Merge branch 'ClickHouse:master' into master

.github/workflows/master.yml

@@ -149,7 +149,6 @@ jobs:
           sudo rm -fr "$TEMP_PATH"
   SplitBuildSmokeTest:
     needs: [BuilderDebSplitted]
-    if: ${{ !contains(github.event.pull_request.labels.*.name, 'pr-documentation') && !contains(github.event.pull_request.labels.*.name, 'pr-doc-fix') }}
     runs-on: [self-hosted, style-checker]
     steps:
       - name: Set envs
@@ -316,7 +315,6 @@ jobs:
           sudo rm -fr "$TEMP_PATH" "$CACHES_PATH"
   BuilderBinRelease:
     needs: [DockerHubPush]
-    if: ${{ !contains(github.event.pull_request.labels.*.name, 'pr-documentation') && !contains(github.event.pull_request.labels.*.name, 'pr-doc-fix') }}
     runs-on: [self-hosted, builder]
     steps:
       - name: Set envs
@@ -362,7 +360,6 @@ jobs:
           sudo rm -fr "$TEMP_PATH" "$CACHES_PATH"
   BuilderBinGCC:
     needs: [DockerHubPush]
-    if: ${{ !contains(github.event.pull_request.labels.*.name, 'pr-documentation') && !contains(github.event.pull_request.labels.*.name, 'pr-doc-fix') }}
     runs-on: [self-hosted, builder]
     steps:
       - name: Set envs
@@ -636,7 +633,6 @@ jobs:
 ##########################################################################################
   BuilderDebSplitted:
     needs: [DockerHubPush]
-    if: ${{ !contains(github.event.pull_request.labels.*.name, 'pr-documentation') && !contains(github.event.pull_request.labels.*.name, 'pr-doc-fix') }}
     runs-on: [self-hosted, builder]
     steps:
       - name: Set envs
@@ -682,7 +678,6 @@ jobs:
           sudo rm -fr "$TEMP_PATH" "$CACHES_PATH"
   BuilderBinTidy:
     needs: [DockerHubPush]
-    if: ${{ !contains(github.event.pull_request.labels.*.name, 'pr-documentation') && !contains(github.event.pull_request.labels.*.name, 'pr-doc-fix') }}
     runs-on: [self-hosted, builder]
     steps:
       - name: Set envs
@@ -728,7 +723,6 @@ jobs:
           sudo rm -fr "$TEMP_PATH" "$CACHES_PATH"
   BuilderBinDarwin:
     needs: [DockerHubPush]
-    if: ${{ !contains(github.event.pull_request.labels.*.name, 'pr-documentation') && !contains(github.event.pull_request.labels.*.name, 'pr-doc-fix') }}
     runs-on: [self-hosted, builder]
     steps:
       - name: Set envs
@@ -774,7 +768,6 @@ jobs:
           sudo rm -fr "$TEMP_PATH" "$CACHES_PATH"
   BuilderBinAarch64:
     needs: [DockerHubPush]
-    if: ${{ !contains(github.event.pull_request.labels.*.name, 'pr-documentation') && !contains(github.event.pull_request.labels.*.name, 'pr-doc-fix') }}
     runs-on: [self-hosted, builder]
     steps:
       - name: Set envs
@@ -820,7 +813,6 @@ jobs:
           sudo rm -fr "$TEMP_PATH" "$CACHES_PATH"
   BuilderBinFreeBSD:
     needs: [DockerHubPush]
-    if: ${{ !contains(github.event.pull_request.labels.*.name, 'pr-documentation') && !contains(github.event.pull_request.labels.*.name, 'pr-doc-fix') }}
     runs-on: [self-hosted, builder]
     steps:
       - name: Set envs
@@ -866,7 +858,6 @@ jobs:
           sudo rm -fr "$TEMP_PATH" "$CACHES_PATH"
   BuilderBinDarwinAarch64:
     needs: [DockerHubPush]
-    if: ${{ !contains(github.event.pull_request.labels.*.name, 'pr-documentation') && !contains(github.event.pull_request.labels.*.name, 'pr-doc-fix') }}
     runs-on: [self-hosted, builder]
     steps:
       - name: Set envs
@@ -912,7 +903,6 @@ jobs:
           sudo rm -fr "$TEMP_PATH" "$CACHES_PATH"
   BuilderBinPPC64:
     needs: [DockerHubPush]
-    if: ${{ !contains(github.event.pull_request.labels.*.name, 'pr-documentation') && !contains(github.event.pull_request.labels.*.name, 'pr-doc-fix') }}
     runs-on: [self-hosted, builder]
     steps:
       - name: Set envs

src/Common/Exception.h

@@ -35,10 +35,10 @@ public:
     {}
 
     // Format message with fmt::format, like the logging functions.
-    template <typename ...Args>
-    Exception(int code, const std::string & fmt, Args&&... args)
-        : Exception(fmt::format(fmt::runtime(fmt), std::forward<Args>(args)...), code)
-    {}
+    template <typename... Args>
+    Exception(int code, fmt::format_string<Args...> fmt, Args &&... args) : Exception(fmt::format(fmt, std::forward<Args>(args)...), code)
+    {
+    }
 
     struct CreateFromPocoTag {};
     struct CreateFromSTDTag {};
@@ -52,10 +52,10 @@ public:
     const char * what() const throw() override { return message().data(); }
 
     /// Add something to the existing message.
-    template <typename ...Args>
-    void addMessage(const std::string& format, Args&&... args)
+    template <typename... Args>
+    void addMessage(fmt::format_string<Args...> format, Args &&... args)
     {
-        extendedMessage(fmt::format(fmt::runtime(format), std::forward<Args>(args)...));
+        extendedMessage(fmt::format(format, std::forward<Args>(args)...));
     }
 
     void addMessage(const std::string& message)
@@ -117,10 +117,10 @@ public:
     ParsingException(int code, const std::string & message);
 
     // Format message with fmt::format, like the logging functions.
-    template <typename ...Args>
-    ParsingException(int code, const std::string & fmt, Args&&... args)
-        : Exception(fmt::format(fmt::runtime(fmt), std::forward<Args>(args)...), code)
-    {}
+    template <typename... Args>
+    ParsingException(int code, fmt::format_string<Args...> fmt, Args &&... args) : Exception(code, fmt, std::forward<Args>(args)...)
+    {
+    }
 
 
     std::string displayText() const

src/Common/RangeGenerator.h

@@ -0,0 +1,46 @@
+#pragma once
+
+#include <optional>
+#include <cmath>
+
+namespace DB
+{
+
+class RangeGenerator
+{
+public:
+    explicit RangeGenerator(size_t total_size_, size_t range_step_, size_t range_start = 0)
+        : from(range_start), range_step(range_step_), total_size(total_size_)
+    {
+    }
+
+    size_t totalRanges() const { return static_cast<size_t>(round(static_cast<float>(total_size - from) / range_step)); }
+
+    using Range = std::pair<size_t, size_t>;
+
+    // return upper exclusive range of values, i.e. [from_range, to_range>
+    std::optional<Range> nextRange()
+    {
+        if (from >= total_size)
+        {
+            return std::nullopt;
+        }
+
+        auto to = from + range_step;
+        if (to >= total_size)
+        {
+            to = total_size;
+        }
+
+        Range range{from, to};
+        from = to;
+        return range;
+    }
+
+private:
+    size_t from;
+    size_t range_step;
+    size_t total_size;
+};
+
+}

