); + } + + if (partial_result_status != IProcessor::PartialResultStatus::FullSupported) + return; + + auto partial_result_transform = IProcessor::getPartialResultProcessorPtr(transform, partial_result_limit, partial_result_duration_ms); + auto & inputs = partial_result_transform->getInputs(); + + if (inputs.size() != partial_result_ports.size()) + throw Exception( + ErrorCodes::LOGICAL_ERROR, + "Cannot add partial result transform {} to Pipe because it has {} input ports, but {} expected", + partial_result_transform->getName(), + inputs.size(), + partial_result_ports.size()); + + size_t next_port = 0; + for (auto & input : inputs) + { + connectPartialResultPort(partial_result_ports[next_port], input); + ++next_port; + } + + partial_result_ports.assign(new_outputs_size, nullptr); + + next_port = 0; + for (auto & new_partial_result_port : partial_result_transform->getOutputs()) + { + partial_result_ports[next_port] = &new_partial_result_port; + ++next_port; + } + + addProcessor(std::move(partial_result_transform)); + } +} + +void Pipe::connectPartialResultPort(OutputPort * partial_result_port, InputPort & partial_result_transform_port) +{ + if (partial_result_port == nullptr) + { + auto source = std::make_shared(getHeader()); + partial_result_port = &source->getPort(); + + addProcessor(std::move(source)); + } + + connect(*partial_result_port, partial_result_transform_port); +} + void Pipe::addSimpleTransform(const ProcessorGetterWithStreamKind & getter) { if (output_ports.empty()) @@ -610,7 +725,7 @@ void Pipe::addSimpleTransform(const ProcessorGetterWithStreamKind & getter) Block new_header; - auto add_transform = [&](OutputPort *& port, StreamType stream_type) + auto add_transform = [&](OutputPort *& port, size_t partial_result_port_id, StreamType stream_type) { if (!port) return; @@ -646,19 +761,22 @@ void Pipe::addSimpleTransform(const ProcessorGetterWithStreamKind & getter) { connect(*port, transform->getInputs().front()); port = &transform->getOutputs().front(); + if (stream_type == StreamType::Main) + addPartialResultSimpleTransform(transform, partial_result_port_id); - if (collected_processors) - collected_processors->emplace_back(transform); - - processors->emplace_back(std::move(transform)); + addProcessor(std::move(transform)); } }; + size_t partial_result_port_id = 0; for (auto & port : output_ports) - add_transform(port, StreamType::Main); + { + add_transform(port, partial_result_port_id, StreamType::Main); + ++partial_result_port_id; + } - add_transform(totals_port, StreamType::Totals); - add_transform(extremes_port, StreamType::Extremes); + add_transform(totals_port, 0, StreamType::Totals); + add_transform(extremes_port, 0, StreamType::Extremes); header = std::move(new_header); } @@ -679,6 +797,7 @@ void Pipe::addChains(std::vector chains) dropTotals(); dropExtremes(); + dropPartialResult(); size_t max_parallel_streams_for_chains = 0; @@ -697,18 +816,21 @@ void Pipe::addChains(std::vector chains) auto added_processors = Chain::getProcessors(std::move(chains[i])); for (auto & transform : added_processors) - { - if (collected_processors) - collected_processors->emplace_back(transform); - - processors->emplace_back(std::move(transform)); - } + addProcessor(std::move(transform)); } header = std::move(new_header); max_parallel_streams = std::max(max_parallel_streams, max_parallel_streams_for_chains); } +void Pipe::addProcessor(ProcessorPtr processor) +{ + if (collected_processors) + collected_processors->emplace_back(processor); + + processors->emplace_back(std::move(processor)); +} + void Pipe::resize(size_t num_streams, bool force, bool strict) { if (output_ports.empty()) @@ -769,6 +891,9 @@ void Pipe::setSinks(const Pipe::ProcessorGetterWithStreamKind & getter) add_transform(totals_port, StreamType::Totals); add_transform(extremes_port, StreamType::Extremes); + for (auto & port : partial_result_ports) + add_transform(port, StreamType::PartialResult); + output_ports.clear(); header.clear(); } @@ -778,6 +903,9 @@ void Pipe::transform(const Transformer & transformer, bool check_ports) if (output_ports.empty()) throw Exception(ErrorCodes::LOGICAL_ERROR, "Cannot transform empty Pipe"); + /// TODO: Add functionality to work with partial result ports in transformer. + dropPartialResult(); + auto new_processors = transformer(output_ports); /// Create hash table with new processors. diff --git a/src/QueryPipeline/Pipe.h b/src/QueryPipeline/Pipe.h index 09931e38578..70e933bcfd2 100644 --- a/src/QueryPipeline/Pipe.h +++ b/src/QueryPipeline/Pipe.h @@ -48,6 +48,9 @@ public: OutputPort * getOutputPort(size_t pos) const { return output_ports[pos]; } OutputPort * getTotalsPort() const { return totals_port; } OutputPort * getExtremesPort() const { return extremes_port; } + OutputPort * getPartialResultPort(size_t pos) const { return partial_result_ports.empty() ? nullptr : partial_result_ports[pos]; } + + bool isPartialResultActive() { return is_partial_result_active; } /// Add processor to list, add it output ports to output_ports. /// Processor shouldn't have input ports, output ports shouldn't be connected. @@ -58,9 +61,13 @@ public: void addTotalsSource(ProcessorPtr source); void addExtremesSource(ProcessorPtr source); - /// Drop totals and extremes (create NullSink for them). + /// Activate sending partial result during main pipeline execution + void activatePartialResult(UInt64 partial_result_limit_, UInt64 partial_result_duration_ms_); + + /// Drop totals, extremes and partial result (create NullSink for them). void dropTotals(); void dropExtremes(); + void dropPartialResult(); /// Add processor to list. It should have size() input ports with compatible header. /// Output ports should have same headers. @@ -69,11 +76,16 @@ public: void addTransform(ProcessorPtr transform, OutputPort * totals, OutputPort * extremes); void addTransform(ProcessorPtr transform, InputPort * totals, InputPort * extremes); + void addPartialResultTransform(const ProcessorPtr & transform); + void addPartialResultSimpleTransform(const ProcessorPtr & transform, size_t partial_result_port_id); + void connectPartialResultPort(OutputPort * partial_result_port, InputPort & partial_result_transform_port); + enum class StreamType { Main = 0, /// Stream for query data. There may be several streams of this type. Totals, /// Stream for totals. No more than one. Extremes, /// Stream for extremes. No more than one. + PartialResult, /// Stream for partial result data. There may be several streams of this type. }; using ProcessorGetter = std::function; @@ -109,10 +121,17 @@ private: Block header; std::shared_ptr processors; - /// Output ports. Totals and extremes are allowed to be empty. + /// If the variable is true, then each time a processor is added pipe will try + /// to add processor which will send partial result from original processor + bool is_partial_result_active = false; + UInt64 partial_result_limit = 0; + UInt64 partial_result_duration_ms = 0; + + /// Output ports. Totals, extremes and partial results are allowed to be empty. OutputPortRawPtrs output_ports; OutputPort * totals_port = nullptr; OutputPort * extremes_port = nullptr; + OutputPortRawPtrs partial_result_ports; /// It is the max number of processors which can be executed in parallel for each step. /// Usually, it's the same as the number of output ports. @@ -128,6 +147,8 @@ private: static Pipe unitePipes(Pipes pipes, Processors * collected_processors, bool allow_empty_header); void setSinks(const Pipe::ProcessorGetterWithStreamKind & getter); + void addProcessor(ProcessorPtr processor); + friend class QueryPipelineBuilder; friend class QueryPipeline; }; diff --git a/src/QueryPipeline/QueryPipeline.cpp b/src/QueryPipeline/QueryPipeline.cpp index 128972b8ff0..9a836f68da1 100644 --- a/src/QueryPipeline/QueryPipeline.cpp +++ b/src/QueryPipeline/QueryPipeline.cpp @@ -66,7 +66,8 @@ static void checkPulling( Processors & processors, OutputPort * output, OutputPort * totals, - OutputPort * extremes) + OutputPort * extremes, + OutputPort * partial_result) { if (!output || output->isConnected()) throw Exception( @@ -83,9 +84,15 @@ static void checkPulling( ErrorCodes::LOGICAL_ERROR, "Cannot create pulling QueryPipeline because its extremes port is connected"); + if (partial_result && partial_result->isConnected()) + throw Exception( + ErrorCodes::LOGICAL_ERROR, + "Cannot create pulling QueryPipeline because its partial_result port is connected"); + bool found_output = false; bool found_totals = false; bool found_extremes = false; + bool found_partial_result = false; for (const auto & processor : processors) { for (const auto & in : processor->getInputs()) @@ -99,6 +106,8 @@ static void checkPulling( found_totals = true; else if (extremes && &out == extremes) found_extremes = true; + else if (partial_result && &out == partial_result) + found_partial_result = true; else checkOutput(out, processor); } @@ -116,6 +125,10 @@ static void checkPulling( throw Exception( ErrorCodes::LOGICAL_ERROR, "Cannot create pulling QueryPipeline because its extremes port does not belong to any processor"); + if (partial_result && !found_partial_result) + throw Exception( + ErrorCodes::LOGICAL_ERROR, + "Cannot create pulling QueryPipeline because its partial result port does not belong to any processor"); } static void checkCompleted(Processors & processors) @@ -318,17 +331,20 @@ QueryPipeline::QueryPipeline( std::shared_ptr processors_, OutputPort * output_, OutputPort * totals_, - OutputPort * extremes_) + OutputPort * extremes_, + OutputPort * partial_result_) : resources(std::move(resources_)) , processors(std::move(processors_)) , output(output_) , totals(totals_) , extremes(extremes_) + , partial_result(partial_result_) { - checkPulling(*processors, output, totals, extremes); + checkPulling(*processors, output, totals, extremes, partial_result); } QueryPipeline::QueryPipeline(Pipe pipe) + : partial_result_duration_ms(pipe.partial_result_duration_ms) { if (pipe.numOutputPorts() > 0) { @@ -336,9 +352,10 @@ QueryPipeline::QueryPipeline(Pipe pipe) output = pipe.getOutputPort(0); totals = pipe.getTotalsPort(); extremes = pipe.getExtremesPort(); + partial_result = pipe.getPartialResultPort(0); processors = std::move(pipe.processors); - checkPulling(*processors, output, totals, extremes); + checkPulling(*processors, output, totals, extremes, partial_result); } else { @@ -370,6 +387,7 @@ QueryPipeline::QueryPipeline(std::shared_ptr format) auto & format_main = format->getPort(IOutputFormat::PortKind::Main); auto & format_totals = format->getPort(IOutputFormat::PortKind::Totals); auto & format_extremes = format->getPort(IOutputFormat::PortKind::Extremes); + auto & format_partial_result = format->getPort(IOutputFormat::PortKind::PartialResult); if (!totals) { @@ -385,12 +403,21 @@ QueryPipeline::QueryPipeline(std::shared_ptr format) processors->emplace_back(std::move(source)); } + if (!partial_result) + { + auto source = std::make_shared(format_partial_result.getHeader()); + partial_result = &source->getPort(); + processors->emplace_back(std::move(source)); + } + connect(*totals, format_totals); connect(*extremes, format_extremes); + connect(*partial_result, format_partial_result); input = &format_main; totals = nullptr; extremes = nullptr; + partial_result = nullptr; output_format = format.get(); @@ -418,6 +445,7 @@ void QueryPipeline::complete(std::shared_ptr sink) drop(totals, *processors); drop(extremes, *processors); + drop(partial_result, *processors); connect(*output, sink->getPort()); processors->emplace_back(std::move(sink)); @@ -433,6 +461,7 @@ void QueryPipeline::complete(Chain chain) drop(totals, *processors); drop(extremes, *processors); + drop(partial_result, *processors); processors->reserve(processors->size() + chain.getProcessors().size() + 1); for (auto processor : chain.getProcessors()) @@ -458,6 +487,7 @@ void QueryPipeline::complete(Pipe pipe) pipe.resize(1); pipe.dropExtremes(); pipe.dropTotals(); + pipe.dropPartialResult(); connect(*pipe.getOutputPort(0), *input); input = nullptr; @@ -486,11 +516,13 @@ void QueryPipeline::complete(std::shared_ptr format) addMaterializing(output, *processors); addMaterializing(totals, *processors); addMaterializing(extremes, *processors); + addMaterializing(partial_result, *processors); } auto & format_main = format->getPort(IOutputFormat::PortKind::Main); auto & format_totals = format->getPort(IOutputFormat::PortKind::Totals); auto & format_extremes = format->getPort(IOutputFormat::PortKind::Extremes); + auto & format_partial_result = format->getPort(IOutputFormat::PortKind::PartialResult); if (!totals) { @@ -506,13 +538,22 @@ void QueryPipeline::complete(std::shared_ptr format) processors->emplace_back(std::move(source)); } + if (!partial_result) + { + auto source = std::make_shared(format_partial_result.getHeader()); + partial_result = &source->getPort(); + processors->emplace_back(std::move(source)); + } + connect(*output, format_main); connect(*totals, format_totals); connect(*extremes, format_extremes); + connect(*partial_result, format_partial_result); output = nullptr; totals = nullptr; extremes = nullptr; + partial_result = nullptr; initRowsBeforeLimit(format.get()); output_format = format.get(); @@ -684,6 +725,7 @@ void QueryPipeline::convertStructureTo(const ColumnsWithTypeAndName & columns) addExpression(output, actions, *processors); addExpression(totals, actions, *processors); addExpression(extremes, actions, *processors); + addExpression(partial_result, actions, *processors); } std::unique_ptr QueryPipeline::getReadProgressCallback() const diff --git a/src/QueryPipeline/QueryPipeline.h b/src/QueryPipeline/QueryPipeline.h index f14cf61aac2..20e58bc0f59 100644 --- a/src/QueryPipeline/QueryPipeline.h +++ b/src/QueryPipeline/QueryPipeline.h @@ -75,7 +75,8 @@ public: std::shared_ptr processors_, OutputPort * output_, OutputPort * totals_ = nullptr, - OutputPort * extremes_ = nullptr); + OutputPort * extremes_ = nullptr, + OutputPort * partial_result_ = nullptr); bool initialized() const { return !processors->empty(); } /// When initialized, exactly one of the following is true. @@ -154,6 +155,7 @@ private: OutputPort * output = nullptr; OutputPort * totals = nullptr; OutputPort * extremes = nullptr; + OutputPort * partial_result = nullptr; QueryStatusPtr process_list_element; @@ -162,6 +164,9 @@ private: size_t num_threads = 0; bool concurrency_control = false; + UInt64 partial_result_limit = 0; + UInt64 partial_result_duration_ms = 0; + friend class PushingPipelineExecutor; friend class PullingPipelineExecutor; friend class PushingAsyncPipelineExecutor; diff --git a/src/QueryPipeline/QueryPipelineBuilder.cpp b/src/QueryPipeline/QueryPipelineBuilder.cpp index f9726339872..90f5ee364f3 100644 --- a/src/QueryPipeline/QueryPipelineBuilder.cpp +++ b/src/QueryPipeline/QueryPipelineBuilder.cpp @@ -110,6 +110,16 @@ void QueryPipelineBuilder::init(QueryPipeline & pipeline) pipe.header = {}; } + if (pipeline.partial_result) + { + /// Set partial result ports only after activation because when activated, it is set to nullptr + pipe.activatePartialResult(pipeline.partial_result_limit, pipeline.partial_result_duration_ms); + pipe.partial_result_ports = {pipeline.partial_result}; + } + + if (!pipeline.partial_result) + pipe.dropPartialResult(); + pipe.totals_port = pipeline.totals; pipe.extremes_port = pipeline.extremes; pipe.max_parallel_streams = pipeline.num_threads; @@ -352,6 +362,10 @@ std::unique_ptr QueryPipelineBuilder::joinPipelinesYShaped left->checkInitializedAndNotCompleted(); right->checkInitializedAndNotCompleted(); + /// TODO: Support joining of partial results from different pipelines. + left->pipe.dropPartialResult(); + right->pipe.dropPartialResult(); + left->pipe.dropExtremes(); right->pipe.dropExtremes(); if (left->getNumStreams() != 1 || right->getNumStreams() != 1) @@ -364,6 +378,7 @@ std::unique_ptr QueryPipelineBuilder::joinPipelinesYShaped auto joining = std::make_shared(join, inputs, out_header, max_block_size); + /// TODO: Support partial results in merge pipelines after joining support above. return mergePipelines(std::move(left), std::move(right), std::move(joining), collected_processors); } @@ -384,6 +399,10 @@ std::unique_ptr QueryPipelineBuilder::joinPipelinesRightLe left->pipe.dropExtremes(); right->pipe.dropExtremes(); + /// TODO: Support joining of partial results from different pipelines. + left->pipe.dropPartialResult(); + right->pipe.dropPartialResult(); + left->pipe.collected_processors = collected_processors; /// Collect the NEW processors for the right pipeline. @@ -634,7 +653,7 @@ PipelineExecutorPtr QueryPipelineBuilder::execute() if (!isCompleted()) throw Exception(ErrorCodes::LOGICAL_ERROR, "Cannot execute pipeline because it is not completed"); - return std::make_shared(pipe.processors, process_list_element); + return std::make_shared(pipe.processors, process_list_element, pipe.partial_result_duration_ms); } Pipe QueryPipelineBuilder::getPipe(QueryPipelineBuilder pipeline, QueryPlanResourceHolder & resources) diff --git a/src/QueryPipeline/QueryPipelineBuilder.h b/src/QueryPipeline/QueryPipelineBuilder.h index 5d273df7068..612e7b1652f 100644 --- a/src/QueryPipeline/QueryPipelineBuilder.h +++ b/src/QueryPipeline/QueryPipelineBuilder.h @@ -85,6 +85,12 @@ public: /// Pipeline will be completed after this transformation. void setSinks(const Pipe::ProcessorGetterWithStreamKind & getter); + /// Activate building separate pipeline for sending partial result. + void activatePartialResult(UInt64 partial_result_limit, UInt64 partial_result_duration_ms) { pipe.activatePartialResult(partial_result_limit, partial_result_duration_ms); } + + /// Check if building of a pipeline for sending partial result active. + bool isPartialResultActive() { return pipe.isPartialResultActive(); } + /// Add totals which returns one chunk with single row with defaults. void addDefaultTotals(); diff --git a/src/Server/TCPHandler.cpp b/src/Server/TCPHandler.cpp index 96c449b7e17..c687a6064b4 100644 --- a/src/Server/TCPHandler.cpp +++ b/src/Server/TCPHandler.cpp @@ -885,7 +885,8 @@ void TCPHandler::processOrdinaryQueryWithProcessors() std::unique_lock progress_lock(task_callback_mutex, std::defer_lock); { - PullingAsyncPipelineExecutor executor(pipeline); + bool has_partial_result_setting = query_context->getSettingsRef().partial_result_update_duration_ms.totalMilliseconds() > 0; + PullingAsyncPipelineExecutor executor(pipeline, has_partial_result_setting); CurrentMetrics::Increment query_thread_metric_increment{CurrentMetrics::QueryThread}; Block block; diff --git a/tests/queries/0_stateless/02010_lc_native.python b/tests/queries/0_stateless/02010_lc_native.python index 6c4220855c8..219fdf04472 100755 --- a/tests/queries/0_stateless/02010_lc_native.python +++ b/tests/queries/0_stateless/02010_lc_native.python @@ -1,227 +1,33 @@ #!/usr/bin/env python3 # -*- coding: utf-8 -*- -import socket import os -import uuid +import sys -CLICKHOUSE_HOST = os.environ.get("CLICKHOUSE_HOST", "127.0.0.1") -CLICKHOUSE_PORT = int(os.environ.get("CLICKHOUSE_PORT_TCP", "900000")) -CLICKHOUSE_DATABASE = os.environ.get("CLICKHOUSE_DATABASE", "default") -CLIENT_NAME = "simple native protocol" +CURDIR = os.path.dirname(os.path.realpath(__file__)) +sys.path.insert(0, os.path.join(CURDIR, "helpers")) - -def writeVarUInt(x, ba): - for _ in range(0, 9): - byte = x & 0x7F - if x > 0x7F: - byte |= 0x80 - - ba.append(byte) - - x >>= 7 - if x == 0: - return - - -def writeStringBinary(s, ba): - b = bytes(s, "utf-8") - writeVarUInt(len(s), ba) - ba.extend(b) - - -def readStrict(s, size=1): - res = bytearray() - while size: - cur = s.recv(size) - # if not res: - # raise "Socket is closed" - size -= len(cur) - res.extend(cur) - - return res - - -def readUInt(s, size=1): - res = readStrict(s, size) - val = 0 - for i in range(len(res)): - val += res[i] << (i * 8) - return val - - -def readUInt8(s): - return readUInt(s) - - -def readUInt16(s): - return readUInt(s, 2) - - -def readUInt32(s): - return readUInt(s, 4) - - -def readUInt64(s): - return readUInt(s, 8) - - -def readVarUInt(s): - x = 0 - for i in range(9): - byte = readStrict(s)[0] - x |= (byte & 0x7F) << (7 * i) - - if not byte & 0x80: - return x - - return x - - -def readStringBinary(s): - size = readVarUInt(s) - s = readStrict(s, size) - return s.decode("utf-8") - - -def sendHello(s): - ba = bytearray() - writeVarUInt(0, ba) # Hello - writeStringBinary(CLIENT_NAME, ba) - writeVarUInt(21, ba) - writeVarUInt(9, ba) - writeVarUInt(54449, ba) - writeStringBinary("default", ba) # database - writeStringBinary("default", ba) # user - writeStringBinary("", ba) # pwd - s.sendall(ba) - - -def receiveHello(s): - p_type = readVarUInt(s) - assert p_type == 0 # Hello - server_name = readStringBinary(s) - # print("Server name: ", server_name) - server_version_major = readVarUInt(s) - # print("Major: ", server_version_major) - server_version_minor = readVarUInt(s) - # print("Minor: ", server_version_minor) - server_revision = readVarUInt(s) - # print("Revision: ", server_revision) - server_timezone = readStringBinary(s) - # print("Timezone: ", server_timezone) - server_display_name = readStringBinary(s) - # print("Display name: ", server_display_name) - server_version_patch = readVarUInt(s) - # print("Version patch: ", server_version_patch) - - -def serializeClientInfo(ba, query_id): - writeStringBinary("default", ba) # initial_user - writeStringBinary(query_id, ba) # initial_query_id - writeStringBinary("127.0.0.1:9000", ba) # initial_address - ba.extend([0] * 8) # initial_query_start_time_microseconds - ba.append(1) # TCP - writeStringBinary("os_user", ba) # os_user - writeStringBinary("client_hostname", ba) # client_hostname - writeStringBinary(CLIENT_NAME, ba) # client_name - writeVarUInt(21, ba) - writeVarUInt(9, ba) - writeVarUInt(54449, ba) - writeStringBinary("", ba) # quota_key - writeVarUInt(0, ba) # distributed_depth - writeVarUInt(1, ba) # client_version_patch - ba.append(0) # No telemetry - - -def sendQuery(s, query): - ba = bytearray() - query_id = uuid.uuid4().hex - writeVarUInt(1, ba) # query - writeStringBinary(query_id, ba) - - ba.append(1) # INITIAL_QUERY - - # client info - serializeClientInfo(ba, query_id) - - writeStringBinary("", ba) # No settings - writeStringBinary("", ba) # No interserver secret - writeVarUInt(2, ba) # Stage - Complete - ba.append(0) # No compression - writeStringBinary(query, ba) # query, finally - s.sendall(ba) - - -def serializeBlockInfo(ba): - writeVarUInt(1, ba) # 1 - ba.append(0) # is_overflows - writeVarUInt(2, ba) # 2 - writeVarUInt(0, ba) # 0 - ba.extend([0] * 4) # bucket_num - - -def sendEmptyBlock(s): - ba = bytearray() - writeVarUInt(2, ba) # Data - writeStringBinary("", ba) - serializeBlockInfo(ba) - writeVarUInt(0, ba) # rows - writeVarUInt(0, ba) # columns - s.sendall(ba) - - -def assertPacket(packet, expected): - assert packet == expected, packet - - -def readHeader(s): - packet_type = readVarUInt(s) - if packet_type == 2: # Exception - raise RuntimeError(readException(s)) - assertPacket(packet_type, 1) # Data - - readStringBinary(s) # external table name - # BlockInfo - assertPacket(readVarUInt(s), 1) # 1 - assertPacket(readUInt8(s), 0) # is_overflows - assertPacket(readVarUInt(s), 2) # 2 - assertPacket(readUInt32(s), 4294967295) # bucket_num - assertPacket(readVarUInt(s), 0) # 0 - columns = readVarUInt(s) # rows - rows = readVarUInt(s) # columns - print("Rows {} Columns {}".format(rows, columns)) - for _ in range(columns): - col_name = readStringBinary(s) - type_name = readStringBinary(s) - print("Column {} type {}".format(col_name, type_name)) - - -def readException(s): - code = readUInt32(s) - name = readStringBinary(s) - text = readStringBinary(s) - readStringBinary(s) # trace - assertPacket(readUInt8(s), 0) # has_nested - return "code {}: {}".format(code, text.replace("DB::Exception:", "")) +from tcp_client import ( + TCPClient, + CLICKHOUSE_DATABASE, + writeVarUInt, + writeStringBinary, + serializeBlockInfo, + assertPacket, +) def insertValidLowCardinalityRow(): - with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: - s.settimeout(30) - s.connect((CLICKHOUSE_HOST, CLICKHOUSE_PORT)) - sendHello(s) - receiveHello(s) - sendQuery( - s, + with TCPClient() as client: + client.sendQuery( "insert into {}.tab settings input_format_defaults_for_omitted_fields=0 format TSV".format( CLICKHOUSE_DATABASE ), ) # external tables - sendEmptyBlock(s) - readHeader(s) + client.sendEmptyBlock() + client.readHeader() # Data ba = bytearray() @@ -240,31 +46,25 @@ def insertValidLowCardinalityRow(): writeStringBinary("hello", ba) # key ba.extend([1] + [0] * 7) # num_indexes ba.extend([0] * 8) # UInt64 index (0 for 'hello') - s.sendall(ba) + client.send(ba) # Fin block - sendEmptyBlock(s) + client.sendEmptyBlock() - assertPacket(readVarUInt(s), 5) # End of stream - s.close() + assertPacket(client.readVarUInt(), 5) # End of stream def insertLowCardinalityRowWithIndexOverflow(): - with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: - s.settimeout(30) - s.connect((CLICKHOUSE_HOST, CLICKHOUSE_PORT)) - sendHello(s) - receiveHello(s) - sendQuery( - s, + with TCPClient() as client: + client.sendQuery( "insert into {}.tab settings input_format_defaults_for_omitted_fields=0 format TSV".format( CLICKHOUSE_DATABASE ), ) # external tables - sendEmptyBlock(s) - readHeader(s) + client.sendEmptyBlock() + client.readHeader() # Data ba = bytearray() @@ -283,29 +83,23 @@ def insertLowCardinalityRowWithIndexOverflow(): writeStringBinary("hello", ba) # key ba.extend([1] + [0] * 7) # num_indexes ba.extend([0] * 7 + [1]) # UInt64 index (overflow) - s.sendall(ba) + client.send(ba) - assertPacket(readVarUInt(s), 2) - print(readException(s)) - s.close() + assertPacket(client.readVarUInt(), 2) # Exception + print(client.readException()) def insertLowCardinalityRowWithIncorrectDictType(): - with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: - s.settimeout(30) - s.connect((CLICKHOUSE_HOST, CLICKHOUSE_PORT)) - sendHello(s) - receiveHello(s) - sendQuery( - s, + with TCPClient() as client: + client.sendQuery( "insert into {}.tab settings input_format_defaults_for_omitted_fields=0 format TSV".format( CLICKHOUSE_DATABASE ), ) # external tables - sendEmptyBlock(s) - readHeader(s) + client.sendEmptyBlock() + client.readHeader() # Data ba = bytearray() @@ -324,29 +118,23 @@ def insertLowCardinalityRowWithIncorrectDictType(): writeStringBinary("hello", ba) # key ba.extend([1] + [0] * 7) # num_indexes ba.extend([0] * 8) # UInt64 index (overflow) - s.sendall(ba) + client.send(ba) - assertPacket(readVarUInt(s), 2) - print(readException(s)) - s.close() + assertPacket(client.readVarUInt(), 2) # Exception + print(client.readException()) def insertLowCardinalityRowWithIncorrectAdditionalKeys(): - with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: - s.settimeout(30) - s.connect((CLICKHOUSE_HOST, CLICKHOUSE_PORT)) - sendHello(s) - receiveHello(s) - sendQuery( - s, + with TCPClient() as client: + client.sendQuery( "insert into {}.tab settings input_format_defaults_for_omitted_fields=0 format TSV".format( CLICKHOUSE_DATABASE ), ) # external tables - sendEmptyBlock(s) - readHeader(s) + client.sendEmptyBlock() + client.readHeader() # Data ba = bytearray() @@ -365,11 +153,10 @@ def insertLowCardinalityRowWithIncorrectAdditionalKeys(): writeStringBinary("hello", ba) # key ba.extend([1] + [0] * 7) # num_indexes ba.extend([0] * 8) # UInt64 index (0 for 'hello') - s.sendall(ba) + client.send(ba) - assertPacket(readVarUInt(s), 2) - print(readException(s)) - s.close() + assertPacket(client.readVarUInt(), 2) # Exception + print(client.readException()) def main(): diff --git a/tests/queries/0_stateless/02210_processors_profile_log.reference b/tests/queries/0_stateless/02210_processors_profile_log.reference index 41543d0706a..f480236111f 100644 --- a/tests/queries/0_stateless/02210_processors_profile_log.reference +++ b/tests/queries/0_stateless/02210_processors_profile_log.reference @@ -38,4 +38,5 @@ LazyOutputFormat 1 1 1 0 0 LimitsCheckingTransform 1 1 1 1 1 NullSource 1 0 0 0 0 NullSource 1 0 0 0 0 +NullSource 0 0 0 0 0 SourceFromSingleChunk 1 0 0 1 1 diff --git a/tests/queries/0_stateless/02458_insert_select_progress_tcp.python b/tests/queries/0_stateless/02458_insert_select_progress_tcp.python index 92240e109c1..fdc64a8dba8 100644 --- a/tests/queries/0_stateless/02458_insert_select_progress_tcp.python +++ b/tests/queries/0_stateless/02458_insert_select_progress_tcp.python @@ -1,188 +1,30 @@ #!/usr/bin/env python3 -import socket -import os -import uuid import json +import os +import sys -CLICKHOUSE_HOST = os.environ.get("CLICKHOUSE_HOST", "127.0.0.1") -CLICKHOUSE_PORT = int(os.environ.get("CLICKHOUSE_PORT_TCP", "900000")) -CLICKHOUSE_DATABASE = os.environ.get("CLICKHOUSE_DATABASE", "default") -CLIENT_NAME = "simple native protocol" +CURDIR = os.path.dirname(os.path.realpath(__file__)) +sys.path.insert(0, os.path.join(CURDIR, "helpers")) - -def writeVarUInt(x, ba): - for _ in range(0, 9): - byte = x & 0x7F - if x > 0x7F: - byte |= 0x80 - - ba.append(byte) - - x >>= 7 - if x == 0: - return - - -def writeStringBinary(s, ba): - b = bytes(s, "utf-8") - writeVarUInt(len(s), ba) - ba.extend(b) - - -def readStrict(s, size=1): - res = bytearray() - while size: - cur = s.recv(size) - # if not res: - # raise "Socket is closed" - size -= len(cur) - res.extend(cur) - - return res - - -def readUInt(s, size=1): - res = readStrict(s, size) - val = 0 - for i in range(len(res)): - val += res[i] << (i * 8) - return val - - -def readUInt8(s): - return readUInt(s) - - -def readUInt16(s): - return readUInt(s, 2) - - -def readUInt32(s): - return readUInt(s, 4) - - -def readUInt64(s): - return readUInt(s, 8) - - -def readVarUInt(s): - x = 0 - for i in range(9): - byte = readStrict(s)[0] - x |= (byte & 0x7F) << (7 * i) - - if not byte & 0x80: - return x - - return x - - -def readStringBinary(s): - size = readVarUInt(s) - s = readStrict(s, size) - return s.decode("utf-8") - - -def sendHello(s): - ba = bytearray() - writeVarUInt(0, ba) # Hello - writeStringBinary(CLIENT_NAME, ba) - writeVarUInt(21, ba) - writeVarUInt(9, ba) - writeVarUInt(54449, ba) - writeStringBinary(CLICKHOUSE_DATABASE, ba) # database - writeStringBinary("default", ba) # user - writeStringBinary("", ba) # pwd - s.sendall(ba) - - -def receiveHello(s): - p_type = readVarUInt(s) - assert p_type == 0 # Hello - server_name = readStringBinary(s) - # print("Server name: ", server_name) - server_version_major = readVarUInt(s) - # print("Major: ", server_version_major) - server_version_minor = readVarUInt(s) - # print("Minor: ", server_version_minor) - server_revision = readVarUInt(s) - # print("Revision: ", server_revision) - server_timezone = readStringBinary(s) - # print("Timezone: ", server_timezone) - server_display_name = readStringBinary(s) - # print("Display name: ", server_display_name) - server_version_patch = readVarUInt(s) - # print("Version patch: ", server_version_patch) - - -def serializeClientInfo(ba, query_id): - writeStringBinary("default", ba) # initial_user - writeStringBinary(query_id, ba) # initial_query_id - writeStringBinary("127.0.0.1:9000", ba) # initial_address - ba.extend([0] * 8) # initial_query_start_time_microseconds - ba.append(1) # TCP - writeStringBinary("os_user", ba) # os_user - writeStringBinary("client_hostname", ba) # client_hostname - writeStringBinary(CLIENT_NAME, ba) # client_name - writeVarUInt(21, ba) - writeVarUInt(9, ba) - writeVarUInt(54449, ba) - writeStringBinary("", ba) # quota_key - writeVarUInt(0, ba) # distributed_depth - writeVarUInt(1, ba) # client_version_patch - ba.append(0) # No telemetry - - -def sendQuery(s, query): - ba = bytearray() - query_id = uuid.uuid4().hex - writeVarUInt(1, ba) # query - writeStringBinary(query_id, ba) - - ba.append(1) # INITIAL_QUERY - - # client info - serializeClientInfo(ba, query_id) - - writeStringBinary("", ba) # No settings - writeStringBinary("", ba) # No interserver secret - writeVarUInt(2, ba) # Stage - Complete - ba.append(0) # No compression - writeStringBinary(query, ba) # query, finally - s.sendall(ba) - - -def serializeBlockInfo(ba): - writeVarUInt(1, ba) # 1 - ba.append(0) # is_overflows - writeVarUInt(2, ba) # 2 - writeVarUInt(0, ba) # 0 - ba.extend([0] * 4) # bucket_num - - -def sendEmptyBlock(s): - ba = bytearray() - writeVarUInt(2, ba) # Data - writeStringBinary("", ba) - serializeBlockInfo(ba) - writeVarUInt(0, ba) # rows - writeVarUInt(0, ba) # columns - s.sendall(ba) - - -def assertPacket(packet, expected): - assert packet == expected, packet +from tcp_client import TCPClient class Progress: - def __init__(self): + def __init__( + self, + read_rows=0, + read_bytes=0, + total_rows_to_read=0, + written_rows=0, + written_bytes=0, + ): # NOTE: this is done in ctor to initialize __dict__ - self.read_rows = 0 - self.read_bytes = 0 - self.total_rows_to_read = 0 - self.written_rows = 0 - self.written_bytes = 0 + self.read_rows = read_rows + self.read_bytes = read_bytes + self.total_rows_to_read = total_rows_to_read + self.written_rows = written_rows + self.written_bytes = written_bytes def __str__(self): return json.dumps(self.__dict__) @@ -195,13 +37,6 @@ class Progress: self.written_bytes += b.written_bytes return self - def readPacket(self, s): - self.read_rows += readVarUInt(s) - self.read_bytes += readVarUInt(s) - self.total_rows_to_read += readVarUInt(s) - self.written_rows += readVarUInt(s) - self.written_bytes += readVarUInt(s) - def __bool__(self): return ( self.read_rows > 0 @@ -212,52 +47,25 @@ class Progress: ) -def readProgress(s): - packet_type = readVarUInt(s) - if packet_type == 2: # Exception - raise RuntimeError(readException(s)) - - if packet_type == 5: # End stream - return None - - assertPacket(packet_type, 3) # Progress - - progress = Progress() - progress.readPacket(s) - return progress - - -def readException(s): - code = readUInt32(s) - name = readStringBinary(s) - text = readStringBinary(s) - readStringBinary(s) # trace - assertPacket(readUInt8(s), 0) # has_nested - return "code {}: {}".format(code, text.replace("DB::Exception:", "")) - - def main(): - with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: - s.settimeout(30) - s.connect((CLICKHOUSE_HOST, CLICKHOUSE_PORT)) - sendHello(s) - receiveHello(s) + with TCPClient() as client: # For 1 second sleep and 1000ms of interactive_delay we definitelly should have non zero progress packet. # NOTE: interactive_delay=0 cannot be used since in this case CompletedPipelineExecutor will not call cancelled callback. - sendQuery( - s, + client.sendQuery( "insert into function null('_ Int') select sleep(1) from numbers(2) settings max_block_size=1, interactive_delay=1000", ) # external tables - sendEmptyBlock(s) + client.sendEmptyBlock() summary_progress = Progress() non_empty_progress_packets = 0 while True: - progress = readProgress(s) - if progress is None: + progress_info = client.readProgress() + if progress_info is None: break + + progress = Progress(*progress_info) summary_progress += progress if progress: non_empty_progress_packets += 1 @@ -268,8 +76,6 @@ def main(): # - 1 or 2 for each SELECT block assert non_empty_progress_packets in (3, 4), f"{non_empty_progress_packets=:}" - s.close() - if __name__ == "__main__": main() diff --git a/tests/queries/0_stateless/02750_settings_alias_tcp_protocol.python b/tests/queries/0_stateless/02750_settings_alias_tcp_protocol.python index 48b27d434ec..1736807410f 100644 --- a/tests/queries/0_stateless/02750_settings_alias_tcp_protocol.python +++ b/tests/queries/0_stateless/02750_settings_alias_tcp_protocol.python @@ -1,217 +1,23 @@ #!/usr/bin/env python3 -import socket + import os -import uuid -import json +import sys -CLICKHOUSE_HOST = os.environ.get("CLICKHOUSE_HOST", "127.0.0.1") -CLICKHOUSE_PORT = int(os.environ.get("CLICKHOUSE_PORT_TCP", "900000")) -CLICKHOUSE_DATABASE = os.environ.get("CLICKHOUSE_DATABASE", "default") -CLIENT_NAME = "simple native protocol" +CURDIR = os.path.dirname(os.path.realpath(__file__)) +sys.path.insert(0, os.path.join(CURDIR, "helpers")) - -def writeVarUInt(x, ba): - for _ in range(0, 9): - byte = x & 0x7F - if x > 0x7F: - byte |= 0x80 - - ba.append(byte) - - x >>= 7 - if x == 0: - return - - -def writeStringBinary(s, ba): - b = bytes(s, "utf-8") - writeVarUInt(len(s), ba) - ba.extend(b) - - -def readStrict(s, size=1): - res = bytearray() - while size: - cur = s.recv(size) - # if not res: - # raise "Socket is closed" - size -= len(cur) - res.extend(cur) - - return res - - -def readUInt(s, size=1): - res = readStrict(s, size) - val = 0 - for i in range(len(res)): - val += res[i] << (i * 8) - return val - - -def readUInt8(s): - return readUInt(s) - - -def readUInt16(s): - return readUInt(s, 2) - - -def readUInt32(s): - return readUInt(s, 4) - - -def readUInt64(s): - return readUInt(s, 8) - - -def readVarUInt(s): - x = 0 - for i in range(9): - byte = readStrict(s)[0] - x |= (byte & 0x7F) << (7 * i) - - if not byte & 0x80: - return x - - return x - - -def readStringBinary(s): - size = readVarUInt(s) - s = readStrict(s, size) - return s.decode("utf-8") - - -def sendHello(s): - ba = bytearray() - writeVarUInt(0, ba) # Hello - writeStringBinary(CLIENT_NAME, ba) - writeVarUInt(21, ba) - writeVarUInt(9, ba) - writeVarUInt(54449, ba) - writeStringBinary(CLICKHOUSE_DATABASE, ba) # database - writeStringBinary("default", ba) # user - writeStringBinary("", ba) # pwd - s.sendall(ba) - - -def receiveHello(s): - p_type = readVarUInt(s) - assert p_type == 0 # Hello - _server_name = readStringBinary(s) - _server_version_major = readVarUInt(s) - _server_version_minor = readVarUInt(s) - _server_revision = readVarUInt(s) - _server_timezone = readStringBinary(s) - _server_display_name = readStringBinary(s) - _server_version_patch = readVarUInt(s) - - -def serializeClientInfo(ba, query_id): - writeStringBinary("default", ba) # initial_user - writeStringBinary(query_id, ba) # initial_query_id - writeStringBinary("127.0.0.1:9000", ba) # initial_address - ba.extend([0] * 8) # initial_query_start_time_microseconds - ba.append(1) # TCP - writeStringBinary("os_user", ba) # os_user - writeStringBinary("client_hostname", ba) # client_hostname - writeStringBinary(CLIENT_NAME, ba) # client_name - writeVarUInt(21, ba) - writeVarUInt(9, ba) - writeVarUInt(54449, ba) - writeStringBinary("", ba) # quota_key - writeVarUInt(0, ba) # distributed_depth - writeVarUInt(1, ba) # client_version_patch - ba.append(0) # No telemetry - - -def sendQuery(s, query, settings): - ba = bytearray() - query_id = uuid.uuid4().hex - writeVarUInt(1, ba) # query - writeStringBinary(query_id, ba) - - ba.append(1) # INITIAL_QUERY - - # client info - serializeClientInfo(ba, query_id) - - # Settings - for key, value in settings.items(): - writeStringBinary(key, ba) - writeVarUInt(1, ba) # is_important - writeStringBinary(str(value), ba) - writeStringBinary("", ba) # End of settings - - writeStringBinary("", ba) # No interserver secret - writeVarUInt(2, ba) # Stage - Complete - ba.append(0) # No compression - writeStringBinary(query, ba) # query, finally - s.sendall(ba) - - -def serializeBlockInfo(ba): - writeVarUInt(1, ba) # 1 - ba.append(0) # is_overflows - writeVarUInt(2, ba) # 2 - writeVarUInt(0, ba) # 0 - ba.extend([0] * 4) # bucket_num - - -def sendEmptyBlock(s): - ba = bytearray() - writeVarUInt(2, ba) # Data - writeStringBinary("", ba) - serializeBlockInfo(ba) - writeVarUInt(0, ba) # rows - writeVarUInt(0, ba) # columns - s.sendall(ba) - - -def assertPacket(packet, expected): - assert packet == expected, "Got: {}, expected: {}".format(packet, expected) - - -def readResponse(s): - packet_type = readVarUInt(s) - if packet_type == 2: # Exception - raise RuntimeError(readException(s)) - - if packet_type == 1: # Data - return None - if packet_type == 3: # Progress - return None - if packet_type == 5: # End stream - return None - - raise RuntimeError("Unexpected packet: {}".format(packet_type)) - - -def readException(s): - code = readUInt32(s) - _name = readStringBinary(s) - text = readStringBinary(s) - readStringBinary(s) # trace - assertPacket(readUInt8(s), 0) # has_nested - return "code {}: {}".format(code, text.replace("DB::Exception:", "")) +from tcp_client import TCPClient def main(): - with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: - s.settimeout(30) - s.connect((CLICKHOUSE_HOST, CLICKHOUSE_PORT)) - sendHello(s) - receiveHello(s) - sendQuery(s, "select 1", {"replication_alter_partitions_sync": 1}) + with TCPClient() as client: + client.sendQuery("select 1", {"replication_alter_partitions_sync": 1}) # external tables - sendEmptyBlock(s) + client.sendEmptyBlock() - while readResponse(s) is not None: + while client.readResponse() is not None: pass - - s.close() print("OK") diff --git a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python new file mode 100755 index 00000000000..b3d04d263dc --- /dev/null +++ b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python @@ -0,0 +1,93 @@ +#!/usr/bin/env python3 +# -*- coding: utf-8 -*- + + +import os +import sys + +CURDIR = os.path.dirname(os.path.realpath(__file__)) +sys.path.insert(0, os.path.join(CURDIR, "helpers")) + +from tcp_client import TCPClient + + +def run_query_without_errors(query, support_partial_result): + with TCPClient() as client: + client.sendQuery(query) + + # external tables + client.sendEmptyBlock() + client.readHeader() + + # Partial result + partial_result = client.readDataWithoutProgress()[0] + if support_partial_result: + assert ( + len(partial_result.value) > 0 + ), "Expected at least one block with a non-empty partial result before getting the full result" + + while True: + assert all( + a >= b + for a, b in zip(partial_result.value, partial_result.value[1:]) + ), "Partial result always should be sorted for this test" + + new_partial_result = client.readDataWithoutProgress( + need_print_info=False + )[0] + if len(new_partial_result.value) == 0: + break + + data_size = len(partial_result.value) + assert all( + partial_result.value[i] <= new_partial_result.value[i] + for i in range(data_size) + ), f"New partial result values should always be greater then old one because a new block contains more information about the full data. New result {new_partial_result}. Previous result {partial_result}" + + partial_result = new_partial_result + else: + block_rows = len(partial_result.value) + assert ( + block_rows == 0 + ), f"Expected only empty partial result block before getting the full result, but block has {block_rows} rows" + + # Full result + full_result = client.readDataWithoutProgress()[0] + + data_size = len(partial_result.value) + assert all( + partial_result.value[i] <= full_result.value[i] for i in range(data_size) + ), f"Full result values should always be greater then partial result values. Full result {full_result}. Partial result {partial_result}" + + for result in full_result.value: + print(result) + + +def main(): + # Request with partial result limit less then full limit + run_query_without_errors( + "SELECT number FROM numbers_mt(5e6+1) ORDER BY -number LIMIT 5 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 3", + support_partial_result=True, + ) + + # Request with partial result limit greater then full limit + run_query_without_errors( + "SELECT number FROM numbers_mt(5e6+1) ORDER BY -number LIMIT 3 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 5", + support_partial_result=True, + ) + + # Request with OFFSET + run_query_without_errors( + "SELECT number FROM numbers_mt(5e6+1) ORDER BY -number LIMIT 3 OFFSET 1 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 5", + support_partial_result=True, + ) + + # Request with OFFSET greater then partial result limit (partial result pipeline use blocks with less then OFFSET, so there will be no elements in block after LimitPartialResultTransform) + run_query_without_errors( + "SELECT number FROM numbers_mt(5e6+1) ORDER BY -number LIMIT 3 OFFSET 15 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 5", + support_partial_result=False, + ) + + +if __name__ == "__main__": + main() diff --git a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.reference b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.reference new file mode 100644 index 00000000000..211a193940a --- /dev/null +++ b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.reference @@ -0,0 +1,38 @@ +Rows 0 Columns 1 +Column number type UInt64 +Rows 3 Columns 1 +Column number type UInt64 +Rows 5 Columns 1 +Column number type UInt64 +5000000 +4999999 +4999998 +4999997 +4999996 +Rows 0 Columns 1 +Column number type UInt64 +Rows 3 Columns 1 +Column number type UInt64 +Rows 3 Columns 1 +Column number type UInt64 +5000000 +4999999 +4999998 +Rows 0 Columns 1 +Column number type UInt64 +Rows 3 Columns 1 +Column number type UInt64 +Rows 3 Columns 1 +Column number type UInt64 +4999999 +4999998 +4999997 +Rows 0 Columns 1 +Column number type UInt64 +Rows 0 Columns 1 +Column number type UInt64 +Rows 3 Columns 1 +Column number type UInt64 +4999985 +4999984 +4999983 diff --git a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.sh b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.sh new file mode 100755 index 00000000000..1ed15197dbf --- /dev/null +++ b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.sh @@ -0,0 +1,8 @@ +#!/usr/bin/env bash + +CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) +# shellcheck source=../shell_config.sh +. "$CURDIR"/../shell_config.sh + +# We should have correct env vars from shell_config.sh to run this test +python3 "$CURDIR"/02833_partial_sorting_result_during_query_execution.python diff --git a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python new file mode 100644 index 00000000000..4d869b05580 --- /dev/null +++ b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python @@ -0,0 +1,125 @@ +#!/usr/bin/env python3 +# -*- coding: utf-8 -*- + + +import os +import sys + +CURDIR = os.path.dirname(os.path.realpath(__file__)) +sys.path.insert(0, os.path.join(CURDIR, "helpers")) + +from tcp_client import TCPClient + + +def get_keys(results): + return [key for key, _ in results] + + +def check_new_result(new_results, old_results, invariants, rows_limit): + if rows_limit is not None: + assert ( + len(new_results[0].value) <= rows_limit + ), f"Result should have no more then {rows_limit} rows. But it has {len(new_results[0].value)} rows" + + for new_result, old_result in zip(new_results, old_results): + assert ( + new_result.key == old_result.key + ), f"Keys in blocks should be in the same order. Full results keys {get_keys(full_results)}. Partial results keys {get_keys(partial_results)}" + + key = new_result.key + if key in invariants: + new_value = new_result.value + old_value = old_result.value + assert invariants[key]( + old_value, new_value + ), f"Problem with the invariant between new and old result for key: {key}. New value {new_value}. Old value {old_value}" + + +def run_query_without_errors( + query, support_partial_result, invariants=None, rows_limit=None +): + if invariants is None: + invariants = {} + + with TCPClient() as client: + client.sendQuery(query) + + # external tables + client.sendEmptyBlock() + client.readHeader() + + # Partial result + partial_results = client.readDataWithoutProgress() + if support_partial_result: + assert ( + len(partial_results) > 0 and len(partial_results[0].value) > 0 + ), "Expected at least one block with a non-empty partial result before getting the full result" + while True: + new_partial_results = client.readDataWithoutProgress( + need_print_info=False + ) + if len(new_partial_results[0].value) == 0: + break + + check_new_result( + new_partial_results, partial_results, invariants, rows_limit + ) + partial_results = new_partial_results + else: + block_rows = len(partial_results[0].value) + assert ( + block_rows == 0 + ), f"Expected only empty partial result block before getting the full result, but block has {block_rows} rows" + + # Full result + full_results = client.readDataWithoutProgress() + if support_partial_result: + check_new_result(full_results, partial_results, invariants, rows_limit) + + for data in full_results: + if isinstance(data.value[0], int): + print(data.key, data.value) + + +def supported_scenarios_without_key(): + # Simple aggregation query + query = "select median(number), stddevSamp(number), stddevPop(number), max(number), min(number), any(number), count(number), avg(number), sum(number) from numbers_mt(1e7+1) settings max_threads = 1, partial_result_update_duration_ms = 1" + invariants = { + "median(number)": lambda old_value, new_value: old_value <= new_value, + "max(number)": lambda old_value, new_value: old_value <= new_value, + "min(number)": lambda old_value, new_value: old_value >= new_value, + "count(number)": lambda old_value, new_value: old_value <= new_value, + "avg(number)": lambda old_value, new_value: old_value <= new_value, + "sum(number)": lambda old_value, new_value: old_value <= new_value, + } + run_query_without_errors( + query, support_partial_result=True, invariants=invariants, rows_limit=1 + ) + + # Aggregation query with a nested ORDER BY subquery + query = "select median(number), stddevSamp(number), stddevPop(number), max(number), min(number), any(number), count(number), avg(number), sum(number) FROM (SELECT number FROM numbers_mt(1e7) ORDER BY -number LIMIT 3) settings max_threads = 1, partial_result_update_duration_ms=1" + + # Aggregation receives small partial result blocks from ORDER BY which always sends blocks with bigger values + invariants["min(number)"] = lambda old_value, new_value: old_value <= new_value + run_query_without_errors( + query, support_partial_result=True, invariants=invariants, rows_limit=1 + ) + + +def unsupported_scenarios(): + # Currently aggregator for partial result supports only single thread aggregation without key + # Update test when multithreading or aggregation with GROUP BY will be supported for partial result updates + multithread_query = "select sum(number) from numbers_mt(1e7+1) settings max_threads = 2, partial_result_update_duration_ms = 100" + run_query_without_errors(multithread_query, support_partial_result=False) + + group_with_key_query = "select mod2, sum(number) from numbers_mt(1e7+1) group by number % 2 as mod2 settings max_threads = 1, partial_result_update_duration_ms = 100" + run_query_without_errors(group_with_key_query, support_partial_result=False) + + +def main(): + supported_scenarios_without_key() + unsupported_scenarios() + + +if __name__ == "__main__": + main() diff --git a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.reference b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.reference new file mode 100644 index 00000000000..a813b18f24f --- /dev/null +++ b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.reference @@ -0,0 +1,88 @@ +Rows 0 Columns 9 +Column median(number) type Float64 +Column stddevSamp(number) type Float64 +Column stddevPop(number) type Float64 +Column max(number) type UInt64 +Column min(number) type UInt64 +Column any(number) type UInt64 +Column count(number) type UInt64 +Column avg(number) type Float64 +Column sum(number) type UInt64 +Rows 1 Columns 9 +Column median(number) type Float64 +Column stddevSamp(number) type Float64 +Column stddevPop(number) type Float64 +Column max(number) type UInt64 +Column min(number) type UInt64 +Column any(number) type UInt64 +Column count(number) type UInt64 +Column avg(number) type Float64 +Column sum(number) type UInt64 +Rows 1 Columns 9 +Column median(number) type Float64 +Column stddevSamp(number) type Float64 +Column stddevPop(number) type Float64 +Column max(number) type UInt64 +Column min(number) type UInt64 +Column any(number) type UInt64 +Column count(number) type UInt64 +Column avg(number) type Float64 +Column sum(number) type UInt64 +max(number) [10000000] +min(number) [0] +any(number) [0] +count(number) [10000001] +sum(number) [50000005000000] +Rows 0 Columns 9 +Column median(number) type Float64 +Column stddevSamp(number) type Float64 +Column stddevPop(number) type Float64 +Column max(number) type UInt64 +Column min(number) type UInt64 +Column any(number) type UInt64 +Column count(number) type UInt64 +Column avg(number) type Float64 +Column sum(number) type UInt64 +Rows 1 Columns 9 +Column median(number) type Float64 +Column stddevSamp(number) type Float64 +Column stddevPop(number) type Float64 +Column max(number) type UInt64 +Column min(number) type UInt64 +Column any(number) type UInt64 +Column count(number) type UInt64 +Column avg(number) type Float64 +Column sum(number) type UInt64 +Rows 1 Columns 9 +Column median(number) type Float64 +Column stddevSamp(number) type Float64 +Column stddevPop(number) type Float64 +Column max(number) type UInt64 +Column min(number) type UInt64 +Column any(number) type UInt64 +Column count(number) type UInt64 +Column avg(number) type Float64 +Column sum(number) type UInt64 +max(number) [9999999] +min(number) [9999997] +any(number) [9999999] +count(number) [3] +sum(number) [29999994] +Rows 0 Columns 1 +Column sum(number) type UInt64 +Rows 0 Columns 1 +Column sum(number) type UInt64 +Rows 1 Columns 1 +Column sum(number) type UInt64 +sum(number) [50000005000000] +Rows 0 Columns 2 +Column mod2 type UInt8 +Column sum(number) type UInt64 +Rows 0 Columns 2 +Column mod2 type UInt8 +Column sum(number) type UInt64 +Rows 2 Columns 2 +Column mod2 type UInt8 +Column sum(number) type UInt64 +mod2 [0, 1] +sum(number) [25000005000000, 25000000000000] diff --git a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.sh b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.sh new file mode 100755 index 00000000000..e70a3c53ec4 --- /dev/null +++ b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.sh @@ -0,0 +1,8 @@ +#!/usr/bin/env bash + +CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) +# shellcheck source=../shell_config.sh +. "$CURDIR"/../shell_config.sh + +# We should have correct env vars from shell_config.sh to run this test +python3 "$CURDIR"/02834_partial_aggregating_result_during_query_execution.python diff --git a/tests/queries/0_stateless/helpers/tcp_client.py b/tests/queries/0_stateless/helpers/tcp_client.py new file mode 100644 index 00000000000..fdc4ab28e04 --- /dev/null +++ b/tests/queries/0_stateless/helpers/tcp_client.py @@ -0,0 +1,313 @@ +import socket +import os +import uuid +import struct + +CLICKHOUSE_HOST = os.environ.get("CLICKHOUSE_HOST", "127.0.0.1") +CLICKHOUSE_PORT = int(os.environ.get("CLICKHOUSE_PORT_TCP", "900000")) +CLICKHOUSE_DATABASE = os.environ.get("CLICKHOUSE_DATABASE", "default") +CLIENT_NAME = "simple native protocol" + + +def writeVarUInt(x, ba): + for _ in range(0, 9): + byte = x & 0x7F + if x > 0x7F: + byte |= 0x80 + + ba.append(byte) + + x >>= 7 + if x == 0: + return + + +def writeStringBinary(s, ba): + b = bytes(s, "utf-8") + writeVarUInt(len(s), ba) + ba.extend(b) + + +def serializeClientInfo(ba, query_id): + writeStringBinary("default", ba) # initial_user + writeStringBinary(query_id, ba) # initial_query_id + writeStringBinary("127.0.0.1:9000", ba) # initial_address + ba.extend([0] * 8) # initial_query_start_time_microseconds + ba.append(1) # TCP + writeStringBinary("os_user", ba) # os_user + writeStringBinary("client_hostname", ba) # client_hostname + writeStringBinary(CLIENT_NAME, ba) # client_name + writeVarUInt(21, ba) + writeVarUInt(9, ba) + writeVarUInt(54449, ba) + writeStringBinary("", ba) # quota_key + writeVarUInt(0, ba) # distributed_depth + writeVarUInt(1, ba) # client_version_patch + ba.append(0) # No telemetry + + +def serializeBlockInfo(ba): + writeVarUInt(1, ba) # 1 + ba.append(0) # is_overflows + writeVarUInt(2, ba) # 2 + writeVarUInt(0, ba) # 0 + ba.extend([0] * 4) # bucket_num + + +def assertPacket(packet, expected): + assert packet == expected, "Got: {}, expected: {}".format(packet, expected) + + +class Data(object): + def __init__(self, key, value): + self.key = key + self.value = value + + +class TCPClient(object): + def __init__(self, timeout=30): + self.timeout = timeout + self.socket = None + + def __enter__(self): + self.socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM) + self.socket.settimeout(self.timeout) + self.socket.connect((CLICKHOUSE_HOST, CLICKHOUSE_PORT)) + + self.sendHello() + self.receiveHello() + + return self + + def __exit__(self, exc_type, exc_value, traceback): + if self.socket: + self.socket.close() + + def readStrict(self, size=1): + res = bytearray() + while size: + cur = self.socket.recv(size) + # if not res: + # raise "Socket is closed" + size -= len(cur) + res.extend(cur) + + return res + + def readUInt(self, size=1): + res = self.readStrict(size) + val = 0 + for i in range(len(res)): + val += res[i] << (i * 8) + return val + + def readUInt8(self): + return self.readUInt() + + def readUInt16(self): + return self.readUInt(2) + + def readUInt32(self): + return self.readUInt(4) + + def readUInt64(self): + return self.readUInt(8) + + def readFloat16(self): + return struct.unpack("e", self.readStrict(2)) + + def readFloat32(self): + return struct.unpack("f", self.readStrict(4)) + + def readFloat64(self): + return struct.unpack("d", self.readStrict(8)) + + def readVarUInt(self): + x = 0 + for i in range(9): + byte = self.readStrict()[0] + x |= (byte & 0x7F) << (7 * i) + + if not byte & 0x80: + return x + + return x + + def readStringBinary(self): + size = self.readVarUInt() + s = self.readStrict(size) + return s.decode("utf-8") + + def send(self, byte_array): + self.socket.sendall(byte_array) + + def sendHello(self): + ba = bytearray() + writeVarUInt(0, ba) # Hello + writeStringBinary(CLIENT_NAME, ba) + writeVarUInt(21, ba) + writeVarUInt(9, ba) + writeVarUInt(54449, ba) + writeStringBinary(CLICKHOUSE_DATABASE, ba) # database + writeStringBinary("default", ba) # user + writeStringBinary("", ba) # pwd + self.send(ba) + + def receiveHello(self): + p_type = self.readVarUInt() + assert p_type == 0 # Hello + _server_name = self.readStringBinary() + _server_version_major = self.readVarUInt() + _server_version_minor = self.readVarUInt() + _server_revision = self.readVarUInt() + _server_timezone = self.readStringBinary() + _server_display_name = self.readStringBinary() + _server_version_patch = self.readVarUInt() + + def sendQuery(self, query, settings=None): + if settings == None: + settings = {} # No settings + + ba = bytearray() + query_id = uuid.uuid4().hex + writeVarUInt(1, ba) # query + writeStringBinary(query_id, ba) + + ba.append(1) # INITIAL_QUERY + + # client info + serializeClientInfo(ba, query_id) + + # Settings + for key, value in settings.items(): + writeStringBinary(key, ba) + writeVarUInt(1, ba) # is_important + writeStringBinary(str(value), ba) + writeStringBinary("", ba) # End of settings + + writeStringBinary("", ba) # No interserver secret + writeVarUInt(2, ba) # Stage - Complete + ba.append(0) # No compression + writeStringBinary(query, ba) # query, finally + self.send(ba) + + def sendEmptyBlock(self): + ba = bytearray() + writeVarUInt(2, ba) # Data + writeStringBinary("", ba) + serializeBlockInfo(ba) + writeVarUInt(0, ba) # rows + writeVarUInt(0, ba) # columns + self.send(ba) + + def readException(self): + code = self.readUInt32() + _name = self.readStringBinary() + text = self.readStringBinary() + self.readStringBinary() # trace + assertPacket(self.readUInt8(), 0) # has_nested + return "code {}: {}".format(code, text.replace("DB::Exception:", "")) + + def readPacketType(self): + packet_type = self.readVarUInt() + if packet_type == 2: # Exception + raise RuntimeError(self.readException()) + + return packet_type + + def readResponse(self): + packet_type = self.readPacketType() + if packet_type == 1: # Data + return None + if packet_type == 3: # Progress + return None + if packet_type == 5: # End stream + return None + + raise RuntimeError("Unexpected packet: {}".format(packet_type)) + + def readProgressData(self): + read_rows = self.readVarUInt() + read_bytes = self.readVarUInt() + total_rows_to_read = self.readVarUInt() + written_rows = self.readVarUInt() + written_bytes = self.readVarUInt() + + return read_rows, read_bytes, total_rows_to_read, written_rows, written_bytes + + def readProgress(self): + packet_type = self.readPacketType() + if packet_type == 5: # End stream + return None + assertPacket(packet_type, 3) # Progress + return self.readProgressData() + + def readHeaderInfo(self): + self.readStringBinary() # external table name + # BlockInfo + assertPacket(self.readVarUInt(), 1) # field number 1 + assertPacket(self.readUInt8(), 0) # is_overflows + assertPacket(self.readVarUInt(), 2) # field number 2 + assertPacket(self.readUInt32(), 4294967295) # bucket_num + assertPacket(self.readVarUInt(), 0) # 0 + columns = self.readVarUInt() # rows + rows = self.readVarUInt() # columns + + return columns, rows + + def readHeader(self): + packet_type = self.readPacketType() + assertPacket(packet_type, 1) # Data + + columns, rows = self.readHeaderInfo() + print("Rows {} Columns {}".format(rows, columns)) + for _ in range(columns): + col_name = self.readStringBinary() + type_name = self.readStringBinary() + print("Column {} type {}".format(col_name, type_name)) + + def readRow(self, row_type, rows): + supported_row_types = { + "UInt8": self.readUInt8, + "UInt16": self.readUInt16, + "UInt32": self.readUInt32, + "UInt64": self.readUInt64, + "Float16": self.readFloat16, + "Float32": self.readFloat32, + "Float64": self.readFloat64, + } + if row_type in supported_row_types: + read_type = supported_row_types[row_type] + row = [read_type() for _ in range(rows)] + return row + else: + raise RuntimeError( + "Current python version of tcp client doesn't support the following type of row: {}".format( + row_type + ) + ) + + def readDataWithoutProgress(self, need_print_info=True): + packet_type = self.readPacketType() + while packet_type == 3: # Progress + self.readProgressData() + packet_type = self.readPacketType() + + if packet_type == 5: # End stream + return None + assertPacket(packet_type, 1) # Data + + columns, rows = self.readHeaderInfo() + data = [] + if need_print_info: + print("Rows {} Columns {}".format(rows, columns)) + + for _ in range(columns): + col_name = self.readStringBinary() + type_name = self.readStringBinary() + if need_print_info: + print("Column {} type {}".format(col_name, type_name)) + + data.append(Data(col_name, self.readRow(type_name, rows))) + + return data From 7f778692164d8a9169d83a7b30c712e00bb1bbeb Mon Sep 17 00:00:00 2001 From: alexX512 Date: Sat, 9 Sep 2023 10:11:10 +0000 Subject: [PATCH 251/433] Increase amount of rows in requests for partial result tests (cherry picked from commit 54bc3fea9a7240bda99949aefd3fe8b0fa2d15ec) --- src/Processors/LimitTransform.h | 8 +++++- ...rting_result_during_query_execution.python | 10 ++++--- ...ng_result_during_query_execution.reference | 28 +++++++++---------- ...ating_result_during_query_execution.python | 12 +++++--- ...ng_result_during_query_execution.reference | 18 ++++++------ 5 files changed, 44 insertions(+), 32 deletions(-) diff --git a/src/Processors/LimitTransform.h b/src/Processors/LimitTransform.h index eac5f9e8d6d..cfacc9634f9 100644 --- a/src/Processors/LimitTransform.h +++ b/src/Processors/LimitTransform.h @@ -76,7 +76,13 @@ public: void setRowsBeforeLimitCounter(RowsBeforeLimitCounterPtr counter) override { rows_before_limit_at_least.swap(counter); } void setInputPortHasCounter(size_t pos) { ports_data[pos].input_port_has_counter = true; } - PartialResultStatus getPartialResultProcessorSupportStatus() const override { return PartialResultStatus::FullSupported; } + PartialResultStatus getPartialResultProcessorSupportStatus() const override + { + /// Currently LimitPartialResultTransform support only single-thread work. + bool is_partial_result_supported = inputs.size() == 1 && outputs.size() == 1; + + return is_partial_result_supported ? PartialResultStatus::FullSupported : PartialResultStatus::NotSupported; + } }; } diff --git a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python index b3d04d263dc..39d81438c1b 100755 --- a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python +++ b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.python @@ -64,27 +64,29 @@ def run_query_without_errors(query, support_partial_result): def main(): + rows_number = 2e7 + 1 + # Request with partial result limit less then full limit run_query_without_errors( - "SELECT number FROM numbers_mt(5e6+1) ORDER BY -number LIMIT 5 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 3", + f"SELECT number FROM numbers_mt({rows_number}) ORDER BY -number LIMIT 5 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 3", support_partial_result=True, ) # Request with partial result limit greater then full limit run_query_without_errors( - "SELECT number FROM numbers_mt(5e6+1) ORDER BY -number LIMIT 3 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 5", + f"SELECT number FROM numbers_mt({rows_number}) ORDER BY -number LIMIT 3 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 5", support_partial_result=True, ) # Request with OFFSET run_query_without_errors( - "SELECT number FROM numbers_mt(5e6+1) ORDER BY -number LIMIT 3 OFFSET 1 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 5", + f"SELECT number FROM numbers_mt({rows_number}) ORDER BY -number LIMIT 3 OFFSET 1 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 5", support_partial_result=True, ) # Request with OFFSET greater then partial result limit (partial result pipeline use blocks with less then OFFSET, so there will be no elements in block after LimitPartialResultTransform) run_query_without_errors( - "SELECT number FROM numbers_mt(5e6+1) ORDER BY -number LIMIT 3 OFFSET 15 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 5", + f"SELECT number FROM numbers_mt({rows_number}) ORDER BY -number LIMIT 3 OFFSET 15 SETTINGS max_threads = 1, partial_result_update_duration_ms = 1, max_rows_in_partial_result = 5", support_partial_result=False, ) diff --git a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.reference b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.reference index 211a193940a..dd3a343560f 100644 --- a/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.reference +++ b/tests/queries/0_stateless/02833_partial_sorting_result_during_query_execution.reference @@ -4,35 +4,35 @@ Rows 3 Columns 1 Column number type UInt64 Rows 5 Columns 1 Column number type UInt64 -5000000 -4999999 -4999998 -4999997 -4999996 +20000000 +19999999 +19999998 +19999997 +19999996 Rows 0 Columns 1 Column number type UInt64 Rows 3 Columns 1 Column number type UInt64 Rows 3 Columns 1 Column number type UInt64 -5000000 -4999999 -4999998 +20000000 +19999999 +19999998 Rows 0 Columns 1 Column number type UInt64 Rows 3 Columns 1 Column number type UInt64 Rows 3 Columns 1 Column number type UInt64 -4999999 -4999998 -4999997 +19999999 +19999998 +19999997 Rows 0 Columns 1 Column number type UInt64 Rows 0 Columns 1 Column number type UInt64 Rows 3 Columns 1 Column number type UInt64 -4999985 -4999984 -4999983 +19999985 +19999984 +19999983 diff --git a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python index 4d869b05580..aa9f80c751f 100644 --- a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python +++ b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.python @@ -82,8 +82,10 @@ def run_query_without_errors( def supported_scenarios_without_key(): + rows_number = 2e7 + 1 + # Simple aggregation query - query = "select median(number), stddevSamp(number), stddevPop(number), max(number), min(number), any(number), count(number), avg(number), sum(number) from numbers_mt(1e7+1) settings max_threads = 1, partial_result_update_duration_ms = 1" + query = f"select median(number), stddevSamp(number), stddevPop(number), max(number), min(number), any(number), count(number), avg(number), sum(number) from numbers_mt({rows_number}) settings max_threads = 1, partial_result_update_duration_ms = 1" invariants = { "median(number)": lambda old_value, new_value: old_value <= new_value, "max(number)": lambda old_value, new_value: old_value <= new_value, @@ -97,7 +99,7 @@ def supported_scenarios_without_key(): ) # Aggregation query with a nested ORDER BY subquery - query = "select median(number), stddevSamp(number), stddevPop(number), max(number), min(number), any(number), count(number), avg(number), sum(number) FROM (SELECT number FROM numbers_mt(1e7) ORDER BY -number LIMIT 3) settings max_threads = 1, partial_result_update_duration_ms=1" + query = f"select median(number), stddevSamp(number), stddevPop(number), max(number), min(number), any(number), count(number), avg(number), sum(number) FROM (SELECT number FROM numbers_mt({rows_number}) ORDER BY -number LIMIT 3) settings max_threads = 1, partial_result_update_duration_ms=1" # Aggregation receives small partial result blocks from ORDER BY which always sends blocks with bigger values invariants["min(number)"] = lambda old_value, new_value: old_value <= new_value @@ -107,12 +109,14 @@ def supported_scenarios_without_key(): def unsupported_scenarios(): + rows_number = 2e7 + 1 + # Currently aggregator for partial result supports only single thread aggregation without key # Update test when multithreading or aggregation with GROUP BY will be supported for partial result updates - multithread_query = "select sum(number) from numbers_mt(1e7+1) settings max_threads = 2, partial_result_update_duration_ms = 100" + multithread_query = f"select sum(number) from numbers_mt({rows_number}) settings max_threads = 2, partial_result_update_duration_ms = 100" run_query_without_errors(multithread_query, support_partial_result=False) - group_with_key_query = "select mod2, sum(number) from numbers_mt(1e7+1) group by number % 2 as mod2 settings max_threads = 1, partial_result_update_duration_ms = 100" + group_with_key_query = f"select mod2, sum(number) from numbers_mt({rows_number}) group by number % 2 as mod2 settings max_threads = 1, partial_result_update_duration_ms = 100" run_query_without_errors(group_with_key_query, support_partial_result=False) diff --git a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.reference b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.reference index a813b18f24f..aea61fad42f 100644 --- a/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.reference +++ b/tests/queries/0_stateless/02834_partial_aggregating_result_during_query_execution.reference @@ -28,11 +28,11 @@ Column any(number) type UInt64 Column count(number) type UInt64 Column avg(number) type Float64 Column sum(number) type UInt64 -max(number) [10000000] +max(number) [20000000] min(number) [0] any(number) [0] -count(number) [10000001] -sum(number) [50000005000000] +count(number) [20000001] +sum(number) [200000010000000] Rows 0 Columns 9 Column median(number) type Float64 Column stddevSamp(number) type Float64 @@ -63,18 +63,18 @@ Column any(number) type UInt64 Column count(number) type UInt64 Column avg(number) type Float64 Column sum(number) type UInt64 -max(number) [9999999] -min(number) [9999997] -any(number) [9999999] +max(number) [20000000] +min(number) [19999998] +any(number) [20000000] count(number) [3] -sum(number) [29999994] +sum(number) [59999997] Rows 0 Columns 1 Column sum(number) type UInt64 Rows 0 Columns 1 Column sum(number) type UInt64 Rows 1 Columns 1 Column sum(number) type UInt64 -sum(number) [50000005000000] +sum(number) [200000010000000] Rows 0 Columns 2 Column mod2 type UInt8 Column sum(number) type UInt64 @@ -85,4 +85,4 @@ Rows 2 Columns 2 Column mod2 type UInt8 Column sum(number) type UInt64 mod2 [0, 1] -sum(number) [25000005000000, 25000000000000] +sum(number) [100000010000000, 100000000000000] From a12a375a1d2fd44984c1b0591e8d3ed533c9d744 Mon Sep 17 00:00:00 2001 From: Igor Nikonov Date: Sat, 9 Sep 2023 12:51:34 +0000 Subject: [PATCH 252/433] Style check --- src/Storages/StorageDistributed.h | 15 +++++++++++---- 1 file changed, 11 insertions(+), 4 deletions(-) diff --git a/src/Storages/StorageDistributed.h b/src/Storages/StorageDistributed.h index fde414f3123..b8445f5ec16 100644 --- a/src/Storages/StorageDistributed.h +++ b/src/Storages/StorageDistributed.h @@ -185,11 +185,18 @@ private: /// Apply the following settings: /// - optimize_skip_unused_shards /// - force_optimize_skip_unused_shards - ClusterPtr getOptimizedCluster(ContextPtr local_context, const StorageSnapshotPtr & storage_snapshot, - const ASTSelectQuery & select, const TreeRewriterResultPtr & syntax_analyzer_result) const; + ClusterPtr getOptimizedCluster( + ContextPtr local_context, + const StorageSnapshotPtr & storage_snapshot, + const ASTSelectQuery & select, + const TreeRewriterResultPtr & syntax_analyzer_result) const; - ClusterPtr skipUnusedShards(ClusterPtr cluster, const ASTSelectQuery & select, const TreeRewriterResultPtr & syntax_analyzer_result, - const StorageSnapshotPtr & storage_snapshot, ContextPtr context) const; + ClusterPtr skipUnusedShards( + ClusterPtr cluster, + const ASTSelectQuery & select, + const TreeRewriterResultPtr & syntax_analyzer_result, + const StorageSnapshotPtr & storage_snapshot, + ContextPtr context) const; /// This method returns optimal query processing stage. /// From 2fdc700da2cfedd51c07f48ab4bb67e558ff11e1 Mon Sep 17 00:00:00 2001 From: Igor Nikonov Date: Sat, 9 Sep 2023 13:33:18 +0000 Subject: [PATCH 253/433] Style --- src/Storages/StorageDistributed.cpp | 1 - 1 file changed, 1 deletion(-) diff --git a/src/Storages/StorageDistributed.cpp b/src/Storages/StorageDistributed.cpp index ddc625745b4..a9b19976033 100644 --- a/src/Storages/StorageDistributed.cpp +++ b/src/Storages/StorageDistributed.cpp @@ -75,7 +75,6 @@ #include #include #include -#include #include #include From 1298dba0c6bbad1551ce819da043975956218154 Mon Sep 17 00:00:00 2001 From: Sema Checherinda Date: Sat, 9 Sep 2023 03:11:14 +0200 Subject: [PATCH 254/433] fix the bug, add useful logs --- .../DiskObjectStorageTransaction.cpp | 68 +++++++++++++++---- .../ObjectStorages/S3/S3ObjectStorage.cpp | 4 +- src/Storages/StorageReplicatedMergeTree.cpp | 9 +++ .../test.py | 51 +++++++++----- 4 files changed, 99 insertions(+), 33 deletions(-) diff --git a/src/Disks/ObjectStorages/DiskObjectStorageTransaction.cpp b/src/Disks/ObjectStorages/DiskObjectStorageTransaction.cpp index 0ae577602b1..b42612d19b3 100644 --- a/src/Disks/ObjectStorages/DiskObjectStorageTransaction.cpp +++ b/src/Disks/ObjectStorages/DiskObjectStorageTransaction.cpp @@ -7,6 +7,7 @@ #include #include +#include namespace DB { @@ -156,14 +157,13 @@ struct RemoveObjectStorageOperation final : public IDiskObjectStorageOperation struct RemoveManyObjectStorageOperation final : public IDiskObjectStorageOperation { - RemoveBatchRequest remove_paths; - bool keep_all_batch_data; - NameSet file_names_remove_metadata_only; + const RemoveBatchRequest remove_paths; + const bool keep_all_batch_data; + const NameSet file_names_remove_metadata_only; + std::vector removed_files; std::vector objects_to_remove; - bool remove_from_cache = false; - RemoveManyObjectStorageOperation( IObjectStorage & object_storage_, IMetadataStorage & metadata_storage_, @@ -203,6 +203,7 @@ struct RemoveManyObjectStorageOperation final : public IDiskObjectStorageOperati if (unlink_outcome && !keep_all_batch_data && !file_names_remove_metadata_only.contains(fs::path(path).filename())) { objects_to_remove.emplace_back(ObjectsToRemove{std::move(objects), std::move(unlink_outcome)}); + removed_files.push_back(path); } } catch (const Exception & e) @@ -213,6 +214,12 @@ struct RemoveManyObjectStorageOperation final : public IDiskObjectStorageOperati || e.code() == ErrorCodes::CANNOT_READ_ALL_DATA || e.code() == ErrorCodes::CANNOT_OPEN_FILE) { + LOG_DEBUG( + &Poco::Logger::get("RemoveManyObjectStorageOperation"), + "Can't read metadata because of an exception. Just remove it from the filesystem. Path: {}, exception: {}", + metadata_storage.getPath() + path, + e.message()); + tx->unlinkFile(path); } else @@ -238,16 +245,31 @@ struct RemoveManyObjectStorageOperation final : public IDiskObjectStorageOperati /// TL;DR Don't pay any attention to 404 status code if (!remove_from_remote.empty()) object_storage.removeObjectsIfExist(remove_from_remote); + + if (!keep_all_batch_data) + { + LOG_DEBUG( + &Poco::Logger::get("RemoveManyObjectStorageOperation"), + "metadata and objects were removed for [{}], " + "only metadata were removed for [{}].", + boost::algorithm::join(removed_files, ", "), + boost::algorithm::join(file_names_remove_metadata_only, ", ")); + } } }; struct RemoveRecursiveObjectStorageOperation final : public IDiskObjectStorageOperation { - std::string path; + /// path inside disk with metadata + const std::string path; + const bool keep_all_batch_data; + /// paths inside the 'this->path' + const NameSet file_names_remove_metadata_only; + + /// map from local_path to its remote objects with hardlinks counter + /// local_path is the path inside 'this->path' std::unordered_map objects_to_remove_by_path; - bool keep_all_batch_data; - NameSet file_names_remove_metadata_only; RemoveRecursiveObjectStorageOperation( IObjectStorage & object_storage_, @@ -274,11 +296,16 @@ struct RemoveRecursiveObjectStorageOperation final : public IDiskObjectStorageOp { try { + chassert(path_to_remove.starts_with(path)); + auto rel_path = String(fs::relative(fs::path(path_to_remove), fs::path(path))); + auto objects_paths = metadata_storage.getStorageObjects(path_to_remove); auto unlink_outcome = tx->unlinkMetadata(path_to_remove); - if (unlink_outcome) + + if (unlink_outcome && !file_names_remove_metadata_only.contains(rel_path)) { - objects_to_remove_by_path[path_to_remove] = ObjectsToRemove{std::move(objects_paths), std::move(unlink_outcome)}; + objects_to_remove_by_path[std::move(rel_path)] + = ObjectsToRemove{std::move(objects_paths), std::move(unlink_outcome)}; } } catch (const Exception & e) @@ -320,25 +347,38 @@ struct RemoveRecursiveObjectStorageOperation final : public IDiskObjectStorageOp void undo() override { - } void finalize() override { if (!keep_all_batch_data) { + std::vector total_removed_paths; + total_removed_paths.reserve(objects_to_remove_by_path.size()); + StoredObjects remove_from_remote; for (auto && [local_path, objects_to_remove] : objects_to_remove_by_path) { - if (!file_names_remove_metadata_only.contains(fs::path(local_path).filename())) + chassert(!file_names_remove_metadata_only.contains(local_path)); + if (objects_to_remove.unlink_outcome->num_hardlinks == 0) { - if (objects_to_remove.unlink_outcome->num_hardlinks == 0) - std::move(objects_to_remove.objects.begin(), objects_to_remove.objects.end(), std::back_inserter(remove_from_remote)); + std::move(objects_to_remove.objects.begin(), objects_to_remove.objects.end(), std::back_inserter(remove_from_remote)); + total_removed_paths.push_back(local_path); } } + /// Read comment inside RemoveObjectStorageOperation class /// TL;DR Don't pay any attention to 404 status code object_storage.removeObjectsIfExist(remove_from_remote); + + LOG_DEBUG( + &Poco::Logger::get("RemoveRecursiveObjectStorageOperation"), + "Recursively remove path {}: " + "metadata and objects were removed for [{}], " + "only metadata were removed for [{}].", + path, + boost::algorithm::join(total_removed_paths, ", "), + boost::algorithm::join(file_names_remove_metadata_only, ", ")); } } }; diff --git a/src/Disks/ObjectStorages/S3/S3ObjectStorage.cpp b/src/Disks/ObjectStorages/S3/S3ObjectStorage.cpp index bbfc6609079..0d9670efebe 100644 --- a/src/Disks/ObjectStorages/S3/S3ObjectStorage.cpp +++ b/src/Disks/ObjectStorages/S3/S3ObjectStorage.cpp @@ -320,7 +320,7 @@ void S3ObjectStorage::removeObjectImpl(const StoredObject & object, bool if_exis throwIfUnexpectedError(outcome, if_exists); - LOG_TRACE(log, "Object with path {} was removed from S3", object.remote_path); + LOG_DEBUG(log, "Object with path {} was removed from S3", object.remote_path); } void S3ObjectStorage::removeObjectsImpl(const StoredObjects & objects, bool if_exists) @@ -368,7 +368,7 @@ void S3ObjectStorage::removeObjectsImpl(const StoredObjects & objects, bool if_e throwIfUnexpectedError(outcome, if_exists); - LOG_TRACE(log, "Objects with paths [{}] were removed from S3", keys); + LOG_DEBUG(log, "Objects with paths [{}] were removed from S3", keys); } } } diff --git a/src/Storages/StorageReplicatedMergeTree.cpp b/src/Storages/StorageReplicatedMergeTree.cpp index cabdf67a315..d1be8bd249d 100644 --- a/src/Storages/StorageReplicatedMergeTree.cpp +++ b/src/Storages/StorageReplicatedMergeTree.cpp @@ -1593,6 +1593,7 @@ MergeTreeData::DataPartsVector StorageReplicatedMergeTree::checkPartChecksumsAnd while (true) { LOG_DEBUG(log, "Committing part {} to zookeeper", part->name); + Coordination::Requests ops; NameSet absent_part_paths_on_replicas; @@ -8788,6 +8789,14 @@ void StorageReplicatedMergeTree::getLockSharedDataOps( { String zookeeper_node = fs::path(zc_zookeeper_path) / id / replica_name; + if (!path_to_set_hardlinked_files.empty() && !hardlinks.empty()) + { + LOG_DEBUG(log, "Locking shared node {} with hardlinks from the other shared node {}, " + "hardlinks: [{}]", + zookeeper_node, path_to_set_hardlinked_files, + boost::algorithm::join(hardlinks, ",")); + } + getZeroCopyLockNodeCreateOps( zookeeper, zookeeper_node, requests, zkutil::CreateMode::Persistent, replace_existing_lock, path_to_set_hardlinked_files, hardlinks); diff --git a/tests/integration/test_replicated_zero_copy_projection_mutation/test.py b/tests/integration/test_replicated_zero_copy_projection_mutation/test.py index 3f594158a46..1b68aac08a7 100644 --- a/tests/integration/test_replicated_zero_copy_projection_mutation/test.py +++ b/tests/integration/test_replicated_zero_copy_projection_mutation/test.py @@ -192,6 +192,8 @@ def test_hardlinks_preserved_when_projection_dropped( first_cluster_node.query(create_query + first_node_settings) second_cluster_node.query(create_query + second_node_settings) + objects_empty_table = list_objects(cluster) + first_cluster_node.query("SYSTEM FLUSH LOGS") table_uuid = first_cluster_node.query( """ @@ -239,6 +241,7 @@ def test_hardlinks_preserved_when_projection_dropped( sleep_time=1, ) + # make sure that alter update made hardlinks inside projection hardlinks = ( first_cluster_node.query( f""" @@ -255,8 +258,6 @@ def test_hardlinks_preserved_when_projection_dropped( assert len(hardlinks) > 0, ",".join(hardlinks) assert any(["proj/" in x for x in hardlinks]), ",".join(hardlinks) - logging.info("hardlinks has been planted") - part_path_on_second_node = second_cluster_node.query( """ SELECT path FROM system.parts @@ -284,6 +285,7 @@ def test_hardlinks_preserved_when_projection_dropped( second_cluster_node.query("SYSTEM FLUSH LOGS") + # make sure there is outdated broken-on-start part broken_parts = ( second_cluster_node.query( """ @@ -310,6 +312,7 @@ def test_hardlinks_preserved_when_projection_dropped( ) second_cluster_node.exec_in_container(["bash", "-c", script]) + # when detached part is removed, removeSharedRecursive is called second_cluster_node.query( f""" ALTER TABLE test_hardlinks_preserved_when_projection_dropped @@ -318,21 +321,35 @@ def test_hardlinks_preserved_when_projection_dropped( settings={"allow_drop_detached": "1"}, ) - res = first_cluster_node.query( - """ - CHECK TABLE test_hardlinks_preserved_when_projection_dropped - """ - ).strip() + # it is an easy way to read all data in part + # "0" means corrupted, https://clickhouse.com/docs/en/sql-reference/statements/check-table + assert ( + "1" + == first_cluster_node.query( + """ + CHECK TABLE test_hardlinks_preserved_when_projection_dropped + """ + ).strip() + ) - # corrupted - assert res == "0" + assert ( + "1" + == second_cluster_node.query( + """ + CHECK TABLE test_hardlinks_preserved_when_projection_dropped + """ + ).strip() + ) - data = first_cluster_node.query( - """ - SELECT a, b FROM test_hardlinks_preserved_when_projection_dropped - WHERE b > 104 - ORDER BY b - """ - ).split("\n") + second_cluster_node.query( + f""" + ALTER TABLE test_hardlinks_preserved_when_projection_dropped + DROP PART 'all_0_0_0_1' + """, + settings={"alter_sync": 2}, + ) - assert len(data) == 5, data + wait_for_delete_s3_objects(cluster, len(objects_empty_table)) + + objects_at_the_end = list_objects(cluster) + assert objects_at_the_end == objects_empty_table From 0a05831839d552488cf77b4c6d6eb00a74308f21 Mon Sep 17 00:00:00 2001 From: Robert Schulze Date: Sat, 9 Sep 2023 16:48:17 +0000 Subject: [PATCH 255/433] Cosmetics --- src/Functions/DateTimeTransforms.h | 556 ++++++++++++++--------------- src/Functions/base64Decode.cpp | 2 +- src/Functions/base64Encode.cpp | 2 +- src/Functions/toDayOfMonth.cpp | 2 +- src/Functions/toDayOfWeek.cpp | 2 +- src/Functions/toDayOfYear.cpp | 2 +- src/Functions/toHour.cpp | 2 +- src/Functions/toMinute.cpp | 2 +- src/Functions/toMonth.cpp | 2 +- src/Functions/toQuarter.cpp | 2 +- src/Functions/toSecond.cpp | 2 +- src/Functions/toYear.cpp | 2 +- 12 files changed, 289 insertions(+), 289 deletions(-) diff --git a/src/Functions/DateTimeTransforms.h b/src/Functions/DateTimeTransforms.h index a1c880f6956..0fb4edc1375 100644 --- a/src/Functions/DateTimeTransforms.h +++ b/src/Functions/DateTimeTransforms.h @@ -49,37 +49,37 @@ namespace ErrorCodes /// This factor transformation will say that the function is monotone everywhere. struct ZeroTransform { - static inline UInt16 execute(Int64, const DateLUTImpl &) { return 0; } - static inline UInt16 execute(UInt32, const DateLUTImpl &) { return 0; } - static inline UInt16 execute(Int32, const DateLUTImpl &) { return 0; } - static inline UInt16 execute(UInt16, const DateLUTImpl &) { return 0; } + static UInt16 execute(Int64, const DateLUTImpl &) { return 0; } + static UInt16 execute(UInt32, const DateLUTImpl &) { return 0; } + static UInt16 execute(Int32, const DateLUTImpl &) { return 0; } + static UInt16 execute(UInt16, const DateLUTImpl &) { return 0; } }; struct ToDateImpl { static constexpr auto name = "toDate"; - static inline UInt16 execute(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) + static UInt16 execute(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) { return static_cast(time_zone.toDayNum(t.whole)); } - static inline UInt16 execute(Int64 t, const DateLUTImpl & time_zone) + static UInt16 execute(Int64 t, const DateLUTImpl & time_zone) { return UInt16(time_zone.toDayNum(t)); } - static inline UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) { return UInt16(time_zone.toDayNum(t)); } - static inline UInt16 execute(Int32, const DateLUTImpl &) + static UInt16 execute(Int32, const DateLUTImpl &) { throwDateIsNotSupported(name); } - static inline UInt16 execute(UInt16 d, const DateLUTImpl &) + static UInt16 execute(UInt16 d, const DateLUTImpl &) { return d; } - static inline DecimalUtils::DecimalComponents executeExtendedResult(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) + static DecimalUtils::DecimalComponents executeExtendedResult(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) { return {time_zone.toDayNum(t.whole), 0}; } @@ -91,20 +91,20 @@ struct ToDate32Impl { static constexpr auto name = "toDate32"; - static inline Int32 execute(Int64 t, const DateLUTImpl & time_zone) + static Int32 execute(Int64 t, const DateLUTImpl & time_zone) { return Int32(time_zone.toDayNum(t)); } - static inline Int32 execute(UInt32 t, const DateLUTImpl & time_zone) + static Int32 execute(UInt32 t, const DateLUTImpl & time_zone) { /// Don't saturate. return Int32(time_zone.toDayNum(t)); } - static inline Int32 execute(Int32 d, const DateLUTImpl &) + static Int32 execute(Int32 d, const DateLUTImpl &) { return d; } - static inline Int32 execute(UInt16 d, const DateLUTImpl &) + static Int32 execute(UInt16 d, const DateLUTImpl &) { return d; } @@ -116,27 +116,27 @@ struct ToStartOfDayImpl { static constexpr auto name = "toStartOfDay"; - static inline UInt32 execute(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) + static UInt32 execute(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) { return static_cast(time_zone.toDate(static_cast(t.whole))); } - static inline UInt32 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt32 execute(UInt32 t, const DateLUTImpl & time_zone) { return static_cast(time_zone.toDate(t)); } - static inline UInt32 execute(Int32 d, const DateLUTImpl & time_zone) + static UInt32 execute(Int32 d, const DateLUTImpl & time_zone) { return static_cast(time_zone.toDate(ExtendedDayNum(d))); } - static inline UInt32 execute(UInt16 d, const DateLUTImpl & time_zone) + static UInt32 execute(UInt16 d, const DateLUTImpl & time_zone) { return static_cast(time_zone.toDate(DayNum(d))); } - static inline DecimalUtils::DecimalComponents executeExtendedResult(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) + static DecimalUtils::DecimalComponents executeExtendedResult(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) { return {time_zone.toDate(t.whole), 0}; } - static inline Int64 executeExtendedResult(Int32 d, const DateLUTImpl & time_zone) + static Int64 executeExtendedResult(Int32 d, const DateLUTImpl & time_zone) { return time_zone.fromDayNum(ExtendedDayNum(d)) * DecimalUtils::scaleMultiplier(DataTypeDateTime64::default_scale); } @@ -148,29 +148,29 @@ struct ToMondayImpl { static constexpr auto name = "toMonday"; - static inline UInt16 execute(Int64 t, const DateLUTImpl & time_zone) + static UInt16 execute(Int64 t, const DateLUTImpl & time_zone) { //return time_zone.toFirstDayNumOfWeek(time_zone.toDayNum(t)); return time_zone.toFirstDayNumOfWeek(t); } - static inline UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) { //return time_zone.toFirstDayNumOfWeek(time_zone.toDayNum(t)); return time_zone.toFirstDayNumOfWeek(t); } - static inline UInt16 execute(Int32 d, const DateLUTImpl & time_zone) + static UInt16 execute(Int32 d, const DateLUTImpl & time_zone) { return time_zone.toFirstDayNumOfWeek(ExtendedDayNum(d)); } - static inline UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) + static UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) { return time_zone.toFirstDayNumOfWeek(DayNum(d)); } - static inline Int64 executeExtendedResult(Int64 t, const DateLUTImpl & time_zone) + static Int64 executeExtendedResult(Int64 t, const DateLUTImpl & time_zone) { return time_zone.toFirstDayNumOfWeek(time_zone.toDayNum(t)); } - static inline Int32 executeExtendedResult(Int32 d, const DateLUTImpl & time_zone) + static Int32 executeExtendedResult(Int32 d, const DateLUTImpl & time_zone) { return time_zone.toFirstDayNumOfWeek(ExtendedDayNum(d)); } @@ -181,27 +181,27 @@ struct ToStartOfMonthImpl { static constexpr auto name = "toStartOfMonth"; - static inline UInt16 execute(Int64 t, const DateLUTImpl & time_zone) + static UInt16 execute(Int64 t, const DateLUTImpl & time_zone) { return time_zone.toFirstDayNumOfMonth(time_zone.toDayNum(t)); } - static inline UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toFirstDayNumOfMonth(time_zone.toDayNum(t)); } - static inline UInt16 execute(Int32 d, const DateLUTImpl & time_zone) + static UInt16 execute(Int32 d, const DateLUTImpl & time_zone) { return time_zone.toFirstDayNumOfMonth(ExtendedDayNum(d)); } - static inline UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) + static UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) { return time_zone.toFirstDayNumOfMonth(DayNum(d)); } - static inline Int64 executeExtendedResult(Int64 t, const DateLUTImpl & time_zone) + static Int64 executeExtendedResult(Int64 t, const DateLUTImpl & time_zone) { return time_zone.toFirstDayNumOfMonth(time_zone.toDayNum(t)); } - static inline Int32 executeExtendedResult(Int32 d, const DateLUTImpl & time_zone) + static Int32 executeExtendedResult(Int32 d, const DateLUTImpl & time_zone) { return time_zone.toFirstDayNumOfMonth(ExtendedDayNum(d)); } @@ -213,27 +213,27 @@ struct ToLastDayOfMonthImpl { static constexpr auto name = "toLastDayOfMonth"; - static inline UInt16 execute(Int64 t, const DateLUTImpl & time_zone) + static UInt16 execute(Int64 t, const DateLUTImpl & time_zone) { return time_zone.toLastDayNumOfMonth(time_zone.toDayNum(t)); } - static inline UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toLastDayNumOfMonth(time_zone.toDayNum(t)); } - static inline UInt16 execute(Int32 d, const DateLUTImpl & time_zone) + static UInt16 execute(Int32 d, const DateLUTImpl & time_zone) { return time_zone.toLastDayNumOfMonth(ExtendedDayNum(d)); } - static inline UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) + static UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) { return time_zone.toLastDayNumOfMonth(DayNum(d)); } - static inline Int64 executeExtendedResult(Int64 t, const DateLUTImpl & time_zone) + static Int64 executeExtendedResult(Int64 t, const DateLUTImpl & time_zone) { return time_zone.toLastDayNumOfMonth(time_zone.toDayNum(t)); } - static inline Int32 executeExtendedResult(Int32 d, const DateLUTImpl & time_zone) + static Int32 executeExtendedResult(Int32 d, const DateLUTImpl & time_zone) { return time_zone.toLastDayNumOfMonth(ExtendedDayNum(d)); } @@ -244,27 +244,27 @@ struct ToStartOfQuarterImpl { static constexpr auto name = "toStartOfQuarter"; - static inline UInt16 execute(Int64 t, const DateLUTImpl & time_zone) + static UInt16 execute(Int64 t, const DateLUTImpl & time_zone) { return time_zone.toFirstDayNumOfQuarter(time_zone.toDayNum(t)); } - static inline UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toFirstDayNumOfQuarter(time_zone.toDayNum(t)); } - static inline UInt16 execute(Int32 d, const DateLUTImpl & time_zone) + static UInt16 execute(Int32 d, const DateLUTImpl & time_zone) { return time_zone.toFirstDayNumOfQuarter(ExtendedDayNum(d)); } - static inline UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) + static UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) { return time_zone.toFirstDayNumOfQuarter(DayNum(d)); } - static inline Int64 executeExtendedResult(Int64 t, const DateLUTImpl & time_zone) + static Int64 executeExtendedResult(Int64 t, const DateLUTImpl & time_zone) { return time_zone.toFirstDayNumOfQuarter(time_zone.toDayNum(t)); } - static inline Int32 executeExtendedResult(Int32 d, const DateLUTImpl & time_zone) + static Int32 executeExtendedResult(Int32 d, const DateLUTImpl & time_zone) { return time_zone.toFirstDayNumOfQuarter(ExtendedDayNum(d)); } @@ -275,27 +275,27 @@ struct ToStartOfYearImpl { static constexpr auto name = "toStartOfYear"; - static inline UInt16 execute(Int64 t, const DateLUTImpl & time_zone) + static UInt16 execute(Int64 t, const DateLUTImpl & time_zone) { return time_zone.toFirstDayNumOfYear(time_zone.toDayNum(t)); } - static inline UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toFirstDayNumOfYear(time_zone.toDayNum(t)); } - static inline UInt16 execute(Int32 d, const DateLUTImpl & time_zone) + static UInt16 execute(Int32 d, const DateLUTImpl & time_zone) { return time_zone.toFirstDayNumOfYear(ExtendedDayNum(d)); } - static inline UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) + static UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) { return time_zone.toFirstDayNumOfYear(DayNum(d)); } - static inline Int64 executeExtendedResult(Int64 t, const DateLUTImpl & time_zone) + static Int64 executeExtendedResult(Int64 t, const DateLUTImpl & time_zone) { return time_zone.toFirstDayNumOfYear(time_zone.toDayNum(t)); } - static inline Int32 executeExtendedResult(Int32 d, const DateLUTImpl & time_zone) + static Int32 executeExtendedResult(Int32 d, const DateLUTImpl & time_zone) { return time_zone.toFirstDayNumOfYear(ExtendedDayNum(d)); } @@ -313,19 +313,19 @@ struct ToTimeImpl { return static_cast(time_zone.toTime(t.whole) + 86400); } - static inline UInt32 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt32 execute(UInt32 t, const DateLUTImpl & time_zone) { return static_cast(time_zone.toTime(t) + 86400); } - static inline UInt32 execute(Int32, const DateLUTImpl &) + static UInt32 execute(Int32, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline UInt32 execute(UInt16, const DateLUTImpl &) + static UInt32 execute(UInt16, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline constexpr bool hasPreimage() { return false; } + static constexpr bool hasPreimage() { return false; } using FactorTransform = ToDateImpl; }; @@ -334,27 +334,27 @@ struct ToStartOfMinuteImpl { static constexpr auto name = "toStartOfMinute"; - static inline UInt32 execute(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) + static UInt32 execute(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) { return static_cast(time_zone.toStartOfMinute(t.whole)); } - static inline UInt32 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt32 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toStartOfMinute(t); } - static inline UInt32 execute(Int32, const DateLUTImpl &) + static UInt32 execute(Int32, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline UInt32 execute(UInt16, const DateLUTImpl &) + static UInt32 execute(UInt16, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline DecimalUtils::DecimalComponents executeExtendedResult(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) + static DecimalUtils::DecimalComponents executeExtendedResult(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) { return {time_zone.toStartOfMinute(t.whole), 0}; } - static inline Int64 executeExtendedResult(Int32, const DateLUTImpl &) + static Int64 executeExtendedResult(Int32, const DateLUTImpl &) { throwDate32IsNotSupported(name); } @@ -369,7 +369,7 @@ struct ToStartOfSecondImpl { static constexpr auto name = "toStartOfSecond"; - static inline DateTime64 execute(const DateTime64 & datetime64, Int64 scale_multiplier, const DateLUTImpl &) + static DateTime64 execute(const DateTime64 & datetime64, Int64 scale_multiplier, const DateLUTImpl &) { auto fractional_with_sign = DecimalUtils::getFractionalPartWithScaleMultiplier(datetime64, scale_multiplier); @@ -385,19 +385,19 @@ struct ToStartOfSecondImpl return datetime64 - fractional_with_sign; } - static inline UInt32 execute(UInt32, const DateLUTImpl &) + static UInt32 execute(UInt32, const DateLUTImpl &) { throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Illegal type DateTime of argument for function {}", name); } - static inline UInt32 execute(Int32, const DateLUTImpl &) + static UInt32 execute(Int32, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline UInt32 execute(UInt16, const DateLUTImpl &) + static UInt32 execute(UInt16, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline constexpr bool hasPreimage() { return false; } + static constexpr bool hasPreimage() { return false; } using FactorTransform = ZeroTransform; }; @@ -406,7 +406,7 @@ struct ToStartOfMillisecondImpl { static constexpr auto name = "toStartOfMillisecond"; - static inline DateTime64 execute(const DateTime64 & datetime64, Int64 scale_multiplier, const DateLUTImpl &) + static DateTime64 execute(const DateTime64 & datetime64, Int64 scale_multiplier, const DateLUTImpl &) { // given that scale is 6, scale_multiplier is 1000000 // for DateTime64 value of 123.456789: @@ -433,19 +433,19 @@ struct ToStartOfMillisecondImpl } } - static inline UInt32 execute(UInt32, const DateLUTImpl &) + static UInt32 execute(UInt32, const DateLUTImpl &) { throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Illegal type DateTime of argument for function {}", name); } - static inline UInt32 execute(Int32, const DateLUTImpl &) + static UInt32 execute(Int32, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline UInt32 execute(UInt16, const DateLUTImpl &) + static UInt32 execute(UInt16, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline constexpr bool hasPreimage() { return false; } + static constexpr bool hasPreimage() { return false; } using FactorTransform = ZeroTransform; }; @@ -454,7 +454,7 @@ struct ToStartOfMicrosecondImpl { static constexpr auto name = "toStartOfMicrosecond"; - static inline DateTime64 execute(const DateTime64 & datetime64, Int64 scale_multiplier, const DateLUTImpl &) + static DateTime64 execute(const DateTime64 & datetime64, Int64 scale_multiplier, const DateLUTImpl &) { // @see ToStartOfMillisecondImpl @@ -477,19 +477,19 @@ struct ToStartOfMicrosecondImpl } } - static inline UInt32 execute(UInt32, const DateLUTImpl &) + static UInt32 execute(UInt32, const DateLUTImpl &) { throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Illegal type DateTime of argument for function {}", name); } - static inline UInt32 execute(Int32, const DateLUTImpl &) + static UInt32 execute(Int32, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline UInt32 execute(UInt16, const DateLUTImpl &) + static UInt32 execute(UInt16, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline constexpr bool hasPreimage() { return false; } + static constexpr bool hasPreimage() { return false; } using FactorTransform = ZeroTransform; }; @@ -498,7 +498,7 @@ struct ToStartOfNanosecondImpl { static constexpr auto name = "toStartOfNanosecond"; - static inline DateTime64 execute(const DateTime64 & datetime64, Int64 scale_multiplier, const DateLUTImpl &) + static DateTime64 execute(const DateTime64 & datetime64, Int64 scale_multiplier, const DateLUTImpl &) { // @see ToStartOfMillisecondImpl if (scale_multiplier == 1000000000) @@ -515,19 +515,19 @@ struct ToStartOfNanosecondImpl } } - static inline UInt32 execute(UInt32, const DateLUTImpl &) + static UInt32 execute(UInt32, const DateLUTImpl &) { throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Illegal type DateTime of argument for function {}", name); } - static inline UInt32 execute(Int32, const DateLUTImpl &) + static UInt32 execute(Int32, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline UInt32 execute(UInt16, const DateLUTImpl &) + static UInt32 execute(UInt16, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline constexpr bool hasPreimage() { return false; } + static constexpr bool hasPreimage() { return false; } using FactorTransform = ZeroTransform; }; @@ -536,27 +536,27 @@ struct ToStartOfFiveMinutesImpl { static constexpr auto name = "toStartOfFiveMinutes"; - static inline UInt32 execute(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) + static UInt32 execute(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) { return static_cast(time_zone.toStartOfFiveMinutes(t.whole)); } - static inline UInt32 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt32 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toStartOfFiveMinutes(t); } - static inline UInt32 execute(Int32, const DateLUTImpl &) + static UInt32 execute(Int32, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline UInt32 execute(UInt16, const DateLUTImpl &) + static UInt32 execute(UInt16, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline DecimalUtils::DecimalComponents executeExtendedResult(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) + static DecimalUtils::DecimalComponents executeExtendedResult(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) { return {time_zone.toStartOfFiveMinutes(t.whole), 0}; } - static inline Int64 executeExtendedResult(Int32, const DateLUTImpl &) + static Int64 executeExtendedResult(Int32, const DateLUTImpl &) { throwDate32IsNotSupported(name); } @@ -568,27 +568,27 @@ struct ToStartOfTenMinutesImpl { static constexpr auto name = "toStartOfTenMinutes"; - static inline UInt32 execute(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) + static UInt32 execute(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) { return static_cast(time_zone.toStartOfTenMinutes(t.whole)); } - static inline UInt32 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt32 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toStartOfTenMinutes(t); } - static inline UInt32 execute(Int32, const DateLUTImpl &) + static UInt32 execute(Int32, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline UInt32 execute(UInt16, const DateLUTImpl &) + static UInt32 execute(UInt16, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline DecimalUtils::DecimalComponents executeExtendedResult(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) + static DecimalUtils::DecimalComponents executeExtendedResult(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) { return {time_zone.toStartOfTenMinutes(t.whole), 0}; } - static inline Int64 executeExtendedResult(Int32, const DateLUTImpl &) + static Int64 executeExtendedResult(Int32, const DateLUTImpl &) { throwDate32IsNotSupported(name); } @@ -600,27 +600,27 @@ struct ToStartOfFifteenMinutesImpl { static constexpr auto name = "toStartOfFifteenMinutes"; - static inline UInt32 execute(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) + static UInt32 execute(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) { return static_cast(time_zone.toStartOfFifteenMinutes(t.whole)); } - static inline UInt32 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt32 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toStartOfFifteenMinutes(t); } - static inline UInt32 execute(Int32, const DateLUTImpl &) + static UInt32 execute(Int32, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline UInt32 execute(UInt16, const DateLUTImpl &) + static UInt32 execute(UInt16, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline DecimalUtils::DecimalComponents executeExtendedResult(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) + static DecimalUtils::DecimalComponents executeExtendedResult(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) { return {time_zone.toStartOfFifteenMinutes(t.whole), 0}; } - static inline Int64 executeExtendedResult(Int32, const DateLUTImpl &) + static Int64 executeExtendedResult(Int32, const DateLUTImpl &) { throwDate32IsNotSupported(name); } @@ -633,34 +633,34 @@ struct TimeSlotImpl { static constexpr auto name = "timeSlot"; - static inline UInt32 execute(const DecimalUtils::DecimalComponents & t, const DateLUTImpl &) + static UInt32 execute(const DecimalUtils::DecimalComponents & t, const DateLUTImpl &) { return static_cast(t.whole / 1800 * 1800); } - static inline UInt32 execute(UInt32 t, const DateLUTImpl &) + static UInt32 execute(UInt32 t, const DateLUTImpl &) { return t / 1800 * 1800; } - static inline UInt32 execute(Int32, const DateLUTImpl &) + static UInt32 execute(Int32, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline UInt32 execute(UInt16, const DateLUTImpl &) + static UInt32 execute(UInt16, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline DecimalUtils::DecimalComponents executeExtendedResult(const DecimalUtils::DecimalComponents & t, const DateLUTImpl &) + static DecimalUtils::DecimalComponents executeExtendedResult(const DecimalUtils::DecimalComponents & t, const DateLUTImpl &) { if (likely(t.whole >= 0)) return {t.whole / 1800 * 1800, 0}; return {(t.whole + 1 - 1800) / 1800 * 1800, 0}; } - static inline Int64 executeExtendedResult(Int32, const DateLUTImpl &) + static Int64 executeExtendedResult(Int32, const DateLUTImpl &) { throwDate32IsNotSupported(name); } @@ -672,32 +672,32 @@ struct ToStartOfHourImpl { static constexpr auto name = "toStartOfHour"; - static inline UInt32 execute(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) + static UInt32 execute(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) { return static_cast(time_zone.toStartOfHour(t.whole)); } - static inline UInt32 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt32 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toStartOfHour(t); } - static inline UInt32 execute(Int32, const DateLUTImpl &) + static UInt32 execute(Int32, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline UInt32 execute(UInt16, const DateLUTImpl &) + static UInt32 execute(UInt16, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline DecimalUtils::DecimalComponents executeExtendedResult(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) + static DecimalUtils::DecimalComponents executeExtendedResult(const DecimalUtils::DecimalComponents & t, const DateLUTImpl & time_zone) { return {time_zone.toStartOfHour(t.whole), 0}; } - static inline Int64 executeExtendedResult(Int32, const DateLUTImpl &) + static Int64 executeExtendedResult(Int32, const DateLUTImpl &) { throwDate32IsNotSupported(name); } @@ -709,26 +709,26 @@ struct ToYearImpl { static constexpr auto name = "toYear"; - static inline UInt16 execute(Int64 t, const DateLUTImpl & time_zone) + static UInt16 execute(Int64 t, const DateLUTImpl & time_zone) { return time_zone.toYear(t); } - static inline UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toYear(t); } - static inline UInt16 execute(Int32 d, const DateLUTImpl & time_zone) + static UInt16 execute(Int32 d, const DateLUTImpl & time_zone) { return time_zone.toYear(ExtendedDayNum(d)); } - static inline UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) + static UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) { return time_zone.toYear(DayNum(d)); } - static inline constexpr bool hasPreimage() { return true; } + static constexpr bool hasPreimage() { return true; } - static inline RangeOrNull getPreimage(const IDataType & type, const Field & point) + static RangeOrNull getPreimage(const IDataType & type, const Field & point) { if (point.getType() != Field::Types::UInt64) return std::nullopt; @@ -757,19 +757,19 @@ struct ToWeekYearImpl static constexpr Int8 week_mode = 3; - static inline UInt16 execute(Int64 t, const DateLUTImpl & time_zone) + static UInt16 execute(Int64 t, const DateLUTImpl & time_zone) { return time_zone.toYearWeek(t, week_mode).first; } - static inline UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toYearWeek(t, week_mode).first; } - static inline UInt16 execute(Int32 d, const DateLUTImpl & time_zone) + static UInt16 execute(Int32 d, const DateLUTImpl & time_zone) { return time_zone.toYearWeek(ExtendedDayNum(d), week_mode).first; } - static inline UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) + static UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) { return time_zone.toYearWeek(DayNum(d), week_mode).first; } @@ -781,19 +781,19 @@ struct ToWeekOfWeekYearImpl { static constexpr auto name = "toWeekOfWeekYear"; - static inline UInt16 execute(Int64 t, const DateLUTImpl & time_zone) + static UInt16 execute(Int64 t, const DateLUTImpl & time_zone) { return time_zone.toISOWeek(t); } - static inline UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toISOWeek(t); } - static inline UInt16 execute(Int32 d, const DateLUTImpl & time_zone) + static UInt16 execute(Int32 d, const DateLUTImpl & time_zone) { return time_zone.toISOWeek(ExtendedDayNum(d)); } - static inline UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) + static UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) { return time_zone.toISOWeek(DayNum(d)); } @@ -805,23 +805,23 @@ struct ToQuarterImpl { static constexpr auto name = "toQuarter"; - static inline UInt8 execute(Int64 t, const DateLUTImpl & time_zone) + static UInt8 execute(Int64 t, const DateLUTImpl & time_zone) { return time_zone.toQuarter(t); } - static inline UInt8 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt8 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toQuarter(t); } - static inline UInt8 execute(Int32 d, const DateLUTImpl & time_zone) + static UInt8 execute(Int32 d, const DateLUTImpl & time_zone) { return time_zone.toQuarter(ExtendedDayNum(d)); } - static inline UInt8 execute(UInt16 d, const DateLUTImpl & time_zone) + static UInt8 execute(UInt16 d, const DateLUTImpl & time_zone) { return time_zone.toQuarter(DayNum(d)); } - static inline constexpr bool hasPreimage() { return false; } + static constexpr bool hasPreimage() { return false; } using FactorTransform = ToStartOfYearImpl; }; @@ -830,23 +830,23 @@ struct ToMonthImpl { static constexpr auto name = "toMonth"; - static inline UInt8 execute(Int64 t, const DateLUTImpl & time_zone) + static UInt8 execute(Int64 t, const DateLUTImpl & time_zone) { return time_zone.toMonth(t); } - static inline UInt8 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt8 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toMonth(t); } - static inline UInt8 execute(Int32 d, const DateLUTImpl & time_zone) + static UInt8 execute(Int32 d, const DateLUTImpl & time_zone) { return time_zone.toMonth(ExtendedDayNum(d)); } - static inline UInt8 execute(UInt16 d, const DateLUTImpl & time_zone) + static UInt8 execute(UInt16 d, const DateLUTImpl & time_zone) { return time_zone.toMonth(DayNum(d)); } - static inline constexpr bool hasPreimage() { return false; } + static constexpr bool hasPreimage() { return false; } using FactorTransform = ToStartOfYearImpl; }; @@ -855,24 +855,24 @@ struct ToDayOfMonthImpl { static constexpr auto name = "toDayOfMonth"; - static inline UInt8 execute(Int64 t, const DateLUTImpl & time_zone) + static UInt8 execute(Int64 t, const DateLUTImpl & time_zone) { return time_zone.toDayOfMonth(t); } - static inline UInt8 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt8 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toDayOfMonth(t); } - static inline UInt8 execute(Int32 d, const DateLUTImpl & time_zone) + static UInt8 execute(Int32 d, const DateLUTImpl & time_zone) { return time_zone.toDayOfMonth(ExtendedDayNum(d)); } - static inline UInt8 execute(UInt16 d, const DateLUTImpl & time_zone) + static UInt8 execute(UInt16 d, const DateLUTImpl & time_zone) { return time_zone.toDayOfMonth(DayNum(d)); } - static inline constexpr bool hasPreimage() { return false; } + static constexpr bool hasPreimage() { return false; } using FactorTransform = ToStartOfMonthImpl; }; @@ -880,19 +880,19 @@ struct ToDayOfWeekImpl { static constexpr auto name = "toDayOfWeek"; - static inline UInt8 execute(Int64 t, UInt8 mode, const DateLUTImpl & time_zone) + static UInt8 execute(Int64 t, UInt8 mode, const DateLUTImpl & time_zone) { return time_zone.toDayOfWeek(t, mode); } - static inline UInt8 execute(UInt32 t, UInt8 mode, const DateLUTImpl & time_zone) + static UInt8 execute(UInt32 t, UInt8 mode, const DateLUTImpl & time_zone) { return time_zone.toDayOfWeek(t, mode); } - static inline UInt8 execute(Int32 d, UInt8 mode, const DateLUTImpl & time_zone) + static UInt8 execute(Int32 d, UInt8 mode, const DateLUTImpl & time_zone) { return time_zone.toDayOfWeek(ExtendedDayNum(d), mode); } - static inline UInt8 execute(UInt16 d, UInt8 mode, const DateLUTImpl & time_zone) + static UInt8 execute(UInt16 d, UInt8 mode, const DateLUTImpl & time_zone) { return time_zone.toDayOfWeek(DayNum(d), mode); } @@ -904,23 +904,23 @@ struct ToDayOfYearImpl { static constexpr auto name = "toDayOfYear"; - static inline UInt16 execute(Int64 t, const DateLUTImpl & time_zone) + static UInt16 execute(Int64 t, const DateLUTImpl & time_zone) { return time_zone.toDayOfYear(t); } - static inline UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toDayOfYear(t); } - static inline UInt16 execute(Int32 d, const DateLUTImpl & time_zone) + static UInt16 execute(Int32 d, const DateLUTImpl & time_zone) { return time_zone.toDayOfYear(ExtendedDayNum(d)); } - static inline UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) + static UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) { return time_zone.toDayOfYear(DayNum(d)); } - static inline constexpr bool hasPreimage() { return false; } + static constexpr bool hasPreimage() { return false; } using FactorTransform = ToStartOfYearImpl; }; @@ -929,23 +929,23 @@ struct ToHourImpl { static constexpr auto name = "toHour"; - static inline UInt8 execute(Int64 t, const DateLUTImpl & time_zone) + static UInt8 execute(Int64 t, const DateLUTImpl & time_zone) { return time_zone.toHour(t); } - static inline UInt8 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt8 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toHour(t); } - static inline UInt8 execute(Int32, const DateLUTImpl &) + static UInt8 execute(Int32, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline UInt8 execute(UInt16, const DateLUTImpl &) + static UInt8 execute(UInt16, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline constexpr bool hasPreimage() { return false; } + static constexpr bool hasPreimage() { return false; } using FactorTransform = ToDateImpl; }; @@ -954,27 +954,27 @@ struct TimezoneOffsetImpl { static constexpr auto name = "timezoneOffset"; - static inline time_t execute(Int64 t, const DateLUTImpl & time_zone) + static time_t execute(Int64 t, const DateLUTImpl & time_zone) { return time_zone.timezoneOffset(t); } - static inline time_t execute(UInt32 t, const DateLUTImpl & time_zone) + static time_t execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.timezoneOffset(t); } - static inline time_t execute(Int32, const DateLUTImpl &) + static time_t execute(Int32, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline time_t execute(UInt16, const DateLUTImpl &) + static time_t execute(UInt16, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline constexpr bool hasPreimage() { return false; } + static constexpr bool hasPreimage() { return false; } using FactorTransform = ToTimeImpl; }; @@ -982,23 +982,23 @@ struct ToMinuteImpl { static constexpr auto name = "toMinute"; - static inline UInt8 execute(Int64 t, const DateLUTImpl & time_zone) + static UInt8 execute(Int64 t, const DateLUTImpl & time_zone) { return time_zone.toMinute(t); } - static inline UInt8 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt8 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toMinute(t); } - static inline UInt8 execute(Int32, const DateLUTImpl &) + static UInt8 execute(Int32, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline UInt8 execute(UInt16, const DateLUTImpl &) + static UInt8 execute(UInt16, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline constexpr bool hasPreimage() { return false; } + static constexpr bool hasPreimage() { return false; } using FactorTransform = ToStartOfHourImpl; }; @@ -1007,23 +1007,23 @@ struct ToSecondImpl { static constexpr auto name = "toSecond"; - static inline UInt8 execute(Int64 t, const DateLUTImpl & time_zone) + static UInt8 execute(Int64 t, const DateLUTImpl & time_zone) { return time_zone.toSecond(t); } - static inline UInt8 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt8 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toSecond(t); } - static inline UInt8 execute(Int32, const DateLUTImpl &) + static UInt8 execute(Int32, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline UInt8 execute(UInt16, const DateLUTImpl &) + static UInt8 execute(UInt16, const DateLUTImpl &) { - throwDateTimeIsNotSupported(name); + throwDateIsNotSupported(name); } - static inline constexpr bool hasPreimage() { return false; } + static constexpr bool hasPreimage() { return false; } using FactorTransform = ToStartOfMinuteImpl; }; @@ -1032,23 +1032,23 @@ struct ToISOYearImpl { static constexpr auto name = "toISOYear"; - static inline UInt16 execute(Int64 t, const DateLUTImpl & time_zone) + static UInt16 execute(Int64 t, const DateLUTImpl & time_zone) { return time_zone.toISOYear(time_zone.toDayNum(t)); } - static inline UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toISOYear(time_zone.toDayNum(t)); } - static inline UInt16 execute(Int32 d, const DateLUTImpl & time_zone) + static UInt16 execute(Int32 d, const DateLUTImpl & time_zone) { return time_zone.toISOYear(ExtendedDayNum(d)); } - static inline UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) + static UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) { return time_zone.toISOYear(DayNum(d)); } - static inline constexpr bool hasPreimage() { return false; } + static constexpr bool hasPreimage() { return false; } using FactorTransform = ZeroTransform; }; @@ -1057,27 +1057,27 @@ struct ToStartOfISOYearImpl { static constexpr auto name = "toStartOfISOYear"; - static inline UInt16 execute(Int64 t, const DateLUTImpl & time_zone) + static UInt16 execute(Int64 t, const DateLUTImpl & time_zone) { return t < 0 ? 0 : time_zone.toFirstDayNumOfISOYear(ExtendedDayNum(std::min(Int32(time_zone.toDayNum(t)), Int32(DATE_LUT_MAX_DAY_NUM)))); } - static inline UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toFirstDayNumOfISOYear(time_zone.toDayNum(t)); } - static inline UInt16 execute(Int32 d, const DateLUTImpl & time_zone) + static UInt16 execute(Int32 d, const DateLUTImpl & time_zone) { return d < 0 ? 0 : time_zone.toFirstDayNumOfISOYear(ExtendedDayNum(std::min(d, Int32(DATE_LUT_MAX_DAY_NUM)))); } - static inline UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) + static UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) { return time_zone.toFirstDayNumOfISOYear(DayNum(d)); } - static inline Int64 executeExtendedResult(Int64 t, const DateLUTImpl & time_zone) + static Int64 executeExtendedResult(Int64 t, const DateLUTImpl & time_zone) { return time_zone.toFirstDayNumOfISOYear(time_zone.toDayNum(t)); } - static inline Int32 executeExtendedResult(Int32 d, const DateLUTImpl & time_zone) + static Int32 executeExtendedResult(Int32 d, const DateLUTImpl & time_zone) { return time_zone.toFirstDayNumOfISOYear(ExtendedDayNum(d)); } @@ -1089,23 +1089,23 @@ struct ToISOWeekImpl { static constexpr auto name = "toISOWeek"; - static inline UInt8 execute(Int64 t, const DateLUTImpl & time_zone) + static UInt8 execute(Int64 t, const DateLUTImpl & time_zone) { return time_zone.toISOWeek(time_zone.toDayNum(t)); } - static inline UInt8 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt8 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toISOWeek(time_zone.toDayNum(t)); } - static inline UInt8 execute(Int32 d, const DateLUTImpl & time_zone) + static UInt8 execute(Int32 d, const DateLUTImpl & time_zone) { return time_zone.toISOWeek(ExtendedDayNum(d)); } - static inline UInt8 execute(UInt16 d, const DateLUTImpl & time_zone) + static UInt8 execute(UInt16 d, const DateLUTImpl & time_zone) { return time_zone.toISOWeek(DayNum(d)); } - static inline constexpr bool hasPreimage() { return false; } + static constexpr bool hasPreimage() { return false; } using FactorTransform = ToISOYearImpl; }; @@ -1126,29 +1126,29 @@ struct ToRelativeYearNumImpl { static constexpr auto name = "toRelativeYearNum"; - static inline auto execute(Int64 t, const DateLUTImpl & time_zone) + static auto execute(Int64 t, const DateLUTImpl & time_zone) { if constexpr (precision_ == ResultPrecision::Extended) return static_cast(time_zone.toYear(t)); else return static_cast(time_zone.toYear(t)); } - static inline UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toYear(static_cast(t)); } - static inline auto execute(Int32 d, const DateLUTImpl & time_zone) + static auto execute(Int32 d, const DateLUTImpl & time_zone) { if constexpr (precision_ == ResultPrecision::Extended) return static_cast(time_zone.toYear(ExtendedDayNum(d))); else return static_cast(time_zone.toYear(ExtendedDayNum(d))); } - static inline UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) + static UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) { return time_zone.toYear(DayNum(d)); } - static inline constexpr bool hasPreimage() { return false; } + static constexpr bool hasPreimage() { return false; } using FactorTransform = ZeroTransform; }; @@ -1158,29 +1158,29 @@ struct ToRelativeQuarterNumImpl { static constexpr auto name = "toRelativeQuarterNum"; - static inline auto execute(Int64 t, const DateLUTImpl & time_zone) + static auto execute(Int64 t, const DateLUTImpl & time_zone) { if constexpr (precision_ == ResultPrecision::Extended) return static_cast(time_zone.toRelativeQuarterNum(t)); else return static_cast(time_zone.toRelativeQuarterNum(t)); } - static inline UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toRelativeQuarterNum(static_cast(t)); } - static inline auto execute(Int32 d, const DateLUTImpl & time_zone) + static auto execute(Int32 d, const DateLUTImpl & time_zone) { if constexpr (precision_ == ResultPrecision::Extended) return static_cast(time_zone.toRelativeQuarterNum(ExtendedDayNum(d))); else return static_cast(time_zone.toRelativeQuarterNum(ExtendedDayNum(d))); } - static inline UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) + static UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) { return time_zone.toRelativeQuarterNum(DayNum(d)); } - static inline constexpr bool hasPreimage() { return false; } + static constexpr bool hasPreimage() { return false; } using FactorTransform = ZeroTransform; }; @@ -1190,29 +1190,29 @@ struct ToRelativeMonthNumImpl { static constexpr auto name = "toRelativeMonthNum"; - static inline auto execute(Int64 t, const DateLUTImpl & time_zone) + static auto execute(Int64 t, const DateLUTImpl & time_zone) { if constexpr (precision_ == ResultPrecision::Extended) return static_cast(time_zone.toRelativeMonthNum(t)); else return static_cast(time_zone.toRelativeMonthNum(t)); } - static inline UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toRelativeMonthNum(static_cast(t)); } - static inline auto execute(Int32 d, const DateLUTImpl & time_zone) + static auto execute(Int32 d, const DateLUTImpl & time_zone) { if constexpr (precision_ == ResultPrecision::Extended) return static_cast(time_zone.toRelativeMonthNum(ExtendedDayNum(d))); else return static_cast(time_zone.toRelativeMonthNum(ExtendedDayNum(d))); } - static inline UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) + static UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) { return time_zone.toRelativeMonthNum(DayNum(d)); } - static inline constexpr bool hasPreimage() { return false; } + static constexpr bool hasPreimage() { return false; } using FactorTransform = ZeroTransform; }; @@ -1222,29 +1222,29 @@ struct ToRelativeWeekNumImpl { static constexpr auto name = "toRelativeWeekNum"; - static inline auto execute(Int64 t, const DateLUTImpl & time_zone) + static auto execute(Int64 t, const DateLUTImpl & time_zone) { if constexpr (precision_ == ResultPrecision::Extended) return static_cast(time_zone.toRelativeWeekNum(t)); else return static_cast(time_zone.toRelativeWeekNum(t)); } - static inline UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toRelativeWeekNum(static_cast(t)); } - static inline auto execute(Int32 d, const DateLUTImpl & time_zone) + static auto execute(Int32 d, const DateLUTImpl & time_zone) { if constexpr (precision_ == ResultPrecision::Extended) return static_cast(time_zone.toRelativeWeekNum(ExtendedDayNum(d))); else return static_cast(time_zone.toRelativeWeekNum(ExtendedDayNum(d))); } - static inline UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) + static UInt16 execute(UInt16 d, const DateLUTImpl & time_zone) { return time_zone.toRelativeWeekNum(DayNum(d)); } - static inline constexpr bool hasPreimage() { return false; } + static constexpr bool hasPreimage() { return false; } using FactorTransform = ZeroTransform; }; @@ -1254,29 +1254,29 @@ struct ToRelativeDayNumImpl { static constexpr auto name = "toRelativeDayNum"; - static inline auto execute(Int64 t, const DateLUTImpl & time_zone) + static auto execute(Int64 t, const DateLUTImpl & time_zone) { if constexpr (precision_ == ResultPrecision::Extended) return static_cast(time_zone.toDayNum(t)); else return static_cast(time_zone.toDayNum(t)); } - static inline UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt16 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toDayNum(static_cast(t)); } - static inline auto execute(Int32 d, const DateLUTImpl &) + static auto execute(Int32 d, const DateLUTImpl &) { if constexpr (precision_ == ResultPrecision::Extended) return static_cast(static_cast(d)); else return static_cast(static_cast(d)); } - static inline UInt16 execute(UInt16 d, const DateLUTImpl &) + static UInt16 execute(UInt16 d, const DateLUTImpl &) { return static_cast(d); } - static inline constexpr bool hasPreimage() { return false; } + static constexpr bool hasPreimage() { return false; } using FactorTransform = ZeroTransform; }; @@ -1286,35 +1286,35 @@ struct ToRelativeHourNumImpl { static constexpr auto name = "toRelativeHourNum"; - static inline auto execute(Int64 t, const DateLUTImpl & time_zone) + static auto execute(Int64 t, const DateLUTImpl & time_zone) { if constexpr (precision_ == ResultPrecision::Extended) return static_cast(time_zone.toStableRelativeHourNum(t)); else return static_cast(time_zone.toRelativeHourNum(t)); } - static inline UInt32 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt32 execute(UInt32 t, const DateLUTImpl & time_zone) { if constexpr (precision_ == ResultPrecision::Extended) return static_cast(time_zone.toStableRelativeHourNum(static_cast(t))); else return static_cast(time_zone.toRelativeHourNum(static_cast(t))); } - static inline auto execute(Int32 d, const DateLUTImpl & time_zone) + static auto execute(Int32 d, const DateLUTImpl & time_zone) { if constexpr (precision_ == ResultPrecision::Extended) return static_cast(time_zone.toStableRelativeHourNum(ExtendedDayNum(d))); else return static_cast(time_zone.toRelativeHourNum(ExtendedDayNum(d))); } - static inline UInt32 execute(UInt16 d, const DateLUTImpl & time_zone) + static UInt32 execute(UInt16 d, const DateLUTImpl & time_zone) { if constexpr (precision_ == ResultPrecision::Extended) return static_cast(time_zone.toStableRelativeHourNum(DayNum(d))); else return static_cast(time_zone.toRelativeHourNum(DayNum(d))); } - static inline constexpr bool hasPreimage() { return false; } + static constexpr bool hasPreimage() { return false; } using FactorTransform = ZeroTransform; }; @@ -1324,29 +1324,29 @@ struct ToRelativeMinuteNumImpl { static constexpr auto name = "toRelativeMinuteNum"; - static inline auto execute(Int64 t, const DateLUTImpl & time_zone) + static auto execute(Int64 t, const DateLUTImpl & time_zone) { if constexpr (precision_ == ResultPrecision::Extended) return static_cast(time_zone.toRelativeMinuteNum(t)); else return static_cast(time_zone.toRelativeMinuteNum(t)); } - static inline UInt32 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt32 execute(UInt32 t, const DateLUTImpl & time_zone) { return static_cast(time_zone.toRelativeMinuteNum(static_cast(t))); } - static inline auto execute(Int32 d, const DateLUTImpl & time_zone) + static auto execute(Int32 d, const DateLUTImpl & time_zone) { if constexpr (precision_ == ResultPrecision::Extended) return static_cast(time_zone.toRelativeMinuteNum(ExtendedDayNum(d))); else return static_cast(time_zone.toRelativeMinuteNum(ExtendedDayNum(d))); } - static inline UInt32 execute(UInt16 d, const DateLUTImpl & time_zone) + static UInt32 execute(UInt16 d, const DateLUTImpl & time_zone) { return static_cast(time_zone.toRelativeMinuteNum(DayNum(d))); } - static inline constexpr bool hasPreimage() { return false; } + static constexpr bool hasPreimage() { return false; } using FactorTransform = ZeroTransform; }; @@ -1356,26 +1356,26 @@ struct ToRelativeSecondNumImpl { static constexpr auto name = "toRelativeSecondNum"; - static inline Int64 execute(Int64 t, const DateLUTImpl &) + static Int64 execute(Int64 t, const DateLUTImpl &) { return t; } - static inline UInt32 execute(UInt32 t, const DateLUTImpl &) + static UInt32 execute(UInt32 t, const DateLUTImpl &) { return t; } - static inline auto execute(Int32 d, const DateLUTImpl & time_zone) + static auto execute(Int32 d, const DateLUTImpl & time_zone) { if constexpr (precision_ == ResultPrecision::Extended) return static_cast(time_zone.fromDayNum(ExtendedDayNum(d))); else return static_cast(time_zone.fromDayNum(ExtendedDayNum(d))); } - static inline UInt32 execute(UInt16 d, const DateLUTImpl & time_zone) + static UInt32 execute(UInt16 d, const DateLUTImpl & time_zone) { return static_cast(time_zone.fromDayNum(DayNum(d))); } - static inline constexpr bool hasPreimage() { return false; } + static constexpr bool hasPreimage() { return false; } using FactorTransform = ZeroTransform; }; @@ -1385,7 +1385,7 @@ struct ToRelativeSubsecondNumImpl { static constexpr auto name = "toRelativeSubsecondNumImpl"; - static inline Int64 execute(const DateTime64 & t, DateTime64::NativeType scale, const DateLUTImpl &) + static Int64 execute(const DateTime64 & t, DateTime64::NativeType scale, const DateLUTImpl &) { static_assert(scale_multiplier == 1000 || scale_multiplier == 1000000); if (scale == scale_multiplier) @@ -1394,15 +1394,15 @@ struct ToRelativeSubsecondNumImpl return t.value / (scale / scale_multiplier); return t.value * (scale_multiplier / scale); } - static inline Int64 execute(UInt32 t, const DateLUTImpl &) + static Int64 execute(UInt32 t, const DateLUTImpl &) { return t * scale_multiplier; } - static inline Int64 execute(Int32 d, const DateLUTImpl & time_zone) + static Int64 execute(Int32 d, const DateLUTImpl & time_zone) { return static_cast(time_zone.fromDayNum(ExtendedDayNum(d))) * scale_multiplier; } - static inline Int64 execute(UInt16 d, const DateLUTImpl & time_zone) + static Int64 execute(UInt16 d, const DateLUTImpl & time_zone) { return static_cast(time_zone.fromDayNum(DayNum(d)) * scale_multiplier); } @@ -1414,25 +1414,25 @@ struct ToYYYYMMImpl { static constexpr auto name = "toYYYYMM"; - static inline UInt32 execute(Int64 t, const DateLUTImpl & time_zone) + static UInt32 execute(Int64 t, const DateLUTImpl & time_zone) { return time_zone.toNumYYYYMM(t); } - static inline UInt32 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt32 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toNumYYYYMM(t); } - static inline UInt32 execute(Int32 d, const DateLUTImpl & time_zone) + static UInt32 execute(Int32 d, const DateLUTImpl & time_zone) { return time_zone.toNumYYYYMM(ExtendedDayNum(d)); } - static inline UInt32 execute(UInt16 d, const DateLUTImpl & time_zone) + static UInt32 execute(UInt16 d, const DateLUTImpl & time_zone) { return time_zone.toNumYYYYMM(DayNum(d)); } - static inline constexpr bool hasPreimage() { return true; } + static constexpr bool hasPreimage() { return true; } - static inline RangeOrNull getPreimage(const IDataType & type, const Field & point) + static RangeOrNull getPreimage(const IDataType & type, const Field & point) { if (point.getType() != Field::Types::UInt64) return std::nullopt; @@ -1463,23 +1463,23 @@ struct ToYYYYMMDDImpl { static constexpr auto name = "toYYYYMMDD"; - static inline UInt32 execute(Int64 t, const DateLUTImpl & time_zone) + static UInt32 execute(Int64 t, const DateLUTImpl & time_zone) { return time_zone.toNumYYYYMMDD(t); } - static inline UInt32 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt32 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toNumYYYYMMDD(t); } - static inline UInt32 execute(Int32 d, const DateLUTImpl & time_zone) + static UInt32 execute(Int32 d, const DateLUTImpl & time_zone) { return time_zone.toNumYYYYMMDD(ExtendedDayNum(d)); } - static inline UInt32 execute(UInt16 d, const DateLUTImpl & time_zone) + static UInt32 execute(UInt16 d, const DateLUTImpl & time_zone) { return time_zone.toNumYYYYMMDD(DayNum(d)); } - static inline constexpr bool hasPreimage() { return false; } + static constexpr bool hasPreimage() { return false; } using FactorTransform = ZeroTransform; }; @@ -1488,23 +1488,23 @@ struct ToYYYYMMDDhhmmssImpl { static constexpr auto name = "toYYYYMMDDhhmmss"; - static inline UInt64 execute(Int64 t, const DateLUTImpl & time_zone) + static UInt64 execute(Int64 t, const DateLUTImpl & time_zone) { return time_zone.toNumYYYYMMDDhhmmss(t); } - static inline UInt64 execute(UInt32 t, const DateLUTImpl & time_zone) + static UInt64 execute(UInt32 t, const DateLUTImpl & time_zone) { return time_zone.toNumYYYYMMDDhhmmss(t); } - static inline UInt64 execute(Int32 d, const DateLUTImpl & time_zone) + static UInt64 execute(Int32 d, const DateLUTImpl & time_zone) { return time_zone.toNumYYYYMMDDhhmmss(time_zone.toDate(ExtendedDayNum(d))); } - static inline UInt64 execute(UInt16 d, const DateLUTImpl & time_zone) + static UInt64 execute(UInt16 d, const DateLUTImpl & time_zone) { return time_zone.toNumYYYYMMDDhhmmss(time_zone.toDate(DayNum(d))); } - static inline constexpr bool hasPreimage() { return false; } + static constexpr bool hasPreimage() { return false; } using FactorTransform = ZeroTransform; }; @@ -1519,7 +1519,7 @@ struct ToDateTimeComponentsImpl { static constexpr auto name = "toDateTimeComponents"; - static inline DateTimeComponentsWithFractionalPart execute(const DateTime64 & t, DateTime64::NativeType scale_multiplier, const DateLUTImpl & time_zone) + static DateTimeComponentsWithFractionalPart execute(const DateTime64 & t, DateTime64::NativeType scale_multiplier, const DateLUTImpl & time_zone) { auto components = DecimalUtils::splitWithScaleMultiplier(t, scale_multiplier); @@ -1539,15 +1539,15 @@ struct ToDateTimeComponentsImpl UInt16 microsecond = static_cast(fractional % divider); return DateTimeComponentsWithFractionalPart{time_zone.toDateTimeComponents(components.whole), millisecond, microsecond}; } - static inline DateTimeComponentsWithFractionalPart execute(UInt32 t, const DateLUTImpl & time_zone) + static DateTimeComponentsWithFractionalPart execute(UInt32 t, const DateLUTImpl & time_zone) { return DateTimeComponentsWithFractionalPart{time_zone.toDateTimeComponents(static_cast(t)), 0, 0}; } - static inline DateTimeComponentsWithFractionalPart execute(Int32 d, const DateLUTImpl & time_zone) + static DateTimeComponentsWithFractionalPart execute(Int32 d, const DateLUTImpl & time_zone) { return DateTimeComponentsWithFractionalPart{time_zone.toDateTimeComponents(ExtendedDayNum(d)), 0, 0}; } - static inline DateTimeComponentsWithFractionalPart execute(UInt16 d, const DateLUTImpl & time_zone) + static DateTimeComponentsWithFractionalPart execute(UInt16 d, const DateLUTImpl & time_zone) { return DateTimeComponentsWithFractionalPart{time_zone.toDateTimeComponents(DayNum(d)), 0, 0}; } diff --git a/src/Functions/base64Decode.cpp b/src/Functions/base64Decode.cpp index 4060aafe1a3..e1a24756705 100644 --- a/src/Functions/base64Decode.cpp +++ b/src/Functions/base64Decode.cpp @@ -10,7 +10,7 @@ REGISTER_FUNCTION(Base64Decode) tb64ini(0, 0); factory.registerFunction>(); - /// MysQL compatibility alias. + /// MySQL compatibility alias. factory.registerAlias("FROM_BASE64", "base64Decode", FunctionFactory::CaseInsensitive); } } diff --git a/src/Functions/base64Encode.cpp b/src/Functions/base64Encode.cpp index 773db7e09d9..4c8674055ac 100644 --- a/src/Functions/base64Encode.cpp +++ b/src/Functions/base64Encode.cpp @@ -10,7 +10,7 @@ REGISTER_FUNCTION(Base64Encode) tb64ini(0, 0); factory.registerFunction>(); - /// MysQL compatibility alias. + /// MySQL compatibility alias. factory.registerAlias("TO_BASE64", "base64Encode", FunctionFactory::CaseInsensitive); } } diff --git a/src/Functions/toDayOfMonth.cpp b/src/Functions/toDayOfMonth.cpp index d7689ef00f2..c20b0b75797 100644 --- a/src/Functions/toDayOfMonth.cpp +++ b/src/Functions/toDayOfMonth.cpp @@ -13,7 +13,7 @@ REGISTER_FUNCTION(ToDayOfMonth) { factory.registerFunction(); - /// MysQL compatibility alias. + /// MySQL compatibility alias. factory.registerAlias("DAY", "toDayOfMonth", FunctionFactory::CaseInsensitive); factory.registerAlias("DAYOFMONTH", "toDayOfMonth", FunctionFactory::CaseInsensitive); } diff --git a/src/Functions/toDayOfWeek.cpp b/src/Functions/toDayOfWeek.cpp index 06343714b9d..dc508d70814 100644 --- a/src/Functions/toDayOfWeek.cpp +++ b/src/Functions/toDayOfWeek.cpp @@ -12,7 +12,7 @@ REGISTER_FUNCTION(ToDayOfWeek) { factory.registerFunction(); - /// MysQL compatibility alias. + /// MySQL compatibility alias. factory.registerAlias("DAYOFWEEK", "toDayOfWeek", FunctionFactory::CaseInsensitive); } diff --git a/src/Functions/toDayOfYear.cpp b/src/Functions/toDayOfYear.cpp index 8b03f1a4211..0cbafd6275a 100644 --- a/src/Functions/toDayOfYear.cpp +++ b/src/Functions/toDayOfYear.cpp @@ -13,7 +13,7 @@ REGISTER_FUNCTION(ToDayOfYear) { factory.registerFunction(); - /// MysQL compatibility alias. + /// MySQL compatibility alias. factory.registerAlias("DAYOFYEAR", "toDayOfYear", FunctionFactory::CaseInsensitive); } diff --git a/src/Functions/toHour.cpp b/src/Functions/toHour.cpp index a6a57946e33..fc9ec657adf 100644 --- a/src/Functions/toHour.cpp +++ b/src/Functions/toHour.cpp @@ -13,7 +13,7 @@ REGISTER_FUNCTION(ToHour) { factory.registerFunction(); - /// MysQL compatibility alias. + /// MySQL compatibility alias. factory.registerAlias("HOUR", "toHour", FunctionFactory::CaseInsensitive); } diff --git a/src/Functions/toMinute.cpp b/src/Functions/toMinute.cpp index 25939870554..162ecb282df 100644 --- a/src/Functions/toMinute.cpp +++ b/src/Functions/toMinute.cpp @@ -13,7 +13,7 @@ REGISTER_FUNCTION(ToMinute) { factory.registerFunction(); - /// MysQL compatibility alias. + /// MySQL compatibility alias. factory.registerAlias("MINUTE", "toMinute", FunctionFactory::CaseInsensitive); } diff --git a/src/Functions/toMonth.cpp b/src/Functions/toMonth.cpp index 783a1341e23..422f21e7df8 100644 --- a/src/Functions/toMonth.cpp +++ b/src/Functions/toMonth.cpp @@ -12,7 +12,7 @@ using FunctionToMonth = FunctionDateOrDateTimeToSomething(); - /// MysQL compatibility alias. + /// MySQL compatibility alias. factory.registerAlias("MONTH", "toMonth", FunctionFactory::CaseInsensitive); } diff --git a/src/Functions/toQuarter.cpp b/src/Functions/toQuarter.cpp index 2268b6402c6..3c301095ff2 100644 --- a/src/Functions/toQuarter.cpp +++ b/src/Functions/toQuarter.cpp @@ -12,7 +12,7 @@ using FunctionToQuarter = FunctionDateOrDateTimeToSomething(); - /// MysQL compatibility alias. + /// MySQL compatibility alias. factory.registerAlias("QUARTER", "toQuarter", FunctionFactory::CaseInsensitive); } diff --git a/src/Functions/toSecond.cpp b/src/Functions/toSecond.cpp index 2fd64912c0f..372097fd488 100644 --- a/src/Functions/toSecond.cpp +++ b/src/Functions/toSecond.cpp @@ -13,7 +13,7 @@ REGISTER_FUNCTION(ToSecond) { factory.registerFunction(); - /// MysQL compatibility alias. + /// MySQL compatibility alias. factory.registerAlias("SECOND", "toSecond", FunctionFactory::CaseInsensitive); } diff --git a/src/Functions/toYear.cpp b/src/Functions/toYear.cpp index 9cf2a260921..75479adb82c 100644 --- a/src/Functions/toYear.cpp +++ b/src/Functions/toYear.cpp @@ -13,7 +13,7 @@ REGISTER_FUNCTION(ToYear) { factory.registerFunction(); - /// MysQL compatibility alias. + /// MySQL compatibility alias. factory.registerAlias("YEAR", "toYear", FunctionFactory::CaseInsensitive); } From 1ce84774f0e829742827c4a420c310c56637c859 Mon Sep 17 00:00:00 2001 From: Robert Schulze Date: Fri, 8 Sep 2023 13:27:24 +0000 Subject: [PATCH 256/433] Implement toDaysSinceYearZero() / to_days() (MySQL alias) Fixes: #54277 --- .../functions/date-time-functions.md | 36 +++++++++ src/Functions/DateTimeTransforms.h | 25 +++++++ src/Functions/toDaysSinceYearZero.cpp | 73 +++++++++++++++++++ .../02874_daysSinceYearZero.reference | 13 ++++ .../0_stateless/02874_daysSinceYearZero.sql | 22 ++++++ 5 files changed, 169 insertions(+) create mode 100644 src/Functions/toDaysSinceYearZero.cpp create mode 100644 tests/queries/0_stateless/02874_daysSinceYearZero.reference create mode 100644 tests/queries/0_stateless/02874_daysSinceYearZero.sql diff --git a/docs/en/sql-reference/functions/date-time-functions.md b/docs/en/sql-reference/functions/date-time-functions.md index 6bcc9d654c5..ef0ccf84f73 100644 --- a/docs/en/sql-reference/functions/date-time-functions.md +++ b/docs/en/sql-reference/functions/date-time-functions.md @@ -720,6 +720,42 @@ SELECT toDate('2016-12-27') AS date, toYearWeek(date) AS yearWeek0, toYearWeek(d └────────────┴───────────┴───────────┴───────────┴───────────────┘ ``` +## toDaysSinceYearZero + +Returns for a given date, the number of days passed since the [year 0](https://en.wikipedia.org/wiki/Year_zero). + +**Syntax** + +``` sql +toDaysSinceYearZero(date) +``` + +Aliases: `TO_DAYS` + +**Arguments** + +- `date` — The date to calculate the number of days passed since year zero from. [Date](../../sql-reference/data-types/date.md) or [Date64](../../sql-reference/data-types/date64.md). + +**Returned value** + +The number of days passed since date 0000-01-01. + +Type: [UInt32](../../sql-reference/data-types/int-uint.md). + +**Example** + +``` sql +SELECT toDaysSinceYearZero(toDate('2023-09-08')); +``` + +Result: + +``` text +┌─toDaysSinceYearZero(toDate('2023-09-08')))─┐ +│ 738772 │ +└────────────────────────────────────────────┘ +``` + ## age Returns the `unit` component of the difference between `startdate` and `enddate`. The difference is calculated using a precision of 1 microsecond. diff --git a/src/Functions/DateTimeTransforms.h b/src/Functions/DateTimeTransforms.h index 0fb4edc1375..16cfba69275 100644 --- a/src/Functions/DateTimeTransforms.h +++ b/src/Functions/DateTimeTransforms.h @@ -925,6 +925,31 @@ struct ToDayOfYearImpl using FactorTransform = ToStartOfYearImpl; }; +struct ToDaysSinceYearZeroImpl +{ + static constexpr auto name = "toDaysSinceYearZero"; + + static UInt32 execute(Int64, const DateLUTImpl &) + { + throwDateTimeIsNotSupported(name); + } + static UInt32 execute(UInt32, const DateLUTImpl &) + { + throwDateTimeIsNotSupported(name); + } + static UInt32 execute(Int32 d, const DateLUTImpl &) + { + return /* days between 0000-01-01 and 1970-01-01 */ 719'164 + d; + } + static UInt32 execute(UInt16 d, const DateLUTImpl &) + { + return /* days between 0000-01-01 and 1970-01-01 */ 719'164 + d; + } + static constexpr bool hasPreimage() { return false; } + + using FactorTransform = ZeroTransform; +}; + struct ToHourImpl { static constexpr auto name = "toHour"; diff --git a/src/Functions/toDaysSinceYearZero.cpp b/src/Functions/toDaysSinceYearZero.cpp new file mode 100644 index 00000000000..c4233416195 --- /dev/null +++ b/src/Functions/toDaysSinceYearZero.cpp @@ -0,0 +1,73 @@ +#include +#include +#include +#include +#include +#include +#include +#include + + +namespace DB +{ + +namespace ErrorCodes +{ + extern const int ILLEGAL_TYPE_OF_ARGUMENT; +} + +namespace +{ + +/** Returns number of days passed since 0001-01-01 */ +class FunctionToDaysSinceYearZero : public IFunction +{ + using ResultType = DataTypeUInt32; +public: + static constexpr auto name = "toDaysSinceYearZero"; + static FunctionPtr create(ContextPtr context) { return std::make_shared(context); } + + explicit FunctionToDaysSinceYearZero(ContextPtr /*context*/) {} + + String getName() const override { return name; } + size_t getNumberOfArguments() const override { return 1; } + bool isSuitableForShortCircuitArgumentsExecution(const DataTypesWithConstInfo & /*arguments*/) const override { return true; } + bool useDefaultImplementationForConstants() const override { return true; } + + DataTypePtr getReturnTypeImpl(const ColumnsWithTypeAndName & arguments) const override + { + FunctionArgumentDescriptors mandatory_args{ + {"date", &isDateOrDate32, nullptr, "Date or Date32"} + }; + + validateFunctionArgumentTypes(*this, arguments, mandatory_args); + + return std::make_shared(); + } + + ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override + { + const IDataType * from_type = arguments[0].type.get(); + WhichDataType which(from_type); + + if (which.isDate()) + return DateTimeTransformImpl::execute(arguments, result_type, input_rows_count); + else if (which.isDate32()) + return DateTimeTransformImpl::execute(arguments, result_type, input_rows_count); + + throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, + "Illegal type {} of argument of function {}", + arguments[0].type->getName(), this->getName()); + } +}; + +} + +REGISTER_FUNCTION(ToDaysSinceYearZero) +{ + factory.registerFunction(); + /// MySQL compatibility alias. + factory.registerAlias("TO_DAYS", FunctionToDaysSinceYearZero::name, FunctionFactory::CaseInsensitive); +} + +} diff --git a/tests/queries/0_stateless/02874_daysSinceYearZero.reference b/tests/queries/0_stateless/02874_daysSinceYearZero.reference new file mode 100644 index 00000000000..9bfb9b78910 --- /dev/null +++ b/tests/queries/0_stateless/02874_daysSinceYearZero.reference @@ -0,0 +1,13 @@ +Reject invalid parameters +Const argument +719164 +738772 +693597 +738772 +\N +Non-const argument +738772 +738772 +MySQL alias +738772 +738772 diff --git a/tests/queries/0_stateless/02874_daysSinceYearZero.sql b/tests/queries/0_stateless/02874_daysSinceYearZero.sql new file mode 100644 index 00000000000..0db1a305ea9 --- /dev/null +++ b/tests/queries/0_stateless/02874_daysSinceYearZero.sql @@ -0,0 +1,22 @@ +SELECT 'Reject invalid parameters'; +SELECT daysSinceYearZero(); -- { serverError NUMBER_OF_ARGUMENTS_DOESNT_MATCH } +SELECT daysSinceYearZero(toDate('2023-09-08'), toDate('2023-09-08')); -- { serverError NUMBER_OF_ARGUMENTS_DOESNT_MATCH } +SELECT daysSinceYearZero('str'); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT daysSinceYearZero(42); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT daysSinceYearZero(toDateTime('2023-09-08 11:11:11')); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT daysSinceYearZero(toDateTime64('2023-09-08 11:11:11', 3)); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } + +SELECT 'Const argument'; +SELECT daysSinceYearZero(toDate('1970-01-01')); +SELECT daysSinceYearZero(toDate('2023-09-08')); +SELECT daysSinceYearZero(toDate32('1900-01-01')); +SELECT daysSinceYearZero(toDate32('2023-09-08')); +SELECT daysSinceYearZero(NULL); + +SELECT 'Non-const argument'; +SELECT daysSinceYearZero(materialize(toDate('2023-09-08'))); +SELECT daysSinceYearZero(materialize(toDate32('2023-09-08'))); + +SELECT 'MySQL alias'; +SELECT to_days(toDate('2023-09-08')); +SELECT TO_DAYS(toDate('2023-09-08')); From fb0f9ff5650c19ec2027e82ea98bb0ede70b8227 Mon Sep 17 00:00:00 2001 From: Amos Bird Date: Sun, 10 Sep 2023 02:38:12 +0800 Subject: [PATCH 257/433] Fix aggregate projections with normalized states --- src/Columns/ColumnAggregateFunction.cpp | 2 +- src/Columns/ColumnAggregateFunction.h | 2 +- src/Functions/FunctionsConversion.h | 39 ++++++++++++++++--- .../optimizeUseAggregateProjection.cpp | 17 +++++--- ...rojection_with_normalized_states.reference | 1 + ...gate_projection_with_normalized_states.sql | 29 ++++++++++++++ 6 files changed, 76 insertions(+), 14 deletions(-) create mode 100644 tests/queries/0_stateless/01710_aggregate_projection_with_normalized_states.reference create mode 100644 tests/queries/0_stateless/01710_aggregate_projection_with_normalized_states.sql diff --git a/src/Columns/ColumnAggregateFunction.cpp b/src/Columns/ColumnAggregateFunction.cpp index 3ebb30df87e..0ec5db6c69d 100644 --- a/src/Columns/ColumnAggregateFunction.cpp +++ b/src/Columns/ColumnAggregateFunction.cpp @@ -73,7 +73,7 @@ ColumnAggregateFunction::ColumnAggregateFunction(const AggregateFunctionPtr & fu } -void ColumnAggregateFunction::set(const AggregateFunctionPtr & func_, size_t version_) +void ColumnAggregateFunction::set(const AggregateFunctionPtr & func_, std::optional version_) { func = func_; version = version_; diff --git a/src/Columns/ColumnAggregateFunction.h b/src/Columns/ColumnAggregateFunction.h index 7c7201e585a..e37a8a5f109 100644 --- a/src/Columns/ColumnAggregateFunction.h +++ b/src/Columns/ColumnAggregateFunction.h @@ -103,7 +103,7 @@ private: public: ~ColumnAggregateFunction() override; - void set(const AggregateFunctionPtr & func_, size_t version_); + void set(const AggregateFunctionPtr & func_, std::optional version_ = std::nullopt); AggregateFunctionPtr getAggregateFunction() { return func; } AggregateFunctionPtr getAggregateFunction() const { return func; } diff --git a/src/Functions/FunctionsConversion.h b/src/Functions/FunctionsConversion.h index 00e2ebcda43..535723bce4e 100644 --- a/src/Functions/FunctionsConversion.h +++ b/src/Functions/FunctionsConversion.h @@ -33,6 +33,7 @@ #include #include #include +#include #include #include #include @@ -3188,14 +3189,40 @@ private: { return &ConvertImplGenericFromString::execute; } - else + else if (const auto * agg_type = checkAndGetDataType(from_type_untyped.get())) { - if (cast_type == CastType::accurateOrNull) - return createToNullableColumnWrapper(); - else - throw Exception(ErrorCodes::CANNOT_CONVERT_TYPE, "Conversion from {} to {} is not supported", - from_type_untyped->getName(), to_type->getName()); + if (agg_type->getFunction()->haveSameStateRepresentation(*to_type->getFunction())) + { + return [agg_type]( + ColumnsWithTypeAndName & arguments, + const DataTypePtr & /* result_type */, + const ColumnNullable * /* nullable_source */, + size_t /*input_rows_count*/) -> ColumnPtr + { + const auto & argument_column = arguments.front(); + const auto * col_agg = checkAndGetColumn(argument_column.column.get()); + if (col_agg) + { + auto new_col_agg = ColumnAggregateFunction::create(*col_agg); + new_col_agg->set(agg_type->getFunction()); + return new_col_agg; + } + else + { + throw Exception( + ErrorCodes::LOGICAL_ERROR, + "Illegal column {} for function CAST AS AggregateFunction", + argument_column.column->getName()); + } + }; + } } + + if (cast_type == CastType::accurateOrNull) + return createToNullableColumnWrapper(); + else + throw Exception(ErrorCodes::CANNOT_CONVERT_TYPE, "Conversion from {} to {} is not supported", + from_type_untyped->getName(), to_type->getName()); } WrapperType createArrayWrapper(const DataTypePtr & from_type_untyped, const DataTypeArray & to_type) const diff --git a/src/Processors/QueryPlan/Optimizations/optimizeUseAggregateProjection.cpp b/src/Processors/QueryPlan/Optimizations/optimizeUseAggregateProjection.cpp index eab4d3f5d43..0b24d2adbf9 100644 --- a/src/Processors/QueryPlan/Optimizations/optimizeUseAggregateProjection.cpp +++ b/src/Processors/QueryPlan/Optimizations/optimizeUseAggregateProjection.cpp @@ -143,12 +143,12 @@ std::optional matchAggregateFunctions( argument_types.clear(); const auto & candidate = info.aggregates[idx]; - /// Note: this check is a bit strict. - /// We check that aggregate function names, argument types and parameters are equal. /// In some cases it's possible only to check that states are equal, /// e.g. for quantile(0.3)(...) and quantile(0.5)(...). - /// But also functions sum(...) and sumIf(...) will have equal states, - /// and we can't replace one to another from projection. + /// + /// Note we already checked that aggregate function names are equal, + /// so that functions sum(...) and sumIf(...) with equal states will + /// not match. if (!candidate.function->getStateType()->equals(*aggregate.function->getStateType())) { // LOG_TRACE(&Poco::Logger::get("optimizeUseProjections"), "Cannot match agg func {} vs {} by state {} vs {}", @@ -249,12 +249,17 @@ static void appendAggregateFunctions( auto & input = inputs[match.description]; if (!input) - input = &proj_dag.addInput(match.description->column_name, std::move(type)); + input = &proj_dag.addInput(match.description->column_name, type); const auto * node = input; if (node->result_name != aggregate.column_name) - node = &proj_dag.addAlias(*node, aggregate.column_name); + { + /// Always cast to aggregate types specified in query, because input + /// columns from projection might have the same state but different + /// type, which can generate wrong results during finalization. + node = &proj_dag.addCast(*node, type, aggregate.column_name); + } proj_dag_outputs.push_back(node); } diff --git a/tests/queries/0_stateless/01710_aggregate_projection_with_normalized_states.reference b/tests/queries/0_stateless/01710_aggregate_projection_with_normalized_states.reference new file mode 100644 index 00000000000..9db43c50b71 --- /dev/null +++ b/tests/queries/0_stateless/01710_aggregate_projection_with_normalized_states.reference @@ -0,0 +1 @@ +950 990 500 2000 diff --git a/tests/queries/0_stateless/01710_aggregate_projection_with_normalized_states.sql b/tests/queries/0_stateless/01710_aggregate_projection_with_normalized_states.sql new file mode 100644 index 00000000000..f7be9239c11 --- /dev/null +++ b/tests/queries/0_stateless/01710_aggregate_projection_with_normalized_states.sql @@ -0,0 +1,29 @@ +DROP TABLE IF EXISTS r; + +CREATE TABLE r ( + x String, + a LowCardinality(String), + q AggregateFunction(quantilesTiming(0.5, 0.95, 0.99), Int64), + s Int64, + PROJECTION p + (SELECT a, quantilesTimingMerge(0.5, 0.95, 0.99)(q), sum(s) GROUP BY a) +) Engine=SummingMergeTree order by (x, a); + +insert into r +select number%100 x, + 'x' a, + quantilesTimingState(0.5, 0.95, 0.99)(number::Int64) q, + sum(1) s +from numbers(1000) +group by x,a; + +SELECT + ifNotFinite(quantilesTimingMerge(0.95)(q)[1],0) as d1, + ifNotFinite(quantilesTimingMerge(0.99)(q)[1],0) as d2, + ifNotFinite(quantilesTimingMerge(0.50)(q)[1],0) as d3, + sum(s) +FROM cluster('test_cluster_two_shards', currentDatabase(), r) +WHERE a = 'x' +settings prefer_localhost_replica=0; + +DROP TABLE r; From b610faf685d672517377c70f5efd65d077d6f193 Mon Sep 17 00:00:00 2001 From: "Mikhail f. Shiryaev" Date: Tue, 29 Aug 2023 18:58:56 +0200 Subject: [PATCH 258/433] Revert "Revert "Fixed wrong python test name pattern"" --- tests/integration/ci-runner.py | 53 ++++++++++------------------------ 1 file changed, 16 insertions(+), 37 deletions(-) diff --git a/tests/integration/ci-runner.py b/tests/integration/ci-runner.py index aa89ccf11b3..5c3a7695119 100755 --- a/tests/integration/ci-runner.py +++ b/tests/integration/ci-runner.py @@ -9,6 +9,7 @@ import os import random import re import shutil +import string import subprocess import time import shlex @@ -430,19 +431,12 @@ class ClickhouseIntegrationTestsRunner: def _get_all_tests(self, repo_path): image_cmd = self._get_runner_image_cmd(repo_path) - out_file = "all_tests.txt" + runner_opts = self._get_runner_opts() out_file_full = os.path.join(self.result_path, "runner_get_all_tests.log") cmd = ( - "cd {repo_path}/tests/integration && " - "timeout --signal=KILL 1h ./runner {runner_opts} {image_cmd} -- --setup-plan " - "| tee '{out_file_full}' | grep -F '::' | sed -r 's/ \(fixtures used:.