src/Common/format.h

@@ -0,0 +1,178 @@
+#pragma once
+
+#include <base/types.h>
+#include <Common/Exception.h>
+#include <Common/PODArray.h>
+#include <Common/StringUtils/StringUtils.h>
+
+namespace DB
+{
+
+namespace ErrorCodes
+{
+    extern const int BAD_ARGUMENTS;
+}
+
+namespace Format
+{
+    using IndexPositions = PODArrayWithStackMemory<UInt64, 64>;
+
+    static inline void parseNumber(const String & description, UInt64 l, UInt64 r, UInt64 & res, UInt64 argument_number)
+    {
+        res = 0;
+        for (UInt64 pos = l; pos < r; ++pos)
+        {
+            if (!isNumericASCII(description[pos]))
+                throw Exception("Not a number in curly braces at position " + std::to_string(pos), ErrorCodes::BAD_ARGUMENTS);
+            res = res * 10 + description[pos] - '0';
+            if (res >= argument_number)
+                throw Exception(
+                    "Too big number for arguments, must be at most " + std::to_string(argument_number - 1), ErrorCodes::BAD_ARGUMENTS);
+        }
+    }
+
+    static inline void init(
+        const String & pattern,
+        size_t argument_number,
+        const std::vector<std::optional<String>> & constant_strings,
+        IndexPositions & index_positions,
+        std::vector<String> & substrings)
+    {
+        /// Is current position after open curly brace.
+        bool is_open_curly = false;
+        /// The position of last open token.
+        size_t last_open = -1;
+
+        /// Is formatting in a plain {} token.
+        std::optional<bool> is_plain_numbering;
+        UInt64 index_if_plain = 0;
+
+        /// Left position of adding substrings, just to the closed brace position or the start of the string.
+        /// Invariant --- the start of substring is in this position.
+        size_t start_pos = 0;
+
+        /// A flag to decide whether we should glue the constant strings.
+        bool glue_to_next = false;
+
+        /// Handling double braces (escaping).
+        auto double_brace_removal = [](String & str)
+        {
+            size_t i = 0;
+            bool should_delete = true;
+            str.erase(
+                std::remove_if(
+                    str.begin(),
+                    str.end(),
+                    [&i, &should_delete, &str](char)
+                    {
+                        bool is_double_brace = (str[i] == '{' && str[i + 1] == '{') || (str[i] == '}' && str[i + 1] == '}');
+                        ++i;
+                        if (is_double_brace && should_delete)
+                        {
+                            should_delete = false;
+                            return true;
+                        }
+                        should_delete = true;
+                        return false;
+                    }),
+                str.end());
+        };
+
+        index_positions.emplace_back();
+
+        for (size_t i = 0; i < pattern.size(); ++i)
+        {
+            if (pattern[i] == '{')
+            {
+                /// Escaping handling
+                /// It is safe to access because of null termination
+                if (pattern[i + 1] == '{')
+                {
+                    ++i;
+                    continue;
+                }
+
+                if (is_open_curly)
+                    throw Exception("Two open curly braces without close one at position " + std::to_string(i), ErrorCodes::BAD_ARGUMENTS);
+
+                String to_add = String(pattern.data() + start_pos, i - start_pos);
+                double_brace_removal(to_add);
+                if (!glue_to_next)
+                    substrings.emplace_back(to_add);
+                else
+                    substrings.back() += to_add;
+
+                glue_to_next = false;
+
+                is_open_curly = true;
+                last_open = i + 1;
+            }
+            else if (pattern[i] == '}')
+            {
+                if (pattern[i + 1] == '}')
+                {
+                    ++i;
+                    continue;
+                }
+
+                if (!is_open_curly)
+                    throw Exception("Closed curly brace without open one at position " + std::to_string(i), ErrorCodes::BAD_ARGUMENTS);
+
+                is_open_curly = false;
+
+                if (last_open == i)
+                {
+                    if (is_plain_numbering && !*is_plain_numbering)
+                        throw Exception(
+                            "Cannot switch from automatic field numbering to manual field specification", ErrorCodes::BAD_ARGUMENTS);
+                    is_plain_numbering = true;
+                    if (index_if_plain >= argument_number)
+                        throw Exception("Argument is too big for formatting", ErrorCodes::BAD_ARGUMENTS);
+                    index_positions.back() = index_if_plain++;
+                }
+                else
+                {
+                    if (is_plain_numbering && *is_plain_numbering)
+                        throw Exception(
+                            "Cannot switch from automatic field numbering to manual field specification", ErrorCodes::BAD_ARGUMENTS);
+                    is_plain_numbering = false;
+
+                    UInt64 arg;
+                    parseNumber(pattern, last_open, i, arg, argument_number);
+
+                    if (arg >= argument_number)
+                        throw Exception(
+                            "Argument is too big for formatting. Note that indexing starts from zero", ErrorCodes::BAD_ARGUMENTS);
+
+                    index_positions.back() = arg;
+                }
+
+                if (!constant_strings.empty() && constant_strings[index_positions.back()])
+                {
+                    /// The next string should be glued to last `A {} C`.format('B') -> `A B C`.
+                    glue_to_next = true;
+                    substrings.back() += *constant_strings[index_positions.back()];
+                }
+                else
+                    index_positions.emplace_back(); /// Otherwise we commit arg number and proceed.
+
+                start_pos = i + 1;
+            }
+        }
+
+        if (is_open_curly)
+            throw Exception("Last open curly brace is not closed", ErrorCodes::BAD_ARGUMENTS);
+
+        String to_add = String(pattern.data() + start_pos, pattern.size() - start_pos);
+        double_brace_removal(to_add);
+
+        if (!glue_to_next)
+            substrings.emplace_back(to_add);
+        else
+            substrings.back() += to_add;
+
+        index_positions.pop_back();
+    }
+}
+
+}

src/Databases/DatabaseOnDisk.cpp

@@ -6,6 +6,7 @@
 #include <IO/WriteHelpers.h>
 #include <Interpreters/Context.h>
 #include <Interpreters/InterpreterCreateQuery.h>
+#include <Interpreters/ApplyWithSubqueryVisitor.h>
 #include <Parsers/ASTCreateQuery.h>
 #include <Parsers/ASTFunction.h>
 #include <Parsers/ParserCreateQuery.h>
@@ -55,6 +56,9 @@ std::pair<String, StoragePtr> createTableFromAST(
     ast_create_query.attach = true;
     ast_create_query.setDatabase(database_name);
 