*//g; s/^ *//g; s/ *$//g' " - "| grep -v -F 'SKIPPED' | sort --unique > {out_file}".format( - repo_path=repo_path, - runner_opts=self._get_runner_opts(), - image_cmd=image_cmd, - out_file=out_file, - out_file_full=out_file_full, - ) + f"cd {repo_path}/tests/integration && " + f"timeout --signal=KILL 1h ./runner {runner_opts} {image_cmd} -- --setup-plan " + f"| tee '{out_file_full}'" ) logging.info("Getting all tests with cmd '%s'", cmd) @@ -450,34 +444,19 @@ class ClickhouseIntegrationTestsRunner: cmd, shell=True ) - all_tests_file_path = "{repo_path}/tests/integration/{out_file}".format( - repo_path=repo_path, out_file=out_file - ) - if ( - not os.path.isfile(all_tests_file_path) - or os.path.getsize(all_tests_file_path) == 0 - ): - if os.path.isfile(out_file_full): - # log runner output - logging.info("runner output:") - with open(out_file_full, "r") as all_tests_full_file: - for line in all_tests_full_file: - line = line.rstrip() - if line: - logging.info("runner output: %s", line) - else: - logging.info("runner output '%s' is empty", out_file_full) + all_tests = set() + with open(out_file_full, "r", encoding="utf-8") as all_tests_fd: + for line in all_tests_fd: + if ( + line[0] in string.whitespace # test names at the start of lines + or "::test" not in line # test names contain '::test' + or "SKIPPED" in line # pytest.mark.skip/-if + ): + continue + all_tests.add(line.strip()) - raise Exception( - "There is something wrong with getting all tests list: file '{}' is empty or does not exist.".format( - all_tests_file_path - ) - ) + assert all_tests - all_tests = [] - with open(all_tests_file_path, "r") as all_tests_file: - for line in all_tests_file: - all_tests.append(line.strip()) return list(sorted(all_tests)) def _get_parallel_tests_skip_list(self, repo_path): From 5f0d94d5236ad441a61b69a7db36ba9bd77f61a2 Mon Sep 17 00:00:00 2001 From: "Mikhail f. Shiryaev" Date: Wed, 30 Aug 2023 09:31:26 +0200 Subject: [PATCH 259/433] Disable broken tests to have the bug fixed --- tests/integration/test_concurrent_backups_s3/test.py | 1 + 1 file changed, 1 insertion(+) diff --git a/tests/integration/test_concurrent_backups_s3/test.py b/tests/integration/test_concurrent_backups_s3/test.py index b29058865c0..312ebdba5e3 100644 --- a/tests/integration/test_concurrent_backups_s3/test.py +++ b/tests/integration/test_concurrent_backups_s3/test.py @@ -24,6 +24,7 @@ def start_cluster(): cluster.shutdown() +@pytest.mark.skip(reason="broken test") def test_concurrent_backups(start_cluster): node.query("DROP TABLE IF EXISTS s3_test SYNC") columns = [f"column_{i} UInt64" for i in range(1000)] From b17dc582a548daed6650b484df1224fcf3730684 Mon Sep 17 00:00:00 2001 From: "Mikhail f. Shiryaev" Date: Sat, 9 Sep 2023 20:13:32 +0200 Subject: [PATCH 260/433] Disable test_disk_over_web_server for analyzer --- tests/integration/test_disk_over_web_server/test.py | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/tests/integration/test_disk_over_web_server/test.py b/tests/integration/test_disk_over_web_server/test.py index bc60e5c865e..2edc05c4d7e 100644 --- a/tests/integration/test_disk_over_web_server/test.py +++ b/tests/integration/test_disk_over_web_server/test.py @@ -13,6 +13,7 @@ def cluster(): "node1", main_configs=["configs/storage_conf.xml"], with_nginx=True, + allow_analyzer=False, ) cluster.add_instance( "node2", @@ -20,12 +21,14 @@ def cluster(): with_nginx=True, stay_alive=True, with_zookeeper=True, + allow_analyzer=False, ) cluster.add_instance( "node3", main_configs=["configs/storage_conf_web.xml"], with_nginx=True, with_zookeeper=True, + allow_analyzer=False, ) cluster.add_instance( @@ -144,7 +147,6 @@ def test_usage(cluster, node_name): def test_incorrect_usage(cluster): - node1 = cluster.instances["node1"] node2 = cluster.instances["node3"] global uuids node2.query( @@ -186,7 +188,7 @@ def test_cache(cluster, node_name): (id Int32) ENGINE = MergeTree() ORDER BY id SETTINGS storage_policy = 'cached_web'; """.format( - i, uuids[i], i, i + i, uuids[i] ) ) From c7e058fa6e017a7539e702f87dd7b94d8c53c797 Mon Sep 17 00:00:00 2001 From: "Mikhail f. Shiryaev" Date: Sat, 9 Sep 2023 20:16:36 +0200 Subject: [PATCH 261/433] Disable broken tests --- tests/integration/test_concurrent_threads_soft_limit/test.py | 3 +++ 1 file changed, 3 insertions(+) diff --git a/tests/integration/test_concurrent_threads_soft_limit/test.py b/tests/integration/test_concurrent_threads_soft_limit/test.py index 2f76f44ddc2..d1e233ee12f 100644 --- a/tests/integration/test_concurrent_threads_soft_limit/test.py +++ b/tests/integration/test_concurrent_threads_soft_limit/test.py @@ -51,6 +51,7 @@ def test_concurrent_threads_soft_limit_default(started_cluster): ) +@pytest.mark.skip(reason="broken test") def test_concurrent_threads_soft_limit_defined_50(started_cluster): node2.query( "SELECT count(*) FROM numbers_mt(10000000)", @@ -65,6 +66,7 @@ def test_concurrent_threads_soft_limit_defined_50(started_cluster): ) +@pytest.mark.skip(reason="broken test") def test_concurrent_threads_soft_limit_defined_1(started_cluster): node3.query( "SELECT count(*) FROM numbers_mt(10000000)", @@ -82,6 +84,7 @@ def test_concurrent_threads_soft_limit_defined_1(started_cluster): # In config_limit_reached.xml there is concurrent_threads_soft_limit=10 # Background query starts in a separate thread to reach this limit. # When this limit is reached the foreground query gets less than 5 queries despite the fact that it has settings max_threads=5 +@pytest.mark.skip(reason="broken test") def test_concurrent_threads_soft_limit_limit_reached(started_cluster): def background_query(): try: From 6ec783a7467cc718649f6b740102f6f6b7ddd325 Mon Sep 17 00:00:00 2001 From: flynn Date: Sun, 10 Sep 2023 03:48:12 +0000 Subject: [PATCH 262/433] Refactor IHints --- src/Client/ClientBase.h | 2 +- src/Common/IFactoryWithAliases.h | 2 +- src/Common/NamePrompter.h | 2 +- src/Core/Settings.h | 2 +- src/DataTypes/EnumValues.h | 2 +- src/Functions/UserDefined/UserDefinedSQLFunctionFactory.h | 2 +- src/Interpreters/DatabaseCatalog.cpp | 2 +- src/Interpreters/DatabaseCatalog.h | 2 +- src/Server/HTTPPathHints.h | 2 +- src/Storages/ColumnsDescription.h | 2 +- src/Storages/IStorage.h | 2 +- src/Storages/IndicesDescription.h | 2 +- src/Storages/ProjectionsDescription.h | 2 +- src/Storages/StorageFactory.h | 2 +- 14 files changed, 14 insertions(+), 14 deletions(-) diff --git a/src/Client/ClientBase.h b/src/Client/ClientBase.h index d877905302d..f98f6d9c83b 100644 --- a/src/Client/ClientBase.h +++ b/src/Client/ClientBase.h @@ -63,7 +63,7 @@ void interruptSignalHandler(int signum); class InternalTextLogs; class WriteBufferFromFileDescriptor; -class ClientBase : public Poco::Util::Application, public IHints<2, ClientBase> +class ClientBase : public Poco::Util::Application, public IHints<2> { public: diff --git a/src/Common/IFactoryWithAliases.h b/src/Common/IFactoryWithAliases.h index 07440dd2463..e406154f87f 100644 --- a/src/Common/IFactoryWithAliases.h +++ b/src/Common/IFactoryWithAliases.h @@ -20,7 +20,7 @@ namespace ErrorCodes * template parameter is available as Value */ template -class IFactoryWithAliases : public IHints<2, IFactoryWithAliases> +class IFactoryWithAliases : public IHints<2> { protected: using Value = ValueType; diff --git a/src/Common/NamePrompter.h b/src/Common/NamePrompter.h index 65a2c76a102..1c2d4a2706e 100644 --- a/src/Common/NamePrompter.h +++ b/src/Common/NamePrompter.h @@ -95,7 +95,7 @@ String getHintsErrorMessageSuffix(const std::vector & hints); void appendHintsMessage(String & error_message, const std::vector & hints); -template +template class IHints { public: diff --git a/src/Core/Settings.h b/src/Core/Settings.h index c969ef187c4..62cdea92bcd 100644 --- a/src/Core/Settings.h +++ b/src/Core/Settings.h @@ -1070,7 +1070,7 @@ DECLARE_SETTINGS_TRAITS_ALLOW_CUSTOM_SETTINGS(SettingsTraits, LIST_OF_SETTINGS) /** Settings of query execution. * These settings go to users.xml. */ -struct Settings : public BaseSettings, public IHints<2, Settings> +struct Settings : public BaseSettings, public IHints<2> { Settings() = default; diff --git a/src/DataTypes/EnumValues.h b/src/DataTypes/EnumValues.h index 2e6628adcf3..5189f7a56f5 100644 --- a/src/DataTypes/EnumValues.h +++ b/src/DataTypes/EnumValues.h @@ -13,7 +13,7 @@ namespace ErrorCodes } template -class EnumValues : public IHints<1, EnumValues> +class EnumValues : public IHints<> { public: using Value = std::pair; diff --git a/src/Functions/UserDefined/UserDefinedSQLFunctionFactory.h b/src/Functions/UserDefined/UserDefinedSQLFunctionFactory.h index 45196759d3b..a7d586061b2 100644 --- a/src/Functions/UserDefined/UserDefinedSQLFunctionFactory.h +++ b/src/Functions/UserDefined/UserDefinedSQLFunctionFactory.h @@ -15,7 +15,7 @@ class BackupEntriesCollector; class RestorerFromBackup; /// Factory for SQLUserDefinedFunctions -class UserDefinedSQLFunctionFactory : public IHints<1, UserDefinedSQLFunctionFactory> +class UserDefinedSQLFunctionFactory : public IHints<> { public: static UserDefinedSQLFunctionFactory & instance(); diff --git a/src/Interpreters/DatabaseCatalog.cpp b/src/Interpreters/DatabaseCatalog.cpp index 53d43f341a0..2e82ff464d2 100644 --- a/src/Interpreters/DatabaseCatalog.cpp +++ b/src/Interpreters/DatabaseCatalog.cpp @@ -61,7 +61,7 @@ namespace ErrorCodes extern const int UNFINISHED; } -class DatabaseNameHints : public IHints<1, DatabaseNameHints> +class DatabaseNameHints : public IHints<> { public: explicit DatabaseNameHints(const DatabaseCatalog & database_catalog_) diff --git a/src/Interpreters/DatabaseCatalog.h b/src/Interpreters/DatabaseCatalog.h index 6fdee351567..edf1036b438 100644 --- a/src/Interpreters/DatabaseCatalog.h +++ b/src/Interpreters/DatabaseCatalog.h @@ -30,7 +30,7 @@ namespace fs = std::filesystem; namespace DB { -class TableNameHints : public IHints<1, TableNameHints> +class TableNameHints : public IHints<> { public: TableNameHints(ConstDatabasePtr database_, ContextPtr context_) diff --git a/src/Server/HTTPPathHints.h b/src/Server/HTTPPathHints.h index 708816ebf07..fe6feebb727 100644 --- a/src/Server/HTTPPathHints.h +++ b/src/Server/HTTPPathHints.h @@ -7,7 +7,7 @@ namespace DB { -class HTTPPathHints : public IHints<1, HTTPPathHints> +class HTTPPathHints : public IHints<> { public: std::vector getAllRegisteredNames() const override; diff --git a/src/Storages/ColumnsDescription.h b/src/Storages/ColumnsDescription.h index fb1eeed3127..ee0bb5efb66 100644 --- a/src/Storages/ColumnsDescription.h +++ b/src/Storages/ColumnsDescription.h @@ -98,7 +98,7 @@ struct ColumnDescription /// Description of multiple table columns (in CREATE TABLE for example). -class ColumnsDescription : public IHints<1, ColumnsDescription> +class ColumnsDescription : public IHints<> { public: ColumnsDescription() = default; diff --git a/src/Storages/IStorage.h b/src/Storages/IStorage.h index c2fb82aceee..a6d990a3eff 100644 --- a/src/Storages/IStorage.h +++ b/src/Storages/IStorage.h @@ -91,7 +91,7 @@ using IndexSize = ColumnSize; * - data storage structure (compression, etc.) * - concurrent access to data (locks, etc.) */ -class IStorage : public std::enable_shared_from_this, public TypePromotion, public IHints<1, IStorage> +class IStorage : public std::enable_shared_from_this, public TypePromotion, public IHints<> { public: IStorage() = delete; diff --git a/src/Storages/IndicesDescription.h b/src/Storages/IndicesDescription.h index 862df6fe23c..e56642b8c76 100644 --- a/src/Storages/IndicesDescription.h +++ b/src/Storages/IndicesDescription.h @@ -62,7 +62,7 @@ struct IndexDescription }; /// All secondary indices in storage -struct IndicesDescription : public std::vector, IHints<1, IndicesDescription> +struct IndicesDescription : public std::vector, IHints<> { /// Index with name exists bool has(const String & name) const; diff --git a/src/Storages/ProjectionsDescription.h b/src/Storages/ProjectionsDescription.h index c48942eb0ec..30af80d6d85 100644 --- a/src/Storages/ProjectionsDescription.h +++ b/src/Storages/ProjectionsDescription.h @@ -106,7 +106,7 @@ struct ProjectionDescription using ProjectionDescriptionRawPtr = const ProjectionDescription *; /// All projections in storage -struct ProjectionsDescription : public IHints<1, ProjectionsDescription> +struct ProjectionsDescription : public IHints<> { ProjectionsDescription() = default; ProjectionsDescription(ProjectionsDescription && other) = default; diff --git a/src/Storages/StorageFactory.h b/src/Storages/StorageFactory.h index f1c1c237393..239f1bb63ef 100644 --- a/src/Storages/StorageFactory.h +++ b/src/Storages/StorageFactory.h @@ -22,7 +22,7 @@ struct StorageID; * In 'columns' Nested data structures must be flattened. * You should subsequently call IStorage::startup method to work with table. */ -class StorageFactory : private boost::noncopyable, public IHints<1, StorageFactory> +class StorageFactory : private boost::noncopyable, public IHints<> { public: From caadfe393bee6da65fe00a01943925d87ec5f22d Mon Sep 17 00:00:00 2001 From: Robert Schulze Date: Sun, 10 Sep 2023 12:55:15 +0000 Subject: [PATCH 263/433] Fixes --- .../functions/date-time-functions.md | 6 ++--- src/Functions/DateTimeTransforms.h | 10 +++++++-- src/Functions/toDaysSinceYearZero.cpp | 14 ++++++++++-- .../02874_daysSinceYearZero.reference | 13 ----------- .../0_stateless/02874_daysSinceYearZero.sql | 22 ------------------- .../02874_toDaysSinceYearZero.reference | 13 +++++++++++ .../0_stateless/02874_toDaysSinceYearZero.sql | 22 +++++++++++++++++++ 7 files changed, 58 insertions(+), 42 deletions(-) delete mode 100644 tests/queries/0_stateless/02874_daysSinceYearZero.reference delete mode 100644 tests/queries/0_stateless/02874_daysSinceYearZero.sql create mode 100644 tests/queries/0_stateless/02874_toDaysSinceYearZero.reference create mode 100644 tests/queries/0_stateless/02874_toDaysSinceYearZero.sql diff --git a/docs/en/sql-reference/functions/date-time-functions.md b/docs/en/sql-reference/functions/date-time-functions.md index ef0ccf84f73..898d57c64db 100644 --- a/docs/en/sql-reference/functions/date-time-functions.md +++ b/docs/en/sql-reference/functions/date-time-functions.md @@ -722,7 +722,7 @@ SELECT toDate('2016-12-27') AS date, toYearWeek(date) AS yearWeek0, toYearWeek(d ## toDaysSinceYearZero -Returns for a given date, the number of days passed since the [year 0](https://en.wikipedia.org/wiki/Year_zero). +Returns for a given date, the number of days passed since [1 January 0000](https://en.wikipedia.org/wiki/Year_zero) in the [proleptic Gregorian calendar defined by ISO 8601](https://en.wikipedia.org/wiki/Gregorian_calendar#Proleptic_Gregorian_calendar). The calculation is the same as in MySQL's [`TO_DAYS()`](https://dev.mysql.com/doc/refman/8.0/en/date-and-time-functions.html#function_to-days) function. **Syntax** @@ -734,7 +734,7 @@ Aliases: `TO_DAYS` **Arguments** -- `date` — The date to calculate the number of days passed since year zero from. [Date](../../sql-reference/data-types/date.md) or [Date64](../../sql-reference/data-types/date64.md). +- `date` — The date to calculate the number of days passed since year zero from. [Date](../../sql-reference/data-types/date.md) or [Date32](../../sql-reference/data-types/date32.md). **Returned value** @@ -752,7 +752,7 @@ Result: ``` text ┌─toDaysSinceYearZero(toDate('2023-09-08')))─┐ -│ 738772 │ +│ 713569 │ └────────────────────────────────────────────┘ ``` diff --git a/src/Functions/DateTimeTransforms.h b/src/Functions/DateTimeTransforms.h index 16cfba69275..a351d7fdf30 100644 --- a/src/Functions/DateTimeTransforms.h +++ b/src/Functions/DateTimeTransforms.h @@ -927,6 +927,12 @@ struct ToDayOfYearImpl struct ToDaysSinceYearZeroImpl { +private: + /// Constant calculated from MySQL's TO_DAYS() implementation. + /// https://github.com/mysql/mysql-server/blob/ea1efa9822d81044b726aab20c857d5e1b7e046a/mysys/my_time.cc#L1042 + static constexpr auto DAYS_BETWEEN_YEARS_0_AND_1900 = 693'961; /// 01 January, each + +public: static constexpr auto name = "toDaysSinceYearZero"; static UInt32 execute(Int64, const DateLUTImpl &) @@ -939,11 +945,11 @@ struct ToDaysSinceYearZeroImpl } static UInt32 execute(Int32 d, const DateLUTImpl &) { - return /* days between 0000-01-01 and 1970-01-01 */ 719'164 + d; + return DAYS_BETWEEN_YEARS_0_AND_1900 + d; } static UInt32 execute(UInt16 d, const DateLUTImpl &) { - return /* days between 0000-01-01 and 1970-01-01 */ 719'164 + d; + return DAYS_BETWEEN_YEARS_0_AND_1900 + d; } static constexpr bool hasPreimage() { return false; } diff --git a/src/Functions/toDaysSinceYearZero.cpp b/src/Functions/toDaysSinceYearZero.cpp index c4233416195..e569c5cb1f3 100644 --- a/src/Functions/toDaysSinceYearZero.cpp +++ b/src/Functions/toDaysSinceYearZero.cpp @@ -19,7 +19,7 @@ namespace ErrorCodes namespace { -/** Returns number of days passed since 0001-01-01 */ +/** Returns number of days passed since 0000-01-01 */ class FunctionToDaysSinceYearZero : public IFunction { using ResultType = DataTypeUInt32; @@ -65,7 +65,17 @@ public: REGISTER_FUNCTION(ToDaysSinceYearZero) { - factory.registerFunction(); + factory.registerFunction( + FunctionDocumentation{ + .description=R"( +Returns for a given date, the number of days passed since 1 January 0000 in the proleptic Gregorian calendar defined by ISO 8601. +The calculation is the same as in MySQL's TO_DAYS() function. +)", + .examples{ + {"typical", "SELECT toDaysSinceYearZero(toDate('2023-09-08'))", "713569"}}, + .categories{"Dates and Times"} + }); + /// MySQL compatibility alias. factory.registerAlias("TO_DAYS", FunctionToDaysSinceYearZero::name, FunctionFactory::CaseInsensitive); } diff --git a/tests/queries/0_stateless/02874_daysSinceYearZero.reference b/tests/queries/0_stateless/02874_daysSinceYearZero.reference deleted file mode 100644 index 9bfb9b78910..00000000000 --- a/tests/queries/0_stateless/02874_daysSinceYearZero.reference +++ /dev/null @@ -1,13 +0,0 @@ -Reject invalid parameters -Const argument -719164 -738772 -693597 -738772 -\N -Non-const argument -738772 -738772 -MySQL alias -738772 -738772 diff --git a/tests/queries/0_stateless/02874_daysSinceYearZero.sql b/tests/queries/0_stateless/02874_daysSinceYearZero.sql deleted file mode 100644 index 0db1a305ea9..00000000000 --- a/tests/queries/0_stateless/02874_daysSinceYearZero.sql +++ /dev/null @@ -1,22 +0,0 @@ -SELECT 'Reject invalid parameters'; -SELECT daysSinceYearZero(); -- { serverError NUMBER_OF_ARGUMENTS_DOESNT_MATCH } -SELECT daysSinceYearZero(toDate('2023-09-08'), toDate('2023-09-08')); -- { serverError NUMBER_OF_ARGUMENTS_DOESNT_MATCH } -SELECT daysSinceYearZero('str'); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } -SELECT daysSinceYearZero(42); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } -SELECT daysSinceYearZero(toDateTime('2023-09-08 11:11:11')); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } -SELECT daysSinceYearZero(toDateTime64('2023-09-08 11:11:11', 3)); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } - -SELECT 'Const argument'; -SELECT daysSinceYearZero(toDate('1970-01-01')); -SELECT daysSinceYearZero(toDate('2023-09-08')); -SELECT daysSinceYearZero(toDate32('1900-01-01')); -SELECT daysSinceYearZero(toDate32('2023-09-08')); -SELECT daysSinceYearZero(NULL); - -SELECT 'Non-const argument'; -SELECT daysSinceYearZero(materialize(toDate('2023-09-08'))); -SELECT daysSinceYearZero(materialize(toDate32('2023-09-08'))); - -SELECT 'MySQL alias'; -SELECT to_days(toDate('2023-09-08')); -SELECT TO_DAYS(toDate('2023-09-08')); diff --git a/tests/queries/0_stateless/02874_toDaysSinceYearZero.reference b/tests/queries/0_stateless/02874_toDaysSinceYearZero.reference new file mode 100644 index 00000000000..885332ab835 --- /dev/null +++ b/tests/queries/0_stateless/02874_toDaysSinceYearZero.reference @@ -0,0 +1,13 @@ +Invalid parameters +Const argument +693961 +713569 +668394 +713569 +\N +Non-const argument +713569 +713569 +MySQL alias +713569 +713569 diff --git a/tests/queries/0_stateless/02874_toDaysSinceYearZero.sql b/tests/queries/0_stateless/02874_toDaysSinceYearZero.sql new file mode 100644 index 00000000000..2c35920e569 --- /dev/null +++ b/tests/queries/0_stateless/02874_toDaysSinceYearZero.sql @@ -0,0 +1,22 @@ +SELECT 'Invalid parameters'; +SELECT toDaysSinceYearZero(); -- { serverError NUMBER_OF_ARGUMENTS_DOESNT_MATCH } +SELECT toDaysSinceYearZero(toDate('2023-09-08'), toDate('2023-09-08')); -- { serverError NUMBER_OF_ARGUMENTS_DOESNT_MATCH } +SELECT toDaysSinceYearZero('str'); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT toDaysSinceYearZero(42); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT toDaysSinceYearZero(toDateTime('2023-09-08 11:11:11')); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT toDaysSinceYearZero(toDateTime64('2023-09-08 11:11:11', 3)); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } + +SELECT 'Const argument'; +SELECT toDaysSinceYearZero(toDate('1970-01-01')); +SELECT toDaysSinceYearZero(toDate('2023-09-08')); +SELECT toDaysSinceYearZero(toDate32('1900-01-01')); +SELECT toDaysSinceYearZero(toDate32('2023-09-08')); +SELECT toDaysSinceYearZero(NULL); + +SELECT 'Non-const argument'; +SELECT toDaysSinceYearZero(materialize(toDate('2023-09-08'))); +SELECT toDaysSinceYearZero(materialize(toDate32('2023-09-08'))); + +SELECT 'MySQL alias'; +SELECT to_days(toDate('2023-09-08')); +SELECT TO_DAYS(toDate('2023-09-08')); From 2a8e2a0882bf46d3c6db682d78db2b027b3f2517 Mon Sep 17 00:00:00 2001 From: Robert Schulze Date: Sun, 10 Sep 2023 14:12:12 +0000 Subject: [PATCH 264/433] Docs: Mention behavior change of splitByChar() in v22.11, see #42673 --- .../functions/splitting-merging-functions.md | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/docs/en/sql-reference/functions/splitting-merging-functions.md b/docs/en/sql-reference/functions/splitting-merging-functions.md index 7336e53fc24..c88643ef7cf 100644 --- a/docs/en/sql-reference/functions/splitting-merging-functions.md +++ b/docs/en/sql-reference/functions/splitting-merging-functions.md @@ -33,6 +33,13 @@ Returns an array of selected substrings. Empty substrings may be selected when: Type: [Array](../../sql-reference/data-types/array.md)([String](../../sql-reference/data-types/string.md)). +:::note +The behavior of parameter `max_substrings` changed starting with ClickHouse v22.11. In versions older than that, `max_substrings` > 0 meant that `max_substring`-many splits were performed and that the remainder of the string was returned as the final element of the list. +For example, +- in v22.10: `SELECT splitByChar('=', 'a=b=c=d', 2); -- ['a','b','c=d']` +- in v22.11: `SELECT splitByChar('=', 'a=b=c=d', 2); -- ['a','b']` +::: + **Example** ``` sql @@ -63,7 +70,6 @@ splitByString(separator, s[, max_substrings])) - `s` — The string to split. [String](../../sql-reference/data-types/string.md). - `max_substrings` — An optional `Int64` defaulting to 0. When `max_substrings` > 0, the returned substrings will be no more than `max_substrings`, otherwise the function will return as many substrings as possible. - **Returned value(s)** Returns an array of selected substrings. Empty substrings may be selected when: From d326554caaf8c7536463c1d3d73d6ec7c0f64c19 Mon Sep 17 00:00:00 2001 From: serxa Date: Sun, 10 Sep 2023 14:55:40 +0000 Subject: [PATCH 265/433] fix keeper build --- src/Coordination/Standalone/Context.cpp | 10 ++++++++++ src/Coordination/Standalone/Context.h | 4 ++++ src/Disks/ObjectStorages/DiskObjectStorage.cpp | 2 +- src/Interpreters/Context.cpp | 2 +- src/Interpreters/Context.h | 2 +- 5 files changed, 17 insertions(+), 3 deletions(-) diff --git a/src/Coordination/Standalone/Context.cpp b/src/Coordination/Standalone/Context.cpp index 84083169df3..f0ba16e50ee 100644 --- a/src/Coordination/Standalone/Context.cpp +++ b/src/Coordination/Standalone/Context.cpp @@ -268,6 +268,16 @@ ReadSettings Context::getReadSettings() const return ReadSettings{}; } +ResourceManagerPtr Context::getResourceManager() const +{ + return nullptr; +} + +ClassifierPtr Context::getWorkloadClassifier() const +{ + return nullptr; +} + void Context::initializeKeeperDispatcher([[maybe_unused]] bool start_async) const { const auto & config_ref = getConfigRef(); diff --git a/src/Coordination/Standalone/Context.h b/src/Coordination/Standalone/Context.h index cf0fceab34c..95f63d40560 100644 --- a/src/Coordination/Standalone/Context.h +++ b/src/Coordination/Standalone/Context.h @@ -118,6 +118,10 @@ public: ReadSettings getReadSettings() const; + /// Resource management related + ResourceManagerPtr getResourceManager() const; + ClassifierPtr getWorkloadClassifier() const; + std::shared_ptr getKeeperDispatcher() const; std::shared_ptr tryGetKeeperDispatcher() const; void initializeKeeperDispatcher(bool start_async) const; diff --git a/src/Disks/ObjectStorages/DiskObjectStorage.cpp b/src/Disks/ObjectStorages/DiskObjectStorage.cpp index 5a70b2ca579..466a1d3d5dd 100644 --- a/src/Disks/ObjectStorages/DiskObjectStorage.cpp +++ b/src/Disks/ObjectStorages/DiskObjectStorage.cpp @@ -491,7 +491,7 @@ static inline Settings updateResourceLink(const Settings & settings, const Strin if (auto query_context = CurrentThread::getQueryContext()) { Settings result(settings); - result.resource_link = query_context->getClassifier()->get(resource_name); + result.resource_link = query_context->getWorkloadClassifier()->get(resource_name); return result; } return settings; diff --git a/src/Interpreters/Context.cpp b/src/Interpreters/Context.cpp index 8074fd8afa8..0b5294b8d00 100644 --- a/src/Interpreters/Context.cpp +++ b/src/Interpreters/Context.cpp @@ -1294,7 +1294,7 @@ ResourceManagerPtr Context::getResourceManager() const return shared->resource_manager; } -ClassifierPtr Context::getClassifier() const +ClassifierPtr Context::getWorkloadClassifier() const { auto lock = getLock(); if (!classifier) diff --git a/src/Interpreters/Context.h b/src/Interpreters/Context.h index eceed31ae86..a4019d214e4 100644 --- a/src/Interpreters/Context.h +++ b/src/Interpreters/Context.h @@ -581,7 +581,7 @@ public: /// Resource management related ResourceManagerPtr getResourceManager() const; - ClassifierPtr getClassifier() const; + ClassifierPtr getWorkloadClassifier() const; /// We have to copy external tables inside executeQuery() to track limits. Therefore, set callback for it. Must set once. void setExternalTablesInitializer(ExternalTablesInitializer && initializer); From 6a2534b555ff38f740f4344ee9d4f1381a4a8ab7 Mon Sep 17 00:00:00 2001 From: serxa Date: Sun, 10 Sep 2023 17:43:04 +0000 Subject: [PATCH 266/433] fix keeper build #2 --- src/Coordination/Standalone/Context.h | 1 + 1 file changed, 1 insertion(+) diff --git a/src/Coordination/Standalone/Context.h b/src/Coordination/Standalone/Context.h index 95f63d40560..d9bbccea2e9 100644 --- a/src/Coordination/Standalone/Context.h +++ b/src/Coordination/Standalone/Context.h @@ -13,6 +13,7 @@ #include #include +#include #include From eed11c36a4b2c95ec162d029c6c78e20efa151d5 Mon Sep 17 00:00:00 2001 From: Joe Lynch Date: Sun, 10 Sep 2023 23:11:24 +0200 Subject: [PATCH 267/433] Add jq to fasttest container --- docker/test/fasttest/Dockerfile | 1 + 1 file changed, 1 insertion(+) diff --git a/docker/test/fasttest/Dockerfile b/docker/test/fasttest/Dockerfile index ad24e662a6c..fd7a5640964 100644 --- a/docker/test/fasttest/Dockerfile +++ b/docker/test/fasttest/Dockerfile @@ -18,6 +18,7 @@ RUN apt-get update \ python3-termcolor \ unixodbc \ pv \ + jq \ zstd \ --yes --no-install-recommends From aac974b38d320ec4a2d827457078062413393a14 Mon Sep 17 00:00:00 2001 From: Alexey Milovidov Date: Mon, 11 Sep 2023 05:35:15 +0200 Subject: [PATCH 268/433] Fix strange message --- cmake/limit_jobs.cmake | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/cmake/limit_jobs.cmake b/cmake/limit_jobs.cmake index acc38b6fa2a..28ccb62e10c 100644 --- a/cmake/limit_jobs.cmake +++ b/cmake/limit_jobs.cmake @@ -47,7 +47,7 @@ if (CMAKE_BUILD_TYPE_UC STREQUAL "RELWITHDEBINFO" AND ENABLE_THINLTO AND PARALLE set (PARALLEL_LINK_JOBS 2) endif() -message(STATUS "Building sub-tree with ${PARALLEL_COMPILE_JOBS} compile jobs and ${PARALLEL_LINK_JOBS} linker jobs (system: ${NUMBER_OF_LOGICAL_CORES} cores, ${TOTAL_PHYSICAL_MEMORY} MB DRAM, 'OFF' means the native core count).") +message(STATUS "Building sub-tree with ${PARALLEL_COMPILE_JOBS} compile jobs and ${PARALLEL_LINK_JOBS} linker jobs (system: ${NUMBER_OF_LOGICAL_CORES} cores, ${TOTAL_PHYSICAL_MEMORY} MB RAM, 'OFF' means the native core count).") if (PARALLEL_COMPILE_JOBS LESS NUMBER_OF_LOGICAL_CORES) set(CMAKE_JOB_POOL_COMPILE compile_job_pool${CMAKE_CURRENT_SOURCE_DIR}) From b72a49f090dcb5b90f6585d9097872ea0d0c865f Mon Sep 17 00:00:00 2001 From: Robert Schulze Date: Mon, 11 Sep 2023 06:32:14 +0000 Subject: [PATCH 269/433] Fix style --- utils/check-style/aspell-ignore/en/aspell-dict.txt | 2 ++ 1 file changed, 2 insertions(+) diff --git a/utils/check-style/aspell-ignore/en/aspell-dict.txt b/utils/check-style/aspell-ignore/en/aspell-dict.txt index 2e45b885afd..34b1766d8eb 100644 --- a/utils/check-style/aspell-ignore/en/aspell-dict.txt +++ b/utils/check-style/aspell-ignore/en/aspell-dict.txt @@ -1997,6 +1997,7 @@ privateKeyPassphraseHandler prlimit procfs profiler +proleptic prometheus proto protobuf @@ -2353,6 +2354,7 @@ toDateTimeOrZero toDayOfMonth toDayOfWeek toDayOfYear +toDaysSinceYearZero toDecimal toDecimalString toFixedString From ea038380b036987714125bc8e6f2df8f3c49ebbb Mon Sep 17 00:00:00 2001 From: Robert Schulze Date: Mon, 11 Sep 2023 07:12:49 +0000 Subject: [PATCH 270/433] Bump re2 to latest main Now supports an alternative form for named capturing groups [0]: (?re) [0] https://github.com/google/re2/commit/6148386f0c8f03da1ab0bf982a4dbe080b4ea7bc --- contrib/re2 | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/contrib/re2 b/contrib/re2 index 03da4fc0857..a807e8a3aac 160000 --- a/contrib/re2 +++ b/contrib/re2 @@ -1 +1 @@ -Subproject commit 03da4fc0857c285e3a26782f6bc8931c4c950df4 +Subproject commit a807e8a3aac2cc33c77b7071efea54fcabe38e0c From f9351fdf7be58dcba4b073fcea92e78c5df750b1 Mon Sep 17 00:00:00 2001 From: Alexander van Olst Date: Mon, 11 Sep 2023 13:52:14 +0200 Subject: [PATCH 271/433] Correct default value for parts_to_delay_insert --- docs/en/operations/settings/merge-tree-settings.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/docs/en/operations/settings/merge-tree-settings.md b/docs/en/operations/settings/merge-tree-settings.md index 8889e04dba1..e746719b629 100644 --- a/docs/en/operations/settings/merge-tree-settings.md +++ b/docs/en/operations/settings/merge-tree-settings.md @@ -71,7 +71,7 @@ Possible values: - Any positive integer. -Default value: 150. +Default value: 1000. ClickHouse artificially executes `INSERT` longer (adds ‘sleep’) so that the background merge process can merge parts faster than they are added. From 5061da4e8ace9664e880cb67070e295f76477da5 Mon Sep 17 00:00:00 2001 From: vdimir Date: Mon, 11 Sep 2023 12:12:02 +0000 Subject: [PATCH 272/433] clickhouse-local support --database command line argument --- programs/local/LocalServer.cpp | 2 ++ .../02141_clickhouse_local_interactive_table.reference | 1 + .../0_stateless/02141_clickhouse_local_interactive_table.sh | 1 + 3 files changed, 4 insertions(+) diff --git a/programs/local/LocalServer.cpp b/programs/local/LocalServer.cpp index 5c522b678ef..3fa2532181a 100644 --- a/programs/local/LocalServer.cpp +++ b/programs/local/LocalServer.cpp @@ -881,6 +881,8 @@ void LocalServer::processOptions(const OptionsDescription &, const CommandLineOp config().setBool("no-system-tables", true); if (options.count("only-system-tables")) config().setBool("only-system-tables", true); + if (options.count("database")) + config().setString("default_database", options["database"].as()); if (options.count("input-format")) config().setString("table-data-format", options["input-format"].as()); diff --git a/tests/queries/0_stateless/02141_clickhouse_local_interactive_table.reference b/tests/queries/0_stateless/02141_clickhouse_local_interactive_table.reference index e4c93e9e1c5..4c1d5dc829f 100644 --- a/tests/queries/0_stateless/02141_clickhouse_local_interactive_table.reference +++ b/tests/queries/0_stateless/02141_clickhouse_local_interactive_table.reference @@ -1 +1,2 @@ CREATE TABLE _local.table\n(\n `key` String\n)\nENGINE = File(\'TSVWithNamesAndTypes\', \'/dev/null\') +CREATE TABLE foo.table\n(\n `key` String\n)\nENGINE = File(\'TSVWithNamesAndTypes\', \'/dev/null\') diff --git a/tests/queries/0_stateless/02141_clickhouse_local_interactive_table.sh b/tests/queries/0_stateless/02141_clickhouse_local_interactive_table.sh index fc71f779fa1..934d87616ac 100755 --- a/tests/queries/0_stateless/02141_clickhouse_local_interactive_table.sh +++ b/tests/queries/0_stateless/02141_clickhouse_local_interactive_table.sh @@ -5,3 +5,4 @@ CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) . "$CURDIR"/../shell_config.sh $CLICKHOUSE_LOCAL --file /dev/null --structure "key String" --input-format TSVWithNamesAndTypes --interactive --send_logs_level=trace <<<'show create table table' +$CLICKHOUSE_LOCAL --database foo --file /dev/null --structure "key String" --input-format TSVWithNamesAndTypes --interactive --send_logs_level=trace <<<'show create table table' From 1494eb3a864caa53cf8734e13f97f6e08b24a47a Mon Sep 17 00:00:00 2001 From: Antonio Andelic Date: Mon, 11 Sep 2023 12:34:16 +0000 Subject: [PATCH 273/433] Fix version parsing --- tests/integration/helpers/cluster.py | 14 +++++++++++++- 1 file changed, 13 insertions(+), 1 deletion(-) diff --git a/tests/integration/helpers/cluster.py b/tests/integration/helpers/cluster.py index 7ee86d74d35..dfd964b04d2 100644 --- a/tests/integration/helpers/cluster.py +++ b/tests/integration/helpers/cluster.py @@ -4260,7 +4260,19 @@ class ClickHouseInstance: if len(self.custom_dictionaries_paths): write_embedded_config("0_common_enable_dictionaries.xml", self.config_d_dir) - if self.tag == "latest": + version = None + version_parts = self.tag.split(".") + if version_parts[0].isdigit() and version_parts[1].isdigit(): + version = {"major": int(version_parts[0]), "minor": int(version_parts[1])} + + # async replication is only supported in version 23.9+ + # for tags that don't specify a version we assume it has a version of ClickHouse + # that supports async replication if a test for it is present + if ( + version == None + or version["major"] > 23 + or (version["major"] == 23 and version["minor"] >= 9) + ): write_embedded_config( "0_common_enable_keeper_async_replication.xml", self.config_d_dir ) From 4a8f3bf36397311b6049b6e918211f80a95f36e1 Mon Sep 17 00:00:00 2001 From: Dmitry Novik Date: Mon, 11 Sep 2023 13:01:55 +0000 Subject: [PATCH 274/433] Add SYSTEM FLUSH LOGS --- tests/queries/0_stateless/01165_lost_part_empty_partition.sql | 2 ++ 1 file changed, 2 insertions(+) diff --git a/tests/queries/0_stateless/01165_lost_part_empty_partition.sql b/tests/queries/0_stateless/01165_lost_part_empty_partition.sql index 8486a6ac2e6..a1db1c27bee 100644 --- a/tests/queries/0_stateless/01165_lost_part_empty_partition.sql +++ b/tests/queries/0_stateless/01165_lost_part_empty_partition.sql @@ -12,6 +12,7 @@ drop table rmt1; system sync replica rmt2; select lost_part_count from system.replicas where database = currentDatabase() and table = 'rmt2'; drop table rmt2; +SYSTEM FLUSH LOGS; select count() from system.text_log where logger_name like '%' || currentDatabase() || '%' and message ilike '%table with non-zero lost_part_count equal to%'; @@ -25,6 +26,7 @@ drop table rmt1; system sync replica rmt2; select lost_part_count from system.replicas where database = currentDatabase() and table = 'rmt2'; drop table rmt2; +SYSTEM FLUSH LOGS; select count() from system.text_log where logger_name like '%' || currentDatabase() || '%' and message ilike '%table with non-zero lost_part_count equal to%'; From 297306b4b974188ec19d8d83c5f1f965974619ad Mon Sep 17 00:00:00 2001 From: vdimir Date: Mon, 11 Sep 2023 13:36:02 +0000 Subject: [PATCH 275/433] status emoji in the beginning of title --- tests/ci/commit_status_helper.py | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/tests/ci/commit_status_helper.py b/tests/ci/commit_status_helper.py index 9788e629cb8..ed686dae11c 100644 --- a/tests/ci/commit_status_helper.py +++ b/tests/ci/commit_status_helper.py @@ -146,9 +146,9 @@ def update_pr_title_icon(pr: PullRequest, status: str) -> None: if not new_status_icon: return new_title = pr.title - if new_title and new_title[-1] != new_status_icon: - if new_title[-1] in set(STATUS_ICON_MAP.values()): - new_title = new_title[:-1] + " " + new_status_icon + if new_title and new_title[0] != new_status_icon: + if new_title[0] in set(STATUS_ICON_MAP.values()): + new_title = new_title[1:] + new_status_icon else: new_title = new_title + " " + new_status_icon pr.edit(title=new_title) From 73ce1993fb524ba87a9097ae6df2f3ac593feb08 Mon Sep 17 00:00:00 2001 From: Robert Schulze Date: Mon, 11 Sep 2023 12:40:49 +0000 Subject: [PATCH 276/433] Continue #45070 --- .../functions/date-time-functions.md | 95 ++++- src/Common/DateLUTImpl.h | 222 ++++++------ src/Functions/makeDate.cpp | 335 +++++++++++++++--- .../0_stateless/02245_make_datetime64.sql | 2 +- ...new_functions_must_be_documented.reference | 4 + .../02876_yyyymmddhhmmsstodatetime.reference | 85 +++++ .../02876_yyyymmddhhmmsstodatetime.sql | 119 +++++++ .../02876_yyyymmddtodate.reference | 84 +++++ .../0_stateless/02876_yyyymmddtodate.sql | 112 ++++++ 9 files changed, 900 insertions(+), 158 deletions(-) create mode 100644 tests/queries/0_stateless/02876_yyyymmddhhmmsstodatetime.reference create mode 100644 tests/queries/0_stateless/02876_yyyymmddhhmmsstodatetime.sql create mode 100644 tests/queries/0_stateless/02876_yyyymmddtodate.reference create mode 100644 tests/queries/0_stateless/02876_yyyymmddtodate.sql diff --git a/docs/en/sql-reference/functions/date-time-functions.md b/docs/en/sql-reference/functions/date-time-functions.md index cc8469e82ef..8f128f13de0 100644 --- a/docs/en/sql-reference/functions/date-time-functions.md +++ b/docs/en/sql-reference/functions/date-time-functions.md @@ -1290,6 +1290,8 @@ Rounds the time to the half hour. Converts a date or date with time to a UInt32 number containing the year and month number (YYYY \* 100 + MM). Accepts a second optional timezone argument. If provided, the timezone must be a string constant. +This functions is the opposite of function `YYYYMMDDToDate()`. + **Example** ``` sql @@ -1312,8 +1314,7 @@ Converts a date or date with time to a UInt32 number containing the year and mon **Example** ```sql -SELECT - toYYYYMMDD(now(), 'US/Eastern') +SELECT toYYYYMMDD(now(), 'US/Eastern') ``` Result: @@ -1331,8 +1332,7 @@ Converts a date or date with time to a UInt64 number containing the year and mon **Example** ```sql -SELECT - toYYYYMMDDhhmmss(now(), 'US/Eastern') +SELECT toYYYYMMDDhhmmss(now(), 'US/Eastern') ``` Result: @@ -1343,6 +1343,93 @@ Result: └───────────────────────────────────────┘ ``` +## YYYYMMDDToDate + +Converts a number containing the year, month and day number to a [Date](../../sql-reference/data-types/date.md). + +This functions is the opposite of function `toYYYYMMDD()`. + +The output is undefined if the input does not encode a valid Date value. + +**Syntax** + +```sql +YYYYMMDDToDate(yyyymmdd); +``` + +**Arguments** + +- `yyyymmdd` - A number representing the year, month and day. [Integer](../../sql-reference/data-types/int-uint.md), [Float](../../sql-reference/data-types/float.md) or [Decimal](../../sql-reference/data-types/decimal.md). + +**Returned value** + +- a date created from the arguments. + +Type: [Date](../../sql-reference/data-types/date.md). + +**Example** + +```sql +SELECT YYYYMMDDToDate(20230911); +``` + +Result: + +```response +┌─toYYYYMMDD(20230911)─┐ +│ 2023-09-11 │ +└──────────────────────┘ +``` + +## YYYYMMDDToDate32 + +Like function `YYYYMMDDToDate()` but produces a [Date32](../../sql-reference/data-types/date32.md). + +## YYYYMMDDhhmmssToDateTime + +Converts a number containing the year, month and day number to a [DateTime](../../sql-reference/data-types/datetime.md). + +The output is undefined if the input does not encode a valid DateTime value. + +This functions is the opposite of function `toYYYYMMDD()`. + +**Syntax** + +```sql +YYYYMMDDhhmmssToDateTime(yyyymmddhhmmss[, timezone]); +``` + +**Arguments** + +- `yyyymmddhhmmss` - A number representing the year, month and day. [Integer](../../sql-reference/data-types/int-uint.md), [Float](../../sql-reference/data-types/float.md) or [Decimal](../../sql-reference/data-types/decimal.md). +- `timezone` - [Timezone](../../operations/server-configuration-parameters/settings.md#server_configuration_parameters-timezone) for the returned value (optional). + +**Returned value** + +- a date with time created from the arguments. + +Type: [DateTime](../../sql-reference/data-types/datetime.md). + +**Example** + +```sql +SELECT YYYYMMDDToDateTime(20230911131415); +``` + +Result: + +```response +┌────toYYYYMMDDhhmmssToDateTime(20230911131415)─┐ +│ 2023-09-11 13:14:15 │ +└───────────────────────────────────────────────┘ +``` + +## YYYYMMDDhhmmssToDateTime64 + +Like function `YYYYMMDDhhmmssToDate()` but produces a [DateTime64](../../sql-reference/data-types/datetime64.md). + +Accepts an additional, optional `precision` parameter after the `timezone` parameter. + ## addYears, addMonths, addWeeks, addDays, addHours, addMinutes, addSeconds, addQuarters Function adds a Date/DateTime interval to a Date/DateTime and then return the Date/DateTime. For example: diff --git a/src/Common/DateLUTImpl.h b/src/Common/DateLUTImpl.h index 6d0ba718057..0e72b489ace 100644 --- a/src/Common/DateLUTImpl.h +++ b/src/Common/DateLUTImpl.h @@ -71,14 +71,14 @@ private: // Same as above but select different function overloads for zero saturation. STRONG_TYPEDEF(UInt32, LUTIndexWithSaturation) - static inline LUTIndex normalizeLUTIndex(UInt32 index) + static LUTIndex normalizeLUTIndex(UInt32 index) { if (index >= DATE_LUT_SIZE) return LUTIndex(DATE_LUT_SIZE - 1); return LUTIndex{index}; } - static inline LUTIndex normalizeLUTIndex(Int64 index) + static LUTIndex normalizeLUTIndex(Int64 index) { if (unlikely(index < 0)) return LUTIndex(0); @@ -88,59 +88,59 @@ private: } template - friend inline LUTIndex operator+(const LUTIndex & index, const T v) + friend LUTIndex operator+(const LUTIndex & index, const T v) { return normalizeLUTIndex(index.toUnderType() + UInt32(v)); } template - friend inline LUTIndex operator+(const T v, const LUTIndex & index) + friend LUTIndex operator+(const T v, const LUTIndex & index) { return normalizeLUTIndex(static_cast(v + index.toUnderType())); } - friend inline LUTIndex operator+(const LUTIndex & index, const LUTIndex & v) + friend LUTIndex operator+(const LUTIndex & index, const LUTIndex & v) { return normalizeLUTIndex(static_cast(index.toUnderType() + v.toUnderType())); } template - friend inline LUTIndex operator-(const LUTIndex & index, const T v) + friend LUTIndex operator-(const LUTIndex & index, const T v) { return normalizeLUTIndex(static_cast(index.toUnderType() - UInt32(v))); } template - friend inline LUTIndex operator-(const T v, const LUTIndex & index) + friend LUTIndex operator-(const T v, const LUTIndex & index) { return normalizeLUTIndex(static_cast(v - index.toUnderType())); } - friend inline LUTIndex operator-(const LUTIndex & index, const LUTIndex & v) + friend LUTIndex operator-(const LUTIndex & index, const LUTIndex & v) { return normalizeLUTIndex(static_cast(index.toUnderType() - v.toUnderType())); } template - friend inline LUTIndex operator*(const LUTIndex & index, const T v) + friend LUTIndex operator*(const LUTIndex & index, const T v) { return normalizeLUTIndex(index.toUnderType() * UInt32(v)); } template - friend inline LUTIndex operator*(const T v, const LUTIndex & index) + friend LUTIndex operator*(const T v, const LUTIndex & index) { return normalizeLUTIndex(v * index.toUnderType()); } template - friend inline LUTIndex operator/(const LUTIndex & index, const T v) + friend LUTIndex operator/(const LUTIndex & index, const T v) { return normalizeLUTIndex(index.toUnderType() / UInt32(v)); } template - friend inline LUTIndex operator/(const T v, const LUTIndex & index) + friend LUTIndex operator/(const T v, const LUTIndex & index) { return normalizeLUTIndex(UInt32(v) / index.toUnderType()); } @@ -172,12 +172,12 @@ public: Int8 amount_of_offset_change_value; /// Usually -4 or 4, but look at Lord Howe Island. Multiply by OffsetChangeFactor UInt8 time_at_offset_change_value; /// In seconds from beginning of the day. Multiply by OffsetChangeFactor - inline Int32 amount_of_offset_change() const /// NOLINT + Int32 amount_of_offset_change() const /// NOLINT { return static_cast(amount_of_offset_change_value) * OffsetChangeFactor; } - inline UInt32 time_at_offset_change() const /// NOLINT + UInt32 time_at_offset_change() const /// NOLINT { return static_cast(time_at_offset_change_value) * OffsetChangeFactor; } @@ -221,7 +221,7 @@ private: /// Time zone name. std::string time_zone; - inline LUTIndex findIndex(Time t) const + LUTIndex findIndex(Time t) const { /// First guess. Time guess = (t / 86400) + daynum_offset_epoch; @@ -248,34 +248,34 @@ private: return LUTIndex(guess ? static_cast(guess) - 1 : 0); } - static inline LUTIndex toLUTIndex(DayNum d) + static LUTIndex toLUTIndex(DayNum d) { return normalizeLUTIndex(d + daynum_offset_epoch); } - static inline LUTIndex toLUTIndex(ExtendedDayNum d) + static LUTIndex toLUTIndex(ExtendedDayNum d) { return normalizeLUTIndex(static_cast(d + daynum_offset_epoch)); } - inline LUTIndex toLUTIndex(Time t) const + LUTIndex toLUTIndex(Time t) const { return findIndex(t); } - static inline LUTIndex toLUTIndex(LUTIndex i) + static LUTIndex toLUTIndex(LUTIndex i) { return i; } template - inline const Values & find(DateOrTime v) const + const Values & find(DateOrTime v) const { return lut[toLUTIndex(v)]; } template - inline DateOrTime roundDown(DateOrTime x, Divisor divisor) const + DateOrTime roundDown(DateOrTime x, Divisor divisor) const { static_assert(std::is_integral_v && std::is_integral_v); assert(divisor > 0); @@ -336,7 +336,7 @@ public: } template - inline auto toDayNum(DateOrTime v) const + auto toDayNum(DateOrTime v) const { if constexpr (std::is_unsigned_v || std::is_same_v) return DayNum{static_cast(saturateMinus(toLUTIndex(v).toUnderType(), daynum_offset_epoch))}; @@ -346,7 +346,7 @@ public: /// Round down to start of monday. template - inline Time toFirstDayOfWeek(DateOrTime v) const + Time toFirstDayOfWeek(DateOrTime v) const { const LUTIndex i = toLUTIndex(v); if constexpr (std::is_unsigned_v || std::is_same_v) @@ -356,7 +356,7 @@ public: } template - inline auto toFirstDayNumOfWeek(DateOrTime v) const + auto toFirstDayNumOfWeek(DateOrTime v) const { const LUTIndex i = toLUTIndex(v); if constexpr (std::is_unsigned_v || std::is_same_v) @@ -367,7 +367,7 @@ public: /// Round up to the last day of week. template - inline Time toLastDayOfWeek(DateOrTime v) const + Time toLastDayOfWeek(DateOrTime v) const { const LUTIndex i = toLUTIndex(v); if constexpr (std::is_unsigned_v || std::is_same_v) @@ -377,7 +377,7 @@ public: } template - inline auto toLastDayNumOfWeek(DateOrTime v) const + auto toLastDayNumOfWeek(DateOrTime v) const { const LUTIndex i = toLUTIndex(v); if constexpr (std::is_unsigned_v || std::is_same_v) @@ -388,7 +388,7 @@ public: /// Round down to start of month. template - inline Time toFirstDayOfMonth(DateOrTime v) const + Time toFirstDayOfMonth(DateOrTime v) const { const LUTIndex i = toLUTIndex(v); if constexpr (std::is_unsigned_v || std::is_same_v) @@ -398,7 +398,7 @@ public: } template - inline auto toFirstDayNumOfMonth(DateOrTime v) const + auto toFirstDayNumOfMonth(DateOrTime v) const { const LUTIndex i = toLUTIndex(v); if constexpr (std::is_unsigned_v || std::is_same_v) @@ -409,7 +409,7 @@ public: /// Round up to last day of month. template - inline Time toLastDayOfMonth(DateOrTime v) const + Time toLastDayOfMonth(DateOrTime v) const { const LUTIndex i = toLUTIndex(v); if constexpr (std::is_unsigned_v || std::is_same_v) @@ -419,7 +419,7 @@ public: } template - inline auto toLastDayNumOfMonth(DateOrTime v) const + auto toLastDayNumOfMonth(DateOrTime v) const { const LUTIndex i = toLUTIndex(v); if constexpr (std::is_unsigned_v || std::is_same_v) @@ -430,7 +430,7 @@ public: /// Round down to start of quarter. template - inline auto toFirstDayNumOfQuarter(DateOrTime v) const + auto toFirstDayNumOfQuarter(DateOrTime v) const { if constexpr (std::is_unsigned_v || std::is_same_v) return toDayNum(LUTIndexWithSaturation(toFirstDayOfQuarterIndex(v))); @@ -439,7 +439,7 @@ public: } template - inline LUTIndex toFirstDayOfQuarterIndex(DateOrTime v) const + LUTIndex toFirstDayOfQuarterIndex(DateOrTime v) const { LUTIndex index = toLUTIndex(v); size_t month_inside_quarter = (lut[index].month - 1) % 3; @@ -455,25 +455,25 @@ public: } template - inline Time toFirstDayOfQuarter(DateOrTime v) const + Time toFirstDayOfQuarter(DateOrTime v) const { return toDate(toFirstDayOfQuarterIndex(v)); } /// Round down to start of year. - inline Time toFirstDayOfYear(Time t) const + Time toFirstDayOfYear(Time t) const { return lut[years_lut[lut[findIndex(t)].year - DATE_LUT_MIN_YEAR]].date; } template - inline LUTIndex toFirstDayNumOfYearIndex(DateOrTime v) const + LUTIndex toFirstDayNumOfYearIndex(DateOrTime v) const { return years_lut[lut[toLUTIndex(v)].year - DATE_LUT_MIN_YEAR]; } template - inline auto toFirstDayNumOfYear(DateOrTime v) const + auto toFirstDayNumOfYear(DateOrTime v) const { if constexpr (std::is_unsigned_v || std::is_same_v) return toDayNum(LUTIndexWithSaturation(toFirstDayNumOfYearIndex(v))); @@ -481,14 +481,14 @@ public: return toDayNum(LUTIndex(toFirstDayNumOfYearIndex(v))); } - inline Time toFirstDayOfNextMonth(Time t) const + Time toFirstDayOfNextMonth(Time t) const { LUTIndex index = findIndex(t); index += 32 - lut[index].day_of_month; return lut[index - (lut[index].day_of_month - 1)].date; } - inline Time toFirstDayOfPrevMonth(Time t) const + Time toFirstDayOfPrevMonth(Time t) const { LUTIndex index = findIndex(t); index -= lut[index].day_of_month; @@ -496,13 +496,13 @@ public: } template - inline UInt8 daysInMonth(DateOrTime value) const + UInt8 daysInMonth(DateOrTime value) const { const LUTIndex i = toLUTIndex(value); return lut[i].days_in_month; } - inline UInt8 daysInMonth(Int16 year, UInt8 month) const + UInt8 daysInMonth(Int16 year, UInt8 month) const { UInt16 idx = year - DATE_LUT_MIN_YEAR; if (unlikely(idx >= DATE_LUT_YEARS)) @@ -515,12 +515,12 @@ public: /** Round to start of day, then shift for specified amount of days. */ - inline Time toDateAndShift(Time t, Int32 days) const + Time toDateAndShift(Time t, Int32 days) const { return lut[findIndex(t) + days].date; } - inline Time toTime(Time t) const + Time toTime(Time t) const { const LUTIndex index = findIndex(t); @@ -532,7 +532,7 @@ public: return res - offset_at_start_of_epoch; /// Starting at 1970-01-01 00:00:00 local time. } - inline unsigned toHour(Time t) const + unsigned toHour(Time t) const { const LUTIndex index = findIndex(t); @@ -552,7 +552,7 @@ public: * then subtract the former from the latter to get the offset result. * The boundaries when meets DST(daylight saving time) change should be handled very carefully. */ - inline Time timezoneOffset(Time t) const + Time timezoneOffset(Time t) const { const LUTIndex index = findIndex(t); @@ -574,7 +574,7 @@ public: } - inline unsigned toSecond(Time t) const + unsigned toSecond(Time t) const { if (likely(offset_is_whole_number_of_minutes_during_epoch)) { @@ -593,7 +593,7 @@ public: return time % 60; } - inline unsigned toMinute(Time t) const + unsigned toMinute(Time t) const { if (t >= 0 && offset_is_whole_number_of_hours_during_epoch) return (t / 60) % 60; @@ -630,11 +630,11 @@ public: * because the same calendar day starts/ends at different timestamps in different time zones) */ - inline Time fromDayNum(DayNum d) const { return lut_saturated[toLUTIndex(d)].date; } - inline Time fromDayNum(ExtendedDayNum d) const { return lut[toLUTIndex(d)].date; } + Time fromDayNum(DayNum d) const { return lut_saturated[toLUTIndex(d)].date; } + Time fromDayNum(ExtendedDayNum d) const { return lut[toLUTIndex(d)].date; } template - inline Time toDate(DateOrTime v) const + Time toDate(DateOrTime v) const { if constexpr (std::is_unsigned_v || std::is_same_v) return lut_saturated[toLUTIndex(v)].date; @@ -643,20 +643,20 @@ public: } template - inline UInt8 toMonth(DateOrTime v) const { return lut[toLUTIndex(v)].month; } + UInt8 toMonth(DateOrTime v) const { return lut[toLUTIndex(v)].month; } template - inline UInt8 toQuarter(DateOrTime v) const { return (lut[toLUTIndex(v)].month - 1) / 3 + 1; } + UInt8 toQuarter(DateOrTime v) const { return (lut[toLUTIndex(v)].month - 1) / 3 + 1; } template - inline Int16 toYear(DateOrTime v) const { return lut[toLUTIndex(v)].year; } + Int16 toYear(DateOrTime v) const { return lut[toLUTIndex(v)].year; } /// 1-based, starts on Monday template - inline UInt8 toDayOfWeek(DateOrTime v) const { return lut[toLUTIndex(v)].day_of_week; } + UInt8 toDayOfWeek(DateOrTime v) const { return lut[toLUTIndex(v)].day_of_week; } template - inline UInt8 toDayOfWeek(DateOrTime v, UInt8 week_day_mode) const + UInt8 toDayOfWeek(DateOrTime v, UInt8 week_day_mode) const { WeekDayMode mode = check_week_day_mode(week_day_mode); @@ -674,10 +674,10 @@ public: } template - inline UInt8 toDayOfMonth(DateOrTime v) const { return lut[toLUTIndex(v)].day_of_month; } + UInt8 toDayOfMonth(DateOrTime v) const { return lut[toLUTIndex(v)].day_of_month; } template - inline UInt16 toDayOfYear(DateOrTime v) const + UInt16 toDayOfYear(DateOrTime v) const { // TODO: different overload for ExtendedDayNum const LUTIndex i = toLUTIndex(v); @@ -688,7 +688,7 @@ public: /// (round down to monday and divide DayNum by 7; we made an assumption, /// that in domain of the function there was no weeks with any other number of days than 7) template - inline Int32 toRelativeWeekNum(DateOrTime v) const + Int32 toRelativeWeekNum(DateOrTime v) const { const LUTIndex i = toLUTIndex(v); /// We add 8 to avoid underflow at beginning of unix epoch. @@ -697,7 +697,7 @@ public: /// Get year that contains most of the current week. Week begins at monday. template - inline Int16 toISOYear(DateOrTime v) const + Int16 toISOYear(DateOrTime v) const { const LUTIndex i = toLUTIndex(v); /// That's effectively the year of thursday of current week. @@ -708,7 +708,7 @@ public: /// Example: ISO year 2019 begins at 2018-12-31. And ISO year 2017 begins at 2017-01-02. /// https://en.wikipedia.org/wiki/ISO_week_date template - inline LUTIndex toFirstDayNumOfISOYearIndex(DateOrTime v) const + LUTIndex toFirstDayNumOfISOYearIndex(DateOrTime v) const { const LUTIndex i = toLUTIndex(v); auto iso_year = toISOYear(i); @@ -722,7 +722,7 @@ public: } template - inline auto toFirstDayNumOfISOYear(DateOrTime v) const + auto toFirstDayNumOfISOYear(DateOrTime v) const { if constexpr (std::is_unsigned_v || std::is_same_v) return toDayNum(LUTIndexWithSaturation(toFirstDayNumOfISOYearIndex(v))); @@ -730,7 +730,7 @@ public: return toDayNum(LUTIndex(toFirstDayNumOfISOYearIndex(v))); } - inline Time toFirstDayOfISOYear(Time t) const + Time toFirstDayOfISOYear(Time t) const { return lut[toFirstDayNumOfISOYearIndex(t)].date; } @@ -738,7 +738,7 @@ public: /// ISO 8601 week number. Week begins at monday. /// The week number 1 is the first week in year that contains 4 or more days (that's more than half). template - inline UInt8 toISOWeek(DateOrTime v) const + UInt8 toISOWeek(DateOrTime v) const { return 1 + (toFirstDayNumOfWeek(v) - toDayNum(toFirstDayNumOfISOYearIndex(v))) / 7; } @@ -777,7 +777,7 @@ public: next week is week 1. */ template - inline YearWeek toYearWeek(DateOrTime v, UInt8 week_mode) const + YearWeek toYearWeek(DateOrTime v, UInt8 week_mode) const { const bool newyear_day_mode = week_mode & static_cast(WeekModeFlag::NEWYEAR_DAY); week_mode = check_week_mode(week_mode); @@ -836,7 +836,7 @@ public: /// Calculate week number of WeekModeFlag::NEWYEAR_DAY mode /// The week number 1 is the first week in year that contains January 1, template - inline YearWeek toYearWeekOfNewyearMode(DateOrTime v, bool monday_first_mode) const + YearWeek toYearWeekOfNewyearMode(DateOrTime v, bool monday_first_mode) const { YearWeek yw(0, 0); UInt16 offset_day = monday_first_mode ? 0U : 1U; @@ -870,7 +870,7 @@ public: /// Get first day of week with week_mode, return Sunday or Monday template - inline auto toFirstDayNumOfWeek(DateOrTime v, UInt8 week_mode) const + auto toFirstDayNumOfWeek(DateOrTime v, UInt8 week_mode) const { bool monday_first_mode = week_mode & static_cast(WeekModeFlag::MONDAY_FIRST); if (monday_first_mode) @@ -889,7 +889,7 @@ public: /// Get last day of week with week_mode, return Saturday or Sunday template - inline auto toLastDayNumOfWeek(DateOrTime v, UInt8 week_mode) const + auto toLastDayNumOfWeek(DateOrTime v, UInt8 week_mode) const { bool monday_first_mode = week_mode & static_cast(WeekModeFlag::MONDAY_FIRST); if (monday_first_mode) @@ -908,7 +908,7 @@ public: } /// Check and change mode to effective. - inline UInt8 check_week_mode(UInt8 mode) const /// NOLINT + UInt8 check_week_mode(UInt8 mode) const /// NOLINT { UInt8 week_format = (mode & 7); if (!(week_format & static_cast(WeekModeFlag::MONDAY_FIRST))) @@ -917,7 +917,7 @@ public: } /// Check and change mode to effective. - inline WeekDayMode check_week_day_mode(UInt8 mode) const /// NOLINT + WeekDayMode check_week_day_mode(UInt8 mode) const /// NOLINT { return static_cast(mode & 3); } @@ -926,7 +926,7 @@ public: * Returns 0 for monday, 1 for tuesday... */ template - inline UInt8 calc_weekday(DateOrTime v, bool sunday_first_day_of_week) const /// NOLINT + UInt8 calc_weekday(DateOrTime v, bool sunday_first_day_of_week) const /// NOLINT { const LUTIndex i = toLUTIndex(v); if (!sunday_first_day_of_week) @@ -936,28 +936,28 @@ public: } /// Calculate days in one year. - inline UInt16 calc_days_in_year(Int32 year) const /// NOLINT + UInt16 calc_days_in_year(Int32 year) const /// NOLINT { return ((year & 3) == 0 && (year % 100 || (year % 400 == 0 && year)) ? 366 : 365); } /// Number of month from some fixed moment in the past (year * 12 + month) template - inline Int32 toRelativeMonthNum(DateOrTime v) const + Int32 toRelativeMonthNum(DateOrTime v) const { const LUTIndex i = toLUTIndex(v); return lut[i].year * 12 + lut[i].month; } template - inline Int32 toRelativeQuarterNum(DateOrTime v) const + Int32 toRelativeQuarterNum(DateOrTime v) const { const LUTIndex i = toLUTIndex(v); return lut[i].year * 4 + (lut[i].month - 1) / 3; } /// We count all hour-length intervals, unrelated to offset changes. - inline Time toRelativeHourNum(Time t) const + Time toRelativeHourNum(Time t) const { if (t >= 0 && offset_is_whole_number_of_hours_during_epoch) return t / 3600; @@ -968,37 +968,37 @@ public: } template - inline Time toRelativeHourNum(DateOrTime v) const + Time toRelativeHourNum(DateOrTime v) const { return toRelativeHourNum(lut[toLUTIndex(v)].date); } /// The same formula is used for positive time (after Unix epoch) and negative time (before Unix epoch). /// It’s needed for correct work of dateDiff function. - inline Time toStableRelativeHourNum(Time t) const + Time toStableRelativeHourNum(Time t) const { return (t + DATE_LUT_ADD + 86400 - offset_at_start_of_epoch) / 3600 - (DATE_LUT_ADD / 3600); } template - inline Time toStableRelativeHourNum(DateOrTime v) const + Time toStableRelativeHourNum(DateOrTime v) const { return toStableRelativeHourNum(lut[toLUTIndex(v)].date); } - inline Time toRelativeMinuteNum(Time t) const /// NOLINT + Time toRelativeMinuteNum(Time t) const /// NOLINT { return (t + DATE_LUT_ADD) / 60 - (DATE_LUT_ADD / 60); } template - inline Time toRelativeMinuteNum(DateOrTime v) const + Time toRelativeMinuteNum(DateOrTime v) const { return toRelativeMinuteNum(lut[toLUTIndex(v)].date); } template - inline auto toStartOfYearInterval(DateOrTime v, UInt64 years) const + auto toStartOfYearInterval(DateOrTime v, UInt64 years) const { if (years == 1) return toFirstDayNumOfYear(v); @@ -1019,7 +1019,7 @@ public: template requires std::is_same_v || std::is_same_v - inline auto toStartOfQuarterInterval(Date d, UInt64 quarters) const + auto toStartOfQuarterInterval(Date d, UInt64 quarters) const { if (quarters == 1) return toFirstDayNumOfQuarter(d); @@ -1028,7 +1028,7 @@ public: template requires std::is_same_v || std::is_same_v - inline auto toStartOfMonthInterval(Date d, UInt64 months) const + auto toStartOfMonthInterval(Date d, UInt64 months) const { if (months == 1) return toFirstDayNumOfMonth(d); @@ -1042,7 +1042,7 @@ public: template requires std::is_same_v || std::is_same_v - inline auto toStartOfWeekInterval(Date d, UInt64 weeks) const + auto toStartOfWeekInterval(Date d, UInt64 weeks) const { if (weeks == 1) return toFirstDayNumOfWeek(d); @@ -1056,7 +1056,7 @@ public: template requires std::is_same_v || std::is_same_v - inline Time toStartOfDayInterval(Date d, UInt64 days) const + Time toStartOfDayInterval(Date d, UInt64 days) const { if (days == 1) return toDate(d); @@ -1152,7 +1152,7 @@ public: return static_cast(roundDown(t, seconds)); } - inline LUTIndex makeLUTIndex(Int16 year, UInt8 month, UInt8 day_of_month) const + LUTIndex makeLUTIndex(Int16 year, UInt8 month, UInt8 day_of_month) const { if (unlikely(year < DATE_LUT_MIN_YEAR || month < 1 || month > 12 || day_of_month < 1 || day_of_month > 31)) return LUTIndex(0); @@ -1167,7 +1167,7 @@ public: } /// Create DayNum from year, month, day of month. - inline ExtendedDayNum makeDayNum(Int16 year, UInt8 month, UInt8 day_of_month, Int32 default_error_day_num = 0) const + ExtendedDayNum makeDayNum(Int16 year, UInt8 month, UInt8 day_of_month, Int32 default_error_day_num = 0) const { if (unlikely(year < DATE_LUT_MIN_YEAR || month < 1 || month > 12 || day_of_month < 1 || day_of_month > 31)) return ExtendedDayNum(default_error_day_num); @@ -1175,14 +1175,14 @@ public: return toDayNum(makeLUTIndex(year, month, day_of_month)); } - inline Time makeDate(Int16 year, UInt8 month, UInt8 day_of_month) const + Time makeDate(Int16 year, UInt8 month, UInt8 day_of_month) const { return lut[makeLUTIndex(year, month, day_of_month)].date; } /** Does not accept daylight saving time as argument: in case of ambiguity, it choose greater timestamp. */ - inline Time makeDateTime(Int16 year, UInt8 month, UInt8 day_of_month, UInt8 hour, UInt8 minute, UInt8 second) const + Time makeDateTime(Int16 year, UInt8 month, UInt8 day_of_month, UInt8 hour, UInt8 minute, UInt8 second) const { size_t index = makeLUTIndex(year, month, day_of_month); Time time_offset = hour * 3600 + minute * 60 + second; @@ -1194,28 +1194,28 @@ public: } template - inline const Values & getValues(DateOrTime v) const { return lut[toLUTIndex(v)]; } + const Values & getValues(DateOrTime v) const { return lut[toLUTIndex(v)]; } template - inline UInt32 toNumYYYYMM(DateOrTime v) const + UInt32 toNumYYYYMM(DateOrTime v) const { const Values & values = getValues(v); return values.year * 100 + values.month; } template - inline UInt32 toNumYYYYMMDD(DateOrTime v) const + UInt32 toNumYYYYMMDD(DateOrTime v) const { const Values & values = getValues(v); return values.year * 10000 + values.month * 100 + values.day_of_month; } - inline Time YYYYMMDDToDate(UInt32 num) const /// NOLINT + Time YYYYMMDDToDate(UInt32 num) const /// NOLINT { return makeDate(num / 10000, num / 100 % 100, num % 100); } - inline ExtendedDayNum YYYYMMDDToDayNum(UInt32 num) const /// NOLINT + ExtendedDayNum YYYYMMDDToDayNum(UInt32 num) const /// NOLINT { return makeDayNum(num / 10000, num / 100 % 100, num % 100); } @@ -1241,13 +1241,13 @@ public: TimeComponents time; }; - inline DateComponents toDateComponents(Time t) const + DateComponents toDateComponents(Time t) const { const Values & values = getValues(t); return { values.year, values.month, values.day_of_month }; } - inline DateTimeComponents toDateTimeComponents(Time t) const + DateTimeComponents toDateTimeComponents(Time t) const { const LUTIndex index = findIndex(t); const Values & values = lut[index]; @@ -1283,12 +1283,12 @@ public: } template - inline DateTimeComponents toDateTimeComponents(DateOrTime v) const + DateTimeComponents toDateTimeComponents(DateOrTime v) const { return toDateTimeComponents(lut[toLUTIndex(v)].date); } - inline UInt64 toNumYYYYMMDDhhmmss(Time t) const + UInt64 toNumYYYYMMDDhhmmss(Time t) const { DateTimeComponents components = toDateTimeComponents(t); @@ -1301,7 +1301,7 @@ public: + UInt64(components.date.year) * 10000000000; } - inline Time YYYYMMDDhhmmssToTime(UInt64 num) const /// NOLINT + Time YYYYMMDDhhmmssToTime(UInt64 num) const /// NOLINT { return makeDateTime( num / 10000000000, @@ -1315,7 +1315,7 @@ public: /// Adding calendar intervals. /// Implementation specific behaviour when delta is too big. - inline NO_SANITIZE_UNDEFINED Time addDays(Time t, Int64 delta) const + NO_SANITIZE_UNDEFINED Time addDays(Time t, Int64 delta) const { const LUTIndex index = findIndex(t); const Values & values = lut[index]; @@ -1332,12 +1332,12 @@ public: return lut[new_index].date + time; } - inline NO_SANITIZE_UNDEFINED Time addWeeks(Time t, Int64 delta) const + NO_SANITIZE_UNDEFINED Time addWeeks(Time t, Int64 delta) const { return addDays(t, delta * 7); } - inline UInt8 saturateDayOfMonth(Int16 year, UInt8 month, UInt8 day_of_month) const + UInt8 saturateDayOfMonth(Int16 year, UInt8 month, UInt8 day_of_month) const { if (likely(day_of_month <= 28)) return day_of_month; @@ -1351,7 +1351,7 @@ public: } template - inline LUTIndex NO_SANITIZE_UNDEFINED addMonthsIndex(DateOrTime v, Int64 delta) const + LUTIndex NO_SANITIZE_UNDEFINED addMonthsIndex(DateOrTime v, Int64 delta) const { const Values & values = lut[toLUTIndex(v)]; @@ -1375,11 +1375,11 @@ public: } } - /// If resulting month has less deys than source month, then saturation can happen. + /// If resulting month has less days than source month, then saturation can happen. /// Example: 31 Aug + 1 month = 30 Sep. template requires std::is_same_v || std::is_same_v || std::is_same_v - inline Time NO_SANITIZE_UNDEFINED addMonths(DateTime t, Int64 delta) const + Time NO_SANITIZE_UNDEFINED addMonths(DateTime t, Int64 delta) const { const auto result_day = addMonthsIndex(t, delta); @@ -1405,7 +1405,7 @@ public: template requires std::is_same_v || std::is_same_v - inline auto NO_SANITIZE_UNDEFINED addMonths(Date d, Int64 delta) const + auto NO_SANITIZE_UNDEFINED addMonths(Date d, Int64 delta) const { if constexpr (std::is_same_v) return toDayNum(LUTIndexWithSaturation(addMonthsIndex(d, delta))); @@ -1414,13 +1414,13 @@ public: } template - inline auto NO_SANITIZE_UNDEFINED addQuarters(DateOrTime d, Int64 delta) const + auto NO_SANITIZE_UNDEFINED addQuarters(DateOrTime d, Int64 delta) const { return addMonths(d, delta * 3); } template - inline LUTIndex NO_SANITIZE_UNDEFINED addYearsIndex(DateOrTime v, Int64 delta) const + LUTIndex NO_SANITIZE_UNDEFINED addYearsIndex(DateOrTime v, Int64 delta) const { const Values & values = lut[toLUTIndex(v)]; @@ -1438,7 +1438,7 @@ public: /// Saturation can occur if 29 Feb is mapped to non-leap year. template requires std::is_same_v || std::is_same_v || std::is_same_v - inline Time addYears(DateTime t, Int64 delta) const + Time addYears(DateTime t, Int64 delta) const { auto result_day = addYearsIndex(t, delta); @@ -1464,7 +1464,7 @@ public: template requires std::is_same_v || std::is_same_v - inline auto addYears(Date d, Int64 delta) const + auto addYears(Date d, Int64 delta) const { if constexpr (std::is_same_v) return toDayNum(LUTIndexWithSaturation(addYearsIndex(d, delta))); @@ -1473,7 +1473,7 @@ public: } - inline std::string timeToString(Time t) const + std::string timeToString(Time t) const { DateTimeComponents components = toDateTimeComponents(t); @@ -1498,7 +1498,7 @@ public: return s; } - inline std::string dateToString(Time t) const + std::string dateToString(Time t) const { const Values & values = getValues(t); @@ -1516,7 +1516,7 @@ public: return s; } - inline std::string dateToString(ExtendedDayNum d) const + std::string dateToString(ExtendedDayNum d) const { const Values & values = getValues(d); diff --git a/src/Functions/makeDate.cpp b/src/Functions/makeDate.cpp index 1e4f3604c94..bb8c302d139 100644 --- a/src/Functions/makeDate.cpp +++ b/src/Functions/makeDate.cpp @@ -28,7 +28,9 @@ namespace ErrorCodes namespace { -/// Functions common to makeDate, makeDate32, makeDateTime, makeDateTime64 +/// Functionality common to +/// - makeDate, makeDate32, makeDateTime, makeDateTime64, +/// - YYYYMMDDToDate, YYYYMMDDToDate32, YYYYMMDDhhmmssToDateTime, YYYYMMDDhhmmssToDateTime64 class FunctionWithNumericParamsBase : public IFunction { public: @@ -48,11 +50,11 @@ public: size_t getNumberOfArguments() const override { return 0; } protected: - template + template Columns convertMandatoryArguments(const ColumnsWithTypeAndName & arguments, const ArgumentNames & argument_names) const { Columns converted_arguments; - const DataTypePtr converted_argument_type = std::make_shared(); + const DataTypePtr converted_argument_type = std::make_shared(); for (size_t i = 0; i < argument_names.size(); ++i) { ColumnPtr argument_column = castColumn(arguments[i], converted_argument_type); @@ -63,7 +65,7 @@ protected: } }; -/// Common implementation for makeDate, makeDate32 +/// Implementation of makeDate, makeDate32 template class FunctionMakeDate : public FunctionWithNumericParamsBase { @@ -72,7 +74,7 @@ private: static constexpr std::array mandatory_argument_names_year_dayofyear = {"year", "dayofyear"}; public: - static constexpr auto name = Traits::name; + static constexpr auto name = Traits::makeDateName; static FunctionPtr create(ContextPtr) { return std::make_shared(); } @@ -80,9 +82,9 @@ public: DataTypePtr getReturnTypeImpl(const ColumnsWithTypeAndName & arguments) const override { - const bool isYearMonthDayVariant = (arguments.size() == 3); + const bool is_year_month_variant = (arguments.size() == 3); - if (isYearMonthDayVariant) + if (is_year_month_variant) { FunctionArgumentDescriptors args{ {mandatory_argument_names_year_month_day[0], &isNumber, nullptr, "Number"}, @@ -105,10 +107,10 @@ public: ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override { - const bool isYearMonthDayVariant = (arguments.size() == 3); + const bool is_year_month_day_variant = (arguments.size() == 3); Columns converted_arguments; - if (isYearMonthDayVariant) + if (is_year_month_day_variant) converted_arguments = convertMandatoryArguments(arguments, mandatory_argument_names_year_month_day); else converted_arguments = convertMandatoryArguments(arguments, mandatory_argument_names_year_dayofyear); @@ -119,7 +121,7 @@ public: const auto & date_lut = DateLUT::instance(); const Int32 max_days_since_epoch = date_lut.makeDayNum(Traits::MAX_DATE[0], Traits::MAX_DATE[1], Traits::MAX_DATE[2]); - if (isYearMonthDayVariant) + if (is_year_month_day_variant) { const auto & year_data = typeid_cast(*converted_arguments[0]).getData(); const auto & month_data = typeid_cast(*converted_arguments[1]).getData(); @@ -133,8 +135,7 @@ public: Int32 day_num = 0; - if (year >= Traits::MIN_YEAR && - year <= Traits::MAX_YEAR && + if (year >= Traits::MIN_YEAR && year <= Traits::MAX_YEAR && month >= 1 && month <= 12 && day >= 1 && day <= 31) { @@ -158,8 +159,7 @@ public: Int32 day_num = 0; - if (year >= Traits::MIN_YEAR && - year <= Traits::MAX_YEAR && + if (year >= Traits::MIN_YEAR && year <= Traits::MAX_YEAR && dayofyear >= 1 && dayofyear <= 365) { Int32 days_since_epoch = date_lut.makeDayNum(static_cast(year), 1, 1) + static_cast(dayofyear) - 1; @@ -175,20 +175,87 @@ public: } }; -struct MakeDateTraits +/// Implementation of YYYYMMDDToDate, YYYYMMDDToDate32 +template +class FunctionYYYYYMMDDToDate : public FunctionWithNumericParamsBase { - static constexpr auto name = "makeDate"; +private: + static constexpr std::array mandatory_argument_names = { "YYYYMMDD" }; + +public: + static constexpr auto name = Traits::YYYYMMDDName; + + static FunctionPtr create(ContextPtr) { return std::make_shared(); } + + String getName() const override { return name; } + + bool isVariadic() const override { return false; } + size_t getNumberOfArguments() const override { return mandatory_argument_names.size(); } + + DataTypePtr getReturnTypeImpl(const ColumnsWithTypeAndName & arguments) const override + { + FunctionArgumentDescriptors args{ + {mandatory_argument_names[0], &isNumber, nullptr, "Number"} + }; + + validateFunctionArgumentTypes(*this, arguments, args); + + return std::make_shared(); + } + + ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override + { + Columns converted_arguments = convertMandatoryArguments(arguments, mandatory_argument_names); + + auto res_column = Traits::ReturnDataType::ColumnType::create(input_rows_count); + auto & result_data = res_column->getData(); + + const auto & yyyymmdd_data = typeid_cast(*converted_arguments[0]).getData(); + + const auto & date_lut = DateLUT::instance(); + const Int32 max_days_since_epoch = date_lut.makeDayNum(Traits::MAX_DATE[0], Traits::MAX_DATE[1], Traits::MAX_DATE[2]); + + for (size_t i = 0; i < input_rows_count; ++i) + { + const auto yyyymmdd = static_cast(yyyymmdd_data[i]); + + const auto year = yyyymmdd / 10'000; + const auto month = yyyymmdd / 100 % 100; + const auto day = yyyymmdd % 100; + + Int32 day_num = 0; + + if (year >= Traits::MIN_YEAR && year <= Traits::MAX_YEAR && + month >= 1 && month <= 12 && + day >= 1 && day <= 31) + { + Int32 days_since_epoch = date_lut.makeDayNum(static_cast(year), static_cast(month), static_cast(day)); + if (days_since_epoch <= max_days_since_epoch) + day_num = days_since_epoch; + } + + result_data[i] = day_num; + } + + return res_column; + } +}; + +struct DateTraits +{ + static constexpr auto makeDateName = "makeDate"; + static constexpr auto YYYYMMDDName = "YYYYMMDDToDate"; using ReturnDataType = DataTypeDate; static constexpr auto MIN_YEAR = 1970; static constexpr auto MAX_YEAR = 2149; - /// This date has the maximum day number that fits in 16-bit uint static constexpr std::array MAX_DATE = {MAX_YEAR, 6, 6}; }; -struct MakeDate32Traits +struct Date32Traits { - static constexpr auto name = "makeDate32"; + static constexpr auto makeDateName = "makeDate32"; + static constexpr auto YYYYMMDDName = "YYYYMMDDToDate32"; using ReturnDataType = DataTypeDate32; static constexpr auto MIN_YEAR = 1900; @@ -196,11 +263,11 @@ struct MakeDate32Traits static constexpr std::array MAX_DATE = {MAX_YEAR, 12, 31}; }; -/// Common implementation for makeDateTime, makeDateTime64 -class FunctionMakeDateTimeBase : public FunctionWithNumericParamsBase +/// Functionality common to makeDateTime, makeDateTime64, YYYYMMDDhhmmssToDateTime, YYYYMMDDhhmmssToDateTime64 +class FunctionDateTimeBase : public FunctionWithNumericParamsBase { protected: - static constexpr std::array mandatory_argument_names = {"year", "month", "day", "hour", "minute", "second"}; + static constexpr UInt32 DEFAULT_PRECISION = 3; template static Int64 dateTime(T year, T month, T day_of_month, T hour, T minute, T second, const DateLUTImpl & lut) @@ -233,14 +300,35 @@ protected: std::string extractTimezone(const ColumnWithTypeAndName & timezone_argument) const { - std::string timezone; if (!isStringOrFixedString(timezone_argument.type) || !timezone_argument.column || (timezone_argument.column->size() != 1 && !typeid_cast(timezone_argument.column.get()))) throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, "Argument 'timezone' for function {} must be const string", getName()); - timezone = timezone_argument.column->getDataAt(0).toString(); + + String timezone = timezone_argument.column->getDataAt(0).toString(); return timezone; } + + UInt32 extractPrecision(const ColumnWithTypeAndName & precision_argument) const + { + if (!isNumber(precision_argument.type) || !precision_argument.column || (precision_argument.column->size() != 1 && !typeid_cast(precision_argument.column.get()))) + throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, + "Argument 'precision' for function {} must be constant number", getName()); + + Int64 precision = precision_argument.column->getInt(0); + + if (precision < 0 || precision > 9) + throw Exception(ErrorCodes::ARGUMENT_OUT_OF_BOUND, + "Argument 'precision' for function {} must be in range [0, 9]", getName()); + + return static_cast(precision); + } +}; + +class FunctionMakeDateTimeBase : public FunctionDateTimeBase +{ +protected: + static constexpr std::array mandatory_argument_names = {"year", "month", "day", "hour", "minute", "second"}; }; /// makeDateTime(year, month, day, hour, minute, second, [timezone]) @@ -268,7 +356,7 @@ public: }; FunctionArgumentDescriptors optional_args{ - {optional_argument_names[0], &isString, nullptr, "String"} + {optional_argument_names[0], &isString, isColumnConst, "const String"} }; validateFunctionArgumentTypes(*this, arguments, mandatory_args, optional_args); @@ -329,7 +417,6 @@ class FunctionMakeDateTime64 : public FunctionMakeDateTimeBase { private: static constexpr std::array optional_argument_names = {"fraction", "precision", "timezone"}; - static constexpr UInt8 DEFAULT_PRECISION = 3; public: static constexpr auto name = "makeDateTime64"; @@ -350,9 +437,9 @@ public: }; FunctionArgumentDescriptors optional_args{ - {optional_argument_names[0], &isNumber, nullptr, "Number"}, - {optional_argument_names[1], &isNumber, nullptr, "Number"}, - {optional_argument_names[2], &isString, nullptr, "String"} + {optional_argument_names[0], &isNumber, isColumnConst, "const Number"}, + {optional_argument_names[1], &isNumber, isColumnConst, "const Number"}, + {optional_argument_names[2], &isString, isColumnConst, "const String"} }; validateFunctionArgumentTypes(*this, arguments, mandatory_args, optional_args); @@ -456,20 +543,179 @@ public: return res_column; } +}; +class FunctionYYYYMMDDhhmmssToDateTimeBase : public FunctionDateTimeBase +{ +protected: + static constexpr std::array mandatory_argument_names = { "YYYYMMDDhhmmss" }; +}; + +/// YYYYMMDDhhmmssToDateTime +class FunctionYYYYMMDDhhmmssToDateTime : public FunctionYYYYMMDDhhmmssToDateTimeBase +{ private: - UInt8 extractPrecision(const ColumnWithTypeAndName & precision_argument) const - { - Int64 precision = DEFAULT_PRECISION; - if (!isNumber(precision_argument.type) || !precision_argument.column || (precision_argument.column->size() != 1 && !typeid_cast(precision_argument.column.get()))) - throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, - "Argument 'precision' for function {} must be constant number", getName()); - precision = precision_argument.column->getInt(0); - if (precision < 0 || precision > 9) - throw Exception(ErrorCodes::ARGUMENT_OUT_OF_BOUND, - "Argument 'precision' for function {} must be in range [0, 9]", getName()); + static constexpr std::array optional_argument_names = { "timezone" }; - return precision; +public: + static constexpr auto name = "YYYYMMDDhhmmssToDateTime"; + + static FunctionPtr create(ContextPtr) { return std::make_shared(); } + + String getName() const override { return name; } + + DataTypePtr getReturnTypeImpl(const ColumnsWithTypeAndName & arguments) const override + { + FunctionArgumentDescriptors mandatory_args{ + {mandatory_argument_names[0], &isNumber, nullptr, "Number"} + }; + + FunctionArgumentDescriptors optional_args{ + {optional_argument_names[0], &isString, isColumnConst, "const String"} + }; + + validateFunctionArgumentTypes(*this, arguments, mandatory_args, optional_args); + + /// Optional timezone argument + std::string timezone; + if (arguments.size() == mandatory_argument_names.size() + 1) + timezone = extractTimezone(arguments.back()); + + return std::make_shared(timezone); + } + + ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override + { + std::string timezone; + if (arguments.size() == mandatory_argument_names.size() + 1) + timezone = extractTimezone(arguments.back()); + + Columns converted_arguments = convertMandatoryArguments(arguments, mandatory_argument_names); + + auto res_column = ColumnDateTime::create(input_rows_count); + auto & result_data = res_column->getData(); + + const auto & yyyymmddhhmmss_data = typeid_cast(*converted_arguments[0]).getData(); + + const auto & date_lut = DateLUT::instance(timezone); + + for (size_t i = 0; i < input_rows_count; i++) + { + const auto yyyymmddhhmmss = static_cast(yyyymmddhhmmss_data[i]); + + const auto yyyymmdd = yyyymmddhhmmss / 1'000'000; + const auto hhmmss = yyyymmddhhmmss % 1'000'000; + + const auto year = yyyymmdd / 10'000; + const auto month = yyyymmdd / 100 % 100; + const auto day = yyyymmdd % 100; + const auto hour = hhmmss / 10'000; + const auto minute = hhmmss / 100 % 100; + const auto second = hhmmss % 100; + + auto date_time = dateTime(year, month, day, hour, minute, second, date_lut); + + if (date_time < 0) [[unlikely]] + date_time = 0; + else if (date_time > 0x0ffffffffll) [[unlikely]] + date_time = 0x0ffffffffll; + + result_data[i] = static_cast(date_time); + } + + return res_column; + } +}; + +/// YYYYMMDDhhmmssToDateTime64 +class FunctionYYYYMMDDhhmmssToDateTime64 : public FunctionYYYYMMDDhhmmssToDateTimeBase +{ +private: + static constexpr std::array optional_argument_names = { "precision", "timezone" }; + +public: + static constexpr auto name = "YYYYMMDDhhmmssToDateTime64"; + + static FunctionPtr create(ContextPtr) { return std::make_shared(); } + + String getName() const override { return name; } + + DataTypePtr getReturnTypeImpl(const ColumnsWithTypeAndName & arguments) const override + { + FunctionArgumentDescriptors mandatory_args{ + {mandatory_argument_names[0], &isNumber, nullptr, "Number"} + }; + + FunctionArgumentDescriptors optional_args{ + {optional_argument_names[0], &isNumber, isColumnConst, "const Number"}, + {optional_argument_names[0], &isString, isColumnConst, "const String"} + }; + + validateFunctionArgumentTypes(*this, arguments, mandatory_args, optional_args); + + /// Optional precision argument + auto precision = DEFAULT_PRECISION; + if (arguments.size() >= mandatory_argument_names.size() + 1) + precision = extractPrecision(arguments[mandatory_argument_names.size()]); + + /// Optional timezone argument + std::string timezone; + if (arguments.size() == mandatory_argument_names.size() + 2) + timezone = extractTimezone(arguments.back()); + + return std::make_shared(precision, timezone); + } + + ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override + { + UInt32 precision = DEFAULT_PRECISION; + if (arguments.size() >= mandatory_argument_names.size() + 1) + precision = extractPrecision(arguments[mandatory_argument_names.size()]); + + std::string timezone; + if (arguments.size() == mandatory_argument_names.size() + 2) + timezone = extractTimezone(arguments.back()); + + Columns converted_arguments = convertMandatoryArguments(arguments, mandatory_argument_names); + + auto res_column = ColumnDateTime64::create(input_rows_count, static_cast(precision)); + auto & result_data = res_column->getData(); + + const auto & yyyymmddhhmmss_data = typeid_cast(*converted_arguments[0]).getData(); + + const auto & date_lut = DateLUT::instance(timezone); + + const auto fraction_pow = common::exp10_i32(precision); + + for (size_t i = 0; i < input_rows_count; i++) + { + const auto float_date = yyyymmddhhmmss_data[i]; + + const auto yyyymmddhhmmss = static_cast(float_date); + + const auto yyyymmdd = yyyymmddhhmmss / 1'000'000; + const auto hhmmss = yyyymmddhhmmss % 1'000'000; + + const auto decimal = float_date - yyyymmddhhmmss; + + const auto year = yyyymmdd / 10'000; + const auto month = yyyymmdd / 100 % 100; + const auto day = yyyymmdd % 100; + const auto hour = hhmmss / 10'000; + const auto minute = hhmmss / 100 % 100; + const auto second = hhmmss % 100; + + auto fraction = static_cast(decimal * fraction_pow); + + auto date_time = dateTime(year, month, day, hour, minute, second, date_lut); + + result_data[i] = DecimalUtils::decimalFromComponents( + date_time, + static_cast(fraction), + static_cast(precision)); + } + + return res_column; } }; @@ -477,10 +723,15 @@ private: REGISTER_FUNCTION(MakeDate) { - factory.registerFunction>({}, FunctionFactory::CaseInsensitive); - factory.registerFunction>(); + factory.registerFunction>({}, FunctionFactory::CaseInsensitive); + factory.registerFunction>(); factory.registerFunction(); factory.registerFunction(); + + factory.registerFunction>(); + factory.registerFunction>(); + factory.registerFunction(); + factory.registerFunction(); } } diff --git a/tests/queries/0_stateless/02245_make_datetime64.sql b/tests/queries/0_stateless/02245_make_datetime64.sql index a7b3a3d23c5..62784cb9b75 100644 --- a/tests/queries/0_stateless/02245_make_datetime64.sql +++ b/tests/queries/0_stateless/02245_make_datetime64.sql @@ -86,4 +86,4 @@ select makeDateTime64(year, 1, 1, 1, 0, 0, 0, precision, timezone) from ( select 1984 as year, 5 as precision, 'UTC' as timezone union all select 1985 as year, 5 as precision, 'UTC' as timezone -); -- { serverError 43 } +); -- { serverError 44 } diff --git a/tests/queries/0_stateless/02415_all_new_functions_must_be_documented.reference b/tests/queries/0_stateless/02415_all_new_functions_must_be_documented.reference index dd843058281..7928ac23438 100644 --- a/tests/queries/0_stateless/02415_all_new_functions_must_be_documented.reference +++ b/tests/queries/0_stateless/02415_all_new_functions_must_be_documented.reference @@ -56,6 +56,10 @@ URLHierarchy URLPathHierarchy UUIDNumToString UUIDStringToNum +YYYYMMDDToDate +YYYYMMDDToDate32 +YYYYMMDDhhmmssToDateTime +YYYYMMDDhhmmssToDateTime64 _CAST __bitBoolMaskAnd __bitBoolMaskOr diff --git a/tests/queries/0_stateless/02876_yyyymmddhhmmsstodatetime.reference b/tests/queries/0_stateless/02876_yyyymmddhhmmsstodatetime.reference new file mode 100644 index 00000000000..9d010535451 --- /dev/null +++ b/tests/queries/0_stateless/02876_yyyymmddhhmmsstodatetime.reference @@ -0,0 +1,85 @@ +--- YYYYMMDDToDateTime +Invalid input types are rejected +Result type is DateTime +DateTime +Nullable(DateTime) +Check correctness, integer arguments +1970-01-02 11:59:59 +1970-01-01 00:00:00 +2020-02-29 11:11:11 +2023-09-11 15:05:05 +2106-02-07 06:28:15 +2106-02-07 06:28:15 +Check correctness, float arguments +1970-01-02 11:59:59 +1970-01-01 00:00:00 +2020-02-29 11:11:11 +2023-09-11 15:05:05 +2106-02-07 06:28:15 +2106-02-07 06:28:15 +Check correctness, decimal arguments +1970-01-02 11:59:59 +1970-01-01 00:00:00 +2020-02-29 11:11:11 +2023-09-11 15:05:05 +2106-02-07 06:28:15 +2106-02-07 06:28:15 +Special cases +1970-01-01 00:00:00 +1970-01-01 00:00:00 +1970-01-01 00:00:00 +1970-01-01 00:00:00 +2106-02-07 06:28:15 +1970-01-01 00:00:00 +1970-01-01 00:00:00 +1970-01-01 00:00:00 +1970-01-01 00:00:00 +2023-02-28 11:11:11 +1970-01-01 00:00:00 +1970-01-01 00:00:00 +\N +1970-01-01 00:00:00 +1970-01-01 00:00:00 +--- YYYYMMDDToDateTime64 +Invalid input types are rejected +Result type is DateTime +DateTime64(3) +DateTime64(5) +Nullable(DateTime64(3)) +Check correctness, integer arguments +1900-01-01 00:00:00.000 +1900-01-01 00:00:00.000 +2020-02-29 11:11:11.000 +2023-09-11 15:05:05.000 +2299-12-31 23:59:59.000 +2299-12-31 23:59:59.000 +Check correctness, float arguments +1900-01-01 00:00:00.900 +1900-01-01 00:00:00.899 +2020-02-29 11:11:11.101 +2023-09-11 15:05:05.101 +2299-12-31 23:59:59.101 +2299-12-31 23:59:59.101 +Check correctness, decimal arguments +1900-01-01 00:00:00.900 +1900-01-01 00:00:00.899 +2020-02-29 11:11:11.101 +2023-09-11 15:05:05.101 +2299-12-31 23:59:59.101 +2299-12-31 23:59:59.101 +Special cases +1900-01-01 00:00:00.648 +1900-01-01 00:00:00.000 +1900-01-01 00:00:00.000 +1900-01-01 00:00:00.000 +2299-12-31 23:59:59.000 +1900-01-01 00:00:00.000 +1900-01-01 00:00:00.000 +1900-01-01 00:00:00.000 +1900-01-01 00:00:00.000 +2023-02-28 11:11:11.000 +1900-01-01 00:00:00.000 +1900-01-01 00:00:00.000 +\N +1900-01-01 00:00:00.000 +1900-01-01 00:00:00.000 diff --git a/tests/queries/0_stateless/02876_yyyymmddhhmmsstodatetime.sql b/tests/queries/0_stateless/02876_yyyymmddhhmmsstodatetime.sql new file mode 100644 index 00000000000..1b5f6220b50 --- /dev/null +++ b/tests/queries/0_stateless/02876_yyyymmddhhmmsstodatetime.sql @@ -0,0 +1,119 @@ +SET session_timezone = 'UTC'; -- no time zone randomization, please + +----------------------------------------------------------- +SELECT '--- YYYYMMDDToDateTime'; + +SELECT 'Invalid input types are rejected'; +SELECT YYYYMMDDhhmmssToDateTime(); -- { serverError NUMBER_OF_ARGUMENTS_DOESNT_MATCH } +SELECT YYYYMMDDhhmmssToDateTime(toDate('2023-09-11')); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT YYYYMMDDhhmmssToDateTime(toDate32('2023-09-11')); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT YYYYMMDDhhmmssToDateTime(toDateTime('2023-09-11 12:18:00')); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT YYYYMMDDhhmmssToDateTime(toDateTime64('2023-09-11 12:18:00', 3)); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT YYYYMMDDhhmmssToDateTime('2023-09-11'); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT YYYYMMDDhhmmssToDateTime(20230911134254, 3); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT YYYYMMDDhhmmssToDateTime(20230911134254, 'invalid tz'); -- { serverError BAD_ARGUMENTS } +SELECT YYYYMMDDhhmmssToDateTime(20230911134254, 'Europe/Berlin', 'bad'); -- { serverError NUMBER_OF_ARGUMENTS_DOESNT_MATCH } + +SELECT 'Result type is DateTime'; +SELECT toTypeName(YYYYMMDDhhmmssToDateTime(19910824)); +SELECT toTypeName(YYYYMMDDhhmmssToDateTime(cast(19910824 AS Nullable(UInt64)))); +-- +SELECT 'Check correctness, integer arguments'; +SELECT YYYYMMDDhhmmssToDateTime(19691231595959); +SELECT YYYYMMDDhhmmssToDateTime(19700101000000); +SELECT YYYYMMDDhhmmssToDateTime(20200229111111); -- leap day +SELECT YYYYMMDDhhmmssToDateTime(20230911150505); +SELECT YYYYMMDDhhmmssToDateTime(21060207062815); +SELECT YYYYMMDDhhmmssToDateTime(21060207062816); + +SELECT 'Check correctness, float arguments'; +SELECT YYYYMMDDhhmmssToDateTime(19691231595959.1); +SELECT YYYYMMDDhhmmssToDateTime(19700101000000.1); +SELECT YYYYMMDDhhmmssToDateTime(20200229111111.1); -- leap day +SELECT YYYYMMDDhhmmssToDateTime(20230911150505.1); +SELECT YYYYMMDDhhmmssToDateTime(21060207062815.1); +SELECT YYYYMMDDhhmmssToDateTime(21060207062816.1); + +SELECT 'Check correctness, decimal arguments'; +SELECT YYYYMMDDhhmmssToDateTime(toDecimal64(19691231595959.1, 5)); +SELECT YYYYMMDDhhmmssToDateTime(toDecimal64(19700101000000.1, 5)); +SELECT YYYYMMDDhhmmssToDateTime(toDecimal64(20200229111111.1, 5)); -- leap day +SELECT YYYYMMDDhhmmssToDateTime(toDecimal64(20230911150505.1, 5)); +SELECT YYYYMMDDhhmmssToDateTime(toDecimal64(21060207062815.1, 5)); +SELECT YYYYMMDDhhmmssToDateTime(toDecimal64(21060207062816.1, 5)); + +SELECT 'Special cases'; +SELECT YYYYMMDDhhmmssToDateTime(-20230911111111); -- negative +SELECT YYYYMMDDhhmmssToDateTime(110); -- invalid everything +SELECT YYYYMMDDhhmmssToDateTime(999999999999999999); -- huge value +SELECT YYYYMMDDhhmmssToDateTime(15001113111111); -- year out of range +SELECT YYYYMMDDhhmmssToDateTime(35001113111111); -- year out of range +SELECT YYYYMMDDhhmmssToDateTime(20231620111111); -- invalid month +SELECT YYYYMMDDhhmmssToDateTime(20230020111111); -- invalid month +SELECT YYYYMMDDhhmmssToDateTime(20230940111111); -- invalid day +SELECT YYYYMMDDhhmmssToDateTime(20230900111111); -- invalid day +SELECT YYYYMMDDhhmmssToDateTime(20230228111111); -- leap day when there is none +SELECT YYYYMMDDhhmmssToDateTime(True); +SELECT YYYYMMDDhhmmssToDateTime(False); +SELECT YYYYMMDDhhmmssToDateTime(NULL); +SELECT YYYYMMDDhhmmssToDateTime(yyyymmdd) FROM (SELECT 19840121 AS yyyymmdd UNION ALL SELECT 20230911 AS yyyymmdd) ORDER BY yyyymmdd; -- non-const + +----------------------------------------------------------- +SELECT '--- YYYYMMDDToDateTime64'; + +SELECT 'Invalid input types are rejected'; +SELECT YYYYMMDDhhmmssToDateTime64(); -- { serverError NUMBER_OF_ARGUMENTS_DOESNT_MATCH } +SELECT YYYYMMDDhhmmssToDateTime64(toDate('2023-09-11')); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT YYYYMMDDhhmmssToDateTime64(toDate32('2023-09-11')); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT YYYYMMDDhhmmssToDateTime64(toDateTime('2023-09-11 12:18:00')); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT YYYYMMDDhhmmssToDateTime64(toDateTime64('2023-09-11 12:18:00', 3)); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT YYYYMMDDhhmmssToDateTime64('2023-09-11'); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT YYYYMMDDhhmmssToDateTime64('2023-09-11', 'invalid precision'); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT YYYYMMDDhhmmssToDateTime64(20230911134254, 3, 3); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT YYYYMMDDhhmmssToDateTime64(20230911134254, 3, 'invalid tz'); -- { serverError BAD_ARGUMENTS } +SELECT YYYYMMDDhhmmssToDateTime64(20230911134254, 3, 'Europe/Berlin', 'no more args'); -- { serverError NUMBER_OF_ARGUMENTS_DOESNT_MATCH } + +SELECT 'Result type is DateTime'; +SELECT toTypeName(YYYYMMDDhhmmssToDateTime64(19910824)); +SELECT toTypeName(YYYYMMDDhhmmssToDateTime64(19910824, 5)); +SELECT toTypeName(YYYYMMDDhhmmssToDateTime64(cast(19910824 AS Nullable(UInt64)))); + +SELECT 'Check correctness, integer arguments'; +SELECT YYYYMMDDhhmmssToDateTime64(189912315959); +SELECT YYYYMMDDhhmmssToDateTime64(19000101000000); +SELECT YYYYMMDDhhmmssToDateTime64(20200229111111); -- leap day +SELECT YYYYMMDDhhmmssToDateTime64(20230911150505); +SELECT YYYYMMDDhhmmssToDateTime64(22991231235959); +SELECT YYYYMMDDhhmmssToDateTime64(23000101000000); + +SELECT 'Check correctness, float arguments'; +SELECT YYYYMMDDhhmmssToDateTime64(189912315959.1); +SELECT YYYYMMDDhhmmssToDateTime64(19000101000000.1); +SELECT YYYYMMDDhhmmssToDateTime64(20200229111111.1); -- leap day +SELECT YYYYMMDDhhmmssToDateTime64(20230911150505.1); +SELECT YYYYMMDDhhmmssToDateTime64(22991231235959.1); +SELECT YYYYMMDDhhmmssToDateTime64(23000101000000.1); + +SELECT 'Check correctness, decimal arguments'; +SELECT YYYYMMDDhhmmssToDateTime64(toDecimal64(189912315959.1, 5)); +SELECT YYYYMMDDhhmmssToDateTime64(toDecimal64(19000101000000.1, 5)); +SELECT YYYYMMDDhhmmssToDateTime64(toDecimal64(20200229111111.1, 5)); -- leap day +SELECT YYYYMMDDhhmmssToDateTime64(toDecimal64(20230911150505.1, 5)); +SELECT YYYYMMDDhhmmssToDateTime64(toDecimal64(22991231235959.1, 5)); +SELECT YYYYMMDDhhmmssToDateTime64(toDecimal64(23000101000000.1, 5)); + +SELECT 'Special cases'; +SELECT YYYYMMDDhhmmssToDateTime64(-20230911111111); -- negative +SELECT YYYYMMDDhhmmssToDateTime64(110); -- invalid everything +SELECT YYYYMMDDhhmmssToDateTime64(999999999999999999); -- huge value +SELECT YYYYMMDDhhmmssToDateTime64(15001113111111); -- year out of range +SELECT YYYYMMDDhhmmssToDateTime64(35001113111111); -- year out of range +SELECT YYYYMMDDhhmmssToDateTime64(20231620111111); -- invalid month +SELECT YYYYMMDDhhmmssToDateTime64(20230020111111); -- invalid month +SELECT YYYYMMDDhhmmssToDateTime64(20230940111111); -- invalid day +SELECT YYYYMMDDhhmmssToDateTime64(20230900111111); -- invalid day +SELECT YYYYMMDDhhmmssToDateTime64(20230228111111); -- leap day when there is none +SELECT YYYYMMDDhhmmssToDateTime64(True); +SELECT YYYYMMDDhhmmssToDateTime64(False); +SELECT YYYYMMDDhhmmssToDateTime64(NULL); +SELECT YYYYMMDDhhmmssToDateTime64(yyyymmdd) FROM (SELECT 19840121 AS yyyymmdd UNION ALL SELECT 20230911 AS yyyymmdd) ORDER BY yyyymmdd; -- non-const diff --git a/tests/queries/0_stateless/02876_yyyymmddtodate.reference b/tests/queries/0_stateless/02876_yyyymmddtodate.reference new file mode 100644 index 00000000000..e3c6a9e2d7c --- /dev/null +++ b/tests/queries/0_stateless/02876_yyyymmddtodate.reference @@ -0,0 +1,84 @@ +--- YYYYMMDDToDate +Invalid input types are rejected +Result type is Date +Date +Nullable(Date) +Check correctness, integer arguments +1970-01-01 +1970-01-01 +2020-02-29 +2023-09-11 +2149-06-06 +1970-01-01 +Check correctness, float arguments +1970-01-01 +1970-01-01 +2020-02-29 +2023-09-11 +2149-06-06 +1970-01-01 +Check correctness, decimal arguments +1970-01-01 +1970-01-01 +2020-02-29 +2023-09-11 +2149-06-06 +1970-01-01 +Special cases +1970-01-01 +1970-01-01 +1970-01-01 +1970-01-01 +1970-01-01 +1970-01-01 +1970-01-01 +1970-01-01 +1970-01-01 +2023-02-28 +1970-01-01 +1970-01-01 +\N +1984-01-21 +2023-09-11 +--- YYYYMMDDToDate32 +Invalid input types are rejected +Result type is Date32 +Date32 +Nullable(Date32) +Check correctness, integer arguments +1970-01-01 +1900-01-01 +2020-02-29 +2023-09-11 +2299-12-31 +1970-01-01 +Check correctness, float arguments +1970-01-01 +1900-01-01 +2020-02-29 +2023-09-11 +2299-12-31 +1970-01-01 +Check correctness, decimal arguments +1970-01-01 +1900-01-01 +2020-02-29 +2023-09-11 +2299-12-31 +1970-01-01 +Special cases +1970-01-01 +1970-01-01 +1970-01-01 +1970-01-01 +1970-01-01 +1970-01-01 +1970-01-01 +1970-01-01 +1970-01-01 +2023-02-28 +1970-01-01 +1970-01-01 +\N +1984-01-21 +2023-09-11 diff --git a/tests/queries/0_stateless/02876_yyyymmddtodate.sql b/tests/queries/0_stateless/02876_yyyymmddtodate.sql new file mode 100644 index 00000000000..fbacca1e2a2 --- /dev/null +++ b/tests/queries/0_stateless/02876_yyyymmddtodate.sql @@ -0,0 +1,112 @@ +----------------------------------------------------------- +SELECT '--- YYYYMMDDToDate'; + +SELECT 'Invalid input types are rejected'; +SELECT YYYYMMDDToDate(); -- { serverError NUMBER_OF_ARGUMENTS_DOESNT_MATCH } +SELECT YYYYMMDDToDate(toDate('2023-09-11')); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT YYYYMMDDToDate(toDate32('2023-09-11')); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT YYYYMMDDToDate(toDateTime('2023-09-11 12:18:00')); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT YYYYMMDDToDate(toDateTime64('2023-09-11 12:18:00', 3)); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT YYYYMMDDToDate('2023-09-11'); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT YYYYMMDDToDate(2023, 09, 11); -- { serverError NUMBER_OF_ARGUMENTS_DOESNT_MATCH } +SELECT YYYYMMDDToDate(2023, 110); -- { serverError NUMBER_OF_ARGUMENTS_DOESNT_MATCH } + +SELECT 'Result type is Date'; +SELECT toTypeName(YYYYMMDDToDate(19910824)); +SELECT toTypeName(YYYYMMDDToDate(cast(19910824 AS Nullable(UInt64)))); + +SELECT 'Check correctness, integer arguments'; +SELECT YYYYMMDDToDate(19691231); +SELECT YYYYMMDDToDate(19700101); +SELECT YYYYMMDDToDate(20200229); -- leap day +SELECT YYYYMMDDToDate(20230911); +SELECT YYYYMMDDToDate(21490606); +SELECT YYYYMMDDToDate(21490607); + +SELECT 'Check correctness, float arguments'; +SELECT YYYYMMDDToDate(19691231.1); +SELECT YYYYMMDDToDate(19700101.1); +SELECT YYYYMMDDToDate(20200229.1); -- leap day +SELECT YYYYMMDDToDate(20230911.1); +SELECT YYYYMMDDToDate(21490606.1); +SELECT YYYYMMDDToDate(21490607.1); + +SELECT 'Check correctness, decimal arguments'; +SELECT YYYYMMDDToDate(toDecimal64(19691231.1, 5)); +SELECT YYYYMMDDToDate(toDecimal64(19700101.1, 5)); +SELECT YYYYMMDDToDate(toDecimal64(20200229.1, 5)); -- leap day +SELECT YYYYMMDDToDate(toDecimal64(20230911.1, 5)); +SELECT YYYYMMDDToDate(toDecimal64(21490606.1, 5)); +SELECT YYYYMMDDToDate(toDecimal64(21490607.1, 5)); + +SELECT 'Special cases'; +SELECT YYYYMMDDToDate(-20230911); -- negative +SELECT YYYYMMDDToDate(110); -- invalid everything +SELECT YYYYMMDDToDate(9999999999999); -- huge value +SELECT YYYYMMDDToDate(15001113); -- year out of range +SELECT YYYYMMDDToDate(35001113); -- year out of range +SELECT YYYYMMDDToDate(20231620); -- invalid month +SELECT YYYYMMDDToDate(20230020); -- invalid month +SELECT YYYYMMDDToDate(20230940); -- invalid day +SELECT YYYYMMDDToDate(20230900); -- invalid day +SELECT YYYYMMDDToDate(20230228); -- leap day when there is none +SELECT YYYYMMDDToDate(True); +SELECT YYYYMMDDToDate(False); +SELECT YYYYMMDDToDate(NULL); +SELECT YYYYMMDDToDate(yyyymmdd) FROM (SELECT 19840121 AS yyyymmdd UNION ALL SELECT 20230911 AS yyyymmdd) ORDER BY yyyymmdd; -- non-const + +----------------------------------------------------------- +SELECT '--- YYYYMMDDToDate32'; + +SELECT 'Invalid input types are rejected'; +SELECT YYYYMMDDToDate32(toDate('2023-09-11')); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT YYYYMMDDToDate32(toDate32('2023-09-11')); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT YYYYMMDDToDate32(toDateTime('2023-09-11 12:18:00')); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT YYYYMMDDToDate32(toDateTime64('2023-09-11 12:18:00', 3)); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT YYYYMMDDToDate32('2023-09-11'); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT YYYYMMDDToDate32(2023, 09, 11); -- { serverError NUMBER_OF_ARGUMENTS_DOESNT_MATCH } +SELECT YYYYMMDDToDate32(2023, 110); -- { serverError NUMBER_OF_ARGUMENTS_DOESNT_MATCH } + +SELECT 'Result type is Date32'; +SELECT toTypeName(YYYYMMDDToDate32(19910824)); +SELECT toTypeName(YYYYMMDDToDate32(cast(19910824 AS Nullable(UInt64)))); + +SELECT 'Check correctness, integer arguments'; +SELECT YYYYMMDDToDate32(18991231); +SELECT YYYYMMDDToDate32(19000101); +SELECT YYYYMMDDToDate32(20200229); -- leap day +SELECT YYYYMMDDToDate32(20230911); +SELECT YYYYMMDDToDate32(22991231); +SELECT YYYYMMDDToDate32(23000101); + +SELECT 'Check correctness, float arguments'; +SELECT YYYYMMDDToDate32(18991231.1); +SELECT YYYYMMDDToDate32(19000101.1); +SELECT YYYYMMDDToDate32(20200229.1); -- leap day +SELECT YYYYMMDDToDate32(20230911.1); +SELECT YYYYMMDDToDate32(22991231.1); +SELECT YYYYMMDDToDate32(23000101.1); + +SELECT 'Check correctness, decimal arguments'; +SELECT YYYYMMDDToDate32(toDecimal64(18991231.1, 5)); +SELECT YYYYMMDDToDate32(toDecimal64(19000101.1, 5)); +SELECT YYYYMMDDToDate32(toDecimal64(20200229.1, 5)); -- leap day +SELECT YYYYMMDDToDate32(toDecimal64(20230911.1, 5)); +SELECT YYYYMMDDToDate32(toDecimal64(22991231.1, 5)); +SELECT YYYYMMDDToDate32(toDecimal64(23000101.1, 5)); + +SELECT 'Special cases'; +SELECT YYYYMMDDToDate32(-20230911); -- negative +SELECT YYYYMMDDToDate32(110); -- invalid everything +SELECT YYYYMMDDToDate32(9999999999999); -- huge value +SELECT YYYYMMDDToDate32(15001113); -- year out of range +SELECT YYYYMMDDToDate32(35001113); -- year out of range +SELECT YYYYMMDDToDate32(20231620); -- invalid month +SELECT YYYYMMDDToDate32(20230020); -- invalid month +SELECT YYYYMMDDToDate32(20230940); -- invalid day +SELECT YYYYMMDDToDate32(20230900); -- invalid day +SELECT YYYYMMDDToDate32(20230228); -- leap day when there is none +SELECT YYYYMMDDToDate32(True); +SELECT YYYYMMDDToDate32(False); +SELECT YYYYMMDDToDate32(NULL); +SELECT YYYYMMDDToDate32(yyyymmdd) FROM (SELECT 19840121 AS yyyymmdd UNION ALL SELECT 20230911 AS yyyymmdd) ORDER BY yyyymmdd; -- non-const From 61fdb63aad07084bf918a676f96bf9a17d31acf9 Mon Sep 17 00:00:00 2001 From: vdimir Date: Mon, 11 Sep 2023 13:53:20 +0000 Subject: [PATCH 277/433] fix --- tests/ci/commit_status_helper.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/tests/ci/commit_status_helper.py b/tests/ci/commit_status_helper.py index ed686dae11c..c3b1a4d0f95 100644 --- a/tests/ci/commit_status_helper.py +++ b/tests/ci/commit_status_helper.py @@ -148,9 +148,9 @@ def update_pr_title_icon(pr: PullRequest, status: str) -> None: new_title = pr.title if new_title and new_title[0] != new_status_icon: if new_title[0] in set(STATUS_ICON_MAP.values()): - new_title = new_title[1:] + new_status_icon + new_title = new_status_icon + new_title[1:] else: - new_title = new_title + " " + new_status_icon + new_title = new_status_icon + " " + new_title pr.edit(title=new_title) From ec37e934611f7ad91f28baf2f382f2c30a5529be Mon Sep 17 00:00:00 2001 From: Anton Popov Date: Mon, 11 Sep 2023 14:52:59 +0000 Subject: [PATCH 278/433] fix use-after-free in MergeTreePrefetchedReadPool --- src/Storages/MergeTree/MergeTreePrefetchedReadPool.cpp | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/src/Storages/MergeTree/MergeTreePrefetchedReadPool.cpp b/src/Storages/MergeTree/MergeTreePrefetchedReadPool.cpp index 465dcd6a7ab..75f2fd26600 100644 --- a/src/Storages/MergeTree/MergeTreePrefetchedReadPool.cpp +++ b/src/Storages/MergeTree/MergeTreePrefetchedReadPool.cpp @@ -61,6 +61,12 @@ MergeTreeReadTask::Readers MergeTreePrefetchedReadPool::PrefetechedReaders::get( { SCOPE_EXIT({ is_valid = false; }); ProfileEventTimeIncrement watch(ProfileEvents::WaitPrefetchTaskMicroseconds); + + /// First wait for completion of all futures. + for (auto & prefetch_future : prefetch_futures) + prefetch_future.wait(); + + /// Then rethrow first exception if any. for (auto & prefetch_future : prefetch_futures) prefetch_future.get(); From 2d8f33bfa2663a174c3d2f932525f1c4ea9cfed9 Mon Sep 17 00:00:00 2001 From: avogar Date: Mon, 11 Sep 2023 14:55:37 +0000 Subject: [PATCH 279/433] Fix parsing error in WithNames formats while reading subset of columns with disabled input_format_with_names_use_header --- src/Analyzer/Passes/QueryAnalysisPass.cpp | 4 ++-- src/Formats/FormatFactory.cpp | 17 +++++++++++++---- src/Formats/FormatFactory.h | 11 ++++++++--- src/Formats/registerWithNamesAndTypes.cpp | 5 +++-- src/Interpreters/Context.cpp | 4 ++-- src/Storages/HDFS/StorageHDFS.cpp | 6 +++--- src/Storages/HDFS/StorageHDFS.h | 2 +- src/Storages/Hive/StorageHive.h | 2 +- src/Storages/IStorage.h | 2 -- src/Storages/S3Queue/StorageS3Queue.cpp | 8 ++++---- src/Storages/S3Queue/StorageS3Queue.h | 2 +- src/Storages/StorageAzureBlob.cpp | 6 +++--- src/Storages/StorageAzureBlob.h | 2 +- src/Storages/StorageFile.cpp | 6 +++--- src/Storages/StorageFile.h | 2 +- src/Storages/StorageS3.cpp | 6 +++--- src/Storages/StorageS3.h | 2 +- src/Storages/StorageURL.cpp | 8 ++++---- src/Storages/StorageURL.h | 2 +- src/Storages/StorageXDBC.cpp | 2 +- src/Storages/StorageXDBC.h | 2 +- src/TableFunctions/ITableFunction.h | 2 +- src/TableFunctions/ITableFunctionFileLike.cpp | 4 ++-- src/TableFunctions/ITableFunctionFileLike.h | 2 +- .../TableFunctionAzureBlobStorage.cpp | 4 ++-- .../TableFunctionAzureBlobStorage.h | 2 +- src/TableFunctions/TableFunctionS3.cpp | 4 ++-- src/TableFunctions/TableFunctionS3.h | 2 +- ...formats_with_names_dont_use_header.reference | 1 + .../02876_formats_with_names_dont_use_header.sh | 10 ++++++++++ ...with_names_dont_use_header_test.csvwithnames | 2 ++ 31 files changed, 80 insertions(+), 54 deletions(-) create mode 100644 tests/queries/0_stateless/02876_formats_with_names_dont_use_header.reference create mode 100755 tests/queries/0_stateless/02876_formats_with_names_dont_use_header.sh create mode 100644 tests/queries/0_stateless/02876_formats_with_names_dont_use_header_test.csvwithnames diff --git a/src/Analyzer/Passes/QueryAnalysisPass.cpp b/src/Analyzer/Passes/QueryAnalysisPass.cpp index 348189854e8..9bb9308a609 100644 --- a/src/Analyzer/Passes/QueryAnalysisPass.cpp +++ b/src/Analyzer/Passes/QueryAnalysisPass.cpp @@ -6341,9 +6341,9 @@ void QueryAnalyzer::resolveTableFunction(QueryTreeNodePtr & table_function_node, { /// For input function we should check if input format supports reading subset of columns. if (table_function_ptr->getName() == "input") - use_columns_from_insert_query = FormatFactory::instance().checkIfFormatSupportsSubsetOfColumns(scope.context->getInsertFormat()); + use_columns_from_insert_query = FormatFactory::instance().checkIfFormatSupportsSubsetOfColumns(scope.context->getInsertFormat(), scope.context); else - use_columns_from_insert_query = table_function_ptr->supportsReadingSubsetOfColumns(); + use_columns_from_insert_query = table_function_ptr->supportsReadingSubsetOfColumns(scope.context); } if (use_columns_from_insert_query) diff --git a/src/Formats/FormatFactory.cpp b/src/Formats/FormatFactory.cpp index c5a4cf7a5c6..6481cc2d125 100644 --- a/src/Formats/FormatFactory.cpp +++ b/src/Formats/FormatFactory.cpp @@ -684,10 +684,18 @@ void FormatFactory::markOutputFormatSupportsParallelFormatting(const String & na void FormatFactory::markFormatSupportsSubsetOfColumns(const String & name) { - auto & target = dict[name].supports_subset_of_columns; + auto & target = dict[name].subset_of_columns_support_checker; if (target) throw Exception(ErrorCodes::LOGICAL_ERROR, "FormatFactory: Format {} is already marked as supporting subset of columns", name); - target = true; + target = [](const FormatSettings &){ return true; }; +} + +void FormatFactory::registerSubsetOfColumnsSupportChecker(const String & name, SubsetOfColumnsSupportChecker subset_of_columns_support_checker) +{ + auto & target = dict[name].subset_of_columns_support_checker; + if (target) + throw Exception(ErrorCodes::LOGICAL_ERROR, "FormatFactory: Format {} is already marked as supporting subset of columns", name); + target = std::move(subset_of_columns_support_checker); } void FormatFactory::markOutputFormatPrefersLargeBlocks(const String & name) @@ -698,10 +706,11 @@ void FormatFactory::markOutputFormatPrefersLargeBlocks(const String & name) target = true; } -bool FormatFactory::checkIfFormatSupportsSubsetOfColumns(const String & name) const +bool FormatFactory::checkIfFormatSupportsSubsetOfColumns(const DB::String & name, const ContextPtr & context, const std::optional & format_settings_) const { const auto & target = getCreators(name); - return target.supports_subset_of_columns; + auto format_settings = format_settings_ ? *format_settings_ : getFormatSettings(context); + return target.subset_of_columns_support_checker && target.subset_of_columns_support_checker(format_settings); } void FormatFactory::registerAdditionalInfoForSchemaCacheGetter( diff --git a/src/Formats/FormatFactory.h b/src/Formats/FormatFactory.h index 2034e6446fb..9b6eb97456a 100644 --- a/src/Formats/FormatFactory.h +++ b/src/Formats/FormatFactory.h @@ -123,6 +123,10 @@ private: /// and the name of the message. using AdditionalInfoForSchemaCacheGetter = std::function; + /// Some formats can support reading subset of columns depending on settings. + /// The checker should return true if format support append. + using SubsetOfColumnsSupportChecker = std::function; + struct Creators { InputCreator input_creator; @@ -133,11 +137,11 @@ private: ExternalSchemaReaderCreator external_schema_reader_creator; bool supports_parallel_formatting{false}; bool supports_subcolumns{false}; - bool supports_subset_of_columns{false}; bool prefers_large_blocks{false}; NonTrivialPrefixAndSuffixChecker non_trivial_prefix_and_suffix_checker; AppendSupportChecker append_support_checker; AdditionalInfoForSchemaCacheGetter additional_info_for_schema_cache_getter; + SubsetOfColumnsSupportChecker subset_of_columns_support_checker; }; using FormatsDictionary = std::unordered_map; @@ -225,9 +229,10 @@ public: void markOutputFormatSupportsParallelFormatting(const String & name); void markOutputFormatPrefersLargeBlocks(const String & name); - void markFormatSupportsSubsetOfColumns(const String & name); - bool checkIfFormatSupportsSubsetOfColumns(const String & name) const; + void markFormatSupportsSubsetOfColumns(const String & name); + void registerSubsetOfColumnsSupportChecker(const String & name, SubsetOfColumnsSupportChecker subset_of_columns_support_checker); + bool checkIfFormatSupportsSubsetOfColumns(const String & name, const ContextPtr & context, const std::optional & format_settings_ = std::nullopt) const; bool checkIfFormatHasSchemaReader(const String & name) const; bool checkIfFormatHasExternalSchemaReader(const String & name) const; diff --git a/src/Formats/registerWithNamesAndTypes.cpp b/src/Formats/registerWithNamesAndTypes.cpp index 2dee107844d..674865a3bed 100644 --- a/src/Formats/registerWithNamesAndTypes.cpp +++ b/src/Formats/registerWithNamesAndTypes.cpp @@ -12,8 +12,9 @@ void registerWithNamesAndTypes(const std::string & base_format_name, RegisterWit void markFormatWithNamesAndTypesSupportsSamplingColumns(const std::string & base_format_name, FormatFactory & factory) { - factory.markFormatSupportsSubsetOfColumns(base_format_name + "WithNames"); - factory.markFormatSupportsSubsetOfColumns(base_format_name + "WithNamesAndTypes"); + auto setting_checker = [](const FormatSettings & settings){ return settings.with_names_use_header; }; + factory.registerSubsetOfColumnsSupportChecker(base_format_name + "WithNames", setting_checker); + factory.registerSubsetOfColumnsSupportChecker(base_format_name + "WithNamesAndTypes", setting_checker); } } diff --git a/src/Interpreters/Context.cpp b/src/Interpreters/Context.cpp index ee7efdfeb1b..8695669a7de 100644 --- a/src/Interpreters/Context.cpp +++ b/src/Interpreters/Context.cpp @@ -1684,9 +1684,9 @@ StoragePtr Context::executeTableFunction(const ASTPtr & table_expression, const { /// For input function we should check if input format supports reading subset of columns. if (table_function_ptr->getName() == "input") - use_columns_from_insert_query = FormatFactory::instance().checkIfFormatSupportsSubsetOfColumns(getInsertFormat()); + use_columns_from_insert_query = FormatFactory::instance().checkIfFormatSupportsSubsetOfColumns(getInsertFormat(), shared_from_this()); else - use_columns_from_insert_query = table_function_ptr->supportsReadingSubsetOfColumns(); + use_columns_from_insert_query = table_function_ptr->supportsReadingSubsetOfColumns(shared_from_this()); } if (use_columns_from_insert_query) diff --git a/src/Storages/HDFS/StorageHDFS.cpp b/src/Storages/HDFS/StorageHDFS.cpp index d2830d00748..7d144814803 100644 --- a/src/Storages/HDFS/StorageHDFS.cpp +++ b/src/Storages/HDFS/StorageHDFS.cpp @@ -838,9 +838,9 @@ private: }; -bool StorageHDFS::supportsSubsetOfColumns() const +bool StorageHDFS::supportsSubsetOfColumns(const ContextPtr & context_) const { - return FormatFactory::instance().checkIfFormatSupportsSubsetOfColumns(format_name); + return FormatFactory::instance().checkIfFormatSupportsSubsetOfColumns(format_name, context_); } Pipe StorageHDFS::read( @@ -878,7 +878,7 @@ Pipe StorageHDFS::read( }); } - auto read_from_format_info = prepareReadingFromFormat(column_names, storage_snapshot, supportsSubsetOfColumns(), getVirtuals()); + auto read_from_format_info = prepareReadingFromFormat(column_names, storage_snapshot, supportsSubsetOfColumns(context_), getVirtuals()); bool need_only_count = (query_info.optimize_trivial_count || read_from_format_info.requested_columns.empty()) && context_->getSettingsRef().optimize_count_from_files; diff --git a/src/Storages/HDFS/StorageHDFS.h b/src/Storages/HDFS/StorageHDFS.h index 8d9976a6a35..ffbf4e93ff9 100644 --- a/src/Storages/HDFS/StorageHDFS.h +++ b/src/Storages/HDFS/StorageHDFS.h @@ -76,7 +76,7 @@ public: /// Is is useful because column oriented formats could effectively skip unknown columns /// So we can create a header of only required columns in read method and ask /// format to read only them. Note: this hack cannot be done with ordinary formats like TSV. - bool supportsSubsetOfColumns() const override; + bool supportsSubsetOfColumns(const ContextPtr & context_) const; bool supportsSubcolumns() const override { return true; } diff --git a/src/Storages/Hive/StorageHive.h b/src/Storages/Hive/StorageHive.h index 604df70f4d0..a3c47d400e2 100644 --- a/src/Storages/Hive/StorageHive.h +++ b/src/Storages/Hive/StorageHive.h @@ -65,7 +65,7 @@ public: NamesAndTypesList getVirtuals() const override; - bool supportsSubsetOfColumns() const override; + bool supportsSubsetOfColumns() const; std::optional totalRows(const Settings & settings) const override; std::optional totalRowsByPartitionPredicate(const SelectQueryInfo & query_info, ContextPtr context_) const override; diff --git a/src/Storages/IStorage.h b/src/Storages/IStorage.h index a6d990a3eff..fcf7675d15d 100644 --- a/src/Storages/IStorage.h +++ b/src/Storages/IStorage.h @@ -620,8 +620,6 @@ public: /// NOTE: write-once also does not support INSERTs/merges/... for MergeTree virtual bool isStaticStorage() const; - virtual bool supportsSubsetOfColumns() const { return false; } - /// If it is possible to quickly determine exact number of rows in the table at this moment of time, then return it. /// Used for: /// - Simple count() optimization diff --git a/src/Storages/S3Queue/StorageS3Queue.cpp b/src/Storages/S3Queue/StorageS3Queue.cpp index 08cbff96cd0..ee840ca2ba1 100644 --- a/src/Storages/S3Queue/StorageS3Queue.cpp +++ b/src/Storages/S3Queue/StorageS3Queue.cpp @@ -162,9 +162,9 @@ bool StorageS3Queue::supportsSubcolumns() const return true; } -bool StorageS3Queue::supportsSubsetOfColumns() const +bool StorageS3Queue::supportsSubsetOfColumns(const ContextPtr & context_) const { - return FormatFactory::instance().checkIfFormatSupportsSubsetOfColumns(configuration.format); + return FormatFactory::instance().checkIfFormatSupportsSubsetOfColumns(configuration.format, context_, format_settings); } Pipe StorageS3Queue::read( @@ -187,7 +187,7 @@ Pipe StorageS3Queue::read( std::shared_ptr iterator_wrapper = createFileIterator(local_context, query_info.query); - auto read_from_format_info = prepareReadingFromFormat(column_names, storage_snapshot, supportsSubsetOfColumns(), getVirtuals()); + auto read_from_format_info = prepareReadingFromFormat(column_names, storage_snapshot, supportsSubsetOfColumns(local_context), getVirtuals()); const size_t max_download_threads = local_context->getSettingsRef().max_download_threads; @@ -363,7 +363,7 @@ void StorageS3Queue::streamToViews() // Create a stream for each consumer and join them in a union stream std::shared_ptr iterator_wrapper = createFileIterator(s3queue_context, nullptr); - auto read_from_format_info = prepareReadingFromFormat(column_names, storage_snapshot, supportsSubsetOfColumns(), getVirtuals()); + auto read_from_format_info = prepareReadingFromFormat(column_names, storage_snapshot, supportsSubsetOfColumns(getContext()), getVirtuals()); const size_t max_download_threads = s3queue_context->getSettingsRef().max_download_threads; auto pipe = Pipe(std::make_shared( diff --git a/src/Storages/S3Queue/StorageS3Queue.h b/src/Storages/S3Queue/StorageS3Queue.h index 712fe9e530b..98d46e80ad2 100644 --- a/src/Storages/S3Queue/StorageS3Queue.h +++ b/src/Storages/S3Queue/StorageS3Queue.h @@ -125,7 +125,7 @@ private: }; std::shared_ptr task; - bool supportsSubsetOfColumns() const override; + bool supportsSubsetOfColumns(const ContextPtr & context_) const; const UInt32 zk_create_table_retries = 1000; bool createTableIfNotExists(const StorageMetadataPtr & metadata_snapshot); diff --git a/src/Storages/StorageAzureBlob.cpp b/src/Storages/StorageAzureBlob.cpp index fb790c50e52..bdefe781ee6 100644 --- a/src/Storages/StorageAzureBlob.cpp +++ b/src/Storages/StorageAzureBlob.cpp @@ -687,7 +687,7 @@ Pipe StorageAzureBlob::read( query_info.query, virtual_columns, local_context, nullptr, local_context->getFileProgressCallback()); } - auto read_from_format_info = prepareReadingFromFormat(column_names, storage_snapshot, supportsSubsetOfColumns(), getVirtuals()); + auto read_from_format_info = prepareReadingFromFormat(column_names, storage_snapshot, supportsSubsetOfColumns(local_context), getVirtuals()); bool need_only_count = (query_info.optimize_trivial_count || read_from_format_info.requested_columns.empty()) && local_context->getSettingsRef().optimize_count_from_files; @@ -792,9 +792,9 @@ bool StorageAzureBlob::supportsPartitionBy() const return true; } -bool StorageAzureBlob::supportsSubsetOfColumns() const +bool StorageAzureBlob::supportsSubsetOfColumns(const ContextPtr & context) const { - return FormatFactory::instance().checkIfFormatSupportsSubsetOfColumns(configuration.format); + return FormatFactory::instance().checkIfFormatSupportsSubsetOfColumns(configuration.format, context, format_settings); } bool StorageAzureBlob::prefersLargeBlocks() const diff --git a/src/Storages/StorageAzureBlob.h b/src/Storages/StorageAzureBlob.h index 706817670d2..448d8e20e05 100644 --- a/src/Storages/StorageAzureBlob.h +++ b/src/Storages/StorageAzureBlob.h @@ -99,7 +99,7 @@ public: bool supportsSubcolumns() const override { return true; } - bool supportsSubsetOfColumns() const override; + bool supportsSubsetOfColumns(const ContextPtr & context) const; bool supportsTrivialCountOptimization() const override { return true; } diff --git a/src/Storages/StorageFile.cpp b/src/Storages/StorageFile.cpp index 8fd888f571d..0e154803602 100644 --- a/src/Storages/StorageFile.cpp +++ b/src/Storages/StorageFile.cpp @@ -803,9 +803,9 @@ ColumnsDescription StorageFile::getTableStructureFromFile( return columns; } -bool StorageFile::supportsSubsetOfColumns() const +bool StorageFile::supportsSubsetOfColumns(const ContextPtr & context) const { - return format_name != "Distributed" && FormatFactory::instance().checkIfFormatSupportsSubsetOfColumns(format_name); + return format_name != "Distributed" && FormatFactory::instance().checkIfFormatSupportsSubsetOfColumns(format_name, context, format_settings); } bool StorageFile::prefersLargeBlocks() const @@ -1433,7 +1433,7 @@ Pipe StorageFile::read( if (progress_callback && !archive_info) progress_callback(FileProgress(0, total_bytes_to_read)); - auto read_from_format_info = prepareReadingFromFormat(column_names, storage_snapshot, supportsSubsetOfColumns(), getVirtuals()); + auto read_from_format_info = prepareReadingFromFormat(column_names, storage_snapshot, supportsSubsetOfColumns(context), getVirtuals()); bool need_only_count = (query_info.optimize_trivial_count || read_from_format_info.requested_columns.empty()) && context->getSettingsRef().optimize_count_from_files; diff --git a/src/Storages/StorageFile.h b/src/Storages/StorageFile.h index 97c3cebabbb..f1464b90ab4 100644 --- a/src/Storages/StorageFile.h +++ b/src/Storages/StorageFile.h @@ -74,7 +74,7 @@ public: /// Is is useful because such formats could effectively skip unknown columns /// So we can create a header of only required columns in read method and ask /// format to read only them. Note: this hack cannot be done with ordinary formats like TSV. - bool supportsSubsetOfColumns() const override; + bool supportsSubsetOfColumns(const ContextPtr & context) const; bool supportsSubcolumns() const override { return true; } diff --git a/src/Storages/StorageS3.cpp b/src/Storages/StorageS3.cpp index 24b2279bfdc..341d8b3f768 100644 --- a/src/Storages/StorageS3.cpp +++ b/src/Storages/StorageS3.cpp @@ -983,9 +983,9 @@ std::shared_ptr StorageS3::createFileIterator( } } -bool StorageS3::supportsSubsetOfColumns() const +bool StorageS3::supportsSubsetOfColumns(const ContextPtr & context) const { - return FormatFactory::instance().checkIfFormatSupportsSubsetOfColumns(configuration.format); + return FormatFactory::instance().checkIfFormatSupportsSubsetOfColumns(configuration.format, context, format_settings); } bool StorageS3::prefersLargeBlocks() const @@ -1017,7 +1017,7 @@ Pipe StorageS3::read( std::shared_ptr iterator_wrapper = createFileIterator( query_configuration, distributed_processing, local_context, query_info.query, virtual_columns, nullptr, local_context->getFileProgressCallback()); - auto read_from_format_info = prepareReadingFromFormat(column_names, storage_snapshot, supportsSubsetOfColumns(), getVirtuals()); + auto read_from_format_info = prepareReadingFromFormat(column_names, storage_snapshot, supportsSubsetOfColumns(local_context), getVirtuals()); bool need_only_count = (query_info.optimize_trivial_count || read_from_format_info.requested_columns.empty()) && local_context->getSettingsRef().optimize_count_from_files; diff --git a/src/Storages/StorageS3.h b/src/Storages/StorageS3.h index a7dba400f37..ee03b9f18c2 100644 --- a/src/Storages/StorageS3.h +++ b/src/Storages/StorageS3.h @@ -388,7 +388,7 @@ private: bool supportsSubcolumns() const override { return true; } - bool supportsSubsetOfColumns() const override; + bool supportsSubsetOfColumns(const ContextPtr & context) const; bool prefersLargeBlocks() const override; diff --git a/src/Storages/StorageURL.cpp b/src/Storages/StorageURL.cpp index fcc9a3bbeb2..a0f5379a1fd 100644 --- a/src/Storages/StorageURL.cpp +++ b/src/Storages/StorageURL.cpp @@ -817,9 +817,9 @@ ColumnsDescription IStorageURLBase::getTableStructureFromData( return columns; } -bool IStorageURLBase::supportsSubsetOfColumns() const +bool IStorageURLBase::supportsSubsetOfColumns(const ContextPtr & context) const { - return FormatFactory::instance().checkIfFormatSupportsSubsetOfColumns(format_name); + return FormatFactory::instance().checkIfFormatSupportsSubsetOfColumns(format_name, context, format_settings); } bool IStorageURLBase::prefersLargeBlocks() const @@ -846,7 +846,7 @@ Pipe IStorageURLBase::read( std::shared_ptr iterator_wrapper{nullptr}; bool is_url_with_globs = urlWithGlobs(uri); size_t max_addresses = local_context->getSettingsRef().glob_expansion_max_elements; - auto read_from_format_info = prepareReadingFromFormat(column_names, storage_snapshot, supportsSubsetOfColumns(), getVirtuals()); + auto read_from_format_info = prepareReadingFromFormat(column_names, storage_snapshot, supportsSubsetOfColumns(local_context), getVirtuals()); if (distributed_processing) { @@ -951,7 +951,7 @@ Pipe StorageURLWithFailover::read( return uri_options; }); - auto read_from_format_info = prepareReadingFromFormat(column_names, storage_snapshot, supportsSubsetOfColumns(), getVirtuals()); + auto read_from_format_info = prepareReadingFromFormat(column_names, storage_snapshot, supportsSubsetOfColumns(local_context), getVirtuals()); const size_t max_threads = local_context->getSettingsRef().max_threads; const size_t max_parsing_threads = num_streams >= max_threads ? 1 : (max_threads / num_streams); diff --git a/src/Storages/StorageURL.h b/src/Storages/StorageURL.h index 504b0d5de40..6c5c50af326 100644 --- a/src/Storages/StorageURL.h +++ b/src/Storages/StorageURL.h @@ -114,7 +114,7 @@ protected: QueryProcessingStage::Enum & processed_stage, size_t max_block_size) const; - bool supportsSubsetOfColumns() const override; + virtual bool supportsSubsetOfColumns(const ContextPtr & context) const; bool prefersLargeBlocks() const override; diff --git a/src/Storages/StorageXDBC.cpp b/src/Storages/StorageXDBC.cpp index 1715cde9d1e..0ba8838d4c3 100644 --- a/src/Storages/StorageXDBC.cpp +++ b/src/Storages/StorageXDBC.cpp @@ -146,7 +146,7 @@ SinkToStoragePtr StorageXDBC::write(const ASTPtr & /* query */, const StorageMet compression_method); } -bool StorageXDBC::supportsSubsetOfColumns() const +bool StorageXDBC::supportsSubsetOfColumns(const ContextPtr &) const { return true; } diff --git a/src/Storages/StorageXDBC.h b/src/Storages/StorageXDBC.h index d7a1138c710..1c1651cb333 100644 --- a/src/Storages/StorageXDBC.h +++ b/src/Storages/StorageXDBC.h @@ -68,7 +68,7 @@ private: Block getHeaderBlock(const Names & column_names, const StorageSnapshotPtr & storage_snapshot) const override; - bool supportsSubsetOfColumns() const override; + bool supportsSubsetOfColumns(const ContextPtr &) const override; }; } diff --git a/src/TableFunctions/ITableFunction.h b/src/TableFunctions/ITableFunction.h index 028bbb99c15..1946d8e8905 100644 --- a/src/TableFunctions/ITableFunction.h +++ b/src/TableFunctions/ITableFunction.h @@ -76,7 +76,7 @@ public: /// because we cannot determine which column from table correspond to this virtual column. virtual std::unordered_set getVirtualsToCheckBeforeUsingStructureHint() const { return {}; } - virtual bool supportsReadingSubsetOfColumns() { return true; } + virtual bool supportsReadingSubsetOfColumns(const ContextPtr &) { return true; } /// Create storage according to the query. StoragePtr diff --git a/src/TableFunctions/ITableFunctionFileLike.cpp b/src/TableFunctions/ITableFunctionFileLike.cpp index 487826dc363..b88af855309 100644 --- a/src/TableFunctions/ITableFunctionFileLike.cpp +++ b/src/TableFunctions/ITableFunctionFileLike.cpp @@ -32,9 +32,9 @@ String ITableFunctionFileLike::getFormatFromFirstArgument() return FormatFactory::instance().getFormatFromFileName(filename, true); } -bool ITableFunctionFileLike::supportsReadingSubsetOfColumns() +bool ITableFunctionFileLike::supportsReadingSubsetOfColumns(const ContextPtr & context) { - return FormatFactory::instance().checkIfFormatSupportsSubsetOfColumns(format); + return FormatFactory::instance().checkIfFormatSupportsSubsetOfColumns(format, context); } void ITableFunctionFileLike::parseArguments(const ASTPtr & ast_function, ContextPtr context) diff --git a/src/TableFunctions/ITableFunctionFileLike.h b/src/TableFunctions/ITableFunctionFileLike.h index 902c7b01c4d..5fe86587797 100644 --- a/src/TableFunctions/ITableFunctionFileLike.h +++ b/src/TableFunctions/ITableFunctionFileLike.h @@ -27,7 +27,7 @@ public: void setStructureHint(const ColumnsDescription & structure_hint_) override { structure_hint = structure_hint_; } - bool supportsReadingSubsetOfColumns() override; + bool supportsReadingSubsetOfColumns(const ContextPtr & context) override; static size_t getMaxNumberOfArguments() { return 4; } diff --git a/src/TableFunctions/TableFunctionAzureBlobStorage.cpp b/src/TableFunctions/TableFunctionAzureBlobStorage.cpp index a4ece2f6a07..c29bea2c5c7 100644 --- a/src/TableFunctions/TableFunctionAzureBlobStorage.cpp +++ b/src/TableFunctions/TableFunctionAzureBlobStorage.cpp @@ -269,9 +269,9 @@ ColumnsDescription TableFunctionAzureBlobStorage::getActualTableStructure(Contex return parseColumnsListFromString(configuration.structure, context); } -bool TableFunctionAzureBlobStorage::supportsReadingSubsetOfColumns() +bool TableFunctionAzureBlobStorage::supportsReadingSubsetOfColumns(const ContextPtr & context) { - return FormatFactory::instance().checkIfFormatSupportsSubsetOfColumns(configuration.format); + return FormatFactory::instance().checkIfFormatSupportsSubsetOfColumns(configuration.format, context); } StoragePtr TableFunctionAzureBlobStorage::executeImpl(const ASTPtr & /*ast_function*/, ContextPtr context, const std::string & table_name, ColumnsDescription /*cached_columns*/, bool is_insert_query) const diff --git a/src/TableFunctions/TableFunctionAzureBlobStorage.h b/src/TableFunctions/TableFunctionAzureBlobStorage.h index 83c4449fd17..e1759740f7f 100644 --- a/src/TableFunctions/TableFunctionAzureBlobStorage.h +++ b/src/TableFunctions/TableFunctionAzureBlobStorage.h @@ -49,7 +49,7 @@ public: void setStructureHint(const ColumnsDescription & structure_hint_) override { structure_hint = structure_hint_; } - bool supportsReadingSubsetOfColumns() override; + bool supportsReadingSubsetOfColumns(const ContextPtr & context) override; std::unordered_set getVirtualsToCheckBeforeUsingStructureHint() const override { diff --git a/src/TableFunctions/TableFunctionS3.cpp b/src/TableFunctions/TableFunctionS3.cpp index 0ae3a19dc7f..f26550d0f70 100644 --- a/src/TableFunctions/TableFunctionS3.cpp +++ b/src/TableFunctions/TableFunctionS3.cpp @@ -325,9 +325,9 @@ ColumnsDescription TableFunctionS3::getActualTableStructure(ContextPtr context, return parseColumnsListFromString(configuration.structure, context); } -bool TableFunctionS3::supportsReadingSubsetOfColumns() +bool TableFunctionS3::supportsReadingSubsetOfColumns(const ContextPtr & context) { - return FormatFactory::instance().checkIfFormatSupportsSubsetOfColumns(configuration.format); + return FormatFactory::instance().checkIfFormatSupportsSubsetOfColumns(configuration.format, context); } StoragePtr TableFunctionS3::executeImpl(const ASTPtr & /*ast_function*/, ContextPtr context, const std::string & table_name, ColumnsDescription /*cached_columns*/, bool /*is_insert_query*/) const diff --git a/src/TableFunctions/TableFunctionS3.h b/src/TableFunctions/TableFunctionS3.h index ac932260792..a38ea5ba56b 100644 --- a/src/TableFunctions/TableFunctionS3.h +++ b/src/TableFunctions/TableFunctionS3.h @@ -47,7 +47,7 @@ public: void setStructureHint(const ColumnsDescription & structure_hint_) override { structure_hint = structure_hint_; } - bool supportsReadingSubsetOfColumns() override; + bool supportsReadingSubsetOfColumns(const ContextPtr & context) override; std::unordered_set getVirtualsToCheckBeforeUsingStructureHint() const override { diff --git a/tests/queries/0_stateless/02876_formats_with_names_dont_use_header.reference b/tests/queries/0_stateless/02876_formats_with_names_dont_use_header.reference new file mode 100644 index 00000000000..0cfbf08886f --- /dev/null +++ b/tests/queries/0_stateless/02876_formats_with_names_dont_use_header.reference @@ -0,0 +1 @@ +2 diff --git a/tests/queries/0_stateless/02876_formats_with_names_dont_use_header.sh b/tests/queries/0_stateless/02876_formats_with_names_dont_use_header.sh new file mode 100755 index 00000000000..ce06ff530b9 --- /dev/null +++ b/tests/queries/0_stateless/02876_formats_with_names_dont_use_header.sh @@ -0,0 +1,10 @@ +#!/usr/bin/env bash + +CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) +# shellcheck source=../shell_config.sh +. "$CURDIR"/../shell_config.sh + +echo -e "a,b,c\n1,2,3" > $CLICKHOUSE_TEST_UNIQUE_NAME.csvwithnames + +$CLICKHOUSE_LOCAL -q "select b from file('$CLICKHOUSE_TEST_UNIQUE_NAME.csvwithnames') settings input_format_with_names_use_header=0" + diff --git a/tests/queries/0_stateless/02876_formats_with_names_dont_use_header_test.csvwithnames b/tests/queries/0_stateless/02876_formats_with_names_dont_use_header_test.csvwithnames new file mode 100644 index 00000000000..bfde6bfa0b8 --- /dev/null +++ b/tests/queries/0_stateless/02876_formats_with_names_dont_use_header_test.csvwithnames @@ -0,0 +1,2 @@ +a,b,c +1,2,3 From c270ebfe5e65bf5cc526d813a1a36a99212a2fa4 Mon Sep 17 00:00:00 2001 From: Dmitry Novik Date: Mon, 11 Sep 2023 14:56:29 +0000 Subject: [PATCH 280/433] Update reference file --- .../0_stateless/01165_lost_part_empty_partition.reference | 2 ++ 1 file changed, 2 insertions(+) diff --git a/tests/queries/0_stateless/01165_lost_part_empty_partition.reference b/tests/queries/0_stateless/01165_lost_part_empty_partition.reference index 6ed281c757a..1af4bf8965c 100644 --- a/tests/queries/0_stateless/01165_lost_part_empty_partition.reference +++ b/tests/queries/0_stateless/01165_lost_part_empty_partition.reference @@ -1,2 +1,4 @@ 1 1 +1 +2 From b69a90d61f1cefc97a0c45a703b1e837119765b3 Mon Sep 17 00:00:00 2001 From: Dmitry Novik Date: Mon, 11 Sep 2023 14:57:25 +0000 Subject: [PATCH 281/433] Change log message level --- src/Storages/StorageReplicatedMergeTree.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/Storages/StorageReplicatedMergeTree.cpp b/src/Storages/StorageReplicatedMergeTree.cpp index aac85b4afc0..0de89dff8ed 100644 --- a/src/Storages/StorageReplicatedMergeTree.cpp +++ b/src/Storages/StorageReplicatedMergeTree.cpp @@ -989,7 +989,7 @@ void StorageReplicatedMergeTree::drop() { UInt64 lost_part_count = lost_part_count_str.empty() ? 0 : parse(lost_part_count_str); if (lost_part_count > 0) - LOG_DEBUG(log, "Dropping table with non-zero lost_part_count equal to {}", lost_part_count); + LOG_INFO(log, "Dropping table with non-zero lost_part_count equal to {}", lost_part_count); } dropReplica(zookeeper, zookeeper_path, replica_name, log, getSettings(), &has_metadata_in_zookeeper); } From b5cccc5f8d12752bdc713429ef2191d52d25f10b Mon Sep 17 00:00:00 2001 From: avogar Date: Mon, 11 Sep 2023 14:58:02 +0000 Subject: [PATCH 282/433] Remove unused field --- src/Formats/FormatFactory.h | 1 - 1 file changed, 1 deletion(-) diff --git a/src/Formats/FormatFactory.h b/src/Formats/FormatFactory.h index 9b6eb97456a..48a1869d563 100644 --- a/src/Formats/FormatFactory.h +++ b/src/Formats/FormatFactory.h @@ -136,7 +136,6 @@ private: SchemaReaderCreator schema_reader_creator; ExternalSchemaReaderCreator external_schema_reader_creator; bool supports_parallel_formatting{false}; - bool supports_subcolumns{false}; bool prefers_large_blocks{false}; NonTrivialPrefixAndSuffixChecker non_trivial_prefix_and_suffix_checker; AppendSupportChecker append_support_checker; From 27ac4be8e2bc0511bc5fdd2da0ca9d4c1a8d0e42 Mon Sep 17 00:00:00 2001 From: Robert Schulze Date: Mon, 11 Sep 2023 15:30:06 +0000 Subject: [PATCH 283/433] Fix style --- utils/check-style/aspell-ignore/en/aspell-dict.txt | 2 ++ 1 file changed, 2 insertions(+) diff --git a/utils/check-style/aspell-ignore/en/aspell-dict.txt b/utils/check-style/aspell-ignore/en/aspell-dict.txt index 6e26a8c8d17..5d85c86ee79 100644 --- a/utils/check-style/aspell-ignore/en/aspell-dict.txt +++ b/utils/check-style/aspell-ignore/en/aspell-dict.txt @@ -969,6 +969,8 @@ Xeon YAML YAMLRegExpTree YYYY +YYYYMMDDToDate +YYYYMMDDhhmmssToDateTime Yandex Yasm Zabbix From a1ee226969890d2f31015b732ab146d41a38aa18 Mon Sep 17 00:00:00 2001 From: vdimir Date: Mon, 11 Sep 2023 16:13:11 +0000 Subject: [PATCH 284/433] Update PR status icons --- tests/ci/commit_status_helper.py | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/tests/ci/commit_status_helper.py b/tests/ci/commit_status_helper.py index c3b1a4d0f95..df78366b9ab 100644 --- a/tests/ci/commit_status_helper.py +++ b/tests/ci/commit_status_helper.py @@ -133,10 +133,10 @@ def post_commit_status( STATUS_ICON_MAP = defaultdict( str, { - ERROR: "🔴", - FAILURE: "🔴", - PENDING: "🟡", - SUCCESS: "🟢", + ERROR: "❌", + FAILURE: "❌", + PENDING: "⏳", + SUCCESS: "✅", }, ) From 960ca5ebdee8fbe3210af923972396fa36d5b22a Mon Sep 17 00:00:00 2001 From: vdimir Date: Mon, 11 Sep 2023 16:13:33 +0000 Subject: [PATCH 285/433] Show PR status in label --- tests/ci/commit_status_helper.py | 21 +++++++++------------ 1 file changed, 9 insertions(+), 12 deletions(-) diff --git a/tests/ci/commit_status_helper.py b/tests/ci/commit_status_helper.py index df78366b9ab..eb9247c2dc4 100644 --- a/tests/ci/commit_status_helper.py +++ b/tests/ci/commit_status_helper.py @@ -141,17 +141,14 @@ STATUS_ICON_MAP = defaultdict( ) -def update_pr_title_icon(pr: PullRequest, status: str) -> None: - new_status_icon = STATUS_ICON_MAP[status] - if not new_status_icon: - return - new_title = pr.title - if new_title and new_title[0] != new_status_icon: - if new_title[0] in set(STATUS_ICON_MAP.values()): - new_title = new_status_icon + new_title[1:] - else: - new_title = new_status_icon + " " + new_title - pr.edit(title=new_title) +def update_pr_status_label(pr: PullRequest, status: str) -> None: + new_label = "pr-status-" + STATUS_ICON_MAP[status] + for label in pr.get_labels(): + if label.name == new_label: + return + if label.name.startswith("pr-status-"): + label.delete() + pr.add_to_labels(new_label) def set_status_comment(commit: Commit, pr_info: PRInfo) -> None: @@ -193,7 +190,7 @@ def set_status_comment(commit: Commit, pr_info: PRInfo) -> None: comment = ic break - update_pr_title_icon(pr, get_worst_state(statuses)) + update_pr_status_label(pr, get_worst_state(statuses)) if comment is None: pr.create_issue_comment(comment_body) From 4a80a0f1c07db495879e94ab1953f171995b1308 Mon Sep 17 00:00:00 2001 From: vdimir Date: Mon, 11 Sep 2023 16:13:52 +0000 Subject: [PATCH 286/433] upd --- tests/ci/commit_status_helper.py | 16 +++++++--------- tests/ci/report.py | 2 +- 2 files changed, 8 insertions(+), 10 deletions(-) diff --git a/tests/ci/commit_status_helper.py b/tests/ci/commit_status_helper.py index eb9247c2dc4..54f8f7361fc 100644 --- a/tests/ci/commit_status_helper.py +++ b/tests/ci/commit_status_helper.py @@ -263,12 +263,6 @@ def generate_status_comment(pr_info: PRInfo, statuses: CommitStatuses) -> str: result = [comment_body] - if visible_table_rows: - visible_table_rows.sort() - result.append(table_header) - result.extend(visible_table_rows) - result.append(table_footer) - if hidden_table_rows: hidden_table_rows.sort() result.append(details_header) @@ -277,6 +271,12 @@ def generate_status_comment(pr_info: PRInfo, statuses: CommitStatuses) -> str: result.append(table_footer) result.append(details_footer) + if visible_table_rows: + visible_table_rows.sort() + result.append(table_header) + result.extend(visible_table_rows) + result.append(table_footer) + return "".join(result) @@ -292,9 +292,7 @@ def create_ci_report(pr_info: PRInfo, statuses: CommitStatuses) -> str: log_urls = None if status.target_url is not None: log_urls = [status.target_url] - raw_logs = None - if status.description: - raw_logs = status.description + raw_logs = status.description or None test_results.append( TestResult( status.context, status.state, log_urls=log_urls, raw_logs=raw_logs diff --git a/tests/ci/report.py b/tests/ci/report.py index eb6e3bc1206..2c9e38fb1b1 100644 --- a/tests/ci/report.py +++ b/tests/ci/report.py @@ -578,7 +578,7 @@ def create_test_html_report( colspan += 1 row.append("") - rows_part.append("".join(row)) + rows_part.append("\n".join(row)) if test_result.raw_logs is not None: raw_logs = escape(test_result.raw_logs) row_raw_logs = ( From 8bb0fbfd2c69c206b210983c85701d6477c20e71 Mon Sep 17 00:00:00 2001 From: vdimir Date: Mon, 11 Sep 2023 17:03:28 +0000 Subject: [PATCH 287/433] fix --- tests/ci/commit_status_helper.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tests/ci/commit_status_helper.py b/tests/ci/commit_status_helper.py index 54f8f7361fc..5bcdb11ab6b 100644 --- a/tests/ci/commit_status_helper.py +++ b/tests/ci/commit_status_helper.py @@ -147,7 +147,7 @@ def update_pr_status_label(pr: PullRequest, status: str) -> None: if label.name == new_label: return if label.name.startswith("pr-status-"): - label.delete() + pr.remove_from_labels(label.name) pr.add_to_labels(new_label) From 790d61971c27d2e78603254d20dec0b3159359ec Mon Sep 17 00:00:00 2001 From: vdimir Date: Mon, 11 Sep 2023 17:06:16 +0000 Subject: [PATCH 288/433] sucessfull items in report can be expandable --- tests/ci/report.py | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) diff --git a/tests/ci/report.py b/tests/ci/report.py index 2c9e38fb1b1..73a2fe6f880 100644 --- a/tests/ci/report.py +++ b/tests/ci/report.py @@ -110,8 +110,8 @@ p.links a {{ padding: 5px; margin: 3px; background: var(--menu-background); line p.links a:hover {{ background: var(--menu-hover-background); color: var(--menu-hover-color); }} th {{ cursor: pointer; }} tr:hover {{ filter: var(--tr-hover-filter); }} -.failed {{ cursor: pointer; }} -.failed-content {{ display: none; }} +.expandable {{ cursor: pointer; }} +.expandable-content {{ display: none; }} #fish {{ display: none; float: right; position: relative; top: -20em; right: 2vw; margin-bottom: -20em; width: 30vw; filter: brightness(7%); z-index: -1; }} .themes {{ @@ -164,9 +164,9 @@ FOOTER_HTML_TEMPLATE = """ tr.addEventListener('click', function() {{ + Array.from(document.getElementsByClassName("expandable")).forEach(tr => tr.addEventListener('click', function() {{ var content = this.nextElementSibling; - content.classList.toggle("failed-content"); + content.classList.toggle("expandable-content"); }})); let theme = 'dark'; @@ -546,9 +546,8 @@ def create_test_html_report( has_log_urls = True row = [] - has_error = test_result.status in ("FAIL", "NOT_FAILED") - if has_error and test_result.raw_logs is not None: - row.append('') + if test_result.raw_logs is not None: + row.append('') else: row.append("") row.append(f"{test_result.name}") @@ -557,6 +556,7 @@ def create_test_html_report( # Allow to quickly scroll to the first failure. fail_id = "" + has_error = test_result.status in ("FAIL", "NOT_FAILED") if has_error: num_fails = num_fails + 1 fail_id = f'id="fail{num_fails}" ' @@ -582,7 +582,7 @@ def create_test_html_report( if test_result.raw_logs is not None: raw_logs = escape(test_result.raw_logs) row_raw_logs = ( - '' + '' f'