+    if (ast_create_query.select && ast_create_query.isView())
+        ApplyWithSubqueryVisitor().visit(*ast_create_query.select);
+
     if (ast_create_query.as_table_function)
     {
         const auto & factory = TableFunctionFactory::instance();

src/Databases/DatabaseReplicatedWorker.cpp

@@ -179,8 +179,12 @@ String DatabaseReplicatedDDLWorker::tryEnqueueAndExecuteEntry(DDLLogEntry & entr
 
     if (!task->was_executed)
     {
-        throw Exception(ErrorCodes::LOGICAL_ERROR, "Entry {} was executed, but was not committed: code {}: {}",
-                        task->execution_status.code, task->execution_status.message);
+        throw Exception(
+            ErrorCodes::LOGICAL_ERROR,
+            "Entry {} was executed, but was not committed: code {}: {}",
+            task->entry_name,
+            task->execution_status.code,
+            task->execution_status.message);
     }
 
     try_node->setAlreadyRemoved();

src/Dictionaries/XDBCDictionarySource.cpp

@@ -50,7 +50,7 @@ namespace
         {
             if (!qualified_name.database.empty())
                 throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT,
-                    "Dictionary source of type {} specifies a schema but schema is not supported by {}-driver",
+                    "Dictionary source specifies a schema but schema is not supported by {}-driver",
                     bridge_.getName());
         }
 

src/Disks/IO/CachedReadBufferFromRemoteFS.cpp

@@ -392,8 +392,13 @@ void CachedReadBufferFromRemoteFS::predownload(FileSegmentPtr & file_segment)
                 if (bytes_to_predownload)
                     throw Exception(
                         ErrorCodes::LOGICAL_ERROR,
-                        "Failed to predownload remaining {} bytes. Current file segment: {}, current download offset: {}, expected: {}, eof: {}",
-                        file_segment->range().toString(), file_segment->getDownloadOffset(), file_offset_of_buffer_end, implementation_buffer->eof());
+                        "Failed to predownload remaining {} bytes. Current file segment: {}, current download offset: {}, expected: {}, "
+                        "eof: {}",
+                        bytes_to_predownload,
+                        file_segment->range().toString(),
+                        file_segment->getDownloadOffset(),
+                        file_offset_of_buffer_end,
+                        implementation_buffer->eof());
 
                 auto result = implementation_buffer->hasPendingData();
 

src/Disks/IO/ReadBufferFromRemoteFSGather.cpp

@@ -44,7 +44,7 @@ SeekableReadBufferPtr ReadBufferFromS3Gather::createImplementationBuffer(const S
     {
         return std::make_unique<ReadBufferFromS3>(
             client_ptr, bucket, fs::path(metadata.remote_fs_root_path) / path, max_single_read_retries,
-            settings, /* use_external_buffer */true, read_until_position, /* restricted_seek */true);
+            settings, /* use_external_buffer */true, /* offset */ 0, read_until_position, /* restricted_seek */true);
     };
 
     if (with_cache)

src/Formats/FormatSchemaInfo.cpp

@@ -85,9 +85,12 @@ FormatSchemaInfo::FormatSchemaInfo(const String & format_schema, const String &
     else if (path.has_parent_path() && !fs::weakly_canonical(default_schema_directory_path / path).string().starts_with(fs::weakly_canonical(default_schema_directory_path).string()))
     {
         if (is_server)
-            throw Exception(ErrorCodes::BAD_ARGUMENTS,
-                            "Path in the 'format_schema' setting shouldn't go outside the 'format_schema_path' directory: {} ({} not in {})",
-                            path.string());
+            throw Exception(
+                ErrorCodes::BAD_ARGUMENTS,
+                "Path in the 'format_schema' setting shouldn't go outside the 'format_schema_path' directory: {} ({} not in {})",
+                default_schema_directory(),
+                path.string(),
+                default_schema_directory());
         path = default_schema_directory_path / path;
         schema_path = path.filename();
         schema_directory = path.parent_path() / "";

src/Functions/FunctionsStringArray.h

@@ -259,7 +259,7 @@ public:
             throw Exception(
                 ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH,
                 "Function '{}' needs at least 2 arguments, at most 3 arguments; passed {}.",
-                arguments.size());
+                name, arguments.size());
 
         if (!isString(arguments[0]))
             throw Exception("Illegal type " + arguments[0]->getName() + " of first argument of function " + getName() + ". Must be String.",

src/Functions/IFunction.cpp

@@ -181,9 +181,12 @@ ColumnPtr IExecutableFunction::defaultImplementationForNulls(
         // Default implementation for nulls returns null result for null arguments,
         // so the result type must be nullable.
         if (!result_type->isNullable())
-            throw Exception(ErrorCodes::LOGICAL_ERROR,
-                            "Function {} with Null argument and default implementation for Nulls "
-                            "is expected to return Nullable result, got {}", result_type->getName());
+            throw Exception(
+                ErrorCodes::LOGICAL_ERROR,
+                "Function {} with Null argument and default implementation for Nulls "
+                "is expected to return Nullable result, got {}",
+                getName(),
+                result_type->getName());
 
         return result_type->createColumnConstWithDefaultValue(input_rows_count);
     }

src/Functions/castOrDefault.cpp

@@ -231,7 +231,7 @@ private:
             {
                 throw Exception(ErrorCodes::BAD_ARGUMENTS,
                     "Function {} decimal scale should have native UInt type. Actual {}",
-                    scale_argument.type->getName());
+                    getName(), scale_argument.type->getName());
             }
 
             scale = arguments[additional_argument_index].column->getUInt(0);

src/Functions/concat.cpp

@@ -52,23 +52,21 @@ public:
     {
         if (arguments.size() < 2)
             throw Exception(
-                "Number of arguments for function " + getName() + " doesn't match: passed " + toString(arguments.size())
-                    + ", should be at least 2.",
-                ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH);
-
-        if (arguments.size() > FormatImpl::argument_threshold)
-            throw Exception(
-                "Number of arguments for function " + getName() + " doesn't match: passed " + toString(arguments.size())
-                    + ", should be at most " + std::to_string(FormatImpl::argument_threshold),
-                ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH);
+                ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH,
+                "Number of arguments for function {} doesn't match: passed {}, should be at least 2",
+                getName(),
+                arguments.size());
 
         for (const auto arg_idx : collections::range(0, arguments.size()))
         {
             const auto * arg = arguments[arg_idx].get();
             if (!isStringOrFixedString(arg))
-                throw Exception{"Illegal type " + arg->getName() + " of argument " + std::to_string(arg_idx + 1) + " of function "
-                                    + getName(),
-                                ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT};
+                throw Exception(
+                    ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT,
+                    "Illegal type {} of argument {} of function {}",
+                    arg->getName(),
+                    arg_idx + 1,
+                    getName());
         }
 