{raw_logs}
' "" ) From d5268f3fb0e61b143a2560a9434a28970cb40151 Mon Sep 17 00:00:00 2001 From: Sema Checherinda Date: Mon, 11 Sep 2023 19:57:16 +0200 Subject: [PATCH 289/433] rename a member --- src/Disks/ObjectStorages/DiskObjectStorageTransaction.cpp | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/src/Disks/ObjectStorages/DiskObjectStorageTransaction.cpp b/src/Disks/ObjectStorages/DiskObjectStorageTransaction.cpp index b42612d19b3..160ed766c7e 100644 --- a/src/Disks/ObjectStorages/DiskObjectStorageTransaction.cpp +++ b/src/Disks/ObjectStorages/DiskObjectStorageTransaction.cpp @@ -161,7 +161,7 @@ struct RemoveManyObjectStorageOperation final : public IDiskObjectStorageOperati const bool keep_all_batch_data; const NameSet file_names_remove_metadata_only; - std::vector removed_files; + std::vector paths_removed_with_objects; std::vector objects_to_remove; RemoveManyObjectStorageOperation( @@ -203,7 +203,7 @@ struct RemoveManyObjectStorageOperation final : public IDiskObjectStorageOperati if (unlink_outcome && !keep_all_batch_data && !file_names_remove_metadata_only.contains(fs::path(path).filename())) { objects_to_remove.emplace_back(ObjectsToRemove{std::move(objects), std::move(unlink_outcome)}); - removed_files.push_back(path); + paths_removed_with_objects.push_back(path); } } catch (const Exception & e) @@ -252,7 +252,7 @@ struct RemoveManyObjectStorageOperation final : public IDiskObjectStorageOperati &Poco::Logger::get("RemoveManyObjectStorageOperation"), "metadata and objects were removed for [{}], " "only metadata were removed for [{}].", - boost::algorithm::join(removed_files, ", "), + boost::algorithm::join(paths_removed_with_objects, ", "), boost::algorithm::join(file_names_remove_metadata_only, ", ")); } } From 803d8dcf85699ec5a8f6f2e9b948e5dfb64898da Mon Sep 17 00:00:00 2001 From: avogar Date: Wed, 6 Sep 2023 12:05:08 +0000 Subject: [PATCH 290/433] Support NULL as default for nested types Array/Tuple/Map for input formats --- src/DataTypes/DataTypeNullable.cpp | 3 + .../Serializations/SerializationArray.cpp | 15 +++- .../Serializations/SerializationMap.cpp | 5 +- .../Serializations/SerializationTuple.cpp | 10 ++- src/Formats/insertNullAsDefaultIfNeeded.cpp | 86 ++++++++++++++++++- src/Formats/insertNullAsDefaultIfNeeded.h | 2 +- .../Impl/NativeORCBlockInputFormat.cpp | 8 -- .../02872_null_as_default_nested.reference | 66 ++++++++++++++ .../02872_null_as_default_nested.sh | 15 ++++ 9 files changed, 193 insertions(+), 17 deletions(-) create mode 100644 tests/queries/0_stateless/02872_null_as_default_nested.reference create mode 100755 tests/queries/0_stateless/02872_null_as_default_nested.sh diff --git a/src/DataTypes/DataTypeNullable.cpp b/src/DataTypes/DataTypeNullable.cpp index 41a9a1de543..448caa8275a 100644 --- a/src/DataTypes/DataTypeNullable.cpp +++ b/src/DataTypes/DataTypeNullable.cpp @@ -3,6 +3,9 @@ #include #include #include +#include +#include +#include #include #include #include diff --git a/src/DataTypes/Serializations/SerializationArray.cpp b/src/DataTypes/Serializations/SerializationArray.cpp index e01c1aea0e9..d6f36e45e64 100644 --- a/src/DataTypes/Serializations/SerializationArray.cpp +++ b/src/DataTypes/Serializations/SerializationArray.cpp @@ -493,7 +493,10 @@ void SerializationArray::deserializeText(IColumn & column, ReadBuffer & istr, co deserializeTextImpl(column, istr, [&](IColumn & nested_column) { - nested->deserializeTextQuoted(nested_column, istr, settings); + if (settings.null_as_default) + SerializationNullable::deserializeTextQuotedImpl(nested_column, istr, settings, nested); + else + nested->deserializeTextQuoted(nested_column, istr, settings); }, false); if (whole && !istr.eof()) @@ -604,7 +607,10 @@ void SerializationArray::deserializeTextCSV(IColumn & column, ReadBuffer & istr, deserializeTextImpl(column, rb, [&](IColumn & nested_column) { - nested->deserializeTextCSV(nested_column, rb, settings); + if (settings.null_as_default) + SerializationNullable::deserializeTextCSVImpl(nested_column, rb, settings, nested); + else + nested->deserializeTextCSV(nested_column, rb, settings); }, true); } else @@ -612,7 +618,10 @@ void SerializationArray::deserializeTextCSV(IColumn & column, ReadBuffer & istr, deserializeTextImpl(column, rb, [&](IColumn & nested_column) { - nested->deserializeTextQuoted(nested_column, rb, settings); + if (settings.null_as_default) + SerializationNullable::deserializeTextQuotedImpl(nested_column, rb, settings, nested); + else + nested->deserializeTextQuoted(nested_column, rb, settings); }, true); } } diff --git a/src/DataTypes/Serializations/SerializationMap.cpp b/src/DataTypes/Serializations/SerializationMap.cpp index af1d96c4ca7..7588e630689 100644 --- a/src/DataTypes/Serializations/SerializationMap.cpp +++ b/src/DataTypes/Serializations/SerializationMap.cpp @@ -192,7 +192,10 @@ void SerializationMap::deserializeText(IColumn & column, ReadBuffer & istr, cons deserializeTextImpl(column, istr, [&settings](ReadBuffer & buf, const SerializationPtr & subcolumn_serialization, IColumn & subcolumn) { - subcolumn_serialization->deserializeTextQuoted(subcolumn, buf, settings); + if (settings.null_as_default) + SerializationNullable::deserializeTextQuotedImpl(subcolumn, buf, settings, subcolumn_serialization); + else + subcolumn_serialization->deserializeTextQuoted(subcolumn, buf, settings); }); if (whole && !istr.eof()) diff --git a/src/DataTypes/Serializations/SerializationTuple.cpp b/src/DataTypes/Serializations/SerializationTuple.cpp index 7f3e7619b0d..5c9487b97d4 100644 --- a/src/DataTypes/Serializations/SerializationTuple.cpp +++ b/src/DataTypes/Serializations/SerializationTuple.cpp @@ -135,7 +135,10 @@ void SerializationTuple::deserializeText(IColumn & column, ReadBuffer & istr, co assertChar(',', istr); skipWhitespaceIfAny(istr); } - elems[i]->deserializeTextQuoted(extractElementColumn(column, i), istr, settings); + if (settings.null_as_default) + SerializationNullable::deserializeTextQuotedImpl(extractElementColumn(column, i), istr, settings, elems[i]); + else + elems[i]->deserializeTextQuoted(extractElementColumn(column, i), istr, settings); } // Special format for one element tuple (1,) @@ -366,7 +369,10 @@ void SerializationTuple::deserializeTextCSV(IColumn & column, ReadBuffer & istr, assertChar(settings.csv.tuple_delimiter, istr); skipWhitespaceIfAny(istr); } - elems[i]->deserializeTextCSV(extractElementColumn(column, i), istr, settings); + if (settings.null_as_default) + SerializationNullable::deserializeTextCSVImpl(extractElementColumn(column, i), istr, settings, elems[i]); + else + elems[i]->deserializeTextCSV(extractElementColumn(column, i), istr, settings); } }); } diff --git a/src/Formats/insertNullAsDefaultIfNeeded.cpp b/src/Formats/insertNullAsDefaultIfNeeded.cpp index 767892718c5..c42b8c54d73 100644 --- a/src/Formats/insertNullAsDefaultIfNeeded.cpp +++ b/src/Formats/insertNullAsDefaultIfNeeded.cpp @@ -1,16 +1,96 @@ #include #include +#include +#include +#include #include #include +#include +#include +#include #include +#include namespace DB { -void insertNullAsDefaultIfNeeded(ColumnWithTypeAndName & input_column, const ColumnWithTypeAndName & header_column, size_t column_i, BlockMissingValues * block_missing_values) +bool insertNullAsDefaultIfNeeded(ColumnWithTypeAndName & input_column, const ColumnWithTypeAndName & header_column, size_t column_i, BlockMissingValues * block_missing_values) { + if (isArray(input_column.type) && isArray(header_column.type)) + { + ColumnWithTypeAndName nested_input_column; + const auto * array_input_column = checkAndGetColumn(input_column.column.get()); + nested_input_column.column = array_input_column->getDataPtr(); + nested_input_column.type = checkAndGetDataType(input_column.type.get())->getNestedType(); + + ColumnWithTypeAndName nested_header_column; + nested_header_column.column = checkAndGetColumn(header_column.column.get())->getDataPtr(); + nested_header_column.type = checkAndGetDataType(header_column.type.get())->getNestedType(); + + if (!insertNullAsDefaultIfNeeded(nested_input_column, nested_header_column, 0, nullptr)) + return false; + + input_column.column = ColumnArray::create(nested_input_column.column, array_input_column->getOffsetsPtr()); + input_column.type = std::make_shared(std::move(nested_input_column.type)); + return true; + } + + if (isTuple(input_column.type) && isTuple(header_column.type)) + { + const auto * tuple_input_column = checkAndGetColumn(input_column.column.get()); + const auto * tuple_input_type = checkAndGetDataType(input_column.type.get()); + const auto * tuple_header_column = checkAndGetColumn(header_column.column.get()); + const auto * tuple_header_type = checkAndGetDataType(header_column.type.get()); + + if (tuple_input_type->getElements().size() != tuple_header_type->getElements().size()) + return false; + + Columns nested_input_columns; + nested_input_columns.reserve(tuple_input_type->getElements().size()); + DataTypes nested_input_types; + nested_input_types.reserve(tuple_input_type->getElements().size()); + bool changed = false; + for (size_t i = 0; i != tuple_input_type->getElements().size(); ++i) + { + ColumnWithTypeAndName nested_input_column; + nested_input_column.column = tuple_input_column->getColumnPtr(i); + nested_input_column.type = tuple_input_type->getElement(i); + ColumnWithTypeAndName nested_header_column; + nested_header_column.column = tuple_header_column->getColumnPtr(i); + nested_header_column.type = tuple_header_type->getElement(i); + changed |= insertNullAsDefaultIfNeeded(nested_input_column, nested_header_column, 0, nullptr); + nested_input_columns.push_back(std::move(nested_input_column.column)); + nested_input_types.push_back(std::move(nested_input_column.type)); + } + + if (!changed) + return false; + + input_column.column = ColumnTuple::create(std::move(nested_input_columns)); + input_column.type = std::make_shared(std::move(nested_input_types)); + return true; + } + + if (isMap(input_column.type) && isMap(header_column.type)) + { + ColumnWithTypeAndName nested_input_column; + nested_input_column.column = checkAndGetColumn(input_column.column.get())->getNestedColumnPtr(); + nested_input_column.type = checkAndGetDataType(input_column.type.get())->getNestedType(); + + ColumnWithTypeAndName nested_header_column; + nested_header_column.column = checkAndGetColumn(header_column.column.get())->getNestedColumnPtr(); + nested_header_column.type = checkAndGetDataType(header_column.type.get())->getNestedType(); + + if (!insertNullAsDefaultIfNeeded(nested_input_column, nested_header_column, 0, nullptr)) + return false; + + input_column.column = ColumnMap::create(std::move(nested_input_column.column)); + input_column.type = std::make_shared(std::move(nested_input_column.type)); + return true; + } + if (!isNullableOrLowCardinalityNullable(input_column.type) || isNullableOrLowCardinalityNullable(header_column.type)) - return; + return false; if (block_missing_values) { @@ -32,6 +112,8 @@ void insertNullAsDefaultIfNeeded(ColumnWithTypeAndName & input_column, const Col const auto * lc_type = assert_cast(input_column.type.get()); input_column.type = std::make_shared(removeNullable(lc_type->getDictionaryType())); } + + return true; } } diff --git a/src/Formats/insertNullAsDefaultIfNeeded.h b/src/Formats/insertNullAsDefaultIfNeeded.h index 3e4dcd1e74a..874f803a14c 100644 --- a/src/Formats/insertNullAsDefaultIfNeeded.h +++ b/src/Formats/insertNullAsDefaultIfNeeded.h @@ -5,6 +5,6 @@ namespace DB { -void insertNullAsDefaultIfNeeded(ColumnWithTypeAndName & input_column, const ColumnWithTypeAndName & header_column, size_t column_i, BlockMissingValues * block_missing_values); +bool insertNullAsDefaultIfNeeded(ColumnWithTypeAndName & input_column, const ColumnWithTypeAndName & header_column, size_t column_i, BlockMissingValues * block_missing_values); } diff --git a/src/Processors/Formats/Impl/NativeORCBlockInputFormat.cpp b/src/Processors/Formats/Impl/NativeORCBlockInputFormat.cpp index 18316a6ebb4..1b6cde11be7 100644 --- a/src/Processors/Formats/Impl/NativeORCBlockInputFormat.cpp +++ b/src/Processors/Formats/Impl/NativeORCBlockInputFormat.cpp @@ -850,14 +850,6 @@ static ColumnWithTypeAndName readColumnFromORCColumn( if (skipped) return {}; - if (value_type_hint && !value_type_hint->equals(*value_column.type)) - { - /// Cast value column to target type, because it can happen - /// that parsed type cannot be ClickHouse Map value type. - value_column.column = castColumn(value_column, value_type_hint); - value_column.type = value_type_hint; - } - auto offsets_column = readOffsetsFromORCListColumn(orc_map_column); auto map_column = ColumnMap::create(key_column.column, value_column.column, offsets_column); auto map_type = std::make_shared(key_column.type, value_column.type); diff --git a/tests/queries/0_stateless/02872_null_as_default_nested.reference b/tests/queries/0_stateless/02872_null_as_default_nested.reference new file mode 100644 index 00000000000..dfcaf696d38 --- /dev/null +++ b/tests/queries/0_stateless/02872_null_as_default_nested.reference @@ -0,0 +1,66 @@ +Native +0 [0,0] [[[0,0],[0]]] ('hello',0,0) ('Hello',('Hello',(0,0)),0) {0:0} {0:{0:0}} ('Hello',[('Hello',{0:0},[0])],0) +42 [1,0] [[[1,0],[0]]] ('hello',0,1) ('Hello',('Hello',(1,0)),1) {1:0} {1:{1:0}} ('Hello',[('Hello',{1:0},[0])],1) +2 [2,2] [[[2,2],[0]]] ('hello',2,2) ('Hello',('Hello',(2,2)),2) {2:2} {2:{2:2}} ('Hello',[('Hello',{2:2},[2])],2) +42 [3,0] [[[3,0],[0]]] ('hello',0,3) ('Hello',('Hello',(3,0)),3) {3:0} {3:{3:0}} ('Hello',[('Hello',{3:0},[0])],3) +4 [4,4] [[[4,4],[0]]] ('hello',4,4) ('Hello',('Hello',(4,4)),4) {4:4} {4:{4:4}} ('Hello',[('Hello',{4:4},[4])],4) +Parquet +0 [0,0] [[[0,0],[0]]] ('hello',0,0) ('Hello',('Hello',(0,0)),0) {0:0} {0:{0:0}} ('Hello',[('Hello',{0:0},[0])],0) +42 [1,0] [[[1,0],[0]]] ('hello',0,1) ('Hello',('Hello',(1,0)),1) {1:0} {1:{1:0}} ('Hello',[('Hello',{1:0},[0])],1) +2 [2,2] [[[2,2],[0]]] ('hello',2,2) ('Hello',('Hello',(2,2)),2) {2:2} {2:{2:2}} ('Hello',[('Hello',{2:2},[2])],2) +42 [3,0] [[[3,0],[0]]] ('hello',0,3) ('Hello',('Hello',(3,0)),3) {3:0} {3:{3:0}} ('Hello',[('Hello',{3:0},[0])],3) +4 [4,4] [[[4,4],[0]]] ('hello',4,4) ('Hello',('Hello',(4,4)),4) {4:4} {4:{4:4}} ('Hello',[('Hello',{4:4},[4])],4) +ORC +0 [0,0] [[[0,0],[0]]] ('hello',0,0) ('Hello',('Hello',(0,0)),0) {0:0} {0:{0:0}} ('Hello',[('Hello',{0:0},[0])],0) +42 [1,0] [[[1,0],[0]]] ('hello',0,1) ('Hello',('Hello',(1,0)),1) {1:0} {1:{1:0}} ('Hello',[('Hello',{1:0},[0])],1) +2 [2,2] [[[2,2],[0]]] ('hello',2,2) ('Hello',('Hello',(2,2)),2) {2:2} {2:{2:2}} ('Hello',[('Hello',{2:2},[2])],2) +42 [3,0] [[[3,0],[0]]] ('hello',0,3) ('Hello',('Hello',(3,0)),3) {3:0} {3:{3:0}} ('Hello',[('Hello',{3:0},[0])],3) +4 [4,4] [[[4,4],[0]]] ('hello',4,4) ('Hello',('Hello',(4,4)),4) {4:4} {4:{4:4}} ('Hello',[('Hello',{4:4},[4])],4) +Arrow +0 [0,0] [[[0,0],[0]]] ('hello',0,0) ('Hello',('Hello',(0,0)),0) {0:0} {0:{0:0}} ('Hello',[('Hello',{0:0},[0])],0) +42 [1,0] [[[1,0],[0]]] ('hello',0,1) ('Hello',('Hello',(1,0)),1) {1:0} {1:{1:0}} ('Hello',[('Hello',{1:0},[0])],1) +2 [2,2] [[[2,2],[0]]] ('hello',2,2) ('Hello',('Hello',(2,2)),2) {2:2} {2:{2:2}} ('Hello',[('Hello',{2:2},[2])],2) +42 [3,0] [[[3,0],[0]]] ('hello',0,3) ('Hello',('Hello',(3,0)),3) {3:0} {3:{3:0}} ('Hello',[('Hello',{3:0},[0])],3) +4 [4,4] [[[4,4],[0]]] ('hello',4,4) ('Hello',('Hello',(4,4)),4) {4:4} {4:{4:4}} ('Hello',[('Hello',{4:4},[4])],4) +Avro +0 [0,0] [[[0,0],[0]]] ('hello',0,0) ('Hello',('Hello',(0,0)),0) {0:0} {0:{0:0}} ('Hello',[('Hello',{0:0},[0])],0) +42 [1,0] [[[1,0],[0]]] ('hello',0,1) ('Hello',('Hello',(1,0)),1) {1:0} {1:{1:0}} ('Hello',[('Hello',{1:0},[0])],1) +2 [2,2] [[[2,2],[0]]] ('hello',2,2) ('Hello',('Hello',(2,2)),2) {2:2} {2:{2:2}} ('Hello',[('Hello',{2:2},[2])],2) +42 [3,0] [[[3,0],[0]]] ('hello',0,3) ('Hello',('Hello',(3,0)),3) {3:0} {3:{3:0}} ('Hello',[('Hello',{3:0},[0])],3) +4 [4,4] [[[4,4],[0]]] ('hello',4,4) ('Hello',('Hello',(4,4)),4) {4:4} {4:{4:4}} ('Hello',[('Hello',{4:4},[4])],4) +BSONEachRow +0 [0,0] [[[0,0],[0]]] ('hello',0,0) ('Hello',('Hello',(0,0)),0) {0:0} {0:{0:0}} ('Hello',[('Hello',{0:0},[0])],0) +42 [1,0] [[[1,0],[0]]] ('hello',0,1) ('Hello',('Hello',(1,0)),1) {1:0} {1:{1:0}} ('Hello',[('Hello',{1:0},[0])],1) +2 [2,2] [[[2,2],[0]]] ('hello',2,2) ('Hello',('Hello',(2,2)),2) {2:2} {2:{2:2}} ('Hello',[('Hello',{2:2},[2])],2) +42 [3,0] [[[3,0],[0]]] ('hello',0,3) ('Hello',('Hello',(3,0)),3) {3:0} {3:{3:0}} ('Hello',[('Hello',{3:0},[0])],3) +4 [4,4] [[[4,4],[0]]] ('hello',4,4) ('Hello',('Hello',(4,4)),4) {4:4} {4:{4:4}} ('Hello',[('Hello',{4:4},[4])],4) +MsgPack +0 [0,0] [[[0,0],[0]]] ('hello',0,0) ('Hello',('Hello',(0,0)),0) {0:0} {0:{0:0}} ('Hello',[('Hello',{0:0},[0])],0) +42 [1,0] [[[1,0],[0]]] ('hello',0,1) ('Hello',('Hello',(1,0)),1) {1:0} {1:{1:0}} ('Hello',[('Hello',{1:0},[0])],1) +2 [2,2] [[[2,2],[0]]] ('hello',2,2) ('Hello',('Hello',(2,2)),2) {2:2} {2:{2:2}} ('Hello',[('Hello',{2:2},[2])],2) +42 [3,0] [[[3,0],[0]]] ('hello',0,3) ('Hello',('Hello',(3,0)),3) {3:0} {3:{3:0}} ('Hello',[('Hello',{3:0},[0])],3) +4 [4,4] [[[4,4],[0]]] ('hello',4,4) ('Hello',('Hello',(4,4)),4) {4:4} {4:{4:4}} ('Hello',[('Hello',{4:4},[4])],4) +JSONEachRow +0 [0,0] [[[0,0],[0]]] ('hello',0,0) ('Hello',('Hello',(0,0)),0) {0:0} {0:{0:0}} ('Hello',[('Hello',{0:0},[0])],0) +42 [1,0] [[[1,0],[0]]] ('hello',0,1) ('Hello',('Hello',(1,0)),1) {1:0} {1:{1:0}} ('Hello',[('Hello',{1:0},[0])],1) +2 [2,2] [[[2,2],[0]]] ('hello',2,2) ('Hello',('Hello',(2,2)),2) {2:2} {2:{2:2}} ('Hello',[('Hello',{2:2},[2])],2) +42 [3,0] [[[3,0],[0]]] ('hello',0,3) ('Hello',('Hello',(3,0)),3) {3:0} {3:{3:0}} ('Hello',[('Hello',{3:0},[0])],3) +4 [4,4] [[[4,4],[0]]] ('hello',4,4) ('Hello',('Hello',(4,4)),4) {4:4} {4:{4:4}} ('Hello',[('Hello',{4:4},[4])],4) +CSV +0 [0,0] [[[0,0],[0]]] ('hello',0,0) ('Hello',('Hello',(0,0)),0) {0:0} {0:{0:0}} ('Hello',[('Hello',{0:0},[0])],0) +42 [1,0] [[[1,0],[0]]] ('hello',0,1) ('Hello',('Hello',(1,0)),1) {1:0} {1:{1:0}} ('Hello',[('Hello',{1:0},[0])],1) +2 [2,2] [[[2,2],[0]]] ('hello',2,2) ('Hello',('Hello',(2,2)),2) {2:2} {2:{2:2}} ('Hello',[('Hello',{2:2},[2])],2) +42 [3,0] [[[3,0],[0]]] ('hello',0,3) ('Hello',('Hello',(3,0)),3) {3:0} {3:{3:0}} ('Hello',[('Hello',{3:0},[0])],3) +4 [4,4] [[[4,4],[0]]] ('hello',4,4) ('Hello',('Hello',(4,4)),4) {4:4} {4:{4:4}} ('Hello',[('Hello',{4:4},[4])],4) +TSV +0 [0,0] [[[0,0],[0]]] ('hello',0,0) ('Hello',('Hello',(0,0)),0) {0:0} {0:{0:0}} ('Hello',[('Hello',{0:0},[0])],0) +42 [1,0] [[[1,0],[0]]] ('hello',0,1) ('Hello',('Hello',(1,0)),1) {1:0} {1:{1:0}} ('Hello',[('Hello',{1:0},[0])],1) +2 [2,2] [[[2,2],[0]]] ('hello',2,2) ('Hello',('Hello',(2,2)),2) {2:2} {2:{2:2}} ('Hello',[('Hello',{2:2},[2])],2) +42 [3,0] [[[3,0],[0]]] ('hello',0,3) ('Hello',('Hello',(3,0)),3) {3:0} {3:{3:0}} ('Hello',[('Hello',{3:0},[0])],3) +4 [4,4] [[[4,4],[0]]] ('hello',4,4) ('Hello',('Hello',(4,4)),4) {4:4} {4:{4:4}} ('Hello',[('Hello',{4:4},[4])],4) +Values +0 [0,0] [[[0,0],[0]]] ('hello',0,0) ('Hello',('Hello',(0,0)),0) {0:0} {0:{0:0}} ('Hello',[('Hello',{0:0},[0])],0) +42 [1,0] [[[1,0],[0]]] ('hello',0,1) ('Hello',('Hello',(1,0)),1) {1:0} {1:{1:0}} ('Hello',[('Hello',{1:0},[0])],1) +2 [2,2] [[[2,2],[0]]] ('hello',2,2) ('Hello',('Hello',(2,2)),2) {2:2} {2:{2:2}} ('Hello',[('Hello',{2:2},[2])],2) +42 [3,0] [[[3,0],[0]]] ('hello',0,3) ('Hello',('Hello',(3,0)),3) {3:0} {3:{3:0}} ('Hello',[('Hello',{3:0},[0])],3) +4 [4,4] [[[4,4],[0]]] ('hello',4,4) ('Hello',('Hello',(4,4)),4) {4:4} {4:{4:4}} ('Hello',[('Hello',{4:4},[4])],4) diff --git a/tests/queries/0_stateless/02872_null_as_default_nested.sh b/tests/queries/0_stateless/02872_null_as_default_nested.sh new file mode 100755 index 00000000000..8f91d573b89 --- /dev/null +++ b/tests/queries/0_stateless/02872_null_as_default_nested.sh @@ -0,0 +1,15 @@ +#!/usr/bin/env bash +# Tags: no-fasttest + +CUR_DIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) +# shellcheck source=../shell_config.sh +. "$CUR_DIR"/../shell_config.sh + +for format in Native Parquet ORC Arrow Avro BSONEachRow MsgPack JSONEachRow CSV TSV Values +do + echo $format + $CLICKHOUSE_LOCAL -q "select number % 2 ? NULL : number as n, [number, number % 2 ? NULL : number] as arr1, [[[number, number % 2 ? NULL : number], [NULL]]] as arr2, tuple('hello', number % 2 ? NULL : number, number) as tup1, tuple('Hello', tuple('Hello', tuple(number, number % 2 ? NULL : number)), number) as tup2, map(number, number % 2 ? NULL : number) as map1, map(number, map(number, number % 2 ? null : number)) as map2, tuple('Hello', [tuple('Hello', map(number, number % 2 ? NULL : number), [number % 2 ? NULL : number])], number) as nested from numbers(5) format $format" > $CLICKHOUSE_TEST_UNIQUE_NAME.$format + $CLICKHOUSE_LOCAL -q "select * from file('$CLICKHOUSE_TEST_UNIQUE_NAME.$format', auto, 'n UInt64 default 42, arr1 Array(UInt64), arr2 Array(Array(Array(UInt64))), tup1 Tuple(String, UInt64, UInt64), tup2 Tuple(String, Tuple(String, Tuple(UInt64, UInt64)), UInt64), map1 Map(UInt64, UInt64), map2 Map(UInt64, Map(UInt64, UInt64)), nested Tuple(String, Array(Tuple(String, Map(UInt64, UInt64), Array(UInt64))), UInt64)') settings input_format_null_as_default=1" + rm $CLICKHOUSE_TEST_UNIQUE_NAME.$format +done + From c9612de81f1b96e2927efab4bceb50b7c780f8ca Mon Sep 17 00:00:00 2001 From: avogar Date: Wed, 6 Sep 2023 12:06:48 +0000 Subject: [PATCH 291/433] Remove unneded includes --- src/DataTypes/DataTypeNullable.cpp | 3 --- 1 file changed, 3 deletions(-) diff --git a/src/DataTypes/DataTypeNullable.cpp b/src/DataTypes/DataTypeNullable.cpp index 448caa8275a..41a9a1de543 100644 --- a/src/DataTypes/DataTypeNullable.cpp +++ b/src/DataTypes/DataTypeNullable.cpp @@ -3,9 +3,6 @@ #include #include #include -#include -#include -#include #include #include #include From 8cd6153c8395af520530321132a59c654488a759 Mon Sep 17 00:00:00 2001 From: Azat Khuzhin Date: Mon, 11 Sep 2023 14:39:35 +0200 Subject: [PATCH 292/433] Add modification_time into system.detached_parts Signed-off-by: Azat Khuzhin --- .../System/StorageSystemDetachedParts.cpp | 17 ++++++++++++++++- tests/integration/test_partition/test.py | 2 +- .../00502_custom_partitioning_local.sql | 2 +- .../02117_show_create_table_system.reference | 1 + ...m_detached_parts_modification_time.reference | 2 ++ ...stem_detached_parts_modification_time.sql.j2 | 16 ++++++++++++++++ 6 files changed, 37 insertions(+), 3 deletions(-) create mode 100644 tests/queries/0_stateless/02881_system_detached_parts_modification_time.reference create mode 100644 tests/queries/0_stateless/02881_system_detached_parts_modification_time.sql.j2 diff --git a/src/Storages/System/StorageSystemDetachedParts.cpp b/src/Storages/System/StorageSystemDetachedParts.cpp index c5d2ba94e09..cec54cef914 100644 --- a/src/Storages/System/StorageSystemDetachedParts.cpp +++ b/src/Storages/System/StorageSystemDetachedParts.cpp @@ -2,6 +2,7 @@ #include #include +#include #include #include #include @@ -230,6 +231,19 @@ private: size_t bytes_on_disk = parts_sizes.at(p_id - begin).load(); new_columns[res_index++]->insert(bytes_on_disk); } + if (columns_mask[src_index++]) + { + Poco::Timestamp modification_time; + try + { + modification_time = p.disk->getLastModified(fs::path(current_info.data->getRelativeDataPath()) / MergeTreeData::DETACHED_DIR_NAME / p.dir_name); + } + catch (const fs::filesystem_error &) + { + tryLogCurrentException(__PRETTY_FUNCTION__); + } + new_columns[res_index++]->insert(static_cast(modification_time.epochTime())); + } if (columns_mask[src_index++]) new_columns[res_index++]->insert(p.disk->getName()); if (columns_mask[src_index++]) @@ -260,12 +274,13 @@ StorageSystemDetachedParts::StorageSystemDetachedParts(const StorageID & table_i {"partition_id", std::make_shared(std::make_shared())}, {"name", std::make_shared()}, {"bytes_on_disk", std::make_shared()}, + {"modification_time",std::make_shared()}, {"disk", std::make_shared()}, {"path", std::make_shared()}, {"reason", std::make_shared(std::make_shared())}, {"min_block_number", std::make_shared(std::make_shared())}, {"max_block_number", std::make_shared(std::make_shared())}, - {"level", std::make_shared(std::make_shared())} + {"level", std::make_shared(std::make_shared())}, }}); setInMemoryMetadata(storage_metadata); } diff --git a/tests/integration/test_partition/test.py b/tests/integration/test_partition/test.py index 2517b2d1ae6..97d21d14184 100644 --- a/tests/integration/test_partition/test.py +++ b/tests/integration/test_partition/test.py @@ -431,7 +431,7 @@ def test_system_detached_parts(drop_detached_parts_table): ) res = q( - "select system.detached_parts.* except (bytes_on_disk, `path`) from system.detached_parts where table like 'sdp_%' order by table, name" + "select system.detached_parts.* except (bytes_on_disk, `path`, modification_time) from system.detached_parts where table like 'sdp_%' order by table, name" ) assert ( res == "default\tsdp_0\tall\tall_1_1_0\tdefault\t\t1\t1\t0\n" diff --git a/tests/queries/0_stateless/00502_custom_partitioning_local.sql b/tests/queries/0_stateless/00502_custom_partitioning_local.sql index 3d5f71429fe..a116f8bca24 100644 --- a/tests/queries/0_stateless/00502_custom_partitioning_local.sql +++ b/tests/queries/0_stateless/00502_custom_partitioning_local.sql @@ -18,7 +18,7 @@ ALTER TABLE not_partitioned DETACH PARTITION ID 'all'; SELECT 'Sum after DETACH PARTITION:'; SELECT sum(x) FROM not_partitioned; SELECT 'system.detached_parts after DETACH PARTITION:'; -SELECT system.detached_parts.* EXCEPT (bytes_on_disk, `path`, disk) FROM system.detached_parts WHERE database = currentDatabase() AND table = 'not_partitioned'; +SELECT system.detached_parts.* EXCEPT (bytes_on_disk, `path`, disk, modification_time) FROM system.detached_parts WHERE database = currentDatabase() AND table = 'not_partitioned'; DROP TABLE not_partitioned; diff --git a/tests/queries/0_stateless/02117_show_create_table_system.reference b/tests/queries/0_stateless/02117_show_create_table_system.reference index d96a02834b4..9b633314bd3 100644 --- a/tests/queries/0_stateless/02117_show_create_table_system.reference +++ b/tests/queries/0_stateless/02117_show_create_table_system.reference @@ -145,6 +145,7 @@ CREATE TABLE system.detached_parts `partition_id` Nullable(String), `name` String, `bytes_on_disk` UInt64, + `modification_time` DateTime, `disk` String, `path` String, `reason` Nullable(String), diff --git a/tests/queries/0_stateless/02881_system_detached_parts_modification_time.reference b/tests/queries/0_stateless/02881_system_detached_parts_modification_time.reference new file mode 100644 index 00000000000..15d2c113fda --- /dev/null +++ b/tests/queries/0_stateless/02881_system_detached_parts_modification_time.reference @@ -0,0 +1,2 @@ +after detach 1 +after detach 1 diff --git a/tests/queries/0_stateless/02881_system_detached_parts_modification_time.sql.j2 b/tests/queries/0_stateless/02881_system_detached_parts_modification_time.sql.j2 new file mode 100644 index 00000000000..ded17b4e328 --- /dev/null +++ b/tests/queries/0_stateless/02881_system_detached_parts_modification_time.sql.j2 @@ -0,0 +1,16 @@ +set mutations_sync=1; + +{% for id, settings in [ + ("wide", "min_bytes_for_wide_part=0, min_rows_for_wide_part=0"), + ("compact", "min_bytes_for_wide_part=1000, min_rows_for_wide_part=100"), +] +%} + +drop table if exists data_{{ id }}; +create table data_{{ id }} (key Int) engine=MergeTree() order by tuple() settings {{ settings }}; +insert into data_{{ id }} values (1); +select 'before detach', now()-modification_time < 10 from system.detached_parts where database = currentDatabase() and table = 'data_{{ id }}'; +alter table data_{{ id }} detach partition all; +select 'after detach', now()-modification_time < 10 from system.detached_parts where database = currentDatabase() and table = 'data_{{ id }}'; + +{% endfor %} From 4f5d132019c011cdee07faf90a329536c11a7f6f Mon Sep 17 00:00:00 2001 From: pufit Date: Mon, 11 Sep 2023 14:46:40 -0400 Subject: [PATCH 293/433] Fix timeout keeper-client --- tests/integration/helpers/keeper_utils.py | 2 ++ 1 file changed, 2 insertions(+) diff --git a/tests/integration/helpers/keeper_utils.py b/tests/integration/helpers/keeper_utils.py index 767416a3dbc..a1d20e0392b 100644 --- a/tests/integration/helpers/keeper_utils.py +++ b/tests/integration/helpers/keeper_utils.py @@ -78,6 +78,8 @@ class KeeperClient(object): self.proc.stdin.flush() events = self.poller.poll(timeout) + if not events: + raise TimeoutError(f"Keeper client returned no output") for fd_num, event in events: if event & (select.EPOLLIN | select.EPOLLPRI): From 9e56cff58889395d1375ee0cb2875acba5dde264 Mon Sep 17 00:00:00 2001 From: Amos Bird Date: Tue, 12 Sep 2023 02:49:27 +0800 Subject: [PATCH 294/433] Better lambda capture --- src/Functions/FunctionsConversion.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/Functions/FunctionsConversion.h b/src/Functions/FunctionsConversion.h index 535723bce4e..8d28c9f0a31 100644 --- a/src/Functions/FunctionsConversion.h +++ b/src/Functions/FunctionsConversion.h @@ -3193,7 +3193,7 @@ private: { if (agg_type->getFunction()->haveSameStateRepresentation(*to_type->getFunction())) { - return [agg_type]( + return [function = agg_type->getFunction()]( ColumnsWithTypeAndName & arguments, const DataTypePtr & /* result_type */, const ColumnNullable * /* nullable_source */, @@ -3204,7 +3204,7 @@ private: if (col_agg) { auto new_col_agg = ColumnAggregateFunction::create(*col_agg); - new_col_agg->set(agg_type->getFunction()); + new_col_agg->set(function); return new_col_agg; } else From dd1567f473adc72ba786a5e0c194c4a30dc7d0d3 Mon Sep 17 00:00:00 2001 From: avogar Date: Mon, 11 Sep 2023 18:54:27 +0000 Subject: [PATCH 295/433] Fix fasttest --- tests/queries/0_stateless/00748_insert_array_with_null.sql | 1 + 1 file changed, 1 insertion(+) diff --git a/tests/queries/0_stateless/00748_insert_array_with_null.sql b/tests/queries/0_stateless/00748_insert_array_with_null.sql index ca36352c2cf..ac55d4e9d8c 100644 --- a/tests/queries/0_stateless/00748_insert_array_with_null.sql +++ b/tests/queries/0_stateless/00748_insert_array_with_null.sql @@ -1,6 +1,7 @@ DROP TABLE IF EXISTS arraytest; set allow_deprecated_syntax_for_merge_tree=1; +set input_format_null_as_default=0; CREATE TABLE arraytest ( created_date Date DEFAULT toDate(created_at), created_at DateTime DEFAULT now(), strings Array(String) DEFAULT emptyArrayString()) ENGINE = MergeTree(created_date, cityHash64(created_at), (created_date, cityHash64(created_at)), 8192); INSERT INTO arraytest (created_at, strings) VALUES (now(), ['aaaaa', 'bbbbb', 'ccccc']); From dbd2c3f7d986a44b9107623eb077f6d416501923 Mon Sep 17 00:00:00 2001 From: Kruglov Pavel <48961922+Avogar@users.noreply.github.com> Date: Mon, 11 Sep 2023 20:57:49 +0200 Subject: [PATCH 296/433] Fix tests --- .../0_stateless/02842_capn_proto_outfile_without_schema.sh | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tests/queries/0_stateless/02842_capn_proto_outfile_without_schema.sh b/tests/queries/0_stateless/02842_capn_proto_outfile_without_schema.sh index ae57cdd7543..f74e662520b 100755 --- a/tests/queries/0_stateless/02842_capn_proto_outfile_without_schema.sh +++ b/tests/queries/0_stateless/02842_capn_proto_outfile_without_schema.sh @@ -6,6 +6,6 @@ CUR_DIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) . "$CUR_DIR"/../shell_config.sh -$CLICKHOUSE_LOCAL -q "select * from numbers(10) into outfile '$CLICKHOUSE_TEST_UNIQUE_NAME.capnp'" 2>&1 | grep "The format CapnProto requires a schema" -c +$CLICKHOUSE_LOCAL -q "select * from numbers(10) into outfile '$CLICKHOUSE_TEST_UNIQUE_NAME.capnp' settings format_capn_proto_use_autogenerated_schema=0" 2>&1 | grep "The format CapnProto requires a schema" -c rm $CLICKHOUSE_TEST_UNIQUE_NAME.capnp From 0a8138935908ef9aaa6de59a70fb1b6a2ca12b9f Mon Sep 17 00:00:00 2001 From: HarryLeeIBM Date: Mon, 11 Sep 2023 12:14:01 -0700 Subject: [PATCH 297/433] Fix endian issue in jemalloc_bins for s390x --- src/Storages/System/StorageSystemJemalloc.cpp | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/src/Storages/System/StorageSystemJemalloc.cpp b/src/Storages/System/StorageSystemJemalloc.cpp index 4348349ebbc..d874f657859 100644 --- a/src/Storages/System/StorageSystemJemalloc.cpp +++ b/src/Storages/System/StorageSystemJemalloc.cpp @@ -25,6 +25,12 @@ UInt64 getJeMallocValue(const char * name) UInt64 value{}; size_t size = sizeof(value); mallctl(name, &value, &size, nullptr, 0); +#if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ + if (size == 4) + { + value >>= 32; + } +#endif return value; } From 667426f1f222cb6264a61be92b95ad8d9cb9b359 Mon Sep 17 00:00:00 2001 From: Amos Bird Date: Tue, 12 Sep 2023 03:54:19 +0800 Subject: [PATCH 298/433] DataTypeAggregateFunction::strictEquals --- src/DataTypes/DataTypeAggregateFunction.cpp | 56 +++++++++---------- src/DataTypes/DataTypeAggregateFunction.h | 1 + src/Functions/FunctionsConversion.h | 13 ++++- .../optimizeUseAggregateProjection.cpp | 15 +++-- ...rojection_with_normalized_states.reference | 1 + ...gate_projection_with_normalized_states.sql | 2 + 6 files changed, 54 insertions(+), 34 deletions(-) diff --git a/src/DataTypes/DataTypeAggregateFunction.cpp b/src/DataTypes/DataTypeAggregateFunction.cpp index b580ecc9592..be60886d74b 100644 --- a/src/DataTypes/DataTypeAggregateFunction.cpp +++ b/src/DataTypes/DataTypeAggregateFunction.cpp @@ -117,6 +117,33 @@ Field DataTypeAggregateFunction::getDefault() const return field; } +bool DataTypeAggregateFunction::strictEquals(const DataTypePtr & lhs_state_type, const DataTypePtr & rhs_state_type) +{ + const auto * lhs_state = typeid_cast(lhs_state_type.get()); + const auto * rhs_state = typeid_cast(rhs_state_type.get()); + + if (!lhs_state || !rhs_state) + return false; + + if (lhs_state->function->getName() != rhs_state->function->getName()) + return false; + + if (lhs_state->parameters.size() != rhs_state->parameters.size()) + return false; + + for (size_t i = 0; i < lhs_state->parameters.size(); ++i) + if (lhs_state->parameters[i] != rhs_state->parameters[i]) + return false; + + if (lhs_state->argument_types.size() != rhs_state->argument_types.size()) + return false; + + for (size_t i = 0; i < lhs_state->argument_types.size(); ++i) + if (!lhs_state->argument_types[i]->equals(*rhs_state->argument_types[i])) + return false; + + return true; +} bool DataTypeAggregateFunction::equals(const IDataType & rhs) const { @@ -126,34 +153,7 @@ bool DataTypeAggregateFunction::equals(const IDataType & rhs) const auto lhs_state_type = function->getNormalizedStateType(); auto rhs_state_type = typeid_cast(rhs).function->getNormalizedStateType(); - if (typeid(lhs_state_type.get()) != typeid(rhs_state_type.get())) - return false; - - if (const auto * lhs_state = typeid_cast(lhs_state_type.get())) - { - const auto & rhs_state = typeid_cast(*rhs_state_type); - - if (lhs_state->function->getName() != rhs_state.function->getName()) - return false; - - if (lhs_state->parameters.size() != rhs_state.parameters.size()) - return false; - - for (size_t i = 0; i < lhs_state->parameters.size(); ++i) - if (lhs_state->parameters[i] != rhs_state.parameters[i]) - return false; - - if (lhs_state->argument_types.size() != rhs_state.argument_types.size()) - return false; - - for (size_t i = 0; i < lhs_state->argument_types.size(); ++i) - if (!lhs_state->argument_types[i]->equals(*rhs_state.argument_types[i])) - return false; - - return true; - } - - return lhs_state_type->equals(*rhs_state_type); + return strictEquals(lhs_state_type, rhs_state_type); } diff --git a/src/DataTypes/DataTypeAggregateFunction.h b/src/DataTypes/DataTypeAggregateFunction.h index 83c9f10f407..6331c23222f 100644 --- a/src/DataTypes/DataTypeAggregateFunction.h +++ b/src/DataTypes/DataTypeAggregateFunction.h @@ -60,6 +60,7 @@ public: Field getDefault() const override; + static bool strictEquals(const DataTypePtr & lhs_state_type, const DataTypePtr & rhs_state_type); bool equals(const IDataType & rhs) const override; bool isParametric() const override { return true; } diff --git a/src/Functions/FunctionsConversion.h b/src/Functions/FunctionsConversion.h index 8d28c9f0a31..ae5b3305baf 100644 --- a/src/Functions/FunctionsConversion.h +++ b/src/Functions/FunctionsConversion.h @@ -3193,7 +3193,7 @@ private: { if (agg_type->getFunction()->haveSameStateRepresentation(*to_type->getFunction())) { - return [function = agg_type->getFunction()]( + return [function = to_type->getFunction()]( ColumnsWithTypeAndName & arguments, const DataTypePtr & /* result_type */, const ColumnNullable * /* nullable_source */, @@ -4003,7 +4003,16 @@ private: safe_convert_custom_types = to_type->getCustomName() && from_type_custom_name->getName() == to_type->getCustomName()->getName(); if (from_type->equals(*to_type) && safe_convert_custom_types) - return createIdentityWrapper(from_type); + { + /// We can only use identity conversion for DataTypeAggregateFunction when they are strictly equivalent. + if (typeid_cast(from_type.get())) + { + if (DataTypeAggregateFunction::strictEquals(from_type, to_type)) + return createIdentityWrapper(from_type); + } + else + return createIdentityWrapper(from_type); + } else if (WhichDataType(from_type).isNothing()) return createNothingWrapper(to_type.get()); diff --git a/src/Processors/QueryPlan/Optimizations/optimizeUseAggregateProjection.cpp b/src/Processors/QueryPlan/Optimizations/optimizeUseAggregateProjection.cpp index 0b24d2adbf9..0599a0fa369 100644 --- a/src/Processors/QueryPlan/Optimizations/optimizeUseAggregateProjection.cpp +++ b/src/Processors/QueryPlan/Optimizations/optimizeUseAggregateProjection.cpp @@ -255,10 +255,17 @@ static void appendAggregateFunctions( if (node->result_name != aggregate.column_name) { - /// Always cast to aggregate types specified in query, because input - /// columns from projection might have the same state but different - /// type, which can generate wrong results during finalization. - node = &proj_dag.addCast(*node, type, aggregate.column_name); + if (DataTypeAggregateFunction::strictEquals(type, node->result_type)) + { + node = &proj_dag.addAlias(*node, aggregate.column_name); + } + else + { + /// Cast to aggregate types specified in query if it's not + /// strictly the same as the one specified in projection. This + /// is required to generate correct results during finalization. + node = &proj_dag.addCast(*node, type, aggregate.column_name); + } } proj_dag_outputs.push_back(node); diff --git a/tests/queries/0_stateless/01710_aggregate_projection_with_normalized_states.reference b/tests/queries/0_stateless/01710_aggregate_projection_with_normalized_states.reference index 9db43c50b71..25aa9dc5dec 100644 --- a/tests/queries/0_stateless/01710_aggregate_projection_with_normalized_states.reference +++ b/tests/queries/0_stateless/01710_aggregate_projection_with_normalized_states.reference @@ -1 +1,2 @@ +3 950 990 500 2000 diff --git a/tests/queries/0_stateless/01710_aggregate_projection_with_normalized_states.sql b/tests/queries/0_stateless/01710_aggregate_projection_with_normalized_states.sql index f7be9239c11..e023c0991b3 100644 --- a/tests/queries/0_stateless/01710_aggregate_projection_with_normalized_states.sql +++ b/tests/queries/0_stateless/01710_aggregate_projection_with_normalized_states.sql @@ -1,5 +1,7 @@ DROP TABLE IF EXISTS r; +select finalizeAggregation(cast(quantileState(0)(arrayJoin([1,2,3])) as AggregateFunction(quantile(1), UInt8))); + CREATE TABLE r ( x String, a LowCardinality(String), From 26b17f88d5e98df38f67c96c69d60824eccecf33 Mon Sep 17 00:00:00 2001 From: HarryLeeIBM Date: Mon, 11 Sep 2023 13:52:58 -0700 Subject: [PATCH 299/433] Add comment for the fix --- src/Storages/System/StorageSystemJemalloc.cpp | 3 +++ 1 file changed, 3 insertions(+) diff --git a/src/Storages/System/StorageSystemJemalloc.cpp b/src/Storages/System/StorageSystemJemalloc.cpp index d874f657859..207b378c240 100644 --- a/src/Storages/System/StorageSystemJemalloc.cpp +++ b/src/Storages/System/StorageSystemJemalloc.cpp @@ -25,6 +25,9 @@ UInt64 getJeMallocValue(const char * name) UInt64 value{}; size_t size = sizeof(value); mallctl(name, &value, &size, nullptr, 0); + /// mallctl() fills the value with 32 bit integer for some queries("arenas.nbins" for example). + /// In this case variable 'size' will be changed from 8 to 4 and the 64 bit variable 'value' will hold the 32 bit actual value times 2^32 on big-endian machines. + /// We should right shift the value by 32 on big-endian machines(which is unnecessary on little-endian machines). #if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ if (size == 4) { From df71dcd94d893f124040105ad35755b750abbe9f Mon Sep 17 00:00:00 2001 From: Michael Kolupaev Date: Mon, 26 Jun 2023 21:49:44 +0000 Subject: [PATCH 300/433] Fix data race in copyFromIStreamWithProgressCallback --- src/IO/HTTPCommon.h | 1 + src/IO/ReadBufferFromS3.cpp | 21 ++++++++++-------- src/IO/ReadWriteBufferFromHTTP.cpp | 26 ++++++++++++----------- src/IO/SeekableReadBuffer.cpp | 5 +++-- src/IO/SeekableReadBuffer.h | 3 ++- tests/integration/test_storage_s3/test.py | 6 +++--- 6 files changed, 35 insertions(+), 27 deletions(-) diff --git a/src/IO/HTTPCommon.h b/src/IO/HTTPCommon.h index caf2fa361d9..04ca85925af 100644 --- a/src/IO/HTTPCommon.h +++ b/src/IO/HTTPCommon.h @@ -62,6 +62,7 @@ struct HTTPSessionReuseTag { }; +void markSessionForReuse(Poco::Net::HTTPSession & session); void markSessionForReuse(HTTPSessionPtr session); void markSessionForReuse(PooledHTTPSessionPtr session); diff --git a/src/IO/ReadBufferFromS3.cpp b/src/IO/ReadBufferFromS3.cpp index de38487642c..c038523bdaa 100644 --- a/src/IO/ReadBufferFromS3.cpp +++ b/src/IO/ReadBufferFromS3.cpp @@ -221,13 +221,12 @@ bool ReadBufferFromS3::nextImpl() size_t ReadBufferFromS3::readBigAt(char * to, size_t n, size_t range_begin, const std::function & progress_callback) { - if (n == 0) - return 0; - + size_t initial_n = n; size_t sleep_time_with_backoff_milliseconds = 100; - for (size_t attempt = 0;; ++attempt) + for (size_t attempt = 0; n > 0; ++attempt) { bool last_attempt = attempt + 1 >= request_settings.max_single_read_retries; + size_t bytes_copied = 0; ProfileEventTimeIncrement watch(ProfileEvents::ReadBufferFromS3Microseconds); @@ -236,14 +235,12 @@ size_t ReadBufferFromS3::readBigAt(char * to, size_t n, size_t range_begin, cons auto result = sendRequest(range_begin, range_begin + n - 1); std::istream & istr = result.GetBody(); - size_t bytes = copyFromIStreamWithProgressCallback(istr, to, n, progress_callback); + copyFromIStreamWithProgressCallback(istr, to, n, progress_callback, &bytes_copied); - ProfileEvents::increment(ProfileEvents::ReadBufferFromS3Bytes, bytes); + ProfileEvents::increment(ProfileEvents::ReadBufferFromS3Bytes, bytes_copied); if (read_settings.remote_throttler) - read_settings.remote_throttler->add(bytes, ProfileEvents::RemoteReadThrottlerBytes, ProfileEvents::RemoteReadThrottlerSleepMicroseconds); - - return bytes; + read_settings.remote_throttler->add(bytes_copied, ProfileEvents::RemoteReadThrottlerBytes, ProfileEvents::RemoteReadThrottlerSleepMicroseconds); } catch (Poco::Exception & e) { @@ -253,7 +250,13 @@ size_t ReadBufferFromS3::readBigAt(char * to, size_t n, size_t range_begin, cons sleepForMilliseconds(sleep_time_with_backoff_milliseconds); sleep_time_with_backoff_milliseconds *= 2; } + + range_begin += bytes_copied; + to += bytes_copied; + n -= bytes_copied; } + + return initial_n; } bool ReadBufferFromS3::processException(Poco::Exception & e, size_t read_offset, size_t attempt) const diff --git a/src/IO/ReadWriteBufferFromHTTP.cpp b/src/IO/ReadWriteBufferFromHTTP.cpp index 7e5c0d37c8e..d0384cfa524 100644 --- a/src/IO/ReadWriteBufferFromHTTP.cpp +++ b/src/IO/ReadWriteBufferFromHTTP.cpp @@ -594,16 +594,14 @@ size_t ReadWriteBufferFromHTTPBase::readBigAt(char * to, si /// This ensures we've sent at least one HTTP request and populated saved_uri_redirect. chassert(file_info && file_info->seekable); - if (n == 0) - return 0; - Poco::URI uri_ = saved_uri_redirect.value_or(uri); if (uri_.getPath().empty()) uri_.setPath("/"); + size_t initial_n = n; size_t milliseconds_to_wait = settings.http_retry_initial_backoff_ms; - for (size_t attempt = 0;; ++attempt) + for (size_t attempt = 0; n > 0; ++attempt) { bool last_attempt = attempt + 1 >= settings.http_max_tries; @@ -616,6 +614,7 @@ size_t ReadWriteBufferFromHTTPBase::readBigAt(char * to, si Poco::Net::HTTPResponse response; std::istream * result_istr; + size_t bytes_copied = 0; try { @@ -629,17 +628,14 @@ size_t ReadWriteBufferFromHTTPBase::readBigAt(char * to, si "Expected 206 Partial Content, got {} when reading {} range [{}, {})", toString(response.getStatus()), uri_.toString(), offset, offset + n); - bool cancelled; - size_t r = copyFromIStreamWithProgressCallback(*result_istr, to, n, progress_callback, &cancelled); - - if (!cancelled) + copyFromIStreamWithProgressCallback(*result_istr, to, n, progress_callback, &bytes_copied); + if (bytes_copied == n) { + result_istr->ignore(UINT64_MAX); /// Response was fully read. - markSessionForReuse(sess); + markSessionForReuse(*sess); ProfileEvents::increment(ProfileEvents::ReadWriteBufferFromHTTPPreservedSessions); } - - return r; } catch (const Poco::Exception & e) { @@ -664,9 +660,15 @@ size_t ReadWriteBufferFromHTTPBase::readBigAt(char * to, si sleepForMilliseconds(milliseconds_to_wait); milliseconds_to_wait = std::min(milliseconds_to_wait * 2, settings.http_retry_max_backoff_ms); - continue; } + + /// Make sure retries don't re-read the bytes that we've already reported to progress_callback. + offset += bytes_copied; + to += bytes_copied; + n -= bytes_copied; } + + return initial_n; } template diff --git a/src/IO/SeekableReadBuffer.cpp b/src/IO/SeekableReadBuffer.cpp index b83e382db01..5d83f4e1b4a 100644 --- a/src/IO/SeekableReadBuffer.cpp +++ b/src/IO/SeekableReadBuffer.cpp @@ -64,7 +64,7 @@ std::unique_ptr wrapSeekableReadBufferPointer(SeekableReadBu return std::make_unique>(*ptr, SeekableReadBufferPtr{ptr}); } -size_t copyFromIStreamWithProgressCallback(std::istream & istr, char * to, size_t n, const std::function & progress_callback, bool * out_cancelled) +void copyFromIStreamWithProgressCallback(std::istream & istr, char * to, size_t n, const std::function & progress_callback, size_t * out_bytes_copied, bool * out_cancelled) { const size_t chunk = DBMS_DEFAULT_BUFFER_SIZE; if (out_cancelled) @@ -82,6 +82,7 @@ size_t copyFromIStreamWithProgressCallback(std::istream & istr, char * to, size_ bool cancelled = false; if (gcount && progress_callback) cancelled = progress_callback(copied); + *out_bytes_copied = copied; if (gcount != to_copy) { @@ -103,7 +104,7 @@ size_t copyFromIStreamWithProgressCallback(std::istream & istr, char * to, size_ } } - return copied; + *out_bytes_copied = copied; } } diff --git a/src/IO/SeekableReadBuffer.h b/src/IO/SeekableReadBuffer.h index 5770948be20..e21513e0ea2 100644 --- a/src/IO/SeekableReadBuffer.h +++ b/src/IO/SeekableReadBuffer.h @@ -98,6 +98,7 @@ std::unique_ptr wrapSeekableReadBufferReference(SeekableRead std::unique_ptr wrapSeekableReadBufferPointer(SeekableReadBufferPtr ptr); /// Helper for implementing readBigAt(). -size_t copyFromIStreamWithProgressCallback(std::istream & istr, char * to, size_t n, const std::function & progress_callback, bool * out_cancelled = nullptr); +/// Updates *out_bytes_copied after each call to the callback, as well as at the end. +void copyFromIStreamWithProgressCallback(std::istream & istr, char * to, size_t n, const std::function & progress_callback, size_t * out_bytes_copied, bool * out_cancelled = nullptr); } diff --git a/tests/integration/test_storage_s3/test.py b/tests/integration/test_storage_s3/test.py index ba0af711151..6b05379c712 100644 --- a/tests/integration/test_storage_s3/test.py +++ b/tests/integration/test_storage_s3/test.py @@ -1018,11 +1018,11 @@ def test_url_reconnect_in_the_middle(started_cluster): def select(): global result result = instance.query( - f"""select sum(cityHash64(x)) from (select toUInt64(id) + sleep(0.1) as x from + f"""select count(), sum(cityHash64(x)) from (select toUInt64(id) + sleep(0.1) as x from url('http://{started_cluster.minio_host}:{started_cluster.minio_port}/{bucket}/{filename}', 'TSV', '{table_format}') settings http_max_tries = 10, http_retry_max_backoff_ms=2000, http_send_timeout=1, http_receive_timeout=1)""" ) - assert int(result) == 3914219105369203805 + assert result == "1000000\t3914219105369203805\n" thread = threading.Thread(target=select) thread.start() @@ -1035,7 +1035,7 @@ def test_url_reconnect_in_the_middle(started_cluster): thread.join() - assert int(result) == 3914219105369203805 + assert result == "1000000\t3914219105369203805\n" def test_seekable_formats(started_cluster): From d5ea047ab8cc11fd60ea91fb876fc98be1aaec99 Mon Sep 17 00:00:00 2001 From: Igor Nikonov Date: Mon, 11 Sep 2023 21:52:40 +0000 Subject: [PATCH 301/433] Parallel replicas: cleanup unused params --- src/Interpreters/ClusterProxy/executeQuery.cpp | 2 -- src/Interpreters/ClusterProxy/executeQuery.h | 1 - src/Processors/QueryPlan/ReadFromRemote.cpp | 2 -- src/Processors/QueryPlan/ReadFromRemote.h | 2 -- src/Storages/StorageMergeTree.cpp | 10 +++++++--- src/Storages/StorageReplicatedMergeTree.cpp | 11 +++++++---- 6 files changed, 14 insertions(+), 14 deletions(-) diff --git a/src/Interpreters/ClusterProxy/executeQuery.cpp b/src/Interpreters/ClusterProxy/executeQuery.cpp index 34be2636917..0890801062e 100644 --- a/src/Interpreters/ClusterProxy/executeQuery.cpp +++ b/src/Interpreters/ClusterProxy/executeQuery.cpp @@ -266,7 +266,6 @@ void executeQuery( void executeQueryWithParallelReplicas( QueryPlan & query_plan, const StorageID & main_table, - const ASTPtr & table_func_ptr, SelectStreamFactory & stream_factory, const ASTPtr & query_ast, ContextPtr context, @@ -328,7 +327,6 @@ void executeQueryWithParallelReplicas( stream_factory.header, stream_factory.processed_stage, main_table, - table_func_ptr, new_context, getThrottler(new_context), std::move(scalars), diff --git a/src/Interpreters/ClusterProxy/executeQuery.h b/src/Interpreters/ClusterProxy/executeQuery.h index b663dffb7fa..5a88f5a5cb1 100644 --- a/src/Interpreters/ClusterProxy/executeQuery.h +++ b/src/Interpreters/ClusterProxy/executeQuery.h @@ -65,7 +65,6 @@ void executeQuery( void executeQueryWithParallelReplicas( QueryPlan & query_plan, const StorageID & main_table, - const ASTPtr & table_func_ptr, SelectStreamFactory & stream_factory, const ASTPtr & query_ast, ContextPtr context, diff --git a/src/Processors/QueryPlan/ReadFromRemote.cpp b/src/Processors/QueryPlan/ReadFromRemote.cpp index 78da19f48a0..f389e5a9e1e 100644 --- a/src/Processors/QueryPlan/ReadFromRemote.cpp +++ b/src/Processors/QueryPlan/ReadFromRemote.cpp @@ -305,7 +305,6 @@ ReadFromParallelRemoteReplicasStep::ReadFromParallelRemoteReplicasStep( Block header_, QueryProcessingStage::Enum stage_, StorageID main_table_, - ASTPtr table_func_ptr_, ContextMutablePtr context_, ThrottlerPtr throttler_, Scalars scalars_, @@ -318,7 +317,6 @@ ReadFromParallelRemoteReplicasStep::ReadFromParallelRemoteReplicasStep( , coordinator(std::move(coordinator_)) , stage(std::move(stage_)) , main_table(std::move(main_table_)) - , table_func_ptr(table_func_ptr_) , context(context_) , throttler(throttler_) , scalars(scalars_) diff --git a/src/Processors/QueryPlan/ReadFromRemote.h b/src/Processors/QueryPlan/ReadFromRemote.h index 43e1904c443..a2486e1eaa1 100644 --- a/src/Processors/QueryPlan/ReadFromRemote.h +++ b/src/Processors/QueryPlan/ReadFromRemote.h @@ -74,7 +74,6 @@ public: Block header_, QueryProcessingStage::Enum stage_, StorageID main_table_, - ASTPtr table_func_ptr_, ContextMutablePtr context_, ThrottlerPtr throttler_, Scalars scalars_, @@ -98,7 +97,6 @@ private: ParallelReplicasReadingCoordinatorPtr coordinator; QueryProcessingStage::Enum stage; StorageID main_table; - ASTPtr table_func_ptr; ContextMutablePtr context; ThrottlerPtr throttler; Scalars scalars; diff --git a/src/Storages/StorageMergeTree.cpp b/src/Storages/StorageMergeTree.cpp index ac2c7baf857..bae91ec8bb1 100644 --- a/src/Storages/StorageMergeTree.cpp +++ b/src/Storages/StorageMergeTree.cpp @@ -237,9 +237,13 @@ void StorageMergeTree::read( processed_stage); ClusterProxy::executeQueryWithParallelReplicas( - query_plan, getStorageID(), /*remove_table_function_ptr*/ nullptr, - select_stream_factory, modified_query_ast, - local_context, query_info.storage_limits, cluster); + query_plan, + getStorageID(), + select_stream_factory, + modified_query_ast, + local_context, + query_info.storage_limits, + cluster); } else { diff --git a/src/Storages/StorageReplicatedMergeTree.cpp b/src/Storages/StorageReplicatedMergeTree.cpp index f9538f60760..fd2ff0ca4b2 100644 --- a/src/Storages/StorageReplicatedMergeTree.cpp +++ b/src/Storages/StorageReplicatedMergeTree.cpp @@ -5187,10 +5187,13 @@ void StorageReplicatedMergeTree::readParallelReplicasImpl( processed_stage); ClusterProxy::executeQueryWithParallelReplicas( - query_plan, getStorageID(), - /* table_func_ptr= */ nullptr, - select_stream_factory, modified_query_ast, - local_context, query_info.storage_limits, parallel_replicas_cluster); + query_plan, + getStorageID(), + select_stream_factory, + modified_query_ast, + local_context, + query_info.storage_limits, + parallel_replicas_cluster); } void StorageReplicatedMergeTree::readLocalImpl( From 60c9dd226dfd811069bc015c99202d6e9363fb10 Mon Sep 17 00:00:00 2001 From: Alexey Gerasimchuck Date: Tue, 12 Sep 2023 10:10:05 +1000 Subject: [PATCH 302/433] Update src/Interpreters/QueryLog.h Co-authored-by: Dmitry Novik --- src/Interpreters/QueryLog.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/Interpreters/QueryLog.h b/src/Interpreters/QueryLog.h index 733be5840e5..fe9b7cbdbc8 100644 --- a/src/Interpreters/QueryLog.h +++ b/src/Interpreters/QueryLog.h @@ -91,7 +91,7 @@ struct QueryLogElement String log_comment; std::vector thread_ids; - UInt64 peak_threads_usage{}; + UInt64 peak_threads_usage = 0; std::shared_ptr profile_counters; std::shared_ptr async_read_counters; std::shared_ptr query_settings; From 30c213a65eca96cddfc0aa2a4acde2d466945168 Mon Sep 17 00:00:00 2001 From: Alexey Gerasimchuck Date: Tue, 12 Sep 2023 10:10:20 +1000 Subject: [PATCH 303/433] Update src/Common/ConcurrencyControl.cpp Co-authored-by: Dmitry Novik --- src/Common/ConcurrencyControl.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/Common/ConcurrencyControl.cpp b/src/Common/ConcurrencyControl.cpp index ebc1c3f9f42..c9fe51550dc 100644 --- a/src/Common/ConcurrencyControl.cpp +++ b/src/Common/ConcurrencyControl.cpp @@ -57,7 +57,7 @@ ConcurrencyControl::Allocation::Allocation(ConcurrencyControl & parent_, SlotCou *waiter = this; } -// Grant single slot to allocation, returns true if more slot(s) are required +// Grant single slot to allocation returns true iff more slot(s) are required bool ConcurrencyControl::Allocation::grant() { std::unique_lock lock{mutex}; From 4890a516b422701907b59674ce90074e7f43a17e Mon Sep 17 00:00:00 2001 From: Alexey Milovidov Date: Tue, 12 Sep 2023 04:50:23 +0300 Subject: [PATCH 304/433] Revert "add runOptimize call in bitmap write method" --- src/AggregateFunctions/AggregateFunctionGroupBitmapData.h | 1 - 1 file changed, 1 deletion(-) diff --git a/src/AggregateFunctions/AggregateFunctionGroupBitmapData.h b/src/AggregateFunctions/AggregateFunctionGroupBitmapData.h index f92f8c1b5e5..7ea1ebe7749 100644 --- a/src/AggregateFunctions/AggregateFunctionGroupBitmapData.h +++ b/src/AggregateFunctions/AggregateFunctionGroupBitmapData.h @@ -151,7 +151,6 @@ public: } else if (BitmapKind::Bitmap == kind) { - roaring_bitmap->runOptimize(); auto size = roaring_bitmap->getSizeInBytes(); writeVarUInt(size, out); std::unique_ptr buf(new char[size]); From 44a9dbd2b701ec299161f32974ec3d502f7eaf83 Mon Sep 17 00:00:00 2001 From: Alexey Milovidov Date: Tue, 12 Sep 2023 04:53:51 +0300 Subject: [PATCH 305/433] Update StorageSystemDetachedParts.cpp --- src/Storages/System/StorageSystemDetachedParts.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/Storages/System/StorageSystemDetachedParts.cpp b/src/Storages/System/StorageSystemDetachedParts.cpp index cec54cef914..a9cd5f2610a 100644 --- a/src/Storages/System/StorageSystemDetachedParts.cpp +++ b/src/Storages/System/StorageSystemDetachedParts.cpp @@ -233,7 +233,7 @@ private: } if (columns_mask[src_index++]) { - Poco::Timestamp modification_time; + Poco::Timestamp modification_time{}; try { modification_time = p.disk->getLastModified(fs::path(current_info.data->getRelativeDataPath()) / MergeTreeData::DETACHED_DIR_NAME / p.dir_name); From ed7b687a696b53d7443432122e954b88bc30883b Mon Sep 17 00:00:00 2001 From: Max Kainov Date: Tue, 12 Sep 2023 08:58:43 +0000 Subject: [PATCH 306/433] parallelize infile --- src/QueryPipeline/QueryPipeline.cpp | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/src/QueryPipeline/QueryPipeline.cpp b/src/QueryPipeline/QueryPipeline.cpp index 9a836f68da1..4250e9c21a3 100644 --- a/src/QueryPipeline/QueryPipeline.cpp +++ b/src/QueryPipeline/QueryPipeline.cpp @@ -353,7 +353,8 @@ QueryPipeline::QueryPipeline(Pipe pipe) totals = pipe.getTotalsPort(); extremes = pipe.getExtremesPort(); partial_result = pipe.getPartialResultPort(0); - + num_threads = pipe.max_parallel_streams; + processors = std::move(pipe.processors); checkPulling(*processors, output, totals, extremes, partial_result); } From 3e16bdbcc7a437f7186b9de05703a71c2278ca5a Mon Sep 17 00:00:00 2001 From: Robert Schulze Date: Tue, 12 Sep 2023 11:10:17 +0000 Subject: [PATCH 307/433] Fix UB --- src/Functions/makeDate.cpp | 14 +++++------- .../02876_yyyymmddhhmmsstodatetime.reference | 22 +++++++++---------- 2 files changed, 17 insertions(+), 19 deletions(-) diff --git a/src/Functions/makeDate.cpp b/src/Functions/makeDate.cpp index bb8c302d139..735fcdfd657 100644 --- a/src/Functions/makeDate.cpp +++ b/src/Functions/makeDate.cpp @@ -16,6 +16,7 @@ #include #include +#include namespace DB { @@ -217,7 +218,7 @@ public: for (size_t i = 0; i < input_rows_count; ++i) { - const auto yyyymmdd = static_cast(yyyymmdd_data[i]); + const auto yyyymmdd = static_cast(static_cast(yyyymmdd_data[i])); /// Float64-to-UInt64 is UB, double-cast avoids it const auto year = yyyymmdd / 10'000; const auto month = yyyymmdd / 100 % 100; @@ -691,7 +692,7 @@ public: { const auto float_date = yyyymmddhhmmss_data[i]; - const auto yyyymmddhhmmss = static_cast(float_date); + const auto yyyymmddhhmmss = static_cast(static_cast(float_date)); /// Float64-to-UInt64 is UB, double-cast avoids it const auto yyyymmdd = yyyymmddhhmmss / 1'000'000; const auto hhmmss = yyyymmddhhmmss % 1'000'000; @@ -705,14 +706,11 @@ public: const auto minute = hhmmss / 100 % 100; const auto second = hhmmss % 100; - auto fraction = static_cast(decimal * fraction_pow); - auto date_time = dateTime(year, month, day, hour, minute, second, date_lut); - result_data[i] = DecimalUtils::decimalFromComponents( - date_time, - static_cast(fraction), - static_cast(precision)); + auto fraction = std::llround(decimal * fraction_pow); + + result_data[i] = DecimalUtils::decimalFromComponents(date_time, fraction, precision); } return res_column; diff --git a/tests/queries/0_stateless/02876_yyyymmddhhmmsstodatetime.reference b/tests/queries/0_stateless/02876_yyyymmddhhmmsstodatetime.reference index 9d010535451..ffef2f3e360 100644 --- a/tests/queries/0_stateless/02876_yyyymmddhhmmsstodatetime.reference +++ b/tests/queries/0_stateless/02876_yyyymmddhhmmsstodatetime.reference @@ -55,20 +55,20 @@ Check correctness, integer arguments 2299-12-31 23:59:59.000 Check correctness, float arguments 1900-01-01 00:00:00.900 -1900-01-01 00:00:00.899 -2020-02-29 11:11:11.101 -2023-09-11 15:05:05.101 -2299-12-31 23:59:59.101 -2299-12-31 23:59:59.101 +1900-01-01 00:00:00.898 +2020-02-29 11:11:11.102 +2023-09-11 15:05:05.102 +2299-12-31 23:59:59.102 +2299-12-31 23:59:59.102 Check correctness, decimal arguments 1900-01-01 00:00:00.900 -1900-01-01 00:00:00.899 -2020-02-29 11:11:11.101 -2023-09-11 15:05:05.101 -2299-12-31 23:59:59.101 -2299-12-31 23:59:59.101 +1900-01-01 00:00:00.898 +2020-02-29 11:11:11.102 +2023-09-11 15:05:05.102 +2299-12-31 23:59:59.102 +2299-12-31 23:59:59.102 Special cases -1900-01-01 00:00:00.648 +1900-01-01 00:00:00.808 1900-01-01 00:00:00.000 1900-01-01 00:00:00.000 1900-01-01 00:00:00.000 From 72d5a5d094772d390115a67419b0400c04feb5e2 Mon Sep 17 00:00:00 2001 From: vdimir Date: Tue, 12 Sep 2023 11:54:35 +0000 Subject: [PATCH 308/433] fix report.py --- tests/ci/report.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tests/ci/report.py b/tests/ci/report.py index 73a2fe6f880..dc31314d031 100644 --- a/tests/ci/report.py +++ b/tests/ci/report.py @@ -148,7 +148,7 @@ FOOTER_HTML_TEMPLATE = """ Date: Tue, 12 Sep 2023 11:55:55 +0000 Subject: [PATCH 309/433] Revert "do not merge this commit" This reverts commit 6a51d9fd7d2a373741f79ac7f9100225d79ef8bd. --- src/Storages/StorageReplicatedMergeTree.cpp | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/src/Storages/StorageReplicatedMergeTree.cpp b/src/Storages/StorageReplicatedMergeTree.cpp index 840a77868d6..e6edb99e301 100644 --- a/src/Storages/StorageReplicatedMergeTree.cpp +++ b/src/Storages/StorageReplicatedMergeTree.cpp @@ -7656,7 +7656,8 @@ void StorageReplicatedMergeTree::replacePartitionFrom( entry_replace.columns_version = -1; } - if (replace) { + if (replace) + { /// Cancel concurrent inserts in range clearLockedBlockNumbersInPartition(*zookeeper, drop_range.partition_id, drop_range.max_block, drop_range.max_block); /// Remove deduplication block_ids of replacing parts From 8db991faa0f781dafb11cd81f8cce4d26a73a5e8 Mon Sep 17 00:00:00 2001 From: Igor Nikonov Date: Tue, 12 Sep 2023 13:41:00 +0000 Subject: [PATCH 310/433] Fix style --- tests/ci/fast_test_check.py | 2 +- tests/ci/stress_check.py | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/tests/ci/fast_test_check.py b/tests/ci/fast_test_check.py index 3d6d43a2d54..13dd6dcb5e2 100644 --- a/tests/ci/fast_test_check.py +++ b/tests/ci/fast_test_check.py @@ -150,7 +150,7 @@ def main(): with TeePopen(run_cmd, run_log_path, timeout=timeout) as process: retcode = process.wait() if process.timeout_exceeded: - logging.info(f"Timeout expired for command: {run_cmd}") + logging.info("Timeout expired for command: %s", run_cmd) timeout_expired = True elif retcode == 0: logging.info("Run successfully") diff --git a/tests/ci/stress_check.py b/tests/ci/stress_check.py index 9e607f7da82..4e79bdc769d 100644 --- a/tests/ci/stress_check.py +++ b/tests/ci/stress_check.py @@ -179,7 +179,7 @@ def run_stress_test(docker_image_name): with TeePopen(run_command, run_log_path, timeout=timeout) as process: retcode = process.wait() if process.timeout_exceeded: - logging.info(f"Timeout expired for command: {run_command}") + logging.info("Timeout expired for command: %s", run_command) timeout_expired = True elif retcode == 0: logging.info("Run successfully") From acb559ce8bd385d311c27aabb3e88fbf4a1cfa22 Mon Sep 17 00:00:00 2001 From: vdimir Date: Tue, 12 Sep 2023 17:34:52 +0200 Subject: [PATCH 311/433] Update tests/ci/commit_status_helper.py Co-authored-by: Mikhail f. Shiryaev --- tests/ci/commit_status_helper.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/tests/ci/commit_status_helper.py b/tests/ci/commit_status_helper.py index 5bcdb11ab6b..a5fd27efb6b 100644 --- a/tests/ci/commit_status_helper.py +++ b/tests/ci/commit_status_helper.py @@ -289,9 +289,9 @@ def create_ci_report(pr_info: PRInfo, statuses: CommitStatuses) -> str: to S3 tests bucket. Then it returns the URL""" test_results = [] # type: TestResults for status in statuses: - log_urls = None + log_urls = [] if status.target_url is not None: - log_urls = [status.target_url] + log_urls.append(status.target_url) raw_logs = status.description or None test_results.append( TestResult( From e406467346471cad47507bfdb0b5596df238b0b0 Mon Sep 17 00:00:00 2001 From: Robert Schulze Date: Tue, 12 Sep 2023 16:28:08 +0000 Subject: [PATCH 312/433] More UB safety --- src/Functions/makeDate.cpp | 12 +++++++++++- .../0_stateless/02876_yyyymmddhhmmsstodatetime.sql | 6 ++++++ tests/queries/0_stateless/02876_yyyymmddtodate.sql | 6 ++++++ 3 files changed, 23 insertions(+), 1 deletion(-) diff --git a/src/Functions/makeDate.cpp b/src/Functions/makeDate.cpp index 735fcdfd657..5090b39977f 100644 --- a/src/Functions/makeDate.cpp +++ b/src/Functions/makeDate.cpp @@ -22,8 +22,9 @@ namespace DB { namespace ErrorCodes { - extern const int ILLEGAL_TYPE_OF_ARGUMENT; + extern const int BAD_ARGUMENTS; extern const int ARGUMENT_OUT_OF_BOUND; + extern const int ILLEGAL_TYPE_OF_ARGUMENT; } namespace @@ -218,6 +219,9 @@ public: for (size_t i = 0; i < input_rows_count; ++i) { + if (std::isinf(yyyymmdd_data[i]) || std::isnan(yyyymmdd_data[i])) [[unlikely]] + throw Exception(ErrorCodes::BAD_ARGUMENTS, "Argument for function {} must be finite", getName()); + const auto yyyymmdd = static_cast(static_cast(yyyymmdd_data[i])); /// Float64-to-UInt64 is UB, double-cast avoids it const auto year = yyyymmdd / 10'000; @@ -602,6 +606,9 @@ public: for (size_t i = 0; i < input_rows_count; i++) { + if (std::isinf(yyyymmddhhmmss_data[i]) || std::isnan(yyyymmddhhmmss_data[i])) [[unlikely]] + throw Exception(ErrorCodes::BAD_ARGUMENTS, "Argument for function {} must be finite", getName()); + const auto yyyymmddhhmmss = static_cast(yyyymmddhhmmss_data[i]); const auto yyyymmdd = yyyymmddhhmmss / 1'000'000; @@ -692,6 +699,9 @@ public: { const auto float_date = yyyymmddhhmmss_data[i]; + if (std::isinf(float_date) || std::isnan(float_date)) [[unlikely]] + throw Exception(ErrorCodes::BAD_ARGUMENTS, "Argument for function {} must be finite", getName()); + const auto yyyymmddhhmmss = static_cast(static_cast(float_date)); /// Float64-to-UInt64 is UB, double-cast avoids it const auto yyyymmdd = yyyymmddhhmmss / 1'000'000; diff --git a/tests/queries/0_stateless/02876_yyyymmddhhmmsstodatetime.sql b/tests/queries/0_stateless/02876_yyyymmddhhmmsstodatetime.sql index 1b5f6220b50..84b3bcc4491 100644 --- a/tests/queries/0_stateless/02876_yyyymmddhhmmsstodatetime.sql +++ b/tests/queries/0_stateless/02876_yyyymmddhhmmsstodatetime.sql @@ -33,6 +33,9 @@ SELECT YYYYMMDDhhmmssToDateTime(20200229111111.1); -- leap day SELECT YYYYMMDDhhmmssToDateTime(20230911150505.1); SELECT YYYYMMDDhhmmssToDateTime(21060207062815.1); SELECT YYYYMMDDhhmmssToDateTime(21060207062816.1); +SELECT YYYYMMDDhhmmssToDateTime(NaN); -- { serverError BAD_ARGUMENTS } +SELECT YYYYMMDDhhmmssToDateTime(Inf); -- { serverError BAD_ARGUMENTS } +SELECT YYYYMMDDhhmmssToDateTime(-Inf); -- { serverError BAD_ARGUMENTS } SELECT 'Check correctness, decimal arguments'; SELECT YYYYMMDDhhmmssToDateTime(toDecimal64(19691231595959.1, 5)); @@ -93,6 +96,9 @@ SELECT YYYYMMDDhhmmssToDateTime64(20200229111111.1); -- leap day SELECT YYYYMMDDhhmmssToDateTime64(20230911150505.1); SELECT YYYYMMDDhhmmssToDateTime64(22991231235959.1); SELECT YYYYMMDDhhmmssToDateTime64(23000101000000.1); +SELECT YYYYMMDDhhmmssToDateTime64(NaN); -- { serverError BAD_ARGUMENTS } +SELECT YYYYMMDDhhmmssToDateTime64(Inf); -- { serverError BAD_ARGUMENTS } +SELECT YYYYMMDDhhmmssToDateTime64(-Inf); -- { serverError BAD_ARGUMENTS } SELECT 'Check correctness, decimal arguments'; SELECT YYYYMMDDhhmmssToDateTime64(toDecimal64(189912315959.1, 5)); diff --git a/tests/queries/0_stateless/02876_yyyymmddtodate.sql b/tests/queries/0_stateless/02876_yyyymmddtodate.sql index fbacca1e2a2..2d4ab46f9ea 100644 --- a/tests/queries/0_stateless/02876_yyyymmddtodate.sql +++ b/tests/queries/0_stateless/02876_yyyymmddtodate.sql @@ -30,6 +30,9 @@ SELECT YYYYMMDDToDate(20200229.1); -- leap day SELECT YYYYMMDDToDate(20230911.1); SELECT YYYYMMDDToDate(21490606.1); SELECT YYYYMMDDToDate(21490607.1); +SELECT YYYYMMDDToDate(NaN); -- { serverError BAD_ARGUMENTS } +SELECT YYYYMMDDToDate(Inf); -- { serverError BAD_ARGUMENTS } +SELECT YYYYMMDDToDate(-Inf); -- { serverError BAD_ARGUMENTS } SELECT 'Check correctness, decimal arguments'; SELECT YYYYMMDDToDate(toDecimal64(19691231.1, 5)); @@ -86,6 +89,9 @@ SELECT YYYYMMDDToDate32(20200229.1); -- leap day SELECT YYYYMMDDToDate32(20230911.1); SELECT YYYYMMDDToDate32(22991231.1); SELECT YYYYMMDDToDate32(23000101.1); +SELECT YYYYMMDDToDate32(NaN); -- { serverError BAD_ARGUMENTS } +SELECT YYYYMMDDToDate32(Inf); -- { serverError BAD_ARGUMENTS } +SELECT YYYYMMDDToDate32(-Inf); -- { serverError BAD_ARGUMENTS } SELECT 'Check correctness, decimal arguments'; SELECT YYYYMMDDToDate32(toDecimal64(18991231.1, 5)); From dddea9219a1327827d01bdabacd4f879c28aea0a Mon Sep 17 00:00:00 2001 From: slvrtrn Date: Tue, 12 Sep 2023 18:39:03 +0200 Subject: [PATCH 313/433] Address the review comments --- .../integration/mysql_java_client/Dockerfile | 11 +- .../PreparedStatementsTest.java | 5 +- .../docker_compose_mysql_java_client.yml | 2 +- docs/en/sql-reference/statements/show.md | 2 +- src/Core/MySQL/PacketsPreparedStatements.cpp | 52 +- src/Core/MySQL/PacketsPreparedStatements.h | 38 +- src/Core/MySQL/PacketsProtocolBinary.cpp | 651 +++++++++--------- src/Core/MySQL/PacketsProtocolBinary.h | 47 +- src/Core/MySQL/PacketsProtocolText.cpp | 5 +- src/Formats/FormatSettings.h | 2 +- .../InterpreterShowColumnsQuery.cpp | 2 +- .../Formats/Impl/MySQLOutputFormat.cpp | 2 +- src/Server/MySQLHandler.cpp | 92 +-- src/Server/MySQLHandler.h | 14 +- .../prepared_statements.reference | 5 - .../prepared_statements_test.sql | 25 +- 16 files changed, 477 insertions(+), 478 deletions(-) diff --git a/docker/test/integration/mysql_java_client/Dockerfile b/docker/test/integration/mysql_java_client/Dockerfile index 75e3c9de138..83c17551481 100644 --- a/docker/test/integration/mysql_java_client/Dockerfile +++ b/docker/test/integration/mysql_java_client/Dockerfile @@ -1,16 +1,9 @@ # docker build -t clickhouse/mysql-java-client . # MySQL Java client docker container -FROM ubuntu:22.04 +FROM openjdk:8-jdk-alpine -RUN apt-get update && \ - apt-get install -y software-properties-common build-essential openjdk-8-jdk curl - -RUN rm -rf \ - /var/lib/apt/lists/* \ - /var/cache/debconf \ - /tmp/* \ -RUN apt-get clean +RUN apk --no-cache add curl ARG ver=8.1.0 RUN curl -L -o /mysql-connector-j-${ver}.jar https://repo1.maven.org/maven2/com/mysql/mysql-connector-j/${ver}/mysql-connector-j-${ver}.jar diff --git a/docker/test/integration/mysql_java_client/PreparedStatementsTest.java b/docker/test/integration/mysql_java_client/PreparedStatementsTest.java index 9efb3330fd5..dc72f8d3526 100644 --- a/docker/test/integration/mysql_java_client/PreparedStatementsTest.java +++ b/docker/test/integration/mysql_java_client/PreparedStatementsTest.java @@ -33,8 +33,8 @@ public class PreparedStatementsTest { } } - // useServerPrepStmts allows us to send COM_STMT_PREPARE and COM_STMT_EXECUTE - // to test the binary protocol implementation + // useServerPrepStmts uses COM_STMT_PREPARE and COM_STMT_EXECUTE + // instead of COM_QUERY which allows us to test the binary protocol String jdbcUrl = String.format("jdbc:mysql://%s:%s/%s?useSSL=false&useServerPrepStmts=true", host, port, database); @@ -174,4 +174,5 @@ public class PreparedStatementsTest { return String.format("%s type is %s", columnLabel, MysqlType.getByJdbcType(meta.getColumnType(rs.findColumn(columnLabel)))); } + } diff --git a/docker/test/integration/runner/compose/docker_compose_mysql_java_client.yml b/docker/test/integration/runner/compose/docker_compose_mysql_java_client.yml index eb5ffb01baa..529974dd4bf 100644 --- a/docker/test/integration/runner/compose/docker_compose_mysql_java_client.yml +++ b/docker/test/integration/runner/compose/docker_compose_mysql_java_client.yml @@ -3,4 +3,4 @@ services: java1: image: clickhouse/mysql-java-client:${DOCKER_MYSQL_JAVA_CLIENT_TAG:-latest} # to keep container running - command: sleep infinity + command: sleep 1d diff --git a/docs/en/sql-reference/statements/show.md b/docs/en/sql-reference/statements/show.md index e94718394de..4341cf2998e 100644 --- a/docs/en/sql-reference/statements/show.md +++ b/docs/en/sql-reference/statements/show.md @@ -208,7 +208,7 @@ The optional keyword `FULL` causes the output to include the collation, comment The statement produces a result table with the following structure: - field - The name of the column (String) - type - The column data type (String) -- null - If the column data type is Nullable (UInt8) +- null - If the column data type is Nullable (String) - key - `PRI` if the column is part of the primary key, `SOR` if the column is part of the sorting key, empty otherwise (String) - default - Default expression of the column if it is of type `ALIAS`, `DEFAULT`, or `MATERIALIZED`, otherwise `NULL`. (Nullable(String)) - extra - Additional information, currently unused (String) diff --git a/src/Core/MySQL/PacketsPreparedStatements.cpp b/src/Core/MySQL/PacketsPreparedStatements.cpp index dcff294ebad..654c28379d9 100644 --- a/src/Core/MySQL/PacketsPreparedStatements.cpp +++ b/src/Core/MySQL/PacketsPreparedStatements.cpp @@ -2,33 +2,41 @@ #include #include + namespace DB { namespace MySQLProtocol { - namespace PreparedStatements - { - size_t PrepareStatementResponseOK::getPayloadSize() const - { - return 13; - } +namespace PreparedStatements +{ +size_t PreparedStatementResponseOK::getPayloadSize() const +{ + // total = 13 + return 1 // status + + 4 // statement_id + + 2 // num_columns + + 2 // num_params + + 1 // reserved_1 (filler) + + 2 // warnings_count + + 1; // metadata_follows +} - void PrepareStatementResponseOK::writePayloadImpl(WriteBuffer & buffer) const - { - buffer.write(reinterpret_cast(&status), 1); - buffer.write(reinterpret_cast(&statement_id), 4); - buffer.write(reinterpret_cast(&num_columns), 2); - buffer.write(reinterpret_cast(&num_params), 2); - buffer.write(reinterpret_cast(&reserved_1), 1); - buffer.write(reinterpret_cast(&warnings_count), 2); - buffer.write(0x0); // RESULTSET_METADATA_NONE - } +void PreparedStatementResponseOK::writePayloadImpl(WriteBuffer & buffer) const +{ + buffer.write(reinterpret_cast(&status), 1); + buffer.write(reinterpret_cast(&statement_id), 4); + buffer.write(reinterpret_cast(&num_columns), 2); + buffer.write(reinterpret_cast(&num_params), 2); + buffer.write(reinterpret_cast(&reserved_1), 1); + buffer.write(reinterpret_cast(&warnings_count), 2); + buffer.write(0x0); // RESULTSET_METADATA_NONE +} - PrepareStatementResponseOK::PrepareStatementResponseOK( - uint32_t statement_id_, uint16_t num_columns_, uint16_t num_params_, uint16_t warnings_count_) - : statement_id(statement_id_), num_columns(num_columns_), num_params(num_params_), warnings_count(warnings_count_) - { - } - } +PreparedStatementResponseOK::PreparedStatementResponseOK( + uint32_t statement_id_, uint16_t num_columns_, uint16_t num_params_, uint16_t warnings_count_) + : statement_id(statement_id_), num_columns(num_columns_), num_params(num_params_), warnings_count(warnings_count_) +{ +} +} } } diff --git a/src/Core/MySQL/PacketsPreparedStatements.h b/src/Core/MySQL/PacketsPreparedStatements.h index 5aca459fc3a..2c5ab8b5596 100644 --- a/src/Core/MySQL/PacketsPreparedStatements.h +++ b/src/Core/MySQL/PacketsPreparedStatements.h @@ -6,27 +6,27 @@ namespace DB { namespace MySQLProtocol { - namespace PreparedStatements - { - // https://dev.mysql.com/doc/dev/mysql-server/latest/page_protocol_com_stmt_prepare.html#sect_protocol_com_stmt_prepare_response_ok - class PrepareStatementResponseOK : public IMySQLWritePacket - { - public: - uint8_t status = 0x00; - uint32_t statement_id; - uint16_t num_columns; - uint16_t num_params; - uint8_t reserved_1 = 0; - uint16_t warnings_count; +namespace PreparedStatements +{ +// https://dev.mysql.com/doc/dev/mysql-server/latest/page_protocol_com_stmt_prepare.html#sect_protocol_com_stmt_prepare_response_ok +class PreparedStatementResponseOK : public IMySQLWritePacket +{ +public: + const uint8_t status = 0x00; + uint32_t statement_id; + uint16_t num_columns; + uint16_t num_params; + const uint8_t reserved_1 = 0; + uint16_t warnings_count; - protected: - size_t getPayloadSize() const override; +protected: + size_t getPayloadSize() const override; - void writePayloadImpl(WriteBuffer & buffer) const override; + void writePayloadImpl(WriteBuffer & buffer) const override; - public: - PrepareStatementResponseOK(uint32_t statement_id_, uint16_t num_columns_, uint16_t num_params_, uint16_t warnings_count_); - }; - } +public: + PreparedStatementResponseOK(uint32_t statement_id_, uint16_t num_columns_, uint16_t num_params_, uint16_t warnings_count_); +}; +} } } diff --git a/src/Core/MySQL/PacketsProtocolBinary.cpp b/src/Core/MySQL/PacketsProtocolBinary.cpp index 9e5e2993595..13014e5a46b 100644 --- a/src/Core/MySQL/PacketsProtocolBinary.cpp +++ b/src/Core/MySQL/PacketsProtocolBinary.cpp @@ -20,344 +20,337 @@ namespace DB { - namespace MySQLProtocol { - - namespace ProtocolBinary +namespace ProtocolBinary +{ +ResultSetRow::ResultSetRow(const Serializations & serializations_, const DataTypes & data_types_, const Columns & columns_, int row_num_) + : row_num(row_num_), columns(columns_), data_types(data_types_), serializations(serializations_) +{ + payload_size = 1 + null_bitmap_size; + FormatSettings format_settings; + for (size_t i = 0; i < columns.size(); ++i) { - ResultSetRow::ResultSetRow( - const Serializations & serializations_, const DataTypes & data_types_, const Columns & columns_, int row_num_) - : row_num(row_num_), columns(columns_), data_types(data_types_), serializations(serializations_) + ColumnPtr col = columns[i]; + if (col->isNullAt(row_num)) { - /// See https://dev.mysql.com/doc/dev/mysql-server/8.1.0/page_protocol_binary_resultset.html#sect_protocol_binary_resultset_row - payload_size = 1 + null_bitmap_size; - FormatSettings format_settings; - for (size_t i = 0; i < columns.size(); ++i) - { - ColumnPtr col = columns[i]; - if (col->isNullAt(row_num)) - { - size_t byte = (i + 2) / 8; - int bit = 1 << ((i + 2) % 8); - null_bitmap[byte] |= bit; - continue; // NULLs are stored in the null bitmap only - } - - DataTypePtr data_type = removeLowCardinality(removeNullable((data_types[i]))); - TypeIndex type_index = data_type->getTypeId(); - switch (type_index) - { - case TypeIndex::Int8: - payload_size += 1; - break; - case TypeIndex::UInt8: - if (data_type->getName() == "Bool") - { - payload_size += 2; // BIT MySQL type is string in binary - } - else - { - payload_size += 1; - } - break; - case TypeIndex::Int16: - case TypeIndex::UInt16: - payload_size += 2; - break; - case TypeIndex::Int32: - case TypeIndex::UInt32: - case TypeIndex::Float32: - payload_size += 4; - break; - case TypeIndex::Int64: - case TypeIndex::UInt64: - case TypeIndex::Float64: - payload_size += 8; - break; - case TypeIndex::Date: { - payload_size += 5; - break; - } - case TypeIndex::Date32: { - payload_size += 5; - break; - } - case TypeIndex::DateTime: { - UInt32 value = assert_cast &>(*col).getData()[row_num]; - LocalDateTime ldt = LocalDateTime(value, DateLUT::instance(getDateTimeTimezone(*data_type))); - - bool has_time = !(ldt.hour() == 0 && ldt.minute() == 0 && ldt.second() == 0); - if (has_time) - { - payload_size += 8; - } - else - { - payload_size += 5; - } - break; - } - case TypeIndex::DateTime64: { - const auto * date_time_type = typeid_cast(data_type.get()); - UInt32 scale = date_time_type->getScale(); - - static constexpr UInt32 MaxScale = DecimalUtils::max_precision; - scale = scale > MaxScale ? MaxScale : scale; - - const auto dt64 = assert_cast(*col).getData()[row_num]; - auto components = DecimalUtils::split(dt64, scale); - - using T = typename DateTime64::NativeType; - if (dt64.value < 0 && components.fractional) - { - components.fractional - = DecimalUtils::scaleMultiplier(scale) + (components.whole ? T(-1) : T(1)) * components.fractional; - --components.whole; - } - if (scale > 6) - { - // MySQL Timestamp has max scale of 6 - components.fractional /= static_cast(pow(10, scale - 6)); - } - - LocalDateTime ldt = LocalDateTime(components.whole, DateLUT::instance(getDateTimeTimezone(*data_type))); - - bool has_microseconds = components.fractional != 0; - bool has_time = !(ldt.hour() == 0 && ldt.minute() == 0 && ldt.second() == 0); - if (has_microseconds) - { - payload_size += 12; - } - else if (has_time) - { - payload_size += 8; - } - else - { - payload_size += 5; - } - break; - } - default: - WriteBufferFromOwnString ostr; - serializations[i]->serializeText(*columns[i], row_num, ostr, format_settings); - payload_size += getLengthEncodedStringSize(ostr.str()); - serialized[i] = std::move(ostr.str()); - break; - } - } + // See https://dev.mysql.com/doc/dev/mysql-server/8.1.0/page_protocol_binary_resultset.html#sect_protocol_binary_resultset_row + size_t byte = (i + 2) / 8; + int bit = 1 << ((i + 2) % 8); + null_bitmap[byte] |= bit; + continue; // NULLs are stored in the null bitmap only } - void ResultSetRow::writePayloadImpl(WriteBuffer & buffer) const + DataTypePtr data_type = removeLowCardinality(removeNullable((data_types[i]))); + TypeIndex type_index = data_type->getTypeId(); + switch (type_index) { - buffer.write(static_cast(0x00)); - buffer.write(null_bitmap.data(), null_bitmap_size); - for (size_t i = 0; i < columns.size(); ++i) - { - ColumnPtr col = columns[i]; - if (col->isNullAt(row_num)) - { - continue; - } - - DataTypePtr data_type = removeLowCardinality(removeNullable((data_types[i]))); - TypeIndex type_index = data_type->getTypeId(); - switch (type_index) - { - case TypeIndex::UInt8: { - UInt8 value = assert_cast &>(*col).getData()[row_num]; - if (data_type->getName() == "Bool") - { - buffer.write(static_cast(1)); - } - buffer.write(reinterpret_cast(&value), 1); - break; - } - case TypeIndex::UInt16: { - UInt16 value = assert_cast &>(*col).getData()[row_num]; - buffer.write(reinterpret_cast(&value), 2); - break; - } - case TypeIndex::UInt32: { - UInt32 value = assert_cast &>(*col).getData()[row_num]; - buffer.write(reinterpret_cast(&value), 4); - break; - } - case TypeIndex::UInt64: { - UInt64 value = assert_cast &>(*col).getData()[row_num]; - buffer.write(reinterpret_cast(&value), 8); - break; - } - case TypeIndex::Int8: { - Int8 value = assert_cast &>(*col).getData()[row_num]; - buffer.write(reinterpret_cast(&value), 1); - break; - } - case TypeIndex::Int16: { - Int16 value = assert_cast &>(*col).getData()[row_num]; - buffer.write(reinterpret_cast(&value), 2); - break; - } - case TypeIndex::Int32: { - Int32 value = assert_cast &>(*col).getData()[row_num]; - buffer.write(reinterpret_cast(&value), 4); - break; - } - case TypeIndex::Int64: { - Int64 value = assert_cast &>(*col).getData()[row_num]; - buffer.write(reinterpret_cast(&value), 8); - break; - } - case TypeIndex::Float32: { - Float32 value = assert_cast &>(*col).getData()[row_num]; - buffer.write(reinterpret_cast(&value), 4); - break; - } - case TypeIndex::Float64: { - Float64 value = assert_cast &>(*col).getData()[row_num]; - buffer.write(reinterpret_cast(&value), 8); - break; - } - case TypeIndex::Date: { - UInt16 value = assert_cast &>(*col).getData()[row_num]; - LocalDate ld = LocalDate(DayNum(value)); - buffer.write(static_cast(4)); // bytes_following - auto year = ld.year(); - auto month = ld.month(); - auto day = ld.day(); - buffer.write(reinterpret_cast(&year), 2); - buffer.write(reinterpret_cast(&month), 1); - buffer.write(reinterpret_cast(&day), 1); - break; - } - case TypeIndex::Date32: { - Int32 value = assert_cast &>(*col).getData()[row_num]; - LocalDate ld = LocalDate(ExtendedDayNum(value)); - buffer.write(static_cast(4)); // bytes_following - auto year = ld.year(); - auto month = ld.month(); - auto day = ld.day(); - buffer.write(reinterpret_cast(&year), 2); - buffer.write(reinterpret_cast(&month), 1); - buffer.write(reinterpret_cast(&day), 1); - break; - } - case TypeIndex::DateTime: { - UInt32 value = assert_cast &>(*col).getData()[row_num]; - String timezone = getDateTimeTimezone(*data_type); - LocalDateTime ldt = LocalDateTime(value, DateLUT::instance(timezone)); - int year = ldt.year(); - int month = ldt.month(); - int day = ldt.day(); - int hour = ldt.hour(); - int minute = ldt.minute(); - int second = ldt.second(); - bool has_time = !(hour == 0 && minute == 0 && second == 0); - size_t bytes_following = has_time ? 7 : 4; - buffer.write(reinterpret_cast(&bytes_following), 1); - buffer.write(reinterpret_cast(&year), 2); - buffer.write(reinterpret_cast(&month), 1); - buffer.write(reinterpret_cast(&day), 1); - if (has_time) - { - buffer.write(reinterpret_cast(&hour), 1); - buffer.write(reinterpret_cast(&minute), 1); - buffer.write(reinterpret_cast(&second), 1); - } - break; - } - case TypeIndex::DateTime64: { - const auto * date_time_type = typeid_cast(data_type.get()); - UInt32 scale = date_time_type->getScale(); - - static constexpr UInt32 MaxScale = DecimalUtils::max_precision; - scale = scale > MaxScale ? MaxScale : scale; - - const auto dt64 = assert_cast(*col).getData()[row_num]; - auto components = DecimalUtils::split(dt64, scale); - - using T = typename DateTime64::NativeType; - if (dt64.value < 0 && components.fractional) - { - components.fractional - = DecimalUtils::scaleMultiplier(scale) + (components.whole ? T(-1) : T(1)) * components.fractional; - --components.whole; - } - - if (components.fractional != 0) - { - if (scale > 6) - { - // MySQL Timestamp has max scale of 6 - components.fractional /= static_cast(pow(10, scale - 6)); - } - else - { - // fractional == 1 is a different microsecond value depending on the scale - // Scale 1 = 100 000 - // Scale 2 = 010 000 - // Scale 3 = 001 000 - // Scale 4 = 000 100 - // Scale 5 = 000 010 - // Scale 6 = 000 001 - components.fractional *= static_cast(pow(10, 6 - scale)); - } - } - - String timezone = getDateTimeTimezone(*data_type); - LocalDateTime ldt = LocalDateTime(components.whole, DateLUT::instance(timezone)); - auto year = ldt.year(); - auto month = ldt.month(); - auto day = ldt.day(); - auto hour = ldt.hour(); - auto minute = ldt.minute(); - auto second = ldt.second(); - - bool has_time = !(hour == 0 && minute == 0 && second == 0); - bool has_microseconds = components.fractional != 0; - - if (has_microseconds) - { - buffer.write(static_cast(11)); // bytes_following - buffer.write(reinterpret_cast(&year), 2); - buffer.write(reinterpret_cast(&month), 1); - buffer.write(reinterpret_cast(&day), 1); - buffer.write(reinterpret_cast(&hour), 1); - buffer.write(reinterpret_cast(&minute), 1); - buffer.write(reinterpret_cast(&second), 1); - buffer.write(reinterpret_cast(&components.fractional), 4); - } - else if (has_time) - { - buffer.write(static_cast(7)); // bytes_following - buffer.write(reinterpret_cast(&year), 2); - buffer.write(reinterpret_cast(&month), 1); - buffer.write(reinterpret_cast(&day), 1); - buffer.write(reinterpret_cast(&hour), 1); - buffer.write(reinterpret_cast(&minute), 1); - buffer.write(reinterpret_cast(&second), 1); - } - else - { - buffer.write(static_cast(4)); // bytes_following - buffer.write(reinterpret_cast(&year), 2); - buffer.write(reinterpret_cast(&month), 1); - buffer.write(reinterpret_cast(&day), 1); - } - break; - } - default: - writeLengthEncodedString(serialized[i], buffer); - break; - } + case TypeIndex::Int8: + payload_size += 1; + break; + case TypeIndex::UInt8: + if (data_type->getName() == "Bool") + payload_size += 2; // BIT MySQL type is string in binary + else + payload_size += 1; + break; + case TypeIndex::Int16: + case TypeIndex::UInt16: + payload_size += 2; + break; + case TypeIndex::Int32: + case TypeIndex::UInt32: + case TypeIndex::Float32: + payload_size += 4; + break; + case TypeIndex::Int64: + case TypeIndex::UInt64: + case TypeIndex::Float64: + payload_size += 8; + break; + case TypeIndex::Date: + case TypeIndex::Date32: { + size_t size = 1 // number of bytes following + + 2 // year + + 1 // month + + 1; // day + payload_size += size; + break; } - } + case TypeIndex::DateTime: { + UInt32 value = assert_cast &>(*col).getData()[row_num]; + LocalDateTime ldt = LocalDateTime(value, DateLUT::instance(getDateTimeTimezone(*data_type))); - size_t ResultSetRow::getPayloadSize() const - { - return payload_size; - }; + bool has_time = !(ldt.hour() == 0 && ldt.minute() == 0 && ldt.second() == 0); + size_t size = 1 // number of bytes following + + 2 // year + + 1 // month + + 1; // day + payload_size += size; + if (has_time) + { + size_t additional_size = 1 // hour + + 1 // minute + + 1; // second + payload_size += additional_size; + } + break; + } + case TypeIndex::DateTime64: { + auto [components, scale] = getDateTime64ComponentsWithScale(data_type, col); + if (scale > 6) + { + // MySQL Timestamp has max scale of 6 + components.fractional /= static_cast(pow(10, scale - 6)); + } + + LocalDateTime ldt = LocalDateTime(components.whole, DateLUT::instance(getDateTimeTimezone(*data_type))); + + bool has_microseconds = components.fractional != 0; + bool has_time = !(ldt.hour() == 0 && ldt.minute() == 0 && ldt.second() == 0); + size_t size = 1 // number of bytes following + + 2 // year + + 1 // month + + 1; // day + payload_size += size; + if (has_microseconds) + { + size_t additional_size = 1 // hour + + 1 // minute + + 1 // second + + 4; // microsecond; + payload_size += additional_size; + } + else if (has_time) + { + size_t additional_size = 1 // hour + + 1 // minute + + 1; // second + payload_size += additional_size; + } + break; + } + // All other types including all Decimal types are string in binary + default: + WriteBufferFromOwnString ostr; + serializations[i]->serializeText(*columns[i], row_num, ostr, format_settings); + payload_size += getLengthEncodedStringSize(ostr.str()); + serialized[i] = std::move(ostr.str()); + break; + } } } + +size_t ResultSetRow::getPayloadSize() const +{ + return payload_size; +} + +void ResultSetRow::writePayloadImpl(WriteBuffer & buffer) const +{ + buffer.write(static_cast(0x00)); + buffer.write(null_bitmap.data(), null_bitmap_size); + for (size_t i = 0; i < columns.size(); ++i) + { + ColumnPtr col = columns[i]; + if (col->isNullAt(row_num)) + continue; + + DataTypePtr data_type = removeLowCardinality(removeNullable((data_types[i]))); + TypeIndex type_index = data_type->getTypeId(); + switch (type_index) + { + case TypeIndex::Int8: { + Int8 value = assert_cast &>(*col).getData()[row_num]; + buffer.write(reinterpret_cast(&value), 1); + break; + } + case TypeIndex::UInt8: { + UInt8 value = assert_cast &>(*col).getData()[row_num]; + if (data_type->getName() == "Bool") + buffer.write(static_cast(1)); + buffer.write(reinterpret_cast(&value), 1); + break; + } + case TypeIndex::Int16: { + Int16 value = assert_cast &>(*col).getData()[row_num]; + buffer.write(reinterpret_cast(&value), 2); + break; + } + case TypeIndex::UInt16: { + UInt16 value = assert_cast &>(*col).getData()[row_num]; + buffer.write(reinterpret_cast(&value), 2); + break; + } + case TypeIndex::Int32: { + Int32 value = assert_cast &>(*col).getData()[row_num]; + buffer.write(reinterpret_cast(&value), 4); + break; + } + case TypeIndex::UInt32: { + UInt32 value = assert_cast &>(*col).getData()[row_num]; + buffer.write(reinterpret_cast(&value), 4); + break; + } + case TypeIndex::Float32: { + Float32 value = assert_cast &>(*col).getData()[row_num]; + buffer.write(reinterpret_cast(&value), 4); + break; + } + case TypeIndex::Int64: { + Int64 value = assert_cast &>(*col).getData()[row_num]; + buffer.write(reinterpret_cast(&value), 8); + break; + } + case TypeIndex::UInt64: { + UInt64 value = assert_cast &>(*col).getData()[row_num]; + buffer.write(reinterpret_cast(&value), 8); + break; + } + case TypeIndex::Float64: { + Float64 value = assert_cast &>(*col).getData()[row_num]; + buffer.write(reinterpret_cast(&value), 8); + break; + } + case TypeIndex::Date: { + UInt16 value = assert_cast &>(*col).getData()[row_num]; + LocalDate ld = LocalDate(DayNum(value)); + buffer.write(static_cast(4)); // bytes_following + UInt16 year = ld.year(); + UInt8 month = ld.month(); + UInt8 day = ld.day(); + buffer.write(reinterpret_cast(&year), 2); + buffer.write(reinterpret_cast(&month), 1); + buffer.write(reinterpret_cast(&day), 1); + break; + } + case TypeIndex::Date32: { + Int32 value = assert_cast &>(*col).getData()[row_num]; + LocalDate ld = LocalDate(ExtendedDayNum(value)); + buffer.write(static_cast(4)); // bytes_following + UInt16 year = ld.year(); + UInt8 month = ld.month(); + UInt8 day = ld.day(); + buffer.write(reinterpret_cast(&year), 2); + buffer.write(reinterpret_cast(&month), 1); + buffer.write(reinterpret_cast(&day), 1); + break; + } + case TypeIndex::DateTime: { + UInt32 value = assert_cast &>(*col).getData()[row_num]; + String timezone = getDateTimeTimezone(*data_type); + LocalDateTime ldt = LocalDateTime(value, DateLUT::instance(timezone)); + UInt16 year = ldt.year(); + UInt8 month = ldt.month(); + UInt8 day = ldt.day(); + UInt8 hour = ldt.hour(); + UInt8 minute = ldt.minute(); + UInt8 second = ldt.second(); + bool has_time = !(hour == 0 && minute == 0 && second == 0); + size_t bytes_following = has_time ? 7 : 4; + buffer.write(reinterpret_cast(&bytes_following), 1); + buffer.write(reinterpret_cast(&year), 2); + buffer.write(reinterpret_cast(&month), 1); + buffer.write(reinterpret_cast(&day), 1); + if (has_time) + { + buffer.write(reinterpret_cast(&hour), 1); + buffer.write(reinterpret_cast(&minute), 1); + buffer.write(reinterpret_cast(&second), 1); + } + break; + } + case TypeIndex::DateTime64: { + auto [components, scale] = getDateTime64ComponentsWithScale(data_type, col); + if (components.fractional != 0) + { + if (scale > 6) + { + // MySQL Timestamp has max scale of 6 + components.fractional /= static_cast(pow(10, scale - 6)); + } + else + { + // fractional == 1 is a different microsecond value depending on the scale + // Scale 1 = 100000 + // Scale 2 = 010000 + // Scale 3 = 001000 + // Scale 4 = 000100 + // Scale 5 = 000010 + // Scale 6 = 000001 + components.fractional *= static_cast(pow(10, 6 - scale)); + } + } + + String timezone = getDateTimeTimezone(*data_type); + LocalDateTime ldt = LocalDateTime(components.whole, DateLUT::instance(timezone)); + UInt16 year = ldt.year(); + UInt8 month = ldt.month(); + UInt8 day = ldt.day(); + UInt8 hour = ldt.hour(); + UInt8 minute = ldt.minute(); + UInt8 second = ldt.second(); + + bool has_time = !(hour == 0 && minute == 0 && second == 0); + bool has_microseconds = components.fractional != 0; + + if (has_microseconds) + { + buffer.write(static_cast(11)); // bytes_following + buffer.write(reinterpret_cast(&year), 2); + buffer.write(reinterpret_cast(&month), 1); + buffer.write(reinterpret_cast(&day), 1); + buffer.write(reinterpret_cast(&hour), 1); + buffer.write(reinterpret_cast(&minute), 1); + buffer.write(reinterpret_cast(&second), 1); + buffer.write(reinterpret_cast(&components.fractional), 4); + } + else if (has_time) + { + buffer.write(static_cast(7)); // bytes_following + buffer.write(reinterpret_cast(&year), 2); + buffer.write(reinterpret_cast(&month), 1); + buffer.write(reinterpret_cast(&day), 1); + buffer.write(reinterpret_cast(&hour), 1); + buffer.write(reinterpret_cast(&minute), 1); + buffer.write(reinterpret_cast(&second), 1); + } + else + { + buffer.write(static_cast(4)); // bytes_following + buffer.write(reinterpret_cast(&year), 2); + buffer.write(reinterpret_cast(&month), 1); + buffer.write(reinterpret_cast(&day), 1); + } + break; + } + // All other types including all Decimal types are string in binary + default: + writeLengthEncodedString(serialized[i], buffer); + break; + } + } +} + +ResultSetRow::DateTime64ComponentsWithScale ResultSetRow::getDateTime64ComponentsWithScale(DataTypePtr data_type, ColumnPtr col) const +{ + const auto * date_time_type = typeid_cast(data_type.get()); + + static constexpr UInt32 MaxScale = DecimalUtils::max_precision; + UInt32 scale = std::min(MaxScale, date_time_type->getScale()); + + const auto value = assert_cast(*col).getData()[row_num]; + auto components = DecimalUtils::split(value, scale); + + using T = typename DateTime64::NativeType; + if (value.value < 0 && components.fractional) + { + components.fractional = DecimalUtils::scaleMultiplier(scale) + (components.whole ? T(-1) : T(1)) * components.fractional; + --components.whole; + } + + return std::make_pair(components, scale); +} +} +} } diff --git a/src/Core/MySQL/PacketsProtocolBinary.h b/src/Core/MySQL/PacketsProtocolBinary.h index 1a201ce3a4a..d2fcc783825 100644 --- a/src/Core/MySQL/PacketsProtocolBinary.h +++ b/src/Core/MySQL/PacketsProtocolBinary.h @@ -4,42 +4,45 @@ #include #include #include +#include "Common/LocalDateTime.h" +#include "Core/DecimalFunctions.h" #include "DataTypes/IDataType.h" #include "DataTypes/Serializations/ISerialization.h" namespace DB { - namespace MySQLProtocol { +namespace ProtocolBinary +{ +class ResultSetRow : public IMySQLWritePacket +{ + using DateTime64ComponentsWithScale = std::pair, UInt32>; - namespace ProtocolBinary - { - class ResultSetRow : public IMySQLWritePacket - { - private: - TypeIndex getTypeIndex(DataTypePtr data_type, const ColumnPtr & col) const; +private: + DateTime64ComponentsWithScale getDateTime64ComponentsWithScale(DataTypePtr data_type, ColumnPtr col) const; - protected: - int row_num; - const Columns & columns; - const DataTypes & data_types; - const Serializations & serializations; +protected: + int row_num; + const Columns & columns; + const DataTypes & data_types; + const Serializations & serializations; - std::vector serialized = std::vector(columns.size()); + std::vector serialized = std::vector(columns.size()); - size_t null_bitmap_size = (columns.size() + 7 + 2) / 8; - std::vector null_bitmap = std::vector(null_bitmap_size, static_cast(0)); + // See https://dev.mysql.com/doc/dev/mysql-server/8.1.0/page_protocol_binary_resultset.html#sect_protocol_binary_resultset_row + size_t null_bitmap_size = (columns.size() + 7 + 2) / 8; + std::vector null_bitmap = std::vector(null_bitmap_size, static_cast(0)); - size_t payload_size = 0; + size_t payload_size = 0; - size_t getPayloadSize() const override; + size_t getPayloadSize() const override; - void writePayloadImpl(WriteBuffer & buffer) const override; + void writePayloadImpl(WriteBuffer & buffer) const override; - public: - ResultSetRow(const Serializations & serializations_, const DataTypes & data_types_, const Columns & columns_, int row_num_); - }; - } +public: + ResultSetRow(const Serializations & serializations_, const DataTypes & data_types_, const Columns & columns_, int row_num_); +}; +} } } diff --git a/src/Core/MySQL/PacketsProtocolText.cpp b/src/Core/MySQL/PacketsProtocolText.cpp index ebfff1c2370..b26346a9049 100644 --- a/src/Core/MySQL/PacketsProtocolText.cpp +++ b/src/Core/MySQL/PacketsProtocolText.cpp @@ -198,9 +198,6 @@ ColumnDefinition getColumnDefinition(const String & column_name, const DataTypeP flags = ColumnDefinitionFlags::BINARY_FLAG; break; case TypeIndex::DateTime: - column_type = ColumnType::MYSQL_TYPE_DATETIME; - flags = ColumnDefinitionFlags::BINARY_FLAG; - break; case TypeIndex::DateTime64: column_type = ColumnType::MYSQL_TYPE_DATETIME; flags = ColumnDefinitionFlags::BINARY_FLAG; @@ -212,6 +209,8 @@ ColumnDefinition getColumnDefinition(const String & column_name, const DataTypeP break; case TypeIndex::Decimal128: { // MySQL Decimal has max 65 precision and 30 scale + // Decimal256 (min scale is 39) is higher than the MySQL supported range and handled in the default case + // See https://dev.mysql.com/doc/refman/8.0/en/precision-math-decimal-characteristics.html const auto & type = assert_cast(*data_type); if (type.getPrecision() > 65 || type.getScale() > 30) { diff --git a/src/Formats/FormatSettings.h b/src/Formats/FormatSettings.h index 12f24d0ebe6..db35f8925a5 100644 --- a/src/Formats/FormatSettings.h +++ b/src/Formats/FormatSettings.h @@ -295,7 +295,7 @@ struct FormatSettings * https://dev.mysql.com/doc/dev/mysql-server/latest/page_protocol_com_stmt_execute_response.html * By default, use Text ResultSet. */ - bool use_binary_result_set = false; + bool binary_protocol = false; } mysql_wire; struct diff --git a/src/Interpreters/InterpreterShowColumnsQuery.cpp b/src/Interpreters/InterpreterShowColumnsQuery.cpp index 6c55ef0e6e9..54039a15807 100644 --- a/src/Interpreters/InterpreterShowColumnsQuery.cpp +++ b/src/Interpreters/InterpreterShowColumnsQuery.cpp @@ -37,7 +37,7 @@ String InterpreterShowColumnsQuery::getRewrittenQuery() SELECT name AS field, type AS type, - if(startsWith(type, 'Nullable'), 'YES', 'NO') AS `null`, + if (startsWith(type, 'Nullable'), 'YES', 'NO') AS `null`, trim(concatWithSeparator(' ', if (is_in_primary_key, 'PRI', ''), if (is_in_sorting_key, 'SOR', ''))) AS key, if (default_kind IN ('ALIAS', 'DEFAULT', 'MATERIALIZED'), default_expression, NULL) AS default, '' AS extra )"; diff --git a/src/Processors/Formats/Impl/MySQLOutputFormat.cpp b/src/Processors/Formats/Impl/MySQLOutputFormat.cpp index 2f056daf8e4..87c8c742f8d 100644 --- a/src/Processors/Formats/Impl/MySQLOutputFormat.cpp +++ b/src/Processors/Formats/Impl/MySQLOutputFormat.cpp @@ -25,7 +25,7 @@ MySQLOutputFormat::MySQLOutputFormat(WriteBuffer & out_, const Block & header_, /// There is no `sequence_id` stored in `settings_.mysql_wire` in this case, so we create a dummy one. sequence_id = settings_.mysql_wire.sequence_id ? settings_.mysql_wire.sequence_id : &dummy_sequence_id; /// Switch between Text (COM_QUERY) and Binary (COM_EXECUTE_STMT) ResultSet - use_binary_result_set = settings_.mysql_wire.use_binary_result_set; + use_binary_result_set = settings_.mysql_wire.binary_protocol; const auto & header = getPort(PortKind::Main).getHeader(); data_types = header.getDataTypes(); diff --git a/src/Server/MySQLHandler.cpp b/src/Server/MySQLHandler.cpp index 219e9c15ac2..6297af18817 100644 --- a/src/Server/MySQLHandler.cpp +++ b/src/Server/MySQLHandler.cpp @@ -318,7 +318,7 @@ void MySQLHandler::comPing() static bool isFederatedServerSetupSetCommand(const String & query); -void MySQLHandler::comQuery(ReadBuffer & payload, bool use_binary_protocol_result_set) +void MySQLHandler::comQuery(ReadBuffer & payload, bool binary_protocol) { String query = String(payload.position(), payload.buffer().end()); @@ -364,7 +364,7 @@ void MySQLHandler::comQuery(ReadBuffer & payload, bool use_binary_protocol_resul format_settings.mysql_wire.client_capabilities = client_capabilities; format_settings.mysql_wire.max_packet_size = max_packet_size; format_settings.mysql_wire.sequence_id = &sequence_id; - format_settings.mysql_wire.use_binary_result_set = use_binary_protocol_result_set; + format_settings.mysql_wire.binary_protocol = binary_protocol; auto set_result_details = [&with_output](const QueryResultDetails & details) { @@ -386,44 +386,13 @@ void MySQLHandler::comQuery(ReadBuffer & payload, bool use_binary_protocol_resul void MySQLHandler::comStmtPrepare(DB::ReadBuffer & payload) { - if (prepared_statements_map.size() > 10000) /// Shouldn't happen in reality as COM_STMT_CLOSE cleans up the elements - { - LOG_ERROR(log, "Too many prepared statements"); - current_prepared_statement_id = 0; - prepared_statements_map.clear(); - packet_endpoint->sendPacket(ERRPacket(), true); - return; - } - - String query; - readStringUntilEOF(query, payload); - - uint32_t statement_id = current_prepared_statement_id; - if (current_prepared_statement_id == std::numeric_limits::max()) - { - current_prepared_statement_id = 0; - } + String statement; + readStringUntilEOF(statement, payload); + auto [is_success, statement_id] = emplacePreparedStatement(std::move(statement)); + if (is_success) + packet_endpoint->sendPacket(PreparedStatementResponseOK(statement_id, 0, 0, 0), true); else - { - current_prepared_statement_id++; - } - - // Key collisions should not happen here, as we remove the elements from the map with COM_STMT_CLOSE, - // and we have quite a big range of available identifiers with 32-bit unsigned integer - if (prepared_statements_map.contains(statement_id)) - { - LOG_ERROR( - log, - "Failed to store a new statement `{}` with id {}; it is already taken by `{}`", - query, - statement_id, - prepared_statements_map.at(statement_id)); packet_endpoint->sendPacket(ERRPacket(), true); - return; - } - - prepared_statements_map.emplace(statement_id, query); - packet_endpoint->sendPacket(PrepareStatementResponseOK(statement_id, 0, 0, 0), true); } void MySQLHandler::comStmtExecute(ReadBuffer & payload) @@ -431,7 +400,7 @@ void MySQLHandler::comStmtExecute(ReadBuffer & payload) uint32_t statement_id; payload.readStrict(reinterpret_cast(&statement_id), 4); - if (!prepared_statements_map.contains(statement_id)) + if (!prepared_statements.contains(statement_id)) { LOG_ERROR(log, "Could not find prepared statement with id {}", statement_id); packet_endpoint->sendPacket(ERRPacket(), true); @@ -439,7 +408,7 @@ void MySQLHandler::comStmtExecute(ReadBuffer & payload) } // Temporary workaround as we work only with queries that do not bind any parameters atm - ReadBufferFromString com_query_payload(prepared_statements_map.at(statement_id)); + ReadBufferFromString com_query_payload(prepared_statements.at(statement_id)); MySQLHandler::comQuery(com_query_payload, true); }; @@ -447,16 +416,49 @@ void MySQLHandler::comStmtClose(ReadBuffer & payload) { uint32_t statement_id; payload.readStrict(reinterpret_cast(&statement_id), 4); - - if (prepared_statements_map.contains(statement_id)) - { - prepared_statements_map.erase(statement_id); - } + erasePreparedStatement(statement_id); // https://dev.mysql.com/doc/dev/mysql-server/latest/page_protocol_com_stmt_close.html // No response packet is sent back to the client. }; +MySQLHandler::EmplacePreparedStatementResult MySQLHandler::emplacePreparedStatement(String statement) +{ + static constexpr size_t MAX_PREPARED_STATEMENTS = 10'000; + std::lock_guard lock(prepared_statements_mutex); + if (prepared_statements.size() > MAX_PREPARED_STATEMENTS) /// Shouldn't happen in reality as COM_STMT_CLOSE cleans up the elements + { + LOG_ERROR(log, "Too many prepared statements"); + current_prepared_statement_id = 0; + prepared_statements.clear(); + return std::make_pair(false, 0); + } + uint32_t statement_id = current_prepared_statement_id++; + + // Key collisions should not happen here, as we remove the elements from the map with COM_STMT_CLOSE, + // and we have quite a big range of available identifiers with 32-bit unsigned integer + if (prepared_statements.contains(statement_id)) + { + LOG_ERROR( + log, + "Failed to store a new statement `{}` with id {}; it is already taken by `{}`", + statement, + statement_id, + prepared_statements.at(statement_id)); + return std::make_pair(false, 0); + } + + prepared_statements.emplace(statement_id, statement); + return std::make_pair(true, statement_id); +}; + +void MySQLHandler::erasePreparedStatement(UInt32 statement_id) +{ + std::lock_guard lock(prepared_statements_mutex); + if (prepared_statements.contains(statement_id)) + prepared_statements.erase(statement_id); +} + void MySQLHandler::authPluginSSL() { throw Exception(ErrorCodes::SUPPORT_IS_DISABLED, diff --git a/src/Server/MySQLHandler.h b/src/Server/MySQLHandler.h index 240057e8ba7..94f86f5e5c6 100644 --- a/src/Server/MySQLHandler.h +++ b/src/Server/MySQLHandler.h @@ -30,6 +30,9 @@ class TCPServer; /// Handler for MySQL wire protocol connections. Allows to connect to ClickHouse using MySQL client. class MySQLHandler : public Poco::Net::TCPServerConnection { + using PreparedStatements = std::unordered_map; /// statement_id -> statement + using EmplacePreparedStatementResult = std::pair; /// is_success? -> statement_id + public: MySQLHandler( IServer & server_, @@ -46,7 +49,7 @@ protected: /// Enables SSL, if client requested. void finishHandshake(MySQLProtocol::ConnectionPhase::HandshakeResponse &); - void comQuery(ReadBuffer & payload, bool use_binary_protocol_result_set); + void comQuery(ReadBuffer & payload, bool binary_protocol); void comFieldList(ReadBuffer & payload); @@ -62,6 +65,9 @@ protected: void comStmtClose(ReadBuffer & payload); + EmplacePreparedStatementResult emplacePreparedStatement(String statement); + void erasePreparedStatement(UInt32 statement_id); + virtual void authPluginSSL(); virtual void finishHandshakeSSL(size_t packet_size, char * buf, size_t pos, std::function read_bytes, MySQLProtocol::ConnectionPhase::HandshakeResponse & packet); @@ -82,9 +88,9 @@ protected: using Replacements = std::unordered_map; Replacements replacements; - uint32_t current_prepared_statement_id = 0; - using PreparedStatementsMap = std::unordered_map; - PreparedStatementsMap prepared_statements_map; + std::mutex prepared_statements_mutex; + UInt32 current_prepared_statement_id = 0; + PreparedStatements prepared_statements; std::unique_ptr auth_plugin; std::shared_ptr in; diff --git a/tests/integration/test_mysql_protocol/prepared_statements.reference b/tests/integration/test_mysql_protocol/prepared_statements.reference index 6ea4b218f56..1cea2dc1b1e 100644 --- a/tests/integration/test_mysql_protocol/prepared_statements.reference +++ b/tests/integration/test_mysql_protocol/prepared_statements.reference @@ -60,11 +60,6 @@ d256 type is CHAR, value: 123456789123456789123456789112345678912345678912345678 ### testMiscTypes Row #1 -a type is CHAR, value: [] -u type is CHAR, value: 9a0ccc06-2578-4861-8534-631c9d40f3f7 -t type is CHAR, value: (0,'') -m type is CHAR, value: {} -Row #2 a type is CHAR, value: ['foo','bar'] u type is CHAR, value: 5da5038d-788f-48c6-b510-babb41c538d3 t type is CHAR, value: (42,'qaz') diff --git a/tests/integration/test_mysql_protocol/prepared_statements_test.sql b/tests/integration/test_mysql_protocol/prepared_statements_test.sql index 0302fe5aca7..0cca2b354ba 100644 --- a/tests/integration/test_mysql_protocol/prepared_statements_test.sql +++ b/tests/integration/test_mysql_protocol/prepared_statements_test.sql @@ -1,4 +1,4 @@ -CREATE TABLE ps_simple_data_types +CREATE OR REPLACE TABLE ps_simple_data_types ( i8 Int8, i16 Int16, @@ -28,11 +28,11 @@ VALUES (127, 32767, 2147483647, 9223372036854775807, 170141183460469231731687303 120, 1234, 51234, 421342, 15324355, 41345135123432, -0.7968956, -0.113259, TRUE); -CREATE TABLE ps_string_types +CREATE OR REPLACE TABLE ps_string_types ( - s String, - sn Nullable(String), - lc LowCardinality(String), + s String, + sn Nullable(String), + lc LowCardinality(String), nlc LowCardinality(Nullable(String)) ) ENGINE MergeTree ORDER BY s; @@ -40,7 +40,7 @@ INSERT INTO ps_string_types VALUES ('foo', 'bar', 'qaz', 'qux'), ('42', NULL, 'test', NULL); -CREATE TABLE ps_decimal_types +CREATE OR REPLACE TABLE ps_decimal_types ( d32 Decimal(9, 2), d64 Decimal(18, 3), @@ -57,7 +57,7 @@ VALUES (1234567.89, 12345678912345678912345678911234567891234567891234567891.12345678911234567891), (-1.55, 6.03, 5, -1224124.23423, -54342.3); -CREATE TABLE ps_misc_types +CREATE OR REPLACE TABLE ps_misc_types ( a Array(String), u UUID, @@ -66,10 +66,9 @@ CREATE TABLE ps_misc_types ) ENGINE MergeTree ORDER BY u; INSERT INTO ps_misc_types -VALUES (['foo', 'bar'], '5da5038d-788f-48c6-b510-babb41c538d3', (42, 'qaz'), {'qux': 144, 'text': 255}), - ([], '9a0ccc06-2578-4861-8534-631c9d40f3f7', (0, ''), {}); +VALUES (['foo', 'bar'], '5da5038d-788f-48c6-b510-babb41c538d3', (42, 'qaz'), {'qux': 144, 'text': 255}); -CREATE TABLE ps_date_types +CREATE OR REPLACE TABLE ps_date_types ( d Date, d32 Date32, @@ -89,7 +88,7 @@ VALUES ('2149-06-06', '2178-04-16', '2106-02-07 06:28:15', '1900-01-01 00:00:00.000001', '1900-01-01 00:00:00.000000001');; -CREATE TABLE ps_unusual_datetime64_scales +CREATE OR REPLACE TABLE ps_unusual_datetime64_scales ( dt64_0 DateTime64(0, 'UTC'), dt64_1 DateTime64(1, 'UTC'), @@ -116,7 +115,7 @@ VALUES ('2022-04-13 03:17:45', '2022-04-13 03:17:45.0000001', '2022-04-13 03:17:45.00000001'); -CREATE TABLE ps_datetime_timezones +CREATE OR REPLACE TABLE ps_datetime_timezones ( dt DateTime('Europe/Amsterdam'), dt64_3 DateTime64(3, 'Asia/Shanghai') @@ -124,4 +123,4 @@ CREATE TABLE ps_datetime_timezones INSERT INTO ps_datetime_timezones VALUES ('2022-09-04 20:31:05', '2022-09-04 20:31:05.022'), - ('1970-01-01 00:00:00', '1969-12-31 16:00:00'); + ('1970-01-01 01:00:00', '1969-12-31 16:00:00'); From 0a1495fd0028ad9ad859af9d85f12a96b58e527c Mon Sep 17 00:00:00 2001 From: slvrtrn Date: Tue, 12 Sep 2023 18:46:10 +0200 Subject: [PATCH 314/433] Revert information_schema changes (new PR) --- .../InformationSchema/key_column_usage.sql | 38 ------------- .../referential_constraints.sql | 25 --------- .../System/InformationSchema/schemata.sql | 55 ++++++++----------- .../System/InformationSchema/tables.sql | 50 ++++++----------- .../System/attachInformationSchemaTables.cpp | 5 +- 5 files changed, 41 insertions(+), 132 deletions(-) delete mode 100644 src/Storages/System/InformationSchema/key_column_usage.sql delete mode 100644 src/Storages/System/InformationSchema/referential_constraints.sql diff --git a/src/Storages/System/InformationSchema/key_column_usage.sql b/src/Storages/System/InformationSchema/key_column_usage.sql deleted file mode 100644 index 43630b8c8b9..00000000000 --- a/src/Storages/System/InformationSchema/key_column_usage.sql +++ /dev/null @@ -1,38 +0,0 @@ -ATTACH VIEW key_column_usage - ( - `referenced_table_schema` Nullable(String), - `referenced_table_name` Nullable(String), - `referenced_column_name` Nullable(String), - `table_schema` String, - `table_name` String, - `column_name` Nullable(String), - `ordinal_position` UInt32, - `constraint_name` Nullable(String), - `REFERENCED_TABLE_SCHEMA` Nullable(String), - `REFERENCED_TABLE_NAME` Nullable(String), - `REFERENCED_COLUMN_NAME` Nullable(String), - `TABLE_SCHEMA` String, - `TABLE_NAME` String, - `COLUMN_NAME` Nullable(String), - `ORDINAL_POSITION` UInt32, - `CONSTRAINT_NAME` Nullable(String) - ) AS -SELECT NULL AS `referenced_table_schema`, - NULL AS `referenced_table_name`, - NULL AS `referenced_column_name`, - database AS `table_schema`, - table AS `table_name`, - name AS `column_name`, - position AS `ordinal_position`, - 'PRIMARY' AS `constraint_name`, - - `referenced_table_schema` AS `REFERENCED_TABLE_SCHEMA`, - `referenced_table_name` AS `REFERENCED_TABLE_NAME`, - `referenced_column_name` AS `REFERENCED_COLUMN_NAME`, - `table_schema` AS `TABLE_SCHEMA`, - `table_name` AS `TABLE_NAME`, - `column_name` AS `COLUMN_NAME`, - `ordinal_position` AS `ORDINAL_POSITION`, - `constraint_name` AS `CONSTRAINT_NAME` -FROM system.columns -WHERE is_in_primary_key; \ No newline at end of file diff --git a/src/Storages/System/InformationSchema/referential_constraints.sql b/src/Storages/System/InformationSchema/referential_constraints.sql deleted file mode 100644 index 8216b8fff83..00000000000 --- a/src/Storages/System/InformationSchema/referential_constraints.sql +++ /dev/null @@ -1,25 +0,0 @@ -ATTACH VIEW referential_constraints - ( - `constraint_name` Nullable(String), - `constraint_schema` String, - `table_name` String, - `update_rule` String, - `delete_rule` String, - `CONSTRAINT_NAME` Nullable(String), - `CONSTRAINT_SCHEMA` String, - `TABLE_NAME` String, - `UPDATE_RULE` String, - `DELETE_RULE` String - ) AS -SELECT NULL AS `constraint_name`, - '' AS `constraint_schema`, - '' AS `table_name`, - '' AS `update_rule`, - '' AS `delete_rule`, - - NULL AS `CONSTRAINT_NAME`, - '' AS `CONSTRAINT_SCHEMA`, - '' AS `TABLE_NAME`, - '' AS `UPDATE_RULE`, - '' AS `DELETE_RULE` -WHERE false; \ No newline at end of file diff --git a/src/Storages/System/InformationSchema/schemata.sql b/src/Storages/System/InformationSchema/schemata.sql index ca4ad4f7310..9686fcbf4fa 100644 --- a/src/Storages/System/InformationSchema/schemata.sql +++ b/src/Storages/System/InformationSchema/schemata.sql @@ -1,33 +1,26 @@ ATTACH VIEW schemata - ( - `catalog_name` String, - `schema_name` String, - `schema_owner` String, - `default_character_set_catalog` Nullable(String), - `default_character_set_schema` Nullable(String), - `default_character_set_name` Nullable(String), - `sql_path` Nullable(String), - `CATALOG_NAME` String, - `SCHEMA_NAME` String, - `SCHEMA_OWNER` String, - `DEFAULT_CHARACTER_SET_CATALOG` Nullable(String), - `DEFAULT_CHARACTER_SET_SCHEMA` Nullable(String), - `DEFAULT_CHARACTER_SET_NAME` Nullable(String), - `SQL_PATH` Nullable(String) - ) AS -SELECT name AS `catalog_name`, - name AS `schema_name`, - 'default' AS `schema_owner`, - NULL AS `default_character_set_catalog`, - NULL AS `default_character_set_schema`, - NULL AS `default_character_set_name`, - NULL AS `sql_path`, - - catalog_name AS `CATALOG_NAME`, - schema_name AS `SCHEMA_NAME`, - schema_owner AS `SCHEMA_OWNER`, - NULL AS `DEFAULT_CHARACTER_SET_CATALOG`, - NULL AS `DEFAULT_CHARACTER_SET_SCHEMA`, - NULL AS `DEFAULT_CHARACTER_SET_NAME`, - NULL AS `SQL_PATH` +( + `catalog_name` String, + `schema_name` String, + `schema_owner` String, + `default_character_set_catalog` Nullable(String), + `default_character_set_schema` Nullable(String), + `default_character_set_name` Nullable(String), + `sql_path` Nullable(String), + `CATALOG_NAME` String ALIAS catalog_name, + `SCHEMA_NAME` String ALIAS schema_name, + `SCHEMA_OWNER` String ALIAS schema_owner, + `DEFAULT_CHARACTER_SET_CATALOG` Nullable(String) ALIAS default_character_set_catalog, + `DEFAULT_CHARACTER_SET_SCHEMA` Nullable(String) ALIAS default_character_set_schema, + `DEFAULT_CHARACTER_SET_NAME` Nullable(String) ALIAS default_character_set_name, + `SQL_PATH` Nullable(String) ALIAS sql_path +) AS +SELECT + name AS catalog_name, + name AS schema_name, + 'default' AS schema_owner, + NULL AS default_character_set_catalog, + NULL AS default_character_set_schema, + NULL AS default_character_set_name, + NULL AS sql_path FROM system.databases diff --git a/src/Storages/System/InformationSchema/tables.sql b/src/Storages/System/InformationSchema/tables.sql index b3bbfa72517..8eea3713923 100644 --- a/src/Storages/System/InformationSchema/tables.sql +++ b/src/Storages/System/InformationSchema/tables.sql @@ -1,35 +1,17 @@ ATTACH VIEW tables - ( - `table_catalog` String, - `table_schema` String, - `table_name` String, - `table_type` String, - `table_comment` String, - `table_collation` String, - `TABLE_CATALOG` String, - `TABLE_SCHEMA` String, - `TABLE_NAME` String, - `TABLE_TYPE` String, - `TABLE_COMMENT` String, - `TABLE_COLLATION` String - ) AS -SELECT database AS `table_catalog`, - database AS `table_schema`, - name AS `table_name`, - comment AS `table_comment`, - multiIf( - is_temporary, 'LOCAL TEMPORARY', - engine LIKE '%View', 'VIEW', - engine LIKE 'System%', 'SYSTEM VIEW', - has_own_data = 0, 'FOREIGN TABLE', - 'BASE TABLE' - ) AS `table_type`, - 'utf8mb4_0900_ai_ci' AS `table_collation`, - - table_catalog AS `TABLE_CATALOG`, - table_schema AS `TABLE_SCHEMA`, - table_name AS `TABLE_NAME`, - table_comment AS `TABLE_COMMENT`, - table_type AS `TABLE_TYPE`, - table_collation AS `TABLE_COLLATION` -FROM system.tables \ No newline at end of file +( + `table_catalog` String, + `table_schema` String, + `table_name` String, + `table_type` Enum8('BASE TABLE' = 1, 'VIEW' = 2, 'FOREIGN TABLE' = 3, 'LOCAL TEMPORARY' = 4, 'SYSTEM VIEW' = 5), + `TABLE_CATALOG` String ALIAS table_catalog, + `TABLE_SCHEMA` String ALIAS table_schema, + `TABLE_NAME` String ALIAS table_name, + `TABLE_TYPE` Enum8('BASE TABLE' = 1, 'VIEW' = 2, 'FOREIGN TABLE' = 3, 'LOCAL TEMPORARY' = 4, 'SYSTEM VIEW' = 5) ALIAS table_type +) AS +SELECT + database AS table_catalog, + database AS table_schema, + name AS table_name, + multiIf(is_temporary, 4, engine like '%View', 2, engine LIKE 'System%', 5, has_own_data = 0, 3, 1) AS table_type +FROM system.tables diff --git a/src/Storages/System/attachInformationSchemaTables.cpp b/src/Storages/System/attachInformationSchemaTables.cpp index d4775bf0d4a..074a648d235 100644 --- a/src/Storages/System/attachInformationSchemaTables.cpp +++ b/src/Storages/System/attachInformationSchemaTables.cpp @@ -12,8 +12,7 @@ INCBIN(resource_schemata_sql, SOURCE_DIR "/src/Storages/System/InformationSchema INCBIN(resource_tables_sql, SOURCE_DIR "/src/Storages/System/InformationSchema/tables.sql"); INCBIN(resource_views_sql, SOURCE_DIR "/src/Storages/System/InformationSchema/views.sql"); INCBIN(resource_columns_sql, SOURCE_DIR "/src/Storages/System/InformationSchema/columns.sql"); -INCBIN(resource_key_column_usage_sql, SOURCE_DIR "/src/Storages/System/InformationSchema/key_column_usage.sql"); -INCBIN(resource_referential_constraints_sql, SOURCE_DIR "/src/Storages/System/InformationSchema/referential_constraints.sql"); + namespace DB { @@ -67,8 +66,6 @@ void attachInformationSchema(ContextMutablePtr context, IDatabase & information_ createInformationSchemaView(context, information_schema_database, "tables", std::string_view(reinterpret_cast(gresource_tables_sqlData), gresource_tables_sqlSize)); createInformationSchemaView(context, information_schema_database, "views", std::string_view(reinterpret_cast(gresource_views_sqlData), gresource_views_sqlSize)); createInformationSchemaView(context, information_schema_database, "columns", std::string_view(reinterpret_cast(gresource_columns_sqlData), gresource_columns_sqlSize)); - createInformationSchemaView(context, information_schema_database, "key_column_usage", std::string_view(reinterpret_cast(gresource_key_column_usage_sqlData), gresource_key_column_usage_sqlSize)); - createInformationSchemaView(context, information_schema_database, "referential_constraints", std::string_view(reinterpret_cast(gresource_referential_constraints_sqlData), gresource_referential_constraints_sqlSize)); } } From cf662712ea9eb9c61c7497175c7609e3c5c9ea58 Mon Sep 17 00:00:00 2001 From: Nikolay Degterinsky Date: Tue, 12 Sep 2023 16:45:02 +0000 Subject: [PATCH 315/433] Add new subDate function, apply review suggestions --- .../functions/date-time-functions.md | 45 ++++++++++++++++- .../{addDate.cpp => FunctionsOpDate.cpp} | 48 ++++++++++++------- ...new_functions_must_be_documented.reference | 1 + .../02834_add_date_function.reference | 4 -- .../0_stateless/02834_add_date_function.sql | 9 ---- .../02834_add_sub_date_functions.reference | 17 +++++++ .../02834_add_sub_date_functions.sql | 31 ++++++++++++ 7 files changed, 124 insertions(+), 31 deletions(-) rename src/Functions/{addDate.cpp => FunctionsOpDate.cpp} (65%) delete mode 100644 tests/queries/0_stateless/02834_add_date_function.reference delete mode 100644 tests/queries/0_stateless/02834_add_date_function.sql create mode 100644 tests/queries/0_stateless/02834_add_sub_date_functions.reference create mode 100644 tests/queries/0_stateless/02834_add_sub_date_functions.sql diff --git a/docs/en/sql-reference/functions/date-time-functions.md b/docs/en/sql-reference/functions/date-time-functions.md index 3940aef8fbf..dafedfb3adc 100644 --- a/docs/en/sql-reference/functions/date-time-functions.md +++ b/docs/en/sql-reference/functions/date-time-functions.md @@ -1150,12 +1150,12 @@ addDate(date, interval) **Arguments** -- `date` — The date or date with time to which `value` is added. [Date](../../sql-reference/data-types/date.md) or [DateTime](../../sql-reference/data-types/datetime.md). +- `date` — The date or date with time to which `interval` is added. [Date](../../sql-reference/data-types/date.md) or [DateTime](../../sql-reference/data-types/datetime.md). - `interval` — Interval to add. [Interval](../../sql-reference/data-types/special-data-types/interval.md). **Returned value** -Date or date with time obtained by adding `value`, expressed in `unit`, to `date`. +Date or date with time obtained by adding `interval` to `date`. Type: [Date](../../sql-reference/data-types/date.md) or [DateTime](../../sql-reference/data-types/datetime.md). @@ -1173,6 +1173,47 @@ Result: └──────────────────────────────────────────────────┘ ``` +**See Also** +- [date_add](#date_add) + +## subDate + +Subtracts the time interval or date interval from the provided date or date with time. + +**Syntax** + +``` sql +subDate(date, interval) +``` + +**Arguments** + +- `date` — The date or date with time from which `interval` is subtracted. [Date](../../sql-reference/data-types/date.md) or [DateTime](../../sql-reference/data-types/datetime.md). +- `interval` — Interval to subtract. [Interval](../../sql-reference/data-types/special-data-types/interval.md). + +**Returned value** + +Date or date with time obtained by subtracting `interval` from `date`. + +Type: [Date](../../sql-reference/data-types/date.md) or [DateTime](../../sql-reference/data-types/datetime.md). + +**Example** + +```sql +SELECT subDate(toDate('2018-01-01'), INTERVAL 3 YEAR); +``` + +Result: + +```text +┌─subDate(toDate('2018-01-01'), toIntervalYear(3))─┐ +│ 2015-01-01 │ +└──────────────────────────────────────────────────┘ +``` + +**See Also** +- [date_sub](#date_sub) + ## now Returns the current date and time at the moment of query analysis. The function is a constant expression. diff --git a/src/Functions/addDate.cpp b/src/Functions/FunctionsOpDate.cpp similarity index 65% rename from src/Functions/addDate.cpp rename to src/Functions/FunctionsOpDate.cpp index a34ded341ac..b692fbf97aa 100644 --- a/src/Functions/addDate.cpp +++ b/src/Functions/FunctionsOpDate.cpp @@ -10,15 +10,16 @@ namespace ErrorCodes namespace { - -class FunctionAddDate : public IFunction +template +class FunctionOpDate : public IFunction { public: - static constexpr auto name = "addDate"; + static constexpr auto name = Op::name; - explicit FunctionAddDate(ContextPtr context_) : context(context_) {} + explicit FunctionOpDate(ContextPtr context_) : context(context_) {} - static FunctionPtr create(ContextPtr context) { return std::make_shared(context); } + + static FunctionPtr create(ContextPtr context) { return std::make_shared>(context); } String getName() const override { return name; } @@ -37,14 +38,14 @@ public: if (!isInterval(arguments[1].type)) throw Exception( ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, - "Illegal type {} of 1st argument of function {}. Should be an interval", - arguments[0].type->getName(), + "Illegal type {} of 2nd argument of function {}. Should be an interval", + arguments[1].type->getName(), getName()); - auto plus = FunctionFactory::instance().get("plus", context); - auto plus_build = plus->build(arguments); + auto op = FunctionFactory::instance().get(Op::internal_name, context); + auto op_build = op->build(arguments); - return plus_build->getResultType(); + return op_build->getResultType(); } bool useDefaultImplementationForConstants() const override { return true; } @@ -62,15 +63,15 @@ public: if (!isInterval(arguments[1].type)) throw Exception( ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, - "Illegal type {} of 1st argument of function {}. Should be an interval", - arguments[0].type->getName(), + "Illegal type {} of 2nd argument of function {}. Should be an interval", + arguments[1].type->getName(), getName()); - auto plus = FunctionFactory::instance().get("plus", context); - auto plus_build = plus->build(arguments); + auto op = FunctionFactory::instance().get(Op::internal_name, context); + auto op_build = op->build(arguments); - auto res_type = plus_build->getResultType(); - return plus_build->execute(arguments, res_type, input_rows_count); + auto res_type = op_build->getResultType(); + return op_build->execute(arguments, res_type, input_rows_count); } private: @@ -79,10 +80,25 @@ private: } +struct AddDate +{ + static constexpr auto name = "addDate"; + static constexpr auto internal_name = "plus"; +}; + +struct SubDate +{ + static constexpr auto name = "subDate"; + static constexpr auto internal_name = "minus"; +}; + +using FunctionAddDate = FunctionOpDate; +using FunctionSubDate = FunctionOpDate; REGISTER_FUNCTION(AddInterval) { factory.registerFunction(); + factory.registerFunction(); } } diff --git a/tests/queries/0_stateless/02415_all_new_functions_must_be_documented.reference b/tests/queries/0_stateless/02415_all_new_functions_must_be_documented.reference index 1a7ffdb1fa1..82f65df4808 100644 --- a/tests/queries/0_stateless/02415_all_new_functions_must_be_documented.reference +++ b/tests/queries/0_stateless/02415_all_new_functions_must_be_documented.reference @@ -667,6 +667,7 @@ splitByWhitespace sqrt startsWith subBitmap +subDate substring substringIndex substringIndexUTF8 diff --git a/tests/queries/0_stateless/02834_add_date_function.reference b/tests/queries/0_stateless/02834_add_date_function.reference deleted file mode 100644 index 8912c8c8f63..00000000000 --- a/tests/queries/0_stateless/02834_add_date_function.reference +++ /dev/null @@ -1,4 +0,0 @@ -2022-05-07 00:05:00 -2022-05-07 00:05:00.000 -2022-05-07 00:05:00 -2022-05-07 00:05:00.000 diff --git a/tests/queries/0_stateless/02834_add_date_function.sql b/tests/queries/0_stateless/02834_add_date_function.sql deleted file mode 100644 index 1bf4cb0976e..00000000000 --- a/tests/queries/0_stateless/02834_add_date_function.sql +++ /dev/null @@ -1,9 +0,0 @@ -SELECT addDate('2022-05-07'::Date, INTERVAL 5 MINUTE); -SELECT addDate('2022-05-07'::Date32, INTERVAL 5 MINUTE); -SELECT addDate('2022-05-07'::DateTime, INTERVAL 5 MINUTE); -SELECT addDate('2022-05-07'::DateTime64, INTERVAL 5 MINUTE); - -SELECT addDate('2022-05-07'::Date); -- { serverError NUMBER_OF_ARGUMENTS_DOESNT_MATCH } -SELECT addDate('2022-05-07'::Date, INTERVAL 5 MINUTE, 5); -- { serverError NUMBER_OF_ARGUMENTS_DOESNT_MATCH } -SELECT addDate('2022-05-07'::Date, 10); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } -SELECT addDate('1234', INTERVAL 5 MINUTE); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } diff --git a/tests/queries/0_stateless/02834_add_sub_date_functions.reference b/tests/queries/0_stateless/02834_add_sub_date_functions.reference new file mode 100644 index 00000000000..26e4b281526 --- /dev/null +++ b/tests/queries/0_stateless/02834_add_sub_date_functions.reference @@ -0,0 +1,17 @@ +2022-05-07 00:05:00 +2022-05-07 00:05:00.000 +2022-05-07 00:05:00 +2022-05-07 00:05:00.000 +2120-07-26 +2299-12-31 +2027-10-17 11:03:28 +2299-12-31 00:00:00.000 +--- +2022-05-06 23:55:00 +2022-05-06 23:55:00.000 +2022-05-06 23:55:00 +2022-05-06 23:55:00.000 +1970-01-01 +1900-01-01 +1969-12-31 16:00:00 +1900-01-01 00:00:00.000 diff --git a/tests/queries/0_stateless/02834_add_sub_date_functions.sql b/tests/queries/0_stateless/02834_add_sub_date_functions.sql new file mode 100644 index 00000000000..a819c40af85 --- /dev/null +++ b/tests/queries/0_stateless/02834_add_sub_date_functions.sql @@ -0,0 +1,31 @@ +SELECT addDate('2022-05-07'::Date, INTERVAL 5 MINUTE); +SELECT addDate('2022-05-07'::Date32, INTERVAL 5 MINUTE); +SELECT addDate('2022-05-07'::DateTime, INTERVAL 5 MINUTE); +SELECT addDate('2022-05-07'::DateTime64, INTERVAL 5 MINUTE); + +SELECT addDate('2022-05-07'::Date, INTERVAL 400000 YEAR); +SELECT addDate('2022-05-07'::Date32, INTERVAL 400000 YEAR); +SELECT addDate('2022-05-07'::DateTime, INTERVAL 400000 YEAR); +SELECT addDate('2022-05-07'::DateTime64, INTERVAL 400000 YEAR); + +SELECT addDate('2022-05-07'::Date); -- { serverError NUMBER_OF_ARGUMENTS_DOESNT_MATCH } +SELECT addDate('2022-05-07'::Date, INTERVAL 5 MINUTE, 5); -- { serverError NUMBER_OF_ARGUMENTS_DOESNT_MATCH } +SELECT addDate('2022-05-07'::Date, 10); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT addDate('1234', INTERVAL 5 MINUTE); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } + +SELECT '---'; + +SELECT subDate('2022-05-07'::Date, INTERVAL 5 MINUTE); +SELECT subDate('2022-05-07'::Date32, INTERVAL 5 MINUTE); +SELECT subDate('2022-05-07'::DateTime, INTERVAL 5 MINUTE); +SELECT subDate('2022-05-07'::DateTime64, INTERVAL 5 MINUTE); + +SELECT subDate('2022-05-07'::Date, INTERVAL 400000 YEAR); +SELECT subDate('2022-05-07'::Date32, INTERVAL 400000 YEAR); +SELECT subDate('2022-05-07'::DateTime, INTERVAL 400000 YEAR); +SELECT subDate('2022-05-07'::DateTime64, INTERVAL 400000 YEAR); + +SELECT subDate('2022-05-07'::Date); -- { serverError NUMBER_OF_ARGUMENTS_DOESNT_MATCH } +SELECT subDate('2022-05-07'::Date, INTERVAL 5 MINUTE, 5); -- { serverError NUMBER_OF_ARGUMENTS_DOESNT_MATCH } +SELECT subDate('2022-05-07'::Date, 10); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } +SELECT subDate('1234', INTERVAL 5 MINUTE); -- { serverError ILLEGAL_TYPE_OF_ARGUMENT } From f3729da64ba4c1ece9a14318c264a9958ae6805a Mon Sep 17 00:00:00 2001 From: alesapin Date: Tue, 12 Sep 2023 19:55:56 +0200 Subject: [PATCH 316/433] Fix rare case of CHECKSUM_DOESNT_MATCH error --- src/Storages/MergeTree/MergeTreeData.cpp | 5 ++- src/Storages/MergeTree/MergeTreeData.h | 2 +- .../MergeTree/ReplicatedMergeTreeSink.cpp | 43 ++++++++++++++++++- src/Storages/StorageReplicatedMergeTree.cpp | 10 +++-- src/Storages/StorageReplicatedMergeTree.h | 8 +++- ...ted_fetch_checksums_doesnt_match.reference | 6 +++ ...eplicated_fetch_checksums_doesnt_match.sql | 42 ++++++++++++++++++ 7 files changed, 106 insertions(+), 10 deletions(-) create mode 100644 tests/queries/0_stateless/02882_replicated_fetch_checksums_doesnt_match.reference create mode 100644 tests/queries/0_stateless/02882_replicated_fetch_checksums_doesnt_match.sql diff --git a/src/Storages/MergeTree/MergeTreeData.cpp b/src/Storages/MergeTree/MergeTreeData.cpp index 3337e136c16..4b6d2ea41ed 100644 --- a/src/Storages/MergeTree/MergeTreeData.cpp +++ b/src/Storages/MergeTree/MergeTreeData.cpp @@ -4245,7 +4245,7 @@ void MergeTreeData::forcefullyMovePartToDetachedAndRemoveFromMemory(const MergeT } -void MergeTreeData::tryRemovePartImmediately(DataPartPtr && part) +bool MergeTreeData::tryRemovePartImmediately(DataPartPtr && part) { DataPartPtr part_to_delete; { @@ -4271,7 +4271,7 @@ void MergeTreeData::tryRemovePartImmediately(DataPartPtr && part) if (!it->unique()) LOG_WARNING(log, "Cannot immediately remove part {} because someone using it right now " "usage counter {}", part_name_with_state, it->use_count()); - return; + return false; } modifyPartState(it, DataPartState::Deleting); @@ -4296,6 +4296,7 @@ void MergeTreeData::tryRemovePartImmediately(DataPartPtr && part) removePartsFinally({part_to_delete}); LOG_TRACE(log, "Removed part {}", part_to_delete->name); + return true; } diff --git a/src/Storages/MergeTree/MergeTreeData.h b/src/Storages/MergeTree/MergeTreeData.h index 95d8e74f32c..6f9779bde00 100644 --- a/src/Storages/MergeTree/MergeTreeData.h +++ b/src/Storages/MergeTree/MergeTreeData.h @@ -671,7 +671,7 @@ public: void outdateUnexpectedPartAndCloneToDetached(const DataPartPtr & part); /// If the part is Obsolete and not used by anybody else, immediately delete it from filesystem and remove from memory. - void tryRemovePartImmediately(DataPartPtr && part); + bool tryRemovePartImmediately(DataPartPtr && part); /// Returns old inactive parts that can be deleted. At the same time removes them from the list of parts but not from the disk. /// If 'force' - don't wait for old_parts_lifetime. diff --git a/src/Storages/MergeTree/ReplicatedMergeTreeSink.cpp b/src/Storages/MergeTree/ReplicatedMergeTreeSink.cpp index 5b235322394..8e25510b3d9 100644 --- a/src/Storages/MergeTree/ReplicatedMergeTreeSink.cpp +++ b/src/Storages/MergeTree/ReplicatedMergeTreeSink.cpp @@ -44,6 +44,7 @@ namespace ErrorCodes extern const int LOGICAL_ERROR; extern const int TABLE_IS_READ_ONLY; extern const int QUERY_WAS_CANCELLED; + extern const int CHECKSUM_DOESNT_MATCH; } template @@ -801,8 +802,46 @@ std::pair, bool> ReplicatedMergeTreeSinkImpl:: "Conflict block ids and block number lock should not " "be empty at the same time for async inserts"); - /// Information about the part. - storage.getCommitPartOps(ops, part, block_id_path); + if constexpr (!async_insert) + { + if (!existing_part_name.empty()) + { + LOG_DEBUG(log, "Will check part {} checksums", existing_part_name); + try + { + storage.checkPartChecksumsAndAddCommitOps(zookeeper, part, ops, existing_part_name); + } + catch (const zkutil::KeeperException &) + { + throw; + } + catch (const Exception & ex) + { + if (ex.code() == ErrorCodes::CHECKSUM_DOESNT_MATCH) + { + LOG_INFO( + log, + "Block with ID {} has the same deduplication hash as other part {} on other replica, but checksums (which " + "include metadata files like columns.txt) doesn't match, will not write it locally", + block_id, + existing_part_name); + return; + } + throw; + } + } + else + { + /// Information about the part. + storage.getCommitPartOps(ops, part, block_id_path); + } + } + else + { + chassert(existing_part_name.empty()); + storage.getCommitPartOps(ops, part, block_id_path); + } + /// It's important to create it outside of lock scope because /// otherwise it can lock parts in destructor and deadlock is possible. diff --git a/src/Storages/StorageReplicatedMergeTree.cpp b/src/Storages/StorageReplicatedMergeTree.cpp index 2316f9b0225..623d1ea403a 100644 --- a/src/Storages/StorageReplicatedMergeTree.cpp +++ b/src/Storages/StorageReplicatedMergeTree.cpp @@ -1476,8 +1476,12 @@ void StorageReplicatedMergeTree::syncPinnedPartUUIDs() } } -void StorageReplicatedMergeTree::checkPartChecksumsAndAddCommitOps(const zkutil::ZooKeeperPtr & zookeeper, - const DataPartPtr & part, Coordination::Requests & ops, String part_name, NameSet * absent_replicas_paths) +void StorageReplicatedMergeTree::checkPartChecksumsAndAddCommitOps( + const ZooKeeperWithFaultInjectionPtr & zookeeper, + const DataPartPtr & part, + Coordination::Requests & ops, + String part_name, + NameSet * absent_replicas_paths) { if (part_name.empty()) part_name = part->name; @@ -1596,7 +1600,7 @@ MergeTreeData::DataPartsVector StorageReplicatedMergeTree::checkPartChecksumsAnd size_t zero_copy_lock_ops_size = ops.size(); /// Checksums are checked here and `ops` is filled. In fact, the part is added to ZK just below, when executing `multi`. - checkPartChecksumsAndAddCommitOps(zookeeper, part, ops, part->name, &absent_part_paths_on_replicas); + checkPartChecksumsAndAddCommitOps(std::make_shared(zookeeper), part, ops, part->name, &absent_part_paths_on_replicas); /// Do not commit if the part is obsolete, we have just briefly checked its checksums if (transaction.isEmpty()) diff --git a/src/Storages/StorageReplicatedMergeTree.h b/src/Storages/StorageReplicatedMergeTree.h index 1f37416f881..f8ca81c71e3 100644 --- a/src/Storages/StorageReplicatedMergeTree.h +++ b/src/Storages/StorageReplicatedMergeTree.h @@ -631,8 +631,12 @@ private: * Adds actions to `ops` that add data about the part into ZooKeeper. * Call under lockForShare. */ - void checkPartChecksumsAndAddCommitOps(const zkutil::ZooKeeperPtr & zookeeper, const DataPartPtr & part, - Coordination::Requests & ops, String part_name = "", NameSet * absent_replicas_paths = nullptr); + void checkPartChecksumsAndAddCommitOps( + const ZooKeeperWithFaultInjectionPtr & zookeeper, + const DataPartPtr & part, + Coordination::Requests & ops, + String part_name = "", + NameSet * absent_replicas_paths = nullptr); String getChecksumsForZooKeeper(const MergeTreeDataPartChecksums & checksums) const; diff --git a/tests/queries/0_stateless/02882_replicated_fetch_checksums_doesnt_match.reference b/tests/queries/0_stateless/02882_replicated_fetch_checksums_doesnt_match.reference new file mode 100644 index 00000000000..23dfd352361 --- /dev/null +++ b/tests/queries/0_stateless/02882_replicated_fetch_checksums_doesnt_match.reference @@ -0,0 +1,6 @@ +1 +1 +0 +1 +1 +1 diff --git a/tests/queries/0_stateless/02882_replicated_fetch_checksums_doesnt_match.sql b/tests/queries/0_stateless/02882_replicated_fetch_checksums_doesnt_match.sql new file mode 100644 index 00000000000..8ee9d672659 --- /dev/null +++ b/tests/queries/0_stateless/02882_replicated_fetch_checksums_doesnt_match.sql @@ -0,0 +1,42 @@ +DROP TABLE IF EXISTS r1; +DROP TABLE IF EXISTS r2; +DROP TABLE IF EXISTS r3; + +CREATE TABLE checksums_r1 (column1 UInt32, column2 String) Engine = ReplicatedMergeTree('/tables/{database}/checksums_table', 'r1') ORDER BY tuple(); + +CREATE TABLE checksums_r2 (column1 UInt32, column2 String) Engine = ReplicatedMergeTree('/tables/{database}/checksums_table', 'r2') ORDER BY tuple(); + +CREATE TABLE checksums_r3 (column1 UInt32, column2 String) Engine = ReplicatedMergeTree('/tables/{database}/checksums_table', 'r3') ORDER BY tuple(); + +SYSTEM STOP REPLICATION QUEUES checksums_r2; +SYSTEM STOP REPLICATION QUEUES checksums_r3; + +ALTER TABLE checksums_r1 MODIFY COLUMN column1 Int32 SETTINGS alter_sync=1; + +INSERT INTO checksums_r1 VALUES (1, 'hello'); + +INSERT INTO checksums_r3 VALUES (1, 'hello'); + +SYSTEM START REPLICATION QUEUES checksums_r2; + +SYSTEM SYNC REPLICA checksums_r2; + +SELECT count() FROM checksums_r1; +SELECT count() FROM checksums_r2; +SELECT count() FROM checksums_r3; + +SYSTEM START REPLICATION QUEUES checksums_r3; +SYSTEM SYNC REPLICA checksums_r3; + +SELECT count() FROM checksums_r1; +SELECT count() FROM checksums_r2; +SELECT count() FROM checksums_r3; + +SYSTEM FLUSH LOGS; + +SELECT * FROM system.text_log WHERE event_time >= now() - 30 and level == 'Error' and message like '%CHECKSUM_DOESNT_MATCH%'and message like '%checksums_r%'; + +DROP TABLE IF EXISTS checksums_r3; +DROP TABLE IF EXISTS checksums_r2; +DROP TABLE IF EXISTS checksums_r1; + From 46fae52c292e3b0d1c43a6a084812889ba6dee94 Mon Sep 17 00:00:00 2001 From: Alexander Tokmakov Date: Tue, 12 Sep 2023 19:57:16 +0200 Subject: [PATCH 317/433] fix zero copy garbage --- src/Storages/StorageReplicatedMergeTree.cpp | 2 +- .../02446_parent_zero_copy_locks.reference | 12 +++++ .../02446_parent_zero_copy_locks.sql | 49 +++++++++++++++++++ 3 files changed, 62 insertions(+), 1 deletion(-) create mode 100644 tests/queries/0_stateless/02446_parent_zero_copy_locks.reference create mode 100644 tests/queries/0_stateless/02446_parent_zero_copy_locks.sql diff --git a/src/Storages/StorageReplicatedMergeTree.cpp b/src/Storages/StorageReplicatedMergeTree.cpp index e6edb99e301..89b6493e81d 100644 --- a/src/Storages/StorageReplicatedMergeTree.cpp +++ b/src/Storages/StorageReplicatedMergeTree.cpp @@ -9113,7 +9113,7 @@ std::pair StorageReplicatedMergeTree::unlockSharedDataByID( zookeeper_ptr, fs::path(zc_zookeeper_path).parent_path(), part_info, data_format_version, logger); // parent_not_to_remove == std::nullopt means that we were unable to retrieve parts set - if (has_parent || parent_not_to_remove == std::nullopt) + if (has_parent && parent_not_to_remove == std::nullopt) { LOG_TRACE(logger, "Failed to get mutation parent on {} for part {}, refusing to remove blobs", zookeeper_part_replica_node, part_name); return {false, {}}; diff --git a/tests/queries/0_stateless/02446_parent_zero_copy_locks.reference b/tests/queries/0_stateless/02446_parent_zero_copy_locks.reference new file mode 100644 index 00000000000..c3c7c53f625 --- /dev/null +++ b/tests/queries/0_stateless/02446_parent_zero_copy_locks.reference @@ -0,0 +1,12 @@ +0 +1 1 10 1 +1 2 20 2 +1 3 30 3 +1 4 40 4 +1 5 50 5 +2 1 10 1 +2 2 20 2 +2 3 30 3 +2 4 40 4 +2 5 50 5 +3 0 diff --git a/tests/queries/0_stateless/02446_parent_zero_copy_locks.sql b/tests/queries/0_stateless/02446_parent_zero_copy_locks.sql new file mode 100644 index 00000000000..86eda526c72 --- /dev/null +++ b/tests/queries/0_stateless/02446_parent_zero_copy_locks.sql @@ -0,0 +1,49 @@ +-- Tags: no-replicated-database, no-fasttest +-- Tag no-replicated-database: different number of replicas + +create table rmt1 (n int, m int, k int) engine=ReplicatedMergeTree('/test/02446/{database}/rmt', '1') order by n + settings storage_policy='s3_cache', allow_remote_fs_zero_copy_replication=1, old_parts_lifetime=0, cleanup_delay_period=0, max_cleanup_delay_period=1, cleanup_delay_period_random_add=1, min_bytes_for_wide_part=0; +create table rmt2 (n int, m int, k int) engine=ReplicatedMergeTree('/test/02446/{database}/rmt', '2') order by n + settings storage_policy='s3_cache', allow_remote_fs_zero_copy_replication=1, old_parts_lifetime=0, cleanup_delay_period=0, max_cleanup_delay_period=1, cleanup_delay_period_random_add=1, min_bytes_for_wide_part=0; + +-- FIXME zero-copy locks may remain in ZooKeeper forever if we failed to insert a part. +-- Probably that's why we have to replace repsistent lock with ephemeral sometimes. +-- See also "Replacing persistent lock with ephemeral for path {}. It can happen only in case of local part loss" +-- in StorageReplicatedMergeTree::createZeroCopyLockNode +set insert_keeper_fault_injection_probability=0; + +insert into rmt1 values(1, 1, 1); +insert into rmt2 values(2, 2, 2); + +alter table rmt1 update m = 0 where n=0; +insert into rmt1 values(3, 3, 3); +insert into rmt2 values(4, 4, 4); +select sleepEachRow(0.5) as test_does_not_rely_on_this; + +insert into rmt1 values(5, 5, 5); +alter table rmt2 update m = m * 10 where 1 settings mutations_sync=2; + +system sync replica rmt2; +set optimize_throw_if_noop=1; +optimize table rmt2 final; + +select 1, * from rmt1 order by n; + +system sync replica rmt1; +select 2, * from rmt2 order by n; + +-- a funny way to wait for outdated parts to be removed +select sleep(1), sleepEachRow(0.1) from url('http://localhost:8123/?param_tries={1..10}&query=' || encodeURLComponent( + 'select *, _state from system.parts where database=''' || currentDatabase() || ''' and table like ''rmt%'' and active=0' + ), 'LineAsString', 's String') settings max_threads=1 format Null; + +select *, _state from system.parts where database=currentDatabase() and table like 'rmt%' and active=0; + +-- ensure that old zero copy locks were removed +set allow_unrestricted_reads_from_keeper=1; +select count(), sum(ephemeralOwner) from system.zookeeper where path like '/clickhouse/zero_copy/zero_copy_s3/' || + (select value from system.zookeeper where path='/test/02446/'||currentDatabase()||'/rmt' and name='table_shared_id') || '/%'; + +select * from system.zookeeper where path like '/clickhouse/zero_copy/zero_copy_s3/' || + (select value from system.zookeeper where path='/test/02446/'||currentDatabase()||'/rmt' and name='table_shared_id') || '/%' + and path not like '%/all_0_5_2_6%'; From 8a4654c39eabd8de3e700c32c7e2b3c2fd613507 Mon Sep 17 00:00:00 2001 From: Igor Nikonov Date: Tue, 12 Sep 2023 18:06:00 +0000 Subject: [PATCH 318/433] Address review comment --- tests/ci/fast_test_check.py | 7 +++---- tests/ci/report.py | 4 ++++ tests/ci/stress_check.py | 2 +- 3 files changed, 8 insertions(+), 5 deletions(-) diff --git a/tests/ci/fast_test_check.py b/tests/ci/fast_test_check.py index 13dd6dcb5e2..7184a92d36f 100644 --- a/tests/ci/fast_test_check.py +++ b/tests/ci/fast_test_check.py @@ -146,7 +146,7 @@ def main(): run_log_path = logs_path / "run.log" timeout_expired = False - timeout = 90 * 60 + timeout = 1 * 60 with TeePopen(run_cmd, run_log_path, timeout=timeout) as process: retcode = process.wait() if process.timeout_exceeded: @@ -186,10 +186,9 @@ def main(): state, description, test_results, additional_logs = process_results(output_path) if timeout_expired: - test_result_name = "Check timeout expired" - test_results.append(TestResult(test_result_name, "FAIL", timeout)) + test_results.append(TestResult.create_check_timeout_expired(timeout)) state = "failure" - description = format_description(test_result_name) + description = format_description(test_results[-1].name) ch_helper = ClickHouseHelper() s3_path_prefix = os.path.join( diff --git a/tests/ci/report.py b/tests/ci/report.py index eb6e3bc1206..b93128262d9 100644 --- a/tests/ci/report.py +++ b/tests/ci/report.py @@ -249,6 +249,10 @@ class TestResult: assert file.exists(), file self.log_files.append(file) + @staticmethod + def create_check_timeout_expired(timeout : float) -> "TestResult": + return TestResult("Check timeout expired", "FAIL", timeout) + TestResults = List[TestResult] diff --git a/tests/ci/stress_check.py b/tests/ci/stress_check.py index 4e79bdc769d..58453220b6c 100644 --- a/tests/ci/stress_check.py +++ b/tests/ci/stress_check.py @@ -175,7 +175,7 @@ def run_stress_test(docker_image_name): logging.info("Going to run stress test: %s", run_command) timeout_expired = False - timeout = 60 * 1 + timeout = 60 * 150 with TeePopen(run_command, run_log_path, timeout=timeout) as process: retcode = process.wait() if process.timeout_exceeded: From 85b23e94034b76ecafcdc82807c515b0b23837a2 Mon Sep 17 00:00:00 2001 From: pufit Date: Tue, 12 Sep 2023 14:06:57 -0400 Subject: [PATCH 319/433] Increase default timeout in tests for keeper-client --- tests/integration/helpers/keeper_utils.py | 24 +++++++++++------------ 1 file changed, 12 insertions(+), 12 deletions(-) diff --git a/tests/integration/helpers/keeper_utils.py b/tests/integration/helpers/keeper_utils.py index a1d20e0392b..19b128abe2c 100644 --- a/tests/integration/helpers/keeper_utils.py +++ b/tests/integration/helpers/keeper_utils.py @@ -71,7 +71,7 @@ class KeeperClient(object): self.stopped = False - def execute_query(self, query: str, timeout: float = 10.0) -> str: + def execute_query(self, query: str, timeout: float = 60.0) -> str: output = io.BytesIO() self.proc.stdin.write(query.encode() + b"\n") @@ -103,19 +103,19 @@ class KeeperClient(object): data = output.getvalue().strip().decode() return data - def cd(self, path: str, timeout: float = 10.0): + def cd(self, path: str, timeout: float = 60.0): self.execute_query(f"cd {path}", timeout) - def ls(self, path: str, timeout: float = 10.0) -> list[str]: + def ls(self, path: str, timeout: float = 60.0) -> list[str]: return self.execute_query(f"ls {path}", timeout).split(" ") - def create(self, path: str, value: str, timeout: float = 10.0): + def create(self, path: str, value: str, timeout: float = 60.0): self.execute_query(f"create {path} {value}", timeout) - def get(self, path: str, timeout: float = 10.0) -> str: + def get(self, path: str, timeout: float = 60.0) -> str: return self.execute_query(f"get {path}", timeout) - def exists(self, path: str, timeout: float = 10.0) -> bool: + def exists(self, path: str, timeout: float = 60.0) -> bool: return bool(int(self.execute_query(f"exists {path}", timeout))) def stop(self): @@ -123,19 +123,19 @@ class KeeperClient(object): self.stopped = True self.proc.communicate(b"exit\n", timeout=10.0) - def sync(self, path: str, timeout: float = 10.0): + def sync(self, path: str, timeout: float = 60.0): self.execute_query(f"sync {path}", timeout) - def touch(self, path: str, timeout: float = 10.0): + def touch(self, path: str, timeout: float = 60.0): self.execute_query(f"touch {path}", timeout) - def find_big_family(self, path: str, n: int = 10, timeout: float = 10.0) -> str: + def find_big_family(self, path: str, n: int = 10, timeout: float = 60.0) -> str: return self.execute_query(f"find_big_family {path} {n}", timeout) - def find_super_nodes(self, threshold: int, timeout: float = 10.0) -> str: + def find_super_nodes(self, threshold: int, timeout: float = 60.0) -> str: return self.execute_query(f"find_super_nodes {threshold}", timeout) - def delete_stale_backups(self, timeout: float = 10.0) -> str: + def delete_stale_backups(self, timeout: float = 60.0) -> str: return self.execute_query("delete_stale_backups", timeout) def reconfig( @@ -143,7 +143,7 @@ class KeeperClient(object): joining: tp.Optional[str], leaving: tp.Optional[str], new_members: tp.Optional[str], - timeout: float = 10.0, + timeout: float = 60.0, ) -> str: if bool(joining) + bool(leaving) + bool(new_members) != 1: raise ValueError( From e587eae303cab752f378439541434f6b0e64665e Mon Sep 17 00:00:00 2001 From: Robert Schulze Date: Tue, 12 Sep 2023 18:07:46 +0000 Subject: [PATCH 320/433] Docs: Update information_schema with case-insensitity --- .../operations/system-tables/information_schema.md | 13 ++++++++++--- 1 file changed, 10 insertions(+), 3 deletions(-) diff --git a/docs/en/operations/system-tables/information_schema.md b/docs/en/operations/system-tables/information_schema.md index 35fd3a753b5..bdaef8372d6 100644 --- a/docs/en/operations/system-tables/information_schema.md +++ b/docs/en/operations/system-tables/information_schema.md @@ -3,12 +3,13 @@ slug: /en/operations/system-tables/information_schema --- # INFORMATION_SCHEMA -`INFORMATION_SCHEMA` (`information_schema`) is a system database that contains views. Using these views, you can get information about the metadata of database objects. These views read data from the columns of the [system.columns](../../operations/system-tables/columns.md), [system.databases](../../operations/system-tables/databases.md) and [system.tables](../../operations/system-tables/tables.md) system tables. - -The structure and composition of system tables may change in different versions of the product, but the support of the `information_schema` makes it possible to change the structure of system tables without changing the method of access to metadata. Metadata requests do not depend on the DBMS used. +`INFORMATION_SCHEMA` (or: `information_schema`) is a system database which provides a (somewhat) standardized, [DBMS-agnostic view](https://en.wikipedia.org/wiki/Information_schema) on metadata of database objects. The views in `INFORMATION_SCHEMA` are generally inferior to normal system tables but tools can use them to obtain basic information in a cross-DBMS manner. The structure and content of views in `INFORMATION_SCHEMA` is supposed to evolves in a backwards-compatible way, i.e. only new functionality is added but existing functionality is not changed or removed. In terms of internal implementation, views in `INFORMATION_SCHEMA` usually map to to normal systme tables like [system.columns](../../operations/system-tables/columns.md), [system.databases](../../operations/system-tables/databases.md) and [system.tables](../../operations/system-tables/tables.md). ``` sql SHOW TABLES FROM INFORMATION_SCHEMA; + +-- or: +SHOW TABLES FROM information_schema; ``` ``` text @@ -17,6 +18,10 @@ SHOW TABLES FROM INFORMATION_SCHEMA; │ SCHEMATA │ │ TABLES │ │ VIEWS │ +│ columns │ +│ schemata │ +│ tables │ +│ views │ └──────────┘ ``` @@ -27,6 +32,8 @@ SHOW TABLES FROM INFORMATION_SCHEMA; - [TABLES](#tables) - [VIEWS](#views) +Case-insensitive equivalent views, e.g. `INFORMATION_SCHEMA.columns` are provided for reasons of compatibility with other databases. + ## COLUMNS {#columns} Contains columns read from the [system.columns](../../operations/system-tables/columns.md) system table and columns that are not supported in ClickHouse or do not make sense (always `NULL`), but must be by the standard. From 4671901313dbb6c247e2312b49fc59af533d2a54 Mon Sep 17 00:00:00 2001 From: Igor Nikonov Date: Tue, 12 Sep 2023 18:17:55 +0000 Subject: [PATCH 321/433] Update stress_check.py --- tests/ci/stress_check.py | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) diff --git a/tests/ci/stress_check.py b/tests/ci/stress_check.py index 58453220b6c..4d21791c904 100644 --- a/tests/ci/stress_check.py +++ b/tests/ci/stress_check.py @@ -195,10 +195,9 @@ def run_stress_test(docker_image_name): ) if timeout_expired: - test_result_name = "Check timeout expired" - test_results.append(TestResult(test_result_name, "FAIL", timeout)) + test_results.append(TestResult.create_check_timeout_expired(timeout)) state = "failure" - description = format_description(test_result_name) + description = format_description(test_results[-1].name) ch_helper = ClickHouseHelper() From 63f71b719d7a50d9493edbc054cb6ed64e8ef4b1 Mon Sep 17 00:00:00 2001 From: robot-clickhouse Date: Tue, 12 Sep 2023 18:31:34 +0000 Subject: [PATCH 322/433] Automatic style fix --- tests/ci/report.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tests/ci/report.py b/tests/ci/report.py index b93128262d9..d29f274cd2f 100644 --- a/tests/ci/report.py +++ b/tests/ci/report.py @@ -250,7 +250,7 @@ class TestResult: self.log_files.append(file) @staticmethod - def create_check_timeout_expired(timeout : float) -> "TestResult": + def create_check_timeout_expired(timeout: float) -> "TestResult": return TestResult("Check timeout expired", "FAIL", timeout) From 8eb12c33a89c05289e41589b793e843b9abafa7c Mon Sep 17 00:00:00 2001 From: Robert Schulze Date: Tue, 12 Sep 2023 20:33:13 +0200 Subject: [PATCH 323/433] Update information_schema.md --- docs/en/operations/system-tables/information_schema.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/docs/en/operations/system-tables/information_schema.md b/docs/en/operations/system-tables/information_schema.md index bdaef8372d6..ee03441b9c1 100644 --- a/docs/en/operations/system-tables/information_schema.md +++ b/docs/en/operations/system-tables/information_schema.md @@ -3,7 +3,7 @@ slug: /en/operations/system-tables/information_schema --- # INFORMATION_SCHEMA -`INFORMATION_SCHEMA` (or: `information_schema`) is a system database which provides a (somewhat) standardized, [DBMS-agnostic view](https://en.wikipedia.org/wiki/Information_schema) on metadata of database objects. The views in `INFORMATION_SCHEMA` are generally inferior to normal system tables but tools can use them to obtain basic information in a cross-DBMS manner. The structure and content of views in `INFORMATION_SCHEMA` is supposed to evolves in a backwards-compatible way, i.e. only new functionality is added but existing functionality is not changed or removed. In terms of internal implementation, views in `INFORMATION_SCHEMA` usually map to to normal systme tables like [system.columns](../../operations/system-tables/columns.md), [system.databases](../../operations/system-tables/databases.md) and [system.tables](../../operations/system-tables/tables.md). +`INFORMATION_SCHEMA` (or: `information_schema`) is a system database which provides a (somewhat) standardized, [DBMS-agnostic view](https://en.wikipedia.org/wiki/Information_schema) on metadata of database objects. The views in `INFORMATION_SCHEMA` are generally inferior to normal system tables but tools can use them to obtain basic information in a cross-DBMS manner. The structure and content of views in `INFORMATION_SCHEMA` is supposed to evolves in a backwards-compatible way, i.e. only new functionality is added but existing functionality is not changed or removed. In terms of internal implementation, views in `INFORMATION_SCHEMA` usually map to to normal system tables like [system.columns](../../operations/system-tables/columns.md), [system.databases](../../operations/system-tables/databases.md) and [system.tables](../../operations/system-tables/tables.md). ``` sql SHOW TABLES FROM INFORMATION_SCHEMA; From 031da054f3fa8ab015315f61d0b2339e18108842 Mon Sep 17 00:00:00 2001 From: Igor Nikonov Date: Tue, 12 Sep 2023 19:04:59 +0000 Subject: [PATCH 324/433] Return initial timeout --- tests/ci/fast_test_check.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tests/ci/fast_test_check.py b/tests/ci/fast_test_check.py index 7184a92d36f..f1bbf356fbb 100644 --- a/tests/ci/fast_test_check.py +++ b/tests/ci/fast_test_check.py @@ -146,7 +146,7 @@ def main(): run_log_path = logs_path / "run.log" timeout_expired = False - timeout = 1 * 60 + timeout = 90 * 60 with TeePopen(run_cmd, run_log_path, timeout=timeout) as process: retcode = process.wait() if process.timeout_exceeded: From 83ac2bf508450cc69093e2569660bace79f0dbb3 Mon Sep 17 00:00:00 2001 From: Arthur Passos Date: Tue, 12 Sep 2023 16:14:28 -0300 Subject: [PATCH 325/433] add test --- ...tches_array_inconsistent_offsets.reference | 3 + ...iple_batches_array_inconsistent_offsets.sh | 129 ++++++++++++++++++ ...consistent_offset_multiple_batches.parquet | Bin 0 -> 6953397 bytes 3 files changed, 132 insertions(+) create mode 100644 tests/queries/0_stateless/02874_parquet_multiple_batches_array_inconsistent_offsets.reference create mode 100755 tests/queries/0_stateless/02874_parquet_multiple_batches_array_inconsistent_offsets.sh create mode 100644 tests/queries/0_stateless/data_parquet/string_int_list_inconsistent_offset_multiple_batches.parquet diff --git a/tests/queries/0_stateless/02874_parquet_multiple_batches_array_inconsistent_offsets.reference b/tests/queries/0_stateless/02874_parquet_multiple_batches_array_inconsistent_offsets.reference new file mode 100644 index 00000000000..ba63f2f7e9c --- /dev/null +++ b/tests/queries/0_stateless/02874_parquet_multiple_batches_array_inconsistent_offsets.reference @@ -0,0 +1,3 @@ +Parquet +e76a749f346078a6a43e0cbd25f0d18a - +400 diff --git a/tests/queries/0_stateless/02874_parquet_multiple_batches_array_inconsistent_offsets.sh b/tests/queries/0_stateless/02874_parquet_multiple_batches_array_inconsistent_offsets.sh new file mode 100755 index 00000000000..1d11a2705b3 --- /dev/null +++ b/tests/queries/0_stateless/02874_parquet_multiple_batches_array_inconsistent_offsets.sh @@ -0,0 +1,129 @@ +#!/usr/bin/env bash +# Tags: no-ubsan, no-fasttest + +CUR_DIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) +# shellcheck source=../shell_config.sh +. "$CUR_DIR"/../shell_config.sh + +echo "Parquet" + +# More info on: https://github.com/ClickHouse/ClickHouse/pull/54370 + +# File generated with the below code + +#std::string random_string(size_t length) { +# static const std::string characters = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz"; +# +# std::random_device random_device; +# std::mt19937 generator(random_device()); +# std::uniform_int_distribution<> distribution(0, characters.size() - 1); +# +# std::string random_string; +# random_string.reserve(length); +# +# std::generate_n(std::back_inserter(random_string), length, [&]() { +# return characters[distribution(generator)]; +# }); +# +# return random_string; +#} +# +#static const std::string the_string = random_string(9247124u); +# +#std::shared_ptr CreateIntArray(std::size_t length) { +# arrow::MemoryPool* pool = arrow::default_memory_pool(); +# +# auto int_builder_ptr = std::make_shared(pool); +# auto & int_builder = *int_builder_ptr; +# arrow::ListBuilder list_builder(pool, int_builder_ptr); +# +# for (auto i = 0u; i < length; i++) +# { +# if (i % 10 == 0) +# { +# ARROW_CHECK_OK(list_builder.Append()); +# } +# else +# { +# ARROW_CHECK_OK(int_builder.Append(i)); +# } +# } +# +# std::shared_ptr int_list_array; +# ARROW_CHECK_OK(list_builder.Finish(&int_list_array)); +# return int_list_array; +#} +# +#std::shared_ptr CreateStringArray(std::size_t length) { +# arrow::MemoryPool* pool = arrow::default_memory_pool(); +# +# auto str_builder = std::make_shared(arrow::large_utf8(), pool); +# +# for (auto i = 0u; i < length; i++) +# { +# if (i % 10 == 0) +# { +# ARROW_CHECK_OK(str_builder->AppendNull()); +# } +# else +# { +# ARROW_CHECK_OK(str_builder->Append(the_string)); +# } +# } +# +# std::shared_ptr str_array; +# ARROW_CHECK_OK(str_builder->Finish(&str_array)); +# return str_array; +#} +# +#void run() +#{ +# auto schema = arrow::schema({ +# arrow::field("ints", arrow::list(arrow::int64())), +# arrow::field("strings", arrow::utf8()) +# }); +# +# auto l1_length = 2000; +# auto l2_length = 2000; +# +# std::vector> batches; +# +# auto int_array1 = CreateIntArray(l1_length); +# +# auto int_array2 = CreateIntArray(l1_length); +# +# auto str_array1 = CreateStringArray(l2_length); +# +# auto str_array2 = CreateStringArray(l2_length); +# +# batches.push_back(arrow::RecordBatch::Make(schema, int_array1->length(), {int_array1, str_array1})); +# +# batches.push_back(arrow::RecordBatch::Make(schema, int_array2->length(), {int_array2, str_array2})); +# +# std::shared_ptr outfile; +# PARQUET_ASSIGN_OR_THROW(outfile, arrow::io::FileOutputStream::Open("generated.parquet")); +# +# parquet::WriterProperties::Builder builder; +# builder.compression(arrow::Compression::GZIP); +# builder.dictionary_pagesize_limit(10*1024*1024); // Given your string size is roughly 9.2MB +# builder.data_pagesize(20*1024*1024); // Given your dataset's repetition, let's try with 20MB +# +# std::shared_ptr props = builder.build(); +# +# std::unique_ptr file_writer; +# PARQUET_ASSIGN_OR_THROW(file_writer, parquet::arrow::FileWriter::Open(*schema, ::arrow::default_memory_pool(), outfile, props)); +# +# for (const auto& batch : batches) { +# PARQUET_THROW_NOT_OK(file_writer->WriteRecordBatch(*batch)); +# } +# +# PARQUET_THROW_NOT_OK(file_writer->Close()); +#} + +DATA_FILE=$CUR_DIR/data_parquet/string_int_list_inconsistent_offset_multiple_batches.parquet +${CLICKHOUSE_CLIENT} --query="DROP TABLE IF EXISTS parquet_load" +${CLICKHOUSE_CLIENT} --query="CREATE TABLE parquet_load (ints Array(Int64), strings Nullable(String)) ENGINE = MergeTree ORDER BY tuple()" +cat "$DATA_FILE" | ${CLICKHOUSE_CLIENT} -q "INSERT INTO parquet_load FORMAT Parquet" +${CLICKHOUSE_CLIENT} --query="SELECT * FROM parquet_load SETTINGS max_threads=1" | md5sum +${CLICKHOUSE_CLIENT} --query="SELECT count() FROM parquet_load" +${CLICKHOUSE_CLIENT} --query="drop table parquet_load" \ No newline at end of file diff --git a/tests/queries/0_stateless/data_parquet/string_int_list_inconsistent_offset_multiple_batches.parquet b/tests/queries/0_stateless/data_parquet/string_int_list_inconsistent_offset_multiple_batches.parquet new file mode 100644 index 0000000000000000000000000000000000000000..ca0e2cf67627b8438ad9f38449f2321bab1e1a1a GIT binary patch literal 6953397 zcmb5#V{n{*!0z$bwry{0JB@9t@y51oCv9WfYV4#j8{2l$=ifXpPUp;db9Q!ScXr=h zv-6qn-1l#lL{tEn5SRhmpqSt2zhEX1V1l55fDlK+fP6l{IoA5YNTT8(el0cIt^NCa zIM<#yww>ZCRDr$f`I8@jFz^$h?;FC<7DBHQ!XyjAxFfEXT5o6gBxf4-Dkre1V2RxCP)d!j0vVt2}YF(`s)aW zuXBDz<_wJH^jYUj;pU8L=S=eCjC<<+4AmPL)axtK8`{_FRo9#3&>MHvn?_t8kzent zS|2)J@6BBw{Jq}qyFLxhKO)9ILC!yB%HNyJKRCzV-@re7+xb(jbArBe%%*b+zjIW# zv%jNr_|@lUxX-|_PoJgF6t>T(rq3jo&$#>B&-k~2>9@X`x1r0oUcRtanXWs(bSjINQ~3Ol+#$1(-gkhjK0|dzuBCx*%I2##L~@J z)6LZV!ASAJg!sW+=;06qX^#Zy00n8+F>XsbZpSKaTPtoCLSc_W;Q&KnpIzY)VQR;A zYFlq=*N16KlxYWzX`hwp5UzTUuKED4de_T-OVWPF!hT!Bei!LtkLluo=we^+;t-T* z$CzkanP}H3Z%Za`2RCn@C+`r_V2{$^0MlUCbz@6@W5;e|TYqC0TyT$E@Bm$KpH1)( zzI(@}dt0}A*V}JP%5MkTZ=cKW5cy+|`Qw22W4~vR6$G3W1)L2gm=#Big+dI#-p|6& z58x+fK_h2_D`3SdV1qJc#oA?I+64%rvyh?#0Ld)4$!w6SteC27h;yujY%Jt#0Ja7e z`UU`>GYhse8}cnH@ofVLd_78ZElqSCw_GiSTpjy(Ex~vL7=1k&eFI!YJ)TW1l}#Pz zX)V)fogjWKDt-fOc0F!(1Eg*}=2|V|TAcu2EjeEuTSqN^M+3BXJ+^lP@@qXYv?B?$ z6HB-wO}G=clp}_eGr};CVAvUq8i-cnNLS*-Yw1X3>BM>HNOb57iVZ}?c7{y{;%Yil zYC3T&IWjIe32-^0b2-De0P$O#q1}PlkB-cbPD036B*@n+@mDzUw-Aaq7>c(D(>Dan zS7glBtTk73HP^fjS6B|WNS8N6m$#tAH>mknwE5ROhF6q^*BqNygqyeELO19_x9~kT z`2JVa{?}Z8u7v*EmV^@o9KaIHLK6(&{@fO!pOv7W4UC-ib6WtfDGQY;fb(-(P;^#Q zbT-&z*3WGL9CIv;a{z(QZQ&bO@f+BnomoG(1qi{{lEBxod~ORNSC1jrfG}SFxosV5 zMJ-)L9q;G1NT>BgrwyR^^`G0;@#xl4>eg|5ZVS#=kIvTs-%ee(q;_}W`n&Bl2=WaI@-0mK&F8k)?9*2a)7Si;+rrh{;MLqhIoy11do4(OMM``P z_}muK@CMWH7IE|DbK7gSo-6vEYrfBIk^k%(e(-5Q{RP1Y%J?4;%(?#o!DYzk9}vvQ z{sG}@vGG43m~;LEf*+RAKOmUV{sV%!hVefjm<#*^f^)OcKOmS<{{w= zf``J^KOpQe`~!js)Am0g?DPHug4@N`KOpRp{sV$_-u6Er>~s7Bg0JA#KOpSU{{w=h z-}XNs4EG>>+d@wW#K{oBN$JJOA;pRMjvHo-8@G)ch>9DjjFUQplMBE}uEWW8!im4m z41mpyh|CO;$&8`RNGi+7vd&05&d9^fh|J6k)y|Av$qeSnjP}e(f6mB<)Jh1|%8<~C z8PpD<)Q&3B4l~z|!_-Pl)yh=YN?p*(<^p>K_SZjFES41o5G2=@$;@{FPONGkEjvh+wh^vJ{Zh)nkk)%1*A@(kwkjCS`(fAq*l zeoBac%20fYnSKspevYbn4s&>pBYsNEf66p`O5J?Q6?%&9c@F<`&HM+`7@UO!9ProL zU^WObRtzyVgimY9$N{VcEOZ5czt)mf?XrP>T1%S@;8A6vR0aICc8(34jTN1Z z4gS+wYG(k~EerE)-Ct|r8(^aAaiSX_KCNXKuj8k$C8MwVYi&gXlubRBO#{-WwWRoU zfb3e@?7F|!>NX&*)f28YfPY#`-%-csT}$m<_t)B2XAo#03bZrKr?nJPPVB>u48u-; zt)+H`D*@t_I759}%XH`@i0w#;-|IbTuy8)j`S@~f30Jvp?Lk*+UZ*`<{LETTeweasT{63FRz#`um4(0d<&a@gPVT~`DrcV z=Cy#(6}iy$Uu%1Aq5W^L{cn-~+-&wsgFu;)Kw124gfJuNFcYgVW3A6dppdk{kThqP zwES!YW4$3$A1Wi!&qiRSvV=upnKO2FZ(*m8-oQ>1+vk{DSn@zpljHEsr zfy>Pj`N53&!QyWtAZ-aF?HD0#D}6QsP22%a+&*{Q;b$Xkt10ZdOl^JnYy|$PLolX2 zGNyySjZnQMTfOtOdRynS5vc4Bud$f56e;eV0 zSQ{kMOqd%41QP`Fn+9eM4Fl#kZcG$RC`>2<9GGe7|7~0x7|{QH1A+xY^Z)qtzc2q! z51%id>!L+~f;dif(Eb1#xRV#^I-g5c|IV`E5^d)Y-zPj+jZLthY{GAIv#I_KbDU-t z)+uEHV2Ii_9C|h^>BS=6o|6v`v=lvsum>&|vV|UP> z(?o8XSi`cXepLPgxf>1yWPU`GtsGRCi3D>}J`f5-n4b&MWT%)6yvFRdt)_RzQut07 zZhlB!B4a9GB~a5j&@w>FWfDtbtBL;&7iCfn901XsIxUnuDxM$u3|1z+hRZVU;^xht z5C*kqcKfS|@7u00f44>Sx*EJxTuJ(k&BpAv>GL1)!AH$_5gRV{T;BWz1RhvlMQ$Hx zhS=z@ER7A&k&SVt%lGnq`68;to`-sE*H+GzReygxJ)PAGC$l_#t2d=rUO?*jgjmSe(g$g8IWS0@um=B_~=)A#Xro|CSYH&Moh@<&nx(>FU(7v z8vsnK=IgmP=nRPp6PYEcn?c=n2M5$yfx+pAJ1jC|u{Zw|A}KyoLf+31#kf{3t^q1LZ!p+f{&-f_K=~GeJ){w} zka{n8QpBv(NSx$dT-C)hjY#zQOeM|l>V}N3kvYLH&yM!1wsOb@GWwKrXE39(*bz2r z@3UgCgl1CruaA?~_qXGttyJvmU(BE+xLi$Ixk>1f;F0xsBMH1sBzhT)DGPnkH9=#2 zIHh7kVhH+>nxWA^5$vbsokhkmM<=4|pajl&xMTSAJ$UAb%)uoiQ)7l%Cy_C{_GSUp zq3gCe1CY4%rtZoK|Gm9W2CZdPEoYA3PrG~ zgKX$YgT-InqU4G&Z-B;#gF-ZUf@=Eu!1eVPC1Z2pZUFEBobg%(+O9Qtk?U1A_ezDj zVUNnxHqF9O@5BUglRqazzzS^CwKHjooX6P(UwMO(_`(cpA~hCfUmz<-er0#trbI|Aau8(;kEwI`Zrsqv(bf4}WZNb< zD@L(eWi^#$H-WTD1}PR{jnN143fUkEfwzZRUMkc|mxM%__a8)}Z;lN1eNu?$92MsH{Awx~`$R z&|O>0xrWlB7NNQ;E_+F55=={$Lk5i-nMrh%ie8o4C-=-!Pq?^K=M=^~CGPBC7@Pc( zuh+yUG2_MzQ%}-uPmyW={oX7%yk4>Xl%ebAS?*lC4r6IU9qvn{-G4L;?8vA}%7uh` zvm`bQG$-n4>@7m~1{IZ~DvxJDJ+9cUpba0|MxIpVGiosA!kOts>N>vDj}cl2!%0U$ZGA6;WXG>F{yhDtQ+ zis>`U*(!irCT7`ev9ulT_A~9-19+MyxrRc}#6W}D(!CHzyfLVSHLG8hV6b*&KUH7K ze~ULC)pDlH7)?L2WK_Tj-`C1ehGFQ<7t-?MZCs`KboC))vR$anCog*?EnBkI*odB0 zu+45*qXRKor;P-?rK=&?h?-U71hbm+$)F>5OuMagsxvQ~oExcB1Fv!Kv;%3CkbQTI zGWY4FLkd&+=lHF>RiXl&)moRvF3xnc~p?x4#p=oQCQ zec#y{rOWNnY3i#gNduuZrH4YPMt1+6+kENoe%5M(xQMk)WK_Y!Cr#a*>W6p0>tRW9 zg(ir$jtibsAyjiQluBbpRU9hO*YmH<;<}y5PoW8Qi*T+sb>4)A(mXaxzy|v*rva(^ z@}_3~F~l8_Ve&SobFc0_sArmMxm17oV3*Brc`iOHNqRO?maZv_sKhOOi%lM~!-0~+ zrifqPCCdtTgWYyC31jY&+B+BTd3~|e?&7}8KF=u3W(1?2YqauZ zdgq!FDcn`_o@zkG?fDl*Jy_AU#1Xo@bPXK|PA*Lmd{G6PK>!L?gfG-ewx)z3s+ zGl1D_Cq7slLxR!L_WpQe0*WuD?@o{BbAZEJ?9pX05UpW(a%h$2L3!cwYKsc`gmz}Z z1jdDh3ySz)q`^t0__&rO4kSTL;t9yj?cEV{-K-rk&w0MS;A+^B_b12hySOJBPqYUL zdl9b-X~lM`7Kecb&;+LJXc86NAwHXIc&1gdm@8=rOWVY6B*t|^QmTr1rq_RSTnoaTMcll3cu+ML^=X@Jg3if@|tCWQXo z(D*u9<1*`=D@E%ZLoHL&>s)hM=zVb|37m&mJUyYgm1?$ovQ_hSFYwN5b1=0PFTb=d zVIbgZ+?l1luwQPObO&VZCdev1&u$%3Ocp;6s@BAPA$J#;wSiHz?5c z4tShBj*UGGSVAmj|5|9V)&ric>pfbf?%;NwJ=Zi@)wQ=+&eicSdAxeCXaKZX$ku;7 zaItf?VQy>U`h%?h%3QUdeV2d<0z!49Mt}){%K-v~g3AqIZSUd?VQcN|0&Q+vA6%1=jjcS6P3g-W(HE|jKyBZ6E3HCQxQT}`RuU#Q9YY&cKFlQrvL4Cml z`@gQw668t&Po=RWyz#r>J=h@7HoUolA4xjb#3sFZ5f{$wF|z1Rs1(L5&HCH0@o*n#{L)4_71b{eX}lX+ z)DM3axrz(#zo!4mG))dx*D4zAF_pM4u%o;Xckayfe6?XFat>SY^tB@|Yq4}Sx1F$D zgt57PnRy%Aw*RB@-6sr)*+F^#P5DUETLj~6Ve_cri~i--ZyMokNs)(Wt|lNwlhCPe zkwSc6d`j3}YB4RjF3b>tHfCId9jvCk5`tY4n0uL3Yi?;$wD&jR%TjXG6lv1i@&k+Ir0oXzH z8_8<_Dd~{QdcgL5O(BCf5|^Yba%*lUhIPGx7%IO( zR*AC$ECJ)=kwOIxt}4>ebg8-+?*~NF)2L=BwO$c5-OtLVck6mgnAgWz)BMC`b3ML1 zhC`hNXfcD|dMu6X9sAoTO~LeZ8KSDtI!H4ie5^qEVAiTq?fGpkkv_D!cDv@*KOVks zj#aQp_vKlB5omSp}h%ozxEZ^H;fX!CmMYnC(^MZ^yYLi@2xL(1q_@kLbMeH5yC1efK&T z7EFGOzuVYg*NQjo{OXxw*kuD*r~y(LJ*HJ5Y8bbWn4#=$k7oC{7_qLQkw4ZTN^rQVxUj?z=9T1Dh~yccScDp zhVsp&GW)r46e`KO#{9sm5pbVRTJ2%Bx9o$@Pm|kalMn5H)>py$rMh(9Au!zoh)8*P zvz}kUwga|EstU1m&#st*dGfNX5xO)16!Fu?TKzB`{y4R*2-Ee9;@_X214rgz)1 zrbMYQoHKebs0&c-tIXVElU^)63#2LoO>ikpnRSe7(8qEvJoJ}p>tF3MqC_+w?<#Vh z+NOJ>oPdOV9fAwu0ynUxY5DyHIT=lP_cYO)Nc+~UhB{`uTI!V(O-WLQ}onCcF z+1dAWUr4;qzM3FjaOJnro|f;8npv#;TDlmmA&d+nW?8*F`xYwESw!gmtesH__5Mpi zMamw*3YEtbEEONS#i4+u@Nmz|#*QYk{+NNkdOTJq^Yq7XcMcBsu8jSU${dc`9oZ_i zcX93pj=eyT2UJs}oqlzh#)v|?jb!r87V~=Uhfv2z#7TU9s-1TrGGOaGNM7gM!YCqH zpf`ziWfcotBI&o;?4HDvdlYDRv+L#bK64;fPq7%O%f?Go$f^rVAA;5NkwjT)Vnx~M z5+4W!=Pg8|<}K-}$@f}j&A2SE?s*%=eeWU}9Xg#e3wBX(3v`Y)1_x};0q%pZr2eKy zJYIjQ3|>*D7)+wV9_l@{tqyk5QA<#gP|qHA)q`xHGg3L;ze#Gdd+cd;2^Gq=lPkT< zaQeSvxvR#!n$s?mokKUesiOuhcvG7OHn9W_(fSqK>Ps(+L20{NAc^Rz6c? zB+`&eaw$;=@{((qi|QKf{yNZF{?%bC8+%+Q3Xmq4>y%i@9+4gy6!PX#Roz8-K8i&( z$%}T1!LI1LJZ%UCXz^GI5yrCD+PYX;OXPp>&()DV9c`;o9)l;?XBN5I)WJQZzleWi9JDj`CX0C;p!RyIEQSrRM(K>Jl<%dCydDnYt(v7CuYZQv_w6G~+w@9qJ z)Kyv}Aa#6kw6_JGX)Yfp_n(4NG(8|QaiL$!YP8*Bw*!{@CpDBj6H}!j!@L8&&jm#_ z1c$Dpk=|;q1&)u&p2T)P-ltaqt7Er zT-K!ZlrJM|(YGW)3q=tsykXKTOCt8I7aVKW;&l9gW8;#=6xt))!&#LrQdeezbj)wX z!6fCc%_VQ7`DENM+2yTHjb;B1@pdg9jM#wp&~l+JCEukgMLvJJl)T?Q~0h zXke-AwlAx=1yCHRqPa8=LvwPaq|}AK@eFgFCU;p(75SxefE`CqwfEZwRUq21u9m~f zK1h(Atg&^G){6akB2g0lnh}OQ2rZ!?Iq-(gW1B$D*E%^Hlnk!^AXsPF> zmcONOs+=nmU21Ny^@^=|_|sER2Uk_N*!)r)aIHZlO(<2Wl^ze_8Z`CwhA8XEj*z&! z3}L^Rf36w!y4#h8_J*i-f{1=WN#{*1>Ia`4te1hExs~+3Ooi0_ozzK@Ai@}~;jrvH zm{3y&0}W>7Uw`tsG{SVQRato8~*0wYhOLk#AEccYTbj zmog4}a>>c@9sT2A^bla|d5K~wE=~QQE@L8qWE8SLXP36On~~91&Utru$32V1&#V#+ zS_)TF6$I-+P)oPa1F=b^PmC$d=kdDf7fEpwGsy>C3#b)obfdeY1JJL6o*PLeg@s(m zxkd|vsv2VBu@l7z&(1%h#rAcua{ZRJZUTn-NZBXgLukn!(nTPnCKc2(l*I$VXx^@0 z8-Lgc=#6=i6t|p{xt0|HpsDtgQp1Qa6gro^dQ}QP)Dou#2#&BEB%`j-$xS$s5ziO> z9$FW&?bK9S)Ogqz)kEXvF=c_8tjcND<|rKoqSAJ0O@y4^l~zqARn1%8axST~ejutG zB2(?je@T%p!6D77m&dFY5%L+itibp5S;@QThPp)-Gla`mL#o_}JcxJlKk#)>lZN*^O|BUf@4tp9eOvP<`Ys;X}FM8+~;%ZgS zF&mfk;l8;_<&zjwfUjp#p^@U^XaB9WD3p?Z#0(m9!SjpU+$V^GBb54QG2L!gtIN!I zgz}EuYK<}`YtgH77(51iBB-P+XJJv~#hu{9sE$=az6=31T7fTD=O*65Bq~6VM_Z)H zu?yU>*mcgwT@vx<&3gTR7vaJwd)#qh>Nt0;9UqCS%C$;}!}5O~x1gm<{dB<2ePecO zcLL-yZ~uxe1taWtOWdc*#p`3f?Ceq+>NX_$S~8HP!_8(fw@I;RBMw+$NPwFbaY{}W zc)5-H*oNar)r)yps;#b)8!`EY7Nu?9C46}V4JE$(wI&7t0A^L zBtq!0?hHVyYt3wV(&r?)UOyrGY`qrsn00ZrbAjcrS;B`!p=c0D(XtXL{#`k12|6Nk zEo2AfvG!~GDGEz?wk^E+?Zh`)f~$#N)dCC$01UKenUs~>rWyy+bNnEuLb^MVjq}b& zgcxcJ+b>DP5qlxSL;`mh4?XFGs6KiUZ3lBDk0J9{Y@WJewZ^IW$-@PHfnVFADvnhz zgBLh~5VJY9ZUR5!LV&O!RpQD*S|Exs;G$%Pz z7ytuB4%i%)6yu!YIbDxv>39X>R>8=aW$&wTjSK24^wvUZTi{o(smndG&kaKAl@OSW z)6{`u+Z(@Z!LvD>94b&87hY<`o3iBDQ??p39iE!B| z63c9Bn5)A^6Ir3im16&SIt~=NF zB8TOzfHpAr(a9+SYUebnqN3&geY$Xj%V@5(cYPPWw(E6Wv<{+G{Fp$rWY;q=wp;C7NOUtpl&jdLR4PG{!i$R*kTmI$ZTV zWKQ8$$Yz|#1N*@c?7N9s3-%OdwujXbV}ywBYIG6?VkkT>IC9-)ez<*M71zmr0}(kD ze+jwI9ffgoruOMoTKVseLouzJ1?|w?d5IQP6CRyWy$aOq^@*l6=4GF#Fu&++k#&z{ zReiT*J4#M*Nt8KuicgH5O5v<#*b0Ii*73~wp1bpLBc?z#wMf$_S_3P)f5yp3$Jma-e(0oQBbTn!i_Gs zeu`a?lf7D-JkCxI+|@ti)(Ca&TV>AM;x1F`yVv!0#;eIKsE@_>v>it-!1JKCAfSc? zwG4TrU>SSM{k;M$l^by_v7plmGo_Z8*#)<&LU_yqVE2r4CBfecW_ZD%Ps7RE+fM8S znMecZ&TRArx5v42nJuc9kY}JYec#Xjt%#deHKD_WmcOoU zTV6_QLqCj!3{WgEr}3v@8isCO@kv}@O&lG0X4aVl(~b$Q<$VYYr5h&yS?a!XF?uRXBC;$x+7w5HKP&I#3R>fxggU#-+Dt= zAOVQI;oDzg<>m+Pt<`pya5R*vz1-FW8O{+sLnL!lS(U*(10-|P3?hGp zSxf(-mIxC`^wQvrbF5R1?fXkQVFpGL`KldF<7}7aLQ>2d*S9TKC6%?SMl|LX7@0&Y zuP^vb4>5ym7b%De+!P#c9Zjr|+blA0NAvq}h{MBAi0A8Q$A#}TH>p1k7zy1+icula zN^))-h=u|>M)3;rf;x-(ByPzHAjCJmM-4N{GS+;hGx>qI9)Q0md|ytk$|g(Tm#}%! zP%uu~rMUen2cU{& zQB!8`FnBnnV2hk?xxXRZIP}N6BPpI;rG_sV&5+@aImZ4jX5PFcMjbNh*l9S_2{~?6?yj7djwr8{gkU*OOVY`&dp$xn zv>SwYhruvvQzKFDyQ=VBgGzuv#n*F_J{)x_iRCIYv3Bg(-+OlH=oHsx-9lrMHqzDA z^v7tyX#Q#_U-H0l8)YL*qHPn+y4sB39nk1J7DYt4neJ0H{B{_qaY~w|c%?ZGzsHs%bG#7+8cR?-fbnO&MW-E1;;zbA zaKOHmW<=l3WI2!ug9lN}viKXTCeP2-o1RK8fE=8nK5D*F9~Q>dJJ^q@pU5bgY)3*$ zYhbVna}g0%l`_pBV2_BmAS2(0Qi4V^Co3XL2aMA6O+?~woc$hic7ihG_eE85{_fGG z)HOJ=5=6)JiaY6b2{(i2I=3lCaq?vmcf08+?g5=$dUtPr0@5wMt{S2Fi7#`4aM-QH za!(eo>FpIxB9PJ|r~fpu0Bbz5QFF$SRo*6RG$(EioWi~jTc3HF)jiq$p+kH0haMaA zlC_A)`<(SR_VK#vQFEIPwL^IJn|0CKAU+SdL65V=1it}?siV^AS&NZ9C(keXl#yCE z#2-QkiR#~%`nl8_@~C2U)!mn5~mA+Mr=?umkw!iQ(f65S%g~Vq~CZS3)#6D zqK+Dl_*(tMn6v@X4Uq$b?yN$#4cp0}R2*0aK_V>hSY`Y%6bFB1;rBl{}@!u|{&* z5G%l|^{obYr%vag<8-ebd`GK1@93zD!DuoLO6FQld{t2tkh{F3N!pv44$#H($I75a z*x?Y zE~{{{aJy|#ggx%#_v()zU(D~N7RxSC}bupw#hF&F{J&wV&ClZmq}k3^CaFV zeKx0}KR9L7f%bTM(VYuukTJSC9m#yybE<#j^?SG1d z^-}cY7PNl)z%I7zjbo{(4kFnN$QL{uR)lcMD~Z@96MVoDRHZ0|2$24!n1tbqmyqWG z=X#s>U_osxlljoge%+_Z3!&SgH`qfoetk04b~fkd4z59z(b*I~st|zAc`{Zw5!9`3 znAmf=LpVfUY(+Fzs90F9yw&2?>5>_<kyIBqzWYxvqpP44QtucE6w?(QfK@CidEv}2hd8uVQYy4H|r1|d*MM+m%U3;#omq6eGv#!QQ{KmHu z`yl>D?o{hvi!>y3+%jzgJAB+p!|@aa%=hw;{Zt)mbzexb=$7&7Y&L8Ml~U->5$wOk ztgfNxI}&EAjrdH*=r6FL8oOZ2Yo{s-rrw7Zm$Kv8hYXayGe4k^8nq^v5AP=%<+pwF zK-EIe6Fc_msdib=NAADlPLK$C6*f=RDSQ(})uQaSE^0m}u{<9z z0=r~~MBk;2_1qA#xKxO7d2cFtOIxZ!B*>7N;zeb=S1Hn&qPDw0xOh`|nQBx@#K|WD zzD9?*njHs0S!SoAo{XGUa^_}da!MFjQgyi20~W}t$DL&%j%m?8zJQznwpE#!V*X2; zOMBv~6PhPb&1hJ>@UxLZdl&fAn;)-4lNcgX6XRZqyfx*vKRSqE-lnoac}2`66Cmec z&gDiJtenL$so2_vIrq*pO=V*QgxGLh-oN3t+MhhGKRbwO(mPHe+xlNRz)6Ko8}wyP z8IDiJi9UNaOBzR@pVvF+VZzDhinmJtyq%s|yLY5*NrvOh5bilw|MO~fxj1}?vWEyJ zM5P8Kg)86QHe4vsd@nr|-Fyok&Sq-(euC{oPDH!lK$5k6ro5>yy{$b{=h+ZwZ#gN| zuKmsI_YrIfs%Et%S1c7BD{-6O$>S{c;v%yCf;5?3x*!GuoW=#oTlaW9dSNNqECNF#;kRm%VKELm%5oVN$Rk~&HxSs-5Gae!=*@aS`0ex-DAen)GYXPGreKTWu(kU?u7x8r!kZ7 zhCn!#6EN5G9K<3~I(q0)%{vlqa3bJ?HDaC&G?JTuLg zm~q_-ac^K_PUwZ0zf@_vDME$eGo~LK8mR_)p_jzfj?x} zr<=Td?Y(c7wTsb(Oq}@}BWxBnapur^!XfK+0lb%Ap_w3(F2Zm6SbTmZ=z2A~%}@~< zvHpnGcSkpT#fPPp=4#upX=v7FViRV~U^f z%k)4K)0t<|3i)k>>dj|6Z7k%}fhIrZ5bLWC=RD_|*vX;FQ;;Ml9xRksz_R& z?IH?FYXodZLA^f8f(wYbNOu2-w$5GD091dO7ej%7TX++6#IIaDwBj{gK2ZqS$7Q$jiw&a=$qR1O}0Duj9_&N8J>lA1>u5G5XW?4(Lg zvlf}IyFKA#O?(8#F=02lf|RO`ZR{w+6PSL`mp=R$YOFq|cK3$P5hGzoZ=d;u9+$25@_=p$B#UMmRRJ=0_zC3?Gh+li0nOeOvT=FQp2KVZ84v~rF4D$R`1@NK)c_OwmpHq!Dg#d34(op00YzpG5Z{$Rzr<%Oyj zLYd(!N2LrGhAnpBM+#HLD}8{R9V6+xH_!LdXB&8}L%3S6iGiWv-Hw;d95`I~Xcs^i zDoWy1jC0y&u8ljbSH(%uGo*-H$>3THdlE+9aq=nPKWG6{z5v+X#(yi(o$HAsR4|

+_4@hS@&N&|_FG{DSDTp>GV-hnGG+|%idvYZG6vTSlD z^-7Lm-BdfsXaV5=9qEuwyhwWnv#4HVtBgNgzDgWXa>bV?5ZA>NZ^P(MsqrR)%Ak*= z!_vi?IsNC|JeX3+lH28`3Q6wX^O8sc;wc{N>3etZIFv_ZI`aB1M%j4ryA3mveoW_f za0pp|m4@^sae;V$-9PYkdt6YUPRwi~uA1XxYCXkXrT>gAm%12IE7HKc*ql$^KJ1rx zMG-Tr43^hq-Ox#ePREP7DrQJ-kTUH-D|#|^$k5g$z}gWkS`SSfW9CGJu`p8&_1)NV zW~W*JGlB!Eeb0~SRvfRz*cIx&fohB*6ALxCD^TYI~nUOTO z0NEVW09A8kk2%9RZv#s_TZXJEjdyZ|S?DPkJSnqZ{9+Ybxs2$EB-!eACU{*!IX!I{ z0MhmLC*Nr*>souA8B^dF z3g&n*G1tmrj8hY0o%8U@d)ry%uqlv}OPVYEq0OSxN^p0kJTLgwAkZYcEs6Zll14>M zhSRf2ngg_gtXdC`NADmh%rwrLO3$OS>NcJa{sLtO!|InV%B1$^2!E)zMZhXzn8l?z z#PLhte{H8P#6|VmJXQ!gWqkR(&z&%^96So!!kQfkkJ*bDqvGo5hjrXRLga2=aX5IY zz6cAs5)7)-8?iAGYOXf-v%=F!$Q5(@8WvjT@&1OKl#z#nAGyuqQBQVVW*D2|w@6yb zaODa!^HDYq_=#1Ap*bA(_q!&b|ssQngtCD##__DdSa zA%k1^K{=5(yMld5uJP`@DzBYFbqzCw{iDwh#OUE`9IfK94?(8)6T>L~#^uS+d8@{$5ZJfYYR&Glube9Nt)p5c+O_+N>#|i7uBR+dbzj8*xX=qv7Dq@&hebfAt z(>!H3ZGeC6cR4!ro?W(FgqNdzf4!%?O#X~=Gg9e_WB+H3#_i$wWal0c`^$ku(JJE} zB0@Keb|}M9erUs8@k{vh&8%Pg0CslJh(}`iHle?uBLoz)qk&DmL)3+Wgv=ByIm>LP zDyhuu*9O&9iIcqw7V87Jmomdg-l5x0RN zz-Xen=yhZN>u@l-ZU|E9=H{87DkMZQuVj0U$Pno`Ncfpz!l2g>V+nBAPVU=*MOedx zTbJ|w8J`LI%D7w}c3NXn=Jw)2!4FOsO$9|Lkf%<|B+TgC8Dy-smN>c)Ff^X`(${=USE`;Ob}@Au zp+Uu{QJbU+aJgP37#&?<-=m*n^Q$oB zCirFXo31kx{2MmY3lih@eKK!8c0ocZM+TE!n=RnD>0#*D_;MCc(jfOs;=aFw!sn(a z@jkC|J+?;VXmBE7v*@CCGd#OblKvcH#{%1`XVxcjm|(}?md-htTDs^_lB;>Fmfk&z z5l9P2JC)5F=X@KdUE_{uN5!1rw?+D$n!WZ*e=~({_AaW4kl+jZN7newvV$T> ztu221`^LmYoo;2iS`=$RiOktrN?NO5`QydR{P@tH3k#`{HE#0@4?6{0cvLAJ-8!p~ zag$`EZ0+DKc@_8H+RD%UcYbpGQk@)#DhU!Uj*Avetb&Xc0`-)cReo2%kghQBQ!s9R zt!!^Rj6crHrsh>Xc>C2{cwda?$fDuKgDY6?Lq-JRW+%C=pz$TV?zOLsC*Wu#s)r`c zrJ_yh;|S^wl3DRB$Zcyqu*GM)5-xBaMtSi8X>bVZz zzJ|^It#i=|vYV&V+9M$qwo7sRdu=V3)oo!89x?UN6vy^>Z?8U^QO}VQj$e{R&Mu*) z@QOIDvHPMwq}##wwlp#@1zKhal_6c$K4m;2Z2e#ZM95S-8c{RZ(-%4W>1~oj;VG%b zMs>0&3?FH5K*`nUHXB6rRYSIK!#)@=-S1 zFL*}`zw=iv#{#HnZ1GaMWk{+L z71blRGW4JniT=4xJNn5GF+m0N6d{zCd?4n7nj&hO71qt$lQ3J;KsQaDnOxExt7Grs~7`57ea^j6j zP1SgBj`R=O%Yefc0~jvse18K0P#WeV4~QJJMi^Pmwcp#4jbQrRCgt4hvdd7KS=^tn zxj_yArxBi{dQZpz{srIi~`e?wo$NzYW?#3t5+*LDxq@O+jF8bG23=q5vnD-JGz`m625M> zw7MPX1P`$A36i!@<@QpnLGkB?&+Crhn>#68N`9znjV_Xt!3AIm1Xc#L4Rsianbzu#;g#IpQf?JEf34>#Vhom8{oPeOesIB_ zRQ#~Ytq{vd?whPJD@zmN6ODSrwqo~%IY1#-u%hoXi3$-%J&-HPH`fmn5Rsz}!;Z0z z8BZ_(%z4whzRe-K=urF*03AT$ziaAHajyVB(8vR9xO336QQ8$!UaDzv>MBuLRd}n;ezS0#3Dw%gO zG@Y0vwBEe~J;rdiUJ*a5JSHA&Tv?IBb$hn4~h;5o_=jQ1QYQucK)Mv!9EkB~I7dx&P1k95b!T91%s4VTQ5O*6)2Thbs( zwBK|6__D{AGJV9KIwUr40%MvPXk>RBKk3|eFe~NOI&qp+*S*tT>`oHg5?kDL!qI)? zC#yW7cy1&t_rm;RZ>R}u=UaIHVN^&PKW7)jeM^laAu(lbt6glKec1s`Z79A#LhH2& ze4F}ng%p$rr=ezPQl}Cckcs0jde?ca{-Ay<>Ud>EY3MrmZgTzw*SPoLJ6F%ROf ztg3+9I~0^#20#7a64oTRb9Y-T@6qm^aH0+8fBNJ=jv>Q!4=^ZDD( zBFS`f3$re(hr%^Q7M+bQ?^NvW!sbYW=}6}66I(d}h*6M^uiD&TY34{Bgqw{}1j84K zP;@6bhR7ou=J%Y|h2^|gt@;eLrML1JPbtBS4}J!I34QX-J|=Ey*dGY8j)!iB5aJic zv+|2{VGf2v4Y^D^=J-@Z+bNrYJ9Mb*vAab=d=OnB_PINo(DvEMw;^}jY|$#P4=uBr zuRx<%A@NYf5;~-zpWTQlJhDoM?vtCua6u;q%k~C*CS)iPP6Nk;S_%FzmU#D05*XD? z6sXK39;FS*cus!gWX7#bbVUI*jB7lX0G6|k{4@m!!D(VvF!58;0>v8fvSohpb)+fz zlT5?b(g8E_ZWo8GJ9#YoZ9^Ti^K`J!@wXi3->YmFLFJJtb!?V3eQ3u3Kium0Zm;ZD znmf$gQbh2p4O3xS4=qEV2!=nY%Jn(aE&TehO}Yh=ps^iZwJJxFqMtI@5P!0yx_*}T zXQ@ZR-|ap4W6-;OjeMM}4H$1h5Dw|?VW|H81a$nq02uI8+OyqUqpAN_>f&Kk0UvSBTKua7>-Qo!_gn z-)@kECBEbT=x!m<2ieOfvteF(3q z{T@|tblT6fgq0L1KBya#0>A0LXFT_Gd@XA`Zl~xc<8;kR)KOy?P}!5+7pVX~?o*2& zk|ug7br-c0IuON>^Q=yl7F1x*E^II~;tR29Jo|{xNhz@Ti9* zKk@c{GwyK&%;m@3aGl4k;#UUU_AvlIx8X0TFItaGcNrjvJTLv%XkbmKkwUzBzjfxm zOV?QAz6w4Mk?-TCju{_|XBz&$~x@Kr5Ts%%7UP{~H!8BQBPTq)!#E_;(VdnY;8n08=!F>T)`bZwe zu#186EUHx6h$AOnHzuF)d`=Y3f2pEY0HG$b)EE=RSfxT7f{$im9ZozQ8!j&yoAYe> zt1OAzQ$sjqe8tuVtwv-IR~!>N`WV!s0{sTWid9Y>BeUb)$A#ldMI`3#wsxWrhK*Tp z;!_GAclEzJ7vURu8=~I2JQay>(#b`uKYSmaZtU_{4c8uHXX!pPX|Ny}TSmikI#W__ z$ZU?-V&)2;ogYP|<;+2|nVfmH}Y3YaD)U=Q(&9MFMJFiVQ z5g)+=kGaaC5gZsg%>3G@{%j=D*o9JWhl44Z-dMdP^%R2)bI`nrXXquduB2Hu6|lLF zQvU5dNA3PODV3B;9l`b@oPIkpI1+1QGThrWTC9)YM}BZZv)|*$`^Ry+z>oF9WAPhU zsr=TKkBq)ezan*=(LGe=JniRYt%K=T^b9rCd%AWWt$c8>ahg`B-g>?@7ZWPftpzE~ zE+=23(#q@i*UM?)s; zf^BYeUkJG=NYncH;$5L)JZ=*=X^5a^)qRIzVesWfsY=T0WudY@_TF2`(I-sazaknO zw<&+j^(pdIF1IewGylZrXH$sHzMuO9C=vZQK@?{0P4pjIlm`!O;TY(dC3)(y5wLkW zPk%rd0uC9*TG32(r$4-WG5T0dJEMs6xXJu@r!3?db)ZH|V=0X>AS0?YKOCQjNNQIV zaV`t)VajTy^F9qRO?f-{ZOI6~;+rr%>f4hU=?pzAG-%1aUwBUR^o^Dl^S;(=!ukbk zrVI4+He05R`|||kG#sd-lG<4j=og`DBr&&;oda`(V-Fy+9_4s?SHe%SmzbS)os8f; zQ`YXntJ3!aXU){s4Jw04i9Gb7l&}Zwpd`$> zV)h#nRk!?@hQw4yHt>ZEpr7%|pa^}fj|DceHCY5!!L>(OLL`d2dC+?;VF6)*e(WK{JW2EL6Ti0}DE1`FUGCcn!YPt8F$>#SL&XLZf0 z9=mH*0fsqsR0iIX2WfM$e*i_p^tc za%9H}S{F8Qr%Z3(jBv>T@8px#x^?18)>lIHlwO$!QC`IbPrH2|_ zoQJTnCYPJn*y$#22|--XTuRc3dQ5%!EEi_%ch>LrQwjA$boyHa@7mqa9O9mw48%< zGiKsw!6A=RuT^?^wIoCOY@|PYua>)zN|xCyZ11Ss64Cq6Z=1eK_X2gSM$mAXy0Z*$ zAhQHz=rBnSwHa*9MnkRtV^+mx0`o+rC;I40i<=-^R3h?YQK2N6b~_rLWBDalm+m|w z?eqk&%X6s?$0BK<_!bLQ+87~J=$^Uo|6QU(5^g68k7*^bgFbVIr z7gG~-L9Gq0Qc$UEA8F3}qdqyh!#sLn_J>wi4N8T|jo`=fNV%2G`zGn2br8l>g|sX7 zMsJv40N(Hj-;@!n$ngxlL)=HEtoxDd-;)O(Wk$EqBk>5_&&M}HFQA_j&pd&C;x-5~rGykaJ)ubw7lHdJR+WHbG0y`hP2ZXd=EPVQkTRZ@F4w*fs4-l~Cg48KY96Va? zd^!zwHeT&=?gTw=CY`dz>)s9y6j_1`C|a_fWI0)3`SACY&!B0 z>d(BccW;_SFil;iSuM;~gcAH15y{yPaa^w_y_d5|WYnr4P1+r~}>b#n9b#r;)jAW5TV0T5^$gq;XTl-BZhq zIrqFNIwMfYyt6!VHUOP-3qKH}!KS^hkSsgd>H4q&1b)h4cPqh5!=|}`=tqyoGK=zK zgsp>^;NpXvIW~H9AYaP%Wje_&Q)Zcj@r1EUGAXPF>GB_}$8m2bEQ2&UkGiMWSM?!Z zV!m+}nwuutj-tKUU!avqU>nM=t;-%woGnWzCp5$WzVGkRPwOXfg}j87 zN`3uJQ3*-<4-ShynCs@<&9cQSC5?mTgc3GzZ3Gs%sT&@l7 zSbv3;H+sFwvw6k@j3z#E(%CZ^V5od&H@iEc8{@{K7u#iGZTQD=%Eo@qcgMP!QmtAD zjc>zucv9fV37%Q?b#^F34`XqQcOw@0XtHPoc}Lt=__4!7pRX4t-Q*eq{jds&GkQVO z!wX4}n1np%)R}^G68No`JyUe;67KL$wI`RMF}#Vp6Pv=KPtlE zwKR2`AaI;UR>Kvg;nz1m9~69KVgApd-Zk{Ng00pzP7v4)>Glg7hhabt3ltK6*WaFg zDduIwC)?F5rX43fyK1wDC=11EK8kG|nE4m?(k3L3(+Q3<#A6#3<(5;5cePt1A3EzL zI|Z8fb^4ZX{jrd|je!?Psw;5Tj!h8=vP+w>POrYq2UC4v`^_}V31Rni3gCG0E>gv|oS*J1 z0~of#&fW+=o#?R_Wn6}0GHd9GwZ`}NlDvd;twW;_Fv#RE5=R-8y~M@Nqbcu!+ZE~T z*1yS$F{Cqh`N4o4$4)(Adq}L~)>`vP8CByVi&A9+Fx<**#_A@>0Je~ZX)E<4katjT z$4SG=i>co+kPfej5Usw2)VSm30EPu^6LJ}#sUi2%vleU_Bi?e&apZ;ocO>7uK@vDZ z16H|q>6Q84jh`s$c#{`Ps4v#?=&-IhXg?0T%fg+lOo%Niw8NEDXvd`Zl1{$B!O-s2 z<};RijlbsK^T*+mYFVmJG+XO*ew2zqaTLPS$tJ!QkJy=0amFJb4*G%L&XwKM{k_!E zx7Ib_{MX0E&6$`Y7|pCd{p5wK?@Bhyj7RB~dlTA-i1ScEcxQO3#F^tkYwi?C)!^4l zIeQgVpnLk|AK2`Gky`O-uaMcq>#!fQXOF*Ai&Y~fs&tu0CtTQvq*PSV3J zC5JMI&9-@F+?EUaPGc1su6jB(W`V?hU?C~|Y0HvdUxOgDSL|u$D>~24+5U+;Jl*)7 zG)ey{aiRV>SENm*r3#8lbucuPQMZw3iq7lw*^!PE(e%k}={p!cM^j!sc7HbX1tw6}%7PltHl$MOR+!Vtzy{{}$%%=Ug7Q2)(8<)lhcjt)lGb^0EB-fuy3l!D+&g$xnUzhd^RMT`Az54AgQc^v zdPk>RHABj?Z}Jigu|^N0$rbSXE?7XH4eV14NpXAy9NKw%M;+g5lO>8OO4UUC2)Rh`3Bfbur_^__C z=S37;==#ZXC64w!QUgEawKdCuG%`4B3mV$U`dxa_g!Y0FapUqW!Hqyr$5f=IRH5x9 z=~hjh9Ryln0t778$4JL8e9q1@gpXB$^Gi+U}e#A4*DaA&^EIK$XS)ed;w@2fFI zUDrtscaaUg6hBDMoKDt0K;?&b>X2EvqWY5tO^JJ*-7Zuf6O-;UNG5bcppJxE>h@;} zihy?AaW21YJS}Z8h~02^6vKM8m~s;9Igfed`Cy*laRcbtfaRO_%N~;2)xEwi8kG8o z28F>;>nJz{k64%PqY|!+DtXo7=W`0}(pp0uAwx8uw^QLon8+}YUiG)`9D=PjLb2~t zjXBfdr_F4i(BFJOH^NDmW-dvt*j~lQh_cJVMT%@lg@sh;T`hYOsBh)$-Q;~}K>RGB zD1(v!lnjm#*}{Km+!nf9_oUIZ07fBXi~Gi^OU0=&rl+tGK}t^-l=xcto_phqI5V`o zuPQ6!iuFKhOfFs>KTl6uRiFS<3PJbuN_Nc40&@vJwo4!8$?ER6b5{+;q;tENW;=4l z0yCgL4fmc{9yjl|Yu@EeH&3)Fwn}ry;g$ug{_x4foBh+5< zT(fN}rldRnnE8N*;Chgf zE3J=Md23;%v?-o`Qlg*j_=Th^V!^~e96;e}bEn52mZor$S|d}BD6RMv7f+Zh z(t_%_FaXxE+Hb@~5J{84$P`R@)$68B0qJk)2KPc`Xl0|#ZU*Rz<#zSFANHfj?3>ka zTIglNuQf76}1uwH5 zaYtF`VFd4hLfST*Pq1ju+D2ce8e5Ry2+$T32(79CZc%EX;8k*1?+(O!x^sQnE6J^; zFc+_-)pyT6vK?H12Niof(ZrmKbJ;sTFQAqHOZB#?tGvaR)y1Z4TAp?F-U?Xar}l@Dby(nuU`T`OEY&FmPp9WyEEaZ zk`~rscF#dQWa7THMX6vKexQDWQR5qgTT2XgE5{+M2XAY#kh2@koJnd~}RR*k_ z=khVSFwtXt@SrGad zauUOy8N~uTO_ccFIBs~dB_o+J;VrOO@n&df1uhwAc!Tk?6`P zw-;VV6m)s=;AFDhc$52bvY;7Bmis*`y^UUJYVw#Y99u-y_wKU94 zgYPZ~V`#%`b|||ztKjyN6#bTs-^cUpt?G56l}R?<{c0&*H1iAd6Qkcy&~pYM-QM`T zZ0PXZK;j&~YYcJ;8-;T6q`7L;^#i?aLd|LCAecjmzuOYiRBd%89rR$OK%SNOOUSs+ zm7erq1I}zK;p6zxi}Bt^q9CIgn9?U7-8lEt7NcYB*mZnydE4TM^A(qvOTwiy*apLB z&SU9CKuyBLtvAV7EHd$`yp5Wu+u~%hLOnQJQ1$~4S^j+&6eh&#sFWZUCufsZ=OOm> zIR*Xud&0Gz+vkvK`a6=|MSt(jp0y*ql0?y?g2cliSd>jQ3?$g(MDJg<{KVccJ~in6 zC><50pOG>%+D3+-ZFW;JHr>MSq;eWo&JjB}O@kiwi-Ovei-+K8hIQIIQI~E@PoJdE zSmsj0j75r=V9VJ@zLBBVmZl|+7}E$t7#>zJ4j{mMNXI}E_$s%{1AU!0;%(vZ8Q^?8 zw|X_mKo$i(jDPx!O(_{W&UTwGuBVlx$5Y$rV6dRP;1$E`P2^!|wY211HyuGH%ola= z@LRL|OAA=&qf>D~@L>e|3*fYtrJ0aeSOxOeN;HTe;KF@4dcuy`CTqkiEW)Y>D8$ zmGoakLskutIWjORYA3A4gP*gr1C7cdMmBRC9&k+Q44UWMT?>(Kj4|9Rv*Fc>Ge-)9aY>v>(VEC2YZ5E$oB`p>^Pm z3k+Au+f~?iv=v{x5S5tRWYtO{tKzH=As#+r^GhFFfz||!o|G!cxzq(Xk9trHACj<> zqcocu>UJ{PM56DU!V*2tz0V(1U*&|we|V^o@4gdHNj$lCRFm?;67E>D_Ah@+ zAGsyQmIk^Ty~xOoNN&szJTB)k?36~|ul_b1|J6NOC7JOfWef;3Dkg{2=df|%VwS?L zQlgC$zt^#@Wj-Dbk@_EVCNur&m-?~l#&khLeYU7B5m#3BZ#{GfDQCXZ{ftBWojg&V z9)+S%Y8<|7y9viMjrdXCtTmPw>SKKM=6w;8FG?zj>D>>?IBajhRerTD8nf7yyQzno zQ<8OO{1TQ;FK`x(ySpJyrKUB33^e4IBeF0IL&d(>LR(av(o_>slPYtL@MxWxMrmwm ziu&BQxMxjR)ih+kVse_&I<;l9J*`K({W}&9?bEJu~=*;m~e)dcWW5C-Mox zkNWr;xi+>8K9u+`f6U(qx}<1WAwzW3%w@(|ncqCSsW-Pg>}!rLq0kUO{D8ULidlyH zla3cgug;KK!9!b2Ii%!;<~Z}0osO9i*3rR^&eqDpQ~e1c({sY7BQ*RrW6ga5{2RD| z+|Ipjv0Jwt(n9%9q%}EYHjK93I74+~bdzE`lSL+$l zCCK1>e}jT_R=}tlFULgoTXsAt*jX81`zsjrO?BE$n5|0}43scFx}Til>`J@nJD!44 zb{w8gGfb!xe#KQ!urSB|$Y#2O0OO(|1{jWPhx4u|DAH9HjGW!gyD(jw_xtch5JPd= zGVp>vtYgtoZ%@OZG;%RT`se8t@_NRKq(^843m{m7W`u@crj{QAVnQ5 zVJ^m?XPiWpWI3m-5gRvVg^?#xuK(TU#1%m)=r|7?uk(Ma6BB3r(l$-20RJ$>2&>Al zBgXYDFpU`wLD!YEB4MIDGA!S^@uqBu2YRwLVOqpx#dV8Q^?0cRdilY=hv_61D5EfT zR4#VPl&Onr5e|1^EeAJZ@z&@-EpX?rq+{S5!jK-p(CO^#G>YjC^XvC&HG7_$jI>J<;Ebb%{@ z)qdjQm3=CIbfyI)_h~P5_NQtN+CIdrd4TB5vPf@=(k*}KtV_Zi(AnW82dC5pmgCxx zJmVPit3p5?3^tQhl8V2KJ92WzX2YX!e|O8#zf1P;GV+WDul}88>go8oh=HdrbwNu{1U+}Tc{NC zPXds(mtkc^VvA~uw^{T_)2i9HQX`q2Wp0!D;Fp_JEt32Q?>*4CeskAc0qQ&JIoEm1 zaCH9IK_WJ=Tvu}&q5eiT60K)7VN-KaC%*@y>%z81{UD0`4YO+M=^$-S@ndjP(KG9j zyu4Y4MkJiLu7(VaRy^HuUT`iOs-YMpgkjH?MYgXTqrsP)5UL7=93;RyZ?xeQL^NB( zDvqCPz&-GXB5JM8zM)1x2YDu(dqk7;-QZ1I;OVqH8xDy4HGIM=BY`Ln6QN&z(x_CG z1$QZtF6c>Z+_Ao51Q1e@8j&QcQ0O*~L^L&wntzJAdcq?ZL%pV{4Xttp1l-mrpO#-C zB+|DusJ7T6zOABhTt&;5&`B*c4@HV!r5A)QFg$HJbrDe6&|zz@IzH%^+ddo2-XgEl z=I>G1yWoYkaE7tH0o2w5AFx1kty>iY?S(X~ONfdZWGqX&nrU_<&z4F7br$lzOqB|0 zF|_;ZpMAK}j7$ICplpK)SM{9iX!ZgfLdbgvQeY9UAw0Ua61r9P-(zXtF+JocaRl1D zEyl3Ss@`VE>L%=KUpAlcjx$ZFsm}I&4b@;kSqJ5v?U1-EDyS3IZ?wSnU3R)0Ef9j$ z<)HiiNee%)MH+H3D(K-l=SY?&yb9rO$_tSTx8%Bi5ui0yJLM9|avS`-(_j1k8 zvoU|PhmA+n+42T#a_UPpX$l^@Th{pNM~Y6t)2-Wa6B1CB&nEQw7^3gvkoXb2oFUES z37i&NtSn*|c3RyV<+nOYYP&a*fg-=e4Zf7^&z^84Y|Or$5ud@{S>1`d`U>G%I}f+h#+bX6^%BWNgGEP5 zw@Q;amAm(wJ>ztYUec_AdND_4`exM4|Ad$DBK@)8)5S4Kv*)Vyl0aMkL_&E-a>?U9 zm{-;w*~AzHiyu&dE;EmACQeofF|CL4ql7NF?Mkt7KWKP*kKJ<&63v00>o-@jBK(lu zoVC{X#sc8_=01~vs+Ar&>dV0oP==efPWymn%=dnIxO3g3Gd!}DEIb)~Rl}KCF;U!SS3H|* zFR6(w?__mhjsXg^tJB)sJW@4>W`Zo8rL^N8o4(gj@)EF zPn(rSzf~NSxjRAxr}m^aV!zX(fD?|3SZ2+~e!3e_krwgojrdaNg2H4p<*4dV=`_25ePN#DRhWvbw?sH9Ef*9u|nGcKRFmr^6Y z;MSTlMq;8;9#M-Q_xX*5MbADp3U%6cedyTI;exp1q20fNEksRXuF=A(-#Gb6ih{=;q&|)q9wsT-qgGy(yY5TTk z)L*G)%t9110(W(9;jSOc_@cFdAYHyU@FlR;vavty6|0MuneEY|S;W`s#@~M9`%L3C zIgyzn%Xho=8%NiL(rX;-yg&P_kj4vYwh&`T1bFC@*L-YQiv$n`AJaH;r0eY%ePc4Y z1x+~n1zlZ41);`yyZVG6B{k+B1k^Wj*<2%M<+oE2cJdpM-u`9dxoC;hWEXm@(xhkS zVsI5)D1IIA?kwmujkT}oaed*Jos!LR%sbZKr8^(-kn2K~=d!1A zQ8wydh+>KBCeGpMcTdoUU#tt?BP=16^7}W?^5xhDF#ZbIqG8h2N*8J>-1k2lWYK26 zDM)tVx8?fN;HB;@n>)?l)CSTnnXsaJDeEioBkA1DNqnv%*V0h3PXVl;-3DnJ(aS8g zC3-&L=FT$U-B*5gEjlr#celTaqVCOVkSK-_8hc@$UATH9d8-4~RiZAi+pxYUMej7l z4$^R@d`9SUo_g1)WCRz3At-MZF@}Gm%W~Ojx3b^XA(?8qJISZ19E=$G^3pJ<@8>su3MusNP55{^|!$8Vg?VZ{wc5p@#|_jk=hJyy!RL zOIci?Edk5MEn)HKFF(SYisU%O#^A|sOYt9q!ZF0yf=^=j0V@LqbIl-9sZEZrKoaIf zVN{e8>3sg4%S5{Vc&9v|TkziG@OO_>LkSITt%JarrNEEmrqMklvJxT^;|B5dY-S#} zt@=CZd_97Z{!0Phw9-+lyk<0I7_8cWlcSINawk*I*;nh3DY$qTEZi* z^hU9NY2-JA2T?yQ}BA z50;s^vSUZ6sgR$jO-Ie|tY%fu_gzE7l9W*!i}bhcZ&^}OKOD%|5ouf4S}bMKjig)E z%vmj%+7(W1@0hhG^d8JV;aA4BL*C8gk!^8y2GJYnXzoLqWy-0=?d{yCQ_|g&5WF2q)8=mp@LnC%w~6f-c@PP#!jFFzb|bK4sg*@~%^~`+V|13~LHNw+E^| zJ%5Lo=fJ_n2DOx18#vjAIK1o(fM>=0FtXKqH+Lisj zStS=HG!0tV0f#?T^-=s(S%#|f1yuFj-y1cgY#$ILjK@oL z)wj8y6T;fW%{z#6Hy?}%Q&(@7E{1UJq5z;dF-0(?Hr5dh59GjoqMT@6Iu)?oZ#L#T z$=0%@Y&o{*NxX~t#o8^GV&Oyw(_kesu!*m=wWXO-Zm-ro5*E$OY52AJ zBJ&trYbxpPlCAJ|PB?s7b0#tFOCwL+Yyzy<4cbmdd8wR}p#6 zy7q$lCQ7Fj&9NgpW@j5pnIRlX1f{R|)VL(LP_5;M@DVZfUbc;`8I1X#Y zB4Hb|E8hazY?6)@W1KiVW!bDNS3lPz6iU*&2ezcNtuvgroBikJmfXBI!w4MKkwr1T zGQd_wIZSS=jI}^uLzOp64q#4n{ZhQkyy3`|*1}suGIcNIU`2!qX(FgGrSp^+VH&d?pvuF`(Q!A6B7I~?=D0iah?Q4B>ff`?u1}LRDg(*AzAEw0@j!pX_H0TeN9yJqEyIE@ZNsKJcjg&*js^c>@H z94;RnAGYcD@@qWoL^M`=wNf~<0c~8NX7x1yY(Ao@H^95D#(i!lG1oF>paY<+QQJw{ zzLjb}8q*6@`!}GR1Rz4|g%R-dPwou|5vS-#+X0L8G%#&jb5&>o44Y;qFKVQ{jXmwD z`tCNt{jp{2WOlF-2kRwasP<5ai^+W>=Wac#jdw2|`?n@^zKZT{BxkPVCRhMFhZ@-1P*7M>BoC}A9%HaaF@-s*w$8%;kHi^1u4Nhxm)QtrtKQdM> z8w69kyVn>yLE)RN8tyuIVp8|tB%?BMT*jL6lTb1$oR^X8JWZ(%rY&)0#GkzD-7y8ejOq$8G0`^%X|J{%_(QfQC@H!2*;*+VuG&BOk z$o71t7k7h8a+7HEA6h8U8P!Gq@DWCm%vp-IgJ!DEjF?(dHS?_S@ib;EsC5P0B zzxOI6`R)D|Omh%BoYLFsR5U;v)IV_4NFJ4sOid|F+dbR3e!K3s`!1Z7^J^1Xtc68$D`&%khuEwFgHdX@%n5?A z=a;9;)JF(aw-eI)#|HW^H}}Jd<~%; z$WN~W``7-a@N#_L;7cfSIXVjEU#a_e9#%%GbVt{LvJi;<90Sv94{`6C5VC${j&u$% zV04+(Wt;7F;z{^!QG_u^Bb5LWD#OYDeY~njC?%m&P*g?)oTJc_s5WxNEcu3|6u+zG zUe3!4b$x6v^YM9*nsp`( zluqNWoqd)&x9;1q+{wz^F*)=joVRsnOh=jRfnm<)AMW}2qMg>^*M74=!_t-fQFPXS zgrAWDgg~Kxzia~+xuegb3M#8NhV^WN)V6QdIe&CG?MxZQlVR?m_JqatiTvp2&0&7P z8=ZM}4H>2(pQeN8ynoKk!e0BjC9h#2#7fkrC%%pX@MCQfapG_}Jfqtt``DdypM-@< zQb{ouclr7Q^tBmD=Q>E)li%SITAj^wbq@Fz!z)W=0tEZHbe#2|X!+IaNaWiItdx=t zG!UfPb4@&rqy+gHEsVulq{F$3|6^u6MtH#rcIj>_$S2;=)@RqIKn}Yw4vlJQ4NCwW zE=@tdGF6E@5ESM;smOiW7U!QxXT|GM38SrkU=782{gPwC+p8r}{%B6@+&Kw#0X|2J zdFH2nN&VRI0hUOu+7sQyDXXOnKO9icN|b9z8`|JI!3|jSmG*x>o)|(r>Kf*;{B>-t z#ky^0G(sQ9>mYBAFVd9SZaw3d`sJ^T)))>5d?}zI+_uEOktK@+ER8MqQ=C+pL}>!) zI)81W;@m!+t9+E@0S?t&jYlDvD1p&_6i`mnuoQngd=gp~H@vO(^6p1Ktu$?x;JCAD zF?QL!Cn;op!w}mf2TE7KNQEjHMU90%v_9hYfsiu~(-RZZp{#_9p-#NA{ffZ}cJPvj zilY)mj#HSlwAvk&>RfxN&%ff)Am{#;)&wKMxWNkp4Byu?ycmZ2KS$@W94Qb)!C%5N z92@Vwzu~?2eEp)mC{~JC4bxP;${YdReUQAPE+^ZM*8L)CVQd@X^rK4+l->@r*Qu4I zjH9Ts-g6pC^fFLIwc(c~O8&&Y}o^zyafT zl(1W(DJ~j_;G7@7{jFUb7IA&2x+jzV9F7*mVzN=P*^>;;7g{z&;gWEKgWv;rOur7z zsWV|s^>sAAtX`MDoGg)seZAwLFFq_neiXYA4*oluAv*;Lh6fRTFze8P&xZaQglA`} zS{%D;hRdC>(u)BkHO4-_PK8w{PIM0nnz$Mz#u}Rrpm}Vi9pe0&VoH2?5H~7KI zwgKhAO51bg49fV6cUunv`eacel*vB95NrF9ied5VzELhxbBF{B%o_198RS@}o~dgY z*7B8pXmW(q%*Om{eRO^Vw}ok`)M8FCiK+p#`mFZ9hZRd(*mnmk-mp(=fM z6d(E?SIUDm1m&K2l& zuPc%a$aMOt>=WzIQ?#=NV*AF9X$DVT1&JXBVMz@BX}luVYL+L>w88;Jc?!g}1>xf~ z|KiEGh#Om~EBlPXfNm8wgJWf@Kc4iH!^7y|z)jZtYO97@vB4f5>fWRd5z09zi)u*z zR_(zbf)nNzy#Jt8^mHIsB;x-)`qrTXxwIMk{T*y;BK8-;TC4yA+kcDTuL@GXO;>xJ z=^n+gUzp%?g}y8CG0J8)gF}LC(cfolwAdxTdkXNM;Y0Q#?58C?5!y~t%K4u%cnx6 zz`4QDi?eCUM4`JUfZv8+9pfOt!saU9%24X2W zs_7|@9WjBTAnxj^N{}pf`Ju4_eo9!0dEF$!2JQt&cv-KywOxbYdl02|aRQ@H&~$;* z8ki!`XXx)J!ruh z{{Isi!9r_sO%(R zGaUrY1H-QG{2FMj!Q|=ERabO@%PGM3*qQGsPaz1Jr|@xHZ8`FEc-oaBC|jo0z}FQs zt<;SA9aII_QxzP0&4T1DuKr8F?S|*9b=dEBKSE{JD>WgLi5m#!o@{Lw#Rc0fR0?+9 z4U6CG^Em<>5^X~>|iBJgk(J;_S zi4>_R^>@W@#JH$gC!LDQD4Us@Rdh?M{v<2oz%=tTml)lBNuYMF6pMBo^Wj7T`8S6* z15szk&bifg`8}Fxss=bMVvHYYR|V&WhOP*N19G%0-@E;P7{Fz(v9b@ zW|!;2{V@JA#lKPU?m&gyHN?hI!VD@u4CVO-Qo?Q7ldex8`erTQFcCaIP2b3ZjEZf_ z(Z|U?^3k0BsdSI}=GmIuz6Lo7&~vXxZq3Rpw02s?HMy!{w3nDy7ma1HR2{|=gDBoC z97T+hOB1e1^d{EZnwC^k2PbG`UOh=JCcTi+sy^-}V(uQjJ;G3m!E8bD{(DPO(OmYd zRwZEa`jlKyom3v(9r;6>U>BRlQXMf6hKac%eT3S3&+oJ-?|pf$U!$eJJzt0`rZ(J` zZ`(PwdzB5_&+5=lcsqK-DsA+S1NK}$N7;3A#U~kJ-SV0(^;yV-B@T?=5zAE`US;Jo zTVGPqxT)(|#9sX}7!7u1JUu#d?F*6!_?sF+Qa_fX z>6SA;;Q~yh{~5=HQY|J~Kps8$#~#My!wf8is{UbPcm5RP_v1GEsKBG8J)}E$UO#?; z>+KbtA=*77mN#u%Gh5)MsIZ+(&XG|fa8O8u2$Cuuj@iP4X=R|*fnqP z>r%>F3qgEPs;p&kU6?G!OpFxY%vG4fuMBJ(jC3K$W&L+u z9fdUdC)@ax_Q1#a(N#1?azwU`-X zF6O|EG|Cy*UDF=g>Q)q&Y1ihfV%n(ARF1Fjom*{kf;cbl(q zA#DIfK)S!NVTcy8)9(TD1x63?lN+fp*omqwxEfrRup-h#i@_XwVs{RCxzw(K+nu^a z2kL^$rut>#(1i!Z_Of47H8dFqiZ1!5g)X`fXBkhH!#?{-%%^YU*M5Y!>)t|El;S|PR@C5tQWh{PrXwj~A=$$3 zx1M;yyablc5F$SbpwVGMuX+TUD?`rF!2TW=fVU+9i3dUjAKY@fms!$}!7(8_cA{0d zLXlf{pC9-j6=my62I%8nxuV}4W)jW!AGmLm@;4LL-v8~tWlp;lV;Cd% zH^LdYfxWL+X3Na)S|Bjpi$BL+ejuh}w5Xb@Lg%^Uy71y~j74vVg=mWoWnbDy<0)*L zjNifIc^$rUGYG-Qu$su8DG#+xLx}9L7_OmI*6fp%5!I(GS31RFf#TB&N%k%@yeaMF zJ)cddthD3JF!J4S6on;YtPI#ddzF!%=1_J@`8#^*^j*s9n)-S!)=C;_Q?bz(1CZl|iw4HO zp>i^-;1Oe9bl8#kkh115EI}ysdNEm{i-aV^gfb!Vz0_YRr{&y2#UPrx z4SyOoNf)8<2b7r-c_+ioa>lO0gbOJMbw-Kz{Beg$~q$rBf;c z7pl1ED1w!(xp82tRj}q!pr?q^-q9(DlDxtW%YNSDw_GM?!SA)?yhqtcz|X^Zi#42e z!zF-Eka4x&?)j6W{2T40heA}t#xmLroEuO==8cN7>||0=E^#mi=aQlaZ4MOvz=?1s zbIw>zr^q4GUA>Q(PxEu{^E*CaE2I;K=%PWK*?Ys;_L3DkwlI+~G4FugP?(thC+&X^ zl0;0_s7>#h@}~dwVBwFSj@u;V=G_N4tvM$l0>*IRHYVCXg3l^;Lq{Z1Vh|aRWV3_S zy!|*I18Wv_M2JM}IHHL@uXr&!c~pvVy&=QD4-)76b0>#qJVZ81Mp!lJRx$dIr#AOD z0O#|H3hRx_Z%{HahK{^qGvGII->QM8zDhU|yGIL$>Zyo>?51QFg^Mfo48qvzFx}{? zCGaOvlHZ+y#ihCx*fe2tJndh=M_c@m*BS%%r*CWp3!^^KUk9!N%2h7 zHCzS8_cq;o>*jYwe(}~F%c+XquK!>rSh3D|YARGDJhJB4&DK{5JvijiGvgV7k#J`7 zQiurgfn~Vrll|pXjrsN;pJcInJ65G{F0%@oi^uQiblW$=kSG%Ti#ga*O= zecfC6zsy6eZ-=l>5_*!YGZ^qT)}6lGf&202_On`x+lky-?j;Ez4i)na z617Jqd55}y))S^`0ZjAIO?t;$9e>H6Kk7 z!mAT2VO&0Alg)o4x-b~zTj)4yM$31AqVEaCO!zpoWz|MLf_K+KeJ*dq*FjwUsA=jm zo$ZJWcB)|^h4$1I9_vZs-gqA_^DRt+qenQP*e@`=I>DOZkpkvirDWx7qQP;=7JiMV zx~FnzX5DN7R!<|Qv$vZ-ei8`N#lvJ_gdTNhC$dnch7$%@3w#H#sB}Txt@klAudFur z5!8t57UUVKAG|ikQG}a*#PvQym&>*HaK2<+?@G=Ll8OKf^ zU5lu=$22xO;}X^AQl}0TeKCg>hc%d_utx$w00}Jvay(7VDdhmWo{|#t(i4#V-tl)Z zr=DT&y9m%duwd;T%JF$qu1jZw`5|uq8RlgW)?MdAGqBGdSK!8#R>5w2S+*>S%tsOC zuH7LK(naHHR^^}$tlkJ{^&>`7KR=%*NacEme`}S0l7JRdQoo8HTEY%mJ=xr*Pr(Ge z>KLYBh=0$Jy2P&aKV_JS2a@s^M^X*eXv4g4Y>gP! zK`PLn!)iipBKNKo?x#MM6xez?maB_XD)UDH{-7~DNur7TXBa8KajD%1LH01gH#>^_8wB06A&;4;1A6wxmBq8dy7Ni zq?Y$pO=90KCzE46jK4lIV(zIi+gf`5yfi_`?(@_kq6$SSD_cA$J@}G(oalLjnJI+k z>2df*2UVD-{TBkfv3sRY&4dSOkA9#Ku_V$W>4w++_TN#Met%HHO64gsK)bfA+CY*{ zwpU&sd0yZ%KuZ?kqOv1cd4cBm(U)%T8@z>bK~UVy3@&dCklY7D+8cX=I;3Qm%+WYk z!3MWc>sVO#QyJx0dHe0nesY`mW%mLAduk!Wmvg6lA`s2?uFt6uw6oA)5_%)~?U=T^ zLQ{l@b5<=WT(B2+FlK5P+}YL{tf@La6m`6jI5r z^e7@c3bv{*22IJ*Y%RQyEWQ`x7&}|{!>Yfla;9yLeX$GNR~Zp3S><1wT20Aq3({}i zOmI?r()&Dkrue~PPL+V027AxA4$7dHWRWGvmK)X5OM2UKR<8{eepIpP=@jo>pZFn3 zpjvIgydz5z+)+zc?j7y20V(!1pU^&qkKAuOm|xd?X|^tB2#7i@XAlJTt|l}gF~WI&j)rA#evbS)?LxOPw&Z1^KY$s{>qa(_}&FoMDbDZoz&~{p_d~|eoiJ9 z(`M_w2#$)>(G2cTf-2Sqz@X97aeUm^XR4`a&j3g8eWIYd6?Lt0mboyram;VuY7GP0 zCuY=^Mw?xcxP4DF=8aTO8;VQID+L2#yerK0HX@KL780#{Y}9#x4x=3crGb$B&U$yz zuwiXWtyrfYGPgy-)2BDRQ$=oS*kDr}|Mxyax90qs;EA{>)q?$snALUfHE?n7TJTbX zC3w|n_t3L8<^(qSsN|h)O^j9C%2DrQp|_9kv9~lJdh2^EN~q^#!ij$gIuqM3({!Y8 z)Ro?|K&Vm?y$^Q#;>=?xLu@Qc*4H?+2m1GvjvhJ z3?ka^rD{mpAX6ftq*yfXE2Y0zh#)8!A->p#AAK*cWDhEq&S|g?2O{u+B306=w&kKd zo9QeB-g`L6L1vtkAeY~7ah;|Y)>6lmCq`#yVk%|C(96HGx{N#AH&id{@WC1_8j~Cg zP;$kH`zhFqelG-4&C7@ufL<~aHC*P%Ei>DWE+z0PE3SKGdOH)!Tr#E?t8iUypE?|V zUei3u9hzx8SYMcX+^X4;8i(~A4Bo-4M1$lV^;2ox_rJ*7fW4?`Cnap$pD=(+*9}bJ z2A&BE7R|^;qvp+tl;Iw#!lOarOxuEI6n%tbTfSXA^mC7hZW6?7CX@!}&aK$m0o9fH zu2Cixuhk8Pl%z809|LWXrP)fhcZ%^WeU2Ox6BaFb+(g+_Q{>O_ASS-CEsO}?xZERL z`!#z`A?g-e83K)mn%_S>l$ZF1X2qq>(k4nXN!>;desk`L;Q|z?*j9ZPwEUg4VOfex zAzgc#{8M7oT>6>5Z4J5<7l$7I9k9Iw@8t_~&N!7sC}^^oiR@Osz~89q{~WDW$>$3Y z%uU&o5N?Dnp0TV}`|)T~%5DxJ6I~E~KHF`?+=6j1&epi&%#8C0g=q-US=`+xW$ab6 z5h^yv>R%7>2p$Cb;ZPy8xP3~`S`z+kr5e~Hs@%a-jf`C-OR4J<4AKrLE*C2bGg2o zhonx>FGA$eU01Z$BRZoaD*^|MSTN|I5RZ~S{O)L1Ch83(6>0zk_XAu@kuuVW`Yw&A zPUTwtG76*=vYC+fo}gHzQ8?DA(wTlSCFxoAtrlW{t=o6r)=R2j+yop+=cYukQxyL0 z0KN~0#z~qW41%c~sMh^6N;RrXSZTDTl~#N!-M1Nqi(4hlV^2!cK7{QIND7WMDO-cO zDgSn*sBt9pxjnk4^sM6kHY@cfVq4t7j$<~LK)?mB3=eaUok?}pT4{bmrMriAj{xvR zFH_^wnxf_N_O~b4^V^UgCx#lG#fF|@T;nY@CsATp`e?m8r)d53keFbPTD>&#to9fu zQjZ7fIJUK)tFnE&tr+`RyBU=-YRnkK%q6uHQ$HOiL?v*v4VldRx3f423G76s%5~Ec z-yDxnG46<2sPe5K$!s`RuAjr6bpAlUvQTx|!lNGSb38)wlts6z-}=)!479Y7U67y& z_gQhu&#`J#3|(jWs+LW;g(vd)X|1;~WU+Ft=h(v^Vwa+!T(+U)3lOH5Fj1r-UG6)~ zTQYpU>{?y)&8)s#NKE1prbn6QqWjA^q!BRK)kAWm;REXAF>6n4AL^CM30?G;dAT+!!Etz@cGwWl)k7)&L>+=qy$}tU3(m9S?1YK!p?0&g zaW~u_J@0jIv6(Tw2^z`8jo+3HuvT$H`7v)@!fUf-*kh>I2!CU?5JUZ%M;$7KXgX4= zfB37;IK*int#I5*q2>yv;!q`cy-qCmsP6I0FGw5l2}f*3uQK`TT<$;v1r}O$2nYlX zd70sEKK2z2_dV1to@Z&gi?15L%L-{GOimkULRO zWQUSlXGc2SB2~2hh zHe|8h5H5w7FHeuFYm6io<{v?lwoVI^pH)0Tp>G`*ri&YQc{a{m3XoYvMjs00H1lqs zi1{k^jce8maWgGahkRt#Um|-+_IK&Z{@dYP8VoM-VC|3 zrj+NFd9@@$0&#A@6*Z!(R2Ok|6;_axc)u)TQ9eTuuG-L}I}4C!L;H0BTF8RwIKf$& zsh$engCufSvF16}3b)h}o8z#2@B@oC9m~wyaj43&@nR%?r6i?oP`6dV zJN+`^*SAwaT2vYET_0VX%H0d?EsGp%$Xu(y^7Rd>YX}4kli_{s;Lk%!zmW4qmfjDo zPo3?Ag!>uC+NypLqoi%oV|<)fo99l1)JIFmY9X7*CJsG(?{%YmuQX+PpVaIrf}C3r zso~v_my%7nut9p7a)hak?ZwHb+LC$<+nFKaV)DV`ok z3|m1$HKM{TGIQVZ49c<14>nBDv5k`kS{lBw;gZtlnAH)fh4l!^q*f#5Nib!hZ7yTBrevL&G%}v544QwhF^$f`XK2W zsDYMmtY8sY9x|UU`FDAuWY({_qpUz@w2jSphoP^`M|v)~@wO4UrNudGwSxC({7tmN zM5d?Oi8R*INvpc>OC#TW2*09KZmFUuDa`De$dA=TNgn6yU!fVMp}~eWi`Z%CA32Nn z?SF6|6i_uH89Rg7chLM9r10PngXSYPxbei;50VrhG!H)=@a_VKp`1#vyt9|q5^A(7 zlXJGyrDxn4eA6*jX@_TLw_DD{V(F|Il~WBJjtNv_@y4=FSr5AYHob?)P23dRBUB|x zXx*;+uxIA#IRib-TsM)UIK3KmwKQ-tg*xv1Ip170wv;TEnylP+O5Sud!OvZw*0z>Z z>o^dn-cQ?*EFq~EE*fdhM`L<`4#Hm1P%EJx6@i#|16V>)tt9NN5dkNg@s)!nncNfGEih?2(r83OYgaI@EFM8H>cZ9B}tk;rSJg z_OOuW9?BQFLNtxjxz!P?x#fMB7C&P`!{VCk7T$@$gKM4E&P9=p$eEXQat3i^rB6}E zdAcjb033U>FWtT05OXoTJRvJLQtU!xdL_x=kUmAlffMvSBY$zv#A4Ck?SPVWYCKnAs3>D*#z00 z<`|K?^u!T)k+4sd<)u<8t+I0_lGAq@B$kTjIrNIC2{yxhPa!gvI6s_(_7}r=Z6D} zk3gko>ZCXt7iX|F=NbB`!=YdzT8|P8_d8!nwD7b9M?{)>P8_;xe}jbhpv$)={t?lBVvAXiceI<_9y#AYoyT(TKBAN= zua~%^{Mw7|R1n!fs_x2ud^=A%+$**=KB5Qd;eHBA;X%ExSDMctc;!nmrI=Nfw;(X) z3WSDaF-5(22A1FRLdn!sS6+SQAnf`q7;>z!Gd2|hCc*o;A^ZWN7swTaxq}D=p*537 zXHzDgg+46*?X>|A)%$%}B00dO4Q>3gKYyYu{$fa$J_}sD#F@%@r`mhZ3 znSVjO(SYvrl*dr%LM1;pLL3+cjW|()JAq{UNSc(*>XROIRH!sZ2=j+1^bw=yrHEoQ zon+hxIr*)T!jL#4#;9+2oa#iM=Knp&wCLY`i0_KFObY9VTb1k2SQEqzuaGa2^7B(R za4X#DjO<*TvWNU-z8*PhJG2SE#1cwWGibdhuFW}@pCqxYyDZ0`|u18e$BKj?d zA&O(C*G0jTrii&Di9;6S3x_!^#YLE*`wIpVLUFtQ+k1D7Q%jFpc<|#q8_4m^!|ESlwaH(DmEN#VkhL!h2g`<& zlGd#mHdfmnXXx_VLMhs-sr5+IcCo0mA|~1yEK|T6l(iMUmss^+=1z@Q!(zsW;$bOu zIOD+D8QWx>V`Z@16pE7lsAcc5UrLnh-kK8X?(?73u{B3fj zcINFl8H4F-Y~PB|tJ+?9&(xsHcT1Ymjec_kHT0Z$u4MLTYc=_Gg4T~`TmyPl+JXiW z7hfHq79{ah!`KhA!bajPlg1xk$U5QbYY-8HxZB~}*sg|JC^UCl6#uxGlC8hsqCN*d zt_GgHjV9eWwj#-{P>kM5KwVYln+%t&OHv62QP=j)Q905FUG%0jQ1)mjRj0w2a3?Gj z^V%>DY!8MYXkD4#+gE9864&V0d))K z`w1$fUVUoOy0;S;Bwx;nocFt&f>C?muX!!S=x`JpNmNq_{Rlqt*9<5#hB~XeaJ}kV zAXHtKAHT+?udx`TSuw#Rwve`6W6g|$JCKQ@vie2z8S!R{xCM)HwmVP8pWFnf;Zu%y zj@2hHg!OX|IEkjAiFEwjlB+5ft(E)9P<>Ur-yV_kZ@yHOG5UgPi>0Z4;r50?1!n*` zr)N+#zi#d3(p9PXyCQIbb905+S{&)Yg(#-5vq={37L9JtQ_sXI^P$+#j0{H`QkF35H3Sv4b=Syap~jCkCO&eg9?3CJ z&A`6r_isCMCWT2C*6K7yiB6+uJ)e;uyI4r{#K6((G^qmlo+TnnsA78S;-HgdS%le61Q2spEJG6w# z&&GP`bGHf#YT88tn@;dd#4LRGL)jp_!n~1pZ8H zp`w!vpg{E>wJ_KCHr`g>bd8PoVI5Jb*#<c0m9cJc!;!}Y zfz(9WY-o4oz=LHOIie;+tw@2}_x#F`WCB!Lpntlz~~l-x@reL>Z2JH zsf$Tu8Vi1DR?m8}@F7(&(8gs7Dmlw}5D4bU0`?17Bz;q&l}>ud?HefOJ00jPOwrW9 zEMc+XxUPddQyM0u;XNo<6svf3{adj#Sp!N5a##B(p(|+GRS#Jk&F(n${SJTD z;|3m(k1G%GWD0~&=y-0ol2%~|7ddn~L?L%aIDMt*$+W=Sn?|u77n)0q!%<@?X6{>_ ztV7U2-`4FZd$(y%M5%49(()OP$m4@?H0tQj=X=odw8jR;_p*HZ@73iY)UMY^ey3_m zzXV^Zm0lLsS2edLE95v$$(mA==Tt-R?5Nr;&WXT7t(e$1igcyNr|oRbqP(-hr?TV@ zEEAmwxuzQhzN08N2SE~J)03RB5mO>Zz1vD@fh=wHqGA^}VqaG*^lREq*BAV3Fu6oE zY`IJgqt--=thpbmShMDd6Y4EEjdXDBWih+O78^n0NGNJ)M94|-UxcK>*~31(C_U*_ zP6TVeXCxQ*3f1))nieJOkMfoV8xo*EhLku&mVf+Ms29VB47i3qioTV}yi0}}Ni?V! z&&!LaLEVSx?10=tp$!wIMM#`kGTqD9r;P({1lTdX<9^4UR%9U2%TEn#S_hp@xA)J36GJL7gR%z1>U#dcWm z3+2mr&M1|n{9WPSH)5{dX)JZVrt4j6q2CtWwpacz!@+hAV0NvapZ~9Pkx$I_6@SO0tir~BOZ{|d z%aO^*;B4b=9zO;lDE!h_3GjE9C2>e}YzV&e+9tVVz;X{%JAN;iFz0kMeGZ@YXOPbr zeo>&iQ@$EiS`*o)sR|dZFTBEQ5(Vm4gR#6_+>a`?I?BWWfzb&>;Ry`UOjR063H~S3 zs#2fV3VAAj8ENd=f%?^Hb+=caZlJ16LV7d=Q0%XM8+w+|#7!2@-brhS3#~w1qIyMp z@x>d(XGdq3xwZ^!SD! zF-gi5!qZDc71mF6bAdClTt8J_>X+*NgoaXQwzqPzi{aQ6NdK4F!$UALMszzB~k_bIYAde4IO-qlBM%R(2_^{)&%0of`{J za6ERnA$i712(J;81E*+eMpM~EN(h(!3dUKOlEN^KSe#QTEcs()QN#t%VCe)Tk`ZLL z<*06f7OD}5L7*H*viyBnKdz(u36o8_@^CLr-Qeyyj=Sv364Z$hoX1kdNTP zC8?hF3bBiXJ((8%A-7Dw;lM~oOIgO@zhxx1$6{^sryt?ajUGkO8@n-=M`z`i{E-8f ze#MsroU%^kI?^t0s&R`tcx(ho>L^HA(dy$Z=T&d$A~j9DEQNl*V3W3?)ZgdFSBrLE z-C^^MWKsr+1XB-@6Q=PISDv=iv^%kB%F}MIo%m8$=v-ZR2i*o;S%g03zU(oNPBE60 zx19Tc(q;KL)P1laksto*`Y^kt=`(na3{xu8kEJY_0CU!6F}ON|cbtnl&_rb6jM@~5 zmE^&7wmbaqY2Vnq-7O}>1d8kBoBYLXq7yDZ3wbl9*?!fqwARc^Sd^$eFYfY68rOnA zQv0jU4ezNjW5+OVU48i0PCk<71;P2LEwK*w^2X}0IpmM(c{ViqP#PYgz)Sgq3hKP~ zD1KsTu~tCorvx<_&l%i}Ge4Wh*Jl^a#4K&rZD21##ADL5#QvS4pPLqGv&MS>OD9zO z787?H03;fDaPM-i$pQ3JaW9;!yHBe#*fp*`LgsVOx>gZONy$HTXO#y+}b z!8JA`l zi#>?1UT?0)OC=CA^)*87SZ^mFWip^b@f32Zcy3u~6AL ziS44F?ZX;(lSg;mb9-UY+R$anFaLfGsllv1<^wjtJzCcxdTn2Sy43o^tzkn`D$ba1 zR_dMdz9_-xe%pg*O$@QEY{#-FGOnd{xRi@fWhDsApXhbDaMc*BUAvP>-+MdbdmjUp zX%dGPrfEzi*n0#~U%sw?AY6P_>$cm{)Ui7H#f+g^6DH(N zcg>abs7e|Pb0=<%*JK|05T0#(xVqm-UR3q^&UsFDdNd=CcD$Mel5MmZ&D(}nN99&Z z4!Lopsfn>{bk1B+m`w2slV)=}ef8+G3KGIqBisT~#rf4XAYgX##ffYVIw*0I?&Fld zy2zQIAhyao2Im^a-MgGDr@bp>tBo^FNCQXraZ8NHr75^MEx~O!B1fb%C$$9_Fm1+9 zfVk*Sl$N+qX&Y45t1#L!tHfi$G=gG&p!(BP;m;t6!A-Vfpv8+_bmfXm-WQQgRyDyrdoUyUde|n%SmVQtz_)QZFy@%cQOE zkTR)ET6uS^68tA`5;Df@Gas*Bc~#=#X3iGBj6=S-<#otVZm4b2YH_mf4Uq!!b!0U@ zD0@esc(+8AnSM2XU`<=kx+c`DHAz@W{B_2kJh=DMSy;1UG?ErjpQpv|qZGGi3!K)$ zwR5RO&K2Ty;|je%0GX&_{G|EJ=1V51O#H@Gvy^p0$g%RcwXmHMI=fP_+x_!d(G^NHqXk*3}HBl}sE~rqoWGV2E&Fs@fh{lFSUp*sCX? z4D$=ePo4D8E3(KqP!gJ6-$9O7LqkY^HHe~(niRBv)MhrVBbBTVRVs0xo!1I&nvCrV z!O5>lePlisD~2`X{LxBPe@M_YajmWj4{6q8pg$EM;Ye_Qv0h39YV5Io355SzI`2Oh z{wQJh6+lS9O3xv0THn`Uh&M+6%205~0<}*Z9#ZYW_BfQsLF}bmTkak~-B8NQz^Q}p z+%Jr8YplAQwdM*i$mYQsC(~5jQLGb=#sd_eJs4;-9Pt$;>_SEv`sR*&E*sh`N8dO4 z3AI~4O;eS=Rx{58^)f9#+Y9>T=UTnGeuoo#|8TawSN@!QgJeTJvicc*=1(FM7VLCvn@YV2-m*)c2PIF(N}LmM-(7`P?ku8|MDawe)-O^;sd=Q-x!_6ej;=k5q6r>&)n5jV$A? z{-TDBrXaE`S57rOFuTa(TOgAvh=F&q=quawoeR)tN;LtB-CLL6T6Sc!u}3M(PJ=+?O)X%Y5b$bvtJQfm_aCPs#AvIN(7?X>)j99+oMn5 zUPzZP4P;E%7o{uKxzzweqFGVB=#d9M8R%=bV|mK$Jsi&kcHO^xvmr*v!l!?$PzAb* zGxt8bc5)q|66I@4VX;BJPuMlTWjbp+oCM1I;BM8Oo&w!tTWL=gpxyw6P>Y!EFRA^T@1$VcqF);t>c-PDq}8QNr04Zda9KF*m(^yL!^`--#6G zs*iSKbn2H;7t?pNpNwhWAdnnHp1l#iU8AisKNyq_@xGx) z4gJj-goD-HUm}Ylu$)HOL_;h&Ws{5=1vHDW27#^-F6}2Ai1grxl3=X@N?+I)6lh}X z14X~(5AtODM?}f`z`Sm!KAF_kVO5#I(u!q=IgBB?Lh@8Z?wM#8-mPm0dtw^YJ;6+6 z9L*J;AX&S^>2Ldp3^*j|{@5V#HCir{0 z)zH4d+C&iQKVRR5M{Z5{FU+*b2~Z(^`#L{O!rLIdI=Rt!K`#2Pcea*)qgfz6@Md(G zo46j#ADZy@=q2SVrwNnq?btR$fW?@j%leq*Eydd7XQ1T7W@6L_rTpvT@<<|lZVNH8 z@4{3vqHu!pVw6H?zr~#qz3Iz9Q92U2f8zUEke<4Ou|CL}%;|QJq{p2zHTz>wWNh@y zu3`M9beJe#-vvNT%!X!uDTJo9S*2>aSU&?vf}i3{8ALzW;U)$!OM;_bRi4w5Ez2rX z!c0K;wM3sFM+bQg(C7H2+uL+W*UMCxX~?a`dPbFH($rz0U41_uu248 zM@cXYaXnL!;-zRWwsu`FLvJ+>+3^zt=!natt{!-#{1rnB>-8*>C+>_kxiRO!iJdVr z>PJt#vLW50PEO%EW&q{9Szy>NkeSm~|L$R&2ULsE-Aw2z4syxn#@`GNb~M0GBjL!x zoFpKuuPY;eX zL;Ecxw#7!*I(Qz!8r*@n8iDV=W!703dVsAAZfP6?1J!*-ZBl4J;DyoVfr$rgD{8nf zI*wg(lQba?R1l&j*0{mvE5a|4g)D`_Equ(r!+zq@u4BfB%dekt^cR~qa^qn)aJ6vf zYq%j$GVc|hlL@}YFH}-H=YFXr@IMk!S$!+r?V))74y9fG!OTk)_`B38u}*<2dTfSa z*{~b_pp!^L>QZ-5WOmJfF{4i##tkF(a8P7Yfum8&!|SZamN>6ntV8J0MiKK@3vz@# zWffC;((+mLc35|cqc57VBmXkEvTW8}0)yftS#*$_0jKv>aW3d1LkVsQIa&)Fp^HZI z3n)~~Xe&G%cJ^g%ewB91R^^P$_zTJ57|4aZG-&Y}p47RJD{+Ir!VsfY4V0R^<*m7T zwAZ^h#YUnXkcC9GOrkTw8L?8}R7GAr#?FUY@|1R_N{Bx;sIdF-kU5&s*;cf65RChw zz7fY@LJ&KpaNMe^Q?73_ThaGE`-KXuR*7HtPMV4t>}zp!@|s@>I|Vqw7E&36oq-9Z z`n$J*d+w;7qMAxY+(rxX#@^cZn_Ru=ly%;?H0yJ{BVKAyUeHpYdr-fs?O$(U7=73Z zq}aNEpJk6OeyBz$Gw_Gr-$w1>yB^R1Vz8_aJ(!_m5;w6LR0g~!^%P|2x6g-3~BoV58NjNW9~{oyNfQjgZu!^br~alNm;HW*}KV5+Y6a{5Tqq2#dx#jO@J2 zsV>P;u45K|d>!@riqhq|LN{rm)s%GYnIWjcepkm1x5oAEAz`|&kA8r1>MW_iw&M7m z=LQp7YIeBo0q@bSpli|{(zRq(hS$GaYQOe(gZr?Ub9ORyKwkq*&0q+#^P4{gWdIHq z&GEimer09Bxv6d$i!-JxD97F$)Uh(d7D+++ zr~dBx@1dEOtoFy+?b@Feq6|;l$F4=ZqtspIsOEE-1^R9-m(d zm+u@D)$NC+4BgfNdsBZdjEnObvLv4tzW*~-T-<#&rHd6$RjUMN?LW zvJC6`nr3nSEb}Dpc>2dT^Y<%>%nE_Q_61fLD3hHzdy#SYOFsglt+NiV8{xe%S|u1G zTv``jT&iU(r9wp|{WY%Yr?-*6+t2i{`)x|DkHRJM#9(JEGIcLj&rH*t(ZG3!y?-jP zYyMbFiSy%0F}aqdSPq`#Gd{_j3iyrnWEODy6&MTifuBI#sfofo1T!7pEh{gE}pH^oNd_hGugp?;4hp;{p7#OWAeM2=ays^{ z!;m41t^-2$J8oz3uH;RuFZDG|sSY7^OZnT4`ToJ~&ty7c#B(1B!e0?8X}*=MB1|5{ zrpJgAM`xcwQ z#0ndhl47YHJB@4&hp0B;7QqJe>W1!>bpyEP`nfh1#^`{HQ1d>CRG+NE@1rhOiDV9HPjHW(R7onY?fp##mi6Htj62pVmKSc=t( ztx+5MDEceL<`B_Ua_1p`Q6}Ld^uXr^WI!f3r^k`GiC5_2*M?7{x9tL=kwm|Au&4d| z$nXp5D4|*07Lddr3G1>~vJ-a{_lPnB&Ydquu-ibd$Y#6vg^=(p&qKww9b%)?7Kf(+RnD@v1Gfh?pii;i^pRA- z#2gAoMBKicYSskmst}S$IH*?_l{Mg%%C0$M*=+W{T`uy{(D$!JyjE3*j|spM5mWgN+A-%6 zIoe9me2J2shN0z;Ozc$bD8C{y+$iTor!~kBq%(na^b_(4o5`vD_*F~al;<`}E_iA> zJtZddua4_X=SOEy;kz!W!CGl(pCskrq~+!ah{O^yw-H+PQLY7e~0Q;tvR}ssFpcI$j)3+a8d@EG{Bg z6EnSA&REl*mde66uR$BR!z4iSXidMb``#GChZ7TGwG)pxQRckfMpU`F|gaS-{uG+uQcsiJSae >hLg%BH(0J z@Cx_5KT#VhK39@Tz`D5$B#c4Al$I#b8$LeuqYBAN*Hu|$3CS&3tf)BNj=tc=8|bc^ zE+~MqjSC#>ciGiQPJ#o4qz+e0>Qz%m{E`JlIoN$p5ylMi0Sn1QK?h>%Ck*$!tQ*{B z5Eh?-<>MbZ|B z;u_mG@QF!mye`cX2qN7rsx)*fv)<>gaUeuH<2Cc!<=`O=q zTYhIHBGj(>w*}hgCjC%k54$fd#^=2sZQdUHxldKmG2C#yQA@AHtji=E7(`+%P)FVi zj2bWLbfd_`_iCeXdUYm&he*}%l8P0kU#vQvX<>GtlTC4vvnO7;9>}k?Ihu)}#1I6t zjo{UbnODdP+y%!yc|DN|?*2?ph_$Tw>!z+JwuB1#r=3bjG~QwM`I5Y}@G?;ku;HTk z>j+1=Am3rb(7#}<2MD_mSHbn;)o_)?Aa;a(pEqiZVO1K5No0i)ag13#@zd)N^UMn< zsF$~Hs34-C0{K#d=4e-t;ySzKtiC^Ck90psJjxzB6e$=@dMFP?cwIu`>hbU$Gy6~( zFRep|j*5h9xs_>O&4Y4qjuj}2VKK-i7I}vh@lUyjRlNs#oEOB=Z(}L_$6rAZQ{{nn zytxW1HK43f6k8?=pvxKs(l*Tv`r~<}*kRjSD`jCyida`i0Ym7^Cy`xVwo$fKYpIHm z4D>5F!1O-6r>~n9x}wb_X!EOJLFYeFng^TyeZ*EyqTG)=+W18Z+O*NHVUOAQsmH?O z10JO)&YH_!Q{2D|RTr?`+;G}MX}s6}MmSjnBNO8n4J#z-I80< zo5^y6scbZd6kJ&g65+pcex5MG3=}TEyvp9dP0E#Rl_dJ)8ZW!XSm=Y2jpEQX0qr}h z(}Do$*SV?v@S0Leg`bvNE7h<-vmXPCtegkA zj@{%6?UGyS4b$E1&iE4a!I6y3W!6p-;db4MAz%0N^84-dD|X~f3VyIas?C%uRwpO; zB>!Z%q^z>aaDO)N&Qh^lvW(6k$UY%3OTiZ%Q_9Qc%+cC~$}rZsT5 z;4*Rw{unU&4#Qs~V6_&#Y3%b7$Wi^|Rm+-=k=0S$j8HaEZ!ZRc?JwR_MLk3;E4+dl3=%{ufun7>W!^@ldLcAo)2kuLE~)jYvP{b%Y3sh209?DPu!E zUa0qJ;^1$#6SlfsM-=D$q!bNW)2vjrlBfJ;PL@sW(aWTKDN8(?rm}E3|DOji8L<{%OMF{h?bEz=9&u6L4&iwgpWUMhfK z`neQsqlu8@Rd}FEcFKQwbut`;Rbxqa54;ExnY6rjhOh`?`a~jygoLD+5u?MsT#x5+ z$JQfAuHcSwV}8lsLAg&YE4kK)?{XAlAmtTy8-dVZGG{Y={wnRzCThP>TClliwE}E} zu#DfodL)}sgFc%&UFt=v&Y0b>FXP?U(qZNmGI+j&h1Q_yn{8DfI(jSjttVlcU=hwl zbIMdpqTdoHD=B*uyxrL2fTzo-5T|+aE7pp$0?zRFQ8bhf;lK=ZftO~X?8^r-dt6`# zDqb(HSFstuG{QuZD@53u`t%69LG2!Hl+22S(gBU%Q~s26aj_w8z+`mxlnCEdzOKm@ z5=;ysh$x^4P$ail!h5tyGA+n!PAR^H6lEZ=(f<`}Vf7)eeW#q1-!L-068BYOP^|P2 z^r~rLhV63E9{56H&GEK()w%nfzLVj_fYAG&BRltZHiz}fMO;z>>Dj2AFql$iKYeaZ zsk^ceHe?IAk5RyLyUykh(S+pT%YLjSDIwHJFuoG!@Nqh0yrUv6sPZWV!kit;OI|Q_ zk?#p6JRU1oac|`$%-EoFTbZN-i=vN?aWP#+Y^&m*bh6;Ei}WcmbcH=JZ&vY1CemsS zjVaeIUaDalBU1DnAHU0{RWAlGm4=aNFcH#Ocn`vqZ=tP{1utelbcWxDg!S_-kml%y zC-qLxqxAJhd#HX?iuh^gQ?}<9@N1Xa*Do^1L>aI)7O%TD4tIq6NH`UmLC;K2R54df z5f4_rgc+%9`$+wT@)qN*b{=|b4?FZs=FSDdv1kPqfA%6_FEPu-&U0#v?7Rq6HO<`z z&Z_3MOSQDr5F_Mrhcn-8@7@XsPo?YVln6rvz0d7Idsx zD;nH{cgRw?xd-IZuX8bt;7%Q9Zst61@MBT#a>iYQ(pA6p2O{#KqlRU^ab`}xG5ZQ- zjgTtMoWkQ$F|E192&_I*=hw z<@&KQep5Z=!-sm#I~Ow2{9GyjneEM{m5jsl*=k&}?`~%yX-|;xoxpVOujMnITONPY zs!+vP7IiyE3D@Di_X#^{@=M51ZeoFJeD5ym`r}K@`7JA~+gAWWJUJa7M5*06En6#pPGeDnS~K5Lx*cv6wfCMkBcV_EW`C-cA5Gw!g-~q7 zlVH^iV9aT0&gG9(O1h?RQ5$8H1D)OuLl(zK@yBz?u5jG@6b*>nP2SB`by$h$fd=JU z%eORW8Pkf-4l8^U8(pNwY;{7}{|7da;~M{bnTc)!x1M_yR82171|{kvCsO z@v$&TOPS7lQBA_H2M(4X4;kxA3S)J|(X=0Fue+j&xw;FBLRW&ne_!l%q9R>Ox83ES z(Z+-N*YO<~;=O!(RM`0uxZ=p3_tn_dfGLiwNUsoZFK-F`V#6ganY2d7Y}E|A##K&z zD{}MqIsXk+H`HHx>r~R)KzK69VE6Uet_ejB(@Wr=0(Jt~xT(|zotjR(XFkm;L^f zn!%P;kZ7y4K3!cmHo~l(RS-rfoQ}t>Gj>#wNT-!SpYb|8g~H?I3Qr3`me*h-vQFO! zsyFeNN#bFez6v9{V%7e+$i@avWLZIbqW0*qmQg*U8OexuMH-9q>!$AVc}2)!Y9vtV zi>|SKQXCgOs2mH7!$Qbg%2QbJJ#fZN^Cv>kkDoTzCl0 zkX}oi@+%|JnuT`PpO^0G^d08}^`L8%myUQ3PXW>^qkT5C(<7+c!Acl>o zn2#@ETB0W;pwz#MxV95>f<7MLNunFSn}>O#myL6bVm?t+wD!pEh?uztw^x@A2gXk# z->FOYIejN+upy;7_^l;KTTFIrlIEA5JEfuHi{X_Xk#)u!?fELO?Kds*z=cYFE{hTg zWpNHE>+!WzT`lg7Z>H=r`%BZSLIHTkUvu&fMWLn~BI!npA0vA+W@p7_eS*E~c#b>r ziy{6#$?A@(kWVhNwM659;&K_MTbk#`z-(1>_sZdX+If1(4^|O&fNsty>euY(Ue51_ zm0Cyi5HPE=aHD0iZjJhBYch*!?V*SjMI#j{YR{YjU7Tn`kNstvc@sJM1zByKlw;g6 z_F;opr$SF-wC%N8Q~7wQUYs2wKblUydbM%1RKPjunHA2iPrVAw3&2{`un-!;%?t?lMLR0MzkvaV z77wY`hrnYB=(RCgSIh#d9j5<6jYr{3^JuON;MWQ!gj|7F+1@0g)$iB;W}o4*zZ+f+ zl{5$B4DT06GaA(Ze$UPL){D6Iir5_~!?lK!gr?6FUBt@oT+*IGpUp)82r8*%R-j6I02)M$637@PfNLNukVH2a1Ycue2uvo`ORv2;ou_fl#kFGlp#EKu&G<-Gou z;_LT%uV*J~BQ=#0JbaT8k_rs*q{q;0udv2WnDLkHmnCrK-13|wDh_NVIthJ_y)Hgq z03(@~UaM|+YVv^^w3M3I+0r$sO?yXVW$|^KF;7?B(v#o`i-r4AenC?z z8!cX-qSy0e*?9Vj){_OFBGu(rD&22czgW7wIF^+{rI7@Et3Fd=u3-zoze*5*if5V-oPsz zs3$b|`g~I6>6}R^t!g7!`t{Gi?yfn=2dha-E4Q88*KzR6{IWcQYgrPe|AE!r}xuk5${jrjl23@9y^} z2Q5)6G0a>}9QV@69n;|%rQ0vwZkN&Q3%GDRUm_*z`f*pKXsyc;Wo5}ihVr223e48#FLnRMAew7S?rjV&`J@|AGCV82{&WXl{mSORB1$OWGLElF}5UziTqbGy`(+UVFP z{#>p`9c??!msnF?RMIT~o$L5e${%$I0JU*!%&x0WRb6VS!rh{|@7!%>6A7{|=L`oY z-U9)MbN99Kf;zhvlsYY~;*a$Ne0aq)JD(o2;vdH^qPRDgPF$oFm8kN)wBHzaE>0DF zJvb%k0p)e0H+{%Rcz3hhCAwQuq>!^m&^XDsGNZ|)K@y*dO&V?`fJK)e`waydwPiTs z2%csc&VfahPpo>31@`*r%g13n7#(dHLCcXBdxME>ZVYqLulc0QlA$+ntv1y3B-<8!}@6U;`ChlBYfA(vspO z#lJ-Y)4;G?Z@6-Exvn5O+Fbo@<%e57nsiI`M2o)n{WNE5=%%y5C#i@qcVM3JTI)2X z@54l=qkJmg$&*Mfs{btn)y8Zm7b}U0Zz@Bx*z|k4O zTN!UY6X|7e)m7FI&CH~CAyF%C?7DU<11(W) z3MV?8y`svhlqE&<$G57qyI$t1Y*Zn~lo1a|1KUsmX!MFU=|z~fd*2xERaE^A_4T%Q z1Ga`Ore)Pai3%|6akZa6XL~pC$A@eB_0!p@4Wfiibp4#GUwL7Lr>`!7{_Rl0aDH9+ zQ^Y)LD~}6BCp+IXrZgPEkJBZ~N>%UxCf4Oa#ngu!+cY6(UF^spSvoN^y1@Mnk1Q;# zNod3Pv;JiO{0yWIAJzWIUJD$w(r7$#~fe)`!! z{=0&$Lb%@l6nr85PfM4$4*TD@F#4|`x0hj7HD^yp*ChzC@nQPg3d|ER0JwfaKZy zM|M>aUHbqlV^-7nTiO*Xo|02GUZX(VVb&IDe29yW&Rx%6mHRL8xG-m_s;s*2RL`fn zmQeYK1hOmbPNa$nP7E-JOJ84;*CRTP@E8cmeC3h}Qylv-Z@R~cJMS46sXBy_O?NwH zOhU7t$#re+u%|6hR;`fzI6p3(bR`$Mb$&RF* zwqf!siOGqODj1n+dA3P&=2H6D{fNKssTxUBFGV3=kNf1jytl4q>PUOz4Z+L-CM8RX*yo|b5o{wUxMqV?O16g+I zw>alCBo=20wS~>la?zJK=DA?m>6**Q;}Eem{Wy#>p5<~3A7wMvqj-d+rwh@+!a24? zhV8gO!2TtQ{j^I`ywe>@q!Jau;{LuF(J*QII-kD?oQLsN^+h{U5^82s?dHnHtNjAw z@m04y$!tj)Ba$1v`Z1voxC+rLZ}+RU_N^`tcGv_V)vblpQEVj-TLf{qlH;5s0c5g!5yfoSc zzKp$MtRt$%abKFOToP%YxoQ^cjPjYh)Y!U#GNU0VM;kqt5FWL6j;xtR-y!!81X9MD zV3{jV;nRy_B0S*00mpQ@yw5GqxYA5EJHKexmb=M+sV|iaxqv2fksnt89#df*l483R z`q{j^f`q2bze&z{sB^eC2Pbzn0CfT=S@8eVlrM$nZS~slvqSWKro@d2A|tq_+$WKK zAsn!p^4x`)Px>J$W;G%g+XT40+7mB63IEJywG<_NDu!7Mzc-m;GrOy2W}Hx4*E>FY zLCMfF<=+ayug5v8S780_%C}ph)JlpW{4}IeVhgvpN_xqMBX&Iecgoc)RxX8`;b2^w zQhoBl67Anp0t1%otfaB3$Ta zZgxi4)Ez(52&JxdkAOFc(17X@krH--St>Jc{4<7J-II3*%LH|{OGbWZc^rLn9enwl z6(Tw%{H{d7H#jCFRDyhQ;u&Hcw5uRC86yui^@3o z{3+Rr)K-+mmT!KNYh+%K)v`YOP>$xzIFf(tcDgM|>{E z6&8qmY&Y2XvEb+TPN=JT>OhL-d*`nx;4i|S9R46x8YYvp{`?N(IA=*V2F_BuviqtI z>?SRiSi||C5Y2;Ta`K56I$WQe?L&JMSu`u95ZP{wNQ;2Ox>Ni`Q5eohU`_>2-@?s4T4Wqhf)3Hh%Oj3f5e(f2X zHS$S&%fKS}arZL1{X<8*kpG;;!L0S_E$GvRtPz!Uw}JTMf8YG=*bpavMmaX--MyE! zzmo>edIV>3vKbHk1`FKgyM8H0@}gm9Ke7V2XWGR`MXspoD`5>j@i56sEo6C%*T}lgh}$x~9HcwErXd5Wj}>dxmMGW#E%q98MOkjrORUS;I+`uQX>W?J|z_lMsKv zT@Yzs-IK^RE?aJKh)#doZ4}iW7tmwEb7TZ9b8^TVD)x>g^BG6c_ylWP`4Qg*N~kLa zlD6z4N6jjP)q@WmOfrkuc@q>+#4tC_EVNL`Sq9m;PvMX>mdVHV>#|*);VeEfISBIM z$4H=Gykh~kQGdA~;)zNrPVR08Su(W`!?&=OVlR|buF79+XjMNe_RFw8QH*$ak*SPK z5{2k-L`)jVvBWrvxJm>RHb!{}3zC zdw&T#doJ1ys{tYOIy(I0=sdRL1cE5~OF$Esy5my7xvKznyPyDjx^nY z{9~vh&?n)8S3+z~IB}zHQJCyxil-cl`iA+YQPgC@o9Ty)ltUukqNl8!eqyzljK4vY zsG8t;28d)|9wuy>dW^3QkWpe1WxU1EZcA8%p@>(~PcsE$gx?SO)O*B!`KX=s zgG(fdZeg@gy22Y7@>bj9X18vUv$l_Wqxc7BXBjl~S6B30LWoN!F1*3q!`mvJlc$SC zet!PWA8c{4jp8h-r<*7K!uEJwC-&>7vJ>4TxPgg<`CYba-eTIc&%?Ww&S7yjT~6;_ z7Ohy{)7Ze4+rPD9n@Lzygh5uN?FNizDeYjyv;2_jHQn>tGee1x-Z0Jm-Id1iSx>FBO@U7PDc6}zS!ffRz(Kn|1gW>tTw@be&?nDivk8qgw z6m>z6ZH9!MWQ?qn*SiQq2~|i$xtH5`hBNOS=)2&3`EXhCiQWi8bQh-pbLE>0tke>! zGCgTSFyOg~md1O&!NxMbb=(puvUhTfV7`_6tIMSJD>fQ=&qZcF?49FZmnT>@>=LqD za`WTmrPDNm+|QBneiY?nFkbs zUEV&r8k?I&>E@Kck{*1Ft_hy%660l&u-~UxvD%*8{~T>yr_4b}wy2(=q-Y?2RDqc? zz${CXd1npY+uDZvQw14J$JfQay@ic^e;g3(IB@uf4%Ig1w-cUhFONqcW`=$k?xz@C z&TjDw67x>SqUI7Z&Cvj{(iMQhTNk^84AT*f=Wq6+hp|&6pjDo zJ(W9Yj51jt11y?Xhi-AJgc3q6-EnXHzM6Hf-C2nAIJ+R0oRZZ+gMZ38XbVDyT^JG5 z(4D~ho+wFP(sFoVenW<;4uSyDDqQtQ%Wa{aD|0v!v3?&;>e5yjvol5(slkM*?@z7| zwbJ(twq_h1a2v^wvS7(E>yKQ$vJq{gm)QY!7-B>!nRI>$`lTLh23vX+yk9e1#SAlV zY%B3NHeA4uFp9fE>a=q*^h}DC4fUveDD;;%E1qLc=@o-0ue!HM$-P+6QotlKVkq%l z11p&dtLzr+mV6G*k`2v#$oVrCLZ;x_W-x1cg7elbf9_8b3VyX{&32^uR8k#9m43tZ z=HLm=uu|=O4B(h2mh7`BI821PC``ZOv{ioPnGWF)8+%#Ejf;A9zKNEdJ;l<-H&C~s zG4T$DkWwA)p+=&D1kF5~d@P^`&OJzoTmm5Q=@`@bGQVxd9VU3b+Dl;5&3eDR#r6>lr`%AzRL0tF^*%Nq)p8Q%?|R*Jyn~! z>0Zw#D&px^LCI#Q!u2RStcQzTli1qxtzqnAB3Aa1tNeiug zFgR&TOgRa<;Q>q#ALbp1SY#CBoIls$9HnQ9{1ki6;v{Jx^KJvV%be0;Xz6Dg3iY#J zi_N^*LO%}KU>oTJ+pWRuLao8P<@kBC5unCYZoVn>`LewNTx?f%R^b}s2PF-h@E(eb zIJq;mA)91>M`ie%0?~;i&h6?7B(jexUfh|&*O`}D`i`)wf#;#r_gQ6v{uj>#Y$9j8 zwNR$OI8_VJE{2`ofh_d{+w84lS9*&2ro>MWZdX?s~2WAqWB!{q#K5&n;9fL1{*&AJ!fvf@n13C!~xVy1fVum^Uv<=mcKJNH;1&LtS1hr%oir-!t59Aha=$`rKlL;c!h~# zQGmfl%fzgu?0~CzPLSHD=k)F>l{cGmckL*}pBCfFtD)JF{tnRds*Ril1Zy%pES090 zL&G3>$b)fTUkRESHp3~(ZWW2Z#S-u9=+z2Sn$%Ety2#8~Vt7wD-fr)r5JukGhZn7m z)nU<@A{AvCpR1>#HVP?Hj~p}P&4)zVa-D~$mFnDEXwo-ml6YoUTRM+bW<+3NDCU&r zWu=G8-*KRS`{ajr4S|EJqC|U@>*jv^k3A*3byj^ER7Rj*cB^0F)gWH42utac{e!B6 zMCq>*{T*^L+7Uns^qYI(2hiKTY5Zb-{;OSvIl-gCIQO>hHSsz9%I2~meMhF*D z+t?5YRu|p9+79O9XQ}RJ%8{^dY#_i6$4x4iPi2w9Nkk`tf!jM%*GS$p8ytk#h}*bY zW(RWseda0|Tgr(ue_<~zEks`}-?Q2_zmRk}A`Gi=A;=@4!kAsAuArl10i_VK!R(6) zh4)jv@1<`CQqNqRbD2ohcW72~QceCTllH_6qk5ABNKUZ3RDbWsArS?96#Z$GUV9aT zQIGzBx6uzSUUw!1J0e=@Bf$BofAUTc1XyeJ{>(vnn4!X!zI%4Y95F*avA<1ZhFKFU z!*tybdGuoFE7V)|;<>O538GLhJ}y3=PSio_mf>j2J9e`_vK`D)zJgvz?yp+qa$ZAx z7>%iA-g#)`Fjz;I`X!pVQUc`8e%zaffeiz8s7Ie7l%_mq@o|MP9R;c{MXGmDRIA}} ztNGuQdMmXeZ(LRLo&?K#s23F&Gyu!!$TP$e8hrTl1#_+AARCnQb#_=+WV>|WB;xh9 zLKUB1EESM zyf+G+wy(kw;vD_7G$sAbE`D5QiFlZ#IV|{?9LrGj;r(QUpK_;=;X;kyOxOD#Uq7ZUOnL%Z3;*kq=O?YBBn zglOpB=jU|joD^~B!_p2tJx-XQ#-5jMpgd#u^-UJo+x=yd(zRdm{Bs@|9esZXOpg5> zdE=I=X=m57eLlTREqFzH38gPW%)Lv5!SeH#_~}R0z9+JO7YekKK z1Gf>$T?Uq~-1*Y&h!DujHZxozd7_tC>}Nd#F=gWN@N##*a}@0_-WY>ALWcQ^Jyw4x zbw%|ozlMypPq(VnqT7`m_ZoJTmnfnA=nd`=Q)YEpn)zDgC_&D8Sb#H?? ziEuj9K!npxS4vhEzmH8R3qJ~`>i2gp}F z2RNG2>}|AG83+GufiG@UkcZhzr=3&MGcsqBl&%XcAgrRPnNq^3p<0`0G9&3U8{XfKX?B$4tIk!~mbvc% z#m)Qmg%kuQ=M0JL#|7ENPgeO;vI|H?ig-m)jZa8W7p*UOK1-v$!j0q|7X%+5`?t%^ zdd?q7&8WmA$MFo+*J=1;+h(qvfrOqj^(a;9>dD4}53e#{)%#|JfC|gAP@9#Zv#NQ; zR1}OyZJZG-nQA-VjG+{<%4wgne!VXlx`Q0t2{BQT4oLl~=6JD_9CjUusTM!SquPk) z3awVUWIaM&#O;_P)@iL%Ll8m406&%C7D!Kj-@f0KcuuZcdv0{0E>71+V{1r~4!Rdb zro5)$@6JBJ>+!wdF~@T_sV4NZGx}-{%bvj0c~@)M=O-+KS!eYmSS9^AU&c|R<4nL- zhF*^Lc`g`(yrd+gh$ivkG}G1$xY#{~w+GincE*xQ3WR0KLTGY6pA4E~Mf0QYcap1c zcl)&I(uNV0s~Naxv3Vvvr_Qoh(gi0HGK83Cz#T0g8yyBi-?D0bXnXkkDnjzA26wu% zZTaz`k&xv&SsSk0jyd=79!!t5Qxfdh{0!UDXPBWbvI(iC4bS}go?(CV_31sF5*7B_ zyYOtY?;fc!LnCQ783#SP{pLIl@yn~pSWR-z;y|3j+@$3VjT~xB#O{8PPLg&X5_3=h z4@h6OE@gKqS1a{Ui*)hbX!^Q51j{54eUp#W(Q#}xzdDC%lTEnJ|=s90U2&%u=3 z_P%w79HEYSmYAOk!5z;Iie_U}`z51@1 zqHzG4>d}ri!6;O{dzA#$mpxCpln~*vr2mzWBmVwA21fXAM%r9WE2}p06g=4Pg)=|+ zb>?u^aP@aWa5GjbE9=`OGVZLgJH%RxBlKA&!JnxM*3GS2{aV4){JeK zAn)f+jltpu8=+8aSMYi8N%~WaJ60lJ3_tAcFTTV@=n!)Smf`-Xp9r%fbLQKi)d@$= z_k2w_#}P$C&qkGAa%9m{R02eEz-*=2<$Bu9I#?fQ;xAr}i=dNIsKeHf2tU&G2C61Y z1%8B2`qvQ}UJ4UnmtTLKC#wFeBk7s%cd#Nu-FUls6DODQX49g`~oUF?Y;Vje4-H;!H%0%827bb#xX$KU4)O zwlNLWzf9KU$wiqM2qQc6U^SXtcwM2l3N-lqvR}P~wzK4LpB32`GT$IR~^A_krp@s_n0JUn#H*HKEK+76iAm1R}1IYVVqYswU(v~ z<&|G?HxQ~Q_fMAB>%b4ukeV@Fh*9%CR4c5d#W&H7+UCj1V>UM?mJXWd5dGSkpI??^ zQM{009r|RvlXcPE{c38PbA^cuH@_GRUKbO zM`XJ!^4j!1IplJ5PcXYt1LpcGPjA8@jnruKbF$GN<2+oH?gqIU)-atJtWO{c)41|* z3UFx&QP@{F1Q8$F4g2`~#E*_~cRIyY#Aen|smR69svPwnLlQ`2iX_K_CZp4vAerr` zPsioYnJYj-y6G>P6&`|@%at(AWVwvD+1rV6T9EZ(jvN1;yS3P`L=$%{&=Wr&eDz3w z;O5hPj<`+i@}ySHsOXt&--jVgc>?pec}yQBv_>5#ZQb<@4z>B*04bq8C1+xm6HDz| zSRv54RQ;Xw%++!*5S%}1eD0_;(nGq2HM~2(;0O|~4YfUB1E;u734<*rS)yuIJCASX zspOw5IvW|XEY`aAL+Bp{?LHFxynor5SYN*A2&`K#SIZJ7q+h~|ma(XDhOu!=3KAl` zG_JJ#&`hvO?h`{X56g`CM#sQ8KSfLMz$fIXb}2Dd*?LzOBT8Wbb3a7|Na+Uc`I&xP zY9CEqs3onm2otSY5!L~CtC##sJuKDr#pu_*4SndgMcDbF{P@U3wZ8dv3m6p~ZYQSl zcSSEbr$-W`IRMK`Hyxhe6CRd{L=xiz-@bQ{({6%>WB^_ifz| zufla&+{@F4HjM9H7ENhEJnf$MN{^dD->TDIP=gGC5mO9~X|8@d&lu9(XPty&l4&yiG{VfhF30J;wQ<~!Aiw${Z z9$ll?vX)ExS1jQ(mMN7nY>R&E9!|U)bJLkpTW+pi3N?yAY%oWr)_%XP zMH{_4*A{@2D`H76!ZoYPG?f=$gk~Uj{9OV^X9Rp%KG>+Pl+dw&he3metZ7T~r#2WI zVF1}S(3EwtK}8_%8V_5@v|U?$k?GlMKWStLKwm%1ZW6)#Fxe5{6g>R z{q}6pYYXp18c#WC=Jn;HpOuWfOcet4%_XkIE(^bS& z&3Z7g|u8h z!zSfP{^&&_aLD57lR?XI*qfiDz+RneBX*)`90rTB%AC`98Cmk0uzhXs?sxp@?W&$O zZzzj}X@CL$?WTx6RH5xN_gG)TZ&GfmLWL&G zrBcxGs|9^KHjV()!pyZFa{1j}-4L-b&3oiUN0e=BXvFKg*pV^O3rP8E9V8!rvE(J= zY&};brPDLFH;ZSmG@}k4G2!pBnC|{&_!*_X*6I3ip7L@&$P!(ryTDUGvi${<&F@%f z@e`PzgYZ8W93X}X0P>ysJ2VIySPUDs-$wa774GrCd*%1&rkhASMrRY&RZ@39yp+?f z_Ur4XpE<9jHOOW;uu2 z;u*i_fU{v#xIE!;N+QGDY%B})9=ZdXgHhZh`osf!(THCR52aY-=_%nmSPx@X?6+aI*ve}vb;ob_{>lg zU)M#9`(5TY)D4Q}ZS7Kiq=t{Sg4F5bx`isivT!TYh?4#4(il_3!z%rjEb70R>ENYX z@6|}SADopTIS3XiAFt&%h_koS8uBdANyS-6MxAMflLN{KA}!bB9;QX-@Y^KFqHfk9`t5|jquYG!F|T46mA_Uy7NBZ$Clx3;JNdRInq7T9t`LyEos|aoqHz@WpaQK;BVk`j&MI zzZSh$0Bu4iHe}#+Y2}ZpaZ?P7!6{%D1CG8H9})5!F*()D1V5SOo{Z1<&EYt<(}m|z z?it_V;!|mzyEHPvAKrtuFiD~m#& zB>aiS&8qJ4YMQsUW#}0X?UP(R$@;XYe|z|A1922-Rih%Quzt4(G|^uV{)sIE=~YWf z9rpQ1MiKF%?mdI`Zth6sI;Ep++{&T3y!k?wO;Ln4f)(uKI}^MVw3+qTwR+@w`qq6SsQ2gYdb)xXq<1Y7dX#Oze@|OSEPs9 z!u&xbk1Y6{sL_B51&T>JdU|+jd@7||IbFBRT}r!;kj9v_f1?%Ahe#5(DtDhw6HDrp z7EvDvG_^w`1I&WdCBJG;-csV|jcRG6f(zF*igg$@{}x7!J(EYR%z>Rid=fSY>Ls}? zvYDUr^;mSr&M(m;7A+pPCqdt@JW8OOk?NW)Vbc@LkoKaG{N3m`Bqn(&xJP~Ro-7Mj zpPiGk&l<-n|Axwu*|xA8;+I6Xh8r-ba{r zHK;RNW_V{O6sV$gt;1ENHFHzob|;~5FXPSYnT<*MOhVS(_vwW7ttQ4 zw<_z5+YjnJ&Pfw$x0N5`7DDig8MXItwSBX0J~bfdqMnb1x_xSoIF2hID zyYucn{DkVvkL_1MsJp^4-Hv{fcLMp8!DtPyN0@p%XvE+7cl!0ST^sLU`ov;8TR4Ua zn5pM~n}%oIF^{p3QDi=02brc{?YI{<{+u-%^XLj8LFHeF9IeTsg=T#!?Sxip+#5le z&=C|P-e-i7Ur81X@I0Vx<+xU{`g>}f=8w0`3LB!`6d;D~2;W0uu50v)-Ge}ipogAj z+0ke5<#W>r95o_R1P*0SqN12(B8Tqs7~&*(YoB?0x^YX8gZZuk%`Cw!LZViBVZ?+u zemWCk6ahXy>~$_FLO$prgjyDu5)3TTAVFmf0#39k5mH-<(n+GdPQ1Q{z0-j>mggec z`6bks5~=(?K_F^&auz&2QY(l>7}T?Go0<<|d_?M6(%{YKelF&&`*pNLd9@e%&(qGU zwpzYenkL1S?_;OCS!H4PVE-%jREjt@M427L}hL+aA?n$4JSr8;( z9>I4a`3=^4sR}$Q7r6VXU@$9Ou{CRNz;^KB;-)q70$QV`UY2 z>=O>`PU;UHEF_0LoRZu$Z=bLa;amPy!(=pecaqGtglD{$IYIbUF>h{k76DZF2p*b? zh_zM!9_Hh#6;GxIU;N+bsTP(9&Ud$vYhVUhdb4OJ%ycb%MKI~RMCZvq0wOkvbQMXX zmE^X7wKFDLujH)jyqE2}uB)-I^&*|(Mi0yPRiJ;kGe&8v{t>%|WIKYr-#K z&hH}ZGAj5;rejzo;lV7wkMYx4;;>-s&8?XtvpgLRGoi^nW_R1>FsoepK2Xk&^L!S5 zpVd|}YA+s6Jr}*}RwS#`yvgbrf(EnXE3B+UI5;@W~T@3LXgT`d|c&VZL zxR_MLm$;5Q3(gc@`?`(?jsG}(=%gE_61ybysA;ifLfOU0emG|2I-#`$jGlLK<-b%r zpfXrdq-=9>^|iA}vTuS=e@C4X+2kfnx#+C@<`Zx#nj&wr$UL~IxqFvD+Jc;2Ay%on z@dQMyg%jDt^d5)9O*75g zaA2!NxWi}iq$lQC^NmB98cQMIZ0>lk`c3Cc3+fLzZ=vOUVL5xz$n!D%J)9Hn9D2^D zX~OZp^JdeN)>waf**7x9SVkk$J4{xc48~%>2JKUNmx9dfd!`fF;BK27iW@1bL<9er z<0p`Ed^h)*RNzowMP-iOm0qA=Xd}7569v$l{@9uwS*ZPWhAFgo%J}f`Wz4ma>`!z2 z(yN%)mW#x{0tI&oIOlY87b}TmwR?~n-6}S<(|yOR6(b3m)Rp;=wx8j3V;0TJ@$$oz zWEKJw6TaD8FvA$?gcEmh7=E|V0wD~qmqGi+J)Y#KT||9Nur(%1A)IlLnnYbuD9^K3 zRQrg-3Z+yrk>Hx_j@vM5fChzEGYY;$Xdm9!_DQ z07O8$zvH)w)uC+$$j`6+WBP@{vXMej>Qfo_YFORp&+B}f0EWXl)o7~8Y-~QEkd^F? z`|irR!Q0K2qtS0g3<{eS9XR0+^k)*t#Z!1TYne#=dDJjlkgdgVpP`oPRZrKI7j?&@sa zMWLU#D3L@0g?9U;9%&fV7E<(DY?UdFmLOyEhR_Ao#fPc3M9=;3G6KtbE$i!?dBxoP zsP4JVUy}_le$R>;;VZMzS%Nc@zk%J164CeXPNT)TOnn=C38Oht0Xei(q5**pV^HOX0dQ64EeRFr% z6gV0=TiF{mDWQ5m2tHMv|s*m;jzFxkN`gqQ}c4-Sr|M*P9en&3O zyR{}RJ+=-%Dfg{%U4JJS$m0b;5jrB@K1v5}YSg}>hSbuGYe`ml>+pIsT|NjOx~rin z!qzvzzlk+rq~?zP-$yF^!ST(sUihfhDQJJo1Eye>1wRkfDrn_~q$8flUTwRh(W?`+ zuVgkgY>}C@ZJL^kfh%4ZC2BqTy(d0fo!bb=0vFVSnR%2eWqZ%Tf}tU(ZG9%8^0W+l67^*6%iPA9Fb?s+{U) zI*23gH5uOgyN&qY-mprG$w1dx`Eqc^mSZS&hk=hL0?WvdM@HunoiRfzc^dMqI8B(S z@MfbDzXfqjGh*sLpY*67n@%=E3H(i@T^3eL&F4JxO=-OHHrS4(@cbl`0zmT0QCJX+ zV@mYpQitH@B@$26|2h_oT>cpL^qfgDp%2$SMxTF#-pfU6t);jwc zZE=KHqhht0QK=&~YGR3*==wW383}ud<=2UB^{nE71NP;5h|eEgtVY=(yg$1~4?kEpf!Zp-$y^kpaMx~5g)M+scyB{CEj9x^Y0i(g z)pluiRoqFDkqn%Sc&>lSnj+2ZV+ES=KIAq9d)bS^iqO;|>>+<3WiAJNUr(%sry!-O zvzPirw2m^e2INLzdf)E@NeSmjT(= z+RwnuH2=we5W0w!+!|!z3$-#!zJXZ`lQLfQsH(T_d;R?(!bEEjlRz(f8~T<%2%qLT z;mOLer`s?caP=05)jaoxJaF}-2dA*F2qTgz2yUzwYscp-c5ye5=oC{=HQa{1+f>l) z2mt8{1%tn5iGC}Mo38o<$=tp?8P3=wk!aJ?X%ILdOL`{k_@>*Hc2&lq30{I#YPN9F zQ-C;g7(W4CoNKt5{HIjBy~za3{17jT#Y@8&(|!<;du9lzZO8OQaxWus8E{8O0KdNJ4J;V=1*YwlBBzB$-!Ujik`rm{x^h z`-1O!z!E`O@}xc|Dr`)2Mq?sUuJ@;jvVutJevGW~F)n&K_&Ciax5j=og_gqjo~K&1 zuOiP7Y)5qp;yMJ_<9(ANy_KTPub;7cemv3Ro8fP+K%elps5HM{i*=Qo(fSGXbv>~s z21DENCQySc-#~ZoZM$}P)ocB*XS?PoOC={3GYB&>t#5>6QY2eRBvaqD9L@aLkBVmB zH46pCUXyWeJY?%NNoS_o!=c&Z@iA?ZcxEgo-$$??H{vK9RinrXS;>@Fl}6*RbG}6_ zX$Kg_cokzf3y~0c3S1?beDJT?L22^fehT_~3K2nMDA9n+>-V(QtC|Ls_bdwOOuI`? z!%P&0@&I)|y>4ltyLeleWH1$ooY7lLz`U7Lj)y2d4z(> z#3q{dl(x*J@U-u=Sj1tuZb<#Tf%dhWXCRiUOv+vjD3W=4J`0}<<_cDK3TNmR2u-rM zDr5;M&HdqTl2Y9kVIz}A^Ut{AQ68nCw}r8TpdAnBogOPejB(Y~W~dcOx)&#vj#M9Q zCwYy(QJVYK%s}fkyF0+b)$kg7NqqAr$o|a&0M9LFHw%$H zej-JiZ)N?U5ELaGT^38tW<0wlmaDN?GtP)5FQPYtjBbig22@aErQA;}M43A-%L}Vp zY>v($T>7IV!XYT0G?!cN65E%a(u%wSatyzyEw(DdjNnwBYKPH-9!A>%uFvOU{4>LZ z3l%reevzN8H}AsOp=H0#E>Tz`8QHInO3ZezOfa<1Sl;PTS{OQ1ZsLA(?4-i|GM72a zqB(!}H)VM6&n;_TEbqEgji1wX?1Qxd1PP4M@`P`M^!!JTY(_F5ixT1huFL7{ zuRU%CCN^_wXB=Hteo@$(KVX^Df0o)--|V%l%>r^(pg;?PYImL~?u*KYE1Xq4YGu#J zgd9GWU@E^SjQ)m;rnKoZc71oqfQORN`Tgj@mH32GBZf3%J+*DaZq(Kh3Hj*Tn~s)l zU7>9dL-+@h8qp>;b6iAk#_3tdRJgQ5=oBGF<^f7Dk$09TR)(WVpV_ z_;atwr&)NdlrUAH_&1J`chQ(2+g7gbN{x+tQ(2n7(mfW=R!bb-;#&K&ccO8ffh3`y z#+Xm8ij{>WJF`?wzs3TWVxo>gBi7=r6MJl1g_t^ZsRfx?e5C4W= z?71j(NS6)h%Y>T|{qi<9HSLR_w;qe1kD%5qgb6^qW4n7HzmF%{$}}!!7nQUnyq-rv z61+H_HNk`_yJE-p;-?dgWh4qHLry7b=dzGx$tRt!<}Q$j&4hA$!T-qEl3uJED&$8C zY$CtZY!r#8|DcB*iOEnLeyVg0R!|s7(RqAWo>9keP|!rf9K% zHpo=ecLMb+`@kITtb4m}f{J)QCg9NB!@@mT;h;tZm@FHY#OLr0a1?J7rUT~u1_aax zL{OP?b=q+<{EEJ}HrKm%7?+v|ud?Xf*)dMh{xa=2Wc;^TCG+o{#(_X2TQxOS>^m;C zdn>oI6P22y#w~2mEZon<7Vg+VM;}B`XAEiZ)s=03*aP#<-OiygpA4eBue^DrXrW;m zr)k2PPE{YOhh00o+(U$iRieDg-`=^?Ez2k^r1Q93er_JC^rda-Z#0x+g7I1<4jtvM zf*6(&9?KqE0}hS9HBD-j)W{XK6jPF~h^@%*!-!>V1_?S(t|J65cU0l()l~!m1ax9avo%+ zy2VI`-l1KJOnQVIAwCFxlzFARAG7z<2d1bcPTL!-cT6+BRzOch8rMbvZ-+dvk#cFj z3}yCkRD7+q$&fOxRfTz=3&%w*VP#{lnky_(S(Quk?KZzdK~c0z+i~#>i{vL0ugpW= zBbOR$VU^ZV(Jn`WeOL&Ufj7U+BC{TnrD(>8@A1;-eN;;NOOAk*e3u%eCCn&B{Bi9U zYyrf1Cmjuu@?~Z(zDq*|orVX=dn8aNd@zkj214_hhg6Ld;whAa54`Pr37&IQ7%Ds% z7}vRqZL`^17raM(nD%b9*D0#i<->4I)m-(f+_==GNk3!UBXkc*^82TZDTW-;7 z6wQV!5Jes~!^U=gR%SI-{F%WmTWf@Y#2#1A2f z@4$tZ)NxPFI$=Lx*{Xex*_5(>e>3CLJI}5N z_4}^U_WRrWG-!R-*RK@~7GS&ktmGSe;j%4U>ve{=J6KvJY8b>XSOTn%4mLJqwi%~f z|2np<#2!|{oeXACR}L589A6VaYqRZuVkG~D|!udEuS!J!2H{6LGS+Yo5w zxk0m0_AW9F-#hwU&$%d9cLS76uL#WbJYTBFSU(0!8l{l{H4hTCAX|#Xu+#d1-JA2O z1o6bh$ILWS{vLQH{LauD#9|ZK>bL>S879;nATZ&}J)K5j7GOeKb5RNdf2e>a-{8`Z z8Nue^)>H6%g5JpwMn{GkSXYrAZHrs@7TmQaXQy=O za3aY|fHBfa2PhEI!r09d3wiZn;wCCI3>Yxd`+9&`{CCA&WIT_RrH?AroJE^+dadc& zG|h>Bz%bPFgW~L%p6f7j_(JKz)8HY|@DUrFJP>$$*2YjVD{>XY8op$3hN2`GePLne z-QeHMh^UXP@Sp5u*rJQzHktBt_*Ub*Wk{_FnTb&-ho-7UR&E(j6@f<{{?S%Dz7J84 zyl=iDfHj$S=s`TiM!!p{iH7&*;FE;b7dO$)eJy}h0#SiW`kDp%I(y-a{9t2`HfHlG zlSOsF=9LZobT9Y)tFyIkkyf#mU-Zy<`(v5({?Q}40p$F37vs06yBkDSh^`kN`4_&Z z{kr!>bLM8g{GJ^i>`h+Dv0l-j4#l<=*Smq--_*-f)gpgx#@f?vBQ)zHx9#P<$A;KN zK@vE;Uu!&Dogp?9iU3sfY9cL1V!u*y-?kCl)!QAZ$yIS8*`38MWrAKW@$t#0Rczm| zdsELgzTINnVY`86C)Ns~lddPP#+fZW8*R<=`{svyB~XqV7~=dGvg9JLUDtdBFa6w2 zT#6`22nOnzcb%kok_H52`uLDuq8;J8L30^cpV%Zyr=djbF3!@Uj3o;szv|LG()))| z1P=KqiHwVbiP2PnGCV%)CoL&jk&|LHKsESpLH5ODsNchMR**Uorfzk5^Ax!D^K4|x z@6NU2^x8Prxl`JWW#P_kq)7}9Od8C9WmlPFS>B7sxox_EKl2d~4BmVE|*rPu-8N9e7x+UXXRR|lkYkBZ!t+lTJ4 zz{zEkR4jw&W^-mPPQc<;47$uB2=rOv7%{3;ld?*DW?MzhkaC9mvXWs4tl=3MBx*Vb z1~ezO%BcC3Brr;Z@}D$0)!ErNkLSOtW*Tv+;gqMTL}wEeKtPzq(og#}5I1+I%?8JUnkPOssFK87$Rg4)+vdM(7)6Xd}?`fqqg;_1F1$#**rrgG+NaT2Md=NSg$v^qY=O_YCZ0mIBbek|r- zf%%XS<{;PO`0N4DTK2C3@u8m{(md;f&(%g*P{2x5<@ECDpjC)6`7f-)zlb&jZlUR| zg^6eZq)NO%=?aU3qEochjYB$6lP1~gvRACG=l{%kK!N1o8~NM&&~@0D;zkvvrI_~&`ZbbJ3gd4e zC@8Xh0|y-LY;uE}5uwoW$3oc?2CzD*J^k7Bm@FK583Q|iO*4ogQ3`~ARQ#@Z)s5Sc zTc5;4eHdF>e85G$bj`Ev_F1MGnNOhPg^xJnN#hhzM89U26Whut+~3qRBEiKd8Pb?RJj%UKTR@5#9$QAChe^9>_3? zxkfm8x%ySwNv~~)kw8=G!SAY=bOvejh|whKXsGmtWep}{re+JZ27(-Ouw!G@VTlx_ zWBLier}c&~G6Mla%%dmaJOHm1ZTfJn;l-ZyBk7qo9s_+~{p^t?haeey9)0#KyM|q8 zcq^JxFH*C{+l&{7Yg)_WZnV-b93M%p>-1(orlAH+B} z6K{p~7OU0pKEt!VvbjFG1S=xxM9q+sT|}YDu9-YbN56`)&ICD?TpZpY#+QPU@x-)t z7B3pMK_#Dyu$zrn<>|QN*yaa7k^3GC9CLK%ZU_se9mru)ug#ouzg_cI8ao6c#Wjd1 z)LMmq&l}>qTuxt2by*4A=x$6m*rL1q-3y48I^MM`n8v>N0;|iVgJ#*LhmfWsseWx8 z?KZognNklk^0aIiaR^;`_!B&xNnyt`t@lKm(af1&_&y)Z*_rd?K2&0FstNGx;-Q&4 z9h*jMYl;Rce(OHhYq%ZGF{x8lp_MD8;nEGf>t$|n5*9D**vPU~aG2-nNK{EC7*&f< zJaV}ljiQ~x66LEX-SDay{K`Cm6l9ViDLWqtM~^wO8T;3KB~JgIL^Lx)6reMI!A4s) z*(R@t2etEt=!knii|HJ`5a*tZ?%{h;SG(X*4|0_ctq3k}2%(g%M{1yL_|%q;XwEbr zv+ZKZ@CAPexYect1X5;KsF*#C3OzJRTa@Z#>b98+92@8r(fY8>QlU%@rbc?B=-B{% zxaHL9n;9MV_YhCR5GD|YF1b53ltD?{fKtz2LkqvQ)tv=Hy6c5PpU!EH)CWKeXC8c| zy+T);7ZJaBCf%M#J`tS82Cj{{ZB01Af|zC%j{&RCa}Ai;iff}k$LCHQXI6j?`3U?e zvT>VyC};+=(MP#T`lB=ouB@*WqUMz0oHsLT+`z^p;bL}@H}a*;kFk|fQ64|TlK8eU zdi+BC$U|Mz-OYNlg6GkNa%hW{xI+Olm!QPP7E$M6qfh_rAsn>d>CC-{<42A(`<7Q?)PWlSYaUWL-ExBN~ADEyM=bjFi2SgwxlsOc}ELf>mO4y zB7vbIJhMqD-=zI?`KQ{c5zcE4J}~Mx|_yO%6LKB>8vh;D=IA z>mO;MEH%3RZJGBtzcr59briw|Yo2eq&uvP@yd-Q+{xhs9qojegY0-b9;IfTxSm@0; z1$1_ISIgWleJhHVx@AM4qMwqcE3XF)}{*E)df>Zp2GGe9VDhERpk|!Tt`p3`0Nrp z^PUrLdMoi{c0W5fIsG2x_;8V=yEV97WvwE8$Qt=dm!Rnvn=Mp`dlQhg7FXTZCBv*% z0lQZ#uSa7>7E6Fxx56#fdZmn!onUWXRedws>(x%Dd-Rdkc8FEWtD4Lc4wNAHM17!N zAeONjn%_wK4IXzr{_jgy%Cl3#{j*iQ0Ad1_O%5$ zOfrZxQ&$zu3DSt04nZA#ztuqUBZh2WiPNjW71q^-mNWN4A)I&SEcFx_DW`6 zeL`1g3mU5bIW=r*>tdWM(P}yu7UGJpm1YrRu1dx5`7Kc5JaX9mRDWE=4Gybbo?UW19iENa)xr9uj2ZYKCH=5N%v-s^RfqwG9LsXQ-953$&$; zPs*3D(%(iw(=L;rh1S(bl=JGOFqxyp_6%4xF-Q8xtp&ZnhZAv_6voPS8bSQ&^{!K2 zO%l#eYn7UB3{Pqo+Lu@h-)67(x? z{?IDlgS8POKBe}6u}injW)tG&p{TwxIvoM*DHl3n5Oml1d60PC&jiy6T1&K|*Dx~s zlUfogJKT~NPSAG)>G^3Dz)43RvHEq?3XayB%^@g-Devnj)b%}8O0>k&*lyLIK896qqxw0h!;t=cJJtI^W)$QDP0fv#< z=)fn=q`2*h%qqHKe;1@g;d{?{)kTS3xR@N44lML21F)mx8KYBxAlwv<7H~d&Ar}!bN#P@AD5$A~$No7N}fYWH+RT?W=HF-d* zGc(3gb2Dy65N%)L(l6YI72^aPa~i2uvTlMa_v;-WjTC$~N+~p|;tsqMp!ytC!3@27 z<`R**Uo%_5NhZMt)+Yf%3ZpEm>b`utiebornQ@9_*)&fTp596& zu>6%mXN*+kk*`oq=_=r3oV~ohW>XLq!ByNh>e14dZVi3x4H=uB?dAyCK)WA(R)ZMV z6t(G_m52t-U0a_BL#LgXnq~p=zR$NqfY-b0scs{HrDv%~2u`~XxESmo<4~#W!T_N= zN$Ci-iUKrK(8{b}L9v6zR4{p-=Lx@L_8Y~IcpHz8K?WIil7!?S5rfLWUyhGa;^TXs zsc$%o93oT#Hl;cDn(FPYI#e+>^)!(e!*`2ygtt1T!ii0!R|5s~9=v8n^5Q|vFc+N< zOY0!-V~N~uNIY~C!pC5JO>@WhB$Zix&;Ss|&B)!{O=8vsU2DcPE!78^Hzb}Vky_u2 zzDb1kR&evMCk=FixJuF!5)L4##FW7q%p3h9RFkfPT6JBb@8swmO9HoXw5r5a^-1Iu&FTp@?Nn7#AS%Bp8JI33k-o?@6d(TQTS;glns)J_>K&r#wLr<2?4lA|m{5vcsq zZZZ#I6&x!f__1-EP{5H}9DcSA=~LQ!&b($#4Xg}RC>XA_ANLhCT1^(_jBx2uQ?{vr zTJ5o3L=kBpiA~Px@6ttnB0lS@6wynPa!^Z@DYv#JA~&MUh>Cu@&wMec}h6UH%xk{^a8{eZ6i z6rT4_S5uWF7!%Rl2iHTw4vo4m3)h#EMS$QnEjRRJns%I!S#ic9Bd8h?`8&=r;unOk z+yG7ZV(uLS$%0?iv9q-b&A-o!E3s5CP1!8Lr+;!_*azhC#Og|zVQMMlbI|EW>CIA+ z)K3!Rq%xsrOy<8q$K#yZ4&Qt+j?ep|vCX@YC2S+h5Y)udo*u_Qm*eL>!TPszDoW7f zhb0N2N#URRvtal3vIG3 zgUAb++sZnzl`9aO8w=gd;Ld&Bav{uQb{M_+MYt9CBh4h$r9fg51EN4xA?wVKKKo)! z9a9p#=)4^S8P*s(l`;M3@@m%)lo8K=&@d1iOUXJfM8xaZ^q29=UhBcB^1 zK6QWwVB^D^HQV@R*%#2?S0eV~9^ zXEe0&P|N#oXOL*DZWLie|1(>j%jWg!s(ZJP(E8EY1)G<=i6U33%Dj==D9U034G61< ziXl3xoAlLc(mdUpj#mBLpJyXz*-Fo|%fl^frq1pu6|7%aGHi-RuF*vLnsNNLn=aUG zM|UTcD;nemKqz6P@v{1GRZ&F6t<%6FnOQ7q-mm@wC?D;n5fgJ?8Y{=GJ=Cw7v}o}v zB#~H?(0FomR1}c9Kgri^!k}`|>`3*(nW>SSA&NL*EPENIBcCnmbMjJUOs>UhYj(^$ zvT#h#U)e<21vnyon0zKt#K)lSnI77*JrFM4K{`4VP5ac`Iu(90#NS6-Il3{dm@$T5 zsSWGHRjWYH<-$2}d=<5?;Nh#a-FA4)Q@ zo?|o*@BNK-O@QbyakF!ePBP_)EA49(gV;{nc2!4IL{k-S#366&L*spX9YDL~So6mS z?z~`XKwSxo%*}ec zg1Ae@j+mI9Pi`0Ygm~WK+KRm{Aji=Mf+gycG3ONUh1;pyAdz?71`q0B;%4Wn{^ag9 zjE_KYPTlS8zH>@_-=Zve8my`e-Oq+@!E#4RWxti~cP+(>M7~IAI8t8R;F*T+Ifr0PlRzQt4NC}rhc#f%dcRI&iL{5AGI9nuFj*MR^_6 z7>UtbrOb7r3sVJyNeU3lt6b}Yb4OJCwvHc4tT)6aFA$ywydYJ+-dMj#SFbdA9W4x_{-R^da?dFH>Cua@+)t( z_mFFy&Vi*c$9?EuNQyy|+S-rQll8h7Qt)8dlav>;uZ{$nmzANqIMs>E*v-l|XO0^V zJTjBV-ww^mTzX%e^x~AQOt&iO`o4?uk-9?I=bCKACja{2v-+=QKH7a)Po6o{P13v9 zf9#PcxFQJV%fm=v!6AyH1F48f)Nk;Q;;BqA&8Hm=?~~!kGoWi>XuV-~EUA-NWa&7{ z2RAgDU7l}jsSY)btFMhj`Y^Id+X@Aa*UD$8Bk0G-{&k>}A)7jg;tY%=@A$t^uJe~V zx7kgB^m~YfA8AzCMhcqUcg|LLsMUUBj$|j*peFfMie<>94OJGySh(<$ta<@bFRMNS zix4;?A(O&7SQEys%DQJ{Q2M0ZMHQz8KXbe(>+-;ZQ3ez(s%NA2_o;9ek+u0TWqg0X z{@em8bY9uRjKWLxr{ZC)9tQHFEOz(ldz#k5j|8WY4724FbWdJS@pa4P1<5O#&ob>% z4J3a7&Qodq!1uwg*sicnN6u$@Zgs6IB&w)%jPt6*Q}*)x#ZLw>O?YG>WQpPRJI^0U zD)mbznmxAcV)HAO{>&jaiT6!VwxhhJY&TrhiV&8LqlGz>kwF_!oRJpDT>Cxw;6m~8 zt`8TnGGQpTp?AGVpZvIrUK(g4rm!zp6lv&vpP@+Ebv-u)JGMcghX0H)6Rt{cdD>8b zq$|{2wC`$4=*$OlGy<s{}4oB>0b|JStq0cE%4(` zQ&{k_!|6_sJ30?TnZ{$@N8LOU!N$0*f z7E`nf9P`c3dx1%LJ(>-VL0N^~4g{R-@- zM%tv!XXMy;nNf{BshuJ`t^;o2xbIH8&x3EW1z(AmL0iW;$2qaz{>gyv&>PZuRkd3MO zJDi6fD!sEAE!wd%hL!W=S2;tgh~_!;@`w11%(!Bf4u}1rM_eU&d47y=qSLZF>6|9h zA#lDv;ZBly+Xswu9CmTP^~k1%4al)He*~N76c1E7T!2;!1K>YQCDE9^Dld#XM+ouF zcHMxjf2yc^&kf59tLf=W<*3#sp=dyAQT{k9+I{^fX2{E>Z9uJ96JTVRC?+W5JZjqu z)$CM*ax?Io94&C{ePfCf^?0|kh5+?u=KQKan{*+JeMHOhf_{yE9ExbRhH{s5lCB>a zCgiaVSM;LGDwmcf_cupnXFZ{lXgRO5dxmxZLUc5A6&ITvu+|-hg92aogdfvcci8k zRW4o_YgHrh9h*v`*S?Z`W(cuLoo)PhZb=|zvQ&Z^#%;k=v&)v4yH>oIlf4C3y*4HA z8=@b4`Kij0r?`(U<8Y?QR%PC>$?di&qN}Rs&j*ZD;#wb-j8Y)HXQ;%m_p?+nUcF zD9l}8SGE1A6W4--uJNuh_rVJ)oVu$w%Vn*Pg=8bcVpKh^*pv$!M9(r+Qpw**ZBT?! z?KC#q(I$3|P4lqxZbz0BoVKYhD!J$$1+csDeP|JNYH|^d2t(Corz#6b#8&5l<^3pH$E_wNdju8;Okr`X zD)n^nJJVFgHt&b%(^^_Qsae0>?)Q214f9&wt``?p;AgE!Y~WqDL)&v?2t_r_yxmcd z6Md<#&mj;AL_U=j+H$1$jw~TlrV19a5G=YIJ`KZHL}S=i*J-o;qV?@rabPa2-ft(f z&Imi|;d7o@cAf*%cnJ*;l%C*9LlgHIXY|@I-V#)f=<<6FYfEz|KkZu;=X`T0MDd&Q zQ|XW2@|*I`VmD4Vp*q~2tFFz~@~Vf|=!*L)AJiJlk8!kTNuA@BMd8pTljg5@N%7a# z>P$!GG^-b*Ay3%%Zn@1kRLm+*MR42Cm1T7-Pw#7h@Y)jkTzd$Q;G1XJ(-(wJkXP7Wpmq8N+hUQPRw)v~<& zilBEWN3qw(O4aEsgQ-e%8R+I7d65SaJL6tpU4&}&iFeyXup!gN9yMa)?oy4LPDIBx zf~hfBP7pg})@iTg^CyiLs%Sx!>IlUVS+lNINeWwl=cH{Z*`cc}Qjo@n8YTQ2_EmA7 zW&vMdx*pOLT8w4W^qPk>#QkMEPDttJ8NH_N(VSjxaX8DZf7nEuvZQ!W!()@H^7B;T zB?_57ea7Q45z_t0-XsqSCUQJql2O0cN-jpa-o_u56&ms+>ZI<m}v}wNxQ}Yx~6#{jKQu`pq`FmvawWUrr+B9T{<(6{v}8f0Sm&UViip zWaWqcHBMkdj#khP){UMe>V!3b$YAF(elENrW0^pt~lLn9FoQU2)T zU{@qZ+Ej$6TdO2ChB@E%)AllXGVb1F)~l;7iY`3CG3C^=Y5d-s8GETg?2L{#G<7*k+)03suiY{ ziJ81&99n>RyNrW%)FPU4UCbMi5zTh*>4qg}tOJFF z8U%(Ud0RbPMx}>e1T~UJWJ`~U)8eIdqAxx%iADSEk=!azi zH!Jk!PzS%P3k3!#^o^<;94Q1JTIxhjTxB28GFgPvw=T361sv>I_^Y50M5ppKRe@3d z4l#0RXG2$rq>NPI%4}yU>sLj{kWyu%468CVRz55$Z%0 z+>H1L$dSM(oEcTW#r>5|@ORzWt{fWMzx_;W$HX;!{5)Z}P;we{h2gWW)I!eO(7nvlvz;-36|imF09*JII#z2Hy; zhKAbIfy9x`R11AgL3u(7B17vqwnnuvLv}67KOI~*k#g^KIMWampj8cuE$V2%LQmT`~wFwKEE9FKgOAW3@q{%EU2zno<_p)J%C>2oVW`4ZAU&#sDpagkuX+3wrfs+!b;z*f zmuPX7Tm#}EvC38DrzwK{QUz2%OQISk`E;NZ-Zg(`=Cfk`(cLK^GLHqHUkr9Rj%puj zG8co0`M}`S1SAahAv3;x5kbLN-xZ$grQ7U36X6g4hC)n7MzkLcop`N8MYNX<^E18O zE8f0YQ(~mUMi2K?#*jsQczI6!v?7Fmvv!tkVU3PUeP{)n9@|ZQ)5bY1U6o`qItpJ z)bIS3e!Z-$(oXiU?GetY(_M?;^C^t_PnnJiRV29I53{kA%o%%a!p&O2W*sl(;hU&c zW@9+=t|f^x_owaH+=b)UtP*7MuIq3}i6^)5N)0IVIyQ|!6G@U?i^qvr#+!*Msvi*e z?zMU!F(Q2BEUyHDrb2H=FUlm9aFfGKviUGXe|6h$1sN0d(*La9qaC91YCH|c zC3x6M6;Eff!Y4T%(ZJ8%rGBF|pmU6%qp9n`WbJj^q@0pO3*@CA#H<3TvDx4YYiTMG zdbfy)Afi!WgC;8VWGW$2_qFFOF}9wJFi0cu-}Ir! zMyj~UESH<2VepUEPtmim}L%?5!>DjOTIa~4hhzY z&quOz&OL$opN;k7`@SvhOiy9O7^#x9=*iXm>X_wjHa%MYieVyZa4fPi5)vAez9K_E zCTcjy(HPECSesPI8|VJ>3rX8AmQ>}h;bfNbk{nnhRGQZ`BMlPbXeQu-f{ZHKsB+Ke z{4bb6!1Rs%V)nf(7p1mi8P38_6I_vIRp!GekNw-APg@HkcadIi=nC6CA}*K(di|5R zsM}b-2gw+?(sC(H(eGvy^>_dLw!vvJ+cGzX-o(zA7dg~!-9B>+KH3FofCTPIr$Mem z*v!kL2XlCLGApSMvq#!C&J)9fEP5NCT&~jeH`z^PYNm_>wGpNS_w&f_`XPw3rQe(L zzF3Q1R95UhtP!d&5<}Uue6z7~sSOLpwTl%m*bW=ud;U?FT_xB8J*ShCDGj63)A z^B`~B&&m$u#%Ow^x5j;CBHWy+FA*TK(KFng>ptW#i(PCpGYUZ^nL8-C^6wa5_e<;< z%DBdBI-MVm)mSbqON^l0uPTMBm&5rn4P{YY*@FY0Ynti1iuBbh}fHrziZ7|KsSa zwi{Q1DEdoccoZ{(CB2E6SzkY$S@$I?6W`eCsycf+R`-sN`OtYI;v4OmLw^~i_H~HU z+sK&@D8*l9HV8yirhnUeX>SH61SW!1&}J(x{aN0X|81bROF~Cfvq#@f%H3Nwp0_v? zf!?Rk-c%Ew>on?xg32zvUU6!#`YZy_5E0Q~Ut9Ei(vM32gbT zNH@~bQl4++(c+N^`-BxHmXu7yv>N^IJY^90wCOE8_CuRhF`^HU!=A~9xq2y~ zo>D7`1Z{fbtcc}E8{)mSsGpj6%-YJA{SEYdKFROV$qy2|U7dlZcMXvKLXGL-;dtuF>g0Yb_FZU8+%!oPzt1vwjMAAElXMmQHMWdaN~Q?0tBBGXVt&)P;2^LtgpsmUo}YDOuhp(fsKV50 z8f{#n1hT&gb-Q}fV8>s3ct%jtF{uc4JFYpSU|c8K+0%i1y7Z(*nmpazrQ+aKnpqbY z8kN1D91<>_*Q>_vKf?;JF(uVI(0Dbb>|fR%%I4j4Zev#5oLdr%*TRg^IS4xcaV}c! zrsO71q2Y2eFNW-G4BG;5Co>J$vK#Qp;XQ_q1Cr*K-^mfDy*UV;veI+bGxNN_81D$h zs2!p+(y+|e6I(v=%hO+B(-P_J}T-jko-pPo&iCfazsve+9(hOyx4#!P%Nm&f4SaOwECVbC0 z3|t0+peB?xrrnx(ogqGAA1;PoNY3XMLxYDILIAb7Q6!l7sqaqTQt?s<6Gsgzkqx1> zWmGpMt#Q87@yo;zJn^siZ7e>w(YmNhB-xN=KO^(Kztp4UiB1=r!O_7(EYve9@q>>w zAMUdr@bG;;#?p9sv?!~-iUW}L(Y~$}j&dTXTq%})(2Ml3*Zk4V^<_DOtagc7Mh2#+ zL*VA@UVvNv(lRxnFH3m7o_Iu|zBf z8F&e0(P3UAM3ldp&=1O^uSw6D_M2RUrTxa*qk}%JJlSX5SK3w`FmcQyp1B>}3&Jjv zBstTo9m9##HNywdvDxZgX~# z+P`Ia&W!p69uw624Nf8(=gvLh1PX7t zRHZ0+xbM5kj$ltviENsOi#3ysMm7uXpY){@vi4V)(km!vY2zy`uTzL~l-~6W$B&NL z`;vB2epbaNdh-&dhG4H9NdVI}JF=6$BIfArwd8@e*iVv-B1W%{Lx}^JwofIwPM*>m z9PRltk0mvgpWTqb9_x%7;|iw7yYHlNK)IoQpZGcKEN+t$>_yeIs865sz)P#4S~(FV zRUsBENAJyEXJ==m6K$O<{X4fR2`;%f%w7;T9U|YLE5Ea=v4{?$R?t1-k+Cp~FdMX* zh4e}Y(E~871Sw4t8m^|I?t>aj$N8jo6IF#pfCr|>`IY%99Ck7Sw zZ37zEH!@uYhQx-RWBJ)?Kgn|@n_+d+%l}RUpekOsdM%}UHGeB(eWc>Z>Ud(o(HW20q``W(ceKKDf_^(-&N~-7WQl9q0Rq zO*W1?eX$1y+;%MM02$lP+vvB8n3KH+i}>YdVeyf;CE4=EwBMa2)!V-|V@)iGry-rH z@TepM>X&U4I(jyHqx05Ub7rHJpl_+}g6+UqK%oO4e*{f(JU=wr&e-~isiYpox5Efa z2=dv7ZAP=#=78yYleHy(zS#(7`o|nemWP#zMPQ=*XcC&t-BVQc@~aj)l2qaJNigHq zX}`Ct%u>sT1}6FqOoCx0_ldSm>dUnfb#=)gYPXmHcFPd&QIB<&U#_Tt27dy=-c+Of z_B(<05+N85t@=1Ne2uq)+Kn*Sra}$c36#bWdgll}FQo8$j^4z(S3*{}K&=Hc|30`X zg!XTv(S`V}u8;nz0wP;zEU?kzAe*}$AFBo1NQ0++fd1Zpm$q5WOe zhZ>5BROE)GO>!l<779X|Nr0?y$3D%!^8-dB+Ow5{EvYVPT(2&dHU5mM*gLZa#AABz z@36R&QF#wPiBODm;jnlJr#k_F{`^e-bKITzEI+=#wwL zkv{xn4VrepMV%}u*i5oQOUi@e9Af-)h+js8&sq;}KD_9`d3D+m6QCC%1MAwbk7?p* zY`dfm)A%{d05^$OX3B~-w8kIMTPz@B+FBFmry}Av^$QT-Hq|pg@?zK0zFw2hZgC8<`|2VZoA|4n!ouB^Ux>LSS@%vbc!lNz}8Q%OuGAj^4BW-Iq zC&d`SvyV(cAYm*%h+zQl5sHi&7(l+Rd_Q8^VTeCk;4{97gE@i(<{?^1&0Nr)^8upq zbxI)V{Ub5W2kc6oxYGWrhX3Syl`6_s@8{Skj8UG_>ayP zsD*1MqVVLqk(f7f&e{UpD~&PVr@oG?VG0khpkqW-yvxb=Dy@>I&smX`J^k8o4pl)t6|Xy z`CS>lD;(6bLM=_&UST3{7*roW{e8k8>^fX-YU0*D_gyIgD6!&+2)jb5hW#xO?<;nS zR%F#SiH;1Am)^ZrFc3n_EyeIcLa*{vHv`+Pyo>BL*DCHz1ssEf_5@o#7MrIsOvA-_ ziE4WDJ+KGp)klX~E(iZSR z(cPIA(7p>QtID?Xd$eq@+4em^KVqlT_j$_7`Bgr=YLkk}Dr_yuM|IJywLu(XUt+a{ zrZ`oj66in1S!6_&ktJ#h;(qO16D z%RO_TK7oqUqdRHOEKa?9n&Vol*EcDq-I8SYLSE7n541MiXEJ*u^N3dji7iRN0r6A$rQgns(`_(CAOX2e~FQj5# z{a}L@5MAFBF3Oj4=5*8C`b2lkR){#YVX?<}s@T;mLq=^0hV1UvTyy3#SVI&evZ^#`lMK0LhBDAB8$nS)O0Gx(sI-hMV2QjUi>NQ4v zL)RfHJ_;B=E24vLM^c)qpsm@NKA3lXV4YAS<+V_*ONR7fI9sudt0_XGOgXgEn;`T~ zWS1XibyOW33NV>W%a$?_bOKPar|(x4Qltu{w1wH4;eaQIyF zhg&!1*b3eZ#(H~RvEGZjjRrh7{$73_`W(0?W-qeH9|AnCL75F=+N;x_x$MPAUrt&% zk#yUv{iThcIlWhR_0m#vM=_w=$wVdcr0Lz5n7nr6TXQyalC^?;+veP>GdLNpPWdY& zqv>8B?`maAK-o^X_WTjpjp1}=j;hTwC(ifgCLqh@hMvq=7}2Y^SujkRFZ7RoxA8)X z3K44Ic-QQDYZ_#ayoWvf3AEIHDIIUUF{h%s2w`=Txf;QR{Eh_=DKr^>f&Rcvr3wgP zT@llARpl#%mW!@U7qu$w?kXY_8a%{h4@am<=nVaxzNA$#NZtjKdr#+}RnG92f_4QT ztv7T@`#pVQU?OzF=4DHMqgJCluj38;O1+xA#^6~rLSBLDv1nnELc0qfiG%UP+5DOk zHq&sT(hCX&UcZRD_-_P;!lE*6>=1AIXymP0W|F^%i%u;$&{cb$RW0N!#Hz zzk0Xp3HJL1SW+CxWDD3y_DL*0=ppE9C%sI1mKu}N52?Q;-oYd+iGH)~U#*9QBdMDW zK|I^k+JhJy9-X3f1#l>wmNeFWZd;Eqnu&)C*PSSydOCAFYmf6F;v1a8;oN^^hgTo^}7M#%Lo@ zj|OeyeB58vEM@sHXVO))$^!qj1!MAKcZ5n6oPkKmKEu#qc5q_LuiSXNSK=m_=NuB?Jm`T`O2N@1 z=S=e{YxpB)k9&U$-uOs?PtgeSuOTYcOPo4OS*00w?iOvSGV~s&XA!v};f>)aF4XuM zO)hYOhOQGLfrda*fdc}B&pSswy-v?dJwEKDt;n)6o{W^(4Tbe(MM{9hw~a+LF{j+d zlxM46q3^6em%c-O=SX88cX~9H>`16!_;?66UV&thG|5?ACJtz>eLDXFch@qB1L3<1 z)vv?UK}(lX@;Sa8U~he}*R2ml4p3D~ZA+st-Lv+N(=~MOs_BuJz7>yWx%Lk94S;pm zUMbwFax>D;%x~eGj*K7vo8hE@rZK{D5*Kv8-C*&DdPYZyWXq*%q}ql|TWW(sgWblB zwRE-Fd03XV;_7qHf&m-vcrtusN=NN%kWoV1N0z-y~VV6{9ob}eqR@r2Vjc~HU z9H=P|e!C=+Y?HpeY&yCF3e*ea6fDOl*s@NCz%CkiENCHrY&n|M$HYGh+Pb$Qqd`wW z;V*rh@_mW!Otw)IJeXniFS6dtU#ST*`caT#3h0%&Rd=aTqXbcxHp$_Owk4vl3bC-^6~%|{It*QW$Ztm2S17aWK8uR_FA)mr8{cYZjrhegh7oo;uM&v7Lr<>v@PdY9Kgx?_HyFM+}kl-+;@M_ z4abP%MzQQhE_=C98w^{dOxCY3qp&;3wqN^_-BShM%yUH2I9K~8fp~kmcM{Esp85q6 zgv|6zmtZa^2%A90?{o)Dm~;DAS6;13#C@GTL8hVq zUiLBa^AknL4(Re=kA{i6{jdrq;f>SCA}nPD$SF~$Qd5>a3e{+w#fqN@dFcUZ+QF%% z7s?Kc>}H}+_xX4J_FmGR(zLTE3fB)UwN3xK8Pu$KShEY5r|hGrUIJ8+xcT2K-Hy4> zl~w=ktrA!1f|SA%W*V^tt4ItITwLJ4lbf14QxSD`4f@`0Y4%mjHnNVCbG*Vfn}Jdt z^6Av@ReRBQN2cDV3(X?fXTmqk0d6=j!ia>SnwP5?5W&m4uCo?vqbD%_wOBT%K_Kix zj1@_;Ve@QYeqr}a)uTm23Of$A5tvR zOhq(KwwDLI^YbM!6U#U9SlufCKKJgmkHakAM)30bcbjRjwK?CQI-e0*wJXV|)xs1U zh&k%E>S;AVM`5pz+|B z&nM^dT(t#uF&LMcqdo6zj7i-?HSy~Iw!AdD$J#eE;i21+){@`GyjF30U0gZ%%~s9- z1Z_vwpcGI3cM88~+S~Xin+%M*Hc{FNmli%a~q+;#0(ZC=ELC6eyyDe~ekV zhvz^7>$lg&IXj8snS+s$1sghzkz|;z%q5o{L`6DFwVY@c=nT_^P%2V(OYD+^$F3%M z*89M}6oe*f1IZ=)QB_lT z_&c#EP{cE@GHk($pSbOW{~QqnH;`y$y_7pYKDNby3^rTy(l7eb;Et^ybi09}-%^rX zI_{J7bEPi-Sv@w73ff4dJX%}r9vXHzMz$M6U<5=!?-D9gybWU7ew#vvyo%IF$-xw| z4D_C_=%PGzW<$0R;gAZ|jQehnNZFk*Uz+o0-1!C4P5;TctwSHm=(I3Mfu|TtWGsYE z;29x(7*ofCONs^O(VSFEU-4um+ku+Oc8r~6F|jollR(J?O0aBAkbxn~f=;3ria$dA zGJSabmLIq2Z<2Ci9Cw{!wwch`q+HaTw;vUnpl@gGqEiQp#P^6VF9{Mzp%^7RRv%w* zzAPqpUO_lHRmAny>X!G$^i(N<(j-dC#)Va!gD2f|{I$J;zO<)cWUN=Vop7^UL2AS{ zH8x${J$6*ZmgQLT!iSJaA0OR|5`D@I787y(*b71CIFAurOp`&PN8nd#kU0%)39N~^ z_nUH*_fZ#`5GHTRwujb&*U2Mz1q(!0^RA zwXR?OG@d{Brj4`)dg80;dH$*r^)7ZgmyshoECm zLbn3lYc4zsC4_2-UPOyL#n|?f8QA1bm|O-f!gHgYP5itcBGTzy|7~)C84M+B>rcaQ zNd*&Zu;;us4`tb_bCU`XD1jFsneq<0*afL%j z5EU3>l51%qRNpZuBTYyV`$;r*zu~&#RT~9Q!R=*U(;qUhE3)r-ITL^zH@orxn;KU5 zD^!+|%tw;3(L}ykd7dYQe#!C|Hk-bHC782N@VtRDUY=03F@|3|FG@k4BuH(OAYVHs(M=d>%2^ zRy2}Zzj=La$}`lwn^pS`9;4z}3Bh!zrM91_uUH!l?6Ue;kcX9<3VHMlxhhIXte!;0 zqxJ~2xvrHaC=cjNXI811E!?DU&N0tkoQ)A--f1$A{3+9f%LoyTHBI-0ku2S5D5imK zSne+aLuOMfLkNq&Wx=jp)1$V_YlR1S7n*KUFf9Bcwj|`iDT4)bTs~Z=qU-Nh`qlGW zV~ah81nZVd{-K>x6ZT%21yUjf;O4YpVbzd-wnQh(4?6O4DSQ=C`Gvd z_UCeG%T|}pSUFSGEqpzp88E6H#Pt`hvg_KBy#2awE3Bvt$XtEspK;4~go+Aq(=7c_AkH6&bqVkQx-(SZQ`5##C z$h6bzpfL%WH5u}TBPX>wGJz1f&hVo0qFILDN$w)T1`D;D$tZ1ldt&=tXewS3`YBOQ zMD0$;Z~#U_{M^2cOmt8yEyc~yIJ2vJ zC!RI-W=E5)?$hYQtuo%lW!Ld6l!8!xCucC4Yu$~4QLF+6+IU?(^cCwg@2mB}XO*4X z0Wna-O<#@kL|8R>bF(>2bCERvQPD!#8_X`XuhE#LMhLyM;L(x*WwNi*@Lp4l#Hh(Y zHZ+myA4Ay5DW|~&vkw0V7UOR3BeoLvzC#8bt^+v^j-wiCJgc3dq-P(WY5hjzJlo+| z;$QO0{)C(pH5q{s>hO(lQtCVxyqSD|4K0+zL*&Hgy+wk7mcilJS>F&%8`8&nuy`?|VER5pR{hCeKszM31R(Bhi7H#t5uJRKI?!a??2c zxf|yX*^0n4-5$ZakTMH$p<$5-MuF;|$-ivafo8AJN(vTivwEBi-LZ*)h55k0^ zNe)bGqQ>xiCbx_V>LJUax%KVL?)#F&dTrHh3DKp%pgWIw*J&cIb!tu6fC293@KU;# zVeB^+!l*kzL`~<=V?6pE*)w_mBc145z~+GY4VL#T#TpE+N!G@#=+_ne{4%ZcX%Z!$ z90G4=VGLjrd0UFy0gQ}67BhD|R%(7KK7;vhD5`Sc!yCd0fswq`D1*y$oy4LqCp9+s z$aBq@+SM3R9U;pjE2ZtEHZx);$_aZw{b#$Sm<1YE9nCQ{u7QNS98tRri3|Iig(EW$ zch%>eEp^;WK+*b@xkJi;uhd2)dLNa2YP@{Z5T03zjSOrL+ix&MX=)c_0z~vj4H?)PecZd zo!|$>xg@`oaDG_^F3x75fOdcRENqIwK-%|X;!Kx zK&1m%8FGH-F&IW^B@#rv{e_j&6A=}QKDMaiNF&msAo%S|6liKRz`bLX>-Ax?BV$iC zd|;|^2G8+*lm2Gv+7cEsLdHMUeM17wjjd*NqC!}ZN>XNvgj^zTFa=EaNmHobN~@Xg zuG{EDIA+4#Gc&S#eoxK&37_@XT=eIyrD?>5n~6XtGR~r}-8U~woTbu2LT*E@@9#3a z?!)$|@cLvxv@orEttO5x72@urU*rlkSDn1{%8Hq8|JuY9eTZoM`HGD=O;w$}FrnP_ z7QL<8J-Yk;y!;q;O@74=I*I<3LgF)sMi@Y#SDE4?tpC2ohg2J;X(mBC;k+=Isj4}N zKQNINwhG0^l4MiL0X@L%iJ; zo%^XY&x&uft9;j>j8S}S2UT18*Da1cT{w8qqTp#qaWB#bE=W)Xs6VF{f~yUUp;q$# z48AdIpBGSG0+CHK%35u_9~_=Fs@$6iGWlJ`{BoDSDfS3v;Vi zw2^eSt_%yK?FwTAHe=4fs;jsQ9#!OFN>ZvHmDvLwR0_@Cr_=Yyre===kL016--ml?k%FY!}d0t%^K^=o6bO!Fl7dv(g4H#osqUrluyb}%i)OE(q-SVlo3Tq=Df zwrXLX_k%a=F6&iMNG?MNlQ^+GP-@|ZDSNieIegTMg)Pj<#thicsZrNBqj${E$qHFv z0?OiHXRQay=!dXLQvK&tv)*xF=vN4KwEWaDQ**iQOO)_;$QH3MQf%n_yVCe7<1t>O zvspupNH6lty!kH2pn`{w@*1O->+Sl!t3bWyr!evYT)yWrYX@kVpk5S+D?pd36CDz# z7*%atQ>AC)0YwfFSD7R|yT@E&ULiHyt85;2_{AFRCmYcKVO3l1E~51W-@LAuxqZQ^ znbq#(fC1&ADrjU&M`(|wGBB_V7ZTU=zn`@$tP&UV`LhtxXJD_ zNE=~%9z?!GhPzQnMX&T4n_tTj`;WqIp=1~|o&72KOTxs}H?N;!{EQajGN5-WElc58AsXz{t5*j7K6mBNUC2CGz|RnqWn_iBFhdzE*6U&0dYTnex(Ja&maRwo$oULAEPuqT&az*3 zNT)~+26ndnp@U49ITY)ty|PE*7;X*ZChL{oreU@cpug*a?1#q4>o0xx88!Slg>Vr3lJ_n@epm?TfI zfwliOw|F={AAWe8zQg(kG^O!@;Y(enl+-WkMgTAum>XXuzJM2?;CCW1yI*yStj>;5 zHduRkl@d3!A`ESGcKX)NV^&r^J`y%a2bNo*g$Zdb?j0tU6HKaL!Lp?Dne}+t89sQ- z{#i7_LTV;f&LWQB+ufxodJt*GF>VHRc z!@3}{ok;g*hU!bpVQ4PUGW1|D8wadM?DN~LIYIgj z-o-5Ws77dH@={umY|bU)pLJl0l}i=^MX9W(-!rlWd1EW?8veTegg|pA8}htNtSyk$`bp1*@h zkBP!+N3w~F+nu5DMG1ea(AfbR4AiIKt@0^xqqrIid2qZ6*l>$1B*lEWKEM#y`8WE@ z9F!;DxVGH7N0#p|K$%rHdx!YNIGny|}F$^@KZ6DT|$GBSJh6@e;JlQxCKwa?!;uhaXmw&=*!&S)Q>ysN!$0yFWCUXJTX1r z{$5}yWWd*m+wm8p(k^xd1*E_ePA;C_sS6&gmz1-bfGO|j0QMjS56sJoUb1~S+$DIC z+y;|oLZ2fgQ^7PzEQ*A0``y=p61t~r{ofr^zIT!zjv?`oxGS?)v!4tu>BU<|JrT#i zexE7bIMz81SzulZgA06`_I(%1pLA?`!AxrInJT=zOC>i&*5L2KSNiXl`@Vr5c088% z@FOWl{|?HI`#L3mRUIjP-q`1mc})`h{CkXKzY%`P)c)SFe@oo16|+HR#jsWjx0k22 zkV1gj<_XFl48Y`m8t}v4GAe6^nuds5UB9i;mBrs=#nhxa3Q7Wo66XDCA? z)DC=F9A(TU+1v&h%luWwrSKL}UC=8u$(a~0@{_$mXBl8bnTx%{G+Xi*;h~u1aVO3Q z`KJ_iZKDa5=gDq2z6bc__~O&MlCx)D&6BpS$+fy(fIkfKvBUF=`%m}WTSx)wmX5$g zD+3oc3l$FIMU>gjXOfV>3iZk~_|*UnrmHJf^NEzWZ%WQwT_8kF4OB&rbl^ZDqtt== zh7Iq&RW;cyT~p^mv5Fa(5lhO;T_#Hpi|jHdg)vcLFDrhK@>n)Y-jLhQFW1t{9Lpx= zSCZY6lx4wF7|{%>jqE2ZU)ab44*eQ}uNPMTh;&RgEG@!}78yhFXWb&YoiikSlyhsZ zV9XqN&eQUccho41&z1={y<{LfV&U`6?)V1q!TYlP?i_l{%&O^un#Sdl6;JJ0AK4#S zFsmgV>&7YA9Gi?j$k(16D<=q)8x}Nb*=?SB3Kw~yi?=F)8N7Nqj_tC>j<#EWmYX&c;DP%bLPP8yH$-c{a{{?75hhnbD3x-(@0lf(8) zsA1FxwzVv|WLJzxlb#)R6R=35$#LKz&2~3J=-?h5s;i#9R@uVds-Ye)1=H6HZ2lZ{ z>{$vw)>-38)D2n7_8MZH=UmHRu*Skuz(PVk;M(4iOJdDJCp}O%@lr=(l_3*wde^^E zhmIC!x^hQ!_&K#^FrMaNE|GI@89q1tC!!xi5!2t>{p*i_?dmva+7;^F`9Z$+cMiL? zvA+s z19cNA20L=c*vMcAsfa7Q-jZnOcaK1ceRX?%nunE!jXlj{it&^i+b)P8I~>McZ+kT8 ziP}RxPil7=TCz>8XFXD1^!Nf74onIKS!}0|8&#Fk%>AR8-h8SFEH^z}JbN5I%*)=i z4>}~9%p1-)!9Tx;xzCSX^8;`5lMdQXGetSnlJA=A`Ki;itH2;5Z>5P}Ox0Z_L8>dx zXvp;?5!mQn;`?Zr#?MOa0oyUejq=a6ZxxdyrDs`< zlS@$J%|>?xUc$vxhEC;^tHE>iwVkzX?|aKOnB&u%hg`2)6;kWIV#M|iw3A$sXO-Bs9nk=~p751z$bHRc;)W|jBGAqtqdg;(KY9>lHvS5%9Qdk1i_HMQ{ z(c%}f4h@>E;U;gTQcMra6BTpk%a1Z>Wl7ZuCf3e$;?nErv4OHXLj@jI)B`Ou9Cm<= z$^@*tFEBcAXO3wFRRa}6q`ND|wwi%P&iT~zNnl}8#IH`|l9^uI@wAoSez+L)V=(iB z4C1S3_`>4ev-s);fvYr^@y-iuRGV0LF&GIx?0Cv0FtO;dc3mYnU#MmAKC{}#^wt$p zwEf((fvyF5<>4wFa>;L`A*rghmXk<=ny0U)(k}s4ABd1H9ztNvSbPZXlQ8J=h|V>? zI%{DQT8(4%arJ0<`MvRvPRDkef?uWh%>E(y6s9L5xT4`XOIvhLE~qn}hAXjWDIOgS z#j0zjiCeDoZN4S##Ga_seA4MY-#~_jkax{qiuxK z8J)QiIc1PP-rnN*7v#P=+jsJj{z0$BuaOU{Uy_Z?!L=7xhjXSV4Q+Mm2r}KxD$w_F z*Y$*&Ag7d&-f5hGYD@w24?#gL;>}(1 zc`XHot)!YCAw|8Q!hCbq+E~)-vEc+-$AohF2*pt{_ugrX6SX}yjuW?3Nu}c3 zDQ(R$?W@^=xp@|Qw`fwLMzxIdYY@#&v<@@9Qezn<8uf7@%BHnENwc-U?=DFde<7JD zrtMx@dF?}6YbK}d37mx7Lk0SKWEBUqm}gXi1AviQNW8T#BlVUztdW76qzKN_!*#K+ zWoq<+55IUAD?I4LA(AsX7v$T7N@gg(?@^g$MqN8I0jGK3G8aV_zS9os#tIj(vfBx0 zS4-TYy>I};1l;9CYR{(3m%^F3-ozFevZcT9eSn%?6%iu@AhsX>JQrCXfn>sR0HK=~ z&@dtQ?cM#PHRU{=^wLBuFZb{Kl@2oYbOc2CI~(>_#Sz>embAV)Uh~c0rgUEir#fwN zAu;&1&m<6<=Bdm5oiM2k?(OBT+BY73qdUAiII$LHo&+fw&kny2&Z4C?z)6wJF2_A6 zRyK{uv{;zRblGq|lHB3-;`W;x=jckvzx&24%(qJx->$PnMTuQ~R}9Ha+Js`+4F2s$ z-NHAxGz(p&QJUKLyoe-(gB{7%-hsB|i3>!KUn6vVFeseGr3^)H&>{?$41ohSTH9eS z9cJsreDWjKDw`VcvdCXcmg%!Z-&mPQZW{aK0{dbL&E_`k`fxdKpCkPwG~)_A=}med zj8o0i(PS--WN}gX(wXHtjJzyb+~-BjG7ks;vg9&jjVLyRv+Fn_;~zuI{s760)4R?b z6b_|G@0R1k{8!L&4$IA9H_EaRVj)^9l zK$MB!J_X-kR;Ms}EwkC`p5|;i{(ZFecm1G&C8(TTTQ8TV`p{!>IuFXqPFE}mH-PcP zY((B~T4x~{)p1r|DXh)00=Z?Y-G)?;7b?j=qDhnR5B|GV_(pw7x?h&+7>>< z_%OqRHzcb&ZPVxDzpQBahC&|EQF=!)l#k(nT)Z~Y<-rQJAL zLVaV@a`oTDLCs?;F+U2rHO}3+>Kl`+elp*+;y*(vRk4^07q+px%9gF$x)R7?BqBf|;B8m085+1#b{Q=&>C-ZEajNcP|Nb>wyY0SO(UA^p8 znKd)UdNK!9Fxa$M4pu?(F5}4)%GFKhSTrI*{FZeqeu}6?it&Q_C3kxxOTtC{ae@#Ag~RLSOdhiK%<-j$NW% zyqXIxC7123EYD4)2bL`0Ip-pKWF4cL$yi--hUKW&VQ+5d_Vkl?Rn*>3wOYG6vXqEBB~ft{PtdqS(gp@nwx!xBifsT zWOITi2M0Dv;i}-#FB(r;>`idV4ouG^tB%HJ#^^RZ3AO9BM{gG9jdsY6x2idMp)sVT z)vUO~VKka*-ka53WVF966t$@h47zw&d!>*i!~EI1K-UsxQaFOprE@}D(P^a#S*Hit zxMRf1x6hPADmBB$uNi+L&yq>(vJ~>@e<~VV{s%qqEM^0FBqTh>ym0nb`a|E$ZIi|y z4uRmS{S2Ap*S645i7d>2+R&NoH81GrA?nOo<~;+7<5*d3duTIv++gLVxJk|Nvv#e! z_e9=`Q|c1*{IQK6)juuMuWDv1!3WL2aNUXxw+~+t;Wro|az3eND#{NuKUVz|V~{Rq zjXR%Hmg0A~)Z(}vYKPt;+B0S(t%)C}sv{$l#G(!Lr;gD)K$k-lAVZUu$qG)cfM7MZ z|L*h!9REXi09G_1PdHgfhSi1olgJz3fkvd3gg}KRuyqHo&ym2N^vChF_J#%8y;|rr zUyf08H**Z95!a(#=^!;k=ekieFNPP<3Yc6{6r;SxKEW@nb%-)HLyTc785v$AG)tgJ zmjtX=@Q}6ANFc}ijF&MtBtL!zz*%cU$` zh0z($Z)i-kB^pdd8#2td=|{vGhjRF4Ae@6wgHvbE0Xp$Jd<;2g9Y1Ge+egdHUi$0WP zRcOR05z7i6rUgn)q7tGC~zU8a|pm%>|G6Pc!0Gf#{6fFm#zxiH} zNNHV4_cOn`7UCxR$=i$g4VcIA?P`A?-ZayL?<-J^Qkdp&!H7H<@MNDy1JWe71%>C! zf4`hC;WF5ML6TUfuLiHuPDU|*(~9?g9XluI{|=5c z<7B(}>il*abKcu>Za!~%@@K3t$?DKBY*eJ8Pq!E6yjc;Fp?aY_#OXf2K~T(YKZRpS zOvabd;_?Kn7_Izg{1ECCr|o9cNrZ?yGI&W{4BNtjol-qPE(&)dn6JWwgt4@+f#i)XA{YK*UFvOSp4IWVl$( zMAqRMmpWjq)MP1?OxO-g#ME&P7~sngeYTCqQx7~3KX>s1SDcHvp3M2l;Q5|dc=7i` ze)IsbfECw9nlW-Dos-_9AC98S8@e+iZld5!YwvZaVx%0ZcnLDL#ej{=(BsK9ZzOu9 zR)$B^;Ec(0$fDh_U);M7Y})iZx%)B6r@#>+J+oIN)WNMs)!5oE!l-DYab~g&$}2^M zg`HCR(N;w4$59#Rb#@~U4{Gp})$lD+L7xU83&l9X;h#g>zVmGCGGy*SgoB?_`)`L%%PI_P{Lguv_fxlrza#S**W)E;cvUjF(Ka} zZ3cqA5xS)<8FN>d5q+!m5$yL)HCjp{*~$WgW0t=Kl>dk#QE|U1l_5!=8^V!B9NlMN zdI|5mpcKT_zHKbr-pRXH<4;@*(Q5m&&qP>-Xi{sQuntV?6M!VVSol83W7)^e8bkSYzl5^eG})d#!uElC zdM&b^f;EZVSeMnDpaV5h=T}}78bFX7@Hp&mKMO}`21!u7!kmwVlJI&2k}Xk&SC+$G z%`T}bk~~5g`XQGYU8v#YQo<#Ni!YrUDmztO#`c?uh7~B?#MZ&@*NPI>1K5zCg;#tS z8+0;w{^>`~&k}$r9D~C5!rH7RNrx}{t@a$MvTDfyvnR_U70D4vH3JDrlF{#$ihF9x zQ>y|5YkbQ>DJm+oa1{rkN~(nW#EYfu5qp7QnoUI3uXF1kR$~pEq?_9qt8}2GhUTz> z$YZs4)Y=OU&VdX&mO9C)n>QWeOXALN;MikwFxlq-IEuG(cewz?IhC zViwc*_h5IIDL)ny&3=1q*SbS|Z+u>z;rCNGfYuupf+<}YKSIV>+5Mr?^tXct{hkp# z^oF-h@AVQGFuW)R5?a%zXMvzS$3w4u2G$R!wdbsIv+d0z?_d1r z>RH4mbqPE~(kN7s$XTZXHs=GlY)9l4}g7$b^WmM?P*NBERq6ya)HyG?GL+PzGSAR zI`Ug6QrW{WfPIgRunG>(c&9d@vPN7*jIVA3f4JO%IYh)|3+TlQOADd^J3z$0g;7*V zLUQ)f!$FD8JjbwDD^BObthWLc}jv4t{gy zYBPSNnNuOzWY%VF=h5coN^?)MBAQAN^%w26q(LcNqZ&@$0U}Y~^!6fv<5>;cP-e>00(5Crln^;@7zoB$g|n(P7H$~XhvnA!Ba|YX zPEy3*p1OcGzM?QXNh@=-SL;C?Qxb8T^tyi^+Qvhq^`ebr*rFTPM*@|Nc~9{gB%G;6 z;%p=;MfpNX|2>S06cxc*MP5vAQT~j&aCU6VV#){wJ{twIM>7sOTRQE$%I;_I1%@BBYF3xeFWNaXF^{n0MG;zEI{taua^M zLrRTHR)81nvdg}w#s6E*;Yq3ivp+kXgzmS$kpa8&#}j6ebza9h^GKWOink;M!?VjD z`?hX?QpwnWoPkb9TjaY+rWuMGxtc=MldZMMTWkb8f6He|y0hVBF(~t~YUi+V$Az>3 z_1^6BDxDbanwq9N#cGGzF_gqVs+#(B>aLG()^y}B?BM0POF?W!d|e04C%uIXqNe<3 zeRjp6t8a~Q?_a4#J=1S?4pk8d^2sz_n;%uTSiXK*XgcB^tD4PhM~Q^CJd2>4JhJju zA}^!C^rYL_<|`26J>{NXFtmKoR=j&%ntkI8uHHcAlqefs`mxQH*FRq{*i41rh+Ak% z;opoE+0^BiKCOxkduKxD3-&(j9}{yPxs{XCx7Jhb%(>PoxO!g1fu~D998)>p^GPeo z+&V_8OTVW&8-g5bb);shvTrPe*C;D`-?X@ONTzK045}(+a~5AxzNYDN_3da}Mj)Q2 z4drK^X{c!6iOgTHf5Q+7O$lsS{; zp#ILUjUNhQe2=p&ub|87V_YuJ0wADr;*4WabgC7v&0SiPlcx%J{Q(~eCO(z!5AH`c zaT~Q)!ht+jZ}k=B0MZ1?P(@z;Hs9;>-@-|X%Plz6Y=|+TNdWBo=K@uzB|OrU@!Ly4 z*;-UoRHbRNn&lwte;H0hVvBEF66kB5k4%hx1kGPt;cdIMg=*%$MQw^>OTVR3t1id@ zi;;Ufzzl4!+L8UHsd_AotREFRR$;0m^f=`(tijd7ZuIl#0>MD|b0b(UE;1cBchrKw z{Jm`Ah=4sjUJSl_xhnXILuR-M#_RMHDQRDqV8t4pg@-F(=0f%pI3r*`Ua4HD66xaz z0}4R>dWp{1>DLSz>={Fi?W=4l?tSZka@oynkvlMAVG$G4*JBNtF zE`AFRa@#iW?);Xt>QlO>;K}YydWqZV-Z4(AtKMKzriO&V2#wAm&~^R?OVcoVQ`3}T zZI3ad|70a2`HVDv{+L}7`tEPa>^4fD0yAufi5AUFq00B7-12=NAL=K!T$5ZFKBRP4 zOa#tNRZ`8Y*cd3QOzF=gEvV=g@Ob{HwJ1KStoz**?RobgmD-64 zXJkOK9uOnxTCS8LooJea6@@UzwvL2GaCFy0{|v?Ng!a2UOFQ-V?vCweqb}VWfiZOS zl#^Q>CWbx)6rDJ zW3cVBFO@tL7)Vd+*}k7o^p?)7`sPN-(v6V;jQBH8ZMw8G|Cqz;^Wy^HnL$+6ksu4 z54geET1*wCSj)>@#Rpx4W)4?DSy1SMtg35jh)q zx`S)iuPUzl>{{_>zOzP@LaAxWt=dQzQM#)+@GR;@tp`7X_I7n)zF%p?O-x9`Z%I2h zJ$%t@1wo&oCRLqKvfew*@kotL@xJ`qckSk`>&(;;`*{1nd!b-maE>I>pAKUuX7sKA zo0V_0Yb-*rX)Or<5_VpROhl$zu!rrEVd#4v7OmGjYN!tz?I`Icl|^)YZ(;^HUv@Z21W230g|P)Iyi z6v|=0s-Zk05|P<@tWTYKMimoIq^$Kk{R z)LFr3Q|41B^B6v))Ru-i5xUTF1mn_D#*UOl+un~J`QoSxYvMuhFz2Qm*$>*0q=wXo ztKQ1<$@}LR66NP#SrD^Jgv_C7QTT4`4-e(vY0=Hxt8=aqRc>#bb{XCYZTCZ*FKzGD zr-le(K>W;xCyg%|XAb!rel;KpZ0@I7D7+mKZQ9!3poy6z< zGnn<$p$?8M%+w^xWN>V!U6GN?4YwM}<;v15)i1ALwqnZ}FdTm$kWmzs5!1xlI#GE& zb+i*i88@QG^UrT15*85<6-3fw?i-D}PHM|+!?QaRNbhi%sdph!Hm(5L>E~M1yTYCf z0d=tu@!~@ZKNvsa#&{@s{kFov;4h(i^jd3lq%g5>c6Z&{^#%}-+w2Wh#h!MIfpE=S za^IA>K&RivO*w`5n>ftmZ<6TZJWB5e#^NcwVgMtCI{=zu=WUs}hvWniE|ATZh_w<| z9nYCIvgyLvZj`7rm-x|~Uxk%D7AILVs+LgMfF<7G?kIaTz z!d$V@puAD$Z6$WfRllbE3V}+zQs~MWs8{yH%v^=8a*@N4hb)k{D0}$xDm^nY!f7Iy z+(vEFLNB4-@CotwH~v8~Ce%}x(U&tEr@mMG-=1#1`nXtGEp2o|tHCsg^L1no?~{}mx>4J}46}M*jAJXPs-C98k{-#U&v>R2M@g7fG?#uT`}w{WF1h03 zu9?(GOud4-)kbCB2A->yk1FQ_Y{KBh`tO>L>OZYLDLQr*53?T5p@ zZO7m#R2&ETJkQeYsiyu*S(yrO>?(0Meh84SpNn2PT#QB#;>V3c`~`I)VinbS5N3U6 z@LL}E(8GU!!j_th83&2{HV_|tSB}~C`(+V@tS&iVqcb1L!;-)2pc2wBUs(4hc_BVE zr%S{8!Rhr)gKSre?$n1;4)k?Rb~>CqWo_d_TG}*3T#0kwad}hzep7eea?-qS3GcD- zeK7|s^!AI<#dM`^5DvnMmJOuX9p^S*-8qeqyP$Aia6J<;%kgw3lNSw{tRHb?L z;<#a_#WBtzjkW2^LN7Ol()g{Jk+bNTCn<#UPp8zT-h4^{kZZ^x(3o-$X30fFqPSu7 zkQEA|@y;6$GC#VC2A9roC42?t6m!`3nK`8Jd>k2=DlShX%7%YanqiHKGcp2R|5D&~ zen{C?f=8d?kC86W#3DWJ5l0u*<>5RqL3#$?97jn<;vvcd&EW6WZ_4e`$0rfU1`p() zpI1Q^ago`ZiCy;6%;xo3dS@CTna`wMoe~<@FRWAH4&V@h?T_7c*mYUYJ6dg zE%VX`Ngi9_twO&f`P$&k-hdVCag3Mq^U{@-i}UGcUj2H?Z_GJ32Vq1cLG+I>rDKt@ zwBnTEF$;e?_9$#sTRP6M%ctXSk9|rftV(Sl6E_zSBK4hL(otFkEGP0#uq`M@< zMS(U;6gi3Gt5}@G28n@oMbvRAS-8FkUX~iNT=qQ)Pv^ldIOBnwD#ph>0c*F^YF;lH z;m)_qX#MeGYk>YiMA8dL>N99*D;lHhwV+dnpTp~FGo%>GaFqdlxD(roC}GJeat!rde!hV9D38hK z?_DR`n~h9mi>7*f(1b#pfA}y&I`lwc2!^UKQ7E)%5M7g?AVxSQi4joGt_;7oKm+Bi z#7&HCKCU~$f@JEvd^@A=OvVt7A@$G^ufKc3Zj!v2d>u!!G5Ishk3ARLg4_+((V$(W z+vT~_kj&IXhXV`(OczyJed;~h%_4J5q$|%unt#G_Bi?P| zo=_@H1?t$#IHhf|5ke4_9dy%fbd(GFrM6$=mz3T^lE{w6nmBUE)1x$}Q3h>t>hW<~ znG-DO_nLFz{`i}^{$jm$ne?+&i_ZgE{B41WT@Udwin9vipKX#6n-F@o+{)Q(Eyi7BSK zn{H6Kcx<#?jskRMUIy<~CAJfGWse2uiaLD-0gszhlOfU;RD@2S_G}@T1cmeX+3)yp zPkyG1xMR_Cg4J&{+R1L3A=;_ucRJY@W7vNwDX=Eo3uM8!BXAdJNpHJr;9WV(e29kzpJI zHoCyiO`qdJBsr^m-)rOHY{*(}m0d?VRbtbEK#IrecQm>WSqI_huXa2l@AAn~P}fbf ziu$BXxLLIM;m1xk6Z&oMue$g$+IKe3H#mWKL|^oQd({e7XMvaD-WmVGSg6q7t@40o z7m9{gmK%nrpH4=xkd z>*tFseoh%Hqs<<4*M~!y@kSx2>fS%o7F#p87)XoY2wTd+q&KOudltJNZKeOK>o!()kc=q{Ub2H;7X{a+u-VRrYw}#(bP7Jq2lP=>%#+G|=&}-Lm`@2bB-duRa$Bu`~lgfF~s@WjwiMu6a^Fl}7Q7ok6 zhr3aNP2o@Ft}52i9rmqX-o)vbMx{y^!M#P|*sBZ@aTjv4b22*pz7kI`?-#FmxI!y` z_a4ri2*d$Y9ZxH)w+%C#g>xn==&~(@&!d_AI_Rn?naPi?ESNAw>B^!jpgOSu1EkZ5Y3we!M|R zN1HI^NPMR%ho_&#-1m^+(ZD1TA*Cr#y*+BV=tZHK_@Oc^nKXMmU92>UvqK zVXBhrw_xR9o5?1?$o>%WdU zm2JlGN z2$#VX)_yKv`tLoM6av~VN<9I_InN55H`A)iuYyfNiz@B*o%Mq5lS`lv*@&2q7>7MP zO^gF2623IYr6uoC&HRvqB^r7~t7e&HL+$)BrpDyL2o0n3gm%Z#JCMr5HpJU)YDUV4 zNjiMTq0R`Gvd$!fadY}@Vn)tWEY#D^mq=KHXt=!z9>u@mI$!K14@fH!9lzV+z2xsn zgbdfh!wsHgD8VkhuiLSg)MuLSnH^HMF_l$Z#=WiisG^rBC{4p@-Mc5%iDiE5&znPP zA`%Q@?8ciu!62VSd&<{r{HUiPGbO0T)>Q~IKYdH0F=MSxgVAWe5FZ5@@$<&Nuz0rwOt}Lw1UrjnZdMW~~9bI_M6uEP29TRAi7eVWZj)RT-J72`+aIxTnC9xbi0nKf+{OpL<*++@q#m zex8CGs!gXK^t!Fw=%%UkyFUbDV&tmfg{!r1iNXG^K6pkp8wl#)urY!*FB^>TqcQX6 zSEtRAKlF!NUC1=EWK+l_$jNx%U2(z%}i~7A8 zuFfia@psf4tTyf-Mc{u(^ZIt&kJAd?-<~T5Y{ISaQGvKv7DRXX7COZ(BBB}y*`w=%xeqU7p z76`^E&?;uLo5NM+U^K39Q_rvF$ zS4J3?Exbv;w*pXgzX9dTci7qJJccGTcc=%FH|s79et9-Ey$LG(x3R9jkDMwuucwf$l;yJ{Y84zx>oUebvvgmJGiQmAtwm`n#)wzc(!K zPPbQDwfaPXhp}WvziSb$r4{R!;p%Vx}s021M1^^>rgiNF_*tTZn|SA;A!QJ zf~b)rHOe13L@)8UN=8*tF;z{m6~TUD6?kdGIJuUgJLW;wNRO+Bs@yBYKokrK`?aXU0A=p|TOvm*nB&)Vaex zUx&1f%2Gf5w?eo5Y2nYJyYTU~HFVbN#03)W@Um-zy`^~*HR(=b`~Y{bVZJmWi4rTq znFEI}Mu(`{U_?keUE!JY0E&dp)!1C{o3f%R#2Mv1RkPf>Yk-=0o>m<{A^}t`rggB* zDG*!wv>{D%gXU9TaeZ>Yi|snCayXr9MMCXd>ldWd_9O(?IkT~x`ba*avNFWQ@g_9~ z^*_#6A1*~hqzx-kR-pRw1z2>>$UgHPi4hZ@H!H-*bM6A#2SJQ2(LGZWEqQUv)?E-T zK@X2g=ezg-d})hlwh)T2+Oms+1@=^W)AZZ$mO;YPJprjot}rcMq43b0W0@K-+8um# zEBBjQHImJYfqU7J=kyKhKrBaL3y_$&26CDXekVekE!*@q;HpJ*5+*hidFsQ0BG2%T z5C;+5@4ClCUUER8hDU+N2i*3PUxZxLkoav?*^vJIcLyPvqORsQ@!^gA$dFDS$#Ovc zotCGVl6b@r;s854ED!r=70%YdXgpv;gt?~oFy4Zn1+HKp+#^SB26uq7|Z3XLHobkE*vP zUh4-}ziIt?nchwb;_)C&oO9Rc;F866fBS|`*;eHqq_8v=SUB(h;#D3NzsCox9N zM+I!}PYc{m0Yx-et!m>vlwqt8)!&h212T$%eOVcEi$l&6KLe6K!Rj!3XOCOd+B4a{ zU+P5oh->jHm%2H&2G4Poe9OJ_YU-@oR1IeF&xkP1h$+So=^;K zy}Mk^P68c+eChll#3{fD2cfu!gSZ__SDWtlkTMA>0CenpNN417zhD^UPuh;dBq|a& zhApI7UewiWg1HVkE655 zavTVv;FZL%6kE(t+KHK2ryupaSxtkXl=(BFZCCAf6vrxc)V5)`l6daRGis;UT5x1} z$!QYh`+=G450qh@-jQzz*)D2+q9XL_c3@b@5?#ncT%CA_t)h+JZGNb&le_IT3pLt= z6}_okns;Ctfgy0=9$(Fo0QaD(vvGqF+6#g!RCQzKkU6B0D6qR(TJ3RWBEJ+t{o8?H zLh8ilg^!Mmrg`xTlj+US6hRge4eFT(nAuNifWOrr+Bngf|7;Z2@f>lVaPJafRkZ;c zZc(m+t8snna37?-RUYMnLbl(>`Q^XGA`+4@!&#;%1NNDq67wW2tdAq2wfx-k0nOhD zK~!_DUh^1?s55fus#Qvj>4SN4e$ZBvnWY)U6zOza(Y$MA+`$PBEV8QZ!q{ih_aAHZ zgb`?neK(Fp8ec@iUGOmsMj2S)X7Do)q@@kqTNn@2X!6yCKuh;fuv~g59l;2R*7+?3 zUNJ|0Xp&m})-X(P7w5gjxG5FVEBz5w=E$YNrjBTHo*VdEXHxK64|}#`F-!=drhf6NkUu zEZ|cBBJ8O%d56cke+1WgQf}JqCeA|3d10NwNTa;jWvM63ZLRDwYfJZTM`!pa2{Fue zH-^JitL;2c7nf$XD8&}d;E`LUS62wR=BM+#9wME4wWBe=75(ZK#6xX}PaO{Lvq{G* zJ5;qW+tVG{(3@u`f#;5h3s6Kz;QR$&mH&Lb@`P1H;B;r0w2|YSb|iPt*J|1eOUG;c zHyVoWnGCP`pNbXT%1zSRiTH>fZ^Q&De_z&Rpz!c_Xlg&QV7~#Vsi&Yis7lL9@lSC@ z$&BW_q0k9_e;H`)3mzJ=302wXq*c4GXY~E0GUXl@=|Ff4(4iG(gV3kam}qcEcmv)2 zJZ5RejaguGEr_H^-vo%)XM0gF22;8tgjk*@8zrGLg2ht}Kn27YlD1UhFK{cj)#|L3 zTf=5eo{rznx%8aC<$=J=$NlWR3r3SNN}=IB1{Vgawb~a>Cz<*BpEXP9RnKbS9%vuqgBTuRa*b@vT*vyw-zg#Z7ORw^~I# zS534Vb9+{AI5#m@<(R?qV(L0kx-zjjwXkJu)}tLJjEALKHSC7C6S#xD=<;<NxL!T|M-tMvFVw9WL`o=?b z88~n(%KsqT_4{E}uZ&tSRWy*Nm)!|7;De9dP&ZRUJ6X~dQZ!U2FE?|2w_n9zVQ5Dl zqk{xxaMN!g!#QFZB|pv@$!nG62q#>f%lXw!J1a}2yVlh*2O2zFOk2hWge)j{Pj8FK zW`2ACO|+cPPu%_hqI-*yBg*XmowP-%=^`C1BG**t))DZbkpr{RdmD}}E0c@>^F!RrP1cSAQMd7~DFtNAHF-Y9l#A z@>OsCoTq|izC8F~{D{0V3{teZxzIZU^Cf$RYD`;KdF}yg*>=3CyS7|mYVu;TRiv0V zE_^Kd9?&i+RGCVif7>%C;`Iuk=;v@m)M@9S+4r;kSlW3Hdlx*WB$Zk@@gcy(g)CVQ ztGIwRi5BaTNi|An{W4fDis9)!sjm%tFJn`B(dXy)?Y)xxZS^ZL^@JSFS@$f7>41*uz8aHATX)GsEpoK* zE)$bnZozGfcbj6H8^}VPbRrnsuv0V}z#(d~+I<)W!eIs$;j=!ZxQ@D)B zqI|SR;-~_Pg*&0Np)xQH8$P@whc1qVUvN7tJRut~V(*>{7eXE^w^+q-%^rk2%($-i zBL)t-T+%{>UX6q|$KkB&67H6wQFOPO5AiCAZrvz6G0Sz0xe%R=RfY|C;Z=f+$VZx0 zlQWC}O$ug`OZ%1GI@(B;OO75bM`m%)NN`Ob_?IM?+Rm3(qL3y~9YU4H7sza&1U^4g z?iJ7kv1;R%aLGwp6G{zEhtjEcRqaPPXO>&sg@=DztRp_b*f6_VN*yY(qH_yL)Ber=0y4_0C8M59iYkAm#=R{_ui+KM4+wIEl!Db)@tmrW^ohRz z72TA)OOG^q~4Pqh&bIIVx;gTXD68+Q?5M4^_{}))*IOGmlaqenUugKhu3i)IZn|k>d0l~( zmPLuypYFRJvbRlDTTUor$9A`mUCl)fW@LOJKngQ69uI@Y#8n#mGENBC-|FA<#7rb^ zeXhrL)HTly|NQi9nXYJ}h~#n?5if`<-bYyqD%+TlVnxfwGKXjg59gK#U&qIO$Aw73 zGy$#l-|Xg1xrS(@Vz6v*IfkOdlgfMQ7*KGAs1xMkfd|>+1Xtu9U$>XEH zxcrTaD}lkMq&2?k>o+9Cd9e&U{NkqYULw4BfRh!_Q8*<##GUPQ{NWjtA*a{q7ApW% z@;2Mv&7jBW-hz8U0GC#fsYJMQ;?1`)L zrz&L?c!t*7fW=V7Hgn+CGXCkYVA1QcIlZJGb8kTta;(RT{T51x%us48R!;+yKvffV z*$WbPxrjoGwRL+cIdI_bRGiE2mBC$nI5jx;6$KXL-I~8`YhUa06i{#Qt&^|{W9B8A z7R{-`B$P&$#ScJ0{~YuZ_{j&K+;w+u1fs>68r337aZ{+&8889SMPb3#biLiHK-~CfhCl#M7VWVH>WS0G3;^5HE2O~3Qi5CqpcEDiTATtYPv z&5{BlZYc%wRuS^Sn<;?W(bT@7xHkPG1n^}9A=z1{T6nnG)$r1?#L=zzFAB=})}vP0 zUv0&jm&D@V=|XDvx8;3b1qYP!LN~Jl~=LB@Di{_)3o?C ze1|ino9Ott5?$-uCKfI9wRp*SmCAnfA@rE0D4@@CUn?eS^hto_Th&l6LyTN>Q?ntW z-OT%|N?NCdF85H2`flzC4baZNP5%|7!___a8_PYSM^p{Bp`|0%fG&<9WdT;Ph(xtD z3uYIzJBwmLoXtke2Y{Ij+{Jw5qLTw|(AX~-5MzI@MoR*UI3^YDchFOWV>k6zFBidh zCoWboipa(C+Y0M^Zoj$wMH{t37xQ1Ceh*xIW^o{5h;s*wq>aYF)>osSq}jl7seryh zd6MkoBx`2as_^*z&YW~xfyB1x!CZZPy;BWLmm=5Y%gRkme|MbA?~XD_Q?*~`>+j8y zJK@Tx`aB5WfDW=egB_6y%Ipda zLA4JSSB3ZtMDL*KxcXLjtYcqcn2})6Bj@)7qcR0OXK#lJi@bfbmcy5VwLHQ)@rlOMCF2ga+ zlHxqi=3GM)rOQL49443`Z3mpb+j`Pe9%i1jv-kW(rY&(lHN$@34Z`Dcdf-$W<$G3h zHyDe0U)(NC74a4E@>c0$H!-IA9viN6+ZJ=t%V_*`UH;npD|-LHVWV3{6g0;edyC9S ze9drK7=5Lma=Q2Lt*tOIQryuL%`M6hX7xdMG=L*J0aDzTn3)y5IYML>jnmCmlNBd(E8pmutiYqjHt^wajA>C01QrPPSUtt@6LprC4J)Qe)} z*~5et8CW+R1dX?kF*U8VU8oFR5H>x!N+fSicgb@`%-0?%-J zI5&hi=Y@tOtz9k?zTj(u!px#dgw&5n_Kf!Hfv#M1%FV0hlPE_IutFIYU)fn+Ds_CY zTz_Qy5Y>w~(&@XvcfDDO4Kzv-)~sr1$^0~kIQeG`?`w&c;ck;oMUzbX>KI>P`|iv^ zwlE7lD)vz<_SO%!W2GM@7=p?k76n8iliCI1$3e(T-N`$-*BY5T^W~E#+y5Q_J+g1x zJhaUBKNK!ZbC|Ckf*An(U!&%N1Z2;u(ou6UJn<|60vqV>7l??`8?@G`=#z2aPtNw8{{IDHPCid~pI=mB#LBZZ}k1+_400ht1jQ|`Wd$ogPh z&92-FqQ)EJj~A6-8BgY%ZbOy{m3z|F-Gd)peJNMic9gD;xhI&hw#^dZ_tl^H#v-{)s0n zy)6C=BW1AvB(ukNbHcQg|0F~ePG_c>qymO4+^pWJ8ZA%D=l7DXMyvdj2yPpgQ|6ou zSq4926UU+;yNXM*e*^gS*9V>t3=(nw=P8!{r6WZ2-vH_Q#*^5K=RU)OzUIaJO0do@ zqx?CT#$r(9Ho4z}!Utg*?MLBLB|x%!sc*L}QVckFWWkGHl(huk!3ar01Lcip?-_5P zAs;{`pK09B4CUn&>?J=-r-Lnr=X;hY{Aftv>Q2B~ zU?VxMjqb=p&4&jsvnVJ82NSaWxwB@uY!E#S>A$=rp$+?+a}4LwrZrzb>t@(6IL_)4W0(^kEjB zDlV=gj>x1JiN)=O)SicLc5(#WnT}s<)yttb6Q|?XcZJsVLvqiLFDC>>7PAeQe z$|bo~Y?GEd9)%ogbHXFcK`i$nbzkzh;{kPI&Uhu;Bi^qWB_mar)WC4y${#$O%ae#fo!s%pUzX2vdaKPPmo>VU$dtl{yV zc>zy+`4#h+n1g06g%VRC2swZi$>(R`FGryayoI?!zh}`Q(deRgZ=QdM8 zI(8yx-f;wVARB+^o?I9*cBj8Ye(RKnREWc>n+t*?V)a-lHBAyEEyeRD`JZIUT6uT9_@ z=~*xl9*&w#6hbK_G9|seazBp`e3*=|SQUpP#|5Ln2sT3rA`X+ARzefzc`E62l;4ZS zQUblSVl*Nb&Joem{&m*+<0ke^HwIa}Cc3%qOlM9rq=T!R3>7U%FH3+O-KAY?(W-o@rYtOU8w!!ev=p z5}^`XP<^S|tsi5Qzss8q|c}QNWWfA}2;l0ywdzW5jFHn8I*nX;1$vS@_G6nnDDzu}cD@MG5+&ai>OB}r> zGLrc_;4}3>`4zD#CCezcxaGe~HCCNf96rRQzgTQ|n-j_yJV$6J;(2R^o>=nGK)VHV z%frL4wnBQO&hL-cv|h5!T>YX0cG20TL4CFPi|AU`?HgBNU!_g}&{PIXmx2`=vPSQE zeZVWeLp(MLSK#umi&QlKQhktnVuF_w&HePc^qYSc-|V&teMz)nRcXx;-cN0(Ltfqd zK>F|zyS}V^&%;;z(Ju^wNN%z?=oC#u;<(pP%oB5*PD|A&59L>)iQk_0XL>fWNs2t& z2O$CiW!80MZiAwj=S|Ppv$(pjxd!xK;}Ns*>(g`g#`Lh9TKOe>yz@m93Uz}m2(0pMGbmqYjGCb(^NyP22Ib(Y5MYj)~+(X##oP7u9ntw_; zG|stJ_|K@du5tT$$rHH~zH2c$6~W!6#(3?L?S4VB?95Lf9b4i3u~VFGx<)ag6W8-O zT}Ma`@-$`r1pbC;eSEp-+N+1;-x_SGdBN5=Y2H{kC4-LfC7~Cz2Ev+OniDsMF6ddy zkC-m*b|f(I(n>`36Xw{I=BrDlS~SSOig55IYe8cH(xdoP!guyGaJYqZBy=40(kHHy zoe4*txVEnO^geY<9kljiL<6TdGn8VKr(KDIgftOT>a8@S&fH%mJ(QNhp1Pdpo^D-t z?ii(-5{K}-cB3t4sHSq?6$J;wlq${)d`O&0-mzr3d3*{XEa-JPCdD-I)Z1+RW^?&u z_mqOKv!%Or2FpmQVC9ag2FOUKaB6kTRxo(D!)ct~UA&`gORQgkvl^3M=k#}%bY#3Kn#n}0CLMGnn4V94Qihmqq|!^j?fc1C)H; z@n3}DdsXI2+E)IGjyshJx$`a$2AhzjSLpED>5!!6P>BpwX!u@;Y}{3??#V{Lyc52B z>5rd^%#VQ@c)HyFd<9dZC{hgLZ(j~%0pz-v{8FMYk6p$&Pv+$t6`Dr-%OoAo;2{#* zYki_92?^RD!Ep@ro2XarzQOWO{o*U=*}Rit>KB=~bi?li|IqOfF_x5v(;$DxT9V7y zE+X3X=;z>wV+GMlQv>7G4?EUPdDR3&xdC5>vzaK~7>vU{Fdx461M5kshoUI{ZXWBl z?YXIK;N!%8&}(tC$9)0ny3kusN^z)v-y&iGS!YN>p4)kDj$gC-%sIxW-CIP$1!*Hb z%xM}!4Ps@y?S8p4U(>3*E{wr@eQW*gX3jMT;ii#K%h&q0x0=f^T}vdBg2Fc$Vpp!N zDuVfM;|}m72YwiBi(12aY6x7s`EtJ(4LaS49f9i1)pScphLv}(QAf%k<~OW@`K}oU8i4#rRZjx=nc-GX;pcMtLytr`v>jmu4fJSd2^KLy7uFF4*N)-%&vae2ivedD=Qf!Puk8(`I&3NwOXyUH&Y>M z@8*^K+3NbY@f@Gh*$T54rYXdl%)WeFvrb>wioJfkDfx7NZ zOSR+Lkej`*i@Xh2>Y%q2v*-O4dUwvs-4c6_OIe@{n=uhn1Ynj~t5KXz%V^hqZ?V+G zcP`3&@?auC#nm;fe=Di{pe_62&ii7D$i$zd<)Mr;`z5tBlTjS;%&l;%YK9MvFMRd^ ztY?q?60&XpLqNR0YQjQEM-E&T{2lcL4H-^XDeTi`$6yvZG}p7uIE!I zjW*Y9!Ypf!^-Ch}%JTBrSf?DAv6GXMbZ8 z!YKY!io(R*px5TYYD0t)X7NkS*ope?9F?l=FWRWC2~fb3-kA&KKrADj0T zwEW7U%`q>-@p4UOx+rYwj^i}WM+p+ax%Q)9-(Jd z5OYZ^iT+3)rhmYO{KOucyQb+-0DbQw7g`dcYKY;=d1@gv#3+de@TGuO&og<7BZ`G- zdQJHo*3(EXdqt^`icI&R@=SV5Qrj;4yO2q7nLa6}H4biU{pk688#93DpE0+NJ6IDH zG=5e%(gocd=LEw(CF;tSduc0_qW4_3oAf$Nedt!E-Umhh4Zl^q#HfpVrFm!6df7+H zXaIusp@y{mNxUp9iq*mjxhQ;N0J_)gSl&>IU0YT@zlK~wSQsV4v~9LoG)`jz?P=?p zhq8xY^ZY(Hulc~85NAYtHUF@?1S-NtTXLk&1~ZC4TGKo@jY{1cFJ4}JE%RBJVkzxG zKT(A>Y2253=qc+N#zIew^(afRn9=G@@8F!BYVRgQ#y-P0!_U%tBu*7K7{ueoTtm0$ zYL%(|?gGi}$x!*s3XPn_?kuIpJjcxAuC2RBYv4_cJweLw#XD0e*aE8|W>Nu6%J457 z7e9ahK5kv$kiV_wkREw*Db>JrXI!THPpu`evsKMCFUMs~cjBfE(##Aqf5Ej|5^I@D zHHq5t1jrwsLUovzlXqw=i#a(H$%`C+7yYM$c=WBsB9IUXCKgKLgJk3P zRN^cDITFt0MRfG(89sRWxmcWLdjl`lnub4@wDXCNb40BLO-&P$4%)4KU4Y$uu9BCk zR$Pi>;TTS>nTY;MViFxQ@ziT9W}7(e*mTn?N9L<@FvrBL`u4$Ahgv~s_jikKJM;;Z;#W*+pBdu1u1G;O0`>qQ7k zHDx&{^v5Ke=NMICZ^~N{vAQlW1NYg>cxZdS?pWAth%2MlH17T=i>YVL4}x_^ff8k~ zJRFhby|@FEv7T(c1Kt*pzt`wQh%QfXJmK&d7FCNBE!WeyH7%%syTUYy_z7;HRjTlp zWyUVh-abz9GIymUahuz?IO+KEGvR%9vW+#(U~V!$dAgxrp)W0>yDl=5Y~pWIClklB z3lPutv#cxXZdiGB0kTE9x76;2G4^ZlICpV5%uJ)BLrmR7vpCJ_!67RuIdioU_~T|` zxTsS(zFEecEz%L1+%0hBlz*3j zUM0Mp2#ub!?p~)Y=0I!&>-AIa*0&~!SggaMWWei7^+>{cYY`>V6d$=HnE^8?Tq;mk z1}7>tiWMJHeO}v#^8D~dkvrbLT&P1yF5cX$;bBWxG<#{*z^uY-$|_yT@hz?7C;NRq zg6*yn@v3^745yb-HzpP0SR(p7HeQ6e#^tP^ZE=?!*4b_9mSyT1dp(wOlTrX7ItcaeoGF%mPM9^+K%vX+Hv)~S+eAq?4wGgID530gv zDS_8kQhR&g=uU*P0+-@-U$TX=f;+olz#pO|y)EAqG`*D2;?^yL7#A>L+2)xW#(muG za{fSnv66NOyc?hTzP$qmH9Pi}*?i0+@N{M>A2~iP*V@qmX1zqmi63FQ@cqmjA7g)~Ww*G-O+BjX znZeji?#*4RkxgSK;M*M^x`TBv_X+X-Q5=(+v zk+F^hC!WdjiHK*0`E_z3UXcCPyO#n1C%oqr5A?AmH}k9&uFnACX1ZDUNk76-6Vx~y zH>b^=+R8AOb-M&#vC?OoLZ)P394%X0P8zk(-&ss8Va}n;B>j%D3TY&JS6l^C_|vg2 zr0bjKE}h@nI%C?l)du~}{uAj?#a*}*QeLk#B6=Tx6d(GC=GX7Zs9#LHK4C2JxIFXV z1+HOCjUDC3dA?bL(b-r@ZYTJW?aa($XSD|tL)GPXwze``#3Fu=NDTjP3$8anBU3pP z(r`gEPiTKNo%V_0kMW+t?>jgYwgTuWTDO}Q%hQ8+bfm5p1^T<$Wvg$7!|?r%!wOH~ z>AQqnYhQT#Tktz3%EPnV;-oc|yZg1@@F*sD_b9eruqkR_4Xy%Q6Wz(@aR6C5xfqDn zvPOCR+r2LIv=R!&lp7Wia}2OiDcz!1H#{6=LPtw2E@&(F^eNS1MD%QM4984x@wz)LCWM-~3u{KW?9GO1p?QxbR`Oc4J z!6RANyi_J2Lc5FjJaGJ&V}r5}8jX<(b+Buvr+e9MI*Qkp!hNz`lxL16qH9da8W@?u zWvbdhm3>yK`a^75J3NuA4=hp8SbaDWDjPvHK|lrM-T%;4x;$^+0!+|#C`ZsM8Qc*@cdE~*Xu?@na%O1ujy&YZP{Jd9>1qAKges_9`~Dt2Dl zd&mGSanVMyh5C3P+$WKeGawI6I0q`Hso~PwT*?56Z5*MEcs?A6KIbwCE2KUXD$wR{ z%&Sx|jV?*W2R~r6az8*2`*aVhZqV%gZmH!D&1Flm?6n$ z2}w1_WE7j9PIQ#nAl?x}d)j@_e#F`CMV-!~SXfXSkSZ}s6cs}|_~UJYgAoumym)UK zl)YO!mrO@#%Es=DcZK;ao{z7t&98tDXHztwg2$wJSi z)u@>TZs1|xj6A96&tfEOY4U!5Y_MP+&cOK^b##b&sm@P0>&+Sx$y$Wr6K;$LHaR(j zjTQFAk-*ZJIrL?wavAGAecpk# z(txz)I$JaPxO#R{05*6JIJ`Dm_K1SY<@jo4uvBg|Xa0eF5mtkmC=_Qe5uFh{k_YAn zN+aFfH;>r*<7E`e?Y7MWX;wug4#v2P?_*tnDH9c1R< zv_}NNFZRoSHZAT?ESvPB$9->xpKP0*ZaKZ;Apz6R`{VX&JYyXsU9cSP0V9%VJZf^# znH6Yf<-gT!w&n_d z-JsJ59H!lTob-7HZU1m^N;AJ-R>JcVypUQIVZG9gWAU1w^?WmTsv%T>+=gTsd$6W^ zgMpggP1KM;g?&{(Y`&2URc_YeuZu2MQhXF)el-M*Jh@GFV}0|Pv_h5sJAqIpP%!u{ zW9#d7u-f!ORXr&M-z)P$vx;*|LoGrdolUbUIWBryte94sDBsb=vrfze^YdL5h)b7; zMW}Ko48~H_DCU8y@Vg)@RadhLss#$>B)S)9y+`?k`;ns6qK>L zPljLd9Y++gKDZ9i=u1?|-!;&R`^sOOYvSI{gytSM>hZ`X+J;Bc@8tF>LWgvciNPL(hH6^|lflLQ3_`Ia&+G5carZRrY7i5Hm ziuo{L__~|6D`>@}Kxrp_6n>&EPd6m4E8@E@Ld#v#-i%FHRoA|289U&Sj0nYi1ANlY zxm)2v#$A#B!Sie^2vGSrzwLIsi;2+^GT!nM=af#+ z#16Wr$2(Qz7hm9M_9Coobjzy^P4^s7jB^^Q?VZ)K=x=;4NZnh;M`nP+fZuj!;_+C! zhQ$-cnwe0L!;tt??!Kkw3LNAib9ZjcL|F!RDHqFpdY5GJV)^_!d+NFFB?~s{Q1Dh~ zgfu3Ml5G9*=$@`(HufQ@yM&s#vmxh-k38Ft*-oa!@g|AP zKt-)F!m^m%$T$f18|+e`U(3&$%vO%*YiBA~`-BHC9)H_SmFQmr4ZN`K2LDj3w{BAV z>n&H0qh4sTVzuv!6cq$J7oIxNqoWb1dtLYZoQ7@UAqrnH(oYLTUF$F9-TDX-XYHxP zZp5;OaJ=OC#GIDY)n5~`sl2aJW6{P3lJ^Z^E9q|AWp|agRS>03C*3;q@XlveKlq>{ zkp)Am`_$Z09S&sa-8xI^A?sN2laIi`2TnwU&yvtH*#RBL_8e5N&l(aLW7v6gCFp!1 z%{mYJ-3h)~!);J)YX6><^v1KZrqrR)_N^WIkOLt!Q)Y#C|6n9Z3S)aR3h!~hMlFA@ zZ~WdD8F3vhG=?mb_2nGNHL~pnvWPxf?ldre^z7+Tax(EPPa;qYDEn>otlEgM%Z4}= zjQB=i{1++X8pYx_D0c7obo+ehtx&L?yU2s!FEhxDps}}Lc(-0=8-x&Dqq%mWpd-!7 zHeNi0!Iu|JZB*F1)hCX&Ng}!t=Gqu5RG(T@8$ z!=nB+Rb5nt6aAZS21Kg8+S+4bd6;FPS;Tf~eo=NVRVWgEs4>#XjQJ>+RWwhLHH4pFPl&WzFknpIQ(F!jf*vR!P}oH+sskip$XQ>7nnT; zSqf!T2<}vUAeCu3>Nk^GD<1Ok^(qs2yB1L8Q0yZsbBIWs!FWFUe%|?}^xw5~)RVRda8n%v}Hnl)fLxdca_x-5&J z#dZ9!Y5E1ioR0-#$#2gL#g0p-?D@OaO*Du7?tQ?h1l>l9+z&(%%;HjFV@JC_!|D2g z2{bm@JPbIMyY&|upf67n`sw6Suy_F@!-D1UPP=HWVAK+p7>w(9lXMa17~2741W+|v#~1vk<{-{a+#~a7w_yw`qpav9b3u!gS5A0>xYOZvvfL;){E4nFxO?zU zWp3DI>qN2arfbH>CC@gpYHkm{nz}erE0Zss`IR@mfBC^^z2>i}a!y)V9(_EX%0&r- zkt~Y+^AW?I^7H}woyhmrpX5hZLcP~x{dZgi652++hd#FOS&fzRM}WI*lwjJE*oRz& zy?F#^1!j4)(eE~4Y>_i?lUn zsmE9$ju1xq;7Z7DpELc;-waS|u=qG=pMF)JY$e0TUfgdiLLRm8^LG&-PY?B(bRJ~) z+3q(#Z7CK;zolBBqW80YOiqD>`NaS>k=f)H%wGpq1BJM3Wq2|bt8#0R^EvXX>wO{3 zdvMbW?VRxNkG(OZE`5J|uaZb|!QEy2@W#+*dnPnKV2-##SJ)K$Aaa?w4E<$Cu$o}( zIPnoYd=Si@Y$b6$#ii$_z%bW*$dD&iN5(FO#u(;)Z5(FUsR`5tFLsfiKwy<{y z%&*PpD@yArA*1F-sXw}~X?0bVvUm&vk`SWxws&y2=&X*aip!H#x^-&1&}k;5t5|6d zo9)sh0hCD$Fy7j?Xm#|SV{Z2O#rYF;!#6AKzBC+VrNc6z`$P$*miSPw&gsqAGqO{q z-bS=U4$+;%7X(@61H4eSj$Xe84q6eyAeGLFzrE3PUDQCOM!n66dh)@X4&Wird)ffn z%JEWBH7$UK&o^HmoG~@@;-{vZflDVDBn*)Y;6{@%WJ`|onYV=mlkVY6-(}7zT`WHo zGuKg^zAWP-7vG&jXLqILofKiohuJc`He^Nat1M8;i3k_nVMartIo?Dx@$I43q{|@D zdkA*Ywa-)On11l#xCOKWMn5hYTu{ntq(C6W;$#?wU+R$?+KA+nKB{bi<2@5A;lxQF zb{~~!fy2B!uM@mwsQt%PjKU8JoffAOJ#SS3C|XDlyJe^U5{E4{4PRu@~H583Z;yCGbe#4q`k8}Fe}_AF4V z`|jPgaqHcP4xu{Y9IK`tVf-%sgGl`Az7G(lBhs@4!Q`;a5WT^3TshFClUzp(lEeKB z&32#S`Kjm-4GT-Kar2{1dQHHh;fYLc&MrYHx*NeATPIHauCWXX3w_3$W;A_Xeu`Nx zb1`^t3 z*7Z;a;&u2+=fJxjsHxs^+VC`IOzEjoe)y{8juZHJFg!7N(#iezJcN*vPE?|Zhr_@L zrRf8s8xXk<416Esi75k`iCp``*5JvM?dN*5wU%yo`J8|;wPI6cmuOydg4`MRa9G`z zc^;P`y}Ge`yl9K_2bR&ONpl*i>a2fDib2zul$1QWx6|^Tc&E+Q9X4;F#qI71?Z(O- zu=S7Sd)lnJ@5u;A`>p-)5pK6pR)cAH9G`6iPg^K^e(35U;yHjx?6Yy5^&{#R9$Jbk zc7m%p`vSTE}W%HTQv~ zGV2*fL=1g5QlX-Io-l;SSDWSG$8XV^P_>0Fs&twdq--3!mPMKBS<6wwS8?#LY*yY8 zCwEaTx4I+`%xH$H#Et-l!Nc1}1I_qmlugfZC)R86NC}+7gN1M{=-L>%3@@#)G)%Ie zmQ$Lyxs*f;u8CH$Y0h?G$FWrLt&xuV)`p%`uAJt8RGVvn{&SaE7M}%3t{Tj8%U(jq+Nzc~SoZffoc&j#m#E7V>rAbBx}*hU2JA8Oc{}FYP!5MmjKdz)YgNA zS(VrhCn{$kdkwh2ZQkI)<1#Xb=SIXH(zYx@)sUxVzkXxXCTQ2o48rNBYPk-Bu}}4s z6lg9KM9%v2^Bn$x30aCiXHCopgfcMO%VsRX{p>4aWU0XrG}Njv83?}c_10z=ZO<`M zTNm8Iw!F|=hePiX_^d6T9R4Dp=%)g#b%^8vLSLIR=+%$TPBMaH;8RHV4sY|D3(=zS zdgCF7xT=1e7e4Mx?pGTKyUNB?Bnoyqlw6xm!;3!v-A-WR;fN2XUq&Hi;aHVO;yQHU zdmNdw9hq%JD6B1f`&%e3mlz(ZQ*~a9$ndHcL|P_}g)oG0Cpz^niP!=;8<9f_nG2*~ zRSDXUEkf}gWaUK&B?hAlmtMJq*`j2^QydnJVM`#4^@BI=M!YviWeX8772aq`x%61x z6=!#d=;?cDg-N>=Ypa-{ANtU5cwNlIpFfAm4i~9b^}A;nC&h9BEAACRReoUpErCB6 z_3{{UG*~&T(Akc6p8{y^VNky&5*pFS=6NrEFiA$sx>4PskN9rACe%~9eo{q+@SL8_ zff@-HpRv_S^WyALV^46@J^$eSYPM=XqArlmL9Nxw2IaSPTn8=Htq+niHV+x37Tcn z4;>dzLVro#6vae)aN0+?r^fUgN!GJ=&9NIvqNtKlBustNzeH?)>yKe6a4uL)I+%3n z!vGItn(G__4md`onF@X-p)v;j-l*?CjMU*cJtSyK4X(qg+WCzv`bAASVmzIpAKD*D zfsxZ)S{8ph-F;JsxA1B^wXr2P<0F(St;0=vHrd?>IU5T0_BypRZx6b#n45U0J`TUo zv;#XdqEH}z-{l`{@wAhZcyej(g^u)%EI|k}`3g|%2-)WSWl?_&k8jxQl%=qpq7!Sk zX9bM(0id0PGcT@LHC6Ak&p9Z_ZXD602OWnv(ugrH|H=(6E(hs^$=``aw?%dOv`xc3adtKajQoXu2<53Txl49Y zg)jDnER!WBlFA(XZ4aAD#Im$BFmQACtTyMT)o~yqUN#+C?N~LSkZH1^bBadO&vSC^ zZ;m&rpBTmxK7vCoy#^dx7#scXz{34Eoof=Xtb*+3yvlslXS9EKAJHMvjIKLV>GnC@ z=iw_5=QbU?Y(+voEs+;FYj3D+7PWEQz?j;fF%e&9SdZ|iZlXt# zD7-4ZxgN`2u@ihFKD$;=jtKkAU^U)t?r1MxykF8Y&&k+pq#0DEK09j4?v5pYo^x_a zxjRsg*|p6m!xvKt(S=t6MTd@r1%HOx60qkg6Tvt%tsS9=qQ*q9PSR{V8!;rfq7=s$Ij1Z% z(;-ANwj+)3MI)`%in8HhNr3RmGAMZkGcjFN^X_gj;FqDt+Whi zzd-vY$3<6xRsY6AaRsgBhX@&W#&tiIj;c-9@LgtXGnevZxKT_K^ zb;fKq~Rl{qxhEE4b<4rgV}>+1Wttg=hF zydl^!ARlxk5Zf4qU#_@)8D0obB`BOp&!WRSF`vpsJ?`bAS`xgTA(%vOw5EQLT?vj( z!UT-HY#^T7?w~*E_sY?=DEqM=%3-x2)+F#zB}hDf;KVGB@}4auf!04sYld5KcddQ z{&TO|G^Hs12{Yp+ln;LC;}JTi4twTzVcAWsOwrq&oeM_jO0Fx5(O!3z>nn>Dt_geX&Bh3 zu&>qt36f5g{nfG(xng9zue#DDw;j4MexTv>@T(=tp*zx!cq>qQ)H#@r<8czb(ku~v zRl;*?ZzYykW`u-D3h|Z1vTTSw-){ zjOtfe7h&Ia_Yl}1`>s$s31}~GA0Amm7?&nY zPYLQ%s`|cY?F3q}cnPN^a+YOdm3qf>vZoKGm3aC302A+v?UTXY?9Q{xoD0KQkt6~> zwte`rZIBpkAYS9@NkJQu!nH;9xeeGd((nk0K1b8DV5BC!sG5%tYbO@m=1J}NJKBB6 zjIU zm7QWiRl-9V{A<+uxH~butD1CCjYmYLT+Mz9J;3L>`VrG%L@x}gZ}iSYKnj@#OsZwR znU3&i$aekYm#2*9eWuy_0f1-+R;A2Yr_YtFeGcqmrf$7*b`zz;3%(H+hR=OmH|YwC zMFDvmLU^*&y_o<-2jB0l3;8jT>T7>5hAY*RmntiCo}9>WzFV_;2-!RNYD{TK0k*+? zb1%hvd@bJBCD}MF=P&zyg_&Kjo#&}Mjh)q!xz@P|bhYVyiKgj~k21n&MQw*5r=>1^ zoh1;ZqT;oUFm1R4{B?=VU=~FgTGsSh{eU(^c2?@9cREn~1MAr*r+l)##gyonISZ*+tA)L=e+7++ zLsl{u7Xg>J0@6xZPnqD~(~#d0B3I(|>eL~xKN9HlZ-Cm% zc_#b<%@pJE2)s_A^n}pQC*tB3TlaE??~dxR3eGV}CZ9W<7Kv~fba_b;5w_!Bll#72 zcl`YbW5E>W#;01fGkX%3>^Lf-U10=VL!I46gi^VJ6uBfe@3?57zLHaHz}b@aMVj95 z%b}8P4?E@dJeEh-aeYu<|Dq}$WerM>B9fm*PAuG7qo9+MmJ_{IE4B`L=O6WXhrdAv z##H9;d~|Xq2cIj4*7&zl(^*Y-tDzqAb)R)QMM=Ow0*WE9>|2lggMGJro31_Nu z??erI2=%8pNkYhBL%+Y+Izgu9dL9$|R*`fh_$vt6$Zxjruas*oizP#nnkoQ{fQrB11J<@6-?ndrs(vt{F7kA8%J;cuD7`C`pw#-xvg zX8uOeI|^sao1NzSjHUVAMbp~UkY#c z{wwx)#$QkDFg*Pmd%AeuEqv;(;WjLALpy?a!NOE#F9n_#!_I#ldAl`;?WJb-Q73U2 z9*S%&gz>nev^Nfi)s&=jgZX+RxA8~YIvz{CwTfa%>Kq-2zUzDhsuNZfVZ#|Te2a;< z|BXbuXOh|AvpP!OU!^Rc8TWk_jxwNyX{b?#nfnt0~G8lZL2PQv*8gFbxz%OJWc;=E6rktw8thh8OO= zfZ%fDROBXl9opNqt70J{+>?c79vIok;esUsX+9?v!=S%4cweJL95adQoBj&jIvh7? z&2B>~SP6AkYw-4J4(hGmo+iBG!MuI*eBBWJwKej6`%1z!)^XpGN_>3jM@JXKgEKNp z5T>RP<<=Q6*vkyM6P2#a1={<#@}|SeRw`0}kcnI6XFg*?)-fx&ByW$JDPYI0`U2RJ<;vVv%IsjLqu zEd{)2XSVj%yeP~g&f@C39E*#z_s<-uKQv@hCq&8V;kcK=V_GW$ICUs&VRM}lJF-F4 z=?tKLeU$|LO+c~F$5OR>TwyLad`)Ait2(hLNXU4?JAU}6k3HJ!BZyx_n7-3XrqvV-yi7IBQ@CLp)t}_#S zsUw>?9jLzJFUuqEW!CR8sWBqV))-=!A!NU|x_c|c8Jic~wSKd&73-QzjZ$Y}hIY3JAU-fM>jPMKWH%qk{igWGyhv1b_EgOVM*P#gAPt3$Y!J8#IF zw}f{CF0su%WG`B#_zQx(|KaUc0CE=#skJV1ciG6IWD(4^-%kn3;!W9_4q~F1pI?lTH zmoi8Prnw;Jo;=7qXBHosMpz=AS?%xnwRV4kadowL+>wveLu26;Bd4gCJFb_`6j=V( z+B6yD%Dt`d2G`xwHtckomM(sYk|Lc1FSI?Ca#laTyXoygeeEO~(tSmC58*171TFpd zP6^vD(bqp}%XOEvgU{Q}2(%a@5dgnO-d^)!_dSKP4O>RQ+=0r>E|z|Ea)ntKcFFAY z9keVdh6uyJL>D#qtxYx=`fR=XdFm zuDyVSKFdRMMXP;0=oQt$GX@OH%9MHvZbQqMhMj-y?K7u03X6 z^EXiR1v5Nt2*%26@4UC$hU!1Fl$Tg+O|BD~=2qW%8}E5p%G%y!yLBORvpIQ_IL6lo zRi0QO(Jrgt<|h+c;y2b=_!O@sV>>6Hq?iCfuS#J8bJHqcELlFk6LF}6k7Wr*+eR2@ zE3}qXt3&?8dlrd0qoxy&G+xlVxbeeLtjP}f^i;Me*SL@pWG|&7+Ry?-o~Rui!IJ}_ zm}YR##XdjRu_1K*wk_#m7;c4>cDO_|C2y22oGEYWSA=z&#Y=WcttKx>TsE*Q=ZR+% zQ+=-D^8Aefjrc zuGCy1QXLiaw=Ps-B1}%d5&5i*vc}#Nh!DFi5ub3M#F$W8E28C+RA1ULMjDuTlnnx* z2%3@!!V*(HuF-)s3rzT_MX1@Hp2LgT%0eSCHUGT_*&1_KygbfH2pX%qL>diJfZYOZKC;2XP$0r0;t0z$qK4g*@-tO`>f3_j!n-X*v-_uSL*9Nfukvv zwNQ-8C_N-dr}>l_WBrf`w4p7TTgTe4oAa+v>mOCMCqoimc+1TNJ=CvsS^mRCZeeF& zKEEyu!tI*@GP_6E94kAC1{e0*o9X=KpLyUb*7lsio4JvV&L8mfM35Qgoi=I?6ZNgE zBfUpA>lg7#*}u><4{^)W(5z`WcQiKt@+Vi3CEpwRa1Y{*-VHI`#^sI6R`u{}0{AC? zd5w(x-UE5ysp9Q)hBk>bkC007Qv-%YlRr=Ct}1?!=qJ9vRlWd@zPV~I{mmg(CuepGqP8l7AJ15iX?%MlJr05g$^;`fC7iAhI z@1;+;P~nDn8A(?O9d)+`FI<0b=e3jCexF*b{VVcOYbe2wU8p9{eCsB-mAl++~VQ`o;n<>8GH>hTPtx?24V;c*mJAh-tia(?l$ zyJ&M?{#=$847G6lx!vFkws7`WIo=1e4LPJO!mNlxKIV3y;`+lY;aMgk<1l%u^irbc z+v#7alfF9;2wl^s6IU>pfvCVB>g2eLD0Ck-VV`6ZmT^3wP@=A@()#UOYk#Q%hb5v% zl(A}KNeo?AJ%U7UYa}*pe>21V-&hYqgP}HR9kEoG*a@7IA`SgW`F*1k!XJQml;^F> zK#t`Q8LKWWkWJSZ9$Nb1p`C+EJg4Q{Ru-P)AJskxleA$ zF+}6eP~ypB73-DjMYpez=Xu+7=8+Ocwk{XUdGY8nF+3ue80~oSbU3=H<7L=pi6TYb zMiecnT_|GQg$jz0WXwy#lBL>B%cdK!n@?Thkyo?xcFM^;`S7tmZ2q=#B(Y8~h8%(_ zEv(>KElB;v-ib8t-kI=$0-4)2DkQ8^8q=|UVrEOqyQTQs>6q&`4AI~3(oG|68PGf3 z(^n7Q1@EIY34RR-|MLz1t>BHj$!*m?f61H3=ib|=C(vtviDr*qif^i9u85G~-O|u5 z9{&0mp7|sm)R~fk&-3Zt@08xo zUE`_FlZ?AVnXDoyehZgw_H*$96-Wdn#rD*fbME7LtP zb3J=#)93h6Au2I04jj?9LEjjrb(^4(U6xbIVQ9|&131J~gF%hLFr-R=%A40*#7>m|SGyC#~ridQIcdNpn3CrHO@QyiofPRYK`PL6UPif|B2aV)!6V}8ITA%^HBZdAeWlB9T)Lav(CYYL z`e(4aAIfjD2dzG)RXvsd=X9WQ#~>&(VD8G!jT>$!bioUjp7wX72en&VMtuVk@IkNk^h?AB8HO_QwtqNH zu2NNWIhbdP%*-^vng5ZeW-UDU@bENOASGwIRKNIgUMA zTwTH1UB2xVqnE}8%GInb`*{8mBb)MtZd)S9gyOxFjH3G3AYQZOWxX8CdU5A+oGIT^e-6zz{8wXDLi69(zqiLvheSVc^OPBAug?{zTXpWtDa$6bQnqWtr zb@*AfR7ZywAn~%NuIT=IWl8VD!Zg>=Cy9)j1d~=~j~Ey4mD>C18Q;sW3{Z88RGO@wNC>C1ImFTPhT2UBIU6*(^&H(kA6rgm}V^VW1MrKCJiJ9 z*H&KN76UYSSoAgeR9CNb~}cq8?zB6pWa z-!E(8p8L&P(JnAU&!qe`fVh{vOoUz1;z(7cg(+4+q_AwMtyTQ$RL!zmtD{xmYn2JIbBU5 zo_z-!t;|~ed)e9}8#{P`@VBWT=@l+lmyqg04~*^U{DzbD2EG#W=Wsl1)@20FzFP6) zjyRMJMKdzW2|yBI0(z5Cdp=y$5B@UMR^ zbVP#=c-AuQG))Zr9Xt~S+j|_OOSRTRE9LrE%maxtL5VzYJtK^EMUL!p(j)T9h2xaD z8ror9MXKifRD#L%t7A6wd8=TLxvTmYEXt-@I^iHIPXq0>v@+cVG`DlCRiM?Y4@bYq zU~M=f4!loa$w*_OGR?$Z3PRDsKbS5PW!Fh+3hwN6mM&NT`{=$Uwy<#^?0U8Eizu3$ zE>8iM>I!dbS)1@@#gIR4OIS;SYWh-dFw_@WBPLEZKxYjepi17kVa4dQ)<+r z-hK12IPH4_AP(mqH-c5D*FCo8hEyp`Dfa_Pv2OB;byy=7h|dOydff`wDVk(pRu_ z#m>;KVD*Bbd;w}#cI(uwre~!zZxp%!kmGltYUic75a+Z2%iYMwjqYF)OY*yt`hID z6_QJ@ovkI)(3@*DlV6c%l*IMJ>BIDyv}Tp*1AP|R8c2n;F5eW#IEZXL5iMq#{gtPl z9rYkIjIvIS?NCw;pZ0}71_c_0<#_9qNw!bfUu64g&;zcD0E$*aSouT+-K9@7 zE*zZvG9C03ODbvD5{I`}m7!#AwOk!T8|FkT__acEqPb>a=xbQQ1jmZt^IsL)P62KC{#v*DA6QMX8tt{;LpYW41(j6mykG;wdQzk20D7x zCY{Fp@2@2SW&jgjkNQF@Dw3&G=4DpRItUT4;SBjzh*A6{mh~aF>IH4?N*1Z;PB8>9 z%d3R}edT0T{U9qhHSlzYvSHP&-io7RUj3d&)XX&Uf{RjhdRUO_u3Ys(Uk@%@S9$BH zxK5^^A5Qtvjg=Ua#D?i_E`$faGCD6UKd62fP%C9Ip#l_=A_f;*_s^zul66HGnwhQ3 zCrmh*hqZP6Woy_V3454smGrCVH{xlCGW)0t^jQ`FJwU?0S}G<5BFP7DpE+1YAWV6> zFP_lTW<8oqqpu(g-VpPZpQd)FXrATtj?9{sSR@=#GM$mwhOnZq+nZ_5UOC`9(&?Z% z*d*&Dnqg*v%qCA+r{VoFvdC$r`6+FZJjjh)3hc&mj_0Zo_`MsQIbn2@m!oH~)@KWu zgk6m*@?Mg8%axQuTwyd>AmN`#)AT6aMLRTYQlM?|s~yE{mXKo_#$P=+aCUOEShQU! zs*57n{ZkFKnqv|}kKSCCPS~O=I(Uc>&JxDqS}kuJv~EdJm6BgKuersMm*nS9VI5c` z$KLg#k_`pvdI||@CC;1PuxH^%ULpmAL1_Hap&H9n^Y+7f2xIk~~)2Q^C;SdcDL0;Nz5f>gad`^m!3T3fB zIzg>KoqmWDlU8e8HV~;fEj>lm_eAVH<+AIzZ`?t7lI8~?T1M;%#>#ReAwD?1Qd99? zul+JrPf8argz-=}8DuE|i3Z;nnV4uW^Q75p6d7f2Q6|Y~SGHc8MB#4~YF8KoO~cfD zx~z9?Rs9f2Ks7nH_Eno77RC_|kHh*B%T90OgUdgHOi(jRY(Ii}zQ+0P*^!`NNP5j~i<8#dZIp@mMvy)QD+JW1zn4N2}ottJiQt*+_wQRa-DAxzUDj#h5^JcV(3-W4in%Z<5ER@ zu|{VfVIT?ain`vojaZt{I+3-(K~S-3iQ@N7)7Zl{ z;+X6nV(fu|*+iRUL)0-?vQP_a04&l-DAMX^l%Vft)NMzzrv#6YPtW(>^cP z^NQx(y3&JG8_QZQwOzRe-J7swXVcuqoq zUq*aM&GFoZE~4No|LP(zJoQ_kUvxIc81rsLH@tZ;bc_YchOmd~nDo1Ow5Q<>fVC2< z`(*LV*^r6VvGG((CDOGjD~FM|mMOvnicaJ_`Pr`H6NSFcmFsCHFn??knJ9CtX#LYl zl9=2dIaXoFMvSAdeSK?MlE%{QZ+@BSjZtY_L=iEFHN@}VN~dyX7#HK+v0rz5*QspO zHoEs`wtGQsFV}_AJ^A9DgIFvvC9Mmx!h^MSYp7}~$jC$0w9{F3^Lt5>Vz{dtC#{h2 z5v;-a%2Bi&+m@l@YrMQRIh)&%%c&>G?SJC$5F1a}3zG72xN6ue8Ez!A`0$PKZ2$=+hk*R^BGQy?QW9-@JpmqOB!7~?>sfpmC>UG1&V^Isuq=N^q0ZH-d5y1EC> zGxVG*?x@Vc{!w%m39bSG5WNykSmN%^4tJ-q(+{)!GK92Ly_X@7m6UydT7b)&W(Xk! zq61$?5-Q&1%z7LL)AT5FW{Wnm7b(`}P?-X>6LIJfaCcMD91SE|p>Ah+`h6Q^z97cX zuw+>zJ(P#RH`8`3R`KS3zEgN)cdO#uLn_02F7m11B=~@pa~{(jh00s>o5F{-(abJ7 zxfxubI82%YrPsQrCMcUtYbKh4bz%dICL$lMh@B zBEgLAu`t;}do#y==IV!amsW*Si=7LRT)V*G@FTL_;66D*I#Vub2qm3Sxyt1=7&CN= z^po+v=h?@HG{%B~;K7?{2Zkw*ppPY_^iQp+yVpsDQt>7mHERMtKf`K&`mA}N*T zv;8m`XS?^(DOjIQf9-D!asD#M?-x+%()F{zbNhn)!Y}u9$bt9M7BRGN)B8)f!o2`nV68ll_Rr#}^vmJT%8Fu^AtnVsXYLtHfOutKGuJ%~~+B4D}0 zwu0_BN%e?r{HDcmv^DOPU>i&EBHEg7&7~HBYmc7stiW0mFJZR*{an90LL`=bjcK3j z#4AfWu-;?uS-DmkAUgWoCfCte@Y>bM#(nUMv7DaRmOP&pMgx&1Gqc+SL5Nb7-w}VM zx=s>fs4bd^pudJG^Ri1QEp;>{b0!? zXaC1vu-NF83}L+?zMtRA$FGQbnNP-c3Lf>C>zK-mE05hFyRAU^9Z{UcOsuyn+lP1O zK<7aGtevh=+3#w(?9HTEMAV*4N*#!MhR&JUv*LaE7-_UZRNQ53mnD{l>Jj7GlNlLr z`!lgwr@4I%Lk-cPG^L+ok_3gCHaMu%Gp zCBb@i&t66iyd474TgI}p#k@nIn~*75Zt8{X8SUx(l5Vned&>ORdI#6`HUc|-W=gE_ z+5XJP%rqGRo-S@_n6Wu%E0=Oh*~i{76r~`w{VEV-p__+@V+l-l-E&db;~RSsg5NFO z6b_45ozA$?c25>U^&q{fwdD5v!0&gumH;EEY%78>3%##Q<5%<%3w;xscym)qoWM5U z^+Rjo@sG$xtUDi|hpBf*6>k;uU-maBuslQ}DuD_UIV3LBDS9`bMRA!^ zQaQ=J=S0-a8D!g;n22&5POu_G6Q3|f@owXQ(y3ABKjr7S3DHNDLl&R=P-V<}fd7d0_As=BQ?_on z&_ywa)+k;rT$PTF-$3^^!>BL(u4-d%)=26ms5uR8W;1C1Z(88KXG3iX_c@5#(Z( zno>03EEUz}VK1&PHQQ)gg7g6fZ;9qSd9_WmL9}nKtUU`&f8dSVY9`OZ-CkG7 zIj8|z%QP!sn&zOKlW~h~jP`0zq%#w0^Bn1oUQEZvnKEB7E9vBq4NnBHu9S$(=OsM# ztIyUi$d@AlsA>@j1Nz-y-~qcP@GB%#dPglLz&rTzR1KGQA{W2W-j`qmS({ zGwfskRq#p-VtY^P^^L^rKA8apE7HBc*g16e(tAgFJRc524@pNtIgi@aE64`?ZCF+4XkhujeeZs6v`=6L3*APN9@zj!$!S>hFMI2*6&LY<4gq zVVAU_ZsreHdpuXzY+y${5)Vc1U-To<2C4B|0CvQ4v0`f4r_Lzg{2oMtydq$Abu3Bd zg_v8SM(nO_7!@bkAOf~11rvr@xI#4*MwpREowWOK?Dzj5abtjoN?tOT^RXA?^oB;#p;Yc|M|Nafu(4FT8S&#gVcQ zOH~6U-0$iQ5BsWDsNhCR8}d$hayKY6<&s?1GKy~Og6+;0q8{U}kCHt|72@qUWqv-rSSS;q$p4T9!>jUUI^q_^=-YJlS%hk9O-YM{hRwGt0+Z=z zUe9TFuEtCGK{xlblj2b0g5nAy=BPUgSNxxlYj*SuMGw!eF)ek7tH7f<8HN&iHCvq}WB~ zHO$j*TLwk6#@ws;F_>e44@oLv__u}%+V1YV4$$=ufYW55$4$GM%AbrSc9OoRKI~zB+l0iLL^dgN8f>`RK86-hh!>58|e*?=F@#pddBUf zEuPGE529%s-8yj-Z;vaULDsl^+i*0*1y(XVX17MctKYLj85A_EhPFh#$kpBoX~lEF zXkXWAnB)VrRirqXK<#Xx7FeBP6xkrGd-1y3T7$GEB-8HMX3j(IaTZb!KXb49c70EwojzJ`xp{pIjCdEp8JbpkG3sgRt&&Xj?a(uZ!AW z!5t_FYc*hPex|e-X9SmIJ@?8IK5XMjFyi=7{E{N2J_uyjuS^p3#m; zYH9WGJI*&|O>sL+4gUZ8M@OwL?Fnw(_%c1>y2K!@?3o@+FV=_!L-#9~uS0?`SIIn2 zBep7g54C^h#3df|c+&eT71AHbLvb4DxA&;oRT+`{Epr=*iXt8_+gtm>GpYz0Vh&9i z0hfsuDR+KLJ8e{F=eFg2Q|uoBrp=rBp;l$|)hMv;WB2EhpuPkd4zBFwQ!v<-amYP- zm`$_P^|TXO*!3MW1Bu(ra`jd>Sxrod_g)^-Z17#Zm4L+d4-Gto#|5tDn>ZpiwKCQ+ z=XZ^PEbCmo5JlL1j#m`8u=J_T__MSn%?l2NKcGmpwtCf_yhwVOw&JvEj>^4F+33ch$U4%)RC1+H0WiPLa8 z&JF}3%k>G=Lt$M+$C#RNjd&$fx!6~XL!!hMk8C&AcOaKEB5;hipr3O`%2F@Kw&J6d zJ}%#~V21mgQKAE*rEYPXuu*>ah7aC^tnkw^F8ikN|HDUP`k~`32*p!KCiLrPGBT^8 zchsM?f!N#8{&bzvPa5T((eJp$X81hD=T!d<#et$Exw2<4F!Z-B4!dhjGoBm_h|!P- z!