         return std::make_shared<DataTypeString>();
@@ -125,7 +123,7 @@ private:
         std::vector<const ColumnString::Chars *> data(num_arguments);
         std::vector<const ColumnString::Offsets *> offsets(num_arguments);
         std::vector<size_t> fixed_string_sizes(num_arguments);
-        std::vector<String> constant_strings(num_arguments);
+        std::vector<std::optional<String>> constant_strings(num_arguments);
         bool has_column_string = false;
         bool has_column_fixed_string = false;
         for (size_t i = 0; i < num_arguments; ++i)

src/Functions/dateName.cpp

@@ -112,7 +112,7 @@ public:
             || (res = executeType<DataTypeDateTime64>(arguments, result_type))))
             throw Exception(
                 ErrorCodes::ILLEGAL_COLUMN,
-                "Illegal column {} of function {], must be Date or DateTime.",
+                "Illegal column {} of function {}, must be Date or DateTime.",
                 arguments[1].column->getName(),
                 getName());
 

src/Functions/formatString.cpp

@@ -45,25 +45,23 @@ public:
 
     DataTypePtr getReturnTypeImpl(const DataTypes & arguments) const override
     {
-        if (arguments.empty())
+        if (arguments.size() < 2)
             throw Exception(
-                "Number of arguments for function " + getName() + " doesn't match: passed " + toString(arguments.size())
-                    + ", should be at least 1",
-                ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH);
-
-        if (arguments.size() > FormatImpl::argument_threshold)
-            throw Exception(
-                "Number of arguments for function " + getName() + " doesn't match: passed " + toString(arguments.size())
-                    + ", should be at most " + std::to_string(FormatImpl::argument_threshold),
-                ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH);
+                ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH,
+                "Number of arguments for function {} doesn't match: passed {}, should be at least 2",
+                getName(),
+                arguments.size());
 
         for (const auto arg_idx : collections::range(0, arguments.size()))
         {
             const auto * arg = arguments[arg_idx].get();
             if (!isStringOrFixedString(arg))
                 throw Exception(
-                    "Illegal type " + arg->getName() + " of argument " + std::to_string(arg_idx + 1) + " of function " + getName(),
-                    ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT);
+                    ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT,
+                    "Illegal type {} of argument {} of function {}",
+                    arg->getName(),
+                    arg_idx + 1,
+                    getName());
         }
 
         return std::make_shared<DataTypeString>();
@@ -84,7 +82,7 @@ public:
         std::vector<const ColumnString::Chars *> data(arguments.size() - 1);
         std::vector<const ColumnString::Offsets *> offsets(arguments.size() - 1);
         std::vector<size_t> fixed_string_sizes(arguments.size() - 1);
-        std::vector<String> constant_strings(arguments.size() - 1);
+        std::vector<std::optional<String>> constant_strings(arguments.size() - 1);
 
         bool has_column_string = false;
         bool has_column_fixed_string = false;

src/Functions/formatString.h

@@ -4,8 +4,10 @@
 #include <base/types.h>
 #include <Common/Exception.h>
 #include <Common/StringUtils/StringUtils.h>
+#include <Common/format.h>
 #include <Common/memcpySmall.h>
 
+
 #include <algorithm>
 #include <optional>
 #include <string>
@@ -15,15 +17,9 @@
 
 namespace DB
 {
-namespace ErrorCodes
-{
-    extern const int BAD_ARGUMENTS;
-}
 
 struct FormatImpl
 {
-    static constexpr size_t small_argument_threshold = 1024;
-    static constexpr size_t argument_threshold = std::numeric_limits<UInt32>::max();
     static constexpr size_t right_padding = 15;
 
     template <typename... Args>
@@ -39,165 +35,10 @@ struct FormatImpl
             format<false, false>(std::forward<Args>(args)...);
     }
 