$msHiHOWChI5-THf;-rxzFclL=xpAyzM&ypZ&`H*C$(GPIUTA!}kUozF^pBZ*Hqu zYUgLLD5hM48?2SG$QvukQw75_oQ^|n6Tb|d#YoIM1v$5iVi()ahfEJqy4+hF-Zt;N zTnlm#i~n|B$GL5d$&SSqfkL_b9CO_V#7&!dIHCVS+O&u*Jo1fItch@P)tD>FtY&HZ zbVqAw`a1<^ZwSx#f6oIo0y1-54Rh6pNxVR_&l(ZTjyJ{(R#$KLJy*WHYH-*Ss;@eK?IA&VCe80$uzjjmPcqXBr^)#ePVTZ-u0%_&vL8n^i%qd~^v_b^b>WtB2;3WP3 z{oYYvd~fqbshhwy)! zM6BrXsr{PNPV?xDMx*7>c=`QiVPk(jyWP22hm9=h$Suk_ZXy!Jnr&{|s7_^(FcPuL zJSG{5_VbRAw_Z^|zvK$P%lAS1W zz1#Puw;)a#Z**ipedOmmS7QZK)`QU6$B&g#r%E!!^$mDemFk4Ax;rVB~P?6ix`2q9jOc zp&%FMeq3|z4Vw=7oGWr3DZhHqyake@p{5iiS(1K1Nj12U5^UC@COK$nQ3GmU+h)+A zF<?8S%`%iONvC0miIq*KoDJU`kQL2LJ#+vG?6=X`N) zD18J!#Cx|~o*S!O*4?WT%jB3NCTsV;dh9RSO>~C2KHp!osdYHu1`T)i`Z^M97%3Et zE3=|U^Z)3g0w7Y81MnOAIY&AB9Fr;S14jJ+R3hF}w>2p#eosaWd8#eiu9jyEqQXHegh^k}5 zJ1l~b_;O@Cz2mV528zexwmntCI0nTy@DrSv8;9Z@t-<8`8x`w&SdPQI6Z6_xXE1WV z|L^U5%p3dw5}&e(O3hr|5XXSagHi67a5uaYn-!3N?e&=x$ zWL7`)_ups1=9{zYtzZWR?zKkz>67xYc zyEJNdu+6kuXxywhB{ux&ry!|PGr)oFpxR2!K&Dr&7aiE0QQtX;qt*|wQ~2STjguk- zM-_BWvRI&g;&;^14`M!1*2=e|usyroKvu)AIc6%r?a^W(o>!Aw{gv&8L1U~m&Mp5| z4)+INQU|1g6v8f6w1NSCnOk)uT-~7;qTCc4K~+w3jjtNUh9gE&2k-W)sy(s9Ca$_% z*tB4e5?TEz*E;1B1c8MsRA$~zk{On>bABo`6p62?81*A-n`w-)+IkoHJTZX_$(juX zGX%w8q#MRR!<2O^eVhE-l$}?kT{Q)*dT22!w2~IGW1mNP?+01P86|CS+I!;`*c~(E z>xLjy1szODLeQsFrJ*;;knCDU+u$4%vy?!x59Fg--f_uK-$0Z`7>x#Jf6c9jL~cDt zJpohZY0cwVz!fx+!Vrn&1gYLfJS4T!nJ-0i)teZjy)DNb;WELo1)+KTDwIa^t%pi2 z53S6@=aX06q+X)TaFLx#5*RkpI$J!&ROIbGR=r(s9Gj-lRfHfeowptJSv$`PD=y~K zA{;vzjEd>Yg>`rP^A53*>8MBHhHbb6w_h%$%0rFNR9!7KXpLv!K4V%!_0bH8apY9% z&|5|7L#C}7#Ct9;L-yH44pp+PhMo-Lx^FO;De6keBxDMaN*^m;_&o3D!kTF1D9KgT zv3qe+>Wr3~AGkm+)Pep{67R~xMmJv2K^wPsqv2lJ#9BMqt4ceOhO;MFEWR`S%wJgi zf9hRJFe!F#-c&e~7<$VVasZNB8&=kshKgiV2<6k0Gk+B*Z*y=0)MC!xK4|t@!JE9r zu1z5##qSJxjg)HU>K+%S>!2l4`aUsU!y)Nq8S9gXzGgL4LDR8m*}0NzQo@CV#KApL z#u{TPVek+%jk#XKzIh5UQoCUEINQGqX&H=dS{4#ij0lko+t~h6x2CV7@1v~+IUJ&2 z^`HL{W?D1km(kVg*k+Shh8paI5yF#(dGN-5%AkKhT@4%B#*(!uEKSQAt=Ab^EZ+e*zzyS^a z2;4*W;LF@KxZkXysNyxaws=OZb)mJ>G>BnSLw+Z5+Yg>r_x%V=Zk;^HhmA?9(q_7qyt2*FMApNc%LvgDmAZR1R z{~rgWo1Z34Yfs=@d2aeK)AeS*Zt@WJJy|_2NY1`9Bc3*1w@p=a0*>}7J&}8m{(GWKXbx1k&sf{3_M9P-+w#M(jYDWZsd zCth`MQr(tY7Ovh%^1eE62Uv;JEItpN@$hY_b|Vv#A3*gWlImf7x%1)!v45QOf!**q zirUb*x$XP7XnSH4Cm0@f{9YaqJ_oId;QD`c=)_b?DHNMm~mwtI|a z=y8IDQs}Xls&l8PNUv9x{qojyUtm)C#opLqrF}AIUiRD2I``hM5ygc|*3(frJ=U+- zvz1N_`Z%e|)mRThizzCG)&)CecQ<2&dpl8*Y%LxY>)^hinZgc1I$NA%+hEYKke;75 zC-pkmuBVq`l@C9I(J zd;Ze;eHF1^WgZ^(pamnciMvAN|t5al|qT3NyxuA(Bu8Y$TM^btl>bw zxMw}7Ji?~C(VlTM!qFz*Lg}QiSc$S|pV>i5_JTTV8|yGVo#k6h3xllOEZ^{FMA{nZ zSgT^fMlEO+Gqq9QIzyPUsBin~lAW+4p;92_cmW zpW~v^^Ax)6is2h>jfeNOL42xnKp!DrEAeF8pG)Qpkv5x~XCPg+;zR}1F9EgJ7M!ZM zrgQ?D%orw8wp65h$vyF(@9mNpM$&?b+adBD|G_*~@0%u@F~>^7Q}8^C1})rG|CdCB z1cPD0rX7h&E>*a^V)=Z~dZ?-d$5h zE8&Dk-q;xn8d~iQf){_}XQHWrSvU29RYqlzq>4s_#q5jsrJ_F4!VuQR#BH5BS@ImG z1wEcT&O9^{=YzO~W`<}Ey#RUWQ)RGRBpkwHY$Y!~o zu+xk3O{8`fyv8y|B>80>@wE37S@1ju6N3Q$-Qy?)9+@X^y6NvXA1dycMZxmP{mBWt z|6es08$vcQq>~C3Kcbk>!k+t{*zB zp%w?G-KOBh{r@*7&L78fNB(d!@hSjC=*BFjsVsQ2}*%J}q z4S05WRD9X#ok*ed5p**k$2EL|7n zgmNABJJm11-5S$^gL$feoo{U<+jW;^fs0mAvWAZf>8@0A^$sDeQ?an??*7EUDkEh# z_~$TkAlLT}?wum58*`>V@o!3NRD_OUd+Ihlos$zNpOBwY=*a#{S6)TVH_!Mm+2v#E zmguIRM6~*W6IpFcETXFAj_jb{#vI6qE~I}a&(q0uFr`$XNL$j0JsW+8QHwy)jX2oF zLkeHa?(>)bZ2?4iTmNEgZ{)O=)8U#=k?3g<7n4{GucuXY2>^|K4c|DBvLm6+pv}P^ z;9-@&C)7K;;$>pv7Y3N7E~{AQozfGFC8(;BWQ7kM%7?3-4Gq(H(P{S@Gp{qA+AXQo zySJ`!VaGt2ixn%AMRV9hVoM6ae~+k#uzcb!G*fww<=tkM+cr=@ustIH-VN7#CH07| z=#%4!R{U~mG`!62RbKpl1$Gj*NU^!L`m5gg9iP-niQiOJJo1ovi`5b|^RT})Ap2pM zmaZO>w=CDUQfaESN7REz%sg517Vgb{%ss^AKJRfE*k!W-Cn==)cN67f(ws-QmzJKm z>599(0sR&`MB$gi-qz>xR18b@I*3^X0?hpQ&!~H}Z&~jD#Swl>4Wv-#l_ks;*;}D( z+WBJh26rN*0}n_nFn|JQD1(8IvPpZZw`OUYu9$(VFob3fG}C%gb7a%|ym@_elQtQi zq$_=D(#i3&W}bDQSx5Nrq-4@YaV}0$ znBw)cUlpcQI}Rsp{Ld{mf2XLOP`FTgZc4Q1r*(?lqEcUY4r4OmG`kk?|AHp9I6x{$y*R zfPDA#UZUN0-DuvHa+z$7;4W82Tqp>bSg!owrCy<$X<8jb^vXDF7NQl{N?_=x3i&U8*+>d9LX0FdsWcy6!J1y2{H3$TP^Wv~zrhEkn z`;vTocHI+QX0iqupS_64COtg;`WhN5T`*xOJC1yI;@QRV`sQ@F+2hlG^fdERmpa1H zU?%DGDauea44%Oc%@3O+#7?Ng-rksrlOeway6n6%=GsT9TIGg3 z@ZQ93=``QG24)C^F}6G{3-myhF>ng1dM^B)9!uvPcM5Y?Y`{C-doJk)U(pV z-*P_o^Ch%0*+70BzTp>m-T-b|m}X7HRJ4Qt@Fv-@;3uZQc&u zG&NeW5}@WM?+XXjF<2FKa;TgIoW8(lFmYC@7iPBsG;w^6)Sq~OLhCS1`w(|7F;sox zzj(b?OxY6To_v=zC5EcK6>D?iXWKVhc~&JGS~WCs)->TDj)^y2Kan)V%7au)V+@HG z_%VpO-N1upM3A#^WAyRk%*2RR<^UOtU0yeP4G*qDC(wq;s_43k__gG?%N2eUZKbJu zBleIwC-G_!#PAiqM(0dg(ho*0Lnx(VWyT|!D&08BN6wOvrk_+A_dluxjeJ75W!E&I z+x%Wa@{J09Q;&d74(T`62b61k6-zD+H$>u0TCW7cSeKsi#1<`r&TzTIT%R|A4>R;r1j!+sJf?XnjBd_NQO8WIGJ zW3sIVi_)_0f+pZiz`N_^XwBXB*^#6h1J31VvBKvGnoOMX;7K#0=kQhS!fT@-{rR21 z{|I9B?I}u=&J1JUQ?T;b2~PhYpEwv_t1D!l8CPvDwGhGN;O&MeJxS2>6)t3E$%dnq zb$sjeGQ&b@<5wN+$D5Us1uF2ngcg!lCXiC^_Gg|aof(Y>a^^L8r*QBSiQxSC2gO9o z0IhCHBay$;o<0BGZ={!F(g^?a06Ad}m?UvG>^z>_KRM^pXW*c|r%ME6Ubor|?nQVw z8J2(Dy=(}_jU@y9NJmDrIv`LorJtLVo{~?OJ)^RoQM+1~!1eP?t%bj##TB5EC-0hQ zwGd1MnJ8Oo>zg)@`dcM!{)JlwyG=15cU)G-?NWCcI}*9L*uA9!{tFu&Q>O-*({4$` zfv5Sm3#}G%_XCefANkNx^&JUjQlj*zrEtJxQip^j>F?`&7wfRf^bxvURbgvOnUF`n zVmFm6rG^QLyDZ9Tbd09z#W>)vuHZQ8o)07RTor;lw+yIRh~beoD0nvsXvsIy?m5)b zpGN^j);w~cID1A$s=XyNO1yiKV`#QL0;Tn$=Y4sliT%mS2tHVbGV9zyMX=@p_hL>^ za`@{~k#5nrFq!ri$Le8Ibcc>eHYkJ~p0adpUSGPxb#Ld6J<|5;(a~DxDVZm~n=PuW za!XELCfEkm%nJw$t~b4I!vx;1;@(2JE}JQX#-7%YLftPvu3zz*Dt5Zs*f<6fGB(3$ zC{{Tk zQJQhIB4AXq(T?pE4EbNt-~17C&-ZSk>8VTWtQXmX9oHEyB>FOdiR!|euP&G1(xSm{ zf%6hiY4K|&+<~p(b9_^_aJ1LCF(MB#b*+f1Arwr{@vzH9a}O}>ZLT?gRxd^{wuQEn zYOSRnldQ2hcCBaPQ|GPS$>Eagg{-wfnkpna;EZSV&Mp$H!$r=Ejx!D4%`4d$_1hz+ z)b-iDkEeG#8Bnv56N}p-Pgs@Kz$EEuvlqhfV~c*a&rU1%*1sZAI<)5UqRFO2@F!kM zn|!W6VQ8Kz`G_y5&p<{g;?%7AwopN;Zr1Xaw%O4B+cCB(=pvfo7Bum5uj;@f)DwEM zfnUUyZ)b9jP0HRXGmD$jPR4hLsLEy6RG;)sb@qwnkM0g={{aD9;YmsM;%$zq#?*Ui zoJro#QHJI^dwtbm3=%5lyuQD(WIKs*Of$_Fr54yihG)eIR||yrk8u_df+aiq|#mqH=2ft#D-6ad6w13CCx;yU92ZO{Z(ud|zZssS4V}6D?GnIVspN z`dR8>bTm~lhV5l})|;76zp+dMhvJ-uC^f;)QC_uufUQ3QrICly&6y;+3l2I{zGm8R zj(MSypM;A9dm*7MaC=mfGwc-V2#(?emxEtX*`rZ-EYBiYNwwf^$la|SM- zRS4~afRVjXnm!~N+_8$`F1@vrohgv2xS$r$y46FxCoV+u`qTN}K7~|vJ_mjXK1+~{ zH_oaE7K-Itj-#e<@71Q}Q|{N7 zwmjM#1^mzgt8lop`w?D&A5!*K&s$T2L@-=Y>Nd#=On5tDxtH?xj>^_w9liN|c^#Qags~ z8O-?!pD+t#(mM|J%rC?xL(9iM7DGS<_Zo++b$aRw*RuYWz;@JP&fEf!VQCI?{rOES z02)8D3XpIN!LBK7SUCiXkjmI<-OGY9Auaj+9>Lq(hLqfAI^Wb0a#OpF$rj`Dy&ppS z6hAdb6!lipx!V{H#9Nb^>w__v7A}?=c;n#1PA75DA}G7p_x46IICoO&7RHJkEz%Yu zune#TbCcr9&w1?(u3SO;Le!eXzB#dLev2Q2k)ppejnc8*rykdjnu;Pu_uUe{(Z;yg zY@T}u>3`C%PzkTg4T6Q)>_(ZPubEtwIOC$*aj8Sq_P5=rM++e$S@FoA?usW8C+z4k z{%bTBUUC{mVRJKjv^w>mBa^ErE;492MLCm0(AiG=4tsV-8)hO+j}{Bg0yj^$HA1_E z3?x0#ncUG~?bkIBXfwNa_;E~n*waj} z8b;)^&7|OU2zO>|O)V+hW?Zt3qGamX3YymoP~witS53zQZ-zRuE?C-J*b*e7o>6sv zT@xr_=GQWK5dPr%HL23T$Xgy)6?e}cSIbLO_%TUk)#eM#u7qP@i3HU9MQ2mV6n7Q{ zVfC#r35qC&sd@0~X%Jd6bq;>MVFq|*7I#>IC6YVLgIuPHRHr)>Kx0$E?!Qig?B73e z3}q^*L#Ug;*i}(BrmMOm+;ny z!quyySO4#`zt9hT$Xt$u(Rx}02i~M422aytLAJ2%3-scjt2-;?)Hp{7+VYX**`Tgv z8DWLl9Kxj`tr*B&$3;%j{KnGfa2MEQ{=ZYWP6bl^=6;uUk|d%YNs9YdI(<2z)P1F_ z`|_bPz8}SVu<#%QGa9vNX zttZboC?-@b%)$oH-%x&>A#Vxj^AA<~q{S;y#+Mj!LQ(UXtV&^KS3l6-LopSjvp<$S zAnJZ*M{Y67>dZ+tN5~}ntf_LK#{q2NtwM|3YNBu8)kR@ zT7SV)hd!J@NL4hOLv|3$!Cl=^L6DZ5niU1fbo(e_57;?>3as*^U@0YjO;L$3x1EwR zop!fR_WT`QAiPG^s1WGO<&9bJA)&EW&Ik+aA#wcwjZ1z8U;M7&fCR~2<98Uy--%vM z9QXydaaO=9_C%LT@!Zl; z!p5537ga+}jXsVHMz_5%OO;4!=QZBX9YmN>I8RYttlNm|_1^U|(rj8WeKu>ii8Jb_ zl($1$&KpL{1(R!)l(60PGovGgks>&0jr0j5Vj;ifCw_PP2l>ep0eh4DP(zy1mH_+3 zDUQ7)$^Xw(Fgj?vkUi2K8K)2KrY@+3_*|t|DD@G!tSTPWRO&c(S5_P0s0>eF%3r=_ zPrYWhj?4XfJML#}^y^1;j@BVt#>9!j#V{&^ZMx{OwFiSzrQEdIdQIO!3*O$ON$jgbi(wdiNk!bg%zw|!fF4+ zCtqQSeo%#$)#xcH5;4|=mdil2h8_`42+!bq%#8LfcAN!2G@3O>9Lyz>LSyXO{p~U% z7iRRyB~Ai5DQVBJn#>YioigkijVC7PYc?*$9=NlZF&~FF?nON0`@kV;=qlOUDMnSF zGa~3>*y3F%0~8W%D$Wywd7~H*U}R{c$DJfb`{*a_rwL8-3p-+uM71QL z`^!7A`}n4)rPexCc?tyk=aZg!l@RmA$_+bYe_i;E104)AjbTG3siMSQZ#!TeJW?X{ z`L$YR50O0FDB_1+=W~PO{EEF!S52N55yGrW@Ea(q{F{P2=Mp{d%3yd#o*C3+jdFKG zxkIKkv{#&;zMWjf*j^5v;A}isQWX;MJH+JkXi^|Mm=W8q9I<&%__AwAg{kH~f;{Gp zUJ*!Xj4w7MN|^Eo=X76Xu9WQo1g8p2pd5XOZy*+}5xVwj_q%?mU6{nq_gmH}b!%IM z7BMG37&+(zBhSQFM)%@2IX4imIo0;GQ5h`tPDixsBw^-(k^JaGcp<84P|Gnjp4@-luU~xK^XE(AfKB*glwfF=@I)V9CpBBrS8HcGoA2i{K zN)BncYYB+HXIKCAo1j-TZ4)yOHVdsm?uh)KYlK#ZJk^-r4|_P;Z4xTv+m$CdVhKwLq zavTPIp{E$4X%D+i{2+u7kD|X} z?k9mUupY`($&k-~)cFQr#D)g;Jd*ZquWIwF?M&dMI<5+$3OhBLK@kPgNmL>pD63S zjXquwelmun@}>vs5KP63=HuDV0^xJQEosdbj^)pUveMd$g>bYF9MdR~>|Dwf_RUP# zy=}!7Ns=5$*bsQLh2kUcaB}{#Gkt1MmT5poq+BMT@pU?xcqNTJ$$9`P+LM_ zm*tQmk4c7j13Y9&{b*Dt^#a{hgEZ@Rxs8oUSkPTA>P#ZoafbbxQp>>XNWKiQd>SI6 zJ0pcaQx<*8e^cQ`no1a9OXZ%udIuDPSkA0*?vRKDo^&!5eml!|Id>G4&-|u$8#d!N zrK6{HA?auRtDi}*_Y@BDSuCSg zL~IZ!#w8=?{!%FZNgFFI_2I&+b<`rUy1PecSKc7oJjclX`_Ve1lwXb{oBuO5xpX7A zO|uY7axOg@iP^lFO2bdIySD$YmU$aWEPKeZh3(P}ZQ+)8luWJ9;hkc_f(^yu?HKc+ z)msZ2&004;BmU7~=CEfiZpl-aE$9;}DzueWYT8!Fwp88AOABxW{bIX~C@M^FWMXWa zfg(g~+?PJh)4wAlNIw~vf{kq%@qXe@<1v=J(?Iv5l%-&YTGGrBW3dIlo{J&7o4-1` z-v7_?8(~H2=*=X?G-~TUEp!st)~}o?C!IqS-o<{lASXXJk6y#-)s+wt(;u(&XEGQi zmXv`145e81j$8`1Ir51cFqH#?SOt<)^9L ztkMM3kYOWkK*C)U#|%Z+SC?(2-^Ghh9p>Qc_rBkO7@A!9TwF~7e< zFXv8e1Rm<|@n#t{5ev95FFQio328FMUV|0el7aMOALGf(m;gMeXEO*)^6PRzW#_+4 z{QBI4tgfOvF4m|`&8A!hQIaTke5P4)UBn$ghu%p~I3N{GBu2Y@WoKuxX3HsY6Ze$m zZIwgZw16Mx)E(_2tURG%_%rZ1Ji7OcQ<6Cvg%7c>{EyG%wbsNpq99F{)5&ff>tC#- zj>EOy$%bK1syBx%NEFtr$%Uoq@+;_L{by#5%UY~G# zleH(ilv7o&WbtKzwYJqwzrd$UCO^BEy}P~mMW&1>#jJx&)sbA2gI927SwA5r3V(C? zjy-HvmR%c{8C%Gww5xR&$naj|2osGMh z;f!h|X=Yw~Li)RjO_OSDP`eAPk>}4-+=kLD{ zPICAs_5CAopXt+KHm{ti#c{drY@K3Uh3>7|#O-Q|(~?~L8A;X7-L}OyF_jQ{;4HJf zHi7kqyP8>NO6;(%Ubjh_@DE?Q<^Sr`yD3dMVNv?|{x|AnQ)*mf!n|25p78%Iu8Sx> zr^W8dar}f)c;ICTCVO{vC_-j&ivh%9 z?8ci5wVWkEAp`zWS*EbV+m5fg`a3HJHuhXC{wqYJQZJ>q30Tx%1p<$R{|L5;EdVh< z&cBtH7=PICkLQnO9D#8^gWuo?C5*t;*j3VF;c%TGz65hY)J9iy$ms%G-lb2=93=BX z5h?L!``~ZO5X@ieRt`eTE5VXa_|NjU&_UGXDrpm|>0<}&e_ClYI(sAGoMz3M;SuPHh^V&v$C^VFeMt0%iGC(XLH%j+FwDpH1E1WVX3SQ1m zXEWO9U8lS;Ti|gZteL~Z(9dN+6>$~DFzVXs0oCC{h5#Q`=8XIiSpqzL5oxCDuoDU{ zbqSW-a*euRPqM0N!sRjICW#FVUYv?qK<+6n0f$?Ekgucz8PC2~VI`|nd0X)cF6}y2uc4s+bds$c4dUNxZAg^^HhNLuFIDYq2_0mQ$c?O$^qY!$=D{2G&aWMIRuB}u zH*_4O{H!u2T;J=L0f?o1-pzq(AFE3?{08W*hI#yahuWs;!} z%);Nt$9N69N{PUcE=ZQs@)1b%jPoE+kRUVU)1#d~f9Wgq zLfHB*WViPU)$hBKt|I5%?OFhd0zVXAsPP45kgY0;mTK=&DV@CDszPM;3YhuuGjNOz z6;?>oa5YP3J;x$YONsU5G>Z@xt7yl#F&NPp2X*n)O1ApS4%HR_6z- z*zm=T6G^ApgZ1|49rQ*@-1@!B+mnTSoT6mYbu+l*y|e&WIIJ-soW5{?pHbnCfj`P( zXnD;oO*Hv%J#Ld~E4g`8nDaY2Si|1#<^Hbf)DkpA81k+-4b1PjF)|=Uh(V4eL=-f? zHM)tGD`MwDKSbIQLR+xrnq9<+J~Qi+E>RPIm7NVdAZ7$;0ftDo4Bv8tPfp{RU<4Er zPdqYu8zLmK9Cu%W(N|XD3z9K#tz}E^gU_liCJZWWhn^uBE0vqc7Ej^Y^^grLkwWmh zlKUmwelnH)yr~j*uk=7yp8AQ4Nf7|p4%rN(n`toIMjx~zVcIJH{BlU0Ac9_R`f0SX z_Tq!n5?H1m6<^U?_wLvRrtjih)eF%Pe#~jECyL&`wf~KrLk-OTGXyeZJS;~V#p!rd z_}uO3hyE_lSAXJ`FU>rnAapnRV34#mRXeyD{afC;p$7mLBVwpdjgSn4GxZxUa&LW* z$>+x{gBQ?uqB>sLM9D8Fp?ow8s7nFC_1oVbU)$I63O5qF_{Ov94LhZMgS z&9eGsr0xYzoK_SFE3^z2Ock}p{1_{`K7X!yhT_WLUT7u(jFCQ`Kw9m1NW}0p6pX;F zuvZSK3Cz{nmfhZ3@&(o0Y^3M3!YxSd5R2uohzIASs* z&=@MD$M!M@I+uO`zKp<{HJM`^Zrd&5^5{2a>L31%)irR`aVd|9rRjZ}cs(DYHOL=S z&vo2M0i0ak;WM=Cn1!$=K>daY<(xRxRHHdw@ltq{qgZOIwDg&%=qaTz8%fwWA$l5# zl%Yr_SEoAhz>e!za?-iug)r8hSgb4s8;nD+08Qzah-J7i;(WN>(s;3aU7OILoq`%e zjvtSSU#dUve4@vZ!o!%oktH}*9i*^5u0dd0bdN*vm=$EW$cpDt(Ae5pYmD4xO?;G* zvmc;TH5OR*e{YRoKUfAshkPF~fgjz9XX*{lICXdSt8syHMCxKvyL8>Jww}<+od#>k zQoxQ#NXo3+b0bZoJ7k-58MvHWk}InJLUB+HNl&Wy>qp9?A5??JQ!Hyv?~_N>=!3-5 zs$ijG#5GB^<^~;@KJL7o*}ruIwS|XaI5wQGk+o>xyjx;i!eLRF?f2~u`z6A2w`&Xa z)c9(?aJf73`PD+OZb5wWMjAit1f)IEZ*KcxC^KT9i04~e5#Gc}u6*zj>XFpaEx{Zr zHPi;Ogwq?M)hj;|rr?|xsxuwhV_(bVE(A%2oEeoGxq1U{ z?axQizG&;fPZ}_U&%HKdk#LNl4kg1gKL>z%SfZXl?SPSCk6g%uHXJW>NfY@CjeSJ4 zg&yxO*2KqFux34ZP^}MYf3HSTZ;=fe#yauV7{7k)78+PwlU_4)f`q@C--QO|N)kvZ znO<|dzjkYt^^sIB#kRh+v{=K7JuvLA{caZMx+T!4#kil9W&YlONg7y{f;u3An{uS7 z>v8aoR_@n130M&llpzP_&if}wHP3c0h+SeP2bB3-;-T3R>fE#{$sT-mi=yUKGxl~#l8deCUct2UYX!dUTGeA=Px+kX3 z?>bEB7W*1_;jb( zT)134@!h zr|rML3K&yN6cZhInzZ~fWq6erXWQ()5E?iro_TiK(l95wysKF8JTGecoLlsNZIb)l zs)uhHF?R)!u!mZCVNjm9B)f90JzmAJ(euf1zm(e;#lCH1gzM_9$s|ZUXl^>MI6rT) zHCg1uX|^uc;Fe0J;i2nAYImCH|yQ`&NL0}`(;wChax3v+p05YuB2gU?FB2Mkz2CdU((|O)rRE<^JH(Rst1KP_qE%snhk?4R9+{b(-1^CGfClTW zn>o5wEE`Qfh%h&{D%O3A9%>H~QfqJxKS1!-x~pGZ`D5ve^p52&)-nf98)c|eoA zkSj=!HObEnQJj8qf3}F98t75>Gw7@%%$6q(ovodRFi z;>~pqPIMT#R+s#mP^Q`{tGoFQc{qLW|Gd@j3_mypE@QH{nwpHg;?9Q^EA&}Hwlm1i zj1rV#aO+#Ez76Q_hsk7Z=T<=YzX2A8y&I%-bcfhiWLRu8{*#P))ci0CkI9co(jh)x z zo2_c)g%TNejIdoruTIEir*^w+F|m7A2X!PoP`O&P(_-2(YYS>1O_f0yjilkZCHw16 z3jb^mhBY};WZ#ZEK+I`)FzFPL_?|dL&9{I|dN3sMn)I=-3{ms{KH9y_2PcSLy?@wl-gf;qT8FCz{gAx1 zqf2K6G`+P4A%+MqJXlJ*8cmSh9V9vAqKFi$VV#lVyMyYKxVJ{9R_NAq2#z-qdHdPG z{i&#tAl@39p%WNHkTy`yW$Rk8RSo1xV){9+6vFqe$i{S)LSuY%n`)9=u+62YvSAYw zd~@j0v?Ek*#Bb%xHxt^(mTI?b4zP=~%!V@tvpS8@rwTMWa`>YMtHd?4*V!A~1)4?J zYywLg`B72`(&KMK5!>=3?KTbhycL+EBim;$U%5yzVuSxV& zw}V|+y>w{EgQET-QP^U~?m9Zxa*drwqQeQ#$kAHA85A`PHWR{!l}L?Qz7Asup_6 z0x!fMUNo`Ov%<_JWqO1D6l210#&gU}%p$ihV_2aJ70vu{I{I}oGd`1M{9mw5+9wl$ z=tQ1S9*bkpL;il#HgzVA?qf2EZtn zKB$7xJ~XdH=^6Hvy!8!}9ajM6YT;FU2A`XjMai3jdhBHj9VJ=q0S{gCY_c7_xD zk}MlzrH$zvmn%TQMy3bSdU%-!JtSlk(^yB7AzT~8k z-3^UvkjfnK#>Rbk1|xh`ag%H`qx1`aWOl>Y{NdND6*+7Ug3c-eI={jod<9OeFH*tx z<=fpRlmNA2e=aMPQ^F1%Ximm*LQY`&`Z6=Y;S=w{vJP~Ojb?mDYY8Cw`2n#-l}wxt1n$qmt@!RpHjhYMC;xg(jz-!(<1 z46XqAD8eC)o2~HlHhU(-P;hZ#`9q!D$NABlx;hEnZ}aU*j3e#OWEIXo8&9h$puqQx zm4*V|o6e1y?lUA%q`I8S(h(5D7#8-|2*iRJZ|BZ@pvn+AG;!G%)4&c&k~o^dVHgEN zIS=*h8MZ}2V6VIGA%@TsaLYowN|J+&lgZIt(&&*`i$L9Zo=WC&6vO@~v9^|~_v2&? zypHTL-)RyajxOGipmi=G&6;&bSJ&qwy@bP#3M1JUo=~qDu_R0PvBBbN6{7Gk8frg( z?q1E%E-{B4^h#Za;Fj#!4G&lXGseg?CWzFs!cmkLyJ#ypym-)fVi%R+iRNACxgy zaqGa>mu5RvcGXrJr{i%!>{9X?#B?jN4Vc+!VrPnj1U({u1a|_VKP+$imE#rHvZAT- zw2?L?Cu1O4kz-F6YdB;vCB|r&32Ef$F%lLyXbD4qSi0C+#68~i?XdVHtirl&kez7G z)0=*-qwIlGP#bvOq6Ye*e6Yr}M&z^QgZ+7a9o7Jf2nNROC8Enjf?uQ?hP&qXd3nk# z>3oddfx;5{{tD3vwgaXwj~X3M$CcTBpQ(=oS1ac#z_zaUa_Tf|hZyLr#}bcJqRf_^ zB7ih8qmhhXQ`bc#V@3jsB4vWPH+=CFGwi3<^rZ3k{cN5sJ`bP|0^UavXS!d;?Nsmu^;e*6 zv*%VAKgw4ylM^xqNTQyx7%ec;#Esy=7Bb8j?Mls9RbfE9a?IvR(WY z?ZX6HCMTT*TKNVN4Dfu%un`ObY=KCIV+`t^B<*m;FG8>C?Zhvv;N zJ{%QD4-(StqWE%YQSa{1j{J*wTNbOdqTJzWVX5+jP};fI zOJG>Sayt(yr)bUi@6RK9IiLp{PB2ajlk_Ghd; zQ&7yz(T0Rylbe@>bXhd}o>Z;)SS-8wwiUl4f^klHt`)l<_(TgXFAXWhB#qF7^&yi8;o}URIN!HG}j+~qyN4Qqh93!91d%vq?G4fV4G8F9CjIS|6T8> z;kgZ6U1Gao(Jj(I-p}I*j#|8Ulz~?WHOL}3v{k;eq~)7PQhH`J@Yq~3tF}_`g@Y$;>U}A-_ScB?Du+BH9VTg(Voj8MF!o+`3CkoPK^u~_ zRxv+zU2vpXfKq>p*~0zASwu$AFjT1|NIF-sDOg`sEj$ku-Z zGkLF%?T;*8b%bO@OBh`2$9b~vd^AwoX``o3=#_XtZlHhenb;T0 z{D0;w6{&Udfd>k!-q9N|+qG}6HQ$EIa=}E>OqMCcRBK!Cl<0D0<`ycOZY*PlL&Z`# zzPB0OJ;R~J-;LNH=h_x>huz=#@#Qk9CK-jenQUtmTLD^X5!ff@dpfK{bM8Jrs>6Z$RX9g(XPs3 z;zL1Gt>1}GG`r}Iu2c%8Y_emOvru1iY}hS79XLvWjN%nHIK#_>vpnwKZ4X$$iMjBt zNWek@QOA&byt(f4<3p5OvQjZ$D?*^B10req4&pa=Ft2j9axpIV9@j^Gq}aAW8=gU- zC?Si_rURfAglildVep3g*{HBn;3BfFKv|&GJD&rtu$5~oO`|79C8y2Yf0Op^<;SI5 zaJy~-L@b38OIrMktmO!CdwMm3s@d^mz{>|}A0ivbQR*dJ!+gDa9^>GiIH}(M~Z<@sYKTmIQ10tRMq8Sdl zu;p_plV5-q2vG~D*wF+Wai(1RRMyWzXE}PNR|bU>TDiDsJSU3XyRC6PU(2d%1H)d{ zHn8FT7J5pQ5Ficj#U|?9H4cqTDH=@qyr*K^WayXE^p8k8tLrO@Y>|Dom<3}F{k+na`@)Tb-2;q^9d43XH#Q-b z*R(?Qb>~^3b-g27c^N;DQ4EIrenn&~t6Gk%OR9Dm-|pFCWwRl1vM`p~N77j{?z|WV zC>2P6wx&3-Ai8X1;Jhl%VOTONAUqQW=7f42@%LU6Pes_Zw)qXJt27#FYOJZZh&kxJ zbGad}b0FH9+@!AzTsVa16|cfZts^6~j}~UxX&Cx!VGAYgQpIq8Xb3xC$Ws?|x1fZO5HLS{O<{IOM5g`*m_h0*<%_dlTW_w$t?j zYXQlmfZek_Q!nC3E^&jwNl_IO?4QLXO1JF1Q20>QPq;Blt$F@HVBK8Q`M;laxRWHT z)_@cC*ckiKs}}^5glHN@8)b-4nc<=RtpB^~@|I@@vsN>kA|R>@tzqhxfpIK1Ma}L( zyE_nvP58F#YK{0lNodY8@W%*J&IK!^0!tynN8|2{Z)EV>?!i%eao9ccp zz%rE6fghfRwG`9thdYiRbcl+x#};Hp0Of{hPsWkF)q7>0SlmX9|0V#{GlqesQZudc zzw~8(W?#yOxwZ4Ea&Gdg3~anm>l-Q!Zv0r!)U4f(Q6>T?SCS=7Z2n|g(cJEzB* z4sDCh?UnSdGt&DBAciBedX)=!AxCtRy&T4+#JRDk4O2GT7nNInCn%=Ysj6jV;|p?7 zt04EKrp*fSdMWC2+744kUKEE4bD9wQUFdz$Hg{ZIgJN`QdMTVc?}CGzfCy(ZDx%A8 ze;bmH`NqtN9x|Pi;V_s)bBjMZwAv+95->%g$*?CA%UV`|qgi750@XPMU-+q=#wZ&l#$Ydj(_}DKsF)$}X_pbYOP~MhH=BYPCtNtUkN{t!dPN~*9 zRKbeO|0m?C%fIEw!z-KF9`Bo8*n;zMKl4VLU}I3+IKbRSzubU|t-OQM!$%qyT}Ool zzTz`s0(th;LR3DsktZcmPZ-@#Z`2^Xxn6bveNd<$!J~PgX^d8mu6ka)#cYGHDr-pE ze8Cd#Z&R=#@ZEsYSm=f~r<>fT@{T0A(k^Xz;#0VKB%)O{y0V1)X&}3u=5YBuf6R^A zj4MKzfJ_rAGmYRODg#pLNcTfWEUa=>joGM2{KW9t_|Id9Aot#2U<-eAG3dN6f_4TZ z<&t6Xps$YHcdQPE+0I3)4_Jaz>X);Hnvud7mJCIPur~d~@pILThg#HYxWvWv_kNTX z@;2@R$ip69+pvBXwp<|QBZ4z}euABoAk%tLUQ3&t9%mq{6;yRolsJ=|!cqe?M{lR? zK0?`R@w^@DZeLvuPrNKCi*;rhea+U!^PfUf#5Q4_BQ==X8G?_9q0*KF(bhzT22me| zx_=DDf=echyPWXYk*gmQHOA{@DsvwoL??7hd*&<*ALH5kjXNF5t)+NaC4iVtPL%jn zAQ#y#Nm8*)4Nv_oT4-J|6?Qg*(L#S?Yh5*U)NFE@9}*THQU(h0QGrc zgM#^+>0gcriyZh$K=p(>-cIM39bG4cg@hK1+r`8|uQQxMHMu=3!)Elyzo|-)`Tz-V>d> z8vKY3qGUb8{ULw;FWXlU28VlSpX*N27G(sl!itl+4$Njq%foIW03Yu+K)eN`iArjn zYqc;wmXfd3yr1r&advj$ZBLo%Jm?GQ*R_e!L9;^cChwJ3m0E@I38c++npsuY;YTpgzbVO{IM? zrz8sC zxO?j$Qf%C~iEpW$A(s&4bIENyoIQcrw=$`3a&i=#vB&YGer~xLH&9H7s%0;L;^5*=W&nlu&DCTZ+iSu*6f;bq^;XeY2gwBEPaZ0Wtndh2q zg;ODotAu^FH9)`0v23Kj?T#~ORc}KfYUCY}IJL;Owb#xqLF_Dqr67@9=ML?tT@!iv z6e3zN;_)cKZt?N!&%PjQ+IY9 zSY|H?W!W+U6M0)Y%5qM6jPJsNyfAgL-wdP$Ori$`RFYk6h;kdH;fsPrAw(w-0-JC*e9A2gv-}}k3>z&ZT?O8!*36voHk3Qygdn3X4fPg)!bQz zBsPtK6i)C8Q-?Fq#yuOoOyNBA+l^wRVmCwxih z3OZ{;i9L@_&o}g6T7f#3mcK94^x;tRRVujI5Vmtemba``4 zVszE<X|8;|G^a=G-5yc=?HV2@5&#NZz3-f3aA;HVOOx#474zt)nbo zO7KUs%@or^P&-7~W)fYJvwn`BOYIk6W`|Gk38%Xd!;(QD%g*`Oj+)?PHpbJIRd6+8 zP7YJ1Wra0+&RNtL68VCfO5BX2jFO_cvqqVPPkZrZI&iwwl?k8<$*kv@RzBl|1*CPRZW zEt}~9O?gx&clp+(dTH-7gohZGo+WO=xf1vYR6q;|uNo|+%dd`iyrU<~@;S2=$A>}b zrW?j+_~^I~HUkYC()n0>$D{l0P>eVsq(5$z4SzAFc~~54@=9Dz2@#u>)dI4VuId;I z_cMp^mq1UnHaa0{iT2(+46+j4y{9CyC%(?{mmM^GDo2m96!&p(W)JLEFMgbw(r;cY z!D+Q@!G86wcn*0n(K;c6(TkXHaK_ob*162@I%Gv&P;sW9-V2Y zV~wLOm8+)wPbjhX0zX?;6N|`G{v**+(jQKjx@$i%-UwVf#-C#zfH8R&$Q9k!2DX#>r50#d4R-nO0dAu$--3)`F8i`eUFYm&Tzwa9xNC;lB zzK86AK^wv=z?G*Wwno+5Ay-7Tb*jK@|NZ zF)YQ**xremSzjMznK|P`P*v|<+v>hl%+A`S?QveeawH}n%2t~>`k^cru*Cj)w(0#o-I85{K+nsA>Z(hhI`$crs(cvmD9w_8@jHCELrmB6{7Qv=peV6jhipGz(R9ciN z)owHIY9RS-omh98SJaqMUea9cv?A4^g)K7nIFxh+ZPEwCS-m;V&5GMOPK`vv)bUhU zaFEeU%-fOAiN56)JOnQGEKKKf=YQ5|%jqrgMJKv66WcL*#ko^KD=Mvswfo_8j0`aI z8LQ8_^q&VtY7izo%g;waPI_V7R+7}E6Jus>#qMo8*6Jf%cYmp$1tq5F%-3KK1HLyH%*H&)tPkbW_O7el^j#p6Ajz1nFr-9cp8?5n>jwJj~Y#j|MkmLswtkvuqXywX?I4UZpAg`^>CD^Q0Lh5 zt||kE+CwAagcE4g&)s&Z?P3VtzuRmmTd6O~y~Rsn)%QvSQrbpSvE;!fOjW{c7Xmg% zZ;%r6Q}%~U-J8YTZzBL_=u9WMX6`N(iP3vu>G@(5bGj(tLFvTKvx^rAgu;sa><>}l zD-ngTr*Qq5fL;-~GY;{-@{qZzR9T1G>}gEu0Nr2HFUbQ7Lrzx|q+5SgVVfs`2*miS zTJ)SHUY_xMB-5c(5>l>h%nex48dfirr9J7(so}MO<+ww3;@nbVE0i zlyY_@FGQC{rk3W}Zr!-KTxVP4i2?&byGXWO7^ck!MTGOv0?Mwhf09kJ=Al88aF*1g z4AGtb{~YKTkfz0L)i0~>dg6j*GY5Aj;jY$N#rDvzcYj3f@z-_=Eq6thnCMv|>oC6QBi*zbqm8rZw%_u}5A zYjgM~pTer}h`m%L_sq9v8qbt05-g>o47m!emBDx3ws|_kYHNG?>+(NGU$vq>T5!@yj-SH~?(#^Bt4SbNTSq-)fQN;(wa= z0&lzAw|6H8?Qu}D!QG`g{uM~wjhW~R5V!l;#8-QD5J4^FLvSeAnv#z|p6(gIF5Q0m z-)L(t$@49IVbWtzIRC$~oV}6LZ`e+LZVap+gfh^}j<=D4+}s~Rp3I3VDJG<*^)dtO)tJX>4hN5h&n zjxE28{mgZQJvO6fROwO@3H5Odt(vr9XO|^*pVgv9Rjo3{f!}N_JBdnw*`^Ck@gdsu z$~IROx}LmSpe4WIj8wamq-~98JZ3AS7KjUtO~fNxbUzEf^P0!OJf}Iv5G(IrkDpyj zWjwCd^m0@oYEZyUT0TW%{2Iq2k?GzwyNF)W*KQH67$TI!@?=5xfDTjcH+lsrw9-WW zB;ui!{Z|EsKUNr@AeoNI(A%pP21B(!|OI5vW9cA4{J4Y?JpaQ|3Zmym%G# zFC@P+G0G4cHvk8|$47a5>z#)?dKAk2SB)_24rD}6cT06bp8hXX?(U}t6$zc!0u&7`0jj51Hh@&aPN~XG>nm4p3yA(>F7(T3PYD+Jq#q+9osz|b* zt;K!D@!;3K+wJMqSFx4bywox!N7C$5U5_W`@pr2$)YYd?uJzn%1~H2p1(G53#a*5OOYm#1{hSt^6X6|~-Z*wP zd#YpZ=KC{a``s0;j{G5XBOdcC(dai3K=SkT6A|&k6)vN(L+v|0S<}Owd_SY3SY^v; zFOkGRHCiqkknzrD2k}8W<{+%Fdr3Yje9yI(tMfk1Di?o&&5DcERuX!&Z~PJTwjoja z>yr?WdfzFzed*ZO@;420tRJ6&BDn+MHBzx=aCT#B(fePNyfA4cO3JRQ#N>$gLn zYQW>T8Q_IZ?}=S70c|AATW}V6qSDFJwcj=%p0wfq3WSKE#p%m!j9;EZ{~zYBw$Ff8>#hT zxg8z!mJYl_S)|a@3qkFgEu%^t=|-%ae-BKe^~a^(@wZLk`~C9f=sV89hEUo0QJ#g} z6;eN2*X`E=&RA~qIm}klXw&%;`Vl>gV|*ltPhdYLHPeHy?{@@wus}3bvn|cy@QH3c z?pTKwB%{5)?vm(bB}$~lkNw-2G(H|f(YV&sR4+P0tuYdc4ma-9jNk>+p$4HQ>Rqc| zQF`C-+>WnDkf(xoQS?hx5e*pYdQOJ8RnnHxo+~SCa|$QGR{w<)X&8XDZ*Ecn)pzp5 z$74ytD3rKA_{4T*g5P)()W+^?<7QROacDD-#eGT-;{)}oY6#8$X6Vz_mw&^J@d3egH_ zZ7;u0eM8A(6$7d~LlMrRccJ1rspI=>y0*p0*+>RKUzOwtJUx(`8v0P`QS^!xRZfH& zG+J^)=m-vsj;JcR!w5I7ojLw}Ma?feU560U$>vM*L0>V4n6OZ(sQ)Z55|`+)P5%K% zrr>w~!+tu16dFeu=7x03L&}*x8`Zg@|KOV+73kOBvvKcle!>dH&cxB!YOVI3=uOV@ z^5+x_hn-h{*ke+FEEgtSBV~ep;yN#{Uu19N;nlSM-m`k+wWcKt(CFL~q4NFY@{Ssq zCmeY|OKO{NOw!9sWxOwIwD%lQZ=-;UC-9lGwbko6YGT|vdQHl zfeEh%87PFu!oi&_3=1LsNRc7}-mi7mD?cVuoU(k_P-Oc-R;^H@ri|BeUzi-Gr*>`vwq#njnH*KVnU=-qPvY z6wBdqnzrPZN9#j6+I%V$q!e8#S6;Cbzuzk{^e`#T#VTkp4B2Eab%U#+5a_V2W}c8O zVj0j##4w|!BQF`@BmsN0U&%iLPIaY$ID);PV)|kLckXT}R#{bXRmz1!pye>a>h79p zR$#BB8k*r=)wB>(UY4N}>kHeqxRgSJ=8YQU_H7Ly<|h_&{7F|8_ty}QlTJIeHL_z< z#SHhW@TR;>V3D;|#=RB1&iRWjHVRY(ff|?+U82V6NWVTA=CeLc__IcDO5 zsr&Zc{rZuk0>8Vyj0h1Pe#a@{k*Po0*2%260T1`g&+YD7VKYobu{gbyRn@uwhdB;A zFPP8KR-Tpf=#_)+hum-9KvqJ5ENS{jiwJ>Q!qR>xNrK#1+Mjp>V-klO9CEcUhE9>b z94sc{ISZ;z>CEChQ#!j;TN~{sym>7kWG3H9FMpe<#wA_`sV>*m@ zQXo!Y7VT^Rl8Onbj<{=l9?iom$PPa~PfuQeMq@iDs7XgNnE<>g`hdAFTx^ z8N%40)vU?EW$d@mhiOVX9?hNW+2!e<5Jc*jpV=QuOnN88FG zX1@QR{Nm7}3~pqDU3EzLDsXK9z2iy#Bo>RQHxq%MWuYvjZk<6q%^2>qQ*=WQo zjTMLzYw)PTqeVm>rE5&LCt4=kq*6D5J!H&J=eO!q824{oiX%lSNDXSmX6h$nk)*c( zH^Qjlwt~yK%O~f0XELLZ@mGk;LI9zmp&|h3zSG%$G;NFk>jIUiqyzr2)DD&zX0+FL zMSs~cc`#IfV`3)-?TV`+WtUPmO@8;J>uWO#AX(2fV1$GeNKdl4Ek;<)qcfR6w4Oz? zEgymL8|rHD`P&0QkRSSzqO6`$7tjN2u=vFJZq_De)GH5Mh{PH$ucsO|({S5BzP!P5 zLmuS0Ko=EZh3WrB9!NnYLEOi}FXbE)%-9BjMCz}`LyVvjTS=zZV1N;Ly{3!$4vHy@ z%)>SR)fJsw-nl72s&N-b)SNQWNPk>n*Afqn3sFBY94P?y9@ORMALz%?9!Wmrqr}IB z>Vs5#z=;Zy5%eo+)Yg)&lkhANR(Sn3;7))3fMN*l*KSWf3JC4NXP)`gNHtk?UsUj8 z;#Ox*mQqe6)eNqO)3Uq6@Mp%rU?lt2#AF%I9BHe%#|j+#U*;uarRb50c|-R9LKj^d5a5+viP|x=zNNfmiRe|$Elei&*XcsNo^cSEFZgvnEE6Q>_7+iA1G# zU}q053W85xV5l^k!Da-_)cd)Di$_(UGGw0ijA;J3CkiV0~ z*)@x=zvVGzlwEg{3P~m#ES_^sVocg`hYEe+2|T;Dpc|rBhir4vyMnx$F%@@@T;W%t zUq*O)RnwSgcXwV&@?2rU|IY{b9FbQ*_$;gv2h%qM<-Es_({gHENm+P#Iu6p|RYAC# zTC*h_9aUrhT`tV|alaJZMVuBMw}8Ta7>ypEe#;mnKVy|BE!=~KjFDTYr^;qP>f2=j zwDj^6r>$jS+Kav?UFISuC#OqCqhZS(HN@C9Hbln2xy*7@5c@}_2>O^zZs3_|CC(uD z%%Fsf_A{5ZCE3=5Yo$}Crrmd*p37VmN4e?_q)5nwTPmU-n38`{?|SQu;iu>htDa*x zeKa@gB_>J-n;C^Q-b%I?a5HX*)X7+?>-gzH)oNa|Tr9kSa4*4J*;`<5H@=z6(ZPPC zvF@zHeF4ooZc;)PTD#X#Ts^&{PdVlUg?g4XuDM`jB5jpx>Vqao#CsNV!)$jd_Z!;C z`qhL>UxzbT+=|nG=;$^XplPP*#J3v=UYs{!Vu=q$>;5)2Vzaje!_NQR8cgG9IzARi z3SLiKbcB2OXicx}C`(pjy0>M3P-c_ysbEP;2yD7tzC^GBYqLn~RWfvbwt}7Y@v_#W z^@&7%3;k8}T#xfT@2D?7ig)+M9FAH||4g!*Uf-%HmmFwmq1}eHAclquB@l;n;b7>% zyc;=a>m=H~S>Z=%KKOQ`*x(rz(Y3LXSK2eLaDT0v)kCu|f5Wwl8G*6Y^!a6ZXFms{ z8>LFd-$N%h*r5yFi*~o)7uAXdK^~!owa3-IG2y5KyAx95AW^g%>*h<)b(ivEcug** zU2-A>mPK~uD;^h(+%pN3^y@_?_99~{9$In+qZ zAJE8i6D6zjG!*3d@GZCj&RH*!rQ@*h!7YDY2l%!*KIiM(~w(L#ehv8}gR zM~~|y!)eeXJ;>gti7M0bulgrvZWR4U5p8H)GCDXxiB)cO;Uf`Dt>jzehu>@}ySU_$!ovApbmIu7KvW!EZai#}W=#5mts-lYX+M6B4~;}AlXhO+ zj89Ie)-;5!KzWcHmggafj|lbb00zw0srq1DAv2xIDirtT+Qnkr>DP6@m&1Ho*WofN zOfY|l9@pi%+os}#&>{NQJgpBhaD!@O07(hQm>rv?*qDaPfVUgBEd7H`kE)jd`T6vRl$Gt1) z79+gb4uFWbcUn2Q;I9;yg2J00uzm{Za%i+8brA%$KKSp)K*<{Z4*yfEb5HLLa|8(Y z$B{<8BsS3jD#f46@4eoq*%s0UsYTy*{D$fVfXBrDbEbL zJZGWJT{dGd!f5npOkN1{J-2FNpA|4U6v+=KU|t0Z5U1ZX49WC{47-`|5cCskQN4{9 z5qVf)*W4Tc`9 zE(!_`#y2c*_@_ljT-*=+^CdQu&|fX-FGna_-Ndy*&QT0U<0TEltB9iz6^|(}`0e4i znrF83%(WnQS(o1gS(1aKSDD>eUj6DBZPcQQUYZi)Is+a_5kar&6Z#8ytvIQwBN0PZ zzmy!aYJTda@5PZ-y&)3CZmLPB{_Zba;&ufg4&e<+w?3*qGI;%F-lm@tzH+QI?4Ml( z1#a4qP)vFPlfk|0XN{^~R_90;bjo?61AC)AO9idk1OmsI`gRE7vDeR?!L9(=IE4z? zC>sl%H0Y#l2ZE&OJPvB>ps5vgA+UGU*Qw~(CDfo4>E1Q+0qVv-)?)r(s44veA`yRh zsG&VWB0l|>3sV*faCizj72};yVZ&$msX$>xWB?j`@eJRvu&9B@XDLuWjt|!dVtlXb zYRr!Jucmg9ARm!1K1<&Y-u4sWn?2q)%#-ut`*YK=rr@zU<%GJl{0^>Fjx32w!BlXm zUdrsVMjVr%CISn#kY|+e8L<(``JzR9#M-{biI<>m1@*}}R19ry>!(>?QNi>{T4K+5 zgvjnZe!Oc=zfhJE>|?i>Zf|10o2BxMe!s7HM{yOx*2Vv+&lNDUGv-==&VKV|ke1Kw zl+?aeoT$3war_eiLFvkc_6Taq0=gkT>9-UFezJV_$j+yJXvtRgZ1p(-#LyKo&_iQg zyIm}7A6|8Q+and@ilfsp{|}F%R``t(Y*f}k9DlDbto=c7oJsKS8IQ~VS}x&%eXwSZG0*Ex_QsN$0;EJ1zOkGA+H-L zu8Q!YPJ)G1;Q59Lq3%FJRHZf>&htE^k;pqSXooqy94fz|x+*%ahgZrn`72I?{@oiz%`f2tY#LV>`2>>O<%FN-bg& zR8>&6oVuU;m83Cvr8e7#`j;V?b6y*R z|8NkVW=*m~o|Xq9FcVZ<7)m+a_p_T@vOd}$R*>bbtxKXuoFf`{ z~4U9(9+x`l%V;w?A9--P*XEL9?S}S zmC50W%O=*C@a6pctTy+zInFiUSCeJYo7=ILcp9?n>y4^P9C}WB_u3ta71+-(7cPq2 zLSJ*-lQO-R`TkK+<28|{GNQaoN|`So>~{G0thS;Z=pV6F9!+DocoFJ=;kMsEy&V>@ z0_Qxb*r{GpR9T>Ugu~@wW7CTJDK?uX%)&jA=I`#8F+7Ppw&T5@HJMkEYH zPP($3GAu+G!1as2YP#l97;r<`mCqMR5yqJ*q69 z+M!J7)wRGJOeHWZNi52+KG+sWG^=BImVVWBYvr=icyTO!!<=W?vBFx{8CCwY3(?W} zK1Hx=(iQYp8YNAKZ^w7&CYg*2&RE#oO`J z5~M;#A#sGD<4n?Jh+j;g!7aa*=rC6C)CDxUmDRFWli|z3T$lyL+5$n%C(8XT&%RL1 z2%@yi{9`%-r&=FY+A2MmL&!RRy>J_;CE3O1EK``!5(1*8Hz@7PJkIwEwyv?(1exHM zqdJBKg>8FE6c|vj^01y1q`1W`43DCk)UMmxzDe2ukZjiS?`P38 z=v(cHs4M9Ls~PbFW>s6ah?~edujWl3|E%!FDVr=^WT#_C)ga%)z@_V*_IYR@zA}xq z>^<1eBII9Z`sa36C(IAK6Ti<`_7%e8)tqEMp^?U9)cDNXUH6yY+L;W>X*0d)7xn6u zw!6=eY>N9s@lSE`#{-wuwTVg@cXiR2icmnAZk%UlKYWCqr>t~@?g&FDM{>jeB07P# zVQ5J#=aF8!o*<_*uRhmfAu7v8q)UbxY$lEZvvI|@^qeD#W4gOA2S2?@8xH+-kZIY^96bth zF4uv6@XU%eQibxq2>L`j(jgb=Jv7$gi!rK#H>j1XtJy4^ zgdeDdLPfg?uxh`8?+)|@B}ZgYBFF^j%T)ib{Z|8_;Dp6`T!Z`b(RgJUcZWR6ZlRq_ z%#iD*_rsG}T?K=7oPa(;7sw6O=*Wb5{F+VTecq2dEk3sHH?z)+AS zj_Na$Vx_8-By=htNHLh=Pf8rNqqh3_waBDlc__5aW(`MQ)`oce-NJc+qzlGhGupqr zfb(rEY5fl0j@CMr-=fcFo$E{zfoC`M7cj*Qi*GGDzMU%PTIuc}AX_IS3}00IO( za@*S`H*#Bg{?(s5TMibzM+37|>k9gV5U61gxJ+7>lll1k4U|R{lhZ=*-e8ok(O>D- z1cABaA5zv{wKVU&H_y12E#C+)3@4>=`;k^B|S(e!m-a{(KU zo}LC}mtaM2b{QeQ;0F@D=5usG0cX_8El=iHM}hhexSzL&tp0~%N|e502E&6|U2&(k6nYa6n< zNU95RnH(I=Gn$5GBepD~L4*d2f_5zyHFB8+Oi0?Jen55Jo{_&dR(nQLX7xc3=l53? ziom88~8y29Is8gg^yZ5mv z#i{TMD{t-Zl41A4_B!&>79S;Ka8*zvG6(wHngu!kv5e8w(4K+-Xy35ZXzSTv%*HA- zDa-Rw7{^G&w_gH3*AcUX@G25wUw64`)5eE&qmR))P*xZ#Lm3M;er$*rM43FUszy%I zIcYfaZaOx8s3eFG=NQ;3{s)WDHAHNr+khxRs@M`L)>6Zx%F>N^&->IT<1WSAobv{Q&B>1WIO3E zTQ_Z(H#3ZU?&^1`KazdWiI7EATJYUICSKd`!ZE0yDkRzUsCu*DuopILcq*{luGS_7 z1PKxO5cmUb>~QPw_^9+9|8ponr48PfckBeaErA-sJQ#94U(0B_k&jdVU%(4p@kJ9| zT$lSC8Hf-6B%1xyqlXy+S%bQQbetW+z_F!s7Yu!G8RngfS`*_PtvLx(H+Sz;I5=rs zKcx#ITT~c!i~WW!Lj{|MM89@BP5tUauY-h=jrhPla((NDtzMo*TU*|!t&IBWkm0!+ zl~By*UZ#HXv$n28u=&?Cn#CE}lx4*5B&8u8@MRWeyc;_qx9Kq7xA z8SPAyfZ|MHX0xtJuJ?%V(Gs^&67bsI7J0BFu8`8_1OI!P){b8(%HlxuTmh(JTa2C{ zFrTv42SZx2i{+RFYMNK16;`j9d(H(kNS$@=?Hyw=!?^WRZ=jj1vT3@Kv{{HW$1X!Zqd5sI7t?-rAN}?|BZl~E>vS@T~Ok}_b?PS9kkx@V$27G zyNAh^HQCafV!l(!2U5LwdekT61A5I9RG_QCT2FuM@pXykg>e}17Ee>Ysh_1HK{)on zE+e8I76RK3h`Y8;B$J! z6Y#Ds|Nl^}dLk=|LO&P;CUUrD6JXyZs>-?=_cRM80XuXeYM@|CVKAB7VAO4XR#B0E6eA;3Lp zPnMxRxQ9r!lEDtf_t_-TpJw)U4fJG8I%#vDz2sdd34csQ!K413ng|P`XtlW>__zz{ zX&)A5QK^hEd?(o(9P(Iq!Fzmx8$-!L*`2Iu($r*Hg2?Gkh|Ytf@&lWYd$&vYnL(m$ ziW~BRfpHu$r9U&;APzN8FGnj)c!m`sFp(hE9Z>UpvN@->-u*h%CfoU)1{N$p1 zxr=bLfnis3>$PMpNWBaP?52M!4OHL18+@DNp%%GzS9XbwZ$f+Gk3uxqxCGT2gj{%L zL0|{O#!*88Lb46FO(%}Lg|h$r6dQ zu2SyB70DrX?J=Hw=>@e8bW5vxxTdJ1o5I#wDz53o^l0PuKczBuyw9|J5Q53G>H0Ce zi)NJ-Tw#GSwwner5`L4D%jX`m1B)4IX~|EoHQ%e^c0vB7l3nR~OT=xzxu{_i+iq*4 zzM78-b4mfXd9OJS9q2vOGQ5V^tJJ|F(u8b<+?Rw^SvACw!}TsSy2z^&+g+pQcUg14 z_1iWgwPvtN?bfI{SNkpT+o*;_D)mr!KBnWqmDN^`D)RPePn636#uIZ*tsKW*J0sj)t()d zX&eWQ>iCG{Tc(PbG&IoCG1W68b#Y3@@vl6JWPQ8rgEn*8nonBo9R;VIb#LJv7v?p+xfL27MmAroK|tByRBEMP#vAOCLs__Mm8e8ZUifpz0_r5 z*`7Yql!dFUVP5X~FJa{CO_oLr%poNn)_;1F`N=N))~jqUq)2uNNl&f8?m`ex*c$b^ zewWqtp1T7?FM4@Ms$YLRY>6ul4`=4ldd>Kl730Sjqy@1lxQ@@wD2iSMW$kc|@x&#D z5Eu#W>uoS*=ECrViP-S3ppU&Qwe9VWTyLy*k~eV=;819Ds#CucpEr3jGjuESWX}?{ zCRwP&x9`^X6rEd0)CJf`V$*w)#o32DOU#dPDH*NH2B3sQUE0K(sV3DjH99 zk7zUKdfD^bpLKeXNQI>Wh||elGzod0oF?auviOb2;u38)7^&3B56||Fcw`Lj{O>9( z7^9<%hT@4mTyD`--D1}j?|~0~9hkrVKo1H#v@{~F+aYGMcye22`dy%}zA6!Up*-C6 zi(I3lC*BF4-{aD~f7LDHIQyNJ_IX;#n{RSQKoqAzFEvBSZq>&pP9ifYjjsR$!Z>OR zr>F((ejtV0Qera4u*7O^iZR7#%_^jy!_vn3K5b4o_4%*|!G8Qy$^6~lQuyP%EbI3% z{J6Rxr#(V5zPbE`!PVV8l^$ku+IznmD)i-S8+gUtZ}@VUFml2@vFWqmjcpB$bX7eX zZWOIyPUui(I^V=tC8^7dSr#LcQpoW_JwoK}%*mUKCo+deu2b zH96~P3ASC)cdMVFuMx=VjT9)SwQ6Wx8;QfZBfF3iiD_$i%)qp$*Eqj?yaZ44#Nf{W}w8<@pDH{K}YXH0DkU4B0 z@OY@s^bT>H0fY=BQxwS0B>B<}Q)!a2iQ#HjVWD%R=2-nMe1htIJ)}EUjvh)T7vHhF z9);F&G~qWd_WZ%vG?gm#N1fnC(mq{vm>mRf^aOP^e~mH9-8lE+Tv5aC7A%3bQXYqx zHI5jUe?@NlY2n-@Lr{&1vGCPi;(PG+Js;@zz5cFiN!XCjM*tNR| zBEpnctf}{|>sd@V*6_&DB6f!COu+ww)glMDkbWH>Ajey$jZwgtJ5TJQG-X-FJvUYJ z=l*64RqYVHE5tWwl~0!TMgpZzCLAUWrVW)-CG$*&RyXusSJFBQ^gu3A@4_T6mLP(U zQg()@OB5^SI2F)JbY8*bise$l5>={1LnEPC%medrk2FJVa@x zBO-z(E%i@>+JqtM=9YhZM0XN~xb%4o96cnz|97v1+TSLubCjsy^mUJ!+|hT(jX^j; z8WUCRgcQbuVA4B{r56*<@yWo6`>|Qj_sO~09Nzb*Mm6DaL@fan$Jhrq z%bHt*0s@a>t^RdCCE)7j!@Zf<4J5mek}``a*k0o!Azs-xZRN5OVFi`);(iqns_}7z zFkG$cc_|;PBx-;7kZzN122JuaUm7yvP81Y{!3P+IUe-b!*$UTshwxt*Qlapnm>eg) z0H&rt_A}KUiT~OpF1R9dAs(&qPTOO#$dSz!?u=baZFLPckYLn}5U*!Up^Oy8fa2A0s^Ry~;wq)gp?=D<^>tD^*;fHIs8gTWUuFWeq9+ zC6a2#1XH5^?&44Bp*+=M$b^6)#&+U}N3aDJna z}D*MfOHdwy-cF<}=d}rKd*AU?!5rA8SY=;suQA2tR`k?fAB5h~CSx zCgMvn5{_bF>Tcm58_ZFKN=tec2KlQdN0f-5G4tqx1r6N>J+yJsT@u!^2TK!OAb{Ff-66L6p9vlSMkon-e5^u`}4 z4WFJ0Zh@p;?yeoAcpb43LJMt6N%@82)W3^Via=eaKxno4zUV#4$Jx(!Tws+)64~k- ziMN6Aj;`o? zgge>2HsYMWU1l*o_t|vpH@L(!b;IyW$L3jxh?5pJ`S_Y&C?nEM4`EezN6w>MeCfm9 z+6Ebb!4V|#wQ_cSMQ>u=x3f@9l~Rvpi2on0Y8d9ECgESDNvW&u;doW-M9uZ`%m8kF zIKoR8P|>|x|F5H!qiAJzT+$7-It8WIBV-#ZMeGG|;H0}i6^dplKT?oJ z37fK6FwY-cNbPe?vm6|eXY+##2EpV-ae|_KF?E~a-=Xx8mu@1t9udV-KkcFPfb4&z zn^f!%*7P&}x!$c6N{CvT-gm(132!i$#?RRYNZ8HAST4vAZ!*k1sI#2u{NzAgvu0nO zQmshljLy&R=w9g{%(Ch&?*3b@HQ*RRh2W@N#7p<{#?n%NylkA6aQ2V}vQYAS^zt-- z(-XD%2dnByEFoi)*!YM>YR*(?&stKak~2wzFU=XvddoEQ)=#Y%u0x-8*bu1)7EC;H z+4T`CI4ML=LOBvb<`2Is`oX%bgWiZ-ZQf7HR)1D@+%h`R)_%k!k^dx}Ku^B!t8iTC zHCXL0v{@avdE-UqCI;9C-ywZ&Q_#RKju=p#5OJnIqg9t{_)u@xk@EHbwjdCYgk~Q8 zN@g-65^MwQjFQsCL<4BKZXfz&NZ#6xk-6#fnjyyNgN;-KJT92SV8`{l`QqpE+x5u2 z=D`-0#$=^XyGH66zqyB~l>QG;1@Kq`ME?saVCSer&#Z^d5(qE4YKoG9_7}^$GFZj> zom*e{Yfh`uH%aRs_L}GpJcL>Nd52~~nt^&kSOCN%Jt#q3mdG+cEq~EzyU-Gia#kcDdNd-dI(QeBHjkd^Hw|3D^W#OXoIDp$&+2G-1W1X-Woc;*2(2i6$h}4p`kpZy#D$ypCj;CAFe_M;L^M7o8i|(Ocv>GH&UK8fkEEX}=)E9r+Ss zHn~J0cG%w_sUPfY^O8^mNLP=>b0J_@hB?Ah{ct0g_*Lu0EmheuI{a2QXLx7b8P_ys z(|Q`E7$Zx9OZ_coPl3P{e$$~|eF#d1pNB2iVwA3Y3%MoiVXw!xGMyWt9uI&sWzz@n+kVC!-qZ>nUD?`BYQ7>n3kv9;*Jgh7^g|js6F_Pm3 zf~WlE6gO@7gU+=Tg4lN{+6ru|L}f9N&>i3NKzz!XX_w%I@Kg1jEmK^t&SYC?qXe2| zNFw(;A0%hlmm*53l*I2m|@xxgkiu`=RL4K1+Z1=&GF6rV!bCB3Ou0g)H z8>NMRfzbt{y=Z2r9n=DNA5BIZ9tQFyEY;wjW4BXvCnD24&E=fsh|PxL!VwW_M-n;p z&AAiu_X=FH?m5+6HZ|eIdfY*kjZ41Ud`w;dO83^h49;_`5{^|+o1r=O#f<4p&sU=k zH=>Z@=`QA**k%*Yl-_&l3mvgOLV1_8CgWp1OIW^H_lq_SKCKWDkU6z?DPUQ(hL9!A zvyK0uL#(|nn91gKtZxW#uB)4|Ep9(;Cc`vp?-Dz2&?;%%;kL`VT!FYK3Z$!4aP~Ko zCVk(<3IXD=$C)X)*

eouc112nbX?+twT>b$`dPd{F-4U31sp43HIzpfKkL zlZlvHfXc>H@@-NeDu&YSeV9>Ykf2Z_L-{MUxGfK2=}r& zl8OuvjrR$7-Y=w-(j|p{yAKc=%p_KrJwn0LD9ENdtP*tGXC!&Zn(?fXDd80B{K5iH zahBQCS`#aLS4XI+XK1XWOScz%KMX=HG(-5!ve0oNbF5KpSOT;qs|xUaP+O=6VE1d# zY?HjUelnLDAR?uqUV*xdiOsN?n2YeS_L6al>Yx-I6`nrEU6W7_ERfDs38iq*PQR@w z)odlwXQ?v&tr{*giG%~X-*2Y1s z_P$&QU3RFoMjV<}pw`VhZrvLBj>GL^)5M|lR(2^%?D1KGpvnoNS0Sp2u*Puxi`z50 zsn{GaAD2W}H%dxruW}5sow0d0`%qd9$t23(2VHIKjV*3>t7utWtn0X_dBD_8TTWC|5|Wylk`W#c1aw#VmB_0Os6tOczO*iYq27 zvCdolwy4`+>k4d6O!qFJ3CNG-=Awh>xX)=5#4JMj-zV<+V_#s^#q7Czm7@7BwgDR; z2EVi~DN(wpeA5_;*rxRzkqJ91f5McKn+MnD-w*wv`ppPRrbUTg0yb`Gl<`pF+Jnzl z75V0P8U|hVJ5ogexo*qjMy8JEn+s9dk5~>DF&(Vf14VPrI#&|I+B##NTjGTd?bBlH zPPTy%Ikf)0T7MW;*ONypVbI15-FZfHmtuyBeKNklhZ3)B!xea25=>DHzh}--PL4&3 zG1a&L^*v8!XKvQ?7CW$$JJ24J1=9x-&s+2>)EupfAZlrmA>oIYqgvEzqL#8EQ@1kq z|3-}tU#o~S&8Pa^lcVw*K7`%8-Fj~1*@%yR14+99TEI9f?u}ns$03pd>{W=pb?mOd zFWTR!3lV-A0&*h<2G`is=Ze#bm5zZ1n0wJZiU_7-k#xl*ofd2 z!n)H3rQmN_+q|F2MJEe-Qk#|NiaiZ&e-%cn8*~vOumu_|(yK(ddECX(XC{)Hp@u*h zlqV=Xq*IGSN&8ydj#wfth2CImEbf1enoWdd5rj{CMKLvhfXnhobh_~FMB~zfMtBh% zR=!F66e1%Ax!{f;6c29(vBLVY#YVjxhkvy`|j@_SD?i$OecX=`#kVAn@cUIoM zSm5$;Z;4iL>Xlsf7@Iq2VWp$CT=w(1fu=jCOSQg(GABQNUS1*xt7|(Ps*Co(4#qAV zQ1H7n+eqh4FI1A?yiHzsiI+dK+w(iK?S8$%@dERHAt;?fa`lttX*98W*$<&!?WTDN zg{X7ZR544S(5E0NHFg{+bYy!g)!4*dy|xu~L}v_I$M3TJrR?EBmSqnJ*)u;2 zHfs5QRTa5S>;~0;!uxw*JBZgQ_P(g0NMt-y1^{Ixq&^8IETK*fXRL@f6u}e7x+O?IS+Wk+c?$8#QIl za^3U+E3{q8Xr9LCYtY<{x3o^yD0m?ZLS9R|2Y&F=^nC@KPCsyvPMHs@oV||-^I%*T zIW*pQSNeEK5X0(xzaG{;O&I3OgMJv14-U@iwdEzN8FofN#nUl3O+2;rpVA>+K%%6o3f@~$V-2S1doRxPlImo`N^ub4{AGJwd$Q*K>jg9dmL&Mj8 zo90UGuYMqtv$=i9#QoG`I!@ruWB-0a6vg?Q5aeP-G@4qx+=fLD;aB6s z8RxNL3GD2|{fyG2rV>*!ph0b(__}MSAj*biFQ|oHgUQ{L{KWGK=Mt)_oO_YXMTTu-X!X-rm*KC^ zd}0{MRGc-|c+YX0@UzN5)kKtt_O5V3yp=fbkxtNDQudN{l~8Y7`9^K*oDo0kw7RNL zWuLPcmI@btfv)!qy^Z1j*Tdiw134Dh!;dLkM&ckPxT=ukH>edSS*ILU-^AjcdSb{Z^K8I1$)7}HdWpdG-@yP>%gr=0K2C?Wapcc9FOuxaJ-40J^|gQ3rENG z@9RjHJlNHZ;lR~AfU=^?a+lCY0ciT%1<%d4^W6QKxE~-(KVeyCtX1~F-mz*`^UDCg zfV!>p*btav4>CMZ0)dY*)pYBy5guffk#s6 zY!tU4<4HVFm~c>gRe`%q#`9XX3&D=!pc`(kt}Jgf@c-s@xz)`~l=-wfjRg5s#%&K6 z;2$g-7uRlAt>tV~zv22S*q{aVQ^Ci~nFQ+=5j^s-SX=d#v=*i@AJ}0F%rvT#fscC7 z&Kzj2w@G-}?f6m4U!hc7#`NZ55@@Oh5W9W)*Rf7cu83tBJ`!3jj57$@*Pr2kG0A$N zVf2zK&AJ)rbjx`>RN@Yxkd83|Jj>s}5jMX+V>@WIZ?5Mig=s(kr!$W!lyoZ!Ppx#d zBMYVldjT48gCywkVL2w&{${w-EvP;0zOQ$ST%0#+D+81E1+#;_uj@!sj5ct^>J)Fr z4;p6rIjnoPXBN)mPy8)nSRd5=z^Tu0r1=1JNi^rjsY!yPa+i*LCLNb%;5i0eY(B z-O|!LhlM8Rp$b8Sg2Q%;VdZ5iC%A5WbKBMESS`?VTRr!JUysOE!AJVoZSFW9#!sdO@8ITcO{)^DN&uRdo>Ssrb#VkbR9L7n?Rfl&i8J@7P7h( zkCo#jg9xizgeM}UU~+rdA*$6v+Q;|hC(911Rq__D$CtX8Fw*&*wzVM|507F{f$M=x3x@Lh7R`-u}}#NqCC75GRhOzAB0C0cKmV zU+Po9YJ+LcH*D6`H`fFZ(pN*!U(uDeJrEN<(CPBP7?15bvch^ zVI;OydM+vDszA#wcwPfkn}v{PxGSR_}_A`n=^AYCK1o*6~E+kpDQ& ze25KJTN9JLwIBDCCu>AwOo8GPCBT%P*P5)BG~iwk5h?$die%26Jm$XZ7%%M zo9!97Ux}Cur@Qw!X%4#GJHP;a5nVJ}IgUJ97+Ssme5M^H`X!fI@F!JxJ!6Bg(PTRUz9QGw_N4K3~EDRCcmb#r(dt22T6km&s6OWUkwe~pP-UhIW z?d1Ne*g=TFE@PK{joQ)GI$BSC2Mb#V!9U92rtk zknf%biBOy(1dqfZcxY%1wb;+3aq_3SOb0z7z8?z{nq|#)CMP~jgU5Pgle1gq!0j1o zU@sE0_)Y_-F%CzJwsvxYh=l*S1M)>aW>3mg&#A=+CsHOsM4}kzb+_XY6dfvPDso$=Z6MNwf@MrOuz% zRvvbeoGmOsN=8$OJBr&*EbkDP%_(sLdy{fcPsiLa&4?PJ>j8-(xE&dK(w&l`YqMpv zZNJfXo7gfTOoYj~if_d~^67v-CH#L9K2__-Ayx+K(ui<4q}kWN#hwk-eU zBTgk#+4wA;odF-TPIr#U*qEK)Wzl6)J3#x&&}@I1;3f-J_o70rj|4o2*)u$vP9y%8 zf6+7c5xp_KHKkcdc88#NJBWuYD(3*xzw-TR=lE`CV4Ke6dFM6v z)^(Ka#71sURvWOb7D=Q+jH|;nNrdm)O4QBGgq$jd@=NGC-Y34+`d8Ub+)G;Py-FNYoO{a6N*nI>~IPqYD zW2bvR3%>d`+@JAurgcM_c`jHxH#wcl+%c;d@dJDNJcAYEpD37C=Uxv43)xkKq{kHu z#A0a-;Wzv|(+A?y2hv~lLyFW@v&-`rRSLver$UJhhLlhVVpBBdT0&5Kv%d4Hr7)s{ zcUwN>$FxdHsuPp3^J??|L(;;Xw3B8e1h1>Q%MQ0!n&OO*dXFVkJFT<#QT;br4+D!_ z$+T$%G22B#1PBtGreBu6uZA2=gx-;=nD&OqYU(r8tNSlk1cMI$K<pc|~A{4tJWxJE?6lkSH2@ ze)rs$+|&4@T^tp2R~}G0Bxv_VV4kjWv%9%XFx?f zKCjGOQTpV}^vIp}a#R=bN0cDZ676!6hw{nXbOECeVty1;CPd8|<{OejCXO$qE+X>6 zrp%2SB$^U(Q_)fniX5rX{Yz#-_wkdseq*m!& z^w+~^j!B-QXu1!hYUx?y-`Hq=;xQzno*7ib7OMWxK!v-uPDu5-x<2G>JWWzSpoefa zy;oBI%jg#=R>&n%`23`azL$?IwaT8?p~gEsYI8*1wcK2T-!_(4Axb@Rb1!+3P?Ejg z{TMyo5qSBbbzJ&E>7n~^nuvXK9&OkFoiG#Y0dz*Qx!Gx;N6==%xp8O39%&tklvyj! z)G)FW$sloiLF8BdiFjSdw4R&M?on7&{(tk?(j-ZhehB_-K57j+o;_>rVJ6W!VWS_7 zfWhP9U&vR#y|uG`H)(@c4oxTaa+y`Xwzgbdzql^?d235H$T@sKPgm%0`F%E=$hMW# z$`0o4ms7XmAqoR$;jZys^qehgES&gA+1iaumqR}80M~}0ILnGSP7;xY#GxkU7+ig~ zp!hED|F`N#=?uxYv}X3 zgP;vT!9Z`pK=EY=)LDxbUX`XhJ$2W6=2m;uM5!Dw6jza97?-X^GBHRq1r)1{9EF=b zuSTm(k4*1>hdp-ot!G@rh9^46rGmUL^-!t-!Il9+Zhb2fNcu#K zjVqj%t%vxivje(^fzZv$*MqVdV9KevEXh;f)*g_AIjp*PI-DVKk6cgisfu$a8HPvO zw4G5-0e5p(kEL|+L=&|Qf1@p)HS^;|M`5FbVCiz1dG!+RBw;?EGnfGaaVTjI8@_7 zQ69F3!;jyE*yhoow^)v)nP6zE5ZG>srr<7i(`~WF_hosW{06Pt$JFi74NcXW>F(9f z%k{fDQ{iS{1;x=A{WImNg>*B}g34))e2RXT;RB)|K(bM(vRLGtL-qgnqK^HP`(LB- z#;=-C?z&@#aDW6)(|vf9kaixQ6|4{Ruwo2~Iukt4e>^p5NLL2Xl0^>(ea-+swOy#c zElZ?~4^6U9gtTq&@~14dD>d}TZ~5rOdwe5qCw=kWA8;#8-f|dr)0bPbZXX#SlZW;q zelD)70w|fcrFZF_;#S12HnYyAJu7L15E)iaeyqn&0cI})8jBhw%3kb~ebj~lX>lZ) zoi+s33wq72Sz`J(7%*%-48(zeX~fBj3M*z{`AeoL{K4VnJ)Hk&Q!vAkSPE<*8={C@ z{DS+}j}$yf_ve@54~E!OG$DSuNIq=zOa##7IDNpqmq2(E+{dl!+zkvOwMD5LxmB9I ztI58(3TQ#f0b_1?oooe3i1Ou@BDPUsv2n!k%LQakx11kpsTqR~XlbK&mllQ%eabM> zrP!v!xZWu}Acu0dvXOceoke?#vyWnu-Hq#K z+p|8Aq-nZ1BqUWJcT-0gZD6Fcq~aEmd->^y0{}ljz`yyc5(q^13tjoTVM z+fTA*gyn)%&=!tXQoQ{tA0GKTYWiyDZt~6`kD7oMEw3Majp(Q-w|97z_`Fs;y#l|n zBh4b%enY_gyxxIIlbpyWf8Ei6>SsowG)>#Qx{bW`;jVsS6aNed$t!HVHy6F{0Uc{N zJE;aYjt~;NYU45@2KCZu_?{G8_^C`>zBX?ME~R3s;YyZ;p*)0_>kqz#$rFm^=P9Ld z&Xdwv*4pbo^@FTCQ|C|%|DSse;5(M8{a4d@f67+)nv!ZBrDVb1U&FCvttjASO`9UD z{8mz&0}@QoRm{cApuc6)(~Egv6*H0$<~3#94gDAPWa5~>p}dNRcWxbUd1V2Q zYQncSG=_Clece(w7_oXdoC(se6$-WwfSy;Ya=Da>TpM2+oi_^_fZl+KMLkdW;5S4- zyL!u@bi>4YUd+_Cw+I)#Op{f@bvJa0B{|cYsGISvq=BooJ{S zgRU;FKo*)S6uR(3;K&8aYWvc`-!6jPuU!9rDOt8;?wAuByslfuY603ZlymHCA9hw4 z8h7i2(lihVS4Bp|FZ5b`dViEA$}(<+v3=HCc{o=RbVikjIgy4bx^Ce|`zCpybrqeb z_vXE?zzsnH{K7E>`Tuh(^Hy(cv~w_4tj|7$LvgRU1`Q8dq*i)_^rtE2v@ce<#*FLA zfv%lEN?0KAr&Oxb8T3eyrbmmot1<0u9VhmY)RDT7Tfl98&fPy}x1 zJj&X*P1#NrEWMg^DMj?usAnvssT=B&#I&Q!I*f5q;Ez=+lFPY{IFN5)9r8dhTQ1K7KR|_Mqj?)$A`(#W2P8qN^%H~ znSJ@H@a*E&2(}Y3vgPF_{ib$Ly4HD6glVzMC?KUB+bbt_xStSxN0dv33)ULj*7g&HBX zY|FCkp@N`nH{_=UnTedHc5YbBHYPKhLLYEKG6Y4SCqM5X15;6|?K$CXRIF?=m)jZ-rS}x-7a9=kN9bF~NHjp1*;OmR zmRq3~7qApd!2?u9{!zzMbih_ikQLTMxjF>xez~)gi}J_ped?d^D82(v!D-M1;BsMo zxeYSu>a=UWCb$sfOLfVqctA{_h=EvcSF0I5=s~tZ;UiTJFz^SKWR_0dfF@5-~3;T zTzPxmWF%UQWuWT-UywR7T7mohcekEI>a4^3W1ZdhMKBg$CxD`cYWHGg3+f< zrTEY>)J+@m7yX)jpl4mte*;LIi8{A0)2XL9mW^wjy#Y1Rg3*S~?17Yglgt`+RMRkv z>G+^1IRqS44%Sb@n-n0LiV36FaO?HEF4e5K{BRY&>6GkZehdy>Pi0^QDExU#$H%bC z@YHQn+;Ulvh3HCe!!TPt*0Rfd8$fqoLS?gmmjW>UVZY_BZ*>Lj$gEaY@ z9ulhAv5|eW`rTbp7$=am9WynDJ4MGrHVfG=OkNY&u@b}|L+*%30rObU4@od?tGl3} z(hwrp5DnG#e@lmIk+;E=i;q&D2O-o8;DHtsZVK#KNKq{(?^KlTDO5tq!zId`7?K;cTAAzCy4T& zcrt1b|L;_&@UQ;Mq&IgyXEl3wG0n#h4Rm)~Co*fU9uvaBU~jVrS)pRZm6YTBtxUM? zVqm9jX(zk=)!y3E!Vlxz7e<2Yk(At{-(EPDDf<*>E4oOpkW3@?-6?Xifkbl`LA#_ zZ5NU?*3gke(qV1-;X*2hR4beGCuyFV!2&xw5lRKxJ^Njsq8%@^VE>Hw!=Ef` zg1BPx_P3Nx_BcBSaq65QD{qp>JQhmAxLYMFsIg}Z4i#nEX?r0pE@fcqNxzQaWxC4n z$Lz*cTpU#2IWa!<{4J6~&;~cK58-t}QQi;Sas!!a`?f+#j|uZ;ED_$huau+TFsgo} zf_`Mok=8J&1px=Ww2MbZl>LszmQ5W$@GI@ptZAt0i0X&zNcvF1h$ov4QjvEwYlrdF3E_3^8{E%^k5l+;@#LT@st;k+8#aNTE_;P>nz}`; zT3X8@ULtHgrxYu^R`!P8J`~)kMX7SWrN+WKCa8buWUGcG0LdPsX-%o#p#d#-j%;PI zH5zNK&!=vRb5K;l$AcUjBJWQZ=9#D&wr@tfN>AxvDs&x_t1QBvT3@I+JtAkjIefSJ zq?jq>2-&H`taQeTMEHC02+3HMobjk9Uxi$CJxbd~>Oyz`(N2XIRE4($?p?G%kD5LT z6kWRO9F6Hd9qul-@b4$HX~YzBrBT{br~_4G8ozZ<@RK0VV~dKLgog6w$KACQ$zOg< zbCLZHLuNZWkZJ&nI7gMVG(lv%zpd>IbR(%=zGnL`xOH5+l)V}URef@&X1Tsyb%S#c z>RROIgy~JxD?By)M)GK{SEjYBvWygGY`vHiU^!R8yeSRVDD-%CSCusq&_D^B?Z27r z(tjMC$FiG15Ji6p7~+zgb7n^pIcL7!Jg@93yCgJp_dN%pp+7wC@f|60$dhYyc*HE# zxmKF%pIbVPtKIE@&+^sgc7Aa6?U2^IG~43*D$)C;0Y(Cg6GD29uR4$Hi^sdB0)@#@ zX*vuis%P){wxkeF4kIYZOkmb;?eAT0CASxI{OK1+_|g42F1BwHTkMRCv^=uU^)51T zCY_;1mvc_M<#|`0oe+y>6oa`CeBMh!OWu_tI4qaL#%mH2);avPiVj1M5@!ldS3e+{%n}o2ZSs zOA5(zqwjP$QJ~AEUqe&eLKLI;+~%1dmY`?RwXyrHO-^DX-I|*Z%E6$dk}ugINr=kL z6gwvUn}4UN5&0{{Lg6i^xD%~+`}^TH+f2U==Zh3wz{_yfP{mv+*sO@LG$lJ@gU`8! z`2=Q{2=fpP->%#+^h#mm6*8`8?zZ@jiH|Nm}S=HUC5TAY>y6KCSo-ZOJRFfeIs2sT-9`JAQdtC@yQ|yQ) z+;*C_sBP$1D>!?4NMx-nrBOYN6HQL)z-JzM-LqcLgP(?@dDhi(waxpx9lI^}v(^Q8 zJ=qm;(}Gha2GtBg&y3$R48FE7h@=$4>DP$$Iku+Wh6e|&bH}7utv`1B1fo_l`dF$> zd+qC`>Wy8a6%Bx};k@qbVGp6!%g zexB>U7?F*hk~gl8<7)@&?A5>-jV@np*~FdR4|SA%$h;*_OK_%45z`ozhiufloGX!Z zVfit!6{<|AjtY>|rRHIVEwv9bob#|g9M(&%>2VWVT7M1htP5Z<+z7)1?!Gfbz}OoZ zr}?FX`)p#nBsE2N@7-sSdRio>JIJDqkG+4^nZQohV(#b&vpPh-GNQbkB;}B#M8Ss? ziaBlJqrNDnt!uqOY(@}z_e-p9eaCZ*{IUN4EMraXRZFDyVr4^~7Ao+&hi4S#e(FS8B#%3B|9L2! zn6mja_>eZ^JG2gN+T3p{i8YMPGb(L2ZJ@6*4Y)cr4~+^4R}W>h2iJ93**iIxvO8mhp zUm9q`*q&l1bHEe2s2GoW$Qk8cZb=IpCc&Evd24;b&hR0yh$p4^yk(h>3c~40Ti8l9 zzwRBmfUa|tv04b)`fg9?fuX#}B`1ZP#+ofY-TpaG*{mGFBf7sG<9c-x&0d*K-V7$M zZ@6vF52;|K@oAi}MmiXr>}~%a4c!A^ap!f8&i8`=h&-PKwMcNu|CL zTID2*svWL#b}OTc&T{TzyA&l8|25F6d(ZW~U}a;{?kQ5dy!Z*2t_eqi=>Zdny|!A)^T%Z%8~BU{s=v?yC4lStTYT zDtv27@gwEBJTj_R%y&u%Rn7IUlh1L$)AgdVkH`g?ez2O~EgzCN7j*f-OQ4Neo?8@{ zq*;s}T$6cxo$Em9^{1f~cc`&;6pPcz&C|QCtXUo&YC2*rLFTY`s>5QHkavF6vi1Jt zLV#IGAGRQZMsqPb`?aKzReqgb{@)C9JE*Dnc-a$`>MW00-&;LdA^uxpEJQxYXO*WZ ze}RN&`>MRK%vG%NXt*)xmF^9rLh6cz0bCI;Ef%tq2c-EFfoD^moR(V zH{N>-b+oml>GLKDvsV%~uXWTw(p@PmS=BvM{l?^G^qc4L=bOGqe8e030!JxI`<|5e z1rsUAG0V0-K-}KjP{X~pE!3(;VQ&i8X(g;f*I?#mz^y}92p{_XZ)WS^s+p(u8_;ok zs^z?6th9v;v{#fPDLm3L56#nzAtS(%*^md}#_d@dAoRB~l$jFYNkV}D-HYD}#$?aU zLgzF|Dg_>Ogm(1)0g3XJEMosN8VA-s@lZN2`NbD>ydt@9SsXJry!JQ-e!HHPC2@Qf zOCaQemY2((RF(r<6!!|M;+|`Nt4977=pcUL$*z9dk94)0Sqhx~sja(IhJ9(Zr5P8>XZa0?0ob-uR!-q(8LG7yB!J8Gtn#8%kTSCww zgWXYKGb?#`8w2Qzk3XHEal}HubhmHkvtDjxiB`e6(%C2CbxQJhobJ$^=VMT1IvWfW zi-bn$8G?STWz zIStoW%C&X!HXF$NZxDPg;B=(T&Ay_H##blElscO)o-diIm2{SHDU3Vfz7XAgsb)-# ztv8BU#gBj2$J(WAMXQt?LUdRZJ4Ho3xw^>y@SFz;&!CzqMu?+y9OmOJ7*dx<1<`1Vtqi)FQfaQ%)*>PqrpEbN|r}pcQPLSc=l+Khb z=~pts`x}+GydpLZ$pWGy9%Sljg5{D$88l0Ex@~S0!u|I5LzIos!+vY`JblH`1As%I=a+<<;_F(vkz{lRU4JWpcBIp3wRYT7E6YGLz=Owi;gOQHSY}QKxYC@d3_PDO;$M3h1xrrmtT5NFXxz4#k{vM?e_6S zp0}>aK_JD^GY#_wPZEjpKIv3xtsKdgkJ3VVRt!l`2WWVH@=u*q{!Vh0R;I*$e1cU% zgo+udD}e9>y_~`CS0y<4%fa6b$7*W&Baeqfw@R9f;gPVHu*utjNt3?Lw8(Q@NRSm5 zR^6ZQBRhSYmo)LO8h{Yg`BRPhm}L5op(0!AC|&U-WCuTCB(yOYCw>1tAp6>1P>D5w zw~}uC7LYAsQVrcVI|!hgXW4QVNdEi+L9n@@jT-6lE6<#+x5CPbDQ37@vJ!_w4i zM~2bQ$Jl%-3eC)1S6DNQ;cVx2_cOEbsXuU=>BvxH@ouAyGQT)%W__Ghq`+zqZG zSB`M1;U5fftf!Q%d{WHMQOHq3!GSghiM~Twv+(cpEgP3B%sHshKjps#n{6otq&; zq%Y+$wMOal@Al**iBz(OW8Z{^;QEx`v8NWQar$gvp=7HuX1 zQ8xJ@AhG5E9GBhW^|(bkg!&@tSeZEhCNfrcI{1^n0``R1fa!NlCLJ zgfZ`_>QdE15HK;DYCcxCnUu3g_oNUo0UGiS+o6@_c`$*dw(sgp?jHZ-x`vkswERSOXejF|5 zP0~I0Ep2W(gjRbX%bT2NE7if{5`&|Rib_*amk1Pe)6uOnwDOyD#6nN(orxsJ(3iVQ zMYv%b7v+q7cCuFb0UWPf@&0`S)qGJYUrRibyC?$+RY|m_aM%MWtLtmF8; z1=y@(BGtJj!6?ETwW!%XHl~`GMfZBtdGew3P?4BWx6nWJUX~N<2si0*>CK4r&JzRh zXygG~aiD6+LO5o>Nu!*KshqpAg9)^_ zY(}#M8y35)SgqZB3O+E%{@5)H)N}H&=NmnYOTzVdkZ$GjeI7D-68$!^y9X`2t%wxA zI!x`Z7_=JR3E`$lK1tv{@y~lOkHQ)7;rfWWt<_I>6YdmG$788s=b16VkdP zGdlYhkJFi@N^$nbwA}s{o%`?i8`n%{dL#V7j1Sc`OX}7W{f4O&om)V;iokLm?X3H3 zPe$q|BrWfDtG0CWR!j((l8K(21-WQcfG)GrPg*6PSi0|#Z)C!wn75V6abt!Lq;Q%m z6jFg}d~H_lPo4nt?^a!LIwvi^UgsGV*|R!}`NSii!W5N^q!{<8fr3s&B-o`r2l7M|teZLuZPoapn;PrBxK`+1Z2mSM3DP0Yz-e|x2>AIX~I;&Z~JQ`t*$JVSEy z5r<44@^Zh5zKUr}ku>Z|E9^!Xio@jo5`q~2oEkoV1$V%`nJ7AwiAOSlu2>=7#i2qe+e zvR#HKaIf@Vy!YFaGQ~0&L`u&$uZ4Gz zRP&;1hqT#0++JG}LK~W4+MR_zU>&aNCUZ>@ql8hy)GkwvhA0F9GtCPM7M@1v1^dU$ zRbp}ddtkk*AY>OCUSbACA;b}1Vwz4Z!VQKOTe}P>T=ok^mXeB+R31M-nU+{2!eo?a z!0kHoY~b;^G=Xp7GUqXCST1^MU%(D=n<65Mr|5j_LofQ!KHBRQ8#8>kel}HchuE=l z4DZ`@Z1*Y6RY4ze`(+x&XXFu9NaF}J&ZBV<`M=(tlSp^SubRu?CQiTL+#mY;ap)!* zQ)f`+=r@zx^UwZC?k#a|P6=CQLM7y~a;Y29Iv2t7DF>5lW52XgOn;{S14($*KlatW z=4TGvg?kUaTv^*v{2trB0uo>FKE!bo_ zhh*O`MIyw z3TTT+EF8+DYf-lveFn2$mvq-NR|e7jvf| zCe7gI4g5g+Km73&88*YlIp6{wr#Tlt>~6342YPZ4pV~bjmG9L|K~%#;`oMTieu<$g z$uOr>EE(xz$@r-^ga@KQVwdm)wQe94e=53*O4UZQQ6)dVKH13ok>Zak`C!EI*{6Tf zM}hl+<-bg*V|st6%z+R>!q+6q%8SKqrF_Br@z1{dkiv?Py|6AmC8w~TtS~F?93c-s z$f!|}dE5|Ol@33Yr5a;>^%ZjTkF9@{%DB(QRzwef9yNCl0nb82 z?8AIzjS?AlGuv6^RyT*M?2d$2ShnF6B09r1rAAwfFr^N>m7CCDvbDvzOtme?L@`dT z0j9ZpXBU!R{Z=VVgn&*ptZdR|QJ=lkuMfc7cMKggSX$-a?Ai(Khq$hsi&;O3D@w?- z4X~1vHi*R_HHDl}11~S^LR2Qp5{o*oy1KuLt|bjx)8 zwbVzu6|U82`M1`6ROTmYrc^v@mo_z&_}NCv_|M8A=8^F~=nF>bB$Ez4tv&f7`K|4e zp>NJJ49VF|iN4NEqqdI9M7+)IoW2%=L)GdDWP342Ul8`M@?p2bWU1C4_*K)1 zmqIcicK`ajMT}7IPdwbGwz`-4yzyqnv--T`TpS2gYm*0p)mwq%j6K1UfL{W&lLF5@ z8~PN*pE*1)U6dhzJuu6uVr4HQF2K*Pil12HObg1~F0(rBBxkt%A|_E)_U4hL4%*k{ z_gj;XNaH*`23zv3x^_2DMmV0!Bk8wlRXY)_N2MEHP`)r_L!qB);N5V!V=mXqw({|) zSuRd;^2l9v#`yL^1ws=#!7Wi{>+0pyEH77=yaOSyph@*Oxq)Unq$&YT1l#cK_@gg{dHpC{}{X`L=(epj< z5Z+&CznW$V3S%q}GttuDiaSXZ&gInVr!+eLmkYD&)99nlG97%jw+X_d;l`!!;c+8z z6Juki4afB|J6xBY;6YFY%#RmZO;JR*AffrHWmVtsT7%LS&Aw0Ui{>z=hih`7AVfrg zK4MjcXmSEww);GQCr06zhn%UF{_tLISny>c<4T6hto6&gY#tgY67b z9{E&Xsv`@i4q+X@9(_aXTJPcs!C(fOgDoF*@*3VL7fWmHWd7W&s12VQ z-`(T73lnw1s)?IrahW8<8JDvJUhovPJm)k1{hS42KBg|?+C|cWFW5f0W4K!owwhxP zf{yU$IwqX)gAp>->##NT{DP3VR1 zUbFcdxBR?*JS>yCRUK8K*5#39-+X30d}B#Rs4hxR_QO#<#7pCtiE4L2>P}0zz6gJr zC$@>$?Y;s4C9&3hrzMd(B(G5`hYoGCdC07^`rr{wdaWp2q;+fhn1(}cCX+wfXiefr zBx8zg9g0bm-#+~Fg~S-UC$do@JEmF_{@UyAvTyv1)G%NdLsZKsz;Q#x#v7#!+sB+B zKyffO35R-F-7Z|7c0USTKhWN&j%iAaawCPm4|y;kL8e|W8GS~GR7|8rs$`|lMdsgg z<<4xqmMc=jURDxWR;k13l5q>Y*n^N@j3f0o&CvVtihEaA$=^zK`;(+&UiJ{wI!rdo zhl$lU(J`Qw=4qL#pOhA=)*dkbuDO^fs27`qmDOLqGfUxdxZ7AM<*Qg%yK_|$LJPVu z(i_}5_I5g5^|*5{ll0D9Q0pm`xstragZ$eqhPOav&g)xB)b@J{e3Iyw8BUa%o+26r z5++vdHf69+SRpSLH_Eb@_IdM1G#OEABo5U+M2IR)-d}=`YSUv0dUL*zeXy9m)XEoN7C4V)ioHP#jx&1K{ z8?8aw(pg%{FoYt^bBM8xhAUhwS_-862=lIOOfZ$E!8l#2=ILf6Aah)9rw(=SyVE83a0FnjEa6{onAMZ%Z=qCV zm1He>M+)*YYOMCoY zPkCq?TtjRwoSiw{7WSp-D<=sM&&^RS*afmEAM))ku)kP>wdAL_u4d^R3tsfO*-`wSboE(b0 zS>pBn#6phTLK>50RyF^}C-DsRx{v_`&8O}lb^zgmCIapWjsX|t9W6K#Dt8dYpViAZxnYj4D&ObN z@57%1f5=?c>PJ-aDa*#qF`hY>vbg8t@oX53J8LKP*E{{3uCGv1kTbrV)W(#yC0a&y z(#yr{QaXKQy}i7g1(+Z=7IhQSY(RLrWR`JrL`iJ_wPo$D zd??~qn)chFhNHJSW_W!4e_THlAuIJF zp!>+?r0E)Zb$uH(j%WJjeTwCJhC(%lqx%)ts|!m5W{4w_VWV<=4OGX!Ja@$sek@{` zBW^H44b9F4pWuoR>nOsRI7rFiO&D^LY<|my*mX{<@H3zWx0B9b9}9(1TJmu-2cP(ao}j?29HKGO}*oPmNm z#9JE)W`Z5Vw@$fjc(H7hs?ySKMiE|FHOmQN%`6Z5Et_eh&cwFX#-aLp$71>i*<3y& z%}N~fuL^UoMWA7x)r-+UD_D1OE?DtV%1T5xwi;?HJRuk&lV#>+GNY1Ze5pMV@f+$5 z0Ue@qSam{9Kp;&=*iAmmd(Qtv##Ef_b@adCLj*B1W~F9#31@&~Qx7_2-`Sr>3+X#6 z$P;1D&_0pH8#ijT<{q7y6^q-Ju`qRxq&xWM1#E$UH*EO`p~?f}<9AQY-y+*56nyQ$ zByLENi;tN%fBPCVZt@7N++#wU@l~w#9f6IxGb{u+Y$s3vxhLC2T)j9*_2h-gtW5(> z(4J6K|7!RPAU__Ez|){+T%<9se;-ELGnWhHP0>mC0)hyWA=&6(l7!P>R6klw>oH~! zi2L#Bk?AG$;P;X6?2bmraV76)7ZWh#TCSKP`RrR4o@ez`Wq0d|dy~-4ds7cqv_8@M zp=U3!oWaxvIAYdxGlO%}o6bJ>AtEL2OS`3Uek-30{C+o^E9?HR52UQW)Uy6*o8lm zl~f0*yQ)Z$2>;Nkb1QkSU(YrpM3wI}XLv2-@k7uXzeYaqJ1(Bni6PDw9<*hsUSMEi zGLt(-i6rQkf;AGgbCrW3OfPq@#xWDHY|M%7H!Tiqm%mkSl`*{YRn~uv_pKLRM6rvW`2`vaQ87Uwuoi(bwh#A3V_Z+ne47V#CkWKN#f5R7 z2~3{KIv$$k_jWOtYo$C(k}DE=O}qv6DPz`$)IsZtMA!qEM~=}0VP`|u&vljW%~wE5 zztUZ^MaRBR_VmNX?fTs;@rlv3yFSkb_>;{1f~BCcKv?PeDW-J_QGly)>8nfp((lzz+_I<}6xQn=l$?Eo5o z5?7ejtkyO{=s4Q?-!xC3t!n26eAId zHNz$V&CwKDJ!<>?=u^kf@ZK4?>*S%5b)j_o082tmgA)IHL2=6Kjap2dG z^J%pxti2JqL@)GU1|&wHA|;Bf^^jfGc!T`5qGm1)O4p15{r#A3v<8h}&NQ?u9NB&$ zx@<~*o>7UlE~KHyz#H*XfAuJj4jz?m!E<9GBfDIv7`6bZ!`^SW~%kic9 zp6cJU%S5aT&WBiBP8Q~RE-=2`RN*Q*yzTM>OOh#T6;xjvNoiQ}bFKA3R40;G(djBr!i8DpDo8;L?)u5ht zk5mck;kfBAy{EC*kVonp$wjQSlt}=k82Uvv08JCD3>6s)%ZUt`hbp(4%jN->Y)+s* zWWN|!2dy7OLB4LB)uptKpk@BiKkP-LEj>=$ zI_%Ed7f5@ub)zOZr0g^cR;tOQ#{o8y-uIMAU44wUX~Lf5Pq|2oI}HsErFXXZL!Tmf z8ole=lCPG}!ep7#cSugF93VLH0UTdwo2ymy?T$3vM!aoyOT?<+z2I$mD*S%IL317R zu650Dh#~g^V*o_WzD4HdO=*D$M zc4Me|nju>QH+dPcO~am+eGbXpr*FTB8`~GJaQb|>(0~k!ZQm4}S*0W2YMI#CFP_>+ z9DdERMe*xBF;M>lk995Ai97FGW)$J7oB0ry27?G`vN zaO{pq)H$pj{18l7B4aMu?`SSsE!h0Mx6=q868XfsD*hzBO~WJ-fAmj>sia}fDR!^v zqs4DT*{R7XN>#t{Pv(Xj8~Tnxy-ed!;;GHnr;Am!2&F)XiNQovdt+S8y{F3CcNm!$ zn@mNVFU!xa9ghb!0JtHR{^pQ&n5R@U7};lTY$^KOiS<}1qmDm!p9z#$1a(pQ zT2hKkp%3abq)4}W*6zS3ZRjOd7R-H^Fc82?wY}<{Ch9Q_i;+o!pRjL!C8;Oc#((Nt z`E*>3N&K!*BYPW*vpg%V2@3i2X~MEhS|DO0wCug5*@w94s*8}%eQr1~ zH^wuoiJ7d)u;GNll{~djqK-TmoChltt2C)e4)avlGFAm45&fKMZ-UBQ39pfUy4CyR z4;LiIRbxk3i-UQ2yxpBI(GNC&yljntyqlzYj(QTb+^TJKc2%okdgbGO5ZHH#!P*Sh z-WTuJ3(9L<@py+aL#UREJ}z-HC+G!&_&9|I;D!4BhjjKDL?F3;$E!sNQ$5kq6r)>V z{t5*xs9jT}=GMKa-Co>!^P(5h%#Q;a8!U0hR)X*6KF7+a;&*muVh!<( zI-2dhHaH#c+v}8xxL3BfRL{2YAQ`5fTSz4Xl?_>PU`$cn`U3;O>eNv~humWs&lOLE z?N~m!l(!yjw(iiBiM?iaZ)}?6p;kZFwoK;S&X5(H3bpC;G=A0Eby20e!Jx0Bi@Sdl zlH%ZpOUQ&6FPNyumO8QC(0%>B)-cRCUNt>mMqe`nun0s7XSNGOE z{GgYH_SL#ELa^&(${+0L$UcZ>^>R6QMGQtp8n4*aY$$cE&_e0zhIbgO!I1jIIICrB z_x{??p&ka{mE!*_mk>Vfj}LV&|9(`X^%$#!<1+Qx;T%O<066eh)f6bN$g&CVM$G=lCax#j}P zFb%)um{A~m@6QbAlgw@SvvV9d>OObWp z0q&>b7B(dmIMtL3t?Ia2X_*^HDNg@1I1(=g>m)i8aVl0ut2S55T-wB>y7_FLl?E>ub$)cCIK>I3<6PrYXOb zyA+x%RlXw#EMdy+!Vf~>ej8~a?!*&pWBIh&Sz$rqsl)UppwCZc?N46Zmj%JlC!D0# ztC3&O)O3|{_n3&tzfdLIeZb#DK!zi(?Q|{A@hrEJYT~#dda*2ftA8-C{Zy}95j&o! zSkmI8rY8T8tdL}O^Y~a6!$uf2{l@!QFjT56_&ws6+M9aCgQPT%8CTkiSuwOV$wr=f z1Ki4J2XlNM^>(m_*7hzmU1HC(ItZKi^y@YyRl2BSA(zO_)kj^+gx!4zgxEoJv*a)7 z=dVbr>&7noZlO8xH)eA^k7jBgQHSb2(=;oB3ZerXJ_4%4v1v^xO$lK(ksWV{4>5kI z0e0f3bUkE@(;y1`hi%tq@5ZTuv_8{Vi&nkg&uRqV^we%>Ms*vHnx13{RStw4@!QVh zJxhG0aoDW9XcwM(u{fi#lE1Rj0EZ)<#Ss)PHvO~Sk-H^khz1;v=FgRGvs80=Olm^K zYhIZ8=c1QThV4uN^x$V=w8_~!zl8keH2WAuJDDMllrN6hE;8AehPf&`=-cLIPtnlx zv$BUpKS@@FnNw;+6%9Kl&P8|Lc8QHfYL{;BFoCN4n)>>!8YN0SvRns<2jzm6n&mdR z#_izmtFLyr`et0C$>+|mzR?}}yXbh@qLuKtIc4G%o2P+e_*79Ep9cEw|CZ^z(#J-i zptcEN2%aT453dX%?CD<+(i&j=*n%IQ;U-**wur@yQl|Bh%d0CFiN9R|e+tUXC2n<* zlfq)Et8y<)S$J|Kt%bvET+iCE~bytbt4N>rg*pGn;=V?Cv zP`lZ3A=73qtXM1~DyA#4nH03^N;6|S+OIfwhKd@M?U|5Hzs2^6$0KM<9DlEYa-*$? ztES%KvwiEUC zRZ!e#YQK$6UJ0wHRKlQcmn&(1#_@9VP?4^IDq^$IYRkx;wYL9EM5NL+TbI01=NePmIvEUhr(8$SlD$)HF(^fp_kDr8eQo#82Ne z`>Fxc7i=v1&=Y8?M+Ax?ZcPiS>=Sv*!nQS)^vtUJNV4hz@l-v;Xcb$?VSHhsrFE|e z0fnqTds2Tz_&oD_v@G%~-rL$}%Jz`DyTRFsnqK)+`DR^YK4n!slx2Z{1=d0N8WY%2 zSYaIUGyJ827J|CiU0sRa|wda(uAL&I`{bgmPc_%c*R+?gUN;G8%@9n!w*W_rm!e;SCwQ5DUHOcxR}7N zQR;TYS&>AANZ1a{x!91TQSkK|7aKYaZ@#UYb)csLb$-Uk%wn@NSw;xn{&}h(0eL`^Lo)H(At@ny-?}>V>CFo}rtAqr zmY9Ucbl5K|Q0WHqP)j*u;u_AYxcQiExGG^WPpKfc+e!Ly&(3lQf^zZSkA0GKO4izA zdsGm)F0WcM_1NcH6pQZ)&~F29DKJjwH>wmqnDd{pJy;EWUW*LmGEFzCKQ~-HAl~LY zJEC7vupi+b&c(Z(vB;flh!ajE+bA$e==tA#rvg!=v(Ey1VcW~ zw7D?za3KxC@r-HcnJZeypV(kfAsR3tqOVmn`@{wi5{87s5GAGwqy|gw9Mn+0jn8F< zt9&K7bQ&Cz9_YxHm|f{At}?wJMohg$R7KNQ=WDvoRIR2*sfUE~XXtviA4@_%qcRW~ zjS0$3K|FY}q^z>8>zf)-v3>Si9 zIyZyTXeS_t*A7I~7oB1x&+qtPPcL=E26&0W3x_L(E-p0s6LDugI(S;T&#;9PfC)qd4cS zOYA9V4+a_g553bIf@YE3WM|E;qd^V_nlXTq2mGlDG4;^~z3gw$GPUXRnhBqm_F;*!*^=$v$QAJ$HF5N_4l+ z*AU9WP_7dhn4}W0R^qJEq*e^OJl|MSzKz0{*-dm;Uh{e50ATsNrzvh)> zQMQu)ITzFgFX{Ttj0Y!N5ZIVmT8-iRl+GeYG@6X@N7ZFhWb*t3j@a~J{8fr}^)9VJ zsMH-rU;(+cl*wRXKpa3M4u`a?GqF(7kyFSz`Mi2YNk(T=c-C=TOF9Vj#!C;bN;O#& zg>|shx6ipG3i|lNzS)`liE_PkYdw%SZzp43s=B>3=&&d|VJgHx7*mnK%r?&?5($VK z#viY6xv);2Us<@*?;hwVaew;mMXbvzy%Tx9l*3ph9sZSEF+=>+bL9EgJD!aK4(>+s z{jQYaNbzqL1Dpl5vHtR*@xm~b3nhWRQ&{#jeL;zhtC7%xF;;eHon1;YntdjAmLPg5 ztyNW1#g=r^wM$-lcKd*V)o0+7T5)YE)C7W!bgICXl;$X}Bp5@h88j2quPl3@K$<7H zW!5i2ZK%SunW`jYsV>;T8Bjt49cdi9#$-Z@lnMw|A)^9k(warX*baZ+w@d7NS?uJg z99RW`(1;~w7_3-zTEjABf~DKsS=%5M${!kZ*eS%bwaaISc^{G@vf6=B{4caL(eQCb zhoIF`sXQ)4fskL&@QqRkQsN}>P1DvZMkR5rgz7nA8*tMG%vtK1r6gRH5W!(XGX;o3 z8)l+WGkRG#8~3*G2Xltgy4Z+D67Ec4nEicSSBq_Kvu`TT<8t@h`F}+ z5}c0FS>*`G*I-YZ;$GzPo;`04K~OOMeQ9pmyzR9=(J}u{zFFJDT-)8yCEgcg(XXulT7$b`2e$Fci}E5sE%F{Gjai8P-rRz0Y#-zjG) zM|*d>!CJLY016D$v=e*R)#nf)qb+O9u_0%wsXVWkbrKdZAQ9Wm{ToK9kF{jiF^f*-KW!Rp}t_28JGcAKdn$!M+WO zX~tYUX*oT^7djO`D#W&1)1%fo-WTszE%wK7+V8rk0Gqi1dJcaQx=ln*xwjSBek6{P zn_o9mFMMN!_2tc!(nbKCa|wDpQ6;LKnB!Q+oq=H%{>rBqjGDimYbMQ70jBl0Rd6w9 zzs(~X2&YxP@WJ2XODxqu;#hE*N`4y`tQvHcXIxpgfYY(nfbd8;|8aztgQ2!LNrHl`LSYri2(}m*4Ceh9#5v|yL4mw}Xq8+6la31O zWZWF?_e>zOY#%HU^!^H_9X_=~3v#0*T?HgvrnN%f1~t&1*Z)E(k3Dwix7g&6bN?#% zow>+5w@N+?YaSizVBbqbCsmzZK5e=Cfa*q{gIJh9mrC>b%axd*azkxJ2lCkNgSu0r&sY1@d^9} zpCBhfl6bNDLND7VF?nhVsIbL@d`M%iD#MiG_}dzeb6!xw2&&rQejl@D`w3_1r$e71`>vs^Y(u=AP8isS>MHp$EGLw z&CMkCs*FvWG_IXdAh-2{U1^9?X{d^-^1%*V0+rkoc77r5k@%aFOS1S@i!So!!ekK4 z=YtM*xWJWRy+rYtx86WXUqb;5(#W;Bdp~5x83C%zErBYyEn7cu^2IC8mA0G0K`c-F zW{nCH0$;2Jr8=QXK3uXyVP1c36Xx%y_li|GFx;+MfxOi$1cNbMj8~g{aW)(wn_+>b z(t&@9Fa91Ay-qs+U``O0Hl#jZn8vJ6SL~6H4~I9885da>Nq)IaLqhOQ=X&8Sl~x1& z1_zhvz;^CW_S+52Z;3y5N9sM(7d(sLGhnEr!c34EN5_;nAv{+?QbrtNqDyxn<&PKo^>t<(_W>Wg&R6KtHw0gb4g zl`O-Rm@Qe#LzlMWnrMa$TflhT+4Q(CTU}M7bDyCcgG{)|QSJ=ODL8zG>M3Z2iaCul zxh>f9H9{tl=hAA|23X}Bmk~wCok+Q4&i05KuPhLQttkTH;6``cra&}=i$}h%yh0VL zU)T?f95d9hqJ7yy_Ga(d=S%g}RP)yvwS0{PRr}Ge{#GOtA1i=*n}& zg~KscJf@jFV}%@*5?3RNK^WDfI-SpI*REW2j*1xFa+=Wm1%6X(nR|w48}yNj2#q+4 zXQ-{+NV($bNdpoesvUKAJs|dHq0m|sXKp-FgvS}aWdhMfaFoHyOHdHha=&5^^6}Bt zzJ5sZbFvQ`5+-_RFSsz33^s96dvw0NmR+L9jfxfMH$M4V(P9U!B`-f+b_5Jb;^qAk zUgY8%LSF4YS-B7~8|hWHqWN-{>P!F?^o^i~NN7u=%g~HMw_RDLX~@2`G5sbeHVG*e z`qBFOp7?OrEOt^o?a0i`#txmtv{sg_12_GOe9?#?-A`U|V_%1jN8a9eg7vf>O;m;Z zLqM{FTlN$!g$~8pk``%Sp%N>1$-RsL+%twZ&FFER4 zmrz_sy3P0c@;2)o*cH}C6u~(P-JtDDOLP{@c;3Q-YU?WLFR4Hz7`Z!sAro@i4DF?F zQ;!#y6(LyKG6H3-{OC{3O*Wp2rs*RAL zi{G|u@_S^QW-|U~G<7#R8rBrYo0T-$PWzV<`ER7(u1*>_4h567J{G;eO73-qZWDk% zQRTqlRrj#Qq~C{0kH6d%{Avf8794uU$-kPg&V0*K`pHl+33}qg>l4G+Nr~wh<*n(h zm#{jz=C1Uyg3)`Xu*_8$u{#;zHLNP+B=c%}k~Jo9g9Y^AvTjVe=u1AX)}EuV6r*7)D>f_h5)NpM(uvX&Ze1C-4VGT|AsYq;Yi~J> z6RpaQy|4kr!df{x+smBjj60}}qK&SvC-&{z1o9JJQ^rBvtj(D_KVA~IzJ!KCf8hC} zgtU6tl^6RE&Z3o<{2@zTDMg;e>DAzd1&m2F^eEQjm@>`yCa{LW;iN0zUIg}0iB?0l zcYJ+JTq(WCd8T@~aEJRb{P14*rM5|q>f>pVZIgi5pA$Wa-7E)E0TOAyQN%*5JA4r+ z+(Dk2@Zh(JqhjZ7SpXGz^9~H(bCeukWjkK6RqeZT7&pX?^s@SU#(VbbKwdh|?4z)F z!F+P1opc+c!jtY24p-PY>`(9zPV?~Ya=y3ztcB@o<27-SD$HZ`NsskJqs9wcY+7mV2*w2O3<($~3jPJsz1 zY=DM_aaZVP(vx#uX*rB)JYEh`>sWUs{1hD=4ou(miLs(7>B`uka*5nRs z(WM~g{ie@Tw^Ko&G5K?76Vq1NSZHUUl8RPgQg;^EJu8ip(ZwpaO|ql_iKKZ=eFL8}pM!uH1jXKxr zNJ$RWi-b9k@dYd1Gd+Ikv_gCRX}*)9{xM#8=Io&4(%zV88n6A!oWxf4*R6Omf9b6% zW>$AD3fMu4VQR*l^IOor$$^SD8;b66s?SnIy-@aUb*{!Qf&B?{H1yV;4t@c*GC!hH zwo+IJLFWZ4KQ?6P^jH(&ziAG_qH7J z7??E?{fLk?8UY_8`~CUJQ~U^H^p-L|Cp9Ab6LUz-pds9bBs}`1zU4|6sY>cMlwfpU z*$ay!3#Ge&foaOzA-pjj+sBFB2{z|u%%w=I3J3un4uY8XL8@Tn78|MBN=r})oF~LK z>zNlb^es1NseTYd$No5ZTG=A0xly*{uyznT!Voy42x_4Wog2iG!uVLge0Se+V3K&gP_I5b!%ItXNq}O(eD{1ZxC$LR%K7 zn&Bz!)=xm3MDbHPuVzVkOfi4+^_BW~-}~BCQ_O)5=%p_fqDMy0t_!o8X!XH(e?X_V zvSet#DwV=ypb2V*ifGkafH4h^w@KTg|;r=wRj9`)sj^wxTzgj4` z*Y(dvJY12wrOEJ-xDVPl!z5*zJw1@W$i3`VZkiv#e*vB61}qK(l)!s}_y)ks!7=RtE!jBg)YQ(z}~mL=^^$0|-7K*>h}(sM)G zt+nxHjTiJQTMq8C!q%?rGTUyHhwFlU0oJ2oFF4q942U3}1hxfHPZvBbw7Xi>M+RO# z3_6WdSS&x3Wqw1v-gVm)m5nvAr@ouUE}!he*JOFu2Kx+ay@(F{S3C-=iQxqCs$+n)G!;a@sLGwBTQX^v?d0Z zRSQ|qGZQu4*iG~^CztLmY~GI$0R2K7zTGmONO85y-Sx`1l%vTy#E`^p>$acCGyq{q zuzYsB)kZFko07Ri{Oc9hw%74eqdNwvKZZyn`P>duRx!mLE(pxe1s9n!<(k-b$De{E zo}aBO5p~OT>Tv>%oV9Xn-tjM14MqfiRpa}LDw}_sPDquRe(fUaxh}M4kWmI`z78Xz z5T^&b!g}z7vO?D^`a3b4oTfp|*#7=#=296qd!z2cdYZ7qlwI0;c1_uI#l5ub?(9Ruv}0TP6Q zWv19eS;6n2Wkj$VM%}~T)V}1jS4<>vtGd6HmS8R(sNT;>%MN+TQ?x8P(GZ0#hVr`M z_j4nKW#*)n^o`HPw(J(_Hw+ZRtj`pT)V)ILbH`?G3=&QFiGl~x!wvJi;DHl-?@^o2 zl~Ea0MuOhvukSP;`v}v7jQQUl^>U&LbBn3T5j>?Os+a9(%axBFY&9*Loz|kC-Cotr zzGQXthQxxiT5|rf>_&=fS_rm$XmsAN3^3?%W8k=&9xwe@qyz0)(QUttPR< zg-N+-bAAy;yp&+~3k0F)?``4Lr;lkZk`CYa1`EnugY+t+0!pDVW~MGEqX#J6w2r+I z#XLi-a~$&sScXW9%W|{!lYo>0K7qFzTEBN2-V-`>9TL_WCkPqL_JH&8hEKqYVWkKc}`lESR)$SEL zlnO#S^iw^rR2^8YggY)@26#{f8Rf2g_C~!xI}WhJAccRdY=tR(c>|nddCiG=f^)3_ zmebIlJhN*Gom64jjX;sAX7R`ay8kx9XOeI=q0y=HidlQ&u&OSnRlUVVY+6~pD9h&& zT}XNY6x+GFG2vcolii=4Qajw${a#}af;Kg9 zla6v?wJy&WoEWVmtPDk4>EdVqo_(m!!JfBmA!0B)0f-R zsI}BsgA_lvoO|Oyc}4YEtAy_M;|=4-n&!)ahaNi1Q-&W~xZ{3-gf#4-8sy}l-e>HO zpOxW>a!InCaAo1msX(=_qTWsHXUc9K&w(etu6N$ez>2WY+6He3>~wL-^_=wt3)DgT zm?o6*!$t|n=~ak(NLDXK?|J{PtbgjdKCU% zCf$s>RbiwC31Ljl#Nk`UiLjGJ_StA@w^}mc(E<>T$~Plq^@y>@+dr%U5AnyX7nek| z(iGX7%rG`82F!<^4&j=?^%L2?Kxvq>AAEaH;Hsp-__Mw14*cv5f=jfV<&^r;L4?2saT zXunw%QDsKD@QH~|KS0(z&yv4e+zK4s9oU>#^hMcl2*_qYAx$@u=}WsTa{L~o$uf<5hn4aRQ;t^bF5)4-c_aL!zsdlMgYm-%eAZ0%NGM1x!1wm?*Jx{R6( zuIRQ*43?>EW#Yuty6X*g>IfsLg3>ate9`CqUd&>I1L&}>9ze1A+P`^D5L7nc^o_i7 zOCs;Y^>MHt6b&9=^p{mYCh{SNX9oEubha;HOSqLS+49TWGHs zrEd8`3Ril|A3!-bwP|qPQJ%h#(u_Ngq&7s&bya1^q{_oM!!d8{R_Vbp{qC+X1#(kw zvLpUojbL9Zu##_T7^r~k>hEvK^68DgD596kJ`qiKe4IN58l`cjXJx__VMx@0#7t}Jj7Uq~ceD4Z*DAq#1gavq*yu-X% zM$Ps_FXUEtI*OXGM3A9TA3%VWt5IChc!Q+$oGwKqw6q$8Xnb97wZ14Q-yHI6(@il| zb`1fUxV6Zm-&wmYkkywkzR!VIGrKqIE@^sH?XC8vtL|j_i@3$`Tda7Z5^v%~CuVO5 z)fp>sc7`81!>G<>eT3z7`wZ;7_ z5%&4|Oo-wThJS^-jtvL&6O#F&R))Y;MP^5u$_`}u7{_`{yGp_*eM{(3__f*o_8~gX zIaEhTq+W)xj?7k-8AQvVU6!HJZ%Fv5YNT>B$~k&RV?9D|`N-}2+IHd!-{Ux3wH@^( zb@KJ+d=NGcL|E`!x|@t3o_DaCzmB1$mEMS)lZeJ@ zIe*iiCXSER!gTq? z%-zve7jmz7pjNl4)`~v3OD%W#g>vpPe;=^D8Bc~|xzz#11n(@;BI!l2x3sMCv^HCe z;IfbPfIUoTRc=8BhCkjyA2K<4`ydGV&IN;jh=&B)Q!5N=hN8EyJb^*d-V<_I|0TmF z`b0=pp9L_2Q?N|VqlbtA_&SnAZNPQ)H`ejuLxc^BiayD&KTM|X(jsnB{FyHr@I=6K zi&>693wobePy`(R2z4bi4Se0#?cyAGR9GU)NT70 za@d7vkribQ_|$BR8w-xGzw=s}JSCDQ^eWvH=GxXmf$CaWH*RZr7*<_Dgm~crIo}Gf zX@~f|L&a(Kj1uZXeYNsJtK&x);v8fw*db^rKJI?xB|+;>CvCImEjAPLlqh3o;pX5= zXO^qM&lU$MI*$`Vble}&*B2!?oKXcp#A)Uzr)Gu=k$NDSlsD7kuygm^Ea=J`Z%%d~ zcnDt$2alF$GE7D-;h|rz#0Gz6hvyvc>C4E8Pu#L3;NlYs-w1h$6Mn@H#&QgF?m5WB zs(C_Xf6%NTq7j7K772Ti??gYHi%MWwGwt&|ikIshzNcfQCpto{qwjxDqO=2xXQcNy zYN_3QNIU8!vqe{1GW}MFYiciEGbUMbCZhDaY!U7rq^chZr?jE?7-(bdLM9aY_{Y>4 zIR*Zi(*6qVUMWj1I6QFTr-inWn?&ISf6*nJy%2BS8Tz_?99w!Mz2?-XApJ}*hgb&r z5$zt8+6zp#;+ID~bT0zIauB`v=0V5@i|?bG{mWJb#Rn!{6Cq`B{AGW;%)s2&8j(4) z33IyJfv@%~G6m5NMexv4$4{nrIxbe#pL$&UZaC{gU46Evv1mSUEPRPVY~&8nhA;bx zN6;%JL(5NK-CZHrafp=?5(+unbz^{rfWZidw4J7Z!^e$NfSd9wkVIY0V@PQxPV7c$ zp!xYzd8zcb@WnkJp`(3za&J?hSp^Oip082YHmcI27{Ef4HRwsrz0&7y;Gq=~LG_hx z;zij^ojeyV5jYj9``)O*l$zW3U1CV~y0F_|PCR#pQYl`=0Y|M;HyGt`GiRDU* z-rgx2wN7aL3}zNQTX-pMT$4CE%Cmc23t<7hzs_d`3(V?7fT&mX7VUYgudPZ@!9k)F zY-rf6IzIAzUbmJAI&9Ya1Wz$54}|i9a6G_*>f+z3{d^XSwXof(Jom|A4AHK%Eqrm8 zK=jnK|4DjRLDb)!WrVoAOg>IMe`Th{FqWs4!%vnJ<2M9Ka0Mx;1-ppMqyoEv!fQdT z2CkEMyRPeYPHqFjONaQZ{qOQif1Pv(UZRPfKfjwKoq%u&-$^^qV3+y@G)cL-lH4Wg zVnB$RqFUfZRsVMCU0Ms*aiwvPA}3Qw`}+Q9s~tBXY=e5G@B)3d-%i{LbulwG$&E=R zJV|%fEE+MiY8&K;!k^e(8yx5eMBdQ%GGXJU;wjDZ^*O37G(wHYttht9RCJ=3q)4wB z3Pv{*7w52YLwFw-c%LX86&89!_Yy>2XRN5PP*w+SpbR;qq0to9;Fnk-hAw3+CnvB2zoj1zKiV586w|}4)nrhnx#XyzTHHDTk z9~rWSB6B!mU()A+0%^`amis)=vMpAT?KR4Kz7$bUL}F>eE3p*UK6PS}t0WPq#%_P) zV+m1+n)_=#c#B%bYCQ`TSaX@Z#)?5XrT$P0o@8E@Tf}LnURzc|JMM{SXulwtAElDe zgN_Vpz%DV@l6^*f9f-DH6lBOMD01s=c!q!tCzIT_bV8zETgKEcWPTwYX+u{b_+S~a z7Y3=T=y=m&s%@{IIA2?gwtDuLf4@>={zhY#-4t(M;&PWy zIMc)5E>9Iv2snUkA0fUHV_q+i8hK?I6gg0J^4eOS0>aAs*d9%tV2dQ?DgQ9YKvk)3 zSXB5bV942UqFE~SlL%d-T5P!#-OOrtBJdPVEDCFv{9tnS=d4c=-R$%xFdee37lz&Q z)6g8{QHWWR5Rr!ALTqd45C8cjiCi1I26XoD(%fhTz*gvZF6;;Ql>4k8e67dMrxXzj?Qs_)~tTZ zYU`~jFze(hCMrxTq9#ROfk5XusHEuvFWfIIZCdB{Gmg7G)210q`U*#{_)OiL9h1vO z=H6o=FqV^mhE6vnh=_(AbWZ%_q)0i{9)D>%ThPTC)NYHKD?clp+eeH3D$V4T`FVe9 ztYWr(er&xpDBV;D`a}GlaFqQ037N;&QGGi#-?&MDW+kkkYl`WEIuPdg(_=V0;H9Bv z?yd8B^r~IlW+kAc8)5w6h{!{^B(}(Mq1U(m+$Pz7V?5pGM%RSQ@0Qr%bvnEk8 z$WS^#N2OW6IqkzP6`wuz>s#ED9`RdzlD{c1B)x9hk*MW1U}CYpG{dG>{Kn?u>KlH* zDZN{smq-0WW{#fC%kca2vsk;pAKo5JF4r$D>^@U=@T{erEPby@5k5#|v`vD$>v;cm zS6ek!<0$;S26j88l9HQGJG+;Pc+ZbFSmgwJ-8x8hURe3yJsY2{Ll?l}%uHe!eo$TI z7mXu$!tt6T`!0ra%59~kyrc-VeoQJh8bIHolP^t--&P9wy#A?n$VLDq5kk7|8=Lz zsUvUhb+oniC?NDChD`2@@@O&w4ab#X*o=D@p9T?6#cs7cJ>Err@AGp@ETuX*20wH$ zGK;6USyK~yXrb0%O=sdhIQga~Z)o{X=TaNt6@}cW3nKPOblljW-YRQwxWGy_6nBB+ z<)8E(?Wfl@Ih-}A_(pr9vXs)RH(0>qSSNcXRL5!%wy?w|<=Bmm?@abM$}Gq4(x#1L zV$hoM>fEAO!uE)@dO}IAe|K7@cB^o;5v^V|x*I2Bz~s{D4dv!?7x~C|Idpz7D>nqN zHqKQfDYdD<&szt-n-_`k1?@uwu1n3Kt`jVD!}`o8lf7wgT5lJ{$6Zt6m0v|1W9(1m z&FZXAk}1k?YGr-vq}tUFgSa9H9F5cR3w~d-$-T%#I^~BJd(U*l-u<%f@1{gi{p@&J zJ9}O`gy>?g22DCzWC(&=Rg4>Yw{zwi$uu?1mV#MvBG+l<2fCK+QsSta_VU^R%UIhx zbC@($X*&4LM2?<6{oNf4b2RV_ydv@Bf`zwER&fiAi%Dp(Fl`rc_-vz;YQ|SNO7Sbv>&rTGMKzbcR#^VV%QdY*8{V=|E+I@p1W1QIwZqLn3Fj z$?lwHjo#q=*1aRU*l{72uD~Ms1k*uFlvQA-{a~-ml%3bFzf<>4$$ZGsh-nEW3lO_n z<3$P*t2@N@tAaFfF1$d=LN3+W>@7Ae4KRtTHmx%jM|!%Abmkm>jtyCQDu0k#O#rzrOWDC*JqAqNXw#qUf*2UU9sG-i2Of zDp=M0Qg8U?)f#TCJSHI)S6rHE!1$Fkboo)JiRUIgy%VH#KDS41*AH^>c-gg)2<zz5GRfM+7Eq7ioJ=;o2@5Nnf?4t zUk2hu;ZZx5;3J_dB%p4O4_)k^SkW1_+`DmwkNL9rfWtp3BMEs(6m;LVD?rDXGUi{H zgSa`R+8s&eQ>b42`FH!vyPQXliNKW#L_&D&RVaIlvUP7U3|!tgUWR8B%}R^9{!2l6OA$K8PaW9~Iwj zZYPg?q1&k{M~Ruxn<1yoWb&JlzoHkd0Y7Wd4wSZ%7oxHi#I9oCU-RWap^N3i@5@3d zSTnuGCZd;?f-64a_q%n|ysT7d)zo%iUG!XIn}6fp*{RoD+ z1MH);A^xBkBB~!D9V}+=<8MoQ{Wj^}C_q3}19%wPcAHR4fX~HPA8C8ZX_OzngzsY< za0IxyF{~*SfqFAee+|uxAm>_~NeYi$rSX?~a(lhyISjKnq0j|*N$}(DPjepbR4Y~W zSl)6s&gPcR76_*-s5>_1f17vbxO>R+QXLI;qV}n=?znmf8l2d?y)Uwk12G*@-Q~=7 z=k`L!R)`qrLrLuE3?b(pn*@Q+1%MGBz5ur#?Baz9|IVbooObqijy6&pi`>qIo2lwF z{)D`t!vem{jD9ceVHi$02|7r(npb7?Dy~?6y$nJG6Pze#Q6>;^aKRM|OJPlPV=O<9 z&cm*}=k;;H<=OyO_8S)93lN!oirt4_V)gsh8R(aL%dpm9-8_WIg*m{K5?N9QUvP0^Istk+iDpm*ZaHzMj0Q40quI45X zopskJD0t$Q0Gk|YOo{zyQ@v3 zAeIT}%bM~GtpfcStXLw1`|(CCjx%nOm@|pX!E9s#Wll=zmhMZeW6uQyBYu$ai?1KI zLrQOc37BDt?T0@F=65dO-kt(?g!AbOx$^B$S$<2+s7Q7?O_VEsI*4TFcZg%dm}0gv z!4e2fA|b*49tPvFhKz(Viu{BiQ-fud#xv3OJEnc17SJp|F;6geVLTl;(wdrXm2R*C%Vj zal><~wO}h1UJcNh0xo_G;_>BObj<-}3pGLprVfIi#3GS=^BAQJ4)>3u^V+c!hN9>% z(aR&fTzZE$dT*xl^}|R>q$rxX2K$`7)&%T$+4?vRWRgx;DLO*AuS}Qj(4c@7G9B8k zo!Ud+AeU8ve3DP3aKSf0bpV}kR6Dk!OdYoJ*{$s_GJ27DTc&0Ll&i+Jh9MfG5lfsiX=6_Q) zE_vXpSHbG3X-RQ6HfItxg3%6e-M&f6Z9mTPJ|jm*Iv|y*GR468sn4MF1h6dm;-;LI zu;DDco%OFWH3`E|cfzxcoZse9z7+GRD9Dwr1a2Oh(DVBTP?zy#OyOeAKHNh2X|ho7 zeC>vDT)V~TRU*X1uM*gk*BO;aZ>soI>l=nkpLGG1I#~n6^4w0EiOrmi zxl%Db_YtD8 zayk=ZEEd~K*K3U*xflFVD7_8k66}z%fchtiPrl2}UA@qS`bX>$M-?UdsD)B@(L9x2 zZKDw=8Ly~>IS`3a2~9;Yb*fic=^TYOmoJ((gt6{O#LmZVtL0WQSoWaD`}8j8gjBS> zg?kUFwJ2q&?I4Dt;Xqq>ZSrJ;h{>CW9n=vMm5llPu{F>W{;;hHJiKua_Ni^&Qhd6X zU$<)LdHebX-N`r|O6A}=h^6IHiLDlfbTZO=N`+#q!9Ng2#tIV2HzEHY#ci3M>r|XQ z!B_vmrj@#2_-%N~zh^}LpL?*B-0^M~I&*s7EmJ_y`E17r_vMw|6Yy7nNy5P$1Pwx5 zT3*4~A^|U_4$s%*l7?a$Muf$vb5*(3zTTfuP@aFxxBAy~J{B2uHIO)6%8*+n#-#nibhN(ZBr@OmlR$Yb>OhjU(H% znfQT5$682;$<0|I7mQ+18lHEPKqZ)%&^9gyH?z#LDYx&yYZv%X!Ni$&Duk=`biJ#J zxW4RmW(tF&Qgv=ZEG%o<3Tlrc2CKS>_=NheNgix!p<_-B&&f%cxI-qX@)>Ny<}OJs>w-+pJ( z+B1V2!!(tk9iSrTtgYft7}5(fm@UoPVCh~Gea4P;I!Hf<`8_D~Y$$`h!yi@o7S3KB z`!>9%M(Nqp)U+M`0NeE`IIX(q7#`>ODY7nSfG7j$JN4R z=CU2B$Yoo(A@QES3!ksy`7*15mLWz5$!*z6ffA)hW`1kx2YuHs7z;FGMtM6Cv$QZ2 zin}O?g>?7PQ)94vVyA^TZdvH{a8mus>#jBz_qH6ht)}y ziZhUqLvJKwi*Bn@f4j%cMsTh+mKX-5c>34nV;(;JZ#d>`(#MLkG8Vj_p~=CogW20@ zO?3W!t^~!*{d=w4i*KZc;_pMFAkV4cWSCx>GjDV=j@Pr=NKWFHjsoyuUIC6+VP3Z% zsph(7sZD|N_&lCT{>j(e+;9kv|Gtzm+FI)kEr;L~<4_L@#;XoFI2+rvIV}1ApcvJ8 zA43J9kG�A*m-trAZ1)5u^L7W#ZJEf%M`MHqIl!`KrEgnRB)(M)WuYtEh_IP{y2 zzSNcPvn?+I2KXvG@rf4A_MwBBxXMHb?!t0;#atAXVyat!=RNc{n?3Zw*D~Wm1a_}4eq zqj+D)?7UY6oEE-vG$j+mvdFJGVWa8UywD<^C|Rn5}f6Hct%PF8E`{8OP(Dvaf*4sjJGHV=5s^EpT`(uoem zI2jREC)}Ijgi0@l2?smhO|83q7N+#IuKJJfQT#``=T0>mJ+n48yLF!Oydv*x8%j7= zS(hNM2sM&lSQ$Acn@Fw%`NJ;-g=~>eqmwa!5S16NmT(GE&$X~6Lp$2w)ISV@#Gp+{ z3r`M|!Oux#a(Dh%F=m6pYMoHaj{j$kxYs^&t}eH%?V0F;66>vlEj5vjOR&s@aA;t| zN>J3u>J-B|(3eAK{7vkzS{|$*ePY(81=dS5XM)U^a#dkInRXXj-<33%A}~zx<_STK z#KbbxDF-O4h*iek$UEzf=fg2nSllMIgE#dDqR}?PEx(+CrqQpUCr-eimaivAJdH8K z#XWVAZlSAZ#CDA?Nra>dGKmy*>jPye_ zP#OA&nTLiL0nePJ@^I6*{TR_P)G59k#EpDJa-aW=DMD7-nIcZiOw?11B0IJE*_w+s zr*^Xav8?8qy&VS=h1JQI<|?Wg8G5}@YUrSuotgzM4MMS#uWE87mIzUI)aus&Jiamo z-no0w#eK=C*s%-uZu!n|cA@f>LT%$owUM~RkRJe6xz(vNC8hqJHbP#Z zb&=`g#%+hYi1V~dnuoe;QZ2V}(>;Et2Mj-9?#J4g)~l^u@>aCBirW-^QPXv`RnX5Q z*`aNZ0lBJ)EfVqzZeh1EDXhdy!r7UfaqJMED%UyWU!bN7KNB^B@VkwiGcx@`Xx0c) z@HKlrL&H+JdlL@VYhi8U!<=q|;;V_?mLsWp0b5XU8_#i>L6iNO?ygx)bRUuG=+x)t z8@Pt+CfXRIt9r}i@B<_( z@vC4>x6iaEc^F2{_SJh@fUQ9(P5IEACQFnW#YJqsjY?pshY6u}ar9qjz0B7-P*}5q zo$&4UfG+dx{4r<8D}*?|IhvrS0lo&6JHez=b?;ew!Dl_;>mh!SD3E?Fr=Z?dLS~oE z^P*6B@|_n^GJjRIPgQ3qN@j$MZ0tjOA`3CFf+g|vwtxx(=(LP2!!YIK6p6I_K8wG8 z+yWqZ7_wzA;CO%LQK>H~EBxHx$rx_grbGagconYEn0~!_m5r!w;NGeFXkrEYicnf>Tpft=CSu zXThm$7?keT$w za`RQiHEuOox(t9c;$M9_wS2$!v1_tVk6lh-cOlJ?t^?;N7_BGbgDCDwm!ieqcO5-p zj;*vyiEGCRn$e3E=5~YN$W&k0n&`JVdJ=YHzc{4n7h~5=Xc>e=J=Vy*CD2(Ti}Omvx%%aBS=@+dhRyr!%MVhm0+rIK zrC#Fe+h=~U4K#(}G&$R2&@(2|rxlcCw~@MtJo=Lm$% ze&{OJtSy(|7Iy~Do;8PPXh_dmv|^);dMI%vYI@x4$yC($&Q4mU_8~4W39$&4>B_;a z{#)Yd`~)!?)g9aR5&pmLw{C9aNg?4h9#*w0giFgb+(-o(3h!*-R$bFuA#Wm!!PU4d z!d9ofH4)e(#D+fD-bWa0cLT( z*K`yrwKv}P_k>DyGNbLA-uIn-b&~%bG=m9nXesO5PcW5!x8-_`&71TbpXB*m-^PJS zJ>g#K2Hg+wfI_*@?N1_1s*amyS$iX~rADcU2E&59_i?o)n&W>(C+NMXRbY{d#AbwXzY`8C`jaVIwrt(8j95nsBG1pKb;fFGpp>LAr~r9^yIJVrqJ)P#r)R={6%UE(7no)OO)g7ratf_QSDYWp2ErR7z#ZsNjj|kmri> zmG0%Ujv)L8`_#Jm4Pe))>%bu%Zgj((T)T1Jk3bUshrIOo&TD9D#b_+RM>j-4JXAq7 z;)0OvC#ui=Yh}>_17+9mTwo&q5i=aa?gh_%J$2L)tCkZL?DVO|W$H0Ay&<|c!B_SZ z4iSes?X>Gaxn9NF5#W?}ly0vwZ?_JetpBB>fny`)npMNQEB zaK7gB$5APf5~XFkRfA1%Yj%Gv!}alFR2ljvF- z7xJCH>_YE0LauJZ8y#KcAEw533BQzWgm1a%2UO{Z?m!pyoA$KT0P6g9lF;9IyXs%1 zW)ytkR6ps`j-1bn^pNr1wX|MY;vZylPm-i$y@eZdlueE^KMlUlk-^9X9pI#i*x=BW zm7PN^N07hgG6cXg}>!i!plcands#U(zxh83TyU^BZGIxVU|dAen-eiZt?8E~lGlw2g}8Cz-4n7>w1K7U>TrL3 z3Ktpf{^e>s4e4#ec(J?-Bb|cS_?s^zUBWNMLrCKf{&~754UslvbO?vHH{@eh?ab$G zq<)|DJF_>Pzk&$nF)`f8{E1s6^Pl;0_Pxbe7*{o(L^&#KZ`@&VI&efm=vvI*Iip?l z?g7Z?QC66)`YZb!zIwNeJ|KCQPKvhBXD8SLHd6?Li0FAat4Q)n);*Gx%q8y$%t$!v z24z;s!(@*S;p=)HjxG>|W3k-pucW^smi=NIv6vJzqdgD3_8N%@k#BwcwybF}!rG^H z&>T&^3fvO7YnH$-DzD?nqJc?lCENjY4-~*`g&#jf{#1vGYNS5wp!a%ukjL6sXt*$O zh9f(7q;A1d`GlxxjPI8~Y?bKDXZ9pP;B4s+EFrWld=&4LP}vF8|Ek-LojaT-0?^-Z zRErCOZ@O`Dj0cC6i^YbgO5vX zO4HroH0~wnUIVW$5=J(u6ZP~gk*r_0DBS*fu-2<1&hXhy>%0c7kQ`$0PjEpWu$B=l zJ=cevmcxcSr||bx&1fF~hask&Jqy!=Ly;|ER^58r+fHJ!Bm)K&U1Nqu0tgLhcQ%c= zywD5_H5<&mz_>*rsx`V+%)-N}RY2<7I3ORNr8zHrxYuy%hG}Avx~ib94nfG#ta+~e zzGvmfipADhVg5D)-1<5Wpk!pbfWP}}@^ewMtiKNwQ`m8ghU?u5@{BNt8s~5CPjf81 zs9t7?sXl43oE+6_TmSy{;^T0RvsEdXV&lfxVuBRvrkPTf=GIx1#h1%?xS=VT zzD497Mw@tKn2#5Co9ehKi?OR}H73;Hg(I=Hjiq0BZ8!!AB?CuR=o%EY5!naujd8z} zM8E;+RPX|?9c0|ah9bR`Va{($;G}SX=JPHe{Inc<#1ICS3lTh}*+s?5dHdrh%+IrU zhbtQsMlO6(&ut5C?3u&1@uv$)o&MAX(~w=BOH0w9Zh#~2zy?)WPu!zj^X@-qUecAO zej@&uCH6(ttk?Au!Jc+}OWC-fSmu)jqbiHDfIVex%sO8A^Vp64sxl*-M+D+TvnHu9 zRsOkWN;e>6l3Mfg)a<8^BUS!ym3b|tLxw%a2#c?WOq$E>JxD+K|5!_6 zc6IeKxdVp6}cB6Y2WCn zs-UxN8`2w^E{*QB=Q1@Ry^JE7LvE-V3=~1@~=9k zfL|r|qjGNV+3FW+J&7rNYUtGNG;|Q8QMa=88Q^!i=kq%KL?LE<>>;fyW)m z41BR{d=#=XnnpEA_0$rla@fVBhvNexniU0!)Lfwlaq*`f#%(8cuZve6el{N!HPz4d zub|SIp)96IkUiueKL~R4;^eE@Bz9-9at0(jc;|%sN!5%&B0#b_`Ug2Yq)RXwC&O3j zQ1L7Js7oYWnSx^C=Yys@_Dj?1HAYG_EnRxX7^|8?r%EwrOMhi~zmaO0M@D^TCxbBF z{8P_5_O<2X=oyo&+7+ijWYxW0P6o$mM>U`i(7uGb`s|3BG)t>&$v4aV;~5fNi!IJ;n#;lApqf2LyqqJtRZ8NFb=gW z<3ypjk2dS3Z9Zh|j0pz&JvpWt+bIx=A{zL=3imp7A=<9cdCVnXZg$jKhvF?wCRp z@~&1R2_w~~_CCAx*CaUdMM-cO8L3i?rsoxI{$!Smh-8G5I0-xd~bX4gLDr*1fJ^<8bGI79Jr zoFik|0^~i{{n+bj{S<5brs0c&KF^+I%1nF4#vYTM$X`^?iKhL&jL%H@67HTp9tgk7(jx2>CCVBay6+@SV#ClRt2J~dW-Ao_)ZR^sS~T=j zFpgLYMJ`Ye;iZR}=zQ_iR3|BnZ}&HV1OofCEyh_a1$xr*e|?Mkdi z#X*!y>Jo|%elWkyJarwO8#}($thWoryyv1Bv%-j|F`rNGsuuLq-lEJqy zE&wA87VqEXm`GGdGcBP?E)`q5>o1W%mBERJp)Y?J5tL@Vu|Mef#qGKEM5}nO-hW^a zruUB>)^MGqIyF=%CVvMx6X)Cq^4Oryy-ak%wtAo;gN-3!l$d zQy+RHuxcVfxaf&HJ=6(YVFUNbmPO;` zSv@V2$;d~QF~E+e!WgKM;q!AU#T5QZR*s?ekcRXdrBq^`qqHHE&*dm8N`r5)NNikmiFkIxN3wv?V7gNE`(mGz_XZ6rRPF$gox8buzLN$_;IiA7UgO ztNxipgG}M@?h%Az0Ufl~af!?~*49~+( zSf+s?m`WwN`zZ@WdP4X3zq_5+HV^V~%~YJUA8FnV*Zc^vTI7a=4pYHzuEEIEKdc`2 zr~uOq7tbWc>DiwuD=F7j_S0FrH(T)5eAEzINt&SRP5Hy*4OY&K&a&b+Jn&vUmzZFS z5N?`Y?RM!L5n~Ilrev1%3eoB-RE)LZOxep}6otPCt|dl`LitxqLQs(s_t3}^Y80E| zQjEh*!8Y-Id4dLtx&~}9H#A=NKAhh*i)X3~HgYuC*8>d2);TA+kxTI77C0U6<`?9@ zw;dl$j{VFLEI$FNe3j!!y2>OA*FR5@Qqhub%uwZ@%p`IVeLU>)J#ij)Xh}HF^1`wQ zm20Vl_66Rr$&cfUmvPw*>2jvq2rhy%UQLrL??a@37K_WM#nmrN+zX)We2?AA>E0jC zd^Y3GBIYyY20IrU)cHxi2PoOf`+#4Rc5f410U#^TNM&Z5*Kc!dd$oz@a6q+R^Aq#Gf6k+=2;?`tT2zc zj&K@cD1%F}@z0=I?3B3?)#|bnDvZDRh9*!5KYnHMP9~*AYWBZVoi?4XW=8H#M4Vnm zk?{Ctv-=N%=foFPeCMqzkC@OgO9tLB8gR*iS3^%s6ALBD9m7Ju_-LeA@Y^-h;fb6| zg9&i{ph6h`ojKf3S0!_!mfz z_}dYf*s5i@uGjD3iF7zz76#YwOD#o(4*8g_TMh+GSEW7A79ru+mz&Fik%v@PAlm^0aTQ`gwcN{VTl^-frPu8cf}zIL<45MH+z5@oGGd%DmI5RdG#|4tN|nPjannkjQP>`&INNx z-$)E3b4&T-L$!r92|dd=^VmLY=01$eYdu=0h$yd?De@2FZJOTO+}18ccrGSv}H_V z(|j|Vj~>C%G~^f$w|5W`s&M3@!99Or`cS$~k*4$_hGTrE(Rv5Ic9K}x7{!N;h{SVJ zi$s)Nks5JYfv6+HiZ)mXBZ}cLkhWV?uQPH|Ej$`%@+~gE_tB0K)Jf@We1`qtadBqv zgo+LNpl*y}we;e_NSUtW#wd3W*N=I)E4$jIt8n3wD{xTtt~jHOTn|O{_lB7ujZk=} z>yH0nESQ2DhW>fl#ZV}Q$Ke)Qt2?g$kH3j&#D|#{KXcXf}%@7$D z#eH#h3SVbz73s;mgBiC^zEU^yk})J&epaZ%j<-s5%F_ent8d>Q3vC=pMpI+o)*x?V}yZ`ppeBtMyzy zhh{KN%X|C8{<@I*DfaLhhp)shJ%-Pn$BuaJh$@axiy+8wO*@7E6zI1Aoohk@y+>I1 zz(Zx}0yy{h%*#c#)~zrOsU7Ec>m4Rg9T~AxQLPBFE70r{tnPZ5h*%*(Z|NFCF}*=c zb}t}glOq~y&CE*VU7NA7z<3zB2xT6ugI=h@=4BoxA^f}=>^Dqg>-?@YiE$S-y6Yj= zkN-2|TLg;T=|gX<3pQaD4CqwE6DYqpDjbQrtoYVw{#xH-(0?_0c*IG*DfR(x>NQ=d870+}>JPe|jy{{1w?%OCW?OV9tqN^I<@w{*8Q_~IBA3nP zaxxdpC2NdKG&NfJ7Pfbi$aBSXM&u_)#FcSL^KjSo`hRToJl@7xcCmT|N_=`FCu`4@ z>5=`doe>RXht3fsil(GC0a(G(3qAuCEn&8%y&l@s~VlXmLl#-b#MP4S8cwO!^hrvsO@fBnB5BDVgBW2Y3bR^Jn3B zqaHO!JZtOnN#R89(rptZ`IBePzk&e7xXPj1-I>8b3~yEbaD{=m;Rs}E6n5>EWSAXw z7Lv;=m&e=7<57V$_&wc54}Lu4_&|_U#{GSBjrFP6-4fNk(bw5E?uoW^Vcl5jyad5T zGWvwe1lH7MZ=ePpB*{f0felovHBeU_}greH+%PsfSj2a!xSM3vfh!Nb^}sW&#IO^QKAO)F*^=kF?c`DvaJ;Su#3rgdtSw-dk0=9^Y@(g_A|2@zrF~k5EPE%?mM|on>-wiIVU+C&ovFfR*z)ZjPWaDfZ zQ5AOhNTSavo?uOv{id?ROk_29na4o1v+Lk?_Y=P#HF<_g6j;0yAJnGxFKkrhQDbDnlP1-qH)aw{JIuQ= zPxXw{rIt4_1kNo4hYHhA+e#;cSmmhh*?;J1fP>i^+yb zG6y!~MCxN=b4?;Qp7&6P)+)fV_6tIE14xbwj+@gjB~I%pvlerS=|i8Y#Y!Lcq4?qf z&E5t5SPU96e@kiqBns?b>qKEYTxi2QI~0DChM=C(5hnH>_{Z(i=`5Hmx*4uN*eCh& zEazdtwGNp+8kJ0RH)j zWdC_zWEXoQ6s3@}=sUe_mTJC|9Hgu>rA;;;LO)$WhZpun$$Ql-MdFS%Pua!}8~iwk zYW4yne=)+JOxHD3^C{_?PR`HX`H!nc9=B~4&LPZ1!gEOsNigeWfG)qBgT!Qcv5JD5 zyZ!xCW64r86^>llR$e*624-4D=qeg>TfmLh8mYHJP3yZX(oD)y&v-;_BnU{>dC0M# z4?2~t%73F2BYSo!PNn&y!spaeJePpQ>a6dPySVbPBR+CL3d~>=x@~F7xEUe4O-I*d z-tchr&f~eB+>_&pKgjPJ2*bn%d!luS*{eW$$~up5ITdZwEJX6|hCpL!4p|3vRap&n!YgV;Oy564Yl$1M^&&f?djVEoN{-Tu^$E)_mk{{ zkNTo3Rw5Z!UYH~M7UeX_nafPq(#tL99W8;N)zRD8V1g0Jn|X~Ddg;XWPO|6Gei<>6 zUU4?8pq*VT+$|ZcU1#vHg`;8%L!X0b)^S4>jkqX*5DJ6idvM5-dvIrYX(cmM=GqT{ z$;MJoURN79h+pV!upavn-Rgu<^k>&5U!quj4wrQ1ks-y&I=0LtZZ(ofuhLDFM{CZx zRUS&9?&bzbkFp4>zWgZB_5YEuIK@7ex$4!awb>tp#R4y*Y^jL6-apqab zYwkK79$P?{)u+^myirgg8W%+`^N$})j;Ib(<+Ky^r+@GPfejGa@%e%Zd z-11)V3Xfwh%(3 zd4G((ybt>`pi?ohTW#b!RhlAhf9v)z9Z-uaYNFVC*B<5`wLsEUu$A?x9f7|F#L>C! z8U)Xbj?R!Gyq*K8IHk}$dXb^@ulrV9*50!KCP+wlZ~+rMy=*sU$!jC8thTX>b6yb{@-kGtdq@(ZRU#5Z46=d7k|S-Y;w3r_D~aA= zAK)cmVgR3Au*rG)b8;S&rSJQtKiHb8dDr|)Rp@REx?8~qvbDUCwRpwQ{I5-t4~6bp zNn;O&Fx5N}hYXVvoeVGe*#fIZJsULFwk#z7jHR(Ep^E< zyD!GD2y6ZPmi8tW2}%ik?t&8v>t z<%s`m!%5d3JNUQuw@*}e1EF?luDn^P1wygHNI|{q#GJj)wyu-1JZSvad?BhsdIy|x zLSU59ymj4uH;l~e?e-+LosFou31Z02Clt}GBT+}}tsu~UuF$^OvhWWFuyjE#nDIne z4=NoOEY*fA{C)$Gg|R$-s&?Y%I4LRJgWjMDtsNwU)__wF464CcV=geHno|v!5|g~L zsdSb6T64S#%lEgOCg{ZdDsqeS=+g2KY7!MxY0x?{5Bpp-{Fchm7hO!^GL9Zq(yP)Y z0z&Yf>xjYENt~QX_ri8XaY5(_;Jc}&Q!oO~&n%J>j>%&KcEhmfwh7F&a@_f*y)|+( z)V)cY>(27|F-hKNL`M4mWe|KKl41JIXjjOkV_>8-DNE7W@asl15Y^Lp-wLb*YZ$64 zlN_OOf~{>v&jhjE4aJGBBFDl_tgvFi9w>&#xn0hbkS98|0+D?y*3>Bq_SVWeYx}LK zSfG|^%kW)zA?L0b)mFihQ2K=_AlH90mPBfV$#whoYK~V{J?CrTD3=_(UmHSCm3un< zzNz>gag8qokgK{01xqP({ey|y@Ls@9MvlAEuN{)f8L8jwbCu`RRmiFjCZJ^TKMbaW z@&P}!K2Xk?rMZF07`n^e(!v%q$hnOiJV0JiR@Lx>tm)0~{XqlAft z7xVtCp0`oO){b;dkr7&FyxVT;Eig3_BK|imM9A8dVTa0!tfOp$(#p=rP1K!Vx&&W$M36cu#XA?z{^vkMikmH5L~fegvC`AJ ziATmZL8eEpv4j;Jtqh+TK~!OW%VB_r`WA?yfKv^674!8(Nixx`Nq!zw8wh8oT|)hS z3%SjACpefnQ!l45#(8XQ;vbHUL~Ae|CPKp5h4`X|tLO`2R!jmWO6~18TzWTZ(K*ouox^Vq+PZL7!5l9W!U}|>#k@2|9h-<&nmJmtUCvHUtP#HgnAf=j4O_Bj%7Klp zbAWWwVW@1yHD^cYW-!BdYFyP$X5y%QYaVV)>+T!(KHp#A`l=PEv%N-CJ7l)UnH1>x zeiW(Byo_0dTS!u^P}eYOp89^1vXM^=o~<5ZtjIBt64>kyFYgSKO}pXktS=l9)hxgI=%CR&dsADZ=SUxryaT~Q#1HPbFm z^xHd^qwaNyw_lS#-qUIda1i$2*2eJrKye*1lf1=^4vX3>0 zg{I61DoRBYIz;~Nxu73;1w{OtZ(VCWa3FA4A=kJlK4gNzM)RMIW<3<5=~>ouTLooC zqt5p9tyhbY*Zq)G!>au2kTf1zLkgfp?oStnH+SF%&*&DIn+A`&F?aJb_glIO%-mh4 ze{(1i(n-~2@i8M%(Ppd`Jf~i`rN)BhzR%^8_evz;UPt-l`^kq=>NnGCB&>9N9D$ky zV0D1TJb92T3~gzN(}ozSKldKWk=!nR*wk^GD^J=0R}#{VzL7`?&qTo~Aot0!A#LwC zNrP~06>qX<-78bZ!=&<$G~A*6G_T0+&o>=XiX^Yj2G%ZbE>FyxG5~1H0c(Z;T}BBO z`wTeHW<401^a62-Ig{MNlDG$k(rd9z-;&kJHDeiGD_Day1m!nHndKNdCe{uK` zMkHb>R>FkfrTwby0zm}?=$lo(3Td1Obl*+$*S;g`q3*BZPO`d-}qL6 ze@41h<}qM1me%@?13%ZP8tzlRPvE&CGh~_hjT^gJ!L6R;nv#laSrdv{;=MEp`(5>K zQ>6xGG>Aui1fEMRZ->&Ret1`r?bSwX zP}3lV{a(^+mA*!Gsl2ERYA<{535m zsL;a#*4C<~NXhtNJka=71Io`kzr`?L&DvVpx({rq-);?H3sPviBnQ+4VeDuwrqw^+ z&h=FHMx;$R$yB+esJnK>R zB$sq|SJ6YpVzP6b!P4>mXW;SNIf7~txuM_@X(xbdFDDQ0rH&Yr35QwEJG|A6U2lR9 z7y-ctY<~&?Y}5WM9a9+dnp4!an+f{)7zbhDYPft*a;Q)rN52|<>%Gz}JVkkea%?9e zk}H9RH_qJqP>kyJAh*lfK*>EK_!KXpmSkPXwWCi-0p6E@iAQh_()1YrEEKV=;?K-I zXN6C7pLf&25Qi6Jewc4+5e3FFo{Ck1XB!KG!lJJKa^%x#~!e^dTgGIo=$vm34hy zs89j-6$xb98Dp$&1^vW;e|96<^xqBhZ_FjiblShyaffrF$nah*%kQWD2fxqeMlBLG zPBJt-j4H<1OutI#=+EDJBPg{)(Tj5uw3Ua(Y0Qi`iI^a}NmgP&$=7FwBXi$lbn)Pc zQ^u?gW>rt?vn_ga9$pSk@CWR=HjrJ-n1HT zD6qi0-n}O0r9+NwXQ6ib)K~hvsW@!k-QGe!0_1+`*Hv&Ln6%4qoJ~2YluyBS&n z0uMHsNnDm;V-4n8jixOKo-U8l@gP}BWrRO0^jPo0XeW{oDfhdN9;1u9*;emiYTt4*H|6^ztpJ4gPT^!wT`!==h;45z8GN8NjDfk(##y^W}6 zPY*p9qB*72EgAHY`!}dEM%-odGLJ`gge2y1tP)swg_Hi`20r6y%R z*~?Ut?G3NWVo+5BTK4Ff8WdTaD`jbu8#;Uf%8|It(07M4217{A!$P}vlzeaemS1T% z+6su!XtQDvoC*|F{xb9To5gOzd7?z!FjWxPa7im+lORKws8ylldz@I*@9ZwO!K(5D z;ONM^pVN-IHS3-C>{c@5dUjKd>YqX1nwM%KoQtDwP6h=B1t(Zy^p_`UreCh+Jc z;pJECho^82bn;@?gdl8;g)9MHm_`y{^*~`HxdYq~ZILAWPovDJpS~|(VImf~z_&jj zT4i@KZbbFNw2D#54Vrmd&f*MV{mjvw+eGdPb#SClgq$QW zw-QKLuytP)9Ik`z3~{XH#M+n4IJpoT>0swKX@jA0d1J+kJ&F5{ZG@jz1T3O8=lfb+ zW=(pA)l~cCCkHK(Va?b`#TvulFG17S*OK1O(+kV9=ZZ(8>r4?ThJl?)gub0d;-QCa~MIW)C+s!NbhICMjqm^ktnA)Ld-5c0( z{w|htREACgr zctS)O6xS3w(_X=D;rX5<-@$1y9LZtzOY`GuJd+toEW%-86B}ixJuUE4d4R=_iy$8x z9>}rxi=w7?5}|kU`Q2I~>t+@2@ZywtVh`2jWqw{|rpQp&3k6T93(`gdY-g-i!$mf! z3l#0`!*m4KIO;x3KV;f@ny#fwj$q;lFLo}Ib$=Wm&x8YR&ObxnwMxp#NA;hV#*nY) z6~|P%kp6ImUIg)h&A%G9+_1Ii11};HI_YHMKEl@dyTU1*A?E3ElHH4ds+?pX+@7`3 z<$R`7VV-;WO&zRu37mFXDZQx1s#F%4eVttH`z%PhMj1dk0@+-=fr z3O%6fih!wl!ae=SXu*?ueYH-n6Uxkhsq#y5=p9^9=|f?c;ms!Lv@aSYBItDuOSZ@0 z{R4^Qymo1lj-qf)@H+6#U?XnC+d6?Ihe>-R)jIpD3G2%r9Ldmyp0DpdHVCt5JCC#B zj9RtJ`_xA0`PF~XIZx#{ehD08gPuaGblL;z%|PCNIS;$&c!#Kb8{o`#_uggu`ZMQB|fxP;e!}K*KIA?tA(~5*sBK8^5r>l6}K{evD@(`9dZyMPgO?e=mHDwk%SbU?nC_bvTmWc@`zPc~R+LJ0Z&bQ%Vh z2-&rQQIa{bM8<=$1Ge3mbjRD}6>{52%SxtawaV6T^g_$H zq;mgTj0t-pu9f3QF$+9;?WX)c7+dfd4lg52yH>UlBbOt(@q!c`tX2D}WN+9jg+b)9 z7yUj)f^S0=-qloWA8&R}wY{uf885BS@pxIG6ihCks2Z?lMi!ns&O>8qh(qnMZN8qE zzMig9p_>_XPG{;Af$IsbH`LWM1UJZ5mSHRGDRjaYTLhv?bqV$VRe#mC#d}1o#X4^{ zl}e_xA<>Z0z|^Sc(tmDymyjp)`Sv$D6mTu9}gHG9JXQwk}-d$s)ps* zjVwekV&}!wsPW^CcuKDul`k&q@ZH(dA73$i#eNM2bs?eatDpb9%PUGg3^a|<0Q=c? zb2#Gk|d_iD7;*3Pgqhkr&*Xb{Wfy$_(t9!aKyl9`B}Gc;u)rm>h#Ta ztZDB2<+&NWTt)#(({ykiQw6yegv#VO%A7io__M|Qka_ue=7m09QIOEly>}%s~ zr?7)M9hb1}0#$3DWICr+2cspS`czwmC+Z6e^!*?YK77p@J8?k%_Bt$Z_+}<}UXqW) ztZzD&1+#gJ|76J;^IOswOT8y7zFmA*tMp3@TRn?mB`Z)2?AvKd7MF~YuWwgxr z?U{|UmxF9C0byuG$DkMfjBFy=n5bM5w9UR431l*SeG;gbSO~CJP*#Wh&=1<#smfJ0 z*$Q@{f*~NtuJjwb^Zg&8UvEWzk@bHV7#$$;b_Bt3R*Kr~N$&)L82!R$8w#ae*=8tF zZ5rZY_=lJsA4?GqeJ%ap;Ex-YLyT|HoUAPN&h?7|$V^ zhg%Pbk-MLfTeZpV^vI+JV>dF^Ua2x%1S~khx`aUC&MdxYjj4Cd1eaY)PW+2WY{CZy z@FqW*B+8D>PiyfP6{T;AJ;P&q^g2{Yd!pYL4&>0h2SQJqRt{sc?DntRg7=`rW!WUL zc15Nd<1e}bLO)_2j{bw2`z2qW_t#wD%sEOs+vtD85Dbl~P#Yu>K2sZ!lPl^J*p^zo zbcRoT_Wlk}V17I~ju)=*tyG!0xchoKyEbbgl$w<}_1rokBqu5iE;*P3F+LEy9Rde* zTvX-%T_*ar5OTFfb9My+)sBEKEY5BAC%SU>#^f4SRssK|r|0v>Ae^8pKsm*itkAkdXp3?c z;?G%=$Y4cRl@*5=M8z=BwY(o`$DAtS=RvFR zQ?TOt`ijFnXO;RU7#;DMt8dS5RO4u6S57$MWe{v;2 z&CBGMe^Dm6mRso+OIk6iLdI&%WE)~k-_&>YLphhMxV@AnH~%Hv_pg_=<_|5;2!4mV z#+_6b&c+cT$Tlp$%^ayoD#j;;L?KzT#(Rym-9lMr( z6s>$|d+hJ-)O5_)kh4fSDEd`7M$wCK*yQw4Bx@3}=p=kx{uY+_IUM5uMqzMxM27`c z7kn>cqWbW*U@;(UM6E0@cac}?lbp$)k8(3Pl4KSfb%Cq=h`E$tK7|JN0!~pY$j0?3 zu+b_SULjxvm)%HjW>i+tL*_kp>ew7zkpCy)tw@c(R)RUX;qbqk!@ruQ3G17>qlbo# z%xsFi`gBQVb9zx2&wDrK+Nw^R${NK*T(Nf!rjZM4uDxy&t3v(dz&fRC8j9VHlfIv> zKQ6RFysC3_&kV`HhsIPMPkc>jaKA<4<=9l?D%Nl`@NO+MJ zniVfUSl99{q(eOXX>cch6#u{7Js3(D3EG@^&*qY4g(R;lM4~EvwRrWP`Kl{wVE!QA)%Ni+h6G~fU6uG^K-=?wqd>aThNE62 z#jYA*->&)a|IeY@o>DeF*IEVIz^6?0>g&m&U+-B*bd? zV0kBU4=cAO1R`DV$J`*n;csa(kxC@miDJlM${89fZi_YS8B$((3qv*x1&Lm z?RMVQa*ZEy|a*|62QP)E062p@m0EYDw%^^>pgm(X{3t0e;!R z(8)quHMbO=(=5n@@V^DC)-Bp6G86@u#muYDkO~O+O+&6cmR5y>CcCe!At_A9;*|$- zlUZ`Ww|n7CeFH%fbk7&4#o+Dy-(gZ7!Ud95n)*vwBtO9a&uh^DH!KJpaG59G&n>t* zR&b@>iVwCQxT+3l_$`x8;LTEXnLRovrD_6-CpYv}`lAQieiS!mYQV9VmZGr6ol#9~ zDs%cNjgr;U`9$EW>%FcvX8KgEeShsqO@VOU%oZSibFI&O_uz`nmYu)GmXp5M<>U_{ zPGVe`TWHT8TxMWq^&dCRWhI@}(5uqRrqEZahuV=M&1Hr|zl`YDIt^_ZuWmZA)~a|0 zO}js7EpD>(Gn+Na@%QL3@2Ri5mummaiC#LtEVhoPv^R};kAb_n@#*>iZQ0GIZ*?_O zxB8m#nUR~2qH|wom?J8I!}k@VS!V@L-4+ejsE_hPSAEt|dohb#oc91RK+eDXkIhwU z4s#XJ5(#{(=6amp-~$TNvh0xMn4SzNEm?{?%_pDEla2-M)ry^`VcYcTP3gs5tRk?Y znUOy+LpY7Kbv$;_D_!|KV&_evB22}VlsR-bgCQMSexc)yQEWMj;aqsJxgePY%U0{g zHspNiGaibWvHVDmf!Nme31Zo{3Y%`q7KWc;vs949f~fGFbnfAGiYJz*H@Rsu(~*&5 zs9B5(seMb##Vp6?MJZY>C!0E1a^Ed_^;BytGlPf69L`(x<@Kr}<9UDKH-+c*#A{*5tiL9p86o6C<^lqLxME0k4h(-aoqYvJAf zs!6ZIDPu~3%XLrfz}t=??qJ&()oXHc|Nh?%K^6MDiYf>J7aLr^P~`?h;WZ+ykH@lS zRrN~su2mTbUKL;OJ`Uw5i71mH>J1(X=4vbt)jx?R#o1kQel?@Kwbc6MM%!vamrst4 zIlh7>{=V5^>Q*c)q;w7jIuz^G<`gElGW_x!ysJ5cH$FOgs*{gEV0wa;k&i=^YJoOi@v%0`2F~Q2MTo8!$M-r ziuy1vKrYkEHpZTYX|=O#Q$DM!8Ct5WJv{SK{?794LX~h11qcCG*d^t6ykQGY;v@Nt zTQS-pg+HiQdqF4uY)m14jyL&znVFu(^W64{hfy}oc?3Nr!4wQTYa6N)%MZ>@3cQ~z z2>FV!{{k28r;wJe6WO!n!y(UQ<+9$-s@*VFu1(D=L!es9l+-Eew<6c$?upi|09Ax- z8fWEWO=8RLf{o>H7ptHwu>xdtX8dAPY-zgOLP~2ZW97`U=C48Q@tT2js(*uxBR*|9 zo$lf*%Eer7lysWyoae6rw{s>3X>4q)@h#269?hDp<-AC#_%G7GU2-HQ0~0Jzxs;S|Ka7`Suez40QSTn2 z+Mxd$rf?s|I56%DOE#B!KHi%H(+r`ajDt-Vuz{3f9d&Q*J`+1+JYoS8oPs#h>`&|W zptU_1^;9lA)-sO@>LEnp3tJ4}uIA!#;n8MKUc@v9Gsf!^>hjx+-o5|-0C`MV{5nP* z+Qe*!q;1_wm#ophXAWFM=L3Ec#v8aRg1t3{Rwn`eu zDaaftO!d2&ek4~Jg5=46B&73ry`HzpW!~V<8?nt_y7!Z)teRQ)5SB4I^6btwYBYsr zffmi&M|8ct7eLmhpAs&9zeQzCY~D0j(5ShXDc4%NMyQ_tQhf1 zn9oKmkKc1rA|>Jq_7az!CNczF^~A1>I9Nv$cD;Eq5fd}ZoEF`6^%H(KK&PtL{4vCu z8(%ITh!ip+D;N-!@FXPnhhwq&Gy}qgCT4uSW+N`RXXAW9P$^%EYjq(c#+pEuMkbeL zRdKS&g(heA!Pf$b(jmlOBufmh0C!}oz~b#QTA@I>C6a-xg|#$z`*aFh zm4DGTZphgg$}f{C!@O%z(l*Y4I)O#na&jAkOK5(So6emOb@TJn8JlD-bMy%!tu)rO z?Vs5Z%e;NVu`!trs^K~c4LtbKc)Dd|b0)1gOatU(c41XmAq}ZrNAk^-4lU;YuhzB5 zmBRh-(DRNFHfd+|`EbpL5R3y&lHZcHoHi>dr4rOpN%J|B@9AV02r{~mX&H0HYEXIH zMF`lHq0%1JQ|{?-!|a03uH!m>T2|Oc+#q6f_ss_hApI|hSoL^{65TfSZSe{b}mH2V_&O2`aY1+@UyZEJlQVZf51IP^71mI)p?auy> z(rk3dk9901dv+PM2JahIo2UgDbAgtzv|#7x(z618SZ-B61Yj&Uds@n3s`2v6(rR2z z4=NmoMClL`Y#u9PIzzNY1X}@CR-amV*h! z4P`9*GS9bmjM+Fl_hFnQt`Yp+hv^X&xZ}usax5Idm>G^0fv89uE%{{w3x7(%X@yGm zA>9uZe=HkEf}EGXTXkAKjsf2h3A)A;R=lF5SR?e-fmV*_kc-giSTs~zsb75BR7O;I zx{jAh(i_Zf(;&*P6)bvDGg1|c6?<9tiaz2MBOs&B6jt_3n-4IP(o1`QI3aTYe)iGgj0mUk{| zK21~p#y>2e1_`^tJYYY;CSIMTO)uGdg{rb57bm6#mrWa;@+3K8Y#Uy*Vge_y=`jY_1kiX zc8{H)tgqq+%7MoY(IKd(dm*` z>SxP6{4)GNnQU%pJS$=DWdz07;sH*^C>|{8f&Z8iKCF7%-B_z1PH8q@vmD^Rt*1IzILL}o4r7( z*ErDmm8yUTb$(6$$+IIj$YYtV{OkYJ)D7FKI%BB#KR_`QwoE9MNt63reT6Y{IF@-a zDMP2MzlW0ER}$*6W~Wne{=Zx^_-Ri2mTx~oSQDf~A&|~6G1S6-;b?v~nUZ)s#2QZV zKne7Xn>C;lY}N4U81yC_M%4kwhzKe=T6Mdlo(>4=hH_%D1{|Ay*;CxK;Z4S zhlc-{i|a8h?Xx9nr>{$v#r3igW}isnwEt(uM3eE$N;mZ`*t@Ft#~oET{FLX1`PrG^ z`*Ys90)flRm;{>4b`KRk_x$u_@hI+PQk&wU7X7bYQsVe}%NWSnPmjIXwJ;|U)VY@= z%3JpG!-vJddD&s9O2-yD`XNKsJ2yGb=0`lx*MK^cFVigWB;`cI>iF#XYua29aP zVda}ligFR3p(u0&YeeNyy9Ue&(z<)|t*x(B-F4a2Joe z1KC?=0j(te@j+bWwMB$Lk+ww^?zJLZc%W4{=w5T3fA){OS%&%)1xrIQuUGW zal&)4>V%sOjY70m^sx9g@%LHfpe*)F`VpB=ppq3_KvZtV=bzd{7$fu&-X&D)Eo0Q4gKaUg z%NZyp^9 z`#@P@Wy33q06D>BM$FFBJcuivqQFDZ<{~uOL$kThdFal-DG%%@?;5CnizQGSX ze3>PW>kn7z-5GPjiUbf+%TtP_-r)O!t1H6cZ!1FnR991+s;rzmW^mF8R<;^A%&|n1 z+h?VtvjW2s18Tp?JHPCPt4SZB?kNLktqKKTW@DMd!Od@eOIHoO`>6;)J!-`KgROe4 zE4Imt%$l9ZX~tq1vtiauIw_)uCUD0<)T30?%{s{1f+5(V^IEd-mBKJ|FgAFUuJFO{ zf+cm+J|FlA!lKYR)5&yS!gy@^#1BZ36(2RQ1metKX;Z%2gn=yi-ARV0x$QuWi(|zK zCK)C?MwSx1O}UPnpPSVdTOiMcwYGq9r5rGg$>kqd*D*EK){^6-ch6EW!diEv;MK#! zRKZ;X83-C;-pnmapQ<-lmmxvZmOnVQ*qM-aBv8G{=G1!0X_R{k=WFB{hp|j>@3U}~ z@HxN)G?c5$Xg4=c6fr~S0&_Sb`@z=zJi8C_w`{+wu|K77e;&7^h1*a}dU94`%b)-+ z;A{_-cD;9HgRj$3DoOYoRk)D>HmxbVpO~jBHS&K%?;@1d^OFm<)^5L;o5$xBl#e4P z>RgO{C*}?@G)+*~rAoqKZm2jMWwVmeX1hx96*(DwQAX_pjdtUveEpGwA(~{cI!yg7 zqgd1oYoR#I`vp!Bs8?(Iz5j;UJesLfTLihgPX4Vr_}G{LM0?Ffw=N`WV}@1jwh*@u zt*L(g&fe*jb(79-PxWMit0KOC`{gTi&JfoNb87?Ys$E(Lh_W7*)Q!;5MTT|W!Tw_4 z5#OQ#6RW{4?AxWQXBS@T9$8S^yp9;dV@CmX;u-AIL=`wJ8kb5^1TxrRsv1~|iCm2N%p8o(`ZC7)mAWz%K#=58k zv!cj}PuhZa^#Z@ge7`BgNQBn!tY~SeGiUMODH~J~%vNiU;7t_i_@Gx4-Rsj-aO7I| zwW$)+Zu$XLcj7EW&oH$*sccc6*aeBZ2id$z7>o6WrYgzenTRRxhAL%Pj0X${6obh1 z<$cX*8MuzHahA4XDTOYGzzf3h75n8M)u35GREI7cya@wJO_M*0<lV7z(0SLIW&u zm&Rd-yE~^JW;xRWg#W8v(U642!zMWT^g&jSTin=^-2mQNmH(fMrJtZFxQDQ2Xw`;L zBG&jc$>8D}1SMBo!*|~0(U3VXikf)K;Lwtr%!;*Z2m<*n$wZ@155uEpV}>q<0q+rx+#23k@T2o25zViY8ehS=4APBj7fpmmwq>iVTe3bSxI`G- zP#m67UNW*Z4pz3C9zNtw5dm`Np6DS&F=A-m6)S(>6#XP#3J<}QT#V6t(akq>O#*7p z^x~tBg|5u!iw~#HBCT!wPXUxbe!E8p6&e#ZHmFTfGKPBqg+4%c@4ZL2AxKkA8}^+* z`ZYSmw}yAwajT`hmc`y?x|vKeU-kqPjq}?}Q%l38ov_!cLQ;iJcoj{#qp_w^fz0S-!5Yvv-Av+1N z%Xd+sq#X}@P0aAMAzW+hznBDvDPskPgJ^Q zG`ZvDyIUGv4@laHDIDGAR0_>5>T_9*%1|f<*Ft$@5S6{8JhM8akyEvE0_(MNhx++- zKT@tvJBs`(M6D?-*WM%T`J5~e1p4@CnvBpFV!R&aEG5hsTiRdGVhY4d%sRTfkqCXZ zo0l;y?K&SI#*uGUdh|w5i&eviJ$}yb*oUNhZlvgEWmQB;Dq0Vk&Lj!?3GXWRMwmtl zuSSc$f*cvSl1uk$Y|H$K9$=!KKDlQ!bzrR6DJqx=pfmBkePY;3-_(TXOUr~6u{e-f zX?dK6utT)>iONgo{d704`BpFE9&=(c_r4|&+@D=kad{ra`UWU+S%5Viu-?C;7Ui(3YsA(iL_`S$*1c3oOyeb!A2hrkE zwr7nsN=(D~H9!COjib-uh}`arfIHypF9a~Tw8mcsHAU18=tzqQhU~RWt?z9R> zPYMH3#L9yj98zaq8~I%oN5-5Pp=hq7qIjeAzG!CURwu0e=ym&w!rbVNXCZrr+wyDS zdwr|_$t!;^tPGr>WFU!^QvQui))U@(fzMPG7I`y6lQxlu+#VgY6*_ZF!GlaBedH=B0=I} z_&n}Q%|znR#|O)@x_nPUID)cD=tbdN%Z;S|a&6_t8;a&6iUsXNU6lyiMxkg>7)A6I zqBJdA9xnEkq1$w&p~k`T&k-zQ%Cz~Zn{Z9_10f^D!brnn9AiBS9Q?M$=Ee+CQ0__5 zNRaNsu9jn8yVQ{dX?HKnAjIJKU+)k^d++mC zVXNQb&Ty88JQD{yX8aqgTRp6s!!#k++#Shw*_p(-D%qFei7FnLaw#}#j4`P|f+G`U zs!k82jZlP*i22?=LitchO+tR-#ax0bgeF#vT&+po4J~Uxu81_QQMrKYt}zXIoL>(U zgl03;oma(po%*bp6RyyuAk$eO8*N@^FHsYc<0yK|hGDZ5xGXY#eXF<|*1YE~-l8V+ zxqK~f?SJ{6|Dfe)^QzpqGM!&P&^6hxf@px;nWiDu`OBGm0be%*DKeJ<0 zSL6zpCC=%Uze@i?Ej4YDXbmAp1_j~R4kI7?q4pkpcx1KBR|(4@$t0oaAz}`#x|tO> z))Venijjns7#+f_*)&~ixa~f~5P9q^DEl~NaE19yb~Fsoyv(Xg0i~XG>(qXeIFyHO ztjXcz^ecvxh_s!IDv}+Tfd};s+m2L|?V-Y}(7*`QxS%VaGn$ty7e%YFO1H$`*bLU# z-#4Vq`ColdR(Gy`FhMO%{SdWnB79Q=@g52!7y=BFk8_Kmbz^pZd*6Xa&!YsrfFGxS z=@>fwX3m3q_f2~#N2KrvF!3#%-|C~@E>I=8+*@%mQ;-|&Irpy;QnnSBEcza}@ zR8BgF5M`a=EJ<#SsNE-^c!C@Y4aqy?E%Mh45>&6g!1p^eh(4$56}Q?Lyn`XkK8akU z(3Dg$cUDe!nMKJBr$z)xf)Q!QLPc$QC!XU>IxvrIqr-fyyvCyH>MP1{%(~@Vfr5Ub zZe;X6Tt5`-a-F){X@zf5Ad~Z%BuC`@wm@Vtd7ciyLaFHBDEs})rzkwHMs-(#^9hi{7}1}q;U-#)y_lsb9{dp za=%mCF$Wfn4@puV*n{cBk*>WMsDfs`+cX8D!ei8G)xeDCfuMFoE zd92v?i?-8~9?^jEKmqfW+d-wSwLaoHf&&|-t$(2p25Rl~a~P(L4;*ID zO|Cw%hFo_iQEpqf#j~)8gsA$%ToT3P;ruUAe0x_GR8j&CR-w+i_Wh@iCe}uYB#o$A zvO|`{k>{K0-Y_z99WY31M0v{ru#uu)0v+s??b4WNwI3=CU1177cR#_S*_KTIPZeZ@ z%2Ie+2N7U$I1>`PEY=SFSW<|!?zQ=vXK1zTJLN=`twV-G>9nIAE`Ga@y8~{FC*Dvi z*PUZwSqe9%ttDK{K69ZNB?QxYS?Bncrk@N;vo=zdj+#~W9DDZa$J}Z{WuMuXhh9IA2UdXU(iGhDTOIIb>SYs@poO96H6a~tpPAqey)i_Bd#1#N01sFv zI>AWm)Yi`}HM6So#M&=S(;<25U%xdIRobLaI#!e#BFp1Wu6-cw`kyeVsjn~w>K@lu z6po2Nu|zThOzP*2+Xu{b{TO>-LMshuxYpF#Zq|=<&eX!6rOb+q8Hj zx@+=rM=~?AnEa_ne&>In#ret{J!LyX1y_~=2RGXfc??G$BXz?xwCH*p->y`-IgT8< zHixFDWZCtq8Vlb9RsTu#^&5@_Tg9VmM|NsM|2U$@c8xO}qCEUuH9(DE9k2;Oa=0s3 zm#EF)kwZK;L!UaxG+8ks1$u0!of}cS{()!@Ik}RR=-IVHsV#(jP!-`2?)kgYI?`w` z^>{yGY=&_hcaza^)>P27zOaU!owyPPw z&6Cw{FUlCqJ9tY*>D~;bCUKpNWHDxYu~ND1BNK2A{sekm?nPiMY(?vy`BSdzHaAMp znv6d)L05gem(O2Up`U~}KkBoGM1r3+k?fyei%CLZo@yiQ(kn*#KW%ejLSG_~p5I7| z3uo;sU;flb;jhqN`a|CM1?W&vh^2CQJ|f{Mvt5+nlvHX$$kOok0L*K06M#O&>_`Q^ zNtmkH9B$^aeodgxmu%;J%*lv)C_8zT*dk8MS_`j@Q?qSY0@%hO=7kp{`wW)o!)QGf zFGGJFJ*wKX5ljt3;#ya{fl!~KSfuljR}xo}f#gt$j4hsZm0GzVX&mYD#~jxyZ?{g^ z3@*b#@WY9eSWQbl^(X52POHnY_tDu26R@WkMi{7NiUJNEwW=T3DBz(xxZE}s_~*>Q zkPx%F4En<*#*JZ!mMP(0!as>`DyT$PikF%zHk;lV+T2lm;=ZM83n>@w)6u{U6!Zf(>`aX4yvG`#k(Qv2}MB;#U541-@+SjpO;0xUV5Ta6S z?q-Wxh}WND;Z3fW)Bi*U?#V|MW!{MWXv;%X%-_r|2M$Etq$;D5=-IgCD{`r?#O zH?~))BjH`>`)eH;{ZZ$uSqR)2U98VvM^iEp>08vvv#DTtt3^$})cENej&qnVvx!eN%x7Kui_OO%MV#sI zF8Q!}^>s+IQ75VWK&;7X&*_n>K3SBF&o(3Iup8WwG-JJqy{qW=kdso-U){1d%FcT8 z#t)u)kw~RdYWylgY;`4EhU>DON!N51D`AfC{qdvSe*2$E=P+n^ zcRjTfP5UEeM=bK-9Sul4FD^G59nnt1=M(??&U9x+w zIDA@u*haOM8EVqa1PYeEfq!~X8rJ4HweBFZx~?#zSR(oCgu&bOE2@ckmw$Na0$J<; z^{jEn{Pqow9euaYPprRFJ7Z6n%hu@81^X5<{)9?2N1rB^;!uP@ny((capi|rJUK+v z7&IJOfG)$*BB*7Hk8elVNtg(LEipF~Q%wVT72HjPagAa7Na#%f_CEVe3Bs-uy9gXq z7wyqD!vlE)qO|a#KFWQ$S!4AG^=5)_>+CcspmbaNVp7@zudj}Qy`zw_U+fekF$Sj1@4r^3Qm(aKl1@1IzOCJ<2kof4%ot`x>{V})Vfnc3g$w92-prrIRv*ZYDhz`rK+>Me>8=F9v$oe#T}v_mj< z+SK6Iylcb$X>Vy@b(36z4!j%L&)29?RgRxTCtvPja!ft@;}9}bx|j{9 zPzHsiO^2a+atviO`UzO~)#0!Bl%!B9bz*k$VFzFTtP(x!EM%gExgd`N?GF4MoB(}; zMVpm}N!?x52%mHB(=2tKLVB(g#ojS@ub@T&`+g-s7jD z8%t5Eqz0+*QL(Pg&dFRlLbbiPC0m9;Hl47soFf43(HF-Nak?w~r1d+)yVoQdKEc>nV5bJ0?)*0}Y0W>iH_v zm2T{m5&IzvKGM~F+wruMIx?{@UVRX)#KadLD2CXmo+FIvsvAqAi%&SlV=<_Wda9%o z5KrO4LGF2F2R4m8a0yme@cS%KwT1ypi}|zZ50>ime-qkFN+h8 z0iV-=Z=Ox;VsQYKH0L38xp1i+(i4x=WXQ7WP?l}r4J@+e2pjsXol&jPjZaott&2fZ z5-z%Z-m1;sk$8-RAo4j(mE*;4S|OaLQ;7U!#)D!xGt)b}rJ61y=iCOf*fy=5>0QvanbqTqdAGw$Bl@xIN9&++lL&TV<-AwOi z1OWHcbSmy0TJr*W=cC}?t^ak1Why#H$cmh+BIAd>A!T2JK2zCB99R>sa3qn)3=wryA&1i`e~P!nbHY|9mV+&vkC-u&2_dLj z0V!PkdU#k`zqY2|tnKFQi{|aBDAfMahP`Z*Ikdf@K_Le~X`9VN&BM{1|aGw9vAhr`*n!u&{6s-~>^E@88T~ic*?zs;VOhe7=VUhXoo< zRWpwnc9O+OFs~D{B%rFNaqeuMHsJZ|&x9s>;Ij4JVQcYGxNYe65TJ^Yqewurm}F8i z`MQ_y9czY(z?%3!->&hukPfgC#5{`??M{NWf=V8}eDYEaxzZF%;cK0gHQLF~Fjeu8dv7Rl9zwC?9HG2)s*_d0?F-6?X<-_M6UJtt>kFS$-$G$5K( zFo&Dukrk|B*!9)vYvT6(JQQE1CHU5VPup3`*cqK^wBxZUFb}ciV6Gl%;0TqPz10^R zB+rTICH1HatT+pdt_Wh2^Qv@5a?TFR*f%&auX`k_KzjsSYPB0wjET40L3qq@cq|`k z3|;X1X`RzBW)G#iU**VvsIc^kN7{He4A(!`MTh9|wdrVuKuDP7vaLPSF<+l5KD@~U-Fw_?gA>2Si@5rKdZA>bN%9frtuyBOtXStOAxqcz z&{86gBS~YeUBgs^5A|4(=l#{f8^S+qB-`8FCpnE8{EiqF9L5A637&ynZF6O1e2w7D zNaJC9^d+AiiF;2`1Ny)zw4F8~T`8@Eq#oZB46}i>T{Jg1v7PBNUT_~LNVv=woM@29 z#X~G&RI06>NR1}?+P`kh`^WHuY6F`t8AFtZXe$=F$*;z72M*%q+jx0|ViO4NdD#Sg z_{C8;Ie>&CWiGS=vL7v%Ho<$8ie^ z*=lai2RWjzdQVe>oOfC#d`#~0bJ-{de=!bu1p=TB^<0q;K23?Ea)U6X?u^kkj;`!rhp#g% zB`|I3;po>jDAnd|o7&ned!lu=q)`#|6mmNEvaMl{y;l`>$^Yz75y|*n;p6N=PB9)_ zYCcQ?>26Co(=S zC!~!YxJYj~?-ZeLY-DSiIiMYi)Y>b3=n)6w|G%&v_I*8H;jdv-Y;TAiDk8zf;@7DM zGomKD3sie*Fn4B}^kbInWAhnb z?lihxC9;Ojeh&H@b!VQB#eaL4$dS{L&!kGkmDeq2W9SHC2;#+RrlK7Q%Eug?pmkym zvDJFxxEBGBG57u{nz*I|?1wb0V1F zOv}i7&e_s46WfPXwF2n82$A%N0#hEQ*dR15kZrK4sAfrS2~AgbSW#G5Po4C5!tNvY zi63RvnX~Tq!4}~%NRN`9OsFzR?2^s31m0UEqb=wJb|>hO?~><9yTZ^*pL=cpL}#rR z_ImtK8d9saS@m)ZdA{7p6`R`;h2Y@WL9K&Z{lk5ZTFAo152{nE$&sIXR1 zq5L+vGaw}JHNE;DebtL{{S2$|Cu!IHS!%J)vJs>|uN3^cvSZw%rKKV38J3=^l{KV< zBq;oRs7QnMS`4H_!HheW?K+STMP<+J_#%1~#3`1IU_p?dBag}J3bfWS4oGgOutx^c zq87r|v0p4n{SYDU$U~}b_VwhfehEt4zcC;py~Itl8wtUvdqcrw{Nk{K=$XxJ?k||) zoFdKC>!Y;rRdp>3(&J zwyX*D>cdWxU%gFbj&a|!jY%>UF}if#ULt(v_QWY%pF9j==Ri-&ku{$}ww;YMcdDkD zDkYj+zoEiVD{?y;M*F^r#rLpBxK;EBe6@E)+B%(~SJe_FR$~%X;W?RO&j()Qt3~7n z456mw zrD%Tk!VQn(j%YNE7>e2lXlziia?{$+2-DwLp6@QvIyruyd2BgUw3a}*Of1j2tfRDZ zpca~_bhnCo3Z<>}qQaFxSN&iJ4DhZTSG9$sC&{qyxl1o=|38VZ2Q7xUZOnNu3_GA| zUQwD|tA(96KWUI;b$g5nRhdVt5wW!tQkoVD8_NZko)*D{`~EFe(ZtRg9G?dow1vCU zn#3?dXD7%;@7@A9CZUj23K*$f6XJquo2)VO>z~Hon(NckE8+L&V`$K%iT4{E0&%0I z-i?nDR5X}+@~c#*qfgfpzHm}<7#2NbOfR=b<&F?PH)mg9W`n^o*N#4b;C zW_8#5d^dwrv8lOP+g_Xd{Irs(yI3X{e%L90z0Z(QhZ8PE5MShUDHKCWN0pN`80a0M z$nmW|apTaDSpwD%Bh8piL=xsp?1>LCdS=P~vwbe=L}DLf!T%t<6zNWc?Wx5KcXESAtG0gcvgUpC`&6QKMWJ@ctkcSH$_q*Bt}QSng7!Bzw4@7*6@49RT}PMPwh@RAsrdU|<$g17NPSU=sxCTF zJh|W}V~Gpd7`@E1v`^C5?{ywM#{!3{)m{*+)oElKneLOj%#1YFl^301ae_Nnd2p)4 zwqhDqp_9$kDgx2=MZP^rUkSyZFZ+ERg@0ikIcRj)bKgWFMy%oC*W3a z+{fjkHC!|K)34_VWIEyBjwux>hZWmG9Z$b<)2(Ky(8fI6So5^E{I@_q*I+rSc(d~I z<<9+K!eC&^T~!-_qABKEsbdmP8%x{G1Cq@7%8R8qAcpH&$C+hYMF)0@S6A>LaB;WSux% zfdhZ7_zT<2JpqZ$swtZPa%;DhXG!FDD**q@oJtmv?s-Q0zq^Qfe@Tq6UThaj$-o9V zb`4r$QK9;nhYB4-=lK0_4nWOvpVXvGR!Z_N(eja$t8+JEuq_qsre6aXFEM%;va`w0 zQp`scHn>aqMxGyByzU+2b{b_#@n<%t!KI?Yrka2b9UzSs0{cOZOadykgMRi=&B~@U3N~myQ79C z*^EtsFfNxmy!X1eDmpsacb~pKJAriIu^kIBHr^~yNMF65 zg?DsFu@E|(f=R&!*N*2KB0ccam%0wIZWfvs08?>BZ@*x(R?wOLq(d=MEgpz=dQjv# zRdzJQs2hv^jO6O>9STFnQM#vc_DKTy zChE#I-PbwNJ4X)AR_);MZSLKe*j0}eWxL#QS#$25HVKXXcosoG(N#3sFnoEP(`(3U zQUWVU9B_(j9V;x0m2?WAINyvz3%PM?Od9-jdpt9G>dft)M>y+rZ8;pAFnJscOPiA$ zrv$%cA0@Va$oMOF<++WTElIESTm{^);-3D_`>kNU`f)(P9FA1TLx zCx8;Flbnl_%uX_114^z3ziDwaj0+qY)(!nxIX9gruR~%XF7R6n@Pm&eB{ySnm5%}j zi2mBPIt{Ai;yqX2rItL?!L6cdZ~mIj6n`I2R!w+2xFL4EJ@~$R+Bd;)U8A5YdN81IaBHUA zYoBFFu4B~)xyn{*y80HMq;N2%l{xp+U18HC1~n%7Yva(kt?8#uZuVM>QGoALb8I<+ zAl=}2reVWa66j7TkRItmSA`TJuhR9RYi z6xnFsSVmVY>W}?-Sh?5@S_oDOjcTaq`1j@GM{6k5P+Fy8q@!j!BS8Z*-66)(+!)uA z**eQy(aS8&k|=D6TAtXgP%EK~RiLnloAX_R1DXeMMwVKb5~*9~XE;)``Q ziDZySvA=tXN0CM=OelYOhPL)e$L)sE5g~+&PVFX0pM1Fl!3Y%`kLIw;&?5y0CY#_T1X8ZeL5-R(QeV4)K>YN%@Th^VBOwHr#aTn zjvIG;Lwkr=cyiM(#g26@XzqDnEBC^Jj3Ca^jb@Uwrz2-z#ho3=?s2V9tquTsgNPGc z16*yT64jG*5crwT!$Ra2Lt?rCH{722c_xJ_!JLYkS7XqVc3~WHqnKpEDpuqE5{U*G z#akh(;7cXe{gdCvZ58uPZoig<-Jdw9{2e#b#pZJNkEfeH*OKI@nSvff`f$6patv$b zvWjI48MCxz2+SSp86?Ioo;WB7**>2<%zdn5Npdg2(4tIt%J4`TZ2U`MaaS+TnJvqq`6<^g}9&E!nDTDcE6E(P$@tX1 zR!Hmv9vE7n#JSbua%_Vi)vu)nd)(_#q$Bg}#biAO8`sJ5CxlF|qWBPl-c1Bt?|L2F z3`nuAkm%Px&bvq|A2*Ao+ttjo_|eN>(pHU9GGQWlp;}DDG!ts>ym14C`KvrhY17Zyj2FA_noyQjV?^XS22u z>XJ3rzay-Ebl`7JIbvz4FpYwvcZ+5Tb97S1nMm#qF?+$rNk0Wk%S@-GiWm_+j97e@GGI>EY)rI#GUpa-f%&W)RxkOgh7QPWHO+>PtMI zhPK(Y@ld;6AL>$c4+4V&vBn8!epA@R3q*nUi`go2TUVsrp*D9Hrj8#V?wDYDPM87t zS_wub34@9#eqp?4HzbE3RSmB7gN<8l|78+1VSzFcE*4CSI?`V!Q{m8I(uLAVv$F@0 zWNRwl>mAE(r25XWTbB8fvZ3wOnHx@hLCZ)CMO@M%R?8EL`bzUzZ@|dXFFzgnAZ~Ut z7G~sFuf_vHsWM3xGP$YCXyT)>e0V`@SwyAwc=+uTLCQm1L@Q`~hYE@^qczg(e_w8) zmFPS>rxNRtSOjFWF2~8(7QqseW90o;$BTccqF4;n1I!UJD!!2El?NKCh#w&~<4fpU zuh$k$Geu`cUwBrRgF$uuIfTBJVY`3gtQ?xMJ6Cg*z5F}|@Ik?~9U8Ujvj6ijfS^@@ zN~b7xM$D-2$eDDrz_$&<>E0I=d*M?%;Z;!7P<34O7PE$kS*i^G%mVUFgGp+^S8(Af z3&IT`b zY*7ON;)&oH650v~@b%CLMC&g<#pi!7f0m?UF{ujl6-hk$Ujlc40S06Rd$ zztu3sW!Nt`luIornG+d9f;v@tX^r>P-RlO@S;Pt~4OqJZ#gVz{?9!^gf^S{}@r$Vu zyi+-5%@s#xKCt2d-bOW+^?P_+wy%7y{dL%lZF{lhtg#5aEfDp7lao0z+oK5Ixn8%U zmP8U@0UzV;)*rg1l?n~(052}Lys z$ZT&tYS;H0g=QNmm)3&|SL{{X`2bOQyI;!E@I63o9Ydk8g<^j3iU7jL{@G$r`^69K zx0A+z8e4%-qJli`J|_GkXEFrGI`$alK8>&o3=#6BqYTv0HA7QJ_BMRD{f-hfRz)cT z0zo&f?TrorN!84@&ZNQ)e!C3ASAm5~M;c*oUkF~3os#frB>qpuoaMjU^gjQW6#DV! zp8Fki(F>m6Yo5(;O^$p7&ZoccEd&B#yRI|48B*0UHX25L@LU=17=eZ~Lp2$)4y4_y z6yY$b4}FrQSu_{2an^w)bcF|q;*C$i1-}r7UW@tO_OStlrkc=PW~Abqp9<RYu zFC_H78qSyKWA@O~Y3nSLxN`37+yYFXXZkAsE~$rx9qAO;RnE1?*ing~3Fe1R?n49xh|pHI?PdEF(n*7I zUZ_~?@e_?=>?yL-wdxXY)u=pD8FYe%)@j4b(xkeToDA4~w3v;XRs>l;bL!WdL;Y>a z739p+&h{4Apftg07rD~hT0VX*&`c4Bnis9>b}<9U)NCc+@_mpdS11bV2p!2vLus1oy&ZHsl5qz zG&A(-zW!8x8oXCeO;3O&at;QU9T&T!-{2S!gdEKFaj3418EAswX|Jk-q%fOUrr^AJ z@2g_EVb)z?Wm@A!@b{wkcE-Vodmj^^f>JM+rg|vm6Ct;(Jo^itywUAi_ zcz~|QB+(4!PaPeWUGBc;BB+9lFkU_wFF{wVR_TXN6MLfLSmv)4V_##_% zRHgeme#$muOgV)pi@$4KHw(b??CwKOjDaM6lC#6Vl9WRgMe5go`e%Ah3tEEyw0M`8 zc+w4}0c~Bw912a#(j+N1rVm@fDK*(@E|uP3!uF@WxlUCOdHX7$C?>>7Jy}BZ_(K}eXniQXzY_QESLHKAX>!7N zZ;qB@_N9{f2{1V$;foau22?S|M#IF;L2JV=am0=Ns8iMtacE!p#TYRdid1ll#Wh(9Lvm2~cv=sa4~UfGGw-z%I*7Wj;8tz+n({_Tlh^o67KLjGGQne+x&67s1VFwI5r!0VTI%rQaKWRIh zg9p4$=^lA`Ds_0%N!CwSzS6WO81#!uug67%aAMDjTvOVsZE|r7YZ+5+Q9tw^we8`^ zlc96b($c;Ym5N*%U;+i}7?;1zGhq1yYq)QJB*43i**(>~1T2Fn`cXVaF^Qp)<{Eo$ z-j7DYgyR`{OkwuwoRhJiFKM@&hIZE%Tw;>nSuBbv6OE8%R7-bU>h(0f@Z$K%twnE} zwaT0jPBUD3U>X!L2$uui7T;)H@MDWc?sBJke^SWrWurSZRK`84)w3zqkdkfrsuRb2 ze=j1Sgy$F4V|MG7DMO^EAp1A!`g@d^n*OhISVdaZ7V|S8`RwFaN1ON~nh@2C(texysRdkEi9elPX;I|aL=78XJqHvUCyBIHK0I|*ar zhbA}s(Jev*fis;TpHzLc95-dtG5D{c`uVY2=so zQ)XJU;FMw)mZC-t4>SpqXjI?4O4Uq{s*!^F1aO_Ol7p>jPU0vCq&MC_>%4vzNC#GE zzxrX*3p967Gl?B45II4gYWD~!RUZ^P@2#z{j`JkdX;GGF0Ywqo@*3*VexzXt+R@^H zROJnrZfYwcW=?Mh+jh%jN+W@a=Z7(OQ1FSpUh~T$2JVx>F=MEI;iU%@qD})8bBkX{ z>kXy0xtWwIx~q}F1gk5*sUDZbwuJ*Zjl?d6_Yhc78u|5aWVS6klw^kUs1GP37fY+; zpOErefh+AcIohUhL}ByV5bYHh7U3swWmD6bN#gjcIH=Xq#ry3SZXZ_HUM6*qRx^m4 zd>2t;X??+DPxNZ)C_(*mw5$v!I1MqI?ST4*<=djKQ>%EQYif?c`>hW9OH!P>0ThsB zmy}@NtCJT$W_jh)dyezmY#luLoA&9a3#oSzE#%KP;i6K_>?6+P=H(%lpfn$dl~(G% zk%c1tl36KfP*1aMQ*G7KzY+m}f)aZ4^KNHtnj=i z8U%|=DYE&MzCk;ioc4%CJZ@foFE{#gwKFw;412%|ls;OGn+A`w!n+gILGXw%G=j8k zX}Wc!YDZEujLoXQjJDwF+`IkCgY+ITJvXZGJL$j76fdVY8cq#{WYk&#ww6^oWc4yH z#1~Xea$=M9lF7j9_pqHNQ|KSkJRDxkXHzV{yS;psjvgxwV?+znK{d@gnh&r*nuAPy354RV-8T{@st0 z9{Go?RrSb6>&!zP!|WHq4;NR=Qh$eW>2#FwiTT`*TV%zb1YG$yqfZxbxw8B&!xvj9 z#WjT(f{ZI6ZCBH1n~Xv*mGma(1fWOA)XUJV@-^WBhV4P?2F6dvik*bN{!S)d^w23` zLv0(`qso=uD<{JuvAXg*@AZ~nT|G12as_1qrqC|%eBsV=z;d|9l>J&?Q^Ch9!ky32 zmP~rF5DLl}w;fg^%p>~Y2Sjk4c~Gvb-L`~{;IpYkNs(ZK#lN*l8O8mBQ9A4btrN|C z7zQsJNxO7+PG{_uU?{4;Dl4@bL_Ee{;a8}IzO0FNqR|hy+~i7`K$NHld-w?}14*$9 z3v8NvM0nVb*U7PznPv0bzX?+qDDO;{_UtnWj9$)mg1NyIE42HQK}m5#ME)4_D7=#R zshB3hcyN2lN1aW%lFTKFi<)6{MYA}*!e(IiUn zNOi_l{5GVi$HjNxch69Y?@@)HJ4YYrRdJB(fvX+fvAc<{#Z~(1`s5PW_%#&;ntHUM z-=rXcu;aA2zuDhxm z)|BxZedUE3hgv%r%3%p})*^@I)VQ{I53?-mHK>>U&t+l%_$g*92s(33iJPXdV0cm* zbs&haz}#+i=E+L*sfmuhYO#3!HnxNsi8N~tL|cf2OMXfP-q}FnUJPOGLFHZ$$rY)9 zpdY7=Ukg;NlE4Y@&IkLmbR%3}>+^9;|37O8v0#1b+)=i}1(yk~%eMlF)*s2I#!%(! z?NU%^>+mNtbxd{?$Tlh@7Tl?T=#OFlvWS*=Na8koQox=y|;LVUT#f)!>Z5&~hSc6@A zuC<(QF!zL7GSqN&hxDm>kWy=e5YrJ?(og*nn8LMg*XafRC=7dsyjyEq4#hYVcE7=Y z_X>Vzw^x=-SeA2(reGS>hZ7Q!r@H=RnZn zrTBe&-~Q7CB5zLU_gMaNKh9^sx@O2#DtpX`Vop+?aGs3GKy)>bY_jpy!@7?61>)Wm z2G`qF>WUDZVLSaY(ALjNQ2$iQ&}z~P)h|OeA`wHZ&Cct77D*S!>psN2=}nl;o0Mr{ z=h|-XbaKfjGVCyCT~2aQW0vo%Pc6pPs) z*4UaxWuRYWTZ-oO6^y`blnXdhaAX_;n5xr4EzXQE5hCotJPf%5ha;7^=t@J)728-6 zH>8JEC>!$5mD~b?BzrSvjK?z=LcOV-ghecG9-GpdOOCi$t8oHjyZ6(XyPae>=_e=X zI+9Q!M~F$S$B9?vK#+Mvy_~2++;s&-EwXN6&;?UdpNv@VaxKblJd9aaub6i1s%(ufF@x`zozl;_#QMK6X`j`1D3KJQ*g z9}c4ANp6H|qu?JvHCH#$kza#U?}w}UJeM~64BQvzlcuPrqc8V% z@cWQz%rD!3Ff!I}Msk%&b?Ueuj;t^-l*B%+F~rWggcc7r623`Tj8QZ-XAqN*DqlU( zo(V4;IQbrN<8!1E`EJ@OQ6rX-YX_S~y0OyMk76s@JCC*xx@h@vrk}&b-G?@RyE(r| z)7uY{4AN$Ru&*uTc)99aa9C1LVzUUeyO`Yuj3rz_J+kdKDmZzr`pF!+7wW{6zGxB$ zJf>vgh$<1%JAiX*N46B0*m4)ka=+;np?g&aJ25VC5q+Y+ui~Bm7a}bJdeqKMqDN!x zUfoa>Egnh6c3+fTi7c7>xPHcW9c_X%n{&$&YH0kx^?++rZtii~Ck};VzX?-8;^6ek zSnbD#{XErl-u1)FY7x5ARW34*v19GNOkCF3=r}2N&CVBDPoZBS=rjq}#V(E*cZL5v zcgZ>#Qmmz>kw!OPTnpC7Udrc%+Zmu7OqLa{9FE5%-Gyf}Q>c`RFR$+iE5lmgJCI(O zw{8ERtTNh3Z`k09Y$Ujky3g`2i+p5J83dZY%EFu+?b^eyqLXbqB~*I1OkTMkdKwyN zr>};6v)2hN8zth9y43oL^_~|HYNiG;Mkvktxlcbh8x~K9qB=9QaW-^)NqvGcz*p$( zz0hB{_u)icOB9sz;ndi16y;bQE2BqU`AbE!b^`gX+Qe zIl-@}^utJQa`ut=O5{Yg7?V*Z^y{}~m3bZoWj_(Xe5$?i=@?i`&ck`XT0R?TcIvi> zwF;!h1$nC70r{xtA?Emks&(}?T9RfvwIq@?QvnU5;CSN{qkb_KzNBL+FG z7;SqdQtfnwiJ#e?Oz?7F_0n}4O1Gz?U7nlRmJqrDj4@6gVNEk1GN6S$R^B>2<%gA@ zFY1p}oJ8gSFmSE+Mhk<@vPx#7Yl_T>%Q$L-9zjJU-)9<#z?rfYHDLL1UrT&|*~I*y zKfifIa-lbKF3i%&nN>QuS-ThV#jjr=s}hS~?5&Z~UZmp$RFo1ww;L2{m>kcGopZme zBrmz-{}J=w*Kz^VAE@t;HkFR0=Ivi@=%P@V!M;r0cq(;cW3$>y?KbF5h^6@1WHq5a zU#s72Oh9gqS;X`-3;zH7H^Y{cZ{%SVq6z3-*f=-0!|WJQ#+=ek4W;jNp;4c%m4M%Z zVXpRI38hj=mHIZ&GtC~O5UI|=XB?+W@S^J3#P`|7g4nL5s5J`O47F17&MW+&XJW7; zDtruTAuzs{Qs*Or1%meje;+Ax^`(MaWFOIep^6EK5;4MevOTGMk!pjIm z82hnoj=b?u46HaZ2zv)E-L$fbW5mv&K;9v6?z_AEm7%*0BD0!fM6Zt9fxeFpMGRfh z>-3FG6tTW+i;!lX-k6h@DhNu#lN2;Z2$sfSoU!($m)^b25}q8!~Crtb8G zK6=g)`jUO6Y*&2SC8|&X?*&4bLrz!sH&1tIc&~KxvuXA#og1&FaGNDgls&ztFbZMJ z3)7L>E`GG(zP?askY;8~lNkx|D4oJpvwPL2gO$=phDh`y!vC2)*;sFEr}PAbXgj-E zMIpPe&NRx1Bj8VtOXiVJ9YA;wuU8!KFPNX+Vu%HTRalkAh)YZ^Elw{sf%5<5 zEpy&y3(=aLks#wfC`OTuAHcf`>fH@Z6!kOESAwX>Xdh71Qv?z@t&S|6dm4uIlVq1| z7{F^;Y)E#qQ+W!^DSqL5(#ZA`kgMj?2tS>P9o*qGs$wIIr3?M3$s=1=QDGk#swDW~ z1{uf-`1pY((@P)+Uc1pYy7>Lb^G<0{e0^Mgd=|hLQ6wPrXQ(I;o5xC4vO>4#otzy+ zmA`m};WK(G>wgVr9jBj3aPxpbV5cZvw>`{&gs3~bR+lC3@!-GLVWT>t>i&(f7U&S- zE{vos2~mo1L_pnWr^zDMe%qO094;&D3UJ8B^W(EHg}Z#5S9(6cbcvQ9(W<}ONfCG@ zk7`McW3cs}e;?-???s7?HU!JN$@j>VL|`T|9@Vi}E{7 zS{&eS`x@qzH*=~ALY>yEJB0ljQYD03fLw7ja$RY6IK2B$iF2Jl|1a=Dh~RD46#C4e zg8XZJXTo+YO2n5gfJAV~t+!fo$}C08MOtIafo(@_Ui&hh(8Zxln(De~DGNSa)bFw~ zt2vG*EIb0F$8BMu)tm0D2D94MS1p%CDC>ciE2Lt{UDvgLh@7FBb;h1Eg%do2ehot%RVMAR82d>iYue9cr@Hu^)lI2OEZHT8xr zRxJ;V5k_AxoR z**SiO5YJc#uFl-UrLjGL|qzREy^AzDd{bkrzxefr^+@8{4?4o;ZDyco$cYnd4w8J>?(hdWc+Nar~e^Zz*k zJRr)&WMTcnQuif<;vz$Jh>dzROWDh`A4`P@BN31WTq>0?r4rqZF{Ps&bDWHLG&|2? zbp<%)*09x|Hk)VEHcUu-es z2>BS2yJQQ%1d!$uhQ_O`Yi0O?5-UGF&WL2*gFkr6awofqV5Mno&#L z`DJ|^D=a4+NT}4JTK|IU`g)qwopm!F(4=CMFJg|>3)#A=2?=FL&-9C4A!BE8>J;$Xd`B;S=BZ&~Vw#q-bqziVs-uzN&pTg*kA^icGR!sDo@3FWo6v z9%k^gn?zW*nyhiT=OtPW#|=umw3WZS9>Kv9Ra8y^ysRJz@Z1(@HCpJwt7sdEm}`Fu zy;|}SEsu_&&7O!sdNj98-srjGn3cdAn6op@yj70_6-4#%x)f1)~;gdypWnZizc=D4Q_qxQ(+^8+~RQhc|?*AL7t=?+ z;dX_OcODwi$_LeDepfomK|pE_Sem$AY|5CERG>Pnw9OV^*pu!#Rq*uQ0qo;2UPf`d zOdN5T*Kc%`u}L@bIBZsOn4#PpG7lx9bFp`4lM0*;%QKQ6iu>gRl-yp8A3@^z)<}4@s1?R6E4xj~HVfY4@HubzDdyX;47}UwGQPIsEu8!xzI2q{lM)Rj7-qXVggKf9*q(M;lk$l#Lw? zf8fTrj0fQK9^=#dS@G8AN-ok5^N{_!B@}H!wLcIh_B?zPo%bOIIH_^CEr}_M6|ir* zn0=^rvM{M!*?yf~L*&ocG?^H@R3f7h8jWafx5M3rIX^bF?A#&y-#fj2_Yo6eKnhVf z)i-*lE5)|9n+`LVrawk~O7>q+psMlW2chh?n|fP0{&@fNHxyQElNW4WV#zSIw%1Unj+ZoulEWZ&?eq+I9>Eo3c$EG=m5pH|a2=SMN)a_es8Hr*)CkgBx4ev1G9|={yd*@Z>HJ1^@@U zDL7H}>D<>qQUhjH!x%o45k29>h#L`Ekj-31{fLvQemf6d;#A#$eiG;`U^W~^U513e zxR>8tL1E|g6?SGIhp(#vk}p_i6%m36h-4(Enj!&B3mwJs&VK225~VdE#f|zMHD`j{ z99s8`UEd>SH1i!I-OU;dz}6_ zJ+Vr=C~php*Wwie(;m@E8~f?=~`i2Wh{$qsGCSVbQSEd5 zt?h7+vR4dqETv1t8F^U#*0QaM4={s1OtvrHgpt+M>fR?#!q}F9$-GS3_|N|A>>I~n zf{~z0R!uv~bfA7$+LAB244spF{)}&ErFjel%aA$s5wF@R{BU&VBZd4U&vGbfnTssO zf)skTuagNUb`&fzhEjT}T<3s=j+cOTlIxf+=T3D}UxJfP!#lRXVr zQpPI`c&>t$|JjvV^eG{j_@%R6g&?F0lR=wiIVhW}c?^0ya@eTgmORVd&qygQwUeyFq-oJG=`|B(GmlY*LlPaC5VK35Hi3}>!KQE^QT08m{OeRe1whfJ% zb2fvOB=jSBCKB=8wh_Ou;yYReQXU-V#+gvs-l7 zJ8cHZ5+Cy)N$lAv3z1t{@Cl^v7xf;UkH(10JF|Mc&Evp_TzQ}LFH)fGTZlW^USFo@ zI^$R0qJ(o2J7o6k27I)KF4!|7L$YO4YTfUQ$kFX^&mzE*9C@VCZF-Fd1(Nfb;CXfU zwZa;!$jE{ge2n6v+(z3Ff@PUKJ7lFk(9LWQ_$}mt_iXh zXY*Hojx7FM!#i_iaXz~iM`|%QGO-mfTgnvYN+ zPvsu+aqVZ-Ls2eIR*x=&m8(i9LHT6ksHT^HdUWVc*c)yvL<|FMiwDTNO4l5M5qvvU z!ej^*n|{TV=myujE#}WADZF1|t-e7kVQWaeR;4&43%Y@=kh=HsY7*t?c!)u<_hVR! z5x9h8E)M;P)a;P2_5S32mYaPRK>GeVJ#3aw!392-l0bw*T<3#l>{xa))~+U#%?+oC zBq3`oKfH=yrQ<%Jm0Lxtee?@_#U|G!i%u3`KBl7-0|LIATuAhi&dx%|64DMhvvzcS zv_K#lA#Bnl>&~R*u>IWm&0U%T(dW9KM~XyC0gTex3qU6tKOTz?W&01z1<7se-wp+Tpa)_CO@|q14R~#g7N%Z0DXNO^lr%m-B|p zu21^9Oa^{77Punz;Ck|bZ8uRUG$X3CCY;`xjB;E`kP~WNxSUPhtKEq&0I(Oab5L@!_Eh2*8nwwd_60jtLry( zdeTFRcfUp1Xk&zeTr2g-kgB}cSk0? zwDHlpMTnv|43FvS-0}L`yK2JP(@khF3qP2b%g8(t+OF{;=pZr!gPhs%b z{b(0<_X^uzUsnqB{voYzzIX}@Jw1i(wrl3btjoESiO5bl7Me456n~@CdEDffbIi^A znJ5p<_A%w*D5dxQiOf31=a5S#lV|C2j2u}_?O}ZN5l+EoapeQmAZL|6P9em*F!lG_6Q9=GY4!{;VMN4vhY*F! z`DeI-$~uCKh(*s)lm>SRA|@02j0mMke21koID+vY*AmGr9DIQi9kA|NgwpXS9H7tR zAL>l~NHH$@F8S~>=O}e(iOWCMv-vgS2b))O)|Jy5*rL2qyl6j~o{s;U*#LhC{#r8P zScKJW35uc5>Qc$FYc90Xu; z&GmfgSeN3nIEFD3*@kuIjwGL^7e6(p#|K|9G!-VE*FN{$U?$f$6{Ff%l_Y`F14r-e z#O2Q&?oH?_!@WU<9x4=E%&+m0tfqC0c2}j;sOd7UBH7%(> zOLxetrqZ~P!=G*dMk9X{9{u>jJJThAxXwiZ1|A@^j8)r|vegRUd5~ zeiw!<6a;EV(WA+1*8Ul3sVbU0*%LKHyUTZ<)^+dAzSy3RE{-K0tMd?AxRb&3%=F|% zI~#~s#^OCZLalM#Pen?~m+_}FgyNuM?-nR3?&?N=QM1#d3fd{P%9ztHVN_RSFggLq znVcx%fg#q&mqLE%mayQ+-|zPb+(ua)l_wBO_%?K|5jj0y-@`^MJcbvQJQ}11Bl0(P z&v{N?ybY8!E_0O>HlqV3oQ*gC<-{w}qr!yITGkI5mCEvMQAXJNN~W#-cG=^+5dYz3 z(Mj<2w0KIp$tJ)O)y=Ha#o~w>evlZ8wxs_A4vL<^(ayZsKKJq>2s?uco^zu#b)423 zQ)@33jgFt+Wbzg;VP1TrztPL*S|x2m$x>uwY^jdjE+l2fytYC1rbvsk1hiSDWr;-R zR_ebHo!VsmEwC@4qIusb?d1MO48LX~EjfTI=|LzNhUKN`dQzy*p# zeCZ8aqi&N(xQE+NZhJ0LcYcSDU&=N(g|f?n&2xOg3t30XvvjgEBFnGwM} zAHR~NMxnHVg?RdfhXZ1FkCMGBkCeQpgxQF=>>J0MpX?|hZq%;hIgV?>9RGu7N+2lK zK+j6fH9GXv2?3$tr*`P&0V?w=lvgA3W7Oe}m{`RZe!T}r)CRyVElA@C4cu)`<~y8A zaEMhF^dMs2OBsG{hy}f?u^IVr)oueQLUn3FkeK51Rjb)kNZY(M{EFCs6U#<~`mBUd z5FzoJB-0ITT$w5y^Y36!^HugQy=fh%Ml%mb?x?AGstpXTZ(8CU6HP?}m57m2DrpM> zD&Epp!8rEP@5sg^xWImWRzb{D->m@8j+ql;y|i)&iC%WH8JRJGUUA}@Y{WupiGM|QA${;SN&zud^&d6 zJ0x?xHg>epj^t_#CgqP#Swk+h&kv~(Q{rpF#2J`Bn7#@)(JJ-bE``csz63MF`hcn+ zR)sbS6)R`pB=j3e3Q&IRU-IkCex~L_E;~`QASggrQI<=zQkDwH2R+m$#d*b9XCA{J9*`E?+NqMk@K$B&FjXk4&9^dElA&}`ma?OJYU|A=2fx6Q zR@VZ;7yvmJV`Y-bMNB*E>OEdI!m>C^*v`1kuinnkAHV`q$2mN7@lH@EBcw9#oe8R> zyXMy(uKTu6w~fLFaMI}kDzVgQuxY8(ivhvTtKE~XQ8?&@FW`MP85*hfJ0hTCBri&8 zV=vA$%ouJeHgqpFjVc;%Mb@axk(MwB#h&)q<&pA#=Ns|Tu?~lqRhC13d zgHa-KWDLy12lmU7SV^~E$^CNl<19a*}l06GgQ0Fq_>`Jg(kp)Dqxm+3olX;lDiHfXI6vGxl6e_=3*{jDH+q z)G)*NsB>(%Nf`?I#lai~D5=4>HJmUKDbR$6kAz+^#IcCMj*dBYIQ}TTamY!180nqJ z($t6S!T9uc)OjXuqkC+o>2twq+i(h24PsDjgv_KPQ zZNn7vxy(Md{8ZhJQU2_f)904EKDL67P5ZnHcm?P_{<$uM@LUQVMWP$60TC$9L`7r{ zRjwz`1bU|qxnK)p_4%lok0E0Cm{>v(8?Uds5W{>+g9-STGRH-5_-W>jaeuf)JJ_xkv+9Gp?g-KUP4C|W8zvzLrK}XpO5#-XH|_f zMp8C$O|Scu%5=FbW5VRuxsiQ&qb<)&+5i-{m(Rc(CgcV!SB#XC%2Jf+l_lhQ z`|tC|(OGOs3IkB|m#{*XxVy81;O_SIL(gu`>6-2$B=6l@1tgnXj~vuYrKi)c^`i>H z-w?5s3emoOHEIuuB{HL-aB)ZLj+v0s@Z>VG%Q7yVt`U(X>a-UR*9_cHBhL+@$Phi% z%^QvbP29A)(3kUlQ*Rch&24kZilO@jod|_ii5oyrY6fsUDd}?*dv6ghHG|(y<@22; zJabSrA)}^bP(@S!rUPB|3upyjdm(FPq|`OQHF0h;XMaRtB;hHN@C>0<&7V}T}bnA!x{T0bG*53{hjr9=@f%gguOp!oWNPt13AsX=)MHnMjjRa2;GV;vKKmM;5?Wd1?iHrHo zQLPbAUZMA+Gj0;M=rG|Xb?S2{UaE#GABSnrju|GIeogC9w8%#^F=}{eUpVrzJhx)Y zH;V~`ylQAz_~qa-VlpF*4${!6^i?0@s;ap{2KA9 z;T*>Sy1Lp-E)sp($#Q3$<4$2DE38xR)Q0`2(H{Hq>^O9`K&pf|)qQj^W05L2`lwmd z5teNMvI;T2NM7q*tC5TGFazLnvPw6aJCi|M5+8h0BvFJb$0>f*dKuC-ZNXoQ2*iaT zzM=4K|9;wGG}+jSNZv<3BxOT%mTn&;%pFqgF)?O&l-0JSklhGK$3^20_2;Bt2wdZs zGYzzfI%&aNY(2iSr&bDmvQn_4!}*Dj+06S&v~D=%KSHZ2KH`p_CYGPZM=&z7>D(IH zL6^>Z-W@y70e{iCv-o>dPldR(pOyWzgs>ChJhf)Z-OeBTU4>Q=OI{j z{SJ5YMbXAq(-Y zm9{ga23bt$ym3XtUki;ZDqNwKw4V8w*tKS}Oxy0-FnO4*Ao5gDT*7ROmiiFa*#4r^ zK2J5mm`K8TGfLwV2M`5mE6Q~O#Lpi3(AV&W?(IUR5HTMC`FlK*N<@yszvF>;(Ok3O zSMMBq;=2C}BH=ZQ_Piy!!1AY!y=v!;zdgC-j1QMMwE-$!d8ED^MbnV3^@>PpaAb+> zt!-nd*FVX2)Z0ZPg?8c3V`A!na8`(3ywPx^_ZqiDI!u<5!_OK@Te33Q%HIuzVU!J} z!A7Qr&7GcbIi;)G?PVRrT{KbTaVeGJ^sZd}!!)e5R#BI(_R*?4OABpW7bJ>L=vP8I zZxk*zpJw!1*bR;@x-FUqAZ?_L!8_A3+&^k}D%afxV@-DaZ;wr)|4}Av=Qiz3^l{W0 zeuV2~2I@1HkNl&neZOW5@shG1f^AA%e4s)YKvoo$bD5hvLfqTtdfwmG@!d}&&U9DD zWAX+Omcsm%Y4X5dyhL}tU}+Fi-PO)Gmn4rbHR)OjXI|g}WuRMVQHX#}iWQ&K5CjrI zL|=$9q;3yYeM{frZO;ma=Ht`e_(e54bESe^Fm8KLcPEtmx2Hf;vj;11?s(SOUb3Lk$rJ%L2iY~F%V zM5Kk5!7q#N)$f}>HjuJ6)KzePiP;xIA$EJHb8$29m;U#7LC)#mQG*iQ5RmN}z$@d< zR(YGw)f=m@Vv5#1&a$_28h%=M7e7)4O9jSkBxbe;8AvJM+6HzauFcC9Rd-w z^tdRYq#ida>~s)Vq_s`Kqi}&wdD4uqNItD1XtuH6m&==B!x`Tvpf6KKxb+IuNq8$w z?wlNft+WpzV58OV9(yM>y9^b=4GXbfTCs_S*+w-z)%Zjr{h@~ASAqr*8^h%EuwjYW z%Os=g&gF71;;t`5yc{asA~_h7f>|f_4N)rLF>bFz!(UesDSV$Y_;p{aeexUe;!zt^& z{jCi@zVJ58Ml5d*2K`o<0q;?T=I$0)->?&QU5*U{*(vrmbmNJWI_{_hSC%U9sNmEs zE-TFwUud_8r&#dBck-3Y9ckw#Z#hiff<1miC-O-UXNYOrDvi&g5osccbzO@jVP_rP z2L#F%|5Whm%IFJ^zS3IDBFw6^IGcqb*Jb3S{aW%W@!i=mRa`*({5+{P{2NJe!mlZ0 zjg-k_EbM+OC{4($3qkJ_w>kinIv0=QB}@+`Zi*94e*CkniQPZk`-GDHXByCky$FjN zsL8%*jb_D(031f8o@*Zm!EKFn{x({2&~KPxcDf_t9v;eLx7sHvn+IYDxtaJns(YnL z@bx*<`I)8etJ(ZA&*AJg=p*s;HO=7dgGbdD7R~la$Wt{cVj7gHlON*mgQt$FM}*TF zw|`C7XXDqB^&{CZxGRYkZDpE8(73k-qUH4W;6|!ow(4Ll_bVKU#J}PNrs#u_O(=_g zV77-H>EMy*$r5U?06Rd$zxpnvCoLN2@QBROPfg!j*#c@(J6D099A{4lQC8C-VXD7R zUh*E>Y=p~?d19B&i_`OQ8fa-_gjacxrV;$`jm8uSQ^C3uv)QX9qA$>l4q=E$_P?vx z!=La*Af=-YJYn|Y1qc4FSNI)c=JIb|Y1Ar;et(-1^+^C{JK^y;mc?gR@zB~9_+wu3 z+q*-)xFGm(_FWeu$Xx|M>76x4|MYMH>_d50L>gO;f7vlQ3@$b%Tr2t+5;GT&u zuYF_9cY$Lp0LASYzu1raY<+l|VKwkt>cE93PK0lePyapcX9s^Dk;3pmT@)qlg3a{( zSdmYKfoT5Qj84Cw-AU_T&VL@?wnq808Z&`RQ-cGOD(C=tp=^DbaAo~ZO**E!OacnD zS&chAwcbKZwMClpQG`@Ck%$aASND5KrFvbUl#YhE>#+yBt0wU$2yN0rji9rc#>_Mn z>BO>=YY`&Fa>C09*!=b=Dp3?Rr17va#xu1rX3Cvo&=$H<8#kJXxq=qf`re^9$zNNm z^PFv6VEe5&H^13I{DVwgQ9>&mCnuBP*vqQc{%$X`HxP!M&7f@fYT!akV}G!@I;y}q z4SI-#k28I4TQhu*xkh?EPxID~YI3`J%eg+2k>Fps9P@9`{iTDlj(EISQSgl`_bv|K z>j))xEn90)VC1N&v6|o{jx3@SSA2`=qU!>D;6*`A-9WOeMNE-JRd*$qNmP2KgIeGh zVR(k(z3%Vhm2aBc`U{#sdh6&sMf3h#=~_znLsg7H)FI+DhZQc9A-DN)-Lh&(yE~F2 zjTC0Y+0DBjna*y1oC4q*s>DqEWkT~bU)4N^?2&s#-61TmNgFP(x%X;?;F-M0hN z4xff57q`2?8JTf3vzr?lTQvN z(~hYuHlF&eRZ1Mm`PB|=J(*<>n9IK03iMdVsn`WIs+HhkAPPiE+d)xex4)Apk1fB{ zs^z1zq~R>vA)(M(jP;MDAq{Uy&}2CFiy0g{!0GT|YyDiM%dC)~8V9bk!LUxah0Fo! z75@5(2r9JB8+`NsbyfpvB&plY9u}iEe#rdc6^NH$y?f$FgHRy|E%R&sUptB3_3%lPJum-N_LD zAKTBe&`*Y~yY!hH=wVyVT0POt&?kY+TuWRu9#=S!ygMFQG6Pb-@+5vKyb=;BB>9gv zbS_slc^co+%3h2la3{G2WeOtvmZxA=yNx^S;vE6e0NOV`6N{Ri8p##OTxLWcg)G&K z>y(flgGFKZcWAn5QmB#JF#OJHsZrsQ%2ZfL{K>z;s(wbp2|m?&UGb1$g1l53v*GK zu~~>4a?X=ah3DdkG;g}wp}ws+%KlR)y~)8a?9~BWl>-a^K2>B8*wx}oeq@-1($^u~ zw}^P}A>}4-uVL2p$KyXHK;h{^_5z`4wd@!aTMC6&EIw`7?Gh=#-vULZ9{Iq4*lJ~G zH4*#>;(Q~zWk=q3#o96KdM6IZnLi(s?a-vwH-|b=Cg3*kwjXe#x;KqY|nH-TkU0X={ z{!&}Wu`dz665jh-Z%)Mq*ZogEYd=V5(a#^(TM@*0?r~puUMMN0jb@NEcl>XWt zJvoCR!@T+ik$P}WbHltN>12$ANN}hn7Mw3z8{ysM=Eu4IBz{WlLqCgMhf_pzG_?_G zl&5rNgB{q~zCiY%AhX_IIUHfpGgg=LuDm%&(W!k<#-}B@9?F|g9c>~EJ=plkxV)=q zr+-WlAwzm-5(~aAUD^9Pf=mX)n*7x?f7&Dk{5 zZ0fd_uU8h!kGqfRCsp=axA+XCJJZ>?9MnmUkst_J$lp3yZPrI!buQ~HSD+rs7R1#v zkL+0(P|T6UE;v{sXnjUPg7)tb=g!j>R6^b`n>PEXM)Go%ZTi(q19(d*6=pQH#+Tnm zk4%AmLj{-ZOoig48Y=118N8;9!m~+@)p&@ri-gK&BJAJba}1Q+aESES?hWlrOH(nn(TG(R?q^vZM!bM&Qi>&r*-IgAtMW=Ueh za;6*C+NbDteVP_OhlF+>R{e1v3bT-(*?!x4oonN6#Wob!7ze-ld;OxogrBL3y&g?` zY|D4s?^cy9F2K0HS^U`p{V2!95CtZ3)W``hnBTCL?;0k2S?tOTrn-S#!^dq0Np6A`6F5?6{^v0I2!>a>$9lCR750E?E&) z3xz`>?Ld;K@BA{AXK1fm5vjBOeT^8WI^|(%xTF(5ugW&rOk&*|gR`C6uCVL}l#BCv z(kb6;9~P!xQKh};7$d9ql73c-K?C;wI{(T$vsC%M;%!2DB{FTqe6I9Q#@J2kmrtZx zIDqM>JwfNC!e}%&UbEFMauCPoS?Wrh#{>DIBR}15HCh8G32d$ zKSNd6wl&J9$*ZT?Uvlq^N1R$jK4?qUU) ze@ntBzT#ZVl3N=_h*Bxi+fsbuVx)CPnfD2kSlvmPy2C@mqE^z|m}gkUwVuBoG3;RD zs4sXSh6~PX!}7rOI5sPI>VEJK=Zo4_;`SF1)iFmE#np&-@h|DAB)VNhS5tu|Jwr$@ z!+DSs1_WUzv?r3m$NY=x*`;yv&T)8uj8-F^Z3HfI6-Cj(SnvpwS3uD-3{|6<7{`3r zDL?2IM%31`-tDaTxRa|l(($!b?rBCn^Xjlb0cfI~VgyYHxx8l=+-UgR zd3{@i(d@>rK|dW$F>%=AqDXMg3jC7j@rSI`J0j@Gj(~jW-pL}xAuR9b(hQcGzSX9_ z(uDVbfZo#XYeY`$`da6Z*(M)4tF?mjJ#INpoRCsB5e(l1cL^g2R$h?H6{GM4-3o#S zd8cLQPVilHJ{zbvf)Wfa3;r(uV2~7+CP3}QYF)y`tZ|{(AwfQt+iRW>8ovMq`|m09 zs-!U@yIR+fL=C-!L)Di&hki78(S40Xx^NPRzCuhf=1j68x{mPR!Nrt|J~2ZJF;7)bnkRk^D@kT(D}Q<`j|$ z{x+nU%s)8#+X+m}FfXmmFH{?zc^#JPM*r8g2y#D;BjpZX;V&csJ3ed|KDu26A29qz z!%I7@8W3#_>(_u7X$xv1th{@esE+_>f$N`oGv_pLx4n93(PbeQoE{%=G{xig0LYn( zmU`(mGBusU;(b=C=trF(2!VTVJ!;UAwWKaO&&gu_Bx zc;sJL8t{@CIma>ic=`qs#?n#0sk^f`%MFJk6?GRFDGM`hUAc9GNqXy2SmxmrQ78dX z+71HXtUGJlrOLrRrv-X9I0_C3Z4X&N*(I10bMe;A*=2Ey9NU-^*}aUpAg|WCul19A@iR#h1=vO<* znRw98=?X7$JW~=R44A|;?QOI0G8xj7XnH}jxQojLzAR!3vM680@pc~(iDF)gr}7oD zOUy9rgSk%~?O$A?>Af6j!O2F@Nc;S>vHKQ_>t&05W!Cd!zW!Y9B**p$A)_J$-|_v< zeHjJ$v294E)uen8Z(>a0mOwm*0R~)yCe!EdawvC+9|}Vcxi57z->HrgTRUJjU%{da z^aCFiN>Ry)F8`=AU+fq7I-WkZ3JASWXc`QWIWI6oFZywd`>%F~W|rFeZD@~FChHYM zrGvIug<%V{gGE%g=u9MF%9kONEQoUneWAZEHl1NrWIG-S!Y|Ti%lzQDQ(WU^twX-_ zMY9(!W=;5UXU#yIYpwXr=6AxjP!&4^3?pfRIL4%i$j1VKVb(uKS+#`BB33b_Hm@vg zfhTQS=6fcbaZ8~`RYdyp`UlPIzlRas2ae%!B8m{0h$EOM4))?mnrzFxyBg}*h!PdC zoxLi4^XD#2QxLab>1S7hk-3E(9?m?J2N!Ty1hlT!y@+u#Co6E@9gtHY%N=Xi9IL&N z;A1>^i{%4pp6YQTqH5anEz;#MSGGdYAeHpo4}FgN3>gNRY(V*E#AHZZrlz6w&#=GT zkKwypz2uNVei#6Fkt^C>WloQA@xnO-oF)B6mFuLS<{SyopQs<%ut z{Dp3`Cl_gl%50L{s@v5m@rhfUd6mq8mqWXNDr5as9DrnP`G)A^>FL5*gOG-_G=nXD zNvAPbzMn3nS{RY~yDC_qLIxJ|g^gyF(Dm1UE~9x!b8Gm_hqIiaBP5gy-asmWMaX(F zX_KK)&E2&5o#SFXeomFyX}`zK(Om~YG!rX%M1>-jtjfz**B#^iSk5n3GkYO)RBgx1 z3CZ5qvBm!GX9V2GpjAUful-wltkU0-}p_)DK5{gc5rr7d%uM&tYRrz zSD4iz^~z^6^@j^M9Qv(e=-y08PCvfU%K^BU-puJ!SMY_@?q5nLH{$t$PP^7MAm>CPs^F1wjI8L%=P?AucMN+wf+xv_GO&1$OFrg4mhk%^c3DY6_vH@ZQ7>fcqphv2M33qB zB7#rZJ_BjFjflTD$Cge*%WsAc-WWkRyW?gsRZRjT`qQ{CxbbSE}v7= zl*psR4L`b&^_F6O7TYfK?!E>S{L)Zpcon#lP}u3HXrRHgrmkdE z@v#;0n?hdq%e6;uC=EZRPh3!lJ5*8kMzHtP(+;3ZZn;F~t)a82N;NMYy@*Hnjtm*` zUz9jI;VrT>eoF1<8GNe)Hq~{~`mI%XN>H%f`Ga88rv;tDm?7eM&-P7 z&~rNC#!&UW12ZV2W3J5>UcWg?$e48GizV>DRI@L&>^Ae9(wcC6HgWu2m`!lm4PJ&? z_+C+45|Q!ytE0~O6JPle@y+Nyv(c=~)eGXP@v}(Js_&s6oW(~q|dznBkwZjr}&{HW>&6iXW z8VMUS1aMRceUUea+*9&fm*%umiIon)8()l)12OZDMk%YF4%>1XEC;t)D>^R3eR^(8 zvZ)9sUEIj%9^X$6=8fOtEn$pbZ&~7tk4UUy zd{B6z++@1~J4umOd6?<>ykHb)$DX#D>!B%d1Y5|ffhfg%!;~S#x z$#zvYqj8{?w(C0xk8=%3ycqE+B)SSwf=(L4n=uV)I~1i>rmmy}30bR>2vO;DGtrx% z;fDGfPbg2W4hZ3fExJ7qB%t^`zkWH_P7m1y$-2hnN>1rwNWco+y*2$fBwUb+kTl^q zzHyC~%8b|Q9M4rae3`0D+Aq?%_JLoO<+NQe^5;B}rZ#WJU=m3nyc?6|AHkxaq;fSp z>3&1E?zN-$Bv!>%1I=+pV)I?sea?;e=7@kOVO#Q=AaRFPLk zU1!Cwvb_;EYpnCgey}hAf&T;y&B<2!Rp*;cOx+e}@3De6wx~$rB7t7uXAYd{q@BGu zj9>R;UxtzW4W94|PrkRO^f#hq%E}ge4o63s*DWx*@Ml7`t)0!$QVzBA&nBYI-rqvc zuO$`J=r3WtQg9Zs-#MpSY>fNRCB5FL*rJNGz;pL-0qK4X@OtW#(l&Dst?GKKeL8Ez zn=lJ@Xne`Drc_bURY}1Tr;~W;EMt3Zx4NF>iVL)+N3S5F8fQ{*=#b7a{o4Ty$5TM& z+J}~ewOvn|0xC3+udLwZ8KKsr_ejjEQmU6|d}Ffmh82^_Ca862PR$AKb|fd{`yI?2 z(l9buYlvjle5v%^#4p#fl*TWTxNW+d-yQs?aWxD1Uf4Gc!4cA?Kse(?O1UGbxf*ItrH)#PgF7B`vD!s zn{af~TYkVO2q1 zrS)lDABv{OCaN@IDlj5tV zTqW+B?9f8SLC1?O`?av0JBPBI6M7Yv%Y4Tz5N<6=k3fk=t#}XNI9$%kQY+DYPaBTy zsJLl3LXAS$9TCsmEHD!33%Xz==KS4i|BKo<6ht8!A-EWU@%f3+)AINbiMGgfv4ZNP z)z!i?DDhX@o#_X~ngCW%7UIKo4MSW%8@!MF{3EpxI?VCcU~a2J4@18c=@uz2(@1wk z+HLfz6HQc&P9D9L+i_?Vv84^Vbi@Rxqz}k73yR!ctLXzOfdn@hYGLc46Axstdq5Ql zcD{AjO(^GSIPCyGd&zW$~f-=nu*ejkf$vdhRSzcqPgYGcqww)MwY&vaOHqLk`ay{HT{&-U>NZ z3wkR+P`Yx;Uwtt+oR?UtR7%Fgir%sEiqgiiw{TF!Vp1S;`wHcXveJ*}r6NIf5AJY= zUu*ro(GIn;M$+NZdKG&<5ElH99h+OT+{Rt1+H_!=pM+5#DA)Wm_SNh4)w4KxBAb*; z{PHUrC=>rQ+}sbyMe>cSuE-I>EZb)AzGp(WI+xJzf=P}$b!hJE%Mr4E zh&cHB`U;V=q|LwuafK@)l=Lkb2-zUk`h^g1ki90V#Od6*hnnZ z#2h)WC@~VyZ|!0qnDm+n_^qMS0lmFTIfu>6#;nLC#`xa%U%L0KKiK&3*Wcq2Vhyjc zs)N{R&2nAKWeS^^z*cF(+4vj4&qu`aml%A&7xQ#S4lcPz!HnpB$iRZCc1iPG;x*X( zR7?2n@tjHgB}I%Ur_^m)i}9t4%PJvXx5B4tzoK=75|uLEeVj9Ft=Rq+UmM+d{~l5O z7;hmMi9Q)rA2+r2$!UK%Z7DVHkp*`_!>UVodiWxvj3CFkL*F>mnVQr-M*+_@kH!4E zT1KvJ=(03raet3z1HKAcAWvCl_YgS`fdW-8%A>xFcNU zbNY;uFkn+|6gTYIsWl3SE6u5SUl*t0#YAlH-aS^n0;$@e%V0UPVaP!9jsa{nYc&+OFUr=PCxZG;SlQTnEvK^`GSfA|#=Gvd(}?VBr!^EB zIN>nFMWM_Ugd4?Ys7ti?nl22U5u!_}?$#^Hs}I#G@53-Eko&gTu#$zf6KA7KA;(+w>=Shsj9s(g>=ImfsYGEU8c!$ zaU*IF6Rsh<`nXf6v!G*tJ0aMlS!M80~ zuidGgc5JwJ_eXS@XsF5f)Qc4T`&^4M=l*fTVU24Z z>N6|6G=ZqUFnNB$X!vMEHO;=>l|XPzGlEAmglKroZY$c5;^oWCP$#uZjKFwvr-Prt z8#oncZFigpngBc}!xi0|Ptm znpzwUF{c*oOyeZjD+ULYT-dy{ey;Ol#FDuZJQ5=D+K+4XQp|{6ZL)tC;ZVbvIH2_~ zC?oOq!gOkM_u=4*-xgp{`pdL}twQvuE4b!ZW`pA%`9?!oyIt_P5w-#pveW9`Cz`Zr zFI|=RN3I$%D#eqe8rN;n?-oKZuGJJ{)QJlq_(_(;!Tk>idR=$o}i#X8x zfvNk}ZBr=i-hPgAxRtm-RhZ6^CCv{+v+K z@qI`TJF!|DaM5dW;DO9g2GHo*#N5b(qpL`;z^o#&aX1k^h@TD+({LC+jQxosfn462 zDmhz{c4dW{&#c&K9%8qDBxfTo6N$bS^SDGaTd27m^b@@zG8uCXPzL+DgC}7!$BCU% zd@Y@JEP3cLVxgLG!(_~76wY+(#aX80G;lhyHKQPXfo}7^XomM>Pd(lIoJNmkm!=_U zz3qSkL-O;{2JJ5X?HHS8&PL_9vrnF)W{lk9)mIXEG;0!y4M%ok1&!ZeV)_(!ede1@ z9dEMnJPU8tPefP{sEbZ z1V0{iVi+YI7YgUS$$5ao;odthbAr1uYPE876cuq!XpJZ8eX89=xb2L5&qgZpY?5$R z8bUFhb($yZcZ78^C;#v)La8@_n z@K}_g(d5%FG3HhmNo))MfKG0xZSo$^Xr>KD@`C({Jb8Zz#3rnQaRcwk!M%m_+Z@%d zW0pP_*4uaZfl8a|4eslANY(F-x6n4s01<{N_e($ewMhB~p9B+5ijwP^>hCYIf+A3N z!{0;7wc;bGlxG&ZAI}j=FFx2^a_`hfD!ZG~LR6R*AWVQp@H|N=eO0Mms%I(Jp);X9 zykJ4M(NP7qlyb0wcQK`ZiyZ0(oJ@^MG&Hi=>_HeuorGuULB?R|cJD08sAte;K-8_n zV5+VgH#uZ)t7ZvVdf1w#cJa+L9B7@|WZ2f%9gmcsdpu z)3q*in=%uvDG0^ZD}~2c=P~kJQQIDjIEE*^2kC}*iCaINlYvhFWMz!*N<@?pN;><% zyP}mhGJW02VH?x>_uT6R(jGoFz7ouXIDFJ^_d5O8 z+ErBR2lVSRk43orc?W&9-dDQf&|3LZXq=|N5Tr#-15;hf53VbIgoyh+6z&DWks9n9 z@FF5lXxu_gTcawzL@RCftZ-(OG_VnWybEoSc}K9HsC=h{sR0l;!RP+Qu8hL#V^@ zjNoX2MD43X)SbXM$dK{8iXikg{V<+S4b#|C!_rWtpC)Y1^Z;ks;9D})=AgGb>?Cc^ z^xb5m-;xj9xFON*(*{l`>1${hIzY$YIS;F7sP5SG^o-;SUF3>%nYI7(9b>Air`14@ zqTG{3CoI3|wBn#_>RY`lPNzo9VCJB*ZQo>=R8U1^#et zH~Y~K+Yr|<5R7DtLrG{_4YbjemHymZgmftIB%jIAX>_29Z#|Clet~_6eRkSz`@`W! zt%xga#wlrIc_;B^e*rLmdK<*O#cZ|nnUwqevT*;-`EP@bd;MXg-mUryEIAiDS@Kdd zh%M7v-zTy&W^%G|%L59QgKGBZuuwK|QDo}!KWHXkjiahZtck-4YwJT-JNW%R%rw59 z${(FCYxdS~vkrt9^GxZ-+lPucH+{hTBX2FI_?N9DEb#dw=zPHIRc>sGs20zIhJ{l! zZS_0gUnpO!@RI>s6|6~6`{J|>)UoGd%!!G1PN5`WJL6{Rkt$my0}{BlL?jUN8* zp&s|HhYPPKHJYOfDlnit*=}ee<+dU=OKV$jv}%;|3v*LB2sX(U-T~?BD+-$3Z0Wf0 zlYT$FyTjP7MHLDxmhZ8sXC&|F!g9Ism}>SBhK`VXT=z#6c4CV^>PhS%%*&5V*`HiQ zDVv~*?3n_KRg1ElB)!NF8=2)W5ymkjxvA=^rw`~F5WR8XSv@Ihr6dnDy+-e$ONsNK z90z}n@Y0%p)h~Ak`pC0MrFTl}mB%ESS1)0TFoxDje4>Gza~u%j3)!9n1?D&Ju(xyQsV;ktTf6Ne{CKg;6QwJQyBae~rGhcK z{N3UAtnt*FsD3ifTW)l={W>^b&JRQt?H<=q9*;oN0Md-F*$Kf0SpPS6TwY()1Kayy|o*Ph=zC(ogstJoTPk%jCb}(HuW2u19uh)qahV-D-fp5Vt z$D3jdY^a+=1sN#TdDn?vX-!~4XlA36rcUB);j&IJdz`%ZVPl< zDdCpv5-Q8qM`Bn>Veun$;8jJ%=<`7+G@(gear5EYLJeM~CVj-Styy%c9k^a}owg(> zjj$RW3~79Y&>aWPnLc&CSGsSm_VZpA=17}Z_9J^OZLULn*5${_y02xcs(PmV*g?4l z+y<61wUe)1R6P1ZIfIvn)_>DF;%z5|op&<*?EWCwlKtW#Y(L1-P&9x{5j7=b6)7bJ zT3&ulo)~VZ7JMSE{t>1mc$gy5?k;&(;=7!l@jaWSYw0=K8WkZlQe+U;@86~9@HOar z_zj}7-Hsc1+~32E=lr;^NxBv_AGX_F-u-3n>8LTE3Uk}3ScMawvS^4!&7}DED}-yM zI&&FHNw8b7O&gVf8)~MOFGd+)WK8qv+kEgJ3CnF0u)akRJ16UJ$PZ^hFk}3i9aqIr zlvZmAH6KOdWt%H-r{?x4k{As~xp=sQLJuNMMtg$#RYP>%zvQ?R&_UdmF{8TzyhUv& z<>sv?1T8^Q5};|cdHxgH?o*C}>N9ru>l@}u=OR6P<$RXFArSC9<&}%hVI26Bgua*` z;G;Mk&$x@~9wVTniy$_5z5L4^zW$TTk7@>s<+SGP(7E(kZJvHgM^>slD#vY67o2TY z{eDwkm+49E$8R}SGNVTPq<$xs=AZ}afh?m*`?h>}9vMAE+EqyWbBeWk%V>v|3}C4P z460FO$}ojf*79cDxqiOUa_jG6gQrc*-G#G3FshzekF-8>z0;N2EOe1{GwzoM+l}DZ zWml6LZ}-x#%(yp>AARF;c<(hl_e`?o`~eI`>!5r_vA3zBn6EvvNgeUN8LgE!6squGM$R5YcHVI(2a7P`* zL4Cg@IN1}{6wwSr*r+5(?NQl@FW!P#fKY2>FUd$(>7Kk-P%S#5IFjAcTi zE)onab)nARTO+`4Vt-b99_Mowi~4!b5-}y;9uT#M$MHg z!)mAr6=;Xhap0(&+*V1=#@hCw3T|E7i%|EbDzMlbcf*YM^(Ob4h>lYX) zx>X$4cG+p_q;kNN^WbTFiCzBs=~sZGwUMuMue;@(i$9JwYb+S=zbP7 z>rn>}UhcFDi=p~^n{EKXT3r3`0?Fcq&tK!q`I3}tv`098xB@2kuT5{{Qr$91X;xi3 zR6OyS1l46WqQzf`SK3QCBVl8{U)y7HF9VW|R7;3o6WI<}zUil!FetcfyLLjZl)MBV zkD#JwBS;pGXQqg2-YL3Do17nqfe!iHY2^tBqvER!WnOOs;$Wg;hs@hqABP7GmMU!WeRkVtG0{nM${E-Vz%uIL zc?ZF;DK(Q~)YcmDw0``~4QRM)k2cy++BUVl?iW-76~?lzl3{3+?1IrAos|L)V#{doxDK{^gSdws$Dib}4}e+*XVo5NggY7s9`}SnC8eo4)UXNDdMu}@XyO&YgyqT*#jkMLKVDxrQ@hjy1D z)guAyNR?l#pNLe#Ub+j^7ux3F`Um&zoRP2nu%iEpkONl+5vjvAG;**H z9K7mPX(s0)DD+M;eiAj!U92FQ`K(&n#|f*~J#9|3Zl17bxADsLgE|ar>zqoXOqR4h zw|dA7nfX|lUQm1~(S5Q*nN@O8U}a5rm%|r(R6z+y(>QD6j&lZOVvmV@?2?vYrsF=c|-KL7sAE!-iD20ADXMs zH1;p%Jx%nIpTuU;DgsnJaxFKT!m;loe8wdRpZ?t&5b$Fl^oTP&-osKm|24)~cep^c zCs~c#G)&{bP%cf!>)eskd#I}M`W>9%_v4zyJ-bJB8Ywk;augs>2)(+ zXWoq8DjsNLh0o4zaqp1;dR!v7W(3T6JwrK-&*5 zM{A6xI`q1R(XxX39Yjzadi^rI0b=u;-Ss8#nH3tZoj4_J#odIWl-$&_`sMX5pH2JG zW*C2X6SHNTZS)kIudjqH&XjfKPf0GX14gWezOrACrBYiL+v}f9K+;^rI9evDX}GCV zO1BanGM2D<82@VK-R6b9wu)%yx3Y}NB)DH2r?#`6US&0he1W-k1DZfDyC2egBvetM zmhw|TbJcTmg?F3g2Mxf}D94hwznq=Dz8x38H9Pkut?6p|g@~*nT1qY&w*1|llX12- z2w@&?G)Xz&=*_Zq*M(WcFqAGtXq&QR7~u7K;oP!&js@%TJc9Tj)yEn{z0AUf{# zLRIEvu0JwUf0lM5HnX$F!%P_#EqE8S@0dtzBlKA%v`_!ZUq-cJrDkxIVYsSAnN|j# z&5Cyx9qzQkH0OmQk(YitvwAfuP4eS+B9#PHa1BdRGr=QZ;R?s!GG`G0}3cm zCmVqzkbvxILdpVE-<>NJpgW&R72f1_h-o%_vH8m15cBja4Ux){^jD5K37%#gXae%Q zOe(y^gL^bJ0>|GnpQo8Gy|f&@s2=2+9@C^!hyf7Ho_S}ndQVD~3k0pD7T@OmG)c3i zI)Ry(6dk5RC{4rn)U;)W2NhToRSw6S3DoyjL!G~1n?#MY#IX9M2D2dOb28Cj*keSb zm-+6;Do}#nLdrslK^2lDB%YZ}QP{Q*CTBZj*2A!qm>+qG*x=7<;5+H4DAOJP9DW(C z-h6HiO<)zf*}z_NT+@E{!k<3b9FnX^owc?~7m_V?s*|GI~{HImk$+^jJ}uUbh;$@PWI zCxc?ZOxku|HT9qWl$5%8)}DkmGb+^L2tXpWVa255!gesmHYT1jRKGUeKYW=)+ZkcK*-Oni9dPZU_=MxrQkr$k@8{JRkb6EmC7T?+P zX9`DUci6I|Yg_->x`amyTy=lHcEF;)y>YNF2sL)L!X8~Z+BBcOeM$SlWj9>n#WnG& z_Jm7W?J(~vJ$y{XTI=%6{bgW~@j(x_E_9bV?N0R#HmM}WFP6GYDnn?~^TB zG-yy+c6qRbg6D5vleucyTXO&{J+Xt=XP@@Af%ynW*C zQO1=9xJ-drR1E`pZ3qIEK3pR}CjB&YyV2fdPO0C-fy@tz$XhArPL+IB5X>ol1)ja72&f2Gr7=rBeOi!BnJ(WLkrMGukf9Zvn1i&h6Q=XGr>fvvlPKf4L8XfFyJf7$Vi=?T7Kd zU8&Dkb&uDeZAn3PQ44vj3_(1tAy%RwWJ z*_?PnLf?X2gO8}1tlJ)YP^zMy`~Hx!CPK3t=vhCmqa@A&am)ZUkQ-*oRJQ3@73miD z7j2YZYt7MPJ9uaPJwhhu$~ncZrdh@NSr1r`$+MKx3%CfgZCXmWrWUEyG1XacB_m2H zYHByzl>`*tbM%68)mAFsEyQz5OP!w3vdq5)Bw2I<#hcMO8U<3wW7;Lj1ng@jSe&#j ze3L+<s%W!JRt6u=~T#rk?Xs>{~5$Z9QfRG3KOdp}d#XlnQQFf(4D zQ#2HVdE#5-FiQnhAdzXROYLtOAwGw$WC!b&W>5&Pz+**K{a-!^Eb|3N_JieVk84XZvLyM^j??yt*UQYizvZ;>BB0IUL4{)K>^J22=)P; zE5@lpYU68dWUwA(1?LAxnf90k)dDldw%QlXayTj7DzJ#)WAi{=5&pt_=iaQenqu+%)WfA6_-84b)SaUPF192emDai3 zIZnkG8Ix7?<;IY8nbUO-w3L0d8};1F@9(-mmSbgG&h;=?RnGXQ^%*DFei)tUr*K&? zEyiRH@OZJBTIGlAL-Z2?K1$ zw)b|nA$i#2C~e3|{grAaxr58KCO_Hq#{kdN?2#>99kc2VZaW6MSV};sJ%KSj18c#=aBPlgInmBKXjTZ3*4ua3K?)3Xjzwa#)!rD`zF2 zAKzXZ@dOuDk`i+Jzu&uX8AZJIX*o=LV3RQ=E#Lbb%Elba5=Y}B z^>G&8e;W$dtwP*wCzPzrFF!h(EDAMPe}*?KJwE?!`KK*&ybWTSIGe+{P3u`v3+;qL zusS{4y1R(8fMVhSM6PX?PI+ZeysJLZ9S?+%?W4G=y_dv?Py>$kgvUQY2#BLgx0Gth z%|v^{npGFPr!k(p)0?Lyu<^%Lc%9tmadX~Qyzvx?W~Pfo9^+E2IybQ&F2Wa>m_{kB zj`4@C3myeP+B$Hw;)2uS-)`c>Z@l`5RNE2$oi~ie@`w-cQC8qNx5E(dntlS^)y8#g zyE0NjsXP82KcB2tEB!Q1M$M8}U5dS$yRWK3NxO+kL6!X`#!y+P8$__FA3r5dy5~)^ z`bMyC8<+v1#+h75`){$YaxA@A(HcU%Rb)7B@~s10YpHus_@aYqhnp@4Y%Jg#-oS&R zUx(UFuE}Jwl_} z!~pd~^VCY^$P7S+_)S+%r<|O#!CS|H9p%(zsYhIRLr!Ub+sBU*uKNSdaP=N*@4uI1B#4Q2PyQ&jE4E528bjQ(hN6}j|eNzLYUekOh+ z7mHq&uy!+rba7xlE|nR6+zWXg`nYEPal2#!6Jj9=M2MJ3bhvwCrn~W`77=K<{9`eoF z>h$6@LKX*T{}!o+Ji&{AwR-fIm7S1p3U?4q(cnfzph6Xz($WweQ=L$H|~|3HUL5B4;~ns<7MBzFaYY0wR0 zc>y~fOjR<{4bhe|302yA9J*Zlykt&=aOnq}!{=8GUpP#oi&{YouzIgkVI=~*+CTiO z^fBbDX@yjSxs*v=$2|Mfp4B-N8na9|*O0_Wb;;R3z6j*vd|k_OnK(@oVfTLiQFRFS z9FyBpv(N~^HARvs`;krIPIaOE^suJhNON;_g^Z{1$lo4MYL0#}AS*(V@We-gH}^f_ z9oDBL`NXj)W=)ejl14K9xqok5uuT)RD%}299X(MtIn;Gv<6ttkP`G0aTD)e=uUgv7 zhaX7ijbJ3tJ8#nSc!6=oicy2)PUE6c>D^iNV~WT@r?7C6lr(L2Q#3Wdb&XGds6fAx z3yc`E&-ctW``UccW0qysMj`9}_E29d2n>N0t~+KR^HD=L!d})^fsT(jFv%aq4=;(Y zuQIYqSKWA2{N^BLDcyD(oTdEVY7^A6s_E3~{?;MKf*1What{;5^*RR?p z8HG@~2T0{-j4jAH40Lfkh=v`y%3%4oxIL$ro=Pt4r(_Z6g9et4kXWqnVghiA2- z(X}#|bcc;Ny`Et|&@bgay}MnBVOEU(NCKVKGd-8vwWK#5NTLQ|q19nwPS9l=k3Xfl za;|ciVJx7Y%xKl(po!Ep*m=+T66kK%reuI5&hY~^;7s&RwqY<=chcXpZ#x$*R5J_$ z&corl+vJpmdv|!@L3V#kvWS-2ydBN+oUDhYul6d#FE5|(b4Byw%ACqZFFHaZ_1?jF zoCZUSJa@&D$lmE){C1gv&9drVNhR{N)e4+F!TDbiLZS2X6zVh0une34%OjIN@V-ZgsK22_7a(GbWU)a(`OZjqqDzMK3avi89#hNQ$u$ z9)txMM!qZcb77Y$CP;D2jf*M2sCLpsr6!fR!|;v6e(@vegP!N$@-3#SA|xM9hXU3X z&|C;*$$sw0(d89PAMSBu-gb;mm;5!PrMPXa^t1D=Zz4rp&P<`FQACsqb^5jqn--!> zfg(-!H9s53cr{WUYMgMqA4c8aO?5{V7d^ z&W#A;Qz`_!SznlWKUz{W{eB48NI`C62vjRLIR9>St_mgk$zWBvYd06}TL$kiE%E6+ zaBhkUSQzo`LEMU+Tc23JQF19_B(GL_jG8Y)E-8MaW2Eb8Y=1o}l7tC?z^c65u`(~V zelki@A{|v?MjDpjc#wRAU{c`wA||*#m7Lqj7MQ~~CqFXL8!ew-9AZuF8)T z+&v)4x{BF};{B^f0w|`@om8MnFDK&AJ6?T%NDoUiNpXAY__*`HwRh@Wf})0Lax6az z6Cxkf{+{d|z1+Z5xA)nv%i&PwRaR2UOf9+Q9n5AsAN4j=I7=ED|0Ci{`vY`>%d5C|wfM^(|W%ZQz17%Yxwm1FW+_dGx6 zMrj+BlGhKz5$|~j(QTqQ5B|JkC8&Z4`;~;WtfjpURzj(bVcQ{Wd+c>N__^_VV&p@y zx2KGkqucJi%|4A5CeqFJFsfb$E5@FJ7Mt9yYlgFVm@CZupz;B+zkaELGx(`a@uNp9 zpGrS-YH7U!NbhgB4X?CZ&J}MVQ{p+N0fhb-acWl^_(pY+q;d39Lx0*V&)(e9#eOb6 z{657mHb*tDi|Rv8a#)?7TRXRye@I=B5`7g;#0enLC+lXYWJxXc^OII)8M(%;Cag?~ z`Yh0PX)(_tm%IE&NIW^6Ob#mj1?T!Dr=7bAU}hYeb3vT)pczQNJ*$`S?nkKmBkD`P96q69icJ^9=o7}F!6aP3;NX)52niT-@*w%l-iyHSZNr`h{V^uP$JuZ2sVmU<#l=J))p zQZ!({Wh|;XUyWO3yJdmo>_Kh?;eA$hffFtCQ~RZt@@o{FIr#NLEmoCEXZ^+%53`wN z44yR|g!H-A9AFa+JLVn6X1xe(b4l>aRR~9jXii32FQ##=0$gY6rO1NsPydmP)=7K= zG-|lAC7D`n`1xI(j|v9~y*IY9PPpy0RYAj@0KP+(@YoF@ zGoss_P4<0>ef6ng7v(R_kYqzWr?zI_eA}fiTIez(Yh3%m{T_+8ju1wW`NLaDKec-FdAOo1W zJ=-~Ay`XtF_0CYJQ{oxq6i*0EvhEWOHQR%Hk9%(a+r%Z6rfCW!I(*Yh4Bc>}O*tX5 zwb#$8M9>5`VAWSQsbRM@9fP~>r0uap1Nl|rVS62=){X;)j0=(M?a)(Pa4L=|`mJGI zrY@N>uJDJFFuVy0!7g$DEf<$Ue~&X2g{5zDxiYJOM-C3dyNDV+n>7pBfDlT0Ltvg< zxrC^1^D{^+DjdT1HS{-(3thr4q3;q%ht5J2j9d7$i%XV-(^f7{ARb%G`8gyAzBJq^ zswYJtF@&HS1>p`lOpqfnN<_v}o{1Ml7KK7_g;#IK|G{1>BUOkFYk5t(bhn82QE#k{D zj#N%wG-sRboQ$cI&dDz|e)^P~3+%)VZ)E8}Qc=!)_-$Wb(3%zq$z=T<8)hM}vd^C? zWT+)D9#!2)@fGPN&J{d|Yhp)Q*|XZjk=H}{Qg+Kfvwhv|o#hGYwj+gw6$OS{CLBr9aU?s-;oUs7A^CV*vjpu$A@`pv22lQG=9+x*5yO5DZBqJ@$g{@ zgCse|tuu*J6`M?XhPlNjZZtH*^IFYpX-E&;&^y>|2}>i?Elq%Nw4D3F;>{9Bn9f}& z7WQ#X61L!Mp~`tlCwO)vWNY;Mvn3bJ)C+Nm{dY+!c%IgJ^R$dET;$6Q)>?FGuNL~D zD^0Oebxq=X1~7-nkMKM7HK=uin`qS)sx+P_we~E!BF!QY?un{kIvJw?C(Gq2GIDJHq5JWsxk( zuCZKa&IfcAgQJkI-jlPblgc)hLuPYx@*#^1;~28N5KtjzlEy=T z@>A6Ic!XX(-5^RFK26b+1xL`537uQ2o|fKcR6TWzKBX}i!6}m7>jJ-qWfYm7m4>ce zjuqsDqhlo}=*SY?YW}|GG^XHQ#lYxa)!{6y^ypNy-ZpSK=@bS;vRxSeLQv(EgipHV z8%PGwR$6(p67kZu4rL^$eB_wJ)!^2bK_nm-z6`%5HPT3Im$n=SODfi`OT%fA#utkC zcV*_$Rn1iEpkC))r47hF+9ww@hVMJTqLJYHITup8C*F%qI<24K#Ooi6p|gSEy?$1Y zEIX~}E%{X!THa#xYV_*sU~upOa~4OXsath>`Hp$6yR#D194hD9TIwvWbM#$qtdjKm z6paiIn2U-;ax-ywC$xC>-};*R-?f8m7_AOU!%wQOs)ye>-}|W$wzk%T6x(%%XzyEN zNXm`zK9gNxG&J+EWSpx@QYm%-2djj|g^eKvL9}BJeUoWfwOfAsv@NG(c1y_=!X+$E zsh_d6%ihw&4`c4ltWs?^vLGgFuBAhz3Z(_3oZy_J>pgJ!NfLyodJJ93w?(s`qml?* zi9XBAX%HfP!}4cRh=6uE%6*XTGbueiwgsUnrD1J_m7PH!rYyo2Xvn`na3=`)>O(l4 zcjggT@(r1HIT^G3bO4OJp(p`ZBVaT0F7v-3vS}{Q44+9KJqt|criX>xc6ZB`TSV*c8Cn)DRy8PN1jMa)WX}$a&gHQfZ@6+ zt#pe}<3QZyh1XQl;v)XGGeVA_!o?$42-)66m@Jy@FN?@KXWhnAj9_IbFeY9 zwy!tID4QI(_@aM{7LMUUz1(^)e*GKWtin&4z(S#lMI+D+kJ_uLxp&yY7k|uR0MUrQ7QPSXG;p`x73qTci`zY-U{lEev_kJh~L571_SoOixOTw&K9&q zFUE68@b#(BaUa`uB94hn=WEZ5@OFiK+xHtxTZVNh)0BoAMOz_5)w7DBDbel^DhZ^6 z2}fPhpB2Ldn5V@em2VJ?%BPl$3$QWA?W{^rH@Yr$SOYS&E1(<%wZrN_;lW?vsUZum zi(AmsKqpF;$(B6>5jJLIboVobw~Ia&@Z5UZlS?bf{cm-1Xee_+$L}7MRMV0!+5Fb6 zjALl=yvnJBmF$&D>O<2I0aW(%W>}OQxsT|wJz4cn_N=A83DHIVL6Afs4`x}bOx?mM z<~1Z6NEM6c%s6yS7Jo(|zy#*)ukYrxT(NVXJZVfv;3?d6ey7BWby`^s(+Vf*&H@4=uN# zUv35%4qvRxj4aAsAC+=rbNih0u$)^fmStPDkJ?*QAwW?B9kL=6_v1sfJjZ0EkF269 zI&|26hkUjEq$ylPc~GfX_v@jmWW(c6KokS5Tl*M0_h%t3%kVepNeBW(IDNQ=Xj?C% zUmltHPsnNdnG>P#4r&22M;yqLZ_&edo})7VpiKLDT4^Gq&j6`gJo)t9W0<7?wmT(qyRtg;;sYt3^`AJdrY$8-kF_j<)hhSZSrMEQf5+MSeh98&bH^EFXGjwpRUyN`Q(L$s>^y`3~3}A(Irfm zohM4q;l7X=&UtJ-{8PXv>m?(Ls_Eot!)Ns80U_&=-1SF^B5GMo;fSA)WjdCKG6SA* zl{YtFP=*|mSy6|9hB_4TNZDFw2UV)T5;oXBvzw0=0wogc^E)9Gb>hdH*s!`Y5!AYp z3~}2>)_EVH#zMN8tt;6CkfI)o^LeUP(d*dH{v&^}>(uvQ5-z{$Ej@|x)LToa&9p@w zX+FId1~#vB;|_bsIWgZYt}^1&m(6D=$COiY)oo9~&hz{sk(U+J;5ySzJh*H zWSuDG%g;B>pz`?|KUQ63R~5BTY?L;HZ&4@waX}*M8dXz47%~ndqW?HLk1aW2Ad3DH zIV{OJXLe+0K;-!KX1wyTeR>)s)vJ4ZXo*S@O<6wuRv7~u=F_e0;rAL7RL#aru@TNe z?6N83ipu23mB|*J+M8vDmVb;ZETzThPQi#jOcK?wz!{6&^WU?L>ZFclPZu=WZVb*C zQ<{T;{4L5xRPWjdko;J|az-?LpkF)1A~%o-AAv~2`7wlJC3|&97VQhlZ;Gj-a(Bbk zo2h!RXWB0Bht@yb8{6|evPIi_H_mqW?5F;Gyd(fmp3b{Zo+ps*TEs&^h?vj~CXkg) zKW4cU;-Z4xPi#h`=PhGx=<+#Yc~d#VkdQX~$JJhc%GnVnTO<7jKA=?j$_wKgeQ^bo z!0p6o%{q0A6pM~Xec_r3OvaFNsj^XF!yntZR=4oR0EVZU9#E8ByRGeT*bl)BOC?zd z%)Z>##k%5#^X#c}tE8_ZWI@i?rmgYDR~rp9XF83a>FkRgkt-FcZ@#G9ym26Es~%0S zCm`)?DEyCoe<9DhNge~bjKzb%&Q`qPzS?vffcrV=g{+xj@bO~m4j-lQ6#0qtdD+Na zOVH7b=MX|wGuj%P2k-T%B>eEu{pGUw#d7~;^CUOO=E9%7ae5}={BJPFtKSjZXQPH( zEtGDlXvqw7M0ii0xq~dHB}cAseEaHAo37WV{c+IYm%w*jVd$Q5j88s_7zqP9_HFl}3EW~{i?*sDC}nf_sT z9Ujj!VmA#^lI$<;#cB)3LGAX_v=fpJH5yqx-=KAm(6gUqYnCY4_|G>b6d!7xFQwh+N>F#MNce>)HmxM+GL0+(FKh1SDa zc&1)I(bO;2M0$vR{EkPx)A(s#6LMjHt+GzHm*rKV1(#8vMR% zLQYuv)~FA5i=rvWq*)H0g9^IwkJd>7qBWEYUhaxnCQlgzx!o%W=nB;!PgKGVUMI!! zAn1joR4*EpM~q%nl5T?0QBdJj^O?R@tRBl9Dy!T4y@rcVB`2 zzE>H8c|0jLQw8n8i;^AY$KR2I2|H;{kWQTZtRuQ$0WlfZeSM>KGoXbJ(l1n*V{=n= z!3A34TMds(zd-PA;VNvjLqSt=$eQ8*~-duz71{m!xVdRY9O0YdgAt48j=3Y zpUChu(R$`Tf;=tSn-+f)J6)0j*1P5$HIcTUAD+ZjCc1wQO0vF2&VMa~a(gu^hgE{T}nYx=S{>d&sbo7<#p zRSY_JDq4xoU{GiE!aJ>D{5}m=;rSz;&qe4xmcx07Aj|OM@*OJmGRkCNSVn3Pw-yL{ z$!Dca7>#pv`i&K7TpY`56=jlq4aiz|X;Y5WF=S-*k3{ESP6;#mu%OCgPwWD58k{kv zF~+LEc)ug*PS3TPt!TV-9nhcb>|V>YA16$`k@so?bd5QqbPu0uqk zG8yzGF-fwEFY<^Yd-att*LEUa^e&<0J@j23eKcM+BV5g^M_S66TuV>qY0p}rLg2V@ z;t@ASG=V6*Vc8bV4;SMz(8ROtRTDt>?1ow5_Y^ci_WA|dpvSN@mjVw}@tvQh?s5;< z6SCpWsHu>ZNMs2nj(CXn2FgZpm?eq;xzlD%7OQAJ^g7vx5h6 zM-rLKz5SU#Ns5x=V8U|-av2T0)n}WMK*FW?I)Kpeb>PDpHHWVshS5m?$gI4WOezQVf8UNO>Lfn=a zQ4#^7z_~137DO3q&;oLlldcOQ5Nbo(amg40|>AN*qnfm|9nnAz55G!yI4YWEhr zrfIy&ej)M9+*?i2M@Wa~elu};UuSAt*0JM!KZCFaK&?ju-F{7J^)@lfd*s8=yX^P9 zmu~0hlk;Lt+A*``Z{#E@twZ+JPYrE&;aFKI8jMmhA~l7oeaVxI!%TD+dI>T4gEu^D z{1EAv-*RlYn~f%Vje*OvK2`@y#yI?hI=Unrs1S2w2!VUOl+@5)K0O?rp6o|+nJZ+B zwmM*LKPxOTdN0EKw(SS*72%`bKqq`$ykx>w!t9$Wzo04$NVEW-E-2j?n)1x6HubHL#aMjQSwU*dFDGVTyg#=qtrGb_B^oSgd`S|Kf7fnM5($;U zE%h_Lj6dBv0!a~pILr6V5de#fzi+&C%T12813cFyN=&oDMd5FA4)2f4)28p8!;1Ek zuKX!x_d03CrA}BlEvE?o2+snaiMs7rhkBVc4oAVhSz6N$Ut|GKl^MFHv=3AnW4xrh z;zWELX7>n3-W{K5Ty{4o`MpX8_`03c#0+ZYCb;Z3CLHgC=j?4X_XKEII58UW3S zO&}3OIhH;~!i$`jddP($1NfDWwtodE3?Bh?ecE(zDzuwC?qOhl?@H*7W4#O>J|1g1 z7$0^=3|949VN{vP7A2HzuGXB z@t+HncktP`*1SFnB~4r^`cw=clP8}yGd6Z^$&&z0U`oD~$>;d5Gi!%?W#b^OiHW~36RTUcn5729m zGY{Ns^6{C?F_L#LpS7{?z->Fjy z^tP8Bteuwv+-1Qv%B81#RSs??$`dzB6=^Nv?+ja(ce_idcptm;LKKU~2ts1Hoc1^J za#);=;b<98+}E*?d;kIRy*H{O%Q)gfac)plSQ7U0KDitcPJeRFs8={rHBfp$G`kR) z6d@5SH^W3@cZWzB2*_33LPF}Pnz)vjJL~a#&@xQuDKXIOihLB|V|(3pgGIJKF{kz@ zm9&)*wQBC8mO5syR*>!Jb(WC#Fc~H5EWZ}!*kI9m>>e~n2i{L_YN?b1rNFnUFPN9of$w=KF^mrdi*?)VCK(jn=KgP-o1$9_{CD{HYIuKo6E+Cvmjx*C{rFNVig@uWTnr8d$1^jcOYaqOzk zMjsna^A^cxmlcIGpb5=h$#H7t#EIUv>Yn_rjS!hm>tX$vdC|2umMF17FhMRqaT1E) zCm_?`gU3+FV&HG@((8c*ZG4Chnec4H0&a1rx0|7}r9MHB&>d=uPW|Qej@PF?i58(L zluyMJ?xR#Zs*{&3zaT8|oZY9;k`zSx7>_*uc23Eg7c9!gKKha zi|zx5{A&sqAC@mnl)ySGIkv7htf6`p8yY;^xB~obuwvVF;`)-m+YP?Ck-E{aO#Zj)xy-Xw6_^#?qraW3A_HKw z+#k0=#gNUM-~ap`{EmGfpw(Mb48c)7xPyt-%*JzgyIr{P_`s?=j(*pXhs71n7VR#u z8^`GFjoFIb=ob}d;#Nf<^bN4yuMg!=j!nYh=HLT@7+$_p%HD>k2B&#BGwPCJ;$iK`=2Hz*?nC7dq;11wIE=IU}XEK z5tv_~=f${-ySIV^DCKI0Qi9fa4)n<{rHoGlGLB)8U-JujXcT+~cZ*PbA;TzilaQeT z+=`B8;YMk$RC4rTAEq!9Qrz9fz1KC|oAMIZPoeq-s=bR}(_d?t^Pv!%k7+R_;QiaI z8u^zGUo_V00c*|atHQHK@b9fnq1?5nR3r=%+z?05jyP2xMlW%|P(xTUmIO|Tyb#*6 z1o8Cy7!asNWuDMS3z<@ROqxE{l7RYDt(e@4RmEbhk3^q+^r{`_x;;?Li<{prSdP}% zRZ*oRZRD(gE*W9qLOM1|pCGfxZr5fKOs^Q0`daA#zhRA#hiIAmY#>EH(omn|yL<6j z^UkBvXg1RbhE&7imDlh=iL|)U{Xt!=g-kCOyRG8?*sm?iOdRkf7wL2pc#poAH$iFr zGOowfJvkHyP@Wl%BLabI`X(wDxi4llql{6;x=l=S84{DXE$Ri*R|Bc^t>o0x zm=>H-l3GFAo;zUq{3fc{tUC?bHctHwIED9Y#Y$3e(d79UGk*Vwn+-Kf+^U4^sMisR ze)Cw@I3N+Rz$Huv}AWpP`6b3Fnt&xr~Sx+I~{!KX2mD8}Ulz%!+ zyE+8JP*CNk7iz%JJkv1cQA5qHnS;!52&c;%HY>T(lD=W#bYYd_u_{-twc@&@PD@(h zE`^piUx-!?c?V(gYn>fFerz;-!T=pVR!K*;I+8Ay}=c$k_;Bp*-StJFG1+hAw^-2FGB^5fNym z{rqZCGmibuBpf0!LX$6IyugaH&S_1nBNhR8YpNoR|ke@QF4SG<{t6mjDErig*Q|dU2h{n+) zo1w3)gpWR@H@jpOd^OAuB5gdMl51G%Gg0xuqmlS0I=cN~%+kOCm3h{?x{|A9IWs<{ z6k@WAEdjh2epcujx#LS2C3e}*T!^=^H=fS;?7V zUR|-{Rok0vmDnlW*Kdne8Huvi_uc||Wn1|>Y~U8q4(XV@#)(!^!(XZGGvc;YWDCxS zMPqy}b*Sy?lgpAXT&REh1h8xBn;$D(RzP?imFH6!C8Xrsn@UztpkjqwuuEIB!lYRL0GFkW%w$LahAvf z>V&isuDLbUbpr`zZ8*_5|JGhAm5oz{*p3b*TJB2Eso zQtb-$MMt$C8djyy7j`^;Z9C=Dhs4wB_4rh_S78!ZtpDM^F~jVoK5(*EOhqeyYCENm z!vMVOGzAiaEF>D}_)>WjOsYU1<_Mp~2>KiwiU;~w$iz8UH95tDKh*n0Q{t;^L~f|)CuyMrqBmcKunkQS+*aUI^g+4FVvQlj@>dav;-%nhH+6fJ^|DPn z9!T=2JH8vnff8|cH^e?&sQCgVLAnGERZ>THJcny?KObu3GT(dBYgLipmEh)w6qg8O zone_bFJqK$`3xm?slR;gS?C}hh^RPYI8XPB+z#?EwW->5S&C0@&C97;a-i{W*NG*? zz6;NTsrW?nyQ!mrY&3=?B7i5td)-}JHoar8u_j&y)gc`BTvmCVOF#mFN?WQTDYrf7c%R{Z`2g&(Hku%^|dHR2~zXa zw>zty+bG^jRYmQVGU(S2I}i-MoQJen*lAu4s150W zp|VGR&|57Ub!sUYoCK@(<9-<1Q+lmx2q_Lf~%oE^?#?ZoO(0G z@`v&)fcyd3s1*|KZlYxM?}_x4Cg%gjZ=}zTd23XEEU!k$=!(RM(QtBI;j;Ni0ku&a zayDq{P&F@yS50bJ(&opq4<_a%M}+#$M@Y3S=0JPD6>K@JCaf}vzn<=0Vf zG})ALAZEvC_XHB=7CHqaIsSAQyNtckdul7li0Bq@jZK{HWBKxfqbL#u*E-M2t@NBUsJUWV4jpW1w00>64aqs#zAMW6mtO&yJcZqi;uB z0qTp-MVN6>%*pw1{Ecc|iZwvu5_@sS9Y@RM*s-fvf)%`;Iac-T`RcfI&O}?W&UuHL z7Xhmjq#h5*d0S?Mht8}xV^CTC*V-#10LGTOOnQUyzOzcA6l2x8E!%P@%(^AYA3kpn zLiC$1iq61C-?sXA%34f>t^FCm{B_xntp!Mf@yi8G(J5uH)8XB=Mh_tt zzG;`4zGkTwl(=r&%pEw}l^ZIzY{SDKtuLlnS8ua?!7XC?YnHChQyLykAkE4Ug$P$q zw9Ln7f`KIV3AdqTg=YY$zz zre?Y7$k=X^gN7)w9SkrjZ!c3rlJBC^gCT!D$%i-xpCu@-t6olr;K!3vL1d%b_L0zI zb-Y=Qcy?CI-p7v9=YY>G?Ij?AOvB}^1_n29OFvop>Ke5$6^vt!y4l(M!F|y;N{r6k z1P9FqILv&$NsEHK4kn)FU&p621o91|Z%uQT0L(^FCH5!vLw_-8%y7dAoAamb@5LqNz+Hj|mk*TopmDZbN3Zjqrfn@&tb9P(#<6btfj&`#Z9PjRr-wv_HJQyjZsmi(88} z=X7rn?o^e2n^zEP8{B=RopKjls|Bu*QFuBLBub#zA~F+PT#m92oJdo#-JWfXiZCEg z^a4n>c_`oIL^}2nd&T@Us%rEcSj+KOe;IxAb!YShlOD!@)Kfh5Zyy&FE)2%F$CNRd zit@EvmYOVq6LWkoDwHdA*)#+RoIK~??_XLB%U(MhtjK9m8j$0h`gKj_1Gh(_-T!#| zdPP!hA^7J~<<{=rw(B;kA%k3zp8ip&7vKA1ci{gCH3Mksxt+k$CUI^l#nqzWhoB#EVpK ztid*49=6!(E?O}6TNIb-nr@0!0biThE;Ao7{Dgmn>7moA2K<^}3F*&21UV1)3AlBz zw4p3BP3+cj`?~q4L)@}cCe1)547Q)%S8KUV9kr3~2wT}k%uGd|MsGhPCKnq|ke8=z zFKl}frU&7`Dm&$u&`4SAJ&ImZe?hf>Dca}xgOXi2Cc(P|#A196wSm%Zf!1p_Te0_l zvc3(2Q<>$?Bib?i7!6KC(CsKTM2M&dJeyb0hpf^m{d`wYmr3I(h58UpW^U7gaQKdr zD=%PQ?Ofc%)!Qx4v+;U03;D2_2d?bhy~PPnFrq(wr8#7e!njNa0ENrhmMsLf)fOLL zCm@}dr&3#s*wKNuks>QFz*P$8?*N-8G)J><*7HE3IlKmiW?;NHhl9k*cWfI^TFdnP z;Gs>lFau@1&!uND^(;J@^2u+=fYKPG|YnGTnN46A@AdBFeoNn!s_Sh zE1}DBh{lIhqz|v$@cB?P#>a+3pDrmpwVzTXRDV68ndODvNwdr9Pvo&+3bvXUW_@r} zyG>+HcX#t%6}dnr8y(d^{Z#;9ByK9}hpK${XiNN47r3&jmMCJ}$kfR>((*@I-QhOy z(39DYyUJ=RO?<)2_uf|0VnOR_^H6!9=J zeo>|+Gmox!O-iX6_}6FETA4&)yxR6 zaq9L?Kk8+F@b7+FwG{wfk2lG%Df&Z6QM1HMPe-%%hrQQ+e4{|iDL5}PAZ|XjVyYh+ z1Rb7AwMF2sSw_HA+9V=NM*tBC07!2cdz^Dyg2U}wILk5OyXkI88pt{}G3& z%C1OoOf#Fh(FUqq*uu0G8@;jXZ?Z8>$Rw_9_mLB5nH21>Jk5~7x4qA2qq@z!OT}j^ zF_aLkT!w)g@#MHoRQO)ra(3Fiwy8+WV|R?E0zCrQ1GYegbY*^_ zx$3@2Q|CQ&wCxu0{L|X4_m1m;*%;|B6mE$xphFQ^z8O`D55>ue%y5AZLhUB`G71OA zZ3~=Fb*KsWR~Bu0crhi$`B8n4bxA3$jn)a>=74HBPXD5XxRvpiTHOBoc!WmnD*R={ z659tD3H3{;A-xgaeh{G&_FiDamArcUsT+aYUK)AmHQwYIw z6{1;iVV=A_-u%tSs(7Z}RkO&tV~a3ACC?K2L>_p*=1NC)32Sq#3KrC@j_MWdw!?<2 zUsDTl&%+~p@lfCbF(jXcmuMO{i-oVj{rYF|wJnDk#wgCMi0J_SH4&cZIX` z^V1#2-$uEyyElp!8hWaJEn*L6^UE_|ZK1MY!^4DoI;`@oUwY(6+9xd^np0g9_Qvh5 z9mUwCgj0(&7=~pJIi`PE*lf!lT9ZNeN$hN8U#AjmWw%o)57#hn-qgf34*CKxGOe0u zMPV*R*{kqKS+x8I^Lsm(j5nh*rscK`isj6mO9qzp>;6t_)t-9kjW!3UyZP(=dHLCK z(`~N{Tt>L273xyO$QjYLZT;LY+&5Lou2?Iuoi-#D!Cvpu&{A%(6K(tK21{<}$ZW&R zFp^sIR2NToFDkl$n2)=Ly>_3lFbq+o;(d#mnSKQ5m!VU;k$I z=mbmsdHUpMxpcXO^!#9Hp^S-G?zwshxIxL$LkH1TY;!ix_3#<4DqFbiK0)Hid&IxD zA;}9a{E39AX=b|IUi6#2i#JcPMRWlNdCX)W(~}~B*jZ4uGnhUlsYs!JjY1^j!*bD$rrfu{x%e$`%SCCWk@_$FSwmYkQhJ~^NW~bn~s}_UYuJ2 zsE&IG*|(Qi)yv!kDe9T?*bIrHrJ0-hr+Z(ULdS^T{O(7wUSZdwHc+G>{$zeGJ~Q?* zNb#zhwjv@Lx4A$H&%9%$lax5`5vf@Dkg>?GWf@(aGtBg{2P@f9S@1sx`NW%j*>l{| zJ1+a3l;3U`Xbwi-)-53SDSqnfW*&w5)#LS-?kY3+iM24^^ko9TRwhgz%9!nFe&kk5 zL1^_=9ak`D?05<5+kST36m9`=eF`XtyDkT7#P5>~*y}s1-PMW`;u#my((HZ14O7e`$M+8fo`HeXgmIB99Y0^2KQ^?E3 zgy_Dr6Gutl?^0g!FVb)P4X)_kHA#&R=9>wtYJ%?lRiC6Oc&?LQX5e`VFH}qJ0b}?5 zgtNZXKA!m$AqQ$<@}t&sjedzskG)@25Fs|9NC!=G{)Z5TkD7MiS|7U04ExyqM|BP< zLby{J*X=qC6IdwO>kMv}9TUrWwB&W7pv+52=)nik!*OSs+@H8d<4rTGgbBOAnxwE!b3McqfiUv*clbCO-_qO7sH+^|_XlVLebq95WT)(lVj6 z_Cr1}0*|Pzein`K9vNxmfIazW1RCOt{&Q>bO$j>cmaikF&aqir)I-MQW6_=zQ@grj zZ}%=aOK)@?7Jr@pOf9k0f21=*I|SqE*VTFd8@_!qTt) zR>NJ+Q8zKzQ+^l0;!?6^W;97Z`BUpPNB3lh`WQ7!nlN?`4sNSk@At&Y3XapJf)v?} z{Ez9h%yinaBO)1}Yu!;v0cVFOxEcMY>d6z8zdW~(UdsW&3=;lkC7Tj63NM|X)Q{Tt zuA!vfdoxOsYtF!dyLg8CCqa8p0B>UQ`DWSlgSgYSH@34eBd7uJ>HwHc?ni3SYjdw6 z%(S5-)e43eVs67&^1H>R(YGfTHGXl~n>%sDmlR^w@`*U!PUm_Enfma8)h-L+ZGF#P zR$NW0qcPjCZTTc~JG;lK0`*s5c$V`TGv27z5V7yvXlKW(I=~wYnZD-kiQIl$)xPU2a@K7b59cKkY^T?m-!X=Ku=ng zsdbQ_A(16i@KdyTyMWGAw?IxZBb#Q1W%syy4&fQ4Pjf2@+Q=3p=39qEF7MU6k zTydD|2{B>apYaXSr)JRZc(IL($aAxBK2}5%85?_z!*i>o4ZG`DPwg(pqesyo1u+Z` zlFxD~)&!QRO9B1ti2~NpikY^_g2j8zahrZvhQKXT@^`F=PbHe^D#N25_ zu@KsQ3=yh3tQS7jaiU}O*_T0oCnIUx3rs9ix1Z4bY9JJ(+BbDA^E> zeD>2|ezICWCtKK<;HhZ8ZJ4y3f(7R; z{vD3>di$Pcmh>al8B0<@y@kp4FR}IObrOpN%9aE&%40R%OEFh z2*-(!-*jLXPUJM@$poKV&>&zVAx3YPJfwXIgz}y?OnteqL>i%Ki^y{+O2)T19QM>h zErF_aQaeKXu*6SB{EruuG^6A#4Ul#x;H)s$UJtkA73xdC%2z7Ov3&5SYzWe_TlQ3} zU5WzJ73nW20o)^UTa1cZYv~z5j1VZ)2Tbdp!qRAF`jlC$4r^c0ic|jc*MPmtaQ7w2 zGS_<~j|JmhX1?Ted(zwd+Zu7`ethSw)-nQu-3$_Geeok2X-(72HcF~-9MQR4ey`(u z{{4N!fu2f4iRLYtZmNRGY*Xk)oir>XI>Cmr|Zf}9Fl5fM|{ebYvj)o?Q|wmAFv}VOQlYea68kv)CkQQrA|ob0^~4p5r;HxvD9I7Ne5b zU#Y_Gx{er0tQP`W8#qI>JgGwWXp?7&LX)R_uwowF1VT=x6lq-|#CY`vP?qJs;nxZEMCt@cJ+S;~KyUFiI`o$kS za661t%4osgvp*q3aqPs@OD!5~m1*V2K6U#3{^5et%U>RZ$Vyc0C9Sor;~7pR22nC> zZHAj+9QZGUFrs$NZ){xZ)BNaZh#+l8!TzNOAo}3%$=SD=W{at6PY>$aNbi}Y6Lg>$2H_95 zE2E0N92*X?O-N`DJ}7sT1m-8UkA9JHwF22&!>edf3(#%x;#oU7qHXGRb<_rZPze1z zZ)Q@p&|JUE_gwJ;*yZ$H`uFKf?RrURMrdTFbP6*xhn3NKsyaeh&lm#6^#jjUN(*Yt zVE;^bhw5H=^;$o#GwFvIQF)g;SnUap5wWlEO%x3`cDv7%t2Z93cU#6`T_lkwOypY+ zP9aLh)Uva0qUd>eKSpUz9z5h{JGcQ%9C6h4>wB?{%ppUo1{}AX89{nFzd2)4=sTLl z6H2Gw&E|iSsL4N<^rHx_HBS@cnvfKIMMxA>+ z-wOFKkFLxm0fl=u9Gv-s$k;conQa0kFd5sSPscso4bLUHtqIG+NPs@r!hkZ6;{ZvD zMT8-dnRWo;`mi zPR={oskv(({fPz_4LFfA3U_b^g5pd_-TEO+P;UHiLBf)sWYRtcjIUjRr+a zlO3dE3cjPu2;=5}(2~WfW_&b*relz{Lk@4K>U;+N%>sm%U&iNG8~?ea1#uvsr%wlZ zy=_m0bepzUa62CQURp({fMDlp526X+%SVHL@=SZbK#g(c>8;+RoYmCRrzgTR5_87s zd4wC#qJ9z3be5}`=wm=vET0yZSbW6X_)16owEPO=w(60np$qQbt=%JnN;#JD&Q%IV zylldo<5iXm_Fx1^@O_6a$|lDm(h3fbvFpX+AyA?!gxmCVSG?N}&L-2>)`^xr#%7$( zp}w-2b<_Ax{2&NBcdIqF*ZU^FGoml9-{VXYv}kvV6`?{IO8os;_y_y8<*n`0u3Ke? zm@>eo(LyqQ7C6r5$xY1=_BI1)q|<_Rep5EPG<&YQld<8CKX ztE8Li*snJ%L79uuWi;OIz~$6&6;qLxVfo{+p1$$FHWSE*LW`!2Mk(xmrtrwd7E6;! zu*8yXNi@%?1RKIe#Q7uuV#4?0!7Vgy*pR1jg%AqrA+ATpJBjlmkH4kdNW*D|Q@kn} z6;wAt%Q?65%y;oC+S)TJM+GTWdIe8F0YzJfac5S9^vTh$2@q9D(~ljRx!q&Qbg9w>~1=>aY5TRE4%& z-1#$7*44w99<^R$TMqK>&i;8u7NK&fOtldLw@wRnBhata69%6#HT9eI?iZa&epu6S z;V)f9@N`WbW%QEOP5Ll)RFedCsi}-X(JVw86wqnWe#7B6hHZvYksLr26Kl6pmI$T~ zJ(LJzXC_PIi)+q#F-28KXm5P|BE%Jc?Op66=h7%8&QcBaY%+|_5&hw)uuXDSlkL=& zx_QTE2A6s7=jjoN(q!*J1mf@d&-m+A2dFRy`-pQ^%>xlpz2E>@<8-PRtNA1AMHla* z4~S%McrqeLA6n)yS*0Wv>(5Y`jI9Fj;cn;FjRh@<8q#3W>hSmUPzj@vQe>d6Z@aK_ z%g#DXSXMuu&sZdp*ZaUdQS7Gkj#Sr4sv@)3-Qe0S2s@rPy~ol)#By_uWqDHHm%lq8 zTuO0!WhY%1wc)uvuv#;ncPL$Y^a;(2r2!LJHwi%M zMN0)lS1@w(z?M#gQ|FJs=c_(9>PV2j+HZpE%h9gwDPQMycdq(0+>%?YS z&qMJ|i6TY&NQLse)sw-#Xrr7I;OYSa9b`-77+uC=YW2qB$i01BC?ac{F z)6NyIM*5z$P(@i-8rkicmVS7|k9kC;d8aE%4@;F?V}?e!s6*pR18mmuEBkiN-wPV8`pX?r{KO|XCtErfpR-Gpq`z6<@cK@7Zvgx$gyE!$R+CupzCqV~&ZdG0(R?;bC9Pt`w@h|#oS$#;&& z>ONL1A}ZfSe5gL@x@MAHbKk1=7LHRacKA30m(k%(_<+^B?sE33Sq#+sjOU;D)c z4tKNYjntIaeE=KuPk$0lzp@ea@Yc~sWNLLU7a*sU8KL0Y#!TQ^n!x+eisqf^6Dfz-+zho~{~of~}cLJZSkB9}=wpQz)SQn)u> z!jfz|{8%oC>^BC2-G9k+rY8AhCUnFRReXNKS8i zEFMi_X<<{!V1h0?zo*lLIl9KBQ#7yF9}?6X$6(7*j+rI8JAQWIh*BizHKr$ZAd68A zyNDk2w>Bg?lnrRwuNmpWvB_PaeGB>N4>I5|`l{}@l_156&BS8obSifaO9rGE`LE*Y z*RaHF9yL>-aA}lbp4RnPg48YpY9GOjGHWlU2GrJLx)Y$?PCTJ?wQkCd`bKU@(Dd?0 z=t8eqco3qxeCE+wY~99%n&$S1A;YO;e?kjgw}XhcLgkD|kr=?SA5`UBm}tFx-7;hv z$2(4Dwh+?vf$*Vkc-;2l177ekGr{v*94DAFK!EfpG?Xz~X) z7RJwB3M4L6ijPtfw=UC1M?_YC$5TwkLlzNezRse!&pL+uzzFVoSA2ev=89F2p z@F$tlJ-WBU+y^}~T1=-*+D@1NuA%lxHC=ut=NY^)l#0YT(Fk6MFKH{d)DL2M7Xb9) zTNQZXlO5NH`zwsbAfDu&oA|+sQ`@8CmIfHYPWc-ALcdl?E7S7M+Jm7J(oGcKFPg6G zPNV%&q#T#klB$^MhO=D_bF_D1$Zp1B8jYOvx-#(7@nH`=Udsnx=U z%C|6S`!T9X+YJ6#zsen#Sah`fRI~At%)|*Xce6{UX5Id^Zu<5_&o-=PjY%nRGsSvv z&L|>W^I18TWpVinnYIf@zRqCCi`0 zLhm7X)_uzQxEU~ZPzxZZ1J>6;O?OyHcG~uqzv?u`U)XVY`1`Th*CX3n;n}gOCxMtO zRBh|?;?=vYJ25NhI#!^E5(1s=q$(e(9Q7YES`N$Ymu~g~j~4i6qaU4x4;%UE5IA&O zR-7E$4kTk+UsI{}j7zQ+orjylC0>h6y%T`de#qjO;~$Oz=UFSlz1e>mLd71&c_pY& zDS#s?m{SkO8ig3s+cHHAKwiHuZTQNNuEp;y_lyjBA8VYc*b;m2Mm@$Dbonb^|W zA6fjo$TMgr4tXn^AEaxaT<7)IjJ|F>R?NVo*V8^fVa?(~b;jW21Gps7&u*Zn9hl?L zo^BJiHWc%9JOitDb9In>mo;%eL_ukrKXAw=P4+niC$mMo)4qsbFbd97UMIgK3bFJM zBf}UGDHBS_+r3k!I_*S7{RTuccNP%rDGNF$ud-MD?S;a>%rGS3{4s9n^1w^#TTunW zuQ4^xwx2#|Lf#Vzcq3>>YeL+Pq$|>~CJVi63s^-B0Pg}nHwcrLE;@2JaN))&22sm~ zVbf2n9-6gkBb*)1KAW_|s$X|a?6J?S@z*BezC3*(6Io7cT$zZJ?HgHcexqFG zPNvODSHzOZ+1bm%U5B~n;Ro{$nfk6~bUHOaX%W|^W}qVev6Kk?T)?EIx6S#_#(K|i z3G>~pVMN< z7$d*gj*3fEt_t>L`jFijbRcMPe*8C1vDJ))K zWZeWVgsS>E8+Wh8mXRy8zGM@fgRyaUQV4AAh3WP-f$4eQd3}^$tG2e1(z8)M9}wl2 z6|&vMxr%ElTJ`Z=Su*kKQ}olGb-+o+UtmqLs@daG0Eh4A?W!Z7yd}w zAqvXf2(oz#a$?&6R;4HyOT7(6R@}3Ti!AOz$ zmQ;ytCYel}Aaz*)^O<)0TuwfTjE(6pVm^i~pf3qud!+1wmdnEVy*~FliC`$RNvu9x zocPHsq%3TP%nK46NasBvzmD$X1@KaL*Z+agsBGN0^T~) zlU3FWCm0gRu-%@zm{9r1!aF+hL0M`_#UjpHuLb2w6$TKs9Bnz7c14!*iQQgg(25h)%CI<4}xx&Aamr2j@Uu1TG=a zwNCP#d@bK(%U*liIQ6f?+}GQA_U2-o-t1nMA#~}D)zJ20T&l?a z&H$KxrhKGd6Vvb`Nzh+78}c$Jjx@&n^2%X8GJRKQo9;WdU9PWucs;cz8KMG?rL2CZ=DDmhbL0Cf6Yl^)(GXf=d`=bA_k z`_gD?Z0^z*I4=d0|JyV+uT%B!mk-|$s$k1W`p1FHIr!v~hD>Sd&Q6sa#}o-rqoIB| z1YC0MUpjzBobReG2gM(Yv^+}{%`~hLm{16Y zATX*`2$4pPB&zug?z8&}gk~pOc5*C@!-JLKpncL9=}{>1VuUXO@fc|^AimRMN4+Z! zR|jM8x!BzuZqN1Ws1GneSn%`xMtZd__e#f<5r~CUud&^br$yKziudE}VQ|JY!>^+5 zM-rYUQhUU5ZeS4@vrJU!O)&TJT#`2Kec~yZC*y+}h8b#=*J(V?W4Qmr^oa74m%*df z8yQ~DHG&CB9fcYbE&bDT6oDy?bQN9^3_p=Bf{!Y#nJ6zIa?4U8f{j5Jrh9L2sW#`< zI1Ob|nQL@38etXQW{hPJzr%z`mb251?PQ;;a$aiPZpK!n>aC9>bZ_?(_qxmNOr{sTF}K@)*j!t0gO)O$cJ! z)Iu&lkyV%#&_05*RzLYOFiX{DB{sf&+G(rXS)g@-m$+S^&^-5G-6HO2 zqrAK3489VsmWn$|o&_k%w#L|jNO5EhY}SYPFcy1vl;n_*VRkO*a$O#ZKXKx9R!K5B`MvNA`>2pl)lddg%R?SlhNs!fOJC=_p4E?}5aT7Us}fdqRUs zCX6Hg4vSPJi{N(Y>tTVtjKyIt^zcN%m9r&r9q!He9*;4(P!m`tekoq@A*67V z%BMtorN-1Ngg!eS7ynotWf7me1b(T{>43bv0*qZ}Tuj_#%~~2BtF#_KObknNDQ)_I zqQ(i?&&1qtprWTXbIv#fZez9y^Y;jqU05dzOG9Ucfp;uvmzWi#m^q*VvN>gv86>4O zib+hgpT{Cv_8it^?{6voB1^;2lI?Z0k=tdC9A(OpHOV`wU1!vPaub2%&3c|~n_IRv z15=YurWH_(#t{TYgNQRw<%hL^hLnu7D*o|u9ao(jM;kZZ_{A(XL~8DI+iEQn6n8Z` zdL}NQF2;+T4=g%a>DB{|KR>6LhJJ5AhlmyFmKuG27LWN4NpoN2Q-f~)PBqlm{b~tD zDy#7x=A z3=4Nv8%HiS9TW!LJ8IRKUT?X-PbUjxl+S|GE`wwyI%Fs)i+v>?(IQ{{hmN59nhF6a zxO6YI>gwsEPg12cGDwKP1>00n>2ROQ@8+`@EA}z6 zIzP85)J5xOHGb;rWK0u1Kf=05Yd_)u(H3*MVps9@j>Z;0ESS|XtJfb*$kK*saD{p3 zSIc3o$z2R|Fwofk)=Y2=8jKh|@zQ$X2htks+>rm6P*OKeW3ycBw0K6={Y?;7r9b%gH?xXJ)T98vdFtWtkP=p}b zQJ&df)#_~95>UMwiut?OaY|BY?FI;7$>}BvMBYJ$;0XgsKa6=JVtw60@QwRvWpdE~ z)dcv)LFBZGa8Fj(z|g5b;vtM%7il_xt*RGj*q%V$u3n;0)ucxX_-*vK_Cw(D+Tgq} zR@EM-NLjwpnpaQN;J#(eBK-MZ&RWpfIoASrt4*6-1ef|yCQrHt%DfD)#>Myow&r15 z2cIu>#?zU~$9)R&d~28Cu^R`n>gy4uR9c%Shf@Iz>`gSz=64v{-Upsu4;}L`6tYM-Tv@G$WbQINb3iEX_zeItw$>&IfUCM&nmV8HaWe`T@3I0HT zK@V-s3qM+DdS{wRwPuy;7n)T((n|(Nj$MIQD&0-rUkXNR4clh>hPbzb1N*gA2Bqhd z=3Li?@eZkb$u zO~8s1So*^~F6XX}N zjZyiGspppt(!_RVa0(NdRN$S{!Aa}9IpM2cP>maf#;M*P?wDQd*(;j+4}W|+zt6U)>4Akh;Il@`~b!^geFC1MlvjJqNpp6E|GBP)Pi$$ z6#DRR;g>*?RzVKYOGo>bk{eupJ!6{_&$6uO`j^M5Qxdel{hInD3vA`^sE~D~oIK*^ zi$?af=>X5amLUkoF?9Hd)jg63ux#*n?KAbJRUsLr8OnK_(L_e1fjRS)<4*g5OtsUB zs<+E&W||aEQdccKuNLw%lQ|+w7g1LVNQmejg{M75Bbb1zo0OyZ{mXd{; z-1F@S-@F?X!!D{R}NZ6)NDtievFaxO0aO7Pf_qLB=Ls)t{^H{`Al% zBv6*P46Eh6wS&Z6hTuVZHSQF|0%6B~&aYCWX@ujE z8c#@YTu`j84?1$4NHQ!U>&Rz)4NZ}Je8!yhYp`sb4wqpUwHs$kvDOxGgC5bbaK)%&?LEPopp z7hemg5dIi$p8Y6tNk&pr&BD>@QlE_UfNQ>V1A+M0yWFDnxERosnE_#{A*-4PAgAjuyIuq24<6U~DZ z|2#ht8b}G)sZTA9$zQF@c5~0n<j2oy-03wB>ZkPR=igV>!wpVN_zVZXNKaM&8R|<+%brK!1fdbb3REhM=tYu z(oV(+O4sp}$5m&g_k}Tc5iEB}Xe{p0k~*{E>Li0agY;tKV~(~K`UQSyy-#E3%z1*f z9@JT}ha#9|v{LUyYj0rV!kOm<8C0EMa>$uITBsmDhd7PRiD_$6DNzE7Qej+bw_8y> zb@wkdB6n}xmw=Ea_?myjueGg{jEzyF(~HA-`cL$A$Tqw=)VfgHg;1bTo;g~(<;yxLT{6Wbmk6m;r0UlZDbwS3zT zOr!-@=vl=vJ@&rq<1+p|F|Ho$Rar8buR*{`6IHq!Sl8N(ae1m9c0SW?yuR7DbMD7= zT%I8)L*MKjrjjkbkZ=ofYOx&NTuyd~$|kotx?G!tH8u}9R3lnA`_pa#vl9j=3Ft0fOCZN# zE?NA^{X?Yj!?2JY<^@!sbpyASfo&(9dx4}%tOF9lrS?ag6v)7q6+Z79o_w@T5!a<} zM78p9s;2R-gFa2}ff76HihMS;5P0H#1LVn;j;AzKSoQ>=Xebi`T?Cpv!Qm!80+Bqr zt!pTQJ7|Y-N-DIoonOhP)-xk|M0;<~g2%@}2KZdq%d0knbMb=Zg@EYUwrQqB>U`+c zz%}K(=?tUF!Av6lmit+8``!uhsB!7F>K~+9@QAHb|Ect5&BYI3&8!?6lgR>AW&v;N z%P*U<9`YHS-FVXF)4trts$NB)j&BVU$fzW}O>blun>R=Bfq<9~!-;WE|40`$L1rRR zNb4#ZS`lQG#| z6sJ~uionoF?r<8aQVy9w%G5 z9iEm*_1QYR{>)KIcWit3z5bmdsxX*1rpr1g&Sq*X16O6KK%Hzm(R3T^jtbyr%zZNw zc_#|>3xnP5t|c3WJVsH-(lM_eUa2rPY0(X^bC-X(?}5o1jrcsW4u!cJPlfBijdNAQ z#Rv7~xgZ+?%UH20l%gqg;miumiSDWT%q4FQ#~xuH8*|ApCeqeC!&`+rav5MHhc8Nw z{+c|fd8tmYY`%Ak#cXAjcCJb&A-A)f(`zcPTK&BY5MG%LnVStVJTZCsfk;ld*+ioz z>*5oAm2p-v0uqL-Zw#hLRm6J_I|TW>kxkm=5d9Y8cX}b_-r}DT!eDAy51<%VUT5)T z&Id-zn@ecZk?)LmcD0W@Fm8KA!K@_0r9X_B(DgUF!fh>=GAr6@)*oBnISkS&yFy%nz1;91CznHtgaMe_tH3+q?(d zFk6(M_T;sQDj8>k-0!dzXl(%5!>%D3GF{Yrcm%2InX4f4;uVpk31ipBzSfxeCO;jz zy?y1$;j-Hzrj}}KOt#~o88l_v&-Drehv@PF5hTgy)`hSEDM|;uzb6nMxFig1hpu_X znnLq-$y0@`*cLC3&*Z=3lViif#lH0o9D-&(O8#Iz1c-j=lS0A$m z>}i6&`)Ok-!Ekzke~&g>{w0BCXt{jjaa2rKJPn5K@0q-z%k02REG}WbPzP1rdJB%x z1;oHV{j)}tT*1jTk3u#!9B{v-lh_#IeLtny_o!7R4a6@mS**f%_Xm4ewtK&o25A+YFg64)_KnJlFjWuHIFD+Z;3Htj|=)u-+* z-5>Qja+?0k;R?C;k;QV>TK|yx&)KqwqqJSFz4dUwlKoURy$tMALv(cUqwHH_VipmGO!k!ZfKL*{N=r7g1n;_4Ye|apTF1lhs(EiW zy6Mc>xMr=M4E{D!( zvpP|EhPztJp}2c5vnl#U8YV$o4R!tI#v^E0EbLG|k8%>=ibiTb^QB#9Ff-v%Fix~A zgc>bD%ns38KbY39E$Gnf_Yd~+mjLlHKbY1{RN}UY0>8U52;Li3z&?6(3zFmM2Z}^! zwj+M5@I0BZwmYHvtpCn~FX_S;R5T4KC`MK$T4>j@6GM{{ZkTQ3-sIWRK7Y!O zj8kMLT?Mu4{#iX5Nlp|qps{g%@4aM^ea|!>hYgkF*Q8@D+bsoybC5JflR~4KqY%zz5oTBVnu}aFb;WiwLojEO6JGiE3^J$=MN7A@?apj!) zR8pXsejo-DM6eR_cQ1ML=+Se!icRqj57j?{))8#GAZGv_-?Fe$5^;uPa(apt%g&;hQ|I|H zXI@8D2gImnJ{>zQi0>#2iPe33<7veuAX^e zi+8tV&3kWaa@2>l6~z_V3O~`ob=_o%YGJ0q#hy_71WR~m9&NM03VPobmO}eOK(2lA zA>T-9ohIEcvu05hn83#`E1t6$lk|ZY{3Cnr_t!1{LmTTyoQJt?bz^n}N46kkS@lSC z$&AJW$b9oI7_I;0Wj?Q;>MrhD)EZ|G0R;eZC4@*yqi8X!>u7GlUss9{jTVoCesyK# z%OLTIK(dl&?uXW`o8CRPuE?!#VLEN7seEV4HYUXu2oh%5CU_MyRV6%FEqqv=J8$n@ zd!||HvAervS^aTTenjPM?tg{G5g(hp!KKZ)q2_8Ok@ic(Q?08)SK3HnawHG`eac3Jw}OD@1?>T zjNE1kMd;**wV2)epySm`WOr%bUglZf9Tf@mAnVnidRJa0M^-NfXr(T$qOf3Vss_>} zhn6D98OL6HslM&os5MhJi~U0^rL$LPM)wh_OmM%5Xo%#JIu?)S;p0RLvU-tnMN@6c z6zV9i>SC+#l7USKG87kN-M?4NAmiu);uPsNb+3Ras4Adv8j>^kH_B>W zr-h5N%kZ1g)Mk59C-_|Y9jc2sDm{EgVw&gQ2N0~|BXee%!0%JqUOR1(xDnP!!sXsO zARYUPwyzJo^pxTP{dcGz(}vl@ElDMa?%?&H+SRinj4RW;yVabsGyUq5Axa3d=q~#k z640uEw%SobG9U{N8E75qc~8tp^9}Gb7*?pIK3g8Yg_jaL-kh9Yd+y=fd`Y*NGco98 zQbC7#)^ZBC2=yv$Zz?l+Yz0)@{FL26-(SOz?>TT7p^?mn=2$sa*O2ugNx;g$KcBdR z-Dx)}?0_^e8@HU$AbXjNW&9*adWle4C!wC7)tVpNzI%FOqkmGBg>l0YI^{oYJLZ%a zFF(levi2q7kFm(Mu)22Z1ba(TK_^4s>KR^APk4^sinB)S+(nBG>)3U3=Rb5UrJt{xo-tf)#v;TC~2vKI5!hJ_krh@>v$D?uaUQYabd{i;}&YVY>Ray<1ma=m0L^x+1o!T=&8@h?Pf5~3#eW6&I8oxt@MTz6h*c5?S3R$TM0ODIIzt*WmhM;5PV z_xV%XkA_5_)6cz+xkqbScXI3_YT?zn1GK-A;8WjnVA9N5=9-|8bkd+mw$bmJcO&ZZXuh6a2?>ed?Qz7SW@XS_cu{@#82+#E_afQ^|{4 zASpgFu^daxf1+bB!0%skvKa&Zgm*bpMe(#Q<+BkPZQ?rM@dh&Fo7xQ7HO zedHInKI!=M*z?De``QrqPsTpYh#XexJ!PG3dK3MfE+s>p$M^4b7CZ8^e`n9nqCHMF zqXOkUZSeyg9tkUiqKSs@Smo;6b$zFj&1YOxvn~IKHOBqS_#LH}O!?Q$w>-{b(*kMF z=4yWF4rztcjoa2yxo>I&MdHtlNzsY^GaHt*lXzOM&d;^%*62mti?Ozq-Q1;6hqPI)#>L*%yzz zB?VwbxgAp;d@t1reLh02U(21zNAYTm`_|H7-tFKyH;G7Fs?C{uc+%*UAzw5-QUT-b zOG-E`f#h=m*7iDXG0>4oU3h*_t?dqk5R^x6dN!19$>|{8?Dyx2U7xaQ zFcs}s>ne44mNgP3KR>=qIg6ai@Bs*}Jn^Zzlpa3plkf6X(4cvAGK=UBpi5#JZ$_Ye z`nH1xJ*TUw)KdscTi%f$@e;-bQ)an?)@c*_>9pT+-r0J(P)&x$)`qaSmZ0SdFg)&L z3KA0vb(JbQc}UqK9i?e#Bd^d@eR@QX@j01V94}>Br%AIGzAqr*FwqFn-YJj)!m8RK z-ap~C;}FF(eZ0iKD+SYVUr#ac2>iQbCQVvaG)NLzVK9U&!{%WZY}4{KDE$u5RwS37 z)N%7c*q++Y`C~>2FKy{X@lQ>zw6khBx3mZq5Wc25ZK14(6L^!^*5~goytM%RiJ*?$ zJ`1s_^;x+VL(;i@ZsTH=E0h2^K*ql_bSmP%0sRihI1&X3&8yHDEUJAXvH;+(R{fk1KoTelL9TNWlV@fBhvVbJBNy=R2_{CA{tR92GrF zrR@=8*YD2_X^hp`J%#LmnylLvh@fE0z;_ zZG?XgyGzscceG!bs^n3V*I@XG;@39Ft?D-Q2WM8K4yU9vpt+A>u5p|MUuLHT+bqEX zOazrN`dGf*MVTJoAQ$|c#+}EWKrej*eR$_=(jSk!JocpO^&@-mNOV~~oC`btz3$xC zL&Ix@B<3Mc2ULBw>ELT;SKM7^`=*c$LhsGHF5Ktqx0{x*t~efZ%-j zD_p~1#(f#3BM5#%=9>j`iqbAW_%Z%Y8eGj5Uh+1b8}(p?nshucyZn3&dWK*TsdSvA zs~^M1+jg-DGqzi=A8s?rcp3Rw);900i8TnA6Jc?S`pM4S3I4Fa2M-ZUgi^yRpe&ma zTkJ8!+friHi@&}~#BbizC=g}sK!$M~F&O1~rWxTw>W+!2mY})bb(^hNXPx20g#dw? zbKTHLQ^GL3>r3EF;wU4Y9_+Op#3W(`{kQvSg1cI35(C`IY^#+>S|f=dnB|kOZ?F8F zuxZ$9z@G4j=uyUMGHE<29*4!AKt<)^wU^V^0a4iB4ONijW=g?sjqqz4AhzbtHIX0% zJZcZL6UeQSSoiPR*p`qidu%`TMaLX30pvg2M7Tq6`+~ZC1FyMq?iR9)x}6<`YG(S) zE7BBL?UL`RuW+xqE|C5B;=Ou}Cr~nFiLBC6vME$veG+^$Bk^}iMfhjM4xj0CYN$ZM zSkAO^URh~usVwr#5v#e!WU!BLF!HFZlHo3p-owM z&eOMH_or$*!F6fE5VgepT9JATNvGM7f={CTi3E+`H!&^37O*dlngL3*2}!k1cRo65 zHe#Jo=1tBY$@m&B{JyY92V{=;W0NV@B}W6ypsO&l8x1k@YYH%E&(pLncSgW|u)>i| zqLd<~tK|rvLE&`aYx(XH&B4JysPucYz}#W4Lrh(3951#+;Qi8gx}(ihs)L3RM2&X0 zV;;@L8w2Hiqdx?~k5~75EBQCg;VIZ8*Pwf$m*IoKF<2gJK6}6TwzRLGYy~ar2ITuC|3Q1ZV)pSeRTzS0_cY#+Xcqex z5xHKHRjEw8TlV{sY2l+PrL(e&b)lm5PbGt|F{76tF` zK56FHKV}VD-yz(d-I11(I+&-akudhp+oEpd;zo`;@R|sxGeYxZ_ zQz`Sq-PFhTY61Fpyn20Xf;Sg%_oOC;FX2(yUSPHr8x=ykt!TuO1ag5H0Fi}JoSEEE}b$D@J;?{~z;zZ?~dr{ZPKwOWUy ze4$>NU)hWOfvNg_&YLkSc+eb0t=CJ*rFc)L-Y9d`v*z)E!&J&7exG#bo_fnpg5Hv& zPMV3E2dp+)jtlCWpTI*ou6nH`9^ErL@O@)8TUg-62B8XQ(tw#4TbwOjKVPJ?nv}im*IIM$>)K~ zdgBzgHKZz+xE~hYtZM{YPUG`B0iA;s0}E+hNZcg7wcVgwV7TF&mK?(4|N5Q z)E_HD0hjWQg1HY&9K4FX#y)OOCUEy&E+EJB(DC^ZO8_i3Jl!N?X<}{z&2_z}?DX7b z1{ongxXMjNaU>84-yM5UeeHtzOTxKhR2I@8Ej!{PVnaiI2lNo(ZC&CEc#m4=G+yEK zDbM}E7QrvtS8%gMv5!eNEY;9{>esZB25y>QkLX4(y^ZI zft~y?REX0aqoO9Mcq<+bW-+aiov7<;03Q}9MhUBsdako(3lq4~>=`YnXye;ny70uC zh_a{SWx@CA8tY4mjQ34HpLq>w6f6tY+O=}NxK%&d_**J~9uBdPv9J>|IpP<}-sOR@ z_8His&9$b0nTH1Q$K6RdtqvN6UhH(z9_}|>O0`SA3CJEM21=Y}(f!u;p(?Y)qzX9O zELqjMxc)Lu948grpMez^7rnT2T0(X(q;bOh_%R&ySYn4g^|n`+-&FL# z*`@1|p#z2}oaLXA&Qo)+(zSQn9;n(K>q*3fq{GDz3$@eG3j7Eb8JN{sPg3U*4U+&t zMkXy}Bk1aSSWLf}V!r0-Q+z?S0^J1FcONi;LFBL1a7U(HVI20n(wS1FOET|D=WDxE z6z?8r8HOuD$|`TZ?{se@rpm>$)&YS!5#PLau5|(j8^qjEeL&7of?X05jn=WB`&#oy z@V&w6;k#G`cXY@XXmUggE0%*}QS)>S&J!L5Yd_ z`DQio>H2qU(ZR}Tj=l6EfBd5CtH9LiF_oK!G?lPpWn9BSpG|0UJ(`tp7PrJ<@z~DHG!!2U=C;yYwI~^M>Xps zTM!RQ=D>~FXoEH~D_f~GH#1vgf8<_d*jxho?YMjEBA>Bm6cGeDh7^kHOPYn5@s+1qJ86o%~zttujqL2O9&;GyWWdMXB+zAqT4Gg4f; z>d2u@w8gpMqa@qHImC4^|0Hf^ei2o+{6&TlmqP?nA-fq_c9$PWVP;R*V52bD=&wLA zRMPEJe}+(JG#}IMzD29TY z?<`b+1lc`svb`A@fPBNRNInDIu1734ne-9)_6mt^OHG~@_aYp;6fuZ|F?l5WWTmLY zZ_-c&pwt4o>-cynNJXcJ11b+&ImGR>oPLs9G?e|_<6+00`=-?Rpnvu5%}u5)3|wwL zUhXh|e^%7~AZ%{h(pcC9wS(%q&TTVQ*1*ZY=XIHdjpzs6-}HJHQJXWkH0#~Qau!-z z_orL4qHt?W@_`sIh#WbXQt5u?ey zJ3$=HVqMLCS{;7fioC+e+z7sH_>;fhP-H^*?^M)xk{Nr(5vK5ZXFXWPfo<)1ii2BN zVO+YtL~*K2L=65BIjDF2~8=5J+B0n&7rM*@XO&Iq4fP=B~06wL?NZMSN*-u zXjV%ZFFxXLXG!L%BS4fuEpfMK3QQ~bPg!)DH{HHAr$#-C;~VmmXW3epLZ;AEqGT=g zkk2)Nh(d**aLB-4V~PkD!3m4HLJHWL6iJNJBz_U71IJ$aDrE2J}^`lDjAW;q!=>wxcf1?z2ZsboJ3rNo;_Z7Es;*5~ZD|X>CX8xHcXEv1? zcJKOG$daGS+OEe#<)0sF!l(3!&h3Yda`Ju^YY-bN>-$`NPJRhL*=3tqX@*ch*U>kK zUsfD!n(w38X|l7dFo}*HNGmrthhB3+cd{=KW8Ez@oX^9Zm99i$%#KiYivane)WmB2 zqVFWQLlIc%l%ye~2o?7w9HU`)Mq`df;rNU4>wCIRk_g){3KT9BAT+t$z1iS`2RJAg ztAJMz*396ZxLoR!YrN)PTo(=?1z(S&NOpYkosrouNMrI`6-+IZ6gioJ3kor0r!phX(f5T+f@5-LUp#Kq zHVHS2fe@mk+bAdZem1F!wPgR|S@Gg*PNT+;C8;IK=seL)->4FuIc(-t1eBD0!Ss<2 z8SFVvFE^eFAyv#H=ZiLXOV!gCM-w!Bs$+UqW27^nQY2_qJPw%6x7FU%!!7ot48i>( z*^hRaV!{ba2|qIRIui0ipYWw!N|l9`=!;DH$r*Da|A z?xYG^C@%xv%9zY*r?Hw%5C^5mTAe9( z?e+UY+$Tl11Uu|L0cx9FxV#}E_o=Pm&=2k zZY%$0xa3aBwWQ;4e|%$=5RE|P0p8kq)cqudV%g zcJ+~+Yuku_yFmX5dYoeMNt*P6ka+pcwS9UZlZoVh$mwdl>V2hO0C$_GA=0=S~@voldy@yBj z%&vVRp}R45p)wU_x`(jR6@d>!kx%H!aRz+Q81lw@_ZKp$MwBE=l^#Kn0MCV#ofARL zQafr)kAI_IAfs*@XUWLm-k~D!udMU+NpXx^aS>8#S@>neO$myW2cyE}hju@9dU`e0 z+b-I(xwY7w-ZV&t9(_4L}FY*vZ6h3B@pC1^9UdvHl7#z-DQg#KS23G49z}SZa%TYI z)yc1TuR(N%=OJ)HpR4JlRB{wIA+?Zi$kb-E(>SMc`bQdkf{inn)3QnmSTYH5$+Mgh zNh~?HLUBwU`UkHkW-spT^VOry1Nl{_8BdMTb?fiZiC3aOR4En#oVqn}JbBC^Q#qV%f_ zX>k@Gy?%1k)ttw;GLbl1*HWAEeaOCcbZuH}Ht9CsEZ;Zm*{f?ShS8V^RHvf zeMD~0P@_qhu&&!B5blRBV<(ZpKF>c(*3)F!CB23S6VV5XFAY&)OOyLyqNB{|L;3Jo z*j3hni_9Ldj!Q~H6C8L!sCBpH-ijFuEPq@Bbu7RkHZLcHSO9nr?144H7zyevS zIEHYUDOJKAA=Gj}rG8}oj(ey(nC=f*oLb*RJp1G^oi-37h8O|*-E&1gu_^?0TKg+C zyFFg5szKY1J75!KOwV3W-^B$pjeq*b^fNLE)O{L#Wy*zq$VW~U-xhYL*-x3FeX9Nx z>zwMZ6>+3py>GXh=li@7~#?6t?o{ENX?GQt>W2$c%GeV z#!ew{Z#l6k#5yTt;e1dYE`3O&IL~mZfqh8@`n7xEFkk-we8y94T;J ziN4wl3nDaNFg|>G!VM6F9rG0}GrL#4@Opeyc6}{r`MSJ_(Ie7-lT5*1{FAv6(MfEY66|;AzFggv@ z#MM(*$60b8`V$1M&UjKO_=o;kx(0{3!fBF^h%v2j-05H84Vf5E$U2?sIn9ki@`XoW z5;7kFo9PZMbCgXkk|vwHtR=1Q|IJ+fhh!(;6YK{bF#W7s57>@seU!?@!&$7@ZxV^* z1d2SNAL{Z7QY%M-rxu+)JdNo6AYVX}W~h?ph`gptTZErOEq1Gx~J-@?u ziKnC%Z#W^p!(l}V?d&&O+gh<*aPy|f;m1vAK{oviXPw5|55~jWTX9uhUx=rhg>i|Q zJzJ;tWZXD?1aW{oHeS6m(sTCLjM;rW_F8|mjAQnRy1A%7fFBy0P{nqJK z7l`Dly!xv`1@JUgGE(4mvjo5z1Xg=qQJ(=g{h+u^j?jZZf#U?%mP0S9`j_V^g4<)`=Sye6puF_dbRVojDV~I zQ@InzUYGvvy(;*;rwU*rOSLtyZ0`>tsM@bqoZG50$y<(3)DDkE5B_Fd^gaB%j(!a_ zYv(F-u2}6j%l4UNI$z{iCD66+n``-sowls7)@i&u<*?23$2Y_VLq6@#q7N1cdB3^^ zdR+a0sm~;EhNC`x9k+bol zpendRXPRh^D_P4*A$l0d*3zvvJyr&>?Udg`Wad-YUq;-j?$kXOUd!(w_NYrGQzJSC zcyZ%wsj~V7ZEGyhOQYX|z4DlHG#>od#?&;W?x1Qa|F;~!HeEvmwK}4Wq`1OUS9S5} zg7Mpd%jZ{3(@wg=7}gVmJV-KPzR7pcJ+vGd_FNS++|kR^mU_YXjxb+L;^u@QBOi?9 z!F-ed5-Ac-Nyv?UiS+kpJ@mmbR@@VQsD%||3dhB}1=a0l|E`JQD}STCauUzy6V5`~D*c)J(Y z1#)mtn@v{(RwLt0S$O<$ZCSiO#H-oB6@8YPOLR9sjJ`tc_mH#CT*R9ywq-?TnR^5q z?bISNe$JUH5N}q~{`*wsecppmv>AtaQ?oD*)oS$jT)Xtj0%MW$+Wb3+fHJ3nMvQ;C>L{D_h0zNy$-NSDAvG|S)@|etFte=>$Ni!I%FZe7W?2n{#=37 zP$p=A5!1#lf&yTM^+wN^T9!QNqz=c;lq!#?aBdfT)lFz7+OkU?$K3bJ0wX zw+wri}E%>sXmecj)KV zMR%%0$~1~8yJv90;J`{V1v=mjVX`byDw z;$CMV=d=f^R8}tYELq~(D-HB9I3au}6z#Y5?GTG714~Cl5*ki7Lj?Be6EaTF281b| z^Geb3klr7xor{~$yQ|lra~uT^mK|_=Ih?wL%4lx&6g+;!%s()~n1a->IMxI;);teD zZ?;EAAdkU?1&ESO&dlju-}H_r`INi4bu}AaW^6UaR$bzzN=sw{fv07om1JQ*V~zS7 zN@q7Ov5S4Vy%q!sWs?amy8GHZyZIpO?_wF8bZ;|H-7*YI8Xwye6}&F85~Q^uQjqTS zpmr9CsPwG`J*IT=zLCWm;L;9H?~ zL0A-BH_yr-ghMCmbds%_UZ-xjyH+h-r9S-@@%Ou%E@{?@+#EzJXRP%~WdtFAjcWZm z-y1^c;VO_g@XQb&Uz(*GP1Pp%oNr-rVr1fACOR%F&}`w~QY{op*TD|Gkb3*N8#d`J zJBNw0#pYMJWGaP)k@*rmsM`)54=~m(a^qE#V_tfwtGI4z4P^C^PN_o6-=vSPeZl1dZ1y` zCMrEM6X1G=ETrKAtvB$ttgq1ZMxc?Gnoee~g;{5W*3c}B@xtPz>=01%nd zB%1hgnG~@X*dW5&sACzqfD%+^YXb5F&?BL0UNtRGKm`PL-D=`{jV6h5ccJG~8iGXv z56%WX6oF4iXFZf`70F4JkNFp+%m9zYg!Yf3vSWb2+91bxug4?lP*V-n@H*)~UjHh! zKaXIEmefTzk4&d7$j4g6e-fMoJEGeYKKJ2sJ)V&5+uuQPfE(~Exg0S{cV6#kL9!UR z7R&eSog>94o*hITMi!~2dzwU76Bdtg(=$xgDKWTuYx=tf^+oQKCs1ZH@!rJjlay-H zd^(&YSdO_83R_zA%n2U|#ICwr`J2Q~$h7hJ)gr!n-uXpStz=G}HQSr1Y_Vrj!Uqub zy>DJHFlSefcVcsp+dk!sA!1_GmiZ7+4EdZx_1#?<^3E+DLn9LB&vZYR>P0I4R3ZH!EZy<h$f|{EK?SWu&4gO7 zJ)HRu#md#p@=lEqg!A*OWQ*~t%S_>53U5@9)+^4U);I7>F7uSsPdf6tx4NH~FqRg4L zb{2FUKP#~B9g=!FksGi)Y>RDZuy{(Ky?bn@k!ikuP^I~U*H7=?^qtr@)=9rh%~mZj z0!qt4Qu5O2N*+bd0b0A|Zgx56?6RyV5?9w58f8m=6t%|7M_Q)tk^cSi(?VH% zTj9UiM7ouWwgz1`aLEuiUeJ#aVL|78SA?~et@vM*hxOoRPyFijfMnH>JLj8Zt6%95 zwpk6H4JLx^%O$XW9rTC5x3B@Y0ekoRCbZxzxh1GAxzKXD#JKgi#2s1Xa{hXI8O!Eg zPBVD;;h8}2F+NryN70L&s2Mp1hf5vzkvhzj&zpgAAS^r@w%;-&W?}h2RWf!oa&$@u z=}Ld!5K&NB2`=OmztrP&lbEek?BD0}=4+LEe>2JTK87rxrj=49SiD)qlJh`cuah3} z_|Vz#%kO4=Co@sTJf3MMEf6$af(q96ae|!ZV+UHLG=jd0Qe16~&7q&pa7))+{LyF& zZ2^)R9;t|(K9_^k>j1xk^{9Hn>FRG<9hBH6(&d~(QO`UDa6G4qz;|USqLdwMX zeTGLDI%2#$h0bBgc7G`f>p92a{;BeKo!$&Oy|HByO;Y#bLhwK40?L#_v3t`Q2$r^3@z40k$gykmhl-;;=L zGh8a&8ccG-T(}W454A0PE>Fh_8L0SKBDLWo3<+U7j;b!i!mpjWX{M$`#IXyR7E>31 z;G(4Aucu|kIpY!TWTuz2ZtPifg)eHcQY7C?1jdKMe)I0*xo)jI3f7jF9)uwJ*O~GU zv+C1T$9zR#{VZ)EZT;vL-a1IuL~#?KtJb(<+YhJ%AtG5E`Kp|Y;(>E!Te8~^GUiRf zopimG2+dqec(BB19o*>Et!X`MM984hEO&-=MjZYcC+DggAU0UU56+O9&u1$*d+k8SgbKgx|v! zDVjs^)eDHG_9Zv^BlZqp)!e7mN>eU;Xb&aCIFl+#c{Q<`^&J+%yM8m+zaId5)d2Wxfd^|WNJ@ya z@6#zPGKg3V_gaBn$;R>?@d+dP5P%sB_OQW;I2}kv8d=6x)oB>85VxTOdNmYo4|LcoO`%-vV=c`6C%3tGw~IhF(<< zki;r1@pKY-TAnBWu5i;Z!*8W%obB|7jk{PGiV5f`nt#u&QN`e255ZvFr@H&2-h6(# zII3~l&MRbrXT{-D0hb+5!x#>(!7;E)gi2no`ZkrVxp01Ax77ml>$haa2P;BqJsDrp zQ3s-kq3u472@3W_jr%}l7O|cf3zug^2?=h!e_kpjMcanNHAFcA=vweixrYR!z{Toh zJ5~;rYarPyJXFe;=*4%r6SdYuj6cDFlw=!%fPXUV@RXB-M^!6dRlF=jBK7 zxG&qR`2CJbgPiqelvg+GfGcyRydL;FC8uHgWrFcOB9Vu2P41VT$TQmMp;wmDvSn}@^luL&V#2dl-BRt0dtsE^d+P|Q;BC2H7~DH` z2j?=rUF2>MO+x=3PKcwQRFpFP#7JCs48!8bUeSKH95Uhlmak);VccDfF_%r4eyog_ z9#Zh(SDA0xU&ICC3d|E+TM;!rfg~fr7&X2beDNGCngiFR&Tn}fyFZdduV!i=y#ju} z!e6i)(1b(3Cklkvt=?S2>j?umjt=c&(I!-f0ov) zR2j_Oq(nbACXgaJ3@Y>;*U#kr*GeosdTT^RGxru2^vAnHH zs+(RAS{|w%?9d2XYNTgldWe;`?Nkwn59&`=iG0v$_hHZPPzXxOmw2AqrfYiXsM6>O zpAWP(mW?3h-87IX<<3d#DI)-L9Dn_I;mE~A(M{hPcEic=rQ_pO@9IotjQbAHTL%VVFH>Q-3x;y*j5r&0Stg)na+rn=(Bf* zW?@--PqqKZaL2}4@Ob!xXm7IJVZ@n*Gj-1~!sql-KTQwLA~MHDLQ?z|fbh_|U0!r$ zwlEM$xpO@Q7cLNe&Jn5GYDAE9+S?N20;zPEb;XWHsM1ctoSMGv|OqpVG0O* z%GgfK`mi@ZIh=*~S$Vf$-Vug+^|vsheZ+h6<9^J>dR9&hET!{}nNRk$8Z`n%;j>tB zwP~lkr6`od{bC9(!m~Nvh`73(*&JGEkh~V`L!0{4C0jhdk!ZpYE6>QS6{tAj)|>Ysy#_ zs^qQC(oB;z6!ctih9YPQ$c?Q<`#nEM6{JzFU4gt(bgxX^WTXqx7=f|ars75iaSZYFNF|%kr>fckE!r97B`^6qZ*Jk)+8zNB*<_rq*dQC;D3v$_s+3&Rl+I0 z9E`-B>eth(TLmJ=)$G8B7eVMbNZy{=Doc-|5-x)P`_-Rtg8_8h_`(yDt?J)eIE|sP zg?8G+IG?zIJW=-B4{Dn>x14N0x%>mU9KGa-nC$=&QZ7&yZ|>0tV#=DQcRQOQ%nmAV zpMPe0?jkU!Ur+}D6O60V_dxWz$g#1z(n=&8)=UxdEzb7TRnfeI6r!XEh>-8Aj|ALq zlAh}w8vdy?H$8>!D8C0EilkTJU@aau=iH$=#Y&(=>6UMYVx%RKBcuVZLyU~3h15L? zSex%><0&v)Nb*!vXm=oPC8L_ms9uXu;0T}{Dol(B_lwil|=hJ*vE7lJNRoM5tgcPyi z`@3h&PFJ7nZwWe-iYsOCO{guK-NrGbOPE8!jNrZ`w`ZAM5gwkc42snXH}?Z`GR=q9 zV_g3BknFM0kBlq+KG}le0hneO#cHP9`qW`8yx3QD!EHyx?%RE~^LOY_V%Bu4G|}Fb z+T7WJ#9hBhi6uhqLn}<(FwO)+phENNQc$!Q^7Cr)q3y z^%Ygp4H{hC64H@l>ENQsdauG9BJel(D>f&mtl9}91eU4L5)#)Sg=sarV@VSeW@+HT z?S1*{$M2T?c3`p@%X;&L_Q*Q>Yt}XG@>;WIm1bqmI}lD)O09B;n#08E5*~WrrF7$^ zKO^G}#!--=O$cR4>_;H8Lp6Z6*V%(;P zerud%#oF9UrH@F{eF34B!aBW1>Y*F>G>3fUK1R5Qv>urAg6Y*N20ycInY%(kw;J3fed*|86`7xx2>7? zNoytr23@qEvpRGxpfd3s57ORG4Nan(ZrjD(!zm)?TBV%Us*iZ8V58CKW8a7!U%ImO{_AfyWePDDw4A&!EiB?UeU(aob^P@1bZQ~d=$3C)g}9}; z=0cP$T4LSGe^Miy{8pmnwI}(QBjd~W5=cGQo|W0d%Y49XJ<^i=RB;F1QxakM+9;pu z%L*Gu?9xPhRN-Slv>rZ92IOOj7*Hw`d-i>H?0yklJ%z4?xC8c9Wo&Gi^BXP8GWQmT ziFc;%Zc3l##b?mz;Hpz5FpX}5q;Ww$qsP2W-g}q-2}K6$+zVPnYh&m86s6*7!NGf} zgDGp+=&zh0Gq_PC!=d6(vUmgGTyjN8Aiqj{qwn?TV6Xj_tOHo&V7j!%v}Pw}7a(6S z?0jk_xOo}Y9`K7p#}sBJtnvM{mG4k_A13_--s2_1M%kyz7$j94ILdu$Fa!&^W>V$& z1ZmQz2>PpqrW(~bqV!zN8a26sO;p=Lh{G^`Xb?t=KI213=bPS=U;D$Un`CL>Ott@Lz~@JrJ}oMso4QKn_LHdjMH z@j`LOxXFdz#Mfb}nhow#olb<9nG5>ur^I_kG+7A!{s=vY(`UiZq*{>p{lqSEB@u`z zMl(LgS3bU&;$ZH{{eH7G~_;gcCstmo&e4Snv16YWk(<%34tB%L)O zNz-aCkA@?)nbyNZ$P`1hSx;KccZOd$#@N0*z3~?rn>20x?|Rbp1Z_s1%D3yWg$FxN zd!;pm`{Dbm<@Ex`+2wG*BY)Bd2|d%kj7zI_ys$Cc&FqaK{}@H-QJXY-KW*$^#L7>5 z@*`o<)kgZHrA9oHvaqi(>4!I*iVZ4XdhQNbA5J5cg8p3C=e3@14Qo|5!B*0$pgdS zcZ$w+T`=BV6U1{H?H$Pqv6W z!=l2x&!Bm1FQA)j|IH%%7SplUL#hVC;8H1!2e!6za-V@z=)rudYg@Y8XW2<7vE0K# zw^{v-(PW4O^UD5UZ2l#VRs+X+{xofVUYOi0PRf1TQnB--m90&_+E6ivFCS5nu_DC{ zcxkcu{^{{i<0&&y%1`ewTVCg^d$UDRd`}53)c>kjwf34yEiM)=*`VoT#^6AUmDXwN z2vZKzUuZVFQDF(K5j^H`%;E@b49L7|c6DV%`L%Gkwg;rC01B4&!;6^V-E0l93^{E8emxc5Y-1x%=lIm0A{kq-B8 zJb|YOSj&_u#Xd~3BM%KNJdS2d`!NjCcJ%uDqSuwLgW;8AJo)zxDp%F@bPqeFLsLaD zTw-=R}riwXW= zW+YoWghjPNkH0qf$w`o47=(xSk5MU+*wXOma3%fwvV>AOTf;q#<1Hadennvs5!Qi` z-dlr!Zg$rTth8QO-Lv64M`;TN3i}(3`0Ex=c(*3Ejs*A=SGth1Hy2F8w7XPosp2MV z*(C*ApO{TDq#mS9!zZ;#$%Sm1#*+ONZYURHe0sjDcbvpfW7IY}KINL4pjdczr=U9h z8d;50*-k?fgvE0wu4u9O?g*sN3!5%d@^<$u8#MusXm$?AUI|FRdq`c^c-I}$5V=#h@ zdb1fxUmh{6#4y5>1yqvm4Cn|KnHsMzQ}@7+ICr#NxNXEP_2X;A)c8OqkfkCym@OaU zE2HtXKWV0+y`@+BuzMJAq>Hye4;v_{gOzL{%M^;STTkPD2zQMFn$iF#`i@$dNiNv%mn+DMsiMb+Uy6UFI4p=R!gJ`q+Ik5x$^zKa1eBY zO`HpB3H_itciD80oz1YeD)N(XaxOdrM1L0b>!?qwQiIdt=@lsH{eiU|KolG#vR$N7 z*rV5?p^2SlEhF_IMZ@ux?Y!rlbwjUv`ysy)-dCAfQ5L)Cte&Vp5uf@n`wGDT=>G3q zXjS;c$erauO2SPNoq7@N&sD#1B1!5;zV>OU2$^z-5Q}ql(A^%fjp2k{HCvP~az9rz z%JJZ;pESH1An=0%o}R3}@ILKbt}7XDp-HW>63^pH1j}y61_xYrb9eXb+ULp8n8MC4Vj6^*+LH?dW=ngo#Q}RftU|L-39(KQ1ZOp9iEwq| z^Jpderigc+urs8(fs$Xbt^S}H{gL06=F{CJ_A(~DU)0%0HU?#nBZV*WgQ#)wat}JR zCHw1K&M8ycr_hq6@c8QwRV(|^)I>q=U9^^$`gy2EM#JjMxuprYG2j@J-O)8@OXrzr z%%vRiy*$zyjp3-!fPWNkD3D)DGNN)2uDUMn5Ms98SGLKGHDId&JE=d=@wjxbu zseJj}Z6~Wq8?u`zBrZdgSkEe#@DO10g^E*UFoF!(>&ikvGDD8O8M$&JTRW+7OnmfA z5&ycdq)l}`JY2qlXB+Pl^%uvzMN0s2&uKtQP3TA0K^GN%!`$fU>`#!ns?SZcpsq*S z%bMb>=xACsS`u*B2}ykoWfXXYq(5biVy~rRm?V3;D_g<2P_s$;smra0FZ^Z;_c2lG zm($4}i|zpC`Nb22AgsT0OCC?Puq6Z4jf)g1LOVOFp45kou5{0F(p&dkv}9M@p(PAW zDowg%QZ!n&&8I1&tB3Qq&`UpWpINKnC2?)6y;{X_npCm?GuL!>7Z=1HiKfp8xvzIf zxL@v$faqr0#ooejwCh>xz(=BB-a7Jk18nLehR)4q(GKhf&-RNB(%|C}QLlov+`56J zg%WduQL-kpdYqrw>pYUO@(YK{vW|IGIN>;=K-{TbLX@OnGUJ_Gj0_Un2q=ElPNCEl zgzvHHf=!YAJB|cgBOBSP1T!o0BIld-uh4vOZnYdSkOAv1?B`5d0%ZLmZ=-0bROK(I zJZapoA_GiVN5hr62qn$P*xN3SQlhXYkkZKIKc&Q)pMCQhJiWx)M-?ePQyTk#ro3|A z#U8}TLCu!n%1Vd?vzy?K$F1RY7AG<=}%a$`j+if zL?Gprh-_Y5BTUR#X;nVhwsI09Cf=6yLyfKU*foWIk0rF`XpICAAw*RaPz!uY2mD0P zzh@w5xSP+zL`I1CHc}oDq$;^CB(&Xdmo+hXy7s|Jva+vy_nAiDB-o1)I(yD~buGjZ zJ5Zi+b`KS(t=7au%}jK>1Kc9knI3~!GH*kmePg6mh2eVKRRf%ugn$Do32OuTlVT&# z7C3LD<=gS8DHEyaA?kcyk$H#58C>f@x_MsRk^g zc%wZ6cVAx6D}p8z#sw>uqW zRe1yc*jljpDVNgk3VahKQ^lC(4sCq=U@_o2C}CKJVo>{#JhH-T*6^c6!u8EaIflKB zX+wvHMy(u?>V9?}PtHq~lq#Ww2hyUo9vGFcl*&$`LJ*?;ibknDrv zTT2#y2ZMTdE3|=|lDVny26<_qB4tKPrW_}fwWL)>>1Xw8T!Q=i6#_}T(_vzN4p;@g zfbdNFJX`+|;64z!XSQ#V(a$~<49Hh7mj=79BepnJ{@z4h8}?DKNA{iC08OvET3TG% zqR2mocsQVa#xcRR`L5fgI=2oMGwJ1Pt~MuYITMTe(h@cR-D{q(a7875Yc&(3ZoH<@ z#p%Mw64zZqR^GGF^PbM$4P4)#>K%p0gDV z-ahM{!tNcM9W!)wirdqe$vw7^=gV}iyUyF_fqTw%G4G2N6=v_&T?m_I8Q&{MTqnp9 zw5vugW1Z;qMyJeZ$bq)SI`1onBQ%-iYg!xoj37g+$a+NwT~chP9E5_y+ntMV#4f|+ zC!2Q?TbcyM+B9HmXia_N@mvu2GRfK&*@1W-4~x=DvCx8$UHv;tt$QNG#v>r*rDpeM zxJ(_Fx{#QseK{^naZfu3a)PZMd}<|pFp3e*>k+j2G!D`@;LC$wA|Dkg&3AC1=TyO1 zPT~*U8ABIUv3Ai7IYtnlj%PiZzqW;JzJgCF@U>@m3ffJDGh+xB6y=LVj(`_|ot-(s zmy(%ED@9m}*|(6V9^19tI7iEPfg#M+$<8q>sH}y_C}|dx=I!GyEo%UjAucMipsWe9 zYVYeCbz`;LC|x{)UG3GMSILLupz_4W1z{P+2G3OLJ&cEP5EZhQ|`76o309&?mw zn2NMjbD>4PD6Zi5x_Vm@?-;Xc-P2;X;s2FYA3iIR;CZ2YBCGwZ1oDfy001t-En{+kDC#-{`G{XVBvXB8@qOy5*y{wFmaHD?!R&XW{+* zEnW>GGWBKjGWrNOD;c=9Yl6S?dVOc(sX+l0ktIhzNKAE;e zWTW|FSYgfY<&X+_0GekM#wOf<`-c%ykB&^mYITAv8`6{cFBms~H)3t;_#l*NG2uuy znUjd*GN0glmtlvS7su&?L`PR~YE?JV_$~UmYi@V4{ld&eA#^dHD%PNXztj>Lu8coEW{lj2J^SpB9K8n_rb6fGF|i5+C~j5j93o7RN@!;mgO={ zf#wpsdbVAa&z8}>#g|!0X=2TYmD0mZr+nDQ5X97#Y_<|;x1*i6-!>aBVY%;){TWGK zl47h9d>SRGis;Cc227fXW9A0Clh=!k zh*iE~B0F%I^3=kEIyvsB^wlWmrEnB+k0+qrpiTkJLX9#gN-v#)85%DQ01VOrlM&eAG3`K$53i*?<*xj5e~lR3Ok&UU|ZawZS7 zd=$x+xE}vQb92bV$XZwYL(cH>L3nWw@rAl6g0iJZ#R{t6 z(58^90G8v#^%ASRBz`H0hxp-UWG9%S$|aeG0^;Ed=|qnwFSO^tB*@EA+D8>|HN!=dQv!l``rX{)8$qWd<&?qEJ?=q{~7A<Fs+;zn*ZTOc{|i_dxrkk`=e;%#Cqg zw&uL85-i+CP&4sR#fw^y5I)|lS}PW-xWT#DZvwvS)M%YM6B=7{G34!F3xH)k5ZH>j%4UXp%ofVP9a*j=W(TUc5F8}5r^r!D8s@@TP1;p*S8?>$vXcVxN5&xP3#_={?AI>AP z=$Q84PTLwi@5m(?L{Rqcu+x%mX=#YExMT!%6EEK9_E_JU!AuX zT$xRbmtKfv*e2sEwJm0@2~(q;)roa7rD9lP4y2VlrUdKV4UmyMSupz>;;noPW8(r= zv*1Oul35xnys$T9xsoq`ujkOl0q|33a>8qDjV6j0_6LUqwcT6_a=s$EFIDm9YN#=i%ae}cAnwJ<&Q;&^ zWDJ^qtWZ;M4O34(d1sK>OZwt!31me;sme+LLE~wt=$OO=h;@Z3o<&D+sD0-#QtN!6Vz5_j}>o%-|tn^2`Uq z(l3ubWt8WRhw@@^&lnG%`cfCrvtUjVMS$XPlNLA)BQ69DSVTo{*so)Yj)e1a zD=Q`4NzT#Gy+3CGGhD)Nk+E(+cC^e~Qo1RI`S2p_Z&m53Cf4WZJvp*CUc62E_{Vgm z*ws2@e14)()>=pC2mJp z^(*qEX$QS^e1~dH_(_EwO6+#1clF{1_48ZB@W6PECuxl|m*yVHe@?Aq+SoqYbyr-{ z-rFPNG#7~nP%W2!foH0sYm1O-15Y0CDZsgsU{ zX|ubFFOu@Tswb7-=}(Xdoc{Mxk zVxA!%$RsN1RuDPX=@`2`v5i1XJ^R|#q|K6*H$;dWCifl%l=B+%!oOv)UytUW<^A$b zQnI8kpoMS19CE3qCU!AkSaLUtc91x=Lq@iwGq(weC}#8om5JI7NyoHmB#Jhf+earP zSNn$Ma=+{FWV6`RQI#pb-|qIpzZ^$5sjkVlHqck{Ieu7V~jXT!4|1PvE~{3Xz40uza}AGS^dES z5g3oeKV77X4Cg2)XPre==VhU&{C6}m9|$4nv*GiVXv=V$(l;&VxSBc(;T2T=JfV%F zSb9}ZrbA~;L)a}`-Z7lKU3}vWz76CcXp2iw9|=Y>vyq$|2u9k^*+b<|j9KXYR$P4> zyAeKaQQo-rk$8&Zxn(Q87N_L;{Qx1ecZ5$FT1)qdT1dzQQr!L>aSM-Si?5#}Aa1ZB zv_2**MPU@v-8T9Ku@ZWP;?wFy7}$JD>=s~B@D z4;&*}-j-4~7q;4|fi+cvw!(AY1c_H5GC+WOs`K9lXE}V8Oix20a*q~+tuTeW&@ma! zn_1jiEu}D=hhYS1m9Et1Z^YKzR=9Co0rN)a3dxSuQj^?wWDHdx)yeX_4?3nX@@~$r zF`<|Xes{?5RV&0#Gv$n=v|`3NxRmF(WcB$FIjd#=U?TB-i}n#acCuf|NcaH|CNvln z>^c7S_1Dy*@$+KLN%!1_Y8o!Cf({y|25hMX<85V zH#Tz9K9XPb_d{`SK$-PoGC5DZ#wLc?Wj$pn`Vot02fJD@QjqPM%bISiGq$DODq{ z+3$Ybu$LQd!NVIngEgYWB45oLw(>>p!cR53UlQq7+mH8KMbY^%nFr;5iv<$ z7BpA63%_qOEhD+vQBuAfz8T&Ru6KVHoL0Y&O+UPI!6Yg=GM*uGIDH%u@_tb&rplEJ zC(5sz?p6DdVFX5M#J%m6B+JbhFXCi_6!5C8klZ#r80%)@;sir7;MWjDt&=LJjgq>ZXRbIdLz@+0>cRrdQW~i*K5jAPgq2~UZHH>I9mzAsn z#k^#+n~Tj<%AJ4hWDyOrON;K+8)Ec$nHH_wm4UibiTKTJMpX!WQs{hB=v< zNY)yTnKt{D`XSMFI)_Cl?R44Rv2s#5zgBrPjuMG`yZlh0-FR2eey2(t0kW}>DrMM{GUN6y_0{(ue4_N72HBk-dutv++ z@)Q{-rD05Hk@Sg-k-m))6Tzw%85k}Y%wDgGHu-!N&*13rRgr9<>1Q=gO<_Xb-a)}w z*xuLk@!CSZ6Rw{@3UPnR`T}JtG~&xo$};m?$qP31dTq2JT&&G3HAW)!_7s>dA)apZ zvX1xdT)O6u0Fh{*(sl99kwXzAd;^{(QL8Z#$RmetK`0@R!y(N*osaJRiUmG|*m~AB z5*#osa*0z&g!UIfs0Zq1tF`)AZ;%afsrXmKX-uDY#&5*1iem{P(~SW9ZC1_IK}$uv zW$a64#eZqIGjmgrdG~l@%cmyy82#NH2t}{fh_nnDkg58t6yYrxlvN@+KtV|9X%Q`olVBL&J{LRXx;xi#msnXAaLc9P^%2BZArSCOD$m{ zt6$B=3y>~STRSa8E6b{V^FsXr{?H}P7mT@w^B3#r+zg#CyL{C+tU~K+auN{thV*Jg zK^*(*fTgF`Ijk$vBg`h>d~T0jd#}0Iq8PFiN}FdHn>Bd=`#XEz( z*qHoW0acC5)4Q%0Q&=l$P1M9J!BC8ubX8yIP=n-{Ca^G?R4cYpSN__!+3TwJxtmw$ z#&Q*Pu0|9yNYrNI6BKjY)GneZLX3Q*{^*D`rTeHbT0$%c-x+iTY%`%My30J+LFs0ffN6L9D^1-95Ah?-ZC7J#=S%$r9&E1UwfLrRepMDliE``M;MI?90Z8 z6~bKxTZDr~zv2U3lJ-OpEt~5PjgI@($S!@3z3)n4x%G99_i?L(1vRR8HjZO*(?~fI zup+z=SxyK@+kV-F9;(r##f3$gi@%$wNRJXI-kM>gecJ<dy(YfbiOP>16vcK5D zM&m2M8eV4LKCLp4SbxTI$MKF6P{S&b!`sh+vh{xX(_jD-p_-M{UTPurd}D^29v96p zc9VT*HEOLSQQ}gSb5%o8`&~~65i#0#@|-vE_k~}16nDVuG(kL!Xd-hL-S&4YC)K)X zhf*bA?TWjFLS>-^4qlmSt&VD7DJlTz2e!=P~v> zQ(+h4*R8;~>e`}&D+p(2`{?vqiS$A3)X65X|M^|>E8efnG1HGB0vs9`fvy%AIlt%ud5FqGCW7Z@%*}|K1eTgB0v|2l z8a!&o*~XC;T*S4Ffz#jjq$1sE824$D3MflN$Bi(Acqp@{A^<-5wia}d4EBtyL!D%DoXgXPWf`cOE^-3fUioAe_cO~(9g z>YdG%ZO+@OSAik-b2pnvhxAWgO<8CgBI&pj04~UYWO!j$gGqspkv>oa@6ngn`QjCi zR;sKUa@){)f8RRm7=oxtCG)Es)=>N~B~$FqKsB>ey^|TPpbo(ZYFpqyma`lX6SgN(SY6!KS3TWaoVDOG zIgWUH+2fuKJ`Tw74^9TW5zl(BX+LCN9_7q%MMe_$jn*ji$+dt~c)r(~L%;~qOIY8o z>4~uR@BHCqU1P$sbp`e9Y%fv}+@JD<^4~jkoPd1ZGO#G*R$%(EqT9ATVNd~nJN`b8 z;U-3ecm75&3Vhj}l^hr}K^Mk{vU&?2U&0BM3~0vIJyyN*DEto_C6f4J??*r;JxsHRM~Wun*u0Ok z6z+&^WNRw*)97HGl7tTQ#M;p?FeyR+dh!t|2sSDbc1hqMY&GxW4Y=eo&*m+QB$XDU znC?=YFxvkas>eL0czGdzTe)Ii!*S3!+3wJTUAsk`9hK-wGf=-*u_Zjzo&$C&Bel9I zadCsYnCy;4#)Z;DKdl>0GTZCa6~1rYV|r>28f$eHGEK~U()?!$K9`Sa~&k@aUjwn&VZYIErEnPS2rsDFaOxfg2i^jZjbBbM>m8&Pf~=-X3ry}O_X;ilssC!;ruwQsU# za%JOUND@5ooj6->CtaP^o>?zwD2|Hp_xH&+QWo#l z_JLl+r|<}+_CDC&*J+Me_en$53XOI3ssB}451I)20ZgLFf)%AGUHQUv*dG`_=OOT}i~Av4SlP!;^kr^&a8s*peF_h7_p z+Pbx<$=T1%W=~NJ{}3a8ps0mU;)%SK={6pbZJ7C?Ip@yKqHhrFWT!z+<)B-i8U3Cf zpu9Ax_IK*ItyX^rpe~N$ud^(XYEQzHb}ZxZnDQJ*`O=+74KSJ&a?3T4tB5zB_vwLI zis2+A`G@~UiXZupii zxWdAB-X<49m@bi5Qq7zg?zcsGYvWy)Cq&0;4#895xRu}6P6%a8c3>a8d_`O#}H zIa|i5whG;M(@rouH%$+L>9vh&6sC3e(Z=mySi_A-pR{(tpyxM+`2hoWt}S)s8SJG} z6_=@JQ0BW&?pRG9y5Tct#E=~0#`fxQthp@gQ1)4u)Ysl(nh7&)B-gtlH(O!jX{AG6 zClGC0!$oIRxS1w&?HCr{$i5V3a=O|Zs~04PpFbKemk|A(0}cqKe+l^nhUr94_8peT z0y}FVUF7@KpXH8mptfVPL{F8PA6y$M=9g#Z1SCaU(`^~mZm`&IE^;Z&((K<)!6Ec_ zRd!D-kauin0$V5Xb*(`^v*BcVm-~p4uR53XhIQlX=Y*vZNB2Sb$ zqs+>~f8{k^rqMMw=BL-S4TnzboNtWqW9DT!4&V5qus3Op$!x8b${6_o!^=ixV==`C zmaw|p2uYxKjK1|yW`w^N8~K=3jS)AISkZ_z&Q2Eh;rKPg!==f13!>>@I<7CX-_|K2 zn=aYZl52Zx$<`I__M^{M!byikm2$JxTiXUnY$~b}G(S4NvfFmLr@sN+pl0r1>VcVK4fj zm>ZX)?3wE?h+LuS;u&RB8KEwUk4_mO3vUh)Lve<=A3 z=DIi8$z-AlS2&6$9Nnx9kTy47h8Rk~PL;jP?H{`mr@}+DE%v_0*bLIAlnf6VegyL> z@vgc5yh>Exi2vK4E~HvA$C_A8A9WZB`z-L_)nsrpzQgYkqDrfyRbN?!lfnhV5j1A!(U#C=_t}TZ+q2r4)*K%Zu~(yr z<)7CX-8bVot5~w)PNHw(@}Hq-X0r*=>Gb{zR?A7oDMhH=|9c$%cJbLLbi*niBBD#; zl7EJH3$<1e2f>fyVL3ErnAnf&4Pr1Qg3v1P(_)oi!R|7%7XLyEaw;~g66t|kGy&^) z!#k-tR$$GTzqdXXo%pcUOj10Hm8FG@<))V9Lbe8n0|6ZWBRQ-R24xx$kd?V4p*3jyGJ zLohjE&(SO69%C;z1w>paotsAYhGrtip>5x85rD54c2Vfl_>oe7+fbO1MJ^vw-vaFT$`5@VS(STLOKuZv*fy*&AchmtewLZOgxBGBAl1{pY~RMeC*lD4 zGL#lZJp`gX;toRQMc?RoCSCW@yy&&UbMZA?olt-F-`{K_BcWq{C@)z4jbs|iWjJ`E zzP8<&>A;tUs(`lA5G3w4CxVanMR|wa{G=f$zE6siS^X(6Atab8Q~1jFC_{MW=-p*e z{I=GtC<(@&?C;6y*eNFb%gSj%9kg@k#SN^p!)8ZL&n3=-vwIn#TgqUtzFyT!6lwgx z0m$|qD|#gbHLXBAR6p}T_plTVZ1AkV9iwvP> zj0RAbo<50dfqd?DcmKXsDpXkDuKdC*yc=?T&WCz&S>)~=d20W5tzljOjT(Pn8=_9{uw zM5Lflm8jI2QEm>lpb^a-yFS76>K4B3BMm$4Dd@7Wa=@LMefah`UfN-|eza@B;(hM z72u}4R*XjEv2a&@B1QBWAG?TclFxSL)p5gI{XTBYm7}f9@H!AkrP#?GNZ&8|CE(ELMYt>bW1-vn}xe>+-Jd2Enz-p+OE$zfLPz>EtkE4InTvA@yGO-x#ZLB8bOjv_8e zj4dzmj}?>}J@w*spoSI>*~$<8o`257VD6!qJ9=e|Mz1Nq-QsuuRJ##kq*n=svlb>q zP}aWJIv5i>`1x@Z(q98YFh9$;GUMQs`+Gj;;h6~jFo;t%aa4cqumB0W8en9q$wmG? z>SD=TsSu*Ua&P^N>B{VwmI=}V*KMnz7S+F7?YrIJChas;daTCAQYE6TVap)>RfW#U z+r-11o#IC7j*S@Ho{Z|Wf+4d$>o)u2GO6mw=CiU~K>4V#rU)wnZ5OL|%zqli6`leR zOq>?BkzTzQk=BH;a5qNvn4{l}3Y`kh6wlc~K3fr$lEq@CR=qj}A%#?#VwATfhI_Ov zp1ni`d)1>;J9_Bw?g#w7(|J4?Uthu*J=b}~1j^2qq|Coj`#GY7Fnuy1hfpbqvd`xl z(InF82^!rpiPOMO1v@@QF0$=|bxov;)E(g3J(!4{NJ4r*O&0a`-J~~_D#kT8m^ogO zu{Z8Oi)ZXc%Q8A845To0O~R@3f>x-4;7Fu+$1m6#6hqyw@l}amTI}>Y#1s)saRu_x zxT=}l@3AkHoCT?wX>?|HhB5q}D|P#=vsfO-2Gz8-K^Tk<+ytrqJ(I=l{$-+(On*^c zaCbKrKH>7I&4WyNbdXB#+?$)gPx1wZ6A0az5vFt zPg81kxnFaJ&=7f5)q~TuDblRgz%JRPMTJPdl+i)OkW9UjDt@?FAa!V;lDnKFQ`U2d zRvvWZNX9TBbfuflB)$Y(-9Wag#yI(q^<%C6-ILCI8Q7{xFaFr}oD}Om3X&U1@bCoj^ z4HaCyKZ3xZ8cl|A-Rl=m?f2LAr0~Unc9Qn01|-xZ&=2^YEg~Unh-aHp8ws_+erTD- z5GrYkADmImmIQ46o^U6!XOVpuX!7eKj26vI$QtPmhq;R~R7%5E9BT)5GuS|OvSN3- zbG)pPg-*k!pO1l?+FE%!^MsS|srxf*wv(gAcyi(S^=8;7fza;v-F;XxF+rgtTWMFi zsF>hcgpKd8-%h*c(Izh>s&5T*>%CpFqcvD z+>E0giAIJ>V~K#w5EJgLzRN|mDTKw0qeHC{(|&G$>iAGI;VHJ&cpr#;(1}Y0ldqLQ z(|`hIi>{=hTf6hE?|Ins4)lp7ev#>PZd9|gZA|ZsSp>GRuqi=NkmcCJ1u1ppXu+tL zFUK2|-|Euixn+y!bXR#SrA(~AVBEP>5L4|r{^K?{glMzeB3ey$7ZGxfrS+g3$hTSb z(;bId@QjBw@VdV_=`c;p^M3u>%Gp=~bqGlmC|_NC2)EjWfA9wmvwMeNIGjs6c}Us# zL=;|jh=kaNU|FWFyxU4E8M`dDa3|5TDEF;c&M#l^2qC^`yYUMclihVi7G`%g&jE3A zPbE^99QooCvW1owX*2D_n9NV{VMAox`8{yG9q49&Sf=1HZV~?&uomPiIp_0|gxghR zDK|~og-NhvITRJ&vI}`#Yn2tiCLZr$<@r#7MaZ~1GqbIq^gR-`KpTc0%1s8~XPbWW z=`Nw9EL8`Djln-DR=zG|;P6A#x@v3!+^kNW`=lT;Gbq@;rI37_#kNe<129yr`RPE zRWJCVKz;hJ9^}%o&mNG>nf5?rtDwG5h<$yogo7Yli8)$G3Yl4Jnxha{7v3M&A%n~7 zGfldE!dhaRrIKnSd#v$_nNa|IA8Am%rk^H}i<5KM+TkNCa#`!lj)@B~$oq^K^C_y% zP|!NmoHjM+W_5B>iTf2s)AC?nCzNO7jl$So9b=X3X!+Zti_Q#Ii+B^LQBjMgME-$h zfV~y%&2P;W(p^5@I=|#khXN*DFk7C6F>tGoPpm{!XynE%wGFiIvmG5<)TicCek30F zsikI?INfinUEP=3&2vjZgHG}{I7>Wk3|D9hg-6N7=PT~J4;hNTFfMP&nu-cThV(Fi zK-z2Y7%;T;M6g(=F{1f>R(@pJYQOw{TvAUq*inv!AzEdcMqC%6uQM%#$oV8axdnL+1Oq@b1;SyZQU^{@ zY9YXu?p(Zf_t@nl{7+dC+QJNb+UZAP?gjqplWQr)JiTJ{)dXW+RN^PkBa&f1mog{CZ^1GY$56aI<56>(*{eU3k>Gt>;47H9+5H8%=D%X67vFK>g&Y<`Eu_%x+y1@1fk0C*AWn*9qq-yQdnjr<2!S#n}TAffPLFhygLQ zzOrmph$@o;l!~qGk#MIiST+?cm%VPytBLW&9D{!on9dN)QIQy zS?Dj9p6Dl3r)_g<_)3GggLC#LKOvkGnTzClbs={4b$3RrhfUhhe~M->T#V3VD)91R zMx}ukN`H5ud%>l}0sOoT!P2u@>SwnP1I7qWq@gwppGo5$0?ePrLt!%1pRrHv&f%fvvIei91t zrcvAG-s>L1(ktQ1qsWYqf}kxr=^bp~zgJ4|Y=og^m^8zFzifKVQrSBRT0R zfVM?X)Q3I~gq(ANSqnSwTJQ4!Fs<5sp4*01W{v1_fjvL>q`h|d-r38xYS_Pv{)VG_ zrlwGvt2_pji=7A=x~sOPOVE7z%YNzQWO=89=)=Z{n#Ob;pq2Ytwe_uwJS$tGxpIyV z9BKAef<4|6`{^XYL}qki--t`Lo#ow_lmP>6UpE_zy+7H`0IsXWOriC3$nsQYYq%@H zv0mW;V*VsYt4qHj(F0-l5@Ggrrep4w2?d{8cvHPvD@P27hl7p>^M!H=nFGH6ifN-d zM=%|v z)dduie*}D_8>$0&gkm9;>mbZvb;$hX+_SS}6YE)^W?76;zlAut729KKSs{5EV`YRr z&R^q-VjE=m69_C(=QBzw*RgRJa}{~tH`izd3td#&uQQ7{7c|>0TCL23@CX+7ln~QA z*pxeeNCzuXkBemYy>%^5#!K7RU5snw$$?g~E;*Ck-Fpjh9QI0m-MN~R?*sZE*5E6` zHe|wWtYQWQgVs|=s_o+kRfW|fh(LdPZOXAiCJ?t_s>F~nOkkSR5{Klq|*7sk_^U)I5~%IiGT&F(HU zPm!wv;}`nf+n3*hiA6wl#Kcz^Q9w==JBY2hr6m-7g$I}JM7Y(!pdaVVY~(G$oR)FX z;FEHxepA*hnA9~659ij@)s<3L6zcNx(3)%H`rFz4GI7Y7Bc5)|HK)CA?f2q4dZ=_&;m7gMnuK>l1z+@_B%ua;dX0qevc_<@jgLi# zxvUf7K=)zHcc`x1;uyKNXVj4yU4_;T!zFL<>+G;i)dFGe)%^_Zx1K@BTW~Rs1Mvb%lDy9~@G(oN3Az%CPHyxlMPpCMleji_3K57Y7M#H%}BJ^-X`AlQh zE1x9MP`rlVB7%*NcaJapSgvrx2+MV15_Em8>)1>g5Fg(s(gnP?#|FFTh5IuC0a@(x z5*$7sg`QoLqCeV|YjErp+fpMIZ#a!f>mOkfcVyspE*J=rjC)#o{I974xUpXpa0!Lc z=ALUhQS!Y1YP5#1j-c{BjDIu}nbu~i>4b#O3L~MIQSIIOuZPT0=cEbGdmTJVv*b#R z4>7Bjou=%YZUeJ!e)e$ld6qT}cv3rCl&f&UJ=0Q!Y-CTWG%{&9%&Zpg1FJ$xQiHwo z*6U%d!)7x5AnEY5v_Yx>QxJ4Sy@{%gZ%z$Lz4C(Pcli23#n6Uxi$nDjz7-EnRJmSxi@p8PBCd?0w)kVm}91$3w;<+x{TcgbM#95fv8~3h_UJ)b;i3*yrvn|iK&uo%C zd|0uUt|oh7Qq84ksxfLe{Edo~+(7PcDSfm^MnRGER|Lefe12wPBeR4xvujRziKXAt zt6%fc=5XD>G=CqRSbkcU#133>XYHWATlVy7UF>8mX$A7waWsR=K|sh|k9^ZOq_6hr zyXc$>^}khMZVfQjq^9XcJCOF`mGXnU&*r9w?q!_x-{0@ls*>z|T!KH70Fjf7a>cL1 zx?MSn&Y$3O=|=~>z0k^WaeC5b_?K`8y{%m5pP%#(VzJt# z=jzn8kGVM16N7BOf-@(G*k1QXP(i8KWcnpszT;z8A~LRI+m7%2Ma++1rv40!&(G%F z@`;!jr^<*3L1=_<9?*{r|2B11kdyvh)z4^(I~dfXXc-M%c%}(~;XfRWZ_`Z$;UA8)77bdf> z-pTY)16b^wY8gSg;$h-M6oP_(W)u#cXb@P?L)|Rz~N{PfLzrV~eP&F@1K% z;Fi~I(+0+_FJpx>`CB4M*j$vve zVK}Z>{m987cJ>YZ2DL}m`&mK)A&^HwL~>9i^%ZpWTfzhmN;ut#Bl6xZ_yceG5)dk| zsS*pd5fRA2SCHcc#A25Nx2^&NO!P(TqzbinRVI(bvJ9BRT_{cHQf6rOW5^W#+z_2h z<%`h7Mm7g^;p0PTR3nNeOXLJX?cO5WJ&dbxsdky%Y)ISOnDL8T_TX`EBp(dQXGxjO zO{l(sgL^|s1*%L%?f81W|C*;2R?cA+EPgf?jY*qP-w(f?tVV~?(3}SdYgg2_3+hKb zRHQm=B5U;}SX1V?l=-XHo);r5+bw^`$5Z$_YV96s)N8wxWpMZJD9*1Xc!ub+^d7B7 zMR~2hhU@ZdcP?yBxGk6p(hh&w*QT!OI4=+P2WFn;?ZnBB>vqdC&14$HonPa-u$;kf zrV5})`u^=eD;xtn3&ulGCae46pmcFttX|NTrf)bq(T%WVzUR7Xj_>p3aSRlX(qjA>%vPc-AwF zx}E)RqDE45ghE8$HJ2V7kLSEqO-^#k@wWxNp@2S@9Fbq1dwwctF45gfr;{yzzKNnA zWknR+JKMPhCv+snBaDPyx3t$pgzI;u1@Mq&~EV*{fs4oHud4vAD=-k&>d&}I10ALG`7l7j{JviRHZBDzLP8{ zrCyPt0_Lg)C=By}ldm{#RN}m7JZ2ppP#ob`*3#grfodZ*gJPA@uR$2GKGhXzRy>g$X%rRPz3Bn>4t?U5MmE<8@{Z%wF?_o|>lnhjZ8m8`wACv7gqc zibQfaisTnJDMQ@T1mlnAc5ZO5wOe*kR^JU)Z%CIVtg~iFL2HH1=O1{$Z*XrXvYp)0ef ziRoHtuB2vpRl!`xt2haM&KILh$E}(^ae*GphY$TUyLg&pxOpOW5X(8^p+k zwbZiK`nt0mML7ql38Hz8IcpF!%d%-jh6d5@x^J_J(JX%it)yW7 zjA8q-gae7ibcT6W7=B2+H^b?#&`IX3+>!?3&DG9)UR-q9jBhGJEUd7kGc>vzoej~riJ<>?Zz#(6y9P%C#`5J_G+xNrP{6}wfHiQqc}gN_Zg>tj<$8X z^Cl(273S~W{!6>gUsfq!YC};_>ZNW$2D7w{9jk({T7tVxS*&+52PbK6 zcXH2V1=Oz2LPNd8;(Z_1UEM`rG4rm=ZpJE6_UY#kxO3hgAM#V%vJ5CtS8DiOZ7_|PMNTTcMKBs6_89cG4nvUAjJohj^nrDTBAYym zZw*~{ruY5#vz}4@!Zca8z)UDn@zp%~AXA0{YL5_<6dOgd(D#vIXOEW6w@vWkzjDX} zkFxcSJlx%eja6e+_`a?$7n12fA(71{{35B@B!1a+y5c15MBA>et7cbe-Av4etJ*#J zP<^4;Gl{4YqwlC4;2TAdE`V4{>MP+`A2U_C0jUUo4k}<;r1W^0PmE4$jw3q0V&4Q` zqg}hi`#ZSBt_F;6KeAr~RjD>g*v7BEc75$OtU!}6`>Bi(fk)H_si|IFM=}gmHsm$~ z(MF!IXMmA!JEfF)~vDo8=zQxRZy^o)|8i4UwkTTqT{@j@gtGeI%Sm=&I z6j};(L`JGrny22un4j=801D2(-F-=Ks+n6GlFAN#ZV^Xkt$Rcqq&T#PNF*tM)-@-u2DTmIY$1A}qZHvLQo9)ZD+JU#q zHqNX;_Hh=}rxw@zxT?`Z;@Vz?V)`!R&b0&CiHi7W-bQV)R?Lr450OZc){U~2RVf&N zG~hVn15G6so4)1qsDtD~EKh`u03!m4TK{x5={fuyo28OVHP>B{$ z)8&rm8s|svLZj%?Zf9O&X6$!k|CZz5DQAajnfN89ZlWtfeNhNupsA1@YLZi9aN@dW zFz0^lcaG18C}fW=e!Akg5-diPxe>Uhhp7+q|`uma7zS_Nw1Z2+z&PnYo5i0Ruk=DmkZk$C5 zNYHxN9S$8=J`0JFs+H$RI;9hY)D$Lk`3f>2HE>#~;m$^aU|-l3mzg8yGYM=?jvzhE z9W!5HT2_sU{oEBHfBQlflq5 z*y+)#SJ;hG{m>=YOIN{1SC-Y_3XMCKY7)C5>Ua*r4I|JB#f`=tY~@>{#T^D{NmnW>GiwZ%m} zF)flOCib*Z&*Zcz&Q18MD;BaE1=kN>J3WrV?N)U7H~Ybr%y_=y1{S2J;|U6`XrqT{ zh}XzuCx)X+cY;c>Je0lCO1_8T;e9ChvLGl3w_u9$QZ!fyZz)jZB&C2T(kg+XQP6`F z(EZDg-WhTlNN!Pr7oNJLjdhDkx}4A}VeP(&ke;28M~6(ty(>P>gEM*laC-*-4YDD$ zCF$c~t*iAb@B>+(%~lne*D@d*v3Ebv14ZF0^FJe3M-Car9mA50y-D>-TS$%S zt(c1Dh8pu@2HS5GtVrvT!APT2MjUe~D)no&B>$1kH5Oo2D~p8ql(>ePzLws@AhB)l z$9{~v!81oLqgJDD?JNGhIrbk#x(r&h%%v=@YS?i)X2?-q6Q(1otcj{|OGnWtRu8C6 z0E&Z3mXl*)yj{VesM_bMjx|^6?^ndy!idw3wwe-Gm!Zym_#p0WA?v|+_*Af~B=_Mw zdwIFfjKmUqjP`6&DVZGkPRAvWr4Zh=2-Xpsc#s3!rp9O!bi=V38ZS zG=(f}8?-graOa)?cEvt;bDN?bW&@Ts=cSp#c_OgI3FW27Q}HE)#clU=F`6IC__(^J@lqSM?8@~ zkZa$xax+*nA&!p~fxAWyTiBC-!akq>ekd+*&&V^5j&L}6CdQsg zn{p6!CiI^xqj{^av~O{E^{v>YdnOm{Ou>OxSAp?fU~FwcZ8%D&Q4<4S;h#F=()y9F z<}7e~U?_|0FYm()HN#AcKC^~Z>O2;!zhg`5Ks6$Adg7`E#C;Fu&tu9;l=+%Ck*uH5 z*d2mS^(h;BQXJ6S*To6F@|JRNz+TaXv_k1vW#Vs9ayhTZ1_<4ou{yIfky&aN=1XMz z9+T3)F)V+TJSPAs$j7xK-1v{(d7kdLNr^I$LwEwTwh5U93z)x;yIO<7azf3E%X{J1 z;FC`9C8#%ClB*+P^jUBn*+^o@5Bn66ukop4ZD)Vk_esJWuYQj~-RGMOs1(&P>E9uQ zS#uuM&^!29R&&jM$cyp};zdkZJGLL=#(5r(^ad3}lVAaDAm8FHOt1Hs6THTT+wbOC zzV|DNwe^Ib%kEHA*I*z`{Om=lW43lWeEEEO5NSr2{0gAR{ZOkcsGXmro4N5Ri zvEJOm&5J<7A#1c(O~-i)ZAbL*X;x9PA@3u}p!P-PxwRCFROe5r?g#Bj@)-m;{`5A$GSyCpg(FBd63gE+6S8%8?qAUM1IRYcwi-g2e!}cRy zcaf}rSofYUYAnzq-h_H9uVuZ;@0;HT8l9iN&BjIGduMQtbHm109A19Qjc5 zpOYYGs&}bToWGdgDc)OiT-A+)Z}HXU_y-OT@^yS(X8gce1ai#+oppNW!$CmA+%|mz zB{CK8%0>y?(e{2~+%or^iIggW#-V5i_4z#*mWEMiEb3QNfcZ`5K+tHXkExIr`6xAj zmIyO~t|1)8nXrb0t?NxDw6Qju$)`S!DzFah#R7_VvjV47n<-j+5ts3G#yumgR-%%# zrI(7WV4pGNDa`^O@8T8JQt>fzVk|cEk!G4P-L}w8KJ&x=eH`@_w{vF}-9Xx<<$OhI zyLkC!>&-HJw+MX2w@W1#T)>plVv(!sc*%n9Tx}lAxILT7iUJqJ|~JxvERcBy@6Xps2> z7y=bH2N~Q*-)caCwZ7=rt|N*;h-TF9!fD*24_2PSseP-iz}(LXz1uCAIZTmk)evUs zZ20^dVbjO&cTc(*Swl~tlW`D%Nri;+YeP)8yStsnabH4QByS7QTNT6X+?5sehN1D9 z@uLY2V4W9w1Y&vMcMyMSmW9L5PJ8z@V@Pv^&&Uxm41b|~rK&(=Rl5GJT?Ij8vm${AmnIhO2bG70F!V1Ut)(fPNoSF`5^p7`|aAgQYu`?RRL5&D87A&)~}5^BB`fcCZutv1Hx_Ky%M#{@fM5{!yN`8H&tsKh=xkP zhfBCo#UD?1m@odYkv3(S_4h~}JVwlfPne-z?-bo!<~px5jVA@ih9szdaj>Ja3c_XO z(F7Ew8%_wyOW#W7R`5u+lml*| zVd|RL^biQEBu|Q~NvuPnEg2cU=JPH}-EA|*J$!E|e4TPA%!r3V0FbtXyAL~ADAZo% zz>1y%^tf-E>)*Z`k^9I{D)}@92l=01pJrK15-|9 z6U&+jPTl1e@JqsFxT%4@=-fLDa|LX|26A*wP!X2=6>gyub}Wt?INIgQfM{!e| zRR>Jm1K8R{x~CX8@2s|^PSQS)7a0W0Rf06p{lqhS-pNj(>QeVID;Wx2_x?O5s*hO> zXK3>(t;alk(R0R-LRsx%2%{RF?r`mr>e9t?JM{yGYSCmA(YUaR?vKnmSW_w#5yrEp zH~YK_=hiXKnHSe-gj7};#-+Fs`l=%Vqr6d~BbaCfZ5S-|B3|eX+cr5Kr=nAi%%NT5 zNRTBw&fgFVyz8L4rgtef@CZ^0N@7y- z@o7aXrC8WtRk$9uKSA?_pz;K>nWaT zbkkrTn%sWfpz;|lc0gsQ>SJx?Ww9y+o7EJ#ERBey?}fbPmgG(F9qI#av?BXcNB!@> zhM&01F=dw6=jb+k)u=#!t*Q6zmEb{imeCF;f{qr}2_+q@T5H$eP#11sh%}$Zb3lro znCe5zdHRli4(=Ksl;DT!|D3M1YJYY*VPd8{uQ*UP*MK}q${)*l=8(jotS2a89i6yn%cQ%zTC+JwJAi2*Sy&4UtyeU|-7n~ze;NGo25qDAd zn|r^$uZd2;q)_ye(3^+ zk#hTZa6XO8`Lby6@hp-}VSeo2hBUmP7tGIfG~B%h83Tod%m(2xA60so?Tn`a{imet z?_5`t!W+_)9wgr8kaR`NaV0Chf>9jDa4ZR31$b6PIfgPN#F**JRDF)wx_KYr7Hl zIxgtrlb6)o>Ki?Dzy53J==%2zOX3>37`FI4D@SyN9Llj?P~j?hGx*lmh0T1JNlB+e zr-*L#liSxcd4qMfwge8hG;`>dG?uHP_#Fwdn|5AGJ&l=S6<0s* zO03x#Q*5<0y~GBwN#H)F6)wa*zC~HS6tEX;!F?9JAODxBHN@E-I^DuBnECOJ?wkZU z$xWu!(N1I|<^RgP7JBoha8rx=ain3Ns6;`^J^P{g@znt*Z(SpAE4+Q;VS0EC4n@-F zxlexB7SJbbw*fDy|2IHA+nheVv%;omdgYBi4$jyCFfXw+-qkkC=XD zuh#eW7=JaML1PKpkM(ONw5|*$CJroOUi2u3HE!^dP%pjv*kNylxn)3*)-@WG1hq}>FKN9uR!@D`_hpBkhjm{sY zw9OFmznUF>%1DmNqMELhthi*^8I=3F%SR- z;|iGaLu;Ko!T36vxG;d!ddfauHF`b1pM~>hMXydUkuIMZMz{Fuu*QnV4(|kw(?Myn zH>fq2nTkTRk3GIJHM43d0v7k8Umu3)PkGcMsg)W0eAOi7+bgp1yf$6|-{%P9;+!@0 z$opKHopeXj8slvn{J!=73F!}&=Z|vfB5_K-T1-~b$3&@z?aqF>eRSBqVdlwka0)M!1Z(rm6YpMa!tCjuQGsr_?cR z=3|f{4E;+Qj3IskMoQCJ9fibm5tH&Qv8i-DRu-Y82VWMOmbg=|}gv#%ox|1tc zH40>qdj*HJhhov)SDkjZmA6J+KOSF{cNjbd3pei~C5~bhJu*9OF`PnS#^t1K7HL2tjR?Z~42* ze@v`YgGc!h) zbN<4&gRf{YIp*T7`?CH0LFJ9pv38E`LI&dS4iLu7kb zJmq^rX8VmB$T6|~1dF_y^;|R2oOs5A3G^0wCIK-AzZr?OiJST_sAuKc0M5%4|CK#r zIRiUb4E@l>LIJQw>i~%lju}FKGJDFr@@a_zV^9OdNICybHky<)7)FXve~6SE``^Qf z#{{lLgtTg?tzYA zU2ks;Z5ZCAXBt@Zr3@t`^j_8w!@943f%A!zue>7pygp8GPVB06seI z3DZ#%`kmYR@AGo%RPI!k4cr<(WAfX-4nSm0ynFWH3yhRmM!N@dvX%|x2ux7Ptj=lH zblIPfG4}au+1%N5cccyBe4r`Dc`oUUk_%F)RD1+%ko8=8GvApey$9QKpN9y_cWK;p zCQ%Im_M{GQ9TViYNvmFl8rYM~VgzHwo5q56`Wo9!S;!L$Y1FA$U zq?2gmC_epZE_)?8U@v*9Hi_jzXj9XIRq3)FLR`gnBM->H;szzg=>iEiRKlctUaz7? z-P)36J&ZmZSJvoP%J2CrXgyy!xeod!q)3eaoqC>pQASo#th@aMHinJ6OZx3xu+n|y z-MoVS82|aLtz8hUst_JAL&S-O=k|5SelD4Glj1QY8F6n?%pAjlwTU$zBj)g%$Frvv z-mRVycFD|iqW=4a3h#tUGfQ?uKpheEV2N!=NcuG5yr<%RK|4`qe}1#uJ}RnN092T7 z4pz-sBPko>ykq9AtCVU^HG>B9LvjNCd4at4D1mP}zPHuAZDb%8ufcY=$kK6?a_NtZ z9h{H$@cb5x9=DBVI0|%C%}~H*U2?4p7E+gjX^BJg`um|A4Q*ak8g}+Nw%-cR20MZj zb_;r3y@6C`FatLjhf8z4mJ4Tip+cD+NmM^1q(2GwC%Zi zD8o6iB_2Dd?i?Q;egsVP_2x~04Dq(ry^-VVUBKyk!T_f`Q<4wi7e)yOtQ>-$SIE1`vRuoK_rzD8 zQ=!({rah$PNMilT;GxT{FC&B|Vbk2JV{2K;cIo4H_%_ysE=k~rvJ=+;o1kOWN7T$z zO~9B2IEDptw#a~YwF8z|^&}7EgSJiPYk91yQ>pJe^qiiCD?f4)8SzdBZM_${g$%mF zjtkL|K}nh6B--m-YJ>e~mSKJB@ z4fgRn*_rS=Bt88^c|tTzkLs?LhTV^o1GKL4>GF(Uo=r}*vZ(+GrfZgwbz1P55RuD8 z1uoTW(&86Rufl9aB;4VH0SZ97`<>Y&0|)5Qc|mtWI-6p`_4l}BXdi7qb8n-frlO~I zql@X7ug60tW4KFq>v5?W^Yn*X08#f?TLiOt9?_r zsJ1|4t9Q;-daB;Glw}!RdF)YfA=UwtL0xX)N&D&*VwTcb;d+GoPDz+p&#R-Y zt!3UYr!3pW&Lbug!?wlaBqNn12c5ztN+MZEl`cpVw;0iwjo{{x>2jTYXlWpJMjwr9R)FDRS21_ zo7%lnkb9{k`bQ_kE8qngWmmi{xz40Uu1u`6?wWHyojdC>0;Ku`jP#gjkfgAh#|F_T zrK&XAM}p7YYoS_P<(DLGKyx`A#YBw0@S1+!;v`A4B( z*JFj2F0>p7)tZeLprHV1%2C^AR3VJd!L$lNPlED{-2*Cr@zfd;Hmb$Y+VE}x97Qhe z51Mh${4Q9WU!xY7lsPjGO??#SA)JJazTVuOUza{K)0C9OudeMMkPHOfA&fX6^ux@^ zubu8yrQt5Wp?w2wY(S~z>9yMp2#WFi-yHVBtp}Ml?Z{GHpALbCn&D)f1rcz_4^1%Y z?cwW0j-MbK+{?I&FMWP9&@FG1#phvBEI6Z7#7%4PR~}P-W^kY^v($>C-ls8z7gyOU zZwFh$(umB7-Rz6K`|e{EZvrn@DSGi-Ck8PUv%^&w)C!JKF+N{MLcG0Wm^(UO?TMi4 zT6&2{#`{thvSKy4tqB|&_gwchE4ofKEL+#8rJEYn*BOF6wsrC@8yCjKqh=CY(8Y12 zu6LsruE%LJrm`0&F>7!v$uK8GrsJcjrno>^^^hdD>8L0m9XOPGw#p$@vF@$xnM@Ru zUS&vC1qHP}+)oIoxm%Fx=QHR8&W1+`USux1SK4_ZbL9*Ue9SJ9WJWFY-HDI5e$;J9 zyRl+ZUBJTgfe10v;C5!Uhf7;fo`yEjn3+PiHtbJ^T}-*!wD~P+Mim+*vd(s7*{cgy zmTZXlUD6=Kc*mCF?m^_zzk$72d^SzkRZ&>7IGyA!Gg;s-5`}U5$hy}op_z<+IAF_F z`PYXOxe$*|@0+8Atf$9A+ZTVu?fTabpCUbJe`q$6%+C_ORYle6bn!Ipy@wd$x<-VB zqVQuU;Egg_(#uSHe=1;8NP;{ z%DqFth9dF&E}tm*YlLZ_ zwT=8l>_ST1-wuLVY$k62<_h!X66|uyHHNT(k_gB(VUDe;KlovjQ1UlMr2oFu=uuW0U?+m=|z0Hju%XyTk(lay?RpQWWLb0}>D{I_TW z&N4ky1sq7m`@l}^IWPgkI-$b~FQIiI@kS}+axa8-g2eRCZCLh+yVx3v-nj`^uJq|N zs^qYs_ z7o7DA&pqz$Sh#7z_BUo5?^D0ULwxJ$T$=H{H<)yP4A8aik%(14y)Sy*+uL)64*2o{FF+9^@G z8UH9`B!=2$gp!bFnDX4Tq?4F5Ca?2_Mml*nqXtLWsf}kgBR*C5v{D{xSKr4N<*uDo z?T+16(z-|QsDod7=TS^*>UopHsK*P_8g3gE8YsM*ubJGZe(jA7zf_myN%s;d%Jkoy zaWOV#G7;OSp-@NfH-Qj}7wj-)(AQdhBlOU{e8+lTkjJ<5BPOLFUSOQiv`xCVq3t;2 zA5orE;bX~&ERXvnn1=q$ZoGPLy*qI{;+6WNDQD z&LMi40QZ9)Da1f1hs1rkZbBxCyPlPL`9W8|d2=A1D>;w-;E==N!cK`MIY{+E^`AEw zwc@Mn1j6)j8!Ti~vNU|D%16WyiPk4AofZm2L5K&fE#x>cg){S#)0Pv;D(GxA~3iaAbwZ5{YIG|EZ`_NfVRl%CZ~z(rD76w z4jeU%0e0%fyjoq{k07w1^|!(|AI_8wYqLHW(642RUv+IoA@X{+Q+0fvGEwE)iUr)Z zwQ?Y7d>eWTEuPYhX(@ho6Ak&z`oa}_{z`t9FH5Pu7tX%GpS((tU>FVI^~*#aT8NK& zvAL!sPA{9sb1tJ;wY))BB6E)aD04r4*Ijvz=7wdoUm7Eli>Znq>w0LL4QIXg!Ypg= zcHCT|?@01yjz=|anGuxCuM~;Ys>|`lIU#V0JTpQBLcXY|3uUsJzA8PRVdCq$l1j_y zIZ<7Ir7w-y&R1wQq^A7^7)BO6s?d-SQITx#P+9a8oOV7eOLHQbDEiCB8xJ99#WQ_< zNhfmxxmfv7m#2Hv>|fa%J{u2`#$5pwTq8wXhPwW2*|+p^%wB7}&_sMp<4ya6JQ}06 zQf}QADXuz_h4G``q`eP8PBA5!znW!cCH_F40xZ(A(AHrG^_fD=TG}v1blBc6;~wVC z{IQ1!=Mij=*X}-g%@H`_EA>XBlP6bPpD)x~sK2g+TSAx(+=wp@DURJxKEsH=;G#K& zW6qY@G#P8D;QmGa=kCOgJnLrOoRrNjd8!oWKfW)*w9W+++6Pq;MiXAjoY*$-gBF=o zu{$9H{Mcu?u61JFAl2TE3)2)28z+4c9adj)g-pWWifwZF*7_PImWr+32aaG&ARdJF zvaue#qNh69FC(uc6!IE&5_j6OgvKegSxHn6ajMziYq-;vyv|JGBY2{({qxCy+gLw# zzO?Qp;yb1waGRPtTF2PRB%)X^Ebvh{S|@u)V<%UChvIx@7?$(LJuy+NP7Ys31jQo( zjY&}xx^YG_Rn(o;GD_TZyHHfW-(K^*J-G=5b~!zEdBS7qr3^P7+>(XngesaVD;AY% zK82TI;!}KOGJPqzHKFe2YRa%z)Q814S);~w`m`ieZi-n8KEiXgf=%MCKv1t|5xj)w zU%zfTD*f$oe|AQ1smeP=Yc?y3yoT-H*3RG!1=Xl5+IK=)(>VpwI@MxuML)TI>Nn!q zsxpqx#{a-bLKt?<*>6c-^e(DDuC@0}6TIsasDnDDwv9I!>F%(QRVMhL5!*!t&<$MY zK33G|#hiO+6%BX@dU%W*)@`P<(uM`qY&QSL7nTHU#e9$e#Ob#C}b#n)WgJ^MYON$inx?b zIQc%3?qd9U)JYqSzKud1AGakd{HD-6SHU9>CY!iq%M05qK)>(s@8EDI^B;v$*M{vm z30He~Igm*kL6{L7Jt0U(hZ$G72;3CSn`|+6hZgMs;ZcvTD+pJ>SkCm-N)_khY<)!X z*q?Uxh{xA}&FbnQI}Ap$Ofr1JduLdKa%PafF&-svCH3swn5HT~%W(6#mp6N3U z+e+HB<0-0t^5Vemm?pL=jFRU|TurUT@sN?kL3f*CeSI*(j7+p9Xf!)LntF++1)e+B zf5dY_=_}~KXmd9WzbhvBd-0>wDPBDo-Oz~qSN#R|+w3ynFvP(TPKz!0oVcZ9Yadc5 z`6_n0-Pq)v3o5@UX?67Pb{qY4#4%Dc)jQTAYA+ZBmL@62V^JZ|xm~HA~xKrf;3I3cp#5zD84=;D2 z5iM`QJ)@VB;L?RaHtOphiSvz09QE45WFd0w9f2JuwID`Y-Egs=& zQXy>9jGVsFetD8VUljk~i@{0tW!RXnJD^iceEnxu55EX< zRu}B?2&msScgUxg&$P8ZeQ^!K62I0U%UGl+jgQ7q_J$iNcbSRtdVR(9#@tHreU!S0 z$ig&P(kRccP1CY%pbvJY$1Mg+GlSjWX=x1t@P@}Wk?oIUiOW)cdV1o{xO1`G^1*(l z&8YXS89W>X;U1Md-H6x}NhTk_>T{Gl)NI!z$e?%ga&8pYO%++-I8LSX5HZHF>z$hw zxDxYssskB^N)T7np7U6VS?L|a=xigkQ>%RiQ3?Y@VaAAD-zK8jhE6mWh=MSh!hWhS z03_?1TwXe(rZ**6at$h`Y!X9rSv8rr7GzMA;=KPpYO45qyQYh9)gZAqQ7e$`i zzTnAFirlTL`PMD2%NOotBttF@rj<3IB5!CIh;4DrSliaf{Q2O(HhRxW43LzEmZYwOv!Yh6Ub|T+qvTHjBlpD(4XSKKcBa!;vfcX^;kES1b+0E4#_#!jF7ZqkYI>-vN;Sk*yj(T69e{cp?c#tACLa$k~SBf>C91F}l-3QN7g<-0+_ z$S-3fvg{F?K(tX&Fq_wsUM#fF5$=pT&0K!5vNkOfa3<18#wFvto}TR-e_XkoAC6s{ zB>(^9Sn*;IBBIuy$pVpMHm*&SWfG0^o+aZ2;8?&1HO1$C$mw|kQB&KhKnMwCeg{)i z-4AHK^fqo)<&S4F$36&8gofXCy!yzAF3+{_DV^+TpxUBZJ~o_^)s^}1!gFRA89Aob zw$`no1p(?KCjGG2_3e8vhRuOQv7BKCL_qELWR)Kw zXvC15+D}HTc!ke*(TKEvlt5~NH!6B7K~d&md&w*wxrOG#tB2`pxQsVHZ>JxpHR&f- zyVv03ryA>@ZFv)Ht;X4fY$%B~hile!nC)b8Ow{ICp4xFx+tBAs?}sEm_GZhz1ic0a zvZMW0f!l?q`|5mYUbBbcAC^)yN21ApL2%17o-105Bg8(ZRz!e$mRrb$et}la8usTfcI3G zZ>M$k@$}&?0j3E3hlI9BrB!j22En@hB0W;(B}2}d3J16@pSa$}5Hda7ALhS!#?z)d zG?)H7s!axae|>PKb$?Vq@<&Ba?3fbm4mP5)m`(!0GzrGe$+y`LI;uE$I7UWl3VOOE zVWu#xu=Bs~RRPY1gRp$}sqlg!G!FH5s|^1)3tx8Xd8R50Rs8Z3n$_nWuWl(%x=LvU z@Qd%oPh!~2m8hzS8*vV=DDA+nT?KC1N8$84rJK`)Ihv(&eV}~4#`Tvqp;@0Q2J5KR zY-*C8bRTlZUEdgbd}1$|!yV&PLeI^_Ueddwi)H@Gf__y+Co_+cPV7k0MnUIi#X#nI z2$koO=s2XFo`I8FQ{hbERB~+o`NQZq6a7VmY@5S*Og@A!Pk_O=|Ii4^QmgILKjttD ziJ{;}jxf>LlFN;Z4yx#ddXn#b`Ihn8QrR~;Cw;^W5%;oRDR3Ga>)9J-v zS&f<Ww!Bx#*TbDB^5cCmf+!TVVS6qU{-&Kg2G?nN%GR@kOhq zz1%nK$0!lwy3lj=n79j&DCItJA#nnp1o%VCA$%XrHcK_-_y`k!6x*}ktzoTM?kzrF zbCO|hNkU|^*y6w!gbKK4l^y(9AdFmPTK(dfw{MxcQSRGbuR0o&n^A0nTe7P8HC04V zaOez|_)HJWQYGJWc86(CZ7v8(s%3gH5mdbeEDuD&|uYs2ZwT5nR^h2Bf^{WGR~>#V+C2Qx82EJYMxsi zsan}XZnb0Mn-=#Z&m@A7a?0#qiHDq?Y@2^+l51ffyN3Vah%IHTve#-sz%r{Z=6ROr|8Rzv zmvKN~AH$NRI#odpx%OseCtEWxRke)M=ZnAM?e>tK4u6ja&KzxW9NC-#+1(6#?{Ws4 zZ&19W#BVT^BM!1V5)3Pt2ls{Y;U}F#sh?);jg3aGu*&3_9VHI;;O#XYk5LnD5Q~E6 z(`Iqb!I3Dlea5cY({nNE5v~WU#;#p9G(QPdevXrc%=pov{wqSsTf4a9NDQhGJ6Y`_ z$ksq+TJoHT@$VQ!C;4FzS1{+Elib$~#QC0xv|JsYPw_dfRBhZEmY_JeH1*D(W|jOc zqGC~`rp5(D{-lHIJMh9%21n-KGg^#+kCi=vL~&;OHe!xS4*Jz@JiEPVATa|+yh+gzXFMEN zbikSAkddmO`@v_gVT6FCpPK(z^+HSymy@}&k!s%z8k&qjn5sNtxj}dEPazzLoyrN) zxjb%Xkms=QKH6dF$4+&-j~f9kQmt7c%h5c={Z74UBs9)A1ow~7}j!~?3r z%N1wH9EjVfWWLbbWPV;ac9S|tIfsGe0{zd3Oj?&re!yPv={?095o^8R9DL5ri_C2p4uNR`p*mb!O8fNWR3+7+8S z$)3E^04{LTHFGtFxp-O#g?%(up|XWXw`N#4yY?9&7xWP9)gtc1%y~Qg;fWl=M&>lU z&Gx4TW^oHp6RwskUOTYj=6H7-s$9@!zDKG(@kh`-^9qoOI7F8Z+X9u59H?GWAKg_X z*Wc671s(>p>D?WCUPxh zrdVK}^>pGUOLDXWbUuO}gKpDaliamEQ85@|QNwAQMRTiji&?#R5{&wVQbW_GdvX$p z@!$?_6DiOgZ@bF1iP0*T+?-57Cv|5BhX}cUIgvAu=G^_JFPEwG{Pl% zdaRNDItl$K3&ob&Ue`NPxwm^{^kfsza`D#U6pUV$US=WfCsKNC!#s5}8X{tuonykg zrYFg&&T(dbPIW03v-QYg@xim|dojqR&< zCP}ujV*9n^A;pEh4H5#YL*0Vkzsl$G@K z-{^d3F2|UMiEWYptQp2-j|{Pnirfcd1Yij5=X!04cHjKExC;3EOb=00bqWun>Qw+a zK*qll`w+K2?#EvI96%qULgOOVo;ngco>x*siR+0A5@aTN62ZqP&t+ysZ#)O1GJ6T5 zu(Le4LW7%F$%wEFgcLjT&Q`X?#fz`F2;%zM^Nbwa#;JZ$5!1jw1YkxLLhORr%I#vq zyF|g*q#nP)H^xU)X@28HgtBuQdm-6uzEu15r(BJsebuRMU;89$B$UL4+WHCB*|?nt z>Z~KC`0J-fPTKPz>*Yuh2Is{_A;tiw#tL=DkBSyz;o*Me=+}I7;z3y_M(xV{5zy-- zd$fSQlZZMR&o?B;%^-pDLwrp+OfBWeRE^w239mgE+aPaXqyJi%z^2_z9}}{v7GHT# zCS{cd?uO%=;W2yNIadv_FvaQlM38bE@l^@8uNH2%VIJOL*QEXdUv@~tbVajI0!*tH zPe8mq?5VS8GK~oDs+*l|jZ>EprgY%~H&lzR*-~t}aO8!3xvHi54O~~t3arKeS>6{+ zcErWTco-Ua(VNlF4NofUT_-TtmEe92U)X%h9>pqiupnZI#-eX4VuoQ7@!Uq21b4^2 zr-9iXDquzwOGP3yHehVIO-@1GIzpu1O7APtR@hE;_hB@S`x&1xp5r{;fjb`xDXq;V zL-=oio9wC&q(vaPrqDhU)#a9zhx{BYnv@)2UDEZlAG!EogwGUS`IQxSC}ZH3v|`*> z;&Lc_ss8NdSlx0CB~@cAhO_9)M*tgkLH0D? z%!W(qsY5@oYMd}k>ZCTpKu~43L@A{7@4QiHTez}XZqz%A%9G*a29x>OnCgCwz>{Rj zNu<5_71%i&h`Y~4p?MOaU#hFP)pfg*pFA~X-IC{e)K^izZuWXgma_v-&u78Hi2v$! zbD7-QG&vK>2ka1`$C!RL)q1 z^flD4b=#6FUk525Y(6?~6yr$p9Bda?rsc1y&%>=Z(rT@&(peNq5F;nce;pcAG1PwJ z1j;O|lq3x=S+{54yDT6FP;|}s`vuGRiqxkdS{6>CqG<@8mWsuQdvx1UE{EHQcf^Rvclx`ifn+p=8rtRN?; z2f|oA*=@tS7>RMNeGjSc#uGw46ApvTGi!9?8B4UI=A%B1QA2o#G9-%C0hb;n=3j=K z8q7^$mmoLsEkVGcV^e<-5`^3V?>`?R(3f>H7AIdtsd2^XXt<_|Q&XM249zRz@%)ET zfvtL#l7RLfMdz{PC=f)!U&14fc<((oy!W23pPs|U#>UP_sHv)#DT!vw3{eew#9Oel zzC|sfaZB{#$AASFiV+7$q(L4%@N!}tQhbYPsxH|7@8!akwl*)vC%pB0mI#*I1TbO^ zHGR>I9aJNm?kuD7nm0RNI20HDy1~fMo0$>Q2l?cf}~?gDdaoqj6n#R~Dk)LbWO6npsC) zJK~mp7rZ!7i*#c%ENqvXEW&(wMd}9-p#>0G8Q3~>2D5|+Cx1WLUl=_V-NI}>3F5{= zh3YmOqFs>FrJ3RHR$qBS1eZKEpta8@2i^v2mPALdF0;=f{=}bRS}-fs*Hy7`BICNQ zpCJ&=@$B-4^Q3J4ZZvJgT>K_I?m)_#u&rY=Jp(ol+MuEn_-lQSljbg84@KV5-OW_MoV85)s(WyB~_bSasvt z^pM#$M2;KlF{uGN=H@;3li&PQvU`9;wvgp=6t3HBW%hhkCaq1h)CNBTFf@WBJR1C# zTSv~d?4+#Y`-LfMDO*r2k1Y7)%9ZIAytTbO4(7II%jf$5W%<35l6*a%na1{GGa9e4 zh-oNY9Y!#zT`d>Bs!#I`*ShBJT*gM;Dufz;a?_VeyG8Q2Sp;NlQ%o=dgL9>}2oa|Z za_Z+f0;BWg++oP}Lm+j()+__qJ_|*WzwF@LdsIZ3gX);dBr%Wp)Bu{~2O-~0YkJ3Re0=Hu;3&l^RnsK7_@7*11T==sg(uCH?-N-4>he*M{& zIO0tm^ztk2d8m+55hsvZ;iVn=l4m_n-)+2O!_%-%ACgQt`>vSw2DwRJJ}!+)&x$v| zllJLs`6x2)b?_$>xAS?(Uxq9Ls$D$4rso@m zOchI=4AryoQ3by=vdp3Xc76@qetIyuMtsi~%# z!MR|hC-(n)tc;%mP6E7J2^%~@1XtQKx4urnc@j44W6W;9>y6fvTqrPVV6pmHVC~N zXF(CYXyy-b1Zf%|VbZ+RIu5K3*t65rs6l4@(Af(<20wW(Qd`8U5|s0*ieLDdl+fwA zf2L@cTbUliW?-+j1{DV)+n4~`!0>8=NK|eX@VXOZ(!=T_L+b-s3`Pkx`0!oOM1+mm zJRlik+3y5@T>Nm^iWQwmoZOn~j;#KE-o8pNyCav(d);KikweBl8r04tmDtig0BaE= zvJ#>jgyk0Gy&nby3!g2Kdm~|-875@)JGIKKdd#T@NLXb#9BoV25s`0e;S|$|eO(4* zw6x|}GUf%>kUyhIaj}yLD_ipaJHT4@6SgWs6JN^fBKii^rFkxQztpt^{$Q{7XwJj( z#P5`KCOl-~7USXG3J(;JP8450s|kfAc*aiJ3WhW0!6qb)z-RS9T9jbFprnKJ*D)*- z^Y<@C6Ji9$qTl&YGv=DJ!(!s<6iG^W_n6Jax(FS)7Gt*Ocmecdsx*bz&@p*aa_`n`Ey5_1d% zR;DneGH9%u)ZAw|-6_5?@I*bYVx)1-K9;?~GoP;1dh+(roWSh1ZN~zYW*Y3Ai{FQ9=`~7OFp=I`%hYsh298p&cEKvf%ar>Z@mscy9AGVFrF~&oV{-FHG?%l_ylHdIkmYJoQ-x7Ed_W!PYl7<#v;# zn4Ngd3#`~L82Q7A%%#Hrw=(qzbej8m`qkF&n35=5DVppolW^*7uV@b~vJE#qXo@T} zN0g3Ze7tXFPrZaO)o{ks=er(o%PJ;IvJW42nNOdH8mH zuag}^H;U$;w;tM-XOKO^VE4^4zxH2x0-5Ey$Euuh% z*5IU&16zgz!)=CC8jIVOxTTK}E80|j1ez?S`sInDn0K?$2l~V;1^w`PoTWzO{^BS=wm{gMMp-EU?-?4^; z_dJ9P3nhLp#i(-idrd05ZCN?&+W^>ao0_)&G~>{75~)tAtlRnbcvekd$;ZXWP-&9| z9HVAbycr01P{U1q^zC73D~K_ouj8}yj!4@l5U# zXWK_81mvRp(idwObXem3vNv(+Qk`2_=++u6*=rw;&K_5vYZPkx?k@n-_%-qMznFYR0zyggW;jV1T_SD(S**njZ$wXWHOLtJ{;lp`S$=7 zd(4o%Qu=&!TJR)f0B{EP=>tRGrF^1QBiXMZv)#}_Btc23xtnn_q*I&{q4dTEU(wm& z{Wx{ZC1gkp51eE86kLs~8WLsMPm$Ol|!M zNmGd9GLn5Olww6|8qndLQ&zMhyb<*Y&&bv%)iFA)|F7luIjbpufr=7YLyv>V)H6M% z%FcbwZ~Cc*f=Y=TN$9qm%3zchb?E-g+p;_SYUdo=)iQ_?lS?^npi1t9A8}9q=LZM} zJ8Qtms~t;?VBAcLS+@aKtJOA;`6+H2lRm@AZivkniD=cCH(vFAL!)nCYTL;#uF`Xh zGz>uTLBU;#CpJx`aCGq$q2hp;Qf5 zNZ;V>=8$(HmWMolCQEmU{eqHXb(U{*&N9EPbTfkSFz&x5sC^XKf4cs$7qyY0KWExq zn_(emFF2C^F{>c#45teF)vvo%QN7gGP|=5EBwI1i#8i8EXUyD7G3WjmW>Q^&P#Ny0 zTCgaE$WVJPib8>(A9{Mbh2oeEkDZJntJ8Qvx3_s2f}4FGr&nebi+ml2DyfPu)XEBd z=oxRvDh1x^?7X?uc6=GvhAjE95^?B2=oBEx>~dK!Rvaedv}1=4+~6iwTF_QJ89bwz zW~)X*jxd}+?Xs4dqBln_G6yOBYzM`9Y1O5P=%t#IeNCP|hD5?;x0nH&LtY4&*H|_o*PStOF7=6s?+^|8Yw%#z@ z=oclk&1yKgiogfhg@a#85bfl0`Yq2%GWv1l{>CkzjhP8w={J{^LbkDW%az$^d^5LU zNTPl}IGEEax4^)yB#Eg;7y9reWwJr6GYGv$4+r}`jM+VgDb^tmTKt$`ThJ*?A#l_p z6up|>ctuXrXn%xlCQ;&`_CU()jxglEZo zYbp_aWM^9uH3@M$%d4fYHEjIi*@}vxlebn@eSZ?R>hhI9a}BdBW4BrmKmnDH;o#j;B;Hm&D_Lr2N{f-4#v*w-24UT zRZ7H6qj%*a@%#ePMm}UuW?SU%ga>o4#_a zz%q&V615<25AN+2b3a|*di(7*-z&rg`_Q0X8rg#ZL(Q}DggL*xbfs#d?r-la127k z!B;)#`i|)nXW!~v$XcN4p*(x~!$?pIN zoYHM*(ru*ER)6oqNEJQ0s12>~gZKQj*{F8iHt5btkJeNj4mwZXX5uCSM$uWaRslEf;`$Q#+(cf}9$tLOJ2Q}jX|IT!VJ?HNGkr9_ zh5Bh2EUkg*o|_Z-HwH8CabB$=gF3C{*iO(BUSS%XAH!7-C~9^BgRi)y`4TIDBCJu5 zvwPj@RpMR9)eRQJTB`EhJlw-#AtCJHuIMp&!>MaLUW^)`F*)iK&x*|Es#Ej1h^A0> zYaM*f73kg_1<=~L|x(%kApsP_n3M*cM{G4YDa~)K5CmUzgW9r+%e@naL(6ztvIhaUm zxmQCG=>eL$%%Htp_hCp2suSNN^#cy`H6Faq6H68$$C$aGMwLwmVifZ}8UsL9d$@1c zj^pb+ERjt(Z<0x5lktO)Dy34j4p{{rt9a# zxeX!z-+8>2icVVROnByqen&{{h(A#~EPXP(|0=9IToU@<=(6+bU-w-7T+LzQW!Z2R zp2co+kYa2h7Xt`{HH$?WrBPNWWaT3oK39tm+E5&NsCf1OJI^3eka@x5STnK>gWT(? z(u!o@H?McCq^%j9l%#`QQfI)RAz*Fzq3~$HePog$LVn?Rl{>EdbrYT8V>RN<*c&*f5K!xp{^qEvsqr{s|Cvy&F3Vk~+c~ z|F9FBNaUdC_xGC&u}z06F4`-Mq5X$G1N8yd|LM?iF;aJQox#Bae^zZTC;{_A#l%GJ zxVR@n28)*?4N2E!;qKA8%jO#7{C6HK77SdKghay2IwxqO-kSP(y(FDE?`L-@aYFD7 zi}#RE%Z$T~he*l23Ha}N+B8`v{fNZ-lq-b|DKPIyE++LlRMP-d+sf& z<}0|7OX_9sZo{0)9qbh6@6cRPUSWk#myPbNaE1YsW zt9NJj&!v1>$4h#uWERS!FNJTyQP?i;Q1!0hzW^f+lQp1a1X6#DneqGD(NpuivFYja zwj#gv*O|}sBH)Y|8)S)9>RfCT8W+OHni|C9)Ih22$NkD#823dX0 z*4FYVTJ@RxIBzYqA&A=I>dIrqO&waRzI`5~9(7yy{m8IbSjPK2ynwP!;{m49wK3c) ztMiVA&*%J%BwC|7xes9AoR7j=y_ur-Wwd!Txcjh2?V*L;O9w7FtjQ5H@d3S90G_rp z2msVW#=Q96Xp)e3y?5e91RBGk7jg&Qyk8*q9Wlu-o0LL?P$m)K1!XJ~;&l0Z-q4fkf` z3vbu%&E$=zVb0{Dxg8)x+{UmVxlR$Q`-M2mToHb1%7*Qr{KQ7_V0R)LS`e#LuoTjF z7eH&CT}F{vh|nO&d_X#A&IQzUlZGwUI$7}PTm0nST1hO3TVl3S8^F_(HN8*E)6KQ>3wGuN?VHQcX zj{K&xlvr$7#8L(#x;<>mjtrX2iX~wb%PP4P%xlK{LvG^u%{tF)@eedN1tcwnD^`C| z|D0+qc^#bRoa@8A+gD~EP;7&6d|2bPv561Yu7;HIpvy0Hi&1^|&)NBX2ZPIY9j593 z8}*V7NhQanV^Oi0o?=T*DFVzQ>z<$1T>N@YONzw`2u?WH?4m6Ei7X+l7&T;S7R+^& zX5^zCHdjIGqxh&-MO>DE$8C-jEwdK*+Dhdc{eBeBYHT6z+Q96YX1acHwnLiF8*F*& z=A3bS4Wb72t*L05&M39$D8eKuSD4>EY6@E12vV#?SE+5fNZ#-jtvtwqK!z`3F-1SL zn0@-z$me3R=hgxTvmD#0y?Ef6p*iH)vh8fq15l15}KM)kteGrT}E9klo!}>jz z@vm0~`v+W64vsaL5>%{<;a1is@-6S6JCB@yTl<)o&DqIkgsp`U#wa(o_1F!JJuW3x zo64kh_jF&{%qqv#JeI`+AzB*kz+IM5c>CCLxSIJ4glNoCF ztX#p4p+g`V2y!qkqy2Iic5vYOay?(K)#306v^Obcp@4SL^rUXF$66d+L?kx(8eKmDi!5>N`=axEybooj%H`phfY->S-k)F zx46Uhs1SSjTo?~<92QJW2cZIj-U%0J8DGR%?G*5$EnFxxd5iXVq7xc%%kO4jRy^;V zfu6D(F=Q5YCA1))!7waKlvGNYxAAf~#d6pP=(1^^k+XsjbEL92@wb9SenVN%J*k&U za-4%PpAP42KD{J~P2(y415$L2As!H|)}z}M0YO20z{E}nb42_Om{T?FGF*Iz?AKMB zc82DzcdqsR=DUfxq5+@qLnyCHDkgwUQS1B0%z9NvbyjvaLmUDtCSZxi{qWIrX$ylEnE3|cMPV?v3LhGwzv>AkJlCfPH3%p#E{pFUqOQ$b z^X^b_&?uenXzc&@z)K|@!hf`98*$k0HF+R3;~H|$;P2{SZIqww@k01VMKV;TBom# zV22~!>1M^@6w^i9xz*(x&2ht1IE)_Y6FH102&)gK*;jB!#aDq17{~i&>iEGI2U3V| z_#jLjZ|^aKJ`8Jrf!ziYFkl-_sIZh^xIyQqw$I~{5NivIxNId&lfXn&tQsL^L(o>b z&wp&fheN>5<~}p*Fq2LrJ^i*92b&k=UWetVwG7j;tnn%yC+P}%5NZdPCxfj>(^q^``)1LY5d^_#Om=v6)SyykJveKAQ@vTNn3al{QyDL^I>I zl+wn_1HCHi)?Q=HV;n9 z^f6;yGCSOv5{och<5loxJNvIBU3T14Zr?7;(;h(hIF|}}VV*l$2{vb<6wbfVs+|gFpR=~YeSXiIfU4Gy%xzEO zPhVc+{@#2D2?Oxy_lCuGa-NsBI{zgDnE91v!+h3iw(@ReNW=^;vZExq3O}l~k;kOv zb=c)ZaDD1mcZp%N%?^9+IDhk#BgoBsqQUPsY{*xS%$1wT%pH|mj)i#X4&D`%?gaNe zHF4iw`^rL2RDh;DlgPq!U&>LsV7ni0U43ptHg((1=>@DSA!PaL(Vba=&BL=anhW>7 zl}^HJ63Ny^>!XfHwEzQnZM{ah)iXl4*L;&TE+GDP-=_N#OXCh=lB_yy&7HEExI6e?J6QeKraG4%7Iwg9%FVCD1o?1+& zJM{iooTE|Xd4^>@&$QyO)A+83VLqwq8z@Nv05vgJ(CR8sQrD=~{?*jrUlC=@d5sE< zKUTciAF>blq%tFoi26+Fmwv(cC&7#Ebli?Ytf)3_h6A?3iA}REwk)C8vq~jplb5B& zXm6oG;~<|pK6}MEPNXy7Lx9eO`Y_M!7O*Wx_pt&5gQs}!g zOfS=C}yL`w;B3(5)8(^sJ9yN?FDlvS|zO5?U{3D1owf*g`XUN z2hh(xz(mJ#F$*?LWl94bJsHx?UbJH$Q@D3Ptrt@^kRPEPj4W^FjTW|0@3vNuhpaP^ zOMW>{KbM{?whs+3zJS=0~edm z%H}2Ur_QODS!y~5r_zcs_)*1X@xf>V7rl_=Sl^Vg8J!e4TV~$;+2De5D+jOMA|)8$ zQ`ED-)$OdZt{h|TE5VgZXtJZpZ&%Wrk{=l3VcLG}RUz&#&4Lb_mhkZ%vSToU{y8)D zP(kQ?P6{u}5%(UaTH@qW+Vd46S7%PFJ+O?9KOfG|dN>z3(tVVB2{dIn5iWEP!7OBZ zw+*>1X}Uq|9;%2$7fdZP0*rvI;zL7xeN*X&93nyVqmkK&-L3|g8wsng6q{q!7mlXS z@0jdnoo4B7@#Iw`kxNvk%`|Zmu90V1I02P;bup>wz<9Z}Y8CFM>b|`C11w(Ad?NYC z0=B6Gz3D-2!buD0-6TU9+3^LtpFY}-VppxfM|ww1enl-iGe_?|@ZC_DeKh=L&cjV_ z%skIEC3g_0SvM>&I<4{q87*QZgMF9>HT+XjZ&6dG=DvgEI*aM1Bf7*gQRD9+&246L z0dGngYF$#|YL|7JwdMbG8qr#rq4FZH`cji~tTusE6i1m~X?u#688w#??a zL|aFt0g+wB@R(Rt*ylil*rahLUu0no`f>h`{B1Pucr%+FB9pPm`7#2=X?SNG%X}T& zh2SK%=-t^$9tkgfxH#j_SSGDR$ZKh3RLhH5SHz{%*ycXs-p7 zkmuiB<)u;2i$ha@`4pLSkz{W2_gHTew5KCp!N9a9qB(TMxLJ~#@y8Vo;IB=V$H%iDSX6zn%SOFxvGm(M-5P~;gMS=r3dPMwhxFk+5?e=#A6tlC9^|+}%7;c-J0?2d5hcS(qfd^T?scMeURC z6#jdo)#^Dw{3_5*u~paqogqfE;CtM^lTaEom!q9P_;Jq(qKw}vM+w~XhYYxvxG2zy zg$a668X`aiX89Xnjbj$Ooy&`$hGSSQ>gyu<97+EQ{bjDvo`rvRFB!77lpSx`38mNK z>7f>yPhHxht0E?XpI;X=BU8pO#I7urU6-NlWCVuiybNE*ZkycXX<=ekzorbLaodoR zSN8uMP^M|a<3{tVN5C&+axs?r`6tde1_@0CNBt9aEh zMkxrR`&;)49VzPL3_54cbCUM=RCH1UK2BU4qq4TPK>{OsISoDh@9VH2)2s&jT&}U} zcCXXBm%EE+t`hVlFJPUYUoiqawgIjYDePZu%L(ZjWht0(UDQU#8#P0>P@!?6XNdlWEdyks)J}`Q>MJ5+Vx0+@4qtL{OYN z>C#i^ib1oJ#LeG~HaZh(IPN$qyuYuI_|f`uubqzpIhq-a82mMi1$wFFy()=#Z>G#j z2j_pP+1^#*uTi+oNuJ%u=302sE7|;uYa8>9uT#+ho z@Ft!-CZqMG8R0r*cA<2<;cD%&9htv8_$Ya_`^#)mchwS`S*YX6;UiQnm83tb{^T_^ z^8sd)y$@TVDs%scG}Dh*`Elv@q6T`014?auQZx>YhQB+Sk;dBO!wa#ZgZA!6jSVae z;r}O1{wQKQcQnsr$5<a^UVf8lIx5KfIX=Jtc5m;~`}i+`<%D-Gf8_ zrYY%Jte@Gb#e#Z(9m?CL@=Bc)-C?$!y83Dp)q2+Sz!oJxN zP5~#WF3Ra_Q#kt%v-NImpDLPQr00M`! zK)J=>gA>Dmsll6J{>XkqJ9kikNoYTHbRXby8PT=rU*rOGP}_zrp!^jnm-02QysC-b z7}#1zX7j=!Nrx()RiV;23s0p4W3=W=Rvsk|1G7<1N`2MYBI(q2H1T=O&;Omgm_=gj zVA@lM6`>`N;&ulOWaQ;qq@UZL8mOmoU8@mmtL2Qd&D0lh3a1O(g|ZA`{BbUQY(MSE z$uC2@zHj7{@TslTYdY4#h$8L7+8S_Pb3V;a^vVpaF8iMpqv_tS%KQvmT!Q2Pg)N<| zxBeoX>Yg(Dk>ebDCmtgwW9FwP9+r1ft}8w1I?M{=z(O|PVI1ScN@0YN>kACB0Qu6% z=!s5~Ws~*;vlC$2#egNkqwUFv-o!N$mz!m6pttLo+_Pb{f`)OzLHNHdy+tR3aQCGr zD^Zelsr+Ji!3TkXtcJ5mFnwW)xsJobn8u}rJXTGCJ}P>9G&E9G?i9mj@&3d*V+zMN zx={>-={qOYs0U2xm2h{;V*w~CiRChi+*3*mIUksKxBSg!ffd&6fUm*eZ-vwZM|vEI zSZJ_S6C*Pm6X&Sd;Csw(#{4GeW!v4TPo+yexrFcDD(yZFLj3fW$z8P0I2ya`-l7)i zORu4MaBoNx6HaJ{YQ8$nRi_s&$e8crEG%h1V5;~)hiS|u-kPqeW0kTk`Cv2ZCKQak zByOiOF!rYXUUi7L((oLjv43MnlnLY@g{vxYKvl*4SlDirt@a?KCaiB1=S%DI)FHQB zso`?&cDZ;U%zoYoWv>*`7SZH79SeN`*=j-huG~Wy2r^_Y#wb#6(Mf$XMU|u zax?zuIY@>DOK!5uI1a&jb;(g6hw%mi(e12-$)#ZN=h7`7OiP+5t}J>bQrtf?1jl`=!L#zKG;-}YTbjCl}a-hhKw7xF?UWNzb~G) zA^A=owP8FUci3=gIWR~y>ZaB1M6BL-UMfY5i(xb(S>zTBoKO4J-lcldx9c9R*7RT8 zrY~jOO>AI$K_i9$SFo%94d?u`57Yg`rMU@&?lCKQkX*ACni$#2simJ7xY?M0Klx)A zX_~5WJr;UxQKc5C#dVrnr%$WI;)7Cl}x}UX>+Nv9rfkx|d5CM-sHIf?OyL2j@4L3%?XpHdExMksM9W zVP8x;kp3Sky!<$azma<9-<(cbI@;TBW@ACJ$qujrx$ZC1t~$}GSi;7?-P|rj8LJ-B z2WQi-^5E3FTkt~~w})yU%e&y+IVE>HKDKF;#9Y*x3Urr{5=W^$YH4G|9)sHJGfuys zZF)Z`K<)J6|1YSgvxz~KUW?*P>_o9~5B(>q>u1aN6YnX88x}XxTFMN1$!LbQZsh}< z{@9f|BZ&!4Xc1w(O3cK?p(p4Dip=*!`H%j3 z>?>T^i46%tREK#B5=|xER$bAB<6tF;+q9<6(V#P{!ZT~K3FWfY0u}>Bm9*y{6)TnI ztFq~6wNw8Lb;#d1Zy3xRNyukdJP=?qGC6e9C4A5up)MTioJRzCRBuD7SEPHve|A6E zJ);~WX!^ey#PRb~=-Q!bbafeUd0HP6A}3TNjVC9k*8%)GYD_sULl>V>3Zh0^HwpBX zSGt$UmcW;K$7Lx@-&x1Fr^BbuW&APAxJ5m=o82c#1eGRPTUs0Y3dCF&Wc@`p!Yv_BXc3V+O#s?9I z^j!kJA7g1D{iGBN$99^}${0d|7X*rOoH3+ZQru!PQt$AchPCTUq4q<%xN&Zsh|`kf zclm`ro+ufP1n0&xQ?};ZC7t@p+kWNmUQb!gFT4-A7vui-ln}7FV~{WW0qdHXdh2i; zWVv)!*7qw`(NbSRDwah#C1H*>its1(Dj5qeg=KryY^&&^m~a{6I+=0YP~w*#=R8d( zty_1tgHmXOnUV|oj;AR63K`>XiBli4`=}^Oey&euKWX(;l?q}}g^$^CDXJ{OT!7uo z!vKqp9{D}@j;?N+&{>@4ky`EXG;7SmtcB&sWG75dYnQGGpX-YkpsU|jQABm$W)$VF zFLqDe>3rDB`z7_1qgby0cWIpB9(`PJcvk7A1NL!@zZ!QO9Es6fGKZ6*r@vM>r6Gum zW-IB*l}x$uHVEjl>uR1t{8%jPe^KR}{lnhm)DBG#n>v>7^??t>fTTMth@#D`D zQPPfjx3c=n=*Pn8nXk@uRivQd=DThEWz$H<7Hqd;vs`~YUm`r%zr`Qs5%VH~rmN{0 zQdcU8yYO|+k3a7@3|0@`CESL?e4{nZM~=*I!|sjc4?T!pvxTL%aE-bs>Ni5nbeFfM zn(LKy(H2h=%7`1Yx^Om)#C}&BDgAUcer4ku4%>h=LR@f+6zI1~wV$Kz$*TM?;w{GA z;E%GSDWoxbWmt(&M@bZ3=&IL2Fy_B@HlCVWWgA$Etc)AgIdCG$d<;QZ znbN99%}Z56T&L*(w+~U-W^Q%v>Jr83C_VH&yd$ELYVamL$(1)A95B9&Ciu~oA^!$= zAW&iak&7^`dDvK}73c>R4sGsRawW$c{sWaPcY+HTqbj~@3r=GVg$E}ml_HS-|017w zDbvphddH_LdPkw51uIyiGAhwId$8D6yB5qOsdL57?!0JvPo~#yAVV z>@)3YxKFlqU00A`)D(A$$#{zsYlAe8!ZV-`S!`$}9De{U{Ee5*&zEGfv|@m|?}X#A~b9u5_N%nBL~lNoQsN z27q+Z3WJ8tEO}0j4n*+(iTrH~%BFit3|Y+n6hmXo5!22Tccdj%)0Gvdzt<^vxRWk< z4Os0*a(-%|%2vOdlnpv&5*Gts!All2jGheSJatw$oA=35v3-HS4KA9&qKy)kZE1MT zI*TIo8TvC)V%wg<+igYPZhazx(8Az=5jrw~v9}7lHmG3KhSL~!fZmLAoU;pdU9$FVIfV>H#g z@>C|U2&%1uFtgXal3Sw7q*p25h*bG_NY5sCdV`fF!;`F7X00pYE8i)`iOR9lA zI6UA5GFKz!e!tdS9wcn^b0Os<1cd$l$f88hn=PwRz#?T z9iPLlGHK{2&#R>uSe~f3QrqpPjI5rlc6j7y0bvu7q+E?L!Jr}$QMe0<$+M1|7f;sj zCJyNm#)UfTHnRP)_lj#x4rEHJv1uj`3BipopOFAK^bzF7MY9|w35(qsz`BOpGQOih z7I#4fC4*N{3?Aqi$?gB{h$IpG5J%4ydPR0#+^mfQ_9S*B!fxts_vv?VF%8Us6&dpTCozu$PPy*pli6Fq{7y?Za2ulNJ5KL7d9HwEQ|?R zyGPeuYjQG%gJh;w?e(1ZcM2jVUlVP#Jz|$zs|MpiY^;POz@lBlJu0Ir(H84Z}k+$L*X}>W9VW@5CF6# ze@|6xuSX($ISbVoaGGbpn2!z-$)8(%z7LCHHlK=-5Ug-h>7Q633B^Jc)L&o44&|g1 zL6K5p`oyx@nBI!KR9nPTisbpV@+L#bGlknlm5b-?Ww36{J_j(a=ggOzPAEg2A(u6Q z;17+6-cW&#;A~?g$bsf(Sb_lPwQ`~i>h@ddNGhbYY)O5F>W~V7l;D9)2ewW%efE6n z0Ld`il~Zf(&9>~xr-KZqs(SBmf|umdH|QEKeTgO4Tx8n(k?}|hIPS=Rz@qOmm$?_F z1-!YaZ3^F%QZlb%u_86`v*KOnk-X4ANJ)yeIK0QD3rJxQIx=ucpi`=4kKtorw7RaB zQ>CGTnIUzEk^X`wwkHz)VuO2ZHx|#OZr_?oJpAs5SM)4*>xkl^Fk$Nle`5SNav-_Q zRhr0m7E<>wm}-`-3HlX(PwW_xc^d{H zS(@Vlpa#28h5+D02yKzrhGDh<9Y) zt!V^!d3#TKQ!`UZbdUTV6N_~BXkzm`Y#Ax~7j?I}&Yl?YO;i^-I(NLk^{Z!yV9Y+C zxRb&_By#iIK_tq1{j_Qp z^4SE*nNpA8SCMjj^Nw7SqvtlBiZiI!kJ8X0{i6Q!*J%$A?_Eu@V#7A=IOy!xjhU}= z$iLjsWdAMpNV{uBhsMR7c1uc1m+UAIxtXps$IDqGiny}Q>)y74lSPvTF7;8cy=9Vl zd!_5AT%JKte6hY|BVu144SnKA1>L0T0~G|=cgllosyr)sIY|artH~XgCaIo{q}HF4AR7H4~cq8CPu*jLT2ETaa1t>B?`=j(x1K_xR3T)371q zXacr?_ob9QITiBcQ(PbBsT(KD!)O~zc8C>?q5I;xCc0IR$ywIzs4YhlY&cOAu=&_p zh#2)Gr|A4qVqu%mk!HNXW$3_4g=`J^l*vCUi;o41@UfM)3vC8_8059xr7%aeA5Haf zmJ}txyQrPJUa6;AUj!uMSHFk&D(Jd8;|!%rO)Cm;cK$XZylsx9v)EW`xXpdP}<8RW?c0>tfJizs0zQ7aj~cVwjuEL1O`E? z3ry%Tbubo)obf){M0cU8fe~2q*pBku)rOl1H}#Wo7=tkVQFR8yAL#?#1U0*3cu!^a z951ijvaBkPYIg>H?s6h;i5N@(=^`L?T5yGbZAI`t4)TJWv0 zk=PhVWxTrFQg4S-&YYPvndpcH!8jC^Y!=o2OsGF!NR%#jth|*3jAts~d=7I>V3-_e z6JYr?<)bMfvZk}4`&CqyI`5Ijiff&)jl z{RVC#VnT&ntj==*{r1D-at682U1AZRO4Z6j>UZSM9FoI?D(eonoVg#|(1Q!)%{X>S z=-TPTz{*e|72_1c%WW!@C2(-6%GT>E(|vEMYrJG3GIRVHY7Fx0q^Uy1k$$900sMuRTh1FB zA(?E3C8SLKPagt(wi#1b(2nCgaGrIzQUZ;WEO@5z-#lEsyC6$OtVSjVSz0o?%Qw<& zA<1$0xIVDdV$YRAV48K&qqwb}A{H4f6Od}$VS=Ke7C7?Jn+p!%X_@^HH2Bl*bQE(ruP9sGDUpy;BvzU8qi2(*v`Y9M;_)M`b{_kEDAU9Dfn;<}oOo z-tR>yrs{kJXN1O`%-xUsk+U<_t}(GG!$+r(^<&xSyWT9uh`qNaDdlhG@lsM^&9SUA zrW)Z3)DN{Qgo3tc>j`$m>M0LUnZZ2=d)p;mY)uMN%7I$$h^`->zL5-{>8-`r(8tCO>VI2o0LNQOY9aD$RyAKQ=CgP!GRjgB;c(W3f7hBgp;fb zIJCW9m30~Mxi|&JG>1IM*V1Ix5%lbgCu0v#dGt8MAGY;-Qe)5Zh}a5 ziotWxk8)5WbcgihnFSf~MjUHtDk5P%6Hi4wpY$x?cvc!dwh z<~nM}DE2{_FqOCAIO9JO@a2xvchAwkN^i)?N7$uUt9+uxXvA_xkg8-ihq|$hN;%m( zMq%mFTwLsWEniES2`);xTw0E8Yq!Snp%1Dn6po1HW4`V(!A4<6O=*LsJjZ`n0mVW_ z3+Xf(+-XJyy*%sqq_(=6S(4+3R%=w*plph3gj|}_`=)7#L+tn%h@l!G$`?*sm5-;W zX>o7pXO6@txHKmt&l}9?4CHlH6RESLt>jmI%ET(0df&S1@>C8t4cU{sVm1p(1`=b& z66K$cU;026*>eoln)V3KQBhawY)e8P359MN$TfPfn9E`8v7qX6KrhtZGkq>Fx@eW} zmx?k`;+oabC%z+?*$>Xn+5NS4;e0xhge2R4N?I;7Q!aiMr?E6OIa3Mvy^(H?6(p!HQ31 zL@Mrhwe*eL1%G4*|55qX?d+$fdHtkB92w7T&De%t{yPyl%t%!PAB=AY1pm`Wh3fWU zf_{}<^%k$A83X~(GXC$7<6*ekA<vZ~Ne+>m+gf zvQzD+xmox1gKbqn!ZApO2np1-B4Sk*0o#O1-SpX#fHj=wUO*561bfBLhul&OTti94 zaa5I^$S>F(=WCcw>q8jdnJqaxN6M7UqHyAzMG?qGfGn+lVhKdTMcJ|WzN(cz zi1~|8gHSJys}QdSa1Did2Hrtfc*la;Q(`sNb|UykBBGpCOwZ4FLx1)G`P_z{_Rs3#wk zimpaVxJ{6IWP3b+4N5lIVH9w^%fJ?wAaSJ?S)I!<>@DTCDpkC7#ALO@cqj;?IZvRLO+J|T3fKn7q(gbvx8k?f^XN8zkyS@pmKJVmEZy5>1dZFFF&Q3wmg z=u9;yBqtM;l@|=(Y1UW3NFeaXPfDe!hq+>r4~;6yYhmfHnG(H@>)KhXtGtuGoh1!P z@e6m-$vsE+4f>+JdcsL^u2$bP=nZRO7oHATR^arwY}rS30GmXyLT}sdsoV1lcGa@F z6g$n=r35qn8Hf|_IwH^As0H$(Si-|H;E&Vq?RADX1-l!jW6pPyYk-RUsAwdv4dqUj z%9iFvnF(N^hawHa8if03{onXyR~ou$lZe~xPTxYMA@1k@b8q9wu6ol-whV9#P-gzz zxhg}`L41zaRo7_;u;d~7-1{W%iT0$PWA!_>z2O|6Yth4yKN1+OoABeHjk|@GkRaqr zIqA7?V5ioGc%#&bjRbCgEVr$k0N8ZH4Z4shQZCP|S$Jag0zxC&KeO5L{U9gXkdO7U zTVFFsZ23`%9CerJ^=%GGqD4Eldn7BDK%BvD&;NGHc*8-wWBz86IXusq_iaf%QoeH3 z(2_g(xS4BvSr9;;sfX+e>@)0K;(AhK%c%ut;)Lx>Ih(ox88lybaU;+vS0$2P-fi`{$E0~kJdAsX+$ zI)Acc9Q^RYVJ24G7BnY31%9eq6m4L$D&HSAzmmi496@MTA?`fW(xm~jg?b3u!5Z5s z`>U?jprL-ax3je(odf*g%Kejjm~2C?dHBj{j7H%QXl?{DPm&&c**KDH+$+b)NDork zeT%0@#e{3pS04j237W$U(uAt|j!D@_2ZXF|s&1-AG4Nc&Jx>*h>HWnM8_UN2Sn0kr zB!^pEGnhqbZNC;XGuv{(l3-F-u3tgf9Y-JPvsb1vUt#g2QDz>+q(H7`BC^xjTf0Ap zh)AzA%%?SaaBC85_|Sq$yUq(rD6)I}Q{l5|wxXSL?YCkbEj}zst^5`9jK=p8Ge?8K z`y*B@E+no5bXZLlApTSKQ-?RG&7FqI#xGFN<|MJZvNpQbls30VhIUG2cO`mGh3uE4 z#+`Viw!}6c7r29E*~NcP#?K@td-=r2(3v2*{mODIKk6Q|6N#HHw?H%vAuk~%LIDX4 zF8+Ty$KH{+qzyUra4AiaMPxSeD>0h`4!q^c9oy|h%l*7#V@&nJg7f5V+ z*F{;;MRU(u9|I`Fm$(M^;6Jd?+WG6NDP1ov`QMN{U>J8|mBCx^07Zmk(H7Ia_nD`Y z^(j^UdA~b>osC%hK`n@6T(%w(3`tD#cx+nNG{ax7v5kXp;sT)!Jm;W=4~7pyJ7ve0 ze7%P{=Fk!F(`-U3iv-e6682KEU*C8gi{PlK4r~k8X)It#@_&Eki}605h1QWoMX#CZ ztG9|hN)J&W?`ZM9x<{w@V9KJlt^<$1=cdf^cf714!F`q4fFpD0>l$rIrGZ<@v_=WX z?#h(Uz<|__&j}hXk_gDHOg2KWVGm-ZrV_Sq6sNo-p=+iBd6(ZwLoC8R@R&?>?Lgn6~7x4PArbe;TE9oOQ0j6L975{gZ^UGd#uF$ZBm5~9fiN1L4x-KeYGx(kZA zp4XoCY#;T_rwDqxbgpg8_d!=$RQvy=Fz))x?0({-9?j^_rQE%WVb&d3x!QJil&cN`?@FM798@lR)*$E*Jiqf z%5V;{0BknmS1It~;>@m*E9Gmo8n+hu@VL=m<9a!b&J90Zzs;=(Gtm+Tm2I8U%%2DM zdu_TSbwEonWYltMY`$cf^b%rSn;7`e^-%aAA(>~Z=7Z&}Rp!Sr*Uw>4U=(wQC>UM) ziF+%H*}dhfP0r}R@1o$(FXR(pdS*)_%A3*N+!%>FWW?h8)Hi~1Pv6At|J%2c1+$+F z!F^p+$YrcTpuV;-gu~vpg%XVGHtf{(xcGpF)dKMXV#0;W;bKmqsrn-2A*PNSI0kdp z*JJf7$mFY(mMaulGeRpbjp&9=iMCYC+Zj4W{S~lqhd%F`0tNrv54P#T_a8AKpds6i zzO)^r?aH|Nt^lCQ1>27RD4OeK*D?HU-W2AbcoDc;iqM@42%9PQaF`NrK6S}TbLr0Y zg7@}i8O*r~0(;$xMFeAQ&yZNaN{(@=Qj%9ar7tq~!8 zZ8L!#MMLiK(iMMmBJ@cVu0kU4n=Y61KCK+g-%dEZ@R&86g_n-#h#s$5DpdTso`-jW zamz603VG8*i6N)q)n)me(d-ujhBMvo%O6qe4llnAo@5>J@Z}BJk&?nka>l(__kXH? zX;ZZ}L6~iTfwkeLGkVK{M~t^Os)%i5(CsC~VwD3~KHrC#fD5mEB1TU*CXC!~>9jXj znxZYuOXgYk$l}c~`=ltPcR?KPZHMAE;(wV&BxJgCGngQI8JO=2^$HYXX+3-tj)Wp! zO2zw@$@M1KZzcP~5zyCnRZCDhYo7T^ql9ah{`bWG`s%fWj=6}KqT}SK$e%rJku6xa zv>aLT(KgG-8r^%AwEVYbG~f37@BBCE!)Y{JNJp}NBUD}?T7abf51kXn;jT(rk0NGc4kri^gQ^z zTNioSt>r`1-s}6X| zXcoN*FAXcz$@%jjw{i$tI%R&#XSWYu!o})u)suM;fq(80=j2ten8|zS(UCPdoFW>N zVaMmlh+|rnEKV4oHpEX*P5+i$_eN&;C2d2A+iAC@1fW2pc&FIb-m+aT9B5mAOIpLy z0ZTwopaUUGib3q+Xqr57ErL=&>a(N{V#u^VQkH&Ns=88Mw3_{XbcB6WVVDb6 zg*aVxY$fVSY&)v+GFK&qFq@!o?@lP}>jZa@ln5qYG3jPpdJf|WFry49Y=r^R?lF!= z?=9qBbV|D0yTJSSQtH2%Z>5% z@)yMF+YkLfMN0-O+U=_}(zjRtzk2Hjhw+ZWUx=>!fH16dKQ?*mTlqe8yC0b;*~5J< zXwBwjpkU(61Z#y-*@ls$VU9!KH3oin<#`yK>ri;eONMHZnAkWH_U?-|3a|5iauY~GAD^=HrZMBF$S$(g6bLek213cz z;IEm?(t-kRY8V8j@Ta@-&W&n!#+1ysPf=(;}P#h_8_ha41ArG^J^ z@CL*8JbVmcht`Pt-c)QyH>|?jfjaA>pPtS#=;FeyHA&DZpLGioa%NNk>nF!`qjOdK zPw%^y+_^=A><{o&9_>uP+T(8|lrDXoihUjZkwJDMA)U@nZNe$xB6W~mo`kl2s>N~F z>Bbyb5%%bnkcn!i#f!acooq;3|7$$VBj2v&m_naPo&Dmn!Fevi0v*(mTZ-n{)JL)V zpD_f3SL#_K*mRKnsvcB{)@}ZB^vjaZeIAmDwQQbbsoVUYzdA}|0pS&+V1~_dbWb!; zy}D{U6heryWTN5G(W-|OQVxlg8s7xHhA2TWbz}Df~Murea z!u1;UEJRJ9b0vq6zD&<_ zVj--F1=Ir^l^0UOdvWOMHqXUq4OLs-EHmaNVEy}r>w>mpTs|diLh~E#6oiCmO%M+w z$^I@Z;%F5N+_MAy?<|A-PyRxf{VkM(1X^0q~4eo^>s?kpBByjKlx zF@&PrPLvJ@D!JZfGpo*gad8qC1(Q^h)!Kw4X>e%*im0JvOZK;hz69rsExW9Q(RA6& zL%jPJXWD%S&R$Sy$Lv*0zPfHqJ4qE}doPMiZ%;eY=g=6-8k`mD(mMkeo)E)G=jLIDIpYSI1aaSJAMiVj!_)!L zN)Re|jk*#Fmju}W!(s7&e2>$tpXpaaHaWqaqv@)3EsauW0UiDBuGP2jduNJ!-9n;0 zcJh@_1lnzHg&>za5KHn{Y+UxSTs(7^AD7Ey7I8%U?n^Hp`Y!QMtH4!Q+VZ?3Qel>f za@V!<{~fcHMKlwZO*FbzU+jmMCEe&2&_FC7QN=Y*82HE<1LuYN~w`ZeGJRXo{pKrL&jBN;~W%x zOkssO2FrIOZ}=0E?;ff>u90m#6pK{(EN}{6kE32X9%P}Q8V63|I&I(R$KqKI2+33v zS^3x6J8fBpoiYdZJazU(O@|RTi{#{6ew~Oude&J zjYbAJtsMLP)XvFF?X9~>U>Te05AXupcW!#sA^^9g?C-Tl3xUs}EC?`HG?acu49r?M zEgO|EUa32rJ}~@(l>=>tuP^YuodF_^NUQrO-5f5XU#;cFVn_!v%wo0jV>UAK+*M#5 zgsqHPx&!2?ju7M1>svwj@=c9bt-r z!#5hYJW!NqXhe1>;E!Ev#D|!H1bzlHxD*L$^TFCQe!@K_GVNrS?w?K)7PMQ+X?>$O zB%FC6wy4rH9?iEpqR_?537sbelhD8hno~I_1b@TS*-cub?y?qTPV3-@J`7J+kY*I}z;!qZG4a!nQjeH^1jM*tH!f^*BB^PXWPs&5&B? z^&EQaj=j=UQZyV*rdvBx626c?e;1Ag%d}rtCG1brheSgsm|;cs44?CvTH2CBgN>F& z$;Rf7$uPT18JDP=I26~ySx37$J&#v)aX`^m@W}H!)?VQ2-2}SEh1t#aCT}_WX!zzL zs7fB|@Iwb-hc%A)zB~WR2_Iz)-gkN+Vw@X3`g)@X-|Lw7VV^P6Vs2WbOy~71Cx<&f zF6bWecw&XHN7?)h4?;xXb+J8^V51Hd6jI!VZ})s2vC5>1d4XXyRlRV`cX&Z``y1Dg zhHF2nP*-JSH2hnpY{XtIwpP^Ks{;B0&F)8;IWOELv~koU{_NLIL>bPi`|7d*=u}L}9%SR6f*W z{HtAJ9z8jN7jgm;6+Zvuvy7apG)p5YV3!Is?C9n2F6fXNR#&wn<{=zF;9iQ>M_(^bA-Y^3ZHoPPCDS+X3!4 z=tHw!c^k|5uBargS(p!5B5!Nr##-#JdnjALsO-R8PsVSghRqD*a3F;X;m0#!EahyX zcL#S)ugI0NNyl5hD5E{z@ujrnGzsv~yJgospQ8evQNTWovtE<_-hBHI%Y7`45D z6T3Riy@X`jZpY`A2Mcjf#XD173BJ5V#26(c>hJ{5(8R+{wPEk(W<|#8Ef6%rlvOP+ z&GqiV_#A}p_q9k)@shkLgoDCOIJ>wC81xL zGn0Lmv4UU~@pSRvxrFvOh239vdu+=^hJJ}na!L!>L-w!Y-vnXd_j)o4=kojyD@K}z znw5x4A}`Nx4~P|d1g*$FnlHqq5OEKT=i@hRN)jj;<|2gz;RM}KZX%io3x5{6)DTBu zQ^>Icp(pzNRHJVtOl5;~pF$#}pGK-0^LnFIc80zyREQ2JI0b806xb5LPH4VE5jbLC zAmw%rQ?YnwK)gG7ob){EZ4Wuh7OV`d)=We9FJfhoK8Cv@GSzGHaoFzM$@lS3%~^4P z8e~y83cSb6AaJjliii%~ZH1X>6j@N>={l-%2%8PYSsb!|E zZ0QI)xi*dK?yH{ExI!z=W#g>k^SnOVTJxYv8;<`mK7`^eq;dT`|93+Oc#ARgUMPP9 zrp0(^Oba~p8q0|(huaCqm;0(v$l5|hw3VH|0`1cdxp!omOWEp|5f-3o^gLV4GV~P_ zj!Lfj{3s)$$FxVp2ywB`gSkP^vT#iM=kr`(5?LO)c&0HQJpPQzbZqC>X3dV4l9~F$ zd_ttpDcJdK{dbRJ%N}J6?dL5bFlq!qnZPN=at1;m3+9RkUSkTu)*Gzz0H*0*aAK~WO&kww zw3jKJS+PKYdma;ph{Pzue}-fAbQasV2=?yjqUa?#MZDot&#`9|KbxQA#~qdgVML0Q{&AK=EJ9a}a-1tJzqWoqqU=FP7YDTPh9uQ)wyXq!=h&y*lqJ@@LlJ&nnMKdKZsy zwgCat)c81z1LuA@EDz%l(`6LHsE1TdXRtO*Cgt>AgKJE2LmL(vXEIx>ng&s}aPeoI ztfdX(rYW>4Lylh@8!O^MwY6r}ut{t{$E=1%4yEAniRwgt(#PePf&s|jX?$cbo#=!p znqP@c?~7u@@#1q0pyI`r{yu$NAm&?3d|QK>up<3p+zYhk9JYQBK+0uX8=`Yx{2Oyeys&88Dl)aj)cEJxX7`)OJ5qjB zs(YfqTSd9-l3aN0NU5wl2R~?l96p<8)b@tmu-B2;c2-`liwbH3gCU+@6uuBX}FApL#n5H*f#q1$dSs$ynoMd6tNN+gU) zPiYQGfJ*C`?pdIH8`G8hDKk@*U3?%z(p0aNYg{TtoAv4ItlH#WnHJ;=NyRH}TS!X{ z-o#n0H!)a<#FU<3G;%#A<1UbqIEG;IknveEFU+A5{N2rlSqGC}sR|!VP8NEb`s3)V z*5m|&F#1ZkT;hYf-r?@n@bqDG6ZoXpS-IPO+D^lQl03Y)T_i?3<&AOr#z zLXgUmt_doK@4npMl)!%FHl4QKOW=ngS>h15BaAcdn|kYT-2$fai|U?lJ-R=^>3J8QgVuZ?=wLl&WrKfv@=_ELitkpnIe)%Xbk4@}X)$KaA zje-oO8$ze{AZ91oDyH<*b3ifc;T!6*qTQHSWn80aS<|)k*UO)aNXwNVl+}}WbNd*( zdNBHeSqPR7tX`eEd3+(d2;zJ9xA8Nc`)5TObx+~lLGb3ab&^~~)MO(c9q)A$=fStx zR5VvDxZ30@@mG)!NKlq8hKs81kalNuVZ|FU9p}0eKDCj@i8epR9@92Xm}pO)_ZKe9 zeo*nHHEeiW63cJ3PSoqZ+HV=!Do5QOH^Cj%v4tKxE^rvg+{@IdXh^M zI~F(P6_87WBaM(M{Hd~n{BXnIQZJL`H-IR$|!JGfi zbc~JZlz}8KJ~Lc7K2NCR$@nNLug3gV0z%Dk*+E(V28sTBVXH**&M&6iEVeTS=z|j7 zs`rPE#l=bR724nK6$OJw1wMoOXGdE?DX9LC$5M}O-{2zt!-c*QNygk^d!#ed)pF&+ zMYgoy*gKwI^9E#}0SxSC1MrDLQ(T3_s6dM*p+^#@%NpD)zIwBKBxbJQQO1u^L?voo zhoPcy#6mFxFEw32%Xe7co{8$D6>k{(TzJU~5h!BNsSTyS9?H zJm)XQ5E34e71dA&Wr=Wh`s=tAHk|wenk3@DN!Jlq^MPi;6RA${ip=iO*<+)sk( z#9A1+a-WJZFQu|CoO#Z<1^=+fH6NDzbQZLX+f_|67|tXPRXHzlrSWgs4d**D3s-k) zpwFiN_on2Su?N34>2+ETsDdt*u%EV^mlZVeQDu%Q#cix zF@nVsd(^f=ci-3C(@D?t$D}+l+2Zl$Sm#`#IRqjZ~mYvWNbiMTGmo} z#j8$xISb+1IuAOh4_bBoFT+q?W?|xLf;t`(c z3Dp|tMNIT!qU(X{z27Hk?ab>*ZZ}!hsaLu!2LJD#ZC{L+}^=}Ok7b*o04KE z;%)2S?)3K>1Sz&U9D;Z+5qPR1QX`h51{TZ&zc$3Co-R=(u2@hC5Lxk9B`m=5c7mcI z{-%W;QmpH4i{b7+Ax>O0S-2xsM&+#u1mTP7@8gw(n*;BIlaUsan`(mg2EX7%quBr~ zWI0SH(A73_Ep6L!?Bd-9rQPdM)+vojZ*Yl7SIQu@z@#^tfJc!FZl9;>7)xl3l>r6B z!IAKp=r{Ms@*EmNYnq@virAN{tPctBV~jphHw)EhJeewj?wJ}-vww^)FhXguq*o_{ zEe*9cN;D}cYD>q{$yct>b~ zENhVd_u0sicFm6s?nr(FYwhOC{rXEUnyp&1Ckguto_2FuIv)DmTh)RWx35Jl;|^Q- zdx$9?G5Ua_%0^CjRl#tE4gPj-r8Lr%$`?G?Vuko->rg&P2D1BHG8C<3-m%MRgoRa8 zy$1n-#H$@V!4GcBR?j87KDHJ@YhhYTAph7z%7Q)3$@hIDa*eaC>w}nixhj^(SdS&=VeoRmZNo%YA^(68=sy-1V|7S?QE+ zx<*nHmV2G{07ewF@}MDe18DWMHRMiuy(*l^ttQrG5wxVOvi;1dp2GYT2}(08HwG+iqZbGK*WerwAH;VR8X%25ypM338)$MT;aAw-z4%QT>r;=j@)|DXg(vc-?I@a`4ReJx4yd(u-TP9s4B3lZniltu(EHtiqg=5wsmJM+Uqa_m;Hb`uv9ILVbcY7xe;ihND1Z zK&_;POysunlv^w14J+J1#mrfCU>= zXTq>Pb|5+zMr4ErhSk>bpq1CJg8mJ(&DKzVmW-6mUfH|%L*cGOPwhTYSUpwmX8LyQ zXupE5=UrBX!msOH$G!j$F*_D+KWR^fEOqv3@9pv+dTqTuC%kxvrNs3d65q4Fg#?+p zNKo#SI?MA#Y475B>7}e+Dm@GY-C_TtyG|S4gGr&#<-h$6tL(~cy{s6S7De+$wI%57Xh3-lMz49R=?|&Ju18e-J~#YD@_^G;9g`gleKr= zMm&po5nqdFs5Isqmza7+*JUF8Uj!PO%l2V^h~wN^p9cz@9(WBAu`_VDoB_&q3mqf|F zP#-TN8SSuJ@<6WxpZf;SJv<{MQY2(XqsJBv1H+XRUMx3}X6a!+J{ZBMLhNars}GUl z8nkP}dYTdpPgOV-gX|HO#G6nVmlYoe?M;ot+W{;|;+iC>;%=8x$L=S!1~(`NNbv*8 zJO+$V{7g+7@)|o1#WZJ6G9En6=oUCLk5*B2@YQit;pk|o3kdgUh_5vCcb__<5VT6L z=>sN&^ZAUPzB^@|)A(YzZlmzxw5@>QPt0>?!^ zsY1q@jO`gvY+4vINDd?PW_nnqsci|u0Mu)6*s5_&8c}FscsPl@@`TWq4qHxNFLB;(ANoCe1HZhDCq6cD7SS zrSM$qDE~$1N4G+MH1SQtW+SR=62EeKq3hy9o)>=Q8 z9ASv+z~C(HW5HLsJJQd)iP{F11+h& z$>4<~u&5s1-l6*h-Om~e0tL>qP%CIswx$ZCTu;aMz-|^r7yHq zhY?DCDO-H(*Dzvq5JS}*#?KA~k1wE$yPsB;SN=ug>S41m=Epa8!+7~_-9q?=)Afwr zK3X?Y-StaIm}P`b8ovbYDqUM?^(L869(cl)jC+Yu#>(xTU8!Av&w zkD(62#UIa+4l(cF=}yS(*CQQFHL&*p#|X?5mW+Xm@C^IuoNd&c!)>NDjG7sSy&MA{ zL9Na%pRvr~5<(h}%(g09r%#YPIuDcNgl!_F6v++&hj3w0p6gG~$<#8vNfdXpRmkI{Z>mR<>ieS8`A*FWOPDxA~JayP;?9W%hgnC-CDjX3C; z58B>UW}G5WrlIj)Z-}nQ~jK!RVe>AXW}_vL1Wl75p&%+4n|BF#0d_|t|NCz!lA6~~Q{N>NSfz5z`8wT5 z&hoRadUFDV5N0F0(0i(7GPj?@vMye>eE z7d!hsb8)@4-ArlAryJI4?_>xL<}U+|;JeU}v(~z}U7$sNXnAbQO@a`^;@3Fb7IXFl zwCj3THM#VCG1fQqhAgzFJp)|Td_a9`N$6jI9#vvTcxga`Bt@trA#3_Zxr#ns{u`z_ z4V~9gKA7iAjcVg<`ztomIV6MAylHImo4>!>i3ExC79@Ht?gqeFI9|vNhe^CR`;3M& z`e;l?zgK6UWHx1td%Q1-1S3l8s zeFN9-Wi%9kkGfF$c_H9+p_EysZ7GQcB}r%X75eno^}pdJeljJ+O?z2*>&`Z%9q6ZN z^m@Vb)c=fS6x*|mD&g^=c=2S>x+`EIo}0stz%d7o96jWjs{1Q6xp8{% zppO=uZdS51-;kWNZ9rhuze*MMqA$I$-SBZ8cH0{4Lyw>nZB$(_9~mIG$0F$<+Z(%;#d z^_P!L?XOohEv!ZBVaxKppa?K^Ul}+ zzR0(_w$)=2ra(ifpG2ESZj{CazR<)Xc0+UAci3KDb$wx)k5j;&^IWYY zV6-neOggaTDj(X5b24b6E~5^_0OnL=|75TfN773Vtv))2fjYULidz}YDoaE57)UW~ zDb5VE8)G{BLvqh5&dwM!m}4j#yR4Fe)M!ZDqETU@5y<1+BHptihCz;=qNGC&i;-k$ z@{!bR886QnrqInK#HR#3dQ;fsN%~gnDyhWm(TvrbYIF%OsXA+9>9JP&nMx{A|Jl0z#=TnLUv-9n|M=VjWP1??zh;E$Am@z+`{4^pK<&|T3rlrlmd$56lVr4T_m-4e#5n`vHu2N7? zJFrOcT{{XLpiVvQ`b=bdu)tNuOphDG9eT+B!b?_9k9G?$wXxB$SL%|tUmK1 z%L(G1>|Bbfd884q4IqM2->t;8S6bF@;jSOMh{&skD^6B;Wd1 zFj;PWNJq-3Kt;RuN!svtk}Atru$;#40UCKK8CgC2w~PQmK)%16LC-qNz$K{=X$}nn6eL7RwF&6@=|+x!4rv$Cj6G__@nSPx9q>e1aWUb59*3|rZ{nw zXn=uNY@pSOT;pPVBm((&7{c61m~;9DTu;F~I7nFM0iswF!yLVX%>>nbIWKBJqb`l5 z$eiO?&gfb?r6D}W=Bu+HyN2TMCXrLHZ)S*+5ptzCo+(h-zEhQlKf?y55fvfz(WShw z)vs$V&S_z%!;ciplezP8nvO8(F>`~X%ew}JPR{hX zZq+7AY|rc5dme&5qmtSE)K`}pzaX^CvNkmH)*k1zx3$#5((?J-+4vOn8#|z zD0Q|KQhl_vFji5mYgs|9C=gU(5bqdCq&RxuxgS(M^U{RRPr_EKf+$KvE!V((G}jLi zO5#Pp^0A%pisBXd@I)$z+;T|>LS`>1-mWgx`6ecX0QA4_%K;MK=fSp{-uFo3ejnR zrBtZJPrgZloNKMt?cjmDp{ZL_Qq*TnY#@Fk_yejRY0uaj3^-iMfn%EHxcbpiQ>l99 zyhfUvzWV>SmPZ!3o_M+Q`sX1QaRY58`h`vddHR&nW!MeU*_NSaxsU@rO;hlwKGLa~EFCG9Fga!=$}s!5bgpSz_(6D^ruFLOmX5#) zyU?5p4yM7`+>MdI6Ij+9QJUwa1d3wC$~GjCMLO$DwPxInKr`m(gN!LT{()I2DkeA7 z-Awqr)$dIhp{PMmzPXB+bx`x3=-)S`B(0emTpBM%R8sxn<&)t`Nelx6gMwJdeLh}L z!HZMxOZG3w28^89^Xa;$@yD{sYA7VUX_a;R!rnIZi$d&b{Drhp*_#uqbS17QhNY?E zBP^v*_t5Stm1s!H8Fd)~Kjvw$JD?s7-ookzPbbxP(-};0x4K*JD>l{0o2j679LpVf zljKoNX426lkL~_4Ef!z2$(Z*#^G`!0r>V{_PdKd5WxdJ6>RWz;*#Vnu9U2Bi16`r= zu0Qwh39Ncpf<&&p)kIfp@PCqFDPMLv|&A?RJ?e$F3HV)1*Hmy(UN0X%qL2ht`4wFt762B&@@ zh8uFzim}&gu4ZuI$x(gCAKD+z>=Oo^_JH7!bh3p&(erN76RP{L4t)`oG$~bD2|N4! z!l<6zoTgmU5{jX!s*ziqAk!Byv0u5-w9q`L&tvq~OidCGUT)@0UH-m+31$yI6Pb^u zFXu=@o1a4MvS}5%0S&9c9t&=zgk7$OUJ-bwg<7YMma$42X1?XF?q}~| zaaBl}%-e9}oCc+?6N8%-Y!a@E3LW9F7^s?TsV;7J+m_ zeZQ^g2O;oo^#Wzky6eL#zIsntdx^S6n(67|{6pwa7%Dn3E8lC40%IBf%X=`milU2q z{r1qh80*+fE>wouIqh20)-}Wx=1)2+DA0pr3cxkxob^biLadTN#L4j*d?ug;m{}GqR6f$jlfptrg8$ou0ONwyiDEFpf`c;5Pp2vp@{l50yfe^h|Qul8WNrA_rCQCFb zfWi(#OSdXA1784#4zbzH9vTjIay0$}CGoS4{?hDx5WQ@IM84Hi7IT{FQ~T7rLo2N~2C;XVE zEmd*M{&wo4^Y&&r65W~hh3A>%myA6*`RZS2Sjc>69NZsvzZtEmx?#oT=1FDvW+~3Z zLc~?HB5V+-z4P&w*PA&-TWLh-Oo>iINEHNaF3bw?oM@@ozuc0G$|Jsp$A1VHzvv%# zAw#RykAYOak)}sAp#UvyHkNAvGIgH3qfgESq>L81OB=)pI_<9F3=>8xs>Qw{F90i6 z>GjZFuKf0sP$JYk6_dgP%tU~Jo59VrPr*I|k@Jvk!MN4X!<7fZ+|e#sQy00f*Y#5I zAPArSR?~1JR({8m$y4F6w5!hVQO8}QMmCDh7os5dib%SgXl1Tmc_X)Xdr`uT3lTyZ z(H(`@m~u|Dm8sGYBtn{dJmw;KgXiQ#hE7b0Rl2&e06sfzj%U zF67ktqcu2+%znuDs|Vq52ldQo;x<5GDjr|-A?flh#!pTzzk#@_jxz_%@Z(pqm?c$k z4GK71$2lQssWO{-ZDGt@)0MAA7=F>4xg$ar?s*y&+^=pF5orf3d!`Aa3{wAnBvMS8 zH8$^Iz${bcM4aTwIwI6fa&Mkx(le;8jYl!&zRR}}OQ zpH;S=>|ZHJn8fN}e-xynn{3DmryS z*Q7z_YZ88@h4mE7^M(ua0tIi~{`5!N^9UE8T}F-)o3$#qmv5e{751%L#Ui|;IDzpN zTv_*Mt!&|XX`DZfD2CrQvHk6bhPR?B8sS-g!?7GgMUvalypCj{Jh^aF%WCxT7b3yD zU4tUrwUfan_>grN&$zVzY`Smy(q!J#v^!^J-Y_|;j{cDZrkx|s>FF&9YsG*3vz3s-*3!uFjT=q zJRHB|QK0zeReC~FYQPM_c2G%nWZsigs7WkNzjKreCAD8 zolY8kw1{D}X{kH?l-`L*ZPH3>k7!Y}_X6Hq)a-%AxUEmr?CxVe%%N&ud`8T5 zhteTgP^siMqKy-1(i>$cJW`NtVC1bqwnRscvLf&U$sKaEmVysjF45MjrI;FqjqBm-_s2@gs?Ex^fuWWJaLd~jcb3J8}0z6 zP}0PdA^uhFo3c=JQ?69SL;K! z75x&1*)rQ9PB2ZC8ClBkW^R=oj>zs8=R~vMJ~W9^U2RVtiRqKJMCqxdpOeF8g=T#t zSP0VfmYJDL-GtPTEs4udH28O4i)%l&vWXq%!S?)2B|R1q!@!SZP1aH@4-;vQR0uViNxDX^HIGK< zLhF3VG^=p&q?NTA)md2dQym$1u z%-AeZE-MUwr0MJ7FLPoSb{g=_Ip53QtG#fQ*_+xbS_1#|;wY6;ZQ}6xk zo!oeNsj)+@u}3QkOgLW?6FOttG_iFhe`1o7NHSzy?E|KM0mnmL?8(kVXo7zsC?z~I zOaGk>%J{C)A9E{8%JynDXOZ4G)oY2rMv#8%j6Hr@!mhe&HS2hS9Y#J2dwElqqR=QR z7VmPLZ3)g2$DSiWLr86|9y@|0ZsQ|P%%hjB5A~RC3uPUmPV$1-n5?rPrKHBy9#E9| z81oY!JUeJ>$pA*f3nNh;CDAPR)@5mAYn8<^$Nri%Vw5nBu)t@b|v4CU(gP*+fqZz4D(qoP5N|GGs1ec zNq~pnXrL<%aC~4g$FsOFnM{fJ{u-ADA|WdyNV%kqi?dREtw4(h~!wX(=d&O z(w8*eyxbM>cjme%x^${}uWjroU1)4x&u)kXE|h`!tk0ttoJD|cejm!X6pez$^9Q~M5)n?>x++V$=}pQ z9;b(y6no*$&tHv`?3`iSkX;6B_`B)1YMk!p752t$U;M&kRGDpuYBx?-+UbrlYK;@P&>Ei~h){4E9Rkzk`$IQb1Io<4H8MP2uNS>xnDZ+4*J8JS;s<|CWqFpJIrONnnRxLsI z>fyvc#9PMn(sPf#NwIQ4N?k3ToE0zOV?n!Db8$Pw93Gu<@n5P7k({-gywFtws?^0u zHrTrExS+7oBLd^{QiA#SqHqh5=_}eihs`!YkN)FNtj-{YyfKVxtIbptYI#T*u z$S<~ag@SPD!TPLo(iJ}&iZwykxnmj%(6h4y_tz8$<#E!RNMtvlIb;RFnWCfNCAUIz+5jHzJ9Y4S59Fj?- zONMkaDAm4x6D? zpTt;jdvkFmy|6>hc(EE5->(2Egv0Ey$aJP4`0*s22)|QATB6W`-^#IRHfd7Qn8%@@ z^xSF(azVM&VUii_GioZajzU%u6TcvxQ165dnzKJ0Ts7QaSX7slrXqfJ=Y&^pk_q{Y z+Xa_{M^#bJ3Goe!Gv4`(*u={&eA9zvC%iGn<>1v@-eyn!sP6SPxLZ#qO2K6g?JqwX zYs`w#?%jUCm$~;^z|_}dVEE-wjBK>O^@IrtfFK^PTxUbK)3znl@&rp;WEwG7DJIc% zSL@=B@hVf?Qp`Z}QmpTuBrU&+^!x?+){;36zqI%96)VS?tXclXJ3S=G;G!vY8cEjr z?5Mm}v!$F2?EE)Zt;OdJ-Gz8h_DG46<+#ouKO)+Sne8+*`e(_*4Ut!+g=6R5Rw(qP-7EnH zl8t!bmA}R(l(Qig0{Rv<=KS?{-EPhuS3-3f%;r}BdFWh?p5XOef>3tmQPnEFgH2*yQZ^gd^J>dsxl`#9-mdA$678wc)qI9(-8g= z%PLevUawa~=N~Pn-VQQx@oSJcn4sIpB$O0~qhLn-kFI8M5yzzcwei%ML38$)z*${a zVkXQ-yTS-rpslxn79KO%aj%ZM(nV8!Z{V7XA}a=`J#p&052fB!dA4(EbvmK6Hf;PP zZVW~Ie%iH)_CCPr#3^J*f|HaGHzd7@E}{X`3t zV~tC`M+50id&}L@V@PE-$Nt-AQQmSk{CK$_UF6Zygp}4kQEtKo-HJ4H#E!tQQv9+s zA%sDlU*VO4_M_s9QS4GABsTnhOW zcEJ&x}5W^e=xdY9Yre3yrtaU^?ps(OJ!~*o`l~&bGzJ^JiulehkY+Fc4kQXG?=CZSxzm6awLF8{P zy)1)HFv!g1bs;l)V(u=_YK&SL8rytTO_7expeq!vLT9LP11`}rPlR7x#(C}#%ps3? zmX)=W2{mY!)a_79R&uCIkv!?Z|A-6kc;aEy$u)%`)7 z85#*XUrm+l&Y0fQ_0gvMhmyv0*eaf-xRhJ2Mh9Q#kbZnS!rf zb;oPRAnnEceUxe+mYu_ZPeydeZ|t?}k=GKSYT3+!uKW|yT4|2x(LjsIa-)*sV5Y@+ zJ_3oe0u)0h?L`Se&JR?YSer+FoHx_jwCBWy51p>D-A0ATIedcm(+Rm*MU8@7*##y~ zsFKp>?bIk4!Xo6mQBYBlRvS_2MMGiMi&%pB`uI^w4{T-vNdFRl)R~F&7duRMrdN|m zdSLf18XLPX@IXSVR%^m@_^OKlM!2rn(|r zr^ZGGj$WdL#x{N;ilU= zE2%nHXzqMKjFR^+8Y&}b(VQ-MzY-++{TB)4eLS4yFZ7y>@!g~fR0Vxpb1uD!b|x38 zETiJ73=NdfTG%e34I-#XAgOe@qsn^ZZ-Xu2@UCXxcTdPh%W!Yt!Jm4eXIX})cEhC_ zQDq(oZOZ9AH0g-K+pbpk_z0V3pW&#jq*~W_I$5b zPof4xnfZAmd_7D6Qw@x)RtZDyu{LH+_?7KEtTe+(BkQ)`fSZyXboS${#9XYp!Y3O&?a@&jz&Gz!kt1xhP$` z@BH_!z*U7-9Wz#3*y@wsciHh%!9UyG)8CX<2#r_<{5eZCrsSm)D0nsySV@`Ta(~-x zkSR;v-8FoK_bC>9_3q*^V1Wmq6$of85NHW{`{vmM$cx1h?-yiusJW z*-9?IT~)`KLCeOi9lFU}^-)gr$&wxyZPD;g-gU(Qrw8?*1<^TK64PRdv` zxh^r{X?Uu}=K$4P;%h$h05y25ZDgx&P0!O>?*t}((Chq9OG^fM{QWGxq ztjD{jTtdM)w#_2W99Z7cs4Q&=VuZ-)Ouk9K7gON~aI=XGsXh2G%wS(HOCDikBexFj z6^^VP$+3qQM)%2WR%R^WuG47_N+F!}7a zZzggWsfX5WHEEQYZ2-7jBxni)FP(m!JGiAR6;v0OTLWd0$Xd7YuukIN+>;MT{_xAD-b?h;Fp9Q84Q)Y z*GPJXxQmVE1U(XEYx#z3Aj%}7&KB8DGz+xUGYUn&DihFeaM57k#=O~jepg>>+4L}! zde1r~KOaab_MFr}suEiR_8wF8>oo^oZLw353O6m@@)R%B3Bv&oSQyf?%(Tux0~ zqD1N*|OGt zw6caAUr+)g5i-|^T_@#^3HVzdn5F|58EgUV4ZV@Giw@)8#N(UeeI&d)HBW`EslrM> z*20zcHUKr*fhftwVZx?6f4{_cSY>l*rkbkjF{HWy-*H99@ zHXEQHgnUyRC(3;?wz3w|guM&mFf1k}(AQFSv8jGcZ(Gan@8r}p7T87&uJD^?H)?8# zzl?D!#!!iyTWvLeiJDfjC~XMvx*^E9&gFO9Y{8x#3dfzrr>RyBUYFFz-rMX3^4JPP zpW$$xD?vDQFiABT2xZzP#6F{QMBV4byx%gMmZ&gFw`wYtx<@u6GN81?@4pb({tHFlCmVj1^Fb`?msIWv1Jpcvms5ZH&QVk+=~RT{(Kl;j~;-2 z2n@B%?VZl_i$YJ(7kO*1B|x_5vT3+zECRNOoPwK|EM3~jrbv7`MV4py!t6D6*|Q_u zQin{YG>O!5dvWy#gnIXVo$>YWfpajZko}VKzjqnoP zbOD%oaUc_iz1M&i5mV^u z{o8Q<=dtz4+_Nnd{ZFJ`%HDr1Q`UP`I+u+i(#l&PxQo54sjO4awXxXOn%Ro~<;rl@ zP*Ap{wU8emj*WNMC>^>}yiYux$duyY;vBIzWi$s43<5X*V;;~4FVJSpM> zF!$+D@~Dv2?V6qZULm<}Idr8TDp*a!oC3&x!c%LmR*<}>-h96xWb@5_ADx~~qoy*( z44GKF-S61IaFq(;MY#gpkdVQA)-aVRSQ`iy>kea!&Rsa-3|AtBY|n2URETA!&|NMD znES^hm1$(SRHkE^tQtxu^n5Goa_5o1GhE$-Ax^DzTjn+2T-KjRjTj4P4l^zuv^YVS z{i-m*N>fo1f_p*&ZGAr9;Z1Y_d-rYb4WwLeGD+-e$lyI5-CEXd@9U?yNq7tipRed& zi$BV+Ex$;GIrJVvaTv-N&79LiGZ8fzPisik(5L&LZW#|>2lwWc)avrd(k)Y*#*96R zdm;XS|HMvlZ)Q$NEO=1ctDC0c$4px2gd8#-?Ov>29We~#JK|zNberYpy?!^Eza*XZ zD>J7q8v2@UId(Vc&uVs#!>H|AJK$np8n?ill28-FqWt-DS zHD*`?DN!biHCZzX~$gm$>X&$ zJ}&fCR}x+@)t>l$@5N+dp*M#dgIm(44n8u)sZvmCAdTN_L_Kk?f<%mer>|$hdak&# zTbIITBs;tJ-ejh727fd#v?g4=u7NAJ3znLB3)e-RkZt`(tivZE z11$HRU-J!WY%sgf^P8MDQDIrW&i0bYv%^hLWpR{hd{tI{qm;fEqNo^ATlG~!`+dgN zLzw(wfFuM3PM9OMDzrn;e7=Hn_?O~P;Q=a$Y%Kj6=|~INdDaV0!g;K2_-&0KB82tL z)voaK^T`~{tABwD-6|ZQNgU7Is$Wcs6j-BISx+WGW}@kLO;t>aUHWtQ>*B3#7F?MpOjDzWR+WxLU_t79q3ZJk{JTCwZG z-Xe-5aN@3N7yP=IFfc|T0>P}siD`IV->Ib+lup65>;tqp+Z$00)D_|FPSD_x)#~Gk z7QQ;j@HUfZLRixb>+P4Uoth65`y&Byp2=7{TMh#n%Pa-4#)|x+A?dKxjwN<#`=N$P zHvH}&Rv0R}MFnnlwGS;QysdDru3j<+G;{7Ix|&#mTuQ5YxsFSC*(LL0xwtMZ_fV@2qnFP9A){8!)F`>#r|!GpNZ8-=iLKIu z_nTm`R(HJ8*45}llYhv_13`dT`<7F?e1kTlU=iDBiwHF_Eg#}Tr+ckuwVre%F|C$<5?b7F)5nSjB7{hD;vt<}yHC7$O$ke4*%G&ecaB%o{1F z;hac)V~g>4NL07Ym)x%WFsh!A0(RUmYF=xjdduO~oN}ZoNW!fMR_~03E9#f=`c0_I zOn`LA3*r-CakbS%ni&$!qvIT^P$W6kp(k>i`(D(Sa+HL3+VpSck}t`!t@FNjg~8)% zt4u}&HZDw%?w66^+myB>mYxjUPnq5zzxYofNaX#+v)0yj`d#*&-gw^<1AEK$F^36R z2kJOaxjY^CPC`dc*&oV~xo}gjZjN49!jdksDHK#spY+qv{h0L|IVI~{y;~+%Z%XkO z=4B-jpY$ZmaU~LVa+{>Vge#Wss@XTRRx_gnX!8vX5)KFKAH%_ySTtziGct;KN_YMW+}p+D{)uv)T#dT zi>{#JjULqGlLLpsEQ0@pQ-`Y8=uk2BX%(wSOy*UBL9 zpX1Or_VY}XR`%Z)SQ#kqh6mF84xQ%102uTL!hIwxyb#B4fUCY-k*RpJL-roiB!hiN zdNSZSqNdr(;k_vbP}+>jHP5lN^j*xSJw1prVVb#U5Ff%f}xDnsnqe4N`h z<%r**F&d*lu*vkJDb;_^&i0P2Tm$C&?91?sK)%L))n3}qGu9+;zA^fqc6Es^DVCC& z=Aq|ZCCenPb$jVj$v;dnMb1;o2X8FANrFy0kBQug$Nws(*1B8z;RMq;szY7qGC;dG zpTRr@n0*U-!PKAfu*Uvcv?9IYgjbD~nRISFL-@4N)zoN|mi?8`o7q8x0N1_?e}D9| zeOj{c(f8s{u@lSUfk&Dgd1}`{}lTJOrEX|`mC(_xv||7LKS>Bw|q7UZ!1Zu1IoVW z9XkL*J@Cbm070A?>VC8_fSsgv1#m-=$m=DM5C#0oU`wKtEWT%#esXCnB=H z4HfjuhqGVl3;vp{EWI6|5*G-mp^i%40UF(3DPm2eVJ;uN^Bd->qEBCF$XXi&LK`r@ ztI8k`59$vEuU}CB$E1J8XH=~ZIr0+NcjAMaH>PmU+We!;A7 zvhXP8nDpK1!|C`QXsmGP+c5<4(TPL!ah9%#&=1?j7qYnAAZ;OjGm$E=%V}r*A8uHUfjc+BH?3zurj84*F3{LbySL|Vc$a|p4 zQImE*46TeR2dZejF^i%2Vma4?@r#0&qiOHbup?L@(cRpWDIHk&R%NK=b#Jl-&GKk5 zV_IStceS79q%;mO5-|#i(8pU`Alx4pjOMme7i@n}W*>dLbMfXSqrZpH&+@?ULWkFe z*-AA7?(Yz%aa7y&3hZ*g`t@Qmu;yy=0RVt{*1Ex+^i8+lM9^aBNPVlRt(MbXZ;;mF zr@dCCn4LL4p~^Ve`x~MH(uhKfWY@`eYc+n%!0@VLcCd+^d-dUFbcm@2V{5&ALQd8THZh9+P(%#}Pz`G9eLsc6xvfRI4`wzd_xOIJ^wE~thoW_A4i!n1S0YSx=``%_O7n!Lt+*>rw*IAELQR)6_PU}0c zmADsr6!P^2jBdUuw)i!&-(-nb+_0D!bIM}z@ituVj45p?v%wnu>tJgrM0lF@FTo0x zp%+`KIS+pvoWR+iU?`hly|}dcF!jytCvR^@H7U83<-AdQ{TG;_EjF|>T@C}D;*fmj z_1O0ff(OU5TSJ~1Btniw>m;()L=i5oFn0XAC1-}I+JD9UT-#dcban=^l5mZ40=sKusWf3v_!T87r0R}bAuc`wE)D_LQ&gh|*`jbC;!cOdah z8?oKT1#L_0G;;}RI#ulxM2yOllf(t+YvB0?Xz^YX(Ni3J9|gVE7W^W_y~a3+mwLr7cM{TOk1GkP4wPOce(wvOnYo z<++N9o382-V$SQteTt8)%dk&BU=p9_*<4chNWQvY%`vveScRn?akb4JxLu0C+3YZw zVDW;=N_xv#uvX^Z{pedql7hCXUHWjAaZsNg$wcp1X~T^6C_&CZjrPC)IH!!VPB>ZV zbrki4X~BP1pU>|)-X@V}S{ShYL7JrNLD@AgkNQx4SzUi`L5zN=*ZUfnb|>4!RiR`M zn@wfWdBwU$z*)$w(Tk%G7Rs2`oPbdWQ>2Q{pPjQH}x(=Bn50(;YMrY1|a=td;<^jKnTJs^7RAp5qv0a#1 zqbHPjXjNBOnse`(`rc37gX#`u6AEHZi?M+r--uSNt zh?cF|M|X7?Rj527BRJLKaHN>jfg@~)MqNuW2WsjUKY=5!gs2z@?QR@lc1(r*a@2oF zqLMsHTZNhr5#y>x$_x$mUID%eFpbE7BMn3x7%H97O2o;eZ8CANJ4?h0uzI5VHG}*U zM$8cBvS?-_U2xocKvL2= z%rYzzkbtyi^EdGm+L1627T zDsCuusd-h00jXqR-3ONz#p&&dd_HJ>2%rJ_&QE!e>r8@=5Zdkv+BZ;kxa_k8b|2`soVC^M+ z<3}?NWG^Edu=bdpMwlja_L6a*tV`ibgqcO7;92IS`K3|#7u42>njuB2fK8f%iNRS-4O)VNx{ z)fiBLG}@RnpR*<>&;orUF*Qk=U6|!Y4uWy{Gx&-}?P_}-xOEmgKSeZlhOiI0dw`RejYBj=uOq6-N8&kNZvYxLQswdJB9VSzSL zFE1lYCMiC&JB-qgQ@45wEH4HZLxiwhjG@=Om~&Fh+REaTvWMJT)lG#{?zPX)e9W0X z=H{+iJi0BYw=e<%E8|)5_IH?LP9Nk@ZiKz7n)-y@_*?xG)iJO2D38p}`RIpVFIdGw zWX4izl;W&Ue|@uLk8X^CF|ob_ODNfIvBX_5)*^{2DLPL1Zr)tzt7(q5Pva@jBN(z{ zfT0fLt_`WTS5(o)N1HbMlFYi#xJtTiK=f}*^c;gXb#-QnLvFkG&Ji~#Ka}CP%n=+{+QBY zwN~PVU-zT`wv!8ZcF{tuY>gTkkMUVeo@0J!NUO-#jOFsUn4aFPjW1Mu)-XN=L&HCI z^`1`aw{QTp!I>Y_k4pls-o|{(jaLU=PhCXk#?aWjkkji}$W;wV1eY5>kE=%h_<+VNA_tD2q)a_+h4UZON)3Y> z-X2(nqr^F+f5g=@gS>-@#2Ux)0`)EXUGBsZd<>^&PR4Uw~#tJltig}JYA=62@? zgknmSATet1#upDP?}ss+>kZNdzL$7}w76f}LKi%LIac9i(wKwPU}E(5O*Eq_46~co z74x;_i$i&~KMkq-_)BOC5O`b4L6{y=AKhEHq6at}!i`5dl>9+ZD}%7d5rxU&FdNSj zqVji^JjrRb=fLc5;+>!3RhNH2MsedF<~+xQj;k+fKRCeB3S(oTS3X~`g;rvnJxr8$ z;Z$>vN$4h&i6HL36NHWTjLt{ku0;;?1t5csj0w8YKnG0tLC=b zGA#cxeO=wq+kb?uE!~(8aYLnxmBl1V+g6C}#5{L#>ih|V3nXpx3rD*3W*NE91kj~u zxVH?{n`k1o~QkSTaZy$_maKptu4 zgL!H1Q$g`;e$-GJM%5m<>m>Vlts9gl6EvRPoro-^+y*eHP521YJ*+=62ruN2;jb;8 zGQyyac3+{in)Qda!5B}!R-ul+11sJMYD2M8Cj$v&+fF!=_6k|&QtC!!%03?KRj2vK z{v9DgxI1mV$pTnI+Q0l6X&l?~qf#4lnbBHYma=AYmh5os9!Q*HmQH(KgRUkXdF}r|Fcm%-b zFWzF$C17s80#OyMwR9(U%CO+<1mBEdPwWpY1*&tk+pk<}pfTJVUUB}Ao`=QOSjk-c zYofz8;X<5~@4^C%vxOTg;r(0RTgJpLP1?WH)R>z7BwLED?EcL+qIA_Ipa}8E6sY!Y zDf?Pi_Izaxm78FqeZ*}Wt9#^wzD{~USd0K9LFTa7g%e6j3J{WOtrBaU;OI{ZwCH>3roF4l6bq#aVPU6 zfp?!-^Y;cQwH@dkY+#-n+)i!uPIq9ZtwHvnc?Tm|WC@k)L z+4DDojSkYQrZx77*{pCi9NFLraGRmqCj1slb7ol1WNvV0)ko9dGre9hnucj}e+x!F zW}1C9XB($pY8ff01Dqx^PiU1?bWjp}!3aonJ1DYFVT34{4iog+V$ND9qJ|G0fgy;OuN>iz zWfZA4m9$I!!##Q;tYL*tBV$BDWU%_0@NVQ85AujFrbcKK!MOt`T1n>tV0mHLYzIe8 zYVmNR^HTHKH1mbv8t>V@&2OM!pL$;`UtD!&9d*6%jOP8^7}ORuN?3f(<#0$o=z7M` z&kEI>xhF;a6oelvYZ0sK{;!`JBOLIhr+)|Z^!PWC_YlEsV;?x~_WgZ%>Ma#2+bti5 zRaRAOl9sb5=g?HNs;D*B^JW{jxVf1*QkMn#x-Nc<=PZeWAR(=s7k$IN+q#-(_YwX_ znBeN>97ccZF;^0AK(9|@i-T|Z*~|*nPvQroC?^Y|3FpC3uv19UC7gzKvNo@0`aiSy zl=+*`#e=^T#^s#^D^aknxvj*7?BfDPTZ&GJ7>u;|boz66}V==k+hgX00MPQiGrn|f< zMP)>}!cEF9{y120?UNkuPlJ*LaVN5>PM+L*5~puPZD~WWrD{bN2KE*6u)uIX`B|d& zCBhg|Iv@QfdjI+OhNSuQt_p@5;WMH~(DSK?Z=Z++;MsznMW_}Qzdu3}K!F7|hYF+m zc3;$2>V|FUyid(-N44c9ENQ+0@-0M_wK!M<;mXqW-V_7Vr3*G-?t*vvr{`e-YZ0Nq zHOH!0+Tv27EI!)3?0ixUNDrkaElG^*Ufv4aZ`ih~4jJt8cyHb!U22uOVI%NaDz|T4 zHq~9Z`t!?|npT73Tux|5)Nx=7+*^tg?kY)wLHk2WlgP)h zi7Lx2B0vEz)f@O=BZ%C9m=~Ky9w#GcSCVuJ$sfBT&v~;I=zc=Q59DAcv7oSTim*o~ zy_}y3{s%nnGMo#w^C*V0nDsh^?5Vi!bV7ShTGOaIZ66Xz_N>g^1TV|-e&WE`=QC>`b#c@f)&9% zDdLxN*N+g0BDX{r!O06To8HatAt)2~r*)VY%aJpoI3Cvfc2GXu*oA+;1qF@v)4D%L zaz$`Do-6JFj4JeDetC7+Omd@7TvOAp9AwvA(2XDI*3E*FQg{#J$G592X1)d@MEbsW zbu$4CE-+UE?Y1to3ytednN)3V%@pJL4Yx9C`Xc2*tj*WmPT|B8C2v5Vc%k%@1JPbC z+v2#E5qmt1`I&cCRG+H9zrs*$HYQZdG}RJGpn0amPsUNpn5~C^t#9JyT^hM4gvYP% zh^J_>@yDwL!Dflr zAzFPdIv+MlaQvd8pnf{ODv8iqZAW8+Rn16=Ge^?)`RzkKFIi6MTfVza9;R#VHpMzv z_;u;t&`+^>CWF<|x(r|^tu__IN2O-(m*1GbujRuV`PBji&S-~uxwyFgG_}H#VcAM0 zNc*53<><90MpVXrt=Dy`fa6?TnmkUg0`50E_PJ1-O!;)Z#uSQ&+ zC}yQF!MM8t?>xS2{9$vKaJ7RZypwQYM2BQ>%s>_%p-v)oek17NShP2I%u4 zpzC(UU#*KESO$Cz?jxrixF@1!P{w#H;xDHmsR@Ya<@uqQ^tJja%NLprV(3e0re2J! zXLWLmE*f87sc>w+8#UnAF4~JoU{kMTuPO4jo3jIm0(7iLY7Afb;6Upa% z>#-|uRtDnXThWAM6T*)yUd_+5`JneS44Ru1Q-1R4=i*^?URMOfqqeP=&uQu47((^f zsxGXPG0P9z(k>o$7p9+#ec_{m|{i?jTSk>?;-h{9|r{Zlyu9FVv6`(%(|1l@XXM1@Z7$Q_>@49xJ z!N}hs&tp1Mi`F|AT_RF-8ffu;(1@#z$P$L+XP4oWTz@@lH#T3c0)uaI!AS*(mKnjy zT_PnGRTbZCK-*uL(-PTaQ=dogKZ?#`%~2qXqOXL@5{JTFcDTC*p1$eje3R};p#FQ$ z$pH1+7?I|yOVaFwUch#0jowFg>URQu5jiT(|BuMs8OcJbCp=FffyW&;oYWN2`KWm} zt@qO1=i7`8#ThWGnGj>DMX{oD?&zbl&D8txpPQ?oHA<_Yk=+$hZ~_XPrv0haZ*2Vt zh?X-71|(9zKex`;b#@82qI1Vv-Q_1?2~W+(jHr1uxf(N(oT%a|8p3O{*gxYJwnZ7? z>eagIY!7$vtakKe6xDPFowul}qW-7?m~G-*cv@gJpS0(8-QUF_g=k9SGquy-qd8H- zqy-=`*ZAN?oTSmSuu&-JQKbYiNq5QT0(`2Dn5M2d$p82CY_vvh)Vy|-R11R`?CrJ0N%Mp?u zT)1`aGXyDFoesICY4*}%B!;~BI}*{YdmY~&J);APcS$ufB#QI3TPG#`{5r^~DBN~= zqZKa25lXW_c|^Pyr?%^)&$}`sd7RQ@UGE93QLi6{ea(Qxwnoa6XhH3vlARJa^2`r_ zke;(8C9UawcqZU&b+^OVgFrbkV&fKvFvk@tPQIEl%S@lIA0y5Z_wh2NgNLc{{N%)7 zVpG}ceO)Gugf`=5PD8Yx@N&P5B}r|=tx^ZMR~brs)>MF{hTOOD6! z-IzU8+gGX$H)kIjcceCBY<)ow!#7jM@P9R9nKgQ{CbYipj4_c2gSCnUMNi%_I}`5y ze|*(Zj0BVud+hWRsG7a_Epc0QM?O$fc&fimF~}vh)XRH@gP>D}w%S5DQa$z<_{#~+ zQ&H@MEf%==pc+oD_G_K~=MHQNI$4wzINQ-3B&wL@SyX^cy+h1f@)IbgrdTl zD0Wv1&pPtSNMN8=l9{IPJ;4^@ff?+TwPQV4guiVs6-yQYcziO*<)kQYR`E;F>v_KE z`Mhxqj|GPIr`-Setz_6}7cHda2x!0-ZGM-#@l-D2W<<4(2|ke;=ag~MIKyS;Ffi|S z6p1tZ-TgtjaUrdbA+4yDXB4F!06V$pmU8U5>{%dj@}qBg0!xxiZ_T-)x{1cjKQ1AF z|61An{~gfk=`+;PU9yLgzh#I(URQ zpbxQhy)m(V_>Q}vhO)zNcydhS8+z^t7TM0C?W$cNXC`K4S$;Z24u}u!;F}hX^J<{w z`TV4cy$rGfnHo^~16C@2>rjByl(i&i7Xdxt?3{9Ml@hIvo-PHSYXBV)2(7&DN7gSB z5_uPB9WEMWCiGzXI|pROy2NXeeydJ==4J`WbE>@S#F?tSGYK}#4X&~YJ~_1*aIf5TgBP zI-4B2TE2*gNMlV@+eUkf=76hc_|m}|=Rd-Gp3ejmh)f+>G}d}|BIcZFh<)Q?QX?0WuAux`B6lh0(t!jFfmf>n|Z zRK$qv^t-a?I<{yucmX5ONkxKl77NofCSylSb%~4%!H;f#+?>a~G(50mWAtf;TlHz{ zb0>(g=7$@~<56JqF5}elVK3}#=+|Vd@93&=)vu+cuLI9Ar(!hdK*rLR?Mfx$p4RbC zy3K>y%rAr0BLkBk$126}5F1Uo#u?dEh+pux3C{jj9m1Oug!YO-RzyAt?oe1GH3>ca z;O@KL&Er~a;{Q(DQaUw6UuXIuwm~TyD5Di&F48;u07h&`=p@xC-;EOS@$aWQ=rjf!>Xuc*{vW#h3@hQ7bS!fVNv9A!kHv1d{?V=icg-1O^im0H$+J$+bP!}aSg zhhLA;2MVlY047kuFNQl=< zMNYzP+*C_)b$o!1Aq)-ElzimhgY4=XD)OqdOnQPNlW^ZL4tv~H?&(9)X}F*vkACh@ zTMmi@@IV@u=bAa6sjiLKd%t7~^@*1Ezr)xUT*L=fwLGWxb0uk(Y?|P@g3ofmScl)4 zdl9*zw7*e1EbWY|RKdaX&pd5FBYa)(V&0nvc+9Uc#XhzPw)cBUWB`?dS^Su3&sC$g zhg#FM39UCR{>rN!Q<)U}JoNZ=h4QZEPPxL#lKO6V{`_+(3N^8-E@dlQ%wdvpe%9Sy zUDOkjMDh2NOkk(;j;AK1>I~#NF{fBWuDvvlEhFg+oZXW&Kjj*x3n&andXi=5DR$Q^ zthkjF5Dh-YQpB=Puh5ChrEE_Dp54R#l~g6Du(eodi;Dxx!HN0vLhG#ZfN#2UAxXbN z$q#V+P|c^6tJE`E?0sbP+6P+1qwk;z>_cH3rAKpDTHMSmw7WM$b8mcEsUjuQI;7_L zsH7MKK-WN#a&rlA6L}VZwAr1UtHeV?IXPNp5t0Hj^C2ebxD#vBqZ$~Nn@I~NPuP3@ zeMaZhmzdVj!LS1SyhWcW_YL3Xre)B#^4ym%KBCai>RjVjA^5`P^G{L?TGt}4p{aPl)E|^VS48su zg0Z=l&heJ(c_x$?uo0`} zo%cXNraogZiIQ70P%#^m5owZSIHJy(v!^Q5Z%35~IQ#5ze8LB$Glf0_RFt1$6`g>T{ju@{U-EH2d51RseyJ-|!& zl{aCdPG-b=a4Ao9lutV;K^r#+Ic|Owt+xqX%JyZjKSBA}0mF3OiP6ZRXo=o2_Q0an z-Ag=LqV611zdv~Sh0$n}ACqL8QO~CYqH1>H0u@EHzH$a;WWDVnrw4D-v@^>$RE!PJ zy;EveqDoS$8;fZ7UkAARFbe9jCiX{N-J*%E4sA`ZwTc~zsjhj zrZYQQuUC7{)UjJ$VWdR=2IVVCA|7xK8bObZ(94iu)jMDg_P-C-h^2Ac237urhq%Vf zo*r8UX%uMK%;TQGcf0kvioA9 zwzxxfXU{2WrCbzrpQKxoU^L+{Sf}*{4JWwmgN8Bk%(AxnjPl8Wrfer%^^n)(C6Ef_ z%&v_2zg%}e`ZQC*5D2vxEj)w@!(B(Eie)hj6$ZItR7ocHG;F)kyk2N9JeO5WSUFIE z-I#;^b{gSH$11mgTg~4QVx#zwR@)8T_wbh4VNnJM+kDx^qmv%Ncem$e!H$Dmxh4=ki zVFdlH#7aG)+*1Bq?>1dXtSpdwaJ1(FJ50W%^8=}~)rmdd&$(94dfTy^iWKjJ z`S>nZ965r>|1jEdq&fJOYn(}h2Fi?({fNZ6Em3Ua{KD}PqA{j92(|#>ylMK=HYyGK zOuqc(m@e%{v4us>3Tf2}GbH7h=+KTgDpB`w?^UqrMG$anQoZG=SB;kmi3z0V_Iq3~E1Rr4R3>s-qx|l`Z)ST6gTcW@fLeN#_ zT%dkPE(L=uUF5Dp-QO?aTz!TuC!m2sP9Z$F`gg=;W>Q`5T7-i6spf~R)s*x5tsttN zS-SeJX|gF!&xP$!Lrl*bVu6Wb6JUio77hly@p?W6J7{&L0{o{Eu!ZpD&@*C6PVsPgdN6$#Ih@tPGMqo&`&4#ru*3 zhgUm{9#V#k_@`?HaqK9bxvORnA+m)OZ#UtJ_}$Y6<6}5zin`a=b-5|jkyLXlmk|1w zqM8E*ZAG4*401g{+jsx*9HsrZt|?oeTW?oEoFmg8`2mk9;-lhM`rO-PXHGq>3l7%v z&%W3M@h(xC=gH8$J4Cx})iPA~jvIV`3*&REqP~~^Te!^#;4~=UCSW!GL3vKxelnDe zq~0T%WexbenUvfu?GIX3p$*J=>Z>fqke@e0=dsT`wjVh4!#sBDd0gXf({#J+Reo&y zv!cc~dG3NoK3;Xl56SU2U8CsR7!~{oh3FR6X~)}6vnU{| zsQO4{s*a?GmYb(X*q24x((6#KOcwKK?{&e?VC=X11;;VeEMHfViJ`yIL^5$Roj$Xi z%-&a4v0M+a%4^Ovz2vjebGV*jj7W?dXpOrt>vo#d?{j;AKUgSUR5O2|bce<;hm#W4*m#a^D*RKpZiNdMGHFAM4Nh*<`E zu@2h$=LXEGANZH-X%y=FZ8*&=+KKGDUpYvL@#4P?Z_h9NCqUh^TypaK@cEt#;t>6BKZA|y>q$!(~;V zp~eq)HzVW;!?W4*Ct#K#+-ivW#YY6K!Kn`~`H(=}&eDNVHoj=^^S6QB|C{>iYj>3d zhyk^<4TaqfHZr|V?8_Z)U;csP(6lndY@%@3u15#aN$vVm!jQr5IA-@=x&zVQUh3riCcU{cmT?>X#7PuXQHnYBiY;-? z&?d7`Ja~__1$6ZtLb6gOcEp5Ug+#p^CbCZBOF!$ufg}#Zy};B)9Fp{_fX??FnsndHCGQgOk`|A8D zRZuV$7JLkz4gPKsRdWuJmCNBeC_YU8B<_q*h7Dysun^{9r4+B;DnKUWJ|}<3l|BWUEWKM%#1Ics6P|F} zygJK8fpGRNqep=Ag*z!r9%M1jMTq!VGx623g-E#7-{o3SF##*MJJ^?Q?E{*)S&jW~ z+xy|pruetJ)96o&e5FS@Z;QsJ5|bRhO&sGjPao%jr6V^h#2KG;;4Y3q^S~bpVs69t9}N0 zy#lN*d~RMnb~jXP1P9i*-DP9w9cZ!qmt=g!gdl;YmNae8-^pzA_EVdpDbx zxQi&YSpL+1b;RpIrnPpX`j_jA*P!#4ABk76A!rX@&Fu)V_q1qsxv8JVNvYs&e@$^2{ABrQ! zf4(75bbk-AGaX1ikT%zn^OIih8IzI4Xt18K1uoctJ5~f;!>*h^jdGgX&v15lH0m)0 zYBmGuSzzIWL34XA^F^W@F8eg45M)QciQC^MP&VC~w?%*FyD4hU3oNa8^7Pz4RtnAY z+#Jy3_lzq`bLHDaQIcj6!8bk8iL{pVzoW@E_=ll9twQ7UX-1sL`YwdxZRK1s1KE&I z@!>w(cb>gE3DvZ9=boNFhHZS@fOG%HeG-fd2iGT^4JQ&qgF3Y=dIMeZ;}Q`Qx!AaF zhn>=Ss~R%j;(C+O*sbCiSl3FW|Lvb6)jSIzJ$_PRYlmt?B7vxG!v$lotWzdC8sV_} z-<1)av|@MNVeC(vVz?~s^7~iSh@UHCVWyQDeWQGz^=rw=IUE=Ea#F4ct~c@__vg6MjR zrv}jnTkp0^^R@xILHxplKdC$!Dt0WHCH(!qLZ*Hp(XsusQKd{-rKVqfR!^DCxPvbI z3xCTX`X)-$4}aWzay|X3wEdeJ(qG=Uxy^H~m5`6#^hH&9ugBYh+g1jwUE^qPy~MQ} z^ubwo0=bOX&>nIrc)upp`I^0XXyW zqFo*S>}0-cDS`A~p)y5&Ibm!0J&!l3QvqO88d2;iu=W!OGo{aH3GY*vEneZ0L=iZi zIqTW*bJuedtC2GId=ys+2F(f4mKIRL6d{>a&|WqU6ROT2muE_Wi zR2wL7+r%1g8EiiwAkN$Bu?tcO0?941{E?(jN4GPGWsP3sp!H|Qu4XnHOlPailq7z_ z|2M49Sekv-aZYUa8Qg(b>XtJPYI#a?4kNV!+i!;gAS zF+=`;5X2I&(o8PuT~PDesIh9dfz+ebX9FvxZ`k}0CoQu0IYBsgQj+=Lv ziD4Pi5}L*Db+YwutY<}76|y*j_F1#L28kWiMN9B+Dvq5<@pStsr)9J)pRnb0mIIs# zFa7#h<-<#aEZ){~f)+GY%o`Enu^fS7uV$iN!bwH&L1?ZO$a zQYJB7wwEb3o+3>Y+4Lyx)B#Ns)uXD8jt$s_UgLC*Y@XHMWkYl&Qi~W%TM0d9vNLD# zvU882bo0x|oiTsMD9tRcT=krkBw^lwc)zn}TD3(g?A;_Sr0AK5Kur-l11oSqFgKpY zh=OYp0dEtp2-uF2{=es7*QhKb@4iS{6ON)8rUk)u}3LWkbcbo;5Y7%*th-{V% zb=z3nSmdG~QSuHaw`q0DQ0cm|xw=PJf;~YS8WwE00s#>;fY(zJPpBt~qZwk6uo>?o z!W{&vY=-hP_3AqI<|mV&WeN@(7hzxCwHd8+M5x8gpxmbaU8-EAl_tnG*gN3IJDQjE z8~mt0M)&A~+>C?-p~7!AP!K@NcmyKoS)c*2^O zgHw`VR8;IfCZ_E{(i`;#$PX>pWyVJa*~t^U6eX zRTERN0_$xUp({B_I3G%HJAwFfkYpe#MqTLKlCh;8I(Ul!juc%?1@s=;dJ5IegIX zB}WT$m%%F}ZuH%Et2-tNQ%uazLJfGcG}$}fvjzyDx_;% z?t%&#k0_1Gty&{AcN|OsYN$78iiFC?*Mq{$E zOb+_8Ku)lFU2}enkDGn6SGlcoGK>xRj%{nVrxAzKbTnvZXQy}nKi;Z3TkU~=`?r)P z+sS~i1eU^hZ(XFOJGA|Gg4jw~cNG_}Gv`tc;ing;qX34R<9JijdC=dT6+9#zYZY>Q zlz8PS=#m>?p^$i)O?j6qe0e>p-_ix-UJUNNo&BVg`%mj^b0I?VE-#ja5priSY_{$> z5AWk`-l2SdJQ_S2+;}BCAEGhHM^?#WKa&CmW$?8nt95A11TwwGxPzCjzB#;o;~sRZ zEXm2&pTDk&|BGl8d8Mcu5pEO|$0!Jdu=FS~g@r5eq*zoN3#|dB;;_lzZej?}ZTc^5 z?YO=T9=jn36

CQPzdE|mc z<7P_w*%g^{dCa^%=FBML|E8=7r*N~inS+iW$V28^8mzjHbbS)j z=bYA|`lizjZ#rRV1@omg?zs{X)0RztdN}wzBxqb8tKRHscht27Q*_rt$IU5TcEB*ifc3#XX!)?2}-HFmUERUz>bX<;~Q%S5lcZt2|hU}hT zTH3>A?h`$^)DT3dj(KRXHq2YfB+)BebX8lCE3AiJWV~_BWhzyebW4AZIgXJNXWBYoBoHF~(gxid<_>DPZqnumzAa_3qjar9mw-zU+-Y9M<#$wo4 z@KYJls}TEG3D-`;SMBx`KfQ&WmrjpZYCj|m(Xk#E;g$oohn`vawfXU&0~2uR0ex~( zvyXUW;Ss~wfT?&U3RLaYi69s?M<0~3*mksGI=#zDwRjfJCrVhHB7qltr2)y+SYYdf z{o^%WgrxH{SQ+B&Gx~JFLzvEo3^F8cJBe@xT|P{I47W4jUz}@UjOmKkTc*yOdsPHn zMhHFs9Z6KcQ$(_VXE_DOFPfdcnsO#uB6{McPr1V%n%q-lciZ-;t>oK=5Sbj3T8J0Q zVQC#)922Qi6S>x?%jO~oQY?LvP@9Z41=;^@5wdcNaoAKnH?#5W3sSbjuipX#FU=_V z5=BnLTnSLZN>dD)zzbwwogSYdRKobvQLap|29fW1somG;w^BEqv%02qzspFxaBjF0 zhA@<-q!w>Gth;Y+|Iu%c`tsb*1RjBq*M8`*&K&(#c4DN>-fj&2{)Ttcy3HM!Si!id zV6EMpu((>kwG8?#!^8DcKnQTd#8it?e{=eS`kpyh(Erqw>r^I;T6Iy_5Rq5`GSLTU z)I2%xis`{^Gy0rX{GoAn126dAh%42PsPch1N2$5_xQDUuSTw&HB|~=fh~>UA02xpr zG_0Q3VGNaF9;^%lQx#~^Ck?%fvNUS3;$6-`E64XsKw&|z0WXdyFDHpmpgC?thfy12 z(k#3}p!%jN1?^Hbrp#j%+p(rl>IrrI9+~|wfUwSOrEUJwn6Xxm|GB9UGjSHUgpaI` zO(yW#zEm~fC&0#pP;akf*j6vboS&JD@pA|eY`A|&U z^!>#D-#fq)rbXI*ug;#i#c3%svlI`hRO?sbVwy729lzqtP7j+;Yf^R!zo$Notd!Tf z;)xy)Ht^aSN-~UUkLn!Ti!E97{8;|z*pDTyM zpDv>?!cvi?FHTEk)w784Xc*ZwMwA^vRl`EZkXhP>;ctTaC`hf9m@ zK4)0;i+!IK&6)RV>SGu^_)*FU#%AuKMwgMu{C^Onwyc{6$6<%C%eh8kQ?x0!UJd4p z(fQDbu#}(lo4u;1eCbDwc&_C$KNpVa7zy~o5m_|B5ST|`KdKT4m+)?ePUl`;8rq&` zkag|0!ACyuZS~^+y-b&+X!R~u)$TUjuboW2Me>P0_qypS`c_5LS-d5DY~c6yL;fVd zsJ7@?o{bHAtv1D{%2gc|X7I)WVLn(?RDR2sk>~0siH=83pJHl2O}w|%XkkoAQS?Bg z^4r9aoiW2uj`iY9q7GwC*Qg8EJ*EWwV9Ogli860Y`I61zO$=OCZivv4(RA@HaarEzO$jv>kon2Q(eY zxG*~i#ky>NZc=XWnCPsi$fr)Iwg#J4+rJ$pl1Cy?M4k~Y8gKxejGSI=D16D5&)hF)2)ffB9f74Bx)M4Kg*m}bC?lfK319V*A`YXUL+K%A zgfO<=i|Gy7ucyOT%e(a_nDSbrLPdP;%|RT1T-G{%zIqgEVG>DmUU_y+!mWAY)S0a9 zIpL12=a_m3duDm9kMp5M{A~M(#=)dJBj>|#8E+)9TS!tTqFEhK(S;~CU?<7!w~S@2ppM@oNmhplEaA>WXHTt3m#S-l=;`x`$MJuK-q z*}8|)%YW4_d-}*GUN7qU7mj<_-fy={nU)!`Dw6twZ8Fb_6BuDly0bn@Ie){u{D3&Q zc#_C=-AJENP8^|!?y0R=9XT(d=Xa>>zTgE=S%F7>J$rJaOMF&;Lo&D3URj_GLgYcY zve=KQ4eAv*-3WDK&4soN+To@{5N+p?`3MpzJ@QFOlF#NU!F<1h@>4SBa}9$?hVwIInSd!j>68J}VUBOGlqG1v zSO0gVOFrcVYoeY34>cPuj^)Sx?nDU#N6+WuTy10&ba5A#V97<*WZZ8I6v zW~xt)uNmsT6JbG3l~E-IZ0Yq&RyC8saL4o@K1$NHyhi%Vx+nX7E0h?qG3Ywfo0j)A#x15 zfJ8afEf|F!_KzioPI*I7_*t#vq2+s!z-%U_6jBES?V#7juVM7x^p!L|KAcs*g6Q~I zH{(Y|C)1QBiAX*>EpP@F!IqTe-!htM=lV6(W7YiQ@b1HX^o>##rtX;4-P5)2Bm$s3 z3&~)wm8?rlT%kEjb4${DahmCn+yVC~o1cok3#^E{(%t?(FS|vXnD$hP$!-;!iyAgT z1gWJWv1;QFZYc5@sn$Jz-Bj;m1fPTJsPz9gk|N!&^*wRW+R0)WHgXi1cg;&e5lJq7 z+_ZlT&@yfjg$(q)!<71haU*qOq-qWCe`CH~E-Ivl<}8bt?k|r_h8+{sgN)=SpQ=-@ zoL^v=IAZ6?g${XRQZ(K__(>T2=l9*_Z)l($-%GewwIOHI^ed&X-Exo}c_N#k{hu3* zpGE$7MO)v);pX4evTk1c3{G=Ixzb;YIvu${B${{ao17I)Wfn&bSS(`GLzefj1E&ov zV4QO&`6Z0rG2$x+vrcIonGd_*b#`JZR||-{iB%6uLGo}yP_a8)LfHf}smZb36t;AT z^{+uoOJsCJM}%-fdiBL*kqs)0&>vhm_H0b1M%>PhH=6A(``g@}T~=xC{S_IISRE>> zW{O7%KFFrSZ+bT}_b^|8<`ROGdBJt06caZ2OIVBctRaQVtY&zA8HO8wffz8P`hz^R zpT#lBQIuX+6TkVh3UK;CGm{|zzncfxO_nJK7k8S%3KV<)bd_29!Oj|%zd|XLL=vN1 z+D&BTk&B`u$-x5FaN$UZx*{^`RI?)cTXil^{6^q zNxvczIS5H*bL(>9=>t1;_B`go;oTEv$F@5S68!MdtNu*@C!8+62XvWeoK!jP7e{{P z(vcdJumlvmA9QEEM%#0x<7=h@DYt`9=E2vV1y@_8*WhPy=a`>nA{Csb+Fb5Ky5<5M z3(wLxs%2R2Bwjjsr%@nf7mkLSi$~s}D>l3GgO~g4b@6R->pF8XgW)40&_>$)u;+4p z){=!a=J9j!Qh;J1&AaG+#*F94M7Mj6wmZi}<8`_-yUhNTTRZ~Zc)3Zc%izT9sC${& zYZb~=is1-P^`#DL1va~$FN%`>Dk3aUV&wDD;#D(mib=g*yp52L*gGO6u`s!AS zMu^YaDU@n2DYOq`t(E*X;7Av?JvZM%FG770ZBg|ur{>{uRiC7}q$Ad9s3M{-yHyX` zR9v6Npq>eO>=;oyUt1$}2WgVk|M)JMtp>eH%O0v|(nmb*Rqv!4L+X3yGwLKt%ZGuC zp>^u2Ca3c>%@~mqvk}Ilk#4S%_|IVq>IX26j);zw2}lj9#!&c(dES#1ok?&ed{p=B z^-pX6R;1+CLrsXW+SnfW%$JyPv%GN=h2UHW~;Y50eI{(iia_3 z8=#OYh%uwl1EK(R=W6^oWGH(!|7~$znc5jOA_q9d7!#;|K51^_Su?db63#oCl!UMB z-V0db-o}8CGA}#O`g8N1nX6lY$uS5Da~z~28?n!ot9YO6sE1j}4|wzdMG37^y0e?O z-P?CAq8N)7xHI#0I&=6-2(hIe8@U@i5w&v>jzzKE9y$>5@6-F39 z@-{08vS?yNSHYq*!%A7A-3g1EKSB%3#8v%&UI@G}TC(kYwv5d%6(^Sb@eiL4L(}DKn-oEcPK5Jg^}L5&uvvN@YN zMeWb%-9^u$R@Ftq{)P5=q(#c6i0yJZMPHV;GwCD%T5X!{{59*H(Db=aQ66iplf5y_ zR@H>NWeke;PJMEn<|u*ek(5sZU#{!Jh!BdC8W%)Z4=hEx!bj|cNo>|rnd_2^KN0%lFM4Wj%Czh=r^lE#t}N|LK^R1bY>LdPjdaB!Y+e&+%m)PM>(WMRG*%$cpv%J3N^aKf!~KO_9TO8JH4|=S~RPf9(n~2)pS;1jFXS;*t!}P z7gWJFK*YbU!LOt*&R?(s{QFV=lBa^94WU%oZlTCxl_-Aqn~^+;~czltYXZ(TUa6M7VQ= z2z>;}UjP=+k4z5P6aB&{lTF6)%nfGaka6+)iCYcyK`vLZU*6rB8)o?u75_}l=+;jr zE4=wKj!6^SZ%K_1IoValkTPb58^M!ftwrf;u=m5@?I^8I2#IK0D`Aiad{dJS*u0I# zj>!n$>L~2B_I1Ou{bIltm+cH`Jx&xuU_WhMxL*F1(vQWF!rbR(BAGxT{;Wz38Uq=UnUo2L<&eY@KUz9DVtJ;82XJv8P113 zWGzmH-*7Q3_xs*#t|jmk>O=Qo$Vfgy%2y@cjwxmF?_3JtG^;$}j9m{oMvV{z$d-4- z_^IF=l2s2)Vf)RnNPUuo^xtt z{-~2;F0zU!-vCNudW$bjAZuitPU4e@7?i(c0Y`z{U%=4Q*Fn5P4#<@C?CQ<+u0pZ! zaGY_<*`^FGqQz*eWJ24&5*)Tasr_k0hl%_sxRP!@k0f$|kYlkxH$-W-5>8(17u4`I z;Sn%XOA>2t+PkPrNo$<1CRVJEbd4)ycSs|s{$L&=yClsCm#2MO4n~7hNs-Ff+ZYS% zpZRmRozP#2+Aj+M0XC)Pjn4~6~!{+J4*oSwmbFCe**AV^A)_2Mwtq(AgkD?!hXu$-4hM&qcvpZ1JiID@W!8uNLxGo%=2@VH`@1U0W)L0zW5y>x9h^MlhVc0&? zR(Km_CFKrz2mke84g-xw3LmPJC~J@MqtoT8=kI-`u=XM30@bF5e(`vXH>u5Q-e~9Fj$b=tgr3 zrBPWrPQ8Bz@!nPh!ie3nTgkPzYX;HS^bM}f=y3lFI$a&*)C1)bM6nyzE*s94w;h4F zAa2YUuvJU7M(&F4#K>~XB-6Pp_BgB1;}7&BjiR_FhjSQsBB8s!ysR>iMtSrSk)`?GSXhJ+IM)qDq1=)-8c5; z!VnQ`D2+bRvWa;dI1w?m@eTX^)LCekZdAKjR3;zYnlkPxV-GyfP>t16!t;553(DCz z_UOhacJ4~`0E38KZAL0;Y+;?{y_Y|xgI)Zoq0Fj1-N`5-WjbTMkeeq{herTR_hmbp zW`}7s8lkp1NQ%n)nFR%Uh0utF5kN7kxSiV$x#4!i+oFVRkGn#i%G9o)@f;@1y^OYm zzuiqIH!E?AXfSJTB>m>Oc7_srB)zy?U+}Jb!uDn?6X!s}a=(%F&V*mCh5fLS9o>m2 zJ3eAYT-|b*1;6kU>j#A)i!w5E6e+dJfYFFV_$1jSaAs1Wp*2F#7TY4S+ z)|X+(iJ17;`=jD{>*ztl*)Aq+SI)A7ngAVoKSHu3DIOm7CQyuASjUQbKF2VtJh4(p zLbNijSsk%7iRG;#gjS_@k^*o3vF2TFU10!60^wonk@`)-LIQTVdX21hFG?`+Xr9K4 zbPvxi6O#=4Ct~+LDp>E+Yw2keLMcD(_u&f@=if)d(Wf-uj>7{s$Bo$ib%TbW<#rv@ zrF!`;MK&8Bl0LeK`pXvrV7T1vVJYr)L1N(>-SHE1h}B=~{t(hR*v5cD_tl#J*Jzde zb%#pGwq&S=9j1*IzRU*ZX}or`)bHn+h>ig0hwNov2P-!qKgo(b<%ErIOmBStvWs-ws#q_u=j>oJvHz53&73l ztUr zsL{vkk;6E?BNyHJ;YKExC{5HOH+Fsoo4n#t%6YDnDl%zw#wZasIxv!j-C2|s*BG(~ zAD?NnnRY}K14Y$kDFp{-_z|y(Un~0q!H(v1>lQk&!raZmllceSsE;r~>_(l@x}}JA zI?WIIP%ZrMFS+M@E4zJfwXq4LFFjYZo;$mbRhampO*Cx;slsg~dLwPTH$Bj!iJOXX zavxZKOc$FFku!$JxkdNG3d#azkHow{&|^9ju0hc2{jEbYEE+3@nO$7oc2im+&EB@V zyPnYN%x%{l@#I8u2PV0E-QQispPWqg{;y~7U7oJC662XjbvBX~i34m*vw2H&n5qfj z@|DW)Nv$du&*W_uCGBNtw|e=2wxQOY-VYhtG_qSB1-s4`-jzdFG2vJ#GZRA39E1AQ zMgMOIarTxWq`air1rhh;Lg$p%q6zW6t7pnhIS^~;P+^VGm0==uSdw)tp1L&A1rdY+Jk#N>VY7~Q5to0);435C0u za>Yn=4Y>#QJ$m{mj7tgjF!Vp^OZ;t{4Azzd{YO%%^QfG^)E_^B+wJ02y6>Mh#_EdO8?0}Si!ADYUn8MC@Eey z?Ce3&#s^i8>}{Wk_)Ke8oto8l-l)4O6ul_TR~FV15lXgB@X^vfO1;6Ii4f`{b_S$O zF-;fJADbijZ^7D?7D0oHx4`@+f4|1JTHSsTs_i$DJG*CJx6~@Vio4LJO0VEPUpeIJ z`ZcZB^$&NdCQqttot5no5__;RUCB!7tU(rDk-T34Y}knWEvAt}cH3x3sFAGk*59Q7 zbcGyA23y1tjWe5*%rJ@V_N*a#&73}!4@E_1zKH3W4b<|>02L698(Y**JO^d}ddIXR zHr8iu!3$EpEUZ+-;^$?9Ofb$8Ye3*xGy)^yMs#1Tb;_B9;t4i)3)@9uy=Y!b{=7ty zb%9tC5?bS{>-AwSlEBdO*KNS@Ijg0q?c)BbCQG(O2hFkDJ#|_Zl;wM^uSt-M<9>g6 z$Mh+n9gbF28(Yz7Sw=CCYD=}4nl)`zziO+Zv^tCFx1tw;3zyw@1NgGzV0hstulSwy zwl!e|5x;ePtCNj-#bEO|JF+)By#<$glf#yt1w&Zv@?G z;ZOzQ)0P4oTJ(;s*N6%OG=#rWz1^nSD}fcuGOw{Y)k+Uj;xN)70=G$H4( zP9$tL2`Ms`5jXgqdgR)_i+|D>D?bBr_nhmgI17%Xy^Dv*4Ltzvc|e}aLlpPu=e2BD zlu9qe3=1^|$F=SvnjtPm4whbHIMuV`;Iz#{$F57#mqi50uB@ z?CPfK$ME&bIqSrLKr0BX5~Mt%DGAA_U#QcN+`@#ICxS9wl!LXq^ss6jk(!>(G9I;y zP{bb~5i3$`*l_HxQpxhuY#98DT-qHcw0n5nUgAvfOy&(0l&nisJ zU9$n&)>33*_{46V`9)fuQPn+`QRc)%4Ee)Z)#K~q2S!5cy$KZJ4EDz-&wIK+rVCk1 zk5S!wyv)tkkVA!{vMid4J@E#4z}C=VR*^TwE)TzO15}=SvaZ*Y2LrL*L!o9Lqw&=havnLsZSI*II|nRR!4TuWyH; z-fC=$EBMG&s_anRkua2;BHgDdLIXekqQ5M4!qS`NR@ zZy^k%(Tyi*cT}F1T^f%_HDusfR|ePLIqr2vh%*S%BXpJ~20PaX+gP@j?YQnX|0;1j zcOTeBf?YNHAng}ZlrP-UOJ^>^lC!=ys}VL0o`aa7E>a(TBih@{A!$_E>=_kv-o~`x zEkw=2;OS58+(Jlg!fUkW_4+`$%7l~StgRF5_RM9BZlJ!v0z+ou>A?CdsLva=;*5EZ zft;j~$dT+gk=<&8@ZQzrai2>&oI|?5L1s1QWdka@SpbPsxd05#^84!Y1C&v#3H>{I z=5?I5IPMy)>a6o-_dPQ0%(QsS08)#}>p0Beszo--q;XFMRd>`5b?Y1yl(e6qyS*dY zte(9~;iopIp)vPf^dFyWiiyu!WjTWVkmGYZMw)D4c>c&^(4aAM0y{3YW|fP*riB^b z4H0^$4GYKkJkSbA+vLBrT|KcIanOt<)I$+8T*Ot%bs3nqBc3A3>5Lw5(7LI+t`#qq zd#UKb(gnOH+6NE1!RE2Zy)pAeE1-H|YSk>`r~Y5dR}177*rpTHiq}J{$xdb%@>DwL z;>500z4s;W+D3b5&gN6NyElxyrSA>B{_2fTj;gc4HfCnazsr5`r+BtdTMtm2(hmg8 zS@qGwk>i`PhL2Y&&xju=!1~76D1_7H7{fzTO0h~2$YrSBh(CA4sHeb7=oU=C0Mpsp z({p|fo$+(sG?V8{R9nQ>o=nzR9)ry=_omdkUfGtFPvoR8%l?a19Of8J z_@c40CDdzS7#&bMV5_MaCeM#+>cAlr@Tb-3u>qdMPvxGiuYLkowpOoqC&3ZC93E{p zTin%lISQ?yZMr=WILuec)0~!>*&Tp5B2Oc8ptE(NoTE^q^M)v6RAEHzXN`Iz&Ge;n0Vnh=Z#8Rma z%>=|S>x+ml^PP3UpPF$G%JE{jH**(^O$8ZEQ4REolFExP#jXr1=~;?{KNhXAt0ige z8g~N0BOJ6J*jG|49!u1^Gg@vQA`ZKX26bZ}I>5ndTQ8i$kF}{7ox(tdX3e{B<6Nc4 zz3&pCTs1SykaR;m;hCE6e+YWs=8;Q@UdhL&c6p18qKB46haLT3=`HD9R52?r5LuO2 z2rbg4^R{R1#v7OR$pZ1s`+^f<_}X?dOoXxZ2JXhbfJz}J-E3V#yYd%D4>VRrNNHQq z@7%$gi^f6pE4G8t^^CZa$m*+*-Tj+)rmA${punQSsd5lrCUZ?|~8c*hSP(!d1KTyQ8kAHv%b^*Pl~CF)~KdN~hi5wl{dY zkI=qtz+YkkZ3TN_^LmBLZF^$uiM!c|WbI&G4&PkDAKcVw@~77FR$zZ&MZwctg$9J# zEw?e@0`Gd5#D$Ykz2{ykZY$mso<&hHJT5kPgbNXu)|rChy!YeHHB!J=l)qHO$(5g5 z9)P(-KI{7orDL35*Vu+?tvF6Jzs2};bSVZRMWvmugqiE5-%8P*fjjAoLVT|3yHGJN zMDIm;xPs^N=+;pI;)l@^nUiD#EK3k3N$76yKe8Pjy?t~3u#~;iiAQa4_T<2Wi7s|e z95+@qCE6SzQ?G@>>CvSQe6E>_qr?p_$rA^S0`JOTXS6@nSYrKe{!~=9cE>Eb!qGVT ziJoACq}x2KC4HFfZv$gPq}6jTDk7{Y-Hfd*D@NIL02UhCfy6tTJ_<@)5jG=}gIn}W zrm9zO!la@+8&9SZFa1#da!##B(h$=p*dOfZvcY(nG%B^noND0K+Urp~GSP?qc!}wV z@sq@~$HLTef(7^x!Y zaz9ofL!nde`0J1%P*5+mE$cZ;;D02zB*qGvX91@T9}XuAuP^r0LfoHoY`M zi*NL~m;B(N5rwq*DNW)Ck`NgXQ#b*T3WAtB84s7sipx}huUolWJat1hg)R+b$rXqOFA|&jkMYv^4eh zXT+F(7m(g0bhz6&X~2`c@muuaN_MLkQpOfScneqpS1!qRAzn$dZ;p zjtG6zZ)w5yq?Co;9HgMD`c!$mLqe3ZP|NX$;RQKOF^r)aHM@xKia&W+4Ls&i8_|n+ z`6;q4yFU^0@G`8H6`MjKwMdwWyvlcL@CALFb0udBk^wCfThiVw!&+l`{q7ScGX%p) z5{vQ?4E-e?H@iNxAO7UWg*cPGwX}I8T>X2P@LjAA{CEvQC~Q}edE9criCJJCe$bZ=1a$ClFG0HvE!a%Pjcii#|#{>W4bgXlSf?;A?l+J>kvG%+VSm zp_^=s!*GsfkL}*!OffWM5|Fk>;!54U%X1wZ`fMMnG9inT%8TxR#Ru{^+xi?!$k@av z*`|N|^lNZ>3%{+h#wzz~nBl3yE&9LnY#(yC9MlT`-+P$5OH6vbbAlvftFkh5BqJog zx}jqH+M{s2l^&hkV%lJDa*Z zvflwHkv@O%l@#~Op9xOG)3Vgsxn75j8VqPC6`eU{^ehtOtGQe?f(t_B`oq>U;J$+@ z3l;2iC9FGfAYyJMVPkKc;_JQ|(u6?W%9s&sPxG!eB%KpDU*rgu8gA048}Mmgj_=msTI>1hKm?tC3opTl!iI0WKqk`AO1y z0H~qI{7l!MQCWA3+$D`GAEE@iieQ0VwqHQgz6_&1vt`;3E#UFsrW{j zLLfE)V#oE6xa5blyv3t@f-aesX&XyD^w^dok`IC&>)WKa(x+I8`$6Kva2HGjgXO`H z5kOGcYfK*)aLk8VAQX@T?cExnRi*jzV$!z9#yv%IOWX!XLf@bF|7iE@WK5u&kl`PD zWv-1NU!nOq$@M_)I5iZ|L}N_2zBuWTQ0 z_~{88Y8h`b(YQn%2U(bR1LA|8-fhc~!~P^a)KR2of)c;w@rJ-*nc+pj6Fwwd+Gp^S z)kF7Lb+&7QUXOQ>v_J8BTun$=A?~eMg>?1HMDy7R3Tz9Ki#XImTeMJ_XOrxPFaI;S z`>io*$bL&yHt1d%dPO5Twy!;ZB~}(9*jGqzL1985&qUOUo|*9`6%dA)ebO`M7KWk4 zCE5t_v;;?Cx72x=a~F(Aa^<}rO*s0Z0245E7dMtG28DWp{N z^)Avw8T%Uk%(T1bEG;KbMBlI9eQ_L#*xt;-=DlpUBW6y*l8kJqKZBqn0)Q^;9M0%hbWu4my*V56}uw))=J51h6aQH44x&f0;gekOv z=?WrxdP1lQQ^Jicv_bR-?B@M9UQXsPpB-*`NV|a!oW{ z8N&2&ib>GNLT`Lqb&B-+EX60EuGMiQ14go?s`~{@YiDwFZuAl#502D~iE*kzc-XZ4 zV?eS}Qk2JBxx1Y&@mm%)T`VKvHN6P>AeE|#Bc+taNAqX5YymtsIDWJ0#HGigL?k_SiAV;k1J{GnJ zU}fc#Z;GXSo_F_AY`EAY8evl^dhYE|wb`4+pXK<}Q$*d#AMVMe()7PG`bWyI%HapU_}ZJKjQPQ0BGt=F zOq$E49~$FI=mH7DIjWdl!;k#gq4?QZf^X`HlG-7vo%H+vs;b6Jow{q$yopW=QcK0* zm1KfuC!Qh|eBl%}LWiDC z_AA)dR&vY+sM5Bu9rWL4!D}iA()g{23{P8xXkhoDHmB=IaS7+&!-ZEL01zOOn3#MZ zf3=jNQ_Gf#bo6k>C&4F4jxea#u4AMESM`)6Go6&H{IJ{vmr6N>d1E0VRow* z1`^?T*atxuLSZy<*KzA}yAgg0#Ywf`QSfh74BRrJxPoqg@A!n?=mQocS zToR>tIQ^j|vS8u3k@HE=W-e=QfyM1Oz{~N)fn2sI*4^)`F^t2==%E08P?S-_TiFq- zO2Ekc(VAMoY;+Q}SC9j9H4lbulQH3QhkLP6;eNl*L?G{YtHyb)HhQM!{6Qn7J)|s! zfB>dGOvF#nhkBgtYI6x?8d93&hQtqko6VgcpVb%+*#j!-;U4cBM z9`QXsZHj;tix#h%U{&!QT+1kpf&qnebKc%>cjvluO##tRRLs06wXq^ga;idhNh)&1 zl4K2Zs~tNZ6T zn$$LIC0E1n+s|*4y!1!7(88tqR_!?|^zrO53UJO^n|p7<4(d!nfqIB~LP8i&#SchZ z?)X^Spx7o|!-I}gyMNX46K`4+x2sk^CaP<~kqa%Jt>qeSQ@JEx2@H!&z_8=#`bf8i z%=(NW{y_(G-Wj`>DTzQ90pq=y-$fB-j_CwCHeO!w`7e0TmZ z*Mpft8iQ{L9&>a#_cv_i5817?H5*P@h$jf0#1gxtye1uGJ1wXwL-&!^xWIz0QoSZq zo9kG+4gE#+VUZqzm6KlHW#R0$7KmHPBba!K+vU;=c}>(!-Rc z{gz&~ob4b&q~CUvv_K#UHI^U_4I=I#HRQy) zlRS6M@(<70T+;H=L2*V>Y7hsaQIKH+YQwgn&nk!lrm=c?13&`;_Y8-Y*=SY$VK{F z7TSA_3r1=U^;PS-5Z9Vz@@@2+{F)@T|9v`UP}+~z8^SAHeRPWj3w^WxDeR~uZ$I$=J&>=1At01m8AC|3NX`#P0Eb{iory8F%7JP zP{)?6b|L2Y{z~nN9FEHcs8(x9t*S+0$sSDDGh*37I_&hL8zvF{O7Ui`dL&rHM|@LT zPNGug+cst#R!vf}r_+uyp&jD%}etwWEzeJzhb}HuTXk1;>DLY`lMya`CZD`_Sz~vI#U{%RW zT{4gRX`g6jg{f=sf|0mpU&DCLHo%6h?d2G-zo2#2e3-K^ioIcOeYeEjQKXHAM=>c# zq;}QQ!w2oGIHxYpWqASbjq|qEX8~cef#71^L~W>@defq?1JP8pXg8s{5xr}3S#t8O zTjp0h*#vI7xOj^MdnsrzNcj8b3iT{EUW0h7l2Ij_x$1X~kz~SeL60$AL38uwfs76p zt6$^3p!xkt@Kpe$$1_9dN-ZjMmrIxkgKpIE1Ra1=zz+F@p*yktIeDmzhsKbZrN$b-z#;^54}m^ad3V)wJiS)^=_%hEaO^ z+mv}S5S{Z>!yOee)&%F|t{_UHu1sn?{4K*UL2q$H=pkRKUt-~>`bHC*n6<7D^G(Vo4N<18D;XHNB#xRBPrWp3E z&F@sVV_fN^(T9yM_)tVw>;SU4mVHH*WPb)f0d(lREd5_J+Hpv^287(TuLoYpd2a`}Dl&H#kAT8u(GkvSw*K*jh;% z4!S-yc_L(oAVOV2#fVj+5n;`|8H50nFbfu~4D9y<&jMlj`1<&XoaEGMO&j<1NiX$T zrosqu-y(eTF5{A4I7yae6XlmQ(fTEc$KCf(ZOe+|V zD-q!KSf=|;sB?DUE)Cm)SR>la#i`P9Lpo&A)_V}ZbWtRf-h`$SIhN+0EG*DbyXPw> zBQrY6ce=zU5g6cr!78`6J0Pr7RK(A67g+*_6A*&af+siZf$jb`GQEn849m;zMZP1} z(-)6W2iD1sNAZh>x$E?J{`9InApxG6915qzaK|PCy`&}0PMePgRet;Ru6oHdR9e^p z7h-fTOCCw~y`jHt(|^lwSOpY6iy(jJ_9H+x9P##!+}USV^EH$?+xWC4BSp7rE1`=~ z%xl}iI@QuHEBbP`$0%iBIj^0bDo%=Vz8OC-axaqeau($?F7F+>V@l9D^D_7jqh_^E z{DU3iQYWEcPK=u-rg9&K1yKrFpMG_3HwBJUdcBZ(>77hApIS_e9A9?#01<=S27CY0Ct6J9h` zi9%FLNfH&m<}BBJ^;smNfL)z(op6!a_$eU>V6kvp}qt^tJ!d58@H?>abJRGF=v+XB^|c{EvlXQY$p9ccxkuj z-3kh7@cFbEkhk0;$I!;%z#(-y>8NGy;Y(H7wz`PJRC-BVP#(j#?n>l9?nyci%&TK+ z^_#rdEa&UXTj7jVn1YL2Ba3ED>^8xY zaBkEL8tU8N_kLTL@wMdthLZ5S<&EIg9Fbvb(eS&C%m zen3X#s#CB_6HjkvA;`BCoEF(K^U9-5)EXI#CYT9%xi;8AQvs6GX`STDq7Xt;=r*tV zQS27eQ+?L3Zsk<|tiI#uG=i-SM##~YTR53Trl9iF!of+CGPaTu>5kIY>sr+7da#DYwm1Yv|h~~ge?qgC> zgxAl;-l&W$!&v?yRm37NJwbxpx6? z@X+1jQCH%azZ6$0VC~GYocz*5_RO@9X2I`naXxQXzgsrZ2UtJ3g4i4nTPbaooI=^R z5l3lAqpKn}Pnzk>UZ<;D5OpG*wM3F_er8A@4mO9N>{NM!qLgzfE<*LHM~9+@!DZYS zAth}4!FWSAO}aT<$hVAjF~vPZnt;q`YDBiqjb z8}|(|XU3M`KdlEt4n}=y-EvS0({Z`Mh8io%st+WrdXPrK*a?jm5<(~*p@~)>TMb(A z&Vd?%artJSZ3S|K8P_Ly9ql6?g#>k>-d>$@TcYo#T8k%W`N)_x|fTSVM-VxJj7uN zle4e$h1LdP=Bv;sc|4r5Sp-`O~>E_!WegYSQ86l~!D_t1dZq zo$ZTa9lHE7tRRCD>VM*BN*@yK;=HDFPIs_LzMn@oguheCV^s8k==jiT*9r6TQ3}mK z*_(vhC6!=oTxIi{4G=oSK+?>d_Gg(CzE@$vR&z^O z+U9f!Qy-_g1RAvQzgb-H)(35xSD#!8;#>yBb0hWZdb?tR-NnL8nB_=sC#zyRBD-I? z4^Z&R+Nv zD#eXNC9zNIP3&w>8v?E`{bhx|$xANu2AjtSQNi+_!jTm>rf9 z|8Y`lP=y)@no(#~U;&k}cweM2+oUyW@2`?G8Pe2-O6qR_6!>Jap z`ce|JwW0d%97nXW+jk_O)3fEt$o4@qFVDPzSBn7-wvK=hGmmA$Kkn+Y@lW=BKeP1# zF;txgx_rNdncH#h=cWshvf{m&Ro7%1v%ED|qOCrfOj528a?{LXe|wK_^RE9KfS(Kf z33&|zsWUX_!+Cn0CgL#eMoCI$3*%uA!!?lCZ*cc`>?sXcGAlYv>#PL9j3`@Q)l)OZ z4r_RPi_f;~ZtQ!vyWYHeT}^vTf$s&z>}Y5fI-QJykVWH3`g_ue9H^18*pAlxZD$ z)|wuEQWFMh%{K~mf~O)TsCbXs@^Nm!Q47(Q%)clJ#Dei<{p1EZ))3cC0sCNRzI; zi%_Zf1qM;4&Z0DK`k~=UeWVHrgTyI~JqRMQU%q6bsY;MXhCEX(vph;w!uhTHp`AUt zAH#v-7AdM(?qTFw4%+mVI_LI0bjjY6>6N(rp>KuRt%NOu5zj!;Cg*3onhCN`#dDBbqP^Y4% zxNWGWCr_!l)788jBd$tOmJ*6vSGd#IoNs5(8E+jWV&pjJ9Z%LLg8^SR+b$4grI6uD z8&~I3%X8J!{H5PXLyXd=`#B-rK)r(Kn!aml*63LE;i}p)0N$BQDY?*pto{iMbY|E!K@@Ss6pXhonmkuV`rXXp;ZJ5EBbq}*mkztq1M0JW#IEIPeke?&K42L^%a z_bY56t}N2^6KlZP6#I$;cI`NM1iaYn8T!1xG}bo0%Nb2}9mT$kd!eH7A$Z}NzypyX z<^V;+ga+W>$=2-aP%b=6{%fM;(Vo^Xi>(|9ZCwM zz-R#|*ZlewJg3?8(Sl#dYwP<1GC|J{<`Az{wgVwxKk8B4dTHu4@kN)^FBd0TZp(#^ zSr;KxjGc;p`tqo;s?sMmRBb@Sti-9@WQA@%7y5p42GXhQ;e7pa>u@&(v6YNS_(^gq zFsUVYvWM3DEnP6GC+-c;(PJ8mrpR| zyzjtaVkdRQ4oaI*cs7TIRkBz$9(WibAP==5AMr8#2`Kno87vmkmYIzdtV9~NcB$&# zH-uDUqqRuAbq7CT)F<2PAnV>Yge0ecxMpANRNY}3GHI3RZwqf#({nV&$6qf6dqw<6 z7){f@?O*Bdp}{$Ak#Z-C2vUq*!9`JG%)n}?uiA#&ZQ5&-U16y7I{SrjOW=m$CNp^~ z$LeOMpz6?ZpT##|ssVM;*+w)^ocz$ydp^4DwqS#U%N+~fJ}-nSM@BRLQv!2D?Hvv_ zRttANbm9-R5#$kN~RAqX2j8BlwN>#qrQL|+4iECsw3}=I|g|J zXyN0*5^CK!(9F0TqYpn@MHU2q4j=ro5?*Av9y5U#v=Vq1%bf@ z?XPvFZYN9ED?X(#jz7A)k|r-dUZw4vbjOMe#e2LqB(7?PpCc;yF3ieqUy0g&ne2`i=MFm2rKB}3z!PWDy<@Uu?{!@F$@(#(e%pv#p38|DxG8%wQZl6A zSrOHjhE5YeOH)#dp&cL{$?n3jZc2|}UGJj`qHZ_b8Kqw*CGm?7w>^|*zUFtJ7dv2 zGmEA}C(#{B8H>HOh>h7^AtxLAmrO#6`}7+-$N+9O*alrj5hEMy3-@>^53`v4;xQY! zssH&A<1ykJ;xKd_uXhOIx1=IDQ^=Ug-txhOHPVuu+F}n<4d)F{vYBqd?#NFeuvDVX zMkC*)a87L+au#H2cQ{!aXW8LD(@y-F^!7;STX~wt3`X<)*jElhkWuYFHNX8D zh|Gb0a|8#n!jr8a?NX>%{>`Zxw;wV&yu4@SpRKJT zzMf!*qYH^pheI+(WUo*($LnT~10{_ncfFd1g-KsZx3t^0@Q!szJqyXMn9{|bf+FWv zLt0yoMh+n z>guJN8@GuJYei|SAwQC89T3P=lMENu%dUsGI`_MdeX)amw!4G8ep%rz;`}hxx(ih} zA;5(72t;NK=}h>=i<+vQT4-*ws>aRhvF@G_vxq9Hv-!mA16DKSM1QQc?Uyu+W=`SL zlb>{T2dwWcBTXKOe!Fz6k(<+x1pND0lhJqPmynV;P@hP7^cL_VH{L@J^>%xriQ!uN zZ(Y%>A!Mw0OcDPTx_rtP#xK=ikM`BCQ5Fs!l|tDtLM*mmVWyq#W*ng?I}$S*NPz=Gs-)|iHq zI7NpievsrC!C`u3&jSlxEN6?IjDad>rcoW_&i!;V{Ef(KF@s#o82$9L$O!)?+V=c0R1i zrwk?v;j7pM+An9Ne4^v2jZ2`Ra3JrWAg$k!rVu(ij`@@G=ckg%?$cYeAuFArm6CV9 z6S`+i{M5;C1aZ{1)75sn!)AXZ0oEI~u%e}d!6K#f`*rcIBx-DyZNyjag;+yxz`;@66816XKC2Y|#J^o_vaFRV&P1Z?Xt zc(xr{1j;HPg@=@xLNZH;$^+-1BCn+LSK|t)EpoCJHoA-RWmCv4HvQgGRYVt0NKhdv zwqH@Jv>NMLl~H9xJ=#bu)e2GZbG(=;3UOn6i8@kXEdg)`;vqIE{*mgUVWa0h7b#_6 z*myXnMcE%uzo>50d`(JuanHl^&C2wH%1|A}YA@q4KTyThs!g%Ut2SuH8&>NV`#|fT zSY0|Zo{c@>RmhyjWOXTu3zv4GRP0-TiFa+ zceWuXPrjG-7cH^vBwsL*#OzeJKs9&C zfzh9qd6>e9m?}o@rr~w?TVSTW`}r8SifRep70)9yf{=0!I-`hpz{xjq#Zf%P18pd( zTsW&z{NNWQsPwWV_yi52{kLUG93q18+#=Y5M(v^_&y4_^ak|x_HZ|FZfvqfb{036< zbdCuQdK`WsMEy`Pjax`rUzx~%KNfX%O~_gZjw=R>9%l=drr4*D-RdH}zr(}ZC#lLk z6>fvAt}|uD%@~A3W>YrPQhvjWoi4w?=fTSw&>G zdEfu5=6b(bDWEg6oaTyjn$D9Pe8jNfoVJNQD`SS{uoqGXAVKaf8pNlf=Lzpuc!CE~ zbk|*}o=vqwmk3iP^hyOe?K#L#c=_FUXs#aJsY@ul|85RWnK5?y7KXH6<6XPTCySp| zOLy>s;K3>~QgI1$yc;D>30vRXrXmKy$2J*IcG-|g5=b(ZtO87;Ou^wJK~<)B+OaGY8aFs^;Y$4>L2e#<}- zid1&)(nYeYYk#!4I19RBGtfKkB_*SXq0=Z#cKP(%wHs7n6TuOf54_p+k_U9I(vX51ayD)yGe66!~F$CT}Lw!s}T&Di^ovtlQ$GX1qKE z{|x4oR!l^_=@{hmnBM)d9VEVq@@)GSmqqU=Q*{ugZ#JEWl9Wl=E(e0f$QVy%7=Kja zk-444@eXIOPKDkKxbmlbN3M3!$a3`_Y`2=}@zz1KY=kinl4@CU#xSw`*;DCj3_7u} z1w9mtq3f+2+49lGVcAZNsAEf6s5S0}biI*47t$Q=XZ$MzH=ZmnoK33pyQ^}XW~gRi zk^7$3AE`CRE@TU*QP8eY5*hhThZw@_%@IFoq5C?7JVv!nbS+>j$3ANqI=UbHln=$9 z_9^;mTbusu=C6@vP|v@P{}xtEeyj8u4PTUD*16r`=ZDJg$w?%0_bH3rDz7$%pXuKf z4Vha3hMIf&Ko-}NqR65Z^7Z1%eS*#-nZPyKSFT`EIhPl^OVL^A5!&@;w^`jq(Y`;k zN&-bn`=k}NL0olBX3$(0kkgGy9tFwYp`Ww$n~Mgd(!$3F?5sG|0?QljoQ`@1%Qd{@ z?tL;fpZm3ewbWC&K0?Mm6Zd6Z$1Nw=1*tLhYmaFY}Wid6sB_>8wzhE6^D1*pJB6jl#PSF5?pW}8$9 z@C$z$%Zsv)&y;rwVnA42-Z^=6Eavta38G-;3XTA0ME121p&OqXaRpruty^&*wpsRv zRopiySY8aoNzmol_BGkr<}qy1hH7#f0%F>xgtg z%D2FD216tjNu5NZQw%EH;)oIm-Ux2etz4l$ZBD#kO>PRPfw~8k__G=*#d@li;gt(J zWq6!WV+dz!?~2P>2Bh0>FR-Z-Lzz6K{%a=vW#u@+>sbzLN$ok&>Wkhucy7K}(6f`z zR#{iR^jkAInlIVU5gR|BCcW_D7m4D7hqU=yDSw}AvdeQekLQo`zxsbSA_{BEx3IQ# z=_Co~YQhiM(NKC=T9a&lq=+0zEGs?gHd(<~f1k{b7hh%^MI2(-mn#bjID6T(z7an; z&Izx6%G;VWj@8MWwJ@g==I|s!`j9ovbKdX(i>>1oL~(aJhYec1JsC|Lq-+mta$|ie z0Z!4al<{k0$dLkDB#ZxwP0CV{=w4N&`dePp@&(^MqV&Pz}f{`oC zJb!^LU^L-94aYg36slc|q*R=QSy#vuhG+Fu(8XEF`o#10J02(OR9h38f|{Y(s8;Y- zR{6q3<%@f;*QZQ<=-uL*h^Ucd-j=IQ`lBYOFyga{_YfooEBuV+Z+<2OU(xZbvTDvV zFZA=Fh0YV%8w%|+HKzA1%WH=vlI{sFEa9qUl9LmVsXqt%z=^4)PXF~jRzXb}4#kDS z_wg2Fhs#`Ae9@9g(MiR{L9b&{YB1uY`vHAZAG2&}PreE*qD6Sn++yn`S1f_0z zi|fp{=B-Gj`$)U)W#i;)9eRH_-cl^#INiT5MG$>%7y_f**qHnTe)CUHZWwYq)4ar9 ztw5H&<2KOkiMPTNJyA$8ow4{oRvbG=r&aHmgnNu&RwCmc&duSxM){@Fn9V7YM zxgZ-LZ@J+`4`Ik8BY#lx+eO+EzW@h4ATXh=bcOavR;-zs_!?{UXw?NL$;+AeEtU`o z`)aB8mD%OTy^8K?rkvER_8@r~8Nu1v5d5_`2PoK?LO;@)nho)GE<76+sP-^HUzWPA z012h^C;gv@{fd~LNi*NDzVe!>Dh@G)u;cBGnYL( z6<)Y*!~DtS6rwSUq8IK}k*;yE>U|4FKw;PoqC$f#J5#obaM;q2@Sm~khGHw_`%N5n z5wCW3f^3-*Bq8(Phkvt(Et!h54xX`<)%F-c--f=o-0}|76~bE$?~Quj^r64S2+gxi zxag$E)gd*G2q%1o_}dw!jpL^M#m&&xTh>ID?o-nfH^l>Q;pL@6944AIF6rR>f(uw; za0UTxU2PE2%a=XNJI4|Ktp?Fwv#*MHbc>vn!nQSdo!^HY$Gt7knc&ceWIVAY_!*wxGrC(cF|FoB0E=!EOPW3q zVpYd}^=m3jKL{Q2iDneQ{9TFB_>^nn&7zy7Cq{yMHb$$n6!M{F%GCR^Jd(E! zy^RSv*E7SAe9k~GWu)@^fojUAH!OZoL^Xmsnwda

2*AYVM!U5orIrTmsJ4LO@xS zK@nS)5p`S$T+BHGFHpyX=dE}ExbrZdy(B)@&`EzAQg*$DXi&XSmTf^Y+n!Rq>jmh? z9=NTjC-{O0r7j=V;~zKc9N9PCvx4%i=2x#FtPs37rc4>|5-VWGngxWc~gY$O#5Uq)Db?h&* z$zmCc#izQ}_>fQd{#|eRiwUSAPu;lelVX~EeVX}XknAt4!#zY$VE8^ zFS^45+3loa)ZTJB`td72j+SpMQ6h;RAn1zxZ(*U+F^x zNR`|qo20btMuJ3k=CA&w051K^!Qa$V<(+b$j+%D4CBiITuNTEj=sa^ISe3-(Yy^Co zz0cUVplGw#q&*uC<%;W}7Ga!fdN>x`ZXVNHsghDX!=Q{(7ABs*89Q)YJ?ADd2#9j# z7%nNWRUNBzS;`INJn205-Cf=J7*$aSylGqH=e zFIrPP8V%k?xrXoLJg><`yc$*)2j>{|v3-+v&7e77jmI36tEm|dyC=GiZ}ctj=%SV) zavOw+uZblY=q{-wvA){0KZj=4zxBFwZ`soQ%VLhq0 z&n-{Pb}xE3PFy>Fy?#;FRax ziUl@an^wnN`Xsl8h?|KRQiF?ydJ@&G=EcxM%#xcI_8hVZW*(d8-6aQUX4<-0eCA#} zteDjv=kS;JD!L(};nO_AG1=LW&(F+Sp`@Oq$Ydt-uKk6vRp(PJ(AnD&2UQsrNiJ=4h zO4h)L-Cif(7(1#hU~3iI^9CvgSsdDjrFT-g7w?i zpNaKp66z&V2osKvKX?TU=lYW$K|GyH@rN{1(TlR>v7cnIGEd_9IkriNP5FX6P`nkG zSleINm?`xF=fY=X(T6}!aWe5me2lam z%n#V9_0o2~&}?4tT>|k3%TQFpkrOn?^y?y(#qw*~aajAYuq*b;Mraj_ic?Z9mcy5$G8|)Jh>h?Y)e9mBi|8DlnUFlq6;wSw*MKzJEci!NHFRQjeS? zFbZcPsXUZ@vG$ItvW`Kf_sLhK=sJIp+@I$JvuDL`6*e@3NF-Be)6IT_J{9k25Y1pP zN?}9yAj2v4>YCRhre%+RaqFSMwv||Y^cKfw^K3^S{2f|PKZsD=fQr|jq$Mu}`(CfO z`AjOi7_of6IFp}*0nN(pCL^4+g&W|6gF5^Gp23K)yPFmCC$t~S>;&`CiPN=|v`X<) z$O2Zvm^*B{fUAC9lOZ_;A{6hTcfE&kw@Kpg9IYY@Zw9aax6cf+L;XOXlnkyZvWTtC zHZ9hXi$ya8-xgJ(wkm^%A`YB>S$+TO^UMQganWrgI6)cIKMVozB;tU=ZCx}HF1dam zZ~;9VzBhut)(LW*Xou#0R4CGI;LP|mK8psdh4{g`rLhpVr!sP~|MG3`EXyf?|BEk;q;uN97{X5vNMabR= zGZ|@G#yZE^-EpE}YDJG4YHG|>ACE?U^@ceS;+TQ$#H9-MXNcm-dHKD1wZ>-0v!!q@ z3Oj!oUP?|APs1KiF+0uJccKV*5D1eMz=7-ZufJ#FaA|s>pqbuD@Vx~y5advN%$gHd zN{6`$A%e@nI)-S8ZvK*^j!`&VvVI+CG1-ts#1cHPRE9DzPE+A&tvf1@hEz%^>qlap z!}J^Vx7j-kWWN|kRU+2rA#bx`)KR30=G0JZr;l(Adk|t<0yrqpQ94){DNyvVpEEab zB&u9^*UzGjZrt7*lJZ4VfrgUxzNT;wG=t2#pY86~hdx0^P`If)mN-K)mAW4^5WJ~` z6XIPXZxnJuTSlC4Y(tAHLNb#TEo?Z+-ExX^mh{#FH}ge*9Kr(hi5L|6Vmmdq_)KHV zztF7JuqKJ(ORY3{S_Q)If-UNkOwUZ9eEs^J5+aWX%o99TG$m&?0lonm4lotnuLM^5hCMpw8sEAU-8=)~Kt z8>bN`M$9$@Jf=L&*$k?H3EzT~iKqv{e!3b>JmslxNmG%=O{JOZ#)3`U4l$$cV`Ga> zS-W&J2PxRrFV)Tbj(aFKFk{xrQQ^5$c9GDA%((dZ6?2bBe|;1W>bG_GdZ3}QydoZX z1X{o-%~}y0M9D;0o|BX|?LrcLRe0l&{@8mqhv%pzj(CTVqJrik=U8X{i6X);Cm>{| zqk?Lh*-$ElUlGqoFRLHUHQU}2+QA^rabQOM%K1yn$4{au z+U`Sfh01g@Y#Mjw60*B6q6h(Yj{5z04)^vb8VPqCub_$@#A9xo@=#tVY)OfegC{jNW*cwbeafwK#fN9O>#$UfFUqEz6qNK>82A9uH|9! z`kPzC+cD#ayBAx7SHIt{zJJ8;_GO5)cH?S(aLMrqiz~94J==I;sJ5BJ%#_+#yF&*l z4|Y1x5Mv5@+R)E}BJ+ijE{-eU2QzN3DKmMsLT(gH8m4?ZiLyJEwxqk_HXY(7ux$;z zA;)v!;7bWI_aRCglc>yNCyy7H=J_-U*2=hC5_;lczj`m$OWF^LBa}%+K1{zQHK_Nn zdl5uMSL$49@nsPV3C13=3&(bS%RD~``lS{3LpePADOWY^f~G+nz5-pVQ#vU7BAD56 zHAsDjW}zY`?V`ALuP!7d6t&Y`{;EDY=E^;?1k<6T3#)2_=kuv@*D$_)vVkB4Tvawa>M-OAZ%7zqrku>b6P!LB9K3A>E=rOxHcqp-N3S=$v9V8z ze`4+fvk=2==Ng|5NKx_eYFJDcLCU*`*YT!!&DW%gUTYlAxR~%KW{vs*Y0d~mCAOfK zxQ5GbdY^>htKJ?`MBG?nf0Z0)+^>nQwYu=e+^QT48R~L%&1-Xx^pXn+`@36&=As2j z&E;J$F1mEYQ82ps%`wES6v@v%WKu0@zekLp1XrJ68=MAJnx62!4tm(s>ddEK zQm+Y5oHSJX^h-P=a57})D{if1K5o^%Gw!NMs z8(qkgT&qhV@e|6@jj`+^5^m_djj_rw&a?h~?@jeWhHu@Or!*6)C)Q;TSjd<>4{Ene zP;1N1X6i9FZ*I{s)7vaUZx_9iA=J|z^nC4>1#5$0s0C#Q*s^S|3{4N9QSl{`@Z=g7IQidCB2MP#Tcxt>~uZpd2lp!m$E}r^ug)C7Xg&OLneh(tQ>HY9yGx zIm4S^ZBOMV{APl>M^y!BNXUHZ;I#@%Ebi0}2Wqu#$#N zBp=KahT-%T6abz6zS)B;aV7pNhZaT_-z{G&hI}@Faxss zIozVflYLR`?$G2LvCjM8?AW-EhP3+p{Oz4c2=Zg4zx(IMF-PsWfw(^tyGtUL?uR3E zoaVlx6;S32Pu9!=q%+047yIbvWoR(y_5PD4_BovrVc`01zWPy}T(ubjy`Wsd{Xl6w z?YrKlxRpmf4tBd@(xL126}-J&Fzd;gi{TrwRV9!F9LypjD@qJH2qU8=`2#hsv6Z{m z-gM+XenGca*;)hcO70otQl>&poZJXh23FKM^jsqOkFGH6LZbL~`RQ81pX+gxgFt?v zoW`AsOdRWsX5V-l7JjXfIFI~27K*INqBIYv_5D4#a1OxwZ6r2yij(!Tr2}MePd1iO zLM4v|n8g%@JXYA+VF9db;;F_$Za@N?oqP^2&=QU5P20bP#(wg0dbyc@CPmeVuhe`B z1wf&eP)GuzizfB^0yFnc{(K2)>R-DYd}6tYf{I}%Rh#Jtt*_;lX)HZm|C(+>NJ#1w zn!3qtGup)7F4>Mc-rrM$MQ;4PQ07tDSD}8r;2{lJIbKn>rhZ_7uAeu26$qRS5WA1} zJ8h@NW`wlq z*WjD($Gd1cxeI)wPD`)Em`UTa&wf-lvAnDdKIGkizT;LxL@QJiLZD<$l7pbGQ*t%jMV^9X37JP>B%p|o3?o8(YUuUA3oV#L^7^w`P-dX}SA6<|tSzTAx) zwZ>N5#vu5w*si4`lcJb`MZz#rrVkKPRbdzTVQ$R18r)sRIn8!TT9Sr6dOC^^S{NpoHYL#}Z|FMo zPHjh(!guDb(mbcN%^O1kC1iPW6Pgndu4xITLxcM&#ZZGl-qmAPry4rE@{yAvGX;Zp z1a2nVV9drFCYQwE#&+3i+8)M34%OClA-GpZlcJ8OVMQ^*t3BZns-I{&-`$QmH!PU6 zpX98H8pGgM945Jq7#6LxboYYlT$bfsp7Z=gG7?FBKMw-1%4YhioY%xFo%=FO%P*0w8MpxIkri)rvK)LEZyb)hmJ zFdQqRl?o5&>@;ABo1HctbMZ=Jl4cRU$Bu zH1&c2$2xn{JcwO%?b_9kBC5`d&K$f%(;06!f9Gp zr0VAd`AXpO>3^%`y)LU)8Rkc;loMcozi{Kfa~o$DvqUaDgeqf}wwwIkq6`+TQ<)g# z7{+VK(q^NOq2GTTh;qK;vz4<;dB1FF!H2sXyJYV&_oF|}#>7p5+cOai8oN?4(x;mu zC$qzJADyzKy2{QBd&=D?9qUrWvi=^8up7X88fKVFn}J_BhngomS$pmU;7y!goBb%L z_oY*}IlvYTt+pUzFTi1_A(OnPH%dB%i+JSoE-bJ2d+^cR=I&q5Yy+;p@34Gx>4q?@ zt-Km*85QNsXj=8=tHH$9cON2*>5+C<7xfT!qX(2pmDAlL;u9UyPdNrwR_QI6XkT7fAu$pNDdl%VLL^SW zk3o!dP{nAT#rT0#646u(2S<4%R~P0RQP64@|3usycJ>=InwYwVMI{?a$!wfIw4tVK zq@YstA-OLcp4y9?X%My`dhQ|q_VUayIVhD@P!(6^O1IR1HvK^eC>|+M=+mgd=)I*= zh?8g38$C9w76@s~Ax_{WDmlEXlsTH9g|^HjJn5D1P3|Gnqu#dVhmO_jE~M}eeGYQq zCUbSb=Z)st=*#L<%=h+>le-RX3I@eyef9*aEWOj!}$~fGPeMI>jmnAxUN>$9V~KxXnklJE1>{ zK)5i4hyAjUAXuFry5~Nfz3K*))?5i>G-T0+G#p3xqqlX=FM*@2CyTyCDc@7fB+0Gi zRimsA;TY4T=~;yxTA^xA8CHZLym{i`t|0o|=;CVXGCGXUPhd(gh`E~QX+6ei;Z*Cby*`qH@Q!@JeIbkUnf%lX{mCb`i#rQzz z4*@O#7{1G@T5qaiV%$Fo@3_?CMKfo|JHwc^k$&LVl{FV~O7f?#XKUKQuR%XV#0E>q za5ph8yWSzJg_XwEYBA%ZVGoT~>{zx?G{XL5iGGyAZ1l*YEV!_GhO5r*<)?`v-786$opa45gmfQkoGYlulgs3=T@Z2B!?-Z%&`fm zhygsMJkrCH7sc%J!zoxY&dWFT%7={{cc{?~XH$e}ipt}D%&1X2D(VkPY|rHDSE>rLT3YU{ zFjq5g=e)H`M+DSahgj+Keix5NwC*%r9n4XtAE>v1-4HarX-h{RZHHG!Ln9pq_PNaY zX&d>*Wt*0NMu|8(EG(U%|H!iDE%&|UagEF!<}2kp)oaI^y39qGFi0n)>^d_t6j*c< zT9d=mX5lBm7shZshBW)`urXG?q1cVOri`O6KlWGJS*S$($CvGlFoZ=axy&Z2J<615 zm`s}w9h8eQ!A{jtxIf>mg7jW>(K;d2eXJW+@;nSXe6MQBw9kBn# zB1E6c8#0(r->86~jfTRv8vSjYLu1-sl0JQ*ThG zTnZKwJysknQ=UI4X38?+YHo`*KoymU8r;~&PR*usFu!{+{Xjk=RiG?36xzYP(ZHPJ zQL-T5eUpsZL~gfZk7)|x3!NrF9>u;7S>?Y3f@QWjZo=^)KbZQ*8zh^=3^s?1On5Ur zsbB|AD45CcYRH8mm`V4h-@{hawUqm~I8;ThLZx_}3ZTubug}fPy(N-7i{;&ea~bIc zQ4Hy(Sr)@U2$KSb`lo)PS8Vj7AJ<3yUAt?jWbng?ZQ>2TO07xbrf_S;jMasq>uP;D zEA9gi7BxLeQIS^#?#+587tc+#;G?;cboP^N8|U7>juGIUK!aNgZ)we@ves(T(Yt}# zdpk1cpY@MO_s%u`Qv{5a`6dD|VO6Mp7MGk9lt-bvokQcUO;}B{ zuJjy!Y0p#PPs7b>$)MAQOdqo6i%=sj85>b8Y0+*wSKKa5sR=?D)WFn)^T|@UgC6`F z9Yr121Y@lkE<=T(axz#TTvQtD%eAAjFV)E_Q7SXDe}0g^pm^MV-6b-~R~wI<`xPEO zB|Y0+i9%liP5OSVtdVH~0#YL4=%EKp9ClVjOc2V=vTjX9uC+@@@MDwQ>yX?7mP4YO zWY%ij)JEotTz6;gWwZgs@TDlmohlM+g~{w=94|ZqPdZt_iK22Q>v}Vfmeou#QO>F> zl~(IM_tADF${aFnx$AWgO2*ObL3je6KTOM?w*wd;0dL0ryp-yu_ukN`yrt`Pycj+i zm5IXPv_;$1_U6S|a;N5Jc{p|AfGV5{bSHqEX98|aXfg-JpP3Yv!`?tXLXTB{*iJ}Q z&RV~7uKJmX9Gh(uH}&^v4iK9C0lOVq=)QhZ!DSSvvJA4#M?$QD^>OpZk1b3-6^MzW zm>aEx%kB74Tau2qot7Fdnw*V>6jB8pql@J)j61)Gk^GumT6l6P z81&pwC>kU2h}IqJ-ZVQZxg@|$BWk8VVUA>0@2)3o^QyuMRbXzry%|-#=xF_RfyLm7}&o zz487iHC|+q%&)0u8L(n2{iz|k66N>}xbKCQh0ek*I`LZ8=T}v;?8!14!Bj;VEb}n! z1*61VZ*wdDo!%yF>&^B;I(&Yr(}a(W+?v-tCU99m)`LXl@2~dCEj&agsfz>ZJ}Ogc z{#v;+b;X_}pImBGgPfO}&6Odl0kkx!|%~ZyyRKtJr zaYKBWL$|lb^r7-gxKEJK!Ja?&n9j|DzokT`k`;=-0>+%^iLz;sPWwjWZqRg!XnsP_ z7i-3B#`S>#-MKZMP(vXeiJ0*IOIcNdRP*}HMDzS?Wa4aiVE9i}Bz^SUFmDRmvl->r zu#&&gB1DY1{jlW_y{zxk&gLUOHvZ}8Ks35FTLv09x7J+wZVma->Sm3}bxU@4{&+P{ z-xYYuF~=MlSKni9r)$qbipq0_Fz118ZK(PdAzldyR=}F2U7H}>>H@U4qBC6d@? zrl ztdBA8A((7r-KttS8dPHsMM2G(W|cl#uYM~E?Gp}^4BtH@=I{Yd)PT}N#}*Oy;JvBX z6Yxo&-XLRCtnqx~J{OW0Fi`&H$nu2?T=*d}6idZTxC*I(vCIn*%S9f3DJl)-4Gn(T zR$kaXWoLm@@|;~=&(ufwr%=I1I7H#zK3K8$7~o|{ zqi@u>M3&fCyw3vGnLyJl;3Y5C8pd`X|r5DLFr3!*wn-*~IB5ZrN^8NhX}F zT_7P$OF;SiVQJ(~+6~1D!Jim*16yMqDiPAlK?HBHcD%E?)3US~4)M4g0_^&Aqw#nUXM#;?;8uxMo4swT{K-ZY|07R_+UWj{Oz-9 z;;XQt(sW0*aVGAvvhTHt=G7FkRl+QcuSz?6zd&fqs4!hqN!QMvtLUc9QzE5EfQ4_g z=t;*;{}~`p7f|YMtD5Dv&tP57J1c361lx5FdPVP<0GqR#33kLQ7caxyp72FCe}Bt> zl+cjGXWg3bx|)hGv6gTF4C|eicSEzhtMXJPi&}3uKzT|}N)KI#r}}KQrDH-xxIp`? z2{eNm;S4Y6iT0yfe4JlpX?4T`QArwAT2woJYo|rYu8td1e@vcoeBl}JsNBS=bfq|a z*N4u1k>$c&6ACSw%joBh-Cm!EcLQm(@wzEQr{2WybNF!%xPfK^!w=}$3)GC;5YDIoV~ZD_8Z(G z8tdMHy1C2H{nmo-?*dobl3rh7=}d09&_IOz9`n>@p;T%Oh9<+8(CX__uVJK_9nvi_ zAQ!F%|Nib_9DWlWA_#eyP~4k29S&G~#=nb+KG#kb1~sqnkrRqbM9O0#@i@Cs)^xneHb<0gHu!8~u#jDBf+v5pCFYE72vhU4P&xXdN5Dl@!c0dB*!)TRW6y z`P95Wi7UoM1KAkCJy!TjZc+jFj`A2Q2b5Me_}@b+Xb4rp zQ_QFMHGQYqdaK`~a5GJ$Gd@o1vR(*>rRs@?Gnd3*S)3jYCsX>5IucU6qmh9k=&X|{ zFy%#LXT%~fahQxPX`InM$Bs}@`l&Bpw&BWYJ(_Wc^s90SPYA&G1LhI75z+(F{PZK#6Y#&1XLKFjZuEFYS+xTqg1YQ+?Z^zH6pXs zhKE&lv)*-u#jUM1vyBr{54_(n6#Bc%;G^Te?b?%Bf*iCe6ZD#osdq9rQp6{)Dj@`J$o_x|N41~wPVx*Ey$UYU0ydY+*kI>X#U8d)miT(BJ_0dv6Oa`=JaE?-%4 zlT=>G>+{H3&@J5M?abQ^r&jTK#!;Q}O}@tpO*c`He|%fKUg~BeQD4+Oi5EffjE~K!bh<6yjK&Ocjq7H`@-oYz^18AIuqWEM1)Po7M0qF9U|iWbZY8WM?Avz)pm2;1K^l zkOD6^RfV`cW&&4QUeY~j^y?LWv~+}wpty#U5NAuVcSk1E_~RVoN)s}kMZmrgIRif_ zWvL+Y(oXs@5gNsH#w|4UqL4AmtImWo-DtEG{L)||x&5mTaGXSBp8U-(Fh)&SL;rSL zk2!B^5!;d}#tP2Wj6_Mgt4>V4+8xaR0f92)*MR(j=<2@oe_eY9{l^mDxXQB5wDp9Y zy;qnW`_NmI{1!fk(8gxC8;R zxsK(Mukzt}6{Qpt=G*&2aqx)dO~bQIY)6Qdk=d$C8m&ZT4Fa1L9u)_Sr8M+j$I5Fz zf|Zt(8K)(8C|>T>**dG$5Mb`A54F9cL%%EC>0IXKlakeYid}I%X#++}kxRHrfOYX5 zmG*t=IRA@_C9aM5OM@AWj~E_$6~+4=$;v+S^e4Y*x|h1cAV{zHuE+|No;OLt<+q40 zE)`~DDAV?g9utm%z5#3opn9~c2-t&sx4#`ju{5LVdV85njudbrz8N$pHO%Q+03<-$ zzov6;ol~uf^8w}~5N`kj>+kW7T=g~qt~x!f1;O*xCrUgQg^Z~wqh9YuDqKl4*RkQ= zl0%4L9ON_tXqI*7t(SoIeuq;b+p5sD@n!9KyA`OLSuRpcf>ZEerhKtwGSXbaQxx?K zErD+-$O9KqO}j#&mSW(`Qpph^9(P8LQe%IT4Vu68BR*)S}mUrpBfZ8;&#!2dzsd_Lszd0{a z$kmZZ33QfuD*j`#tEKBMv%G~R(4rTDS30T@J&Aj43+sHAC^Mg1F8C7Zkg*Q@elV4t z!<{Q9mV;9j7x;qI?O2FO@fo%GJ4pAPZc!aMPubeyZ>J20joInu{Ho7T z*WZ6AomfBB6TKh*rIKPE9|EQN<2ANkWv@DMs>i{of92x)uJmUkIiUg=u$cN(eo5ZySax|H zF4IcaPi%GMYm0m&nRbmsTi%U)?cOG@sh{&S;D+qXAb)4V1QFK-$MxG#DJ4hN*MFC; zRA(VirMVX#6A%0>4%MbudcNihKsidgtq`Bu0qX?WJK2tTt~9yzcGMO0Kt}u$)mo8g z!{{$)%vEpE@KlhWAWcuWxbtU^Gk7sItdWX^7#{yP9>p#Dm7w2Lw zlpo27#be}}H2HM5iw|z7{{RqPnqab25sC_2G{g!}+NS9VURHb2W{5G3n8WA5Z$Ige z=p|q^jLRlmwLxYymv4dG*c8Rk#YEy4MXeC{tZ#nzSEwEFOz!}JhD^FL-D38xHV2lV z1-*9sl2hD5{Wia&$&~o(Y$|)T@xFtw2$Ei{>5M5KpSZ}G{VnmhD?Y?A24+3a#%ai# zN`=FkvSJjWNt3eMi1SdCr@wsw7=h-=@!Jr_I>IZ((Js#c!hFF(2&*BH*5{A=tf0;? zB-!c6{FqxvCZvj&-i&~D4A^AEGyD0_fHMu{V~m4ZI!6teC*juJzC7Y z!cXwTTP+=~ULzxOc~{Uk%bl+1@jFV*D+x3374{IP+xqpPFrYy$hV}EHebpu<)@3FU zg>Dq9Ki-a_8w*&*^i4i}>rcXJqcu$v4O?n7{pX@hq#umXbgokUe{LZd;ubC=ty4J>B^Za-| zfTcjh8Nz?@E9z|I;wj07SWzpb%&UH6tQR~!Zq(Dm(8sRTjB1D%nUx_*G(P4U!ydTD zoNqA1_x%WhV8$8|VEDjFG|^{C1y9dy3RX2!d0n+bv0uXA%Fx5)xPd*5zD$cJwFAp= zQ#~Awb6-aGaEbHOcdAODclw%8a;QZ#o6YEb0e5_JS8r)%I1f4X8>kDn2otyNy(Es| zO&2U4E3CSLQR9D8nBT*HWjz5?K`XjtFMHCrmBy!*9x3>*#N#Xf z@#jc;Bfl1hUg~*)ewbdByUc1rNLJEOBKqO&IDrKp587FP1#_*;k z)TYGbm5_O=ACvC8(3oj0x@C_Fsk;ZkG|;eoJf# zYT#*W@o2%a4@TB1mRWqb*1A;LTF~p%dckeqq}!HOP)@7guzYF5mL;s>53-!ZEWR_q zD6&<2#Pj-1eqk}fiX7q#0a6$>{X?0Z>3Nm7=g(-w71toewpQyVOkT8b7{ymc3veC3 z%COhcRSZ|Q7w$3Kf>sAfX66XoW+9yK=qbYsTM?nb}a z7(94SmxS2HC9NHMX0IctnReviQ?by3Uz&Ps^F&9@^%2lTO6FLrkuyW?NTx|gaTpAC zZbX9wnq?TlN}!dxZLi|}%GM`enB^p*Zwie7>?!YFK=oM!``%OBs_0ooY-WF(aFZF@ z6v>@uz~Za(OF4Z11QZ(c*cGd|DQ5=*ilZTmvu-AaYVzUTQgXC0$P(2{s;5WG|K*hg z>+A-w%uCg)$gROn%ZKqY6OM2rL))aCx{<7Ud0XNv5xXGPApX6*&5VWMew@tnw?1;j zw!88`Ph?|j+Nzb#0wFBe{ULV4KFPsCQLH3Gq&Rio;H?rBwDkMTl35m^*jDY_bV||a z?5=y3l3Oj2ow5&{K0s5zyR+q9CM>`KsSJsD=9^eO=+PI3+~W5N*0o_sp3vtUPHh6{TMl^O5u%MLg2bEHAEMv8h{E%gotvpyt+))? za8rA+Zp$t%3g*v5o}j5&h^%=`*64+7w)N1|LR3fY3XFhK-Jt*djF@v(;Sf^!E$cm}L{GMOWt zQy+!Dz;fikAFzgJf5#!4gWI4db0JwrqPz8}vMDjWH!VPum^d0snRnBe&I6Qh;>8dJ ze^!x`OsVVHaeb3LE+q~=~wi&G(>cEb=ggE%omnofO z8X#zF-~iMT%`QXR;H&?2m$xkFo}f4h&p3fHgWCR-+EtEx2Wcu9{lsw z5nHeCnrzw5_LQkDIrT_lKB9!qNmK-kH0ub?n^T{o z*}c0EPL-avecCYE0O@@i)AWH~$l0OtC^|7_40c&qF^3_YSHQH+qDG^5C}z2!M2C1T zheA3^5<04(mA8Z5iiJ!&UvW2sHFn-=xO+7CwnjIlZ>aZjJRCsfSe5S4@+S|jD4>mc?W_rm&UEdFKT02teHFVlb+U` zM1hjQ1sB=t2mJsU%~frD+*KrxY!eCCi6xl%OP^{fK^KM`pZio^DnD|JLFaf?jhE)F zay$@klsVC`Q3~!-z4*;E7eSAN_Ec|1B$+c5Zm-!dbc}WkV;3~6*sq^6>?{B{{ z?u#k~1W@KsnG)rNgye%NsE60qyk>Q%QRnHFtKB5z^$A-2~{}sd?YLgFGK=~QKd%7-$N+uqfx5?s6o|QS9gn38D z4|60-)|GpU7TAeTK{6i-QMAUM3CHj-FTK9^dhy4z7X^4mr3WqpvOqP0j|Hk)a?YFq zlaAHnw3;nGy`=do^mxzCpqA%p7pkA1_Uv;ulfiO$165jku&x%~NH604ugFWC9ymhz z?M=WsGZhZZBJH7%w7q`U!#PddIG$u5WW=r_=i*4e($Y@Ja3fK?PfW-=KCw4t+y!i1 z1MC%9NJx)QvZmqOn;C6e*SN32BQ+*9ySVHKsJ(eOsh)lYM&EmNm8lscRCD;Fuv%yE z9RE%46ppXHX2bwbpE|D(hR~g|bgNY^5Mr{n=qf3NTpKh#=N;kBSA@*m8Wceb_(17)!ec0wp-YC%d zqKzQ`kiJ~KKCgk#4r|xO-gQN1CRO}dz7;au-a-&k5mAMm4PG^xfbgl2c(PCG&!Tss zjcSklG}zyhww`^QB$()@Bbw~P7xWb|&%fP(V@)nvTd*Z*dJRlhcCU^GNVDbleN4-^ z%w<(qH@bt5j7Fj$m$|voDEda%_>oucv=(U+J7WDLa)IX$>s8fiS*66e?wUoaEy}aF zSdM4OTsfw%o-I3e8ddQt%Jg0BUMLFouc7{eP=k5gpNg~jtrEH;w+atq2t!?^mGyy4 z3z@OVkMi!-KOLTP0<`QD+Z?#kL9cNpO2i1CVcxiUIZ8?Ix-;hHohhx2B?=;HV&*sr z{<5LKaNyaZT(~}S{YeE|p(JrJPxA*(9(@W+h_ykRG%l8~*-pZ8D|o*h(#%uA+^Q3= zt4G`%wF$H2>rNKQmJ+H!{zziW4qe5%yq5NEQ|2s=E@O+=h3D*CC~f z@iArlNJd`;`K-IN_sRy&l7w$sYzNz}C^0C8Pu*Vc;fH^Snh!6)lwwaF46K?1=sCm> z9;}z6M$<#fD5hHV2rxDNAy42@zk7dQJFJNri878lx|K|_^ZcRCTNkyNsmCT?6ImJa zOX1P&dm)ln*dgYckGQx-?IKw~Hftbw&5u$=6byOcMS`=ILUUB%%)WjR@Q;oKZOG@X z04G9E-;h)kqjfS&Y(b3QG15ttW*8WFFS$eHql2#5C=WZr(h8DINOMLjEuC!6n@*k+ zL8raKNNqje%A7$5zaxk-@g@3bf80uUdc4xhlZoomxo`G!$m;5v)4}abx^5on%bTfZ zCSLk?fPLUBeJQle^wZCJnh#)N3{uPBLZgJ=KU;}>_GFG5u^EVia<@Hd5sM$YavE<% z>k?CksH{pfXFR5hvt=jgUFBlA6yCc0t>Ltwk+DDxwx#iK(l_U+=6Sd){1Y#-f*<5_ zN3cI+lZed9d-gW_(wTGHl;h}n-pU<4R&^_u0=9=-g`V7a7{DCL3#Vgisz$d$tiVU~ zUIef5gU)9h@ye*y8dq7u@7Fpa?Tr@AeYLv@H|1-apaqKiC zz_rt#oxV<(FHg+K!%-T$HK_AF2sgQ+d_$pd*Ze)ga^M?I@K-31%3>tJoiDPFYSW~+ z*|11t%Dx(n>%H#ppgR5J4us5u@7fy@{@pnrM_4)WF?+&P+BWA__FY~=w|n=589+z> zQu4M)U<&c!b}Fn%^Xayu>8NlgPjL;(IR(w9g4A^bQJiN5nOA+jpM6I^X?Jl|Ju!jGztN^!mJ?INmWyn{i*~ zu8TR%B0Sg`2X;$nzx&GE!8r>C;cRl#noYiy?cDlEVSm^UC49B{3@JH6OZ#CDzNfb= zaC~5Bq&|SP?P|x9$DaJ~mj$e-ETu~_*Iw_TWcQC7&PisTQ6aJ1v97RK%cazziQwg~ zo^26ax_5_PG%3|?^PL9gk;-w-Bckl)@mVJ(;^de-6W7AfCSYa?VfJ0)ttgh;F14gp z3xEe!b=>_Z!yxmz_A#PcDP~Te7iGi``f&cyswBUT`PD9Us0>4>saw!x9J*)=Hx0Ju z_K#uE%C0eVJ@39f?~e(-Ut`lhVmCd4$#$DFCEGoNd}JB87kV^M`Yd%d(s;tIqxM}7 zQfut#N4H!3&EI}qMkpi1h{NMfYQzdVHIcYk|@CvK5Ba2-!-6 z4mYm};oFAq=X?5fS-f!PTKcB$y7s4NpBR!dbFSuRS;JBLz4bZ_zenFy#$JWEWhH;6 z#V)NhRA#S8tvdDyA*(w7O%>{1EQRHxjP-;5ZFsM1NOQyLH>0_}JJ+UaJYO&}BvNb7 z#DV$smcOt$Taz;Ay)_&{Y2|X z@VHnbYh$2sG!VtINmMO!mpFU-klGokBG@8(8{8|SU>`9asgG_@VJbA#9F~k_jI(_u zDrMw4HrrXeSK#T~h6X<#cK|b2WbIGcuN|}e`;pz>afP(>q_g9TW2R+Tn;QZfNry{9 zTq_Zc^=OoeYxMKv@Q`F3cYN`T_8R|;AMf=T!~WPCUh+c0ThIgv(C}HSfbs0y?qTcs z6)lAPv_uZy4&ZIOHxcBu!K=P z1Z`P6_gG!v9@^d&+q#a??~!Nj?Os`tVZD(aO`O&a3K{f+5vb6HRF00;)$rfIk@y%u zp=~|OS=2ud_goTjPsRb2hwv*aWyA^4Bt*WO;AsIrtKWm>xrXnZwStrZK^7)XMi_)TXW(Bk3;zUX$#+*rN&y}%Np&>0cDBqPmS)Q zmM#h%T#IpAckj+9nDh0)hXjFtDN(n!YQ`KcXYb0LSn^}9O?G|3?2OM#5O(_Tewm2^ z#N2HPJx+Cyj2R9JBiHPc?-nVxKS!44Vd*9$(w% z`?Z7bcsLqqo4uhWyJnk;q5ZXUknB91BRzGb9Fb6H`FU20muqvjxBNSCD)Di#A}VBa zBQh7@bQva+@a1uUI*0g%kJC$}GTxfU-mjR^@PxPFLoEXyh994xm8R5smstW%2@nW+ z$lfRt(Zk8~N8DJfb4a={oU3!EdjiwHR9?Vn!rHOLFusr6sNLn$i(zTzU6UsMYpY^Z zaixe)ValemKLYU)2%`PjmBR>gp+BVl0rU;S6#Tv9VHyOxEEo~$)a zSo%rq8IBvG6g<-!(Tfd~eFqAil!)Z9sXw!Z_6yPY0{vldcecRKp&fvccgv;qs*p_K zG8<`ccECglKM!T$HN6^${FJ4zkn^U9;cdC*&S#({LcC^|BNLh`hqLHf3E@qlr|KZc z8ml6pg&NGFSZvJqSyRSz&b`vMin!2dL!s2^Tz_MMm}Wu%)%k;rx~4OT=jz9S*}S4Qs2EA zHe9i)K@%KtP}zWfgc{}rFuHIVzO%l#Ehd3nnnhZD{OSnjmqTF87$xD9ErRN5OhtA&d5_Ui85`wjzi5bllb!qU(O9z~r$3u5l zs6Y$s*x&a4K}qUfvXFC9Wie>j;Ys@JMVJM)62B_F>wU~ZggI(m*VBu|A=S>~!bu~k z9e+cxHO0tBq1DKabc+sl=iaAtEY2|4=R_SX<&ZR!k??&q)1(Abvi3IV+;1CM2*s7Z zDY9Xut$XD(Ck~5U&74wgYw2;U)jU0kZ^{>`r10~Qdeu@=-#3+G_yuidk_I=NGpfIQ7eMqVuFup083UIM=%|p82Gr zIX_E6>im>Ga^TaBj${fh-tS6#zQEA(7%8&lmJ{@`1Nvx!HRjfWAn3?peydUCCRa{9 z%qxkPI3FruHD56S`wX7Ud2k{&_K4f8IEEiJH0|Nr&|VzP%vR>6en|MEAM#UYo$77m z7ZRx1OuOp3>BtdbaitbLsxtC|Sv?-pI~gy0{buihT&=-%Ux8Jd?<797(8(U`ELUp; z)7bg+acxyL?eAKt#Jl@H;G^XnjZ>>$`i+M6# z`qnKKzY--tLhlm~!IegH+|1zVZ_6^Tnop*smO~V}v=0m`<)X9CtmS>dQAS1>v71#cVuTUFeMW={_kB#|)BTT2uDy)Qy_?T(hqv zlfUa3VSrQCRGy3&G`~QsqNqwO$^uF}2_+Iox*wvoBC896t3)-@zeS_uPPACil;_^< zv>WzP#B5DCWE`C!4ZB^31byDPohJA991GCjO6aeyY`V*03U;=go3>CpjsL=-n#kaa zskIU23|+`VXNLKFTMvf%^ll{j!U!b-?MP(jLA_u!u`?L7PpiLA3hBT1ugJqMSBZO& zvIOmt7FIoaAsk9VNTlgY+CxiDCx5h{+dK0x8lT%4XQf}1?YlR_8vpeh=$XEDE9_VY znGV1BK`Fjb=)Ley+qi9U_g;IvV=Hfv&Hl0Ru6`)(zNiTD5MIiJv|30)N+uLd9 zkPK0|_V>)PAe!h{78as!nF7t#bTv&#!ne4)vP+&&TPIO)rNwm}lu4r$iFWw6%3*UN zIwV*cg=evxXWyYR;nmY=(Y31xIsH$3qucvR>*x7Dp8Qhlv$MSf9O7yY)7|a9t~qKV zg^b4$^V2@cgL@*tds?RFIjU>vb+9#qGse#%X&?Gr=%o+2t(dTt5Az3xAmASZEFbWp zTrpAMM>CTGB=S%)7;&6G(1LidHP%&Kh#LwG`0VdxxQ_G2W|1U|Y2>L8{hhfKf7(>Tb?TD4**P$Z&X zCsfX?O1fmzlX6j30lv4O`6G07zYLqrjduIAvOL~sAEOr5*H{@QvzBf-2(-QCSMa9B z6n0#=zY2;dY4km-FGN?q%lB6)4v_a6g4CfFSBGZ;$|_L0K*&Dt0%+j$jBU3}ZqfuC`%1Y42lxSJNAi$ z@J&*`?-xoF=FD5fKLg6EZ#_~_S|4A9u!PTgtZCWRs${a`2suJ}aC#JOPx_&L@zXHb zGsk=w!}HTrgdRnb$k#OkN??Wqp>3A2w-MX(*{tOJs+{koaaR?^J5BO=g0Gu#%`9|N zg^bC zi#zYPS3N-~$FUoTKWSPOw9S2`zFnxj8%W2bCyCFHaIz(RtDJ?d5$c1cz_@intRk1Sgvc*>u;my3a7~l_bTxN!BXtB>6tsB|j zZv)8U;SoAZU-M@qjx(lmjOC~U>!usRX&Aai&}_GA35PJzNlBxy*IHiQXU5Mzy+n&z z^>=s6X^2HMnAy+MN>V$z+d#1BM1Ct9*G&U^ha?K+uu+!yO|H4(zuTZ*`!%3P_75T- zuHOOQPd2Bb-o2EL9GwWGq}ow-?a8?DGTI3Hy8y$If9da<8W{KE`nu`-bg*E>l!a^F z@d=(-ooaO^jp&^#8Ru4gOrL;(+r?l96xdaIp6CxQ+c}k!E%lwWT12|XkJ`HQjv(~E zu$h@(`ZT3R6lrn3b&(8O+34kkP=ic_oN6x8SK3b+o@yRm+YkN)+Hqg++J%U-WscQ= zh4Ssy%6-{g!YJjE*@k6o&yhT7*Bq(ZV8?GPL0Q%X*G@2u;K!{aPr^j6fPR2A4&swqnf zl+{@p!S?EawO&j=)h8PX9=!-<%t+I9xQr1zAg043l>6f!*+7t5NzZ?t|@hb zktx0(kmY-v)WO5cr~L4k(!|SFc9M`C@G$P*&mmw(8;PIeM;9X$(d45A+pSkaz8b_r ztzJAI4oA;WnH%}zo*+wB^mcw4$tK$)(VZFbQg+Dwlx$q0q1)dJnc~$VWi)CbSu#~o zrcuz;NNxqk-=i&H8%T3dV;DcEpK_(QK(rFb@Dd1z-uZtAt~&Lw%N`!=&XbeHFt0O% zqXl%@$1RhpZmo3?U}!E9X)L@##5S}})Q~?1snuZ+_i0)9BR6X4$1%M?{0c+BlT|ju zKYaBtixTwqfCj~RsI_I9S1tpRy&`He6ya2G>w5z!;qU??Dj0A3`<51vB0_JZn9P(e z!L^M9Hzfo=+LqhSRFBZS;DyyVR*%cY)Y$AmnkBUy@x?!A^edMt!?F#l667X+O4j!p zho0B`or*1ZQSM!154kH)AyXfh;-*MMa|a(+xteSefW~F&pK($O!R-|D$C4N>hO&57 zP^gH#zUIAK*pgilsKI(}$3*PFaD2rF#5`ORB1FnNh!{R*(-4cf?JFDpA|rA0wbH2! z=(P38;O8SKmEl<-6BVqv2ePIBqj)EK=wnc5Z^=649e4sFNsYjMuv@LgIi8qA?^m8X zH8B;%cs}}4BMwh?CW`Lxo z975Y_WnqFinh)G|6wl5vyF`f#{Oj;3ShOqUd&8jJ&td#Vd}#0llm<>@a@@gg&{r)7 z6F24S*6xLYm7>t=v7-2pq}T^?s*zM-|9rIJGGn2!HGpsx3s;|or0(PB(|^wm?q78< z749ySx=mV}Cl4PE*x{$YsU*rtu`&~=XFoLa@0UEyZ`f?u`ZWq>-!wGxssX!hi`DvJ z!eVbk+Mu<*R}0B8-nyrER> zfM8KoAW#=?J>(uuSV3{iJiT0KUk=2L*IR(MNun;_VdCn81cuhoyUv&ySII?Uy!Mwfoq z)|(R2(J!IX*S09An8C)FG`CeFsyVuWmZo+81WUk#n#UYNlJ$13JuTP- zeHvGtoTrk$EC;0kEcILe_o!%%vdOP)SPM1nDW%}FH6>dUe83V!{Fso$m+3PjUh_UB z_JBM8vyAf{A9OP~*0a47Qk1XU*9?n-9(UP4;9Nf(sywRHU89C-MvH8v zQK23Fx@j|PQDfTS?@g`+fF!t;qxw9_N=qzJ0q~w(RlznbGkVf>GDPK&vk71zG~@)4 zHon`6)hC*}%+3Jv$@#`W$TU{}ldZLag>mfnnf%PKcJ{ouXDPo#$E>^dsXNN2vgZA9 z`a`nY&+r`c6-8>Z%=QBGhx4{2lt*`7;{IEuB3w15^lRTcSW$MZU$i`Xt)Da+Mm{Bo4YKbsr9xz zL|3u565LwRn~?AI`9`4f*B8a{Abk5V?rm+s2gf3?4E#c|cZ95Akw&z%XwAhS9UAX* z&8fnepyP0XpyE=DYSkN;Yq_~0#p|Qf_34o;BZ+OMdt3zhtSWtXWx*6*ZJss~K9W&& zl0C#5w$(|;b%XwO0>uNo?nWqP&l8&bC}z*F`{`7>Y^=CWpmvNs-F-)o`4}`U_w+W- z%oM`Z-xR^hmZ*Hrg7LNsB3G)B$p(>hIe;6?N>e3*^eFgEUBijQ9uLN5@ph&v@Qi1l z&FcO!W?c_?rb*t5cNRlZV|$iLb9_1#dHuZ#sh3uC4=0*LO`j#+i8+J!yBj?>x{SVP z(U*3rnjvhL6^uCheJ2(sLCK|cnz6X73R?R@?OorF^E5;FU?*q#$R7JOG$l0~InvBJhDm?SY;#4Z$R*E~F{-e=iS|81M+-!~X}*>^hBDtZY-L+)Mrpa;*|ECMRUvm1I`Og1 zpFt=^Y=tM-4%L{`K4DZRf*_zM;%V-Atq{(|aq#C@G%~(v((mL~K=O@go0=O@-7%g9IyL6tj)A7qY6WGQ=frM1dg!r|e)Gi!v6SS@-{;2CipExb5z(vNZ3vC?h z-V%{jCM5H+^!^O$JSAwLQlN=VkJ4*;zF4aEImJMWJ~zsPBob}aJztMDF{-kO1pjn6 zhhbA-vwg(2(88U@Vqw>W8xypcOd)LGO9)kb0BKrcd2W{CDzE9d5R{$L-*o)L2J)54 z^}U)FhLGh4RrEF}P%}}Q_CVWiZJ5rMgKSvk2(Xh1!JWpQQ6qi~{$65s&nm$b9;NJ= zD~ObeZG67S#9D2F8d}nWnAlE}@sC2jAiZ+0&NE6FY8^^R&LKA4GYID8beV&@oG$0O zs!#Y}&D@g~XCpch#*FP_9I4OO_;!(BwHta>2eoRmUp3m zFw+n6MPmszsWuN~T!Ji_na74zWQPnIva!6!>Y(e6S5T4(Jy%KQAi2EJnd%G#`^Q725 z?Z3^^IJ(4AvaWu-gg5!F_9YMlQ@!XmXi_qEA4wCCuq9iV!uevE`!cFrgo7Z3rNT7T zPew6H2(-RN-htvuuaQxVna7kqqssJCYi~%7y$Wk&63J<&niEQ7m|oLPHWcHXuFaF9 zfsE+M>$E$?9o-L}Cjh8UEYAK;Eh)@2@h;j;E$*aZqEq3uFlU_9r^X7+jWZ#qyilh@ z-5&DeX=6v_ra9ve4NRh`q_KQO?H9RTlf5zFN5*0LNI;i(-M@VbEJlupmmc`QJPM=C zKPa`c7HT_bg45EysmRL=enXJsn}49a8|fv3UO+pdVX}RP066Q6F$WX;$SXH~QY1sg z(xEz4V;`qDWtr{N9^-6t=p<`n2 zZ&^V{R(pxI{YeLyYLU2ygRUa`x)Z9&@C!)dWXw1J&d8XFpN?5v{!i(iL8j&KV8Z6SeWSKihyjJoIJWFKbm8YZPN?LtEe$WZ$G^RjaJcZrPo97JrWkY>1gltRR zSxZ1XuVOp0oa^-jUp^hMNh+cti~I5`m&|7*06-iP#Yv+KCJXfwX)&8u)dkkX`4>@& z_Becpf*I3a^jQ<3Za6$cp*pq4UZDD5DQ68!n5?kDoy9-+ir}Y*cX;{^?trNGWs_7# zdP%fwhY%eLT+3Miq!Z$vX!;A|9k4yer%jhaA3iIf9Jd!Ka>CQZQj(m#<(3b&mUzXYW)x#cFQxI5-P1g2EK0l=E!|1|Y}rRLLn zFAUS*ip)kI#m_*v;hkAGKtz(vjrJKL4sq4c*xR3e++JJi1o++0hVu86Z(rKNxuT>~ z7X`H=yoe1#TuK!2_Y06p{qeSliu{{RXbILV->}Zt_$h_5Ga-)C{9&e)TH`yag-YkY zi{V=>0F$I|NUQTdQZcTFpvA4vI}lehrlzTZ0ol7lcd5&c*E!caxhnIjJqoH+nRBLy zdo^*2k=nx@t~O=J_s4o9v4uuOYl~0xTjp?caiZtvayN*Ch3<4FuEyQVUpIk&h>Owl zZppmdZ;o$}1j7KO4Hdf3Es4hqik2}bwVM+dMxDi7Z9L=N{V4$UxGG1gGg4s8(YzL2 z^h_w8S%SrkYKexBLVwbX<+LCOw*oGu!a8blC2|z_@(B5av@|m0Hxs?ytlwDLL{lvr6E>g&bVzhQ+;>#{G9*{fh3u6PTi+kj1Nb#u&2*kZ3SYf1*{OKQbAax zbjtnKSQ3SC10Zb<1T_G9K2>MoMc56WEypU7HN$<($-$~388_AG!C&$ri*_>YI^c?$ z6KLF{jQ6@wSx@>}H>7HBlm3BPq_3HVYoN326NMm{19Mh_PEVS^`+Li8;ma6~SL_v? zY#|IQ@6Ip0TS%fE;@KTUBlRN0{YB~OF%N)~-8aq#HKSk{V3Tl!0e5u+vxn_x$JPGn z)GgN_vrHVB!~S?{mdDoW2krgISG!HRWCLf3@0_}v%pG!d#L001GcT05VB_tRjyc8; zCrWG{#zrX5_no^dl8v$656V5neOMZrN-efC=0+djWEwUq+~{6Vv;J(TX}bv!>3*&w zZlaQklG`{QE0@rO3NDNhVBLzXCx1@06f}elbW-o>y!?IYt;2bZ~FmJ*9|n^>YThwTtug zV^J1*sTm)7lis0NKzLOtF#$e}3bhq~!@wqmn#HY!Iz>y@yHNOceneO{q*OE#hC&W& zl3G*s83?adMf93jZ1Thjv%XI1^@D7_m^NzfW{c7ou|&WbNGz#IIT5lSXfEWz(vCz= z$<_O>C|meVewUG$OFn$Cy|S@h&m?D3xR{CLm`ms^vf`>9!$z#3`B7?^mB-dk-BXQ8 z?4G#*X8e(Ps=T{0;o;5JnMaWf+^fHzG4y*>$Ec)MkWXYmsDrE5bshqNPhE)67F7$$ zDQxpOlqrMiw!C#vm4Mkzk_jW8CbYcaO<(QzVAOHCrm&`&0rei@OiCpLtK(TXA@Sph zDlS6I0JAN@ex7B%K6-nA)8Ho%Q~lW*F`k_o`8&jpO~MDZOeHy)quXR-w8kWm#;6oy z#Qa_J#CwpByrgoG^KZL)8OK{EMI;%vcjql*XDu*kv~ajiLzj94Z8^G(Uwb`$<&jFA z&Dds+W>cXc*Dn-O`jDET4^;IzjcFTsa#;bW;E5!=7ME_S=wQn8%I>UL29f;CwNM&D zh7u}ZVI8b)^}NI#V#^n2Z6Q43F{>vBRIfv{^O{TGzEk;kp{nUV^*=RC^<^9$LP%3e=8nE?k8<`B%6rE7nIa|J0VF1I8{H264oPZ@DZ&6MdV> zEsM=?>OAcEWozd3YOH_^Qws)#C9_I`PtMTq20P%VvU>bom&?z@yW{F{j_k(;W3!K- zxW)w`5T+o_w)-*&W|{8Sqgvy>TvSIk+%A6bu4O9-Ywy{bMz}h;8vrdp(!UWlw&keq zq?wfxOWnucNg3avkrbRQYLCT+@ptDxPec8XQlz^;jG-DEhrA*Y!=YeF0Y@2u+>?IG zRmcF~L0N7A*hX^2T|&}{b7c0UQrXN7(cK_yR^7*^=>21od+q=T+R&sE&k%=dh+ZcU zI9aUY)EQ~>k>4q4616|HCH(!|DX>ho5SUzPFDi^SFIZYZadMEeQ(8BuAuS-8gG4=Z zX@q3y0|S{_@gCgOK@$w`B1OnhcdY%Okg4?c?a$&Zgt9c$y%Y6A{H1KmdE1!sS()cD z&Qmt@BSw=7#)bm95br;(6<3o(D4db!5bpe>GRh=Ay%x zo)^b^NMaT`58D0^S7a&bXCx8QTTKb;N*FE8fOnq=d&wvX=|g9~NDWhq4qE$d;E$NH8QG*m`UrI(1WG5%5 z!zGRlA3cM+>$XWQkyLS!o_g4j(g@3rSX3?fxe=0(TwsghjkkBtuTk(9-y)a$N)CrJ zFe2yC47IHw9oSNYpc?3)tPoExDDT}_$0ZKqatts}KPyzdOA>*cm=>Y7*fnew>H1s6oM3=K8s+-gcnOG{9#c7tgGUc* zUhk#%TF1QWaNTdo_vl!m^oquyy3V7MH+YC1XPA1WA|e*NR8?FfaMXKhzSXFaZI=DP zi*KT?Y4-?~UmfR+Q>h3>_8=zD(}PJU%*oQth*({p_xO@&d~B65ZJNb?XtwRSJlb=X zQ_I5qJiq0sXLxV)Hun)-jEVY^f?wc|%0qkDn`^t=dvJke&vC@;R!!mDwGLA=reFp! znKFl5uA25%j^w!_R@`a;E@IeZz@87t_SQv+J6B}f0YQH0FR=O0_D@&pn2)=h%B8)A zahD=FRYz-rt_!H))%j-NWvOgeaV%^1c(_aMIbEeNFo}z!$u89K))H4IE6qF#vwbpc zu_&E20b;=c=+PN>u6E8Z9jP3YRw9MT5PIw+j`q3{s$42G+Jr%H-r7Fha{2PR_;YE4 zRmwf()WP|COSk98X7S!d>-A`jdJ;q4t#hZDxEg4Sa1Kp&CtU%&ofst9R#w4l<_Bs@ z3!}hxkp?mQ_M+jIXwhvz4u7{~6v|zPoi)SUj6!tHGIECX!8ltR5(uy7pYiPZmWWFw zb7CzFXS%W_1APhi!{yU(T%viz!1?+%)RDIIUKLJkob060tt(4A=v+fSJSp+oz4ZIn zRdB!*6LRsgeXX;po6pFnH2?C%(Ck>~fw;*NXn|5m3P^-+G_~mD0gUY(ugjqS6h+ZN z>7X7nlBIakZS&-wA1YTdNbZuwsEu~iObHZ=k;;3OiQ4TP&cE0MmiK0fHl$z+!UnBNmYl`lrAlt&+Di1F6Xu=&Mjwe>}3)!aYXfWsYl|XpqdOrMm%|K0F>^|ucyy%mO zMf)DUCG)n_;aMhE?36BQ5+!OQE#vx)=TWN4D}iXf!C1x<7I!_x zw-TI@>PtvlG3$*U**xAD+~>KXEu4;7*1li!k-zm6JA~HYiZ3*fVMcAV64ua`?W)Ak z+SRoOE6gFWxr_|;zb(&o=?G$^(p0pBfc z9ebx;8k}zABwQ+TDQ*j1if)iOyfGaC;F~hSV^YHLppL}r?esahUzqYndcN!W{Q+&q zgC9!%R_C@lFAgmuv|qEitzszaeke^Vf#ikzV5hu1ywLl)(CjYVeo52Zc~&xKlXbiM zHkNl%?snpiR*ZpHCegd1l{ZCn*^{zWyDYszD@jF-_NH8~S=at`=)M)@M_DD5w1z-6 zG>(Fin`r-=K1nHCe!BWDsS{l3v*1Tl?sX)=%QlkhcLR}xu_a`p>Nz0-hh?ibv$fd6 z`r)_Jj6uT>SL!Xc{bXs-P%?awkKjeU29q&`;J@TIJpiVKV^!s@m-47!yxFno*8;bs zlNBVIe&oZytHlZU)tuFFw^olKn9V=8E~baK=f}6KyIK@jCB<~>V2$0LvbnOD$_kp% zs}5$B=Bg?fYb zO6@6wFn3I`F3)kx*1g-bZByw~T5?E1ecCblREhG9CD!t0Eb8!xeW1onk~d8A`#m(B z`ybc(jC^Kn&&P07XEz)|J7u!DEQIUgdqx<{$AK+_P?{VGVQo;A4m?|bPo>RH8kyq)^oQXIsII87pS=aE~o!qzuhWJ6)VaWOYDf(7Re(&Cij} z!xDHqwGb6De*@eXA${T6&^YR9eVlGdq2tFDu_Ls=*r-J3I3wYfUVMaEG zWIGzy(2wD}ez_!-NR7kWW+A#GIV0}wsI-c-k!_Fbi1%v)`Q^Y|CT-EClr##kmDi7O z{$h^c)PrEnAFTOhxuoOUA7_qPjwHFE&qWLomzBAasCX6*ReWD$=2|fbCZsQfZX;QB zIUH97$u!UboN!)q{emv9GUPY7r`hHe4Jh{&m>;~uD`K3oi-$HglWcd1yK?eVs)|I% z!_K&SSjph(rsB|iD&q}oucd0dW!pwc(bT?bpmroHC_LsMRBLB%vrFoB@?j}g9o_kx z!$0VAMt+pD7+jEa&bUqp`7~*BEU{wY4o#LXrULDc{vJ+~D8P}|e!_^D&@eFjb8Jn2 zYv(47eZpaqTdL-B7`Bg|;S({d~=gWYe_9GRK4z+}5t4pK2@q4_Y5;?wLKzbAQ8i=K&0h%w9_` z$#|jGGkJA(4h58Y@bpO0)GRhs=BpPJd@4UwDLn`-U49oPN{%M=>sc)jmqSML{3Lpr68ilR+wg*H#}UULEg zeI4@23K?Abo>6P1XHoBRHPx>WnpFYCi0_=8S$mw5_9myGy?d{R#oBi8u(MbBy;ZF| zO6<vfuZiE5a&&66m`HqP)z85OkQU{xd7|F8ersBY<<=9! z8YFo*`ldx-)8c8TK=AO*{-~?<09EU1t&}57w`O6P2jeGwZXmE=WQCzj?#1E{4WEJl zV_ef1Y{f2Qd-eJauA@3+y!1;Zfo=nld7WB9*nRK6b2qTn(tvVwE|V?RBu6(*C}V7b z$kD9XwzBAAYoeR1O}O@BIuTFRIHY?Y{gs<;U2@~7jIy6 zsLS0Pk=4>BHp^#!BFO({xX{b|>c1N8ZpFzZhJY<4;FvK!tSp2v=*&PJAF+;QJ$|1= zk}I(;tf2*Q&CAtVE6HAnSfb>dvv<7nL1e#uPf0_5;GIW{#|0{UB|h+B#de_IuRltN z>;=vmi_l2Ip*yA(*KfeGPnkIil?R}sj`bw^uWeSF)jTtLQPjaqfSilN;XAbBj#-uO zbN<@FxICa?-1A+Wo=a<4s1vF%b&atd{hu;PsPUV zlWt5+@w^b($dG!R4L%8gz5E+@8iQBNlh;eJ$3|2H+ytS^Uj89fT5J73>n`nJ=B2SG zCI+$JkrpZDJVt|`-|>cZ*VhWeo_pUfYC6XrT@g>tKP;v@1HECNxAAX}+H4VZA4AJ9 zNyc}lSWW;mTLd@(;~ClpAqCY>KXC}F+-*|-h{>(`8%pW?l~z4BBe-xcHyYWZ2|&^c zqNKTAzQfqmY<7u$L2O3^71rncF4t@X`_P+NxFHN@G z)oME&wdJBSp1IFQ=D8|#)c7Io`?l_cNxP57$*+ZH2;4osF)#t5vwqAxYD#<7rRi7&*_JVDG`?AqmwLVRK?$Ed z9Cwb2xvM{E2eqv2`P^>gXhjm8fFRnzgwhC2?_w_6SEj$WiWZpvNfct6i9XbVnHN$5 zJMlGlkr|>r-c$6Mwm0kwHb&WYwD~sIQoO9lL>6KLm_6RKgvN*oRjjM3O4{=j}gBm`m-}7s! zkoHF6R_zVo0eqPgOW)KTc@fCO_!CN?IJkqX9>@_#{lIROtZq@TR)@po_ z7rsB!;Nry*6_j!?X7s6_xGf(5Zo28kX-y`*O(Ef?6^BXo%+fU;Xh}$3OsW^^LiO#= zSRB92Z0$~MzQ-jzB;iDi;SO>JdyMWhE|n%THzI zxGXfj>M`bwM?LLM%CKmOk6P9*A6IkO%N&9XW{nk3w#e5Q@UMPsU$_tT+w z?L_d8w8<=f2eMP6Cg^8fOP|kU(KfrPP%(r;@zBDQuIM8I{H#HEp#2Lvv*#IVo$ z1&J+B{AF9;(TYu3i=s&LJrT{)Nu)8p@|3&lb}UL=%1m=N9!*cgCzztU8Oy&zPbxNU zl1Uq%s&?;G_}apIvrj(6%W#sHgG2%~x9rCm$iv&b#hfGLVR&2YN>Qc0%JjRljQZn_ zbPyrmE}Al|Y`;OZDhQaln){-WB4OnX!EYo)W(x-(<5SET1DTA@R33Hld3TD1zemVk zINDm%c1o((yC&s_>jZ8oT#wP}$WmsmgBZp*MD))i%<+o9_D~9OHeB;|c!qylUxhm@ zc9~W;3yFjV4UOunu2soEtpbjV(6@_Dh*QRf8iGRUD#opmfF&r!2ruYvra1Vt_u>~% zjkSRNB4;T>RzjTXddI3GjSEcCh2l;A>@7d9bzdBfvZ`Ornbeb;e&?+z#tNBnGi%iy z-WJc<_E475vojv1Eq8L~*bd?rvA9dn^spa4A%38&we~i2HFL}L^gN>F5)5J$GB?yk!9`G7BA-_c7()UN#VI zQyKjQ+cI}=SD)MWtb8!;G9p(WlYRl$`?eqnEEbVW(3TnyjJ@xeXM+|Y43VmUf3J%K z5>uAtfQ)9QVTdz#ZqJCJW}#)?UZOb)E?tYE`0*s|25ptYtb0%n?*ArI^ff~ef*2V( z&c{CV@}LIYg8nwVUK>svo@Icxj@&0E%;!e?Ufj?T3`GlRP;038Xv&1ZZ_#4_p6q)) z_=F))2@pYG2qv_fjaV>f^c%2?P@o>lq#U_gx=47|%_&*-;eg)+zAT1^^HqTQb&8@p2!=&851iXs>xPl!=hB*IBSXS54nx8jlZ9k#dU4gSAGBKy z!r9oCSlp6j;Zs9_+Ts(yg%12V#n&5$VnFx59_9$eQhfLA%Ay+j*`O&i)t7pBCPp^N zGLfE#KYUG;tq7lo{@Og> zAFr907CJ%dJ-Ey1ev&F!zG(`?Vv}@`bUB$}{F(Edek0YL%t$x4|BeLIg+gLhO?y2& znfu*}`Nc1L^g(cP?pGBh!S$DBcv=~ICw*1(Jph&i-B@5r?}G9_lbVdZvW zNX?G4DA^g+?Nfp?GIQr74`FK-7N5fWJhQ8~LN}xBOmBMM*dev`ILPUp3bB{xocqz| z?C)(?Fmc@?()SlA_3ernM~y%%Ojlgxl1Bgu5pmz!6;gOGjiuT)J6*_rGucGat^Wn0 zH(<3Ku>xZdN9B?zV zy>)jY$81>QYg^}k&jIM6#6F075#!tLC}TacAZt`@sn)wv=n~3lkb=EU=b2E4T7G28 z?EL(i0DGd)ByLJI&NGdddhp$-Rf^Hf9rk+5^mwLE;98VJ3}1<7jDqtu(XL5GTkcNw z-0sj|bN!%pS33$wJZZZEkeOU~wj$irpdf&@!}uxNx>yNDT;fq$2? zeIsK(&q0Ne>+G3%1(J1M1gg-P#XHda@q+XGXaXZj(>9H?AQYR^7B3=|gm0NbzL*;| zmHs^pv2?Fh^@;m))thm(-R!))*Mi%vgsd|HRp8>i(?eiw?(zoUyDqyhp7Gu?Ur1M>9UD%h#bzr zNN>uJz0Q&y1kaugw{Q6sMPg4i&z&pA?~LigO2pmo z_2RZMLVuejkIK1z!0|e_e?LydV+6zbSDK2!8czoA?thaRMjHq=J~54wdWKr+3C&XoyAD#At0a0ibt+Qwfs|7QvI zBWX%F)zeWmG-&R)9HjmZ4i}-{6f`qjM_C?hYKN$_7eI9IT))qTJps}<=r&<{^DwmV z@J6QZUtn<{#WilM;%BWZlP5d(SmMal_lUs3{6cc8{z*ia8jZzQ$)pC9VHrY;{Ch^aEcYp#~ zTZ>JO1aj@vVf4FKdp;#xnGGUfAV{ewa)}9{|F*jO`xFpRUnEN@1I(L*U0BF4$jwzc zXBBz8TR$;uQd`AIE_<~Ani+!9_ThuQKqL}V`ui|wU0vRRcID|BQ-O`e4Sg*)78A)$mfR-PL7hJ4SnI4QAt%B=SqJMoTE5vox<(hH$d zWG_{E#JofTR!${6abA}tRK9}V#q876RLL23Nz=5&b=mRRr!w%+bC`OP)Ap=Kd} zVU|_lafl_?11)``Fosld|2?;*!e9_zY(PiS*@SG;6agx~%vy?g-jmaot zoA#X!MCG<2g4cu;Pkc)+;&iBTZ(V_jDz=DZIWLI6GeHIfSwLulzf5%2Q5HHRPm5T! zTsmZ=BiLIkisg;N^oW}-s>Y+(ZF=b_pK#72UPaCI2Y#Ax2>ppiZcyA2Ik!Fo{L+4L)gz!Wils!mcO|mVcF}XB)u6wHiZc#>+EL& zZlO~+DHW-PYpUIO$^1seEHatqzETx?zi}plXt1(Os%?kI(j`uhHG2iu_F~4Qo;2lA zS#Vr+7F$p%F55iu6@+kszMWq&h+s1s=P)k%z4kN; zNS(*Ajj6~+iYC%mEt(>=ZCr15$eq-+;~IwWV&aJ6RGLtGt*~%oW&6HFjk1)Xj+#vD zZ-%I!fx^4ZEZF4Ww@Ad5RM&pF1cxZwbG3P|w;z8`;p5z|kmlWzZ6LihgwxW+LuHl` z=diZ((^^^MNhE7u`nP@U@7sPSzsu$p+t%MJCMz{bT%1CCAc(o2ZfKk=qOqsi$`;u~ zHN>VUkeCAo@)eGT;+BNfgZ%Xt7OVUFB@aZz8`=Gy+i@V)8#U8r6P_OQmJU<+c9)~$ zU%@3s7m@Z~w}%&r8$x3#@TxqCLB)#e6=7~x8Br5T?7?gBKq+N&=eb;AYxJy4BFj9D zFIz~|6@#xq@ly>%rE|5Z86j0i( zGP-g8`~&WU9J?_(VEZWVzl!`jGMppu^GGU_7k6Y%yU#;Nmvf-=VgiJ~bt-HV(~S1` zSrZWbSid2S=}6iRZgp$iCXkO#CKI7P8Q^d}X>zQkdTglPAEieFTV*CoYYR`iqsCct zW#Cpp>HN|Gu)aMur){#fYtOjTy6~#o`homys!X2^5UZ$3a5Vn`%kXvKh>P>-;>Ek= z%4oJtL8`LzwVU-AxK9+yqFgV&-CZ8}+V^N6cXPkDv1x>L`?G~O%^|2p^nO1mtY<6}-<0|-=YzNT75$8J}jz|rFxAYfRs3thcQiFiFk5BIq8uuO=W zjq#dpO=HGy!LXj^N@o`^l}Oa+8(_>9dMp8Eq_y5UBy>&HfALn9w*g~edvXmz5l!IP zs5S`&IP_MS{)x#Q zenR8&mChQKdE_td_}D9!tgPa;_eVjVW5VQEK0-H=Pp!I@5}FiH?Ghtxns36LeM0~Q z1AJq(cP4E{-{eXyd~$JU7c@w_()kV6i(CLNe%)eDViFJtLkK9t&C{MG@Bqb$J0pmH zRYc+;yeh9d><$Sm&`cgku$J8~Xzi=h$7dMPCGyCDToPo)53GfLUh&-)o?WT-B6p-@ zD`t$s`!PSC+p{cpDgBkQ$fu!WDwL+o$4YlZ@|6YMbwxYpZE~UP--dM7g}c>Y&M5Na zWIAhz=x4Zco?2n~(FWVS@2$Zp?Kiaeg|Me6#!g0Ql2ck*<-{X0rIE{KIBMxQ>E{X} z*{jjw59gdp*_;qjA+(nQMjvIQ4DP1c%IWa%a$q_g3s zA!qd3#%8DVA=veuuA!9w__q!EvNV5uddv(EqRY_w5_907Qz9W}MLks-@uyS|7)Y-I z2xkY}uCQcwUi~j@cM%WGo_y}4LW8SZbr>&PN84VO&9Rl`8m7<)n%No{xmybwj>l=5 zTZ3?N)#rX1oER0~Y6YZNFbC90sIPQ4Vr6DZ6){h|e%o6m4>ZEXUcnV~8EA z;f>K5xGo9TwRiNRIEX3c+TOv5d(~PXzUxc2P!{uwGs;|K9R}FPIi6FKOZ@u6Wz-P+ zxpxrmkLy?VtqRaEg=y$(%$_yoQXa~I#=tUc-mGiS$ zA;ymeKHh-`DfMR;0-7U2~6MN#9A!6Br<>mrV zG>dw{bk~2V6XhNz2_a3-tZ$1uxxdu*4gB@&Q{FkGI>h1+Z{bYB4-FccOLB2paiorF zou0=Ib-x0`b#XLL%uJJJ+ZJtBN#TkUe#N)z<={V{&C%J$+KooGPN%*rM{4Xxto&_; z9nNW{-x&jT^8QNU~A`})eo z%R==$Vxo3~?=p%tsWM*?WbImhvUi;cDHjS_x^=#g3d+N`a{8g{zT)=vE{*z#KU7o- z_uSTA7RxsptF?P+e#gA;gT4?>y*-SbL6?f6_ z1%K#vSFxw8lT0$XD$)RJFhx?>5V)0@TFN%UWPPMDlx;Xa9%kGHQ_NzEjx%d$f>&)uX;5{J72-tm>Hxm?<#y&$c7h0Rc@;ETT9_KvuXcA5GeC7VT7~@>EyWY3L9$!wCp*6pTvWN8?~AjMn<9p=Q%w$Q zaiND9X_FzWL4yyqNSWSwQx$eEj>CrX_@N%noeTg@?~Vd7#$ZI|{C+Tskti^zrj%P@ zt-{EWIC>3(ctlOZk_lX~Hc#TSehGv=F#P>kI4s3Xw%CQ^^+`F%@-Qi4W?JQl{Fe7v zR>JpyH=e;qk`^|OQk^+|m{4cr`gal6elD>7!JxSz@=?U0f~rWzIVJ6QnyRjXA#+*} zE35Y?&<71%{BSVzp7{-WI0!>s`2F!yQ-%&Z)@R`A#8HHP54aVL?)G`kB-F4@?D&n2 zgaDAlVG41mvtN&AkhIbcvJ?HUNORZN8{4a=ISQ~!TV*)-g-K8}EXO-#iKEg6D~G43 zI-*N_2E`$k0RR25t=BHi5U~r7BsLh#l~*ALP%d8J2EzM<9XSl-u`QMyuOf$xfe}_x zK^CtK-8^omj3)YovGJB#-$U0#0QiSX=RDtn(Qw9fg@PokRgu_?MS$`qb)*j);B%@e zh&Nh?Vzlf`BbXJgO~tpN8JvaBX>3EHcTwQMx~(HRO?SO5+g|ob5N8TP95G};mr~Wy z`=1v)41%J{iG(am8rVRFX2L!i?wd`GTIaYJHHI-`5#eD3FL&k@jRO5TaX1B9cd&2^ zBq8zoFRG7l{GE^dU6}3}R8fG%i#R*S+&gac*QHQx{!s*0rdDpd>;i9I8L%{KzCWlP{iz z`4)B^ zyRtAKx(FRNt1KVJQ zauf=DS9n-#Tr(5AivH<0$SYq1V! z5NO}u$0+wOpy>34ERNx9Csf$)XE;%X9-dP(+8Rz)*h5bst1ii5fAVgXEQwc< zh{83y=*Chr-hYpEcqM0gM?o_fsU6Q>2#~)3acl)+h<2)6pt&52tBphSxz_k={%*6v zUyQ(c$#7P3g|esg>2tJnh=uQe)4QcG`suP=aeA^Uswt%|E61FnK;6zhOKEFEPFQ%m-R_`FtN!q zSwT{}Il_SK0zA9RRhOz4LFReUZ%cxyy$th%E9CDee^*mSJA7Rc!&iMAky2h50Fgmj zDc5b8i!Yb7Bp^JqJwMlS->-K*+$}~|&Cp~#8cWaCqZXpPi73>YXo^lmj zk?5x)|J^hn-Q%t{mhZKFI*$;tPJ(_tcez-X!#!^1ZkSlaLz(nSv^2lAuT=!!j?SW> zUTQ2z*sTc9eyaQ1EnkFszjU;=>ymQ^&|1h$?gsWg;m1h^0qf!Up>4%)b5T2zQ1$?^ zP~-F8Glx)>(rx)4B`#wg{bUHYojs!Xm=KQ&A)RevQCvlmv>ct{p;O{BnWJ)s-qr@#slh@^~k0d4H-!Bga}9 zW@@OY{rw3}vgNx6H-l$GLbyyTqe#^j06#p3sprt>=Ku%`y#!|U5CohU6fwOHSyqQv z@A-rMyHH7-9^Sfa)VW#5+eABleJ(iuLcWg<8xF2zr47Pbt{vrOUnIk_cbl8>u?gP3;oWk7zor1z`VyN`+hcl%X6lcM{?>sT*3xo z$S^M~NIt1r`?JVrt7?V9aAh)PMyeJI`x%!eeNmrttVN<2HWV zX7zGC@o^ScVVImK_Hwn`fl=LYOfIRt*kwM9Fw(h_I#>ZkQEdjkQLJ%H`|!GS6_K|H z4v=*zuVJjxYgL>@kSu$` zCsPAGhrxBrws3dD_;D#cHvAz?Hje>7f{1s3z*Q~!BXMyxlg;QoQ1-6kHBJ~wzhtLi z$9~*A9^auAUIiv#vv~)*SsOYhsvL8c5ELvT2F34b-g6?b)k&5(*ROV&mxY48DeT_;@v!#Mt%fl; z$aX}kApnE@quQFaY1F#+7vAy;#cO22gONmu0O;2M_{9>#xSPC6{qP&BWCX4-D_W%_ zKYN3%z`Rqu`@T7Xc=kbaPV|uO`S&+!hI)8IJR;5BD(T5C_I`&9$HzlV3UYF6BM zp(^KZHKxc5Qzy^_x+-h-=Mie{rT8mTA|aq*!fJO^#FdA)iJruP^|(U)z~lD$WAmf6 z6<^MnrcITVH8j+uDoYvz82+PMWX&<_vqMwGF&NHzR{n}}pm}~Pf=HLT4+@R%EPMB} zoGbbkm&Bt_@77;nIUVwkqVw3269%H_FOkEN3{4U_?8rH1zJ8dE&#|X%(~_#*y)-Rh zSGXOPmq|VZ4QREPhHKZUwN3c!bF9SjZl4j2PJ)2^8dD}!9{%)XJ`>ef#~a4W9#hTC zS4}075m{^qvf@#da~Km6J#9cfUC^xPcE$MGSl{1B;a%Y39>w$i+*mp?`r8>pC?6ls z$4!D0u8Aq;6O|pA+b$xRmpnvu$Z&U3u^~jqH$lo?KS0I^$sIAMx{=RF@yF_6%a(*j z-|H_nisajdXvdw4RSr_;Ntx!p(nt{>$Br&PQdJz9X^oPJh@wM(<*fJF zi_MjgkXu^3_e1tX+)pwx-*C`;9|rD$*&(_PWs*W3LtIRqcT!WN$ZZ>0s~9Z64JW>t#JxADe&E=TgoKB^A)e#RP_8n>EGGQaEXPUK?2Z(LW$ zxwA9~M6Tt}b1^CFc=Ro7$A0r?JesK0tBgBm-co}y1L3KZRyU%=wE2Q@g8G62I%dTJ zOTT)O$N4uU*lBL{TFu+2^5VGY!J;V*oUpG`xtQOaarEO>E$AAFJSRA3hyRSA!lVV$v05uhR{>&TP`B>9Y_&B%HXDR&`6^CC1uo| zl_o(BZlF0bGlc28=#ISL(%g&-J2GqSDm0|T_wPcP z`RB=Agqe5$d^~=Tm3x9)<`1vPnr{9$k3rjn;fT#Y*pR7dXd3K1_^w_einWLcmH%B2 z$ujTC=nL|T;}8@FT7$-+)`33axF;4d*&qhj@#IVQcadxJaW5>ir3mpyoD8+)^?s4i zUK+FEnmwMzOyt%`tP3?Fh;*3a2}AkIY7Z5NIZIP+hI+bTB@aAd)5V*eYc)QhTr+>L z?T&DppR$h6oii37^m=}U7KcbxFuBp!nU|Aa6GDG3RmF};KuB&EV2Nh6#zawo#t!gG z9g#oVX@z+9RQnxPK4{Z-irJ%vGvmV}k7*Ui#~2i0Vnkqwk^o5noi^Op627d4xG7nxsA9Z{X?ta-HcBN^+!K39JHn!r&jfzot8|LxR3k(OuJq41N?k9n!1lAypF!<<|>IUv-`d5)w(qg>|{g~BC)r$-x! z%>%1eV}|kcTlFyYU1ruYt~9=YMDVEd`-3g`)KlZX;h!NC#4lLY9QXNj)K5X}Wx?%Ph zn7Eq(Hvb>*%;~MCizaj-S$S%R&L0>z6fCFHy&0fw8`97&kHS&~==hbkv>hW%hjQ5P zl3qCT!YGBZ5*gIvgTdAvGPOm39Yjo`A}3zp)HT21E<1~x!Z^=x(TS;wym}3>e%Ccl z%gA|eOJPBsi46`LBt8{)=8HxzLR)4d=qB=FE-fvKk9s-F(%G zafVUh&Hb7mmJo~YExiytx;rGql+iFw0iE03&4Cp^>hiZcbf5V6Y1e!JK|sF0T)g=~ z;jxTbDmk!0t`cWLtB;iPt5>{6xS(uFo6e>{Co_dy|240WZ(QTuRZUp>R}{;vGESy4 zg2aIFh37E&-uTmJI&v{lY)6#HCl33khpEn@b+FLY2Q*N#;mxOtToY?_fblb2M@@h) zl-l^52|`P*R>9}P{Dj;+l1pa#OkQGtG3vTjHNANIBgAXn2T1H@HMG{&VWs@-DP8yU zOv{$>#{8#VXS6=`g9bvB7~RW;x;R_#SnA*ymfL2k@K;Wl2X?3ra>N2+D0R_@#=%c& zc|I|h6m^^QRWM)vJYvnt>xh!$NzZV;g088l`39zAln`l>=v20~U~|7@ms!4qCE%?s1b$)syM7uLaxR*fWbkEF)Tf83!-6KzMQ$9UF+ z=#`HR-P;L>IfsM*LImCX^q4z;`sFJmW|1F0R_~Ny_rBHm>0v*{A*S>zGm=af?|rLO zpV8tM?B={bg&wCpuF8IUpDbZp9+5rVcI0YZVvD74U{UaOF^lu-PL_%KV&8=SN87Av z1-Sbsw9fMY2tA-iv^b(Nk36iSeoKN^)v#M!u12}f|IOA==)X#&=GSVH@pU!>%TQkb zNP~`zFS>_G)*-o!+^h$ZbzSRAyjK26kd}2@)cUusLbts@Lj&;wY;HYdHciaMKkOZe z#o${m=u$!;A$Q@&n_Y{z$kY~+H0+y_Q2naHNGl%vG_7#X$TpYZ&|k|Te0;*i(@L5b z%!e;*8{1oXabH^hx#y~kkQwO6fP=5inW;zpg#ipx&>^(P;t@RWv> zL;O6tGHc8?A>PC;oR)JVACfi=Ij()D;DZvfuJ}8=b70iqoR>rdvu|JTxvS2k1;dRV zaZkH}qML(lfX3Y&8FZ3e6EGg2Tm=KR&-bZ>UsT^;&EmsDonG}ikI|K`VZJJ|4hkKk zV`L;0{n&B-9QhE(6a~jLV(P|qEuCxVkI+@`%QB=|5k_XGUOnry88T)u-2>^73A$2x z59@2e{9qyWky4>(PfMP~LHIU#=830gy^UWCRYzqzbW8xYFzDXld9>igjm^o2q`DGc z9KyEXVN9Qn;!p2zz7pBPqCnOuHU$NDKv^vbsHhLsnQgZo9GwAHqIIo*L5qHKN+)D> zvu}yZjz8E}74gVGJKrOOZ$4R!r6?6-x zB62^!Rc3NQgN8(Qp4szi(Y@D(ajT_x))q&9XkHCR;$#0N)NfoN!Ft(Rzjz>^;yds)u%IGFcJAK=0 znmKoRVW4pT>%dxw>m|6iKmK5s66KzLhX=;;Sa=)!L4bQ6uvOXdX3oRlRCgk=oK5Y3Mb7Ghm!ty5W3PP3H z4s@eTHeLmUIW+4ie4~DeR*7x@Xirk6MbBXaG>0UQ1B_T6r#fIlvigI9BREE_<#t4` zW+>HZeUbR0=SF`E*sXr5|K7PkGyVGxGes#H0{0tiUDNM3NZ9yj;(I@^2*A6xS3?A= z+f=$=rQ}bpD*L@2T6^v+vzQ)7r)1y_q3d9Tsql3ltDM0|zXn8>N`mmC4v3M{=XEE8 zv%16`sy8^5yVB!rY>(UX*Ghgt)Ux*i-TG(Sp*p3V78&;bS`wq+5Ye7!N(wqPhE3hM zPRJZwnSX7_7ANpJ>+>5#tw!~q6`rPMc!pgomakX`*&y`1^3*N{LI7KoCxhHn97VW& zpwp$$U_qn9^l~twvFutce!|X?P&Ez<2J3mGSvur5)x8Wp8U`F$x1nTZ_!H1zGdDxL;~&SCuaCX6 z2D^xZH!kfp7C>~1{6=~Iv7=$Hed>+lwaOhUW}SqHiQI76AA+fi<;id3 zG>A@ZJKkNmDc|O0LW9M~i={7v1Jw{@)g>dM>Mucz;A&X&Vvy#iHsf8S*d{C9<>%m~CcK!7O~F)vDQu(nDf)0g`)2z6=g|2g7$PN(3g=$1m{@v7?C$w+o{g ze4=wdFG#tGrVDmY#e8L&{RxXA^V$F#j`MCh9?O#yO$`2#M54x?BC28n4hPxL3hwbA ze)FSA<*>Q`eJ(y_$8EOl?DC`^uoacqBOV|~#fx}r27P7xh@VMlo{d7)s+8N=h6)A) zWRO`M4$<8&1_28Otjz9BaggVoUrOZ_y7;x&@65~yUVcB3nrH1y9|cjSc~2A0je`a8 zY65G}pcuKGic?=*5fu)bCaD%5+5;0(`~B1mHOXJ+6b6HkvX_1@10U8_Lp=8wDMIo) zAW=FWT~9o0k4rCV=;h9(s!RnL*2te8cbV7H=JfAy#x1w$^Q+K@akGwj0G@_%5Ljq3 zt)p59(fwuChp6a>xwSq+E(f1xaOiW}?U*c$NnRG0n}rMZa#@8GEVGC{PMt-PMX5u* z>I};mGLc_L;mGl^DVU4-o@5cur*Sx1(O`F(N>Rc(MXs^`X#%gx5)3Z2&Q4_G-eSl0 zDi-*7h?QII$){)>C{Sr!q_ix=pK>*k$EZcfbDXzdt1-HQbzd!58c*Tue}QnZ6a95u zJJ%y9LzaYz(`G6KA&Q*A#BD-Rk%bqlHEzpb^44_$?#Cq`3f}EIr~(dxh&e6~QaXgp$uW1ezAV;!nISb>r-;?rh(PiY z_l2~sSbXW%7WYx9RDMt0`3)u}?(6Om&`4&BdEMSpY{rVl62&QbHeZLMT>1x1<8R_j zH}{-L=aZfcBct9hUS!}Tojo>C$Y~CR2>}P!c3aC~v}<0qqtU3x*!Hkfqd>k=BcJ{& zsi4Ny*}LeO{x3h|_oCfKzkVP0qEJ+1r~EWLLDy&AWA3+@ed zw(39vR$VWyF3Pm>I}Bw7!IP-LQ*@rcbeU^!{3@k|Qr(f&&HmwQ3%IT4kjT>Ge@s7O1P^bk&4yR4}>PK?VyT4j`<(L}&fx|9@9t zY!Exo(-Rg$=knvjM%@xl{pgw~9=b#8wlW~ZsJdNx zL|MZ=5XZ*f-Hsm_XPL*A5kh#fHu6q1%v4qt*l$Y9`G12&Rnq;k2s zYF?X&mTL(cI#LhPHuloWTDv1|$P)`K-)-0qAd-Yk(zfraX6a<|45 zp9-{S4e&!4vK6W@e1xwPT~*hfFOSkvKYx~mnIrPH6cB$vMnL5=>D~*sDnK3!cPVOM z``?7;$MdGdQX{W8dqGBI*5!Tzl7CdYsZt+DZrZ5)EAhm-|7Y5YYsc{Ij7UID z<$LBsIZk#T$Q6Otr;Cq#jkGLVPLh``@Gmlt2{%(a#bMB!<$e(GTpphAn1_WI0};z< zQ)cZ4Jkh^fIK%dM^kw=m42G5AAR&OWyb3YDN~Vt{MmF<^n?%A37$cwQF>;k-=|w`t zUMdD1TU)A9iRO!R(Rj4RXymTKHWOnwz5?@^xr=luM~@)jx!8y-R?H^Ji*jm&4oTL( zYe2-{Ccg?thI!1=qpK`$w}++=>AMhXZ{St{{|Wik-B zAs%SU{dS*Cf%7(?OAZ(6vSw%Z_iuJWwV!dLRL;&brn(8M-+=}%-p)j4&0UpXT~24H z&Zvo|46wvx%nGWe;J#`-`O=AkLJuPIJNcYePv%C@&A>1{FPk69{#0JOhqd6C!J6{Pa zrQUm!lOWY5ab6a)TA~n`G>P}GR5^WJ^e>3x#zOjD>Z+c=l9qBcCY6idH=q=L1sp56 z@mXBva_mP*JEDf=%1*g?+NOuZJ!2wz0`Jnjv*KfK=X*uQi?5pA;k*5t$U_`d3mKWs z*tbiMx9%W48ch47WA$*=BVZ_6FkeNE?pvMMM=uOMa9lP%oy0e@#+c%W$kYT zlmTu)u;=h{k0prEJiw9rnw#3kmeh>T(&J|26Q#=dIE*qLCy37sfmg=r5qa073)5xs z&}O}oL+i%~U3RNK^=oVFSiroFPUxt6g;ni?5M$-%mJ}Y`{R#K>%IxnV$U^sw)Lkis zTzl#6ikp!-RQ}4sJ7l)+H{;jNnD~Vv+}{M%+(VExY{iXR|KERyK!-`DcT_d_q%jl- zh#_j1H_xmrs^b-Z#m4AuZ(K+4vY4l7R*9Zh?uK^Q#Tcp!p^CJA082cy2ItO1M~@hmjtJt zf;MK!UJX44VF(Rx>jg}u?;*J)z>>sTo>voMj0vxd7NO|Dx*3&Of5`Ba#-`268Cy(K z()Zbx9xcQ>QmY9W{TR`VKJRpQIO28avKQB`AkCS_wqfn|*Cb$ zy_(KtiLf^p!MyfdWLm3*Ge|z(L-l%@eq+L#CGWSUEJ2^SUYUk(=Tj{ldL(C0;YV6# zJ-Q<i|M(c`AIr(9sSt(mD}n_4w`VC zJ(ZR^oVoRcpQ4eHnPOOX_Pv~~d@^Y?#R0neDe~Ja>RVdD>(y>y=p7Cw$)1`Yxn8-$ z85^cyS;yzKSZfNvt7G4)8z`NIYfwXFwL3KisszaiyO9rf&ks88;F%h!`9>gq|7kiW zO_OM(Q#DW_-I=a{j|M!hcwD#@OB8nI>3hszlo?D>XhE<=PNGCa%V+bVzpsW=Kk!6) zqS9MSCAf+~mZrN1mESaneryoH+<@*qqjB$_2iN{ZAiz@D;9C?#!AMdmNsXXjtFtiO!~L1{D{*zyQWiR$gdTP zeo+;H!MzI_k}aF;_2-dXO#@CnXxH3?Gj*GZuAjav+`nQ!nD2)M3^I395?yWlVx8#mY)o_(kcj8q}Z+vCX^J|HrmtNk3~ryj0=wO<#0K))cGS z^v$}O2zocuj~1T^lG!3Chipb=A?=di`GdYTTZ%p*L5insX*?6uZyaa)e>dFxDFMom zZ=pVG-=Gm1i7hl4>W}!4lKG%HqePTrLdCqL3D}}Waj~Ny3$w5W*SIJFHzgtF=wv7C z*>xPg>D?1XQcrq2_l}Qh){)eO{BOt_Z-G_`!`)r0a(ZhWwK?P)Uq3l&bsT(~lP5*r1mR zo~Qh<@KJpTmd?CNm+@Ovbm?&A|I?rzOsTC0$XsYt2x_1XtIsK2@>M$^N?@0nlLcqd z`3=AM*BoiExOkd7xy(CVR2{{mA2V>_6KsTPVC80DJx z-%15%f^OJZH?KM)vCSMEw&=hC6kWx_@XOO*%q>B~Htvc3#HqEm!&KA+HS^1=19-Dj zO33EC$kDsm-kPh#B!0xNICw`c2?^Ud)#zt|6yge^{ESUj5|_Kcdj`?}{CUKohwTe4 zO`f7oO2SOS{Z~+P_$nIj_LBN4^$(O(d(hbs+0)2BRCOARSS>gK9<$P<)*=MU^&t?{ zEcS6ZUb|et(f9_JSf8v*cze>ZkKmJ%Umw1C%Y0q;x(Jyvwy)2Pubq`)`M+HT$D7tB z`)etS@#&o#GP57HwTfit8TB-myS^nO9#zlJuWdsb%Ew`y=lih(cU3lgRiITVy>i@m z=;|2vkUQL=+!893Uhq*#u%lpg9#or%bZsp@{@DYzx>7V`P!XN=lw~>7Wu%>U`v{nk)#Q1(Z8b}x3O7(kKyQX7M`;rClFOgFVXet%wj46gXF*YrDs zqUOXBsHKK!!Eg2u+V+?UBMcw!j?Z*FDZXuq`kA3D&NWt)6-I5_oMIT-1|8@M1UXj| zt;z^Ip=-?QPbAS1mm8yRCCDXInBh4*9>Dre#}j)K`d0?*{7eiB zNzXxIEz6~{m`Q^g4TP5)^L*n7pjA?k`dmu&&sD?#@e2f-_AdN_yZr>ypgjiPRJh=~ zG?j`~^fHdoiNHj8(i#29C^(Xh4V<{8t!_+CxZ%1bLP&9IJcp2GUHx$PZ`moM7pZjy zzaua8ip@a!Asa7!9h%vgr>Fmch#?YI|lwB1+&g}G;qM@0`5iY>mr!NZuK)=KU9Ngf~^J&AG>d@ktTP{k$} zlboUwc76!bp_rE#dT!IJ?51Vh*4ktTukzk8dK5EkevZi%Z0hmx0oB?e%#4!qpY!3a zw+)^_$Di&HE>UKfJA20O+K#TQ&4_+zJP$sJ^(};*MpV`)7Vd<0cD{wB?KSFc{m7dM zPAX#(Hejg8TrpgmoTj_rl|ojQMXD3!WBd>AoaKnu0Ewu>YgFC9w%76?cgQB1FDYZyg1y=mz8Eg~Vo79RVK)m+u zxk|^Ut(LQXgHvJ)GX$fzM>m9IB+huO;L}=J`B_X^of}KPFaXQm5$F0|fGI7(X*W*S zpkq3nZFcO5gQ*`!XDpPi8<8*b|88YXLs`EMY&-pZe*=EU3=`Z-@Uf>_9vNZ@f>jf4 z!#w~GM1dGt(4CjzB!}t;4I1K*xhjA}&P0fmaMWBk3WK$JPVRv=DKvEcMEb^a^}-*_ zr~#)>{Kj6^YcyNr-Z$Ya@zg+9kYwT}?jg|&N1uZ)2Yi);^=imwI?Q3RPti<#(e}OD z&rnLz4ojB>*h!<|M42rN&#>*ve4A=^tl`N^%9z%NJt*-gmB8W>H7KDYUXIQ`229A~RI?;rR$mPEU^_BL|bL@1Y_P+WgRe)(F z$jb^%QlY+vx_{#V@vsyFf1qTooD>fbAehID1nsE6gadkT&00r5+PAwEn~!yBc|Md8f)Gn*_?n%arT zZELj`YE4j&CJlYC7SA?v6b*dwE7?H$VOfS~tu|Lq0%ZO=gc0eLW5(5nuO_az)uM1w z+BFnnPwx&W@sih%3y<=BH%@BZaG}~6F1R~awCq^s=~kvYv3cSdm(jW79G>dSvi+K? zu{w=pgbQ?#8b$JLY?YhjVGb`=9;Y~=x!YGhe$C|D-S)~%{@;XHu|qi(Ueu1l-DW