-    static void parseNumber(const String & description, UInt64 l, UInt64 r, UInt64 & res)
-    {
-        res = 0;
-        for (UInt64 pos = l; pos < r; ++pos)
-        {
-            if (!isNumericASCII(description[pos]))
-                throw Exception("Not a number in curly braces at position " + std::to_string(pos), ErrorCodes::BAD_ARGUMENTS);
-            res = res * 10 + description[pos] - '0';
-            if (res >= argument_threshold)
-                throw Exception(
-                    "Too big number for arguments, must be at most " + std::to_string(argument_threshold), ErrorCodes::BAD_ARGUMENTS);
-        }
-    }
-
-    static inline void init(
-        const String & pattern,
-        const std::vector<const ColumnString::Chars *> & data,
-        size_t argument_number,
-        const std::vector<String> & constant_strings,
-        UInt64 * index_positions_ptr,
-        std::vector<String> & substrings)
-    {
-        /// Is current position after open curly brace.
-        bool is_open_curly = false;
-        /// The position of last open token.
-        size_t last_open = -1;
-
-        /// Is formatting in a plain {} token.
-        std::optional<bool> is_plain_numbering;
-        UInt64 index_if_plain = 0;
-
-        /// Left position of adding substrings, just to the closed brace position or the start of the string.
-        /// Invariant --- the start of substring is in this position.
-        size_t start_pos = 0;
-
-        /// A flag to decide whether we should glue the constant strings.
-        bool glue_to_next = false;
-
-        /// Handling double braces (escaping).
-        auto double_brace_removal = [](String & str)
-        {
-            size_t i = 0;
-            bool should_delete = true;
-            str.erase(
-                std::remove_if(
-                    str.begin(),
-                    str.end(),
-                    [&i, &should_delete, &str](char)
-                    {
-                        bool is_double_brace = (str[i] == '{' && str[i + 1] == '{') || (str[i] == '}' && str[i + 1] == '}');
-                        ++i;
-                        if (is_double_brace && should_delete)
-                        {
-                            should_delete = false;
-                            return true;
-                        }
-                        should_delete = true;
-                        return false;
-                    }),
-                str.end());
-        };
-
-        for (size_t i = 0; i < pattern.size(); ++i)
-        {
-            if (pattern[i] == '{')
-            {
-                /// Escaping handling
-                /// It is safe to access because of null termination
-                if (pattern[i + 1] == '{')
-                {
-                    ++i;
-                    continue;
-                }
-
-                if (is_open_curly)
-                    throw Exception("Two open curly braces without close one at position " + std::to_string(i), ErrorCodes::BAD_ARGUMENTS);
-
-                String to_add = String(pattern.data() + start_pos, i - start_pos);
-                double_brace_removal(to_add);
-                if (!glue_to_next)
-                    substrings.emplace_back(to_add);
-                else
-                    substrings.back() += to_add;
-
-                glue_to_next = false;
-
-                is_open_curly = true;
-                last_open = i + 1;
-            }
-            else if (pattern[i] == '}')
-            {
-                if (pattern[i + 1] == '}')
-                {
-                    ++i;
-                    continue;
-                }
-
-                if (!is_open_curly)
-                    throw Exception("Closed curly brace without open one at position " + std::to_string(i), ErrorCodes::BAD_ARGUMENTS);
-
-                is_open_curly = false;
-
-                if (last_open == i)
-                {
-                    if (is_plain_numbering && !*is_plain_numbering)
-                        throw Exception(
-                            "Cannot switch from automatic field numbering to manual field specification", ErrorCodes::BAD_ARGUMENTS);
-                    is_plain_numbering = true;
-                    if (index_if_plain >= argument_number)
-                        throw Exception("Argument is too big for formatting", ErrorCodes::BAD_ARGUMENTS);
-                    *index_positions_ptr = index_if_plain++;
-                }
-                else
-                {
-                    if (is_plain_numbering && *is_plain_numbering)
-                        throw Exception(
-                            "Cannot switch from automatic field numbering to manual field specification", ErrorCodes::BAD_ARGUMENTS);
-                    is_plain_numbering = false;
-
-                    UInt64 arg;
-                    parseNumber(pattern, last_open, i, arg);
-
-                    if (arg >= argument_number)
-                        throw Exception(
-                            "Argument is too big for formatting. Note that indexing starts from zero", ErrorCodes::BAD_ARGUMENTS);
-
-                    *index_positions_ptr = arg;
-                }
-
-                /// Constant string.
-                if (!data[*index_positions_ptr])
-                {
-                    /// The next string should be glued to last `A {} C`.format('B') -> `A B C`.
-                    glue_to_next = true;
-                    substrings.back() += constant_strings[*index_positions_ptr];
-                }
-                else
-                    ++index_positions_ptr; /// Otherwise we commit arg number and proceed.
-
-                start_pos = i + 1;
-            }
-        }
-
-        if (is_open_curly)
-            throw Exception("Last open curly brace is not closed", ErrorCodes::BAD_ARGUMENTS);
-
-        String to_add = String(pattern.data() + start_pos, pattern.size() - start_pos);
-        double_brace_removal(to_add);
-
-        if (!glue_to_next)
-            substrings.emplace_back(to_add);
-        else
-            substrings.back() += to_add;
-    }
-
     /// data for ColumnString and ColumnFixed. Nullptr means no data, it is const string.
     /// offsets for ColumnString, nullptr is an indicator that there is a fixed string rather than ColumnString.
     /// fixed_string_N for savings N to fixed strings.
-    /// constant_strings for constant strings. If data[i] is nullptr, than it is constant string.
+    /// constant_strings for constant strings. If data[i] is nullptr, it is constant string.
     /// res_data is result_data, res_offsets is offset result.
     /// input_rows_count is the number of rows processed.
     /// Precondition: data.size() == offsets.size() == fixed_string_N.size() == constant_strings.size().
@@ -207,29 +48,22 @@ struct FormatImpl
         const std::vector<const ColumnString::Chars *> & data,
         const std::vector<const ColumnString::Offsets *> & offsets,
         [[maybe_unused]] /* Because sometimes !has_column_fixed_string */ const std::vector<size_t> & fixed_string_N,
-        const std::vector<String> & constant_strings,
+        const std::vector<std::optional<String>> & constant_strings,
         ColumnString::Chars & res_data,
         ColumnString::Offsets & res_offsets,
         size_t input_rows_count)
     {
         const size_t argument_number = offsets.size();
 
-        UInt64 small_index_positions_buffer[small_argument_threshold];
-        /// The subsequent indexes of strings we should use. e.g `Hello world {1} {3} {1} {0}` this array will be filled with [1, 3, 1, 0, ... (garbage)] but without constant string indices.
-        UInt64 * index_positions = small_index_positions_buffer;
-
-        std::unique_ptr<UInt64[]> big_index_positions_buffer;
-        if (argument_number > small_argument_threshold)
-        {
-            big_index_positions_buffer.reset(new UInt64[argument_number]);
-            index_positions = big_index_positions_buffer.get();
-        }
+        /// The subsequent indexes of strings we should use. e.g `Hello world {1} {3} {1} {0}` this
+        /// array will be filled with [1, 3, 1, 0] but without constant string indices.
+        Format::IndexPositions index_positions;
 
         /// Vector of substrings of pattern that will be copied to the answer, not string view because of escaping and iterators invalidation.
         /// These are exactly what is between {} tokens, for `Hello {} world {}` we will have [`Hello `, ` world `, ``].
         std::vector<String> substrings;
 
-        init(pattern, data, argument_number, constant_strings, index_positions, substrings);
+        Format::init(pattern, argument_number, constant_strings, index_positions, substrings);
 
         UInt64 final_size = 0;
 
@@ -271,7 +105,7 @@ struct FormatImpl
                 for (size_t j = 1; j < substrings.size(); ++j)
                 {
                     UInt64 arg = index_positions[j - 1];
-                    auto offset_ptr = offsets[arg];
+                    const auto * offset_ptr = offsets[arg];
                     UInt64 arg_offset = 0;
                     UInt64 size = 0;
 

src/IO/ParallelReadBuffer.cpp

@@ -237,7 +237,7 @@ void ParallelReadBuffer::readerThreadFunction(ReadWorkerPtr read_worker)
         while (!emergency_stop && !read_worker->cancel)
         {
             if (!read_worker->reader->next())
-                throw Exception("Failed to read all the data from the reader", ErrorCodes::LOGICAL_ERROR);
+                throw Exception(ErrorCodes::LOGICAL_ERROR, "Failed to read all the data from the reader, missing {} bytes", read_worker->bytes_left);
 
             if (emergency_stop || read_worker->cancel)
                 break;

src/IO/ParallelReadBuffer.h

@@ -82,8 +82,8 @@ public:
         std::unique_ptr<ReadBufferFactory> reader_factory_,
         ThreadPool * pool,
         size_t max_working_readers,
-        WorkerSetup worker_setup = {},
-        WorkerCleanup worker_cleanup = {});
+        WorkerSetup worker_setup = [](ThreadStatus &){},
+        WorkerCleanup worker_cleanup = [](ThreadStatus &){});
 
     ~ParallelReadBuffer() override { finishAndWait(); }
 

src/IO/ReadBufferFromS3.cpp

@@ -1,4 +1,5 @@
 #include <Common/config.h>
+#include "IO/S3Common.h"
 
 #if USE_AWS_S3
 
@@ -42,6 +43,7 @@ ReadBufferFromS3::ReadBufferFromS3(
     UInt64 max_single_read_retries_,
     const ReadSettings & settings_,
     bool use_external_buffer_,
+    size_t offset_,
     size_t read_until_position_,
     bool restricted_seek_)
     : SeekableReadBufferWithSize(nullptr, 0)
@@ -49,9 +51,10 @@ ReadBufferFromS3::ReadBufferFromS3(
     , bucket(bucket_)
     , key(key_)
     , max_single_read_retries(max_single_read_retries_)
+    , offset(offset_)
+    , read_until_position(read_until_position_)
     , read_settings(settings_)
     , use_external_buffer(use_external_buffer_)
-    , read_until_position(read_until_position_)
     , restricted_seek(restricted_seek_)
 {
 }
@@ -210,13 +213,14 @@ std::optional<size_t> ReadBufferFromS3::getTotalSize()
     if (file_size)
         return file_size;
 