Jh~qTVPWCP^_m2N7Ra@?X9Uz9g!epP6Mr8mwX&=iJ;m1i;mJ^ z!bdMuHz)L0gyJJUkzrih_3#C6fY!^9qP0=v9nX?ohEx#}q&82w`a<;$8q;rLY7yZ@ zG7JahN_yg)IoTF1{^_gdap|CT^I8`g0krsbG9FH3Wetd%}ktb7QnBHm~ z4262AWXr_1Ca-048X95TkBCpBc2N)^4=Kt(nOYd$eMbZq$HLR0AMkU=#4; zK|g`Ip9+`Q>8$NKr>=jhA(;o_*0;VP*^YKz{cE8CzH=PpG`hU6xkA0S%f|J*V)%iy zpRbmi+ik`(^J?X1`o-<7Sq9Vc_~YrNa`62KA&2f!&kCX1yX{8T5KC|mNCxQ6{#G=o z<&Nm*a9uFX&#Mk=+eekzOUC(*_VCr_WX&_yVyXIfRrj~ecB)w=V_Ia-o^g5{`;>Xl z?_yU00tf4?z3Jd$uj+>K@yIaZ3lx`GA~9G*!UlZ9;3|@R$>87c@^W8?8^pu~^saWe3&jBN7iHh$6E*1W& zwHI)sOh}t6z2yOI$FtA*!ptvP=>tVHX-qUEFA#~%e9J)D_E7TNI~8O9e7R7pNd&TGEz5M<{bCmb_iAPaDTHWu+Im6+0sVp38xyvU>xAZyAO%kDbx^zRPJJ)~-ZQ`Qypk_&{NyE2TjDBIK?~M}Fv?LPEZKcOI3HWgd`bgciq6jM_zA!A_Uh*~Ee4|Z! zup3M>GtmcnI4}nVS2=A?EmhkJlcJi}|9bV)>VdoO%jo?Shvzw)8h@B#!i6m-MW^_P!>cVb=xNs?)(6XOJUmX);B= zY7`@m*pV0+_hcM-UQSy3YIQ_r)9;W_eVn!ys*)(uPE9!HD!_vVn`vZr4FiyGtMCdZx+7vXsLOUtrgKux=L~ zUVtK!-1|nGC`nlvZ+^j_78!QrWWVe4#*52*q5vvhiq5s%SU9?{SBUQ_cciK(hTP9u zTdUc-T0|x=5uQ;w8G`Vcq~98w>u`K>l+pCg_dKKCKYuu;aW#K$X|v(hp0t))EyOFww^l-ladXnHOrDI>nxhetfc+v zN~XbSj&<%`Z@DzuNI_I8P3i8bZ=5j_Z7lP>r%c2_;%X6@a}y*>b~c{Iu$-?T1as$d zGs_3WYpqi==%o4VQsDL^^i}I|7XCJC6L7x*g-62IEeB zw%~LB@hnfQyHE9i8UkhZa&$rZw))&?6m^@HdR)a2cCgaCp_Ox=S*u`G%~8u=W~vjd z??=|fWn5%B{p$0kn-HFm`F{;8p?2X^b67~{>eAos9Hdw{;$mPF7D|A{eKWEBgauN; zq#b#hqXikWU{W=?$;jiK(zI?&_%)XCaWlt-GJ81e*C;f;5$ntOx+@xyl<9Wjfv`2C z+?=f@AnA=9;x_pGaB#4fajM>6atngS&V#l+Hn1VAy1MU`#aE8%T@OuR(pKY{pkowf z`2Pxu^<#$E+|_IUy-WAn3dz1n&Sr-K(VN^P{@W2s}BM;1qS(Noie|E<`Y6r#L%2yi! zou@%s%z6{}u?$H%%birrZJ@<%x>9X(paydB)x#n8&xD@LYR5OCJlRUO;~_21zQk}{ zQCf-@Pu@8lZ|6wA+(3hPxb-9=m$kEO>x!DiU8JzDckD9>*3foXquD5z%XusC4gVwX zRzk?#f@2$|C#+@nPVLOvnFXZdsMlVQhj$*bm)ctIQ7qJZ`Ne8$M|4R#B?p2jwE1`z zH#J3X<~Ak{d^t$>UUFq{eB4cgMA@)!`Fd6ZEQZhoG$R3=Q8!T9UFRC;k%L6~5Y z?Y0JQInb-VyMla6+x$xW7d?Y!%2pg#EcZ6y_W};-s6D&tVeU^$tf|l@NWK-lUMK`Q zMs!iNOfMuQH`u)3=XWfE(ndJfF6%Yu`1OydKx+ujaN;4Qd7^*v4>EsWFq=#IBdEk1 zBM}PzADQmp>fyBqb)Rc*a4lXeLMh@OM?IYD5?h%Vei1(xjoPf_Gb1G=;-s(8k4K0t z&ZYGM+0?&mmCv|`wz;>N%^l>1lx>$lc)%04*d0sxWhu(q!pC>Qz{k*h~)ofvcZ){iF%XwFU$ zLgGjaC6rRRiu7mfk-5WsyFo3w*12{XH2O$~|6_uej=XJzN-Iwtu6OQc-$tJZ(d#|k zoAJ|Lzy^w63Qw+EQ5G76J6Es1AWzbcz!wx853GQ}5T=XmY}o zX6NSg>hVz**BOIoT}FZTxJP!Pw4QGT;pYdpf5zDv4?ITIL4sfJ++-ZBPNb@vOU+aS&>IIzzghk*1PjAOc96K&>JGs z2@C6N?1i$BNk;EFSH?*}Mh!yD4pJ4-YbQyau_*RB7Zhw^DYp)^VilyY+QF@DE@DeExq+-E$DYc~Xqma{ z^`JM>S2E#7leI;oa^61iwdy&0XIQx-*pTAn#B9oM8WKt2i1d2LeX<}PK?+3XcFX3f zWMhcIYSnp6ujh%vTK}4oKZo)n0`<<$WMnMHD#RIG{!f>xYA|D%-B6^>=y{CoQ&Ih1 z31b;e4$D#S&MvT|NV{IWZjr3PVa zXKS1#`D}6_r%H$L_ZK3sZa(5kavD%$u12${#HOBauHn2D>6w4&$raOsm9lz5p+$S^ zx}fH*Q!jUBqL@u;K=L98!h7Ut)H_rT@WM65Aa4>{iuUC*)dY|V9Bx5fx_6@`iV(7k zRW=V3{Cb2Ay^v2lrpIUXDpfviU6Lgs--xh!6anKZnh^9&eQ8u4YM*M&nSh084KGoC z62ZOr5J2X;)da*WChPrDPF?0{*rD-Yn7|2Bz5A|AR$qqe9Fsb0zgJ(P^j(_{*djX1 zq0FB{_o>0;=1B2}F zDr|&%={J&YY6#G1{&iq0-lm+=r~TdSrwpBm0Zt@P-!JY{7B#^qu>|kB-agk6KxLuO zs`475X*LW4l*;eA#c@ofh(3@Iyvfa?Ug}?((jhSX%vWCN$UHr;4QidYOKpKrW4$*61*p{AJ<$k34+u zcZHK>!^#xK>6A!T(~HHMbRU#N73>YzMXVF_b#Ce3zHai<7_Tnf zWvD4c7$X-0%Mf~$Thu%3b9$XQZ&^yNIt7S~R#Ft-c<=)pxdL)*`(9K&r ztVLe=?Z^?%6t@|LkM+QAU#@cB86I-C(NZ`$3QUe)fW4`LLtk030OEI_!Ae22-~eOB zzS<_Yb$tFlUI%U|p=e2!CXzCPzLnel!c|B4I+eNVg9=xUe<@pQbuAVgNEZ=j;=)*L z%*-KKmx8&EeNC~MlvcQ?uYi@pI?!X3h0B>QSq^raupt>g7Vkz6<)a?~plRQdFTDPm z0}82IlYCc&2j&0av_6(Tl0RZq>1;*`MZJgaBg7W7Te5YfWWthGzpH^Vi6_{42rSo5 zlyyA-!)=gvgl5G<-ffD|oy;lMo=2$2-RLka;&opu%exe721k-Gk^X)h2;#Yo3QsKD z7|!xr@zD_L38SCqZecHz$4HeTimiuZPwELzX_)Oh4h4P;#BJ5l4q3S{Tm_WBUwEukq~2l(fD9mYT` z(l*q2LV*&$ETQ)0L9Lf|>PvpX#H@hskmMuoU?)+2L?kv>(cdL$TLIR13|2#71siTs z>T+DDzNyR)wzz8FdXPYaS&d-nYR(@zF}$4S%}cxW`j{?UvLBZpAeV^*O=3$*TiNuu zmRWwQNQKiU!yzD-ClDl@#Gc=r=i&it3?wZm6g*O`kz-i~&adrBWZjwimuTJgd9zl~ z3p6M)($?$tdU}lfq+Iz}VyoANPCp^RRWbK#!pZ~)XQ&1z)n?!6y)$h|1< zz|0ROG4D23RC?HIgF05)&B$ZeyBE5O`F!LQkr!TluM|D~kZ+n>vATYgCN1p7*E}e? zvcM(H;BC6KJBP!{6hN39i62*`bC4um32;aKn%BXNewsV<*^oYbgeT62buU5~d}_mu z)f`{vZ!$$g1b;lIURT-y+@d#IU;^_7KxyH+Yn6dRd2|J!Fp zO>XPYxCl?>Z+G~HJ!%)h)+JN_%}zCu2;~WU@o}Ywj^2$XW>euBA&2ikK$uG68EeG_ z#d>p+Jk+w*J-_<$))$LgQHfjpbSmEllIdoQC^zK+(Iun!bd~gRvwW|LSTG! z3-R^Z9=dwiP>kzWn}>%We4TyX!}~& zjNGFAO>-$(rLMJ{nzxB{F!ny_Dtf4X`Ba^a51>#9+MGz5rQ%uj_+c>gLnD@A&_V@g zj!^1r@NpmRRX@Q>R%`3m%{tBMR4MUz^u{0Geun(l;jL4cT|YbM5y3-cKYo}YRav^K&cyeYNUzB_IxY*d1PLib zPGScnU3I?@ED(VEE|D5y(?lA+4VmtZyh^SIdd_%1wU&?Lb9;E}kFDD8cs0ZF^f4`$ z;FLP1O?bnK3Gq0zH49rjj+rUdLZLHFV)@>RGC1@#a)iEph}5Z2mMQ~J;uzz@15NK7 z{_A|V(mh6|g?nqNIfCR8-r>??biWr)29wk+8sgub`d-JQCwJRQCbeR+e9Szh#K@Yk z1#5~(mGX|HR#F6wEeMuEk@42>3~3TZegqAXD!Zy`e%k`pEHWg~lWvD=sF=u+M6l}O z6fHI-MSl3M)XmOXvLEc^Q0$Wtl{Q)xNA9!5^QnrwY7;`5De$UxRB9els`@rXLYy=L zQPm&ZnQGNFL?_Ba2qzDkF3fqLb$9W9?2t>9e+jnt{-5)=<+GDWd~i03CNT)ZhRjaq zv8vzr6|I>55p``kHw7>_@of4Cv=>o|O*?*g{^RrUHVuUhh6)5%JamE4rw3L7dKSj8u4tUi9OZuFtpgW_fTW87w{QKYC>-c8TsMxO_E z%S)C`P#9(rr$DsTw4WxZ97q`epS?D|j?Ud{i~gJ77pj+p@dMW~klz#$NSxkKL=0mc zWD~{~1BaE9dx?F%KQosjP)O=aQM&eN^=1f}G9fZf`V8AiR|3B`KuGY4bLCfAS~l7C zUm>#9H-mh$CFjFpz^Z9bztiN=ZrL~9@}oo?*jE}<`oujNWd1%s zId&&eVON}ze&R$JY8cySQfK2Y!lNR2t~)aPuKrpVtKHsVfg6ufhH% zJ6+TuYC^@aft}*itjrAKv4Xe(ZT#0KLgUB$s6B`@h*s0(YW4zaq$=U9Yi*;PcU5dr zO8coc4h83J!j$GytB!L|4-GbHD$)gW3Z{hiFC}z@;g%HsKa+uz3pYS|4b^@pWap|Z zduMq0?J;jwQqI1=E!kMSq*XwRq!P&)KI0`;G9pR4LPu=$<#N_)p{NzeuXnnbh*#8i zyn70jlZaG>^1sEE{U%P9&OA6&Hg{#O-yvS=91x__ixymCh4`r7Ik}U**>DsMK_Gmz zvgn&HzcuoRh5g#Ecoc6-xYK!VsC5@?mc@V{(Vn!wXt$=qWB!m4svrWU9gtWs=b}r& z4(}{dIfvz2Vm;!sa;O1vTn-@YcC}?7ka7us@1f&`4&_R6qJ)nII zF*zpvgKBqsuHa{u1ut|l>a+x*$W{<>+Erk#oO+cE7WwDa^EWpO{xAxc64X-))6!bi zkL~Gjl95FW#JBy@XlTsv#J`aBvt`l2?M~@A7Q)Y)B6KDQAmoToWp;By zv(Tzh#fMs(U2r7W1s95arp)s1HXQaGaX8x7WB6h&J+S+~d%1>;n7xV-K8;Q@Tb;1m z1^DtRrWF-tK8UnqdEYS+cS3^%Lnjya?jMqLxHbQN)vIvq@l|5yy&-PS!H=(|xkEE9 z@c`NUT79OEcu)FhAmaO~H@S%fOZ$kdH#6sw^m#kJSmf!e8RDI`NFJegVOE3uO9;&m zLWdFfJSJt67l?7v{ZnFI)UkQ3 zHav!1{;e!5ORhA1R=?KdA8W3E5+4^AWAnHflsLmuYx1CvC|?gBn30?kf!z3uik-3u z;a>T|TO!*?fU9_q8Jt{QR<3$f= zk*3Qee{BY7QOewqVlQg-^X&S>?Y3PR76wIQ#|=i7m8Yr(t!8p+zmoefv!NMVx<)4! z-B7}VTDIY@MC^Jq=re|Q@5>q%O^`i%0)1GaMPQLM*jY}K`C}3F01h(I!Hmwf9;p0qjw*;YL9 zg7ZXB)S7u~P+Nn+zqW_bSX|F*`l<3L9$6S^0JRSW*Awb9YS6-B=76Rp4}Bhimp}Sj z11?3j1@Q)bYbl66cX;p5T~JIC+}ejJPyBJm9r0S+d4t+V+O-gk?$FJaM%>k8Tb_2n0ATk&!L-)1`$@SR|7D!Ta$P#;jWB!HRpOqx#SyPySt1)Jq>a=d(5fa*~gzyTMJw z1jEDIvtsLsd}*n9!ukPkpLk>2>f<`2IM^>>2rEK+;TKJWEsF|fCa-6TOyx(EVVTlt z;e~_bNKkY|hiY10mR>xE2lSh1?-fi(#z9QIN0YlY9*!ytBsnt5W4h^Mmr(QCi>9XV z#3uJ<7>s~(832r9!WJv^-_AtYvRQsu`tv11a$t&HeT;t0a_^=ze9N(ND>__Xg?2mS zbh1Pkk`hj(zVc6%arK*|enjC4bV_&cP|*QxR!5tBCZeW9T+CD?-IT(q6Q>-UNawFU zy|v;&+~l$s*)(Wy*pV|RI0+7PSI8E^xlSEwq)dcLOQhH8m8bWG3MQ#dEh4L*hB##E z_|(0?A?&n8NKWn#dGFJab3$~)?(XEQ17%e88=#QsV1)ywwrJzL8YJx*y)gs(!ikCs zp=x+dnQzvJeTzI4VSR&fpin$q@KDK^-eb?i5;Q z-;0_DGdhgC*FLA;OOMjGL$7^8`gF6F>Y&P6B$Fve?x$j_EM_;e)_anux{;asP0;i$ zNi8~)b8p{RM&||i*?GBq@8S3B^>=lL!rQb#-G1_t;U*gWHg3ZLb<-W38xMA@1@zu* z+qvA$^1on=N2ByjP0D~8`yhdG%_#O0b^h<0k1hWRz@aP@c~u;@J6%OEim>=SE5S6B zxyA<5>+K{ai(27fGR5lGA0bMr4NXL!Hv8Gq(H>g`HdKgeVpbscvqIuU?Z~6_?moMq zkYAHOju8{9eoRIcJ?9kGe9 z(n*p@EN+teEFM!#en0AiI}YTAmLk7~OGrgfl=Bu-bdV{Y>-JaK=eo}mWkh~{q?ldD{bUx$l)5ShohXmmt-->+r9)6&OVeXE z(&R>c9og>NcUEplaTSJX9>JH<3+Z?=xR_-aCArc$QxC1l;%s3k4N#w5!Zv3aq)9vj zZO)9jO}ExByLAKF_@YNI_y_{TRcPb1ybarewdmUUAa-|8F)o(gvYrsBeFw;KRPcOC zp-T;Y_R7Y9!I#F@C(}AT`gLAqy>8R*BdZ8lSYY5RG9u}6t(f~>M>;=f*~0f-*X?M) z=0TW*`n>U62Ri<=ej6=_J+AnL9$owiu?D!Uc8j;!h@*w@iw#6H~O1d7FEIF_8QzU#N%lBmTNOIgS{nk?nF|)6@ziaN+iiJtjNiCyo zY4Dt&cTLkSEm^>nWp&Bs^W^AtJK5WHXx3Qft$B7H@Xa*r=k!!lP>rRv4f5`$Z48^~dr_RCYU*YV_{%87-iVr3i6jf!AE^#DFgPBJu|N zlBynm&`{1dv$aYtTA>Aev4}~PJG6L&StMQYI$8BEs#h*g(6t|wvxKT5*UuAFUU#K1X6mOrJT+R$i_i1seo;9q3(+r_1@xT zOA+v|syX6yS}a~ao5si31i=4}LfH;kSLqcEwZGTHwJ?<-&9 z|ACVa+6`5tz+&TJWrBO5C$YLxKm}_kJzgnaJ<0s*XWcAWBju1XgE$H~)n3c)>xtsn zg+Qu?MOQJYR|sF!h8$~}+B_(*6!P>+vI;RPwQHVa&MTN^1B9W%b(Ssp=ZV%7Z zhUz_otYv$FE}~IZkd3pM7YX-uqrOuZ%X2aB2b5H@SuRN3LD7p4ua9|yOQFY#ubCa= zd!u^3tC|z`%R7w#^&tfHZ?HIU=5TKt3Gpzn_o{+eK>SDFPuC(e1_fdd)-(|HB2YLt zz!)=|cRymLQf_%Mj!D`%@>#>_{%T2oMEM9w>ri<)59N6GjIa0#)Rfn$v@ExyV?Ibs z-SKDaZN8FHKbNlU<=9FvK`hpCjlqd4OZ0Z0OvG98$~>UPTybEwAO&2k6r;i>&ZR_> z0JVwowFgu9IxiGmR$ih5W5^PdHDRoYd*+UPMG=JUq9uF@bT-A? z=ISC5Rw$3iTN|Vv`Hc{`2EVyler)MRHo@7Y93k%#-uZa?{}HTz;gg5!>!5R#u#Q&A z8679eQS2H|@7P9Mht+F~mnEOIY|$)FPjI5B2A;E$I%$Q0E*B=&gKiJX`STVFU`^sE z!Yc7NgxXmL{52caD{{oj#q!19DR^y|ZdS=6oxLjT(&`ik1rn<&po5TAL1$`;ntUDF ziKUlKTuII>nJ0)*YKj9g`b!~nrv4m~JQ*dWZ+`WZ^C~rwcu);r4?j%_I~lN@baPbs zM6hk(u4p$Z`p7A5TSM}O;{E4K3zheO=pJ1A|`V)B<;n8!Ol(Dj!;XVc3?bGaT<)hKwg`$M`6799EpH4t@>oB%=@ekiX#;*qIAXeGL7V>qa|@(P%b9Y7S8vl6PBjn)3L;p{n^_`{>Iq zcK0x+6b43i7Fc~TR#3}#cNj8JZK)MODwNMUD)D~MWFR4@4U(oD$M0QBkC&W&Z6QL5 z$nKS%%t6Hl{>hTm!1^qS3*CEqC@SdNqhqCr0mNE=C#0LmG>^Nd$!wURMXTIS&P#iO z4w1sP$xn1FXNwu|pw0ZNz(-!Ow(#|YsR>T)Y`m{WStCDK{Z%OL72=+FbfVyLi&LCT zXg5v$L<-9Dz%vr%@$3N_48d<&(nKo0lL$9E(euG7%ThTaTRPO4Qrlo;twHS9k&%Nq zks8D)Vt9BXF|K3S1D$I(>KX)p6#`gQy1gO`bYT3b-B1L$sR=8|b1c6?jhFQ;g7ehZ z5oU>*5SAPuomCBGFFNXDT9+w1kTRk7OU*cw5M1Tg6z6z90q-&++d5`7;O7^0D2~j}^f0bS znFt6A>?FopPZ-hpdP+RSppH6>e4Vd^cNvB@ImH~xv=i_onPJFE=Kf?iL=G%)ZtF+* zR@$+%vZ(jl+gI^Q?w!)Os-U&LfYNg1u4mn^vjLOOMwvN49)E){G^fYbc8N z)?pzgAhnRdGlY~E=qR^cztfiYCoPxf4&Z)HZHr&sPh7wJ(xeb+;rRWgW~H~N++N`O zYz1p{S7A^WUR*Bj(#?rHuAag}Cn?N}3;CIkK)y47Xyl>C^jMP}S}MHzBeF8>Q?vg} zVZ$jM*HH>WC(<_03d3Lzk@QtpeipYMfXxxCqH@aN?FlO-Vl7H zi84g0ro`ZXM5hyDhM1E&JTgP|lI&1w;pPg-CKzH0Q9bm_w=}phh*!85w3tNLGm*F- zey3gxe>DWFQ16)%UL?>-6+hdTSTPNz!AI^j#2T@%9>JIOH80=)O%0jjcn*2aaygt} zl0A6oJ$>Pb09E5(Ya|Y|r>-CMy||^4h^?cQ&H=So8q3Pd2=|!YsfgRWB!!lJ#=AHD z%I`;sRS+K)u(rH#vV9Vf>V|<6mpN=;-vTLx_S~doJJi#T@Nt8n@v?qh{d(*z1pUJz zs5SJZ87X3B%Huqnxzh>Hc}ZN!D9_|32I((Cj%dbyuJu(ZORGh}n@6qS7-q_>$9g)g z3r_1qlmA$rTKj}SoP1rw_G$OsKXX~Xw`U?yAOBaoM~pE zv~5S6D2z27T0Lf?aDHq*7${;`<|cXNXr8}cq!McfG^XaizM$jORE9iBE4#2^VpPDF z*Y>MMMKAgd9$~JIK{eC`W)T8#0Eb-qg}j~QUVRrxiHv0Tn0v1`<>_yp=Ztf!gafnV zKXk+##5fAu7>P<%@C>;4@y?3FNT4TnLfcCD2~yWjggkUYB2YsXD=km>%=vQ$^94~X z{_m%;*o}MV0VvKzuA-&X0F<6j#34P-e~Ml34OwGe4wb7BNU2#6T>j;hMlxYKp?wv= zAD=kf@+%M}9l1Cn@x-T}hZ?dnInUxjCuqca+-`G`UGFC1NHa?Uis%xyjR-xW-6ME= zXL!gUH(4w89*<05d1DEK4E$O+vvyQOfFe^+7)AQ_+Bi5wx~V>|-Sv&HJ|IO^Xk7gm z{y&i9e=2`g4%yn>k~rGZEYes-6&FpwjE$&L-HdXfA0ldeVozx|)D9m|gP#zJvS4Gi z`_orwyS3o$&-mo4alskfSNOvb!qD5>5U5p~MJz8UD&-kn(z$$F(p?I|J*ln6JtE|l zO1}KH0ZXf0hvj2AOnRo~z@{0^a}Q1E&I7^2Ybq?&A9#%I@QO4ZB*S&u{}pa1sNC3w zCiUxk#@NR#uDHwpeJwGA55TZOE;y*Qf7iraid!RPRU~(ra0OG8C+>`@>~BJv@3WmpH-KOODr@QWuX;A6`v4)CApQ0*x-H65TT9t znl&ao{`7(}4q(fvQFEIHDK)4$He&_6Lfh#-&TJL=~P(OZrlX?fxN1R$Y z|9?kc!5U|$Aq(!FuXH!O>`TEHkRlG`rjyoY`|XlxOWDYE0vzy78aw;9*1Y^458yyX z%sDefS0+oiQxN}{qbCY-bdO_I+rb?$T|qUU5^fJ0_Q|l-(>;SnbRiA9J28r0u*>)z z4(D+64AuZC*z{V~Um>FQQo*hwDr1OLowH-Mr_(xGs?s^o@iSRF6*_(?6G+7}1E+e! z#Lq>2c^bG9gwXb)DLkBpEaMG3RY;H4cm);mbbIv;AZd%4t0$0I9WAC0COOZI7N zy7v|%=irnv^F_pJP6ql5^K!1a_GYg*>^7LZ%*I$w>@FmHJc|MW!H(xNwTK8Q_U$If zr&IDLzC^lpWM7~hJHx(S_3xMA$FF*7Ch2fVHh=1+#adgsb&0}{m^@p{-5kD5dJ*n_ z<6gg3v93sEV0!X377Tr5+>d^_oz7pRsVrWYNc*I8^*=2~KgJvTc4NJ&~ht zN5I!mt{EsIxc`sjx{dYydu7x1s$@$_4VVIR0)NWQsy>#s~O ztJZ;R+lwPAf|nRYJnBFDx9UGuUD^9L{pNwrn1Qykr`#DDO><}`-7V7c&enLWsiNaC z!Boj)6L&4wJl1(*fFzq7r}4SBdmC#d;tI&J5M3(y*&})GI}nqv3#X!Q?j`$ z^V2gMJoN(-mB0wANbfN6k3zk)dGpLCxIAS4vMFA@8#ho}QG$_%x8)rxt_E1VC&!KkXS@fTwU(h=S9a`FhkRa ziCSmEj+*zAVV0ehJtt%R-N*r(d;uU|d(i>qn5!YbX6Y;!8PFv&2FT=<4zLKaZ@W>~ zS>}K56h;hQPtrcNcs~Ag}V2laL+9V>Dj);yo>h6Gw~z($^XG=dndZnj%~MHw7zY?jb|ALsnc z`$S&RmV}K|RYywWMw%jl?YZ8DoxESu&G=UMNvgv@KzDGB%|$Dn2z7kf$r<>G4`zx; zI;J_H#@5Hvn1#cj6r{*)yTVMv4bo#%M{z<9#8b>VUM#$4SeiBxWCCK}+B-gP%A z6=m#-!C-kFsS}sbj zZgS=Cd^~+kxVaiB&lWnNJmkmTx}#A17ob^;{ZyeN!yIl70rz2WKA_p*2Or)9Qrzrl zHEl2yl(AN<8x5uFT+nLiuVh+zDJX}2w-AAa`|2EQ4EciD?m)>hSda;`{CfK_m42nd zyJ*(tR}(baAx4l$#6*kaR%XPpG)7HfiRht}hpoNy$`o%xNVch)Hkjb>l7;!;<3lO* zf|hrEyfQrfH{(yUFl5b?nmDzZ;44^_RB=0Dj7oHuU+D?%4#5YjS+)T+i)|4T$#B&~SXqY9*C&Nz1RtO?7r(;{b7Gbe1}T-gYbs zqD^>s+%%3;;4Yk_C5>es<4neU_vXpq7@fQge z){j@1nmYGH_tbCE0;t}CLygSNh5Fpbbi!;g%TQfKZsu$)vav`om7CqA{@3xdMUy;5 z_}=yfed7G4VGzw8GW2F|1k3%ApVafjD#A2`s0JTjxjsJZ`Mrf0MA68nZ;?!}Q&}}r z;Ln1St7VO2RVJe7JPLi=#Mn&TSZZjtWb1%N(w8H}xW(pfJz484a563<7WgIct&ha+ze6&v!O$IhF;oSDTCG-NJ};)z@p znliqFi;7NbJloOjb%45$e!X(O)n*U#Ygl3lU$ni2N*>R(`Dgh<-~_$fYRZoz7mWm? zmg-M&V2&Hz0f9X!0R@B^;Zchaox#xKGYHiGN%VAAK_Mzt9CE#KqTTH7|DC6nF&}EW3WJjl zwPaG>zayeQ*QCw|7x@MPh2f1YDp_Rix>-zyhf^mwfpS4FoxBRb=40<;=0J>G4=VEF zqJA@pIosM(9MHnBIJ+ZK)5eD%jBsTxew|S(Rc;n0vJ>m+=qTb9F-vRckgSI$-^iT| zY-q6xbf2sjs~YovgLOP@avHnhtV2P5Uf754x&TEGu88O*fA!QAwbD(4Mt|_f+!yvc zjUpe9GaKOT6awQGBc!h7_kPrbiM5?9_*WT+yMDVHoi1neS@wioYKIN2P3MhvSg%Z; z@*rA?9)CYTV|nY@^U!D(r7-iBcYN*A=Xfj6ng4kB=6QHA=acZ~CD$*MdLKGGSeSsv z=c8{4Fjzvj&fWRGhX0>|jtf}HF%kEBCPSY8HGql=g$tEg*9+^)xFCYIfH z!i{_>%YNVc_kTa&2SJfiMQ`kw*{-N6uCy2uzg2W{B}$bCS+RlPkQB?-!d_6<#U~bs z(oJgdqLwbL$6l$1&PR!ow4-6Nny5D@MXVdzEqz%A<62zxmYUudcQHcqSiUn>dx@hkihYAkMMLh~Dq* z@`+j}Izpe+GBdK0g2rm<%ap!iOmVIAewb5@lHyf$hFPYLlH$MTfFOK;dx(5KQ1RJk!1d60)rds0 zaOZa`(?e^>pO?kioAgPEOB-ey81Gft_%ocZ)Wu|X)_aslz+)yQStff5L5 z8;J!4+6pSC@QYnGX=EvNIzE^fe7*3)TB}aKjzLhw9C`C5i3=K--$0I-7ms~l_^+vn z!hwRt6Z_P0sU=$?4Zkb2(U@p{0#%II$i^4$BFT7-Vk8$(Vq1FK)-;;)vil3W^t;lX z=|;l!i;s`6aOYNU9d|dMVPqElA@2obxj<7-P&V}>)Nq{qS{9><{G=det#@*M-fvbq z?#(LG62l9MGJ=7V1a@J3zWii%hcKuoCHV-e%C3|ed5hK#>6jxrOkDeN4cu~hPSZR; zIp8fkfxKxDcJnUfT_(0&;ZlTmvH~xX;e&$Hm$QtjoY8C;oHktMjQU%!t0%#q{-i@# zd9%i3`g(jTV*>>kY;%3W$(;xM6>q zn`-m3{ZVuti;V(75d9@!h(|Ia=X4|IocVg=t9LDm851Y& z-bf7vR+2t*EPU(dD$ELKjP2%Ym~c^Iq156AU9+bCVna);BnYIeM-|rd8ysFl1kUU` z4rpn4`E`Ajqgy(6&Zak0Qn!qKTwE<^9mVw1?Ks!v?EwpY_ZI=GzMdueYXe?S(U^wa z!iiOVx9lDZw@FE`DOS4ChY$DzC3ym0PCd6HZay1-Cer%<(BsS3i$GyqdmZHLnq@W9 zUKxBjKK5e*?N*^Br17k@3VYwX>DNaUk*Y6l$Y}!S*noSw`PgQOe?iRXe@sNq4BZydHT;mMbD+lc zR!H-MB<~GP?>sp>)zVdkXbdwa~FaGhr~vrtNdtbtr-^P zBos*NqyPEI{h!JA9mm5WqrzRpBcqGDHMe@mb})y(p!GW4!xqVSqi^>)u(N8%RT z%GEGJ|2(GP5|$L7w&&&$1BhiEB3F&iFIA5jS5}_ZiItEP={?-#WWNieo(bTOjZc4# zq(&ZPsu1s@(rhw3c^oHJiDK}kbnMvcPT`%sQ;R?Q#HP4$|Li#;3^XD%xuE{GntzgD zvRHUc&6A2FPweOcMU_;SOedAT8be0ytu5|ekGh)sJ2^RBT==~aAy?WkLwKh| z`%>Yg6Z)|iHV7L*(*yy1khR9H!i{65U*Do3|6@@Ejo@-#f19u}Ok*`op zp-blXE~Db#*ghS#i(d3<#spuo7T|v9*<%uBzi|FhIbi2vRC8Svk{lZ1ZGK!1sk3?~ zl}d!dyLG2>`#RBBBIiWA-^@5zT7zN|9Zc{lcul-`Hb!jhi$wu%TO&6E7~X{;h3gv_dJyFR$ajL|n?9GQB?aTjKR0y|Nof>@ho~#zf-aNRu4=?Slj= ztpJ;)LivbOZd6i{9+u3P^Nv$}7e6nXWj`&D!s9F!4;)Z&RNB%w{{ue$4fpz9g& z{a&f7ZjaC#EYof};PiOW+s;a(UvNn4^SpMbj_bmf^0%+omSdjRlGVVTJ{JvsNy4-h zwj@{K?<$SO&1)!;#Cvccb+E%D5KWTPn?a;%EowFd0ftfdY4T!Di3kk>&@H0Lm50Uc(z)A88#5#54z4fjGg;;(lg zf2}I@*8WOL9_X(=>tbk^L+#;^9A+o+v(Sj3o*!omJa!DmM8ecj)B(iIo{NP&31b}X z(GTUC=HQ_!0A!~7YO+6#*WTjqd_7rcn^?XucyrOaOuVC~TRmuqrQt41`7#Ihs1h$j zFf=kYXfCRvn`M&u3=qt-GnJ`@$gHJ_Mh-1*+zCuRXJnB=P~RAD{U66=^zz=JG&#qf z#P>C_rQqx88kS(EePqUSMYl?=fiTQlrFH{F6Srfaj>FYYHy)&hJMIR#sGS`0{L1a4 z0m}ghk6U-z0yThME*{e!9TR-%SqElBk^mQaj`CNT+DH}2j=9hJ`tSb)?&elvF_3n9 zAJ4=uIj6)rV6k*jfBMcgysMzJiDlZsxD>cH1$U5#N=g0GrsfNy*UW%8NIERAgssvf z&vrb<_nWg&EX(atbon|_=ThCrIG%~pT#S*ByS7(jDYxLgOsLWDJcL5`l03XZu6KZ! zUqM+x<=QZL*>m;z6)FA^kaZpUz8d3i&C5KuwxP*>cW&~S?@qiLB6aK zgqkqjHF}QP3U1<+uF~Q>dY>3Ud+50=-9G7@TChJ@t*Tqg2?KW>Y)S41_xMF2Ew|i(M;u+ioZ5`&G9!R*l0TIr?$e%xX96{U>(z%G`4t}4o$gFb^qND zvvw!JR0HVJm$c(tr``0L{R zWs)Rg@p3)xfc>P=�v8Mdka|XmaCb_0HB80Z|mONNO})jyDY9irYnZQp%sdqLdu$u zzFb>kbZ%=g1d@Y=>TM*uh0mn9y+^(<>~DOUVAy83U6`N!`X3RD7!-Igd%dA9AB}6K z6N-|sN+DTZ?_AlrcFmu{z0=sbI7e;pwC^dy+ZsXlN>`CQITvigqR-N7`9h=UeF?7? z+~YI89pQ5!8o6WM93RQ(@M+NB%n>6rLw(}2Cn|$5R4R4Qk~|2P);}wFjVjGB`@$it zj0;`eyb@e}o2jRfh*g1qxW4<=ZoyBJFbA$+L*qw}paU);O#gpWwr^CFf9<>4r-L^q z4bTze(&Q`9$N4xmL`$R_-_+$N>&>+!S(BtYSvI{95MQrTavzLV;ldhKVkRKJb=+Sv zgzhge<%6CUBq0Y&{+0~g84uzg1`+H*f^el_3BlP`8-?@4vGoMC=-EZ6Xpw(OP*3(# zPl$Z{94fdhZ7lcA<0RXN)UG*75k6i=X~y&#?GG3ERlu4Amz_irSV-E-Z$;UsFf++A z@!8@1Q9IsE!MnSIPd4R8|iaCs-J zSzN+sNX^G!W0X7s`aMKT&sY}@t2MY_o>gjM!!1EX%yB9 zWa~g{l$wq7c^h7NV3yI>ixX1~y7c8!!m~EWx+T=yr$!lq;WZH$^fC)Wjwf#8 z7U6jxSl>vGaX3)AvZc?l3!B{VL7Pyv-Xt>*fPlBivpdvvj31>#O_DU7^N*_DDdU~wy(BPI=q#QaV^P}#9$4pZXV zvArajZ3lfUO<73;aT#fj3uyTU;PcU%mI!-Ei_OX}T&NOmJg_4yUTIf)j>S!E9!tWR zb50?${e~Ei&mJa4Cz!5(N_>#YL?N-VNBIYcq8L)j>>l8E+eNG~WauQx0}uYwUvkz& ziHqN&uw@bj50DqW`NEFmr3vu8Z9Zz&b*aq^aW|=1zsX>}%V2VP^T{Q~noU4pG?y_Z znZ-?yn=7+M)-zz#9&;@m?<4e39oCW6zP8Tpi#2+BF13PYWg^}CJ+mG+`xrqBDQ*0x zYSH`ouzajMS}YGi2St!RL8<>uN>qy6-Y3Ww{JN&|nn*cEljs-0sH zq+(UO>jw$*CL^Xp5#&l7LoD{sspelgap)(jmzrx;vfbtJy4tr8BbzYqj7m`#MzhmG zGzD(R^d&Gmry}_MzzZRse*PIBHf9hPX~?y0+F|K$mFDb*gfikje+Vt-t%nAgCEa7d++~ghv!gbWEm8YE7!ViZ66*&$eymmDU5~ z#mTdZ$vQF%f~1H>S!>Y?*Ryfk^u;M47;Q0IWqY1_H+P=*mf$BC$j@Ki#z{Y&#t1cW zmhi?{u7+>*U0WIL#F{uG;Sovm*BeuOCz40YAZfS4@e)&7!feKmS-pymmjsvm%s=Q2 z2?El)J(4OH@WHLC6K686)0WRPPH~X$HLYALbH__?1aUMiX(LNxSAmOGXD9-}Z;4!&Oyq&n>3@SH1#! zCl@!8ze{jmVblxu0K#RRzjY?_=ry1zm1I}ZcYWW8HoIY*VFOiY$s7|UnG#A(VD_4o z^N=nK;BfTmwnz3pY!!zq-t*yv%K78gKF(hfG`?^b7AhZNr>;SyG_=ri&A)w7$@(`) zuU^PwqglzL9OyEnGO%pG3Q`uAD))Y?3$rjI;95~BSGcUF)-!7#P~LD+5rId7sYtt4 z2AV5ccFeo?8W0f^Z=v}zF$9rod96R5S3<@L$dF9?DO$abxX+F@(5us{q6;%E3*P*e z*sHI9s=3u{9n&e~J|&7xNp~whA%%|??zNhO214=ziS(5m4;s}a(y*4Q1q*1mTEyrG zlRLqVN!oq_`{kCb8LwKnwUn}%S)1;r7ThXx6c3RA9 zxu0U8T2gFcgT?KzgK}X%3nnm5*i|k`67a$<&2M%6B9_8J8LJ{YDLfh7pgGKNaIL10 z*7X7wlPvK8SC3z!sUJ@%qL$mePPesSX0%D{jx{N);Zk~|buVZx!YQ7yuXs>eJQ_Qf zUBBYzJRG4sgKO(Jxs;EZq;if%p&@2(s^?}D4_^g8bFc@)c?oilc|OwEbD}rc<`BD$ za@BY{)AhruMF3T|K`m$+mkISS#K`R*+pc|~s^yjn-Eaz6D)U;I@TVPqIq6q;zW5vx zN(57;lB%e}*_$Cj>EI^>Kfpo~8=1TTy1PTd431G~I zKQFnXtXpG{LhPbC$_OJY`>UT}$+|?Dy9Xk4Ia67RE~@G4qg-FC@kD9$6@;(FTmMp1 zoIXa_fu6IhSa{PB0N9`5pFu_+shRHRK4cxuh39G)n(o=%)??7>`ipC9rAgACgq!eW zQGHqOMenFCFeGg@oJ-FZwELENX5RiY=3Y>QK4HIk|)!!O*gIeWT<^DGeeYfYewe{{dI_3J%^$ z$z(Wt8Kba49e^Xy18xG=9%J5{W_i)GV0sMJRp2#UkujTLMY$@HWmng>D0KNpV0=yv zF^?5!(GB__wp>uc6^m30z8_zx{t-9>TQAj8skPsXMqH+;LUv_@T^2T?bJ7+`DACrf zLXD7*L;kAHVg0zcd}K5ukpCZAorlxNhG6pq#e9>ny>GLS?_6$U6Pjci6G8*ZfSVw2 ztizA`BwJ@NB8HTo-BNHtwF*qRjsdBXBSl8F+n==A{_qoA-}B*&KO@4N>Z2j_^upXb zTH6j%#NHP<4PPf@6pz}((^%XylE>%+XyGpm-sJz0k8WXgFYqm2-H3dy7P2oBc>Ma< zz1^F9Sp@i#4LvSa_l`VHo+u*}_U3Wy6zoU?S*`B;GaYA7Jda7=YO0z8z6A;cWjL>V zY^1Ec8tICowHmO-<#Kl-WNqw^)FIj>?qdAw;MP!xzi`%XnGYU2mpZba5fBe*V(F(+ zjg}DcHLi8j2X2glDOi{e%Yj@mq9_N9B6cGyUla`T?Kd5a{yz)ot+Zq-i}fvjxHb0- z=9bCl7Kfl^Kk4)Dudd2@DM397m@kI(a>&+EFJY){`g2mt{N;U9*|ESYHR+y{^nZ}0 zF$L6+d~SYFg8Wes-2AvFxFQ8AI}*QCCe-|!_GeHx{0KVYY;QO!{lDh*)FiwlGks8 z2E|2YFw*i3WfEl53>u+-e>`kA6VGJlCUR#pR`tV$cUk6?U-Q{ZwD5|MXO@YnRnp&P*aK2zDvt30QuPK`4=Cb+RpeVk@KpTA$}<LG zMny_DnIL>9F`(qDg6%Qh5g10hCXt%IeN=|wcZ5VZHzfCh&F3gw%+j-Dk|7htFKvFE z+WoE^-ranPRK=aa%-@|NFk`R z&xozyc-BgZi}P35@Jj6%ltDW66DUfkTLT>cLQ?}NTM)J+ z1FF0aOcHx~$Ob;$j^jE}YBZFD2#+clwDDSER&G$IE~Hmcg5USBRPPvV={@cQQ3-&9 zjZ9HqUB)FTFb>>^ncO9h7Kq|rc~E$7kVE{J9lpEvAOLb-N;yQqO2Aa9aKctuGpqob z!Ilb1$4@ke7xy8*sUgQZ{JVl$ zJV&=VNZS*x1zUJJ7(CJqst!7p1Qr9XWV&YYUweGm57~6&_O_k&u$F@Ll&arQ6fI-V zcwcC-@ym7o6=}-Tin~8Tm=&ci!e|PlgA~`?(b=zgw2_n=bG3Z*AKtH39*l23)#u^| zK!GR`eHe&^AfN__tdisT_?+84 zB#DKXPA+t;*@!%ZozT551&%de66?7ev3yh}WC?;$eFCV)8BoQBCxm=I67lz7Fk}X7 z4=hwYnv!c!vUa_tbwTHl>h|0rg zEGTQcRvS|b3qZWD8K|w-6Z7H!yJ|E}(oH>5B;WkLk_t2+Lm>ry| z-Msw~Ae!7TgpS2*=vrF+lP1PS`XJBw_+5z%Rji?E%{L6>6pj^l_`m6sa}FNvcI9bQ zEa4ciA1K&LV|*9yS@VZghavr+o^H5irwy#Aw%kqfy3yy6BG{?Zxku(0`q5EE?PWDP zPO0EdKSFiFi897L@l9yAE_$_f71J&!P#^*^DTBj+O$^wwF8u^YZ}jkpyz${d`?;m@ z4+wDaxSnpD%9y@150tRL{r%l%4fi_O+VB?{<^7bo-r?;yYX4E36jTn`Y}6Mn-niQ8 zvAjoBUH8KxG$!oUQcGVB*15C_`AnIA)yTr|D{ZLjYIZBTOJ%=BcfoOZvyVUx_QTgz z)HFChD&y@~tPnGpB{jI51v*x)oxSGWdQf1d$Uc)s!MVG*DMtw%A?A?#RdpB(wTxmZ@TUntN^@1f0D z5DBu2+FiYV@1_v6DEs73`IhsTDQL<$n9?6_J$q@!7Hmzp+Y_DTDQZAW_f~DP0dQrl z0snM-M)lyiu-R9Ziqc(@V z^4jwKL}f2oe+xR}^q4MmTDuXa(fJp=T8iNPw>Z80!|mG+=ge1`SZ0LO#iv}Q?bsF% zzYhBGNH?Wv*3)-&)W>K`8q((;~7EpwOuX08U%O`s( z{>NAi{cmh(xPK`SBV<&&oo(-BPrQ#WG?e8_U?;awpH56|S&qPPQRNfLz}5*?n$I0Z zDRlWXj=XQdjr{#do zc4lJc<43`Ie_vIII!X{rMF8xYkdG>cyi4|Kn zcID(bYV7d9$A^lt3S2#Ko|9f1zPl1~Iyg?KeKV_TB82Iqc;N8=VI3(NWw!_3-gIlB z+kBdC?ACCf%{Cp0dB{V~I#=9CMP%G+a}qYwt;6{ZXEFPY?iY9Um7d{j^vajekBMP8>?q(CF zBy86sy_JWVd#bvXdQP{%Ze*astW8qZbny2*TB?m^GHFTYG4 znl8$}gBd(rL|d4^6#5xF^)TZ+zo>Nq6EQleR;^C^Oz4)MJ~V6=CKuNmrFPGhNH8QX zf_IOuK2sHq)xRh=WVy6rY|n5sBququlkBtUw5~mpltb?L3}JtYtj&DOA&E6TMQ+BV zS|+=>p+)L?YOQ#V$Fau@hgtPmk}P#g16cV&;yqcfdS|B@OfYaa zx0_nTXJ0-q5pr$Ajye3YB|jzZUmNPAVAFBT0`f0>Cb*$x2md6=o4F64QT{04P6)I6 zi`~;~>mO<(e0#OA@`DF)gG+H#qGgtIm6(~}{}gta@kkN>5ZU93&$kf^=lt0v(ALFH z&6uV)MmL=;TxfecU}5xV%eGT0#wYQxYj>E`gOV_W2r$2#FTU6pAzI9YpOPg}&%E4O ze#tGT9!>Uj3d=tnmL_7W5aKducn&`{{BdeWf>~ZZT~W>w%qF_Qr z@yxg;PcRglHP7Z{G$yNRgo=)>xh^-tuB>_ ztrQZg%*gQPYMf~xTx>nbbac7Wq9UM*CSbCJf3#4hm(X|(by2Qp5F0Uk-Su8Y$wo1R zaN?ioZaW95Za~fXC{$}*L+b&+i6YqUAkEQy!6=XtT4F@%S$yZ$ge%3f@B-W-J94r5 zk=@>P0(7dDYd5|v*I8n@6TNITY{uBJHy1}a9Fgd%vrt^|Bs{Vdl$`8K&`8d%rvch4 z%q7U!R#B)zr$+Zq*ONWZmN!A}1khM=vn&aCw{6#_aIrHI{!cX1U;|4(L^SXs_qu7s6@U>coRV80|vo0SSh znpn%11EYPvZSUFNoo-d}llZ+*&rxsA^M2EQez`-O%E6VI@yl|dYVcoH9{W;+c3Afu z4fTJGc;I$wB-FgQ&PvQ^S9`O_(Qen>o0SlP@qr1UAqNFlZrSEgGkDl@c^r8x`clE6 zFY|XMlT&?#0*94R{iXPWqPC`6nqirelV>|WeMm_sW>UFHTc~;6_Sq$;~lSgl%jfNT*X`WArDld++THby{8Jv;?~e)0X9r(&M41^tmum= zU@sNxTjGCNWpftc>4TV?WEabPND$6jRgI-m!%&pYrkMMI`E0%Z3V9B9s^ez&o=QQ5 zB1O@4-fH~;hfjI4L4;GUxy${2?fPF%I{i7b%C)Adc=E3i+kG z;?eo;m{CzN$(sb9rQh%!s}~`ia1tiAf_#Q3E?{6jHN0vFF>A@d;S{rY^G70W4vgy@ z3bDI?wgab-qijFz|Gdezk(l)+k_;K1A27Qjv!gMag(#?G)bbw{(V44=CeY0N&n$?r zw5vuCW090nP8>}3_dIIl=@;hEu{BDrTa7%7{G?RLn`_qFKqgq)Vy$2_AkO&8dKxga*o6=B8Yl&!>!MJ7{ zp?+V>M?h*;a;u3s&sdN~)h60a z5v5KHhXpd>tv}=nHMXwcX-ihxh*fD$%u8P!zp?1|mmMc0wY9W1=YW+2U2bBSGNy^w za7qV+BhH;qCDQG2L1BT8_UKw}4_S)iwr4GJ!fBiaTD<*_mK*I8m+j2%$_U8{Y3J62 z{b}qp*dL2lbU>=4sN)-nE2HKyk}M`RXDrPVj{oOms^Z6fI=OSG1)llDxB9fcoro)! z-*yt$P}GA`O(uclAit;=k#stW{a5i=^cH^e!<|fJ%^Y4ZwiJlc;n|REro^~?eR7&k z4+Pzqn45Ug4)2PpeTgKNihT4;h3EKs*%bPWpC7MdJ@)hlqSV@5S~`%ECEjQv68^I`);`|C zoQQrRcCmyZ@vO+?lC7KEUzcw+DU zd~+J-L){pB21GTXm&wqZPM>!~S7IIso%f&Ex${?Tv%+A292LNBl3Y5AnqN0_i|(3^ z*WPBEwfeL|o=VK6TyDR)G5@&mFxoDtGLW0iorX``#|iQR#Y&QTn4(-i+w<(nQ$D1b zMmAyz3k9OXFF=xIu=;+;!z)9)`AGkY4{^04*UAkX~xZ-;rkj~nD~ z#l9@(6Q-5jVE)1PFwo9`T~&4LzqPeh5KZEW@A~n=+{`jAx>$E- zHm*vb+@`Sj1Z97*WL=G~49D(oW4Wip!K9xdFXHG*-DoC?&Z{16(&!i1L4?RP!d?B^ zBsjwP!4PxiF4XCi4SwXQH9jr{hM|d!Eek?D7wMb5kFn^{P_tn7ylKrBL`Ai3L#WBJWwtx?7D>^i& z)g!MimD!KX2lD3%S74H@8loymn~4YG*S_7rcNE_|mW9lDjpg89qe8a{63i}`4j5j`z7$428{NYysbD{z}7JN~$T|pNOe>y-cKBK33&Uz-~{_p~0e*WuZB9rl>?odW& zM2jadqVQQ)7wbekHRXn-;F=>66<^$s{Wc}0zF}|YkurVl1k}R+kDNrKiXB@oN|bvj zGrD9QvZffEPsPPs7K-GzZ#N#I6pEOsALbTnXI zu`cU!v!@kF8!5xZZQ|f8`UPdz>%hIipOG11@Im1H=ItrvYzq;BW>#X{SyZN>z_gpS zvyMS~yYn~5o)}m^(KS%N&|*9V#SfRMq%@2LlaR=o9SQZd)JGW0lK#-NFE{$Kp^0?z zZTw`&vn=!|+>60qW|5=(N=v|Xsru>g2Dk6F+fM)N;it=<8c!>02*mfMk-8%Z)?dOe zZ#AghITDpm#arV@TYcfJS=rr9@-O`wVdGaK?eoi#z%AYdZ5ny=7to~@?D}b?w8&l`MfLF=$1Mt zM>V~4<{p-}Hmw8;>120#Z{e93>nR&IxuTF1%(S@8WfjhMLCs`Lu!)*lO+_F7W}Nqw zA5BKk+*&Rld|m5C`5c!LTyry=lx-VQs}V1owcTOQYFmP^mWEbI$}&Q=t&bh}?{sz! zf&b`ooO}~~5Z_b=11}Ki95d~W-qMf=dIbB?p)SKG6hS=-AzLfNQwK@5r^{-=B(^(r zFef&pHoK>Ij2X@V1&1h#;WMWHv$ll1gijgB!Lcd!sY#y#6Gqq!aG*w{Zg>Jdr;5>V725?u6I9 z1^7xm}othn)u`bkA#xGi8gCKh9?Q77or|rXbX5-MQB(V z&7hu495Ei8Ry}i1yTp;Tg7bV9EJfevMncFgCzS+M7UI3|kP4vPtdm;}bO*am^(Hmb z|EQomTe_y6_oVjvt!OsKT|{p>tAu@?<#CL8J~XpN+bwGai~Q{vpVF_}=gU7=;VHU! z%*18r|FLI3@cg@M^g#ES$&WH=UZ}(f9-=eBDtxH;W_%)sDlJh9nUv}K!Be|;Pjjff(l2* zJP6bC`&D+hbCFm|e8jSmUw#mKNjAw*@L?m$|J)Ci#D5%YZ;tf=#(;O#b|2Ypdstz> zLg(=P+i8n|nvj+BdaHkup`HM(;9`{a42s+=~m*e-mr!LLnLY_U4)QPAjJqg-i@ z$M4`L*cyhK3<_P46vxJ!ss(7)mKuw&IQR8Xrg1k?w!CDP8P=@(+)(M^xp6S@YD)OZ zbg&f~OTSf%)_gR`+uY)T(mAZs4-Zj$qNLxUUZ|G`W#Xe;^kaVelMnx2C%5$UwIKeh zqsjR!SE<~6%W*gx+Uqz^8R-6iOT??pMEe*6AU*K450;G6`Z zJOe(c-^DIpZ-t!}Mvh!G80VJ=MBcdYMxCV^C4@oTD}^2MejG1t6UwVdJiA)U&IuTL zGvMiiglypnNKcv{iYtFvsdPK`K+{!do4vLa3M{8{?r zNy*W?^-XK^1-#_{CiURG#-UC9xT4kp<%tS*6k25WF(RHcJ8*<{N{0Q(s3I}~O8(m0 zlD6N#zeL<(4+`Z-g3dYcg3`k}w-9udAd)+j^=|7zioNH9 zA{XNm@1swn8TqFT;7!KH1)ReWQH4H0bvvL6@r5#))jgcpUFM|Qbykr236xx4LHi)} zDI09s&6$6nV@j zZ`AnyWE8(KTR>C?U*J_uJjN|z%W_yiAZ;>BpiN*>WHi9^7ZIWmX{%Z~a!Kb353c*eNUhd>)h@uv-DNTsHfcL~<2A3kY8cEw1Q!>HBG5@Q{z?=;fUk4<7_Kh=hEI!vUl$R%sY zPtH;wmPVE#=4AL zrYTqymAq_K9x8J%mO3~m!1()eXxmbpVhRP-9q{=VA}FVj{#jJKM>)%)G&DYu#VV?d z81|}wVtoCv21Ml)D`F8lg{Eaeh=Y(g>qogqDfcevuPWD&$?+^za)HEf%-r>w#C5!p zVdc)Ub5N0_dM~5#dq&r)Ecbh3?xHFTg&xqm2M=j&LAecdjOFIO!s7e+4UMa+oZr`n z?}#uvL?8(_F4on9XUHh^IPdYw~_Tgns6`+))Ns4My6?&l;P} zgSU7kDyNiBL*EZQM*<6SLfIE^8{VW3hc&_M>s|#!_8!< zk$EzMF7%*7=z>rJiL(bL+*`>Ne(@psU6bgzX<0X%OLwnQ{|a(;>Gx1RmVv}L>)R=e zX`gjdQ~VT_@8jhT*E7@#&5Om~kP$hS5n&T97upeujBn*aU9HH|8ibAGw{PWevxEF@ zvGX&J%gt_$M-dw>4q=ViI|6koJw@V;-HS3(YH&sNn-y6=G+Q9c9BJV2F;tRoRmp(I zf=ztbLYbQEocIy#h37TS%dNoI!`sG&ofI!!(|-c|p@dw$SAG&darcqHG2=$szxTll z)E#8^pf*wOV}sVCkdM!5EXY+dxDq{BTE~$*A#0v%o33F7LiPAr)r6m*1h(sS2zNd4 zGU}Pj!r%L$QJ(R(aQ~4=eqrcYSKUb`#>P@i0oM#V&rt_n_~t zq(<+S5zYo1WoW1ep|}*3VdSFK+l#EKgN_SGVf*16wwbeg(^H=`Hs3zoVW)pWJ_;`!U47Z}b3ryR>t z_O&mIib3J3ZOZ3pBSXocXQk(orO<&Ylhbffz8LjBzL;gU@YA^mWuO8MrnzHUYsLF% z4iIqjL(MTjCci&s;VAw(7t5)qRr;_};)6f(G|J+}!Hy>@+(*3MC7sc6PTHXTh1OJy z0nr$vN8~)$crcArj4m6hQ`1h^B9Q+pbewo43V&M53jd!D673y#`f~NTo*h{v>>8(8 zCXyAQ&&1q8sJ9#>UY)&;H4-~v0o0=jIrYl^16~hBcJ2-f<0D&l0H&;UUeuV+iDg$W zEp#kHURc#Daz#J1PqCTD5<5N#5LOcs?;Mpw7!SKA=K12-<+3dt_-*ZBd|N-ppOH&< za>!nZViw_p!IlxkjiGXdpyw(vWF*-1<OFTv;!Q0!r}9#LV}I$G4k)f`_}IwCDILiwtF0yXOPJHYpRNgZFn`DdwHV&aheT%H za2FFDFkC-f-qITM`x94SkF)Mn7C+<)^An*sBiPcP;IElD3AGsYOHTs$D5H=X(+vHN zjl}Kx1wW>H;j@}c(G?S+4y<9~JJC3ntfkqhjrVL#KdrfnbpIKY$)Dm+DuQo$T%TOM zHuu(bB(sf|W5JUUnfUN$Z$*_#Otm=8UN9YPR9{z~Swd_w1DP97NUuCFiB)6OYm1HL zM=v`cvh~rH3U6(Uo{B_AKDHb_)xOwK%#TnyEcU90Bo4L<02W-Acb97O?uO$>7Nf1nx8DQqG?qBh>m-ps0%|)dsQBa{tNj@2+OB!VwmiCqD95|;fl3^h zO-Kk`GG#qX!QKHpZQN5df8(y3C(N7Y62%_!sPm~2X2?@wILTKF&mc{Gam{nqZ%*%3 z%~ki1zTuIxDes?5s2enxcNuOWqCwn*;z&2u3wmh3B^En0T)S~cRSC^JQ*a&AN55*Y z7~U^K=hT(Ar!$-zh9?$;sFzs8=j~@uWd7wJYC1FBo(_r0T$MkpMjlURWn2EUxK%Oc z`o#lvyG>#?|IZ}j<@V7A1=AI1axM$gFIHiI81P#We9B$PO)N%=p1{TrNRi0c41CgX z2+MS>@6bdYKU9c?w6to9I896#Ie1;eU8)tmqt75vZWi9v72!K77PRAo=P|4!@d}Z8 zHx#!uwfls}UjeqjZAQENyu-Poa|~X-jx>=F+f9<3wH( z*C^>F9#@wM1U#P=Al3h%O8pm=A%7#a6L%D~homu8qA$6aMb)JD>#vp4b(j-WCNi|3 zr`-Rek$6G!W~svF;8yFTytuYc0W-$+(Vb_RU#M~G7t~pYBXoDj{U#T&`vJt<&CAPn zYWHKDM&Z`A00h9ev67F3x6eR#_7K)FHk;Qy&uWn8+!vzS(TdZ$(5p7`c3fN`jy2>$ zO1y?dM)icCst(3}8B{1}A@&E~>&`4YtuE(P>-mX1Vs?}UUd=b{3cC+&_L&2%ddYlt zEY=EjY!5qQS+%kRRpSAr`*y>eS!Q`I0?C5sl=iwt%0r$?E!=@Wz=~ZpHa_axNb_m~ zrxh!~qATBO-h3evYlzJY`^9Ol*_KPM+J(!AF`bWyzOv{D){)C5Z^{=IY28|`W}k`TD$jdY7{j$6f@ zI5j*a*J!3uCOEh*DP4X8Rw&tru&UO(^dNpe2Sn{4JvLY`muuL$a=y}^!YvJZh(2YY zckQ;?9hH{dhGX1bJ@x4P7?Ns^eP9yWz9k5zD%$^@TXZ`5jMPeoR4DdK7})RMvt_%{ z)Nq0QlE^)?)<~^0BNB)sN>jwwmAWxAXj>Yl$1;SdcP@f@QAq1)GiMGHh8C1qV18f8 zecX@0@Zyk@uVb%rIOCLjP3uJ=ZJ&f&s{4$Jx9i`7Siji*Tjn(f4*l&)YVj|}g6ul_ z=a*S7hTVEOf#c6CXUTrKGvNHiq< zbv|TDtrLZkYYKdC;S?Gr*IoxHWZXmCRa@>DMH}05vUJy%*ZU2zrlxJHu6#I4{x|ZN z@745f!z!v_1|muN+Zn5%$%sR}|3{#bd{frA*Fr*#v%&Nkua9CX=BD{;&}~U7ZjeF{ z#+6RC=~r0kbLh8}VYx?YUIjL;-J=R%x zKIRvCd;@-CtYN=!a)mWIi4#y#sqr?f=MO-hx8`yX_Iq@R`?p`=IX~!%#!Y5-xVTFY z5b~45mPC$JMa6SRc;wB^R3i^Ew(DnAV%y}w)hf>s_L0=ixyO_Fk7{MvF<;zmer>-< z>}8-z+`9yR^s`NmL zQ&t3ILF!rDPd?HJ@u4m?(hef+Q5&hVZ2q5&)Io|tWW(#>jeHTB@>^+b$D&X8Fns$T zUY*E*6C8Vu4QimjYXnxMLwZ4TVUo8yD*w`0H{b^P8T95NeiBUE3aE4=Z}%49YC69_ zk2yNmsc=P1Gfb-!wp-+`K0?40lSmSTDE=V)MOC(h}jTZ3pf zo0RD!)Z2niEoN|oS8r502{W5UMyYK^UzD8SsX8GP3n5lc=c!wsK5^imqpicQ@ED&$ zvw5ENR~-$HDNa}ne9gm7qCJUJ;B~b4>F@RP((-tFTQUMLP;Js?S*C&W7Z8)JtN)K) z60VhNks!mV`RgA|Xxlua3GxwKU{Z!6D!{uq`~wnplDv*k#yD&tu)5i@K9t|1x6W9E zZM1EN3rwOBP|xYf84`7vXIiee1C9NVnWW>H0JfbJQ$QY`c9+^ur=To3xlKg&=2Pd_ zDua<6u?xS?JL)(i&PtV27pHW7oef(aaok+Mx%A5~WTgnu)~FZEp~Bb>j;;_tcbxn)+(jdz4QFhvoz&msWM0>q? zop@Bho`n4iwhe3W4inqWQ;Z${L0^~0FF67dbiI#V!@AuZ1Ju4fLT?AI#s+t5FIEcV z0Ba)!cIWf1Bym=6?9GE65n*mWLk_{A#8umRWK)R?s=_@o(-e$feVNC`vPPf7x?hV~ zcmKw&P#jctDlG1qtZ}r9V!te==L{6R97RTTe3VXQk9IIl=fXU(W97J_<{sF#Er!n|=+I%&`p`pd>@f=!8ubwiAvS@A7!8@6nex$Niy~yZQdW{7K4tgYrl)Z+U?^&^Q|2j@ zWisT6M*Xp29JHm0{tN)0G6+#bS~`k7 zdk>s4KtI@UFXa^E8I@Amu9q+VWz^o8poVCI$zc(exAS2hTr*a5nZFI~XVL z#XntLK)&9{_1R_SPTB>g?pA_x8k`T;hJE#G^)CeSGs2V{%N7!^V0*Yx!c~F1%(@Sx zW*5bNM~;EYkUWg;x1|lQ_*7k7=a&mz2D2jE9po{b9-d8c(Lol36BH9oS6?)1A9qRP zK0wDDB`FO&M>!Acu7ULA*HBJ_cuM-sTvL-QpQ>xk`FRx|Aa+;~Ti|MYOwv6S+ou zn_YHTFs9MRo{R=Fzs>sCq%jqZ$Q^|a@#jryWltDQxd=?-k5<=umLy6{8uUyI?~h-R zY-TYp4Wm&mWCY@ioO+bPu_jck(;7|e`KCTTcQo4CM*(3}pp94vtR;KyIyDa2B;UC&eS$wBIJ$2o&8-a@~ zbl{up`);X!agp~q#I!OjkheG)7XdnjmL0icQda<{_(6kV z9ZX zfTHMC%bd?WxcbW;#Uihfaz88Qxp1~?16uFLxz8S(DVGFB(Q^fw*zOLyg5y~ih|*D zYgO`Q+lkt6i_uw0yQW@ECehLxa-?2DM1(^$7QcIzGdV)EU6uH=X`iUIQtd~X)&=MB z7c9p?J5++m@#Cj!8}lRD;4$%H7vTtOgVFg4q<=Cg9@JMe=heV@rstIIr-Nj*kc}@r z?Xe42w>b9eQRT%t5i#i$VXzWDX56sy7cNPc8iftguUFOo|&Pyq!iMf(a4)hxB3|zeBf!`)uk8q4K&wj^>w0);J_w(v%so2@ezu;H>~)MBYxR@q+)%mEe4fjSshYd| zMb;8cHW5P8kwfjsd$iH14aL5M7B0+!_xDr{SAiKYn4vD$N+5GjL`b$d!6>@FbH0kOo?(B>}&aj(*fYa@F)gln^?4q;{ zxZ)Ava2w&d=pPsC;en{_EZZT7$4!m8p=#!<%9IwV&(MW9sG>978zff1&YrF*e{!XXfkJQh(ZNZC!obuOci<#B5*UU%+Qf; z+)CyDdlGCtW7*e>^L1H-HAMs`T3UTKtW%ERpZM&bx#vP|4hMAUkqG?#%+3s5 z%)`^mlt=gviSAWIOH8$j829ir_tPJ?yb1y7I{V-t4<>##j*3l$J~CkpMa=p6H&n8$ z&n&R~o1908zS||Jux)@kjY-upI}Gub^7(*^L&_;?7v)E&)xu=Lz{k%cJJ%zT=5X_y zy7g*_mgfP{&N$m$E?E{VgUN}2zgL8Q_qis3+;|MV+pV+U-Jr9duN?@vlYO(mW$6?O znK5ryo5lZwH27FSVK6{>G6poG?`C{mWZD*jZ>Up?6<0jso?=q7O<@Se9BANo+jgz= zdCD&v|9S?2!!+WP4k*pa$I#2rTZ$wrtL%&YvY2sJ1pM|E>PJa+cFhGrWMGM_PZnW+ zp&I!7f^6y_Kngbb0JrtUn3jf>b4~V$o^SuSYe51S1{+StV-(ESl4r4sn~?KAvHKylMaB1*TcfPN*ybH$ z6}!@SxM0Tduw*eT?Txy7q1~qoc-#~{dF?C%%qZj;OzOZXGx_wqFqmXiTwae$N#LC` z?nf!CuRDA~G5w-Y-(D_F(CoVGvO2D4?>ZKngZD%J8WU6p&(}}dG-j`Ry+27M6A$@+76QTLxUpx` zWEU1()J<4`jK8Z|b)raQXfX}p>^G2O(fl3~-3)`SV2KcJP$Jj7Np99A2%NTnPf_o5;Cr#?iwm~b{bcY3;%=i~W~;*1AX(dR)=GkIyN5ZKUgIwB zceY^Rlt3KTMPMPkenNwL(x+$0ZR?|FsI!QDuxfa*F>iSzyHaxyFQEv(UE8m8nKJ8SBVjZbu1>9|(Ti?6or=eNR?>ZPKVgS$oOi!0z5(9!f>%MsVF4x+~kDsdC zR}DpCz*y(iSHoB|yXOD+Bi>{3y-s_grp8o*T#YKL?=@HbDm+3f@Bch@9oy&dT^}y# zWuece!1(bFN#T}pDWsYB(Aln7UM^j=gN~4mCq@^rKDEu`KTq_IEnY7=zGE@@K{EYi`F)0Y3-=z|L}=VTPf&glc?%iyC-~xbpg=nNXnDB zwne6~m}K~q@n=15o|#Vt3)J5TC!nAnR3R{|7&?=}F222C{G6n^ z@+HRN(_hk<*aOAOXjJ5*y>bwO*HI!5}+McqkNs0H$CI@+=N}LL{>auM4 zA~>)4#2S(UALT2Y&S8Q*E^9l5=Euq6}D52A!A=076z5IK*q!BZqS$@ z=B=K}ekg14zlx4-DAZ*~WPV8L@fbte^N?7M_L-6+_Qj~#M{#k(^&rFQ=nG3JMqeo_ zaD`t~sH5o39~bw>a^K<}T?{ueF8&nl?*%CynUcjb1ZpeRzWQPzItP|h!I#_%ASka?B7CSp&* z`(LvOzhv^h%P5~JFq2#zmQaWnVbnJNOl7qVinhgt;F?W_;0RYxZYx`Z+ag;=HQw!3 zziOIkhsgaomdyVC=|QAj<}fgG_N-q7xKxb>A>h(zLiV-lah$4Z)1ifZ+rPZBFvfwfKptX=mNd8UYxk6;HM3#x zIOAJQOgyJ|v5B&nAhA~V!v?cSI$er2G59@o4bfcup=H;=*4QW%8#5#~DK4`htxorc z3+rq8>}B%Y;9csruS!TT3xqma%$T0ou7fd~QXd$g;1hJpoGIe$uyvHN?XewZ_KBu* zKSOBKjn684%M|>oqdWri^kiIM$^dRp#PW6A(ZkatGJ=5dyG{1w7yYHTXzScuP* zYz(6+%HAkFWb*}eTK92JyZQ1{EHKr8Ee*(@(%qqgk{dB}U?BpiKP7h2oq`mSV04v| za(5Wkd4CnvSLFwT8e$`+^WXLigbb*e}bWNy?m=fa6PZQPKVzYrxmJ z50O&{eJoHdJP{bHb@8`cZRue+ZxkSgFLv*sO3SS2d|$_eOp&LXn7fe+Z^}rlnnk(I z{>E%XuHE^qT$X69Ab0B~>4l9H{t~skwc8s*eHBg+V-fv|aw)_-d}Nx5X)PA-h(xpE z$4Hss4+qQze)#=tv zMv8@0k}>A=Z?WBGb^u9b#tFtBg*B_?Qu3DTUfOc_Z%u`Sg4%o?!l~5~M@+oy`#o_2 zQiit8Tx8;DZ1(RbbrL0u6y|GaL6lEsQhsA$5hitCP{Va+@_FhSSC_O?#76pRlE`Wh z%6s1T+nqjK*A^K-&&%>ZWoGSeiM!=g@IAXIZG5}Qt{ac&$rEV|TK8rxAb&sB|Ct@I zH6(KBdE>@}{!ziUZFGIX@6S=3fa2ZkLy%mKANCp3&pDD0I;EDbF=%w&Do9s1IvMlx zGF`j!Xi^d6YMEfT{JtpE)t${ebRHmtP)F_Ya+YdfoViWkrT|3BtoZ!g4L2Izh+{m@ zi~%RC^7^n7owhh^MCzVkyN$FBR|-$>wO41fYOkD>4){2|`Hqzq%P)Oi#^$dks?wu} z&1`iRPvH7#BhDP}k;|-&>GWi!O*|f-+$iD!v7KG%u}yZkmNliVF+_GH$;3^H(o5gu zR|NJG{;D++{QdcIQF!Z(txtSaX7RO*&@H#vI_h@zGYWqqY^qO2Zr1G3y_0oBCzC#m z{9KB+-?M&zaocb!*jWPlg74Sx8c5iJn(@JuN7Z}Q0p<@fiK^D9kn zN0{t&vC|p*F)7|m)S(|Jt}}<;{5YfI@CLe4mx2$XFZd=4m=fg)+dxvX>&L<=>1XX^ zz?V{s36z?rdWr(lxq0NHC6hkv=++=C3JVAJ$e3M;IVD6^|HFCh&B)+{ zhc7K}$bLkQTJ2G!@k%VAQU1Pr(T!#2)^41Gu90m9CrszO$c9nTy%CvV!2Iv}@WPV@ zCtr|!S&C6DdHpMC=5&Zri*+`ES5Xv(QruO#J9C|gj|%D$HnP_N$V>on`F=b|_1?%C zV=W8~9D_lvd>RdTU~b<80lXzcRl2nxfD}hH(sQGyZ=sX@|L}(l$_xe3z>6H)YZXAW z@r2&@$}W!4 zJH7y|a&C#8v;D&xPL2f@ZyZ&nK1B$AXep`a6knn7n&v#$8WCM4jn*|CYqQ1<5s~(; z^^4!2I)u^KHr+D4_0(;VZG(Fw?VMr1lnJI)w&MC=-brTGO)bPGcVRf5SDaw24!>S- zpvbPa5VshlW%O)~P5Kz(r$Gy0jca!W<8HEbKTPXXli-zB1`ClxrMt*&cLuP;)-a^3 zltZbR$>(aUtaPTXvLQs69x3Psl%e5=atk3hP!8B2+cbH?^mk0|2(#P^X%?p`UBE&i za`s;x60N|z`!roS_8LW!k&wK8gce7dNto@6=(Jz~;o!cso^)oq%Qq7Coc$VNmkwTN zDdK>s?Q(JLQ+}#?-JkOolyO1Eg@2SOZt^OPNJDAbsV_UzI|Hxs_>PaIAHu7IfmN~v z1G;kEr#$awUO0>(bnMv+E&-{;XAhk)m2VfBIqRbin|1g$RP(Hi@FVWzXpYOohBB5` zk!mtp8%Jb%+%a{QG}sYAjge2%tQVGCK&wg?I@U#JNq+&E`?)QX*c6uty((x?Qf020 zWYmVBxh4G{@bA@jR|~Dh{SeXdPpAK%XBehga_0@os4<|{K}=eB8BQeXJ=q%p^z4=PF(nXu3@yQ@b^V;x~a zgLXzlt#+?*$H{-;z1V)^v`QkZR4>-w9MWWdVTW>_p7QXP!5X2oO`jh3t0eiiMcKmR z%|QpGxd8CX9v?Z9hm^N38)<4*eg;fgsWIptejitAn@u69G2IN4bnT0f6wm#(a^5dF@V z6uFd`TQ0=+)2t&Iki!<0UGTHWW?3Ap^P!GHzn?7qq#(&YF_s#znh^uz6XUEQ(ssCn z{APYJw)jPGCkW(8a+8WDn-cJ0SLjn5eFk-EHZ|UJlBe1B(I4Vl6fG1V)KF=>U9&xe z#_hqmQKt2A)ynj6D6C{o-bMuRO847Um375wq9~BAK}VgDcT)->8XHbC%-U-F9H9Ea zpd=FA$b6fv=;hCPhFoE1>1&6t0fifHR*8iY!rk}=2G#ZG$??gRogHNi(O|PvMAGTF z7J(YXO}Ox`Azx|pzWI|3lK_v?xyV8NuHYsx z(Tp$rzM_6(tyq$%LfGZNP9{osT#%KeP4-;=YFA&<;rw`aaJOXTWDiW)6x{f~BIC<-SWM3y-GqHxTNEil zpz;YPL@wC%ajrL~fmNK&y4oC$kAvi=zB?E|^+y-rt;N2y8HKTSvArHlZC;^|pgy1t zB5e8~Jk&4|XH5<@Je*M`b(f#V$peiUQ7CTHMG9S`-91egPwxL zR~SxckiGP=$pPJ&vjsufl3y0?Dq$dkky?DM13 z){FVWnV|ilUfTd!i;-m~Nmi$6_$t5oCL>{y{Y%&4E6 zMZap@FZ=%U(RDuC!Sm7@VuB6&O}3#Q^>Mgc^tVSC%@R?k>(>R;`q3kfM9PecPytyn zh2?r*W|Nf}Rt!xK-e6Cu{8?&k+M~B+R?Uui)HEoQ#VH3dEEiax!8h(U6CMh^>lsJ` zuu!?87dZp59e?b`^_hKNT7$JJL@}^hy=JwlvKcUx2tHgIFn)>r4-zy&4^LdGcSbQ? zH80%7_Zb`G>pk#byQuH#>*8@6`S?`s*NZC(apkceZs`AcDNEkiscgSM>e_OjCd*n8 zFZnQIW0l?si!vr*<2YKqz%D&+48fbu45h88%wV{B3|e=bq5fF=%vOMm)B_F|&~ZX1 zeb>+=Syh#2%Fg)A3eS|sta46Aq|P5z;kq2Rk@f(x{xU&|JtU33d^>1borri;80p@GLAvSmhr5Vrw_l;hOPX6i}*vV zp3{{*k$T57Oc%4`KDT5qG7vFAxGcf z4ku=n-DLmWDgjOC6J<|lSpY@EP8`GqcBGAhb6{SbkM1x<`?8bvJ%FE@K2 zJ;FNa6_)=g7rHcx1N-hW=j4etDfuDdIGd!>jNK&48{}HkP;PgPN8ylcpb>v^hy-=0c&e8(mGIGG> z64Lb#K95}vwhHi0@RbI(Ud#ubjV542$i-QULAwzbO(N<`^Po=ScO?sK06cs$8z>ev z&cHXIhgo_meyBB!Yq6nRfRpSy_Gt*bbH{UtxJyzsDKY$cGOM! zHEvq>Qqz-*Oxl_5ii2c{#{x?2uYwdj=j*Wc-PnANA9(m)7FmAJx($cC2{litfY0Zt z#oDqeBQuJCZO9;DL@}IhYAgHbFt$Pjsm9`hl4Q3j`o5O@7*^oqT}Cqtk-v$jgR7=*v=1kMsC>PK9Z=C2hi><`3na2w)Fts!YLSZ&x6 zK>*X3OFM|}*bEk18C|S~af!fW>i-#i_dEiljRIHe{L;LIwTGmeV;r#`Yww5-`!OS{ z`$API^nIr2zT}=v(Jo(Zn2w-sR}rFR>CGt457fatm<7D zSTH}=rkW+_mkiJ9WE3v%g+bDN!Yk_yl%&;Bto0n_y$0RF^h(9Nszf|6MtBxVEVGWx@2Mi3SY4&QL$vLfh5rQAQx z4_A-c7TZYEy=Zm# zO7pEXS8UBs@Mb(sU-6Cv!tzt>7V!D$dA&sDu{M=9boF+aJEHPN(>z%b!36=S{TDrUTI6}v*;ICiE!lQzcN3fwvAix8s0ha!Nx*IS}eu#YC04@HXF?m^Jp~7{`6qEi)jm8?mIW+CZb2WjPwPsEah#}|r zl~}u391My$JCq`$caQS-N&<-a^);XV4)z2=qhLsY6*EbCS&51M@ZPFb2%XhJbo1Cr zNv|H7;1r$z*z;x$Vrb-oO?Wg7sB=MfC&oQP&(d7LSbw`oWXrHDD#LEP5IVR!Tq%P> ziuf+vwO}PocgM7>wz5@<)vS@|S_?{8r5@T1r@g=&uC}Hoj|-IhwY#Vw4NR>}@}40* zeeL1uj$&Bq+1mK)uyY~ag9$WIF{4`Wgo*`!9G%71oInspUkQg<;_l84Xxv?%KK$7w z=Oi;jcU67&jx_l=-~nouY7-b+Y+mgHLwWADuDzApN&An-s`T*|F%G{f_nc4LSils> zA`kIFf;EPCZ$gqOJ|{RE+vO(|x-(IT=8JIHp>M6Uj=5zm5Sq2Si}Hgp!euRe1$!RbD+{xEi>Mv0 z?~)5)8;7EJAcD9;;zYh5zkQ@cy3AS)y~^Hi-}0cF z!w08w`1GdnhmbRMt5U^F%x$SsvL7=#4LC+u5zIwtf#PYh>A_{dB?-Uu2rhH z*gm-HpoNiiHh1A`aUR;m1_am-)^~5dMKAlib6$rr?1K=iV>dF3=kW(RM2lTuc)Mj) z_|bfkdJokX&)*T}swhz_)2-;6`ApOPunY|ZP@l1G@vOMTWlSPhhqqNt z8NH`1`Ek*EFow6f9`lr@00rZyA|PqJh@B)Mal#`r)TO1(}VeS9ezg4n6zI1C|gFjAYGmt4kNk;*T{-Okpn z?3CpguFvyT%-i9IU9e%F7feHY>yNI0fx~<)y=OLGZG2`3P@v@m09!_Ekdxi`EhCBM zXa3M?uIae+95&VX#gVd*$wW%TLyI_c&OW8;<+r;lgmL;5@%H)(5VQ|4inoS7V}rFP z_9-8*8v)G`{I%mwp6%hQglXd;kR0BtT~uiCC%n*d9N$!Mf|bwpgIFWTDSo0s#ov$` zWz&6xhCXyL@=@6`+Je6isw7esc;}3UP6eefP8^KWoiSvj*FacPP+)03D(iaHz-M5+ zVc$^Veoxl>1L-g?jAy2#k{62JS}pbJ-h2BvUU@_0TA0)RjqrN%A_}<;T+RN(TAbOi zs&Q%%eZ&p+ph@#QN}$$kD@07DFNX(ErBtx2$O-QCwN_MmOJA(D#C#Psv{$X$I%6R& z*1>Rms^Ca{e|6P=kMhy>*5K~ao0RS=>BkHuCjYeV+hrsiM-)Q*{OYpWjVR2U zsmB8|p2K1p^uo$=w2a~qi;a?R1htF}encRjO}bK9`WCbuEc(mET+=$CTFB=~BTIXg z^CH-mXOLN=YPx!TR>c{mzjs$xXbBV~5!S8UW4s~g-bAu>?g$*XTT;foUBqNO(HOh& zjd=@suw#v=a3m2TlUY)ZHnC$cQ}s@GdVyK!f8WF1+G!l^Dwje@zP#{#$MeKDZ% zyS3%M#%9v;gRV{645z`!kl}x!_~|&>C?^EL=^XH_QOmw!Q_F82N}G}$tHOS4g0XC# zZwgkb+5#qoHn)8w#!;=HO;8Gs4rcQ#>o>;7;HiAQk8;lXM%{h;W}kiOmfs%=o!v6} z=Y_@357! zyzg&QIA~XjY~yLsN6r1c+<{0k1gOU1h`$*_9*Zxaa6g7{OInXnihIKVdaU5vz=H%n zn&)iu#T9f;MD)Q!3PGLRvK-#E0!*lKgZLUByr7Cgv5zEh`lU{k(YPWkJTbpobK67M zU7h5w1?zHsI{4FuaOv;`@L3I48nmB+A`AE~4Ia5TvcAkngViv6}KUKh<(=&$wfw^F}bGHNphRbxP$7pP*gse-IF41uolMpM7#CCH!99HxC^K$b{WT6gjo6#gfzCW@!YN)i5;h;T)ApPSfweT;6 z*LZlf)=g#XidmR<#Ly^zXHd}+{!Ly}S}nACg3q}qZY%Cg3=fUvJy~ZM^wQ{YM|E{1 zlcBD!5Bc0Q#uimhu+wqExPLe4Q<_nehohuFjVtTH9?g5H%It|HlMJo?n>*z#vmj#6 z?(ZPn&#i*3R9v_lxJUat@B5jHhUY!p2%@R7q1JV9kj!bV;0p@jI4;dATPlF z9Nq@wc5-6y0S|_A*502#{++G#aCbK?Ev(#&hGC!!^kaX85W+h;i7^?MJKc?7s{7eQ z&knAt+tHcmjWvI(Yv^EcYNs)!&RAIQUlMXG3G39p#R*Fy=Pq7iu-(aeh!3tg!e9l0$@rfKQFfJ<(T5 zrBv_wBShN_Wb&DRu!!C@c-6-*6UUIdr5C--T7n&y2KOs!xzE-LWvb?`aUPL@Hh%ABjy-{bEPw!dxILTwwec{|)mFQ#@75#5j;K z)?_&&+U--{`M@maVZ!`VWVSB>xS{&9@z~x#z3!39Z)=?DR3n$26TMvxu|S)2W6C%k z5OS{yVsbrq1j!9AvA1Ert7)Z#r{GTi#C$mmx{H?7bC2tF^I8Xp{o@1HM*Dl6^B~>$ zbJ3I8RR_~&pN$T?;I#5@z?m4nqEsKL#FjAe4V}Vp07o&zw{2vLS$vGablx-i{Y?3) zlp}&wj<{mxa}`zRU6)u)(cm##OF>T-S^Ypds6zG%-Z+~Y0l!6)a3Rlf&iuK2jWfBG zhDNJdEHoE59?;hrmfmO%26R;#3hVbH@M()cKl<}gLyg&ANtr? zH6k)NR6%cq?-T?sv5_#U6RI1g4x{pJ(6s5XvP3+cJ+Y6hPUC1#t zPtKqi>BttNQAa+?f;hAnP0(Q5);bY{e5Xh{>3!0XhBdR)oxo~aXv{P{xx&+*sXs5K z2!Fcj7ddaYzBld)IbMk5Rb5!_`uP;hr{pkrVITq{Yh;?YJ^m?;p;2+ zaZV%o)%<;gYL{;(>4E7QepBZR)P)7b1QR;w8dLFB6=HcB9dx2KGso{^auh+mHLUwo zs1Zi(&_z-kQee^p$v2~udM4<4ix-%M`l{R%uAr_C_Y|sQ8#IcXav?{yJ)=)LdXd@K z7qV7M2|V$;cWiZk)eNJo_M{u_Md)P--QA@jf6Y6mpO(Z;+GHNdteN`4G-?;&XzQ=i z<{HhNbi+Gjg>J)Snr0M;+bnAg(h@{NZu-ArpS$4&#%rrMCl+rmi6(ruk~A{aa*HwK){)OW;#65wJXP!7%T#i>TY7tj zx=}ajW7zLb#wv-(WIjLHRJU35*hy<@<2-J@L^YgjV<&$Fathjsod#xB=})l6A@I!M zgUhWQYD|`|t|W-?8#l4J5d^J}kHu!wFpoh^ zKkAXvIUi(w+>-v*C)1yc<0)f&BkV3j;IVe2UcUgH?bZy?u7AoG&L8L~Vg@s}+Xn(k z5BMX1@k7jI2T8D1Vm7tN@M+@1ga~>ORmm_1ncD5hs%QU+R{9Y!*kk*$fllcsj{8W< zkxP4Z`d+T?>3dBSu8sac?7Nx zg{F4H%=KC=`MK|pT!L+H4<+9+O(yn$Vx){9OA{Dw#qg9GDW;(Lw{!<5(J-D|Lnz2^ z39vD#O&90?kfhzwsC@)wPE;j^8h2mQ)17RXHQCBOAOv^p-xXw)I?~FDrX!P*;q4)h zP4DtA(Ya2+jz;#&91S-b$vVZEyMd`HNXZvrYJW z_@xD#*ALN84BrlYQrf!B;G0hP4spS78QJ;!UV52&H=J;ikX~=gP{iGf@tcbLt4zjl z1^3U-{ht>!Z)&Q#tBGdz6g?4?kl8>O#d7$A*i?G^t}6O5y>ly`%e9ANp7J=>tRDg= zIZ3Kd{2_o$ka@~{>Cte%NbmU@o^UU91_HEI%MtDM2JwM)#Ih!+hDyQAOGiC+4(ng* z)F9cuX-*ICLd=B@Uj1?p69{DvxLUgKP%&)M$;kd=`T>Tm811@nv@yDn27;7jUIgXl>Zk+yP(W zw!lUWCr~{FrZ~he9Z8sn7uDo_4Z)S`{3$;+p)>{OAexF~6*&uV=)1-Zwp+5-f5Pkb z?3R{@tDh45W}^aLAJrY$Mmm_@to>)*=58tS`1kI>YJii5%|pe0;H2_O=d|44vS-|? z+Je-g#;tR9-``Aq1r|x`b!yMGKIsyKqGF-9`4xTOG+d%OXr`Ka4{RdK^R+AWPnmf~ zO6JfCH6C@)vted;XnA&f>P@=#-|uRHi7}JM<&7P)Ph6tS({$)hemfLk<*E3}cFbHq zsN(?y*O=BlByOLvpsV@(_%J?iJv62!B(ift>0XFe!w&_m6)X_RTkaJ3hS9vho&H4V|~jNQXzz8zGL;H*U^MT~fX@-m%mkuGUQ zbO!`bMFRK+Yw;TS`E&FDD-VgcpNbg*5UwPn8^&kK{xd)Oc@q=q2k2qE`P%VAo+(o*%w(TSg)P^_L+Z~3p)1g!lsJY z%oSxbTv+HNg_UF~zOXVJDq2ddW8Fe%9ApOdMWCS$Ll_-Lt<_E`fG!IVMKsyRTchQ= zNN}yyKlx`EQ|i;nkIqbe=n|vx>s0cE$(_Wia+FV4Y2m1JTqzxskq6GF#`VrE=Vy@; zawDA`S=|Xmsw^W>WVDIxZs%h09G&>5W$8`3L(^IeTGcszY~4)-k7S!T? z@xP{Qb6+Q{*K51WCt=FFV-#*T!b)!P4SqV7PH*)|4{jX^?ro@=vbYA8Iq$XZ;w%ZG=Q`vGQLP@|Vs96-G?w9p2NC;<gYU+ zV{5n3J_6MnS+pf_O871;lkHZkY1CX{7$Hwko++zar9I^%&x0HROxk1xp*)+*7hboL z^EDj#S}PH|K!|6xRs-cv9X-dg2|`mp=d%A zM~3!VU{Xc6)A;Y#HnS0(w{K1kL1oo!IGVBi#(90%LXZbt3dICl{G|_nTo74inIp^V z|Gr040UV!HX^*-J&sFcYvE**u^Gdrsxqflsa_Gdl$won($>0#!8D#(+Hutxx9tBhC zekeb6ESR`@Jj!NMp5KSpVoGPKLaE6hN7>Q1i+Ihq z=Ajg{3>b?Ekq@tDRNU^J&!6ENEz$3z*zOrq(XnTw5Ee5D65M4UC9AU~jSNn(qwZv= zWI>Wj8yAv)uFX)!!%R(OvCi2ja`ATvy6+{tH!V1oXJ`}<|8NbWa1Zskn z9ysmYpG|)~amIWVqD9p@ed~6A=1{d^PCZfds)%F~)cDceqLC%iu(WCm8CHR< zv?w^be8ns!2C>uK)?xT8^Ro%TJoW{J8n67Vv-UuiOC_w$PzD3P4o32J1u}>@4QNZjx}u8kuBL3(82&~CplzEkw(Ng^SFu6A`e(ju^nOw*T`UFMHax< zsfuk-6Rjb|Qlh!@>0-q}Pu+m_?`xGZI9oCY!_z-b`8X2`1Nd6O^mqQH;*CaN{esiT zZY_}}&sq+|o-%DI_3hMKl-KgeZ$j0hEN6=3R@ds{@*Os{JJEEo@dBz7e!fE=4*q=C z?`1Fy3O~;U))p%1!4V5@s!;#9! z|InaDHEx5}3wH&~p^~_KShkvR6p@hD_WY|4w(LH?@eLk9zI)bF=d;j2i}yb)G3=gT zvROki^KEcl(3!oy4C&lqfiflICF~#T*;{ZccKU*o%Z6;1i#0Ybf+J@1cY^DqV(1LS zX??XXWDbIym)5zdX2kdeD6Sh4$^Jqr^vp_>hP%ya8wgobyz}sYdE+IsH+z=d-48cQ*eh`9x&ETd));SRb z%I*3*hI$335Sikn!Mn8H9CpiPS}m)s$~iLGr(R!JCH_YpGfYUMF`nL;_2$xeJ$=bJ zG%)-W-j0m)A9z(EOjA&s?CB+n=NB;(CXMtuWmmQHOOFGm+994a*fKP<`+O7en0;vDgx5)g`U zdBnSgeZV!RjI1fA6sjZa!?7!FPgYEaUi3}1*wixOZo34x2Yd$z zEZuvraRPqsGi*2H-=wBN%IkDdpBWtJXk1j#$_0C=ipEQKAv z2~lF$ebfCAf^iVkM)3s}^$H?=SNq^4mF2qZo6ziqIwSn7ujzZD!*hcS?T2p;+7EHZ z3iTC3vw1|M!LeExuRMCUajB;O#7Zp#ro!NcL8f+2qB^- z#P^^5zje-xho_7*s`OL5UfWSDS>}oyPMQ^-pnGej8)Nl=D&C5kz;(5Th zayPQ(H73!3TIL=$sC*`+C0Ha)#yXSE1wB-RsJet<3G@D}(ANbWRK@m-RsAF1}exgBG&T8$#1Ty6@;-Zx>P zWw;j~#YJT$&OiuC)-SzKmiLVDh#j6B{|@3Q(^M60E`(z$a}zSKT%=^3HfhZt%og>F zJtSQ8%#)w6XroBOoiQ8gop8GpX3lX5LzRXe1?^`9J5@pXEhqh2VK`5@>H>h11gPlh zI5VnNLN0C=|0!36$~tzidKWxL*U*u`BPFa?jHf&U_^kUmrP_U_fpXH*x@5(hP1t5H)d~?YV2;W~Vvra{WpN>pB>6kI4H0oVbCn*lmI1XpMec!QRUXD53Y&$IA zgm;CEE9rY8d2*IG+I{KEsx}r}qSaP& z;3&dVxo1~R&G+a;e3i7IRgYOCB(!G&V(Zmo>@ z0tE7oPI5V@(}QrXW3x_K>B+RT=AAmUd7Fv)7)Wq>p+I;pEC(B5q2u!3PYfbV~;{b<6p%?yRI0$Z_PE2eG3fc z3Z-@HhjY|De*Sz|Ntl3ETJ>yW&VAIZ9Qmjzm|99TMnv!Pm8dQlW*_gZ@QsUt`>K&QSk7Xad-a{~JTRzz3~78= zbp5~^$Xz61TkO`W2H`p`+aXjUHr0hyv2T3n3zpw2X{<_|ru~z|NN)18u9gp!8%(3G zt<^P8RA+TJ!6At`L!2BY3krm2_Ya3-Ooa7C9<2!bE9aDjSK^KT&B~A^E5g0(RR_ZPA2-?NF=jUhIsCx zW0~ONf<9%NTy~Zy=sR8JpHy5ySu2fATK_BmDn9WXcH4W>bHkl+(I;w z@Lv2%nAZy2kW}GB@Jz0`Tt8;n4dq)%?>3{U1sVBEWo$sz%N5!nv`c5kA=|&LIBB=Q z`LWyz=x1n5x1dCxYjgf_;4ZYk3)b!&#Zv|e#Gm-hY3ROb;lA_kq|bm(MJwpwsJ!lN z<{$Q!ato*Xor7k)t%Guu*KzN~dnZ-Abf;&2W$uFmp&q+dj+r@xYBo60p7rbHN3=+Z z{hVlgGQOj)jTlpyH+@7)EJ=c=o?(D8d!yEJg9$ezJ~_c;QTh8`jG2veRPyo6}cVSyo-W zLpVzFFmel1gx7f8F`hXsyL_{6;*}7|v2@a0ycx;u*~^&2_pCcVVt__E6lK3k;=F6p zoL+0h`uE%|jTYBFXDLkhO(jWcTmjymJj{~(W*2N~s)>29*87fbeL0y1CM>idqO$D| zU-1H!Sh3Y0&A-#&8}nzgiy~b{7U52&Q15;vibR!1^^q~*>}mv4rzo%?WhAK%$6xiy zOYfQaQCJ1n-7=PoPY0p1F{<(pTzdmS57TY>>;fe__+be6JJ1TN2Bo$JN+4#e>gA}_ zw;^&|#mB3JeGRiuu@p^&XQw=7y$XUutG7-e%-Cvv!^&Rl%pz-fj;;1XYGMxj8h_L$!_v%qwlCZRVJF>r&sN{!( zNV({nfU&Lu3>*|41T7F4@o7?pwMieTq+l8nSv!&wY5nTu`b;ENCNcvT!}VY_6uXCA&EA8ru5+jg`Zd@X6~zAGjYHLiR^0`3BEAQ%rk}Lu zZ=lN7L#qa

yl@k`0H(JIZPN)Oe!dTRpiaGz7l2daT~3ms^UXm7l~xdQvuA?0`L} z!6$V>V!NoGCs~?!Yvh(+v9hUHOkua{9*Y_Ofiff&zXkWC{2HGz=x{oF1zH)lX#}vT z8aBBNM{o3P?Rl15ZDQ+__pKaQaYeT%xa-2&W9pqLd192Rw3Y?<9Lt-`Wi?pSc+2+( z#(AEU=;Kgk6%_KiHkHv5j4@gUGx?(*wseTUw*UA)k8zNU# zbwO&2H+tB8;HTExou=K8j2{R8(DJAZE%D2Mq*S%r?k66Ld5Jsf1nbgy^33$fv|nDD2X65qxWg1T3 z$Dvs2OmQn&ShrICBz8LOnd?wIDh_dD=xUGEn|YY}$_?3wk+rFXi(Qn0avQzDuW3ql zY+5p`4d|4pg`-|Nzxgq^=uCu~7g}H`{d10=v{&jkO1V<=+Zh73x)ms*tmcQbpC6_5 zRJ}z&n^JN7>f+0R8Hs9#dj@*eN|Nm)h@umQH zct?JYF6g!qmb3NCy(*6iD%~u;O9ew}w^u)O z!+W=Emm02chPny(3`#Kw-y~FKRiTRKz)7wiZDxpR- zFn5r386K(xhtPNmbKf;mIx;XGLabe`g8PT0(>ID6)a;4TOfGla5`oCZl^yKcO=3Np z4&xK+THsfXi3yk%1Bh4dR@kt_8I9t%S*x+)+P6~LQ8LvA8#8qiKhx#{xwy4F{1yIL z{f2y6>IzcSV~y52%zDZqvb0VuZL)PN<6S^#n9h8XozA_#Yi<+ZMK}{{KSAUa7lO1o zlltlR#78SER7c5wV=|3jIkn4O|6l??q?FBmh9e!Sqe%GX&>*5@ctgf|=cGEySaeRk z#PJEZcKOUmqeaPZHIKBSuFjGhwH;>Pi9;@~@NAcjA!*|F?==JwbHSV&p$s5COS;3R!`0>&E@yG67M0o*=g+0y0?(6v#0Ljda1~hDb-TR;^TH{h z*;a|dm0j6Ymf)W`M<6Q0FFFhJqCOW0mZC(N4$kCBG!|G#VIeevD2#Sw{2SZCC0;1H zWvsaoKaJl@@r9AfE+3t8P-&Od$HRJN?xtmClul4T`7a7Yj0BVTi4%uI_K?jf#m3JL zU;N2=teo%A;q8hQrrsDfP3crHB|bBT`NZm2g{xtw23=fR@_1`Ex$`ushSAw{&adrnQEAS=k$NVUcm#w2Wn^*X8`;o8kPgEN>i|*wM5Vq*#|w?%>g$9O&h=@>Qqd zI#Wz&U{2Q;8ME(eGrLVbzyFbPh<%;vFZ7vDtAYr~4d13?@pG_p+_T}ycflYt^VwuR zj0Q9d4|UGXn%gY&torZ>Z2MT6x$f2E6ss^_VLo1G{3l^poHK}s+YcvvmQ$(yfyMD_ zK-06t)K|E==JP>_%D*_-)??B%qcZw~S<~S6jnHt?ojj^;y($nU`l@l@o5 z@wIsEwcPsi4H><bLqd7=~IeM`;v27Zi3O)w%_rfx|oJw)er)a&yed+dG@j_NVe|uN)h&&^U z?nGZRyPvqduGzSYgw2{xd3_H!d*4W1W2uhe1V548@Dd^M)$zvRH9SrhqqGusxMb{K zl1+m4eqNB@8q%x6cq-qZ!8~WLRduBe8)@Y1kPd7*alumk(iUijO7--)2}}8WOhe1I zk=Z@DtCMw|ErP@0i_+AtVxHn;m;CXGdZN_NIJ@&Lh#&8V%Z0#f+YwgzrW+ zb#s;u^Q!ZS)?3e=@!R87@%FJ4ZT}*rl3B$kP`+e2TU(!33z4y#`(3>%Cjsz3N$T8| z?PBGrT(OC{a`E$GrtN$xpe_YF>O+1&H=SOBzE>O}wIZtMt^VhK>qnxBU{N0eQx~Yo zwdgb5M9UzP{h{q+>xTmy2Zw;=b$+eWQ?h1cS)phtk1dm}uXNKB2=u!Iwnf`Kx{A0s z&yM(<%)t;It{dA5>b7IWR%6k()6XTLFP=N#>=M?HQzy`r3v`^Bq1|VjTOb=e{LVK@ z*qfx!c8h(od!j^#RP;I69}2?cc3oA2Kf<$K)aP7IXK;}Z%jFg*ObHjTISQq-t#{&R z_$c5jg{)3`UA8r!);I#Z_n@ zw8ZVZzIB$r(F|$nF406K6|D0D9~wyv8@3Dnu?kp9bYoQ|>AI*~;YD9)z04C!Te?WUDKfD}H;qN};?zZ!YyPYeg;#XYFPGPsg{ZMGvg^(fmw zpG$Ml9Z=9OG10qid^flj3B?8d_@*25_66JL7G-ia5pDF(CUI8oVt(x6?&c}4_)Vvm zsFQ9IbLCe8QAC;5O!J~12)qFa+uD%wmbY{TkuvXj5Z})WA5laqk2l4PB_2kUFz#CY z2#4eRIo@xCTC`?%;mDd|Cq8BVQ+6IlQ7h<{==ofShR_cNDVb@zD1i(<<1L$1J<(N< zz0VPxvn-=Z!V?^(lsB>!EEg;jBBZIuCf`1tb`bGf|L!t1M5VsQI(-);eqYVn z)aezR5(o;ia_gvV2>wKjuE#kfN}Xrol%EPEwjnn8`DJL42i`BKfN9y92n%Yq4qLT) zfGM?5lohG&Hm(OygQU)Fq2477&TN!dF2k}u-|1kEKcHvMK!UJ@n%>WLZ5GA@Jm>W1YQQs2kVPhXE`zFqP0K^l=m+LJ@^} zVN|;0Lc4aAE8V(kUpLNTparaG;Gm`OUY^r)<&B?=)9{Sp09+AVisV6C6@~BW#EA}i z=o3}urA%qW5-kl_CoM8p<>r)?k`)+nap&;H13Ip&TuvphCC9<%kd{}m<_C6+^Z=f+ zA1Wf(ork?Sh5H;0MP3Vm=^g!yk6Pu=G+^*|$}c2p^Tjekq| zf~bpt$|w&eq`r*j8{B6*hRzfABj1z z5`9r)v_rE+EJB5pb8i=5K*cKYOSsg05W}eyekZB|-h|JUy?LC4!R6b2NV3L0T7iVe z#X>=M>BsZa7ZnRI@MK_&{5_{2jf_~X3!b6-Z(L01xuuAjmT zWo(q7xm$3z{9&iTlv7`1>MC9}sG%_Yi_}FAEjGUOiIYiPq5+3npf{kOeE5AVf$f~Br<`v?eANs^@A}b6HtJG|U#we&#YIoMtu$$F z{;i5By6$A4f+w-i0cBVLiPfRpSUlZUY-ez!rS4Q~JCAr+ZThURkZgZs+bh1vt@Rt4 zq@z`WC)s)@PJ~sX-+{i_6WK)_`f+{R#B!;wxUED1dZsgTO6XH+a)Q;^BaGpg5vTI0 zrHl{{pvZ`;>xCmb0Y}aO>ly;Mjh~mrj~UIh;AHx;Zq_&8!Q;AFs>Dd2SH zq6LSaQ4+kBAU#7?DwP7Bza1z~`LTfsl}8&rrTG2j$yu7KPKJg+VLSm5s3IzP!|mV)?ca zwzTvZ<^AY{`}$WsO=6i4;>+#z*kx4I_RH8^VqGEp-3Z14xT`@6Vkl#UDOsuQ(vOPNgJdMpywe#c3mJBvrW6$IEYB$HC~D=lzKD z+KKwD7Q*-xX5=!t@tZ7LL#(hrF^*b}KnY@Yw)TRakh8ySg*6cs)?9Vq+DaXRjk}6J z5tuN(u{6#Tt_aznG}JsN7z}3Y;z7ocWM0NFl?ShZp>~7dHh#6eF&nKDACN)omo5gH z;1rRMo}kHAVsGx8^>_CrU6!cbV#|6vw`qJOkbspM%>|=3Y0U)H^+PUN3Z=9Y5YXg~ zBJzVpImzPXD+FyOa8GPAgtG^#m4hH~ff>qk_LI=n60P0Q6OI;h5pjiRA2dJf$_Nu6 zGyh4uxIZ%BUs^b^&tE+iBIMty=|RuA9!RcjxZS0YhxMYZ92>7#G=iK)-?t7m(JmE5 zjnHx&X>zbJJsy6J1Wu&Fu2S<^G7%+MCKHw4t)ENM%SQzIgh`+|N4&aRnwgSL&~kKH z3B`A$(MhFgV$CLn5C5L4B=^c|Z?~(pqHLMpq7+IwRD#;}iD=eo!aL*C#lMO8R5=i2 ztV?vvFXTsbr5SNiuht*pe;X*5uY)HbD-gh*?(vY$mVKK4)fO}|&X z^Wy`_ZJZ9qw8J0kYBTu_b`;tFPF1&Q<%gr1N)={1%&ci$ZQTqGrcFrL9eIILF+vr~ zIUA(ru@3KqMIt3L!Tm~{Y_^AeX53zI=Y_^p{H*~dukV=Bt}$}l9FNWjCN`mz#lH)^ zigi-S89O12je7LRm@9cz=^W~Vyb1LAV#$)pit46zVz0ksoRv?Af0@!q$;HRP<{A{k zUgvO2?8em3Sgo7UkuJ>h(a#G*^y%|v6Y~i*%(Nvm59VoauG7VRq=(t*Rg7)z>(yXm zp?e9z4?Dm>AtU^>cPO!^KzCDUyObi(cKRu$9fjc53#cbHr(hPtl=h-pJ&aeUhWW0Z z@$6paMFQx=iKk687#RscxO-blgiF~_p!wNAE6a}0-&fne%*575mne5#eHw6Nblr(8 zbbeZ^7AM<7F-BGUd*48p?qYRun&3ngWgyrelN}I$H{>mopwyC1V*%+4C>&Tq0%bpIr-#J zQwqvo(V5elHs+l>7zx_=7{qXj;-gO9uA5YEscy=KjMl>FH>NKEX?HYlhKdUNo}e?F@j*i7}+(_eW1ur zZ#+>78MqJc^Gdv!8xM-jD@e9=jhUz3D?BJM-J+ z@1$NMC~{0a@@t`vh-V)2p0d^yV);I>#XaIm!e>QC5DWcA+dS%8hbRq(4ta!J0;TCL z-94<+&@swkW)~nK{3kcp^3!FcpWz4kLPbOxQ$CVI)Z+v; z>aN2nEyGgZK0b;md4;FG5lRv6g(VJqs`sseL9AhS%-P}XuwXVPiZ;9;Q`^?TUP|an z9}aLgy+NEAY2&E*28)L)3eXzmRq$9C_7egwTfT?vCyDNYS=ktCoU;e|C6Y^H1eHaF zx?yZPsiN3viI~6DC4X9V3GmG}Ol{9Y;9#mrX=HRj=P8O|Y@+b5S=Cg)a|z24F;>EF zh?(^7#uge^`s8XXlk_~r*{ywo&(y~YuD6FcG`vc;>5F|5#(!xUQ|q;;2Ut*AKObO3 zpma>3XGEZj!Fx0%btxzvWb#!VTc5&5NCWf|? zP_c&+TBX`vYe|EwI%^rzU+NYXT^s9!t7%-AQt8w!+GkghtBeVeD*Nm;b!6(#IxsNK zJ3%%fEIELB?9WTi4EZF?4WOOnucUvblgJ93>%IL&`_f0{gTE(ohc^c%obd) zf3PyYxy)JRVl$;I*ncteKzD4t!&45&NbJ+?R7gz`aKNH6Ys$TwrU&?#gA;c`psLb8(!x)K z{H}CLdE)zVc28dVx!HM}aAIjOWqa3*_#XXi|4)J#GpLi=E7*e>doH=iCkV=vzmJru zw7`-dI_t+ZK#0ZlpP(&~WQ(Wl%M8A9y}!$wKgA*t4+g>sXjXB|sN&H@MQr;msk~#g z{)0}b3~Ihs(1?1_8m`6NY~~)0&}MLRMso5EM1(?b_cWi@?OC5ad&j+K+?@@olu;{a zDl>%^$h$EFd3GadScc;t`K+PPw+(r>1_KIjSVF`~GvXL$-nv@O+Y`Qqyu{jL~QSB*WU zaC#-PGD0opmxk3f{PYrgMi})W6}R&`>z%D1Ih#fx_MXM3X_DC(SJ&#AGXDt_I*3gu ze6BAaXgfesd&66H{%HD(eOzhH<8}x-Oh9y^L&SEi4r703Rd;sq6Li!MtbVrqqm3Vk z?aAMd;`|-H#MSMUMzDTNVkg#!?;nN_Nvn30&y?2N`T+O6csb>6_mS7!Rg+txhZ@Vp zs)JCRv`?RX{_+Xom$R$hYWvGL32HJcpflywm!VO8lX4kyHM+^l5Z7Ev23bX!YbG># zKxowB@iI0m4j8^4ww>no9dKW)h2zj~bwWWwfuzW>!j?8Pd;18a`ErcH3plQ0Y%Enp zzV64U?~At5775W%uQIzwOzwL1#T`iIKHHf@+y=g6HB+_p=3~BWYgkAuY+Qf(ZMr_$ zSYK{ATQ`Oj=rmB3M_UC0xX7}@Kpj2tXhv-5B|Na(qlXkM0scMR=V075w z(b$ThqPr%(BxSh70Y6yr9IF}@CQ}ulf=(`nIs&32=HP{VDU2=Y?%giqd`fZffHBwB zE+mBF$VMy3gIVFNJQECR?3Wj?yM6`RPPs&5ap5{7iEpjY~!CjCQTX0dJ_>35Fsa_hyZ^gh-(EFFIveeI91R{ac zI(3kb>|)SDH*+|k8KVnCV@UKX&DO=-#-*Q=$U7N$_o7}};FhBJ%9y!t{!kmp8%sNc zi2OTspJ}sF34BuH8fY@Sh9h_L9ZzPxzp8-{RKH_t2!6LwOS8nNznYXz#jmoSup(A? zo34~uk#tH@mZ|2YUq~0onW$2Q=N%{gfJQy$THW%R)OTi|uH_hu_+=`Zf4%6iDjj^V zr_|nL?g1uLVco04+I|6bronP!QNC`v?p)zvUO7nJ!Fp-_La)1|czdDG*JVzqjCW|? z=Q;d4@%bt`UGx?|_CBvZ`ehQ9Y$1tP!%Vuws1Q*bKv|*K2v558ZIW&Ycl=IhbrV>P z%0>_E`28VtC}N?9=@=rA4qr?o?O(zuJ>Ph_%%P9WK;#FD+-#=!keGk83dIAmjump@ zF)cC)a!}CjeK)VzRVJ0)57SvB<-c?rydlgei<7*mqnZOrR)^~h_+HdQu?ZAE>Fk>N zY~7w8#q7$X9&HU3k0MZCV~icE&!X0SiF0;F|6MtKoaBe+k@wk}+=L1Z*rd@5^_c8& z+*##b>$~?0L8HQo0>gIEyWo~r?Fm)T1C@|rfh5B^Ui8&G%>D?qza#73(13|B5}P3> z$!bP1okpLQz%w2|1dK7>J~&*k8$%IfI`RdaHQ7V$PD=D@R-ZQ7UeMo2`(iEe9cW%B zi=v=kd5nHFuKihP0eUzr+!O_dsOV{a^ez@pC)i%ZK~S_=O$trtlJE@H^`?Rj ztC!`Sqb^wpm?Ss%J2390!wg28vW0XsTqanvb3$AIFmO=Y6vnIsP1Kn-K$Zm{;`AOcxDt=eY37 zT7Or>D=BrIK6+{v;&YFdl!c3rp+xe?+1|C!Ui;zXk70n2G$hiy9&+zWvSQ9}S|m}^ zngQXI!qEG+jPr*eb%98Mr^C&09?0JhBz|oxPHU#mzM8b(>>9@c>vt$;gEq$1b~)pE zxoVty8xO;>`GyFq@a{&fLWM}i$!?uJzG=dNe`Xe~FB)d(Cwn7Hh&sPzqMKWo zN~|5tUcjeFCL@fT(PE0D;zbVB*sK?lyHF>Cg;of0ug3_4r~Gxf+__}kB4`qs;?s(u z3xr*0T}2@`e^cu%LZGIbBZwlfaY8HLlPEz|>^i0aeDEpb5WDtavcHi^MgsVP_yob! z?Qnd8-9@+sV=CUM=Fmw&JW{%D6GZUt{=8)w3MKM^xFCd!uufl{+!r zxh){8yXVI7Um*F1fj8rD#tSL|QfpyARNQSh{5?MmM(&7%V4q$LzfWV!{1d3gL(6Vs z{l9a9q@I*47D)uXPX%$*gCQcKQ;ko6dq1G{*^gzNj;@qyv{pRL;1-LMHhjRHx)##e zwE)+dNqEwr*_A6HJJo!nD>zq&IxtnAevyGceNvEWv-5MSi1n{@*|8xNPJOfLIY#oZ zc@^Y=${H{w84=8MYiPnX z(U4k?ze-=40WU7h$*@gzK))B_9=9-SiOjtjip=FSZ3@#N>Q+kY!5!$1=smT85rvZad&n*vJnPCi*B!=UQRid*l4Q-gqP z6^!X#1UmajuOu+}xJrCJ{7UaCZ*k9ezA2{jTshf;T>&H38Z(9(lStRIrKl~WvTv`N z1$A>-STv%FE@Qz8q%=IHHZ|Jc35Mq<6!H%UV-#+!sW(NUXmToWTLoQZe;v%`>%8(^f0 zTL1QNI6b@J62++T=%uWF&%OM&oegArn#41*jlAIxcStDC1!HBhnwI?qMXL4XDhJ=z zFwGt4+@D79P+`|q`rb#fVc@FNG6=a#|szyfZ>i^ts_8)Urs6+tGJmKHD*Qo}(s zQqN%Wc}M6qo}>wvbkamWuhXoZ$Nh{hR`)ar2$^phnz-&v=#(5fi=WX@7{%5@7xF*? zjqv2(svz zkgbk=qn~bc10q<6e8+y|^oGUR{`y0+I*X7=)Srv-rW`#nQS+IeXKGllDv6|py8%jX z4@Yf0vH161_IhN6maqXHfEJRX{0fegBw_rUA*;bjSIT>j+LqgTws@{>(~N*!b!V*b zw~BGBSUTul7r`~AHv;1ZMgVjK|T)6|Cz5Mh<$V6STn=sx+>c+mlNB$|D<7%VX zDe1(v!Y3w!i*O-3Db0$CO=pRlKq)>AB%fY09#*`Ar3F7JTX^j`6|}*79MtmbpqA!sCyx1C_#h;cCvgeLlcvng)uQLwume{ zPK2L6mi0u>ZjKhPy11jNHTe|za-q^R7rXkbr>bv+21o=1+QlPc<||4Qa1h2{z#t&#P@E+5x4 zi#{OGkNy;`6Nc_Ef~lPN^j8%H1e)%18Ywa_0IU)^$K{AQ$B4T}`t^dSzleR`bVfKB z6ez85_CfqZ5?lhe!)~5}mkyk|Fu!B{Gn?+zSm{U-=BQoLkCHI|ks}4eY1S!Sf|afw zKtXa{0g#0=znEZ40Z)h?>@pj6w8~l#>o*~Pkv#)jqd~Oc zH&H(p5c0l)Ww~^zACm;qI){FvEogI~gLEaQWluEUm4Nw7$_B`f@()Xx&zj8JM{mT2!{Wnq_!62Z18!ur zw+0vU4ZO8)a)Y&$(=0PnIUC(s4bt~I`d<@j>9&j5rnVyqQU2l|ay>cj z5wag1#kLw{Y(cHmdmPZBp;MVjw&wn#knhzZ>mZ-mXmf`n5L~LS*$QQ}nT;)-UMof@ zv(0rk=vN4~{LZi0;h3J#bWFa4_YDM8`MseehcL5lpAqh^zL&qCzYHA73ko405qTNm zcfx(Iw(rXZfPcE@UHEwpwMM@n0xg36gj7+_$`E>(=5Q;hJlil;Z{_C)tb)_3D)no6 ze4x8P6zQgDzwo}BQ&PRrUlc+1yg9zy$@FUI>qZ&_V9L7={a=D%dTSy%{#tlfXVnUV-hk^mFJFbLNOdVVT&zSwfg;QBd~Ud+sA$60*|a zXS$``N@30kqs=ni;ufqHWKtr;VGCs(pDF%b?F!AFtcohW8dN}6a8@vMM&l}9hvcu^ z#p>Y{7xwmp@#~8s`j`oxZ9^+dR}7l3J`I0C%Zb|H5kejPn(yC76U^d$5C5hfHCpwT zr~yy(X1j$MxxCZjG^C*ahJ#9A)VY2|VO35CGm`4vXT0(=Hq_95bU>(>WMSl_5N6MV ze?{VV`7{XfSf>lZ_wPnZ(q)vCUTa?hA)sty3JERRLoU*IK2}&^Ew4C z)tH^Xd*#ku-MUovt~c{3uZ8e`chWu%7EdR~EFy|U#mGng%kT!o?K89NC0eeB*e(zJMJZ~88eOT(0tOx9EL_Z}+Fbp9tXtgD7Z}y2T}n0Qi(!}U$#2Fc zWEFjzor2V=(UDvO2UYQHicH~sHB)ee`<>apEQ@y7eea#{kO&5IQMQ#aE$PZqC{R$Y zhGgH+AR_C`&Q*LWGG2dmiy>7-^(=a6;WB+XeqZeHtIy!=29!36N!D8%UAe!zhL)v- zLL+iUubcR7{sVA&>%yf3RbJRB_)Cl9)-usW25s%O_AOq>3V)OT$Mf^PNI8)~;P$)mSnUh(^e zq!Ep`RC8Ob_7!SLP+*SG{C@nN&YF+uE2^JL3sT}oKMgG4tz$lQ$#z^iuY;+QfJwLV z;cgzW!L9H^X+4@k)eZUl2oBVZmN4vYc<}}q7(UAdI*GuJL%~|qNG}NtC4*HWsf3Hn ztE2XGOINjkgN8_4amkbKke*GOU#6c_tR9#vYxzvoRaPCiC53I0&~cLLTr@B6mph|b zQEg)#x}9=ozo`8h$#DoonDuFSV>kY2OpA@|>^(2ui*<1JaJ6)I#pBA(F2bH8naYM& z?mgy_XXHtG>tYWIwOI5ltQ|?n3p~*$Tuu<4?1sWroyox0qlsr8YckfUaej77s`_c` ztx-_ZA?^*Ry2HG56`NY0p!9`Bgw#vR5H``8UrK9xpLW~0v%tZ&ME&s135l(A2bUyX z@RUZ0jr$7HT`?*j;)m<~siSHsQDa-PpJr?NHp%v4(-eg#9P2P|Xd-sU@O*aM??Rq4 zbNbO6RrZcD$$vevZI?Xgl9EX0NUTNG>L82~UK_Jnxk{!}g*=#~qw~7ua2pTREH&!d z+l)SovVDko%_1XR7N#f#xN>JG8Zq;<1u&}Hg?cX`8IMC`WRCn=^Nwy-0hG7@zWsVu zp+L}HoQ)PUqI*lya1cA2u4rHZSpOY)HC9!|5K~k8R#9C?O*?0)KGdc45+m`1B2zt< zEXa9jC9_hvMC^l~InI9WKVv;uX8T?Yk6@3-m+^gu0HOq2@o8c~#T7ZEL4gy2*)=QJ z*<=P^!-NVlp5Vu5A|$r~VW{uK(W@0D3kSfkB9BJgI(6B0SmMFv0sm*IFZ#{@A*CPM zu-rQYoa$O%*a>1^?mYNLG@Cb%Ji=9hj`ESnb+~zF2sLb7EWFeNy2^2!3+_O5mnq#J z*w58dCf|(pvIl|mWi4NmdFKVu>y%8%k{IQ$k(AnqYekK_Gu57TTeH*s>9tocf>;X0 zpo!6+0<1b_bawG0YxY)9uZvu=l$W{1C)$(KILR;-f+$>tElbf+1XK&b(YO}JF)1hI zfkn|na~+|xZ`2tV6|mlx;UQwzgLZMkyHC5Y^tT8fCt*_5Y?wTSRu>KS>!)ET?9{AV z=k_e3h3Slm!6qYN1xuc3m1$uN13)hxhi=MqGTPBbuOZ=UveUx*T|}NGc7wr|5YY%J zFog0MFumHdWn~xH&)}kgF6*MwY-)*|v#0pNf23EX=LT(u6h=60%Y-3qD zcOo8?P0B-fX^ZLNMqPa-0kb*@Rf^wWOinLaEJd|lU0-KrRcvC0o2+b=CujzVkc-r+pmnC094 z1woMD3nFJ^#rT~K2D@E`DklCz_s?^sx(a`((-zs>n>wOV{1g4~o+!jqL(v;!OjNjH z6wUabW{|fc3Ig%t3emRc1y`dZX=lMHLVQfn%QS0C6@Pb!FEu#;8RK{Wf z*;(}&H@@Q!4&!1HgwV^SDDxtSE@>>Ydg|{+#Imos7ecDitOyinT3wP7<@EOQ)g~us zW~=M?TU1}Jywn%4M?3ekmJyQsD0|xf?yh-KgLYy|Lni61+QgHMRFC#lL$s8Rn!khZ zTPreGKC^mVYPhA$8cZFRWtUb(TXue|mgQs2xNnDHn)$xgBpM1ZF^d?*9GmG^WUD%? zzpl6(pMCwr#x7Bj`5jSCLH+ey2lF_`S4e@EwbB3Q$eZ)a3 z>X9+=%LPTrdLriJY+%zkHV%>6**m;a+0XBET&oIeIKY(1mN*U!vn>-G=48=d-Obc1(qD!!BOrps3$R3id! z6=k`wZ{vr|8S9=CS9tK26gBd@Zi(0H`s|Q+pPm%6z1OlGb6uy(I&9kFwMR#!oUD<7lkqI->-O+fhv6Wy3({MS@6gsHKeP&>IbH4IWN#vl3ZOq6)+~?Y@dC?BJ1)$a11amWZz8^wCuq8f(17D_<-K zG(kc#*FU?k{@i|k(j&R1oK>WRF7ex(H(Ffb!^?f)pt`ai^_Phb?C?2wO=$k!w;{oY zJIsAEiyhVoPqByY_A0@RQ5+do9qw~x$7`~M8i#S_(k=T=Z9)RJT+I^zT;0C*qM-H% z4`LqC&x_Z0kb1PYaF>zhSZ}JpMx=p%h0^yuwEmT=i77|!LAvohCsHcl?~G*Z{JXL0 zoc8<^$c|SwchV4Sj%Y({raH&as9wd{x0y~62l?3%CztNM)?67tMuM;So`qFOuEm#i z%tkv>vL{{3hp6N#OK(g~a&kWOW`DsfnLV(|ElDofZrhY@-&dt@4JTgXn*}cynUhwl ztfW7ep`Yys5UhHUk-IosBEM^wXw#e>e$PvMHJv@B={~=@Kl2UK93%fdZ%VRI`;@=jFkHDZ7`+WA zCROFlyN@Qi-Zw3?tZ9EMqhTSBDR&p9e54{X*I)$ttpO-LOp}@PladwP8^h4`iG#_e zb>P`KSIDoVFik}*&AL8t^S}Wl3Dv-E(VGgf(ug&wm^~?Xxru2Q(W0!K%uslXH7b*G zmXZXekx>`b?2Nrc6mHH3*lbr&~|yrWwqZIJonkqkKC=5KhJjej5cMG zlSKn*`D%4#5_et05kX~6^v&stN6_atA)ZvUcCd`0Sm?~SwQm%^CMJ8$e{yG(^b$rHt(dDv@h&R!ZKLRkC{pDUMQ~=n zUq+tZTsvZfBg-J{0*8C}LS1lz!{uW}d2N`QdA^0%-slo1Uea}UK$~hdt@5#PnusTf zq*v?5Bl+JABi*OlZfm65G`QbfC5Y9bWWCV0XxSx03n#%!iXc zAc#xPHLCN6b$pTRC}qoB#Mch_ zN$1~W*FCp1ktkf`f^XakE2kH!incHp18KzhaXn}v$hZ`30)$!WST{4F)2||R;RY*!<{KKpIApfb_%3!|N=9!pS;>Dl zrf>16-k20m53i%-fQV=D@vg)5q>1`W1$Afv``KcHDiBNUdY5>x{IJqTR+qhiB*EF& zrk31(kny)%eGj#YpHxlBCm44qFOik&ESCm0NIHJ%ek6_z(!4@w1xcrR#F-@U8Zl)U zp(w`Z#x5x~UIwAyLVZ7HK9jn6WTK9e9B95oYK2W67UE7j(wy3K@q&#vbvFtFVMy6X zcZ64q?rZ6oV*>OrM2z2ewwg1(OzuG2)Xk*+NLx)u4*tpx{h_Cy$@QNyA4ku`ahd%nk1c*=!V$krRk_ZyGBHygwAF|q=Ajo^~H@G zj9xPjW&%cXuefeH@ww#u{PSvXRb^`fzE%-7~ znLeHgvxr_{txs;p<;8tq?!Y=qaiaU?my^`iYj8fswdaMG8K(A+zgt**g&&^FJ5w0Z zm33Bu#T!kin$V6yBvpP+Ae9w@GKGlFWRV{>_D5%B;$tw#9m}@5<5!$Ku)?Gq)hYmC zMhdwtELRJeT@HQ@_5+$2MoJiJ*`1lyEZcHtJ-3_I8u+Ky!+1HFJj@q1B@6YJj?HFO zDgyO!V@OU6_60*=xv!(tM`y8!y2a=ba!jO25;oly>OpXv&P^$t|LsMANIEABZ zd(Ttjb|1S52E@I@<2}n^cj$e++HIK%)6VOoJS{drXUqTH<%J|{S~G3DY65AX|eXS%zF!E+#CSO`&QI>2>jHLIM+PI;P5LaGW}5;NX{$b)R&V zeOk@(kE63#lN5%+=qur}#NC}8?pj!Q`k`mn)AMx|klg#9LlsFQaZFqXvYpAanD*@C z6WuXv-TYHKb(`NN$m=Vd!KdtUIBM6et8=hrJ5nh}?1;+^a#*V{&Hxp5=;A1oFP%Pc zh^>_VrS+HVky`2c=QUmsESNFNvH-DYaW8k6UfOiGYcN65jhC@Key|f>J>Op{Zh@6T zKZOnpm-vkYqsf|lsyQ`OA&{d)y2y=8UWr3}>DT@o`3H8^J(H$r>I7yykPQXQ$XxZ4 zg(#HV&1h)d<~KpkmVBKj-=&ZrWp+i@OJBnl8rdsn?x3WExv+<1GOXm-U~-As^?NIt1!TZbuHtM!_mR5vdYmf387BHZR|U)b zmgMfTzP$?%Wm{8I#wJksl-!JxXt`Vsl@HVMu%QqaXg>>H(&C`I`Ogv~Lf6R)-W)S4-Mp9S+X-LV9kaV4Zal+dsg9dlP5~z`dG+pvUetJ0r_i z(XypFO~Rbkw2(3{IyF_*UtWIZ18%g54j6W!rSQHENNW^RgE}-mj0G<(xC<7U@hn5U z%#YM(<;Qq?AyKiV1Ya}`J(NE584nq^IrQOSCJRBa6_)fdUcb%p3&~08&62ZRf1x=y z2O>GxMd6K4BBx#QiSYyE`o?=ACS9xw`LF6+-koRDU1s-iqM4a+#H3E>ca75s4Xc2=bmGwfh*dlmz}T{x(of3tS(8Q^=F zDY$B-D2*cd)DS3vJ2WLsE;FEq8v7L&V#zuNgma#e5iwrH!QMmQqqembb!3 z)3Sl3tTzpq`Xvsq4={f2v)skwz?aLsvc?pP_f08ofyfTC81LQ9-<+Vy;w5I1aJt_xIzViN#(|c6Zw{N4$j5XT!D4& z<%(Z{y;$Cj{KLHyg%8H>bc7Y6)%~kQoA{8Sy>w1qTx~Mqw@T+)`>43khc;6v?D;Jj zT@3?+XYAM@DNIm3#1UA6Gvm{=Z}NcWi%NrL?A1TbEu7k{<`UEsi~P1gy}yRY2f;( z*@feso;NEp9WhZ+Xv-{3Cjs6o?4igR>gT}mSZy=OlE6-y zz0@Cd^+lt)NSh*RL!L|vHdtyC0it&!Y$0jqsc{K|atJYw$lXaD{sExZUV97K$j zKvmA8TQ3_t?p2$)&Ce5J5yN3xd*Y_#{lVozShqH`OQnD)+bvEa@H)cBOiYwc*$|w2 z1|pbHdk{|*B9!9TtLRs?e#3`^0$mANJJLilB?GHp40bmPxslkjDBZbNZVN}YEY@aM zCrV<8)2ZNSNdgj;rx7Xy!j5<$sAS7 z>y!?|sgTPC=5x{_{5v#m03*n9D=H(~(BbU3EmK_3k13ca&Ot&7Z&nGT3O?>B;Fps9~y=UMv80SNr{dIN|l+n#= z1cgsd;}CWIg#OtAox4tUuWh_{#tLDg8v*rKHWU(FQih|`p#H5Ml(lAWJ1tdc=;g(B z5Ro)T%R;H@_`&c0Se;&A#OL*Wz|Jt%%Dl>|FD$k z+Ig8CML}kN{W1lX``nYnStCX#Jt)fb{FyMT`{K7k=J5AdQhI&5@O#lVj^Wt5uVYj_ zOty@C%;P+k2S+;VF0K{U;;3--&cE&l>46FGcS|t&&>5@UYH~^78GG8=xOw;x#rB8GM*?fC z0rh0Cv4)yA>3Ilm8C1lDOzMYFYj@>}wH-Uf z@E1h=fcP0T;hnlZ;hcP_v@hwidZil{o6n}zu#5Xdx7zRF{ zRA||1AF11b7xYc)GGQ3@-6qLR2{I}@6E|NYCYl({{A!FN-Ux~9DUA>om2Gnf$qvou zJLSO`N-DT%G8eO#;ipb?x_MyKp=A|i$Fdw0{Ig-WOH40SZHIWPOg`a(C7v;g=Iw}u z6&{B2>MFl<5<}B12x5X8?A(zCU-twAtsht;jHH;cXT!j$bx$5Ug8S>=SfkMa&Og9Q@q2=1HOdvy~j^4l$~B#BdLn# zt@(XJkq!w{|G453dq1e|3N{cU{h{WYmg^hWOL3n$XRD-0-+Ko>ryUDPn1T8@J#n8% zd%iq3qVB{5CDv9oj7DCy3Vu!S(BiCC`*4Hz6^X>jnphGjK8HTd5jUMGWmWG(^4307 zGoVumpZ={=SS(|%%y}D%+djFh8@c8Z3!Zor!7Z2VwqshF{LWE(K`wAr3h&4WWD4o~ ztMbR9T} zvapwFrvBz+HK48!cG}ih%2gEuamS)P2!Q<)l0rV3<5tQNqlP{%D_Pa;1D>ulz=ffC zep?>Eg%8EhsVeQ($hq+I6@%phGX>L68P!HZt+1zFrdIV*t(Tt6-XCK`lPwc(+u}UzKTa zD#RoB;61u6uM?Xk?H;0?UVwXVWi`A-#q0F`r_#vt{rZj-da1v{K3=Pq?6LG6cj@I zVf7s~E*ILXG-75Bauu_$X1_d-Uh>$Gh~i1cs}9=s1GT@eEa54NdfD@oC4Bcd$?vw{6BI8E7Mhfd_0xrJENLIZwgh&k$=veSYR0Fk`Ok5uZ0g^mh|f z9pj%NB|^@0XJcsWg}@^B^2~cf$F%X>!5+dL4@JWvZ4Ea;i>5N8-e0L0#ftd^7Fu{k ziaJL2TW%+=EonG}3AQLK!SZV-+fRRLs1@{qhql&~;b9=LTamO64PK+bj@WX5&uCe) zwh~E1GclJ6SAPw5f$-hq=3bQ!Iv;L5FU)?3ZANPKr=iD}KANEpyDu%=aji|8vrcpi zR!9dA6Gu#6Y|YA-v8L!|bjTz$NFvGlpj#y9&G09HJ=(*g!<*nE3t6|x^wHFeg3MJ- zsGe>&&jzBXw|yWw3(KKq)0(BAEK^Vr*J3YN@XWW3EUA!9LtCS7;6+{pY4i87L9IM2 zB3NjhSBCNYq8{e*;AL13`vvgX1;PP=YN~35t$~Kfvj14^2h#kiX%f{8+2@ekRS*eB zY&uFoSnI%~rnYqy{N(qBQQ|o4Q(%bEi+qzr&2ToJP4JOO7Dx?i4-D^Dq(&%5`(0@! zn3XLrK?LQ=WU~*ok}IHYXHQAl1}Wxd26N6kcB`SP8HXQ+d4)V!hUsDDZI&G=m+8wS`(}q~#AI|5$oh@x*>3Dli=H$(% zTZD&f5Drw=Htec~p_5mUqFpuQ(2mgK81ryGjs*WS4%e8(M6Xg)79d;0%&AendBGrYB#8DekHE zflA2TgmL13n}|efWk2B=b#px8USVqbDNQ*YUdSzQ)+%e%v-nZrltJXyc&s}R$gScc zOD{@+NnbFusfTr~eCHwx#HcGRC-x5&+g&uS^Q>_%>M06;9-}D4!iL0yQ-;XnD9?K7 z>BjJ`xp&`>?Ql~>T5w}Mn@~zcXsBSNS%^G;oj-WzunAF^+h6w|8P1X^U>{#e`+Zcq zLkky7mNn9=F@S7?Te?OZaJ6h*U+)xRc1Hzc(xaXzS_FLBK>-EEmAP~7LC+*Qk_^+H zMUi+N3fJV2-Pm4=!G+2w=EW3ixXF**-{+9}X>%I6Fw$}5nA*hMpdHDc1xI2T`ignX zT9PT@rD{TxpEHAuX^+>9+6^A?uWXVcPmR3b`urF9_)2$es zopVzw!4pkroCDM9I}UZ%D|5OL*%Mi4ZJ8C~?^hW>qXm{IWLq#l85x$Qr-n?-8T8L$;kC(uhfy2Ey9W=*Lab`T{io~q^s~#G)nEM$wwnq6G zix#?Tg6VFb@w`ZuH#{7sb;Sqh9AnUama16$GM@fKRxwIDj_TS!dW*YmWN&eTYGf4f zdqG1~kjO*&3j#clb3PG6Mx4^p7Z`isn?0GnS}%8o$oHT%1}oU+@pIRmO4#CmQt~ZaW6|dmK{+uP@_1oMIU_#LQ$k|CnVaEe zl(U+jSfI-JY9mCE*uf&rV`s4pKz9^(GQ)pQbS#|2Xyq4gmfOr8rG>#yN-Uc)2aqbZ z;0!(=1qeyg_M$pW&Oty6|ntY zW(#5uM-W#j3B$a`3pH7Of@2PStVVJ_`M)eMmcdvE+2C6dZC1JCO9f<2P#3&v=MYM8fWlQPgO6d z6dxFp?T`|nLL(C5lnJh1^D(xeAeSC@dSS?drg>@lmFx9UVajgyKv|c-Lw%@`DUP+) zZ43C(_CH%jy)GDTy6D)|P=kA;1ETTUg(K&o*QKvFZF3P0LTbCUh^uwEN{Elq5eK@@ zhD>WQUk@oodot;^40#f^2Chz>Q38(6?{wF z%XGw}XcOoNFii=p!Ha*K98zT~|Df@e&`8iDa~>_?qY~|B)^ZBpdXQy3U3(th_@E!X z@R+vMjlq{o&_Sp_ z1{0L&YuP~R{jw{953cYrc$_v5`Gswmza=|qUmzGiYJQ#H+T#Yhh)X}t(Xu5@w3PlT^rRxUo%IpVMp4pPlac_4<^|#&FE71{&di_qvoIleK zW*p^%5>;}x1-0+sGGSI%XNozUm$H^3!Bv=jB2?<^||%-_?6N;+I;gl? z$9n>m&IX7&mL3e1Z@kA?dkWVQj|b@abR8#SNWR6d zxe?5G2fi^0qbkR9Go@Bz5(swDxJIULYFnp{Tj6_}eUQz%$%iog&#m|6Pq#=M@P zqmZ%NTa-2m!v$Yy)id>9t)JxGgECXndpx=+|Vl%qpsyx~G#DsFvJ^2HxasJ3y`cHIs6 zkKP`t4NurwQyil|#W{bepL=jOlEZ|qER%>vE=XQG3l*&p_205B3`FJu4&hQL$i8a9 z8K!*hl4{ISK@NGPNEWA*ZBINoaYXzz;ByCVMb)RQL)I``w7qMn|D2Q)n0moCt1=n~ z$zBCSRNc016?)pPr!>6su1PPH`Oi-)?V2so!Ux-1jw#`qIUv4+=bhU#yg-|{khX~@ zkZDd-yiV`>%%y3j!_C5y@IzXd3(8MI&>aF3t9|3>dL2AOMc79{+1TlopW{*~du)@B zIDA;Z*b154hRciNH50<`{i`GM*nzd@#^R~0Hc>io^%n+a;QzcP20a!5mxs~@Ts&fP z_xZ_l=W@!F9vpY8ED2LE5||^25+vtLKW8es&trID*pcvBdRGZftjn-^aB}XKk!Z;Q zJ3X+o?sDUA#gQ(7I3U`4Ji_$+X}NTk1Kqc}O7<9&te!rUTXl~Gjk{q@?cLEK8z)EfMl#P`Ay>XA)KL_~@CASH^-@Qi za()s(fu9%-_qBTn^|h$Q=h+lYbd31D@|a2r;JZl067AohtXC(}yOpLiM^P&nvIpUi zvIuyDhJNM+LinB&{t$EKFbtOFx98MeVFrwI4*`kg4!dN^@ENO`I6$CLOn|Z}PfT)} zIQj6t1`}2I%M;pH2l&^6;TP!Ip3zji0!!tp?A?o>(;dBM+R!pmof^!zlC&U)Sh8xi zgx^;yZ-uplvZ!JM71ol2gd67;=zzN*mZ00;$aeyblim%xsU7kJ`8ml9*YR9sKI=kH zS;u^u$uNQ22n8QkVT1e%T5BywqQlvr4Z}@6Bu#H;9*e(-X>+v%Q)26>6U9#4NqIIq z8l+#rS)%VT0mt^%scwrBZ!X$2>g#+m6PONixQp=WrXV@4uVu2bw_$Xx!Qrm@<%X6Y z2)0iVe<~DH4}SfVi#%8ZCMrBcW8!ci(`6CXr;My9DZ*Q+Px>+MC>JhyIdFE|$KLa~ z?$~HN57TphVFs_y>QzQEme2hUptt>7sHE;8y5i3DR+1 z?4F$>g5yZ60O@7kO9z@0^63QHKQ~kT0wrX55Pn`$1kYMK4Pc4L%JXdjkLJ3s& zwEy0{eSZWFkMsxjJk|tyjzj9g`UcL2=`5#zTa%@714)F~0Sd~A%Imn_(eLT!9!3g< z67lIKU60COOu0}5Dkj$Gy?+)&$MJ2ksC&_>HV_c5A+Augft=8?mTZLKlLuHDGjGh@ zjlL_AMxF9y9Z}FSV9?WO`wbTs*UO1`q@*h-c19h&|3@4W? zBaN;@vncSoqfur4y?j-;As

9NlHRS(Jm-dd=16r^~74{V+};w-fkF4t%=j;#;)vkl6({(0a8W>cw5Z2_dK{0PC*1=`zh)KuVLp zjA@o;4ZNae{dqF`IPN+ikFwPj1|x%AJ5FN|Cu-5DF-_+;4_fI*zn78kxtoyL?u0uv z4jk#172@;m7VM)S8J_&MsiFx6GQIDWw)xO8;D9*(v62Ue7(}0R;jyP1ENJb% z7pMgGa$LcGNSxkraQ!V^?@emvHKU8eYOIv9=Nlo6LS{6PAJgDy&}Jo#Clc0tqTQ1r z)eCAzms8CXEnP?l4iUd#p%cSuDtJ@sSi97)Xq)nKiM$G($V(qOFP&3*gJX{-*sLb; z6w?cexA|}VIZY@rw1JOxtuSM6ng=Gq?E?*7hZ87+g>I#48bz;t2y zGcOO>f5aKLZ;LUDN7FBZMeQ?bJs-)@w)3^>;yz(SZ+B_g9DTebE2;+~8&*)t9;7KQ z5|LVAQ@V!Hk8N8tw(e^fbM3p)DkBA$X2p%q+xhvIso6F%qwv!+snht?;YG;L%j7RH zWqMrrchQcUXHBNSRtZ4&_gntM>$Uy=w9TCoEq=2frEe*`*~>Rc`I!9G$cKHhJ(}E! z+e&_mx+V!jJb2zyF~IXDg$O!e2Ink_tkB6zjAZvGFcn@&^|8ZWandthf4KZ}fQwaB z5H4*hWI^tdnu<#m)5>jC6CiQ~i|euP%2#ipm~v8>6@u`V{=AuZv38S z%I`E(+Fn1b;e_AQ-Lr}jxNPK6V3>PNLG&a1=&l_6VNUdI8&rdm@oemsOZ(Av(iU{_ zJL9#P{Ax;e^z{K<+jt>%qI=FRpk>KpN#67fmo%Mz9L(x*PVAT<6{&Np3ZXVL89f`O zKz{fU+p^|wX|vqud96(&z}v?g2yfr*`qu43u2)?dxXG4WjoF0cS#BOjbhW*={!hpb z2(KcWc-N93g_6ui=PKT;b%h8!IZ5bk*H{tixT$CIOI}oii5aU!voi|X7h`_QIF%8? z=)G|6e7Bo^*iCr;98t-L!^tI5GxPGQFt(EZajbU-0b8H&fpb>sUQY(4bO=E#1h9p& zhy}~ku@DPysa}wXtBVR_jj-6%_Lo>zwe$0WH;i?Iay&J@f_W(bNfl5ew?XgB#HP-3 zQ?1;0Fl_vZs%Um8r+Ur(>y^FJ1(RlBl-I(mu~AO^3jZv9Z|p2LtA-Zb92(N!2*c0u z8o8cslo=ni?4fao(X>z7txF8I^r_j;%-f0(6fEtj1l9D8oSvB8Urob&7KEAdlB^jArgv9XYs4Q$%j&D9mJ5b@K9 zc7{1lU_*F#UD)$_7Q>S6ve@=*IA4w4Qot3-*`p`+cZcH1wnI^=`~F$)G1}| zJ+2c}o4+ymnPk0*4Jp=`5zM_nGABXtIA|=};!JBe1MZ{pBvFt`kfhEU4b4(WXZ$)} zBG&98?)O~S+dwz2RsI*-Muj;gBg)29?2rr!-Txl1>zVK%@!f- z(2WWtrgY)RU%e08Y;t9MgI>5G8WJlp6z zW0Ff$xRq5k1evt5^@O`?F@Ub*d~(iRIyDW?o$?6X8B$VyIK4-VC(_Lfd5bYNey|x2`@XwtUcG2?Hg284RmKP z;ullQ5m}xNYd!9kX|g$N3i&$>c^W>t&fz!h<;dY+u^s{!7b84HMy1CZdTTI{PDR8xNpt%_63~aA~mtx$JEL(@qNO8&b&x+~$Mf`GAI3{A| z%`z-dkdo%B7eCraXa(w}0#(obxXy4hUXqGr*vi^CI!O6z4mCe7)SaL+cf26_j1ce0 zR>*|*SvQ4B~%8#_Pru!=oCh| zF#+=KBFa^E?SgKqmO4gYxLcS|vk^M#xmzHD^Xo243)`aoG#GluC*h|RGu@dK8tzZm zdy_Ex#1tNgDxZu^X2%yZ>ZpZ z`v7B=)WEy>LuAI~A%(&s+e{(O_U@YXKHhPrM9IxXV;kB`H{0n&@rThJEl868MZ+!EyKl|zXKYjG|o|2^qwj)u6lZmoJF#T$HKELpUR9B!2(Z*q^Ol$^70vf+AP!dTSbkTOrcQrP`qbwZ7>Us=YoY)6 zAkx#=Nl_Koryy-4aWC4O;djJM%Oo-lGGl*6zO>^uMe~sdBTElAdDJU^L~NcDtI{lW zTTH`RxRzsdp{9l9lI__H$N3vqx5xQS2`fR&xcQDt$~w7CbaPiCj~`y2HMU^k+GtY< z6E=g80;0@isL5iah)aB3lUr(BncHGPg${Y;deHCI@SgV6NGAr1_=-tQk#OYH0&*tx zlkqf;fZj?fwX|7ysf*PY6{E*zE}=DNV%`oR9LJ|iv~9-9tBc|)G0-`DOXXTFXrO^)8Nsk zRs1dHiQWGgo&c@KsN$;!S?576XOKhR^!FPgf2-Nau1!y#xG}((lvF}OuUgA>sE4Y* zzF~^g5!SKA`Pd?V_oPYIlP2f~G&?QM&h8fOODGT8vMfsxx&Kjq{gJpIjExz$cO~VT zSvoI7%&w>m5p|9QCpCycCQsFzmhsxW{DQTMoc6;8veIGYwxL5T12b$VVynAB#W#Jz zFCiBxloSYz$RAyiLkw}9-lBCuIoBfinn)|is<7&}IYa@li)fQ=BWo+6dAn+kmb{`b z_z2NA$S6ao8R2YyeU0T|2EnhY0zviV^>9TVPc*xjv}rW6@g)`?vv*7?VfbS|58HN{ zC|tWIx>Nhje=G68`q$;pr-N6EJoORw6O4*#vsboY!d1fIT!bIv1@@ah(2sKVkMdWR z-YgOKQ@3>X29JlAvS=fmx7qEPRQ$eb0=-i}BT>WIbVBWppfw=w%F#Pv3n%5vE?ltjNEM(EL0Dn zcTAqEN?};R1Fd1_1xAr{C>kpS!iTPZhlIt#-Ya<15&k6_N@Kj(9}@rJGDN3lIb>6TfKiOR{PP}9|xUi}hb6dHJeW=Z#^bRWH2 zRIWs9y)k7lElsa_Gc3$oF+v+=R?^MNtEJ%^?l^lnY3NcDbH19uWEnRpEig4|Q?B=+ zGYq(I0v%l6j71h1;CT6J;+p#5?CTe=x1mmy2Goj&%{7jdDw%#(c5`KoWvJ*MGVk@} zE;~?*j2c5{q$ji68lg8ia^1+Lwm1Y>ytL-TiUw6%G8%}i=j%^&re3NpLywahz--L0 zfmeaCYfA;0d=n%#EBmM>l3u>>dAb53t!bUEaeTHz>zp$_n?xMm)LyP8FX8pGDA};` zhzN>?ex*NF>_sHg_a%TBoyZ!KEfZ&yldpfS8e*ya9dM3}Sub~@fb&LlgORT{ogp@lcRTUd5F zy2>XAV(gG6sBj`WI8CMcxlRb~$^+3^D;wabd%Z5px7J4TLI}@8XSxoy1^<_2xw2f+ z&1)C@#3*(+90LRg78RZxYr|h8 z5Z|ekOHKYJ1I9ZYi$ZR6nGuuOY>2dzVMv=`Cc?(`-Fb;bqj8lJ|F)bgD3w0B{w%}e z)v7HoHfH91bAFnIR;+ApV=l6yP*@hKV>c=^c}{U+n3bql-tk{AIqW2C*d#z(dYitx zp60qaL=wq@y#;3$3WOktXr2mN|0cT;w8ty`LdTGtcyAB?2+ai~qlFqd;%550U%?of zInV9Ct6oSVjE8Cd#*?&-YSaU(sKvsz=wnjigFhQTf;dA{pmjFYFIdSkQAM%ZM}<)& zSgfATxR-|)y>#RSEqXcFj8U*F1rE?EmVCn@mc=Qd0 z;J#HHA|Fi1>_9b#%Ow4TVy*-I0guQMCYh)8#g6MVllL3*J8@4g^S*hAw!T$yr?LuS zRFeAJ$3z_C*s!%_pJwx5Gp#TD68A6f-zhE1J~b@D6^#DXflxZw<3e*Us7Nd$UY=9(eEgH;LC@CDyEo)Tb=q zH8`k@U+<7AjKcSVXXRPx%jK=M>QmJ}c$I|;{G^aGYD~xG5|ceEAQZJZ*Q)2r#xBu@`Bz-4}JVcJe z?hE-s$>h_3t0@=9QG@u6_yS&Td)9~)%MT;`t|;8Ns}-Vm38oWpt>Dpyd^Fw;2WQz6 z&lES+T294iN~qtS?rQfb&eV5sf&iqj7^aH^N&~G=qnm@qKz{fuQ}-Cr(IFw|H($*w zbcX4|aH>g4(qxSn{rawhA~|ebi-97>{SEz6MKKYMlJ)z7Uj5?Mlp|IdNX4?oXKeH$W(R9>A#ksYQzCUN=gBP%{ zBcJgzhaHV&-0#a~jM!*$aC3i$ye6hArmYhT+agnv-(s?bD7GGxD`3HOr0f&U*pVhV ze|zK0Vnum$jk-SY{fNmhcXLRU(3v*lo;(5Ped1>oQ7Y|2^y@{of9IsD23Fk54Q4Mr z;Ja8f{C**iX$3=pGN?30{ArNrv&y+62IQ|fa742wG1)7r(9Xr!-y#V>x~ zL3`mVMeB@IiEF4Tyx)}Ba($WVDC*{9H*lb@C|5F-4@5EJ1q(*=;ztP9Q~H)Er`x^xrs?Uw0{f#kr7yqE?o2GeI^ zwY#T1f$tM{LlHpmvZ{qV&IS|RbggB+-m_rl`~G&NNNSPq$jtBs2zml*-L49)Nyy*> zHi&us$dd_EQiTGm7*JPs8*lFI+zqAMW$;!F?}-aXI!%h))O86miH?z%qG#&-4~wPB zQ!6v6#HZ^PSYOgb_}TzWkxQr^kgbMNS!N+B{+LcjHFoW=fUZpf^0}&`LmcLfJiZ1| zBQ5{Ff^#+~Nw0}{-%h~$@A6H%cIkI=Y61c$3QK)XkG0X{dFv}z9K3JHpQT^Lz~9&1 z(W@=5*6S?xT(Vuxd~{l36=!XErjoRxi*-;(91JI8~ZQ_#Urh{J<5@M!yITB9> zj&E$$a-%W6OlATR;|m>iKKv?nkFgUT<PN?$8H)A>NQ2y=ymXyw*~7ZnkU%+a0?&5Ocr+nvNT_3$uLD_a1VNkBUNmD7!tzG$ zcFCZfzS^gD$6`0c)Yn6MJ*Ibo9`bVd@X-s_5~Bf*=7J^JN#0z<;W*VU-0$(J3687N z_rHlB5e?bsql><0AY-SgDOUqiYnX63E=*+a(ro!#bbtKpTJE?(&}}iDymYC*BlBLO zTopF-cwp}`(-B8z&CmkZHqJ)TkIusOJnr6^hH%A^d5_e%pBMYl3Ako=)KsBA4_ngP8qZs49#jkNG>2UOnq)*Vq|h9mO>7TueL22DAseaN}rH0VzzNyW#um!+XpO}G;r(=l>Cx!r$J`g=FI zAW5751T2~O)R*4tsdWYWrHXkFgXZ5I>8mZ&NUpjRTe?+ad-P_Xr%J;mKf1L|1OEEf zSVZ!BZ$YA}<^RC)<87u>tO?y1n-vt6vzClPF`6 z4SX8j{1|Q@@6+z<*Sf!}b18r9mj~Cwe-qymYe89QYZWzS+SKSdM~!OaW5m#(^zgw+ zAH97vMPIr`&bxo#0pbs`?J_$ug3h9PZ;D@Y>6ZzzN0Z|TGhO#Bf4lZIs%|W^p8(3hww?T2~xfaB>xQhEvI@Kb&SAzc2EX* zd!Ei~(JcPEJDnx72p$4_+t{4lt(+U zln|MQTLb#gl-hQ?$wRe+*gy=Wn~Tn{d;Ayfi)6?2 zw0qx;esI$`9N?Z+_3y34aq?bBYe6X<5TxX{oLrTc>jCk>IsNieM6_~R<-Ex*Gy zWfKbnHV>8rCbI(NdT!U@ycvE2CuX{5I>gNA@#zgY>-RXdKLyuUOhG25ieO>-Am7|4 zf5G)ubM#mp;>CZSA*p;t<7Ca1zB2p<7)xaTSqF{DR_?b@{;WrW*+7(-wzA^h?E^H2p zKa*CR6i9N7xLM*=)UU(kR7IpY1fk!%fP*jGw-8p$;P&jDw)*7ET0`+(>MZGJ4q7EN zza)$?BAT+L3_W7p>%tk&pCfzbSMoT0dA>czevQ;`Dux%r4z``#6yBRciQaXBv zFcQTz{lUWR95prB%T7TyhBAiklp&!m!38)}wXLdyRCu$%csWj{)|OxDeX|lvf_`9A z4JKi~WW2&&Wk*Nt7CQ0C85$j#d42DlA&?U1-|S;e;=S-k2Mjit&k^T6Huum(fb6)} z!{l9;s}=5gME_=4ISlz#$R=|{B%aNQJ_)(IZcPh5c}V;Dg`6&4)in-5K{Y;VXmTwF zOts46uZVU-)E0o;f=XK%B43Nc{_gO{m(*Mn7kR|x_vH;b-#VRH-&5W@ zo1M+TN?~F#MJGY;wtQe$$xrK}#*Y4zZ5;|luK;r@t3yt+q0!L>Gs|zTPAQ?$P}B4x zxP^Lc5kapqsyM|@_45p`LiIfSNuD8E^sNKx)Y0c_2)9BsbLij33)m2VL!8SGVWl3u zS^vR#o==ynyONEkYx}sJAq$TxMg>tR+l(=IdWY6e^FNf)55Bkop5eOqE*iJeBeX!( zr)%0HOfX;jqdJsqpsfrE{C54jnHj$qGRMavrmoD!NWhmJdakb1R-Q#9n(sVAHIpGDA2nK;JR%`U+ARC2X}5%rxsEHwjT`_Fu|bR)}@Oz3&HKwj=AV|~rQ2x{M? zI&27SXF?wr%|u8_#59a;V}%q3#d`RQ@oEi7Z_o6%mMM^o9^o(zJgm=R-i+=^Is5Zi z)O#T74Riwf1y#Q&8Uu5wC5XhlO6()QOh%fM{Zm{6>M84aZ`yf zbWHu_x~HS9@6?qy*|G)=H1cmf z$(ESBGue~KefL0*ro!op#eG?7JA7{dTB{i^IEhhoLtsu=CU4a_xKzUMBugw- zcB3`C-YLZ*r)^pj_jZiyEYsh2!?EW#l<+w86c;D6<#p9=>HLG(Qur()%MpkSp$noK zcJ;P3pS1fc(UpZhaj0)bo!(2m(3YPM*=senu50Aj_9&LKsoXs8l3mYF+7?msXp7X6 zt)S`3%V+VcUf`)BSJwFHouaZG#8*BGU%mruV-^sFRi{VJM2XYcko#ox<4j7ojQc(V zKQm0wb&0Uk(A;${)S4$P4J#%KaqFttR<8T$qcJ+D6z>XLJcB|Qk*)!a8rOl*C29Qu;qOV9 zAv@UV`vc+!?W+?Q`L?B>z+b$#D1PU)A-Sp?xBGn%j9FJhF8y|Q-09Auqlrq#5CPX) zA;A5-j3QNl+%;uBL-c51VSa55lSIYPitVTgBN>W9QRC2@W9&ChaX)h z`0fHFtJCOB^rRY$Q~W)s32TLb@vfPR1sA@e`-=|S@bJRz5n}t3XT9zzoy*y zd{c6J(xcbt^}@?N^HJk+H~hfxcnKb zH@Z3Sg7VddBUz+mlZ$=ok;kp#85KT0o?9Yn9@ZpqCZvdI_SZQXt6_e>l+i;h ztlA%to-Qn8Pzf!%F%4N1pCPhUF;YJAZ8Nc2e^3%=1R4H*zkFBx5gZEFTFz&AeVmW) zakY~l?jjHdP30KSOL0zZVGK&9g@(T`I9P{PnL5>8ASX#~UL?3+7z)n2TSj_j$allw zf_-Fgn8u;2%wKP+uN_GG4yD>VO-vOK@F1$BH~sO)6V%*0ITFyhwbMHS z&X>*kThu5!h%dZ~n8eFpgOgGaEmi_7WhFK-gD%GowpiH!vs=OD?}C)AomNe+d4<09 zn@7fCYq6+3A)|qCq;Gj?O|5sz zJ3fs?l9Ek1cJ?~tG#;n%VKEW0r0VYyBaVY$t0nGYY0Ead0i~XPL8fpk<5%L1Ux^nY zB@b&{Rw^1N)4*+03Vhx*ZBXw1=HUy#&XL*gjCLn#ED;j_KI3CH_%sk%c;d4Dt(EU<<37k8O zEML%q`l!2alRLOH1T*ZmdnK!@C!%djo_m-ZLaG3jJWj2lG#v~5#9*G53T2XX*_Zi| zlTTefeiF%^BU$~Bm1MTrXWXh^-PAzW@!qs0VVx&NUpE}O8An)$t<9!DgzFQ7%OmOE zyUTaau9&XHk>g^K5e{)SR+);2@-8|JyB0kpM7IUov}aI{$0t>@+~$|L#9>T$F^@ap zC1LDsjX=w2b9@LU9Xy=z&n!D1vmj)*MjKsjDs=*vZ3rgDDh^49XjvAklzgWwHYYa| znU@yt`G#SqPVP618vLFLlUVyZ>&;&OVCtZT7D}IzH^j9`ehBf?#a_awjSn?vrhD&q z$FkIobM^jx9i&$iskFG^hG9C2K#nK(2sC=^+22BsrW6F^_by*iON1oUA`~&Ddl$T# z%2`YK%!g*J@gaTsOcr%>Bu(xJpM4j29V9ighI$@fCTiB>ti7%6y6~9jufF4($-y34 zc*nHdG3Er}W^&^DaHq8Bs~cCuP)jxDcxA!(p`%-lBoC@4m(LV;Uw|)@1la{-tlH4f zLDV!x=}<`+x^W_n+UM)0J2Eazf=Op2AQ}}l=hKkQaZJzOT{vSZxcV1okfV`Hw1llY zc%)yRz1somzKIiDI&73QQ6u{@r28Q*k9`T^kVh0wSdix;zpb^plB-Ngtx}s?O6Ib$ zVcxT#nK)_;t1?Mz8T@Scrl5b805M5NgcPnDTnB9kOEF8k|IG)*B(gU+2%dc$!p`CC`NiIr4Vv%)?jP6>LMDzMw{ z68RakyE)OY4Xtr@s>>Y}a2>h-V4JhEbyp51Si1Ds-n`U+q6-2EFH5S@ZCD^)`bgru z4@h=YK@@e6MQ8WzZWmWpa#9?7&dLPdM72)y+UwJ|&IV4-hGt;Uv2kwPZOXcz2ks5HTX{tW0Jhr9 zy>7!=sKHWX`3--L&AdEGZ;Njtg`A`B%_p*6)_U5)nilJb;z5Ba7+AoCiw?omu zVQI^aiTI_o^@ZatpTMwXA~J6pGlgw#$DOh)C&_M7O!HK@5`fq8jL}im`z%*2W1mRH z9xHFka+r#BQje!haY@4xe}l>hKU%A$<;sgaI<^yA@5?-#EKKK|fs)%oe%D@#`<2=4 zNcD5kV%gVg3AD~153IS9QEAyJJ4HK2RN$=Q!aGADoj_5u{+UBNn8>@5MOVGh-0Qe8 z<7p=?W(=ixz_AVp8YLxILj08jtDrb4>K)<*-MK2Xoo>WfW`68w#c1T+-Z-k$_d{#J z@MUi4nepct<>J=-t>9sEZYzcD)z*-*9$%tSI~>$$v0RA%ELgL}moh%vg5Z)$O5MtL zPInHaC&vcyHe2KT@_wF9wc8#*cIJQ}fs#_M%Nc8}F-ORL@B6bE*G+)@HM(jKT&~2G zBcP8u5R0mqOB4Rw6{C~HO7%DYNj_^8Z{cfU^u{(~d1J~+AI{IpuQkH-8NCvE`ux|I zQdFpnQ=g$05Wj0zK5Yfr6lGU_n6)G z*^b#3AL+FCt(L8*)7gA_EMy=6D3)v&cJ$8zEkz{|^qbz0@{?BycNG$fUG+0DWJL)H zA%kngC(RuNN8xtqf?-%7)M&+><<);ogTr$|DH}bxYFf@Tr1{A*k4rK8#}5bMLQH;a zrgPE)%!8H7iGr6&u6L0vI2102b7PSppuswi*ou;So`dms)TwH zT0GV1ll_wHv^(6Z4WVY=wZQxPAT+k^nLC1rw;ege2%w)!&Q!_2lB*a6omBVlO{wCb z&yo9PxWv*+qptW?mrgNF-Qn@|1mQ|n$vfTH+=WxIHm~(&j<%f297SyS${l+b@31cm zawgmt%+xBhr)8_8BmThIAkt5FG)Z#Xb=FF@zOcs8p^*u$dlUVcaNy&V#3V*yPDl3$ z%khr`9TuuBf>`gKi$Rl!85k}v_bz>#i$&Yzan{pL&_nz6GGDXu+4grhvxnK5r4FXN z8x3Oj`I+qBnyIBfbq|fic(VY~o?NiS_JJKZIbR}I(ZD;XU2 z?ag*aj)#T!gsO}Im2YaEKE@y`w}LYkKwj5}JOa&h78`7o)KauaNOu!4rJ-`;ferzk zwRU>tt2~OIip}U-H;jqiGLE%(__N0~<=EgH;~M_kYBlbl^+M*?{>12~CS8@$I3g1Q zwkTzr!N1omjirPq_`eYl&x4(cDZer8f&_fbQDe#>0v&Z!m5(E zTTj_4aKI=A(9LksYX`$`TkloMV$Jv&{UBGE(5QVI>P+WJk?nd{d|)Pg7uDw%y$w_1 zi{@$P&QpLSf~P7wS|H8de+c`$nM)K?Vdbc^#y-`wAY?gBqNC7-1Fg$!M0oWOp`KZ6 zSi_iPQ=|8Mv8D;O^b;BpH>|4*Ajpcvu?3;-fNsd-?;gCp1eQVHH8x|~vY0H|NTPKX zZvD+Mv~TceQ43^mno8qK0pXz~=l(%$X9aA1H8u!qJGs&=zXT7HS9Tl&9Z(dNgE__Y5pi{P7=y6 z5)!7iS9FUM`e>r( zHJD_V3X}USh^xaVPv1V`%#NSAblCWCnD(x1(l=u;pPymnaP55Y+>dk+i{v56 z2RQA5P&W^MmJQmqZ995@Ybiq;JA;dO`{Ikzwti%S{Pw$f;POPz6K-O%Jw?xT1Y9AW zan)LD^fpN58j3TOpy96&(#?tibPuOdm5;|~iq6{Nv-+FnJMBXCX(#i?&^jc-a4pW5 zC9*T*TTykj%h$&Bx0k@lUW6|MUQ}Q@LwSRE@wzuuiI!q5k_2=Oiu5j8fa&g3TV1f`jlcShmNQl|2m z=9hmUQFnH6>kOnr_Tj*l`jAS<~@kI{>x z$6*r6;1IOmO6bQGiPbih6$#bpZXuReICQer`QC^gkhBd!5GA)1;Md)fkDY2CQLmK! zJCId3*>M*OXJEg9E#7!JB66a%H%XbT!SF&#bQ2jcTZC1_nevB!(eU|uGy_{MR$1a+ z0Kg<$YmV@rd^B8hXqmDF*o!_`oFDu^QS0n*IrPKM@%rUD5o2anbzp9zx2{L8lD(KQ z1=%MAwyO)KeoHOWjEqpG0=%I|2?na8htaJx_H{pFaW`I^YYr>|1M%q(a-jG7S&Gpe zf$vKeQ0o^3_>OhKMzl40a|ajFw3U&#<(c!f%>rNc91+>*HyIB`$RPZXM8ET}9u>+s z8iaba+(l7*jJnHbt_a*x2BgzvR3f{If>Qr-jbuVZ2A0>aPd|)iXgr?V`6b56LkCGI zDh~dg^i#o=PEuqwk=YOX!XWhr-b9o6KCBgx({k4_-eo{IkxV+vyxohT95=2d%j5O2 z8e954_RJ?LU_N#OUdF`wZZf`&_In~?qgg6qcl%eV0v5lhz9@28Pt#<)c@HpO<9>ud zW*h4L5?6n&NA9MXx&D@dd@I6A`YzC@h7eH!`8-Un5~MywTK~E$?m3V!)7QPII;R4y zuj}e8Hi*H=x73pv#cZsXf3A2rKhVmd=XYt*p~>G4R+ciya{c`^OXW~4)s6M5o?4=l zHtVjxg*QmCtsU3cgyNL-8F$~#B`&(=_>DAxhXFd{W0mnzCypAnyWlsT4poDapIR`! z*^yxt>ovEIJ|+-f!|6_A535r=z5&(-OEP8->zmN|fFCsr+H?A2XppPY1bke zR|i9*aHYz;oQk5vc4IDfdoWar^z71}R$eHet2nKnFz^0Y_dY(R@*K=EIr^4%T4Lfq zKTk3-a5?Nd=TGL2YpE!O7O~t>abJOTxstDaodc5Qh!g}sLeI7ihDquXBba_)k#9>=bcXt zc+0eFN3aSS|MreMru-zT^w%D=}-hYJbkAIRpoW+~<;UT0}^*)jce57A?I zVrrbs`dST*Y`U*pp4KSIBPLB5r3$ZR8D%w-gV|VsPAI2d;^4ThcKBoe`*y66&TXaS2NZrHR1VWPITl z{fUvIW|_)~apQWe4NUD?Ic$yZdL(wlIH-7L@%r{tD7Uy{&I|}iev;wa?Gxb(K%9(~ zm%6f-gscZ*ndtZ|JHzzB<~E$Yt2fPGUtv0lkWIjcm30%L=H6!GN6W&2`Ml;J%Fd%* zL+}Ddj8C8~(c!E?Lfxsysm&$S{H~^E`t%#Yh-AeJhxa7~?j5zV#oBNUnQ<1fbIoZc zC5`0UU<<8^ZWBQ95s;k#l*EXKT+a&?!c`=Gn=q~cDNV(esY6Vf@=yd>SDW)pUbSwB z&cua0B`=I6svtKRyf-QHl6wNEtb6LYn|7PTo!^0KpTu?-mBxb<(5K~f*)S&8XD|Z# zOHX?fxyqFj;}t1QK91Ljh)x8rlpU<{qqpW`yYa5tvTa^AkZe_8S|(Tl5(vC~rf}Ou zKWN+(2=9ntJlfc$){BG`P{G-a^^p%pv9X1H+aGN|Di`fN@?|`w zN40;=vX29^XdgF4x+e^oSs5oL*i(Idnu;K!*h)En%q>30A#F9F$+t$d1s#8J#K@i3 zo_OkqT+IxBeH)8jE+Gm+2p8UEJt?b0UQ-lfS8{!rPFrf;syP~-pP-wwT7%A#UapXz zV5;(4O%1&1i~dxRv1Y@~OLF8-`Mfj~)W6UPm_`T6Uv zu$Wb+Lfq-}HetNa@aB4HPkn~NXP{7^#UZGdJohv06k;yHL}M;*{IMABUfNupj!fT! zLw=xHzabofuV)FeNY6#60B+}3$$LD8X5C)>T31C zJGsGr(#TgyZr*@9tU1(rsucEuLyHr&6bv#O8_75Z-Xk_tCSk45FM`88ckJNsh1(2~ ziy27swHGqAZJS5ADI*6G_O;nt9Sro@OB_ODb`>tW$Q+cmI|sIET?-e)jn|79N04q@ zU19w`@u-%R>3SX&GbQUf2+MLU{JB4h)Rt~={Dsa}vzn~9mc~yyrP>rF?CSYE6p-d% z3K*OuJm~!yfNDE14IzjqfyvoM@IP%s2D0pBY-4MXv17C)9Aw5_jkOxBoEr?|aFP=R zC(xt!wbpBVo}kK(q>txxSS9I;=99>F&+7}wtm5A=t+(VhTGP(`p!Vqsa$Vk50v*>` zHgFvjC(+t@LB!IR3do^xAG|Tvd#nu&=Wfy7SsykORfNKlq4BlyC8U1D2Edpwg7}*} zLaXn-tO+o%pfv;#C62UK1#%p--M0lB7GcGM=aJb`dMrMN_dzy1v=7Cra0$Y5;mt{@!?uUNA~hq$UwHBdxe*i<7*zOIODxB4Lt zVUXU)L1^(~Y4&$3Z&pdLYnH%F@b$rbKk=veeKOIV1up2NU@xI~hfbwS$QI>mHMN#} z&{a0I9hNuqAYrdzR!&{eTAsgLl@Wh3v+zs35Fxvlfm- z{DiwMTa^8#M=_RYo6m8!arDM8@S5C-z&zW?(a2;{hvj@7ViP|gH#6lIL-p@JoA$GI^?9x4Bz?6x%1oE;`3&`Nj4=o;^h)!TZ}4T9^Ff2c;jEQa zMk8{3pkELr$Wlq)oVYqP&_4f4>2X8ig>80C7?-NByO19;c^H9dt;eX~@MQKmKbq=| zKIJ>U>oO&1P@}N}(T)I>V$_CA9tTb|z3w_4%9n_LD5g|p zdtSFI3x@zbDttX4)Zz4e6|oR0`_gc(uHQm;OEfB=%DLGJlLg)o7e%EmOt@Fhdgl!@ zG>}WQ_raf^>5o8au^*e za=#1F!afRMaQ^SJxj9Cn#n)wasswU$n}-V%kw;=ATqtwdPgig_wKW9nkU2LU2 zI1D=q5nZdgL(Lnje^bF6{yY^gw~2i;7{BeHT*^zYU0FM5sI4LcI9mmp_P&ph^ zwBW`kL=td@AvnFy@%;X zZzj|vr}z3*#`0@(4z+WUZJS|CCHzTq`2i53<;04SWl=V#Y@2yd5Oryw` zcJ4|Cr|}B$-gfp?M`%N-{Q9#%Z)WxtjHh8Ir&LyGfHZJ{gC92iLVmdvd8@9qDnT@e zXcARPHhBH|$ms1A*~rFn+HX-m@s=O!V`v%PR_^s@3@|WK+S#J^R_Bdyn#J;olV8?8VAU_*vNnP=49G3|{Wy-`wPgrV>UM>?EtrFiW zmF>7tr6m32lRA#g@+?t60YX1KH>%Jj*(4iDp}vwJFNq9pkN4dJn}YgpDIpkf#0M{8 z!dqUpSCDg=fK%QSXPI}vS3QLW4o{f1&k; z4E5%_)=FJyC0no>z6ZT#kdDy6^otUF>u1id6QX)zP^V;#EVV#Z?FsIfQY@Di5v0RM z37-FRUEg?cu^d7Akt}>(TDlQ_Xl{e-da8C;+=lCnbn&H0*!**6Be-8@-1E|Zpe}~0 z`&!bH+~C|OvNtF>x?`VLbj|8XjoTqjPI1Ga`;PHG&J@pzLjj8~6e$r#TC#GD7Rie_BXRx^2#KP;qXhE>H6jUS(Jgn2xO#50KY^90F zY8(q}u{QHEC?;N6iplV48oa!-Td|@+tP<|orbxctE0jF+{Eg)LUAIceKy(NYF|?B5 z1w-nJ8-E*g?dhtMQYDv>eN+xLq{BOf$zDgqR`Iga^~79~q^Q++O_6!VtGzn! z#5vbsu(v~MZCjv`0d=N zj>ueFRMs4V;O2WancPSifr-D!a!aN=Hh%ZLHNh=eoi5Ax?YX|YOi3A$+{fA~e=L0% z9@jnGn3&i#KSd!_9D2-P9cS1;4$3UPmy-z3^7yAAOM~;TEFeQIU*Prr4bQR8*|-H? z*_ZLktAYYMdbHX|(}@C9G-2>5c2$exR6(>nNwB=;JmzHa9QSA`!!cV8H75F8e8wn` z&0r+Xukn%3t@HqKaa~vv+h%58b?hj3GBm2m&IB(cEuz|aucQrYzs%fGZA!SIbM~_3 z5v*@pn3U1dmJEQiFcW8OKfBM1nfYFn48aeEgwKBFhqKP~fQu_!!_NX?3FK6nEx_&{ zDt;=(TX5iSgx(=!3=hOubno(YozxSzEQyYCLG$_4e@Mj^)9}r~frEiLUf)!*w&DCrub(Vpdsry2mbdC?8Bj65cS+$3c?z7ER(;s8ZJy1&;(KOPMX>1E#xUZFURIQO#$ zcG(p6^>f^SY(q|3YBn#AWSrcLVc}@y+_NE4E-*>nhbO&=UBAf!UTZaP`6Ze}XRgCv zj=>6n_>~A(NSkWWCdn~>!e^g%kzseiTqjF^fF^E^grbuRfm_NmN;q5Gqsh0^;&)vT zh&V{RIfHHiiWF8G_tG;CmI(_)Ij3R9@5()xxNT}s5#sr3>|?MHX?v%CY3%C0AG9qo zDzq@8P^;`AQ3s#%?o}gJgXCj9IoX}1(bABLlG}YCi8#h%09)I zyISd<3&b&l4Gv0We;a5k$yYkmbF1ff@ehx2QQj7n!)+c%e39Efm-rgmWigcuhl06~xw@SA8BNw&BhP{M8Jj6|si$g?A z({3v5ur++lvBu4fj9JNuiv8kaGPd*0LunA7^>(r2tlaMcn^9Y|sARBJGi;LE3nCs5 zl;Uj#Ve*c}*5Wc?jUggzc=Wd<7*O2paIYk2f=_6;si&Hx5mjK1E`Te=ZDTT zgzg*^R`gG0vA&Fg1+UBOzWZp9nF<%(vtbU|)jUgPfT-MmqJ-p)>@U&g@@$!{E{hgu zYe}KBnW0dt7Fwp?qDk-$@JqWQ4!y}@V^RZdh;1r^6UVLA+-JJS@GJj<@SnFAJ5Ljz z-ltFI5v%1hnvqmBPG@O)BupJx=ump)FP{WeSqp0}E97}9zzu?DIg&ni9)~3`@4(lMJu#2c6#eMUD13!sZ|`y<>Qx zKxx{1UieyYAGKfzCI50Gaz-L_E`Z$ML32j%i?DZVb>+Ulsli)1+s_IK?TV|Pp$k*e zwKEi@{}hw+Z#k{n;d^>e0y| z)A_kAy6SD)>_~5BmId=sDlKvFa%5Lnd;A_KskN51h%%hMiBkfF>!%VB>v>8rN#-CB z+RuNHmF}7Q;LaJRCF2|>G!Y5Y{&`qwjTSC7(xC>*J*8pC@XbcIA!sIXrRe>TMHf2m zTiXUsHJw%(`UnP(LhXZLneSJ*KC}nny&;~wQoHyg8VRXi_-Iw;3W{o;4AXUJrk7sI8km&K1zVK2WE6_*_snK!e1+4mKO#!R#E62sz4r-q&<_L|&8l=$6f*Lc zbd~4Wh~lNFY$oUT6VK#Xw*)zUPEEegJ6*28EhZ8P#CrgL+lCdmX=5t(!)|zq<9_0* z9GwRwlTDTQwCO1w5sOR~4JSRhnEmG!m$b%RnZa+uA=l4_K;0WaRoR+Gaj_kTuPlp-DHor2beNs(Vw#Tk9L* zWBps+lt}ByLkt3aeea}!=;aLe1mbrvVyiy6Hoe}U_@$*S0*mpKtX@U$3wUasIO6<8 z*(o9nT7G5E%qDR6Hz5-tT!&0Q`7ctF=fN1ca5I(V43Frz3btwJKIE$O!nV<`ChK|? zkNm_-HAwO)J=gAmRFbzPmM;DN>9t^PtZi9Y{+>_>QIci$@P8bgMV6#c6hya#%Mgde zU1qq$$7S_R^|Y%`3j7ywFNydGN9cW{Tz^E-urtdHR1>JJmvIni_LD=zm|MG~g1n_8 zFamv;w+>?rO<3Kulwu#Rt6?F2bwJu^&R5DAymzMm$?gIWmG}{I_NzlQ63Xlzg{JHg-n>fk1LdCKVC#oF-RVtY&$ld? znORV%OifS0q^TY|vG?ag#gO#;_S@SfTZ%T8fqQ0)s;T%`d+#&4ASZeqPt~Ws zq2F$s!^H_dns{@-Ywx|rtFaYAi;k6zwSM0-mKwu*Gu@n(Q?xbUf(Z=^>2MhY4|ri2 zP=;???H8NX3xau?vuLikoQkR+6UvKj6Nwooy}tceZJYr)jJ9wY);;`sAg#{D-pyH5 zVO9;c~B}d+f809?HqR;~> z^1d}XnbFu?(_oeE|V}JuE4- zTGvy!>6AO;5x*9@D9OX~;;CNxs>sEw^(!wby%|qSBSYCGNIW&UCRUc|{6z_2Xvl$hWAH8D5p08Bqk|h-fqRxV|E0g>yUdOomhdB2?A(Gj+G}fnjfz4bLn2R?SCN`A11_cmpR(i5 zX*^Mvh@UDe^x8pb1*GjkDC%o1^b2A9*?sp&JwBUxYb~UoYmtrP-kdC|aNpa+V1>QM zYBb;%ob&GK+?SZz`aU~Q5-1<%?F1hlF*H=LO6G=_L3}<%fJnPvnFg$xl_of&zoosI^&25HW7)*TU#JG>{!~3A#^bm8kx@}8}w$*Fpg6Ov? zpaZUuCO0$nA7BoqPI|#f!a|5^E*gZF*W1{=O1)j`(W+=@6#KWNWC`TfWNKOm?Au|^ zVhN&6mE8%?A$o0&hlAOEUj5dklha#fWM-VNi&kbUq^t$)JM8;+SDuGV$clMZvOv{hfRHi!thLgX0mLv2EDmKvNUN5{1!^0moE1h*-lwI zmY*+{uP^aECX!4al}b_^Kj9jZ=ngjc_yHGk^@uN&k&AG~9Ld-j>Zwa&4g&+^c5xbt zzvdXEnA`@~Vg$lFW*RX;nApsI#=T!Na6vIwyzrY7n7k=Y#B}R$qh8Up<&-j9l3Wlb zQLu(zl>ILjHdEmnqoaKdrK}(d%tdC<;WckJDwGta2qgWxC)xGw*{UCER^RnQ!Gxd; zg>?3ASdprx=ocOTMrax=*utkRT=NDf;9~KetM>_vLwPU`j-t_VYNeQnNB%LkXqG9% zWObPw1t(&;%mu&W9JQ&b!uWYe8&MPY%j+4)*uEW5{-cyPun!lCR(h@P+d|a?jsp^L zP~OC=Z#o_6Gt>saMZubgD!K#4D+1Cc?}BJFEn$MODqr46T_9oEa(vsBm#w;gJ|wn-)ox+_QR1*cR6 z9>a@4bHKakCPZ@lXpCPT?$a&O&ZbcvwLj8WqU*vGa!-HrE?c}B7RS?-d#<;*CmD?K zPy(9}YwTIE@4D71J6Vzp8r7*4nvt|l9e(f;SDVdQVhIB9M6gv!w#8d{4ig^4zV+no??9W)8dh*}to$;) zWz3GH^N?6bf~cu~pN0ohI;X46P!?9TJ`RSA4em+G4CpY`u_Jjp#p0OGIggM+5!qPY z2=A>j>j&Ra7=NRm9<|$$=Hgq4*is7^ycu!9Y~Wl*&W~IX2&tbF*ApYBEs-hzls+3u zX;c8aUPD;_dn{92kh$-S`ru1PEaWB;q1E+b3;X-nVf(?|Gr=4nL1qRC)8qF9sY>cP zR?t2ETnYE8Maar^6|f`yEE$^uKHY_d-Xdu%uR=wM!`uY5-U^9gGxhT}+_ZC#-Sw*3 zjb`C=4apQGoV_Sb-WA+aDdr$6ffqs!=Rjf6Ib20D+*831N$j!S#h0+8?Y5`gDK6YCH8yz zcFMrlH7OnDLthgLp?ums9GX$WCk8nCBNsk~ZPmLkgk)5@O53lM3PYt)o*hUtJPbcx6=G`RE@YGs%dhO0<)r?k$CgKF(9g#58k{!r zLYVsq>QdK?%?Ij&*hwAg95J+fh28~P%cxfLlv-Gy#y!QNEML?v4%O0J%0!!2ldZ|_Bmpth~OkHyMGjHpI$1i3I+WgFr(J$ z#<+-W%VVmbm?awg2pc+RSLoZG5@s7~&Q>T>-E|RvDZ)d1<1G6hc-Fq%0#xKF`iL&5 z!{BKr>*dBX=pjpHVV00BqZ1v$8&7s?|EhdumFw;nIK!$eL*H%OrBRNNi%GaP2b<%T zC#|=Iuh$Fg`q2cY^E zwoUhLDk!_|RxE4Z+tBbcX)A*rwb%kbSOjwNb9Bjh_MTaf@yb8pJkI0?0Vh^2ULf-1a;j{-|tzV1}_f`iyf^dyBlC%a{n&7rf6HHwmj+ z;NovjhaVBD?yvqRp*Ssxi|tLoexkB5zbFWspF2eFEwOtY$W5(yLF6J3jQ5_LJwEj@>lQ=vGOpWA?q-XqM^6<5-=I}+H~{I4`4J|F|`2Bh}0MfuH^j!OB4_hO8Bv)rtw0RP4_kQI3?!JD-)2u zXpo4`oBiIfb=-YUz5Zas9hwjeU}?@(u;#J@V(Yf}%34*1K^)8?jFcH0Nqp?)u1G5y z?Gw!W?_9Q7U$_zRj^uugqx5&YukIzVsJ#;5na)z%lG1W-oJURTYMIR5>pxLt;QuQ`QV+0VBi3Z!|q$=v+3W+c1A-L^2hN0`BeRAx@<&^1b zz^cgX+j2}il=g6ZAYEwj9kpD7T2UxBzz*$JMX!e-yE5x{kXNa!<4c8LeAG)D2NAKr z2Z6lzFb_dqEcJ+4kUcnpAS>hV-XOZhZUr;@Xglgt1Sv1y@i?~~3GU{>{ zdaCoI(q5ifB`kQ5f|Q$VyK$eC(3G~`Chx~#+Wu1e!OSO0q?&gwOzWX9RYNZPYdR$C zJ{K|TNzY~!KkaW=*oe1&)7A$}zpnx+j?`4CeVND>;iy+%vsE64sI$8}pyOzx?r(Ey zX{7=!44ZlLj6gttBv5k&a7ap9K*Qm;zYh1cK&eP7r%twK5;eOFP8r@-Jy5x#-Vv_N z^Ltbnh|^c0SvajZQ&33lS#q>aQ4rJg0b|gv+ZossY=^514RZ*Xn{jt2#kUDQUz4}6i(ZcIRKNeqiybd8>D4CVY z5f83c+9}B!VKqYk8Cz)cqjful)+UHS-ud6fIyEdxdZ|zgltRJgE1l6nYs(Ek+^4Vs z?F~PcNCRG7>Ct@r8uQS1V}5rMXZHBW)MWO8dY`hQ?p!ArwM?Sfhk2yMwBF%v&P7k5 z_B$Ct{6d-124$+l)ep0@k&Piee!L{MV8#+pzhj^sKMBL*Ch3#fJz9+4)c#dde!)LB zGZXiKm5-;fN0cJR9j}4wmp5qm<(_^5Gwq552?=~O#9>R=;bVLAPOw!Vx5W$xZ3>wZ&Fde}FWvC24Piw&ThO5n+F^>WxtmRQTowa9}GvERLBU>CqW{ z=16WJbhtY;GD{7ALOK4eLlbuFX6E6?2@>m`tBIeD+{ZAQ-m7^#0g$WfFm0}3e8i2R zUoCE5Mf4_91)%r|^ZoSW{zSvsUx9Q2JU!?~Bb9=v6?R@yD!~i*xmdE-SHS*eZP!+J zaqdm81$Y18~|8v#VuzHxp-$ZD|a`*XQHIShNImBvjuL!SAWE9%ZsH*&(Q<-UD zN*#Bz!YbtXt9APWY4j*tAAGBgQK%v^a!1u;-?s_2Ep@`6$z7d_RZTY%>d))X-b zHgo+}D5H9Lu;v3^a@Xwzu=OL-5>~C3Gwh0;>(r~LeM9a-|F%q}*L>UTiAD{1=S(RCiCSZ$~(Zsg3R~IPh#+?k0@j;Fl!R}>i@n;g5 z&)tmzm+V~7mV+6&Rxsum9dL@m@g!!J(V5b1J$IE*9+SANv~ciTTvTP2l3q8j5r)O` zV1g7{A})hPn5Iyf{ZVejCKWiljd2`4ekfZyXE+MS-_h`G(4LnRY*Dk5lJnzq{MPN% z{04UwO7z5sd=74y*ial%8MII1T~P;!{^GfUa-wPAjZ?j6yW~hm7>!q6kbO+U@Cde^-VzSpTUKsYwmdF821s7*trF$x4+oZDD=LKM6Nf zzo?qF6S(p9m;tRrNzCL@eChr}^h-TTUCSn%r!PwfC?=BgK zmHiM|gA<2)#+N%OLp;+qzev|SbgP(zmPPpKkJC7+la9ywrU!?0iX5;wpL?$xl=54> z>x7m2hsjiJ@|!0g9m;(RVk=GcD6_94AWv@gyWM0gXi+ z1?Pe0E4+5~LbEvBxn0n9bU)oM>_Peo?Ene@O8g~b*)WalrYUnBeGQih!eP&zFRuPA zeK4P+b+HMAK+W+@YEyfY5qtLSPRcRw74^ttOq?E4b_}tfP}_AL(BQI~DOB`nely~c zK?t<2hLZlO2oozN_BVvtjDfr6ON_FHVUNT3*gm+hs4P9waD)X8Fy5;hbglf}5PFQv z56sGPGmX3{^0FcXpFc~(GHulO!??{ro^^Nn2fkmVr;ssFWbOqEu^(04=gN8g#F_3& z%M4y2L(BAzXc78KguFnal_aXkzpbk$q{dpr?*x=p4rzK*{k41Rm?%H5P0JdO2@t4b z%e~GfJ1YAJBW^HH^1yC}xNN_XL-;xPaHu`JU;NHS0iRiM6p9Ew$R>HW$o<{s3CU6RgXc3wb1VVpQGN!&}eH8!F}@uzq$7{1Ue;Wm&gRByo)^G z*sy*Y_Icr?N;zBmKP^A~IDevg(6{9Y^9L za|~{2&7Q@JJ?TZ;2dGX^Z7_s~A2ZRmRP1Wj4+x~|sS)-7HG`FyP+-64p-+p!ZG-xE zj5GUj?Pe6HqQg77#4u86I(X>e7gqN5nuysBDp{DylLYlA=S8M9LmIf9!wylIkkC1G zo#G$Utc+_I2YMd41IUeUTai3YEl4 zm_@U)lU+f7=t=6f?xM9DVK()5mCabuxXot0ZEv)~4@Dr#_Qw6oRB z!D7haahQ`O84eRT9Vlc&zhY-Jvb+abZzAhecQ zw5GF0c7n!&?B%UwAzAL(=D|Az5o4Mc``(w8rt`pY@^|LwLn=rA!Loo+s=5@rY?2Cc z_n+co(@zl&*w{F7inSysJ{*+oH|WeOstw)#-BPqSZv_)}Cbai7FoT(lB9Z`SVt55^ zQgNb3{W|Q+`obZS)y5$uz4x6G(w9heuBu>`x7ob<+nfcy#S%uaFl2b0=3`qh3dx-q zD|+?d?3Zvi5>Y><2`iXOF~&60r{H9gL*A~r#}~Xzw|6GUp2I4KLE%bFc$A7t^Vhz$ zlP#r7XQ8wgi=J1{0*mW=C9cbPa3s4SbNA4vUGr9{hnFWTd~dTzp~njbEfPz%LpK{z z7J|jozf0nDt=gUN+d@u+Z=piwmlU=eL01!wN(wRh4ouby21$Xu8#VH2X=QxsLWhCg zUwMsG(-So5)nE-A>VE&@X|74B+@<+o?84r8+cb;R*7;_2!lWOUCQY@?Mq^5TkWjx3 zab;OozKLlwdr#F>r68AUd0$TPRl_smJm&myv-gzCctjDt&V9y^OtP?UvznHl%^Sau znT55dvIg#y!O-fpV7IFVZ3$f!>}dt4Aj}RwG`?#vZfOZMO72M}Wmed6ntz9fJhZm4 zo$k%**GK+ugEWIF5Y$N=@v=;ZjYbs8G&B9VCg+jE+l*eWVc*0ynZ5aO=78<8c+@iG zwHB+=*>l9=_E<0#JQVvGpNE*YV5g~}N|anpGFbY!WQD5U`B+U7$<;)Hf_!g&5*1xQ zoC^C%UdZ#Lnw!qMUw_IEhs+oWev#j+aiol)G*?^Bns9%9-VydTChnR(i(0XYOsp*R z_$K*zn?>5KH;$G<5~oM%E@`=ZY|hkT+zk=U>lkM%v*$Jjfj3#kye&?~EAy?2ViCym z_=Ux|XxOwvJhvhP8yVRa{ak8a-Zg}{-|zz*!|}YHt(3eqe3A=SSeYNjR@=(@pJ<7w zkEbqwy{Rf()7WAI*|&DVS>J-LIa$br&QW@K>q|yoq~sVvInz_x$E8bzSR<(s*Vc$p z2^G#*uI@CZpk}i`v~^z;{2?oCde{w3NlUtiyfWC{ye%S6BS$NNR3ML+H8K&NI~0tp z^=;;^LqZ5LMIB^Ib+0S^1`m3uTBl&sBZU@#&UMubCjn}Ro!l#RXz2`=R11;5nTnvT zsi1SyWZF5pLq;>mBR>HX2kh(H6^MrZ<-xgdM61#*Byu^nq$CZmQ*gW1i{uGi%;>1- z>>*bJRRt8!V81bnDjFg(ufMFF1u^8PAy`9W*D))NXh^~wnvRtp9qh*^1&|46PT^>b zi6xgLenNRSYBN+$?);cHda8k`)MQoJebu_2G?8Ho)#s#_in(w^yE3#so_r(lyUq$- zzZll{xgCyHH-KHE&)!&5NJ)d{+3Q3)y_`OX3Pf6$sc9S?qC5`MD_=yQK|v4%`?^vb zw6KO|5{-q9qmor86#X~4&*7f~e#U;Kr zwwClpD8_{3>Uw)PM8UfE4_BrEdWyU(Ah~ZBx<670lpdbJk3pqTfKAz1VMODo@gvL4 z!9P#$CvnHmuwox>=!A06`|4*NL*SA}&pxpeuE{`owoy2@=XOkcy%1wo<>a(ULlwd= zc_mYG82XEuh(1Ymx%Yyr=hyk^_^t%4M%$_&KruV>KVTB+>Iml>;|{iu!D&%S$~h+H zwGiH5hmfZJc@LRf|J^{GfytDDFWRW$p9}NYPDwVAeSg<{TDtcP&&^rlcA_>S&LyXh zW6ApD1;*RLxPL_JerFK9fc=&G`p|l%8bQZ$zJOr@njXz=a9{s13fDl=DeOIfH@Ls? zD5joZ;5QO+?uo%*`sgxLU0w#2p_i^7Qp;D6P>rF3(^w$#-5>Wb(?p3Z1$wEJ9jDrn z$G3QFeSUs7f<8T`o{#c!i#>5$WCE5NspYou06po@;=QfJO~Z5a%bmx{NOqedwqQyz zZw2l%9e~h&CvQ*I9fp$gIpB}od6}A_Y?9i_ss_^If*aQ#B{sy`H>oQ^F7f3tZ9L*9 zk(U|gX`G`>4wzGHte042g2yy=LJ9E9sUPT_lr?zK5d|nix%-{Jv7W^Bq(G*mJ1t$-ZMDt0aZiDCHnL`EBXA;)vs?S=O5VRw=b#x}59D;jQGv)F70UgXd5IA4@~D)glGM}X^q*B!=i<>z0POW>|>uhZr&HQTyi zRV=eoT0VOFHt}xhb;ph_#bqo@KEs^V_;|4tEwZ@jS8fIGyme3LPiNk4<*{AzN{cyS z;&+##0}>IFp5`G79IEoCg%5t&ntJF_;X*zxrYff(Legu|#!$rj4&m{>xNb*)>*CqT z=cjwsuecI_T7mLT<^-oMUF8Oeu7k(&HJ!@>RPVEtbDr=1U|bVT$tKUMtzI98k%_mx zjJU>LLXjYhD8Q?{qTsfC*4`fDt4I7U&9Zk7^&nC{g+(VM_7@$?M}dKm)weE=Dd^FB z_R$dAa(m468_b`=+!Lyd0Wf0A|Fp$o#{9PYNXmO0M$r@YY}V1YRzmO>f@?0(+JRSS z+j^5{l4(^p*1t>USDB2G+`k$C2uwS{^h-4Rq3*L_bQ#x@k7lhUL&UqIhay*;ahlR*v5H~CN2`GM*KDi>)m$ROna>M;sxzY0?zF% z@9R2o=`kN}F*4K*LY;3;!oRoWwXyPBNCHBM1GEC+^RRbhIf41R9bq7bL5A382sqlo zPKK2HEyd2lNKPCG78093;%cTJb|^7EB%%D|r@`{{uh>jiN)au!`dpP{TMGwR?x~m2 zWUGwb-}<%sb=Eht^ROBc!osVq#Z;-O55qXQfso8fFy;XMP-d@n=27qOI}ga6(f2F@ z+m&d~MxaM_4t`5Jt7@FR7kP2w{NiAfvrwbJTPx|4f7rN_Rr&@_&nRn?QG#O{pzeF^ zF^ZWR_v_^8#_6+z{}O1X@R@i&Eb-<7ONCA!#vZ}tJ}{SU zrouzb%bW|v1wW=!j;>(Zvc7d$dmI;By+khQ=^zaK^FCtnH7t`2PcL^Qq|17ibA({o z{wI1vP*pNqN;Te`U0ym9_)=LslW*JC8wbvF&ZCH*IY+d>QaynR$1ZxvF%?iUgg7Pbl(N_2TqWGFwTaunt`kM0A#znEl8r~o zoyYEVkJYP1Gv3Wn?b8<4hOIqXlIZZ}-bVM0TjUVd)D{}8`_0`f9$u#;BNMJ`PS73~ zeB2Dhlj2Q`Ni@;x%A);Lpo-iTLIM)@W!Avb7*z5>mws!K-49w+@;M5OQGev9@MP@A z99inAYSp6eueNNKDDEL71BPHNcoTe(ExDi&KhoRTUv;tjMB&U+cthkC4@@f~E{WCt zFrzS$LO7w@z>~|jR-!RMO%TH&LA++pLTD~_18sYoCNg8ftS4t$g>fm{1@9kb7xezp zr_yWmL{br%orDEQPDB{-0FMk!@)GFf^73$O?H8kIeA#U%Pjru-D8p#_3`uq6!{2>(LSK9Ria%wQUjdC8uI( z(_?QJ`mcsKIV7r7;mn9 zW77~4oh2t$J#2}Kt;9~Ek}i?FVM)kt9+>*Wn4vSe!m15{g&5ebm>(`vwx(8)l+o7_ zTe(74v`WsgrLJqP6~f9Oll0P{r~-ncPQGT*7uR3`v^cDHmC+2-`e1n(oCA(dvktVR z(*^Rd&aBjb#6b%h(R;pUR-3s{UaO>pG8c3?mXGCFH~c5W6tD3Vi@ASyX6%mNb-gk$ zg8Cx#;o(iG;;xMjWXHn&uh*4{Q)EzXFsN>+e_iYc@yVoHTaz41yj=L#Tj-qDD@8f$ z3AZCr$em+|7`D9?RY{0|;K_PY6~N-)YZ4rfJz{fewx{GgIqH$*H#|ff&5V=mCLud0 zV(-`IB7Zo}D*{k*e{)ZCsynVf15oZ1dc*U1L8Xc+u9lkhjyh zdB(-NqTs$I@C%l`0(|$2%w97!J0Lz_+TAJ(=9}oNR_MJ7A?lD`TnhEX_-yPMV~Wz_ zpHZ@Oh`p3XDzT)MP)*P#)Lw(yy}LzD3vW^fyF~;lA~?sH4J+RtL@d~o4c8p9;Iz|eVoMsITDHo}pL!s6hp^xP) zxAxurLdjc$>?vGCcwPN$5|vu8PCu(N`+NEriE?L^r@p!}1ym@lPJd+D>lZLNvlaU` z-q#^u(ep7_KMZFnvAipuudyq2t}|6mQQn;{vQQz91u3;&dbEY=R!@;p6u(+jk{a1E z9B-b3XX5w8?>e|yIX+GP&Pgem%sb@8#wP67s9d;k-MfWZkgE3T~Z%SyKI_OksE{Gj(Q4baSUC~FP*FJRc zsGA{i9GYD|60zUBF7_2qlk=Z)L9_T%1;Y27Q()scp?AyirtY4c!-^oXT({$*Znm8g zSHhF$x4)WdE%x=}{50CHT9ibUV+DmX3t{6>E2In*wPo5&rtml6h1%D_Z>oIc+0$S)(6N!@r6qMxPKT=p=j2 zbt*DUWhJ{nSFh}cxN-5O$2&OI80O-k(hh;L)Wa_ATl{q=?@`(Pw92HL8%m}h2=G29 zX;_2#Fl$sP$VYw8o|Lx@#AyiDUWp_?j*&DVGgo$Mbyv#Q(XHI}NwU?P8sXk{ew;2; zYHVogE2jZ2mAN*DWJ~p?C7r5N$JTahtl+gMHyg|+qIQK}$wP7xuy14FI2CMfy_DBs zISP_dU$Qhd>SzJe_(r`%#(g~0C|Sj*i&?7v|B_J@mhnZo$cA=mdtD5=1Q$hQySz^xO%dI+}1_A zW6sZs^C=pe2y!){Wtxq~cv`u4Li{yd8;6LDXGwoA6R4X|`pFWsT3Unv2y=PL$=KzS zKj+Fw%Y!!z`VkN63_RmLorHMd>VLJ}@otCE%C=w_QAyC8`nM($BDIUB1v1O`sySf} zJM{*D42HD~oBbgUA#NF7w1o-${t?)H4Pbr1a^<>1dn$-FX@s=?iKZZC>&ub9#ecyh z{Bn5h4SC=^A~QaUv&>-Q(;1S^`!eMBeSCePf4I^*C6(A@>#hVvFf2#=cPfR1iKr__$yCF|Ah2Y8DBsMwM9});vi|iQ{b$(}uW>fZqyAvJJah z!QL*zg>0UTOJY=e|f`#}TN4 zPedWzQXKQp7`uGGePZG49YdkJwCE0U2~5xtj=G8Np%MkzSYCf=9tQuv_E-{%3srv% ztu$gPA9I7!F!Vuu8p`BNg6ZZ&1@1PY zTbdq)uHUOjoa?J(ZDH4}S3C`?up8^=Fp?K&PCSIA?U=egLyH>qmycvjA6!6(eKtra z?n#yBHfSLPQTj7lhP+>m$jA(9cqg?WAmko;M#!Xjvo3n>a9cl4$Qu-0 zGS{oi@ms){zO@4#m;DCos>SK4sCY!Nt@GNU2JvKxR{?g#ry*LLxQmMegsFSw?1V!_ z)BJ3X=1UXiAfK|69Gx~e>H$awpV{aZjnp-)nY2^uHTF|}A#%kdEnhi}mZD(LWPJO) zWbRAx6VB#oLr|pYI|vOaHMb(<9rH@&i<7tnNjl#>{yKe&-%7tW);CToaki-wXQkLA zewRiB)*+ol*V}5Qnh9f57d8vJ- zbw6{-F$sK#eTlxrvNdDy`o+MVxw(XLfwzY<(8tCfsKGMtvOle!4vA{e{mqD+4ECqo z*&%{kUq;Ium+=5ao6Hdntws!Yj8LU91x3qIjn3#FEfuI5MiQ^Zak0j2KR5X#w@se( zLet4*sZf^6>CtYxyHn460XqhR06v?Y<0=hOcLke`TK~5MJ)#-IR$OXaKPADN$)iHE zU_X-6V=3^jF8=okVtS3Q7Fq43n=o z6Vp%~1;d%0scRs7^lJI}xY92*$oz-*pn%iwTP8hAp`H4wxI1MiR~dUdJe3L;f)*?d z4MN*)_*G{E?#J5x)T!KF3K+ztPQ4dfg;H*HEaYB&U-X;5)8Yd+)!De}LZ3Eq)mUIb zqgxPVuF>+-vgWO9%R6r5cppXe;rBtmN!N(7vO-lrc1+3A8biu|-q zv_XPi%tqlKoUmm+v{CkReHt}(Fek*GCW z3QksJW*91`qG9{Z0p!Xr#g7GO+5V>>I;GgEra6c|xKv z0m}!NWp{$ny55Wy;l+=TPBmG@N{t!X+rD|AsVNm-nCMnnRw)@!ejm2b+uGgd zrM1P)g%M=Kqek%jjI#-6O7ApsGZvcvpsu`q{a5LzDGa*Xb@rZa_Gy3S8D9&73PSQu zlZEHW&S7tv?N7=@=XAs^kYd^%m|yt2()++hT*$u`-R+ttmOVA0tF0!vJXuoLs$7dெ-b2i{hm=R_c+4bdtryFrwl#=Q*%05e#4LQYaK&l^97eF4c!s)kIlpkXnR`l@5`I4)NGgF~owt z%Ze}`-K?1+f-IMS!Y-30eD-h3WMmPvu}*Sbzqt+BKU;uJ52CohngJ zX+qQhlylsDkj}BXObNU^8qwrTbkj7RHyTK2)@TIud59V}J_*XuwA6aT{ZiRXd$Y=U3%sn*8~&vG+G;n45C1ZBf<0*A6<|cA z;C)C?eX?0jKFv;+p;x(*xxmU_$^XSrY(c5novBxy{n{h5%O-Mep>9c`n*tqWF(uey zizuTEzZoipz=cGT*VN{GBn-D2d8x>cL)3yJ?vo|u(ao{drw?}GBU$`W-L>4RBhhq$ z1X10Pa>~U+yrf(wnfGJq&sosxsKuN5L?~#p=4X@kk0>c+F1yttS=7np4t_V{eD$~C z?O(Y~xuVFAMt|7&Ng{Nw(-QtXj?HOeNDg1Bdz8xc;&@UD#jV2N*)XvxR<}B)HpJ6i z=NR+rBD@7{+3Wz7uGqeX)6Dq`qboIUZ)sj`__e>$e8jlZyx*YzLi0y=xyeB$GTW>C zmF$yI21VW7TF_$J+U*5pHaP=A>l68_ARJri;gWy3RB5w`dv|l)poqS%JKHInQ-Ka5 z#ln;X4fT-D!kmY`(B2rg?6-xuQ*!1sdOkC8(ch>a7Pj@PFvA4X!_Kvk%k!}8U<+g6 zjF&n{l|HUG!4E5>w3M6BvGhoTiwhEf;fL+ zYrxSQe6Y^ziv<j^PB92ybZ+Yqt@eSY`8_03;56PpPc0%a_n zt+7M*Pp^kgfrK|2t$n=7tS-JB=6HM2!gIx`>3A4a|$Sv=p zxE0c@x&$qs;U%ADYeNA8QxCFR6S2D2p>y32O;1<$HX}P3$M1h8&4_*&?hn)qL@?2C z211`ZRn5V+W_|;u$At-XqVJ^NvyT+rMUb@^X6?VfwV`5tbl%v(Gof(2{Fkv;y2CKIa zl88g_Ea7=GA(s4|k?;UC^Jc@`qG-tXr#ludZ^$)FTxE>&@daQ)N>47u^~>pdBy+9P z2hvQ+y(-Z+^I+kV>9THVH4#C z2{y7bheOn7EOqgFp6VSG2odd^tpwWx%N*tTL|4Hr*2%oDEtOfV0-hfss(~hLudXCj zd}!%}GO+K4ml!U5K*7e}9b4D1Nt*D;oV^zgez9h`@V70n28U!cHU7Kn`&cY=-b`xl z;rO-)J4)$qsI>mYTM5wlUFp7F6Hdlfkxr2jpQ^~h-0BZ#Qb z9ZT@08c8QE?6jTrJq%=h^p~%#DB4x7U)$xzpV9f>>>ojtZzJWIY^>uZ@|Ic-^EFLZ z?#pLO3w_9~cb|5wYvn=X$XqhsvBBCO=@gX6HItJ0(^Vp-s8)B!?8%;Dc^zu$+I9Zj z#_usmk#bP0ETsztNb~$;6t&{N$AE_%JN>qq_TeQSMK0}H4>pD)D3r}De$X7JR}6=3 zm}Xp6yQ9dCr1x`h8UD06=k(AeE0ormXrAwHMjW^jrDjlRBP~VjXYcfd#(Wu;OLe}} z5m`cg!5XkXxD1OB)gW(uo;FEQ>iyx+kVjDL;UUeE?% zg{R|JEsWO!B~dcj&d|mN!>Wn`yAq@Vg(j+WHja2om$Wkl%|8g&_+^OPA-)Os)Z|hD zS04KX5DTkP3cr=z(7W8fpOswsDtj=gsb@@o4{|j>W3>~tAM|yVq%fFjBT;h(OQ!t@j6Kcv@rB(MUYX&pVg6V52l;3G-^uQ*74o?XiHP>B}lp zCcvG{QHtOJ=lM!o=g{OWf2}Dv(l(xL8z;fg0bD+aR6}u?*v~3Jn>zQn(_fB=2e@@k zWm#!oYq`VUb&l@$)h|m>nYEtE!zyD4dOk^osqj4?Z=S+^e>YLBY&vkzL5rSk=%XaP zjA87yb}2jmG&nSY;ZL(y)0&u{tin}afmcgi4C*F%x(y?`(fXYdlo7)Wuynpqp7&c; z@0N%1tBNy0D<;m^IhD2QHzaj@0uxTdd13}W(y|RP@~55qL6)7P6@~u9+Vl6pF_Ja| z>GYuhH9*S02rw^Rze2@m?SecNT`pDG!P|&MgszEX*e@USaVp!UWO8x;xRstm3y}KQ zxwjf5*iS|HRZM?f)sNNOz@KkA7n&HzYUPVw2ty!~VX>Oy-WY|tuhW1A{mepx<3d?W zGG2Dlv7OtG%tmsH5V}%@1IZrB>xmQcJ6bbVoSbKFE~Y2{^+(YU0+f&a5T|%862)gO zlQ(R`tNe{X;49F1b=nPApmXRx!_eJ8em+ymMvf5eb6(iAQ zRcqtQ{UmPnMPLj#tbJ{_eo#N4q`T`vM`RQF4U7VKp;YHJ=Fz073|6zaZpJP|JqYQl z(hymU$R?fGyMn8m-`38)48l!dq` z)PO^FyFbQ$>#&)UhiL}x`o_ax{8YH+llJ!sD!=iAA3%L#iz(S|#S~u``DJ$E*MPAJ z^n@=2%E7^6DzJ{jTk<0HVlK6eoUMCX`-&i-?J`eU&F~^<(sy0e5=!>8`&~&VbtcY{ zouujfBCFEHF6of^%+kJmJN3PvMZ3)yOQf*yDafvRa>D%8)L!C}!FY2P%`(nsFHzBv zG$@c9hMJivs$nq~>l(>^2?VxwIGe4)H>~1F3c2vPUB=Gx-(*o1L)g)YkMw--4l8sM z$LpfPN^nfdB&Ak_Q5~hdK#@Zrw!4L?dPRM|p`@H2#xvtOs9FU-;8M7qEO2;(ZRA}6 zMTtoazY90u21fYSyxEjgx2{nGyZ0v@;!1c(bghTd;8WXgm1b11v~sBcjp5~jsoUg3 zrN5+!c{!M*J8wBS05`sD(o^~}%~r($>w|R)ij;d=U!AA7e9NWL8Hwk|H#Vm!EmM4S zQRWCr?yEQtmbnQ0v9pd-7dp!5KqyeKdp;s#qRgZJtWK z<9DY-r)|8jEeMnW31TF9CeC9gY)~W?9?7|V2FCncF)stOa+oTM`#)U%UbgkQC?A(& zG0h6En2~XHf;61_oXsU6-;j@^d}rTY7gK7i0+t-ZnW3=xvhf&VQbf1fG@LM_e>vH{ z5Q~z$cFVawQj1u^Sa118Gt2udAi}S{5YI*uyS81%fTYPQGH8O#n$A>sAVXuwYZk6b z&JtXTfeC$ZI+Kv%EHiZbczm8A6k1K8l01_&u2yL8P5Cem*?v@yKph7UbnR#Gl@M53 zNxdUC=IPGnOFGSY$?bv#UGOAgVb$04jIz3fb%=(3oX&MvW8+ih%$)q3Fx|hrU899m zx)vmth#iU0gHMvzz{o~=*w1q$Mgo7=f8^|_G3Nub>VPJ`mowN4-8z^&D+-`dvOQ;P zn55&`zf1C-(BJlF$(p3E@IAr+O4rZD<4g>ezNy%x?b3XarZ~^5;p4R18SQr>dyd2; zgiBqrmoAEQOSL(yyc5=__F-G(EQ#@5f(fPY2YvtFjp0m?LU^$b<2v7nr|i@=DrJ?4 z$f|nI3$mfav#V+w%KbZ4pz|Dm-!={Lj#q4G@3{kX0$&uyVQaIzHu6`6XkF$cWQndm z%T_++)Q-KUy{jF?8+~O&mE@~LV&~a(vWZ<7#nbRfMfcROIL7Na7xkM@ZmyBba-j7Z zx-7ALAN~rjmz|5V%kxB&AiNu`$212jztc30+zIzy|31`X-6we5C_Vk@`H0BFh$k7_ zYmr?IU8(I{BXlfpGwcX|@t~1e^4>nMRVI6@6u0-=9>*bNUM!9$EI=@L=*ejo% zTRlNig%EXGgt88iy>lz;#Rouh@Nf)6Mn+;$0v1^w#JXlC5ug}h-)1P&=)DicIrCRQ znvhYd8@;%(F}nJE7{1d9x@a@Vjq&YD6T_ZGhN@9-;OAvK6gpub1Yq+y_=K18$LDp+ zo|tGJc4Qk}orVc5*hLPZgezE&hYQiGy^w2;-~QSvFkI*EwyJ35;{Mr|kXO`;S+OCY z>Upe6&e^YQ!C}V9*h&3jYcbbN6*&Ku$2O7-ci2ahS9py1-1rbaA*t|DruSJI>gMmh zJz69!{juORO+ObrYVYTYZrpoTtI*8E8HrDY844f1Z~UZ|`4WMWQicwB z9lKj;rn;($ka>}9XOA4?)17<$UN6$!%RXKxQg8aki?dul{sDxSAprZe*dtdQ=j;eul4L+*W1skACh`nL zmvM}VmyeL4EnH4-HuFK5=$jkTHzc4Zo`IN|ykKXQ!{zwv%@DimA#J+C?=C-mEY=jw z=IDV4WLbqykluCDVYLNQ5{R_{NK~&kOe)g zi6VI~mG1@HLTjq6Aqo1ZxVZo=Juy%bOYI~Zbh^%l6)U`+bx&3z$tW27XL%8JjuV)@ zoRiJSaPj@I?eF1LnWcjiK-(l44vwA6p&G6={l)i~hi>W9!Q2z#bqBdnryPM;9OJiS zUj4RIh_i#S1%*KwIjxPI=qyoNtw52Z5QaraS3}3lQW>JwRZ&Oi&7DYv`X98l@%4Kx zbc~K?i`{di0@{U~f^hfcY7RWTp%C`vF5kU{_iUYkXF0Kmd}UHqWsznfQBFp&q-NrK zvuUd`>==GMWuqLO^9z^5^@u4sH#vqwgVCBfG{mfLl*p+f(Ai-4bxjsa%oa(5Dbk+( zMQJ$Vh)rcEcJL1pH0yXk+(Is6hD!Ro7o#KOTc#B~*T`@qVy7^pACPkj`_@auK@ap} zkjxMh51tJ4WuQglpO|-|RX2fX_2iUi_m>zbkDYI3z8LT>uEJB2EaK$*$qHJDP zc4OnO5nAw7AM5pg<%jw5xF@|&8~MwBnBMt-=K7v8#;3qv)32Karx!tCNg23$QT_l- z&J2IG*22{aEUrb9T^Kyy_{<>b*M`nFOg{4K#gf1!{Mq~rqw3$YhRFEl@?m)~jSg8N zxPm!Br$9aKK$SDpXrPG0KFN;#yTEhY5HJD?`^i-)EBhC^gZMkkd*vx3?NzEXHb0BZnZp(M5oYdrfjdLcl7ICbgM7+oJ-2aDNmYY8DkpW%Ybb|7bN37 zbUvdvG0;v#S=o7Ys{_XK$k7~O3|3uFb8O?dIu~6CC*pO-pBtK?+?BP}g?afX#+CFX z!y~KTlv{QjXab;h612e{`&H|H?TDg;*L}r2vhgnjreQNefenX4pPxa63i6Z`K|0pH zrDyj!PfT_3FAy>j&FS~JjI4WfBqm-5QIW0TLvMDM-(n@-zY1xT!c*7tcBwVBSw({5 z5mE(LaS8h{i^N_{B-3$-^hCp4plU6y`5l*SDe@(uOOCCaLBL15oLehT-QqXrtp#(^9-oN8CLsNa2fLP zasWzt_#{mVT+xy$C2~1z9Yz1FDl%u?w*o|q#xun?kSVCzvm2i`-$P@T6lweFg~v<} zn;3_O7Z(-6l~+o%DfY25jM}@eUzotnk_bEC_DMrNlstW&&nc#knI(Lj8|ACzNW_Lv zE|(`Zvcd&8zRmpy{cKlfYucFpPHrWWPL}dbnr~R)wp}El&%~%l6DgFVc!WUKIOO!r zJ*H`m13UE)6${#xxf&Ot4(06^-ah6_)On)T(*XIqGiEP!snpPfeEo=AHYZXs?bxsi zNNnO4=+mlE@_d&M8>C~kep_&KPBYY>JE!)0eW~4sU%!Io1^%FL-ANqQ&H)+} zaPo*f>?KoFRn|b`8f$WxW;@x2xMhcH@}=5+Z8aSG(nX_s&*W64%ex#1HGA77Thd9W zw_iN$a9fH?6=3jVQP1ypT7~uOdcPlWxUzBx zk4kXEoVAmU0VSKUFxG%VN^%7RfO8#QJ~T9^p7lh#0vj~r?0q~`E(o9ZK*`CWrKYJ}j6>mhdMcb+^ zDa>-y#-Z;^vIC`dIB6}i#PTfA#meaxVjLJB;;51ga60k|P`euqV$u%tm*4Yi$<|zS zqCf3y*__W2dx-S9>*RoiUz`X7pGbBZfgz&ZaieTQJ-$H{r44WGtgjprZE6k9wsVZ9 zfcCoeRlce7wj4)g0_vF(l>UNc?8#h`17CS%p=EP!?HFa@R;Nc=n{r$QVXzhAS-(#c zd@1EU^3X}g-GDM5A9kv(xtBGzT5Z)Rf|((;Yj3kvgk#F0-ogW&pWi$jDObbg34YJM zE83o3;ngU%8qw1BFTUleL>YBH){zaAPf+Go^3mEP-Oj-oZHSw#cA`K(k9lkk?Y8X5h%GKkHE zYw+@3$JwS0o>*d)DCA>g$#$i^+I(H5@EPn)@!q*}xUYWw4#~e2P+dL6viJd(t+gIi zu}M!-5{4q90HpBC{!GmV;i*XCD);(mffsD8ePI#0HeN8Kg$eBTNYLL@-FB_K)7zYd z5j0jOBd{qY;&W|aoWl}{fo%r_#k1EZ*~SmGEWwY91sm)JxJ<_stWtfxY_)Y3jC>UQ z3Pgz9y+M}2Hkc)SLqG7w#b22UI}_jC3Cdq6YNhJdf_3HdQuanDdMnbKRatf^4H){t z4^qu$19J4tf5kFIia&V0k>#(7Qe_##0+kY<+SB-@(LApA#?(66W!BwfdCg>*e%u6|DU#9P`rpwM^O? z2$&$Ay?ObHyBv5ZP`d&q)fr%kvvi8Wp+WS`YR)hu zfvR7Y?3a+k#;~nb=ZomdirYF=+F}~kRSLJCFbxa!>^BYghn}P(S+I~$1_bAcGQ>~4 zTZ9Mwd<-Xc-m?mMG3MQhgH>!Uc|UUuU66I}=hCUp!ywjF`r!I{e%(9pooQB@%*fD^ zc(3Ey5>LjHXmw8;(LvBiq9&0evr6e6|UOjz}sLL2dS-J6HK`o@EE z&Ltg+Vcs;5p%IbDVt^-PIF3(hxM>tdauDN|Fof% zZ-Z6{qO`H|{3+YhI#rBh*y2hBpQ|H`$UmMdWyNj@NP1XA){*X7Hq+|u55#Ll@K`{TEK~{N zLc7ZusdhXaI}!}r#$Sl`z4U5~FcImGoP6AIrp=dQn%ALeKgR7VF#sErGgBA#EA@GADrFJyEq29s_Mqg;(=}{ z6XLe{BmIT#s^9~dMOEpB)8=oHOr3e`V4iJmZW;R-2m#CKuU2{4HV0HY!fHH(I9rm@ zxuQNJ(h${}K=l0qAAhu8Tl60M49gg>w|tV{T-GvkXMEWjSq4|za%bCUtH95~X zJ|a2N)2@CZmI4gfu8jI>?NqfvV=^`TuEY^xQZCobj% zCMovIiDy5ZBC)2%d&~~d`}dKE`zF6DT0KcPBIN@w@f?NiZB;)GEizqjWr_N=cFFjV zWJg$}?7*qy$ws6t#sMr41Ye}}=!6Yn~JnH^^T3;f;O1r}d!$15tZIZDm zc!@z!qKeB`)Nr;SX}@JI-&3zoshX7r5T~YHAhTO%tW+@k;Ec`|(f*>n<`5!b%QA$< z7KK>z6gD1@euW;078^tOcHJ6XS;-dL%nGCI?GRBu@*<_ju!JQz>_YN+MPO{n^HdAV zvYH9U1g1g2q?P7=GMO919B?XsqSeRs`HmVYum7Rv%c{oc^|pj4Y$|TXwS=pyc^2LX z@R$;*s~2YEFlWZOk>6-)ccfMCBQZIM?LmH^GZ_>S`86PKm`=xOSsU!9Ke*lqF@3OW zT*_Vrs?H_DK&Ob7uV!Q!kZ@Q{Y?DK{X|24F?jKY?LF_YD7w$} zjkK8o80x#46u;Mz5gK(WXkUbs)$^=6Ez|8UIrRFn$vWEz=a2w$STn*N1bYouCUCS@ z#hK1Yl~Y&0+FFjL3j&NYu_|Vngv2Rk*MBa|sBX7;^~twX30sZ|+0{~Yx!6HpC&WSu z!>shBS|qCCW~bOQNraf47}hKhklZaTkoSg4G(aVa6AD}UQHKofW8Vp0W(gUZ!%A+bZTQ0luBk88e<|k-0@C18U&93ML+?Yz+^J!9W(wx`V{6HK5;dP>l z9Bb&{zu7BNIxJaI&`e`dlLmPBjeh{{QIk0Y5lwQ{tygcWNS!j_49a7*bHdKk#xi3H zm7;b5xcaaEwzfJYrC5s&g|*m@pej1V8qkMbWk`YySJL2(BqLtyY2mTYaYKjeTnsi% zi_3>wTR@+`V{Io&yF?R?G~mrv*k|5qACg*3imT!e(OlJy9(bgxX7AvLdh5M{GqH1h z;~208OiSINJb_pDElPtV=6ZJ>gPvB9A|G>v{;8=K>QzsQa=k@55%Z2oFDk!1Qi?fO z*t)Ne;B9@-$1h{73$e}--L)ZN!{r#EqJV=pWv{BqUIl}18bx>T@31jbV1BHh^6e3^J;`8V(R?ehprI2$eP_ka=H|v$((qz2Np@vNK$?)f(O=7S!lS-e2C_hPzTB&hmis4R={}q4` zO%FFjmz9CMwZeK9b^N`?m!5+w89NjQzuOmLzO(RA7)kEX9? z8LJ$>V^=rJ9_nScY0e77_GD2~6#9hju$;@sgG0dBVbJV623f#qrC=thJfo?0-{a#!#acU#?WfpjX7)$)NC)2b}cs(K2 zN{E>m_iKpC@xN?XOv7q9TxL9F6$E%4rQLvZ_U z%W86DdX0&D6?P192(?bpyFIRsD1Z0+Hk6Nhu^{>OrJ=c=6SlA%uTh5X^ZLE9cH&xQ zoonG#SpSqWRka!Lx^iJcEzFu4jo_oTXPTE?O{a!90L*_iR-%vdnakRV* z55maypXV^6atH!IcFEi!#yT5Bja^f*NRo(t?m~IhHuV-n>ehJNpMlaCYtI>%Lmrju zBM~!o%tq4uSv(Pr##kQjUa06AqSPczg*UpjO6AnzvK-Ry_32XO?r&aqq21JjvQwJ# zOIv^YyOFP|`K2*eSXAPfoDS1SNAod_U|hZxtzcJIJXbuyBuJS4nxN#jk^yyl?-5PI zX*n~tYRAm6q6Ft>wo8fg0v#BQg>093M8!PfRGNCampi`vTvcB<;A@iY17V02^0=xd zvY42PSLKMeb34bR2YtY>1=i1EsWz-DHkMn`H8J~=Nw1vjvIGY9G*cx-OW&Qc;@Ukb z?{3abiQImlWLj==lM6WE-jZe>9tt9x=7w4Gt?PrFOKSAWJ_-At^QDvl*X4clONLY5 z9LOBtM;Ch16hl&<#o`tPrA6@rS@epcVvexdEz(kd%(z45-SU$|PxVtv>NRam+(!90 zm@8*4Fec_c%1)O(csXR7ul9Mz(FpZEeCP6Ynn1q`%Ox`i_K9BA7Pxq+20ds<)*+=& zAkPBjTY2c-H)o05^E-1AEp7H>+WR+=@|%UwyHo$d4O-avq#lYEBW0o~*xpTlwNo?A zNsS^?8Xrp|k&-Q%x#YB`#s!R_8#DuHP&`1YsykbnGGhswaJ=PY@wFJtG}GR@^O9BgRZuZtxXTfrwV z@+|WgE=@s{3a6DN|9^bedY(!Tl2E8M^EzgZlIUCZqJxk||92^L{+LHxGgvEBJH2Uf zxl$hsI~OL7xpO>Yi#2*;a3t(w^=J83Poq++{FqZPTC|MtN6s0;%-F^g%4$Wd22y8Tprz09yS&8IH+9@@Eq2$1CBH~}^|OtD9G zSUk+SF!*~%KWh2mCujQugPW5zOycMMmr#o&N%cf(D&GV;96oL-98YbL+^9UmXocHs zBra@z(Ql*Givx(W?KxT~{U?H~=Cp=&U3BO>Xv4#l^;<^`+jE=JWFm(5b*J~e$77z$ z4IDvL9t*mIFHeZHui9O|s9k6Z7CfN#4hUbW8O+LrCS(zE_nd+Nx7WP@zJke2F1C9J zsRMmwq(6oTEkZ1f3lQT@N9&%87o=`eD#5Y!BJxWJ)z)YVx4k$tI>YDqkY%4_Cy}y5 z6)3{CM{tC&XYMA9l$G7Kd zx=@8z<1G{K={g6>O6GfTl~m z@R6qUP6~8)OC1p;>&r3PU~uRbe1I=(DMu2G3A$|!+rFC_ZQ4%D0ZcGe0Gir+Saszj z)o?h2uNeI(aK)~eay+P3oh|3@sfT&`%%yXiNLt)V5|L&2lRO$PZQjUW0SE4JHo15%=(W zZX-hIPnP>nF40|pD>cN^^tWL@Y7KUDG%iNr1@`D!^6+`!rsN<9%xpW~qH5|VnJnp>CcyxwG5ZMz& zQ<^uX^;)~6*Q>9r!-Iors=Ni+nbMBZrP;xH2h`*W^1pf|j!| zJwAl%utq?`)i-X%Vr7@b=?cxu%v$MD%Cyy8oZ2qGU+icISKKQmQ901iSHGd$`3N9q zvvYKqEed-|g4)(RMU4^vH6>Jm&eX?xK3~A$)cZzh1rx`CZsfZptd_ zE%2RGnO>gMe*UPtnu)xF9GzV6h2-?nLzXfik$jbBtRI7RRO5_+f$8o}A;B-t1O%b@ zE+~*rxd=N|?9Opcj{mAJ`)Xb*gIw@}^Ze#6V2b*Z<4kfIctF=%$&(h2*L5ezW&F8# zwx2h%(`P}oBO4kMog1o*3z2#DmD(8U-QYN`lD`hIbcWbp3b*t_AP_kExW~Q|U1b&m zq{)tvq?Xc9`*#aSHt6uJ{LOuY%5=sVOldz^zwl-G6tIXUXg0v34~{J#P=QYxQpr=( zD;ul?=HVKsY?E{+1`6z!v#EC#e3)BR-Eu1ktGJ+`7W~x2Qa86>m!Yll(Ou7i6%O|o z(k~`;Wu!pyIo27KTEJGKps?DkOr_^FTJ!zdb}hY^j16Z?pyV`s4rqtN57_SG>eY^; zb5Nyi3{v#4oOqL_kJIvoocfs*X_lwouQ-zvBWBh`)?!hm;?wIlBibaW2aqxlrJ>6{%Omc{2 z0%~WiTMy9rSW!Ih(XHorKBN2TRRgNx=;OLh0sRoK@e4^vvbm0I;4FRHESx3;S42(M zlYRKWUO$LhHnOmKBMQG4038JW{Vn3INlcR2)qqmOr)SAz#^(-($`9@!+e~Q9j@=HH z8*p(=J5f=zM}J>sW1h4oE?z?uWi+(8`MLST&UnVaqDppL9n>~~)Q#+kOEV;e2wP*( zRoI)Esf^4LK58w0f83#nZldtcR!k9ZPVS0t6>487C!GuODKT<$F=Nss?v|A+plgk^ zw)mi}lD@g_e8yOcyEmwa)dZ!uWoDRfXsU^|a}S0kEm4s}@;E&)wpzeKu$tOFEtHq_ z8Bj9HZ(+2LUe7wxoSNG4_NK2z%4KBZDf#tG?Qi?Xs-o%kocG)2-+-f%_4*O!Re^Isyp~HxdH^kXM1`b!Svb(m* z5pEQNR-_>nhm^Zl21PoHpA<%?dsM?+hwJG)G{}5e+Op=28b>{JZ^_DEk%(VWR%)Be z=VB32?eat+G`!T&mP&k(T%OVwedfolBy)BgMkN!Ecm2ezo9=fWsCj$6`=ucKN1OLN z@_IHgLkyi&T?Z;7jL&?Q?A71=pvaHSel4xSQz5F$Bx4XCITmR4(_XX%)uN-9Ysm8k zcZCv0pBHz~R-T%5$KnMkDq^}x5bSOeZdcCYV&W?=?(VMecZ}?_>+@4?fiycMJV8s% zRLw7lo$BqN73XvaWHs{aRh<_;( zo*_J)805Hm5)hywQIo?kXxqQgHnG!UpN2`HMr^~nuGDCl z`dIs#2Mawd=<*mp?eCE%KcPjxPxj)$TwqZgEsUI}QI87_p0ZCDnbqf?bQhcrn>Vq# z21#ydPhxUPTsS97VdQOKu7Y@O=`9-wE(k$ZTDTB{rk(_+zU3x22vy7WMH~bBgHOVL zBQV+A+H%;TdQ-oLRn}TFV!Rded-J@%*AF=R-iwH*!nk$Kd!|)r+s~VlxE2mU?3UFK z`|niod_m??FM^vaS*FE~kkr`WF}4j6DsPQvea1cOsoStR;x=U9dATQFPuU+c6UgR> z+S#Oh){=v7TkYv)u*{{0kw?yuTZnt_iz^#5(nfXXiRG)LRhFYt%h-;-^i~Gi2PRdEF`FF+vZJ;7Dz90p$Pa8a{rM!yt~h)IiNg16W8U$4hr|-Sv?7)}5}NufC4qnP5YTCzt)%Mq zK7a@QGv5zI7ak9B#Ydm(8=gC*{q{AWeY9c!`jsTv&Al|FW_4M3wqZd%R^706dlYM% zNvF%r`}#>(i#%g0IVcQ%9j()J@eO zzdk|OCH*^CL4#PFrNK5PI%AI{WTI|aMb<|t7S2-1U^2>`2fwE3Fn(E5j_Ez?;_f; z$|xnldDP==43{Uflc4l_t?b>G56mgZno$J!8cRbpX>akNMDAi}3g)A(gGK{^@zJW) zZB3XFGbNl!L!O>fTrbU-jf{XbQk5+BM60uqJICI@wjq~iV<#P2e~DJQ#f#>oAv5X0 z*H{3GCb3F~nU2=ps`uciB0+9{0=I{atT%D>yGi`G8!}qqNjCO4sE@b9!tjX$B0PhU zmn1;U^NUT~kJ7QQ(VCu8P$;sIdEAvcs?H(z@81$Lu{T@8Q8&65^keaBp_@T=3aP* z!Y^me3}DzoVvlfiXm0+pIgUP=^M@u*_uRSK8@3^uu$Za&1(}h%;~}pVnC6`x{20q6 z@^3rHrF5mXaefDvs82?#Fw1D7^1TVS(4cpEWSxPGy|3Yn7IqL+z_m-6;Cdesf|O%K zFH9R5FwN1O=M}h>DI837Rd5<1CQM(tlTq0izSgjiy8Lkq+j*(-9EI@eSOCh{C-K5( z@Z~mM=3vVNxO|+AS=Pd?q;Z_Mh1JC(+5S6e{ej!V%D-MFqVx|y-c#3b{LKNKYS)r_ z{4V2`sU?b~+?5VZ1Z3%s8_y-~HZ~;p1R{6p#wsk|~= zmcEIE#W3^1(csTWanz~GmPcGxbqb0a$gpN@;1pbgedFgK1e-Dyqj5iYQh8&*4JsG9 zo%Tf-{X+fgcK)SR!!dnnzZ{G#mtOj#%?dWROd?K5G9-ttdY^NsvJIPs+i(dCeszNZ zJbSNcb7|ULbDWr@OM}u$%335E_jR}<&VGA0pTlqb)t%FL zb3&p{ZvmIYSnQ}v%I_PjM%yLiY*JF}7rBU!Sm|gdI~#vAz9>8wfmhn<>cc2@hKG?5 zpt4Vdb4qO!@0sd$3YjSIR-PQ5?NgOAYeXmt3DdqHuv55H7-|dg(K_V?OguL_DMH*v zlw4LXh*QJpNF|ch;cN2Hv82chYVdOu=5+YWxf)S>ue?c)fa)*v>WFCrcE_L^K+}G7vkV36tciK*t_o=+&(HN^Zj0fPGall||9? zB zXDK~8sO^?QHuDmhIf~_S6)h_Iu@jN?5~`4ukzttA&8Qe45}7D^wBHP2P%PudJJ;Tm zCY+f01a3D`D(%iGM}22X>#NF`X|c{UM@>AZm@@)9b>E_Ym5tg>PR3aS_*)qY(cdrK zAk=<;RR*PDJ8U85Yd<30I}SQatVwk<(3JBB!XY_TQ@gjx?f}A^s$Osp1yaQC3WiOf zoNCyu8Tr+w(v^bHw)L3?`xdmmI=OXZ&$^H6d5g-MH+7tp?$0Mx95qB4fvx#oXOD-9 zr2m>z_T#qu+?h|gbBg&&V_)A_efe~EYxKQ|0w%6ul2?)CjnN`wUQ|IW4yKC!IO3W2 zR?*{Yb>GIP^@A-6y!!#xbklOUMm&W=5@kq#+WgG-NWM;InlV>H#)JLyC1t|yA*8V} zHYRZ^NPqFSH-YZNQ3}Ia+eXP4M3N^7uEz+LQdFP4k5EM^#lHsdI=Ov~@?^22;n(5L zQdY`;Z`T(b{i`~DW;xFVbm_B^e25_mU&smpe&jKI!^?D|jK+M0hKR$!ewGA9&uA_h z@raj7cMeyH-J*A@-OAy&X{lfS^tq0+$vTwWNp|PkNB^PTb%MmW@_LkSzew|XSxxP+&3t>lkCfq+2g40Ut+=f z#_m45sXuM)9^F!Z1uc1TE;+mu4~ngJCSE+k_>+pBTCbp?``=PNBdC8zh@$X`b-GIl zdP+rl4b=(5BY^r8_nS$pj|%V!oDYxW2V7F$n%?qjQTlcy66x4Hinh*G_%f^UF8L+M zQCBz>1_Rn@_V}og5>!o+35Wn-kpBdxroGh}BbnUf7>437mxnztrsRu9es`(A^Oi;N zO~jrY$#9=d&Nn`l^QX*r6^LV0>G)QKyev82Z4piv@8fxJ)?i7hi!A~(^ecOJn4lJP z+gw25?7O5=Q&)&&UM(^FGCN50>!yA7-Qx0W#qV!r&sf{Ts2UE&+|TsCI0-vP@v_8# zHX(m8y9yx5y#`*|nJ|P8Ocrd}Dk!hCts`25E0BrNJnJ7&n;)NFp~xaoadSiwmsV)8 zoIv87kMoz2Ip?+!6HYvivAF$?^~O1aWcAJNj-RN~~@IDHF3Fjg;q4 zg)5f)Q3B4p!Fi)>BnIC^!VSv)^ z6iw@SMgtL5pm4MMnn90PsrF+@`}`}eESuW8ru?E-;oDRl^_Ej$pji&DOF+UFlfoFm z4Z#;zo;YGIR>xG-qX(AKQ(rNCvTA$2LfEEHY_S}Y;~yv3_(`b?m8FC@lavqwz~g-|#ES>VfbPp|_tgkL zSg23=(+=SuKF}YH-0`6|PJZr+Hehoj$`ICaZ#Q^Y{C(!s^XRt~R;7bj*Qj;>D!oS< z72f*|YS=<}dkEf=Fh2Td6cbTVuF_0(7I7z0K<48E6 z$g1@}J>2{Tx%j_5@i_5^CsN7TZ3r>HwsesAY#ZtvG#S?jLvrJmEN=L|2lc+qOFm8^ zsRbM!PbSb$)Tq4DdbD3kylFjofh|aMW;*7u1HT z$WdVq%zcJ;E&lu*!5(&FV}fRujJPXY5F;KIw)sFh4Y^QwvIUoJb~!n;43pB?SfY|i zJvD|1o;G>}KWT-TiOo-?frKe#erSnNkeUna4*j&Sou2tTngS1VtXoLCo7UJt-{#qT z!k+-@DS0+^MN#p7wN5Qg2wUH@Mjp*VdSutYT^!O8$s4I-P_WMLrG{|6LtM~H!hte$ zoLtH_8*(4Kx0W4ze;s(1IHZqc4Tm^QfanchJMU+@IQ?7LSh+q{NPd<8cKwN^&6F}_ z&WWb&14ciwD!$dt7Rz&dh_({6wsk1QrdM4qGCtaOeJb6L;OBBGzN9v%Oli5yXe%iS zIDr#&`^NF`dJTZy63|VVg|;TfH!PN7J{HHFjfe<09nP)sAeleMN$i;T9Re99^Lx=CAf43PsWtO-P&L>GJ-$@+Hp%mhc5X zjE6~U$(eKjhxWM^+km58ZBwX)7~}}tA^n6sS!i$b9{Y9^$Hh^8{!0}z$Jc8h3as3&`=I zW{CFNz^YUFsMKKjm5VvCf2ei> zmoog|I=(VUX?MXJf+w;ZZQAR8jbH6`RfdK8HikWu2ne0|)d{kygxu;!3mlhb&?>c| z_3gXgxtslL;g|m2L=17);TlbTQFcQjAX65PecF9|>v}mNWp$I)-M`MO}rIncB1@^$CGi_xm^;EKk9D?`v~E zekGuhf;eV7{^1Z_o7`v;@It%^Eif?pdPrY8#CatICDSd_ngsB*H5qHaAct~Bmbm@- zMdD|fY-}%YcX=6Xsheq@$2ieUt}o+j4}!*9dive!1ezx64!S zLB~udh3O*Y(Frb9+p+KR*MEBHDsj{5OaYT%lXI$`iPrf08#c{8*nP$CdYaCx^{!lW z)z$Oz6XYiQYfEfwv!f1DR9v({#pMp?QG(LsN#gh7Tj zkCsvMbsH%Weu0TF9%e#Ax6Sk_R%C>hF5<^ku3_ry`M-U@TX|~2T0*{s9{MYe&6G5S z^ADacagG;JbH3U0KU@hUw8*9TY#8NcK9;=dV3SD?@f9OFbPiNFi_H(VdZDXul=GMftqeu&|_INy}n1vbbFuq_$1;l`D8o=|3h`Vk}U< ztks$A^kri~kHgLsob^7Hxo>?%#3B9t)roO=Q$sRlyXF6-iuW@iIy`2lZKTnuIf36>xj7TR&>RyZ_Q4EZWxds0e4l*3yL6s z*S7)cMyJrleC8ps52>Rj)RaNT%ve=kp>F!ag@({Qr7>o|(A@_;tQu80KJ)f(lp1Y$ z9Ntc>RxDE8Xk~0BHV;E9|Mte+5Zs3mqp{jf^01F%Vs;P@xNs{%FUiVqQ#us8aY*4R znSs!LA^vmszM`O)=q6HTy0IHCjl!aj>9&TK>#hqAwyb1&Ja&55jw5+mx3owJ?+Y*Y zR-ODAf*2MQt}pE&*HY~8GYk5es6w~NF_N&E$*H2}NMVS}8$%yQ5)#OZW(&BG$zLdP z3<=-HXNJ{2wL3}AR6hxV@W?Ez?q9;VjL7RD2>!6t`vOOr5qCLgf=-;EgN;G>bzjvI~FKpMATpX>#B0GTYMfTRkGyQYStkSFOcb6|cCiy$g zMEf!vq}zs$WupUJN{MO*JECe3JA8IwG3)9zZM=+8iHuyMlnup5svihY~?N5IqXiacP4%DbuIb|%PP<2eDfbYuJQcxUB z+XY;xW_ORa1AO1I$Qln$5#ToS1JeeEDZ>b-W2*N8YUa*=&B^T$Qwkd3EFfck}7>Q;!t5wcyWhTDr^po+=U?vbcKZZ_wTn}R293o#v3*Owp%RgP^GSr3V}PntZuLdV^W{R=tR;{T9D8Hw06| z)mb!26J>RMekLJV(OVGagl)gKc|X9G_}ffY%kVG3`XSo!LoHtm&SA#HX)fdJkJ+vx3k58#8B&3~9?ixnnJ;`~~kvTWFvb&&dKW?q?`}Dn+Cs zTll%;&YK=t-R6kLF;3w?Rx)jnnKn1FSn7&s0-vY)L}JXS?#De1 zW2s%jw4a|(kcSCo5ukLjxF?nnTcc?cJDs@}qCB9NYeHyMqVqXjBf2RoFTdo=2>V7R zw9z30cfekb%gkp!ikxm`_BGf%ljPG(N_-9=F%`MT^-<#VM@2K`Swhx{NK=|>^B~ag z9Ql<~{v&2~;H*`^RZbPy^=UmhcWc~*MATMLxiT_@Np$HJzIEpXq3=D>9p((@u~9}# zi0j2T5#9$?Gf{K$Op?0ZonKpT8_JbK7fbJG%w!`#+~*xnNz^90yi)=~+c*L$P)xx+ z{GKQHE#2XOn~a}j!ryAs;g&g(3ggUVPKMvtJqxb;>Uz^1_^6VW5wIVfzHO`+#<)h9 zXeZMCK6WdR816M;&{$c9{{$-|Tn8X%+wxTdz`G!sKRIFM@^7a;?F8AzJ0mZeMXM^U z*R5gM^Z4A~^I&*Q`upYq)WJQy`>@>9BKWk$FNV9J!pfPBA^LYp{b04k?_{Bt1n5k` zB?|d$aMux1jYqA(ynKJY5>kF)@bM6PZXN3g;*BV}wnz#8Jyz@*-A%Pmkx~R?)f0qb z;q{9cXx7`mBlB41i(^FwIcRr@>aXNF2?ixA2+;oi!76teZSp(PiqZ+Vb{Hx$hdm*a zy}^R)FkCkkX3TUp1xDRHr5!u#T*>~S)ogFC9uLZ(K74zm-ovc?<^(!VRy8_TKLiI? zcd{J~?IH^W7?6HN+8dlZoSD2}rWyYpRoKX^g!;|;;w9nB2c(?=R-&@jo)4XFC2mZM zVN)G2g?v+hjITUnt-h<(FX+PVP-3Z$YxW+illSr|uR5W_n#`asKyjwHd!qO}K z!Rp+bhsRZG5weo&v6w`6Xge-?vqI-;{n_bV`_mkJi6))HY8LEvp4ehQGpx$NOeuLL zn#Fklky9r2pcKf=HWCcaBSbCcu6N3eu7FNT9sn;B=@a9V`WxK;ruPf zX2FQdFuz>N|B0W0qQD$od{z6NNQ02}8py&H}6xXCepC3iPD76HDOTh*AJqWQGK z6auU5e}>bkEG30MlT=X8rAGbU1h&eCD?=7&cd#z46Q|VDEt+v8JaDobh^BY<8%U4M!5_A5Oi`El~8WiH@|;Ap;l zJs4xbud-SaLDQy5^NK`3&qeLp@%~G_Vz$AS;;b0Xp>AOg-N%>FuPq1tNzV{TKUGth zggq0vV9(Jbs0Z<=7oC12Geme(wU6QZ85i}P7AZ-%zCf14cj0|A45~BrSEO2|GAo`V zK#rl`=5j%xI;8Gnm%QiluKfbjg7oil>k|TZZ=jRNo0RmJDGrwtZJO?+wH8p3ETxm< zTiaieB1@{DMvs=$PZzXZdpov?BRh-(?V_D>MRtWdUAP{j5enJX>zS=b5)SchOZVxfwdcv0QFrKt{Q z@d?kY#hE95Jdx2(AsD%)7Oix=MVHA7QEfy=^<-bHUY=7!$jsda3Jk$F^(c8Oa0h?0 znyRzMkVq#Ao;9<^6}!jR0&b@==`w_LzO5cB`+1K2E0G;`o5i=d-wrL}_a9!VOqgO? zd+z91&EufpWbQ+k*w1pwmr#f(tN%ZbFk`1-k+P0TtFlFM2F==r7)w_6O13{fxI$Tx zWrUH!^KKRe-F^ob-5oA+IbBjTIE93J`rW*=-kTU-*D?$y$TCS&5nFp|IO zYHZxEE5wXf9kXUJ5iJ%QD|AZw_dK9Umi^JY_9-Io2V*pD4aRCJZ}jwHSB(W2pZ(Jw z%7@Z;a5u7f6S4E|7GOMiXj42I03Cx#`R zvyYRLednB2#qYraTpo3g!rF=Xlg3nE@s>aIBHrAIlS@1kwllteuE_qz#cSM^XfZq6 zPgVhvxU!S&=0tM|m~~B7e_5ChKQNrT1hOo7f4qJ;e_!D*oTffh+St|2u3aZ~Y~f7(fTRYZ z==(&YE=;rCMXN&{H?Y=tUVBAExSb1eT#K=MX-T5r?7_ESu@OP-3-`4eQsUxN+EELs zu69lyv$4`FOmQNW7HdG~C#32vR-t6=dV@`&YupzDNV_O#G?Gt6*16^~e|IHy&Iv_k zXfg}3&8m7uixUL7#@-+y44WFnQ_)>4R)z&{k+1_xvSb3L?`4{8IW4^Ml z9OpQ=@z3&Y5IzGVB~6%aS$}o7Qtt4a@Q@ZQ$QbCaQMO{C;I*~P^BFA9`x))w&dW}; zzsg~Wj&E2cNe8z;F%YR%s2IP9=Fj*zVMDa3+Di@yFfJ10#nJK`30u_~`*=TKZfBPE zNp~enGWpku{fKlI;G0mNM4{q`-&kMLSkfzzI=C2F74){!KFoH(e>eaHm{J1vh?%`LsD zc)Pb^;u~OX8zb3|vA*0QF*3{#JRgQ4i zP*vs8o`}miPsezV>c~YC0Gs+@`E_Mf_v8ezP(K)fty@zADBF&RYp)Tt$26uX$Fq`!+#g4Z!H0qjOOua!4bCirKKF6tc zb>i-==kjogAI%&b_%Wr<&MdIEC32jbgQL6+s>2VFazlwnUk?wJaX|U$R~O{6Yt!Wz z(3POA%{e?^?~6?y*-C$#ptA4Tie^fx;ARep#t>>oc0;R2vFICE9s6(WFmY33z*|f2 zreV`WZgQtA8zjED%8vDGD(XUGWdelMB}sNgr()Cd{Vh!1mX&KA+j6QTg7$Z3DqlIV z=jsKRY73!^gLrJz?JB1Cc@UTm$;-!xHfY`Hr%m_U&3VJpW7ByZPp%{l#iB=Zlq4le zGC^GKMeOZt%+@Dzpkz_1=EuvLpZF^w&Uuk`#Y2gJlrz!B+}PJcw|%gRk1{e^gs$qC z{4u7eJ`d+dB)P585Z5CsEPuo{{-K%1Im^egf%W(6U<0vsQ@dOsJai7(Y5|A1IF!>Gh6!;L@ixGz3PSLCzAN zW|Rbl9t#eSxRpU?^)5f2g@zYa6oVEG_?klAR9Z;9kJi$jnJbXK=$5w7U|)JYGP+B5 zD`%?*qt3mv!P2o|-J!K{v#yet_L9^g+W0s`LpsUZ*Lvom_aky!VC!s}qzA%*)VTqUi;Udr# zFn6vCbs``S+G@}3LSQHopIjd*2vfJ4CVugXbI{v3AZmdMDgAg$b(A%$_R`&lVL#q2 zn3=V(SXIWKF~LV-wmi877l+(@R2@qG5vNIg>GnW1H;xI)#+QyMaM%tyGU_%FMT?P> z0hc+t8^>IPlJLM8Dg1DQER9B)5vR{V;a5HDai~3smc=BOWNLm5(=flbG`ey)(j>$^Nkl+kO%bJ+JcPBx5=CGF?>@I}7$Vf(@3 zS{NBo2o6^5?rSQtn1}J%N%G)R-gw)@RL@I@k#7h+v=w$%{9J5*-H~YVCqmWX+`_~{ z40dmL>|(AWRg{gava|Xa^>8+ufF2=3@_RY(@kCe#igdGos*fyY+-n}=(g*~M-G963 zc$5CO1!rN;3!|alLCL?@Hc33iYU0Uzt-+)v%oaz>8V+gqIYMv4D1v-#ehmbkEYSq$ zTSam5t7K2K{5m30?Rv9L!80><4+T*>a7WwtK`^5F zIhEr8Tp2vLIsLacRdvLcA%02eNUsz7ohcQet{$?D)DO3!RLMpePkN3i55MuW4_9@u zh|u$frs`+Yg)sI8CT(ilAo)-SZQ{<&(ZKtj-XB(GE*S_=>j&JhmLc9I)^~#vWvq|kx$b?2F^?Nse z&##uz+kL+DxB70+m>V?$&spn}HuoQS49V`w19LLchIP~Lcsxn{dg?9bI9xON#`Y7joAs7i4MzyM^YQu??a} zhdB@!&XlwV6g4ERkqviI&9IIXiN~6g)4MbfpWdcKM+btmf|=O;I*oP%cQmkMtWfQp zpPJ`soQ5sQd!-8EKV*U@{TYrKBOYa?>S`oU3@i$tN&BIz_c$n+wl`ew>2>k);JxLG z`{*DM)=@?jQrn%~rceS8$FrM0g!ZjezsB8y5O_7iJ#b!~BE@JR9K}T&l zdB&zQ#l;t6XhA_mKiVqVMPu>YA=k6a>R0dHBmie3GIm}M@@p}PN~v&0x(7Cet9;tbIin`wP-;m;wKAUlJQsLX=D*pKxi?sdEI!|Dxk5;i#A0U_Vp4 z?;l3}dHZkeSPVc>aup+Vh1rWrunTYLOu96*^IvHa3I}C;Frj3BeV1%re4f?y^c6_1 zsHNZ#V3FNyJ@yNA5nl7WsIz?E7)AEympX#l=?YtLm<-fY$+WOgoWj}ag(uo1oj5zC?c&JLNl5JGJ znR0OLceFhII}9$KN5#n!&%HXC9WmS*M*g1G^I)h~=e!W&+wN=0t|x`{7JwO-zTCtm zD;9KEcAWGx2`os+{mBMpjcdO#^Tr8Pi|CO_b}8E;4yx@<>Zg@Ps(I{)euasmX*mfS z=qly<$`=3H%!DBYnhz(=W9o0S;2@$oyOJcAb@60MRryG+2h)U2`2NtJZ5Bw4u)I%@ zBqgmzH;esFt~{DUQWIPj+X#?PJ7&tDxVg__1=GkP$~>Hy)6|1v8j1F>O7$}~1vOFj zU*9;-UeeyiWo5ad5Wm)Q>0ZbJy%%Poi*$IPEJu4IbtPGuMvnA?*qfw2kC>%IrKezh zMM^2jDu^Z1^!kSjY0$DQ^IIg`AEJ2LU$)Xl)kRhdwPBp1{IXq9gOca3IhsSANm`?A z%-v?G_R0OmO`H)}Wjl6GcBx2+#5~Axk9qf`tubN_SzRrLQA-oH4SS&*X#0_??al|Q zgDUixR+l|PCy|PsBYKq8$b2tUJ~Mqp&b*ut)HY9Tqb1F=L@NSn7ACoAhT5puhWP%( zT8%j}k=&EmGA-4IOFx$&1l(PU%Yw_y7-!?^K@ z5GM*XBqt5*dlgiUW|`h8^ZbGA;Iyj^4(Ha>uQEn)EdZy;I@eTa@0x;=On40mF-#Af z#-ElF45PDX_Gv#}t+2#l)80!+DG@GKXQ0{Qccr^1kQ`Y?3=sn7O^2KZd>Yku_7jC` zX0+X5ok%Q3B&f5GV+=09=c?Kuy1B^f5bWy2Go?fz)_&f5b=$V@ z3DZ!XIS#AaA$8YFK~wx~!rizv(hkozF>csbTP-9O`m$2%Hja{i32j{1N5s?4C0oC2 zP=sMDgXg-5&sKV7%D$oZQMnvbcumtn1(ki{_4or$Pc>34#&0=e&9rsn$+;XggDmfI zF>Ns0LWQcUeMVt(M~&q%Q!~OqxV((7}w161C7y8@@7TZ#FrSP#-KE4p@ z`hoj+Cwq4ir_F?O3@T4PcQ$#rbvt6BLKzjijjrs0*85R z5%;k`gKP?0vEkIE1RTCaKvbH~P>Ry2GSEe~h2db;u*46cbv`mFKF}9LX^UNCRvZA7 z%kfsesSPR5xArjANur%@k{ru$h(B(zOvmkHrv4u|sSJ{+S47sis z=e?i%@y4Ik)m?T#{WKzv@W$mw^bcnH5A=K2B4jpmN$7A*JHHYtD^4h^gf% zQZ+fJHGyKSq6f&S-n#Q}Id2!~E-DoM9nr?EGI^Il4(pe>3a!Tk88Un6(w-(B@3-UF zTfm-I@sXB=XqwIGUD0Wv8!P9t`O|;~MrGjh<2iqLrWN&VX`L+uj^sT-DQfdI^Ojyj ztM=v`<4vy(?FoVQI*rW6)DtiMXb^1L8!+4*;br6T7}r%htWa31`ukEta&v=>@znvJ@nBRm5~*9?JuI$Xwx;J~T@!QVtS$vcI;P3R3% zRHgb*JzuZ-2r(>QkZr^d9yqKrF+E7C*IKQ@G=>|Sgq^-@T5-~a32^~~IuZgC&*ZxE zK}(<;JKI&yAx!>6HoE#_%&04Trj|T^G0~*Gm-0g0xhuG25vIQwV2FJ*nncs#<5E|j zS)a)7x#(|C=k;`Xz7Bay<)aZGh_EK@JNf6vjxuIOlHRN9DB=fY=SpPjDX*GRuh!V^ zG_=dyfkn*t22;{YeO z!mY^CJR<-k7%=QX52IXa(@||*NoSxQ$v7c_mBk0KI(3EFn~Zf_b-xBO%-(vmw2!-C z*kIl8Z%L(@qx-k3rASw6m^>6?|C`D?p$ zR{MGF33&cy*fGje-LcXM@Qg95zuQ;GWhHLT#2*JRzD`EH84PO|Z0tQ(Y&JAawFlxt zYi3`e_{KbXoQ%f4;iO@>ibZPc$kg{6d{$d#q^&oGwGiGkF29?Nqe=F&`6;+r<%@QG z#N{{87WEsP(h`B)gNLViN{6P6HQ%Ei^AVcfudC3$1B&L*5Ipri@8FC?5*+vC%oA{8 z&kmP;S7S^*4RbU1;_kn*CrL2L0X zv*j*Bvv;*@pg>p(N%O#Fdp^r2c47Bp!=!>4=QC|GloErs{S|5KU?SmG9w$TZ2(3+4 z@sQ*AQSB)0*9ZcT5Ii{9?Mhmj`5it&8cxCg;#0ie%(T*G9= z*`HW97zKspFW20_nU&GSx7hA&llWa62QOw8(^dKGVebVM)k-2?(;`K>`>el~sxA^l zwIdqKC2J`fVteqWRdL-@7#$b{-3Gy)X_Tu#nf)9*7PyNx8p%S?|!#ZV+ zUptS?dI~;iPrjMD-mFn21LRhb&pC(v@V>WU@x<*T-4yQLco$ngv$+BFfVe!2Q(;7K zTh8k1erbUVp47An}s4#qaRWpgL5#z*4cWvW3ItvK+Y_~IxN_8^QJIolJTI7 zn}^vZVsebKl)ML{3>zYHj#g0%MB-5?@>d;m%uXk%L2D)zG|i+oSFRCWx!w6x0K-QZ8VZ6lbz!VI*6VEJ9u9lOx8n_&?2Ld z?H#dMWH7VTsi~!azA@w+1yg0u^G3Tfz?g{QBTh_g&Np~#BfIAqTU%}V$Dq_k`i*OddQaCoU;Lb2<0?vC0#U!cFz&E8!DArSGw%6 zg>yNK_Tx~`Q$gI5tr;PicHktFb})xEhu%b9nemH+eKFIXbD2-)(1EK6qaUQ%!F>Tg z&9$e*ZCp=H)(Xdu%94J|=9*@k55anCP3Y7%F`#TazBev*+Koh9Y^iqepg z0m1ghtm)~_@S{YpUsH9~Z1g&#$HGi3n|n%gZoS^x$L6kch8Xv5Mh3rGwpE79Q}397 zX+xf&;JPI&{;`CPi`igY@!5y#ufDJk;mwxlN3T&nHm@|VoM{u>v4`uD;`?vwVG#n| zwnlaANORBX9L#?C$6O`P!V#QYlSUB$YXd_s6Kd?LeS2 zVRDh>5u#Rr$Lc+0*EQoM-MS|{5ETJY;v5xvV2<0M9f0t}!vd$%EljQJmmhg>L6J(Vb931102-+%zTv9Tr_fc6 z*_LISzk=`As!ODCNlqUPhmX)t_ z*Dr^7CZ(%qRIz{wr-s)gz$F=6VxHV+0gmV6r1CMYkU{@DfZ%FjEO-2Dl*zG|E?Dv0 zZX4g5oR4oh_BSUYkqAHU8tM#^%(02`xp0<>Ga;R6hc5AMqF@0#%boozx#f0Flsbrn!n$; z8yDGryj9Bkwr-&db-1sxR@-!zQjIE7y>S<{Z4)Lj1yCW*nfHBMz&;D|G)`SX6m{Cp zGJE(8X}$c82-90|UZf`EA5>+|9wkVh525dJYM>nsv;f~)>p4;W{Bt!wy3i$^mEFi3 zRGpK1b^7r^*bY#adkM*7A;Ljxj(F8EjrCUq> zRtr9hgYHK5i+^&Rn|)mnJvNt2p4l3lQ6Vj|*p0-wBx{y<_fK&ln&B_H)%++-t|2n2 zT~g*6B0-L`8xrl8VUGv$PrIU3367pZ7#B%xkkGcty3ZEuf%5VP4jBdwi7X8|M4Wxl zbiWsc%e`tIj~H-0mGw}q_<*fUJ*pGaI;(XRhc;-2lTa^*_;`Gv zaB=;lLc%Av@Ho3M-juDQFlQDu(sBH4I|O6g~DBhbaFQh%&>@Zqq;yUQ1LI;MF{rz zf$Jl}g$v0<+Kz0A@W3e7>U9`u(DaIHVX4_#Vb>|U4i{aCLN18b^Xjf?J}BY$&DxC| zq!)-t8%rhf!J#8gM??k+HjNpW(&2iV|zvF;*wzXc8GMagQ@x$3W#9 z8tpL`2sOnx(j?UK>VVe$sRo9g-|{X6c?%7`__ zwPHn~PB?#r3I+KCpuC60doc4BRLQUWLaoR4ge!!(CPk>-p5lIdjP%22>(T2X>DLN> zI2A!OU@kS)froFH+f4U~JWz4IQkXrJG*U@`qb7F4O|j3rn0m-hP7!}c$&$3Vn^F`B zE7^X5bDiVfA8$;j{#e`Rm_K}9d*8m{ko;g^I`Z8HD}+yK#w2OszQl&HWfPggo}Ia; zB70{h#DQy15v4R!v{wnyBj%5srtCTi&>E4ZJ@2~i(~9kHyKteuDjxB<85%gG@ntFw zIchc)`L#Hdk=mPhf0PHuB9^fuzX^WZ1-dT3Fpu#`H?m^OqTeXOQn!}T$RrvtibR-aICO=m^3=>M5B4~20Z=${qA3v`MUqGVTTss8H z=SWUWlGqB+I7K+BXFx(qRR^JYCOU68+I9!`OH~?^GG3weU2~bH>T>ho0yT1sPbGRD9t=lxIRG9?K zY1rFN`;t}EGedOC9v992d0>h~!4B;OD$-o>Lan;wOS;W-9r30t`v7;p_(c4z`f_=Q z_e)2A&%q$izB|%<<)Md-i}4_LqFg`DQhjGEcdBV9vk}H*6nsoW-RY`Xt9%&Ai&aK5Z7F4mRy9=w zHIZQ_4=XMqbEKa5WPmR_E$3$J4lQ-?k3_Hk}5fYWj?CQwIsT|VPmuc zca~Pu5BktprGTeYqT62J=)HSRKZxAD(YRLY`THHUYMm!r+_IcjTu(Uu`4l!$tn2Qn zlIf5f3@D$(nZXisoH9(ToNysbb^H04F6}($`h{j9pkb2YGhsaKfe)B)nT_6B6}C?0 zmJ+)JpA%@W_|_cXLWCf&<_6kEP=sG-`ZX7oHQ8dT*t4p~c>7V8zeDiJY<0a+R3>{{ zL!Vrrpl_ug2FW+r-noi98Cx?qI&oI<6$p0Q6WKM<`lV^hlpiYj0<2i_rfpG9FXT|R zm`Srw@$v^rQf72m6?fm$<<`dQf$j*{12EfiFZ;N)&mt9l~ zzn^;{C=M{Ha1M;EMJL%vW4Y<3?b zUXS9t9?A3I6pHsUEKdrCAbn1WmkZ}O#4mN^3umgo_@P3PJ}uAG4Elpsn3glY-^Zbp z#RFY2DwF8nVFtK`BhYC8RuG-Ziu%iare9?7|0w90}$sddYQ^4g18 zD#By>nKj1OZo=NiK%dJ;O`%4%P}v%7QUs=f{cXCdka?#!UhW7zsG_yL^_N3Q|6?`` zJ3_d0@ym~VWazU?)csae_i>nu%jh9}dSSJt`&*8TsnXX(*)P*+?0&f1tC{0~ zQePVUK~k5RuXU_Zll83ZRHbkBNeM%BFd&<&rN!jXVV0$q7U-;GL&jetbqtfngQ&|1 z)BdW!euk?zT?@m@Z1FMJn-bp?@SSChJU|2WFhgBdipFh=H=|Y?&Ms9x%P1C*-hYX3 z&5eh@Xv=sr&E1cv5F_I@h*+M1H{&B$J$-+|Q#<4Zno!soQTqzQUbUPKL1I1u6nZa5 zgRw_ZeX6Ji@1x7m_*P16m-1p*jM7MC0g@CHbYY&k5?$I;7S0u;h&cEGP9k5QTCyo(jaPPMhGD25*)MlHp-WC^pGuUcvUtV==fVzDY}XtG`UFwS!5*f41x)H`}h zq^MlQM~ehH&)&sG z7Pl<3^Lq0QOR}GJGT}m698ra!+Od@vhj;ZVaV49$DHGDlxf1nKX$kM~j5E5*hs1BP zGsI)1s5vW?m!)fZ@tGQWbP}C=S-gh}_e~KA{(uzMK-4s-v4=8tH-qt@z>bO7pXgeK zX~K7zd){_D-O7*ci3Zl)(Cq511!=1({!NKU10)s~rLnyJjkrB>5Vp3?mn1Jp(y@xP z@b-F((1SCa^}{=?eaP;K<~YzRp%t9wXg{pnLJCoE;YYimBJ9kjt5MjbVrHfB z0S<+aYrH?0--*3kRVZxggYV2n`T4%|gxY5t+m36#I42w+9adiuYZ_eGgV7ZDR4^Tl zsJDjpRbC(tZk&H(lE>5xSB2>4w`>KX^8HI>OVbvIM^E|QuunAoeI#8xT!$ts-UlAf zBex$K!zZ6lLUTp2Frs;Fxx1WT^;gwo{xQl&BqKq4azNVdget!? z2;$&i#$ci>)+8kVx~!k!lH=!Kgyfw)K4a(-56$`YRyFTIFV?%i{eUVVxGc zZg-;D4<$m1^ryjT?n?3u3%EK7Iutu+ELn&Ny;}vZzE!Y^cFKg8P0a^V(5E0QcNL@Y z%Db5G2}1v!W}3&)rl}r1!YZ@_*Zd_GRu#M_$vae)XSfR}j|;L_plSWlzKj#lrZz_+ zpLt_VRc>5Zmn$IxauJI#I4RB_UYuuk7FEW-;3}p?A)&CWhC2x(Bq{;|KePE-Jc{FB zZeTb0Oa*@qyov{3a|mdujJ&C;V86PBZGAgUO35UBHwasI{H3ZK%KlE;1fixb6L*A@ zWO+H>B_lX;WIx8v)4a|hzjoKNncRnk&Pbsor5^+sYHS~=ME_D|R=$w8iBE zI;$Ef*W5R?u`^kLlNFE-gJ4b%(z)d~I-{5Mn-=g!oM*&?;z1trynlB5YYO6>L*^9@ zqQaoe5ijPE#9J4TmKIA~=8X+pK@LC@iPDTrM{~awkN=#j!=tMKKCbt|Jfqix{d{7k!*+BEB%`gDX9V)8 z_;Yd3I0J@mIun&WQ>D}TlAxZP1IAW)uoS3(v99(pZ6G7{*JT*YP`+*@`bmAEp%SMr zYm&~uo3<1~kHe=eB;BFrYKx0!kC8-d7aKmVgyp2Gj{YcT_EMMjz1g;8huuUxk;bNw zIOb!ncXaJq`vVDbA@GcYDI38k$N1+mi%r4kZS2gG)FJ!NyCwj-Zv4D6 zmgx(_O(v`}&V(XY+X9NxDt@G2BWsvbTI3fTyQj@3&A2xNRYXZ@o_QOi$HWt~zd91v zV~Q7xxoF;;c^C7WY>{2EeanhpW5044pR`vCpaX;**hXXmoL|TkrzUT} z^D0k8dJ3AxOL3HRW+GF6yWkVs?aNiI$02;o@RTyBfiHymImrx@h?uhyv-d8Rt_U9S zd>`Fwt&x*oYyv_?4)F?smfGrY0DMenU`K;*Vcd?EbMfJA4`8G|$jD^aP_&$s=^-tqK!(hs&yXcwR3au&4D=%KB< z$kRyk>Z<7-N^`D1LQmxG4>^)C>8qbWhvm$ty(hP+@C)`xyN4#}v8h;8_$()pgN2m; zv{!esIhOKkn0?anDU@~CO3v9`OT#9X+@o3==QbIIZ(*&}OoN!Qh{zR-LWr>1642! znrfq;`*iG9PbYNFI4CeM6H6fVQ`(a|pJ1C;@c(civTcHhu89w4;Az_B{p{tq01?F8O39$0=k zw%VuwrW!;{XS~<)P8xu=YQr~2 zH{aekc~E)}*iTZ9@Xa<9XWV#>GRfyt$aDo%d7x>l zx~4RL*yiAMY~hqo$H`6dn@Uk}QI6{L@X~qogy6O^%m4d+F-Km#^SvuP(*%^eK;HIN z(#MB18k)j%WfQVn|5Z}=7NO)hmyD2%`D zBj=y+;O*{yn&Y0$=<$Bt>uTXFVmdXhF7R1!g%>s<&uFo_7rtnD8tffPD6pn;{Paqx zq~C=!y(wo>esHd_eQAc>h;0%R3LR_1O;d!$eXII-@|>?;z74AGZqQxIKk#A|X%*;+ z@#LT8`Rr^F81&;5a0d{dSLlAw%{3GdZ);*F)9j|}6%sr&8=ihEJ?4UTWh=6WV5<Uqq;3|MoYOxa{4})QvGA75I&t3=k;n%mGtZW_ir#?{_j@* z7eSTs>zz$MueMV`h5%MTslRu&fm!kbBO&CMm;weV&stDHMal;v z|L9e#S^HPHI0m@a7nrlj_dA}fnbK!p;OmkJ^8^w%TjpcP|Fx=NIuH9U5GkC*P%jy$ zM)#v?3caV`I}_-kq`S3BK4RQPJqOizseLxeYm|I{ENQix`H3<$+W9 z2&i{h`)vBtR9xxVrt5Hu!#_(ex)?KYd$kP~?U!g0X4C<@ZYztu283@b7Vt`(6C-ya zS>Hj9+{rgowH@M#O^HDjfvAwd%8#bd$jAz7`<4!;2a~FU`tfwWg1iXjiyO^?1R7iO zA+q@!?Lh9b@a&gL=&;rB2lgR0gC424dNGc}#^4;45lxG3oVI6nb3LXpLRsTm4jgEI z+VB5qB{yKjv7R4WJNYF^@I?YXkAv_G4 zHY-JrfK0P_zv7}{*+MwpyPt@|6eL91z`9sbP8O6ipn1Di$x{zMQ7!v#ajYLVL}{9Y z8EFn^0B{D>uQZ)}PjLF#))fzP6o+tvjdH9BRl-QAIi$|&J471!HJ5=fjfJ(>-JzYX z?Hm47?jl-4Da-HIMyHK7ITX6LYu^fn((5AIAI2p*a#2lH6N#}4gix@KTZq^cZL!Jl zE~8%rz&9@d#;Ut;EWR3p`lO0|bBpXFenCnkai098Qo|^NDY7?B>S1C1of{Dhox4tu z8A(z4AfovztcGM+6sk!5fiorEy?p-FilEbBz)bve-xwD(DhG<5BCl%vJ4cYyc?0PL zqx~_8tY4_XbTwY0CWAd;wCZDLt9VFGSSqe4;q+5l%@h~!e2Jd?FJc0g<8Y_qz72*TRJFME+(BruF29lKBn!xGjq{l!;t@NdHX z^auD@6xH_MtysyT+T_@D43oMmKFQfmc#jg~Ff&R=R98rMrXnxTps8$9*+YsqEq!8= zVP6?JO-mo`j@Om0pK6Dl5hn_zxKR(HFC~~Eug?Q!b145Aip&&|jJN(%{4f>k^y2TC z!^h`x+)A-u_A};|or5C#LE+lY3jc*5C?a=f+Y8L3S*EjrfzKJ`AN@}A!z z>(N2?`~4O#tPp!p3>9UYSM76Qi~CCh=Yg>@sa~?GPZgb$M82j!H(QJ4h~vo~#ikK^ zNtAQiZds?7F4375n>zq)_6g~tH3*Yk(_}l*;E(Q2UJhquUh;CnG&aWzi$RymTycE_ zN|*G-GxNr{TW@vy>G;DwA``YBm$j?H1IL0L!*Ye!CRDw5)_cTY!c)H1x1dbJI_(@@ z5kBd~Sk(5FD-yCG`Ym0OnhUn*zaAmu6E|HpR$)u9~mJ@BUH+b3W(rwYX5(+smyn zYx>H)(Q~?gYt}^rk-)$3w$o^exE3O~?Y;AWZ+#`BuY@o9-FELNYks(Q!)>nt!Jn1_ zOBSowjc|KkmxO*mz_`WY(-GyD7g)OE)tv8glths~dooH!@tt3={^o@Cey_7zZaVqz z9{s9RvuxC!(Y}j*2C*biCbaVtMC(gpYK3(BK8ek{15hBA#Xnx``&x##&y2uu(1`DB zFOA_mqV(&VOfQUa32;;rgQH=%S%X`(i@Sd9?2x*1({Xzn7iK%}FIo|X+ zPmz1l>gV&r;>c|l9U-TQAYga(Wjfm{5prpV5%bUl<3xc{X)nW);x0D0Y9?&c_D0Bw zVy_&-^L&D$mV24XwAAARKpxbeJ`*O1Xw6BJ2S3ejo5xt8T^PBwixAHOpA4NbGI`9} z!(+K$+7VX~mBsUrty&?2K4Gd;JSuM{??y4-)Vc|JSxk{4zhcT3|K}z8XOc=!&t$bX zt7&ywiNYla@vjwKS}~(C zb@L_os9uM$J|1!)``FGls^st#ai8gjLz+g5A$=K%l*@K3hUU_s6>L+kQiEjS z;)x_ziwjC5k-KjtAzWU%yt8>tuZVa#0v*1Sb0U!Ux@ED5Z&Lc5m_59Zk zpM_BZ=b|v#rTutPIkY7UBzPegHEHlb zPmlB;UL>lC$+d?GmxZr45?Bwt53B$_^OnTM$TAuU!wnadWPwJnbhVc1v@OCbt_-;o z)!{2<#zP!JP(2VP4R|-i=~Vv4pmI&Shq~VaXYJDw{k2tWkB!;f(nVuZ6_W`Q4>J`Z zVHG$nA9Q_Wa2fw)yAee3>i800rj+rCL4eI>G4G`4qieS}vKD_GQ#$g7=Bt|!}O?kDIm}wT%3nMIPJWN?$pGr-V#yAjI*GmLY6Keq6>neC{VH})^uGL zr{#$3zWSh&H-7OdHW1>F%wM}4g)J$$HlqqSEGVkL*bQS{FUNgb7cXC*(P$6{fq~b{ zhnAysxo+nWDIcPPft1$?4|ydQ*9p1Jco&n~Sm<$<3ulhr?OzqO+92bMI^T%;vqYhY zPGXz%;mtf0)a37TBu#KK7?~_>o7bp1CkMP(-beGOMUb8b7ma6e5*Lmx44%=W%{}-9Xpb>!$PVIg>vdC5f9ew zo;9VOsw`S%E`eSCL4~sQ6#WZ&CLNSdqlTG$D(c$kcMfr-LFMr&++UVzNcJ;*QCq2& z0EzMIIv6K&H{Ul3=bCM^ui%zxTY4WJc;r3jGZe?YCBL8b!DLmtP`-L)XI`7XWu-Fb zhW&Q{Pjvj+_|SN4Y9iF}je-h=Im}FxXmr7pZRsj`gB z>ggRS1v)i-Z)KMp?llM$PPM>1i#=()Nte3?{cG4Q^AB*bEby9}K* z4A<#`;-@5zsy46ZTTx``XIWoS8jV-85smBGg|0Q($Z=vQnb|JZx%Le;sZM3-ss3$J zg=F3*G&w5OZkPd&Zz7BM#(-P`dX%EecB%13B(%OGN;PUUJXL1!DukMP4)#Wp;i@0H zkM6c(Sjt9NxmhqZA>tKp>~l<>v;yZcQsLRdV~yDUx1AcXN8M}y?*UbnS!WS8lb>37 zFKbT(P2=lPS)x^OFqU?>l?9aRPn~_e(POPjvtfA4uDu?HW{(}4K_DniF0rpUD7wW~ z?LM5_$9_ATS+C{tu{hY7v(ma>ha*W>hOa!s>Zbb`io3F)qv2WsKo;vwh*##Kjd87) zISYA|O`^%SQ2(}bwPw}KJh5G1LEltk<=o`+3A9wiTcr+zF1Xf%&!oDu?RQD&L(!>e z{tg@UtC}?P8=;l2k--hrDenX%#ndV^&7y(F{iD!9`ttf0bhk39X_X%au|4_^AvkI3 z51YpwI8~A<6&#r0tSew5#(^yvv}pF0wBBV45Q!p>nIrr9*oB0qKWOhA_9^9|7=QZHrqN*!YRtP(j!EbH-~K&xH}Y>hV_aeXmXs$hX*4(mZ}f&YmUMK` z65)HdZ%jhTzpB5F4HOJULhOZwEg{r{wLQB zRqmKtc*$$gla#_pNESHP+}s5rTI{Ip{LohQtMD>74H7^gtL~q<5(}-11BVO<-TTv6 zZmog&Vhraatb|&X7QgYQ+;H7~X{2*Y|7#FG*4PTbDdN%BXy@$IE$`Emk-Ak~7WR{N;_=d9j9A)*evzBC+1n;x1(*Jh1DULxR z&Aa-2w|N9ovyqjNM~F!il?M9W2(lx&DM29EoHnb^mZT5H3hQuZorXl7juE$CT~5RN zYBb4oL$clVMCo{}#s=hL*pkRg8kkz%isSFlbDeazj+Rctyl8IRFG#DE*6osb&LzPv zs}M~7j^X)wS9(85^C&W@CrQ+zK2f={6h^N?!rFTlq}A?AmwK_NDm8qMm6yR!pA0*T z4arg1b*|4kHz)?k)JHN)mgwE2-Wj444Ep)r?(_4|dtn&2mDwW8TQ` zIYBa;AXqD7e9!lRCLJQ0xthWUdGsRYbja#)I!skA3HvbI(SzYxK-L>3meApBH0VaJ#?^Xmb=Wqr0I=* z71x|QigClbpnaFq1o1$D2_u9$;PZU6OjnACqw;m;4j~p88$<;s>Bi7d4_=*|k?9B7 zh;Vur(dxT7g;L$)FI23z252tqn7yDpbpWUz7YPF%n7?KZ7L*n{qDpP__-TW zN{b1LDo4EE!W|j;<>G<%H9pH5(QHlVXVPYNRM84x5;iGmpIoTuiGgXhV*fzqc|s8< z&(RWVvS+Gvrt~%ZyjV;nxnKR~>FTn5)Rj^^W_XGj>%JnXM%QTMGZ8RHg84cYQH6h- zMr^$$%KtZ$^mcrT%hezY01*&3;eIQwxWZJpvdyXgC8~hut-{w((#&plR#8j1H4^uG z%dsL&u-U4A^YDqRpEF|O@&XT8NT#<3U1N+wX)6+t2vayv({?D^i>Szn!7$S?zM%v| zdS$r&8dtE?pRV&2Uek2;w$P}8{sLi5yQRC=lJ^GmdRZt|t_Ptx?*6QHG0E3_%;qn* zVv`k}&2s#%zB`6hXXo5vj5ue5=p z2%aa6c(|VmQ5Y|CkbZF;-EEgDNjelo6^RbA$2{qNcq^vW~u6yFRJq=c`=0vjoMl8^$=16j5)VSY-TEG|{=eTzp4? zIR6UNCx{bWmKDzxv>~MiS zuPU}ooy4M#m4eXIqb*SFllp6bbRxv?rJp0o!On5D;xGGznl~3th4?j@uJYMnRRvVp zR}=C)h4-e`b5FL9m^aCmeE6p{_*n}_%1fR_iO_16?i%eQfMsLog%W+smpe4*up7ir z}dJKV&L5Gec_D8tZ;4QW0>N9Oe%2n`HnODp|{ImN9G8L4!4&U4z5+w3}P zvRdGpu(7)&xCGoU-NY?H<7k?}tw!uZckzj3Ml1tnElo_AQOW!knR>bWQ6GXIcChyj zR^_B9!dcd5#Q_c9fjE-v{h(Wt3Qcx8we2WwVHITz)1aiy@CD0H8(7BMql>Ed4l8Vw!^M4N;Sm6^L_ObXImnB`j;XmmYRRTZN8$c2nuF&*Kwb>z5n8)(x1a z2YX*>e#PBIa+`8_jV?C1=;k2MKTEf@9A9dY<1?I*%NjR(w_P|g7`hODerqqn4$*2# z%>!%YT!qODzAi7|HZzIg@eR%Kl?Yy*!WPt1FRu{R#*SPtRWegyNc8)<+D!%3-HD6u zWf)F4ea7WQj=15i-TYGvl<0de;Y~_#*|Xh_r|Isp<7-?QD^qH0{wEFF^QFL5Y!RBx zXG$C$EYZ1fYwGy=RzoOamv5qx|=s<29ed`;aTjgE+JO2_f#&NdJ)7vH>ujvgoV6hlQd zt+kL3Ea#0X;-&c&=;Wvbg$50UYyvzTb7YpDohYwW6=zMz$iS-65CRO1jzmy( z=jlRPkWU{TT*gH^b9&F&-d^-3hw3hvVit(R>i&c#oWGEY6I?<}aO(R^5|%~=T8#iE6x8_ef}5H+P{f7c*X41IhQ#vY;BEtMss$GWke*)iqq&For&R==q}znzbT;z znUFGU!?y8ZVi<^*qqHZ&*cjy48Nnpatv3ujnmf@q`XRo&&%(e}!l58{T*s!&vv>Hs zj23siKp%)K+kC!2*sx8@1&Q$kD^her7dbVIrCWGDkefLeW|ekIe=4TY;gww`%yaTL z`A|?7SW_&rz1zsz=ABZN=uA;YvW0}Lk>m@JPtChPgJav+o>;5}L_X?@bnMdbGAd)L zQockcjsAYyusKs4lhlO-68`Tks3=xB<^VfjO9(x6Rw$ITY)`y&q?{QcOZ?6l(8D|5 zObi1`b<)>Tg#B6+rf{Ru9O_$}$ATxfC^@)OPPXYZBiR3G!NWOA-!)-2YfHVHX=JFP zxlz1sSLFJJMG^U(6?Vu?YcXdcn5cUnJNk|Z!0emP2M1;NA?}nl!nJ8t29_#Nu|URP zf#Sz#aah-PP`WZl>VJchp9X#B<@o4j8(Xoimaf=WLztB%0jqpoqWn zZx~?$0|gtTDhN%7W;#Ib_$w45MNc3!KFg1-K~H!uXyw@|MRzzUO;B>u2!;Cgpq!|#S7>4TYq9b@;^@+*K@mInR(+UtC3Z!bThD{ zs&&aae4-YtB$e!UarP)jt7=x0(d#bk7qkAG1$Je-ovu}Q_PLOkMXf2|<@sPt>ca|! zjMw!!oO8E2So{Mn_l?o(R>E5=cpCD?YioEr;s+ZrpVQFr$rks$-jw!&%)0r+G+sK+MApwb!XNq-^%tN;CDM;0X%pv^X+x`a zsqjNNnBJpS;u*T)ny-~joYi@?jPN?4t+4&2tBc`>dz@^%Djhxm+?W^rnERLw6L4Iv zeP2U>Efia59#Aa)CX85c7%L+BLsG`*?rY(yYDX@wVbJr*S$f+6XmyNu>y?m z{po-zeVkpNJH)M0tyjDZ(_gEe8EtIPR}QCaKthRi$L=&HUVA4mo%=uebde*%zbLIf z=LC;S;84c>U%Ds2ga;vu1CnNt&u`>8T{{vBnl>SZt9L)m!3{@18I~fnoOeXNx@}m* z?b7p7`?!oTZh%O8E;98p1B~MK^d4ca3BtDXcN_Q2IA4}kv5NNI7fTy1gtEqIiWCRl z3$qo+FdI-SHE%0QphU`U*8j%xh!SI=Uy#!X?;f(|v15fCO#J6ME&c)&^Xs~*SD|HC zmbPD~?{=#`6`f^t1vJ>O9Jy^fGyB$3n>uc8Kw^(rYQA3{@|KsDbwJDDD5>wx34;P+ z(-C2yymx=95RO(+aDulH)2dRwx%h;Ykq&elj2wO#1OMJK1}(9k#>~111)X(GF>5P# zIK{fWjz&=XBZo$vj}k)S2T9= zX(^VO=qOwH?NO{G^(+#<{TnfJnM2oNe5H4bFwY5mTqPMQSW4)ZgAjA9VT;F!?g@^F z*g@DnDRNf{K{Qp9;?I<^B(vo(8ZqYiDz?BN8P%h!VNjz&CziUh!F>X@!fIy1gEfGQ zh*0RO_Ne(sX^1lz&*%kYLOe86>r?(imnZL}R3nq4l6ZY4D5TP~b4<_`BKxV0J|@X& z`4<7{VSRqO9fNszRDXaN(wx_cc=<&o;bm%&k0lBfrIAKzlucQnAhBX!q#M7^@Y71~ z!xyp}rL|9bQ(CusXZF|AU8(48<;=3qL>LTOFzA<0l`_{*_@I^M`{bXv1T_a5ujU_MO4uySj6*Oaoj+GU3dE?Q6|uOIlW3j|L+3W z1*HyCOw~Yzqm2_X_67UY)OFa!cP!dznV~C+S}_2}haCY1q5A^-Q3& za;bOF*D4W5*eRFs3bOR&KC|(5xy8nLrJenVY*_pH5ZuelCmSiI@F$c^ZAH2x<4qom zwC70$@97y&1*2K0*In7Z$rg=6!)x|Hc0k5x70%o4W4Xm~C49QZ-oujun=X3$f~|8s zqmsgAWHl=@=PhoHyZ%{HxK?|!gc@WDkQ-baXq+0lGpm(9qfP)&O)MUBDbuIg2 zPeStr61(wTLMWq2i_KeyWM6{-w6=k5v1Nec<@PvQIm8D)OFr6gO`oQ~kMV zg7h>Mep<=BKp$#2rRB>uj{RCl)*OM7qK*+B3!XeQs@&NL9S4tH6RmgVil7Lytn7@mR(wjV z@K)~!!`jL4N3&lw6I@qc?n_S35+=yFuZ=uTYrR;Fz0V55Nl~x4%A3Ld97z@F0-|T3foAr0t)puDU72$TrE+C_aq81JTp)21}^$|oQy5+x@K2BiR*aik7zC#GuvPiuM4>^C!KlveLl~c^h(j#ep5}@Bm5wKZhKM18G@=tFD@^nNv z_&loHFoq8M)zWFpKP^r%?jzXhp>Uoc3}<`bl_lyX?CcP5p+6hVtK|ywbrG+3V=AFQ&-@Jm7lX0JmlF5Evva)n^`{ZPYG zNj$dK9szQ7{%=*+%#w9WvPt)%+l_M{x`{T6A2D2Mw%`S4`yG(kx_aN==v*i5oYd&f zX~|tX>oew1QRHn>7dxuy4+|J>=a2E@eol+gO%91ADL5Tgd37KVe=JEth7Y@FZ2~(e zyFKm&&L4JCI?d+;|Ei=lk=*;k&8<)kcmD?Ni@TAw5>%+!#r4c$4MEed<}LTU5}kB* zmF92VI_0MXD;LtyA!+ZAXkaU9dLwm($xjoYSgjj4zD?Lr)b9c-)`w`sKn zYa&^Wo2dupb4b?Ynm>UvfH(v#vq0hF8)j5xxP@0jYT+3H9XQsN{YjnYC!)~8Ub5>S z#mmFm91*gv9N%0Bc&RXGI6b=Ef%WyEQ0zB&jm;6mYI-bWY@gnmmB1E={PLOUW}Mz5 z$Ha}i5X%BNSsxE{GqoS!I;2O;WXg{4);TD`BI9LtZfd(qBvhYy5gf2`NcYoO-nX;{=G?T+}Rh(tF{`yIm9xff#2ORt<`MyP4OUhRVg z#dhllyjqt3YYJTqNr)LI0hySjIf}0l{B8XKL#?U;xv)|i@M|#>2CPy_ZW0zB0Xz=j z1nsb$q3-=65U+8sy-#tGn_HlGOZQXr@ss&fsuoJ0<8u=IB_%>o=0;FaH0X&k-t4`qwFgbCz}*`9mF*sc($GorYgV`i;cKxf#}P`V#lv@2&Z& zmmNP*w{IQvqhg)D_kC%cwsYkPudF%9G^6NzfA}IET=F3}ccQbwygMIm3T;pNa7w(B z&Qim9av>F|?X|LzpuWl1IBqj;awjcpdNia9q`z{t&=SAud4+eurUrVhqP$Db*m3je z_;wzBXajf?c{IuG5iV&1juN|E2Pm7|a>qTjq&4X$ISQJ<^FWJaGW<62vI_&&teMv= zS++oD|39K^7Jp@I`pEC2eFLgDqH^P=g+pWxPe@#%KBwEEsF|?TXnxRCdq+*p5ER%j zoq^N9JOuqA(9|2UIbh@(!%P9Tghy&T&QqKm6%W-|;|zlGdZ<{&0dV(RuGqCqseB#o z#Gh_e>78UTq2~l_KAdtZ#DIemUG7a3=g{N%Rf!?c8Oj&=$33+sdZZ6y8qAq#WltGl z8=1?IU=H)Sn5NI;PF4^YY2wBSEL99^VMDesT%ztY+nL`;F3` zhP@(%U0=LF$<3Cxt=s5OJj(2H%4IOl?rlD9F)ZR0Hhd*ug&xbsMlwucqvPBJx~kymy!*qH$+Kwkl#?-t1EZ z!-qvlxM8a^^h=rwi~FkIsN?J_dgGrMgSPm19NW-CrdxdL+mz+rw32~iqOMM>{aDs` z8InaX-=O8Eljmj*@=nmX@K;Sk#&eO6e0gzJy|=0${0;(}$?e>unRA#C;`` zGNE(=(nAstZFFZ@5xjHp4}uNWgO>ZAF;X3pjdTjqbgxCx_?(I zYV4-tReT#|X`qIV0iIf=ij3(vX!~Eb9iHAfUM|3-reo2-*Ir7TJp_6ppPNNUX7!Ma zhb6}|bjJluF9X;W)?larG+2cVHd%UA_50U*yI{n861oWVM5RcibFRH2&aa=(40x;> z5hTUS6`P?r$oWkRa(-Dv=2lI)>5$12EGB^Xsk%+=iXVwKiY6w;oj}Wkl#82oBWUvJ z5>x}MqKW?zc!?w_TyZ6-oKUW*F$6(xa1%m%s;P3a%Mby$4 z5`8Wrg~f??Ht9y#SYG->qC17)|5hfC6!iH;yJyy4 zxf0l?gbzlihSwu_Z~ZlkSJ1qhIrBJ0n84O6fw4fw9y@X?@LBlKkv~@6GKQAa zI8%gMsv((4|IF4iyP7V<_MpGwkqEkQs!o+Ic65<3Rws26JeR^A{1$Y2I;~RE_AKyg z{Wd=y^LxkKP)jSmDgH|gO4G%Ims(_O&6lH16NX5HRYdDIpK`XeHUEzMa!;nj2(Mf;-`Ti0K7RXZ$kT{v`i&XV z>an|iN*>~wU*CLl^Zxm{mb%IM1SRL@EgBgbQT`@vh*rG^`$*m~3ON+&PbQ|OnL%Rg zqsHei7UY}BA2quWH<3^WLbn0dgFX0&f=G>tzEAZ^3GBR|(P3P#cX2FQi7E}rm_*D^ ztOJT_1wuRVbrg{iK(8Z0^-pb5TU`Ao7sx`4YuKeQffA-2nbhYXWTp|cuAo#AWExt@vrL1Y z5mrMsWS5(X=cs-nqrQk5?%SQax}Psj>08BzZ{b`c*Zi&?jiMHc6&85$sm`fgonXnB zYyw&cu-(?@!NJ!I$;y9uno@|xuy}hwQ&C71Pi7(iv}QI5P}v0@%(;Ag#lE$QcCic} z#u-um9EpaNw_o+oM37(f)3846p{X~nHkU;}HT@J6L)~l!^Lsf0QHRJ7r4{RVecXAg zY(_YTS}PJ7e7#C~$$rLXK=`{rS@%;;Tf$a<{1hAYKZWF8R&b?E8QV~DCC`h!_-9qj zMqx%wQr&_kIH2d06J2=J%=Bib3xh;i^1ezm)~lgIC=d#vQTXRncWix;^@SQI9W1va z_q8t+A-7YfnzA;#i}A-i5Q_6%q-Dc{?Illip*df=8^G5rfDbCZt}L1-?w3;=>*(DxU>n)EGi~R6s%)(W2TL?|Y%9CSYEb`*7%q z9Bt92k7H3MAfj=QEmqIu1q%)JH&3I6;5B8mCo0JGVByIB-t&3Q7w-pQ8Ah1mf+ccL z6O6fd=B9FG?ZljwrJY?w7ctKgg~TD2ro!1OGSigS#ym)8c)y!f^g^6WIXB8vZS)4( zEfZTdC5U6%fO$SfAFDE@bU!-Sf;a0^;Q>8CP^=_kRYQ3E?^v|xftLi_8-N1S8;)yr zk!xx;H)yW#8SuKJ9~CTs%Q05whGH0|swnGN((tRU+n16qjQiJ?5m(-+7p^Yp_>!1& zsnXqMs{sj{&wSX%sVu*ZjE4l2V+Bg#TV5o?+?{#l$g(ytKbhLYZPKWwrGk}J*&8DT zOPw*uzARA`94ko!;*HdRoGUbbcxcql*;^Xpz2UjQjB@W4ib;A$k~A{UvT2TS)6ssZ z6YEfFHih<04BUnF?N;ABeAI60GR~zft5T}$Jb!`E(^?c%shaGf3>9h835Co|HD!0X zoOxnHVEv=VXWkNg=lry0v9)OkGqJ@|``XLFh9dI4>*UZMy>Oh8-JV2_&JI0&TcUD? zw~Npg3qiyavlsS{EE$6%f+#-SGuod=dsgZK`Qq_k8Mg-rao-Jk5IC6Y4ac}vP+E7R z(7G5}ckVpIly=}!q0bhqA8CnYzjOWnq2m_*I31cCHu!POP8YMI*G?^R` zr47e!2+np%a*%zZ-oe$d(MW|ip_~%Szx?N`*<1f_wdL}Fg_3>75M`xV#eB;*xlEx* zO*p2%ybPe>5r%edw+pKEMepvkZzu7gzC{hmr0R;U*nMO{gJyn8D;I^dbUo4@)avet zB92w**RSv8Q#+`SJjfF1|KVe~iPi>{?~Rh%-K9ryowO_)=A;z>7(jJlQvR2v#9$H zG3^;Dx|y6)>IXf+#@sV91>pf;uKA(nSgNX~b*cyH1Sn!5;X4KDelJ1hq?iV&pyr=1 z`duProiZ~>uORs8bHW7a6_x}9>f(93)l2KaqPoPGjpbJO!$v1@`66+VrqQsL{8NZ;c z(i&x)|18iWu;nF8w(yW^gPN zB{ztxcvuSp+XU?FZnC%r#T0~tT-`|x!5!Oh^gxvI2p3=?Y@IqPRB>bU-1@-^PAcp* z2kl8bcV|;rrY}6SlVTFIPewL)t4qJ=EQc4@>UGlM!u?KIenP#}Ze`oGs#ms5z<+X~ z?T45tyM2thQvy57YP8Q4E`c92t>vlgt9TcO{p3J#PN`4x>{*S832xhs(Ei-3!HK}= z%G7(Jl3k$BC$|9=CI*>LY$W^KXYmz^Qz&MlEWumw+qE(ygq0Ymik-7p?)IoZdg!&E z>E|DI_CvKUnGu<~%IXQopc$-=6`FEf-pZoXNNNwA9zL|PE_&TpwFqo8qp&xtHKBn$ zyMS_^m#anf{f93CbwZ;7XV)+)u@UQuhY|X~F|i1{LhJ;gud4Ps5?z znW&iRZ-6b1-&?^n=A6UBO?P+n$YpGU-Z7KDU;J_^6yFXQw3ons6>f{a%GUU;Gl^N$ z7l-;y7Ia4nR(mxgPSZFl&B;6zi*^tH^-QTLg0|nN46}JbxanBF54lR${%((L9;{JN zPbR3LJ&kURtWQgNtbyh=-AE#%OB?ifYuh%}`qO%7?rN6m$xZQleF#G9!VdzCAwM=R z>_7%zCpl~V>62j&4mn>k#6BA!*CrdtGS-`-JF^C}6CLx-9bm5ijp7%6EDUyn5~ZFU z`e8NW)3<8$M>+_C9)1i|Mv*=kEY}DmN~@Xkw6kVtUjjAB7kTjh7C>dT=f03NcV7R_ zaIT$E2aMs6y6=5t5fdqKk;i@R0!q{d%jVIckmz=o(Wj=BGo-a{S@7h|6q^6K-j<)a zY)wz?V|S9|{S3n)Aszo-M+!kr=k)IEgP-)Ym14RF$s`2)7&8)@Z?4j+)9-urcc7Dw zAy0h!fN)jhYh-VyC#Rn z!o!$dgpJH{jXiY4J3a!&IV)gt24yPss`8X&OevTBLRe9cZ`3w7xI0-IArn_W8 z4MWSqqL$b?5CcNC&~l{`>8+A%eKu!>M=Ph#Wr-E6SDy|Squ8WR`&HB#aoE~j- z24m~O&n%yI(WP1S^DlW=Ae3jWk4n|x6OFNE7~7j2nRT~ee1r8fU2`?g&1%u|r>t>X zRp^|PCkW@S{E=>XzBuSV$6`?;QbYN^t2AR?N+sLEnK|cL`Y7-)LooRXd@xV~VXBMHuFi_<;b3S5x#n_*s`5Wtt&rml) z>%Fves6G1dXlm$#{wrIFWW^>5=<7ldR``l;{^X!j zbGU2kAhMPpfEV#ySQ@vfP)s{Yj-#a9;RoSD$gk*mO|x!8qX70yNH3p7;NDdY`3qln z99k>w8@0zsFY#BkLr_of9_ryKIn6dw9MU-cB8%&N!H3D9MM%rPVUwfMU9cC z{zkSK!uP4WxRR@}XrX!^*EzxnbcJGLrsiK(gtpai#ALFPN^aCNTf0nDy7_98-xd=x zN^;|HzsrD{0Xpxdb1m+1=!0i-*o_ya;$MewH^u@u|LI~dM@y#6Js8h|c7z@?_Q{fp zdG~qyA^R$-#}dZa4r36TjW-sjS|JUh0X$=niLaY=q3tHh zWPenT&RqG^;AtZbN!}D9GxtmB?bwW?C(|S^C01oHsJ7b%(Hwt;962>6-i*;b6mn&%AK76oSTRT@2Y5)UGC1v99+; z%D!;s6Fxj7?{i@4lMIl8XP~H(Y}Ct-tivbrvARtUBV;f^=843IUWgNio|X*Gn8k^G zPD*VADSTl{R5dDNf9mNoaiO)60AT~jrlQyf7SlzQ3o&ulLmR&;OXTDl8E42O zN9utBQ|^;^)VNn35fQ(m(|+iF?q9?Ic6;0PX>GGxjzHRZ5$jRb415OdO+%QKFVF4t zWu9A2J|fm%%gPOaVrY6Fji2N=w&b2B%e+?qwA7=&<+iE|TaqfhB-L zy4pet;aoWckTeBkn>1bBt7XAlspy9e`LiHu%efog6YK5~d|K3>oa6y98;PrW9z?<^ zqvMqOduxs;!U;Q>(z$f$0bAB2R_uKz$|~Dqa%I00SCRtBI3T-2-h39jt!rzx2f2f z1;tFVL)tL^(#$qmOXYp3l^;H)XdG@cDh_%Y%22IODnh3Y2+79?6yki~uNP~DMiEFq z$|{dWY^s>>#zfVm&&|6@OviII^fs_8kmG2N>=!khI6;?CQTEwTb?tWw?GHTNXs_wk z>I^;!VVq0CtejWj?+uqwU4KfbFPugU?=Qo5EI^&1)tl!cMG&4%Ugtc-!yD!*KgIl@ zzp3t|jvpvr?GNxzTRS2jblI#j(;Ugjw`2rc$RTDqcE{ru1woq6f=ZU88rc;uU8=l= zTldc8PG(C;V|1WUU!X>jd~N}27C|y{sv1ycS@lmAUf9{6?U65HqM=%m*Ulm!I@JZ$ z?7>$F5i2ywZ2%();&ng3#?5n%Y=}kzV271=C;ocF%O1GV)tG&Z{y((I@`c_I;Z(|P zHIsbg;8>C{Im$R9Lg^e+cWgL z(Jc+%imQlY7&k?V;8Ws;42bzsIT8!CX;1E<$6eBQ(PB6n`EeUjFJcYqDk-XH9V^qn zuK&$TvvUKF6|V&6^!jfcyjZ>AUj@ZGLiVnivgdlBhZUoqP43~NSSqHAU*TJWT`~?E zAJleubF6H_bh4gN!85pijc%%B;zaO5KQ*Ht1z(60r|6d;L9}R^(%KU!+2ezqG!Nj2 z=OO=}7ya^_EpSC|(0S-95dm72JGlC}a z`*m#tyKJ!Y^&OLG*g)5I(bGF9!|-)^;D^}NC=lekAwI}D8Ef`CJ9d?pAYk<&_21AB zb6U{w)f0zvcy=YB4-Cxs=1^Ja<4i%!iv3Ajen~5D8^?o;52jIxI|hKxSJ7yK{fGEO zhIN?1>9!1PA``cll289GhHWdFxN{W-%%)$z^|T}h^CV-9ir(pPeNaTXbo746G{1>A zQZ{~h^oI&OLn5B*(n~QbXAYBFJ)ejcsO!m0{qsVuhVVY=vdT%X<1u;@`zwcTxW!AP zr$b$A7HOkwvHbeSphn-g!d!w6w;9WC9UPGc7@}EkCPPl{y{BEJ$%a!ipBv2~kcSJ1syEd^x9BaI&z0N%J}(TCg9BS?O&=5M}>l#94((MqHBt(C}z2Unb4iRGEO zFkZNNeTdrhgD!yNPYd~=Pu)7fqC-3?UIqqIp_-RCD%dGJrFqG7U)M**{Tg9D>Ygy~ zw-Dm+i>_(ju%NXMZ=f09Q}_lLup$z8ctt%&MpU?2X;&hwr>9lNBcsN76=XIi^$X({ zn_f2&3{vU1ylJ5qZfgCPSgI8XJJ2Y*{bovlf(KY-9dJ694jCHi1M3jAHbEHn})KcKvk-NzgVni`=F3d=@CK5^-v7U;@Z%S`wJQPzM zESAx;cUkJQcJDw>hn3@be`i^5tlouR%^Q#bM-EX8b(JJx?SaEp>e&nD3eFXPN6U-0p*uKOpyA+&n@63oPJCE zi(IxT++R?2!y}rrh#E++Ii_7eZJ^2bxR6+vSi|HwUa|9EiXhS1cB09Ky{l7>4gH+Kw8f(JqKANivh3Gv z;$7P9*B5a_8KK9>tApos=3p#lEOoxrqg-CYe4bSU*u`G*2f~tk*vQb`@yVQ!FrpHTmxD;w68o! zs{}jMlaF>S0XM{Y5T8jj*>CdhHoCd*i0|D&KxQ}ldM%4+{$;|Awmfb+-kj|c4=^q| z^^We*dKO3G9iSrN5}Eqm^4dMKP`0Bo>DAUFKVyK4N}K`iFquy;hq!TqbGpYvIyJX= z@C$2hk=2T*Yke=|yna2XXS{_7n?0vdP1e0_@r|;O-8&aLKY^K;U;PN>%%uv)_J~!8 zQLdjkL$vs3*4b#==1+*gP~<$2?}Xnp@`iIrm#QDVO~y$q&IM~Ko89N8%k!*A?;xph zC=95icUc}a7OB3y#Nb&bt*aAq+J?TfrYSW$FaZ(KVPhBx^K$c%R9!Grq~``{9K+K+e{}Pw--`IPhNi80Rbb+X1J2gpG$u@ehrDf zfdH}M6XVk(^i%`ZDjL>|d!w)RR4-l=_tP{d*Dux$Gp$f$608q={?XU<#7d8S#CRmV z$GzNZ_H3hT#ukt1Y}G${uYLdQXIeX-IIDq3c9HgeRG1UPyEqEk>xZx z@)U(ioh3c_MZRtkWSZPvaC_M{G%x-xHYZKoi6!33FC~;@nE@rg6rj966&8G0`*RG9 zAY?NESxikYB8sP5k;vepGbZmo`WI!wm|GCNi#wG;-*&i+r~?1B+(Gc@5LVd|HSH6n zdS5LimiRxFv2FHdfs(H)g zqKoy=3_ygl;SD4x0$y+UcVNX3%k2fFhAC*=Ze z*4|fZKgLX46ok>MHS_XiTUR$RD>`OGYYGQ8+o=(a)T+n^=Q+`(?m(XS%q-mLFbh=Z zr*i%gE7+o-Iz?>p-IS>MAvEu!W^G9OAcp!Fto&XQ2VyW8PVC)8wnYGQlz@2_7#3_` zYq^Z%(0RfhKke0?FKQuI+E$)QVG_Xk}KvQK|>zivRV;kmhldz(_<7L1- zxRpS*=nHWS0>FSzp9b$Bx7R=756~mm562$o)Dx{Ja$fjt@Rbpz8PgsMk;v=w><}Dk z_e#9`a_|>Yd~#$~n^b5hY3CE%PXttV5G~jKF3F>8w*(3U`VrsD{RV1U9~10Typ-Z!NpL&05BVZ+TBZE~oJ2Gb~Lhu$p@idl2^pGsm$q{xlCu z3I|XPMtQM}zpIF@nwvnpM-Q0rBYNbnb>G@eG6UT#a*d#tWl#`D@=Gp}k0r>#=$sJnCunh-qdq{bY&^!>I4a-XZ` z+Rt8j;Jy50vF9OdAbv|r5cZ@sS_hSW%MXkzU-vZr4nnZxxBUu4+mnx=Mi#dV{O6%9E{<$bzof+Bg|GbJqR5t_ z%gNSN?2{|`ABA>0qg!sk!E4lYdL|OhPt$|aY-F=^(q}2!6$=}+QD9>2>?OJmb4kX= zo`d8UI_obMMf3k>&q6gzuf&=SgKm5;w6YR*my(nGg`0w?AvK#0P~un$fuIq5P=&ao zA8pdJ>Po^ak`V?yAsQepBrXB!``6M@09Q#;9P2r#h`zyem?G(+5ocBbypu9WI#^zZ zG6wt!RG5k>yI4Kbtm~djnChj&t|zs#<<&QPq|qGiMDV{eG4RU6}wj21fX6v^+OEYIavpl=l~*ejFKKs%QY_vWR&=|V`72t z<*)sccrm7OO@A{?L2oqQH}Qz}n|CfgsXGK>OFdq?CYcCt__%mpo>&cITI&#{iUwA& z5j+u-;Hq8vjX4V~JFs7*Rl<#?@7mDWsLBX7mr=LitOklf23H>zZpUHT-TqIJEJ)Ot zxaF1RM|chtpvDgrHO5czqU1QCtDmxj)^K)gMku^7c_^c|4oyDcCcM;5co1vCqJ$kJ zx|3-SOnT|RnR*Nelg0`Cu;%fE5cTt%K{&AqG;DuU_q$OwliWB`fT7AZ+2AtshmOAn zt8d&$CxkZk=S^#O^x@0t^a+cy-|YlTy6!uKX+vGL*3^{L80NcqB#<`OY@Je{zo~#g zczj2lRL-n5@zI_t2;%p-f8jF`QTCh%=CPS?%zO3Ry`qJyeQ2r{>E+(w3Tpuv*9l)%%Up>L*K!B73;$DW)NGjORv?HPP0A?enS_OiK-rL-1;n7(BhdsG(FiRGI~~NnMPd#Bl4@$Z0*fQp%dDgN6Mt zR}{bHm*p8oT3a{n8}r?*Ze1I$HnXaMUiNRB1>w}v3QE9wGnDNsE-m&lLDs5d@NR0J zos8LQ(*twcJz(OVGM#2JwP9Q0ahr>I$A&ZsDqWQCRueo=MoX~JmZEyU#(ZjYNw8rF z{kE^ewy;!jB8}kLVU9E}86l@tCdnTbgW+pjU<6aI<}sp(mNe(H9+fWRL{>yhn;yi? zt!yi5A;Bb$TK0m78Qn6ksr7*XdyD%DaF}<~4ZovBcVF`Va>nsSIP>n8UbuRiC2niR zl-A2WhI_JU)3|$P_uhj_Lt0a-^pFcHaN@xcgPI6BkAFC?)<>KO$)AuVn`7I>eb$pyQDz8A zfgZ7TG8!08Qg!KVl1z-!1Uazwu0UNxbP~Sia(ZEf>Ojy@ zpIxvZ@--inbTU7mz}0}8V38fu>a5(5svQA3F%i|abe!6*v4n9PLkLQPN>)rmGLAu4 zm34l>oRQOJ7O&$hdubxz9UX2_$<5R%R)*PbONg%N@J8kfClf0cGI5QLO4@;AAVhX45Lku@l#{l?sKVMlk%}D+QP6Fi^8qQ3TC~H1NHi zUN+bBr*J$+R<)91WhvY+qbt29>g#FMbxj1?`we!OYV&zoD<$6;-I<=V$V0sT@;JBJ zB0G;hj<6F+2+ptx6rViQJELUKV$`{QmeKk~8g$mA?YVk8$Qf^5diJG*-Pdph6;sk# zuhYeadKe)k<wFi?gaLwv zj=?Le8b0>ozl#@(Ra^lLz8lnbY|x9C3P|Ol!LDKwccWiGZSog7#a3W^e%$d{4E8|rmnmiy{mpX>dK zsuFK0L1RaL^RSZhvRnTD$++xJ+qbB?6rN{h`#2s=N#vJj>!g5t>B zBZqb@Zuyi2^0~a3FeEKb%+dT-%rckTeieFK8xi&w`&$on64u~HW+w)Q z9yK4@Lab;m7j&^$Jt@VM95a~@UeT>x5KNo9Xb0mJRy_uS2koyu?xW2}FImx61d@yY zsU{boxD>6lr+xl7-e}>OmfjoWh4aB>la@1`Lg$E{flZz*Hm2|kURQ3bYi*;g%!*tXkwr3G<-gggtkD71Y^WXV{W= zW^Njs6o@d^aF%+ms4m{DGoGJyCrWjV#wK0jxtqngaCO_Iz3JklM zskoQ(8Oga$$$iz+Eg<@se<;wRp<{Jyk~D<@wwH8+g3SueKJx>g$XiXfZ~AULXoU(K zY}GBdLnJ()A3sGF@<@X?6yidW?QSze4L+M)hv&~gYiuf>^|{BN3SohrDyh>fzXT7B z&!Z43u*RR>7j3GMgY|Vomy3Z<-npfcx6@h59aqeg)`6Yu%JWqEq&&!B(+>S*E;T|~ zEO!Y9BlW^ia!9!YK1aqmam~;8MN$6s(JqW#*Iy9f_aC#f%Ch)uJZRBW@4*shkF(~D zoiPWcA{w6TDtc>=BC#d13!#4rO!mfPd*|QTd>Nx755Cmyae)tz7%pKX#nkyG+V*0m zbacvRy3V=A|MgQ0U$z%n7!>>I`{*}7-o8F+dbcR8P8dWughevHxu4b``CVl#ysilF za-B#;Yf5jd=w9Dc%odgqaj)J5i422qPaFI3Mki=VA36IQZ5bTxx{#LXygG2US7m%1 z6LnlHpTn-J-RMJM=S_qgjUuzfh{QbwoV%#!89b5A6QoYBKXO1(4=&2pZm*@-Q(D4b z?fuPUe+5~2oZRlifEN1O_ZTMHlH;<@to-J|5MsMsH)ylZUpfC?-BLOgHsv&ArQf?n zx%F5Te|kFN)=XxsE`eCm@MfU^n;>L(W`(dXV&<+IB>x%2e%LV8vd6WcnA8dxeUT*N z3X)6~X!U0(XOTSR9JyFlHsH+Cu|BJ$s!~Vz*l9B2r>_?Yq_t2>h3UCxe`8%RURBTA z6#pcVTjl9JTW8>oCNd=`u79WS+=7tObEd6O;}-4y>y}}I1G*ALOYjhN0G@bV zub7Qe39|wk=bhGmQ!=Z%gddR<9?wm8C>N%Ckn{{!KwEy$_Ny^cfWMnCuiloxxJB|v zOhlzk8?4}#{@u5DK~@BGwh$shSJiCdXo>=}aV^G{Pq%h+Q1}CNWrVuF&d-^3X-6=fhAb^=c@YJ{a%XJ)|7qMmKA2>2)}a@T>BBx zwJ-M4_0){Y%vaW;5MS#yV@vw4_-#Y_6}js$?wvHv|2+&XOJrGnnA2VLoIJV{H8Z2c z$|O_`Hj`NSTb%@+9CBP4v&BAqHN)qMXeHaq!f$NE?l{V6G^yy6H_rzKecF9{Vhy>b z>a8(!1)8Qgc~a9=mJ7(Q1a}Q+O-({P(qQ8E7YdyBut5OkHz&$IMG^Gdar0V>S!bhm zEACzHaBqvqG_ScIMLyNBOjeDTj(ZjUJ+_N+Ce4e?9_N{DAirJf8p5kRUonwpX9#vmP`Uw!ZgCP1;Z0& zdclNEqZZ9L4@bYTeE)yOMt>-McRqWB!{|^CGr=?dODtU3mEO}wPUDvxAHYm(g&eeE zE#DEz<^qiAl%igQ%dnPT4W&OM@nhi@{waoSxcmFkuvc;+n^XV;oaey23P2o~}UqL6F zHODA3)A*}8mZUN6snD7D3ZavDcg}bSzu)6|2IZ(<(qW;ah3=x@#m}R6gI+#Msd6KN zuZ#<*mRUma9Gt*^(Au*&XubTvrk?3`ZdZ9~U6a2yzc$))U&wDWK85kBZiCkGH&M)) z-y>3(zFX&WE`j_r9M4Lzme0ftb!1nZk7K@D@3UyyWiPZw?n-mBNQwb3x6YoxCwF7z zGB0E1&+0gM&O;XCb1jhtQi?o@F`!EXOxk;87^0q zSxLnfX56#O5cRvAenR;9`p-MCglP4#Q0KQU3LoLyfWZ*%g_($Aj!b(kO3DMO)=H-J zuC<}~g*y+&_tQ*)o5&k_aAR3CutI|FEW0dAXH{vh zbNP>h42z{L{CKZiQn`_U=c&}84rQ%(sE5y+gSw5-1mL?1sXyDjDOyT`NFvkb^sEHY zTbEf<0ew2q1Vs-r@J#W~q#$(hEPq1p2qTs&5a%V$Y_t9#v1%j-igzvbQ(QWj!Ip+x6ih*;&hFoex-7MZ$ypbi1lxgCGVr z_-UK+TYi1SpoA9*E~K-5Z885|PLb5eJoP(`Y@7a&k^1&+^6ux*J*G7CR zrr{_hZ(X3bi&E%LS{pw%dnN8y%r@t0C4$a}7rELTi?!$GJLCR0$Sa64|v{)4&ox)Q*>R&zyR zsQLv(9vPQf&|P7;CYzS#s~Xh|M!;ecv-sa+Q6%cAE2<61dcTkD@Yf8Oa-J|kAich{ z-XeEdLud)3>x;tnTfPNV2r~i*H_l)-20j*P#S50i{#y>b(mlConK)4K3E*j=!e~t~7inN8?d3x_t|`Ms6N_o`&*@8Y zTxeW%5dQz7*}-n;4#tDw?w!YA`CvWKr1tgAP~k$0_tDI%(7u18mymK4G|yoyZBfW0 zW@6u$J_@nOQ!>)XY9nd?o5f4jqWZ?xD;I#YopnV9wrzxAXJ2&!a#`WS+l<0DSf+_2 zkPBXAemoAW>NvS0DVxpugMcJhhv{h%ig`cVL&<&25bheH#eFFs^Fjt900_xjG)C4< z9o==754?`Z9ZCaTq!JH`mvhqPdtV9m;sSpT@=iVn(UAN>qoIZ2>bfnutXP!}Tgb1} zMV6Y`fRFfLr%0ZJS2uKz#}GCfALcPGf2T*ZC$WwvT~SoZA#7$$Y_9XDnscPT4hc#P zzmUMGi-D?6*5Yy7st<3>Sqp6b^T@sH#^Y2t*JtzpVSS~*HRFIi_8%rotDF=_$5L3( zL#**x{9$HTn@7&vn$#=~B{V^Sa*p8mf=RfHVi_uQLMY3+X0f4c{@l4A{`Kb>M85Ym zNq9}81ZwygM5+IU6NBmn1*g(4OmIm>TYF>FFuimD=p^k`)oJB|A5CZCFXEozhuIO^ zO18lEv=~Vj`qp1o7akq&#jX&FC*3cI(vwxoPi@S>r`7g>QV+Fe*>y=_`U?UQ)Bb*s5&OtC;Y^B^9)2fK%CUZW74ql-Vvmr zc#U})sY`Lj3TDvi*u8!ue;m=}P-sq~%BNG3vEDb$?1UiQGkS?>amvF8;TW~&%+6GV z9bFJQB=vtIuQxMB5yLW13E_zbhcGRQseKl{j|+Md%AD1wTQat|dEju+oAJ(hhE>9C z7mVBlB{vY1>wBAm zljMj7PAZqs1?rv_&PKdZ zagk6JC1FqJ%c?gDL%YGDpVN-N54kKv8+kNw zG>G`QFvMo)#ZYl9`!lJGi@tbE`7FxYlnB9RZw`1GD5%y}xsST+zUA0nbS&1DyE9d*bXKTZE{5byttUCQu>I`Q z)YYi{mLBqOxCGW%7u#Nm`sQAff;jonnDnRPT#|0IM8JJZB}ed$#TANZg#`%T@pXnF z7Co$~9@(p_=%3O2@ay-rB_%W;BTHxP7F{#|)CDe+PnA{y5%2H0bdFcmQm9C(SHXY2 z7?3dAhhCV~la$MMP57wOcZi`M>^lss{`1e+x3_DmIMw2x8m>r}oQiK!nTVr_xeO)! z2IQhOCFwW=pGux_>xUYB5JH zzS})fI|@J4Q3ba+08D#L!YvlQJk>mV2-Jz%W)f>mKvx^8O215-fmdBfjQSgELi1(Z zC1dxY3h)dE+t2Xg`6vD7WjbAm2kWIq_;7H*2#Bvk7cRu~GnOy4Y52aUR!6UsCnpCB zJy)nV)sN$@(@kc_6ydzKCEB~Ap%DIh-i+l3`p&kZzqJ1$!#zQ-1n1m4z(zM;kb5W^ zJ@3!mG;QpvXcIXZ0&f}YIatZPIXqrW5IsghI{PQd9GpU`DAxzCg;&*|*#A$fc$x6+ znFiTUIB#57?9++PiR%@7)yD3FN|5wYgIL2)tFV{3JVq4XQ_~bi;n(EiRZWOyFmtFZVU7>A6`KZwxW^b!SZ_&QP z`FPE%K6ab#WS(W}z2>tIG*s`ev(zF|p#J3JN7U3(;DlvRN)juLsPTO+G)ENwy~;1D ze&TBzU_VmWiR6$l<5w^QG$BLSvch#V#&$Md9a~IL_2`>?nC3n}{bcF%H;M==(@Uo5 z;ruTqondymO<}seUOX?pvurb>XVU7M5oFki&RNf>F2#s-fMMs_qcS2;i{qxBR6SE= zJft;pdu{I6l|iGtQpj9TZaS&2pVB$7q#|AB=hjFw#+(w@*p{upbIY9 zggt~iOl8C-mXlfcwbytA`a=8jl+lcI8K~&?W+hl6_vKddK4)6tR@DP4sOTX0rk9Ld zqk1hmWlnP!Rz_tn9OXMY{YG_B^~AbTg0#Kb0rf!VhxLKqmlx98TF5uF-c6|$DI6NH zZ&I6>31s&)N6&G$va1bvj$<%J*ugyYyutiOj1f8Ajz98hN4ue5$KbIa;y&#(6|A8-+lBATPNPZU zhS=-w9j*>y76p4V>icL=HkCiwp6I)s1Q?^QTvas|rqY+YX#?VSTinxZhsI?C^}YWa zX=TwdCgW|Ok?OGG0xxGyD(>r!O()#Ob@ z?7%a}GQG<2Vq_Ds*+VWWOjx-$1o{79$mcdc=$Zwi^{$_hu5ttwDYPz7#Cd$|b}$;Y zO$blcE>B1fAPgTKm$p=U9TS90y#%-g@Lo6Mr^(7J9VN%%r}Li^Fb3sb?hiG2-OenN z!UzyurO-98@ydmF#MnJ?zn(}FO+$nTuNMC^i|DBU-l6lwvo;U-PBW2meUT_Xe!2B) z!fs=tv1{kL$dtmuEHZp|@ro1ONX``nr%{>jjCo<#rSeK%P%2x`5(!EGxJu;+1L_)nE;XywCO1+pEtNUl)^r`9^2VFnrz$PZmyn1iU~zq6$)!IT3vpja zPKlD!xBE%Hsg&@EHxxc@p8Tim*6n%cmQWqaqBlJbp(l-_&3kXKTK&n|I-q#&mMYLj|856iZ(y;pO}mZAjKpm&gzsrM>A|13p*%XL3!ip&fKXlH~&j zEhl@InM^ml=>sO*^?neFVJhx7`%!qOJDMG5AUN0M^0WT@17^ba%DG7tCh*wl(`Kk; zMT@qy(tqElS@KO>tr)I5%@Il+un|1zkowJ zzjQ=Vd-F|L1Gb0k}H9&E+*w#96Q&d98zpGBG&YuPq%g{0rs%C99oMf^%umDrzgXKt6~;ngTZ9>U@{Z!Rr=1GDr|w~!V`#$$SY<(KeVhe+5Z z1lXSTRZ+|6K(^$)vz+B+cnvm=TDHa$ksaRNy7K*2g4lsIrT$nfIxVD#mnVOHp9_yW z0W;8{yqXbLr0{bkM>((w8qC%S-~9W!FVo*gb(PMJ>m>Z0l$Z9s8s3sq3-7oft@vA3 znC;S}dlqU05Nu?-?xz`zzBNS(j!Edju2GABmr<_=l;2`_GgbY|hEvIOdnU{1%B5v1|Kzid_nAnGR6H~3Z^|MLFo-P!egU}(I%P2hjhFHXRr^}Be zD^YswLfvhi!-8!uhXvFg3FqpA6W-0_ zdltDf`6#JXNR~qXdMZF4lRMt2v61NPp$1Q>T^$yK5o)?;`L+4!xfn=wZqR2Bs+Woj z8sivDknl1ItbDi~ER8*IT2L6#dx$mx8NZ2B`ss;sWM8UZQ0OHrL6zOdI}R|*~{FBgk# z7LrI80rhthBkMzFl(*#~!%hSc!^u~A#%A?9wH-d=8QeB5htLeNk9MvqXbr^J{wSzL zEUEM_L2Ad-<54We%8Y_5Y!uY`fn>}fa@$G-?d&!}wI9#1&AlF&1A5^D)&J*MLAWs; zzh#L&#PU+%>?&$2vGZ%gp`TWEPq!kV?K*qu~6P%%|nblNTw8FHsb{?&*-9ojbb zfW?3it=r?NEY1@9v8GKL?1rlX7qw+=OWWw+;l=&%A~o|y-vkWOfIX@JpH8;3Oi_fl z4Ng1Hcdyn~rHF~dIv=WrZ%vtZo1=^@RsNt&1cN`=d4OVYpedxerhsr>F&5@F@%q+a z6pcwHw+ldf(FNeTHViti37eH9-;%P}%}+fg z$%)JSV7v~2&&6*7a#sv%>DS8CAs5jUO}XhgjgfpLOP*}FZ&o@$iIE0ItF|30_D)ZmzN)?+@&HiTFkGD zZ@D~kBbu7HQYSK>g5szHzVAc`HE{@HCG`ec%DR+FX}~(i519o;qLo7AFC#jf=P7dY zhyN0g=>drS?kazZU5%moz7K@;#~JW$TKlz3V{O15<=Gn+9aD|r%y0Jw%L9u_l$eYb zb5tJg9z)qm+qGmkPgt-fUFa2o!%PhtX7>dh&27))&SVZXXzx&f^?$4T<>imSFhpan zoh_6wYR0NVeX9We8EfUXQJC8dtsrbm9$UN~^($+MH^WikSK{roH`C2HQ7xC1xI;;W zXk>Og15jh83n>G)&va?55?Da|#h;ag!SZjYEZK8VklxSE7QyC7*;tBg;wIi-3)s@~ z+%#DZ958gCN4xV$2rLe+8eS3l-nx0A&u$V1S;eKRKDtQQp+$S8@?m^U9PL%v?2U#U zMPX^_32)7w!Ib;PJlBV&YS}z|~KD0f~u^N>6SFr84K9Qe;iQZ1@i)UL({^%yILLA*)QSEl<8X6BoeQ--1Sy2lb zM|J8jY&oAJ*|z<@7G8C&j^2xba#hPSiHH1Z4j=fkuwzenzC7R8RBUtJ_tbH=2Whu> zUS$Zq@|?VSeYH?+Cgoye=PvEGLgC zRH};kJc`)}1vkyk#5Usjk{wIiZpJ!&rJOJ8ZS<7acSe`^HyDr?`4h{>)avs>mj-(J z$<0+LW5cu&Cd}KuRixuIz3M3F{*+8&vc}fJJd|f9Gp+pjR*9UJ4`=&Avacvf!cesQh_aedc;|fkD~+0>t#cfw$@{)3^iB zAQi686twkYL!)ViN}yQdGHBr<3k_L-H4a0{U-)ptYK;82LHNKzUDd-|8yRq=v3gvJKXuy5!Ur5^n@p^V2} z4vzhr5^6bM?89?bTeI~#2e8MNZJDL_<5*^HlkGsiy6c8Pf_tjF$Z)JSoMrJb0P>I58jL)wDMQ5Zb7MFS(aC=ea|^z`#wW=!Wv-ghr<-7K4drw-@~I zR7+@Oj2kax>g_~6+-P_K&WMKbvsI!|Woh>i``LKH4kHsw!=|hE|DwnDxIfAu>E5fI zoVP>doS3LNvd6+K<^ZIcixd)D_gp^3YVW`>hc)qBP4>&Xlg$usdymy;yY?9{pFMq?Uu10K{34aivnbcD?Jt_;s~_`DDI^;kMM;mI5M^h28UUfkn`x%AZUyIGuL`2gO#AdD8u=hurzREr`vKO{V{jpvbH#P&K$ z2~=*K8-)37$8KrOKt9h!SV)z07&?315?K|uHpFitz1)AKX7rHA@PD0IB+ncLIyUrG zb&MkKv_&c%0Hkg^eY2&6+pk5s6;v(mqhMK1pOMPYmztdcB?!{pK)>lFF`M3s8Q)yz zs$iRpF5-Hg((tP3kqY?XSRsERK27>eynbE!EG7+wEj1TU_sh6I_6w1flIuZR4)Wx% zp(t^LuXa_HOrH5y9Z)2)esYd2M)?1@K~csLZKFx#D@NSA;p#PKIO(T_;?<1nWP|R! z$}@dBukzN?MuFKKOyuiDk8>Ab=tz8teG7*Xh$znVi4q@jzFPb_t%F24kW=&6(rBud zZ6Z#23ztJf+t&~hMOh&B z2lX~5<>#_SnJ5^`H4f|XzC^8d%_i(w+@0wUX_T(G&ur{=EghtRW<55K;<#I<2b-*!*+3=Da6|tZ zpF>`YWNoPrI`7h{Moyr1LwCA`j8-(T)jNZb#>TCF8dOeVP#~f%BOc6ReR)mN^Z(cd ziPPxG?iWUf+(uGUc7uw(xQ}3E0^`u&8grku#H%;;yliEovtdrXje)v4W|4WbbEi(| zCMG0=;fYM~L7}_zv3@vMkP6?W9+HqHBY(?Cx*ByjmzrdKRnkvwN-Wd2so1Xsu8PE7 z7tZxWO@gd(?NXt{+qUAXJgfTAb=XHLZ|f9>lM)57!iRPW?9&GcmWdby)#$&oof>v5 z@K;Qkou@tg+v%nRue~aIsgT?I&chQHwHGh4lT6c?fv-R5w7Axy=!#Iu!_4UMY>So6OX!`f{!}7$&kXUQySfO zv-&YOV(H_fAq^WXJMA~K#5$2PZQ|?;gF9~!ktW||9NIL%-UZm57;~3ewU+~Ixsi?& zdkK4qRJTH7qIaIQ6{3$QH(0a+`zwgIDA*-{R^-kNWc-}m77Ns457-U&BUg1 zx%^Gs4$=%|-;RoT5=fi{ogTBNR|Y0Lq3Ka_J;f5ZA{nh;H zT{~m~SrppAxy3Sq+na8>=~++;q1uGyR|j*+i(nBpgi_0)^Z?@&>6dpZZAt!C*RRj~ zJon^AWAz+amo5S`@mP*A6*0N7_qafxlgnb3-d}RkH)BU*q@ARtExzKc=#?G5w*^k* zd}kJm?BJZiyy8}432>z2g`ruTeSA5h&f#&0uB7|iy1)}C>as^$El#Wa>S-1qVh9g~Yrv7dD z?{X=PM_jqqW-L$4Uzr6-m}U^6>d4`n7aD#54#2NOT=T$vJ4mnP?qDdqxf%8 zm;HB3)19j`|5jgg3QE>5hzxt>_%WQfmoJ}y@+O9Sfz*@XfE~rlUoWz_{1K1qw$u8i zt{M2huaknVsGT!~KeWvCcU z-E9XmKX62!5|{pWvgj^2?fW`mH?dzXgb;AS_OxM*EUmi&8sgz3=rNkv$1%)&PT%wt zDqaWu4Bei=T4N_l+7If;Y&=Y^9U~g5b-#oIyxDlN3E>Vv(A@-h4u72yt z+Gzu8BVru{snqnWB6w|gixdm2c|3CS( z3!>%<%`urxsWyp#2b)UOk>ZxLOOOj8wmIpyRZx;@v&^*k3;qC)cnRf7-SoJa zVhT{VahxluIO`c+Z$|EyelKT?GY4#2`J@fZumbKTrNq4Yn=H--z{6iyBB-uQ0 z+0Iz5PEB%S_?li1*}IXufvh`zsl>)VS{^Ai2SsY;N9nL3c zp^_MjixR1r3EzTkG~{}@qz8wM72f%MZ^M#-KANGqX9&oOr0K$~$H&7&PMN&tL7?=E z7~Lf3mv=VhGY2m*8(c8LxK!tN)C8-J7l;seqZQCBYC;RDKmNW8OZ{K|1(_~K83t;B zxBpG1<-w?8Z-ZiWSn_U$r1iIJ_X}~g(k+AV+9TySxof39YM&pR(39w_4w2*&)AhX4 zN==2XvJxNt@a!995Hk z^h?}4{N*(WiO1xP=fd+Q<6G%>ZqLRTlK8a_+#^a=MnRZIUs*%Cx`BS40GhzdL1D;EoN0oD#ak-Xftb~r#+!QgOEhU%SCtzrZ=BTWT_ zs=HcsZ}x*V`lj%jlQzXbU~QJWIa0=y*jSab)PbnlpqG6krs#k-88-2eNTDNkuZtbY z^Y?|iSxbJ+n4Yps5@W(F`nc0oo2Kbt1K_=As`E61pPj?+B}Z!V@3a^no2pgSO@6`_ zqRGNlwe|ZAJUSI+hJvG{u1F!VNy*z$nOlZD8;PJ;276l)NUrjDYNLV*URPw6!RT+e zvL@rj@%cx16XE>o7sG#h!<(0fM((8xT6G4+tsK-D2u;i?H}-4A$F+8uBajR)bKH)g zf{8*o_Q=-OUS6#+UTE(4UILdpbL!R&z(aBATNhp2{4xs+7p4(J(*GN@k1JkCqxh4_ zo=MM@e|5ZaO78EbwE$vpU>kSN4Sziw`z3yP1iTS1=3Ih5F--W9f123!8=|g7Z)* zJ^Jw?*p$RHUQKEtzn7hJJvqs;G%Rm~Fd8jK{-YaTFf?-S>DV|4Psj4h<)Srh>aW>4 z6kd|_udU%(oP7UURe$^qFB=}Yfic5pzQQ+`d!_*;C8}2Ha`E)B0Y`oOcAdej(JKH* zTiEZt*4Hhom)Dyet|9Y&rA6lZ3h@IL++&_jtvt$Vnuq=(u7D-V{GD~J=t&QXQJrpc z6@hDWgAAFNf7@g#V`%Su&B?2CU)U}x`Lv7MnqF%yzDk?;8(xv1gcuC=X@+ zz6!58gkW@c2dVXYi$1HfTr(+wTRlCS8dC^eDoqwFRGM?Y)hd0G?qjm{_S3TUCNUSY(LLZcQMve4s9>A3 zP;pI1<#}4dl^q9}0Na({U=dFyz&_xJQ&~m)+ON~*g%m8SOC4OssFr?}Vt!0BQa`o1 zxc_f@p^C&0*+QAs*hm*%9`49vwG3&Sk5p$)k;Gb?)Vtc=@ImWjA>=Fz;$cVvK zBRd7ZMtH=s^(}la?Ft5Mpwv`qWJ1C2FJ5&p9C>VImN*rmE_F@LZ>jJ%a4c-w5pS*r!>-gMitM!4_3ctcDTxZgzAEa{O5KA0bkBhm0?=TK(_XDAli zj7_O%yl5^dWFhGks9Ix_>XOm_PW_02=ECo)4kZsz}d()esH|QU2 zn&1KSj}?TlLbNLCC2WzeQU5RjHvZ9wAHDX%oOOt2m0r_JRA?ias7CZ+BR8Q`1)vsR zTveO*o|{^t9IbM3^5uDm;8qh1nO~+8?9Aqua8py$qvH$T(6OqHtq%A)MZ(KkAR4T< zGe5eF8dtg2bS4Gs4YgL2aNl(m44;J+md#$a3z0aaFel$s@0=}pt)Epk)y#*f8Zkc{ zE-$`S@p^tDWb>ktq!kZ8^WTxjEfpS^@8$G?!Jr^W-G>g+Xp0$osM;ZeIJ@v)(n3A@ z1gqU$WCx32k?F}l+-JqhGg|65XZmL?to(7!gaS#+6Tn03zP|iSo z&?l79ri9s)G4b`d97!>mN*f9-{HZ__zPqSw+vvr)BAlmm9u%BJV0tR|m3@eiZZ7UH zQKJkk=J)5WOI)sd@P)XEh(CwX@*qq~Q7JB&2TvJfD|6K_nVtO%y}}SYNrm?B^1WTR z_SZ;rt|-W=V03{Yy5OxoYt@T*AR-u;9k(x`sT9q=sUHY z%NmFGvEhwsta-Vvc{=32t|Y_vlDkI?(3G#*>^IE*C+JnoA#bdBYZL>^v{7va9hd(^ zPt7XHgmK-V^5p4nkPi*-gocqMsDGHsNfj)j2U9p6pT$jmfYd+h{$HwI-P&+KB_aV94eCPzRoR!nmSeM1@Vp$VgAi2Q>)K&$8<}eGVH>&Szw!RB$Ya zI%=ZM?nqLxKH5c~GU~vQF&j^hY1WmvjA!K5E1E*`EiCHkzRlK8x*`YZc}!~qsSU|r zn>6Pww!?mKB=0cgldv%SlbmLF`*0IhhLsj}pYa#v8zX&2v4j>Bkh_St4$9P9CaqkD zb7FLM_%JF$hV0l=+o3V8?^6VuMWfRjQlwboUMFm!F4)pej)tn0*d6QJ*$VF`XM3~G zL;c&+su;5cN8NLieTkX{QCsC#iRW5hv~v@Dnc&RPu_*Ngd@>}$FKl7=l(MazbJHcO z4?V9&Oz203_XAlBqlV42xQ*BLKkePi7KSaI3gx$&3we`i7Dkf)&-zwnLXizW*kOBe zZRq*9HJ|%;F!v?_M?eWyE-(+Mplv-J@U2IX*k;EcK_+923BPvC!9(reJZ51D!YWmu zmuN{LETGsvzWRO~MmNF^tc;xU2lsUeHo>$9pM z%ngF-(M}j+a%~dknc(NhT5(OqvW52u5MlJ-6YnnbCJna1+`kD+>iT?xOs8S7pyNfj zzb0MwXX$<1oBy;*#P(TEd&BlB=le-Xt?Kg7Vdq7{5t*q;XF}sW@Lb`eenc?($+%78 zLKBIcX){uoXc5QcRk7b|73emDUXJ{PI=rj&BTUU~=B98XzYt(GQ4k*whB}Fq*dAC< zM*CA2rqtg>%~ihlt4bz8$=7SUM6NiUh(2&P%G!QDHd>71QiDbu&Jvx=Jq{=Rm;D&j{or;) zrxOYDx)WP;GFjebL9@CdCyA<wFmjFSx&O#ldmC>uG~#;cW?~_hbsjR> zh5>xY>y{&y$`B62&GlV(B<+gs1y&@BANTw^c{!_l6rEPb(r_z<{)yo zq1;haYbQB=ayQ28OUN6kT*G`vkf}27+7NQspshP#=&8d@N3}zy>T@y@BXNH23@%N7 zfeN#ve*)WL09#-)rh$)IHIdUqp&|)VHDy09-fcKU2 z=(X5T9<;oWCIXyM6Y0&@r6C9o+d@ua548s9uWx9ZqLk1y3{!S8G|;aGwJ!_H1(Zz5 zK91^??qBIj?>REyq9ku6iO0PO?VdS{Cf&gyL^f$pN3_gAxIxb#3z{6*XKli5Z_V=w zL|MWdnR%Mf(k7#I^--9ZbfXf6-+85cB;vTpcJ*^06K#dwDn_k^p}{c2^Xt;XQLwkO zDZ~>Rdc1=fF0E4~*Ks$L=P!z|>st5o8c=9T@*-E&mgBT4l zj-(}d`1Y)1+oHbNxrhs(VfvTB=UW+PbRBL=ABQ-oXWg81F%)8St`|*>k~*z-W?Sa; z>jE{#=XbVEC;8v9krp6p6=+t3ObL@?Vk5>fmK_WO@+6AHryw+(p%^^hl%;ZuI1F;o!73kpPn9e!`7H&)wVgH24a|SUG{iwh&1`mpUx~iUGLhZ zvo0%txbgnbs@03^jlLPw1)w$GTNjtg)T`OWxnWvSqeFjuht)uSBCWj*8NlzT$T#s* zNvrb2Qpo~ma7L;=QoNl{H{#C2t?bqf-8gz>-urq&9Qt+1|M2r*@|Qv1XP(vnkC|xG zEvHu%-(cN$B&ZUp- z6#CgLJ0e&IdtFeK%{0Qexq z&vK=F-KK3vCOtEV$ex@nqcj;v{=W!M)bQFbn~1O|@^YNJ2h5#52SDrM-J*_KWl)B6 zkZirol-kCWY8lF;Rs|{gKn#O))pHyyZ#;3EZfLlRo`B<~nLr{u<9mB{=jK)Nn&d46 z!84|Yh{1iu=kQPT6@bz=Z3nggs1g;e`gY1${=CFzF`H18+*z^`l7M_NMMD;k^uO#+ z#73A`9xbTxmyPzxmQ{o6RRLWQzj~yBr(sL#Sftl>+B+w@}N} zjC+>OON;IN!<~@;anZQrpvW(;Svb!QSHDcjp7J!qa)jq#GCUg*MJD75gk$RBXtV5! z4_i-21Y1mh%qwzP`y+0ccg_*jsKrz#=)z_59Uq+^)|jUa97|I(4&MK#Spe97LY#Nm zA+_`Le%<*O(t5RPqv77GJv~{4;e33#%CtoGyKg^R zzzYWgb(EnO3V!|5o5Xp;`G&b9!z=NqkP2i7o@WXme&1IC3gQRh>vyaP8EfRXq zi|L$ZfvWs>@ZKdI(?^jFHJH`-iEcUo={}AJ7nxq6&WCN}6RArR^jQ^4NS|yjVgrJu z;ITWi#lK0_9hyLi*6|^oq%wpXlT$k7Tq~C3yx`NiP2M@>NUN)$UD!yTSteRzDu~*G z1;{RaHk%7C6mL3W4r@2c4}O&tnD#7=3=d@CJC*GGPvVIyECl#_Yl5m-x)&xu+Dp zCsPi6qJ9?>O-~}VbR@FRjnwQ`SIup;t@L-OdR3`O0IP!72=}<^ES;qd0ieN57!_qv zq8N;ju3OitUwnI9n+M;$S%xc~-R(>1 zz2<1jFAz#@ZwapX`Y-#tgQMHxuGqt{x}>aA`ic-B@fh{YQYV!2M-kF;B8PHq(;4cHv6K^pokD4hc}stm@ix@`d4$$MX%QLeJ_s zNWa8#^8`e0Q_HyR+nP6c=No97**G!Xzc1a%uEL}!+zKj?)9|SOW z^d1@gR)9w@x%D=ok!K&1X5N@Xa6V0%A}kT3;XiMkf?WxtK5@z%*`{9tJ8WRf%BPNE z++lI;CD;mRXDrcj!X_yv7?MdH2?iyf$!_8espe%!aOt02@R4&tlv&G9HX!)Nokb^6 zq}}BxxyqzMz@%0qNQE^cO=M3d*)=zfzY&~5E9a`4NiNx6Ja9S(@*o$$gdo-1*V4H4 z3g#UUE&|(g&TQ9O2Dpm4TD76tkIR=)Y#G-GiAl?Xj#S{HQkN81l9uIh@GD(n8`0g+ z19qKqM2T+<^6?oRKQvN#IOFY|-o#wVf7CUc-c{JLe7x3d4RP=sGkgk!)|vWB|H8UD zjPF1g-)T0iW(bwCNme-S9m#%rHKs1jD&dd*4Z)69=649kUNwnE2E-(V8r!{NfjcA% zc{RkbvxznY8k~6gA#QrjjGS!;O8}$)omC^DA%{oWH{tlH`|X%X_VxsE3VI0=(&^{A zd~;Z!vp&(YPiYg>q2{+EUvsq%qvzkG-vG;(0|M55_)duJ#~^q;Ii9svHU3-m)J{Rq zzGu9>MZB{m&MX8r>nq`g9kvyTl{jg*1Y(&iq>%nH<=pAMSV4Tqdf~@E06;E(?YDd z>{4k7dFk+?dTFF$&%KC5%FrIKL*r>uvDEB&;XDT(!gj>v%JAd!b7Ke3Pn!hEkk!ov>J zBYH?W*Gb$7J`lb!Z?B&mtsdsbslAvX35;KP<&!}T>63_i%KgMUWkX?M7_&^w7=KrI z!^asJFO&gMMUt%h`@g?_`*eFzC<3W5CgAUiqp0obLsmlf=ZY^ z+cvBk-_nQqQJ-}WajUYsn(Rf`Q_0kAhX3(8^!b`9l;+xE=>6PS4Yi97jJNy2=ZJ7* z{^Libmd^img|3dTZw%#_yuk2+bRdZ~dO9g99Co85(6j)NPgeq2d?n|+34#HIwJYR8 zoz_RF2ueup+>m=S*G_|mQ?PleKON8A!xpOyZTd4nEBBK3Lmm6BB2FWy(@NR1&Q$7J zF9W~wz7`PI@v0_BPGxdy!mIQ)_VAQM!!AM70NJ=X(hjywVmYK%}}u$-=R_J=x=+pIZdqU zSX=}B6S-|J+MBBQtPY%w|6(6)JDqFxj{4}3fzXL0;U+)Yr)Iv_7APXMG!Z%I^8c2& zAMO66cWQ7Yzi!p7G{qrY`L|?Ng2Z`YEgAAU>@|&bw3PWV;|VQ!j>{(f?IC8txlD6s z{rXmfXLgg-TM)8T6SpZRF{QZfMEUocx+rl1@d9Uv*P8QI6xQ8EKR@1@r7MHi3)%)2 z8Qk0Esh^MY-bifn=%ral;1sD1K%j}qt8)qj*`7$luJ8NBX|S?d%D5R?bdK2g`%FMs z#HiZwA?s`8-J_wx5TNeh^x6(b`*WE#mN`y1t0lcVXgEm7ve{XdvuwU7*LzjAX~bPb zv`I_SF=Mmix*2TJS@wowC%~OzXdi%2qxX zxvke9Iktb;2jfON8pVl2Ji)5!!nEF=>^s#dE!}s!k0(5=(Jz-~*fZw8RntacIJV_+ zr?Lo|Izk*BNqFaO{61>&rHE!+8{G@+4awcSn)Au`{*Z{jLPdASW-6%E9-mQsFj+2KQHgJ^CbVkC~ z6I1Ub5YFWN>#SwF+FI8;V2Ii8NU67$=jXjm&ifU*$LofN8Z%{!z6oFT1Vx@Lu?rn7 zi(Xp}mI{Cb_sRT{+YU7a*C6c&PD@d)-4PTuhbaG*3OIj2&VqyNP)DXW(;U&wkvMGh ze{^9kvh6*cq}MTwas>rpWIH2;bQsCV95Q$ zJZWVFYIXlNC%!fWCpZJT0xX2<$&+@mD5W10X@eF%Vv-*ye&yt~JzA2@bRMqZmcTMi z1g8){xdd?He6Cb05XCHuZHWo`INPtoGUEUT$_|H!Zy6tB9yCeG%id55?W#Tm9IzwY zSB}{h(DHm>drHPpftdz8XsY?W5BYH$?M!MGDE~J|vx`gPP@!H749hpuvjfKWX<&TK zBKpwUx_b&x@+|C3G}W_{&iHN)oG>A`t>!xNOWYq@d!p|7-3=_~m32*0C!O+(L3@8e z8uYyqrg$$^%0%X!C_=r>J3TEQVm|2OrO^|@ZxKd{p@kT7a=%*C7qM05XDZE8zHEXA>4Uy0#8 z4I{L3?$~}h2t!E`-|>V|jELSx?RyP|o_V%WFPjQjgGAo9!;R~;CUUuIyk^(du=;F> z#9pT8&tvSYTMh`LnMovvHlNVF0aMpxXT}~6f+`>Q)Sgg}L8~iD zm{==wnLkMc5molAs{-3bQ`iur?8ZP~WGs$*#O7a&g$oD?td2iwsr%>|K8%u9eCKc1 zVUOg9!$wq3HR82Fj>u0P^hqI>k5^aubY{ufvSLDCKQQV*OR%ow6p2{`S#xuWvN%(~ zuR=02TsSA8Srut8$i{-|gB^hbLsg-;ce|%FMa0dle3<2__(3F>YlFHbd&774fKG`Y z;|s}y)6v%&#jN7*?Dwf*Bja2f(Xs2w>GHM(`6xv&SdK!Y8YYLEN~I2!JQ_l*8p3E) z*GLxT=B4m8y>A!;GrZavM+)fSDU)QoQ^h5m_O$(qeNTLAy_XXGOk zLzE2|KF?pjPPZR)qz*HR?Tuigaf*NT95U;o;MEdEd}Q zY?iy=5fNclVT4DkVrKd}xFDIW4bt)J1pU7YkRx`sV8od0|kthza4uCJV0w_I3mMSTw!e8uC~d{##(Z+RF8 zoG{BB(Cur_9&AXgHk$ zJGh4$$sDgb6%0A;ti!Q=#UrP7qq5AXAU4(iyP)U!dgLjWkEA;IoL5Xa0bWBQzx{K> zK=g{x@$)RrRXp$-hEW8o2VQ%e6-0@~N4^GrovU-c0DbmlNyo98NN<6P_{E+$HVb&W z{IcZv{3&TiLSMQa8PDpkhY$5q$Kb;d9^HEGGO~K&W*4Ae*>g`fbxU}(M8BoC_cLA} z++Cq{1u|R0>WpA?k}OqIgoGQ?s=LfOss|GR8{*Uwl}x~sBte|MQ1MuJ4f2+?8B!%r!n_>1@T&|@HaS zl6J-oxiV(1XmA&0*>FQO&g9mK671M!cqrk`08n>LzaM03L<738e_ch zkrWb+thcAdKtX4j#h&m-oHk`<_x^N>E^;Y7YR5dAhX(M)K4D%-`rn8lJN6kJli-jM zbUHlsv!qc=a@0hBUvY%yawohUu}Ea&Gf7?C^Z#Y6b;n@lCL~*W{{z`JnSEDCQF^>) z2BZe_4T{Y_@Q~8T+7Z+?a3``XIaL43z+^)=kteDsqC`pwBR|IRbvK+pE5z+7>r28$ zb6Uxhv8yNXYud=;4R`mLKK?ghB%O&M{j8}{bwbFf@|4Wu&P};uaua#-|62>P5POhN zo`gB1It}{2@0k9ucIlq%K}2I%S%0}(TV(@1xP$s>9>3(H6CQs9(0v?qh)OL|Zph z>N!et9%S}m(P1yTm}6Q2jLm8V$GqCC_T3V?>$ilVvqDW_<<|23m*=0&U+;%zKdF$1 zzG~fV?15$u;gD*E%b>{!a4|el@C4jQB4oN^Y}*Q2G0{%%4l<$Vl8h**x!S1G4@MXI zgV)+FJ14(#6G~);vmv{pm)7jmN2z|G>}ZJ7)Fz@|Hu3z+*HLk?^0{){fa7}}M;mDl z{$%c)$jdUUk7mP;%2&O>-z+$tMn*KTGj(YA`Zh|V_q|1EU?VVxqni}C&JsJCdWS9_C_P|Q+#=mN;=Wn zD?PP_dO+h`lx^3;!&dU~D3`-rBc1L^o3#hTJ;Bm#cajd``YmUPnf zo5Thph6`Yo_r&zi`uU1$f>Bna2h^$C6af#iJov?-WKpB174cEKrFs{7nme*F-X{1% zatV2RXmgNzTTR5IQjyGxlo0ehZ$)KAvwO9g9p%xZ2@M@)EEu-6wIw!(!qnWN2aPPG zVqdL6PIXiB-VM}L7I8PP zBdLFuTF|B^vB!28Bp^rB-$Lc41|uu1mE%j~!)L#Rt%x*k%7-LF@!{Oj#S$otY74*A zALtI!;*`Ha=p>{c=X{sT8VB3Jl-rDL@(2UJQxVtNm}!3^mRYZK1MK`GW8j+jsF~t- zKG7s(qcLz2hXgW?aeCZka@PqiGl)$@`1Vh+>@f|iyOYP%b%LuEvZE5l#V2{KuQhrj zQ)y)Eh0RM1?-9!$@PwBv^*dzw$-$P#mnpf1qO3bF>bJoCoUvgW{6XpRRVUc6nau5O zi(f~q)7ywV5#@NC&+a~jc$q-ocD3woenx&Rr%+xU!hH{oJ0r-c*EC`|c6N^qRNI%) zqQrgi4XDT4^IlbUgbZa;99eob5_LlAy*_HYjQ`*MCaW05k|!#PGri&0Le*!LTs~tV zR+D~{c*v0GAy1ow5=jo%o5dUI@${DN0}>H5O_USc4Ljecc1f(J9VdxAl*B-XYcv;k z$+g{b#CbLkLt{X2ONo<*bL>(j=Fg@$neye<7uk$Yt(JBHi($CXnQrnaFmT<-&^&9B z)ipp>=)AiMg;Wa`f3FWfYn(!8GIu^Xw}z+jEVPx6m3;%bACPP2w1dLV4n?qtz)B^ zm|3FXnMYV~Yw%@xlCa|Lmr)Hbi>wG?c?LSRuliZg_-4DT%a{L!g!$tr98(s9tL6`H zGC%7_OpFU3ci5l(#5(ZoGh}&+ysbT*uhhC=fG4{tkt1dqGfcbli`Pv6^V~I2qw~$% zaLVeh&Y;zze8cY?`IEtMyG&QLXvi&g)lisKNZQ)|UPwpyy4s*0M|oR9*XBdT5k<9A zR|*+T@FrN;kN{9!^THd2cnE%p; zvZo4@d(H6~o#%FT)mWu_83jKzwb!Cmx@37PnKKRD45Bm8+#r#a(hwquc^nr6y6+XgukD&rq$61O|EG9u9 zF|QJ^F)C^u#Tj!P=USMUU#jm}jjX74y$S5SI5VbW<#GtF{hrwPrY@Xx&7BmOS?4+& zp4wqO@^jb_Z(93Eh2Uv*l5d5Myv<(42e=>i+l_8+aUMDy7NkmTwQtyb=yPGG+|1}V zr_Tj<@uXELYFW{#w_g>SEuudzff>_SO;hX}eCM4^XSAeW zjq6@etccJ2`5_i*Ci4Dj!}>5i84JSST?DC(`@v?E%EVjv5<>BFN9eN`So{}UNXoawxT<4 zl(|LDEzwHvjJ!mWb%NMA(_BK+AX0vX%69+%GIa~J{iliUi{dn*Z?C_NX09FDBcT{% zMDYi$_%_AwDw1wVa9pEG7xgo77lI-%7J(8x;(NG0g&x11s=n+b1st=*zDULip+A2z zb^%Hg9uyUcGDf(jFQO3Bd=hsXacsZUGACybd&kKwQul*Rzsc(Ktd(2zALwlz`H_(G zn>EjpHQSFjz?TKurAPtM%VSgOtX}l!-=uT-Sp``H@-}MHp)7ycep7up&`VIbH43&B z8f#Rj=twqQtKVsFY=m~RVU>1Z@-u4aHy2;snq&D(qlD45uqRv#uwL;vWUL}@=oYZ+ zLXd%}nQZBjaNr_O$@4@7)yu^LgmwvZS-zg}dUgSzGR(`*4wH;3%lR8(;h}jedKL8?6j)u;@`SW7)kA zoJ*5tF3d+*ygy8ylGNpo^)_cj8u>d1by6pdiESY_6;;7^pA*ug7^=C&V&jDn+o?1- z;a@n2KimIi#q;2Np^-r#TGP4EdMIuI}mQthLh4-t00f0SS zMBm>GHe}M0ev$#_wBSjBKN!czANwHW_UwCy-2W!LNv27=)XS@b5@z11*^g(qtq$fZ zX~HzLRQr(O57wBinHJ>FjUhnziIQ+*hfO*1%CrA3!rNmO_5$5&n4S9%4nUdN83)G@i@B*{ON=_-2!!xbbbHR@zWtln)w zz4{f1t0ORnoxdi9Q?^EpxF(bwvzt-+61*<+(oCV7N6m6NoL^2gzuLjb8w)u2!vmom zlkNL6dtwQrnuhbcN<9^yFSTxmDt) z#lfdMRI}X5z}Z5qo3sI+_3y&gA`ti{ilZ9cO~Z5`aKrms2C<-DvuPnXx+|oHq+MO2 z-7!cVl~1}{F|#S`IhEP+hwDtwHXeshZ#hMw)_7z0#YzPabw||i+-K4|xM&B9N@1u! zAtIYl?lOq%AN!Z7ivNUBeX<$Gr^?`g@R5w|K51eWd87HlMxsZXbR4qCcA^OLo-(;%dLLfJ-48JSh&piUsXb>b`z}8Bk z1r_1xw~hMhr}?Dq1p1Unhsz^$fmv*LhU1w;S{hNu{VbWV@ugxmmkkNo{t9akCAP>( zy#4Wg-kbbWi$t=l!#&DMbC>EcDzJ3$LR&^zPMr~00qV-GN#+-0qiWRmlr(Pv*(`1z zeWp|+#DS*X7~%~6vBi%xL4m(eZ4E|ZeQvyJM_P&*FB)W!&V)%fMaH<}RTr^(piZP( ztF=S4`F2BcRvmdj5neb2tb5fPu1Ym+nKC-bytMw{$NzfWYt>of!$z)Frv0nIwp>dm&++Sp z{+n|OsM8<9KMyW;i3}Cdo@1>0>~cw0A9l~4Za^K{VV);^R?gokTK*d z4?Qwf!DP}ZEMxyP(Z{9toAHPk?TG~(Qdi;Cwtnsd6=!HQ9UQq|FcLOMx|U-Hj}kh)v?XM~B(Rjjku94WkAp>w@dm{&O`2yUB67fxQi9VBe2 zTK=t7%0)N68LUPqTMMQ-&!q8~>b@a_^=rJBy~K1pR(#}}1ed92`MPPIw+9<%kuup9 z8m{NH$8=RnZmJoXt>G}6Blf^Hye=)xtn@}L6Ks3t6=grZygbMv@r1b9RQgnrsKncR zWX?&n**lW)#L70&#@|K&7QaMaMikJx|E4wQX^^?1R`!SrG3JoOC|r~pu$ZvSH6wW8 z=#DhISs24sO_4w;9Covm*97YvXg^BL+3v_(hjbB19|{)tMLB0720inWiHNBHDN`T9 z9_MI-0O{K;<04c`o}ZPBf@(d7BD6_OeR8iO(+`1yuPPaeFAw9hxn%D%BNQL*BqLlW zFA-V#oF@XZOz~Tg7f7fALGfA9gnzkIsVU#9@dD8P`pF&Jz+ahB-q&b5vXs14KL*pc z6CmG`ZN-pYID1S^Az78!7$F`XIB|A7{u_(hYdA{VBQK5>CaxOv4>>U&DzOIcZS+_Y z+w4TiRY;qq;3K+a2`J*-}*uE5(@n5TuAbbSm8OU?-Mmb;-C ze2>NMmrO8wxZL@6NVJU)q`PdEE|R*wf>?tf25zjI`_7!b=HrJ)&w8c@y!ET}BE;Ca zES(_Ua4HZ~f`;Qpa@_SYWsM&`&B#FXmY$%<#b7peW%?QJ4uqcAQvZ(Qi7yBD)5+~; zp-EJ1MfR488|eZcCFS=<2sGBqLfy1rzGa;l_X}-l-WfMhmET$8L3G3-!65;b+uIn&Vn!fs=V9fo>Sz2o zgN{wQvx#9%IK;LK7^L$(3(W|TsaD5xv}y3CqlalL1R}nzta6hJ2VTIJOr?* zc*qx}>Ra29QxD>Kxk+?Rc6clOv65uZDN^})Y7qUIrp7tKJeMFwQ<*Uvtb zXE)+cf3bX0O?Z-<|GRk|)F<$vTPJ~?Mh^MGkzP|JX;E~mIO$ZUCMGY;jd-$1-rd&+ z#MNdkivPOU7x%d&6-ghU4sRU33BMdV{OX&kz5~#sE;#lCiGiyKr727pc2qU=i5EV3 zY{A)@2l4yry=6K zM~j;1H*SWUT_dDBgUYK75#3xC+!E}G3@FW)x4_5-nP!}x?o5h3q*`zn6Lg?a!ZnX~ zhD@`VJFL6+xNkIO=t{|etA*g6Eh*SSDV!6+S3MjI!*T&Avo4i==sUQvgBkq@xwG5^ z#!8qDefWVp3FzWTr@yEaehA*zML(+6?xT%Ecge-Mu&t9*Xn{NE-+^wj*Z_PIBJD{o zjO}`)ZR2^QB7P}TSjxrR)Ef&p+e6jS!#(i+7swdB*bk z=}h)e;!eTu5rW0n=rm0FLWWaW*un5?8B;PBu$6%tYGBfa3b0*ir)=wUL@XBjE zDq_9Pt__Rj*ZzoksYQ1N(v?5hxai*LACn0;WgT`V9st+pvPd8wL>$p(9~_A@8A0b? z3`xcPt_w>O0>^l)!9TZmCi0U=vNy2pIwSWA*5tmws#)C(bdMG>l<|hxe$NLPsqq?2 z_WOjIX%T71wiU`r*&^tjPA(ExHKS?t<2e~@Q67>sG3@cef+NiQi6DeJ^C60*0WCVS zRd_M%w+#Lz38BYDfC}YMp=RW-=qhq0OrT9^K9D9~{u{_9SIW z8xR<;W?O7M=ZU1#p$RLLggubfdfzy;^C}}ugxM6|JvC=B>$WaJU*dDTEFAgSJDy$pzuuaf+^G+Y3i_WHsR)fCayif`_4cqU>IORMN5{b4C>5Q4GF)ZZhwKcrhBUN`@6`K9r*|Zf z-5@*`hSAf8ra-m3{HzhrP6(~>b|ve(06##$zyBLmcwCOMjhIk$h=LB>U7KPKb=jvR z3^HR1w*Fi_-WURxke1bVR^vrJag!y{)mI;RIT0$k)1kaSma#Gg!^8skB?RI?u35N!~Mi1l%T?RF&*(HT=R! z!_)DtazB0QfFJLR+;+&B3o$KIiaoPe2KuFEqyglVw~a$en6-Br>qSD@&A{3gm2gU~ zv{$wy5Re`r&x7Mt4WIjIKgR)Pm(H2rqCxcQ{Kx{wn((fLtBrqw>ncTuDz*=fZ2jN9 zJXTCrZphy=*Q;*oOFX3<4r2`$&t7lyH5IKZcCR` zDK(Go2W$@I#dxqBw@P`;2VIBGvNGm=wpuiJ_7agZ3T+GVHJf1uZ8#rz%WdY1n2^m= z%EvFd_IpA@9X{q~<i41*Y@%&J5Rp2e|&vrA?b zKmOkc`Th9MA!73jR=Jh+Ht7w=a@=s*3}yabC>P_t@=_Oq1+t`_#KCBvLy7Orb1^4P zq)>f&hh1qe3MU-Sk7JIUo$FzZYACm9*<#%{D#E{$^hsB_UQuTbk6hd#_xD<`l3{fT z&}<0h6yF%M|2HH3gbw6JWT^CWC9QTKYev|K3vr%_Zvhb}#|_5#U0OJlXl~XHK3>BD zUalNQugzms4ZE%Wqsssew9prFL6%_-+J9ENHFIvf^k+>lefDUSyskvdk_XeQ8S%Q8$IDWoZc7%)Ub#z6rDK~+(OBOLQDVu06;#J%UNSYNOqx`m2j@pjGmCEY*T zpc3BSrU@y{YTdDgt~3?-%uuwRExRi*i(MCqvPt3J(z9N*b1K(D40m1+bs52+WDe_8 zO3hvtY8pt4bZ{OMj+N#4`r`{dUuG%qn5R-R2lJzf?qKMm&T%G*vfXf>fnca#4LM+& zD!>z=*UF$hmHVP*o_^3x{LOj|ZpUS^j*&5W+tf{DUsWFwlN^?=g%sCqzAW=37#}*g!JgLD zP|^$?2?kx4J?Ch4GymHHFjb>_T$FDzlR7Eb+h{~$+n_XqN zZGQn%%k@@UOOO_vYFh-;U*YN=iTBp8zAut95p=a=VyXoJItv zhMQ%?hgagruw?lp^obAcnQ>8N7C=@%_r-|3c8B0gARCX|16z#E&A=yG%6wx-ZZuyi zGulWQ;A}j#UTkF7R?&GE^wm$v(ynHf!1HN5uS5tPNKfK`dO{YM8X->UpmTXs9)=)m zyZpOK19Pk{GP?