-    Aws::S3::Model::HeadObjectRequest request;
-    request.SetBucket(bucket);
-    request.SetKey(key);
+    auto object_size = S3::getObjectSize(client_ptr, bucket, key, false);
 
-    auto outcome = client_ptr->HeadObject(request);
-    auto head_result = outcome.GetResultWithOwnership();
-    file_size = head_result.GetContentLength();
+    if (!object_size)
+    {
+        return std::nullopt;
+    }
+
+    file_size = object_size;
     return file_size;
 }
 
@@ -234,6 +238,11 @@ void ReadBufferFromS3::setReadUntilPosition(size_t position)
     }
 }
 
+SeekableReadBuffer::Range ReadBufferFromS3::getRemainingReadRange() const
+{
+    return Range{.left = static_cast<size_t>(offset), .right = read_until_position ? std::optional{read_until_position - 1} : std::nullopt};
+}
+
 std::unique_ptr<ReadBuffer> ReadBufferFromS3::initialize()
 {
     Aws::S3::Model::GetObjectRequest req;
@@ -272,6 +281,36 @@ std::unique_ptr<ReadBuffer> ReadBufferFromS3::initialize()
         throw Exception(outcome.GetError().GetMessage(), ErrorCodes::S3_ERROR);
 }
 
+SeekableReadBufferPtr ReadBufferS3Factory::getReader()
+{
+    const auto next_range = range_generator.nextRange();
+    if (!next_range)
+    {
+        return nullptr;
+    }
+
+    auto reader = std::make_shared<ReadBufferFromS3>(
+        client_ptr,
+        bucket,
+        key,
+        s3_max_single_read_retries,
+        read_settings,
+        false /*use_external_buffer*/,
+        next_range->first,
+        next_range->second);
+    return reader;
+}
+
+off_t ReadBufferS3Factory::seek(off_t off, [[maybe_unused]] int whence)
+{
+    range_generator = RangeGenerator{object_size, range_step, static_cast<size_t>(off)};
+    return off;
+}
+
+std::optional<size_t> ReadBufferS3Factory::getTotalSize()
+{
+    return object_size;
+}
 }
 
 #endif

src/IO/ReadBufferFromS3.h

@@ -1,5 +1,6 @@
 #pragma once
 
+#include <Common/RangeGenerator.h>
 #include <Common/config.h>
 
 #if USE_AWS_S3
@@ -7,6 +8,7 @@
 #include <memory>
 
 #include <IO/HTTPCommon.h>
+#include <IO/ParallelReadBuffer.h>
 #include <IO/ReadBuffer.h>
 #include <IO/ReadSettings.h>
 #include <IO/SeekableReadBuffer.h>
@@ -30,7 +32,9 @@ private:
     String bucket;
     String key;
     UInt64 max_single_read_retries;
+
     off_t offset = 0;
+    off_t read_until_position = 0;
 
     Aws::S3::Model::GetObjectResult read_result;
     std::unique_ptr<ReadBuffer> impl;
@@ -45,6 +49,7 @@ public:
         UInt64 max_single_read_retries_,
         const ReadSettings & settings_,
         bool use_external_buffer = false,
+        size_t offset_ = 0,
         size_t read_until_position_ = 0,
         bool restricted_seek_ = false);
 
@@ -58,7 +63,7 @@ public:
 
     void setReadUntilPosition(size_t position) override;
 
-    Range getRemainingReadRange() const override { return Range{ .left = static_cast<size_t>(offset), .right = read_until_position }; }
+    Range getRemainingReadRange() const override;
 
     size_t getFileOffsetOfBufferEnd() const override { return offset; }
 
@@ -69,13 +74,55 @@ private:
 
     bool use_external_buffer;
 
-    off_t read_until_position = 0;
-
     /// There is different seek policy for disk seek and for non-disk seek
     /// (non-disk seek is applied for seekable input formats: orc, arrow, parquet).
     bool restricted_seek;
 };
 
+/// Creates separate ReadBufferFromS3 for sequence of ranges of particular object
+class ReadBufferS3Factory : public ParallelReadBuffer::ReadBufferFactory
+{
+public:
+    explicit ReadBufferS3Factory(
+        std::shared_ptr<Aws::S3::S3Client> client_ptr_,
+        const String & bucket_,
+        const String & key_,
+        size_t range_step_,
+        size_t object_size_,
+        UInt64 s3_max_single_read_retries_,
+        const ReadSettings & read_settings_)
+        : client_ptr(client_ptr_)
+        , bucket(bucket_)
+        , key(key_)
+        , read_settings(read_settings_)
+        , range_generator(object_size_, range_step_)
+        , range_step(range_step_)
+        , object_size(object_size_)
+        , s3_max_single_read_retries(s3_max_single_read_retries_)
+    {
+        assert(range_step > 0);
+        assert(range_step < object_size);
+    }
+
+    SeekableReadBufferPtr getReader() override;
+
+    off_t seek(off_t off, [[maybe_unused]] int whence) override;
+
+    std::optional<size_t> getTotalSize() override;
+
+private:
+    std::shared_ptr<Aws::S3::S3Client> client_ptr;
+    const String bucket;
+    const String key;
+    ReadSettings read_settings;
+
+    RangeGenerator range_generator;
+    size_t range_step;
+    size_t object_size;
+
+    UInt64 s3_max_single_read_retries;
+};
+
 }
 
 #endif

src/IO/ReadWriteBufferFromHTTP.h

@@ -1,6 +1,7 @@
 #pragma once
 
 #include <functional>
+#include <Common/RangeGenerator.h>
 #include <IO/ConnectionTimeouts.h>
 #include <IO/HTTPCommon.h>
 #include <IO/ParallelReadBuffer.h>
@@ -635,43 +636,6 @@ public:
     void buildNewSession(const Poco::URI & uri) override { session = makeHTTPSession(uri, timeouts); }
 };
 
-class RangeGenerator
-{
-public:
-    explicit RangeGenerator(size_t total_size_, size_t range_step_, size_t range_start = 0)
-        : from(range_start), range_step(range_step_), total_size(total_size_)
-    {
-    }
-
-    size_t totalRanges() const { return static_cast<size_t>(round(static_cast<float>(total_size - from) / range_step)); }
-
-    using Range = std::pair<size_t, size_t>;
-
-    // return upper exclusive range of values, i.e. [from_range, to_range>
-    std::optional<Range> nextRange()
-    {
-        if (from >= total_size)
-        {
-            return std::nullopt;
-        }
-
-        auto to = from + range_step;
-        if (to >= total_size)
-        {
-            to = total_size;
-        }
-
-        Range range{from, to};
-        from = to;
-        return range;
-    }
-
-private:
-    size_t from;
-    size_t range_step;
-    size_t total_size;
-};
-
 class ReadWriteBufferFromHTTP : public detail::ReadWriteBufferFromHTTPBase<std::shared_ptr<UpdatableSession>>
 {
     using Parent = detail::ReadWriteBufferFromHTTPBase<std::shared_ptr<UpdatableSession>>;

src/IO/S3Common.cpp

@@ -24,6 +24,7 @@
 #    include <aws/core/utils/UUID.h>
 #    include <aws/core/http/HttpClientFactory.h>
 #    include <aws/s3/S3Client.h>
+#    include <aws/s3/model/HeadObjectRequest.h>  // Y_IGNORE
 
 #    include <IO/S3/PocoHTTPClientFactory.h>
 #    include <IO/S3/PocoHTTPClient.h>
@@ -682,6 +683,7 @@ namespace DB
 namespace ErrorCodes
 {
     extern const int BAD_ARGUMENTS;
+    extern const int S3_ERROR;
 }
 
 namespace S3
@@ -839,6 +841,26 @@ namespace S3
             throw Exception(ErrorCodes::BAD_ARGUMENTS, "Bucket name length is out of bounds in virtual hosted style S3 URI:     {}{}",
                             quoteString(bucket), !uri.empty() ? " (" + uri.toString() + ")" : "");
     }
+
+    size_t getObjectSize(std::shared_ptr<Aws::S3::S3Client> client_ptr, const String & bucket, const String & key, bool throw_on_error)
+    {
+        Aws::S3::Model::HeadObjectRequest req;
+        req.SetBucket(bucket);
+        req.SetKey(key);
+
+        Aws::S3::Model::HeadObjectOutcome outcome = client_ptr->HeadObject(req);
+
+        if (outcome.IsSuccess())
+        {
+            auto read_result = outcome.GetResultWithOwnership();
+            return static_cast<size_t>(read_result.GetContentLength());
+        }
+        else if (throw_on_error)
+        {
+            throw DB::Exception(outcome.GetError().GetMessage(), ErrorCodes::S3_ERROR);
+        }
+        return 0;
+    }
 }
 
 }

src/IO/S3Common.h

@@ -75,6 +75,8 @@ struct URI
     static void validateBucket(const String & bucket, const Poco::URI & uri);
 };
 
+size_t getObjectSize(std::shared_ptr<Aws::S3::S3Client> client_ptr, const String & bucket, const String & key, bool throw_on_error = true);
+
 }
 
 #endif

src/Interpreters/CatBoostModel.cpp

@@ -169,6 +169,7 @@ public:
         if (columns.size() != float_features_count + cat_features_count)
             throw Exception(ErrorCodes::BAD_ARGUMENTS,
                 "Number of columns is different with number of features: columns size {} float features size {} + cat features size {}",
+                columns.size(),
                 float_features_count,
                 cat_features_count);
 

src/Interpreters/DatabaseCatalog.cpp

@@ -233,7 +233,7 @@ DatabaseAndTable DatabaseCatalog::getTableImpl(
         {
             assert(!db_and_table.first && !db_and_table.second);
             if (exception)
-                exception->emplace(ErrorCodes::UNKNOWN_TABLE, "Table {} doesn't exist", table_id.getNameForLogs());
+                exception->emplace(fmt::format("Table {} doesn't exist", table_id.getNameForLogs()), ErrorCodes::UNKNOWN_TABLE);
             return {};
         }
 
@@ -263,7 +263,7 @@ DatabaseAndTable DatabaseCatalog::getTableImpl(
         /// If table_id has no UUID, then the name of database was specified by user and table_id was not resolved through context.
         /// Do not allow access to TEMPORARY_DATABASE because it contains all temporary tables of all contexts and users.
         if (exception)
-            exception->emplace(ErrorCodes::DATABASE_ACCESS_DENIED, "Direct access to `{}` database is not allowed", String(TEMPORARY_DATABASE));
+            exception->emplace(fmt::format("Direct access to `{}` database is not allowed", TEMPORARY_DATABASE), ErrorCodes::DATABASE_ACCESS_DENIED);
         return {};
     }
 
@@ -274,7 +274,7 @@ DatabaseAndTable DatabaseCatalog::getTableImpl(
         if (databases.end() == it)
         {
             if (exception)
-                exception->emplace(ErrorCodes::UNKNOWN_DATABASE, "Database {} doesn't exist", backQuoteIfNeed(table_id.getDatabaseName()));
+                exception->emplace(fmt::format("Database {} doesn't exist", backQuoteIfNeed(table_id.getDatabaseName())), ErrorCodes::UNKNOWN_DATABASE);
             return {};
         }
         database = it->second;
@@ -282,7 +282,7 @@ DatabaseAndTable DatabaseCatalog::getTableImpl(
 
     auto table = database->tryGetTable(table_id.table_name, context_);
     if (!table && exception)
-            exception->emplace(ErrorCodes::UNKNOWN_TABLE, "Table {} doesn't exist", table_id.getNameForLogs());
+            exception->emplace(fmt::format("Table {} doesn't exist", table_id.getNameForLogs()), ErrorCodes::UNKNOWN_TABLE);
     if (!table)
         database = nullptr;
 

src/Interpreters/executeDDLQueryOnCluster.cpp

@@ -320,12 +320,13 @@ Chunk DDLQueryStatusSource::generate()
             if (throw_on_timeout)
             {
                 if (!first_exception)
-                    first_exception = std::make_unique<Exception>(ErrorCodes::TIMEOUT_EXCEEDED, msg_format,
-                        node_path, timeout_seconds, num_unfinished_hosts, num_active_hosts);
+                    first_exception = std::make_unique<Exception>(
+                        fmt::format(msg_format, node_path, timeout_seconds, num_unfinished_hosts, num_active_hosts),
+                        ErrorCodes::TIMEOUT_EXCEEDED);
                 return {};
             }
 
-            LOG_INFO(log, fmt::runtime(msg_format), node_path, timeout_seconds, num_unfinished_hosts, num_active_hosts);
+            LOG_INFO(log, msg_format, node_path, timeout_seconds, num_unfinished_hosts, num_active_hosts);
 
             NameSet unfinished_hosts = waiting_hosts;
             for (const auto & host_id : finished_hosts)
@@ -358,9 +359,12 @@ Chunk DDLQueryStatusSource::generate()
             /// Paradoxically, this exception will be throw even in case of "never_throw" mode.
 
             if (!first_exception)
-                first_exception = std::make_unique<Exception>(ErrorCodes::UNFINISHED,
-                    "Cannot provide query execution status. The query's node {} has been deleted by the cleaner"
-                    " since it was finished (or its lifetime is expired)", node_path);
+                first_exception = std::make_unique<Exception>(
+                    fmt::format(
+                        "Cannot provide query execution status. The query's node {} has been deleted by the cleaner"
+                        " since it was finished (or its lifetime is expired)",
+                        node_path),
+                    ErrorCodes::UNFINISHED);
             return {};
         }
 