~^5yt-_C(_Tw4LDTEP_008#`>&iCH<1U*$vT&(;Ufu(jnBC(^6P3 zwpEvBSy@;5;f9P#=PPcZ*F@c^i-#o{7r+#l3_K+W?Q#PQdQdJ~q=Xppja7$pey2Zl zTosKXJfwh&tY@M`f2(hXN4@I(%%RvMH1yV$OEAo?w<3FZL(`BV!g}AaVhnwTN{!xS zP`dbHM8Rq5(M!fohaq&{In;7(=d6S+9Tx_izrfzGgd$-(3ao>;H!jB--uEnXXo(P)=vrmv_^}*`7)-TD#Ney6} za7a;hKatet1}1vDX|JyEI`nN$b^V!k3Xedd4-&!FBoeDO-=BachKE=3Eoa4c(lKXyQ&Rmf(BI4Zi@xSBZXxkD8*(DvE{Un-$ zMB~)g)fq*vZrZ}v*+VEzej1-QjcC}ztnVb6h9=YUr z?K}I+xFq*F!d|#1qL!vUG8K-cOGxmmWYk;rkl^=aO|1lqp0|Ua^gk8|3fVR%k}FEI zS0rNIj@0TXe2urx+K$lTE6!krkB2HJa>Xg*VL$zTv82D^NS3QNqU6d>>66VAS5SMZ)a%{8CtrvkW&0QgxVvyZP`9}T-2k0ad!>kL(BawxN zINT5i513(>Az-BPLT1oDFyj1(-u)Jo9=9O~k40p-uZ_61z`AaXMENT9l^SavXgLnX zUes}~L18^)h(#9Opm(8He(_^_-BX1#$&%ON712La(E@8nq`{91UYQ(K-2wryUkv!UwGSTAyxls$0Pi_alP! zMjH(Rxwt1m&hOjN_szcS{AicR1k5vhHjX3cO`CF65$`zP+o**Esjg+e)GupTQfK|_ zgFY-sq!VZ=Dx7^Fl2Q!H3FyFXu{T1CU+#s$RU&AF3p>EasGh|8_; zoJFzt8iB=)EYao)@1q*Pc-LQJDdCq^GxpI;I`76CbMCuF+kRSLPs-C8BnM3g5(#-Q zy;if^E1S)=8XL9YVbyOHZIVNTh9lu?jssk+otGCGj~`5)QdurWf@U1uvvw-~>TPM# z%`e8UxfDnXmn3nXjOL8}*4#145UZidi-OX<3{w^ymvJHAxFsjmJI9?u%TN<_wv>WR zAl;ii2n!$OyaC}x2V0Z9@r6G==A=6MPs@5NL@%Q7=DJoaH>6OQ&f?XEQJ4EIJdl$C z)N1Xw)Hl|@cgP-x1kti#`28ci0g{zu9&7pnM^Wr~R{c^Me{CI%U;Hcwu9(pDv(?1ps@0;oP=?DY@fDsJvzF-eni zi$4E@dM$oqGobwmtx746Sn-WllMPy}1qFhhKWkMy+M;cS*u{ZAt@TVn6J}Fj1`Q8x zeBHVTX+vW%h{9RcWm6|2hbPLA_H<@%euXo%%4%^yJOnL|93cg_e&tVV6nvNBSF*s0 zn8Btgy@}<_Gz^ab4)LNNX1pV=cHJ&CAMD;F894zpY(|;l%MR83G%>oxDs<8zi-4dc zR^@UL-|~*1_~|1OLnX98?+GPA^J2vEkmhu%Yp(Bj`@ZJ0m%%cZ{RrJI_%Jt7l({^I z8kMnq8gMy3>Bb*@%Y*9`d1(M`{l)z#EWhHnClK!acNf zr1Zq`%O%Z#I}>&2iVmD6y27Ak)lo0zO=fDuh6Q(0bOr>j48iq!5Y;tO3xAu>b8s?> zM5=C(yDK2+fK#hQQ@iFn<|X<@E;8dJl!{`Pt4s`YX^}rQ7d5L5<(vKa? zcD&S6@Nx0Ci))hMW3w+HJ(3VDb(QnR158q#qo@{B(VTkLFEv;z1=$9*#u7TG@{vG z829or2o#NQkrumuB^Oyx!60dJ`xqH;zcBafUFMcg^>%?imLRI*;@Y3uvtfS0xaJFQ z!(fY}=H=!`tzVN?(W9(ZQc9&HZF%}9&DUs3LaOk6tez*pX(V6Kb-BJ!&*)S3IfIe& zIeiqUFgUk7GD7}hC4&#Nw*p^yoe`f@eFE_)oQfu6Mg2i|o`n19kDa=p7J;n0TO>&f zbsOcbsypaEpUF}pztDEjO_xiFeKJziozk_W&JK!vUW7EnhN^-4!TrhgdxlgTPBo!4 zVmY=Iep_?R!S;pDP3JAy{hLNa^1D`pA+K~ha4W{yg~iLJ&LY*14>rj&_Z8u%vLG3M z)?XGtSrOjsas*N+d{;>82ccI7S^}=`>K{}1{X%!Zv013Z)iz?}0;wwUmk7VLgz;@( zjqm?_RBDUVPDI`2w>zB3a|+LBD0Z2Xd(*BjlqPuKdT5#&JgY-J@=h{sn*gqhn9mrs z$*EOF=DxNVtcw+l@^${XVQmWQ?v5;lP{8#xZ{(>K`P#TC9qz5`v`IsQ^+*;1A;<(5 z*jgiwSo|-tV+g?*WepqtnuB4Ghw6XGM9rF?qDt_I!l|Dw{=!WRcP~N24M|MycVRQn zAJ#Sjq1tP!yOp^#K!S0|Si-*pGuQ;EX6_pzY&5M8h1l5CZcHugF9xv zM!VdiPoMe8Y+Uz#b{T`TL!Q9xDWWBwO77w#do-cMZ2Aq5KOZ|$#3V#h=kbMs$Mgo| zhGooI+1xk_oU2DQ>u~I+^lFzMZ5Ro*A)(k<7`_(~gqNVHy0y}g)|q}m@Ec&EJXrT@ z`lLBQrwexyT3I)=%v!;dPk>TnrxkQH`k}dHpZ2fDfs^^p0_AfWVwPd*8rc zsslAY=dvQrDP5UQlCdDw;CfD-7(Z|cf~Hj%9f#K>9~MpozX#HCWx6EC6YpX9erMOe z1hi0wzwk1LX$T6p-)_cT0!p<183zi6qV%qnithAw-Ez{rS2!JoRD+$}pf3ZFb^$qdIE$s5L(z%&TWL-LwFC_&w)&mZ`4!sWD7!k+!ZJ(k z3KZk+;mlVmhFbD9>&fdF34~ki|8AUl5SPE;Vw@^MvQH_AWaxc;8Nn%Am?+JtPBWuUGJUjL%b zd}bnosA=_X6NA?1oOb5+Mk*aO`=P;S&%8=LRd)VPf}LWqn9vPL57}S%-t#VYi)u_} z?JFN+J{WIZh2;QlL2i9Q4fvz*=fRmro%cLmI(v(3z-&#Sjvf{oGLHOEeG<2|@z%On z>v-SZzReJgxE8pgPDdO)q2cNBUwSi)O-+ zzHlC8+lcDUIvMI~KfsTS6el~t+G;=Y2Q-jt`|b)#PUQY|!^C=g2{YDKC1hbb2LUcL zcLN>E&n2Gol3R_8W-+g?rAX9fm^;O1mKy40QssebZWr@ciWOwhJ$Wq2B8y+3tXSGR z1w|gsXdHqq5$8c{>pRE~G3aHNz9z46<;q=yz)L1b&k~OHlkoR@9j}A6`ZN^k<~6;J zx(`hyPma9q>`;^%oe3Zxx*&Ra#d%-!S4bd)sfwcw_XEWLrMnm*1%4 zcN9A&%jZyE7qH8WOU|wC{AIF`X3w>dz4QTxE7Q(6NKYq&#{bzD1{%_$2(4khrh%<5iJHF;6*wrZ=j!uhuDPE$0sE zj2GgM0^0Xy(WW&Cl(L!h*=Wp#oLp&>dRbyi;$~dYss~}xeVl=puZNFvL83iri2NT% zXR#zH5Jk~n!evR}PU5h`-JP!=dNva=9Tg?;^4@c*8?p!haJ2Xe=*yVf_;95GTKemJ zX=}dMgU0d*0jLOaclywuixvbIblGzERP zeDW~{cY;AA=g?aU-8Ue0<>4sgP_ld-+n+h~ZsgmXdVi&CXVP)BU$!K<(q;`!SWlcZ z631y9%&M}3ICCOT;g3$XtR+hCJ^d$_jJ7{8`|eBzKV=F`O0Z^ybxiQnMmr}`*3hX9 zwQkl|QHVc=Ae<1tu$HW?w;2RyJ%?ag3#F z;o{AsRS_hhyVj*e6Hdo!%a*Fi##Z4c$1_965v-s)QAjyiA^c+#=#IJgRIlyeauAoKVA&8(p3NEX2&5ij z8H!D#87n9Q6xN6EcI`Ffz632E$K6POWRJjPR2@}o5$JBj2jUu{$FB6&Dp5h`{lqaH zy~S~kZu5{xZAtrE%kRghgD`K+(3sG#P)yFL=_1zfm_YWWN1)rA{cFRKRN1>BbJI zJudb$jNLj3SOa*xQFUlo7MS&OdcIp;Gt| z^70)^KK=i}rAYQq6wbvkrPjvmYL^=Ua{;Hn`&u2OE6I2LY^io7VD1~Z4z;3m$zDoC zL`V$~-dv8xS6pkH`$ zZz!W{}Djt97 zrFFXT7U-HX_G{S;8;zpUZ|{vh88p<9Q1;+|^C}-UZ6tNVSQ^Rp_1AG2dabuVipn}pQ2Q9YBu$ycgp$r&4XQYgVPl<7Lws8jMOERv+$zE3`Pm~o zwxOKDGm%m`2kS)d6qf6k+(4YH@kZF?Qml;*eYlgMfYs)MGTYA=>=cJFHw&2Dg`Xpf zaZGk>-C9vhNQeOrf(9l_jP^wbHBzQpt-hIp|fl(LKydX|~Qg8^MZlsGO1 zUV8yyneOrCm*-@xsg1+t_3zKRhjk}3?yGY>v43G~B^2T-^&UZX`hn!iE2Q_~uRdFc z|Dx@QSt_~+Khk4V?w+Ol5iCusbm%tfOilI#d78`Eh+mI=Fz?KhQE@O0kS)==?ER zrBEBRtYIiWR8U-gVKqAk;4kNQ{E)$WQ6`5ChUf$sB%_mh`WH=gogH!C%FqoqJ@l&~ zy#*%igzhe?NQSFRvF~$@;lR^q!2~z?Lr=8!x)Hp|4`RF@!DSK=I@fu>qA?oeiuauU zyLCC-B=}d(byT#)5kjMTHDfg}dfw!V>J3{h@;}M_5y{f6CAFFlB%RJQ7R}H7?44aA z1~A#*C0TyMHM1bA?>xjG3np$!mt=8hPJyTChl*1S; zQY)*NWrOI67NnY_Ns&N0=@cn_-1h#SVXPLo*nZuq*(p4TnsYf~Le9VJ1Wx38=ePI^ zkPw_(I0}Thim@L*Rr3qCeZ?&ox1Xr@X4><^q`03yVbUi((;0aZa)wY)vLT-LHx+Oj zg}tgR-o#k`u&qO=ccE#Ey9;kH1$;VfoTC$h(!MNNKH0yvOM%Y$rBC`yG-4jO-cD5_ zn#_e_$Z2uj4<^!<5_dB13`~yOQH5fKn%O=PVCUkAJrb2F;dp#_N+>;f1js+XdFEGQ z4!Ein(k;^J_a+)1LLyxYW)D-qnIyezdD77r>Yho z3OlP~Ln7OYt(YccXSJ%EGGOHO8&B2bdV}78sjXH2jil~TNp!UXyO2VXWd6Aiih+Mm zmP*9&JNZ>+fU$GQb|jt+tAi8Dv}k-{9>U}|;j7DM^nyf*{|@G&Jd7|<`y{5ih)@^7 zJGZbl?@3C1^5X^O7jkp_3$jGs+3tGXi91O5gDKPR0q9lDt#ite2^89OSl#0)Z?*E8(qx89(M-@O zKGfx6_nB#}evJ*j+3*m(g}QHMYRTpf|Feih@{*=7s15ICA;^=%yC&FZ(+4K1Zs%0B zAKZPvgK#00(-9=vq7XNT*pb^ck@c0#VukeQm1k5o2)s1D1YwjCpgQ&ChC&#hMbGTt{Jkp1F7dUq8Ef;UPL_MuP$<|;YXX0<^q*=L8jPTfZ~Cn z8~xVv?xeV-CT{40h1T$j%0}|SL3PH^Q*>OE53=0-QP_w5o!fevvtDa!@Y@q-_?E|@ z+-^G;o^oD(=30_Wc0~fh;6Q7#I@*JoLa(r{EG7jtBl*G`?d9y*i6{K%I%{9&k3dhj zdl04@8xa&zND(PmFhPT07(FFe)wXY6rzl3`hDI{Ei+Tqp(V};qv`h2izeR)AUh?j8 z?;Rg2=~gY3D-uK>u3DK~$qV&KH#Xc^nxoSyisK7t#slc^e}@@u8i4ol$nN`L zW*pS~6>e@S1%f5EJK(s|5)}0Lw47e?uL#B^8%QBoL_m|0Fhz>r9M8!jetGE~+1P)yDC53_z18L82HF*A5ljaGp`GxE05hvN}=mX6TF zczxMgFfMNgdP5oX>K>{SoA!r;{G5MlSzn+sjX9U*gK-MoH|I=%#o{*L#U~`vMU0{K zIPn$v^GF7Hw9+l-Yd^$(*7qkwWCUhhzlghJO=yG9C94fs4!Oej_G1Z8Nasx4FR`SZ zzd5SCY_j4vLwENOxm_AyVMn_$T>T9u3UO)5Fw~JQ>9Xcp02ohaW;HDvuN-ZOqR@B! z;_2GXss=qPH&#TsYnjZfvtbf7gbVw2b4v|;2z7Umw!D6A9vn1NKSiM-Mk@CKYidJS zQt-%>-dT4&;f$1bWnNo)K2!OW!c%?tiqmC`t5DiGC$Msm9Whw4NLVIGwgLq~)rnPA{g35FbE(?1p4+6hg3enrDk=UxgTDpbhXe z(O%+aLhpK6@sgoUH`QqE$8fK=_ClG1$J5q|bs66q{S_nUh`(A)-MMKmF~(OFNh4KE zyZkja`iJ7&kF~eimb&`-fVU>MmTrRTvv|9xe{}!~N;3?viO-y8_S~!R-?6pUxa@LF z@YB!(H+{`{+nLlgOJY$g&6v{d7nmTC!GY}3ByXn7Zdt8Caz}X)l7l-qcFPl#Yf9+` zKIlt=d>-6Jx6ivQw}|0T!WX>63JnZnl4)e%c?ZAwyE_MyZV8EM?W;PLS3qO}Gu2># zePO+qI@9xPR-v~^eNMwQ+A{82A46(ZntmrTfxV+Mc(YzaBb2Ivm3TZYnk)BOUE)eZ zSfT4`3Sl~hg+md}xfWC0W!heGPGNauKDcww$jscn?&p4m$ucH}$opkI(~26}R`1S_ zXOlxM6yt#zDd2!>({kc1YkJL97$kVPfgIPayK2#jSe{VN-;vE9iL>R3`8>#ecUhVA zsSbula}Q2q`d6x%1P@#*(e^HGB1ahnIR4n%6uDp2oT@3>w~XIcZdN3QD`=mzr)Ox4 z9>)xa1+T@T4_iNjLe{D6>h>2j6pCayE$$mm(la5TeG>$aHrJfe7k__-WS`K}LV@w`2Kq)Xlhd!pa4HlgOVR16 zTBF%(m2cg}gbiu$KFydPK8pO(b(rEZ+wOvc#U`8iiLXr?E zPn>uHgd~0eRytQ=A*D}RYUo7dTl1RK;_P9e1W(aEyrwnz684{%AQ?pjY z|5o9r0rRKp%3zk~&^Rom`b4MX=h)w|F3EtGi)Tr5!4f7R5sa@!Ts0Eoi1H7^oWbhZ zGSV2^G>EPoBXUZ{22#gNT;-SQVDgz@6NuR-7ox8wlVEzt@<7g&4 z^C||Tqi%?VD~l;(U(?@jHwS{oLOjt+?rB(xlq5Yh;$ebG!}aer>k{os1-H~4bb^~=5?9~$4WOoCGt z#FnIElM}E>`WxC)Y1i{1GdRYJD2&ct)-G17(4$)7NkkeOqI9NK$=Sk-f{6S2U!Y(&hx*O=mg>w{N^gC^2BH>a~4 zjaNT7xxw}*Bp;&8Cp`CqN>lR_{svNA)I%zg+lw|U(Yu{YV{ddA#M9p>Uq<*|Iv*wCrA~#QhX9{)^B0k~v+Z~}bWBKEIi$}b!49Pr| zN(RL#lQw9#6}1|Of2!Z~l96>KUb3iuX^Fnk4fSKDDbo41y-X#<2~FYS$e!;py@_QD zP>%|Pfo3QJy^VE%5u*e|iYk7L-AIl_g=2f(ajH(Lql%kJ;~-%y=SmL}ifWqKrg5t}hA#n#WK*;v)U=SE1Do zuPMbApI>IR)dJ-ew&kCvUQLS^d*r7e1uj$TP1Kok4RbX*=+tLaT&}xdPQezf-}2Hj z&DiZCXS(+`n*PW~Xn;`*S8T&jggg_a^G62!me0|-KABwj7HMJa# zb;|wU?{X_%@9U^c&x;~h8`Hl2q zR4}5Jru$0X%%FHXO!~%XAMyOWgW`F@NoHUa5v^G}{pDMQ@Zck|Rna5j77XR{Gu(`x z;S;#Y(wQzEDyPB2{?nq$pDRuLC9*zYv(=!e9LC#S_YH<4m(^s8ONK-BmBhqUWJm0o77WM z5Uxs`tIAsxjuVENiEP-{D7t(p9SLr3cgyqMz3dh;^%k*w7@ zM5Sq0TETJVRG(^Hk~nk;YoUS$po=8Md>3#62{+;XE3WXp5>~z6BQu;d_KQF=F***` zLdgRng!eX5R*=|2Jg(C9m7hn z`57HL`L^&__ftgQwjb`<$lX_uH+kdONCXTDg^ogAsI*UQz+eUUY`BmIaVnF9=)LZe zle*G<+q*zei*J*L1KOaDEK}m=XENLUfw}v}?QY9L!?U4psYYW$0K5yksA}#|YlI&` z!vyjm9QNabB2~YYJPBfl|2TY^m7bp)(MGJ8NbpCuYY$m*Swq=jjy!piR|l?GTUEZr zXtEu~Kv-t1d~DRs@iIDZxKDe(4C2E&S~g-`3$2vmIWW*gQ+~;?L=_XY**?4tx9}I+ z1F0eNZ|Q}V>Xk=dtQ83g%};I%ZCH42PW#|9fcK{L9aT22?<~kSSm`{&UPh))($o}j zqUyvN;zNAG;LUS{vWC!Hgsp=8wRAheiXWj)O*QNHxKNl+SEZm{4e+CzqR*fVv=ytx zi7gpV^AXp73C0NBkC-{I@su(L(Q_vR<(cYU5xz|5xStRw^f@WD)&3z)bbMEge4#lt zPXB~z|5!`gfn*&99E(9$aGoGa(z1e zZTlu(G;Ll74U>e3;@ewtFcdn~NVl+D=jrm!GXx{FA}NP65w;?uw~^9JGvv9f;f?a= zjg0Z8inHOo(c#PJV|Cw+XzoNaz+P1|xs#>-LJaq1UWYF&_S;xhyN6{En{-ebP*ZEGfVcZoznQ8BoCw zTCj4)!(D052aXWmNl)fXb|Kme`<}G7;C%?v#<}aK9#8VE{$$$Giuna(JGr-mTw}+J z0|t4`@be{)6t!0~LXeJG4tIvWAh$UL)~X9p>h*vsFRiE6gNY@h0vv~+;`;rVwPcGz z##FB9R(OutOj}T?P1~wEONXXlqet8Aa=0=mV^P!Jg|^|PNOa8vo|rxDwQROhV1QU4 zTNN&99b0yTC?X4&WFpifsxAJj^TS;4)}Wc^+J@{?K?RqEi$V6}2NgDdLaiTf`er~+ zS?Z;{#>g;pSInT5Z}NNB#eG_8lz+SBPu6Lt`10Khep^>{?9{uw;+Mb@9c8@bE698F ztAsAS2Yy5)i)R7JcfM)xCYYou{4yBnu?8)*G$^bf5p+u~e_cW({T*&h(X_)UL`KUC z1`>VF?K`Y9PS~dG2gBq}8)MFwGm2T*?AaZoyE*O4QKhUAG?F4P(x_|YS6=)?SUPUn zZp;|(h=@^rh6#tF3chBPAcK_&viV`5$AdtFjy0Su`3IJbw0%_$ z0GqKgbaHe|N7MWje1L_qd%Y86@o&PV9qe|Gx3M~=ls!DPCoWc-HD!r9xAj6UMnjs` z+A@{S44#yT)74PpYj!L{48 z1q~~1YYwWEr}XTZkTsfw+$=w+P?=68u3u>Vd{&9f)`Hu+@Eq5gtt7arBtA>1!-`xT z1SbOW8R1}h00!Ju^{0l;H>kBGug^1%RtGF#GxZd9Sw`KK{78k(oN>>Wb3(reJJlSK z(9~ClsZ#{!N5rVOt_Y^_d&43ONJ+|iFb z5Y}T7{H5MXDQ=hkpjJ;JQ04qi*SI!uARwdz=YCoSDWZ@p3snwxB&T1~^-mYKq>W@> zT~!jNQ{^&qCHJQtK>@;DT;wBeIGi&Pzj+xqaM@o#NDe)}Df~$MFjCPDYJptrxVZGvv| zMv9S5rB$A4U2td_0*w(sGFR*m`*pPkBO7(4CY?Dj-o8j}#?8h3GXH(&a(Ka8ixrC`YbY@RunFpDwgKzwwUxNPIPDD|2Cla9X){_g1Sc3Ig$G1%XptPjJk`B!SkFv0<7)9mUmI3~R`R0Pc^QFl!UQm^0u z-e)7%rR0!4vd1S9k%DrRWF)MANHALBi71|X@+o~450VF{>`GRKV{&QsPD@eAfEGUg zOQJY4DD8)z=`F;|*@1q(VzFoX;qiaf{BFtvXhT9yFp%?45ZL9HP3e4=yLPi^m|Bj& zUQZkfZ^ZJ!K%vtlc5X;AtL6z$*amzJk-sNyPC>lcAMbR9ZZJj1L(|^CaxJ`*7$OVmIC5uJ3 z&FTjr&3J+JlBcAedD)=uE9D6^0A4jYzQ@n_s|j6>l%;tf+ff?6lDd`f@6 zKmgJ;!84>P?fg+vY%OuiwoWV5hcNwJ!(hE}c9T|9E5Vak$d@a3~EdeUzqxBXakaye4)>Edse@`CqK}&sx>Z zN@D&3DGa^$7)-OOW{Dou}( zxA&CL>F)~K@_wN5N49o}<>tl1;{oxjnAJ|?>)bN$0YxFiYX5RcBxMT61I5A!^I4)R z^4N>GQahISV)|Hk%|?-U-ssckae3@IPQi`hInobriX_r~^@xBgSNjC%Pe>WV-Duye zDXTkLKI!-IBOOd9nRJ^VHBN{HBw(4eFL0STWA!{wZ=LQFHCNy8@WI0BtrJm?w)mEc zqqx_Y)m4hsdo>|~>aKIi^pIyuOk(1bqY!h=9>I_w zAbn+Zg7GNT%Y^UCdHYfOU0wfDZKtOe8_IS!TXs53Q*4}d+c;Zz|BRE{cA542MuSyQXjsCFcuhF(n_&!hQu7duG+Uau+vys@##X4HPw7_l6z?u z8iNdMq7xKRuL$^9n2n%nY5iPXGlilnPdZsMPBR&{-9AiTwh~9gVvQA?J(Z4p z_VFh6-`t_yuahMXqvpM*{Ut=C3v0>aR_ThXG(5_b`sok!Vivk4C``BB~)mVIl5z2P%8>UyhbGIV>QuCyjuC4=rVg)GxUY1dW~={fN0x|w~-WBkh7=GNeX9aNrNpi z!wA10=gkmeZK_i0LMNeIBngv_)2tuu{?vbf}p-{s}!p_&U zeu>SR{!{d<()6xSWW??oQR=JboeVFw*mM=3(a-ipDOE2*dYq$mzbaxMQqDXsje|4R z^GF+lBNGY9RyRrnyw(mi%G%VbCROSsapfIB{-x!CKS!BAnj#UhEOaefv@;u3d%4@t z;b}o7*S^MM6@l|IBRwdeuX||JzDx)ht=-a)2}}unf`|l>3sUh)eF+Vir#R+{FN?i% z2{!0h$qsic>D^966C_#Jx-F54za2fy4GOtXjM=#BEYSGLK6(KoX*yX%>K&f{G|16B z#|RP_sVf+NRJnQ$)GGYs8~e$28^Fr1^1ZJ3v_|ty?Oxj}zvYB0DxxY4+QP7mz^Pp+ zUmv*9p7@;;p8f%Kebg$<5Ou~0h=nnt$+&Ge3JVdgf(eF%qzV(hXrnI~1I==*t5@2$ zuUUi|qx;fXfNyo|F}JDVqmE6&`2sVqj5UCUJ5UoI(v^m<)Oj>jJBzO6b(@Vs^V9ed zJ;U$3aQNl>D+FaKo`Lo{J4w;lXp%{sDxIo0Ig#Ue4qjU0FkvuHZu9u(t2FN(YN^)Q zj(ZYGQqbpDIdt2tnMJ(qX|^`g>JQzFxH%xZzQ%sZvd`LtTn4?;ufN@F3>dRx1f#Un zl~~gV8diDW^SoKxG~iD7)D-sm%MQG;6z|baTQr0Q**tO*FKdmQY{0)$;W{Vn0fOm^ zQ<4^Rc*tZ%Ini@}cGRlUGhFzAg`FG)j?7Y(fPZ6AS}4LB&OJ}-)wvR-X!%khkxUpm zWSOv2Qk53?%%mLa9G6oGmv$WIb&Eo(bnHVnvj)B^rt*w-5=Bs0a{{`D<-+tBhF^65 z^$-(}&zB@V<`8YQZFI9pb3$rwNGDdlZ0YFx3q40)OP9H^%b!Vy*&{SLqTpsAa8(^z?h$f8C5jQ7jA>t64=gR&XbX@M=oeVF*YOk2@GYcI zvheuXPee3XZo}De=icmN)&k%Y=0>Wuw+qG1!~m8*&-jZtzQ^nHH9uE&$Z#n#+syLM z4RfP3G`5ExT=U~wkM>0p9|%qD<%qcBTa+Z9c$1OmIQMP{N5=E{V9;|!1@njt|B3O@ zC(_T8KdAc=#)A}Xru*rT;#&0m|2Nv)q-t_}bC%g-k;7?Wy7xRr#?+@XRX*pitG3=H z!x--oh-Y0`qb5u@EwBCht*5N+sXCDi@eF>(bl&qwkMW3}+zb0bNs#DOkiMx9v9Mn) zMh#Nm9mbUoo%~u=9f*)D^bd@);X#jzbDnli^c@TjxWy-<-0e<=0UPn{tT6DmYnGS1IpW{ZGyu5vSkgmjTVx{(vHsoY6?)urW zuc4H4+_G|tFMYic1%eQX!+Kjvu^JHPcd{{?EV`3uTwU^DBZA;rq+|2#pq8kUx-HLT zwd=%yQN;<#e!B*J2EuwxSv5!M;A?%T#ocL^E6$T=`??4=+v4M2EN$8pk6y{m)YaXa zcKGVzoS`shnINKtAgX>wF56)x8Pr>SlZS#Z2LTs*$)Yaex>tcX^p}8X7( zE=EWxWrz^T5i!|x_g?yYivjxtm!1jTMR(#CK&W_%3Pl&rm&&Wk^67?9lhHq@ju*Zz zytZD}e0i&Sb4zL2k8OwxF<+8}fF)ka2J;)hohhZpsyk^CQ}x?*_l{`qKqhMzXV1` z?V=Uwo~ufIz>MaN!sUh=}m@WRxadA263XRX&qn2d}Oma{U z4S`LnRyt%^R9yZJQV>=F+8XAt&7^ig5GNI31QykL3p}h4h+Qi0YQVH206jp$zky2G z7`e-61~mbrz4ewV`t2I^{%hFCLE4fNj6u?U)9Es}22|my{H{aQ>@}htmEmHBY*)Xn zdO@csJ=q7oQ{ls3Ar;7iDE62Chk>_o{*-dT(T8s>4U=3@&GY)%al=+YW#i%z6h}Dj zjSd_4J=mUYqF^*n@{ZEitEONYyKTWG4ZyCyyioH(xnEUr19k`$c)i+$C7{nrVdd;2 z70Rrcc31J&eTdLtM;A*;4|5~s=XbGDWDjF~k~2&aGMl3%RuVj7jP!^&aJGvbh|lhbpMxCh zu`)W>*&Cy3N{U?4>~;g=jivrhA44C*Kz7|QI}6K*db!f=d|w-iVygFDFjW!+QDu06BjzXjYVnN5NxNe zi+)m}AJ2Z6aL{dIf*EWb=$H*9#UE&@??+Q@{oS4ccaZi| zv~CX*{5PYxqBSr|mBBduIE22Vl|P(MI0h10$+oG!L`Ae<3i(ha3{Bk7_3&BC!LR z;fnehU!U-zKoOS-XN->@Z$f-fWEwFTSaX6g^{q^sd6Lw0?_zQ!+YIu32_cggI=(Uz z)*G&pu=+678$qoU<5-tAuCU>*eUpNT-CQi;RPU_RU@Z0z!vY~df*PFr`XdEt3g{Lo z$P)JG%|fQxaGL&9=k$*IrNjm&M1O%0bq&I!C5n%O0b@;xR(ksK70!4rR7kBemLy}{ zqfgczIK1pf$zlDlhw(kS#lRH6u9p8V*)z;2p;4`|Kpf`lq1Cmg5Efksb(-(Q%C?7}(`La%y8-NJyM zZpwf@W&7uWeJj6EJ8~o|fFiJxj~%7xXxF9r`uZJRnrE;V6e{8SqsosBr#o-r=rKCF zfIEs!xFhllXr*7JMqmTW%)1c7!@^4P{{3wM*MJ3i_hQT%hfkADevLS=WLAQ@Il_p) zZ;H%t%myDqJn zoz^XDw~R#VyJUUtXd0cu4U*8eZ!LE;oASb24n$h)3vpN`&V3*f%Vx5*NQe~3XYsKu zrURTP7W%v~m8Q~=VT%3$+6wmR#j@2XnmbeA@d-e|k(sD8^#--O3!^2rY)0N^y2V5n{35Gh*203^uyc2ckvvIV(`i6`2B29Xp5P`AR0AqWNP?zUM+69PG{ar>s&dX1U#U*Yfz^3x@89|b@3r=gOBjK!-pNEd1tJ7^jOlYm zO%HJe2t4B6Pa}o;Bk4P?nkg3xdR;1ACJJvi)~T!9F()S1 z_MGop>^hbsYRkB*d@1hZ$4wB`ibbiEdHo=W1z?I9zA+9^JbvZJ%YSE4YYwAv)912K z+>rZ@{Sck~6b>WNP-d!oKqQysD&5x!56b)uWn!&cWxHe*xo)UXUnW~Ni0x`w#cVBW zyz>ui%26@JF&k!r+<9E?i8@_4f6Rs6oaN4K(hjn3j{YL5E%Q{f%3lrkHXOMcGil%c zQB04&cDvqqfGsW1IT&+r$e6`8z(WY<)g@Zg;(Q_vvlPaBKEY+P)|y>t6Dlb%5{!ui zg%=@rS>QM={DNh}$G%kNMlDB2tt)U7AUexibEXQcl_7lb-kcA1FjWbMYk||#D07K$V8p^REN?YZcMSi)x5mHO> zdY>eF>J^Ud*+9sWnrje)cTj+iO@xNb`_$NhA%rkK3z z8;_1E#BiwY&GgWiGYn8L*Yh_S!?0$aqK_t_dSIAf{K_Pemr|Q|y6%zPG|Snao8@#9 zqn%?&xKgQ$uBzNzKs=cnubT<9NAM@kj@lfe{@K9ZFk=hk#OGm4821jAJ9;YO|L|MD zx>KGAf*EP8v}yQyqG$uHwHO87^%ks1G-poX51RKC!}LCc*Bt-B&Qu&HYN0+B-FAs! zrl) zs6_q9&3%cAt@fiekqRsBa7POmcN0f}#>!IdmNzM&q9=O?rsn%l0Vxm(&-p#>;2Qqx zxm6mioMB~+ogI~S>L4po{Pqx%=C|%FNiEN%oF_y zOOFe_T_a|q#&d5!&-Q~XUh703zqE~f%DAP8gX^d$A!V|dBz~A9-rQj5@)V$ld)lRe zT-a{z+@ThJAetL80pR?0>I-GX#L%vJOXQu7&{*whkV;R%%QsEWTjeC=C%&XJ2pKgB zdTNrF78bWw;Fxc;pRf+J?n5Q%8lEkf`%L!gD7u$7wYeds`O9J8d*wJezt>)P_*zc$ zv|4;>l|UZOsUm83Bxe3z_?v8pNRMU_ctbXaO%Q)pz4)Ynsx!6PWf!N* z@$2lpw?g{h2O7i?Gc3pJ|Lq<9MZXRhGF@K11#1 zApke0SjF0T@D`B~VZ6d&@S>MPvyP!;$`r~-pvx8x;B^MV}E2XW)QCXsP`KsWv$j4%E%<2sXZ$G%=-1BWn5{$6|xRU@Ko-oX8jfrb0 zrh+4z#MUVBR8>^(1yVK_FC|Zr-C(rBnc6Cfw>BkI(wTVj5u*ou#OI0=4p{0Z!gk4< z)1{(0!dB|)FS%;|)MyE&>8r<*Y>79*hwL!>3dqmcBz=JA53_uxkP!miuCmlx9*2lI z$;E+~2}f_x?&+Fx`ZAdb8^<$0P)}D^xYca(gY@*2NQPUtV?1?vNnkM(Db3oSAY;=V z#_jh)0sKYMc1zLpaHGRi0BFqqD{lkr1oT!@3DUO|Llc@WRUBTWar%`4er5-h>l$Lm z9yp_;p@sguy;M*4xnTMdoF@7Er6M$PSN)InKW1Jo8mL;s2b zK+5zg#P{!&O?V<;c!saSC(YU4<)UiQy+$&H3Ss>^mUJMkSJa2=r*&_CAgOvaqqNiE zS4ZgMsYm;U?XununI9klMh1F9g+9=+R|16-Chjvl);ea-)P1w)o80o^i)?r$%xzka zs57I9w2|k(uIZgX@kI~lzFV7@ca&jT^q=m1OJDo~ru-(xn+jb8O%^37POY(BDHN=w zRu0qiU5;yg4^H)STvoJj3TzW}olyXEm|X9$N}SWL5M{Y$`Ay$C!seV!{0(26luu*^ z55p>j!CcC@f;=q0_6}{$Dn6#-jBEj9D%|nVBq#j>D#=U2%vLd_rHo*6zO}EQK>UO! z;+q<>pxgK7>po85H!SCiO%_etO!`oIKMQ0@LE5yE*hEVaTrLs}UPe$NyNEe_!`Fbw zMq1pfnI5}}KGmT5A`s44b@iFAX{H~lR%e0rI84vbTYd_9rHVrQwgN8-`idywvB&c@ z?_9*s>2x@FG(~P1l#5zw@@}VC4X0&yoh4$tsw5;%Xwa3QTm&}LwK@coXg%{aG^R*1 zs8_|W8-*-mo;Y-8l78JqL@tbmy+e(*1_LpvuLZs-^uE?l;}p1drGtD>Ddti_(^)It zjIZd$17tNC5Bh*Ux4mEU0ll;Enf+CrK@Wz&T%fxB{?JUiE@$By2lS(H{VAqZ@EI^5 zrskh0|6|AjpHRSeQ-r%HEj<|0az%rw5e^gK`xBD9mX_8mlOljab8pQ##J8oiuVa*a z#0(JgC?ymm%_XNEZG8Zi01;)vdVcY(J@|396O`+Q*)S!?B`LXBUwf6}5Vt<{OwSWv zUPC!E$T2X(dE1hw-D)u+#& zj=2{Hw_UMIcc_nz}uEefNs?aQa_>2-8Iqff1H)&Pocl zeKJ)E^Jx-E#@#r`=yYUVxZ30^8cg>-vU{evWeeZZkptvv;$lS>zjJq<+|od` z4~jIE8hZUA;4PzP*r4uDSQQV*A)hiV6-5-v%t+pwD%sLY5sqpgR~Hd!riY@b?m)}p zPGhk}Wo%q_heV^$)MLB6ie6_1Rl*NV*#;bPyoGRK6lnvb{eJG}vN3j}S|zNKQxd^< z#8A1{=f;`VrTyk%3{SZ)`%v{QU4TkAlVSoH;QJtbgFKXK(yNEqUt_1=t*a=7X=_Fg zf41@yr{^K*nV-WJJ+PL-B*SPogYqU4Ks-phK70X|WZ+a(TK<;;GIqYVk^=O^lZgT-xWrrr2n4U^I#VY$%oxoc=H{B4De?$7HpI3kVuA(^-HyhL(th~W=q?fHHlkwQWHw; z3JTAC5<=>`i(gu(e{IQRUdmqU!?2KMU3xR|!#dZ6WmBGpL7Fk*A9Nmli)e-@jSqG* zSC1~#_;=4ElaUFEy=AB1NNgDm_e{`_9T(`vG)ATZr_u#nmb1VV6eo5~OJ)Us-uxh0 zBD_h`vt*lhvZNGAD!z_KT{28cmiD&Zx;Gu&oM0v(_c^6(TWl?pR1MUH*1Z?pEnWoW z!*MD~x^*4JwXLq3c{2w2w?VE~W91%Zp``G|*_>CAv3bBr?HfRXaf9d~qw~r(fL!sE zQn$U4h_}3&f>B%aGmyS&8@H@avLV0J=OErEGkBR&tLHLn@HN|QhLJ|nF@L_+Z+4zP zsZPfyV=`e!)^Yf239~7`tUtx^_w((v361Wpo*~blNG-2bJ*&H=I@05Atz{GANxuZI;36_#`pMXGdr$SuF0z$K4jTRJ582N61G zx3tEXo!Z}b5y4q*xXeHWYM0yxZKB8jBq@|wg^US96sjumXC4)OZiC2CgRs=<8ra$H5I2aOi`QiPEdpIm{G zRk`BzB7lB##Q%I(_w7+2r0OtdaSR*`8wK2^xW9cy9xgucPT1q+azpWmdF6RbQ;|y~ zcoY6;!WEP&H=&ojSD~5Pl+>I}6A&;H>#?$ns%hWFItt)-7a9a>y+mP}Ro$U9fkk}G z=f?*acMJIS(ZF^?q(WlFA&42>DD~zUq>huE_l+aH;A1sv>MfQ{4{oBTsU`$GJg)ZfGWKp33lctKY;5Am;o&_Ag zFW#}i=6PDHb}X}`(Q@Ort60i}LBxMs+B-{`5lk&_z8c=gCEFKH2VI3{P+2@~%GN4w zHwWnvoL~12bftx##x~YIQZB8-kA#THHA_r3F&ZgQ(U7GeFH`Svv!jJHiqF0y%>dtl zubQ8n&*DI`_Rm4RW567YAS7}#ZwQ}68gPcr5y;Pm5EP27i)fR?BaY_-J~Sni3Y<=p zhCMulm7v`qW1oV>Ur@nkL zy*hrx5ye5OLSf|Oa+r^;c{R*Oh3e%?5U+)(;t0beBE!#vi4){^Q~&biJ8++5WiRgz zUp;b}Oh}up_Gj!@bueCR$5 z>T(QyQXb(}%Rk=btbD@9o$DCTuWW?iZ1{d4*Ceq4JAEg}w#N8m5F2vY{6Qg0?M>@> zHcS*y+%|Mkoj%J!!OCxnwDcmPotYs}NelJ!sm?q5OA4rb*}pJF^@w)l+}8+my<%*=+>C}AQKrd~4o0s*@zeahHETL$qY=C% zfI1K0oMMu!n3~AF>U*a?qtf~}#~uM#2U#)t7j5N}dL1Lj>567B2Sj_Rz_Lm$eQiRQ zgldBWA0=#j5snoFh4JU&ttc_jG7w+&DwVl!QG{(Ns;1QILDx~+Zui6_>IY2qFkfsH zX0c@(3>V}MGkEJ5bfc)!zn!k!EbiEqX%6uy`#G%1Tf~2UgtP)trUtih9{`gEviXwG z>^FDbE{;^U(jNuJJM)_OJ0ob0kY*~u^Bwhz zraolbIy#W&m31c#5b-HZ2@&_a@okw@#sl^=eOO|ve+TM`S;k*~F-qPJc2B@NH0+G| z<3}r{4E;L!II*Ao*uLrCAC>np`mTd=t?45{0V`omf;sMdajGfxMG4_5Ologbz8vqV z3#JtJ4ynJvo22qlOW106q5xXjT_0+B>!F4G;G*`+u}YJaRB>rku3CCa3cu}lr9mdS zv@VB=9WPb>OQVwahbo&Sq05UcRAH}zeVizcTv}vQOsPov9Sfr)5ypw>5BQH-34~$C z(rJ=QI|&&}(e3Nsk?jBAYgrg7LFq?t-|Xn-F{jzg3O8q%4mlot)wUus?w8*%W8u_E z4A=+@KWO|O_5d4E$O!SC0E&^PMM?b zZ8n{7QT`=^HJ&HhwlyC%^DrXZ>IgEH)Z#;*BZ=5g_XFx5QiLhdvlu0u{zU;7%*0Qh z;(T81u_5*_R+m;6Wkm(!yjzt@#z+jbi;twGCT?ES`T&I6{@i3o8o*Zy%}VmTd3T0`|zFCz_&5p1aaz>7}= z@1+gSc~@zd-w6>M=X&zz0#z8QO)LKjTt=((V3kSly*Tgpismqj3TkP{_rGNW8jeSw zyD)vDhQ8RM;;|H-GQNF1@)>z~zjy!&8)<9C(QqeZ^%Q>qGwmJ{YpOS(Jyv1FukH&X z+a$j{--CUQ5z-xkJ{M%Nr;XU1s>9ZzE2NDy95i7!-gVfuxUriU&(8a04U`*7My-jf zHgkt|C}0f!$*pVZg^%)n_tf=2TPq%2G1%mgPr}YzN&&_0j8c9ztq@Jm`;wuBX8pdqkm%+{H?amOU7f;PLFjC=pwaBDNf9@Yiio)eMxX{aK zs{Houeruo29XQ}3ccROSTeO~;cmh)U)+T@>ObUP;&0V76XRW zIGy{|#Y@&9;Jzo-NM5M$$+r5K{m!C^Lrmj(3}}Sk>Yazf85(k@?;;w+O&zz65GQZn zU#x8t`*OC89^4)RXJbEWJF1?AD|HQBWI*2<`kYu>znFSU%Be1BJMX%I_JbDvViaO> zTLzs&KV?8eQ7ABNvbw}!7D;^^@NPRAWb(8?Kc@^%g+Vx31<;kXGCjjrr4ROy9!OZ$ zf~>HE4T^!?iAcslQH&wKwub{*0tMX-(hfntRBJ^$rRK^BYz= zOch<~4_BTUCB}PzD`P~GVrVn=?fD(YewQn1g$41Fw>kNZSZo}P2Dck%Ih2xqMLYVI zSKIYUk%V|B9Q^b$?5A);T#56RmLBh^Pj+D?9t`<8CyhBkHsfZ&9byRSg^9vtEmt{Qa12Ze3Ho)M~e(q(Wy(91QHxumU7J14ZsuY zcof%g1myea;~`e(1S--()umWOIx?9*6M`JEbg@?XM#PXl4GhMS3KXg98HWyFnOuFq z30YN)gOr7kcr!>9gZa1XuU3#fQAAV{%ay{I+iRkt{^=)JL;8Rw0ub1wOXeUTngJHW zWnaTe-|i+m);4inEaClGQ5rM`i`L9---Q0`z{t1$_uepO#+!z$1z*Slv)J3%1tIz| zRafl-Hiubr@w}b=zuOtg#qz)ekWavzQ$7K7>oRNU^`^USu|gSoxVT}!wZe%(3rfjD zh^lvfm%zlEDdvuxI*%)Prh?PYQW`QGhm$fSY&At?#U*%<4j0576LeR5^aQ2|w<*fZ zp2WB01-ZGph{Ec{KkgDurfED`S$?n$yGP}{$1e<;eYs`TO7l!5Cz^W48 z>QxAXLwGex`CTaHn4HcsY2*!hp7Of~TAg4u*2EKc>#+HhHGxZF=q=npFT^X|b>Ruq zeDlMMmcO8?s2epOX?`M`SA)K5nS90ed{tb-n_1(>CDDlNcH^nUXT{mnkc?R0*WA3_ zaNn?VVA|kZERnT*D z{zMlOY#7 zf8H)j1xv^v^2!2s=tYf2wRObBqUD1^Q^ zA(P-~IpL&BbE>8~u=qi7k+q2@HW!nniRIYdiNR{RsXuayvanP-hhU>5XuF!E74?9C zD({V!D*Ir1i$+2(5^hi+Wl#07UO6UV_$x6M_&TAwFl%BNM>LnFTx0X!EI6=1kL{GUJa?e9vQa=UwjR!E>3G@Vgc{7tE4=>$u@p@X9Tj-R6 z?J^s-Sp(WCv?+qwD0XPa?E2cj|8C*T71D?sKi5h%du+8z=4=X+UWa+S9KD1ReCwSh~g8{wa38<#>wrQ zM>it?eC+LUADL}*3Us2Qc^w9Cje1LlZ|vZHmP^u}oF@xh{^rp)s2g zewBm^nhjs3*4+suo0UfPzsQDTfjieQ)!9mdUgZsD zH>xy}Rms8l($b=9kgchaAwMNelhIA&VPLj-?oGm)XJ4`+ZcQ({MVEKl z)`1mROu*qCs=VJ{!s()!|gCdxnTFvEL*^lkqqS(+@3<{`fE8`sTgSvrRhyT=tO_TaLY}}`ktQ_ zc97I4(0B!GD;o;+^4nM%8@UOt;Iq0Z|Y}<*vtWJibiUuWrwr7NSxpTcA~;;>vR1WJwOgjax0@bDN%v6*OIV zPtogzoH`W<*L)>`%fNy;cAR)XN$EwgPJ=1Xii3B7@CDQw*`k?JkIaGmLd~2= zEfq%9K7xF$-OuIcYZ3U{O=-9)sTk3TIv0ypk%*9?Rs^aiVYyk1_a((LtMcQ{KyR+v z$0LtZ(36Siu+9DHw^c04s5suM>LqeM))60+@-9>IGx*Lrjcf0AW=*-=p)vuolmoxO*Ds{WwHP|}9@d_WqMnpieW#!UmpWWJ+#R!(%!oLD)WI~o(OLQT z2vV%sO-FEkys9Z*qsafmXIetC18|G5aZ~J$A0>^Llt%ca+m!qgB{uwc-y1Qsg+Dmu zq@k4Ck^j2{buP38OxqZBGW*<2!F);9_#Hs289ZV_RTnWJR!6}jodBjm{ZKZ8=M~YU zjjK{enS63w-$#4`8nUiZC-oiJu@DVpFg9H7ZZ6=(fhjMYGXDt5?+s$Rq)xJdALz3F}`ps*?x*w4z2dHT!Vd2cnfTT^a%(f2q{xtA~rjr8ABa29HVpAJ6 zX$R4t9KRF`_W8TyGwaF078X55x%YWs7YKKwD+{Zz;!jYx4L z{vb%D7nAtA3{8aYWTcMg9?~foFxKs|U|g*>iH+k2Q|d#}EcMXeam;bAeT5lN12}0~ z@Y~MODJL<0#a`7vgS@mXZhNt8t$nFoIK^su<9FqY2KX^F(zy?BMEPjnOoUX1^D&@s zh(w9UQ(QWXqudgM#EcSEs7Eq$|BGlkNqH1rDG4uo^=b7-t4a|c(cxm@aS)xwk;~~p zIJKibEQ4LHL%da7U8&bI*D;8Jdy3_raan$JlV1Cj1H8xvfu)5lL_F4_1O0+z$e4yz zNk@!Ua#l+NPRYJn{;A+H4&e^kk$|<#2WA#1f$UhY=oC=jxfXG|Own6QTvERm4$pdm z1#hfb1?<bju);CFaTCI=Ot9ANNn1MKvKHs^TO4CvqDai@go2 zkPU*`h=~`S;L<&+Lc+av*M!HY&>9G@^-LZ)`DyI}jzY9$Ktz)Zq^A_QYQ}ip>aa_j z&A0+*o?6Rhc-^-GneX>*$i_X6tudmsD?*~p_Cg; z{>0Jd-!Zjo3+qgMi+)wkiWDQkO{&g?4)Nqv?Hb0|f&C20#ZBu-a(98!GM2bL;&zqT z@tze>e|I22Pu`sJ=GPG!9;PH)w;)aNj!7^Pu_k*NJwXbZxXM!{t~}(4?>H1Va_pk{ z(H`j-(?{LpB0OUhmo7IlKSD>OZ24Bx5esmnl`<7G_s^{rAxF92TnZht#)7g@lw-0~ zzZ{??6_oe{XY7nF=saIhEXXR~K)0+IshI?rU;y~aSHUD-`kMa53jKG3OD25Sb=eTY zoAku#i5#;Kk6#~J9Bbl$i`Jyz4IWz8qT&ym#@gD)c_-VZt^_KMFNon*M2-9XZpIQa zTKQWysecQpJ^K0%^4{eMNtU%SDP9r6`eCreP=_T``}^W^=}Y;Y?Gn;|_n?9wdr7i1 z`2D^o(C~t+-|$e**RCL=G^DvtRWr7bpLp|*@dL+A5gp5YU>;pc@u+qr+}FqvmR9|o zI=2+Q^K+)GeFj1$_JIb-ry$^_TfHkypU6(woALGIdcHbMHn;m;_9=YCF1vO|XM8($Y z>t~TZXY5Y*LNnJtQKU@jPfPg(%Wc&QzGrf$Ih65SBx(hNc#D+9rp1%Wj^_lML4Vph zNHoW46kHD`m(BF8^>E@*WJM`K_Qmx4i^mxa@3QUPd7U_IRuky4{dP7fp$$46%}q(Q z*Yw_a{guU{k5tRX;~~qNLeU8&oSX3Pe+6c7?%2mmGk;jvO7m4?Q=NsQSm(cRCWV7- z{3Sf$DJC++$_hSC>mcZFi_38z1>961K??0J?nr|bEnE@(c*)+Htvgw=r72th8Po{6 z3(L@8bMg-+PN=iPaKS!rX( zxOA$(s#0nT_FMe44b1Wpm!yy}XbxQQ%v$9*BRZ1;pPh%=q6jFrZWY3eLIfDQxx1H+ zv8CHn+L0c{c5Hb^9UldK<4-9Mg%Z5(f^d$_JimxShRlD>dlfJy`Eu*&kJt=0e+pbhV2(AlX7u0sj52$TCnG(LK%~&!5 z1NPA~#Q3zZ5}cILj^Sc(b0Fo|sp}D_2A`5vk zRxb#gmf0)KT4Cnda#Y)`GI1N@tobAr<0Bz47VBKs#|6ehy-{AS!TQ#MialKYN#I=~ z6=RBm6|zUbJrf=1xh;H8=gTH1{DMuiEBvN!#zEu9sEKzM#|~%Y!XLt|+%TU;e)5|m zzfkZ!P9(bsV?$tJyhMHpF0bZg_~ZjZ)0k}Em_-^Gl%QV9AdAu$RU0k$AITn;)+oMI zVtuHdg>t0VkjVdcBixFKiQi>b?6b%djoAauY!Ui^0(5%fOLXwSKk?|PZ%{U4Ci2tAi1`aP80~kNixetW_yVI6 zldI|UKoL;v`Y0E+RG+dOM+~Aq>EzrnOup*ozRZ=XR7oH#gX8Q%dhzR6DsbF*ZIHJ# z6Vjb)8})RHx5Y8CkKcrx2FUFh5kW%_Y%XE=vc>eUM7gsN#9eVaXIN(2WKSaKIHH-J zd|slMs)(9Uj%Ce+U@#7|s5SF0F~gfm)cT}YAqzN8=W-U#~ z;g;xasY4inFbF^O~daR=K7w}BG<)l8#Av(MdCDi?dBOje&(fE@A zj&>fKL`cQ?Ocg>{#EI==!+yibUlry=q9>*U=qY~`9<)?1GI&KjRjn~!bhHj1RJl=8 zUy(h=BlBUNE9qq#h6#|}IOGzt^Hh8RMa8dA4l6s<#@WMBmf;|ZRj>;K$>>;(Z29#x zZoQ8tsP)p46Ap_4TqEuT$46GZyX#4`?IX#7m@1}A&~wf`GS0H2W(c@lm;pCb&B6|E zBkq#>oyVw!L@iWlxW|xB)<~mo->FA5n7-HO2!EX=dr><4bPAuZWK6FRKqI3%vycxE zc|zgxDu?jYR4d$$0%q(@k?l^M3vq^6jIsE(>uSx4ODo0n7D=qP>R(*au_5`Q2?6{R zX}z~USklJlAh&Da@cUJl5gUqdTooBYJelp_`RZaT!WZ!=Htme<*!Kpi9{H--`yoHv z66fP0-8+R(EjM&#I4;X-#<6LScnV6nUSMd1f>0MSdvP#f&qvj@ag^PyQbo}H7J+rQRm<@PybY5`N8DB|P1OR5%Ux zXee7Ykc}pav<&73Q$#VbvJc7a|IqYXWT$7flOs^MJ$zFl&Or-D=>2wvHtd|wAp^24 zhk+rdq`xO?9HC2^!Nbk@-wmkj@vs#%QXDJnY$SEn;n|Y^U)N&|_VpD$lZum$7UR-( z8RQLPe<>s=k4*?YL!=))&|Yqu|Dd$9RIcS_F($L@i-m}(&s8C*Vx34gT}HnNE>Q@7at;4Z zLwPBYCH(M$BD>#~VSppJ9C!`|NvoilS(Rar5m8Ixj_H1q<4{A{02iE$}R6hj#4KlHXHJ!SgUzQcl9d=9qe+F?pi173Fs+mZhUZj56|^S5|_21UoyBo&B0B>Aw(?HO;+u&2J3j0L%)Q0mjrn*T3HsU(V>vA z_MK4~Ifz~tRe@wae>xEd`8D(5=H&3U$=7^d*6Ap=^~WB^zBZYXe}tJ)p$#~?x`OAC z<<36B)b^f_ad<>Bt#(48%Le?gY`?VNl}k{QWo3w2K&K3s(g`I%IsP+&vi!1Zv%71n zjP-~*PL80uQylx{qY;EC?5RjM30Kq_IGnU2mTlJ%lG@gn-dkGjqfjx0c~gVrkpfjI z&uB+i^?*5r+L*`cmVXnnEV-wJ+Rj0+dKMP%j@3R()UX{szX%JkxKR>!J3+@yW^ zg>;v_G7Yb64f%K9)I|W3L zba_)cs5rLR){XCmMdR!4E+c;dQ{91)#W1#P<~2}-Zijb1XQp|uE}F;2r+H0}U2ni+ zwD99X)W=kq^fp(KXU*&Oq8CuJp*SaVx)iQjua^Xu3xEBj{UTP*_um~tbHPo24|D>3 zG8f@S>Pz=NGE-v4+L&W)Q}>VNF0eaZ{qovSvxx`P+E5aLfe`O)2@~B|-zFIec;c@U zQxZ`vSa_3``C@BhTrp@0L49K~e@D?NopL^`a3&yRV=6J`1?tt<}?tI|8bY(pQvw;_QrU!7i3e_aOEIi3bs)KXECTVbgkQAp)XbizfKsrjF4eC3~MO0 z^*Zo-<6n!Qi<}+m!gC8{4r2xvY`Ui?H*ErG1uPwUGg|);zvF5L3)-0V!=({JgnX<6 zYQ6P7`%JW5k~K9v0bOR;8nr_^Ug9f6sy}+r&LnNm#MN$Wg}x&cx{J=QqNA`q)qqgL}(F32N)NI zIbtVUyvVF?{B6f#54m7zpKbR~%%{Mu z_Luu2%sbFlJ;!HWN*}7nhkR>>)4R8-Jq2gV>Py_W4W;&?P;&HdQ6d-r2O~rY;_1fGC6$+sMXZCZm{*4-S2ky9Tru(kZ$H> zw{F5^PH#_DYj0bL-&vxod@R;czIqL0{Xr-c;QgXRhAw^_u_#^xb!sEyyCPo+j-OC< zSJx|#nWHY|&;_Z$4f=Kx+`D7_@ zFK~RBz!RZ*tqGY+FDiJc8!eiQ` z%fVG|P}euEmgLJG95FT4Jh?>qNj-47AEBs>uYs$dk!XpuAXUVD!`iSd$=hKKbaB(p zJ{h=WY1}CxN{;T-((D_;4Of`dMA0Gd=K36c4Fh~0L}>G)+tKYQ$P8mXBgAy-b^7^@ z@SK`7)<>4Wz%6{-^yqMqFe_hB{_2E7_VPEG@26Ixv-e4WMi1n=WzA||?e%~i@nD}$giNA_qx=0gwLpu%)@atUVz->k{c`EF17OkNH% z;f~%knqj!Fj`|G!u3Cy4Kr6|TTvdexaIMhDA26M`nZhp%UjeVZLeKT4>#H*tXWiz` z9oF*Ne4m8_H~ne>eo!2Zl#Qu&+sUu^&DXsfHZJ{zIC{c5jT)q; zsP)nXCXf;^N>u8&wvFoP{TSdB*HlZ8<0a#A2#&?s~L;R{zHSPr_|3GkS%S!;s!$6&6j$5vh2j>NXJ zdBdk!5Z~vs$6uV%l8XeUZ(GbOI`8H-gT@$e(vxDJ<|>~kN7|yQ=(De%SDRM`zBA1u z*GA2ehTMr74GQwJ)@2U))t86_+NTj>0KbF(Mawo#izv*Nv_vJCQs@^AV}oR9by8BT zGfhbK=DB@Kuc}4AY(4^2h^O5!PKxEB(Xh6C+Q^PtFg-CjGbVi)*N>Q25{IJuJPei2 z2qH<=`_PYP8SkuGZfhGS;DliUPI!onbTQSw(ih&G!mDp|u5JwGc|TPh$>e2>A?+Y3 z+B`QZEzSFh<>O#~EjZU2WniWMuqLug7ZFTaH6h=zv&|H*$O@8gyE0XYQ~mbw3DcEl ziTDL<3i^~FpRe}L)r|yQqk?>zp1I6E1kOz=qC$+36;DtJ_jRhppu~R#?Sk_)b<^sCgz}{co`IHO?{n zH%MCDH>Q4+%G^*vx}p(5VbyOu#8>@{F4RVKx%_z%X@OHmo`CUZBL!H~p2!aDFZ&kS z%!f$$QPia>ieUdlhA+7_kFXVaq{xb_h6Mx;#kV!9?K>($kyidniNUkAyAIo9xZBjl z8mRMv(V%f{{cs@WnumZ(UkyyNZ=>JmXLh-aH6x}3G}?p=m+r-5=k6yYAdPwjj&d~l z%X(TV5!MXV`3!FxjgF-v8`$pP`emO=7Y=4(RZD=?&FF{5TA|7!7f~!5JJe9DfT>2(pXkWFOYQs zTG}q9TK{&NXIw<5rSf6VHgqu+^ZZ70hpC?JDxDjY$Gj)89ucI%n5_|dv2r>+H1{L{ zQ+a*T`@*M-x9F?bCz9iMU(}OuK9Vx7?RojNoO;C!I-y?~m z)VpXY%FMyks=00Tb#w%c_HF(1V`ERI^a9QVMI^4aDcoK}%*lb=#4Z$~DHpenI@6ekm&D9h%( zAmxiXOTu&hZ=CjI{zLMaJWp<0_ z`ZG?-=LxMy6L-bo=Ly>`MoBLczC(uZS3OC#?G=#X!7GYu-!+Y~Ufv#%aKhVjg!nID zuYl>bX`SO3mmeMTACOWSnXhnic&OCp5M^gFAEesd6G=;EDIM*vD7sw7N5ri9DGV+Q zO-!HCn!8DS_z+4dj%BeiiRAIW^idf{?j^=vOGw2Phz&inb74Z#=!tq`;zNzeJY>d* zPJxgOT`EoK8m7uVvr~_o&rCl3E_IkNnSP;HDp5-MkiYG8KWRAoVyIPFDOc&&@8u*; z6k?-^`xA#<_O&ao4dbW2?=|2gz2ur|O3j$L&Pn`?9P=!Au*aY^RYQhDzXmlsnND(~ z4nKKN(PWcqKgd(u>3$c)V*ZfcR0Bvob!^VRM$6ElLiQnBa}ScPdac$}_~EbavW*lN zXMaw+$IV&c8VDGZn?6h2ROZbzHjW-k^xMpnZi>GfX1x~S4J5y=H|?IoZN>K9k0$cx zfPtC6&+8*MCfH$ONorCsK3SY2QBQQ-M-Dq%ePlQhGu>A|^-`a@mdNNI&%E5L6*afm zXi!v?8{7?+Og*9p`%;tGmC9=lfuTi{GnAshTljw1ZhBqo&T(SWPjjR?m<4BEKebKtJ92xV|b& z=+P4x<;qHwcg^2Mc7;|M=$~v0KFgl{%PDbShtwxFC`cJpO}_+9f;jU~j$6lai|I16 zee&VIa^INE>H^ekC35K3^`%#1B|Wkdm>iAXXII{@>h|0oA@x`6(M?qYH(z;}#6kR2 zmCjb|KA&Sk%t`n4iSnbpG3s2Wu2RURCEtH|tY9zHosQm44-S~y^0j#}(zwL^2 z%=1$pMv=<_uDXl3-GgO)enUT$>8(nmV?(03X;uxREhr$*ivF9B(bg7JQ=b4^dJLg& zc-uYV{bhdoR)?5!eT{;oV^dIB}nusHN#eJ z{gB>VUo=`I;24G5s*2tgH~afCxGKkR0ZUZt4C@#_m%9@02XG`_LQy(u zu)55zTn?0P^AJi1!PeBNk?i)V$Uz%gX<`>`s`tFFxGh0%fx4IU>w)czRj1l}DbI9G z*>d1}WYDgj222YNJSgQ)_LB?u03X&68lD81fpT0d-YVb`8Vq){#OBCJkI3Pq^3PlF z^?b2jUO64)n}B8jaw&s$`blb(hcQ8ZQ&Y^YsXNkXXVXgdIA8kTztLzhb>VOy?(ai~ zwO_=M3!0;(1`Rn++<;mC-Y>SHzfhL$WX-?!vTwQaGw>UwXY{Rcva~qeW4Aava6&sH zd7_WQZKzal0&~q_8$srp(D6o?m74KdnJ2SGg05XSEv1+f)9Om+^W9i11H$724rgh- zx@s$zcck15$+y(bct%{_6iF^iz2|NA)y?=S~@Sc9ODis%zPfMVT#5*pdB6dJ_*QAQd)7C~1m2hQ04(9yB)K zLR*~Q7plz$N>mN{x%zdsFjd`l9?5Gm1;eT188%+-hYe z{SFzP6CkM7%F!@!E#+-iT*S%~AUq2S_6`yJnKfcBtj^{aix-dcCYxZX=>D=K(|${N zh`*{rO#ithW--XjOcib@5MeRnf%<+#5{^ACCnUbRy$fTuY)y62?lM>%RV~Q{O2;3Fis#TY z-rjU@dLRx6y^xbm#f8P{w0y%d2s5=C3#KHaMJm#zV~|~vPqmyA6PT?Xmj&>{X5&MA zrW_N4*%JG}5DmUOzlB{0uJp}nzqqaohV5>*q!L`k9-hxWB`R*aJ>}^xo@mAVB&niC zmUR5yK)o6*oM~(-i2M<7_CPiiMWh}(!eA$9qM8O$tj}=IF?09Jg)HCm9B)i1pnvxJe$;RzB*AAbPKiayY0HDuG7cNnfWPm% zST1|*Tu{G@3Uze&b7z#LlD{i+G7nC^uONe69YV?6nfUPszbnzgAUFz>Cz>nR0!>=_ zTTSWHOpHX5%OkIs${@`pDD$!0(OxI&?{41N%DSEGd0iq2%-7gZu_bD{%!J%&*~(kF zuRwGSA8l?K;?A$c8=5-8UeHJ)d{!gw`mu(TP;Cyu_?}Q<_V=*e4bh&!TT#Bg^77tb z5{oWeR3J5+f1xk=$BUpw``c?UeE>+Fo&D3BU+K6HA;Bg2l(wr>+FqNGWvq!CrwE;J zh$XbR5G4`>1BX(wPwOTl3IxO_%E!u5OYAkxgS;02Po;Tb~ zq0|(4CMaqPLUh7|Wqj->sO%F2F4i%sE(m@E#ATqau=JAYFCB+T?=v?^#Mgk6HuDEs zV|#rPZ$j0wq-Vtu>6CdqB z?AYArM#a|rQ7Yk*4SL^QBuQ|%2&Rh0o@90;$r|BX$g583ywjbXmgqk4*Q%OSGBQu7 z3;k5-NmlqfH4zm>8##@nk8PtT@?+Sd&x?+m ztG}tYc5qUNmaa@yMrJ58qM4;+dD08Sm9Utd;eNfk_r0hy-dOjRlh~CE<8`uL?{XDc zLI1`C3CM<|X~l}*6EVX3hl-L{=|XHS+`KZwsvF`QCK;~>cQ(FUxv8ln^HjUd`XC;G z!+&*Ps23)tqhTy~@O%8ZFTne5@$*psnu zgmX^2v#x#gwqz}VfQ$y;?L#V&eYSN-lS_~3j`AE_;$$D9Q$?sO!p@UwUNOFpy-Zd6 zk^)ugb}{yq_D1x{;oB=?&4KtPyysd@8+&zzag_Ypua?IlTqVlFDzWs1AQt~lX&LrNi)EZ%%@!Vsw;RWGTPvd9V z)=S5?ei7FR$n#_4KCK%VbDKnO9}divoX-U8Bev4`@nkCkp~?GAgpY}bW!&tPW?VW* z@&{27BqZ(hfI{EK1BDUbu2)A+_sQjV)wZS}hKfSSC4|GU#Ct9%k9Na+aNaH{;O5E= ztZ+B)D6{Yw!1j<#Rj?JoAyb`< zh8Q2C+`ci0;`dtS&$I+fqQJYMa(&JS|25e&?T${uPft=U6F#M62hL+=C|WRM25#NGOL=tKYjTBSyxENeVIch-CmSrO`Lozg$_J z5=N~AV0Jg2;V9U?iNQux;{I;SAEOb$>naPm#@6-6s;zfGXtmL>D1gm3|A;#GUii1g zxJnqv!W z8_mDJyWTS0(Y+YhnsnNPu7yn9f-bg#_ynb4C_*!~jw=9ZBj~0O_zG^D@@G|Tkm-iA zvq^T8cPaC0s=O?Q8((=I6n85zqCc1O__Q_s%u0X4W)Jv`NXqjd(_|bnqDl5Gx*xGn^Ma^ zv#9I*6hp?{&ptl+O$?xD*}T`V&H~npi;El2iB9{6t<>azFsd;d1e3wcbnB-o=YZ#% z$|;M)p_Hm^+oi+23&>%$`ez6(2!15}W0q;OK2yO85=--PL`XAq-(TJZ;X@r?y6c_n zQe;-^k8UT>`C#5p?ycK%tXv!;vicRfofn8Y1n*t<8kt;i1S=!uTO4;7vBh?7;gwq$ zyr0#rPE96?TBhbDrjxt8Je8sajaSu)SUGA*-+!Qj**>*63Kj&L$VrJ|)su*{i|Mcb zEuC3cu)d6RsW7Ifsxmm5Rvn_#kkb6PWMZ_fU}9xOCgX7|m08q$YjhdE@l@3DVoe^t z`ZI*zmW^#L;OQMwM$Yo zR{BJrzSAeP0+i)_Q91J8eWlwq;?BcCn)7pgp&kq@H^Rz1y`nMbQjHriM`itHm6whT zj7J*J>wlBbpSTAkMAQdsa@$uo)* zX?5@J!TdQ%^U!CZY1ekgq#jDbWo5B0<9-?6Vx^MI?)TjkKP-yye!{Sb&(%btP&pqm z_kEzWLz%~FFX=S)o>94xvG1L|i7%;fd5|rQQAoboEVj(>`cD~%X==wC-L#oN=7@n8 ztIe@#{K7zU|E`N+CTo98PsI7K3YfrUPB&(iAwgI)2vv9-PJdKm>=3U z&nY;Y9-`Qse+9Xfcb!kl)el%RacPU&ef>h2&J$~zWAp8Xi&)I6uMJID!|;rl=u#fq zwQtz%3j{*Fr$q!bC9@)mw$dsl7-`fwnJ^yLhQUaR_Oa$4<~u5#bQ@%)#isRSS(nrk zrk@L4k_#Lxgv2aHs{&Yj2$F1=$^vRY^0q0Zu690&-Zio@K!B`z`(wYm6l!qz|(z3&n5)+FQlRF`^Y~i#w^TW)^m`hG;Lx%^)h8 zTBIaZikGyt*54)C&jCjhY1d$OA%GN@5RXnqz$rda*0d0!julTk zs_W2-GQd7*uTc2hvL*3tq7sKefJ4(P)JW@T?@ALXX-gM&b4ee@ip6u&nM7$Z`)u*s zUZ2_wh8yJ)^k1x~qWNOy+?Qfk7+IeN&b%)+urE83cL`82Hv1G;sQecnDvbN4mKeWW z6sfoL_1pTX$M&7oKuNcAx4}``>93(Ly9iNQ!sHksOj=x;ejo*y+g-^RXg|upxY|yE z`c9AbfWI)e3?QQpStFC&gKi(KQf{5OVbNYc&sSn8jbc~8d>r|L6|ugKX7Z8uJH7e^ z3FV+}jF~pRCXj>afQeESD^%`GP@Q>F`a0LNH)UoOh-0^LqdEoa1VkV9h=DcBXG??d zaQirC)G=#*GMC6JDFP^>HOcbW_%RWNvp<}fdc$>b`8UsR*(^1(Q2xX^hJ>K;0nk8c3Bt| zBYpZ+!DFuYnC`pek5ygvu|`F!(Re3WPW2COIi4nFBP_#sMGw6W6PLn`4@-`CRD9P2 zxYcJ*g=tqUX1?6*#e)c1~mH#B8S6FsdlR|{VIKEbT) zUdCl|-?hA4fYc^;h!JPf*Z`6y!BoyUvC@c!y{w9b@CvuK%A`kjhT`qym`Rwq(Dx0%|YAV7A!B8e8npKM9mWG4nyL^ z&QM{>vzO^&6y*p(I5YBV=^K?;H^%zNN-~rxr5BWw*%A^aRU{0pOzgLEA|CA&?3IZ) z!wB>N{ht1}1*_D|HSD##n+NoAU_z}FJ)^GQzT(B-qZ+YWDenBpNWCj=9bNQN9DVcWxQ`V41gVhxlAg z4APCsxKCC`^sDxiX%wg=-hmCv9n-3+3T_P772cB~IY|yB8mY@gRSd*20zTXF2d(1P zmGl$rniLkZFFD~kroRT=>fV-u>Hh4O^-VH?kXu-@hLGRJuh6bHW))Al~K zvQiJoukNWlWH?Rq#lPLf?EQ?(3cYsVmQeDE(&3`#CZeNm=Td=|J&5{aM(v6=Q;$Mv z@0;4&X*Sr~<7R7--cRfajxvK3p!)AUsKqR-q8LT9e2X2 zwhYuX@9PEIq`Dw24(Ie@67yKdT_EMz8ijHf90B*4vhylwzu3I+1WDhHEebcNW1!JH z2`;{W|3?%B820$rPWc0k@1qh2P0+t4%~%ok*|Tu;w715dDo0Nb5BA9$V=jo8XsI6j zpirH3ysRMRE17=Sy{(t;0biyIUrB{u?_Zr8ikF`PJ~#D-^=759Lf*S_fKqfOLt6d_ zhC`n`lS&kI%R|>L?pq9N6I~trAZYlxAaAG8zTgXdD%lswn)OUV>6KzRhl~AQht$4R z)3|nf=|}p^qjs1-s|t!ih2>dhaAvW)Di~u(GX<}=;Vdj3yc{c1m)3v-z?S&u)jx{1}thBct&ffk#3H)I1nlRkW*S#wJ z0wnaV?PHe3t(m$Joks9B97jd#;t$aJor9k4GarZcYTRPl=~1zy|02#yO#!ny$?1_u zbDR(vP3Ul_>@Av#@-{GiZwVMF15tD{UAG& zge##tZFi97rsaeBp(6=0uv(nlCQ%~n_`cPvf9LSqgZX_S`G(|jiGoh0P|V()LzY$; zj$}9YSw}Wi6;dP>dUjoM0fjb7Ci+MeknS5*hpToOlL$6As98MOkPNj46<=FDoAtbb`C$c7vy#KOh;4AE1AaWKo7W93*dJ5kTfP?AZdf#3Sp_S(l~-^pM3m~h<3l_b4rN&6SY;V==X1JdnB zS12FITkFtBTBZrpY<=D~e$3qAff-qn)n4ystA#2YmB3W9z$;wrDl1X?L}n8u(k`~Q z9947^&e&npByH;a-T^#EY#IbYDz!-EPU2uLllzBnB#TQ3`bQtuqbB&kE~P=eu8tgw zxCoAYc7@WMBmL?dK=(8%86EK^f^XEfq9A&$M5N?|=?1I3={BIFz|ntB29KMjlA-Jy z07Jye=lcl#7~*{54K9_vT0k=IZMAS}x zVc5K;zqhk7f3yO~RWd+{z>HS67RV#1S zq;IIMnT@tv=g6LP=+p{i^Eq~^)Fa6krOA)yuZYF~nNw%ADPl^76n;GIW2vzp;qNeR z`q(70A$&=Ce&a)NZeIT#gBDt2O)p-s09bfTd|!8rcrdsh(_2!Q?LqZ7gUH3PmdNsp zM_b0BOeTO2@#B6-RTlQ&FN0Z-DVe8}R8i*yKV0i1GH2OFl15|vbeEf`+rQLIm^KsG z(ok6*b~J5|?CrM|s`|q>DiQ5Fk-j}s`o9FH)ww7SxL~VE_D|(GXE*cF&|93;&fl#% zk%o$XR`E4|a>1E1UM8^=$sLL?Mm-4DdPv{lM^B7MWohuHXNfm=Zs1jvkCqs#3-fw} z;?eD$g_^v=&ZG=F^EAnc49a&^MCK89qJ>b#yPcSGBfNZZh355M>P>eS95&$15SO?v zdn<4El@N!Mi&O{2-|{@-DmC|s-HC(DdR(VHOiQtxhmV$C8FTKY6@4)nASm#nE^D1f z>Ht;P^Vue!tc?buJ(8W8Do{8cC%fCcCBo;Et(U`a4!}o;Ro85ygGZZ#_}rnKxtr_P zu-O2CQW%0VRM^~nAeD59ZfE!1h!(s{J_RIFa?)e*cbEb!4W}0(%SWXJ9NX7aQ;0ji zF>H`&z7iRvT$OVgYD1Z?nJ-Y`_per6DL@{*{pi<*psLH{;F*3-e$+VGD;vW;OpH^F zBKke?gXScSp~=+FoA(f5Sc^TVsdEd82vHJOxAsfOw{7{(j*{{xGQOSH-_%bJC2fX_ z(Nf;6Zf#S!w>Qkm?>GaLp#Gi@k}_0E#c3YRIEPaEvIoyEk*mKeWWFy_zEQc1ZqutQ zFeA?qhaM7z>*v$mLc#M~WZ}uW7eZ&^(49Ej>N<7TU1@N`FU9q>^d?U<$ZdkGL@|v7 zE-R*k^I-F+l9lX0MKclQ9;5iaC#rAPZOi80I}38%C=665b?a_9$x)(b5N(xKG|!0& zd_nQT3|To6&nfzF(Pk-%bcTiI6!mcFZ1jg=d@FkNg(X!_v>z9mBK>`fqQ^09L$#y^ zSdVI*b;C^&3VMotJ)Aj1otN`c6?A-4;>G?b4(cYl*3Qv@0;a-Z-Q*UZrtYM;!kLL@ zHmeAMAbkYYAWTmy$o>0V)Ufd{%<7>yHbeVT$Min)9cbF%|K9YziI9BqB<};$J*EPO zR0ROUYkOc+01)3|Gz1f5fiXIMSbbGB5;nnZvvP1YUhsf2Gu?oMifE{Olu{OAQ~ODY zwvc%_;G#m4=>4ojd0FTN+0p%`GR=VOqCWvd$UpXi@m|ZqNL~UQM@%%E@9H?vi!>=- zyJ5n?khr6xb^*Tm_cX_u-M?)h;K(P}ZSUrh2o6>e(P!f7VCU z$`y@2F*WUH3nDAyk7udV?RFq1Ze+?KfE4jr?7&K}?3$q`gac*IDj0$kjtikFmoXEj z9;;~_Rs^HeH(Gwcf-(7N8)B?C2B@l`k573PqHs6^l5)-$9wUr5P}D(7#*$s#1SVcO z2noAyC2c29sP`NFM9Rl1AX0R9eRN5R!-;E^l2_xn6IVZ0mBacVio&BeKMdD&ax)?a z9i$VY`l`U}=c6Vp*L8=mAQ zXy#vu{oFHCr;$yJcMOI1b`+DU5|~x>2kNii8JVs!_q{Cm@|SH@KybE++BrI4VAn&2 z#GBcFu+%41LXXn@=(DRQr3<5YDLNoyShI&SZ^X(I+FCm{EvHk?q#h=GWBm@lEcqRO zM)_g5r$hIK+yWBUD_not#NK&cygQ9nWVTZ7RV}j9Bfi=6CDf)5_RxYWrwf_Y^EhkF z7H`Wp=!&Iz#YarHj1M^{`HD~q z^MDF5D>vf?zGP5t@EuLqj(XwMyba+kdt)i8SdHBjZ zg=2xZcsYry7r8R~rUy_V$pB-h4L&`c>9`*u7fgP+bcW{98fdk9siYKme@g4iaoJ()OiK9+fg?bYz;&c{{j>77Sh)bFV z%3sVAK-CW5qEw`Sg$UI~_S79M(tc}nJXAf|zwrd=P6zqGeMG3@fPj-@_???gU zmok0~Co(y)HfZLus258V0_&R?rv@)~Utx;ToB-nSO6GjdVNy`=thLH*k9b^$l-P@*-h zHP9AY9F^<5Y9{~>DrjL%SIgCre?qsnLQ|3puuZ6E^fXp|K)JDk^#Le- zYoXmnN`@}4kb(metkkk7LV&tn;+UKpIP1-)hc30(d(m?ZccQm0i%LcTQ3r)mN99iu zWy=}2vTJ2|wIV%2if_I&{TdQ(N1>!Fb=Br$>T=9nKVW7o!#4=ht{1nBYQHHjNjiWm zY94ZevTFX89a5JCnk;&WK+tWvf>0tY>R3a?4Mxd`k4A?OtI@<{xZCC?hgacUaY@Fe zQ^q;HTb#k}Pg22^TC1t5nldBdZlcXV6_BQtIyZUn?a`mMs6!u4(YLQh{)>4`?!$DR zL{shDP9dXQ6|nU6Ny=@y_XfJU-~-o9zf>Mw31fh_dVz0vnIv3S9vTD;KPm~eeT)q= zyY&_>z0jjc=1*wAYcb|lt6eB~C+(be-TXvmmKY%%YBn~}Wi{F0$=e&^cGhTHA|6&wR?~C%eR6xl$h0KhTO?1)*N97YNbzdscNlax@WUsZD! z91g$DPHwk8c-k=Kp6WTHB#M@d4Ez8t`=Vd+<=XM(RfsTg4qk}4qehsIaSTqUCtb$q zlrEHfoy#J_xJv4>`_kO%GY1aC!EW!p=9Z)CtA}~&Oh1i!7Y%uggG9hiTIO-wTEbvx zgYp(acFI|krJZFEA*$weQ$1NK#StoP0oOD%a8)SK)w-qg4$4DKt9h+m8Dh_(t~h|6 z1bxjuZlJX-Iadmx=6P)&ITLA3Nj|B8iVYO-iLJPDG;Dn)_wL1QP_S>%{_fsTvp1Zt z^#zfoKQ83BJ`;H~y3V+f1vLCw-tI!03U)RlKLxQ5E|&Uh=@hz@eghIk`q)Plj1aC- zYu-4)$Yz{8C&PB9o`S*S$cHZZ#9{7Y@7^A1QcsaN_F8Oar1n;)iBCAccFwP2^ygq7 z6r~{MV zLKg}2z#*_5@i`7N!k4{0#SD~AqCO~Wbq0ihuTL)h$N%AEq+|$L6GS}J+ab_RlPR{R zG+2Bh+aY(Y5wU54gX>LQdc$nZU2P6Bf6ztvee`U}1WSqe`0%_xJllQY^*#Nb`W znB2&74>a*>G+#cOZeT)!aO-kKmlNy9&&38Wq-<8D3{B&KY}yl=is(GHJ9pa>Pfmw& zzZc7=GUS@VF~+2XXS2suAhV8f?6UC-{avHyo74?8o`|wVSx0q8BF-jh!R?z@f1+}3 zcN{RO`r}t93?{4*O{d#@VBNacS4o8+~FveOjzKI@~E)0av zGcInJ1kUfqihX+i9klw_);?HK2&JZRNgXf7_F3=UFUl!q$%p$lH@Pl{pv}NjZfako zaA_#6|9K9AvI~iXk6)(mFNVN)NTwYYAV!vrCY-CJ*}%l7iK5`HZZG)MdF>ZYf(lEJ zmQO~?G8`JcrblhiPs?kv+xPBgBc^DmO=g)wz!3z$uOGn~mG&k-l1CNDcURLlK+x|& z{E3-d)-6Qi%sOA%aYmEc9M?~S&L!NRso#{R6q0utjc1rw``vqUv@)#^$dg}{U-j0U zDz$Wb1@lQ@I;a2L?hQH`H_urvm}M>xZIo)$?03(5EoD|R)S(N9xE!ys(n~bj;2D*} zjoDnd(va6H{qD++h|teYLnbl5ldN)>$K{>+bS;F5sR}qYu&TZ#eB*feEH4y!8=8TR z-!>A$7om0iwJ;(~sBF$zeA;(>F z2@7M6!8}AeLT)T?1Q-cDBXYTJtA?zkY<6}`B?_39m^sHIKpYJ2FA^s?bn?4+&>gdx zyj?V*uLklpX2q1w<4L7>j4Qp*04po!LPU;{v{u52W3k;eV;Qp=r?wVTg3$1bJUx_|(~z^#)Y7#Q4C*^cPFwf-h_|Q#YnsLHVv~ga zgD|P#tM|Y*avwKrrnR_Q{A%Tv2{P>&=CMeokJ}) zcQL?e%##b6B>~t2?q@L-HUyG&=B9|(#TGxTpJRzzar@dA&XhI!j?n&=!JRkQ9Uu znntT&TjVyI}}bzm2rPrYpJB{dKjD7I_7Z`pXQs$^rtDT57J{tEPWg| zwVO91-<;uik=NEsi!6pxD|O4$Vn||ZYyQXiltIE|qp$2fl1#$d-+tN3qM~jr74WqM zbLL9VguhXW%}K3DhJf?W^7YR?iLiRHAvwioapYV?+Hz6Ip<09)5Gj^wflj#J);M}e z+q2-kO3n7-)D*NFQUgD8D@&DxH9DW=)zuvTRXgHoi}x`Egn8$_$N9B2`= zrsgt~l7awkhU=u<=s>;Jkzk)`wfEW)WOzA5XQcR$dDHcGi3aK1J4(W#Eb(;81oA`PD6W}Y6MI!2!Yvf5*J{q5k1 zKoMbgDZ*AMFs*yjW)^1fiw6l|p(xj;X>T?%Ty5QG?G&giUj9U|Jh3^2OL8=EBp1B? z;9p`Cw>p(eav84=AP;S-O`Wux_wf(Et4pB}sibN|#4@ZfnBfL{h)6taN5Y^QivO7b z*DeKt-I5OU`#Xj1{X@e_(*`*%ma_kON0O3o!g(ORJ3uc|9)hY@Jh*L{twZ&LVSO0* z8PU`~OL1M+=^|RLL#aK+#BHy%@wH7WoK5<6USukq{3BI|LpbeKOqkJsWSfzI)g&@4 zUmm);k>tJ6#qzFUoa0bJ53`->)*S-gdZu&qLImLTTs7ti<#^_z_b?$>O`8jWw+J$JNsv z+V`DW_!_MiWR_Igye>BM_NK6utV<|&YtCJY@*Ef`&Yrl}hUmnaz6WDO9^7YFBGJ?M*xte!Zpgtjyn&$)b!2k}m17+_Dt0V`|ADLV2LZo~8aC2%5HC zWlGhcolx4<=Lljh>z5@yvl||(=Xi^+d84^+D>>GGZ0FzhU6PJwR=;8ij-klp69YT+ zw36cEJ)ybtkk%XIT>RVNqIW!c1GgnmY6;qm=GVF`oqB5r;y#y^Ofl4P#x(85Ag7e) z(P21}GdWT5ROW+qyH(m|@APKt;|i@gF55ho%tx=Z#7A&CDdQ? z2S6VOUQ4Dd7Oz)5z>PP>meS0m{a{lu%#>0OEhgKF2msVi>GR~Rgkon3nc7SJ$Pr}G z`}91wNsfI6%{;psfA$a>lcLD^-);UvKeDAeEo_PNe<)&<1r)Q|&q5lz2PvYKYUZ>~ zZiq&~5xF=uQV%2LUbA6(xM$&spG_Q)v%g={e>9zq6M6U#_ZlZ@UH4cYN1_N`RvuQ^ zh(FKH;sgVuLp~wRpVFe?7wiz4it}}tysV-(?TbQrQVCJnOrQG>5*B#~ZkEmmy|fuB zXf4~QdQizz+Fn@EU6WGrb?qg?TN>v0SRnx4=Fqg&)ZZd#NDhhMqH$hW7VXmnsYvj} zhR(d>Iy!IKT@gcx+AE@VsED~r&X+umK`?$FZwY)hjd8nVn}Mp2b8pr3RmSy$t}?W; z=KyiHc`}E{ltf&KF1bmN^?k>6H}vk3W?7F4iTjx}ybhV|oy*x<7`R4$LX4*^r#*)x zbys?l7xkEHnLWz!5H7zS6P>+ON{^VDyVg41F;)OqJ%r1J4H5CeVpjN+`tU6*abfl^ z7hJk>D1QqeM!?{TG2X6!8-|`?frBE$z0t<|Z24#*v2QUAb#Hj0q-5bDOvAB8>7x$r z*0BcUlF?9mExmqvU5hP?_wMEr>kt;=(UTv~EQUo7Hp$=R5Q20)TC#GGA`lKOISMS@ z%%GM9Y$q~wB5}M6tn*j)jIrw9Z2PVJ-qnM6kF%cZ7{u0&M7t>QRp44Nrik1HsKP*g zwb;D_BTpE#7KL`&KI?Gi^7g8qE>O%UbNZ~P*dHJx3pzxLe z)Im05cvqLRelcYK`=}{l&T+b#VhTq<>FqJph z$L)Re5wQ2x^9zcRc*#>{gdyp**yhN{SA-$EmxMwCO(b61@1S)lFIj%Yu2x5cM78st z6WqUZ_-!6*knntTy8#SF%zHI=P9J2x`zo1UM7Ql!wU42r!GwRg6*fKNdj(+*dcQ{} z-3?cF3z5;1!(%ue8{4zQks>ZTC95D)g-ff%TDE9bJO64&Yfja5yd?|er#$lcsQnv&~*cJw2Q3%^JZ*G(%pttw*$MCcMrfKo& zbMkrgsA@2#eje8vzD@~xQLk(nLAW1kSTEpE?>?l=DNBJ-3Pr}ZU0NdZR5JuLLULnM z{yY85!hYPP)ruCfIR@#A21}1r=(Zg~mGUre5wevd^6cqHtOGmvF2*lXjjl+$$QGz? zjHZA=uY~yb!CC{o*De|#ks}s?1m6A;0F#I(P80BgQ$-n{>f|8@KbRyORa;*lW>Sv$ zs(t4g&;m$Lu{5l*QS>Q#3ZPV*GsS5BVVP%IJhdOcI<=UB5gt< zdg%J0=Jp^*T#=t@cgV)Pb=#P4uU@LDVVq;^S=Lj&_+;sKaO%blayJN!Hr}|R8|nygVSGOQZAFWG6$~??Bg;1=y7g9F%K9i3rAi5Lc;H34B~l``cTl zwqKQ`Fs$0FLN9rVl9<0GN9(F#F1pCVD0{Ug-7Acbn{-MviHKUk)0g0Rg^OYhR@L6+ zay`@Lyf<+4a)$)c^HUTC=~d-OYajqeK)Ampr`KU@btcubMYQSy@*^pQ;lw4Rl)w@L zW6TLV?Q$13sFeoO)^G9`FeaX+81YTegdj6S@vy;<6hyoj_l3_F7sMTJ@~o&=7Tz~h zk;?8WY``uJJ_^Cy$|dKFeRUODWU(VCyhVcS;uZcr4!2aph8x<7Sa6mE`5=dQmfh`UmH-9ECMJRu zG+GN{?Zxu#Ye4I1dF0LVqH?ZsUriK;uh%*>!FD#5S44t04Wd*89)%YbA!H_I*N@$0 zTn#77&r+eT897-TVDD@0uoy8^&CAz5jj_vsM&m0YVk#_kDwKy+4%S;$!-@WYkVd*{ zoRgxlR$r)G8Zv%y&!^U|Kj)ys6!zrC7>5&Y>xim04za5G6#sk92aH7Jz`f_3_br#| zaH&&7N|tGPJP)``3#Y8&lZm*ibIlATC!0k=#`sV-(6SDKoC(*ik?!fY&v_sztwcJU zZtz*b+U!X6i|ACXFCN$$l#SemeJ0q7OUODx;}68RngbM~d-$A@|gB5dMBZP?9DW%ykW3oKQ2AjX)_upWs6!eKLL z(=>Fl$D6fUgC1#$2u@uwQEG5<%CX5;XVen;VR?RqYaHa3T$b}VaL!hAJ;h_KgIIh` zQVwx*@YW8PfmqrN2io(XoZzlOPT=P9i;1%|;@z-g+N4a<7i`u29Y823N%4N-hdfXK zIvUQdG#^eOq%YrmA4kYC?SY@R8quAl#qFs)*ld5Cz-B+qPzykd_HP&@yQ;h;YwvJF zn4xWjy86*ZPvPq)afou5lVmAQnWxs1tc3jG){kpoBgqe2`<>W*C|A|6E;G;XI|R+w zZ$_oDxmp$<@GT)iMK}$mva73g=^A8&+e9|knp%+DHy)0DSEox%r$pfq9PH!Jr-E6n z7fWfQpb^S%G<|J_57FpN4oaiT5R6Au;BkraPPt~FA3-NuMssubI8H$r_-blv7IuW6 zJ@wPG&!I>s{n14oW}50pl{G}j&qwm#cWJEH5U?P_rrkXT?-||m{e8|AGiD+G(m?a6 zd^5iUWGqt&o*F_y=NGd86)g0Lb-pDLNz2i%iwX{-SagoKYCDeEZ*{`Of7LdF0N}0Q zYi?%=jq#fYM|^^yd`w}LJDyLk%tVgeC&+E!D*h%lI{DF3R3hQ#x^}WFPbLuj0j{F5 zA&4$t5VU4rkf{=rY`G5%-9rApL~Bt z3#{9%vpwRTG~6z!hH^VkyOnv3BjRyWM%XG%-$C2L@AQ+QhxfZ;n~L*7-6tSD+Px~< zvOwoU*xT)9yz5ENr@4$E7H+6)`&7dGXl%s(9=QTBd_+S^G0ks_UYj2bpvppkGznT1 zEERKkre=@pS4G?f{O{D-CyQ=KM;JB^pwX4L;3ul_f%I;8=owx0bnLM#kV~@U+o;b6 zC|*T}PNH@l>b6c-Ki}u1rTf~{*@Jl&t{T1+jPW|@f0v4Mvz{l@AQAR7oie5954nTO z&|!CG9H)~J4;c}0A=l)?%#62U|5vhj*GDLrCOj9|)eSz-xU=_kLhBMZ4fmkPv4)Z7 z9VINxnVu=|_eS?{@0#SrWMT*t1YDtLgnM-m2so-(;N2fzhM+7^){C**u;xey0nH{!f7Js~5 zL))~xXC*;&YKEc2VXBfTPrO@BoVYB9wA6?il=U!p*v{9Csb+M(^N$Oq3xeC8!>$WW z6_X4F74$M6H&pRjN%<9Dg_X zXlx*I>hsguTB(eL7T@URJx@raHBy+d|8YwiDOn?GsH)j;uvq+ za*;uF%;A>`-`+|9M_z6VttW*TT11jfhn8ldpPYkeMiX6Bv94?+vEx>}`A0>9Su1o= z{#?x2yxOl-OTfjY$w+pK>Dayy`i^*IFPqBsK0LrI$-QtnvhG?Yi6|KHHD!-1 zXY!G8fn9D2j!M?$9h0{Q+->dB`%LD{45GiT`jSU zmtVofW6CX_g-wHB(h<>Q!SY+v)1Ccmxm%`x*CIGow|<=zoeGCLw^yqVv=VN*n~)GS zZ6lib4euoW=2Nf6qp#j|ke9(iRT)>dO~Yw74w85~;9`khgpB@ z4OFTF*WB%{`D6G1XrsG{NAxa^Z0Rrw7W350v;MI&t(~+AsRwOkkQBO8*y=zxzn`tU z!_D+L_xN|4CK%#MZ+YMrj6Lg+FvN0=JBq?I9m-V`l?)o323cGx+?8r!KMl2 z2i*h2hOom1NyPQ0)ZCap8H1EjQU+wnph(8rJ}VS%I#WKMrvv3@NU{F_`DUEUs` z0wWs_^`kF2NjOSg5Lp_lW>iamP$<&&#~`_`(V~V-N zYKJlDOFFC?I_n8g;lZD-?co|sug-O+4@1aURfYCKP`-kNV~=LgDQW=X7g4GM~LvV2z&#ySMpL)cpNO$&_e<*hUnA4lIUqa6UF=RXmY$h0t1EY)4*9$H7~{Bq!Pe0@=FfD<<@4401JcHFP;(j%>mH6?JOdHZWZ19u}mZsPQ4=(6{ z?t1i|SLtRx%d=?5|1p}(6QL1|tf0g2ggrXMt3UJHcYRI0y=_UGgA?3Io!4Smvr*z9 zht1*bZbl1N#?+f$79|bEX{2zayQ^=XBAkY%T(f=!YsspKHw+aE`Sm6p*Cw?~H_V@} zc7SVqJy55d>7B_5l+!gT5V&ycLRboe!wviJ%c;pHc+--?#bcz2RD6l)a?LXoiac(d zMZ1ga)5{!6_MNpgmasef)-((A@7PA8GqkMk4vHXQjvv&v+zWE;7;NwQ&^>X_oY2#Iw$fDiZw=z-uQC*X6)0yMOGT%X=4Brwe z)RfAYjZ`|mJ0+vXouC=>lqq2gY~ow^BIDYLa-Cmwc!Hb@(aRhQ*jDtSD6US%uAIvK z_id~wEM{tbxJ2!`(rr>@s7G_k*vOdUM|K+u;Yj~!%%g>0$?xxrBpku2O&lL&;Vk;r zHAz9QXY=fXcBePUT0>DoLa8pHzH)`9RQlp2AKnAa?pXX6$IRdH59lvEmV#b!@ zyNAe}#a(nFA(NV;b2mW=!3)fMiv%+LK492KND4xiCzR^=Nk{)nC3aj(p_=8fz@6s@ z0{+Fb!(uuE#%}|~RNdy#xsGUlV7>}mB85Ca3wt6_&Rg(wr}BYQ8@WDUdYawFQ_=SY>uV^FI*avMxZ%iIb4LtL&} zX82lZ8HS!Sh&`nlxbdVP3!>Xz3!)^IvZf~AP*k=q#xpkbeb1h`nmcaDb}5TkO)3n&19;5)SCILyVnoE8DlGEt_`&+w25<% z+Rp$UN=24_cQ1+-b?TM1e3DU0E9^p(EQa^TC1$Vnt?fl>L%NO-n-l?hHJRt0j`sY5 z(!^4d-!B`vCF9i%$_=b@?c)?t4-5Hu=gswY`*K+u(@|cEhKu3h{IAFo;m`)eOAJz zGI-+wj6&l4V!wkfp4Qa~#kdr!IYu5W&7&HnJBhlA5g8XZN%VZn;~RvJ!vftMG=YCh zSc_TN1jJ6g!*&Z*LY1;kJw<9=EWC~^5pZ0g1;tU?UU}25H2$LCQ&*#`O`C=eWJ!*7 zTqc=6zQxziJ*ytXlnE1|6PYegEh^qtBx@918_m5c2X zwcZ+p5k*DKg*7IkwWgei0xn(#{&4f+9lX9#h zxQhDe!)e;AKB&M78&4SLZU*&|AV{50@8vo)#@4kmD$#icqVIQHN|>k6jhW}bQiwcQ zFXA#c)PzP6L7=J59``9cJbE6++w%#~-fzE_>4_X-V(O{vDs$<9JgmMTg1Nd}qmDEp zZ)_db`cz#o?0jiCN(~wf7p92PG6`;R#+REpd)?SZ_tr{wM6B_12umT@clq`#OZE0g zqWSiA82b@(73asVY)2paB8xwA_9HEXoh1jQ&}j?eEe9$twk89-m4T|(h8P)Ra(1hP*A>IprZC7U5I zkWYDlSI&y4vErdc`7#?;K|zsP#hh z6Ba?eZ;PjA;oX3n9BIc!_aDgo5@IE03l(V(-us#8L-V5eHIXfgni4JGdgAi^VnM2=Xaxo2@0q)k@$PT%5Zc_ zckldHk>t^N_Zc(fTg|CgX^C_FWZ(9(=(*8gA+@y}K-Mipj!t!0@OrCNdJot7xG^_$ znvEy;7^EH0jbxlJ-e0-wn-<6l8;ES99^A_qSC(c2wKZ|?CUEXiP+FNuU!>YQwNLwP zu)c;;NU8`e@|oe=Z?d30?af5?LAE5NGZB9@5$%@&qB8Go(;Aam>}z?^?jzU-g3(K6 zaA5d#wQh%{g1QER$RutLU(SJ~7W+J^CjRIwnw`67zMxvNbDRX3TsCBpK8b0U3}wLO&nr-I@e`7*ovhGbW*g{TB8uV~b5e3! z=Bj^RMHlPf2Fn#c%%LoooJ+5qu^Pyvg;7;*`aF=RlCQ(X>|y$_y|%yK~(B z>ucPGe0>)<6$Wmm3Xg}Asu&xG_u*C{Un z_U!wxSeqy+Ifr;F4R6CMT#@iQW~3#m_XCAeqBvsh%$U+v9fjg&qY-e{TZdAMk7cCx z1UJfYNmBTd3>ic=na3m24ZmnI)b6Dq>DUM(P1zg4;B+r8`p~<}kL?~oEz1%Wi?|4_ zKTC$<)#Ll7og2l=CTMzCC1!J6cno@BxgokV0WIe>OH<4y%hRBtTRW3{zCP{jF17v9 zuc)thr`@?hyBJj`UZ!|Ly_%BPA+agtOs1x-of(Z1>1U_)OHA9!z7)stK0KC}LCr-% zA4WO|&ox)B;UwfSDQAB zzbmR%aah%>8U#C6H}zbLj3f^%gH_XFg|@#N=|zPk9@?7yl(v(f!e|enOoRcmV=>6w zvAiy6Iuj(}LL6*-8(lxr>fZfY6>cx)@I?=|nuBUTsau+d9@Ie^lH=a9S9s-jxohYn z3JkZT?^TL@9=~75V+p&y+0RVKnSk835c-s-{mR`@4{o5PkBxN82bhn=+TTyOAV%Dq z9pBYTw6*DJEh9NTyx+*faTq>Kgq8_uOZviBS#ukCO>fr8^DV!iBBmgj^JLVeYi;bo zMXGL_75-gT0vD~-AJd**YyQkv^4ktCRiDU=nesBQX)1xcr8eFuCkGinU*x8+%cibq zn~%0(l0IJ21i-}Uwa{)Qs=wDj+ui%2Kq$sL6FL`sk(9xvxuZzeQ|I8AS;B@D2xXDB z#BxYY`L_Iu7<-Jvj>0&cpes6vYxFo+AD63Yo@$PpCf}p3bHmr5cPP_UIMEV3*M7Ky zn~w#Jih~dvy`1N@lXb|w>1%i~O%2Q?7`L*2K@|m7_YIiKdC1Q)P#(r?cO+ zPqixqAk)glKr?f4wH&ej_-!5=D0CxJ4V%1F8IS6zNqJ1F4qW{HRx>WnbBvp{_Bqr3 zh(xA{2(2H9)G9Al=lSuhZ_~5rA;x9Obh9qvd^J0!M<(+_&N3}%USsk4FI!R)!NS)K z#8TCB6u?UDH(}DcrGX7)MIDA*ZdQ!Hf{nIFPu`4b*@-$P?7(e$^Ld6p=t__<%w?Q- z^6{&k^@dGTM~l3tV*5w3kb`+Ya9C6!1xsxR88Vfy!QaY~x6@p_>}00yb4_r*)ePnK_m8 zujWe};CR_Od9-QEscA9F>?-6agxsV3m5D6{y^%(ia+B$IYd@N=+s_#o#Xg`Hk1iXz z+r|8|EJAj@V1W8wuslR_QM`n9@&N-@UzWon@!H}zvGP;L1jLe;ML0YGM}t>I781D0gr^KQoPQ-+nf%|Moax&U~Qr^bW^bT#j^}qht3so-lI1r;ch#+PY-lQ$Zp~`% zawuZ}bqf4UVd%E6^uesRpY7&x)oPOFMBl?;a>a4F0;AL!G_^Q=S~v}d1K!&%dURS7 zYH9m<2`~J<>E;(wCEvUA43b!Bde|}oqFTm8p(mn3-4&H~p2ntVrS7a^oy&$)mDI>o zMqtC}9Ug>JlAKqO)*_qR^<>K+J^0$>d~($+M~@G}vr~XfQNifLOAD1pZ(+bKaXDQ) zc9}mqM$?_VNgj?~jBF+hQeKf_VrCtumAjEWcP3n~8FC=Tn5YNRAIFed?-(W1u73Qf zN&1+Y^27VEYXZ#^_j0S&@3Qr>d22}5(;wGb$Y7>`+XlCf@_iEVa?2=72h<{(IWjxQ zwIi5t%3v$yhM*kcx0Os4O9(&q_|MS3c~g@@WN1bN7YO0I&Q7?Z*SGkX&{ z03qy^Y1?=rAcr+iWGdv205hLr}is`bsOcnG?IHsVuSQPMU2Ab zt&9yrh&|#3d_!)L!uaQx`Jy%1Kc?_hKW#2-)tDiP%}>11qH6=_5Pv8aH%B*x#&+DA zO0;OB=j`nS@t{h_HGC!r*klB%*0JXJJD;&ffBw-LmNHzw#jbb;I1arlxjTv-<+N^F zk>)Y6fk&zoG67vgq`YtMXxf>Es_pZ!-k5mgb%i~mxm*votpu-=INuIb)O=t+f6we~ zgN)-hq>${n{WzdtCTbe}!uy_9YZ9n5cCDjUkfx$ONf|8OB`az~KTCrYj#)L{=qYSw0*5i9 zUbmqqL+i4F_1E;|RORrU)T>fu9mF{a=ZgK#AKI+9+#8D#h>u9}9BWX;aqyffXiH6>nB zP=u%isesSTkJ27b`S3_U_t56uYYeV zSB5P$3TpN~^#a7kvPl%BzYhO;v&+tBQ#@{vS$W2h=pU^(Nz1iD*_)LZ>i3-_>e>5tp7%r}%7QwnD-MVGZ%<0m@FG51nZa}?U51Mz0 zoj36KR=-gJB^oGK5+A3*ok+BRTCpjr=&9z7)Z6#wN5#&?Jf8-JP>;RDWy_@;)n^kC zr+7zRf}d`dLsyq5u@PdyUKzLB_2r>`1t79=irpd~?jsvjKGMZ&sSkS=UE6WMIWuXt zgmEF!8{LGM#n92DX?J{Z4RWL+eqY`y3p$@o@Z;WoBHD_%_Q&G8DLoZ`H#@4ov!Mx! zF6_t9-;1e3OFhc1Qr^EgvbJi_D&eT>FhBP}hdDQM#$?1fcfvfUrGQ<&v}I;X(j$N8 zX7A-p{XVUTB2%a&d(8S8pt+L6wWj=xFg|{ZDH9QWjs!gy@OD6!rvz-vZ{rW&UM7WI zZ+d$tu*00lTlAn*`fPtGhG-D27fuV~AI%sduV;?H>cr2qXdyO}-xLh49K=wP#qHRd zs$m0)?5Jb$8RO!_+i2AckTJz*39^7VQnHTJyzs$~Q6mX!V{HgC%Cu)+nfOp&(Kx}mub?&R zpd`hjc|5ldQz7-U?ue>=pr5P7%)>k4E+-KIoop?p@`U6o87IBP%fY6ng-07w#ikiW zEPisL=Bd1+V}a4*=TatfVqK#euX1+9>dbv-{V5v1T0~gma`hrOcYlc@F00xo!%*Z_ zED_$h=j(oMjUeey#gHFa{Qnwyv#AIB)cx8v4!Rv zVgVUp4##vzMyE{JBcMfZ0P8!|c2% zT?=(n`XOwv+Iogd6U{19;JUBJihB~8LfM!3uU44f8P6lhXd#3+cF3t!1tAON+OajJ z$NsERC<75ZZp>;eoIEEhhVx-!HRrQx046>)9ymTvMg{Hg5 zMAA~(kv+_SlvpHxeEEyY)OfI4y5k^AqD1^KB?g%}^%Q|IV#do4s9a8NL)^PNHcBl#dQp@UTj#j;;9^8r+huM36 z29CUDamA|}d&Uvi$68EAduD64o!%C!(oU@U#u6Yx>_Hhks_dFC#5{x>$^5>#OR8O^j2jB+QPzx+l?ZQ{&={&%!Ynbc!iSDtYIs>-q2a zEN5a)FWNz=W*LXpZbD(NFn3G%=2;CLs7b5h^BlxgEhS{^mE~b@=E`tp^Y&}4I|$MR zD1JvWR9$p^Q;~%voL}pSKBG%@y>F~I_LpcTlr9rp2~Qua@Z1B6Y^TU+&-S>sb$(w9 z|2Hh|+!~3<4CAOvjVP0xBVF+7lS=lnA6WzSnHy<_~noYpve9}h};Y*8S(DN0Z1 zRAH|rS;T6P1Y(!x1`^;00`I-z^-xPcNXUk8G32+BDyb?#a@iGX2hFhDYp?8hMm=4f zt-k&g$I-jqNA%wdhEJEls6Ls4aD3u#msgmrXr{vfWg0Z;7Ycqaq^QIfreZshM$rl9 zd%xoB+?b)CmBs*>`r4TGQ=3PaSy-NOmlsh#{frRC@;>~?MJU6`i*@^;9XEb>$Hl#h z1Y?Mc2beQItUcx;V=Yugg8oo70tq2AiXt#mxKC&`XjkMvdQmE3a#6*qWvFH0fuOIR zR;myCV`7q?+`4Y4qNMpxll)WLQmSEsc_)6TstnZJqz~UeI9!sC;3mE{!gen@562{G zEw}u$nqdnC2-+;ub!M39LU9%HpOBF6Jd@><+Y;&CM^y>olPp_v1wI$f=HFXGReo+$>QqprF0hxlKa^%Fz|k%yCl zgz=P7l~!f9%!v~-!bcW5da`BYRE1d@=^?% zrxV&5ClTyL+ZcULNbKPo#rts^0EWJ{4zkK+wBl?x6`(orl)d&ggW4uD^T}hHs+EFt zm(&6teCQ_!eDLD#VA9Wzk<+N^4JAxMpmSU}6hyfs1ML2EA4(I*Wr)nXVe{oo`JMa> zt-@!&>HGCEjhtaV&_WPdjEVNB_yyG2Y3mVN?+H3xv6osiOAa=~&_#UKuFVba^SHKb zq>TrGuhUyEDR~o$zAG%J-rD&U(20xhTBfO{g}$)RLJV%|)8AcuyM=~H*|GK!qo<2a zdCE_d!cOjo1YfW+(YNhkfdGmebZT0p$eTIM0yWExP+`X zlU0*Myd%mv6HgXq#G!umg(&Sy*;TmN?nY_8A2WaLB5BAdHIgixI^J01&t&fxM?6Q! zQ*3)MGw3CTnX1bZN`_#&we)Lc-(r8Oy z)UuUUgka%wwpG)-PKT`dXUHi={O;Aef4&%SjAt|9q6yWl?pVg~vX6~6pDv(NOHq|I5C+W-p@vGbvkQjh};b)Smqm<6Shf#RaL z@-Go{(Ocgk9uLerIh%%+5nOH8AN49Vr#|o$6TVgOQP?qscX2Q8glq)oAMzspYVxxC zC3dL61}F}bJj&CTs=IV}xe)fP8zozIk*g>TlVdrZoWoLfYGA&BS3l9TKYc_maNi;OUQwS{OU*2GA*GO1{a zBIW_4=vMpwrNq3Ll==Nod(JUfXnJJ9lfB%|C=0HMw*rTY=@B?XKB*}hcQ;!;lJ*qI z=?}+1tGjHXdaheo!$rbqjZ!Uds8ixLVy3&rug4G!-T~wnrhu`k0 ziHRQfmm|pEjK*Y^4V&2R38;gS(I2Zw(J0G=6I@u|({H?jjKYek0!Ad*+Uym=WgW_g z@w002<~Y*XL|@ib&-0P>x*;KXppfzAWxC^7mENl zMofK*5Ld&u6wcFCUlHy?Z)-?&>ZA#!fs>vpvh&^1y$F+zH)I@=JEg|4yU<|w%-|+- zyR}W^9)(HB=i&8qB1*~dz$al}`AX-^tnq0q6Y^a>Bm6)OZY3A)nBm<|DLBbi{zg}x zEw%vX?wN+|XmT6x4is@gTnTrT<>T$us97LUo#wpd zK4rWXRf`2-4OR{%|NW4BkEjBt^qo{-GIh=P)QAV^SE3qb)4rZ2b@GtA;lznaO#Fue*eRy&ATpn~|9KGd}Es&vTA} z!L2jBie(D$*%e2It|n$a^zgnJB2ijh2Un-d3}j_PyfgB+ zw^5i^-=H=Ejax6S0yVIkTgzag**KnCnTD108#M~o=weT77{Z_%Yj_p~Y&M$`@f$Sf z3bwh{e`?0yw59aZTpX5_Wi&e7)L09h*hs{2zjbn6cwKq%Ax9-3e|j%ybnFCspH_WL ze=R|_uR>R$Q4^;e**B+PKZ-2~a*Ix^^u>I+#CR2jQQQtIOp+?qXq9#eQbZwi_bPZe z=pg72mFSMY6L&g2+cZXDB20JtSAi!V2O?uJxj7$NhA|>dcCxvY z`Lx1V|0@MXK{;}QeXVKy=Fq;L)AZk2HRQ$?lG1VvoZC;<)Dz=;u5ek!sP3cyWKZ%< z_)@|w#$)n5KqN*dOyvQ)+%q|}AOm;$`16tu3&&BFoVH3pJ=|P45BBvx+9B=idPlD- z7K^{B*bWk?@fu-}O;{XqQzKO8KkT%hJMRS4KRuJW54t5?bwj1m;VXLz>=x$}ROWw_ zFUytXv%k7sgsq@>&R1^1O&M>Cwj%ntQos`%;&wi)_3Xl&(r}uwO)i?~y`Q!1zkT5g zvB_w0#Tis<3byq$v~WY5>{eA@4Fl@b_~hb zWbevQ7-82rolclX;L0}%dEy70P5JDQK(G1O4&E2d_Mx$b8$^3dj^Cz<234;JMJ&>w zz0UYiRi~th%KA$XkYzs^B^uSAVTMuX-UYK4H}T-nb_(-?F0tA;NQ!|a0srw0`s*DOpP&@bdhNjkb8MI1MVc*NArKegV$YFR6g3iheUK69XQAaI zGb4=X-JtSAbYunxixdvo8iO-LNQikj6TY^iQ+`6BR0iGTRTeqDn1@LsiA6GjeYxcd z5V5oQ`Vi)xZ38I`EOl)zyo%f7a`3*}5Xp#$(l2|r*CyAy#=R2?WNvd%_UKnmooxv5 z_0-$0nTUah$heza^y)5y!sg$zQ9iR8{u8aY*(z7k@JD~rd1`qTJk zVI&mQxva}8gD7~(wM}OsvgDpZWhiajOV3C%N@Wuwb^JkjNw@LF>!fBFXU-xTF4m|hbNQY@m-3mXaz1BE9z!9N){+s# zfqx!)vGDJ>=?j)n*c!QEOXj}qj6JyrIxXU|LEi{IvyBOs6QygCqNdSqX>)SYzXO42 zs1RB~dBeNg&qS}J3u#9B@0@!^BF8$pN{GE@E-1%_lGcfW!@2ZZ%-<&`N1u5rJ&bkY z0fC4(Q(K6M5DNiqE1rs>i9>yx2~ z>JMcCIWxDF>wuFJ8n0XO!=#jI<&m0Q%UtjF`&PyqUKI)YXrp75r+;eHDVEMW+zmN| zP!@P?ovwRb8k>-?e`pgLBKhM_O9Khvj~F`AtS&+0zcxGdQ$xynT)p}QsBXLxFS4&m zO%J*g_~M|}qIoNdS=x_C6_?IsILiv%JlLk9d&On*T)|4zkx6R`)dlrxdi|~GZ>v=^ zn)fnu@dSgx$>l8R@%HWQW52HClzVL{dnCRXEjrp|| zU17Z{$a$D{1r)vK1Wh}-bbm{l%%US3e@B14zSdI|_`da97=i()bgVaKm4c11VG$;r zS!Zyzh?R<-7Efk|RemmB61y7jPV6BIRZ|L8X_6I4u+kpfWxt;-a^LWv4?Fb1&D?O~v9n*|uoC{e6$0mrI#sR%<2$z_pD;+jPC8d2Cq6KV zboz78c0OVptFBLMe{E38s(U*@UrlnYJ+D;y6F=3+w8ThjTI~V+ZjsK4{s=%JMXUGz z?sPxK4B@XS9_MpnV=P#&2lWLF>Uq}@6trrJ`7Nls!=CxaG8>73_{ogAd0Q@l0gY{* z3C3phPW7nvi687m;-3R!%~Gb(Z^P;#MEh7EPhMt1z>O@%c{yeBbBeWiuPkEcH%498 zBEwR1OGmgQQ|vqAPCGpWcQlWjFa9iD+3w5Gk{84JaUrJvrcr%=Z%jy>K@dLvI*#Y( z=M*H+`U38$K4kW#P2Viqn%0;Fnmh(lwqoNt8lnAmW0mwDr?2=L9KR~-=8IzosKqs% zbl4wZ)K!%1Me6WE=Fi^3#ll^EpQ4Cl{js1|!&4)D3to2!6C~N~H3HT%^Jj_hWhjC9 zV3G^Zh&Y6_tCaUx8L2PW3t{jb%46~FY06ul-vyk%&x2Y;#W*u4&3o$*8Vs_+R|90m zy@xEOJYR^%9eH3ZKCK5(Rq{q5Ek1A0g*z8gO5$SNC&aK3%pkzYrpc?94wn>mZ_+H* z1(UcOzk~X#Ly9Jonqs{uifX{yqa=?Tw`j4UiblNUp4rl)I(yWU2#V!8@7Ly0dkRbm zsAVfe+7sx;F+@z*uHV_)eU~C7{o53v-yLqwcemsVFFHQb7-<61L^e|9otiGY#;NJM zHlmW1SL_8n+~42nw0xkxM|M75ZF;4HwdhvqZ*#bUgN-0C``n|GE$-CER|t*phXA|p{>LtOV#h;*ndq9sD9Yist58eOo^wk{{!FvwuxL_h*s|sHi34 zV%?Sm1n(|fa3#oE>(0gQ-7=|t^!WC&y!GQskdIPoon6^hsib*(Um#SSu*!VpDD^N( zCx-34PKHYS=Hg{vYdKGf=du17evWv4ZafnuYww1&LlFlah$a{^8d&dq9|Rkl!=HX2 zRGq>whU3R*&@Z1GixV61t5?d%+b>;3-W5&kV+dwz&%UKVbI*}r=&+5~4k3oocm>?F zkiA`YZ9>KS95H?nsthecGuDTVhoF~ysj#3fdfT;yw^|TAeN+~C=qG{z^k&o{<6SqL z4^^qbxfDF03NOT;11}$A#fM^_LC6Odss0*@kh6uyPE<>}79wWW9ZdJPxgW&WseUf= zWOu)te$wd5z0~RTJPc$#_*rX(XecOunzS)~{vvjEyi2jbmGJ|I6*)N^Ar#}P%l+E6 z5SJVC90H`g7O{M77?agT96QRJ)qt#@`-b}}${x?)q@xGKetTX2-th17a={uKN5O3c z`z|McU~ftk@=G1f5+8+gqsa8B@ZvgXa{J2(r>) zmUQKqCg~8I3j>(1e@VW?hwR;_1Tpk1YXaC=ua6J z|FRmHJHI6PEn7ty?aV;D@_b98$%j1u#4WsF!8gn4SpmD2#jJhODF3j7O|jZ+pmR-& zA2Vwz#bFk#%KfHJnlWHL^1PXvvnTdNMW;>gK9XG6}$H>6@=nhz}n-1hAr6Ji)gLm6Goqh9Kf5kBML z%df|o0SziMh4#mIEVQF~pBi7Q!`^%Da%K}-mVZ-0vV0&33nEfc-j|q0oH(twnUCmWUTTHD-rJcN(fF z9x}s4q>>JGWSxr~*czPF)|`G&6RGk2{uFZNk*TTpz#vC8W*Lu2y%sbtw}Mcy-X1*u z!yr^U_;f@Ix{Demb&Huyd9ZFvxrChV<;(z^s*v2&hkM&v7;~npwG?>Nj*rqWR||tR zd_VMFnf##wWysc^LBNDiV?);p-z@tJ@eh1Ef3wp**H1id68RZmtEw`HWFH-i=y{4vslt64>*fYn%E#HP1pK;dnS31Gs{I0diPxU=#T&s{;7Pmra=_^TRP=a=~W9P}W zChGux<1Uc3Dn0x;$qt7=d(_}XMGH2YrgEiqrEP%9@Es0t7D&d`ryeMLNDb0+pJ$LZt{o#w{0 zef>!Wne>i&9TiOeeIy9$<|m~5D`M}be5>8y`J3qcahoiT?tP=UjJ+}kn($Yjg~%OQ zK7Fwr6cK!U9_cr3KGGy0ve8J5IV? zgz%jnrl<{n(}iabPYT3Xty;IdC+4v=P01zipE!UmtFWBO8j7Owb~l?SB+kJyr7BrV z*K9L#^(X|A2KFwh9|rqoYZ=Y?j1J*$Z7&;NwimVJSI2cBUzS}k20Grmr z^9;4#M6CByFfxc2DfKX5=R-s8PMKP9B(Prf&C#X|1BKk1vw^>c#xy z6!vE}lJJ4OAB%qJuP57<;HwWTZvsVrf}=09#@%+=ZtUO zf;Sto{s|O{q2YKPLTi(y^fO3$nOlYPxgS?5w9J}Vs7h|a^vPh^tgfIm&XiBr@&yH3 zaWETfR0wfF$(x#Z*8D4ya1$o!>!0lFev*}O)xIJx!REmWl&TPBIc${Xe!6X95u{;V z=q+RthU0)-AO{`nn#5h0Tg( zSJCJrX1Sc74*E22TCXGHC)x%DyA-<->$y{g%b2pxd&S}%RGK;qO01gDdIA2D$Pf;X zeVmt3oBR?{1TBo_1+VcjexZ(ya|3|&*o-f>tSH#Yac)9CAq5t{pFqP9ub0xM=Nt^a5!(?5Ky(H~Z5Gn($|4yt6AAAbvBIYks=k_5Anlbm?f?j^YHwP3Z<`3;Qs<9J77^IS%1x4l* zDlrAoPkqo*Pu=ih}o)z9=>3EKNuQ!GNhADnb=|1 z&6vu94ro$N4$JKFzGvs#Mt`pY)9k?KwJ-D%&WMuO`aZhs)`VJ(-uxweL_g6~1fQ8K zPyF}1*U)-erNbzH1pH`W%&=;y_2g;97Ek}_KOR9!!N{!Ef<@3p6}&FdqTS9%xV1fq z8t&T*^a_r+k#pv^lKYoBvLE$$sm z`?N z*(I9sE!;6a<|i7iDIW^76<|0NGg|%AG)CqU7seuBFR=ZyIZbt3Lpb@WDAg&H$&JrC`LMI}CjT?nr&iKi}*AGDZqMuaMH=i$(GyQ^O3EmHJdAK283bZ4)aZM(lj?i$Xg^BD%LbrKc+XpgHfcwk2nz=iCIv^DoW9elmM!B{x27S|GGOz7H|l z5A}Kt6r6?0DkL^0b?!dH>o}rr@jpd4H}xAUH#a8U*b7YR2V@D0Pen=A(sSWgB4>u} zDU{Xvj7i3W!vIr4B+;k>I=mFIf#|zy(53HJ>PB3^cPL6|(1j^v()DC510WMN$l^~& z_n9IN!fKxh$$FAdXQ_o*OUmV4qd;VEC#=KqHDze@_nI#RCh;F(TxfZriBAXY&|eSz zk)mJEwN4BDST+-XoG+qt(Cc^9$4~9+~NqCP+qdj87E` zk4#hB3a@j6wNbuu+auhqvl{6dKwVECWc%@~&@@B4-O`9(HwAr@;ZmaW?cit=9N*wb z6 z$zGPC7+?DQ`6d{)Q-!l(u)I>Vxo;w*yK>=0$_@PvuTR*rzc$sr(a21kck&OqpZv&K z0g_>Bj9PqBOtU;XncqTcrcfm5wR??V(t^0m7mJlF>?-&dB?CSYemOCyXU|OXReArGNh{8(lm|#i4!g14&%nY&6bt zP(=KD*e-H{+X^j>&}JTHv0q%+9&~E;@b^f@^brBqQS`6b4%1jX>YN$v!bzll&esTv z#5ZksIy$$Gm(*IQUv@&bpk34`*;LF# zqm4*Ep9ukmhvsz#v%GZ?TqCKpHICNEwCJez+B8x%`Ti77KjBlkNBeu;`YG%9a7%Y` zTx+zunA}cyfsir}a@l-)Xpi!{;QSd;suj|{&Gm<$k4rnvv0tJm9PF1_IrD4r7+Qct z!%R>~2kym4+V~Sl-i8g5!f02YNf&3W<9BgwwYEDUO^gS>cqeSfkMFMWXJh> z_pL2Swn$O;QgR2l>=$O+JX@?6qip57eMWIG=|?Z9v4-4GqNp`YNbL1n8WUMkQ1Q^b zRmVJ&6FGzxS%!2f%TiU!7)s!L47%&@@qtIe@KbLU@=M=m--qbK?gbTq`T#d>`PuW#`!Q$D} z@b5-s28pT+mE#-of;twyEq5|mSwuZ}xN7V>^$nuMl7A1z>)dPV9myvOF2T>0~@S`{8h0|%4D#vF1!dSHJ0HvtO;5MF$9d0aO(^;>PR#Qt) zCQDO5q-aK6v+=nAlNe%eD9paz_GW$6+_W`y>kJ*3IffZa%+f2pdl{X7Cp@PR=kz^0 zOy0>+>!lPcD`%r8db7nsC+|9RvYrA;#g-&;r>HAIQN%yQnla@~>0{533|SjNI9sJI zpgW*6wzo?r&v@t7TvA9ivzJ=+nw|9gxq*3%Kzfm|%3Yp5iDN(C1X52`oac#)7Svxy zt^6Hg7Cooa4cl0ox5n}WlQ2@r5;bdhhnqau#ep)o-TBN=ud7ty zFjO_IA4W1of=J(Q4e{IMSfjTT90!AjGQ;9pbv-Zjvg+sa>K)NX+wW)%ScwoPV}uVY?lZg|ovzo6@!id$)V5a;P1pI^C|BRR35E5%YqOLSaM(pd zm9nP>_$*X{wCD17)<0Qq5gX5XHIu!3A^l*<(K3_nM@d&2@l&J(nnA z86>(DWx)h93Jc?V=4=*PN=wT(?KukUJmxl%BN``HFHbL4`?19It`glVpDjMS;SGg! zN?jt-aK8&AF_m`sBhPUkO7HnN>D~?zL8W-7CqRfZW!qdaJA5oH2@{-l(79S^n}?oh zYywY<(!5_xoqbhhAmD$Q8l=44c~7=tg3N1GV6r$TG#`ja+=@slfJgCVq8wF`+uW+a zM>5s(8nn;%93mSQtZceY%D#t3`$`DD5oKm9EKD?4yZiCSPQ+J6qjuRL=rsGoNYl1lB&Gpg+W()FYUyhU6{ERf$ zQ#{`;+RKBW`~=DBU`0ieglQgu}U?J*(crs*)~GS-UcyC*FRk5+}6_ zm76$wypuwMjb@fvIF2A&KBwc_GtU(N*XoaZ?)=4zE)nka zANKB&;a0MS&e|%#`h!qP;HEq+h*W7EB(K(+kyj+r*}R^d!0TCm?%3-0 z)l*BwvH$}-OkAubv%JUA?$)DkIEC(rV=Uv8wH*0&{ zl-?7nSf}@G7ZGm0?^I|MLCE=np7&>~&`=+JXv$Yu872qM878Vv{nHBKaAOpqZQBab zxkg==86wsSwWjP&G3SgOP7~!bqL*J^udQELfv-io6KV=OQBA}fio90}@>gTRf$imq zJNVn~v}8&qx-*!=qj4?iaKVM$t^9=z<(6%7%NFQsDP}vB$`ZSAsHC5gn}fVP>!0rJ z{lpPV-v;&U-Qc92UJSL?Iv4K7$6Y7*k~0M!v2$|na9agGcXu2^KgMu(p(diXw~M9U z%{AJgJgAbHp~*>9L-AM{+E;5vt2%$JlP1-nTX~-Y7_LcJGrsoZ4DTwkv#{|>q?O$C2R{@bxkO~t?d4rdM8NFx;`a-A z8=;@L5ZzkScm&(tiZaurue{WlxeVC3M9x4C_G zl*BNNe*YX!LOtT#P(!?K z+gDsDn3&PMm}`Ml56dB!D1LNN(lhQ_2mSByJ7PxYUusa+v;nEKpn&@QT_;Asx zQ&t*P&ciLFhs|2ZsY;l8od~SNK z-7h2Z(+`!1Pii^jx`~1w)95A)I<1rVXs~1r&))OzEFmBSr5q3ypVw{q^Qp{E6HBs; z^S1C3g>$#{((LVeIee9hG^5s5FpdsOp^0`4<&g|`Cmmt;pnkGSj)JL&e+^Y-H5snM z_&EEW&Mv@NpX4g^8N|U!UXbV0*T%|_E@UoE#B2@rG^Cpzx(Kk`I*wK?CWYqb(1!2k zhf;g{r);!%pDZ7n$05X3TtevS>zM(>-h5LwiRt2%U}D;d`< zJS{$$yx?gff{8nASDHz$X$*2avd=4+tQ=d)O$+g;q;goKLyHgmjM5h=m`5w-ED`ub zXdlHL_zKr+rV%k6uzy;4UnP2BziOBU&VS-c+6ajk%DSO#)|yqGXL0goZyhKG4p}LDEd1P`aDL6ZQ zwSV>I`CKJuuR?SrE3G#O%SctN+N`@oOJ`OVa?OxEs?OqXsv`ZeAeUjIPj*oC=p@Ky zRr^-3kE1HZEsVnIoe(tlt8p~x2~o0MkX0o$Y&;$hmPbC1z&q~_L?Ae2ANLQ6MjxWM zy~8uuDjQoS77PBF9to8<*&uFHbp(MP&p%x;+8gYe|9v!~L=C-T| zaA?ahNeDrFoK%XiN`FHuU0Mly4Hpv_&+fWqO|7`oHMWf_^yJMkUcvbHC}ybp=LE#~ zgJ7I_sSSyUi`9$C;u*ig7tXOx_k+Rc;-uj|icQ7y>=XVrOwIs{hcs33m62+W zrfEmFGqocofe!ROJ+s`cmbwJNcR*`wKi_=bbzj$;aw2%fh@(G>J>>!zALsGoFSMjp zao>q0E{iy<(HhFHz#V1Ft_<~Of9AbhTz&yF8f(`kO12X<&iQ-kLLS#^*dU344Klj; zjbzBFc9b+V&U)i8j}5pxmN3$NUS8+nhP6(P{F(PuzqQ#2S=c>D96Ui4gVTmIVSVL2 z<2gywbkFXlz*ofsLx*w<>c}HGs0Iy=^0z64sa16Yg%+d($v0lgM}NAI`-+HLtXNtY zZ2Zx#+?0ydNBCPbb)p$-gmn_NtJjFf641ZBAauDfgt~Gc9uQV%b2r#73E?-TM)d2! z4<8+#!+5pZ!!IUde~RWjt)1-3;!Gb6CXyUR<`!uXjNOep7}1Gz)xZk;UMI{TsWU%u zKe>@vA%afkXUcSkY654gUyvAb_kKT!Rl;Bez3lL&3Bh=Vm_YE4)%eDailQ_}*2tJx z`ki(@|Mcc{_L^#Jd5+e3GbXuachT(Jpl+1Fc^O|WJZ+!N0(=xf^b<4CC>7(}bbW*M z;L7bXFZ5#w%a5;pov*l1e)(oyBva9AdIpxKZp7)IRC%!M=%4-5n#LR?i)XQtAaxMS zDhAaeX+XCC;tRm8mL|=AFEDArE_zy>3ho?vmO;oiO*~sS#_2U)6{+UClygFzB83|n z3tWsHU|j-_)u1yhOdsG+CZ6LtO?P}c^Ik!A_gq>b^#Grj$uO;0{74r3$Ov#5mU)Ls zn{EPv`eee%D?75@<%Klzi;g%jNI&Ain$8gG`U&hx=+Pf~4Kb>4W!d$x+y%WlvMCU^ z%7cyh+#qdyo+f1g_Oh3inU=Y9Cw?{L5+LhWHpw6wKhGCnlEL@#a3IT!b@6;=0Ti}{ zM(ZdXVs|`=5x=zj`x_4{_Z1cNp!QRUhC>rx_Tn?KHowBeJ-M%ID!w)mSKLcpRz93Z zbuNp7C))UGgXOd|e~ad>Rh>we4hNHnDc7#z(0-l-TKik4?6aj$X`9*4lBD$t^pnEU z;+q5sjluQwYHwoPyYwkt`nPvYW19(C*H(cb;0Wnu&o|QU(fhEeG)p{MiLnB|YF9yy{)}hNK=VkBEpBD1!LGB!Nj59(;w1?8F zEz8Vajz%fRGYW0P3o~*VSU6K&#y#r3Nn|Z<^F9~3u-tnQeU7_s@u-MXN-=)} zV-Y`Rd?bS$ZVn(3E-cBkA7wvSFdO*KObkybT>Wb-NB)bxvQo(Jvut!tG!5H(bLjXr z(eovVVbTb0edQX#5(AD8wfl>S#~ALyHKsa|62&VH*t%R^(UHRBt~B!`cSG9Iz}T9< zy{Lke{_Yie*B%}{0uxCDZy+R9pwAIGTJE>GI}rhab3!oSmXl!8++4J{)xT1axt=Yg zy=Em17Y$fl0`ixM9;eE;N~A3wQm0M8zYo%0Rumw~#onJaYWR7)f6f9K#>bT|KMsF~ zzfJP!55ByVcKMV){Oz8DXD$;!ijB$HFn0U7ZBvjhYd9zxVCScGC6u={5nUm*6W)jm zMq-+D#yBxd5H&5zLmRa+;pT1n>*tOk%O>K8@%EW-&rKHYq{t3QD5kCcw0t=b?He2O zr341RPRy!)0mE4m-kjteQ_Z;#d0Ykpz+4B)^1@}_n4mc$#oRF3i`<{*EGdrR2zGI(iW zqBvk`)bTIHZ0X`DP4?5AkV~+YLb0;zQbOln0dHI~a-&;me%R6{@wo?(z^TeX&Ug`W z9%eW;m%xy8iQ9g(m;4n-l<~t%`1e7dTd@=rigV|o5mFaB43?lJZbfoSgPG^UDT>Jx zy740~AvTTFyKu4tDPt8gybtw~N2o1Clo>VaQTCVGA&$JLb7rykcS`6~C~xd&W}Qhv zSxksfAlO~GuQxFEPjTRHA}OTa&20$t{_Q_taJ^Os!76F;&#P>5ebtzG&A;531fC4;7wd-9j<< zdWLY)>LccSDfJ|@j`+1rQD%-Ws9X_pi>jW&0kkSaO0cyv{+(^Rx0kL$(Sgs})b)|M9a?I2MZ1eRT(kHdKc zlBXf*uIEML3I&tf?7kHB;8?cz!|Z~}(SkZM4Vu3}Mq#gp5b+(aSJ%YD#Oxr^vNRd& zbGcHfPuW|9OV$vGL=&0z7p&U0h;#}N&tnnTWY0~+L<-WeT)NHLHC!s76IWAYSGLU@ z^DqIdj^8Z03`HTbOW?A5Ha4cXYM3|{`R^=L@7_4(*rl@*siQNjjgNIfR?NxTUNT>H3DY8FH9mhGT#&AhtdY=WFX>8ijS?U$-+@ z7z)9hniL|27;_PqSlWO0WT45nBp+wjMbaAKyxr?W%G=rII?x{@lQu4$uK0}^l595J z8+z5KU^CPv^f<|g0KS^oYSR&S^gwFSz`i(g&3ano=qz1qwo6LB;fpHKJ)>MD2TNhI z_*e9SX+}d$2Ho}~PN+n;VuBVko}ssCW*Ug92%{aNlu@BAaVAlsyQeZ0 zy34)+hMhgms1|c)G8K=*D5GPKdL5x_AaQCcAe+`jLrt=uF4pRflj6#_nm527q++F81)>fvf3gbR&>7Uq3Vd)q78YUptRU$~2m5X4to*FrpY^r+a{bmc&YTW`IeVuw3#_IaL`mF>dJfoVRy!eF z;(S8G4`V`FPM(hqMt2IpEfcOXmBgWMt88^p)j3^@O}{!ZSyWgtnz$oC1oQd-m)J)kb)a0S|@n;8#%6i1QoSstTME>cyS#RbdZa(=%27=u`rFp>pbLt7y#k}`h49K*&TKj2(x#O~9+o_syQ}Is5 zWS%@#{n$rltJ{_zT)W)?E0n+_$EEt6J}-?su}r3F(3R(|0 z3oXv})XB!2_Nx4@UPs!W*rB&loJ?Z)VtY()G1(DLgqa{^r52BCnO##{cxjJL+eTzL z9b7apbZM?X%%X9sPPdZrt=#W4bx7j@A@hv)NCI_5{&y$~&Ib%x$H+qp5rvTSy^7Po zoO%@1$mo&u```eG4o!Pk&yotdx8^b1<4mYixc6XAtQQ-NlT%= z3R4eE0LespEg7Evh$Rs5JZMyEJFyENbniUp3}Nt}zB~^h5$~D? zRXD%kcH*kRM#m%YFMd{tSAIq!Jb_nbs}VVG>u5Qah-3WCNbdcVKMu7-?GH5_Yo~gKTz-9Dn?!(nUOa7C~hQ_Ku$d^ zK=&d=xV|df1*e_CIUvqxy>Pj2-=@STA`{QG=G<4BoNysB9MrkgSJg=5!xPlgg59Lf z+OMn&ObmCV8_43!MdBBbhg5S!S*T)f1Qj!TohnMVYHsG=ZDFb+)s$~Fta1OIEmz}N zN19gtiBY{{6-1Et-okez*_Z2iWFTj4q5~?~Pu#ZBGIFsVIyz>ks$%!Yb?BeNS0c~z z8F_6Mib5ViQeJL}D`1<9 z)e;_;g*6{EMa{&mmO^=>Q=Kre^dTN}CypnGn9&UY)h*!&7f@@suK7q-bBqM{`US(k{&bRsunLy?43F6NZD~i4T&i8^Gb#s_ z$bGycJ^VG8%mV23%) zBSO@r6z5v-O{!lt-6clbu%pvT%jtC%PDHSgARFb^C(8QQ0sGqU+Nzi)DOqrnlC%>nTeG{_3J zIwW~V7N*J8uQlbS>0ZBd-ewD9+rpEOIZ%W=wPC2c+^NaHnHyMEi5NDzk_GPxYWV$X z=GeEqFTyR&dleG|WXlG~PgWGUwj+0Z-=Eu$qGN~Tvj3gd)~h(;8#}~^kfszpN(H+n z($5~;GHpzBej<>o%z1%q_E7r0=1ixIYqYbg+((Ib!97HxrZB_Rvxvq{z>|lk@lU%8 z-tTb^EST}%$NdlpHgUOX`O>+O&Tb=H2NZ2|4>lDQrfdegyZ(NTs*4uSO7$gMgobHX zgYB>fi3NX1=Gyc}JeN*H9wR?-#(OW-11>7L&d;NSn^x5Aa}U+5xaVWRrzwdL-XAQy zw~KhlC8wlLB{XOf_00k9$0dbNk+zavl&oG!CoM1Vk`Pef7XmWX{Wdoq@Wfvoda7!< zi`hxAZve3*MeQe&MtCbrVNl{sLx7D^nr!R!bI&}um8>pl0re_4C+;1C5fE-X%0N99 z7yIepB0DuEaV?Nr28U$%^;X>gu``V}2nj9%-0+CeXX(aL2oJ*+)2&P$UWc!rbPj@k zLf&|swP5M^@!H_g4y`siYEkD>-J;86TZeit^de9VeQRSO2QbD)TCn6d%rX(k{OJY1G&@o;@>t<4q~&;C6N2^nMFjN99K@&ZyNlWm05`wRDjI z&Z?v=7!e47JnPxTC&i;B>}1Kd8>9-jAxjqP<6Db;xf z;IAD%atnUK%PlI71vS2R%pB-;fUw~nPuffrk0g{wVv=RAiE$(GHXX%X1fg8?1$&Fi zbXXK7F4?z9LpXu4QjZsy$<7GmBflRhi#r0aW z%g}Vkv!ox9bTw^dVa;0Zi8`gP-UHHCpT7!w#$^-8a%Z}+p_sS{-Ac-)CkW2@VD9&G z{k!JGX=MbhX7diEJ~RIn-oDl_!$;;@DlL9AS8?d~d!i-Vbe^=ATX#0z-+wSk0V8lr z$&g#KY@!+0a^v-V2i4fPYcK+t(%$p7W%i10cQvs={cP6>)fg??80q$ga{$WL+}bCJ z9IDvS5cMZY?wny zSQTzim79PS@@^4 zDA|~GAt=l(m>vxgzq_ps%}lHcK1}NZ0`xt_--o3u%4aP zgzBh?MEZi-&jUpLza>hr^i6p@&I~$qRM=NikB(kdF*a>(Mqa31RIo{WssRAkEY

F2=0PTi(i>HsBc6^ENNvG3_bLpMf1um4p5MZKPN;<8l+OwF;HfV`dk*n688ZMr+ z!fB7Z7|p@ekv_HJ$XC^;;VMYerE|tY16%aK)%b2FuTx z(fcaWyY3rbv(LetXy-$H3{>{>$vtTF0n<<6S(ZG7nyTHE==HVHmmbwAupfMn<-=n~ zroTVs$*AoPIV{-R=B>Vw5Rp44w|i&DqPz;_w8F~YE%b(HYBz+GFpRL#VIFriLn-2) zL(THm{`YWQz0GKy`_6q=>ULX1y(^WG?`B^6=@HmSPWxG-6cbhaE^gD1hetdSIZ9dc z4Kk9c3(QRFX4Y}G8J8O&fmM~#(eI*xY3Uq}t)%c29_n$0dr6h3bOEt9O3*6&k$03R z)H#f9y_tj)=hAkW>MMm-EunqjGDW|Uyh1rKp)-7nzbb{ZB447z(L zzFQm8GkQN-np4BS6^+FJ{A?DZQwD5Bui6}EmRy-j1fK}#&A-=CP#I|oVF{eatEphZ zPV>hP?a8`7gzeZlh&u7@D@I|qb0ra?>KKW?w84%=Uh9g=Qo~Rt0$FgfXCCW29A`DJUKrHzS?rAMOA&(!RrN9`< zD5Po5dTy%MA5!ncWauo7%j(K^zlj%o^k&j=A*A}{QS6`d<<86-##~icM}6KRT1LbM zE~ivuj)PVl4I&x^id{RQ=f*Ivj7e$lNz)qH=cqn|_B&fnDy1DnIEs{_+AmmQH2!fZ zb7n>;G6i7q>Qw1r@CL|l&yc{A`e z@qRI;nzLoSSa)*fKtl|nsR*(rxt~u%AA2$j-W4Ig9s77R?ypZQJN2a~uSj8p@G@20m#!XrV)`6J% zQ0aiK33Gl!+eYF0E$$>u2YG*!^b9k% z<>vEuBdEzPASN_J^^(MwPJF``d{cyi;_gHfZ_GO~<0fMw=7u{VTqsGEpZrtqATF5E z7$?gs?{cyZpSOPQ+ohw{HG*M%-mYhuXt$>|;KUXFQ?cpOADTkwKX2{Ux`rd5wV#cPh+Xw$Wk5LFoh3AI?pTxSGV08hDrFuW0eyesE3iHlP+xIN@ zl%m{tgDu(l>#!YeU)km!-n&hHe!ocE2$|_&T4rFJwu0*&rGzEYYEG+r{yC@MhJiK# zJQxK%B`MVc0~>0+HWqLR6bB{fh@yX;yU$5&E?Wg^Z-si?5C?YGT=o2_b3Kzy$f&<3 z4;`zzA%YS2S1Y**NbdSII9=do=SYlL7kyyxZQ`CS9Fg+R*g!*DMYj88X-^By1VgY? z6j`cKQ&}bPBOO`ElH$ZD{3J53S^S$rsspZsAq7Z6j?%CC)yTq5qi^tt+y>%G6eIVm zg`FUCcqt4LaGryC9f(j!xpM)v6-r?k;zt#4)AfxhzxOu+`;q1<*qq=9)IUi!o?2kM zg{p=o-~YDwa>vuBO~eLFsc(BzG1kgUIx!QjpNGd|`CR*>RlBJ`6kl;iw3{j$gK!4( zwj1YGSXSd?T`M%C5KVmH{c>2hip%jvRH>w;o;ZE@@VAwLavNWvrbjp7EAHK$Rw2s( zhak*X*!aD3DjR`vz%T`U^^y>*$nM|EjgEdYi1&EJi2)Qt&mE}iQgj*7VjY~pF3|$e&PuVa? zwC9>a81EVLsHSj4jS#JR3~eBZY&_zz<&?x*p;J&je)M786TKIQxd*K@A`(q3?FT-p z=G?1anC?%hdB-P}0Um_9qI`7Q*$^KHbj05r$ej4FWeD-Y5Od&5$VJ=6h1&$Jhj|9a z@rdG1o(U<&I}D5m>yC-`UR9NCo%DeJM8r0`E{ zFxfvPe%;fK(eEC@n~N|pR|Xn2K{mt&{1qFWPKnv04>MB;m$@}!fL2m za7i>TOJxm=3(YaAP7p<>+P==^H$f2k{?Xx*G0K`lD5bP-T{z9H0F!j>9;Vw* z6O&&(|D|LVoU5rOKF5N&BExiw_*^7m3OnCD*Uh4Dj#C30>&iqDtI{tbl2C}RkHMok zP{ERCvSii)N4*WJusGGO`LaT#=5MuehOx=F6CTrxlME<7^~3X8O3;K|x&7$}y7f1& zYrIyKZ^M@CHzXjaKi!E!?^vnY&id*>vLcO+4*U`DxLcc8>XF3DaMZW^ZsBBdIdWmg zrv?#-;ys(Mh^<4ZW3DQCFn7 zHB2D*Tl6Kp&@7fk|i z{XHX}_&q@}3*XZaZ?-vvD*^M#DZFmn&3~Qap=6?EclSH*SFBvn5E&DePgs*r)8oxP zkU-@Q?jx6rdC1@78^P=ABeg)K7Vt#Y!$;>4=GX}x=C8h)uz+*cO_A0l^%*KyH@I0V z^)35yG8M^WuZO&u?Lv_#X?Tv65InQQNh5E5B8)L)7J)HMJ9tP-tl66#U|EFXvOzb! zYH3jy*(0m4+Q>Cm>{WLXD65%7bNLbXNM;y}t>{k06LI^EaGDCsf2&l?_BL4NtGq{J z61>CYS|0M*bp-h`%%A=32kM3GdJ6kOkQGOU6Gf(ThSU{{^6=9xq-7$+#3@hwdp@by z`lXB$?_!2e4%WNsq?oxaw472US45D{Ij7l(e2HiXF6xD2L_HnMILBxS8#$&*Y_R5d zP9gNLXBO*AuRQ1+#2B7@(PdJUn7hWKdPrcu+AkNu4hgIG?Q5i6q(yy~MOVnmcV5mp z+I*8O3tQcbX+UTwAc_an(raM|is^CGM&@>XR)t*soqz^Jk{lxAn8biv^>*iwOC+1* z&Wbl=(Q_%4W^E6D@wiUywKv0pe}C<1v=bpX&N)}je0G=hJTU;rqcox9>=(HrjV9SWi`O2- zS(x&0n?I>9?^P3kY^-DYAY@aeK?XjO2QdBEz`Ao98hyqYs09?h&^M47ZDkmRl@+ep z=a#>5LPr`}j|2DMW0R7+T%0H@6IXcS$JS3kS|_IMJie;@3Jpnhfb@3MOUb6I$qR0| zonU>!Cwui$HY{NM8PNRs*LDzrj=-P`N&KDj9V=5N*+ePsJW?nL?T0& zhb2LT@}n0VDu^1_M(i`ayVmN&{>`?x-?%(j{ll z#bP&iziGDQ)kkkvuZh;2oxuM|%WwH$$n`yc3La`fW$DcWRlW3~G8Dv1* zd}GnPBTs9qSkvXC3X%G=5DKy0;z_T+iTy6~O8dRrCpEg#$7j-vP=s?j`Gr75t@+{k zougw_p*_i-h|#dauWqu%KrZ_5b;f5P&FT5b@yx8;p}rd&504DB5?(AR!JHpORro){ z+U1zGE4oQbN;8Y*`!u;**&ffM6MrjdxcV0{2eBgD>z-LtC4NS_q(qoELr1NuWt4Fy zpb;0NRM}#UU+srXtRbhHImp5bl~*-7TNj$2eU;axiOvy2mKcPw?e8fvTu8Y!Hdjd{0$NCjti>;F`AY}*UJbGUV$Rsrx(INpEud- zO9C)-1x-yHtR3fC(17V|nPB>}d$SIj(xI2~*=^-ax*c>-3;*NjES4+3 zDGl!ETuKsY4+^+Mv+J994_)qM#mQ) zg;BhJv8%0uVWMupCMP~4TS*4<@VTc3il7$|D7z7m)Wks-N7Bw_Q=qMtSEhmcTZg{P zV6ScNJ3{EdMfZw$570T^HaZkQmuG=lA|b^YjQ#h-=8+Qu3S-|X(%EfR&xrGNk^N{{ zYdnLBSQ8hkUN5nxeDd#%#3F#y?12DNK&-!!c#@FGx~@#*iT+tg+l#cnHvv{21p=tt zI}+ZpZxraymwZ13OBR_v4R*^@NlFfy@M4$zF2`kccwPjdOco>ssKbD7a_StFolqBs z7PN;ZbJNt^upHUx9Tgq#=X}Uf6NlN@l5Mr#tiXwPl3<{7y5SsG+SDT8=4&E~#Su!} zRyv=VIA#dOqD9IkZXnh~5#L$Q<~Q3db0N4jE-34OS`m1roy!Bg!@ZgK#(WO%@Nx-* z5YwZf2!?fJdmbahwbBPgH(5KoE#T##W-F9(pCYJFK$s=6JH}@IHqe+m6 z{t~vg)2Le%aTK1REBWnB?1=lUPjodqWJ3xf(56#|G8SY9)A%bPal}9cxYf+I5@t1| zv{zHTWSrWs;m==ZPf9)&Vv*VeCZgyR+o9M9Ia{L3*wF0V-!m`v(}AJzs6pr%BC%*k&UZNLT#edJ$C)$6S(~ z57$Fp22H_bR9M&&=Y6O7@%j?^M>HE=CkQu7xbvM7r#{sVN#1zUg>QEq&aZ;EG!jQs zzE*lRjAkoMNO|LgCju1tyKz3w8kFB_Aon>jU}QF~Q8nO6dJ^yt96uN(CYaVs1t++I zs%(h-VqDq}g&d#EZB`ZSD>*845N1q#h}hb0LQ>P#3Lh@TBY74@W4=d(5Lt)D#n~?! z(X5Y!&fYVoIm5zT^<-HOcLmf%#70@!Nw@E_&k+)nLZ7&jM&T1wUzC~nBI`i)Q(w{0 zPIgJJ+?l2~%iY8B&h~tAF{gcUKT;#8B{RJUy?k(luU9+W%GNGSlR*bb@Fy2$v5>g; zLsi*o-N-c*BpZli=Q|NHqs;*=(5BSfr=M$`N@D05As|*bi%Fsa-C4V`x}C&IcC?T5BOe&> zwvj+n-dD|MtO+*BoWX&I7$X#mn(Cq3NU>0YJKZcfZvr0Hee0W=tFe;xv8vFGVnRW$ ziWbiGfA6V^PQFke6%LQjfXE9Uz ziW@qa6=q;ymx(%rznMn;eRhxsp|LRA!k@CNMkZ!OZ>3Vft{j?~eLRW7By+ z2Ue|hH0c-`Qzydkqr5^7*+?|(-_KGl^9*<_SL0><5y+t5LZIl|{kw;NWkvRP^VO8S z+2J{8ayeBdIQS7)*#d`5)x|G96eYySmsdR9xBQ%uY2-^;KMnM017XKeW|#bp$3^tS z2e&mU`RIgPS3LOj%z5_44U)WBx6FM9jKYDq(+YUXG9*y%+z|E%X#SaF`8$%0UPEG0 zNjrOzsbAfd!w$l+$6!$`w57%|b6^x_n0jT%o=+u2es%d_&1xUREV{6P@1IpJVZxN1 z$((f7CvF={MH-kc_6dM<@vvF&iGA7?y)inI-@{>kc<(dXBgy$LV+08}OWre*!5Q6m zbSRCFD(Qn{*Q7r zvZ{lwG;cj$tTY*67)vZXXgEUA+&#rZaa0>0`!LsJZlfOm?umu+79DL{9;b4!O52Bg z_n;$>2Wz~1C$)3EP&Hk7DKQ3HU-GzvK}T1JBxP*2KlBoRK#&SKRU%UJ*^@H95_#N_ zwWO$eX}O$nKq4u@(XiCXJ=<+TJT&57ycjoNxOJu#rt>>C29E>Cb!XWxZ}NGvpdU(c zjIjuU>{8))jP0!;!qcAL%=5XOh4*N_HP}#7PB;|>&t-8Pu4`hvNs2GD1I>?Hho5AJ zttK2VN?n^45+TxsAac>XgPD5hBu<-ur9(Kiekc}v({&rZB~^P$KYq!L9xhMKXQ zi~Xa^p!dZt&%E>Ii_XR*j5Fn-$36u~wqm#~8MU;53$U`=HTE#`NnbNL<9l+4Me&D| zffX}aOq7H%u9NfOf#A!*&a5BZm0FkG)wOn;o@P~8+56y6{zl{)17uMg(AzBAWy~1s zu^y@>tcm;pp0=JtRVo+bR7_j(cEygnG_h_s0?}r}4|%$sc{Yd5q^enE1dONVf@D`? zeKVjA^)yJhsBhA?`=G~Py=`DNqQVcEQ-?@OY{}3!VFGI!QCIbX@2GH{GMN+@&wxD? zS7ZmL$&mk?H&Slz`t;D0ld!^0MkJ-^+)IWFquPGhXCy%~rPagOiom+dtCjV|Dm;8i z2R(CQH}37@th2yqbY>3+rYlFg?mS#A7M~gexkERG_QHPlxr5dxIFc(`Tpa4d={Zdw zO~_43{1$GDhV%b&s)KYO_^>v{+WHD5PQ~h8_HJW2Rf#+bEj3Ah9xcYM?f5;OVM~+N6M#jI7p>$_nWEL)X78JcskymDI=`__{ePW*SD*La$ z!mLZk(ti*NOHcvy$>!%trocr7DP7v~krxa_jLb3o3iS2}q*FKt^t2f&4kPh_l|>MA z=&4*7yB_E;+v`|J9Hr4PWP)YQuG`WU>Ordx>bGkU8sS0;6B8jH zD0pBE^TL$+cSN*T87?0oWb%d^W)aU~$!OqPQkO=}VW=*d(orLFjlqpws=rG@kOc+Z zBdhah_V9KhYKlCqL~8j3)-Wz?>w_b81kB$*Jx%e{qDuh&>YGuEPZxAMOXOo|Z(;EMrmVCRI)81zd!$|LM z=%|PUah%AZ`8+*>iDrr|M@Jp>=Pj|X*3-Fw=RA}jd4AMxhqT1{ zorz@;Ezc+ur-PbGcUWrl2Ux4HB^2S;bS=r7etpNhiyu#NipHk;D6+Mq2%;tSeIWtS z!^Y-N(mZ^yfo9m zRf=_d&3Rwf%SpV; z^EJltWfc2@D`ecqJ zv^aL)@%`oV{-=HVQ=jS9C;4@rjynO$*(X>T+Odb^Lxzo%ZNYx9qZy5T7RSaanQzZO zDss1|hZA~piTW%>hKb8!Y0Sw8aTEkw?qvPly^=aZ2cOiGHZ*9iy?=owtV3K4N$pN_ zeffyaOu|%j1_7HxRO6kx^Qyfo)2O`k$3h5-hVF`wY)Ps^>oD)jjRrk|a!*}&HL0!!L2Ej z=A{5etJZb0bM`kjo2-U4j38~t4b_v7*8yX3cZxoEe2U0sMNp;@LAG$M5&!O|=I z_(*osdbqk;eYj^vwC>6mFSXIQ!R0NdV^5??Rf32TKSr#7t($ikgoLM@OsPR8TYr1@ zOXFfq3TKo`_HC@po!}BN6?*enN}j6>k;6zs#ox1I#m)@den*8p(3rky0gU^_sK+(` zuJm^*uewtkfYO2qe+Uh}7xKhld9ed6a* zH%%mO%hFEYe*6T!DMt}~wzI@rnfuepfTWpME(?#TX)Xkh# zDA?bmFNap3qpVvN8X9O_Rc;2!Jlpv+iOet8Uv|4UMM_K#d-{7t{~c0NN@nut#(dL@ zffkc4+|D#+R=)jS?7Hm=c~(^M>f-u0aSq8?wYYpTf0oWh zZtpNwyCN;nH?VY3h+s3x`TTrmN-hf9nvRQv1${O=s5*xGibqwDCv3cHl#OQ!0P4Yt zLZzG`P-mlZ{6?N1QE;4nY~S>^OqKGWx{{sAv(+>|Q9B*3HB30Z zabN&A-5RA>iv`^5mM4zy-z(9sv$o6_qz@noof;b}35!+#x6?*pH}*nOW)9}Gfg+q< zKP{7yw(aWQL?vxA9_z{UOTja+mheewiHFiwW=lV;nO}W?F>TyQF}$%4 zUB-|mi9k*-I&klyhJz&TAfaUS!#N7S?C~eNPJOD}84)Zty3{hhMkUGRt{3njV~>P| zaS^~~$QIfy%88|bH%ZZi3bAC5nV)|l@Y+gN6vC8zK~sBzXd;@}$*hw?M!*wX0Jj$o zvljpHT1L~&Hb*RklfHl{0gUxl@cNPd7QOMd#Q8blWC7jk$oxG9DjmlTyoc=Hagy$k zT3(T*v*XQuc9&nwVMC0)<`+C+z=39e)#$is-{2lYuE&a*k8f7ic?v!mU3cFprD%;$m%t84%eLXeAy}Kb3i8~RX@@Noh z$%7#*>`s-3VQU{O(}W;nsW@9J{l*u5yQt*f!OOg(%KS$ib%zK#lEpk}!1a+=&=o?lRu4#7^Ua&^d2&n~VTG@sF}+3E zt6TWj-JSKQo=3e=O}(!ZZ(+a8v;*G|PoruGXF)qrsgB!&C9o5$m#C#HWM65fl?X?0 zn8ze7sp$oA~DH8NnP62sq8G!XV+ zcx!L1wrIxW@Qt;oTKwba?W$2avtD>LbR7zGlMdwYcT^{Ldrie)56S(#ipxEL!d9ol za)8|$pT(%AYD8%5){p zinJ&(Ok)w7%TXAkGMHs2^6FAn5`v@}r9LNKZ9?_6S(>9Iry>^n-uwjMcjX4_9wp)< z8sUX_@rUszY<+&q924j>OWI6+nJA0kDdz?Vht^ym2$N2*Y1xn0xF|DOfBXsuQa;>df8* zsi*WdB!p^3=8UegW?=&tOzs9;)>=qb8c{+cqni%qN!WGg#5fOeG_lOOU!L(i_2ky3 z@DIHpS@H>z;OuKHr!~DdIyV59S2veeDqZRIGOjnZ-Q>x0({*SODixSEfA@7m@h8k^ z@N5Md;K{=!d|{x2oR!ASYA*Zidzj}p+rPQME{ExCHC0TVWa{yRyYi+TDuZT6MfL2o zJ9c*-O{Vb2iRSX4y-^0zV(Xd6vmtl+s!j$Z)98}Sczn+aYd#yljli$LlltRcSJwBr zcl)m4yj?+2F% zP~z$58)`e1n*ho-T{(Y{_&gmo(^oPKExFVek@%0NtWnQ|& z)S*>vRfSJQ7zQq4-`*m_?aki$wds`MF zizHPzzYhE+rpQz(3l_OOfq*T0Ny%q}19h-z%g>U>iNVr?yXB#4Z@ZIfS`{{`rP(wS zMONk*(P(rdR3AJ8VFW$3#o1npM!7vP5X&Tr0n@6`a=;6guy6(mAEmo@>zjMXE=i>P zLO(oaN3aIOpQi`zG@+(o%oQqh=u_|;IHAwT+2m&7tn+ zbAgj>(d$V#uR!%=ucks~^lGt$RjU`G|gxvN5#;t=t-<4v4qu%!mdN{Rn?j8BH&mzm z#P%@(qTEj+*z)zq9p9DPMVBkx>gAerF$wpgjh-uDIT;mlWJfZ))-s^H^gi^)4+J+b z2y$Pv%nS8UP$I&o()>=ZevFfrobCi~2$r&Nkw8|S;+{C+vpsRO!~yyjg=<)og-7So zq9kt7ubmIukwWoeB_Aeg;dJz?7Wy*eTWvK8(9)n{JAO{}9R9z3&F1P4A^>P}wdvqUa|HpAp&5NzKYaN~bm zYGrn>NjV)6_z#tS*Dxp6ituyHGzaZW@g|6#iwl}IN3otPyZl62I{dHy>rg~i*Sb4;?S2v_p0Dn~3&dHSCpp!wnt z0EsJ{P`?lwqf)}C^Z0$V5E@9=lU2jXW5iE}vkDY_H?9tlV@8-`mt2g%$nB|suoJ$JO zD*H|S#J6?_G@QN0(+--VyZScFENXl0?UtNNH{Wz;^Bcbz;W(V@mm7??X7RZxCf3Xg zZ%>QMujbAf9gdp?cP&%!VbWXf_HS?kk*o2~r7uW|mG2^Kg!WstGI3YuB?6_;xwbuw9T+A)Rfi-Q? zkT2EOt%Ttxn$BDdcj@73mM!~VKkNqq6ffVqn`OxHbH^FtyO=9|J;-Bdz#^VePkO8m zDd8#g%aTk#mP9r$$*tOD#b> z?xQ)@@ByC~iy8bSd;!jmzdmo^$z!z{L>*Y+Ym`R%3Ak?g>4?Mm7>#~FM=)UZNb+*fdRkqCtpog%7Ve-4#Xx96ZM ze1eLMi|~T;JAi#5LQamk*lzfx$7Q~;2qU35tb7sHgmn6I(Oe=FA@4T{BC*=9dM(AOzCme7yU$5|>ifNBNeO(ejCfN8_E-Oe7M{ zmYRgTbc|LbtH4lVXNQ!zdGX}zyx_78abUY{hux+euYyV%CnOJU3eGz)H#T3STS-sB zZ*z0RL9`>g3!&eHuc5s~(c9|85!G^78^SYsIz$z6g^M9|B8Pi?Zk@vXyW^b3DyXM; zyeCX$(KDm!n~lTb(^fnliRDYIcXNqUijIQ_w=#VD92Y8EatOXN?%z;B4sFV5k0x%rc(AYW4FHsg+lck6%BPAso0Pd@7V;#LHPSQttNyua50H|=RSy{?d-sM8w5KxyF-~#0mq>|o z82f|yjv7?Ou&5LmtdtpxXobq`@pbm8^y_?9L%!Bd(f$gjOs zau#Yt<_MtA0h-!c+;cn%;*r6ZgZ&B-|~{Mrhu8e&xNTdWGA=Fg-Ar8F|?U?d3V z3UE|fOscLUa!h}L^!x>*Z}RW%@*!$OLy3&c!iC4g?jOt%(>znEEB7&UG^mA@hq^=mK2Yv%;JU88m6DL$jDtJp0FEn{$2pM zgZq6I`|`)Cp7(-z6A$H(Gp=_+vISi`zIxSq_RLcqx%W*!Y>X_^+R>*wSW5y>O zPv}ln$e(08q91e3x&|qn-R2rJbSEwg7Ubt9Ysj`0x5jF|TfWxWY@IURcCC3EU_W)4 zT_%2D;9=ynS2}++tufUZM`mW~o2 zOCrxResr~HB&g#|{K!+793V^~ngf$rR zSfY4sfMFh#%Hnp)lphu5vcWtw=GQRH-yN4l8277=*){dK=U^SHClDCCxVKRrEddCc zKzD&D`@icAko(S7R#M6+fOHVkv*Jh#msrpPFsV+!?zIy7jvMz;?2q<&c-8SBC?Di5 zLSLT2Q!!s7u@xy0kS-xZM0|7p$lcM)jYS$Ex2o8~;|z&n`>-y%g`X*le|q{MKMb{r zWs7Wqk+80#&7?w|`$}>w@>bQPr}sxSyPc*6?84JMSbeuvZ!=2KL-sK>G&h<$OI8Wi zk!FN*nyBM!>R7VT0F!+nkH&CR@C*01$eZ7{HrVT6ent;~JlkP~WH(isZR)o;k^v%1 zr1k;{$GPbIAiQZ0IFUmCZg=WCyNphDqGie=hvbAyu^-8IfwgD`YjeV4jpxr#VnIE7 z70h)OrY4fxRbqh?aHGkZM+Gs<)42CHFKOCV7$V-qFDb&)cY?sY%|eCM+LDO){dtR= zE#Mm6pUE4)GtIoDhvZI!Zs*hR42kA)b^8cQ2xnXdR2|Uj%XQ35xkP|bUjbrN`ED)Z zOI2cuqZsT5%fMKH#s%kWX1c)uAG;=P2|{ioDR=*fc|}NMB#+%svK?ey7tN6qw2EC>JZd@DS{5|vXLA+z*^p|q zmrXWD*y<#MsQF68b3U@v@%4Igkkm+o)hH^<2mxhdsgT+3&E62AZVGI*VBLToGSK>= zI!JPeW_@N)NWg!4mn0h;^`QivE14r$_1I|Q$I07rcAVdo6rK|0ib7^y@7hn^PkuW* z%py`aqz$7h*TModz2u{Vh60;ABdv?j+A!-?Gr>Rdy?q0mT$x_&d54pBoQ-EA^gTd4 z-UkHLkiSRJhYw|Hs(yytIT>2NiVZ!X%<>UVU`4U2qoPi*CI;7KL)!>Bb?4UB&+_c& zZ*l!j`RzV^(L_DQzrt-<(F!Dq70S0rAS`pzUn^i%I>u6`F84gLHM5tdVhmmAd zop1Xo2ftL+bV2=o;6B?YJG$**X=_h6Jw#0kRlsj0mc`Fz2>3~o8o{0GWgM1+6JMGn zs-M_R%i;UfvLt6pndJqeLhDy0Xh=S33Vp;l!6G=s=UNOYD-_S_9UY$q1=P zxc+|=n=8-@`dhfDmlmofU-?7a?Ta*_!n?=09{@jC1gK@_iANO+m+f&>0E{ijxmgE` z3VFu1Jp#Q?Nd=+LdPh*6VYhfw+B2`&ICy%^l*69x2(O)65D0$Ib_f#>k69nujj&5J zwB8mkv~qCMFhB>LiTzUB%N-n~Nx<{Ykc+r+(zOb1o3<@0+!v8uq`Q&9MV^T0C1(0G z=2Yu*G{@5Jt7Dyzwl_A`ug0-*R;{k6?5k9R(ae*Q=^KT}^$+5DhnJbpE zg%RwAWUrdokv{gu*;U$pA{R420A>oZ*ZmymvYe=dA*zG*A9Ke=8mP$M%Vg;61)XMc z)ivS?@d3qrZrliIGrYU-%TUA&I&V{Mx-j+~RyP$9jn?yIS_4%PziJH`WFw#W6{U+S zY-GonL2o$R&a>Ty7cD)JX!-jvXCLG!hU{EMIIHk>e->>@5lcs8+3~m;h|6h|l7KQL#UBwQqS{Z-k%;=51gyCHMWY4Hw)BeDvbu{r1!cQcH%WWaBOiPXX zcCdAn$RD>?V@+PXf4f_IW+>(7eeqT%N7lMi3&8Bs)x+G(wk#XDgltab`hvz%PQ?}; zT*s_bE&+YTY3|BjG)P;@HzMKWw@UpE#;#~Z4jye_chKG;dG^b_62<2J=I(wDF~`oo zL*L!B$cX*HR^j2f3bKnqEXg5C9!3tDl3rS4PHiPY_}FT0-1&77->F*y55o7`)WqRYVr65c1-4q`?rfdck4Ub z08HP7Zj76+H*DTCob|Awx-TxzaSvQxGc$S$eR51O{u*jpuRDXQYySc@l01{SjWcZ@g7!Dm#D@52(|Mg zoGhJk%}00?bM|)c3}eHmmxZEJ4j5Q(G7958#m1@jv9e`c^tVS>e~%h1VPM)MIi74v zKVgY?2$-5pwa#K9pkjxqU_UA1PK**2RiqTrKc^R^L6a!O4?f=Opp=v$khdl{cclW4 zlI``P@O!=_BdWhHZ!F#$Hg1NOjKFlbsG;$@+WYTq_sY~Cf!cR}dt9%yM;kM2D6j^`Hr&V`fo+FNVwBOJQNSu)H$09KOuXjaET z`&mRCxlHQSOy3|s;v+*Ck%}QpYnb-f>bBj9lu%k|HtpjzSRK(l5m1Z4FHHUxXmg;# zwnugeFQc*9EQ?|9)2s{n4PTaK_jK8@t*XCegET40dEEAqJ!BnwCRPu+WPW<+_7jK_ zj?)(!7rG03HPN0qIzU@dB`DFq6{IT(gEd&=wVASq8H$fY5$D<^0dKE zvM5<4y-FGppxEE`IW%2tqK}B%a2L^WRs%Ngak9uLIFa{g5;hdxD95U+F0Qd^C43BV!oW} zP5^3_;vyW&36Z%T+Ri*pp+!kiruCL1LK^-GVML3a9{KJb z+8lTgO$stD-vOxlnGxuAqr(mzMV@Hc#maoBbqFpkq97DVSf zkmN+?^T{nDi0uPW(116#tUCt8LbRJ(&YQ2o9F=^9Y9jUiQe;GZqa6djjX{t{sFKyi zq6#Mxd;NNG7Do+Mp$ob@X#$nmD8XO{u5`OHTecNyRQ^&K(X~TIvw@$~aPMf|+D>YlcffkWH;^zZ*gm9^Y1qgAJ7OrRbsG*F%iiINOp!p=*i=u=O7qIFd0@^QD*g2;R$;~TD^6sqdF|zg&l2AaxojW zQ8P5$lFNraUj~1hkuP-m_Xp@OnKw{HX9Vk96u$MU9d+azaUT(@FW}s27k5@_SRB>f z_r0)6I8PUAtz$cTbd!9$tdJe+gPFE_L&`A%A_6hiucHB1`I1wTa!C>(|Erg8R2jOuE>uC1JyF8?l96&J3UjmU$&>Qvi%XME5jMqfNt(y!Y)tG# zF1|SBv1ZfFiM1y-i10@M;YQtHt(4CNnAlj+PX$PRZ=iFm?9l>?uRE zs_NP=RwiDE%J6J<$JhPzf~LwGO%tU*G|k|(9KsJReJdb3Yz5Z@d0relE%-;Ark<)# z>nc%qEXHR9IZ$wKU^ISppV`n??VPZ65ogMFPadzfL~fV6xBLi2Gu;pzw6&EM8MOw6 z6j;;J7Twmkz@f0LJih6%tSm*5TdQHgoCpMC4Z#!VHvd|BUOr($Sv~cz zQkd8LebB!xl_K$3;W}}8rwI|j`=O!f~d|xAvs`y_HmT(%b}dW`cod4<{PQdY5(F?aqF z=%-V-!>&_|P_`3^5M>vSX1ls`pM2?s`mh@N+@+3O)Zbu#CDv1Tl=#J1+s$E*RiBNy zyd=3LkUmB)_`FfNE;)Il&pb#+8C&X{35+^m8iGk5FoRIFs;eQaN^| ziPC*UiL)7|M1lhmudlU}W|YRc<%e~g3iBDg@goFtPPMszG#X~ng$&rD$@(c+Wm=r> z{#UzWzyDk@*cPa2R*vBoWJFp&VAA}lbRZmYY9qbLsC`gPoD|^K4xRFhg;Xo*NCCLy zz@nkzMT4tS>rza6EMi>$dn1pK;Gn-w!Xh(v`w485B^#~d7=J^Xy2^qmzenV7$lWq+ z9x3oicIHEA>U~nxn~Kczjbc_4q!o`$;$_fpW)b+@uyt+7dLi}=D;Z&ge%7sM6TegK z28AOAzqg1=tDQ8567V-}X27RR_x!-fngU5R>1` zdewyQ6ioYhu2<1wy@nZpT`LH+=pd$1R7RIlWxAtnAh2x)kzJ>w6UP^*;KmB*0I$dQ%D`8_K-_w3BnR8gRrP|#s2VNS>M06pjm44 zCJJ05Z)HWcDy8fEo&HMIXg@kQ16B9*pgL1f_KZHzUOJH_*qOPH^*IYrcE%kfDu}Hn zoOrxteynacP+rnv?NNBqHTrVoCQPFi+azcdYf;aQf50=@!+6UDT6+?m(ajp;cr1%3 z>=MbSQo!M+t}f}nKhk1r$PFl&!lbl zZ_XA+ukIu098k}QP__PxTS5OVHm3mCxrE`6b2`9pdT^WAr4v4i()K~&W0atwz|C4t#R)_Ch526HR?qj zCothV;f{Q)Ym`$qiAHl1Q83LlH*d_%PS;+*X=bX`3)cH8T}V(oJmR2%6^Ei`?}7eRrR1K7kAye?~@7GKBY5%x1;fX zu>qf`N<$>2fQiR?_pq=0-7fnJtxNQw_OjZqueEr1XcSMXYRM-2-+aSrI7oC_Cw}B4KO8;l z6hYKYrtJ46tTe4XD3v>=M1qu`qllwLGDti2oW4Ir*Lqn)))@pTOa7V-uPE3xHz3v$8zs#1TY@NP6~McymIESxE&Q0OiB{cMeu z=viA#A}hB6F}u|Jd#Ysavb2H(V=Ly~mBcr7Kytl0uSDw^PfoEoCUERiPu;=so{5@z zEZWVT!`sU{v8GJ(D3M1LWjGH3u;vca|F+ z{-;U2TEx}jw(tETu%Mp>Obg4v{<_0aAF3*xPxKFM!1yX3vI{SJv7F=aq76|~(Pbxd zN^To`VjThU6!Tf1BUWArB{}&XhDzB+cpBHtT&2pj;Li|(jYwNEPyV*WsHY$%-!JDV zW3E~uBRn@jCzdn7hcu@m*Hlgk96F`VaVCulKDf%6c891)QcYD8?Gai$y)CC$ayR_- zPtM&fwQF$=R-B+|_RCZCg)Z_Gjp$6EFYnV1iW2=I8;LN&A#q>4qe5FO>noS03~X4G zlW27QLSNni`K&8lS?kHJ{?6Cb6W}CdUb0s0ELYV;O?$A!gR}h6vk84u&)W2H{>5>( z2#!WtxYpD}!`&gZ&&CI=3!>R`I9w)f7!lpx``Gi{Co$>q6oFFL-dJF>qd& zpq_2>?ihys7;irG_)Wivh^S6$&>Ehwu*`fQ&whhDC|6;6070_4Sgw5p6RDX%*t`su zvDH699xH-ym5-8aJdGkW;T(V*(qzhQbTwS9i<>}%70Xd$`VdQ8wGFauVzUmc%kkFT zfVv8aL;CVb6KtZ?KH7E3tc2%!X>`Qcq8}pfeYqGtY%Ks$%%O_)XO)BYfRN9!-2rKN zZC*}tKj)Nk+%^4WC$-dug3%Q)VxAAgR&IkZD5w(NyS2(XKwrBeoSZW!dLX^1 zycOjK1>)qHN@=Y^lwDRyJ|?R>FH7&Sk$an5MK@}r){jXD?zP_Na&MLH9puomP#wca zR59XjmjblBdD*qo(9_;M4mE$p50xpzaCEOKKvFnkNy3+nv;<3-F; z`7v-)T40govSLqwbBfa1rAEX73LPhLIzqGncG6E=naS%sX4xrd`}-m9PVD?k#uFMr zH0`WmWE7%TyT+)P>fP}$XtU?^I zA85!Ed!NhTf!?-C`Sj}vT=Hs2l|C`3QM)Fd*c=5EyWnD5*5a3rdd;hcUWYGJhLrSX z`@Cn*O4jUzVT*z$OS*L+qxX+^?)(Qmq!}42jMvlTmu~K}qMNAr>0Yq1x)!CN;j13( zX%w}>c^vjri{uJ$*q3LBQQi(x#BYt1H-SU3FsM^-Q7y_rI!Z$GjE=sbW^wEZBP0zkBG7tjb5Hw_RL%Z_mL#veVrk{&|{2 zxS7@v^R1sDNXb_y%bCU7f2g-0HI^Gm65xCmU?u6E(9`<1DGYh}RcKcZ7?g8B0JU(t z-dV_67Yv&+*sS?5WA-So={-u{EUS1tW)n&;ud$YfdHfNK4m6z>AaC^N`IuxP@f=I< zDg9z%LjdbQ^EqTK3_@qzuj zlT%DI_V)O8Q{=~k4pjv;hivL|a4MEJpoklO?I>Fq$rlZAogsN~0zT-)?9{68iW(Q} ztsU9Wl};B3{3Zwiv;g}JC)-|X^Gd`v9nDSXBce9xscv>N?|dsM*$6L`2_jM_dzWds z8#KUwg))v04Pygu=343;@$OX$&aavX+KkW zjr$dmu9sJ^Z~U?tzyY|jCx>yn&kODR2?LflD9;63q9Dg840q{IdM-PAFs2)YO5COVlsh zwc75^u7ky5@#?mgV`Fr8mG+5ldmbPRC6q0yn7n>Jlu;e4e}4mXAe3ftfVwbpx=a0J z&INY7*`!KmPO+8&9a%ZIchQ5DD4bg!?7ub$2!wiO_>!9x0tOX($ z!nq|4l!(Y><4}unensutE2XP%f{C913#XWH5g74C?~gXu$JJe2TwC8{niljjc)-o@5o?vgdB$t1#r`C&fkh=s|&J{HZRe7^jrXb)o( zcT-E+R;lUSECE5q?&LrrefbvO7v0DaW5ylTrq;w~MsXmE4Qs1BB7*R{Q)0(=%0Jr+}UKkC=k`{_=cB&tLAG!~A=Ri?^wL67zm0~Mim-;p^=%>1VI?tXtsle0a_!Tki>q2jc;q<2Ef68?U*>f0D}w6MWE zQ8u-nfwesh-+~bDg{nsA>`PZ{Qw|7#G5C+2cpt50NmO84X$-Z+pS{LUQ4)V;y~Mbp zL9aW?18J6($Bx>$k? zo|WC{S5H=Px@A>%wf9YOcvU{QJ! za_3afz4PAYY%q{t+5EG!K{0igpf~ax-bW}lym9%Qg4V1(C2bT=>t*2 zIYYupe!V{(+n6>ZX0N$W^!wGnH>AqVd3r>;<|RBY{@DD11^CYvb?_q|f&BTfz!9ZU zP0dI6&N9x{n)|E&Gcdk6^oxD_yWFGL;9$FF{Y8x(#|4Y7`h7aX=hZbLuc270tidON zzg%>%5)TOF4b&|kwC$>j&V%8sIcVoa4Z$WXlj0a9 zTgQ0Yf7YH42P8&x-@_sE%__fEzoA|?UoBi|&HrG@f5^PKY3mUVX320|kzjwt&KXt0 zIPJXOCc9-h>noC+9geFb{Q$cm~yT$cBjoZXsiNURxSTG7nabNfg)eYST4G zd0x*)06}COY0ZX@frgW=31jEz>Tr&6h}fCWf6jgHS8gSregF@zoYn=PFrj^>?O6Tk9T7cv4gRMqH@*YC`__GPuVpM{Iwxv?$ z!*gZxTrdgd0CdS8MeMuj@d$rpF!ixH64|c zU(b`XK2a2*rU8<)Sq@Rt7JOJYKTKZ{e*AS5uL&pmZV{vpxI50-hKH|4h)2*tY0;(a zY4NBmu~SH0Xy#4G>i+oT2U}u~w4BS`m4b)AVE8yETxc?%adp>;jEHqgWO^Tkk(=bw zWQZRG`wj0wa*AxgSDCmcoZ|COO$GOOigOVLHk+GOCYn5Za>3#}oDe(zgKCj-SaX zA0nh(I@3EL+Nk^yck~awlQl^-YqORuk1Z5U@ee0D5r$9j|cGEYw&wp6m!Dd+M? z5&qsKX5yD|IoEl5s$SqQF34W_cC{C5UC|dKxrIy>)Cg~n>%P%;(R42q@R^wTS2+o@ zrczcDIIPJ4kQaj4i&K8!24xIu|Apr_A&W<@X@Zo$k~Cg@6&4p!*k zmdF}03x2^|-wc5=kW6*Bw2nIFJ>A9|2S=Y4oOiS<&N!n}OSKk3deMk5OBlB?#nWTe2i^&5_yxr^QtwU7=3j51>)};tU5Qt}j<0e>->fGoFL-3+38FK*QA~(X_DYnQvYAS(E#!)a5GmXLDv76x=P?vh zf$E6Je9S4Iokz!8>R2AXyajuY!a^o{xjDaRD23g-w>vB=kZFG(^XTd?I74$<4dyBD zgMI6G((%3_mB5z~o%&6;n^Bq;KkLW2*OiL5veBbOiyySxt8*dcCrx(qlc^kx>PJP& z%2rPP+}L8)t2RF~UU-x2mwpp^?R@`pr;xs;(sF7@aZ^vfcejU;LgO}zlx%u-WO`Bu z7V&!${Z{D7+*9+a_|b-SE%mh8ZB=}uGtHWt^uNOqX6|<&Y76It!Qa8aafBe%N}t^X zbwNrQmvd8?t*F_o9`m@rAn0L82u?=ROJPpE&QFWA{HUihDuxzr9v-*WjA|Q-9mYuI z0Jk@w3Txph>}T5&mHREsPFvjE?HfKE+#D(|90`wSAwYYTy-Rb1w&!)ea6_e$A!3~R z21Mkn&76<@L|JV%Z;ToXKN2Rl$YJdfosjKUNTGBWr}@>Cvc zR)Gl-+c#U)Wkj30496P5v|OB2bZsDsC1nUbfQ;JtK@ITh0%ty&h-d8M<=`7AtN$VI z)8V~oK{6k^`WBraU`{K+${bal+*;cZFVcoj>CKiBDII+q6_$DDdQNh8Eq;Du?N{JC z`g1lDQr`UEufq<(>7|H%os8LYKR%2GlNUm>cq z3=hx8G^!6Do!jc>eM!&2cw4IdO7m8{l>dr*FyDvdgqG+Kw>F0m%pW@~Cu_N*w~lQG z8OUUq>=kKEi4`m=_i%|;R*>K_r>U*ck7%L7#;$p|_EwI|m1?dQwh06D}9)3xzi*9Ry@3Jbf6 zIZIfB+8#rJ=9W`jj!1gWHWK$YfPdjG$qghURf@vlrfo!~ew1|EoQ2)3*kZ#j1(#d` zG+sPT?wYAoLgSEZjIH1+CJY_<`Gf7PhJkJfLM0gCMb+?KyUDYyCkfHYF$+}lsJ$k= zPoxR7DSu;<=3V6XXt(oiPxf4C=^j13NCYqmHH+4c+le)7CFB&Q$JGI_UA7%2Q|=rR zLGxzB`YkyO$eh{X6-a1{e(ncryv@MZw&GshlXbAD(Y}~2?U0b>N51ZR zFgwe}E@Xmq-P#p)YrlK+76N~>g_xD@A#cYlc@W811Q)Ztu+l_xrHD`G#?iqOuDE|s zHH!J0s6L(0KS$;FxrX9W3+SMu3@ZCZwU5W<&RIWy7n(IM-sK`zSRG^gr*j~)zfs)~ zojlbPJ`Tc>uI->uLSzI?8ogJ>`z&DWpMm3t)T^4<|oQpZSfg&j^Q7cU~L(ij;6ce+C%x~Mj zOgoz*UmtC_yJ6F+^r3gPj($6s+(G-K&?Wt7&WCxuj&~}g3e7o>k-N8K?RC3W{X_d% zlSn3MZ~pMyPolZ;;BKSQ-1ha1dI#>@2lRqR+$DHi6`HqcwvLqvkSV5i?3VVKD3 z*$)%7q*x-_2~xD`CgJLG9`U7@L)k-DAm;~;^Lh_2Vp-_K1Ejsto>#!JZ-1LR1V2*i zEDR_$TIUg&D%Rpkf!FP3M@RoM62Oi7V^xR!$`fO;Uz8p% zjV57IrYa(?*cRZdJR(Qmsj}(ss>%v4kzbETy=i}jI#OIs`Y0dMA?;-g1S2FH@MXH| zxKivQ*?-M#CEBuiqu+!(dja9!KnOJb#cp@u7FfBE9apb_B>ET5%-GrHxL*7am&+0x zI);!>Pk<6Oz|N#_(Gg&;V4U_M@=#4#Aa_jc+@V=Vf zp%~dXjhH>5Uq>plir@%;N0=$uYg(|*M4lqgKa~8sS;N*QE40nm2%f*L1l(Zno%D;sHJ7qsPt9fJ%WCJDSOC+|S^)LpYA49Jw@C&Isz;AT7x{qp3oKj` zV6Us=_^NPxfV#J1m~+taGkCQjUUTGrr@J*gU&Z7?2=)C4{3TUV>a%-H2vV~nv2M9% zGSqylU$?cbWLynV5&i^m_&M`bC^Vd8Em%Lz5)<7|2YB^Kh=q>PtwNnfJe2+9|D%s``+4Ujt-RYH z6=d*7f$gm-!~C(2INNcJ=MVf(VX{1G7KgUsPy86loOV8Dypg*7(tW9v=^x%q;)6J5r|)P1N%p_VJ$8C>$)pbgOHW!S*M zKKKT!LHxbbtJHC&*1Xl?KmpOM&Qm|`Y+2Sl6kf@Qt$3=u7A-HA0N?a)O^{Q5V@Mb{ zz$)B?8#3ghY#lB%%qmZ93nus*R791}XvBDgGuz&4>UE_sL^w!Zr(BNb5iib;TR=-w zsC4@L!tSxgWQ#HwkBVjUON?8h;2iQ^ea{%7Fzg^PO^LcqBO%t`CI~AJ>+*vZxLW8p zn-{;FjZ{Iua@hV*4ZdencJI%Wz zTkP+qn|@v#D~$2AES}x7ZkA{+pITm>`fxmw8eb4H4ffNHN44kzIyjcwGAin{h;Ivz!i0c%U_RX{JhhyGkYt!LJwl1F zegr-+bh5u7n8YQkB-u1nxi3ActB4z>X%x0GC;DUuQ63*8cPt8CCe{n;m%a~ByeS1&`hrQ8b&Y6 zL(CYp67HidrrFl}I`kA=VW(J+jw`K>UMc0SECKac8AcqD!@Ai0_e?(}a%fqFhMp+N zCJbVRJc7et?;u1+S;VC=QMc3(8WnjQy?cfABdgmlQT0n+Ne@1=zjE@kW#_@!W<&)x zg=@h%TFufU?{l{pJ?F%J*QblB*#p)JGbTec^Z1R=1S^7 zmR}NIHkXWxxn*}z!26i;G%pS95WPURGFD|W3bW_1Hi6e`G!Z|M+O$b)HrzUoOR>NM z5S*y6=mBdtQXRHBliXB+&B;|GsG4zg9z2Te&{=Sgtlz9_Qv^(be_)hJCYXs%3%;R_h& z)y?OCyEo1aQ?BNFrqrsMn$&0!n5WSn2P=mVor%Km#+hwRBJJ8mSNyK^W-RO^D1C}} z35N*ZV%aki-Zja*n8MHKEHUA?%KYmU4;L4Ywwmhf-+I?Bzg)`&f!Q3WRa4|jA@aK{ zsO(%?mO`hOq58AE464uzk6jNCau^;n&-_5528b?>l8u;QLS*!QO`oWxB;-(opJmc& z-EHyOV|(NVy%Z8)$_59$k*dxQ`n4Q+jT#w7TZNLb_JDU@crj0l*DC1Bd+rjCOwZ+W z`6(K6-Oxu%;$DRpaNKVQOAfI{mxvg0 z^Dz^hr@$did)|I~{XGfS{nSwKrH$00mP^43B)Ni?Ho&f4(hTckGTui!xf6OIJ*7V= z_WE^g>9^{e_-e48RF+k0no}s53$@43F8t=g?qa>AKZG&RHkwjt&+P&HIJoKsh!td6&aI;S`Z7z7RP(j9tx(-r}(IX zU@>HR;*fseY+C5Hn$oU@_NV#q&2n98mg#Hhn$yVg#B#FqUNfhnb+1Cy$r_5 znsrJ48j0RNf%nJ$uHY@FL+%^uWw0nta*fI(*D3|kf&QH9asyAyheR{=AC346!_$32 z$K18135-{DDr1q?q5gd%JhS2JG=~z;{am1%i`28KH>T+EhoKtRMa2D@ryR6sV17Oj$Tt4?_!P5+>V0_^Nr`9 zrLj)MDrDEUB-KEzj!7cgc=jfLZKeR&Fqf!7ryRrjp#wNXZ9FW&OaLnp>MbJE5$zHy zhqWMXI-pLJ@e_me1j2l@L@3ZTaN64C%EQ~nfdj|3h=;PI=Jg+P^;Cow$L`e3O97lU z7DP1JK|z%D`DS_ct0u0d%#-COlK*?%b{H7r)2<;~J{?pgb}_)=`S)V#!>dAfrsZPO zG|qtX2simGa$ZB3WO1=y{cjcExDE#Kho~j=^N~@T19B67wug;Ry!jCRT4F? zK(MKBTp3%c-q0~%9)>Qsn^p41;(irTvy{9Gm`Lnr#||Mjb}7aDV$7({JO=`8e_E~I zD2!vjOs_xkg=i+WKa};Uj>Fl>m4(oL9PzW?G>PF`eo|Yo;F;XR4c5?Pwj*tT62*F+ zF1IP=#h?O(e{NZ@?U2&7i$UY@w+|_W=99vnThQ!HXbsLSQ2uf>4tCrc6gtIA==cNS zeqJhCr8B03^^6W*^jW(U2UqKsAC!5c>VZ{d4dE_GGBuyyS1FY~g@n)ex4`lI zH~f3qAqCoxW6ABb(jVd3U4I&OAerE#W0?v>*O#Cqh;D~A0$r%zLXO(*bq8F+d%Et; zya$9pgSF?qHi4#`e+{g}K8S?yn{9RqIX~Xb@!6}ZOyanEpj1_kyym9wQ*UGp;i4f- zT~G6!ACKZS$lS@Zegjk zGS+0>_+1v`-Q_-sM}2slraYtVL8^?bLLM)a+797PH{{31=hJm?hr{W~EcCF;gNZM- zN&69%9;83#4)_vVjID~DTcr51KUpn~=7SNcpYJE%z|JwYi{|=D>puToiPg{BN|e6X z88fdYOevSPyg8q2Ac~4klo(mz-1Dx=zZhAsp67SU`eG05sV-rinK?)-fhQOKxE7iw zUoZ&@L_#T_$KCr|K@wC*PNsK$j%pdkhapjIhbJ0Xv(yI742tKqZF~C&lai$;0f&vqM+OEG(_|Ew3|_cM^jUe$ZW(*qztND*?*Vm zla1AEaFQD0r-zkdp2)pC468o(D^W{ev5&{?#}MtDQ%7;amFg%8UYNngt;w4`Zzbp2 zckag%C{Ltl`Lj3?wS6?8z$_i5T29;Pq*yH9$v2h_i@57L<1)|^60EW)CcCoI?erCD z)>N6|I7UUQPII+HY;MKhT)cf2EhXcgcI)8i-(;cp<&&u*KIP8Hon*j3s5u?9g z$9Vz4U&6K&pu!x8$((m|`t}>>QH0c!_TYCOOXi4OXa&Q1U>3^VCtOMFrOCr}er+$> zO-u*Pru6lN2wLi?-ny_?ls%afA1N&^WWWi2kVnesh)O0F>GD%t1OI4R&aKPFhTa*o zEZB*e-w?I4<3XBg9bLorbB|eHQH0Z*$%pWTT;$iXIi|HLftiXIU_v^uPS6FclfHOy zP`*XoX&Rpcm$jKD7QetcOlI<9=qu#d>QoO zI+v-JkwAP9m<~tBsbM41U-Str?QuG_7QaV-$>`T7nH8JBB`0P6s;i-NZ{A2_va!al zs#DKMHD=k@)vKTQEX*y<2l*vAES-wL@cic=dQw+P9WA3o(4n<&`)YhL`Zit_@soo= z+8P$h%u?=)1k}qpFBQ1(0Ur|;51(OIIMGrf^`_B12oOX_;Z;xSfv{45b}_HYSN&eT zs--in&c&Ql)E=QDM4+2wxYlK8K6M$M35T#A^hxwmO-(JZ($DYgS4$4VA`#hJh5rp= zT#>{2IVN5VJ;EwqBYDe=M;ZaUGKEq170GyBfm&x#*6VH+D=2fK&bSGhoqK-6-Yr8l z*t9yfBQ_;qrBAz5isYuo6&*ikCQ_UR5!Mg0{?7%*a)Y)a=-EfBZnz}z@)#+ zVrd~_Pvp^+0fw0C4{}!d+4xzhc&Gu*c+W?|u?P_)K)5ZA6s*>(V-vm=oy6#-ne{uS z*EM}w!CSXd;S-v}PMBvMtHuuKohZg{omXfbLr1N;Bq;{j!|@zSpUcg}M@@k?r=){9 zBigR4HPtH>XeK?mC9KLhMIcV!ShmPk1Zror0ITVJ(jWMhm=$JK;u&84ctReCDxJvk z1|SW=k~j6q12tMzf}>s<1W~^o(2-c?u;LBv^uDor&5rdPM0g@(d$61=E7JB+n#ezb zm-W4YR3fTy*FzH7WEnTcYD0)Ug%LZ(XYg)I#ZPOPS9ClqH_ZwNg<3T?> z6(mUR`PD5t7QG!#^5uzsBd*`o+1Hhrw9S~rH<`eGCkO4Or(61EPv2w|o^$(<4T=E0 zJ2@Xn@yuMaV%v?8eHhpRmmqPKIWS*f6nwXaXN?sGw1PY9ffGt%6_Cd^DNfdP6^HTl z+brdK#lx>hNFbzXzGA3?+i#qMjsbsws3Q6#*>ruy3Xs6AjUw)M4 z3LX}-qQV#F2@~N$s8Fs_%?>k@KMyKYSKK8v|dj}ox0MV`kZ^Wg{9~~q99V4KT0O{nzi$J z+Y>*;X}Si2IlEr{Qh5t0Dj+wTHt67F@Z5TteWtf3UInP5Fby zs}4Wr>0+R8nD0+^pR`4OywwM9NNod0B()~&*}|}gH}B1r9G;kA665Z1Fz>>kywuC4 z6%6aV=b60d6_<{tvvNuA%$x|OoaPLXDxu=a#`$ghl&dYuVO5V{9nx#=@b{d~J`b>P zfX$lpGVmm1|AT#c(K7se<@h(&S+W@%EuXT%ku*_Uf!4k)=Qq0xYijDY{qW5ddc1S@ zNb%k_OKMIA-Eermoz$NqIY8ArD`EAuBBWi=eLxU+InOGJ51Ylk&*WQXixvXR>qBI= zg81++f0py#EkE_?Ssh}9{$zHggLa`zhip4i$*3cF%m4kIP7UrdZ#hx+j_|hS9th#~ zA&p1}tZbG;f)TV%TzJYvry+7C@mmgV{$WNvyk>ld>BVjWOrBMklqc`oQzGCsYBNx8$ zP$|gW5~S?U3K_I5ua`}Je-UMPT*Pl~W#fpeKNPiK2bqx>Sfb;%H;DS)a0CYV3U2e$ z;wwT2sQEqZb!BJfwHAqYKKC#zp?W@|=PVYo-+U2=mV}E~v=bcaz z(t1CCC;?e_KhU4YNozX#Rn^Km76Sph>A2lbLEX!`SJjDpA3LpaVSo#Tx>nv+RK4;9 zEavM(Q`Qrz<;*2Q+}C;;Clj%ryIp@!0zOVN zKf6*(a}V)FW1I(P)nPWs4RR%ibo$W!kT($aD~27%S(nttyN;j4zCr4W3(*;SGtk73#HVPuE%oT|3Sxl)UfIW-_e=|@uCjg8BjZ2|%u`EA zx;f3}6ZlnFPVIIqs`x1iDy`@VLEVargu1NeP%du<&u+gBV6H@yiE<}ab#5#*3<0OW z&M~EtwLqow4V{#~;m?^-SCP?P&57>h8U%bd5)Y zHFaak7d^r$Ykh!0_m4N^SP}61@H_ps3%()md3k%;N9v5#3VYe+8n(Mwj!>|8F{xgX zN1`0&A(4T>*L?82wT@M`3_k500v3&H4y~wI$VU~8_cb#6g_E`6QhK8Z(#aqamCGq%39XDo*u1!UQ`zM{8 z`Po^a_~07g8!uGL3h9c_K+@A_O|5I+iprRB(M&mJ5Fd}Hr6cjDk%*2|a);v4KwB`z zV`?_JLUq9QUi+v^Um=?idRl6IyfZ0w6VdmrB&vfvN>#rh63U*PHPNz-Vj2_2nl9S& zdF+xAuwbamW@(Si@ToVCxS+mF6ICxkniQ7>5e-u`9I#a$Lb)1wQ*+0*j`xDrlgJYW zeF0(3!$-LXZXCvoa4n`dpWiqg^vO987Of6x!fMf7DUAo_Zx_Z+o_;;T@d_9Cxhl9M zw^cv?c!JoA>TH9eZ=@hELNP_~Mx6^pjIwE}ZwJl}%tQB%z->v2W&!6vde9AUa82=(QA%hZI^PYYDKdn%H_! zYrSAdjHH&2Etg)%OcZYAFw<6Rz z`ue4pT;@WmRS9}_geQtHqXPD`SWBUvLaTR`!P21v!xFJxf4YC~?}Fn_X&& ziea!QC&(!^tvcPpubJovDRf8afo^PmMBrVKvoS-_ygO`#2vNu7fh_4Ru>*I}V63^( zU$E(rDlrhs;vsj3td^{2h+-;C7iW(huBT`$CeGIS^|az`Zojyg?n6q*+v`@TAy+=KS z^v+!4){Ji?ZbX#85U_;_rZ)G}#S1^W_H*wI({Sy>lktD^3z;n)H#H8^1R3m}pG78q za?Rkg-k3*K>kcBw9-PSDEv8JneTA<(U#?gL@(|i`_1JVbL`JTBm~uuGGFBwUvgNd7 z{i)U+k-OVud|l2>ef*H?P)w4E^QN7MTustLXP|t`bOPk2-0f^ta0RA26j?`UA=4uH zjBH#N&gakecY2}pA&WH6d{3sQMS{i`sZ^3yH5Y6M5k?J#eTKXIE5Q6cUv(g*ryW)k z6wmL)#PwT&N)=Ut(n@hY#!|xRzZv?{77CV?Di&C>wdItXyl4fbTnk*>_PAmU)v1)~mK3PBz$DY<0L$vjV~VSFR{RY%4V&U?nL<}S&1LXi-#7<@q;ed^Mln>+=>G=% z>WEP{G+JR~>sVv{e&ed7(uQpxI^QM9zUxXJ+PJ-=_fX`oeWBZ#lRVy+(c{70g)cbh zEjPs$*)$B}ShqnOTx0}5@|~Y4#ir?s5eMnB_njcw)0iFY{-j{<4-k~ytpbNzH)*)G zUXnIpa-$XxRwpxGg!@IUw89rEe~rFEP)ZZ=r>16Zu#E|rSJL8IZWc%#fTow+V-Isz zHR$+`t-LtM8^SH0gAQjxN(3YA#RehF$|ceL!YDCU^ntI6`bp5WW9?6WRU}0)d7h}I z+Zy{}HB#xl$Y;0GJv1bs^jV3+=+~lTYiEmpSY621+DDz)3wux7)Sdd0a%C=8{jt9X8zs|1hSF%q0R3^I3m}w#ks}b0qtuS8eqTbjmNz&|2C8DV0Is) z`}a-5mhWwjDCe#PwziEgs9yc?CRxvFG#h|E{+&Iq`h$6T9w zE;U~s>N(H@_kNat3p8gAyW4TYRPsP|pDF!$8tVvccJy$k-lv4{%nU{YbAtaJ-*VVr ziwjnDdWkJIU|%nf)rb1v5SKq9!th~N4wrhC#=Fe;P|U*=gXL$&6@KrE5Ov2?hOa1) zCZL%8qr{eP+!#Fkeq!i6iKHH2+B2?T4chov0@d*x+Ai_kvsD&;C( z3S6WG2z9Jm=cDIRHm#>=?GGT)9utAo(Z@{Utn6dDn%)6bm%2tSQw3jEHW}Z_XVN)k zu}Qc*(Sbg>fF8IDK~*$+QX{wu|dE!c6yl0;BEg zWsE3ktZ$X0#?G0%N|m^TX_!8ZdLVSrxjH`v?<&!ta_4KMY#1oWxgzhxpopL25H(2o z5#8rtTU>Rf3bYY*j((i5e%D5jM?x%rjMjD)|ze=Coc) zh)NWII;9yXXG8Id2o{!p&bwDB+1%AJ$v1zkXuvi~@JlxD(P2(Y8lfsQR@5gk4Ib3G zw6xM^B5uF$8p*?UiH{3lIX`)xO8qqVqn+KzJXnsAeS>Xw$BmS2rG{}EDsbdEY>Tt? zm0LoYtXPZ>$z71MGG@LPadV9N(E{hAWe;0DuTee~O}?O3GV>GY+9UV3n#uSa6zyEC zuRUwj|@-&QcDf4n?5JffH$h*eOqc%@C$$k*Bgg_K#+``;j|vzJ)C%#3I~^rJJD8pB({bQ zmOMHuO~u`;{tRx|o*+Qx!s~WZLi5!#U}-S(Sg4n_iZ5$NIOhLU?9wtPG}!c8+6m$3 zYP?r$hI1zJo$`aP@+N}i;V)r=&=FW`x_kZeXQgDsInyU{Svyv=X#_B}(43GZsy1~v zsXXAumL7`(_XK}U$vCn`>(k$we23n_HQlJ`2vIgxny2UDUx|aIbjQ+SaTv&Md8&p@ zk%nOxsvMNFHlLyx_w1_`29Q0;ofL~-pNLs|5sDZUTnS>3+{OB}w*^`|R15L_-YyV6 zKd_4cmQ>CZN$`V>daq}T%o2}WV>&ce$AmfJZ^xyU$D3DIpVF&eS+r22qDj4tLIW(z zA~V=)U|$9O5%jbHLiXmn`7i?*CRG*#py+cgOG##{;5I`ju@5$%#8ZRhQfmD7&bFL3 zGUQNE*R?EdgmF?4Kp->t$i;Dp0>h(xP1oHGs6ScKK(XzcO8Fi!*1rUuk%jwV7{XAS zwl@u}7)l1?{FaOE1Wn9Q4gN(DaxsmDF4pMKo#X*&E$n4@2D=SwX6Jx}4{eR+dhpTJ zM0Sd0HgtbDyThaN+^n23CssMFC7aA1w*(p~YydZbCgw`Dqe)6Zb~Gcwzyj zi^(E3zuwh)vCiRP9?^cFj!w2{nC=o zunB~optM&%31$Dn{9_*LbhpbrtVO(vX3iCrRbY2@#tZrl%@X#Ucs^h1o-gHTqoGGPfN!AZR|y}PiMFn?E=u@WO1MFxeX|8#R)>L@*&Pc((G^Ih z0&_f{U&*+h{Sqo?XtA$a?!nuCSkY_n1Z z%QnfV^VDC`Fr`+uh{3=E$Ixq&1nSVZmF_RRY?Dvltz5b_Ntp{>@FJPo25)U^QK+98 zmiV5Z?h6Ka&A5l;g1fL!b-A$UxzD0O2A2v|Ifk3-WC5%YwvH;7!>CcfbVwI?QZFOk!2}T{Z*~ zt8)K#cGw*G$Cj9OB54F{VVwoK4+Pz z?z4hPtoR{GK1pY5rMpq+X1#a)3E#_`Jm@FY{b-=@%M9ZV`Y$nco??J=XA7Bd0t&Ng#lWvy13CTydU#)c3|xuk5XZ~Q>JI8dI0ly!i#Wsm z-}MV&!rL}~E}s6I1U^bbOs9hI4@|JuZP4F|HchoRNA`zfQ;N^f4?RcjjWGtUDV$J70l0pwh~ zN%dEp%x23o2TC0bz6EyG191{VZny-}P+wBcQgF#IMhz zyT{ixd`Dl#&Y?R;FYFYMo(U@RLw;?hWmCM3fRPD=$@Gv=A!Yg*sH!R`8n9-hIm?*) zlIP%@+$u!4^n!fMT#d)ySNuL>hQ{UGNak#j(I8l(O|vLO;IU-ScZmWOo>4(0{2t7T_ypZjb<;=rma1LZSq;QeOFe$zoyz1>FHvniRh6XM2K zu5l=?mw5%e#4)r0*C{PqiO%eZ&T7SQJs72m`j;@nx(ZBhU@e7qLgm#{kK$y~x=(Rh zD;|la6>sZZX8_>^t0^nzb#9Gqd6U%!;VTz81*1Jh`0Cf@`Y(6*kQ6WIxuCvE1yk}y zkXIQE>;C>2Bu_0#Z4n>4sfDe|Ir|tmhuE z3TCyFkW)2P<1(|~lUW5%`dR#3j%G&dnD=2U?B+GbVYgMToj|-sh?xY<71I&rAnF1$ zgS1TGOFuTC-ef*6ItzBi`ebWTRZ&b62@yxkmgl*~mHdqaaiU(0HUKey^)=dEKmQ1I ztfE!pMD{)(A+0#Jueb@zzw6?DY}s`3WHO#-TJ`V5Shus+nCGaUR-KNt5#!y!h0C)d zdRPLTZ3LYkq9~5Q#Vp6+WnN?J!%(F1p>o`G(H0pl>m)!m* zcO3~&;#-x}`#Xn3V*J??$A)yW;XpYR!n3cJV<&f^SqJ2qoE&|ZNt^nRBmESED%%Hmj1XEuok4jL^*X$JO@i6SNk_}(U*>43EE zFoo&R@14y#9R^Qg`#KIZ_IjG4d3Q~`E>grMCsFXuxA7-sQbOQ~z)x6nJM2U2)Zl0@ zWkzEzvn4>4-mSQcgPobozuZ1L5}Skc$k?zH}zuwD>|kV_!aLD;C@1D z(3ilNtYOSyD_p}hSS+%uQN?AXA_+|EMfir+gABDUU1w=%_C>`(o^~O#BfT$NAMlpG zJ*O$|HV~sCZ&B(W?AU(SN5%Sx$a*s{FMByp!v&r(*=FB&@RR(I-n>CD7{7XHj;FD@vo0yu!hM>|pOWT65v z9T=mBiEz>wsm|+Q5F6~CP7ZRm&NgJ4-`t<3Rk$@qjugDLE*8ao3!i+GAR86f zY&;}AIVHa_7JNWvYqMJ~r^UsRVR!@oKH8GtR%4OT1BD?2K{wTUIuo}=TNHaDuKm|J zqL0mj$TfrX1LLTb;ml$R;#qCnKccgG9L}NIcFueIp&2Tfy%s>s$#Ijls*;kG%BL$n zjtE+6;&DBth{C$GX~kv*&- z7&Eb+5_-y`uT-Xl3tID)2PMo?mx#RD`{1Rwj_gXF)4Yztc;Qj1`%0|bYodTfJlTH! zoRk1OK*PVZy8H3RbjY^6Idp@h8~v%*o>NH0LI#=;7Fo(Gsz=y3`tgV$b#x?62N4p5 zec(>HoXTnDt6#!Kd4f^;I3_*jHnwgu;&rGHD@>)G`?bH|-8*fgX9v1b#ZWF%MG*Px z*$l?LD*LM->OMAndfeHzSE9eXYP#dfFu=ErQIef#(GllcC_DW&R+xxQS3O59 z6oDEL=92rN`UJaHb{)N_!RNr7jwH4aJ!9|E{-;v2zuOBEXhW#%!kF3hlA>lNIS@Q; z^VgxCQWe{xOn*>N4_Q`>oy9~dVlxAiHv|1sdWyR?b>e~=tmGP*zs5Z6vJqY0PUq$0 z$h^V7;UHP=u35+m5#uacH?@Af#SheHuisKQs>5fs7Q07Zo#3|fL#avMj;n}c3DzFV zrbm3ynhc8`WF!0a=6mcqJ7;G)yn8;(JAz)L1ueWS8^L-+V_(VDQ`K4cll`B_#I_N! zcoXNLMNEr_&CRIJg($T@ty>B;W zhgPGoCfHl&m0PLotdun}H*@+htxmGEx&0%l!HDMa ztUa-IQRMeH0~J#m@lJq#NWi3w)R4Ubn~cGYXt`V8^gTaS*6kbF!?WTRk?$K8PcoH2 zUV#O4AUWkmC-;h$jEQ`$Ak?oCb^CIwOQc?Ow69tS(USzi8#hP9uF1%PS=PIWw16xo zc_n`6f^U{uvu)i8BCuEt)rIO8X?aKiSw*B*@aH8+QXE@Rhueu9)j9zHfw~=CFKfYh z^A}nbEZ#%*|An&`FGW0^vU#$?>Ns|M)tU<0UD3R<#NHeMH;D4rCiE&#L_F{%BQv@$ z#3|q-6Mm?+<`5AXRy%4;V&}AM2<5ts>4%>6(mNmOOY$3B(uBzmKv5NN>1}OcPi+o_?fcRp(=4_UiG5<3{Nz)` z>~Z#gW?ud3ag;N-NF$&8%IUF>%F6_s9(jtNjgi94Kj1?B@+cmU@M}RjKBGAGbl`_- z?7YuKV3910Kn$i2;`NZt=_|6!>AJ*;u<4?1uYV?`CeRx~-HJ*%_nPDxSjX56dCsM47w>2L37zvqW)&%rKp0y-+rfFfyFw zLRhZ&$$9K%iM?}v3>_WZRFYCx&?w-CUN5!CHs1X|o)r?c!sh%N^mN-sx2vq~z2tTSZ$Cv9Vb`v$JgGWu*q+y-3CI{U0K zByW;X{mX3+wY=e$y2~-cJR6+)yEZgWsH8LT5Mi05*iCw$ZAR6-7v+#wB%|G(H~`DVlU$Wj_PIzBtC)#E#%}pCRJ* zE*;w2kUf>w;VR3H6)CoVPoB#WE2m(qIH^C|BHdyMA-^ud@pl-$(l$%7l+E*)Gr|w6 zdutx+iUz#%xJOG1={ zIyMo%VqiD&{v=iGJkw=M2+nF`q>E6svQbS4SKgQ>Lm=9gM}V3J5I1C)Qjho(x*5g1 zVfFB;-KthC$|;q@Q#_R%TO&kwnxdu?(5Ok2 z9gkKDwla%78+uIL(ozPhH5afH4VejJRDlAHIaO+aaYg|T??xjNW@rZ>}-4V?%gmFgKq`;Vmu_bb+ot{HrJR3Ye`uY zqZh6obV=Crlt7LF-QDC#{l<(R^L!!6xngvj8`vSlA9>Oqb_gnig2b?-hHN_tEq?PaVHd z?*JKV zaT>RVY3}hIB}1gf;jH~j>zkiDUVP)R!C@r&W^#>)#ic-zdm-+4bUQ=KBn~V4gU~0d7SHS~v%u z#6mirOcMqJ*yA9%IyrJ>M{f!dIU+KmuAk2{Fk|j;pTW3LP*jnq@E9s{8?M~5X1aWhddXyR_Q*s3-U zEY}Yt-ob9zE$$u#!MRN_*nsxN83TpKLu(@wD#pIXjZe@x4v{t+*ObVYenbluw9!+) z3~Y;jUY-y!a6vzY`VSiEL?RjQ7g_;&$oaQ4CnGqu22Phq7PR%9m3v9tPrDfkQ9Y{d zC2<8I2D6IZ(+1UeSE+dM8xfW|Io7Rar-GDY)=z`D6A!f_xM5mRX~z{lqr~%d@LPV1 z7*l&xjV^*vRRU{-XNIq9pS~(v_lz_yBe67lugV|hPJw?Zk%pp+Q7N$!CF69kbQR?p zA}xxza^|gqpZmk?Hiot|o*Ayb3%OXLaV|M8!>ct|UM+1#?Z%3O&*ttYx7&rBC`UG6h_|;Hgnp0A68uqBARPoF^fq}qBbIW zoioZ56lXA&#J|RHS|{J>gBwq@G4YxY*p@K(q{n7M)ev%NQ6tMZm9{&nh}4_W?0gEQ zsod0lO+kOXqjizg1nZz07~eVFA||JrhIN+I?)%>Dsom?{MSPR4dv`W>-jHI4BN#=m z9Qar1YPxjR-Jb`hvUS)jyMD_Aa;O*(Neo@tB{AIJ>^SM5pX`V+8SHK8n5|bV5ge{c zCAKNV4;wq1&(cq~p`0(CA0=CNAywi%srw{oD3x(_UN*O>1=)eN_8T+kxWj>Jk@>|- z3Tz}~tk@r?DXB_X9)afLi1DoY3$`$IG`$r=X#hUN@sDao4<_ga)#hnU`@apazv2n% zQ{{$cmYOxlaa#Y(2tP4wL9MltiQSr_4dyUeJJ5zIe4R(EfLePI5Lr5p~;{ z3{4LuS;vQS|L;rnRy@L4?+{d2Tivsb5+`gc?0)eXauz5}rzQn0<3+;1!mdHb9IjlH zuCqfytBDqcK>6(dJbukVf)aU0?=p=M`(^W43Cr!nVd9&mi$()Udvxwka}H%rMqF2p zAQ;F05_N*Unyzsu#_Lw^z9LA1%^#TeNEjbtly|Z;pkiN!@VLggb=0Ux$DA*+ln`bZ zcgfu08x#(fB8C2Qh->=%^fad8)yTGt_xu-ivnL-}^J+I^1md#r7e|i!h7tCmCBdbI zX*)&~EE00_%_+D0u${z(IUlw_7Vd*SYP$L(sHRD`r2c<{x$P47_5F$4 zyYklrx`wX6yx{Cgey2cklXwz@3FIc3r5`9?lpW2SM=ZmnthjI-?K1kcl$4Coo-HH7 z!Tq%(S#i8?>UCpo@}~D@>sK4%3{}J^-#T?DjvQ>yV1XOF5A4R7MKYM`UoQ$v&d~Ws zro^XZi^*A0ol!1l2!E+m7znd-vDEGH=eWwu!C)Nn=h<4GZ}MP_uFYg6PeKO2Re$oy zDcB{8*-b<%5>U16g!#%@(|wlwUwByD$hVj?|?63=SQlGcLjjzFq;g$SUvQHJlwy(`EVg2A4^cNf+{dK__|HMot`%-n@ z{WdUuDu#H9CX?zy!SHPVh0$2Y_?`>q<)-dpCwfdo0JhlAh~EBvY^8hmSQ147?2I2A zLz=l)4GSaA1cW6PO8+Ro!wQ`30 z3A%NT^j6y5qkZw+J+4;(-8FLA@6g~BuP2;;sI78wY7vgyU>wXPa#pqGiU6cJvq+~{ zR#I1i&T?YQeix!Gg_(bt=Y+J=rLAP*4=zwz6D)pO=sRV(iQI&yUVtE#;b2v+yOGRu zrYZP}zF(C%F`f#r*vX&IlAfSjbT(_E&ljk*`<0A0}mi_Qh$ z7#m(Fhq<-wqJ4qT2@G6Ihr?oB7K>a}D!V2kk9BP+uk9qtc@=kQDH1T>#Ce+4%60#Q zj5+O*Y!O7-CqLHQ!lOaoxO2a0%9R@l(bE)xQuc!To^KmZkDI5L^0{Xxq(TN|@|9i7 z&DhJwR{R(rpJrk9S#+#=CUYH543EHW5hbZN+=kYK$HKY`>GNwb?_(Pe zc_X4zC6eUONi^70hHR^ZaXN{T-v>8hoXNCGyUy2H0;i;Lb3r^{w{H=}=#eh9sCD|> zC*>yU-CpC8n#wlsN{7Nj{966WIMwQO*?m_(ikLdjS7U@(H?+JYm@$I!waD$>ri(9> zyNqf@1f2?%4`%6=tqv{^3h;y`T9f>f)Yee$J*zXz*>Og)<%}C4AtiFAiPT)y@yv_` z!B4-q4tGojO&}njI)2xO;Ea1KX;YvzJP(?o#%RI^sPW8`t*$gaz>V~O3`D9=ZUl~M zlOJ+6{MWAUOI38llW@ihZmN1tDl4a=Z0PWFw~}@GYgsYO34-{@?P{a=7$@tDw-|ApWIK$$VI*fyOovQVwB{eboVuWpTLJiY;#}ju_49H)K z2mt&>QYaXi3Al)wWGjg3H{Tl!XQCEQQu=)w92X&-)b>)3OncfiV`Q36Er}MjSe>K@ zj8UvxX7x$Y^YN6IVXZF@X$FPMZL_XcPm@_|vWnu-jbTCXu(_v1_|l@B;q$<|xJ6;{ zDcsKMWCkCG+{2)N?i2>sKQ$RQ2&=1WsrHcc;fJh8Eyqt)vl6_Mr6it{q&;u9amF)< zb(IM;CEv=qecEf)M8}H}tajb6mhrl_6I~BmtFQ+w=X)EM8c=E4ru;~CC5c>w)p5x` zh$A1;UNx!0;2a2X!y4|AoGNlQU*OA;7A9-$L}xbU1q<*?bG?^&?74RYhLHNuA^}$& zQkO}`D*U*?gPsNZC{R0cxnvXScX7TLoG{3XOix2dyv z_(_bQF*ZT1RLedYk0*ze>pA`MNTW1^ymIUd>e$h?J@xp!@hOePBlJ(v(h2zuYb%`U z5L??ys1V8+SSWV$i{c+&@3u6Jr%jB{4}s15mluW{ezxpzp^7=k`f)a2!ubBtT#(-`;|D$#?eJ`E|d zhWQn(7j6)Ne{sl@@zD<70;jg8$@8|QrN+5bJ%-LUrHrPBB{%Hi%-`a#kB!oC{+fOs zKZAPKL#1mKI?Dd$wt^pbL)ArO0GX92MBUkF3ohie$i3Z`w_7Ve!LYrwTq>}nV1yAZ zeA&EHlMiXvg~bTZ3Q@?b(1gI)d*>V!8<-He806y``DdIumCR2{v%NY@NJ4=8s;PNO z=gpWfAXF>;&Ye{tWTu855Z<^AJ{&FJqHX zdBWTv0h8bE@Wa?>e!zx*;h~Of>UXy1(7uT8CZ`&dUgE>XYghkcnIqD^30uK09uQK9 z=rT)VvAD}x>21*7m&;*f0D1V4m8Xmhaw7bQ;J;{t?k7!&*=N=(%IO5gv;#|vBO@hi zdk6|oQ-!8h2G&xC!-xzs-h8U;lZN3l-jiIYB9G^0KV@xby6FTvO;y?_ujB04uBUI` zxC&?OLYb2H^d~>D7~5s;*{-TK#w)hOGW78Uf9nJ$D-`w7CyAMkk{_a3>u3Rzju~=s zc99d51lv`W!cIv_Ov3+jDmJyxRh!XAL8e#VX_?DOnk1))eX~q1OEyHHp->h|Ta$g= zoQC1_BJ4+ZaH~Ro={sN_gQQ!BE3$>6O&p;1qkbK~pWo>NZC0#!>B6pQYy(#p34y69 z)jjmVnO<^1t8MU8+xvgqlI1a&j69KPd1KLKyzwCDJp~Xuj&6odzsAHbhn4u+VYqk> zP7SI&6=&7yg(WepS(T^_%g*PQZDP9w6H0qO^mtBEPENu{%!g$fx`F|6h{+nXCuFof zYVo&B4~uS8=2=+y4!?_zSFg`)t2*J~m)!iRI0gl&bETIs=wo|XKV;71CgTMQdwL+c z)#{V~Z<;K*iZ|30qq1({bhhY6kHuffiNW)rJpN+uf`w?)V0Sv<1uB}|$;2ks93j_# z&geO2*t<~iI87J(pl-{vIMwPXzajrP%oJXOpx>VqSRb_Hd9p(W>fh(CL3g>(2e?j6 zK<%5%t_nGEOaf$;9|Hz!X#y!;I;DL2x#0Nvh5F(N%gHB@IK4AktWhZ(7o3N3u&y(i zVj&|H#ncpqD3r=C*QB3a5Ca>zlV@{Jv6@(J z$VgJ2Dw&bNP5Tz=W)!b=EIM%Nbe38{E5Q=BeC=5oqQgnLrCLuvqc~Em*V9mmg#F{z zO_i4Dw|t1>_UkL;QQ}h2v&*4$q-gmI^a$xMOx7n_AaAtgF;;6=;g!>SKq|Hg&WYio zlJQR%C6zRF4S43C%xHi zu%K&aG6e`j>!bEFLs?$01_x=(*S8_iro~-8 zQBUTM#dDbn&7?PG^i7NssQMFF$uui8%R3%V)E}Mp0T3@ojHOulm~EXiJfv1fR_{bs zm+EP*wMNW1ED6ZYB=4Qi295kqrkJZ7e434RBZ^o!UypGcP?Q8Ks0;<1 z_(l}$28LkW^<#E|&zX7`H<%24D~^NYHpy?vN5jMM7$ZAlc}R!Egg-8E`YINx=~_0) zguKp&N;;CREb9{y-lgdh1=zdWL{-$ie><;%%Jao0HXnWV*Ez8NBq zNO?E+Fu-qr8`T#LJLd8%e*kNK=VNIDc!i`QrdYt;HBE%BZ{de5c zsA0a3>#%VSu%{56Fv1LBBcG7WlTkpCB=3SvyOxX^;d-SrA&9YdWNepq&^l%9i0xIv zy;38=!q2SPFs| z4~n~MD-Zbj`DSs%J&q*DQRA^W;#q=GSW>oqTj!|2J~>+i9yKg?7QmAv0LsBFNKHRI3#= znSJNoxCLX+W@?Y=_+z~whfX^JSbV>cw{mO-o$vMSS#sn`p*^>Q^Yy zHAN1xSuRErGh8=e@a#t25yBByeJY122ZDf7d~_J=YPi{#clQ)c4GaKR_&#)njv({7 zQfr;#i8o!aBsrDWs@`amjPWkKxcss9RLm{n<5NCJnaz3X1Fc^1(pCgJgc!UJP{U|3 z$c@@-U#0PHfnrpbv5O(<$<~b?vHlQaR0Cyb;f+sn^M?cwlu2(S zq2Q`M0iltnh=-r`ZpE+M^|N+AgNZ6!g?{)N>@wM?XGBFmWA&Ge_MX7>u<(6o5J4$> z1so(SN@TQI@^ex0XAMM3MF5@*rccGyjAFr3`Gn`W$08`0EU-?TBo;Z8$pv|Wi@ zf1WjoO5?+>7hBDsM78y4MLyAWr$g5YtrwEhuZue zMHUJ8Jq&k5q3AQTapZE?<;4~k-C1Wv=)uQ-zf+#l95<>-VlkO%(Ee2gkMd^g7@p}) z)%s;iPotBUKuf}N_aTE?*7>!b%6TCY+|N=fqOgQ!W=__+*+u7Wc`S_8+1@q}bLxri zdEoCHc~PX=G4e2AzqeNiB>jlA(lBN60=KT5;g_&nC=~`*8BhMra1BBzUJ*?uEa}-_X`l9FHS679Vm}L z8{ThS85!4peM9SlWr|m(@7*tj-|FY`yyT|%#@G@we^KzM-MZ4~JS5+gx=Mj9tks-f zP`BOj`gFj&?R*yu&73;d!oIq$69L$*cM~Pt((d4Ve9M{IlDE3D#uNlDVcN}hX%VRK zsXlDs9v8v^oiyr`OgN3q6-#cZ#DHUlB`QK2#{_phcKaT8_1xww3!yUAzL;? zdGqt9qo?>x>il!LaMee)8!L}3n&%T0>Oy_K{C<2WsAV=4Fm`aSeGv1;>QGdG&^?1I z)6N@>=&Zi zwkdJ8`U6_wEo`coX6rm0IztrJV9b}XKbJZc4khsS$1~m!Qq+5MA49s@#H!b;eH4ja zu9{&K8Vk1~_nuZ4%%S4df%0UO+A?+nAG5HRxpT^f2K~H{$-9}90^s~uih!4%qC$9V zi9p>9iG?25)BJDa1fkP_ijaOlNX<{N$5{%ZrH5=S>BdNHR_pquJS=zga7>3I$G0++Dl1LW zpX-%yX=wfNI0cI=_4`Cu1S*36afE?}UM=XAV!+ zYk9vOv?>#ctmo~l8d2@|@zfV>P6eevGmEiE_b_$mi>_J<5z9Hh43)@yv^QOk%X;Y% z$%7I8XEJNK!c%jb5NqbawsP{$rH8;|>Tk7cPOo8&F)LwWj=Kua!-=6x-e0qaC+)D& z>R0Xjqldl5wC%eV0PL8H2%UK*>2ff5rPLxSl6B}FuRgC}w? z1n+X>k+q%;azdTTT04UX6>An9%Y0>7!c1e5FJpe|O6&`Q>A9cdUtVB@^0MsXQ{rUt zBh2FdOGnc(*Uz-&yk=pR^}G1TjPZ@?wV1-FN^%^+0QhxA=8$-GC? zWnqFf(sd8wPj$;ANR>){07+f}AG-rd3CD3@}?UTiSA%0SYWkCTvyI+kh z@f^1Jtx=V%#uSI|6zQyGjd`MP$&U*zz}u(#`+fik=1@_vg;*x#4JNdn2-Xgz>}g{; zj@tJ=H0~v{VVd=lzw6km@}A(f$MZY9Q~b^Tus_nff^j$OyI8VTxpBW~`F`eyM{fLj zX@>Fg7P$lZ$O}C)3fWb5$Xsib7Yk=##|BYqVHaLK^!8Tq%GviD zAjuoZ?k4_n2h&)<8pY*UyD`|&__NR4Dk#GZQYW=)GnoZ;X1T&#zAv>h_S-Yp69O#? zcv@Tb4W(agCpRbXC^wW!j#u@X)uVx(M@I1J&`|OriHN_l4~u!;q4(s`ktt?w2g3F@ zgU@A~v4{{Q5@T$-#j;y=0u3aCim!o{36 zF8O}G3PXW+BDaLDBm9+dVON;MA>He!cTpPr@!mE2!{4-lj2-EiAM7%KeW$zM{dMfk zb!<$^aeSCC6346l{{WkrCZ=jbMa6|_iibi)UK2s|Fo+(CGbrdNj!+y`NUB{9fu+T7 zblE;nx;Wzj+Knjoq@Y|qY~O#6HPdls7M468dwE}VKIQvb?E$<&jZM68ycYc;(s4mw ziRQmtVCW*F(!^lLg;qLH0aT&8+ZZ1t9GMiz_4{QtG+1@*-{3Ew8|2SzkKgwziD~kU zLvx-u$hxb-^R`m>PiAUv)pY1ujjeF+iQS>*c@{9>fmBl{*E416TezxWMYj4&q%m1zEhpF;q+-QjVC z6cgQWBIF!1T<_OL!U$TxC^m2jtBogQW^fh4RJAI?%w$wIlyP|Ej=egn^~07f^mJAW zS`noK^CuE>r6W-*HTo_orh^ykU8%r;@>^XzV^a*!x{8Upb(Gvxub$X~&`ziY<}K$j zFyIU==ZYkD-%yrM>(f`=c|=V*wT2sZqR%Q^EI2_yLdn#Zcfy}UQ=n! zlq8KNUWHYJ`$A4kZ+E1ho5l<>no(aaPgn4gq!BRxL^(?2r{|)USfb90>@d13pwCFH z`84D+z>{*>DGLOugnDc!&@UJFlJ^C)eg(rog;M?BY+4Z}HV9-bv&zFS>;7m-bW*jR zv?i+?XLiRyiKh`-?B(_R#CTN*=?M?J;x`l|FwZ#O0MM{>%czacEsIUN6Iau{X)RN0 zB&qp~*PA!CO$@tHX*_oX6Ihb=J?gl zuQPbK&{Z1Jt!WU#Mh(ADO|Y9M)hsTyjlu7Y&M~XsH??YLtjxp?#uRN(#ve=qYR%V+ zGbPf!@5yr9&U>&V)I$Q+#UXUpA_@Kcq=`1- z`Y0laz3#K24(mO_Ewe-bCA%!_Ydb$W>#b+jxMiyz7^MZydt$!^-e5}{MK{Mp{scq* z)TwTGNVrGXOqRQjYL|)!?rq&NoPX4CC`y$G z90ICr&EfC}q=QOQXtG7C6AP*Tu(=ra^+B{-+GFU{O$16Kh=-NvSCLz3P=47eJDp8> zA=p-l*E(t!v9L3ZZNI)43xTMR1zKs`@&C%u77qdPQBJ)HzHvux5#nk`j~EKGU#D_7QQC&rL;BFR&ph505dXq9p<#{d>gf!fIp zaakHeZQ$glVksm~DKrQrxvdc&zf^PWeEg(;JeN+&QOj!a5DYf_g5qNfATJ#`({^Ep zvJIcui_hUimscjtVq^2hZ24A^bFhB)oSj$xynN-Iw%u7Wyi3r?Zjyq{J5aQ(-=NNh zW1B_FOAxXCpR?F^E_}&nz}s*53qWZs|1Z)c@Tppi^@b1heRtjni>p`O zw7X}$O2nI^)-%k{j|@>2U+Lu>#qGdn33ed!?wl>M;aB!HC9Uj1*tL!KFnPO?i%rBj z3y`VSJo^`DX7s=@|04z*VIFR^RX3Eocewiw7E2K}c{_K)n-A!`+-{jlsh$Zb5|V33 z``>;r^0cz*XcYeMhK#3qI{%JL@kS_Pe|^?>0sTs@KP-yi0({~*=f@IokgP8Mi!dq`$mnzpS zbgcZ7+wAQ$@`x>V0UwAF8mf4aH=oU$e2FidcvjcMyz`fV^C9UG57aV#om#okt=qQW z&(7BO*foe9#g*6PD+#pvs8ux-U+Sl9<*rW%_4d7R`f6Kv`to*WDZJci1>tDfEZ_3K zRR!3CT0u2!M;Jn@}Ck*C2P^dn%oDpqTh^XzIqr%`!zie*SRdAAh>J zjncmYF&;lAb~S`dIn}aPVwZ*yju2-sazPJlD0z(&8a77G%;Q+2A({fWuYl@+PYAxp z++hxzq*o*%r*iGF8<}Zcs_Ohoylou<72n=+bi5?liS?w5hdfu+$72`?oKVyx z@#Cj&6n=psO@g5Q1Nm+JR>>Q{pmk4tsX_HSf~>1kAAWYM3nyKDlDz7yh(!(WP)quq zDSH)4TgEbhVv;Mgz|?8LoX`mE0AZ4L)ID~C=Ly&mAZ&+4l( z80$yR)Q?xL=xtD6qt(UaM#(b#Y;^X~rZR(#z@&Nj@1l9e;z{S+A zQTh6CB05iB(kt~yJ#Y*wp09)ZS0h^!KbAiN8a~ANqogv=r~UwSTW<7uS1l6=nEOA{^G>lEYoCU z%F;`qrX&7piU}DP5#6!sd+@U;GV6Z_8?2HMSsaFVEZlZotBDN#uO);*otw1mQ2d}; z>7^w;ZIcno|7T1E(2bs3&Okk_1Hf=*m5~V1HL>zm0;Qt=Hyq!WEZpP|Ot>rQ)Wu4$na&q>j$jMQhvTNT3eizjZ#Z zZBxNCCOyE zl7y2l$+LkvyE$v{*0)@6{OR+lY{S! zL1KUlVQ&j%&wRnNIY(vPRj;mRr0lL^iHd8knl#V*8jg&Y6vbD_HSQBE9RG0FKRsj- zru+V&|2=iwlS-lxMlazU=`?S9Q0`2fXu`<7q!IGic5bfCO;e}+|xox8kV%l z#;md3qocHo!FO2ApC=51Q>kyM+^rTD3i_il%!&#}Vk<27D@kLw)!jMEa#qOF+mB;v z-7>2{#o9(ZYE4NIU6-w)IX_;MxUsMwyi zH3psJ>c08)N1yuFU@g3tUV)fKA_>_-eYO)di5tx=j4mMkIB~ObLcI*V@FBc|nR0s-`4Bc<#KehK+PpL$%q+F-yTb!d zQV(F|z59R61M{<`cM&H~ca2)I(+*i)R_w}RbeF;IYjnrwYr?q=3imf|7x$Q~jZOF? zg{_mT_s8?=|9Ma~wG>w$`zB4|&z6n{Qus7PBu5g(!G)3se-?S_8^eel`pjvx5XDK! zCH~jBma{GI5~j(i5|(Z?_GNDE;OKva=8v>t`)xUy?Pm)VglToXm#0;#D_``Vof6Ly z%vqb!j)!#X?v@FHrh~3*FGOo%a?W1QgZW9L|98s^6fHl`Ip9HBiWei&V!U=KsN_`X z2oEP5bi9dtaevs%P1`lJ!R>^cGhqilFan`(j>)|S971Y{?SxQV8?#PUvCDz6s2*pM z+k2UY_pS%LFoV>`4Ph4Kmm!!m@p_2zB+jt+R-H$umq+jQ5?$?%iNfWPsc-VEIQW(0 zDp#s+C?-J&PS-6x&1DJMo{YN>+|OL}{r*|YeJ*2zZ9V*P?7&h>d;WeG3r{XrsKS+* zoWibAcflv|Vh=lre&EO3C(&ZuLEjolWiT$)Hg7P^ADx#XNZL)Xpro=#BfP%!aM}46 z5dbb>{HpYKiZ&|18?vX|#g>G#$3>$h!+VX~)dUgJ&}C6~qc8J~vl#?vOVYb1eo>S@ z{6k&b`x?U87}nSlt6n2#*PLy^*b3U?A@-ukmwpC6AhOv-vt3Uz*K6m%6V4ELnbQwQ zW^BS2=iFqP8zJcsqcegY$IRDT&sr9!H4Vkeo3mgFS`=n@y~h@blRWp!nKKLk#oK<= zsy_xfd}4L*hO;wF&hfWCOJF_9(&(6FbT?gmK8BFEB(yzn7Z-4A?aC@fBU+Lox3 zmA(9b+Tlj-X`rfGhp@|;$=Y;(KVrwxM(*`{8Bxfp`ZASDEqi**7X?~5lL~nQZ4A`z zs<603i2Y=3RXf3WND1C4Y*x>C$iYxOWnN_{M)7rldfuy4|Ji9r%`Fr~fI*t4JVg-y z)Dp-idMoRzv5I4bn)Oq+eckog_vArX=JRv#%o`iuhJQVhXq{ydD~yMMN8eKT5@j%~ z-2c0{4vBb3se7qC#$Ox`W{0cWuMB?jE~>KEQf;_7GOuPeT00QQZ$6|D0#_Iw>u3X0* znk`DCpy6ZL>f^2%Hz~*RZr!hw<#YKO=;H%;J4oz!eqxvEt$ zILhuVg_dV7`Gh@}m)cQ`+FN_*0Ge-{UXe{oPTjS!73~%Ol&)C zV4p?Z#^1a~BFSpzC?1WQ$}IG)QnS%}@A08l5r0IkbO=;Q&#zk0D3eHK$v|4w%i%&3 zS6xn&zjbr@U|86_+y}djOFv#R5ksTk*A9|CG9U_RTBD|P=ipubp)wnWGCN@6Y!d0E zO>Q6)SkpSdbmtr2A6rN}7}_Zj{*duJMq(zt%2G^QEmIPUk*Xjk-4N@6HHrZh)A7`FF4p$mew*@60rvOLgnLV?w zW{?jMV0U<#|J{^-I)%$bJ2a#oRbFFyLj1ErKfkEs!V%q)IrNoyWG>+64ZbXp_WG-W zMv{b);-?dcSxdRcnEZBgXqGsRK$lj8Z&F?iW6A&CUWTraq@J*coE>4^lDu%shBXuN zbruh!c@*#{8^2tHjYO%D;DrdaM0Q`jv`G#6HQDD{b3&@#lnzr#=VxaAa*!}8_*=(eu`MMCp&Va zPvw;C0iVOUz-jX}i9^b+@mQ?tvl6Gh%h(%n02b0YEM3t4)VW_(-+kw}^6hCT*w};v z9sx5{L0d4_+P*`*OslX_=f};ns((H?p-Zf#Z&bIfv4uyA7=+3}+)IT5A3a9HG%v6d z`Ut-;AEFVB*Cw3F)FSE#(Y_g+3S}^5j(>^2nazIbj`vx*g7~8+#9Kd%d?D0RDILHRt8=$QD0WSc!tZYqvVNT;5_rT_X((6E}$7Y*dJZrakFv-q*~* zjB?aS=fJ*fiz2e|$GA zlEuK?15fRh@(VH9=2`kuqFJNG`ypW-wb)ACIC1;{CAC37C!!09NJQ&IoZ_w}b5xlO zggn4W#VsOk=AFtXmF7AttX4i&cRHY|4rRicUt{fx1VKLjA#k?$TghLm0~)zWNdDj2 zGL1@sntGFr-^AX!*5_JUuXeWA*&O!Ixza4@D#4Li9xGfT6s4V^x_A8hzei~$w}E`n z8CFb!+9fI^UF%u=WY=6jz5=71`Mn>3-&KMTO z40}3-5$&+4Z(qg`yy zFy-P8!(D0dkm1B6!#hPtwr=H+-OZsvV+Xz%x>c7sJO`^o&r6BvaFv7O`Ukssdg)MB zgt;Nw$A0oe;Ii@3H!yxHZYD*;2}LEo^D3L9dGKYfIna>`HA@(sJ?mSfiB=Y|^FcOn z(cKx2&kGqU3fNDxiI0ArAYd@YK78Rg^td#BoKa$8%3a^jtS*02i?7B4gO^pUAf>q= z;iB#6K^2E(2)z#gJ`nUgr{PvGp#v|JB@08Pmtg1^eu#ZXnl20Q?~T_{)$`V)uB$Mz z59D^Y)lPlQ=~Z;^2mgV)6B-gv_{&)$jk~7$^Xb~cLWj+6t&qq6<0D9dPhUwbZ};wt z@H_#Qz3+Vf&?(0|tbB-|-jw569cjHHzR59WU|sYxrnIXi|Do_&81UFV$AK}K0OH1N zgS;zc+O5eA%)5DV*=j7M^3&B(Suy)sMF&EfF+oK6ix~aP&?(ri^;0gt4z~672pc1& zuzdx)bMsrO_LqkaA)ZDs(q0-XV4}W=?%HCmmZU{w7p{Wbl*3ck@=+aO=l{Ty&2J>=c_BN_PDs|2IOSG{@v zF5}XxRw+i>zgRO6l&F}p2VHW`DT3YLF%$mW^(wtuEBMA%eS96zNoXq(qlO>A#et_^!aWS?uCgom7@dsn}9 zY-sL+(%2>e4e}1rYSMk=1Ev@{|3)63(vXqjLahhx15Lub2@fH@2Bq8}`7D~%P@aAZ z{v}a_!MUp7v2jSTp=>tqcY~>wCy|{F$E6C~hGUm_#`%Iu`+vt_kxe!Hga#+5l%Ywo z@MopNu3MC*UTAeO-JJC8H%BrmcIHy>{tr_Na)ip9eEUmbr;`jUq5z7{*ih-4^K>7F zpYs#o=ya0PAZKf)wOQ@6p(m!Kh(n`1!X?T$Fv&6dy!@i6qfaJJ(N!$s{MuZG!iaBOI<~BT%l&iOt^ZzEtK*tVn}4mu_-~GUP{0KM?kp08!D<+s}~vBLNP}byt1x;k2FfL#5mWDZP{kh zDu;i+JHv&|(cIA9g*CrMmfX}!410>pznI;VXcSL7gojSN^X}HHnId!%>;{#HT=~SU z%ujN8w*nLkyfq;Or4)g=$0+f#(Cp;BBJzH80yHN1I{Mi|^c3uq!My2Q3ERj!q908e zsKzo;zk*D_3kJIagKxtJj>kXTqnq86jR3+Ek;95}5#<=d1EnL}t=M3ENNOd{R?|VT zb&qaGr9>k1{nB2GtpMZpt~T#uSI8XvCW_gY`!l->LAYY(t7S z3~h=vM#fO$!vj5`S|{(ipJ8DoT6^G^_v6DoGa|{?gth01FMh>B?{Via=RiCol|PzV z&YK}dD$-smUVt6=an5B}{2wM(?>Ju5LjRLs`}gU@v31k7#MX5EepO=XPRvkx7?n(i zf`iiKc%{Q8eR@iTHGySyLzaN(P;~l1CXzCW&<+)bbL;p(ZWx|}{1PXAz*m~pM8ljZ zrRCG=G%f(2x@gU9v5a#F5N4B)?O~f(VUvWw96GZB9Br3t6848(t|iM59Ugv=B(?jI zFJux?i2ojQlv=#-SD8QJ&QaXaz2A*_wpNGYzWVr~Q;&=$q@&h}+6!s#suoOyxj9kk zLK$yxjOu2Jxxk_3kW^SGV>gvA|ocm+{!bIDUl=tY~V?HY+ zU&;0cTkUvHT4T4zsx#a<19atd%t+{+xXC2-s;s|mDxqm^ZGh(gp)}zR%ShWQom>yE zVo)>A;|83vD^3_iqypTx!GSP0G)6Y>a_=kWb-#X_>eCMthjasmdfpSotLlPuL)elu z7N4C|9|RWg{+cLu>$_@vlws^9WxW>wGmw3Y7LGAOSBFWxP=3ryd`7Yr>RPQ{d<4|1 zL)WDxxKPZ}&}KO7sF$pgfN;oDxYmQCS<;|GS0dY11_(}6z|yH)*JUULjS)HC5-yg8 z+U<}1Vy_UH_CMaOIc9MgI!;F+B65;%9x7UEmI*7ur57{=Iv3HC34vM0R0EtGd z^A=aVyj_2}-ktHg0!pnda(aj%mX_LZ!NBo(EBaFs4&eJ%@f%AkA|}Q#13d3#GWUwg z_|i1*%P)|~9hCZKZtk9SeWicdf)5nR;`@6N`b0Xe5#eDNR-JWKLbm#JwRo((O9cC} z!>!CS48pCSV1d2_x-XP0YCeV@1K~{DFR;UxGgoR_-3?_XJ>%yF)gG@MO@Z#3op^6; zgXs)m2=X2h_U!5AwP>A^5 zEiuNJyL0ASWDwBYz983`7M^+$!Nw!UF3NKe&FG7!%V z=)f{T$OUJgL*D-N)9uvVv~YK)zrNxOwt6tuPbEY;nu|JCrzN|lc=cS)ysb<5ywl~X zZqg`TCd|)?QuWTR+*2VU;uPX&2w*@aA4)sytkbZsvE9p}Ixo+;vSV=M>2oa0ml z+-vg>Ji~I$?p=-_sTJ*kLc3({dJw=}hUDz)$tB7XDcV57w{mO|YrkOJ7cz@m3VuN6 zNnm=6194Qs^uoH9iRBYFih^_Z-m=4=>v}`Q2Vp}29XQq-SHf=EB1aC^hc>*H=Jmxk zl4G?Z)AH`t!b(krEf~4sw%@_pBkPBWO}ItQ`2sRUdIUw+4cB$@{3 zPlSaljC;LOyz(`KoS#ngX7i|Ql}1zMwgy|1I#cIH3y~$G!Bq?QV&B$Oj*l@P zKqJUWhn8J6T0G71aC0GB`kU-H>D2hvu{B}Wzc9ZbN#D=tm<2hzf^You^g-)oYO-kG z@y2m;E4!7-kUGnb<&Ho}XmIq@-;-$vhMnNUX05>~4e_=LtfV>3V^Euq#wxt$>qo+q zjjhcQYea)`pWiCqhbu5)F$OpTip8hRfAvTmKbAvMiEJH4PT+cd z@uqijTyP>Ckco5$eJ&@(qU1?E20Qx%HUU{@?m_i}!mc}cpWYChU$BX1ZInwH#l zjc5sa^RP(Eh9zVX3PrLSHUNvI$dkQ^!;;-M1n$Y9EZRvd?3e1!0{>)D&0cN|S@2t@ z99I{Q;P*2>J4i0h>0};vdfsmpn$W-aF39+7<=($X@gJsEJ^6dWavY85DIBlfw!`+x z(yym3xrCbC($Qkw=WiqCdv%yt>?>BqKvAVj*2O<+`him4(|W~UTi40+=lkcuBXMeY z%J*py;UMGnlOr@@Z_$9?n|djYLV3m+?NVQ!e72yXe-xAQoL(WT!6_UU2DiATH6E*_ zR0mupu9jSs!z=38W$LC>n`sKdX>i+;s*IoCl|0o;Xcat1#qG$U^H0$@^laa|jyFt> z9;gv;vJ-6foa?7(q^p-)YAi;-Eidf*mQA3-jKnf$PNQ#Y1v=la7l%~gq+J=4n#Sy)D zR}nLh5~u2y$75pN%XUOv^~MZLBtMv|Xf;0g8O7tR>eab@fJ0o6r*2l}h&A_f`1<@e z@)=k3$QVTM=)A1#PC=m42m@n^0#-Jd}VWccw*0Xcjxm^NGEVr z+fAK~1!F}fcjlRIrx5Gf`jH6F1Y0`5k|57#e4GBcy|NR!mKgU^KqjpEWEH2wt!9Ep z>+weblPU#`O+#eZW)irUQrSLX?fE&qRU;O{l;CgrnC{=C^D#hBDvoZ3S}nKbJ04qq36+Q$yuyEej|2VL(1hxf^kGn_@K{ zl&t-iPKWft*j3L(Pnnp?WH@%I%rEh95g9)sqW?GeNWRso%Sq8O5-m~zyzJ9|*izr{( zah_w*ScXgc47jpGVl!8Gb0m2lxhcWi=CxX8Q;B5Ei#9HxwCc8IK$Ued2;^cJQ(L;e z`a34S)bLU__m~~+*UxvXx22{#y-6%}+a(TO&7`{LI-_qqA zMI}>C3mu*nW@u2#I_>K-@25`QUnm;Hi2L%VNib4RTqiE>78#FO;c6(*kr;-lNm#!| z(L6k?$RHnwmhl^v)^0eLvJm#oy$~yZrkF|L9pntMECZ;%nGC8{IfI$O_4yd(X?fR7t!R zc#;d&`gx;IRT#?q;nR?V;)W-y=p;G(LpwO3axl5*ORIol1^u*7 zrG;pDBSUO(rx(Gix??sU4DhI6yH5QIaxAzy zG@tMHxPG1Gd@U4ceOT)NttTeR7G_gzNz%A)Vf5JZT|5<=>$)jKa~t2!aY)mYYseIB zXlCj#nXphFAtF7r-uG+;WDZ}3Sj>3I91qL57SwfcJ`-Dl@xx?Mvw#BZiv!|Tr4UIJ zGs)l?DJrLutuU(s0_?n%?-mlzC+E@Gib|^lZe=dX?gZ zv5E=5QHTPn_O+<49J+z9f!w&rEaF#&-l<3XXs`?&zt^J-eh=||dbF77dW2bf zP-EuzZg#Hx)-nMB$u4$_yz@S9qe9Ta`Lb{t$}}L~OmHxI02ifX!Pk2B9_`jlRAORG zeVIAB>F!DyOCgPw(p_B~s1C(S;a_ff$q@aF7LQHlB$N78;on}~-l@2@lUcAJb9{tC z60tD>7te#dHerQ%&#fQyj%=-Bo_}xMxK{>iDTR|56|E(rlwcy4@<85p@*Ibm;X^Bf zTes*=NwCDUfOK1q5)IkzEF+(N5)D6?f|M|?oPR58bFBo#ei zO>@SrtbRY?FXT*Z88JO%h-)2t%8HPR9U>3MR-w?3nlKphMZ{IDo8xU)%4nKXu+$e# z>^(9z+bSvu?A%4mXvCFaX_Es-uewE0L4gWM;JTrq%SGtD0Ax=#;CB?uBoT}OdD2T% zH9hqsFp87{+p*Y>T_%Qg4@;Ag36aD2C%{q%OVh_bSW3Um(aCfyp{2H}_&Yb@_v}`8HoHFK0$$M$(>8&oaPYRm;u=9sE6*U^P-MSLjCnn zB2DU5PFr8C(1fjq{#X%!$ii4))Mk6}b?JcVb0`^;;B8x2?P1K`3C&Re@-6)yFNnLp zXdI&nJMy#s69HlL`$|=vGFUfOtRHueoa1Q88lWofSv3%L3;FTdQ0WrM-}^)i`f~g0 zoc0S(X@h;Ei_A!Bwp5cni!(n;YBV2XcA&B=o+!xNsPY~_#Zj>G#I_)9hM_DH`NSdY zaM9-}nJl^b6O5S{@*hQEWI@|#V`@RK5(pihZ5n85(aCjrsB8)JNW|V&Q(3FA6Jf0B zF{~biW`iFo5x?01n*W=LOz!#+ldk?;-4aI0OJi|YtcDR%gsfuCp^U{}2u@(VUladi zAV+raHW5k$4XrSIN!&@2m{6A0_fX*H5}vLCIy9^mJm6DxOAUBSPDOqPe-*7{FK%2V4O#UC{CY=kc|b#@Zz8?{ds1EKG$exNKr!< zsVWgHwW@4QVzXCtBZ!$QEvma=U=d|64h;)!IH_4zPoDD!jl9}_hE&Vv?&vBdQ^mj; zQgvjZ&{3;P_ziBI+INHjNY&--wYLw%cV)}N;y4NgXvaX5&|8P@#1B7mcQQd?D)+(j z!q{LB(oa^ifS47~D@oKpG1$U+`02c9GEq5RD1s@S(~B5gvmI-rf`v)_G7iCDBX1&K zS8DD%bLtRHnXoQPecA=}hxwT@xtN(^2>fAlTM#Fh49GmbPBbYZDiI|NMf*{oyBX73 zvq+khJ_GieWcI~yH{v-0DHx-^oJ;YJY^Y#^Dh{(94ahix9{-bpDB^zIMTYBdi;=$H zn8GI?PW<%MVEP@pP;)-6Zia-&dIBnGS&t55mbLx{wg}TU*k4$C>aG+Qu$2 zVBlzqln6ZyV4lzAHCINeS#jx(&S-)vsUQ}oJzZAV_()6!rtZtV&Z$7#7-J`R$MH51 z7WjheW9k_pDX4y>AxV6!Qpojwm4PnxD<30;hapi*!m~edb+@$)jP2(`L|ONJS| z`0P|KzBgY5?}u^D{ypy!jh)zj`nj>7u$Y;7X zxF2C7H@&`Um`d@pcvAvy`7?BgP)B$8olR`8$Ie>^r4<9r>fpB?3gujpcG9Y~9qv(u z?j<5rfhHOZ1%6UC8l~Kz9YuJD{P>XDs+R)wZt82>kXJvxGv{bt*5*yndIB=ohyM;% z#i%*jp*cU0J92$pR37Z-e^uvB2N1^KPBP+L9qxAsj@=o4Y-UK)c<^tKf1XZ40LKxN zkhb$BZF@`UXGrCn9PVw43!(T;_uWZlNMv@0)0Mx@s{>GHDve7LdHn4r#%I*UR$>>DRkEIn;?h_ZON^cX}O;Iakt5EPc;Oo6m%_LtCH?-6%461e^M zcMH<#Nh}Ps1UBm;R>(5Kw2e;b&JbcI#V{4H(?B6N-yQiE{8`p`w$L9*ctWaf5ls{j zTWW-r`BQ=4Dw_`_XC=ikWGhE#@ZokW;+sy>9w1I!5F@W!KN=(M3)yDUB*_sF(%;p1 zAn_C84oi(nw;hh2yfB|k&sz92y%JWq}plOL<0z>$J4JgOx1 z?!k1#s=W#e6R>-f$>h|;2 zBSnbgwte3~ve@9&(d^eeqrYXr+&H%Bc=1%VE_U`U{Akz9V}AXed^%lFP{|_KlUGX5 z9LiZa?FU`rv)|;EIU2IGD>&uLX$p-CJs*6;J3jaY($&&PjR~vF8AM?Ap}0uh4{=ap zU~S0Eqg+-kz~tkREl~L|>bzw#dAqqDz)=^UJ8#8LdDGsiD@`H6RX)`&N2y0qh;^OEeem^p-c_MX+(Vx6xg z=#0{>7xiipga+5$?uQ>`L?qW@iBU~bjRdH0(a@(;Uw>`u<~s=YoI@G^n^UMgs}$w& z{1O&|(G!dd&i9D(x*OAk-F(S^I5H%ZNna2r%W_fjkq%a*#XqlE%r$*!v&+T-wS~~!0hKjH;F@jWVI!D)utNBLQ?_^! zuX>^r2BYPopwzMdidj~g_#|reB}oZv-S1v3$=Hxn;~?P3Ark3_h$`C-j;n<%lD)F! zR&;uqTz;Jc7mbEG+mEMiDT;-2PCpxJjV)2J$L$iL_*_iM6X|!GOz6ou%gFJuAFcG^ zFw)?Qzea=C}j!-FXfP5K%eH?celugz(t^*eWJ_kSNmy#x~S~2jVT9K<@O@t(IStkD%Gs9;$^&DwAWaO(lY;5WMo^&lom8F>IcIL-At|(B$ zIgDF}r}Bhw#HXvAT6H!DKPge}`0LTA6VTJbW9BV?BOrD+EhwN@D~0KWy_SwJaBEG@ ziR#Qj%8@Vqt{B5FB6?Soa!C|e+dIuDw#MHI*l&04%quOTw%OlJvY)=n)@VnlP4hI& zIy}!|L`rzhewOU?i<;2I8M81@Sl78@RAds#{ZVEV%6|};uaOg8@{~ibm|De#$D5!< zk0Tj~9Vkq70udrMHPuP|^t!F*}Y6SVq&xXNd<{fIFK|FNPa$bIdy| zF*=^?e4~+`%J_PTxCmJJO`n|&FpKqEj)@o)&i$KStx>H!?{>Jw!#uo#N57<+h9|J= z8?}v__!ir0<_cgt`8mJ~YM!$Y7m{R(R;o0J2LH$v&e)MzEoHc{y{+jal8)?1Ecr}` z(%6oVK$Nwdh%YTZrtHE=hnlK^ZyP2M7#1;ehio&mU2?O)n$$teT~)@CH~=k)<}T9k zr89xEy@5A@>PUoQs?(>;Z4@V0ocdYK2$Y{#z)Ut^W&U_VPHM^9{ie9&fT^}hK7aZU zYEzLr`gnY9gmxcP^agw&q*lr=yhcOe$bh+NzPim#{XW|SBy=clOCwg%gBTKOQihOqV(ofBvKRDi8?x1XF zbM7A!+htg@p=Yz3xm?wp7)&0-58SeD*!5yum9y@yToog=wi^<2M+o989SI4LLDz)P zKc02w-)0DG^^MPJ<@^sc1oMR_7^_3Yhxu{#Tl;?Mq4z6 z;;GOPyl#7A!zgnw+`H^4rWg6Spwp0-<7wzp~a!ZIc;PaogzsmyfU2Pu-d z?da3LtC!Rmk?_Ez>co%nf(g1z|MFQeUd<%2s#js|#K8PeDCWxXm1$0)t>7V(C_d1~ zz{hc|m7~ngQKK!tbbPJO;yjjuP$SZvkKk?VTYb|5K0tYr##fH-NLGh8967rciXpT0 z`Ve)ZFJBlWvG(K55t}0J`4(6hato4k=*aTTBV*KF+TPk?&3($nlqbw#XLcDliLM#dB&r6IdfAI>!3nO|tD!Pqx zOHg3`%0gA(?!cesA}e=;ihBI}s&UW5pIhesB(M(!$6On4vBra`gg(#hTNfSf`e&fI zp7jgLa3>J)3}L~ z`Lx7l5^<9~{C=@ogJ1bQO9F~DAHU-Ur8G(5(wWR#^wYQ!mvfexrN{3gZuKm`U^pNs znli=cXfWtf!o{JA+fO&C^5t9Vo+6^A)i!HRjLg3E24+yJWxCrca1W$?qkE2&!dhNC zhj~&bTCy8)`27K?hjlv`#JJD1m{7KK9%uPt3J;iAPasiLh0`n>2&#ta+Yo1q21Q6q zpHY>nN&Qs#NuiR1{qrjW4vz&Uu+}qfFDZM-L0HOBi@;)i%TqAUgN0AVj8MC-0kUcHW|iQ`>^vk|Ea%$IJDwJIqgoixN%*}%vk&A4RM@Zk|DWzz6HH_{@PZK$7u z!wIESxj@V%5@ymf-FZxl8sq+54>bSk4_v!_fCD`v?`8#)gU};nc?rES_)|f|D*%Ke zm2F~7M5>=BBm7jmXlNc=JYhV`lwm*YC7S=PbL3_1^bNdhDQ>EID@YGIwS%okEWXMT2bKxz^a- zjN{B1+@znQc`-IQ^gx95I>?J5CId>KfzlY}Mv%}FY}BlxNi0t#5xSTxKJ36Q;+8{n zaA&Q@xv}F%^;jvq8Sw7Ibv;9!)GX|5B9I)hYkHN-oel;995?qap(UE*`w{>6(q1(J z`!S+R3B?0(T9irBEMj&Bmx-g`5S#l*RSm8O%ZqwLo0~`_$iXX6ur+!xcwy~vNOiK( zIhJwTYOX|1jiY!)$kT$kj+xF9q^&)=i|Ws3<6%G6?CGb~m3M~tk$YIs+%0?Au?4+y z7=TPUExa*5Oz^bFjRoooK2j^&&8}|DI*is8ed!YkdqV*o3s6m$e}_bdC{$+a}{ellCznc)bi#{NtRtO3Rw zA~Bf&DXJgQfHd`>-}$+k08nd}uFbRKqHhsF^yPT%f#D6iA&Lk$*z-P+uf?P@M$|o@ z?bhnf`l;mAG5Vy`Hb|fTN=7R+mj2+}9JV+iX_51$iX5Zv2NK4m@MlJaJnmoeY0_M- z`PyXh+>ahuf1xdwmrAdqGPNsXS$xkQbGj-%x zAR?gV0-`oM$gdq$5cI61jlSHuw@mS~f2*4v!;g(IYI=!7q0;p4U3lylxw#c5GScFX z)3$|yeeof<5I@tUQay+CChnYHl_8^`!1jpe;sG}LA@Dn@=s$e#-ItSXtF{?fTO0P( zTmDyl?AC(!zJed)SK&)BiFH0!y$w~n(ox^#FHzFnb8qe~m7`PpHgR4TfDr1k3j`}g z*hP;<_|YUPT8kC_FifFW_5K}2B;zag6V<_iG>E%`z;{LX^4lxNS$V6M5#!a&*NiEw zzdGD233p?|!g>ua5lF^b{j#aoi9_>S)A51^!Ya@*T<>q?Rb$EAYJ!l;o90M*PeaJV!1~0yreREUduNQ_uq#MY4_%DFk6B9*6fQEM9}@Hg4ybEbZ7x>H;Dfr z(b!>Jl79p~V=j+UxDF3Sq`x+f)PsYbr612CSl+r8Ux8VdL!5ZgS2?s>_?pMiTW-;2 zkNUPlDf*U~#dLU*nGjY&!hq!VrJy(%DOPop{oa%i;DE+HCm0_67ih&Jb-p5JC^ z?zC)AK|~4&V|XoKv6kR+N|Nk>RK1FC&Br9RBYyP+`n$Av4F^*{EU}Y66&{Wy@GT36 z#W{$6v&ilbQ~C$vifof+_M|PFGJ0h#<3FIDFixXOgU0(jp8jgUt{zo)MK=cY(-hE> z=_bJC;r?#MW{XCEn_BY&2NhuB9)3LwhbJ<&5CC)m`lg`MQ6KMyNJi5sV4+T zg3JV=c0)!RuSYEh^EVA^Vv@X4Du!!5ov1Ckhk{N0)TBU`7$!b~R371e<^)&lz~(Hu z4v~;zH=={fuRXzF1QI#CL@=Lw_l($Cq_)j+@T;@kzKi!Il1Z|$je?%PRs(vT;!o%i znGAncPs&`UsT2^avohL|=BiA;PI>|-iRoxz zRV(+Dh&$U2z9pNE-X0Qdu^zq4>|kNmD?07V`33V3^9AS}Hjtt;=gtx$H=R3ic&Dh& z;5;<7_^nD@LLYvud6W%F77g2s@XyygqR`7*eedd>mPVF}2GsITt=xILpM_n%GoLit zFRS8>SFP{n+(+#`b@KF8k?C=vhyKa@JfKl7&Q!0cP_Becr^iiHL|IEN*5hPzK5|!7 zO2U(|->)97)zSTwsIq|Ny>W`1k%-yXTP=0_l##DO9c_vF+?vNa#Mg{)O-fqxOQAg= z#1UP<#$Bm6b7{V&3_#1aaQkJpA~Cnjg{H0^8RFr=_<-vCU+Z_KD^ipK*IXLO`?NvxE*qzrwqx} z!UVV&7Ecu^#+2VHCI_j{HwFy|c*5l!*{WR*z-G89P}Y&_oUYdvZwPYjisJ(6R^x}5 z1%rx|GJB1QSpW8)ua6JcUo1AtXj?J-lKN*NEpBJ^i{;42B679=>x#$c z1kP>ecI?rgiVf((WZVH1Sh>QKIe_g_ zB*bs4q$;NKfh4K*I*B~q=<#YdghvnYi)omLVoLd{bahf|8D``s{V;aRGHw%{^=K?| zJ|T2XVVIT&YdPEe_=Ono8ghQbIlT{ef77+O40dF$z9ZI6rSawODWx7fyw|#chQeY6 z6?TuOEM?n$Puog+ztmi6ru>J|{yai%K#8e|`zL6f6>OhOJs(rO}_-(i8 zyB{nILmGzDTd+(nT=J-&s8fOcxdY_sn1UDQ2S+KVqw~EZ9a`BNMh83oqw*C3`;}>} zOm3^vXBLL%mVCPddHJZrK7myGX3tZqV3#6?D-XxQM9k>NwRFmL%hvXln9Jl!I_hcw zL2uLxf(O_>50PCh(a&JvqkGONC0+(~DEBOk?rKRvg6)V9d9U-B7Vg2{sDugU&IVj( zea0MuRc1s^fy{1;8v4OyNM5wN$FD8a?4R)~1Un9Y(Rg*e zk6_@tCVlNFzA#MxnMLi9u@z_Df)V+n_xw+6cG|NLNA&cwrsXs#K3`x=eBS!4l*6X*BpB<;o9CteU#EJ^N5hj3@w>ZR?O9x z5!0RN74t`x)ebr0da)H5Bn_u)EB5vb zl`_e?in;mRn1WTuCPIn|rf6i;T?c-I43dB}R|p|dawV=yuJDs@Xyl!(JU1+EaY|)x zm6#V*s0umPi92?CS~Aw(-EF;_n5P`7n~{B9?bWmsd+b1@796QpUZ6KUI5uwNb|L$G zj7?@oAJ1)UqNhiq=2i)wxmK(!Vg^X+;(x#m+s#mSkRjE16I6C^&ofr#+Afc^W3fsv z=&N2^iMw4ANxF>}-k>%sbmg-r|gsSA`4Dfy}Ac+7!9}tU%5S zT3%@0)Vc7tRqmLsZzS@d59o-?iYEX%$ zR4F~$L8hF(xHyG)Wz#psv)mQZ0- zdLk_vK=usaA1mvwi!?rx+(pHMMV6Fy#kzs$zG3nUmslgqX#N{)EJ5?V>~$&xjc;n7 z7Yky1h>X&UMz7V7b=aj2r<94S45WS-q#YA|MMEUvwqkHQRc~X7CJ&AI^woJ$b@%~i zq=cyNuVuiu(hpZDrNwU=-$CfdO%sGaz_Hxj4khghoWL)|V!&tgv#Cm&HF?L4Xj%=t zU#eeE4#{#V`ytO>fz@f;P~yJdtRQzJEJ=yt+r0xPB95N*vf4CJw-{g+T`X52&VzAm zano$0YCk*iDuvONtXx7`l`BxUikYUzZT1>)fBonf>_r1(H1cdovYC}zI z5XO(D-*m1m(N=D8bv|9=#7(PwC0O;za6%jsmKXW`IeDbzS*de=n08>=N zK1-tA{OhstMrqRC0qRRR5!VnU$#dW8RSq~+H+y3)_%{gncQVjCEnRDgrJIU1yfekt z%f8?HAoVz0?+hmz%FQn;{Op^P!Q5JIjH1W2$QL6{2wi16cPltTF63jHZ ziUvG`B$1jI9ew)K?nLl-Nz9^q^a!f|@pX;nKPk0xnlmFK^N~VE%sD$y_@NXHsg2-c=7L*dFs|>JUX=po!^*3 z!D?8wN+dJdq-ur2y+^USM+%WEpD|;?HnH)eC!{8n;dKv;K9;%`638?8`#J;Ro``qi z(*%lK@yNlP-Vd!&2J}AP57!Ih6|N!2sEOaZ%-cz~{ydxpITN9^nnW@fft#BtV)iid zX_Y7j5 zE;QK_fRghyA6MZkX@YC`fr zTcb^HS+d{NHFWe%sYVgR598Pz&!o;deY+ zJ?SCnl+&>KA^h>+tho2ji9AG0(D*qKJrqTPN=vdE*RV>cV<_mq7g2~L`K>D~5ZtuQ zjhc&OV|m1OPMbu;INr?UjI#n7k_qR!5=(4BL!Qgqne2Em(B5 zAy+d78|#%&xJD1&^qJU_K;l+V5GrR+tm><0#0U6r0Ci0(7eQT)sM)I$3f6jwh*ZCX z%~K?OvJlM`w>zDkapvt!t{*TXV-iG)IDwy^7ITVUizXp`dLhX#hz9hV29 zs`_y{>qiW9i5IzqiN|EH?U`DC_uBhj1B(m2m$0j5l)HT68%?T83($7Az5S(C5Y`Ui zx{m;TN$Bi(56_M|!qtnln%-m}=8>E1v!WHE^c1XSLs>2nYyF#v>5p$y#)^x~swU}< z^S<5b9=&#i&!Dq(vTbdqE+{r$yol`3ct?dJE!EAj;3u(K83RjZ``FO1Eh~htW;FL4 z;!WNz=IL4u(0P}({c3MSd!_vxycpFCR?^Qktj4ej-<}i~A<2hW$OGRb2jWK;{nVo$ zlWV638^ zH@42)SA-wj{R!B)@A%LU$CMOOBms}miyog%&7%Z?lF1lk_G>DeNh;jJaiNXNi$BoR z5cn#kSA@P{bEQj}mms{-kU3E{88D_c0B@Y0; zCkeb%s_6QoZa<>JM@ll^ClJNvK}ionCmC+B3kjh!I@HAJ{d}Avbw@;KJgBByR9(PL zl9*r<@Ip9arj0OM_ht+}Z^)dS{JhCFoOZ#qA`-uxGZ$zz7ej!x(RIdBvZcJM-}oNd{8x1;|h8f4^o zSmipJrTl)7)w+J;s)DAA3dm@JB8_xuce&9i|4I;E7xC8QfE`W|sVi9~@ zji%vg&6vAyx-6!hte7_KC8^{N&rctfiPSfdT-m1}Sx)gCUR3d}{hA|?o^T0rKa#2y zXI-;w?mUy9F8BQMYM&C=8?YdfS~R^aBQw|)tX=h4SFkbK&Pa3DTOleTBAT_vxS0;OZ z-6@MhkQ>Aqw#H>^q7lczLyM}6Iw1F(5~nMeASSI&PxERg&UaTGhA;+M8ivdo{#BHl z>4l3x>;l(moA%3xxme}=8Sz!b&evkcq=<=nJn8H_znq5YP?o_gH)vB+#Kh1s(Z35$7ujV?)#KSTF6-Md@S6-X7 z6zQ0KmOr(@j)}3n$;HdxTHuazX(&mBPSc5v>M@;GeSMSU(DWd}be>HAONGvjg>5>q zIok$E`V@2{6UmdQ*PT(90QMWq68fE|3NlkC3GS)=ZZ9_=GgKtN3a4-_E>5?>}qDVIwbENL{*`U9ig- zNBEGT;K;;~uf{{c_D7MZ#+S$nI~8k;shHHGJ5N9j6rbJfpQ;_?l#Zd96F0T%^|q#n zFGe=PS~#-so;x5}gUwyB-!0$s`_!_O)@?P+r%^q6A6J+nT4Z^FO^VT)&T-i`F5w-i z^D+h_uLBgdiP+vstj!$OYAjX%QCU70h4=W-Mo3rMOxd>^eQ<(yjx9&G_GS zU)`HL!AEwF@)rEUStT>8)+x)2Epo$EAHHS8?Z?Qx9`D^upFi@iifPP-rUO=+8)(7_` zrTu$9h7ZlXovycIcaV2y^aWvi&63US9@$7QrEjzq+=sfGoJ*ba=13o}P=-(SQ>UIP z%>BHAs}ApFtXZ-k>C^;ri$zrZnA~4-P8c`U#hM2BEeFQ*#FhhCUFcM-XvR>wERUKfpG%ESD`F}59km2|zMO6W=%;Ao8=u&69rUnzs4 zQO+*?+(I^Kbj+xPSm`-E=W?Pf$^bch?wzKq&cz)H8*Z6dSXB~<9Y>1Af*>r_qY0I~ zZtTPON3>6}$i6I$u7OaNwBG~kdA@!uZOzz0>@-8`1l>rxTXzYX)Clc3z`u~-Ag0@Q zGhK9aocT&CFELk6UsC>V5+?O*D}79^NhTnW8(cuM&qA}+;XxnQu!xO^o{C_2n)wkv z78cK~@@_E~x6z9uRInTrXhlg*9MUZgMwvb@+bI5C!tA~Qs)cMW>kEw8I&HvhjO{(` zHJ{T!@$L#Fe6PTO-e{? zy%+b#OrHE*@~^IS=v%<4K%|$$#bsRud-p?;F-JHG^4h|ig=g5VHBwygOn$k7mp2b4(A(qKmK8s%*j?Q)rSVP5rJG;w%DYp{%=mj=Wp zsy@Apl7()X3KBL0XdHzzN-gj=$Pn}}hu_kF>&cU_>iq=vN2;AcsKeN{ZBQYTp!Z$M zfrc;~uUrkr^KUql)Yb07)cyW99bb~OU1yjpqRM_=0)u$>R68ybAzuPuog^cmLw2_T zX|L%O6|&IYjnTzcMiL*8BIY&Q;Lem({)63?aK z8S-$JNLb_!(lE@pjvWPecK+s)4h zp~oNn(Lt)JmCc0$k_j2}@*Xe5TVrDKfu?gVq^mN9*rJ;0RQ6t5-A7@vW1j27#2=6M;{$RkTk#CLF1t1WGeFG0 zsVmo!Nz>lVNt#RzV$qe_FjBrm6A6(-OfeK=xlU=~F5PzOun>f^l$#jBVzR#>W6jm@P})PqL5yWVv;~xa)ci-x17GM~ig2GLR;u;Ma!RuB5JM!gg~pj)O-KhaQaJ9; z*_LIsOZe36EhDoq*FxaA%Z0+f=6vv4qB$ZjeSiI#I-(UqO~hKWDYu*ycO$H1CkIeZ;aE)wJH4VrC*>F{uov zQW|;R%RAjyPf&S=Jvg%O(k+&CvtP#=zLDFn<0s)a?{HpQt*JY#HgPYk0jGU=B^yl= zDn#YK&s;;2q?HBKk)ze|H4~Wat7Y~Hc~o#^ZjQ$@+P?}450C|6cILRqXxMU1;I z*Nx^`3_^q%U@h97TB802TiGI2mXpkp^*b7%b-QnihB94hqFMj0*?Oz_TJ5~;T@ zk)rlIWG$|2ADY~_De0`fD)@Vp9nycUGQ8B9JdA-~nS`R9NUgltqN~|=d-AJOFfhQy zQG)QxgBYg@LMW3;sC9Y2{@!xWR7q1FA42OcjYE%LeVAF=N2YIJ0|P51mMY7Qh&tQA zm#o4|;M^?PtB6$^07pXCX>}o5sX6ZKm!%cP2gE_bTKwv)Cn&e060IR2&*)uN@_@;XOfN>rH>E<1QTmu><`5E6qk{ z)!jj6GdQQczh}ojgVD=geTIc@heUe&=iVP0wJ#S=nwiXb@D=V-xER-|piR+oIVt+r z5km-EpLxop`1@5RHAA6i7oRO`w}i8#eHlBRp@3y{jtYLm4sZ&i)26T0 z)hP%#9|JUr&MP;)+gqR<^`TtJD=>Iiwaw($DsVI*gwJ&u){`PXcJHM z88Z?-Eb9zonsKUK#q;r@S2ea($*JAfALL8jjCUGtr?IbO`l@pm1>*V!4@ntiaSz3( zh&dc?+|(uRMHOIeyiwDZmQ-uPj??JFm}&koQtGX!zQih zgm5}Ik`&&sewo1Xph-;R8uqlKoK}ZVKjLV?)TFFfrxYr;Rl5nIz+RQgpC)w2_G>#PL?mO+-<&OP(T<+Ro-Lg&9ZnjLp6m=&d>7pT*FFb8Z+Ad`w8tiSQAvD5v8eS54S9SozGGNjNF;1^)>gk+?*lvxVl+s*Rs^I{XD#C zv5LgKBfbR_izWps6(x=EA`l&_*(u=-N-meUbINcO$y~u!@1bAlzYjVTov2=cuH`F= zH;oYL+`2Pt6zPRqzJmHDYF(wFPc5{dn?Lv!5QS6v2j~(q8d~AD&QoTyYvitBdOeE5 zG`7s`Bo3lf$jpel0}8>T>iP-TT21^_nVumOdvf;y>Y^pHz4(PcK4@{tf8*czXpUdT zWmuqVIJ+1EolZg=VZii=&G7f*MgAuSpr#xH2$56xoV2yWa4oelJlc_TF?N&^W8!_= z)UuCiT+Fbh@!RDyP;EpL?-w4^ z#U=LJy%D;kM$LEU^4)t9#R%iTDQ-(90ebtAp4KTw{T;e7p8Ag0p`k$H6N%=>l$dr0 zcKEz_CL@7VUg8bfjg%w`ca|&9LryPuJb( z0Wzq?Qfu=`#nh99zcW3ub<7U0X!nueT5=QS+)3|Y+-Mo8@$5K{#pm?9h^%&iz|qHv zN!mkKET)t5_kI`M;} zMZZQv#N@{tl`Y|L%XD=Nn9UlmZCG$y+iVcz0vTlFASeVkl5=p0tyz^KXDrBJRss{( zA{NR80Y-_>&Np!3lWHb&h?RQ2(sTqkXAd53e-%5w5AI_Ij>}7O=rs|EPrKm!)^5G1 z9_160rVbUE6gkZQFBsyM^K}^N!|0+OyE$sA7WCJEn(*1lV}`jf%-$Hk-ZwOT=z8tG zoe-{rebxu^`&j5T1+N+4urE9|bz}q`Y~9q~_)sE8a_{6V5$=svqYS_HjT!bhd8j7v zKNmCE>GaP}m%+A4-_hTPc1BatY z`<}1cL<$t}Zk^^laiKwF-{CeQP`S%xDI|etEEM~i^w6X_sup&HG~AF&3@kzhEClyz zM80#o{+1Ag#03~_j6$#ko;h#>D`39C5&IBF?|87(-oV+1Zwf`z-dndDX2g`ds_?UG z*G4?f{1dcVfbws{7CQlqak?Q$_%6S7cQn5l_(gm6fbJ*^f*h4(I}@eWZD~ytlMru= z1v%=D*hpC_8OfYriaL0fC2z%2uAQC@4I7*}&f_Y9eoGd5wN4sgBq;(hA~^+Mp2LyBAt%JD>F#x@Z<)=-w}H9B4Y zqL!Jk&OE`0c2vmMXs}-@L7KuuE)R8zGvKv}Jv5S4gKeq=b(hbYSpkj~s<6vMGmt-g zbA07+MT(aZBiMgkwNY#wX>~I+l&T4NtwBy&e%5>!NJ*7Rr&>n!a;T+G`1!`DVs4IZ zUh-tiLTuNa*Mp%Gk@}}U`c>1gfohtc+aL;iG+t2pM&3efrG@iDv`lD^7GQmZ1mugB z0&(dW_8HsDuV=2HBUX^$Z3=f&0)D~Iyz!KbYi)dZRiZ^jhPmw$L!wxiV3)i@M4$Wp z{aFaqEAW|lgkuZI+bo)oIXOwUHQ`P{BvX-u)I}qesQW%nWLIzqKUk9QJonsrxc1R2 zh!j)67prkLi^axn!8JRJh@9C)`^7kDB`W8=C8xweCWb@QyihO2k39-s*CLOr7hnb~0MURfJy?_v`OT6pGcV52&g+N{6-IDj>nv zuY<|kw##*+_4h=H7%PePYCi`_|5PyKp9vsG!N`DU2|EIbaPf+aa-w9hFN z*ZKfsfkpes?X0?Tx$*awDOP7v{F_>5`g2@Uql1G^tPWC5&#avE8l*l&fARUV6;zYu z-QcAxbi$SO-^6^cj8jNKPW18E@iu%a~mUJL;sBV zC|RnOvfLUgL$dq~b#+vSD{#J$-mob$OwUl4nOY6mU+u1rhvkh<6&p&5RK|*L-kBzt zx+dc4)9tseTEQUV6ix7ZF*RmYXTuIeadSTvkD`kfQ|;kQIGr#I{17f`3($`@j5)=V zxpaNV!Hm<{vzDTS%fE^E?Ms&T57M~O#$eeP^_mswc0W%*#|CT#Q6<{I`t|l|2^-*? zgO?qbl>2IM<|0WosOoA@X@)(Jb8gn)XARq*v~dZSqFuUO!|HLW8;~|9_I|HvwP+N0 zc20PJ&{x!apf{VV@_XA7Nu`Eo&DuVPuZD(jRL+rE88-Z~YV6;5nChnWY5SmO*7g{` z!e+3Wuqq25j5osJ58ZF!LPDk4_Q}{$^M^K@kz-Xiv9^U;WUr5R^6lMQ33jmqzxiCq zuCY%XdJGCUb~Ucd$ovh5eHUJAIB9Tieae!FqPL1K9rwitoajc|eOu-{4TBx~uUI8x zjugN4U3$OM+wF@NzEw`8Dq^KXHp|TGL%kyoRH4SV6yvB#7@eG=bPx_!JzIIApd;*= z{Q*LUF%ar?ap-m$c&j<3YhHVU$DujMcM0!}##YYhLi(zml-6l}@?~@~yjD*6lrbt@ zFT=_sChFYmZ??}@<19E2rAU?`yQ(p_UZUcj`GIH2LrBUSQ%0Px+?9!76sffHQX-q_ zLu+z{@BNn)w|gS)ajRQ<))L<*#DxCPL=62@nhnyP1oTTYd+K~)ltnWZ<=eJU;>eeE znRs4w<8%3#p`GEswiL8OhZ@?Yn8{Y|F(l(URZ2-vpJO~@A7wDEykuDPuzenFD8jtr{r*Z;{{hu*`oZ&b<0jG zNgK(#8k&ihN0P@-{AKsuvQhC$n&v5q>aDt1P;?&}hD{i-b8JK!1(`jWv31!rjAG~g z8Qk?t)=522Pu1f5G8GmJN_2tdT#X9aEM`W4E&)Ru`h$yJC6S8B)GEIqMB>G9Dg1Uu z$4?|QRGlr_Z@|askz-;V!CR|hHPD%WJEj>GyiH&&(A9qMmW7DMt9mTG{XWZ`nt6*i zjaQ`p{D9wur_aeN(})ls)d3RNOiTPbS-!oEKfksZXfCV@i-$s>>QIK^mT2!q63aB{ z6`GFao0HOOwwN&If1}L2DSp@6DBEb6-$_@jdBpY5O87;IFfJB^pZ?)ni6K-r(K3IO+|IXOu6~tR`_a$L zOrHH>Bwfr~^eD-^5>5x@d3?*rQ%kF z1y%Um zcht_+w}bFLWk||Ps^y;}ZoD_(beI1Uv&apxC?#~6ezH;ZYgr>MQ$vX=LGc&h%>X^W z^q93C9Tf%<%*~vKgYnoN#Tk$X?xoHgS9ni=)KO{ZjUf}V#Xa!JZU*g3W&Av3Q>2ZL zn^0ojUy_eOnuYqi@B}E$owY=-Oxw(jQg1HRrKPIsjs9MkjWpPk9Ju{t4kz^+} zb90ExcWMg-;!u*Ewnp{5^tGD}d%%tO8M{edfz~woJ`KC`RN+qjLe}tu(iB$(8?C;3 zw2T0_9xWH*^|??7D6}WhuWckzQ#w?$WMQWlv8UB=2$|%VwAi@uOD`cz5idj{e&*7h z)^-;xK(*<1&@Z^G23Q1hHv}O+Wws-N2Z)Fyf9XzmeG;C+aYT}%{;;c8mv5?RY7=}V zPfyjoQy9z#Kj)CQtTo8|2bZw6G?NqC$dicciU*}aoS*5q0%r+{E$d`;sT7cClaMi# zX5SJvAyeTmWX$63F<(7FK%Vp|v2xZn*AOo05N{%G4Ac4I!L1f@m(A914&jDCa)-?) zF( zqlaY8W?b}$VGFp8f{4_sSOyP|r;XyGT0$<(>4`Xj{_F|Adb%@&)0ZFgq3O%L`Nkf+ z$EkCDbP6X-tq*F;I0TUR+ZU~QDZPQT)kny8v@MgMnR+B`H|ntML3JWPr}lt^UgY4X zFRfL*4m@&uy@Ah|2}J@M-`xF!IuCm)vg!`aKv#|&EP_vwsO@Q+dV&a4!r#4Y)h}4R zFpWOvJ>BgM6E8b-!GZ*=xkMu4tcLXXYHc#)vky1r7Hl8egZA2n5G~`9=)uec1cFm%4Yw*?m2-7)4Gkkg&Qs>Xu1$7!30DB z1yVVcI=o@HMG`G!xfWDOKk8E7|PK!T@(M@cNXzqJd}LF2tF(Hmo>ZwhK(LIP?Gx? zC`Wy(A!Q*s~lOuIxmBxchO2jl*#6kx1bWJF}}b@ap~lr|};dc8O# z-wXJSRjZkE!Omkti5WHZdU6>MPX@I4Xo%d0Y@KpJx&~NyzN};)1biuRta=BVak`(m z;*hhbgnLHtXry)_#&!E{;6T4-GY#$ zF-iIsabL$#Z@#cZd*rR&QU!XIT@MA3x$&BleGd~+o=k`$(gfqBNDlj@G2C~MI>y&& zfL=bs05r`Wfnh<_-ny9R;JMfS3fnIIbtOl4G~?mzJg6_aQ7soYXecPAO8MRI);7;Z zPrd~sq_g<)t)u7!fA{tW3zo~jwFh{LGjE%mU6ybktEbs70)YM%&jtxktWBG%!7Eht zuQNy1nzZGa(hsW)A7(R)Jaq77KQ&4$B}S+=&?J&7?v-F`e&13r4x%^~lYM5W$?eP` z5V(p}_p+aM9xiV@w)TFiv~@2rOzba@d2Y6_>9)FD?8H+OcPg zy@PKD?=C>`d&ZfZJ^ED9Zd2*9Rs?(kE>W7pq%pHYOcFO9$GGxiIBJc*7v}m0kueyb zpf=6t#3w#9JsY@J=2`_Ey)ojMojm5_RM z#N?^l6dt$)&Y}dwv?RXzgZeR1$pnN`Sf5JDThn_|o#nKanLY45-b<$@*KtX-PdP^V z$)%B`n;ur*P}7A{E#`zm_a+TdAI{%#VB>x2cG1s&Wsi|;JUI)`oX~0tdx3& zA77-dlK~}hHQ2QMzlw;xs86S7Io-5!I1n=S0~CP z>4bdy3mgu$D$c0TWlytwG_i4Vv3zFiy6=&Ea}G3c<`+uENro7$y!fQDS5%4>S}m{Y z{5+j-xGsmjbKA-p@Z~CT2jdFYbA=mB+RpoKJkhTkLvj=U?Kx?|wv=PcFHIR!${Hpy zkboxv6l6fnP)!Uh1s9Hrd(1GbjvHdC`pZ{Mf6nD>-(0I(CTZheC@o*vpiZTVA5s|+ zXSdIelIA(Zf!tkgnxcsu|7<-ZJtfZ6@mApm7CR^nsmF3r3hnejKj|n1@5Zn|^*(RZo24)qw?%q06 zgTZD{pGT4bx!pU*BPbNIX|2p806?cw_|cLPhbfHDeo9TMqe zN0j$M@N2I^wZe1Y<`=X;h{d4%PO|9LY<&n9pR6YaQl$Bc;eUeQF8G(>ll5^Xi9~T z+YEv^NfYI=9=W+6k~#$`5@pflMK})`v-LmpiRn|ZCY)!suxq{+n4-2bNM!R%PY5oq zo}B_swu@Z^VG->+n#A(59j9GQ!6!Qli7GroG3#gbvEKPc?9qQ$Te#z75qqJNj_Aw1 ziENl2t%iETfZ@>0Yp6K)s`RaewR4<|{vp||dHQJU5VdA6OEZCD?6Rg2>s!FGoYl^K zpeCouKR95RPf)rLUB${w_WO(l$u;!C2Tv6vWTs+ud>?tECl!{rB-L8K*i+cJwt|y4 zu8LT%g0R0Gzxi>*93Ba*?8z3yj6AybJ2%D1^ZZa-8ho$mYghL|?d{GS`F*ZfC$`>S zS+BFjj8O1I=Nj=t4#a;8i^3R2zeGJ}tnr(3+ajZh37_BO4=~wJGYhZuZ<4 zffO;^apK!}p&F<;ICfDFQKMBf87T$Re^p3SQ%o5p;&p|vi)~vlQ%}+;>iJx%Lx05& zJ#8Lt*u1ugU#rA>i>U9?D-IVri77>x5n+mU6~$24ZyQhyK52!4>W)vVs86_EO!KPx z;$^-f^73UX9K+P2Uv>cY>UAw+QEHXpG~_vhvxu$It{xs`j4c?)G;4jvo*z~F)ydJ^ zh`3A_iI1)oncmGeA56m5$s;(4E?D&eo!+N9y`VLso!&;Uy|+xP@+sM}nVD{s-7s!N zX#Y&*_kJrkZ--kfXyQp+I!swFq*%4B6VTCixC`U)geGZv`^|qwZsa85_9wZFJOT+t zV;qYXB|hulurBrlUSfe`&P0`ngEuoYkwqRKmvjw*wZ^P#jz5ULPdaV#VK-q3<=y7) znJ757;JHUadDO7sA|uY71ft1{Qfu+`SY6>|PC+>C!33BrSuT}(VDmLyi%it4~yssg3dZ{*WCQH2B8WLKqRRg)!aT@HRD;mp#0?KqYa>!{Jt zGt|z$G{Rrg0M8j>;&CvLJJ1whlYzw+#lg7v@2(_WM)%dKSf)(Oo+u+GB+jDvctf zgBIKhrY9yg!Gvv+T}9Mfy;`c8D!NxG@5@Iq=L{9*bN(>Zx#|u}xOYQi-sug2;O)iS zgYIZk5^v;LryU)68GmD_&fWQ%#UOlAHVT-*cnX=GHx6*rqvf@#3M19Xu82eGG)Ceu z`s$(rKPamlcd8`_VEy!~4^^z6yq)b%)jTW6GUA(gv`ObX2X01s?K0;j776wQ;Y%2HkBh3Zx3Urq zem@jM5+-R@*475$PP`qh7pf4*{z7Gg2ft;%ZDvJcOi(}FM-{Zh+H&}&H~l5Xe+el0WW3_!}D+p$N8%PIP#(Sho^>n*n~|QtKD({~Cgx z4CmS3m!K3M5sr_^t!aE8VHcs|^u{L|tU&QTqr|DkAT2*pc9mY@0)-|)E?VYzP<@rQ zU(3=_@K};nlr7@X2Y*6v0r1bAJ67wW3vb>{=aN8Jb0Qfxc`BU5zZKL$8@F}tb~|`g zxOh{-QBIZduyQQ9klR(?9#{xXcKxaG_5%caV&Zv3g!B?zTBXzPXC^BX8W(;mkbFeA z6qs~^Yu#4{)=!_SZ)SQHJ9O#s&=VzvU~1js?xWrg$q-o!k!u)7sNra{ZiqipVI~=N zqXv$lY`kerp|>ZIosi_2=t%x69z0b49xs&$J((y|tSZFD1xJ^_7G}Lgi?+TnB-KibQ)zk7cm1qrj)9M%TzFbQkJ5Q8}GB z7{rco|FXTXCci#vsJIxIX`&okIE#caR5ccYk!-$4#kKqL+CigaoKCp|q(f+M6tK%@ zBlaFKVwb}xZ`fJ+2KT!N(Y( zV)D`j+L#T$3&_-<4vb|m!%tOap%r2T0!~b~DX48w`Ga8FslM4CLJ3O@y%d73k~ASG zhi@aqSgC$kGAWe%ifOj1HI&f#53Y_BP0B&Y(q}XY;5eNlT82RoI8v>w^_0YMinKsL zPHFghwFY^433g7aW1ZmOY~#+=12~&U7JPIN zJItILx+I)egerLbk(?{G3RW+=Y9Tkfe~~dA&!XQ;Qe*=NJsIm|sAroZ6F6~I5y^)F zb%-|1I76M%CQ55!{q)XgNW?hNqsVT3kJ{f|QpM7`3a0)7)Ynm5dmSP*+vRXMaqbAL zDyPFjvt`$R_l+Qmo`v)<%kTJuSmCGtCe_Mpa-4?t~(wABB4Qe5D01c zRjH}|_|#c$KQmvS>`unS^KNK1wC>lMkYVDU4XniYGVt^3082L$P+QqIW@s6L>&(%! z;&8*9Lx-ybjcIQ53r0*+Be9fuBb2n1;a#5E#mCG^yzPtY4J_M6zhjc8nW`Dw4bv6&6Nv1Hn7Bk*)qP4w%!LW>`v7;dud_p`Rm|_`L10 zNwaCdwk6w1Fj!|iXi}i1!8_Yl8y$8+UwmK(UGj5&HoYC)hSee-3sjdcVtGc04oS`h z&0RpBvvyCfSIX5+8rT?S@zLc6{}R20Okdt#+*4V6@XWXp`E?M9zcu3*YPJ1OK;!2I zc6!-ljZ^%$ZZIO2p%h51!3wSNjj(QW9vQGPjdnCIFC`H3 zI1~YlT$@%N{b8TMy|3%aSim3UtQ>Ul(VEwg4(=}EPK!t0UCR@3Sc~YZHfKimdJ$nc z=|nuP==(6Lq{8jIxeN*V5lx${->PopMnzDJCIqMKmnSo3AbQG4Od7 z#!%jYu+n^*p?*=hX~235Ay3 z9I)hyd0D5Aj|M<-pu#WE1T(wCepYZca;&^&W26qLZ>1!+h+b)ctT_rkwQpo z6j?Sf|50=v32p*G6ulB6vn1!7dq>VWbNXSn_E>}Hs`~HAqC0!T9dT2shO%hNeTs~> z(D-f&C8O=YWy2aCsiN1rriHW3%-eu;XdU0Y9>kGaP5~kxbJ6hwQ(_jKp32Y}_ZF<< z2ae%bsypJdYSX10X8z`N5k6b;sV&D%qUuamHK^F{YA*b6Nzr+b9(7DGy-)MHtg`pW zSwvkq>6Sr|86J*tSBt$UHB-x2tQlmD$b7Wag|k7ORCPB%Qv@D&t$k70x9?tEVWNq>!v5cU2DDBJU)?BA?BijJ7T|+i^xk>ReK2OPdmY?h1 z6~T$)X87wve~uNR>mH_IH-;^*rQny`3zDuYnrel;NqQ-Fa1xbR{OrKdadBq`&tMOa4h@zGm1A! z7uGT-YUelXXKt-OYaL5xfph_;N9P36-ZyFUb&)Th&}t%6wCb>De?R+HDuJ242DQTC zt92y6FcbJ68q6g^MQRq+t*19k{7s5;RGs7L1r{^B-_v3X8!y79@Gq#nom!~gqfdI;(Z z8fJ@Z_qm2IXv3-hRM|0R@A>Xn>8YzQqrZcK20bbWm&JcO7IUwhv?0kfX7sxvMTTTB z+1R~#E`MmQAxPKTgI*n`U2_jbQzqOyDt7H*yf6Jn*#$;L1|Sdph@(Y zuICr-D$0T0>%4_b#3OIMyJ9ec7O%+nFL?K?ecWD2VqeK$!{HaILHZFN9JIB9FJ!Nl9XZAzrbdLQD^)}nz2e5Q}Cku2tyqLjf(z$R< zNgr5qXfiDLz`%KQIrYy@Wso%iS9_YF+W1&P`A`XOz zzY`Stl`z}s4ys8b7ojTUME7u2dXQ_IWMM)$&KbEFrf~C+J}YuIGBsG(ekB#Xtue@Q zUZle&%EhWKV8nM7&AR(0g-sk}@U51qwx8iAkZk?XzGwGOId#l<8BC)_nbwf73$Cru z)3{kq-QaY^fEWgT1Ez-@!bSo)io`wG6lcyEADvwR?>$JVEHouI%ziA#!N%ISrtMg{ zl!58t8c(JrYNJHz(3GU<;1DYHMD&^K>5)082Kl;}KjJTi;TAbk-7w9C?LfUY{J*VmnP>7Bzzi;?)0Rj-6~QlEyTQ+`|PG z7iH>aZ1inw!59&4k?jP-IFmd$n6jNCM-~%EMk7c@$c2T_`(Yx%@|Px;!N4^PA`SPQ zf#E=)MABIeD%Ic>sy)tsq&}uZqERZgv$Co@E7RW{3_Z9}&QJ*OZo(IPK4sYru+<@e z&y}e!Om_tl7gBp9%imEZiM9W9-~0~d^OQ5`g@q*29{(o9pH@ zJ5c-ZHIm$AmKTQCQK5G`+3EYyZ!bC?;5k`Gg+MV()Dvxh0TG3N*y6*tX9-LgOpW3pk zY>&oOnw{{`Lr6J#ybTNaSk1f5g58UCAOnW;3-{{+_J}&Va)v2POB1L~!fdl}#EZKI z54uW&g%D*43z4G2&_w_Cf(r7pEh?V#E@WamX^LMqzf5Q_x-D2{g{zL2O52bPhNigv zX_dvm-ek%hbM>m9HHt2G-Rzy#vjU50>2AGdJVARjOb0IN7`hkQw4oSGAltVd#S^%l zBuNouWBPm~c8b158+5kqQm7k+MENh|C%u!JR`Rk1wfyC|-^%`YjtOM$g!knM(G@Ai$Mp3f4Ue!&tNtPI{9tj`8TOx4 zD19@y!0qK(3g%A0)7TsE#i|TB4t`jGvrq9Q0R?00r3IQ)vXzO22pV3LBu=wR5!uvW zqdu#?W9aiszqb#XIHa0LxR|XuHKdu0mCtNhK`-H_9XFaRWnk+^Gn^|a{k&=O5M(gb zLu}Yl0XW=47>4`PH;TQ;EJo>L^p-+N))mlA=+xT&gf{Qwd}Z=z9EDam4hS|CqxrZX z+eWh=M^-dov>UgxD1V|ltCy;A2F`r9G-`bNUCGv} zG@1%5%eCE)mSk8Op?bTZd8(xaAVWvLd~|^e)P3A7wBC?df}s$z$22#6u0GI?%4kb* zx)XRPcO{F+KCcPx=v*%rU3Aq~H7fQ)eOu1%r9VA(dr;$1`1Wj*hl7!NYY^$zLb3|5 zsaNq5uD!WA1S3Nx*LRy(7J0-bdX#R>6J|k$Y>1aG(EonQuh7K!8m}L|uw9G=j6`76 z(+`u=h5CgN6uwY(b8{>Uqt*KM|>thOR#trNn`ON9jdY zAdi2FiW&=des(eDmBGy)8lQnRxy@iCJkmv{T2kAqFLHh86(g!u^9_-2$TOd%n1=6x zIgLIQOT7{=psw*Fz-J$HKc09|{q0|U@*%5JGm1U~I(gw!r;UB*@<7=nOz(tHM3#uq zI{K8{_^`M6Yc4N<4*j>w+CMT6B1W3uh_4zXIY{K5Q_RrCWwi-LZ}dpPrcO&-`5aFB z4Ird01+-9r0emF&4)Ke~d>cZ(ZnS~EOmv&1{a*XugbMU2ZMGPC{04~kkT zLQI@$pvO?iP3bPM?ciA7o9VBmJ;P)Tju$Wc=+%z5o6IHcs+!OAOU?TaAb=L=l+5Vc zNtY}&-@ZpN1rJTbaJ&{~sa+pCVxQ57W2_khmkULsjoHdctiJVk>MK;;a)WVR(7za5c4v~(|hOhhu z=Mxv~UB)UQi#_vQ=V$0c?@ORmVhPBQw*K(Q%jSMrrF;GTD$0TDnquI3R!b)xqJ!)lFBW92_j}i4WJUE1Ci;i2YrHj^P>w#AXa-& z%}dC95z=!lX=A^GH067=s1JG5cPG4iTTNOiuBpST94B`^TjOO(vhJ-CxGqdFJb68Q zKXj}{B!S5uRAHtDt6C39@#f3@1Sa#$Fq4M$4&jXQmQZenZKi*N(8I?oU!KEb ztu|hByLy>ub2lX8DriefyVC0&Sz0l&Z_S=7-S+~?DGTUp&wG6#SF&a4rzK!RbD6iz zL~EDu4#&kF%=emiYX`Np-Z#Zl+K3Z!wWvHhw!awirymsm!t$ucpYSURXMgz^O*`@p z_|^8vw)2T739!7I(B9|kr1EFnU4I-GUz7Q9I;jD=T;a>y{n+diFYpKQ>7n!M;%WZ+ zwmn+E^qxz`ZbwU2A>A@25iKZF^H@{S7W6ip;Rp7dZ$H%-dp@pc64!6RW@=A!v~m=} z^GZ`>jp1YsSsSEI77vL#zEDu>XDG+eeCZn?IrKQyi!8V3pdmuAvcl-QE9Vi}jLh$& z8FlbHPkmvuB$xPgM+qX-NZFcAIHEbIbo7obsuURb7Lsk%o-BiZ$AYe3J43F zKM}Hmr(+1tYOf(2^tVnuk7=)})BHB)yaxZJ`dbFaQD|{12^rEKUg^EmNLR{8q{+OH`KqPFYhE3RR@WT0^HYIQ>F=9= zREL9z7~Z?07PsCKnOh?r%ah7OCL&R(oe^cJe3(1w|KOkuMQCW<_;OX1SEOto(dM9? zd4jg3z3yBk)m;6X`c|b0L6<#r%pKVe&+PXnDOVXtqrA4VQXsMF+noX{l;q9JlGAM- zW$Tx&^_PpPQ#$&T6nEgy`BR(c1wBdDV44T8frQ?C51(954NgutNSBtW%9iR`n9-bG zSR04*C|0@TE79vmgJ&Nl0LTZQQ%e-5H!+9jj(Gd{s;5#cU@D~tIPwhl5j$^# z>^tXgMP}YL5&Nd8%6TY{0hUl)HRSM?(eMZGRnKFg$CuHrYS)li`3H6*Ch2$;(QAEa z*QZSX@>--4&$D|TsFIhia;ss-bsnwOifJBa5`QD{*mIZ$x~?vH4C1BcBiwV+ujS(q zNX5(=1@dzRx_!3SI&a&b=svqpHQe55{L|P!90n3$Xv;9wtegzpF=_11GyCeLF zl5tVGZy4%CppY6S%n!jcgUhxTfpSXVMN9AT6rPfGpy#H2xEg_2il>c9`s<34XQ zvwJ2dMD59Wu~s zB@=-$btAGSry=0v2%-6iVJuzSZCK1Y&E+~~RQC8HWo7EL(ee5lYBOw6(}bX8DL|4z zL37EE9V!cs_as-sb&_N=)l3f4u{iRSknps*38r(?HyV$$jp^ZT6UzCXJwq}BMTXnZF$y9g=f~Au?ri?^po!H*y zodgy6t=2mZxHO)){PgZvx>TWU_p@#CxqMBszT|#J^V`NI62`mr_&YRC)kVLF9e(O< zEeCtS+dk08!X5BP_t$LT+l8-E>@eW! zlfA)lkQ^QNjdfw%+195ZK>q+~)>Ag{qLF!4|AJTMEvcxg_h)5*#MSmQV9*(78%D`u77E7RjivfX z?X6E4?WG*o>$?&k24L8eYJGL^F}ZGbA*yqK54SyIu~Gs{hSG8RxeTd%=3%--+p9Y4i-!_LHZ{&*&!;gsX@Jg5`krwZ8ieS3fp-9npTa7++A_^cIenn`Cv&gWdOX>1(z?pyKb;KQWtOG5jZZ;2{u6%it zdxlr?@@{yN)$f&A#J6aiFNAGwlME>U*X(Z}T(=7ECl;VT4%G;uvM|CoD1iY!uZ|{y zn6jK?N@yOQ@r&_`IE93}De>DDXK%6xx(jZp*95mh17`*H^jNH;E3x9EMNl=?aq z`nC%8kD=}uTANRDNtXVGyuPiC^47bhCpza>caVQ*G$!W(Nq|$KQ|N}Gf1Cq;H$ce0QhOj?=RN0!o|e-)f6kQ|&_F6N_PqEIzK+cFp?A+ev*g{+aNN%`R%1)5 zN%`{Wyme9{J`|;6m?V!u1M5|Uj#}I-amSN<|GpJA;&5&IKE+j(P;8brRrz;R2b)4* z%6%?_G-=As5vR6Wnp=qFz)@JYQpu@%7<1-oIGfQ45KeX0sEffh3mDYk>lg5)@df& zNDp~^iaY#jHw(c@;R(idNWvz~0vak$q5|p2Oq5ta{7 zE#n%KLol(o78jFA%##aO17cDxeP|{o5~Q&Hx567n$5r-u+b!hC++yy*k3R3+N4c$CtN2^O@^F%*B@u!Ia%@GwsQ@PhldosFqIu) zywi`He$pG?%sMD6Biju@Xa9%zI;>0w|Xb%%a~?i$aM0Tw^+f`{bK@q zb(0m?t~64#XN=cV4w&yd0Rl3mI^D@IPC=R(M6m4G;t2|!_L8Bg3+j@v`_Md zqU7bRQKH7ET1M1Wkds z$9`&NiAWuPpjZqwlC_q#EVY4g{LC9%+) zPE$b6e!BJqd$|=8ieyMV>`sZP)A(ShNo_5J#3W1)k6Bj95CDN@A`-#$tP0(a57#3p zg-XasRxR7(gf(RoUnLvjm{Q-cL>CTxZR50&)BbkOZ|s3H_y9*2`kLXE%(08$W4JZ4 z5bM+O*?*pcc+v6Z{mQfhMJttThhNgX>ej~B~ z#C`18he>{HtHv}Ra_GE1;#MGKJHYtDMVnzXKl>JMtDqt^SAlz<%h1DyCu>XsLzd2d z-iXmpuL57PL>|Ey>h8vy>Dox}lQ?SYc*JYNEt{D=DrE4O+%iNuAyZg-qPe#@TIh`Esk~|E66}e4$5iV<(Y2mb@Dkn3#$uUo@&rlp`#W=o+Mc|)jNa){>k|wishj1P^##8cVI@wP4#Rt~8!?~nfUj^x__CxJ9f`db zrnbvtIMK3XY}%3U?Yj26%Iv`vbuqZ}>I3pv!FB?5?s)l?ki(KT9YBwGm8pgj4oy@BIaeZf-xz! zLQQ!$gsSUl0D$_JmwL^O0n^fLJ&ipMV))s;Vo=K|~ z)ZEMB9N`m4(lzWhcbEFAVO0kGapmggI5&o%c^jc@5h4pMs zCkKNl3@thFDDx39QQ*lnGP%PXJBKmpRr}2m;NT7QMHH>CF@}4+oGi~nbIV`EiYaM- zWk9dT7`(5}8XHYuSoSY$N6(5?1qNZ;juD@ktmfNc>n1v2WK=heB*0izZHYuwCcQZA zlK$o|tUX%aQ`F+PI1#f3(dsb^(_d*(tg*kdIB7X|0~1Mxm>qa2doSgl=E|!%7;gBu z3*Nauhg*JabJ1#=O<2T_Tpw_r>C+gbI^lwHA#ODZrS{qBU(w#@br74L3gzno^Zv5x zbJ#e#@tbvGMz@3P;LvS9(7j}?!`a29^B!sap!D#(w-Z`=C-EKXLc(sgH~k^R;Uhd}utD2*uPvPG2@Kg*_N9>D z{T>Exr97l*qiMCw!1yuy-hEQGQB16Z$YJeaK*f!J0&%C~=<&QnOr#;Wa0Z(w86nC< z1J2TgC3AD7itgmpmof86c((O1=WD;zm>zhmpZU&E1T2|ZKw?q1(WkFJ2MsN9L3<_F zaZ!H5i-h-&W>+NbgLGjCPS`+zIJj{r^(85bymEJsek-l0ARdT|>V$ zX?KzijN*VD0!ovO;!&II-(_~*+I%uVdKq6+j^O{It+<(DD4Uq&uJxN5%aQ4Q8Rdz1 zQBloadE4BObo;+USq4@$D)2EE+56vw@UpG+_e1&NXR>+di$hZC?I%bPnx(=I@(<=z z4Q+Mv1jC8%(nOSCJWmLw2Z5EATpsRqnOh>(BXAk9br0_t?jaQ@MT^r}ltwK5!sgs? z6@M1a68KD+n}YjD5|$MXCCbuo7^Ephs100f^*#d75SNW!y-YC5tO++2Qij%*-{2Xk zxN?Z$4Hn*;@X440spY-|^3wsHKdwaRqUfx$8P^k^HP}Gk>fYhh)}#&Mqy?{L7c(oy z)#A?0vF%N5b?O$tx~*dVjNY`>zi8eGmOo>deQkmPp_}1iSav}%Sm=W6Qk|wNf>ks( z=lP+QmBJbjIWK7zzu1C&td}Ip4nl_PFU!iNdPPin>4O7HlgwVaWsGjBLvy(qka%$atjNbSuMJsP3a*r)HB2t10*ipw~k=uEIc8A?m zs)wyz`}ufvYwY;$@+A>NCv4pq)Jnqs=g`A-!b|NJ9r426rX1s^;HpR8Nn_B?%1gzI{*A4L!mM@n%AZlJyT%}2 zJ7fX+;J)MEn|4n%(}wN;&dbeZ`bC2KiX^^;lJu>zWwt@Pwa~UNtG9YzHJSk#r)-Yy zuuvKV?j4POxQ5iixjZtg+G~;8DLgRc91QzUu6#Pw9vF?GI*y_H>T6x!Zx{zE!OVE5 z0Bu`;E$i&s#&6m-X(sjgHOXcN#yrB0FC-e41*x?p)2AZ1y6BbauN?$*bkXC}kzw{Wjg-;3H%+wFZ7Ltaz~+C>Cz&_*1HamZIv_ztIIB!-HDBKm22b6Ki1!nOOapz?lH~#EXryiw zg3atZ$I^+xcD94AMByg!3A;qX)KHT)bGIt&vpB1#&W;A{U!ZJ~SN3$3qh2X9jR2VGc+lIy!FVbd}p%cQRM&bAW^9CfIFJWI zvrnV;CUy;ZL>bat)TuC1y9ipBep%qvYqzO1CN(1BByOf8L$A2NS<%S&VPpQQZZ}UE zUJVG(!n9F2%M^Yqei(tT`)ZQkiu7v--=sQHLEzv4po66tTb?GOPmv+cbPq)89taMd zLM}2*gY@#v(RGP*Wq-^(0l7${!F}#P1(_S7-`ydMG|eX`Z+ul9)?n2Ku;#bwVw&UT zKz3o9Ro}&}So2_&*h? z>gBo4#m{v`?1Smd)+V^SvUDMoE|9`rXA_0}^Vj?6kH$`NrgV`1M)$5Bx=~!AO2V|( z^xfPmY`qD7AFE9B8CzCg>Z@w@8hF{`Z_d5^D0D9%?A)gFy*-oAH;6=wr%0TVBLlx+ zcWf2&G9j6nw6stk@W%9|-kT8HQ1Ud(#5-`M$dnRf{>&7=h%e#FNEHlr)z0ch1(a{1#bz z7A6ZPpP3|$?yurBHD|YG4+5^}LivcC4_1{x|2PP1q`W5YoAGmoBW?yx4`08{UV$)0 zFmPNGlK%MKJ`bls-mg5lxFvls{0KmP!@CHWp_v0Q0E*<4OzsD7CNtLq?jMe<=~ zLcCi0Dkt%gm{8zt)qaB44Fi=+Wg1zWl?Lfm*w6e^Fc>0u>ipqi)ox?HF~ADYu4xU& zU6-z@i=Gp5f-kGU7i*l+LVY(bP{ratqW-Kobd6LiRU&Svkgv8dXeP%+NlK?VZj$cF zrEJN=dT|e1yV`h;K_ciUy2$zt8-XYAE&mowc5IKWV29$mE~6M84h5EVLUp}iMG;h* zgkVX)ZKTjW(^iG^&O&<}^Q4`-q{4 zNA@P@VsGw^=1IaVElqPx@AbS11^WQZz0KV9I)oydzw?O#8bV%2Bs&Q}j{Nd>B4A>; z1Kz&~24pNsml6nO0_C4A{b`_g)7j&rM7Ns)V{Zg-o1w(<=| zB5C0~w5#t|i|hb)eZ7W?{$Ph?L12xDl$Jq@q;rgHmqb0Lp|Pag~o&J#I2{o zf2yf5{Pw%iIaX$}2ZzNi<{%fk+VG7s{+Jbn+~yWm^tHjtpd-P^T8wBqHpx)8(F=G4 z#oN}$CuwJmi8(kUX-~;HTK=DiytJ3jR_>9+Jg!;jCho{6%+`VS*RD3G*wMqG;THVf zxiKuH3YNrqAHga2Rn$e%Kdx=xU4e-|rirar9DPUoKV2wD>ibiq#R_u?FUyY+vkJj> z744dTE7A}jOPQ)tP{)&s);2PAUqgX}?>c6%4qT?CQ0N#Lnw<|3 zYt%oup-m8NWavOUmb|m9(G8h67G$UD`{nS5XWe4HiArOAtz{5)QCuL6R;U}?is!rZ zPFYtPo$s90cqvsX7@#qQ>v1a=As5XNe^UGUmQchY1=anuyvvQqMs@PTIX`B7Not;7 zWgdF_)PczD{^c|*Xm^Oa!%16IY9a=n<=`=0{Q!x~jC8UD$ZW1{znl7x6MB2;&#N?j z9wJ-f7-Gv04d?&i5me)&;Db*q6YZO@i=e8(AglzfTPG0vqYC&!g-@oaLq z)ageF#sj6zJoFEtJyeBht(}O%<4SXRGV36-jqeB(a!MF+&*wVe3Pa|~gjD@4ieAZ6 z>eL6n-q76Rpn6pM>2s!L@N@;)nrXsn)PRjzNl{IU$O@tfYmF;#^VrYENmFHX7;RX0 z5jww09Wx)HtJ95{+NiWr7Xi*EivA0ZOJJMX;5TcF$8HtZ58X2xE)-q4`rcT*rO$94$y| z+ilL4LVb>Pxj;(XpxO@CUki)S?aH{ct!Y;<3#Py@g>aXZQ0Hw)P4oW{^SJV5bvX4| z^enNpx_Uube@v;iPUBJ83j=3)I&&%^F_^%-{G-N*3TzKW==RM9jh_3);C4+%^-U*J zh3(2Wk>cPiFW08hLw6Tel&6qyrgdS{^lrprTduYHyn7JB1t%ifA3eFTE{0V>oNp1x zw~-;3S9)Twv&>y;oA`LNimF?}mlKBOY>KH7ZSk^IJ?^a1zt*tJb({t1K)b$v>q#L4 zTKjQ$hY_F<+GV`)cl#`kS7&exOi{)$tQJ&%jxj4otzhE=72_jRrr1Y? zp`X%4Ci&aj9ba7WQ)+UEeSrO#I9=rsE2YV~3o}3^h?pT&#wIy&@@{TT%B+ zT}6`utYh_sOT*dSix{YRQe?NHk6ct-p1Mthp899lRvqaXi_L;=?XIL@4X#QRI7aDZ z8GjZfGzp^KBsNT)9bx1O*+M=i9O9M>D!scq_FlQ<4qU$A#XBzOi(3AP-p}UKmMIt% zTr3B@!8H?xfjbu|*7tdG`if!*zPhPQG{aKi3fI)M_e{C(SW6l3j7-oVZXA3&`dXjW zY%_E}=P}`Kbr&mqXK8y1U(uJoaI0LF{p?=cskt$D8*Z92gb`jb=5PG#!%H9>(>{)! zJc5xd!f&l?&nJf>%dZi|l(7_F`8n5>bno;{_Xd3*yfw#h*4;^Z3Brv9!XLL0KjRLs zVy8nN=h@+qWET=e7dvVg^(vmoIU7>O`t=uvsvo$~v^??(p30lpBl(r^v*n~=i|k~l43tfd<-n)k*=RXbB1uC`Pr!+&I0{1cgU|%a zK#-JZmwPi_{)B}=0HcQoK@)o{uuE-HsbQE=ZC|K}nhqP;*sdft zHorFzyQPwDWZFvRb$*(hZb$#_9$DU*lLz@KQ$jnH zniac2TKSq03a5P|V)s~8J|Tf#KJjp1Dcw{4Q%FT=kjg(@wKL9Le(EyvFIL!gPq&o} zXO4aJzA%ukUp^YR(W~`!N8sbC8o~j6-fI=mzi=VMTK}|o{{H}XK-0Aa(r5a6NJjq zjxNC^>QB91w{okr4e9ECKaB%F)M4VdFCW*wL@j^X^gVk_nfI`>q%nk1R~6jYP+=RN z*%FQ>)Vb~}jz!Hnc`sx+&0tf^elwCj|L<;NCa6G)6l}cbxW~ki)T}P-*B-Zf1A6u@ zG+Kjm`{Du@CM*vTHGg{!AQOZT{$hQftDzgqOl;?`=^zIyqyQ2@Z*-k+Y7w5BLR9hw z{x&m{x||M3A`z!noA~4Q>g)Q6EZ34^i&Gk_R3Q4@u&tAjLcu0``$DjSJ!?*{##Fl5 z2)%*BF~&0?FsV!Ov>LvZT~gIL;7^zceTiIc?&o1|H9&-3oM{jLa??V^zVB$XRDwoo zjIO6rAZM;+OR+y6%SQb6%Mc9)af9TEiwH*+j0jH3F|ni_U%vQJ-XLQeqbMx-${~qN z!^UlLbe9V1Xp@OLkv*yaIf~H_C5a5k$RHq{uWF4Q$V9RtXs8cbXe0>Y;P@0X86+I8 zc|bUA(UYBBRtrTsaVPG7PX4T@bUgjaDG7$ysMw_^Ue{LtR2hY&hOaBN>P@fRv z-nvobt~C4RGQ-ZM8;}3wK%0d+j5F&^`p{-OZVI>PVQgd=`B|f-DD&Q?&EAc3&|*#A z3L-ka=W-hk!87UU6iEM zM&tiX-q_+qGG+^aYq&x`r+sg=An*mZ)SsR2Th2WEO;Hg^yJ#8+Ey9#;ebj(rIBa`;>5om48)=haC-Rk~(!$a+1Gwq%UXF1wn;!VhLf`QUDp~aSn_+w! zVACtN9t7jN8#J^ra7oV`8u-iyZFhq-+m@@6n(aUFS(Z#_rQ{DVgmHXNN$wQe&fl7sHbye53O z%ja6o08ZZK2}m?_Zkr_}S~QpF*ilVJn>|U?qtY>If$yO*6!U(C5g50dM&rUbcwP92 z^pV+6l4<{~7+vzAxI;Ti?WOrUDrC9(j8Tsjp7M$yacnEY?0fxq&W(vAhpx!LTEf5Y{i&nzQOI&k z8EOymV>@Qckg;gCn}I~3#zO6snBEg3(hPG{lGeHU!=Y=9QeTlAGi`hH*rr-?_s1!^ z+Mj=oPb33~hr6C9N`Id^~fuGl%4_b(X>TD~<#cX@>;k-k2v%Vwj6!&~$ube!!oXOxX zhrxEG;4gSD-$L@g4Az!`bWdtp@Z3?KB`xyy-O&C?u;auMe-VSmqRD-Ej*O+_fV~ZJ zxlac#w@f$JJkW}WO`E0J_J;SMlQ|(afo01Y1({$;N$thD|}|- zIJ}>imOkT$u^T_Lihf8q&|HAFt6ASj#g4oJ#(_P(KtA+757cs5HE*#MHBF(M!eS`i z8=Z3kZ74s2x1lkEFmXu5ov9c5xbUq&&=9Nei6&QuX6&E-BeUK2nR|MoFR{^-cv*$; zmgTgGc_6U=8^ga?;ahG7m#F^ExC#u9^&Yw7mF7)hb{Cl8Y=}0D7`{P zQYrN2Lx;=K8RZelvi%>{O`LvB?JRQZs|N~lY6yEbviih)zZ^rAK8IRGa&M}^JA`jF z^4{|38vCg%Zd`O^t~wN&&-S4&D(0J}Ko3d-j*wWxyV@D7}y%?xhcf6;!F`!UV85DQi^6zE*g27_LBadt`1QWGO zih%^6U(ip)-C2@Lx7pvYCU!BR+Pt`H-ncmgH(HZ{;3nOqP7{KO~$HGmWIjN^Qw9vi&{ zpf=RU^g@%cvD=^-ZPcbEQt>hF5a67LKRL2~D75biWnbEI>8hR6LZVMTsxUMxrw%V` zJnEZ|6}Oy+HHRc2U)I7_3LdZ%`n9=HZyivJ^hr8F)ruY<`-GQ(&m2{JIZt8aaDja+ zh*EG0(s)a23D?EHdH0m!soZ66@xEIHgS4y+gy@6XW`qhFbPdDS06Dv z_THUtjllkb9@CD_691Xj(8GIDHpYXNb&oYpU0U9`YbkZl0cuc=_Hn0iqYJMNffY4f9KJN7ELG9^UHL@6PKnfs zT#J0jPD0GXWiVH`#Op^<+S6Sb?R|4eZb0_h`tJEkKQr@HycrNpXrhrdBBWG$8h^|> zmnxB_^4l9<&jTYJt&cIJS4AES+(!CuDhK-Eh;HAtfGYKcJKyW(xl#%Xbm&#Rys}sq zYi~{i+m*E_St@FH4y9A_;<@nE zm-*MaOgPq~Ue%tE?aNWeMEOIa)E82g1DZON3U_kr9z#lv z7V!5%sN%DNN)|`%f8QMghgw;bnzIyCyS5-v>_#b=BTP?PvZLG`T*z@I{YMNHh*OQc z>Fv~s#YrnspC93noEBI2;oLL)L%9Ku>s#?>+{jF% zi$|A*@MVO($uDHa{9%QefT2B#~un zv@PF^ZMw|&$KKGTA{F|S4`D3iz@FiZV|`f-3mZ19rsPJjIl2rNG+D`d;CI3{vEqbk z#gh}fHqWy(5$_<>+*)AM{H7!s_w3Z1kXW}#Y#J|mDI{yp^iab0(-jvF*$Oic+=L)z zDo%H!WY>BFBU3qs;V_c+@HRYISx^|)ywd@}E?JilAr_x_eI94!DR!`e>NR99oYsyK z>8#>D2~PdjpXh+T@o|*XN(6^xxz*JZO675lk?FU?ICqgVSZG>!HmHSihS(OxMbd2D zo9KyjgTmm&?n9;2e8gQYQIL5l)@5n4xi2qiO#88`npEIy)g1n%P=xBoVj4W9@z&P; zW{Zg=Q>*Kb{?EB}p6x|;PLJxeZldY=eiOdf>$`=fE5a&`Dm936u*GLk273 z4%nHjdV!IVeoqjB*|b3ZnUnf&r8u^SHxkU-v9ikA8Kq)(!!#)5ex@PhSe)_r$;t}XU81SXUc*V^sy|eX`#S7TQGx4d z%xW!Ev5Wq14lJ^fn1QI{)1==ew0S0+_vj{H$Ocka@m-2XCskn%wwJ_V6y>3WZ>Zef zO>+zVVjdpyAxz!Is1~yLhfxcU7&mQ=fBooRmufs2MM5-cWJ$)2xeriO41RMuSRPCV zPfyc=uRQS+>vuzlt8t)?-r>Co4WOpuMP>R3-t--ahnF|2Ze`)*aNHW{nL9j7tw1_* z2mUZ2v4kq5&Hqqm=RrNNY_yUn{Gud|1(5KJ9u*ga%7yv(=EeAcO8B6=cslbI1D2mR z?WZF&swGPo9c1C-?nP2UYm)qx><3-+zjYs5MBl|J_5TLGKyoZb8^tnz4lbSHt+HsZ-T z_q_uHl$psZRkNmJM*k=;zgx61P(jlYHwKRLGOyp*SDeI@v@7wG;!xzog&719*Hf%b z5{{Yi{Ty&H##QfOOp76)N~SH41Y@d)!GNXM9(YONc`PYhrkVkYj(nfq zf*J;Ol`#+QwjjBDVk>VT16y22Q?gE{7t(rWl z!?)_FLC-EnL?7E#%IiT-uyVEfYB%co!F{z*j4U_XoA71 zDXJxOvQgcRu#k{T2i)85^lji9{W#N7NT??+-;t#K6aO*U&6vz=)NkA)A6u#1A={h3 zFf4ri(=f7c;dN|rO5r4u>CX$#6=$@LfQaEqu@0HR4vzm?XM_EV?&_SM!nPRJr;h38 zn|lrI3(w8{`j>!+4xz?IHU?SnOir?X$|pjCbU{PQ-fOrh+6K zM+XdkyRuz;HY3R|S@I>zL-KRY91&Bvz4!`FjfNto{9JzEmaGw`ma2z079J5KPbZm- zJ{B?2W228H=V;8%N;D1xQe)e6@>Q0hW^=mN()(E5|eyNMFdw!?MbQ?*`j3}l7%ofQHZg}SmZ@hB^p{E_dA~? z(z8!bo!(;$kB~B3yTbZa>#L=sbL-s*p>-h1S!fpZ#j0u=UxzPq^V0TaR~crLKG*kd zu?DGS)Rul)0bXgTYgr!z7QbZgHa0>l>gCRFcRM{Mgam;|@F1-bc8H0p;pX3TV=)GE z!%rjTIY}_%2V>~3nbcBYA7zbtOXNpgqt`Vd(WPiHn)%&gEJmk=bG`k6Ch;2vJje*O zDo>$nK#nXV36OgLEUZB5(H!C6={3^K2SwtUoQBUf!)x`?&&T2nN_W&!QYnC~ZV~F` zH)8x(gUUX{kR~mppIn-%>CmIuG!v~xYKYU8orOP*1&0+&y|ngPTCqH2>Dp&@zp(lJ z)5`8+fD5i;`LPyhDl+C)$0s#zE!a{<(UhAvR0kBfh!(5J%3B2v7z>ndrfzqy3m><8EYVqW!kW2kAZ06n6c(#U( zPL{UCVYQ%~PzS3I^WPvE0W%}ybmP>BS;6ER zsv}2g2+x~7&?$%$l(paojU<7~k5K82<7y5qfld4)ow6ruBC*g|aOJ&mTx%5*kBjWu z_Dfs5*v|*OA^`FEE3k1BM2cVUMYJzO?)JwNO7vyckG1#k;hMrck`U{1=zUg%*^9Q( z!*2LLG(I@9(Y|bsl$;k)vJ?5tC@~6K#YlhR*IDJ|^9ntJg^rpfGzuoT^cxy)kELzl zl2)5rpMgVYcSdtOA0k}R(RF6CdPP?+*W{tU==oqI_KAYmzPOA|UHTyL92Ey`7%W25 zaWhW|VY#pEbEn^_txa>Hbl2P)cm2J5CUJfXh4I)=UD!A|WEvp~Fw2m88<(e);d#(o zBM>$OtD)>Zo1E+xG?A6rXnweJ)>V>Bo#8FckIowH-jDaij*W)tkX~VO40J)DKn*fH z@AzN&`Ei=885xk?PsadXQ5N)tPU;j@a!TA*B{xyiR9ykJfCKNHoEeRx06$26bA&n7 z1K8z!Q2P6C{tf0PwN~Y>!6mUOIYMsjx-dYkS12f5C~aSJLSp*WIaG#ktt0y3%~{*} zjN`vLZGOVbR$VQ)wHXKdJ+(Afg8@RJH93VZTZjlnTNObhp!BjD8*nnwwVN!^TR#Ql zWPOpE4MWuo=XtlYFaK}RDJqPD+?b4E!omEa2EV%01wCZZI9 zzDj?JPc_v#eLiR-F*A}OkInkXNpC``gwL@LDma$dwmvwokO}8b!SVwg;rCNV;vCm# zGdB!B3D%Cuq)a7Ivbjl_i`z{P!DAlqS!Zm%roD#NuwL7DA#j%4oF zJFIjf^XaIUU%~;oCR2&hUk~_~md*2$Snbs<hzJY6>pH{DN~&PAS~WmIP|G?}ToJs`#9xpxrtfz7AC) z=#}-U)7BO~@o|3v)x*wPAqP^sSzGzFBivP-zl=$iE>?zA5Ssw8xoReyz9jiE0a_n4 zA}iinsZfHo{OBXWe+MibyU7%cD3@l3EMGSv;%}(|s^9Rb+L=~9V^+HG{E2VV8 z#}pE3dVVXKJJ0{uhT`ffh0pYKIi@iaO2J4$SYQ|R+&`N-TR3ms!&y^=EZII=Bw- ziAj(hOw4eYjSx}77CL9Gz!y&Fww0N-9@cqPRfCd2v8b0pWl{3j9A}?W{$O|F2mb#s zA%;y!OGcXFvlLmg(EuMnj5Pp;R9If!cRX#~d-=Rs_Hg#j(XMm4WRcgi4x3w&80%4Z z7Hh!O1zywMQw_+PpRDkS-?z!p8T6am5I z`{X4}V^GI;DZ`|+iv#n>)3Ar^!q4&BDra5|RD>zzI#&n^vRhniz;-h+2GPkjKT94h zSN!qzDh7hg;q{4)cRUU=S$=qV5`@O6VUot)ugsr@OTO`4Ydb^ zJl|G%XZF-S#h&iD)s;d$ZKZHfHiR$?BxY)b5q6b$5%}6iFh7 z7Vahehp~g9#-ER{X_HE+wLFBbrZTKMa1a+=0D~wemXy`7I_r6x{3X0KB zw%MJKUsr>P5+)JGB5FWZw#@o^9!Sn$#&p7seH4%-wQ)@NlOHCRu<@;)}9 zD>)n=>SKgNo$3N;P|jvIqhz`6WF+k~n2wTnVtand{OVKH({X9y0T5QE_de6BMB140 z=%Q3E)W2Hsqj;D<;$=yA!6mWi?T=)K?nPH~zrtsz8XWW8UMQoQTAJxz)E=C&yI;t+bX%Id~;Puke)3np(!;^s~vTbN*cpUIh=S) zJy3$cMg%PZl_+2k#vr63Kgc64+gpOAm}fCbdUDqTE$kq~n{RNIlq3IU*zRCKC3)yf zq*dfN6D=^`P|hq4XqQhYCbK-9;0O^(A5#q;CywkO<8llD^fI%8fT6CL^E>b}m>i6@ zVf$*9+n^4h53<`?=6F4UGs}X+%f{FJc*rKFXzX;6taEs2I<+MW1?T4zghk%RWw2)t z|39{kMG4o!+Ozi97XI{&n`cH0LskpM{uzx5+owf9kZ+{UX0U_4m$^<*PG0|g7fbBT zyN)9z@$VZ8HhxM27J1#X{w$65RRS9mPUQvXWuZQ4FX{hljs#3FO=goSW#KKQlb0rm z+G3Lm+bIhW?``ARxlaLlA1)#XWt+ibTgW9b1#Qylgbx zMaC@Z$(v2S0pW?)n;K@%e9_jDToLiE<|}*aYidk#Kf>90pJ{R?Lz?Al3c^pm=EaqC z(XT@Q-N%J^v{OnCj3I_(b0A+q<_u9&+Nh(`$Ajgyp}mC{MngK%+4VB6 zk-F({(~4^Jx+Kl(4s*3`X|%E+$pT6Obb>j+8xs2rsHGS0+!c$*we{<}E40;Xm5Z}Y zii98CS|12;^P0D;ip8fHy#!_rkNi^ixIvrY1 zYIAK>q9@c*M{a^NOdr^bWM;Y9IX$kG7ivCMt*M^oj>f+Ub$So-gLv}`OEW$&XRuT1 z@erS%E{z%$9iNd#uvrCf*C~Jbezs) zrT&rD{W_7;>L!PT{APH9x1(a7*=3_pCne1(c`MLQn}6o$2E>}-Y=Cw!x0^6-urL^+ z2z-NgVCOyv-$qfPn>dgK4s=>>NJg=aY%=5B*nJY7Bsj98$`Ues-wsijf%eT#UcL1n z7RV=PBfC9^)s*Fjx|u8zQB^f3>{b9+X6`{nv72aX4=?rVt#(iLEB`KgGVil2-e3MK z;mHRdRA=JUB=vjnN5l2_gTd&;T(xX#ig4tt&)Db_61jZCtzCIP?x;?YM>!{kLnX(v zAg>*MA8!6^afkVCtRsU+gB1FixONVlf%OW^tnD6ZBE+4D1006!Aqz6{rq(iK*oa=l zqBdK-^nO7{b$JIW-KEa5`~?{~L5W`(=iz+jyI}slsio*=a9WlKP;Ffx?8mHRR$Y?e z$Zg$K*=V_D$%m=W$6y4ppU;cd!hD|k8X2OmjvNJOj? zZu45ALjxhBb0p4J>-U6Za~Qve3X@>ZtQuCTltI)5MZeq_#z`nbIf6@X!GVfb^RkSN ztRJtrYJ1>W)H1)a!3A!?n9e}0RPv&e?lM>&3PBV+7*{2Yf+ZV@IOZ?~q`|WGyN~md z^+>*BOn5b|g;-$Gco2ykSk8U)H2=wvYJ!!f6Hh-dECAUu#=DxpJfI zu@Zq6(>M31u>O*w#pxKM?y@AW`xgR`SE?*+>@xYh1hbI7Ks(YHoy`(`cpZuqTkFkk z;}E2aa!8rGHP}YySC^q2g!K@f_0X-}P=f1Gr@acMlTglu#K99;G{t=AmeXfc7Pb7& z`fIeUCwk;X@zu!)tk#VKp_ve@)e^3R1G>msMVXwh3_;DHP*jT~#z^M-wf>zlDl2N< z^Dn2uZXS5O1<{nA%fy5LS3s!0LEoR<1`&~e zZ8lcpPAu4@LfQWRJe~h(&_1aTrFr$Y_RO=2!WBk!+vBb;9Sa6+@xA08{?bL(#mjv4 z;;gbW?g(U)`37#H1fnE_CY!F3xG=hx<@`#<7%PGNvzeYlG>6y9c9JWG85J)qq_kQ~ zSSuWemDr;8M~l~5D9i`jE&qm{^a}1ER1Nb6!kO#Pgaub4S|f{R6sWe} z8^x-z=syKQ#qJ&0Gio>hV67cqJ<*Oy4s}CVome1_{K1kTtM=g4&_om8%omk-Nd|&= zJ?Y+HAsuwlYsTsxU>Qqs}jE<~_VXc3nat2o9Uo#caJ3^(#bJ+v9w zP5CzkxuH+BoZWQ0fsM7Wwy42alg(y(H^qT!yfld4>-lNw%1a>S)5dv^zOmj=aKFJO zJAxLcaidg$o_hZe*$jvFy8maN@iuHIt<#RaCm~&xoWM_K55>sW5Bb-J`;}!YXC#y@SFcYm;wf(pb(Yrd_yf8-cfeaLW@J+UsK* z=5FVw?zA`ZIt~L|n8b&O)w#3#g!L(XQiegzlD-?f{E1JcP3WB1%xQnh7>5Ws(pGGWp1Va)ruoLUQQtYCfy!J!V@)gOb0}`_{i-xL0t9!@^W24 zTXNY9Ec$|QWF^(4rg5}WB6Da<1TS1g0UeRCqpn2B6fN_4ku#vY7tV9o2c$g@A=#ZX zw2aD7A#Q@YTL$lFMWu{ckF!vQH3EhBUeE$V@=~`;0OyaXsPTSP&V-YIrl-R0OkIwG#ICs%bDpF44@%2ET zP`3>tOTg~kGB0I1wM$a_B2;;&@H+ciD5N(=I0jVx^xlWCiI! zUAq_d5F^}=71aTOFPp2wV_v<+u;Y#R9a$7t17PFfOYI&cV3B#>W}%=^>ZN^GLcJLQ z%GC-Y%%yxMdO;GcuJq@$58}gf!BjxHyp0~Mu^x8YbdUhT0YGg%y|#Qf806fX39cX^K90E?NgN#gV2 zNx!Zcj#7`=ag@**iXmJbVTv~uifF#!Q7NO3*e%uelSj5yo`?^ulA3y5Luy{En-Zd- ztb3;`*xd^f9_-yL_y@SOsxt|}u2gDh#zPu(atcB<{TcHP%c3hz8{E{bc{~PsYX6T^ zJ=EL#Z&q_Ob@Ql(VI5mI;{H?c*A8yri^n<(2WP_{(6+H2A^a5t%S85>uA041rw;3TPlii8s=Z0qv{mmc=GGKgeiTZQK3IXP{HiHFJI>*+ zez0k3)?*mr1XoCXpnAD$jv^6e6zfphXKz%n_E{M}*e2c`?=LXi{&;Dol1Eu`aKxGW zR_uR+n+rgHRp=ZlN0x(-MS(*dMEML-;jjAuKj`Eg@W;&?%rNw{h+je@J*C+Wl^a#|K#XkJ(Q{C!V=;C+B-Nl8b)Ys&@ zB6pN3Vf*kJwaFev>FLf!b`rr2Dlf4k2EYD@IyQ{(o$ktdWVZNx;Cm*U4Jplg zYX@YMASKZ9|2=Ujjx^3L}ab*+KUxfdE_X>k8`qQ_LbfE-W`T?3j+ zg>?-21+q~+peD?+g+dZ4=IdmSEYrj<(z3uk5-i7#wf@6P#5+f%MwM-2O8rcXNcbm{P z2LJz;m0qoqYuCc zTQoH3I+E57C_2GJfmpc3Wk=9{SKEe%+k7TaR=uC=vvBZw&tTt}KI?b+w=F+*q- zP?6~=^!;cz#H=o%x~OC*vi1yVd31R-V`r!o)SQNwzokl{wm$_<^2xarwQ;|}Ohc1| zLK9!LOW7J;8|6iWf4w7Gr$NyCD|WZsf1UCEbaIKzUSHCa3n!&PZtAdjgV3UJXSzF` z&aniU%$I40imi%BwmbNI6z|ORbMk$f^PsDkyN@WA_O!T|P$MF*Jh&Ary1Jma%+@JH z3r6C&t$ix#eH?n@sO&TE#unfyG8 z2^`YlGOKOf4^n5BuA^pwSye}JnlvZj;1q7$V_i1yN!18y?`geQq@-kLubqwz?t