@@ -386,7 +390,8 @@ Chunk DDLQueryStatusSource::generate()
             if (status.code != 0 && !first_exception
                 && context->getSettingsRef().distributed_ddl_output_mode != DistributedDDLOutputMode::NEVER_THROW)
             {
-                first_exception = std::make_unique<Exception>(status.code, "There was an error on [{}:{}]: {}", host, port, status.message);
+                first_exception = std::make_unique<Exception>(
+                    fmt::format("There was an error on [{}:{}]: {}", host, port, status.message), status.code);
             }
 
             ++num_hosts_finished;

src/Processors/Formats/Impl/MsgPackRowInputFormat.cpp

@@ -359,7 +359,7 @@ bool MsgPackVisitor::visit_ext(const char * value, uint32_t size)
         return true;
     }
 
-    throw Exception(ErrorCodes::BAD_ARGUMENTS, "Unsupported MsgPack extension type: {%x}", type);
+    throw Exception(ErrorCodes::BAD_ARGUMENTS, "Unsupported MsgPack extension type: {:x}", type);
 }
 
 void MsgPackVisitor::parse_error(size_t, size_t) // NOLINT
@@ -498,7 +498,7 @@ DataTypePtr MsgPackSchemaReader::getDataType(const msgpack::object & object)
             msgpack::object_ext object_ext = object.via.ext;
             if (object_ext.type() == int8_t(MsgPackExtensionTypes::UUIDType))
                 return std::make_shared<DataTypeUUID>();
-            throw Exception(ErrorCodes::BAD_ARGUMENTS, "Msgpack extension type {%x} is not supported", object_ext.type());
+            throw Exception(ErrorCodes::BAD_ARGUMENTS, "Msgpack extension type {:x} is not supported", object_ext.type());
         }
     }
     __builtin_unreachable();

src/Storages/Kafka/ReadBufferFromKafkaConsumer.cpp

@@ -1,10 +1,13 @@
+// Needs to go first because its partial specialization of fmt::formatter
+// should be defined before any instantiation
+#include <fmt/ostream.h>
+
 #include <Storages/Kafka/ReadBufferFromKafkaConsumer.h>
 
 #include <base/logger_useful.h>
 
 #include <cppkafka/cppkafka.h>
 #include <boost/algorithm/string/join.hpp>
-#include <fmt/ostream.h>
 #include <algorithm>
 
 namespace DB

src/Storages/StorageS3.cpp

@@ -1,4 +1,6 @@
 #include <Common/config.h>
+#include "IO/ParallelReadBuffer.h"
+#include "IO/IOThreadPool.h"
 #include "Parsers/ASTCreateQuery.h"
 
 #if USE_AWS_S3
@@ -238,7 +240,8 @@ StorageS3Source::StorageS3Source(
     String compression_hint_,
     const std::shared_ptr<Aws::S3::S3Client> & client_,
     const String & bucket_,
-    std::shared_ptr<IteratorWrapper> file_iterator_)
+    std::shared_ptr<IteratorWrapper> file_iterator_,
+    const size_t download_thread_num_)
     : SourceWithProgress(getHeader(sample_block_, need_path, need_file))
     , WithContext(context_)
     , name(std::move(name_))
@@ -254,6 +257,7 @@ StorageS3Source::StorageS3Source(
     , with_file_column(need_file)
     , with_path_column(need_path)
     , file_iterator(file_iterator_)
+    , download_thread_num(download_thread_num_)
 {
     initialize();
 }
@@ -275,28 +279,79 @@ bool StorageS3Source::initialize()
 
     file_path = fs::path(bucket) / current_key;
 
-    read_buf = wrapReadBufferWithCompressionMethod(
-        std::make_unique<ReadBufferFromS3>(client, bucket, current_key, max_single_read_retries, getContext()->getReadSettings()),
-        chooseCompressionMethod(current_key, compression_hint));
+    read_buf = wrapReadBufferWithCompressionMethod(createS3ReadBuffer(current_key), chooseCompressionMethod(current_key, compression_hint));
+
     auto input_format = getContext()->getInputFormat(format, *read_buf, sample_block, max_block_size, format_settings);
     QueryPipelineBuilder builder;
     builder.init(Pipe(input_format));
 
     if (columns_desc.hasDefaults())
     {
-        builder.addSimpleTransform([&](const Block & header)
-        {
-            return std::make_shared<AddingDefaultsTransform>(header, columns_desc, *input_format, getContext());
-        });
+        builder.addSimpleTransform(
+            [&](const Block & header)
+            { return std::make_shared<AddingDefaultsTransform>(header, columns_desc, *input_format, getContext()); });
     }
 
     pipeline = std::make_unique<QueryPipeline>(QueryPipelineBuilder::getPipeline(std::move(builder)));
     reader = std::make_unique<PullingPipelineExecutor>(*pipeline);
 
-    initialized = false;
     return true;
 }
 
+std::unique_ptr<ReadBuffer> StorageS3Source::createS3ReadBuffer(const String & key)
+{
+    const size_t object_size = DB::S3::getObjectSize(client, bucket, key, false);
+
+    auto download_buffer_size = getContext()->getSettings().max_download_buffer_size;
+    const bool use_parallel_download = download_buffer_size > 0 && download_thread_num > 1;
+    const bool object_too_small = object_size < download_thread_num * download_buffer_size;
+    if (!use_parallel_download || object_too_small)
+    {
+        LOG_TRACE(log, "Downloading object of size {} from S3 in single thread", object_size);
+        return std::make_unique<ReadBufferFromS3>(client, bucket, key, max_single_read_retries, getContext()->getReadSettings());
+    }
+
+    assert(object_size > 0);
+
+    if (download_buffer_size < DBMS_DEFAULT_BUFFER_SIZE)
+    {
+        LOG_WARNING(log, "Downloading buffer {} bytes too small, set at least {} bytes", download_buffer_size, DBMS_DEFAULT_BUFFER_SIZE);
+        download_buffer_size = DBMS_DEFAULT_BUFFER_SIZE;
+    }
+
+    auto factory = std::make_unique<ReadBufferS3Factory>(
+        client, bucket, key, download_buffer_size, object_size, max_single_read_retries, getContext()->getReadSettings());
+    LOG_TRACE(
+        log, "Downloading from S3 in {} threads. Object size: {}, Range size: {}.", download_thread_num, object_size, download_buffer_size);
+
+    ThreadGroupStatusPtr running_group = CurrentThread::isInitialized() && CurrentThread::get().getThreadGroup()
+        ? CurrentThread::get().getThreadGroup()
+        : MainThreadStatus::getInstance().getThreadGroup();
+
+    ContextPtr query_context = CurrentThread::isInitialized() ? CurrentThread::get().getQueryContext() : nullptr;
+
+    auto worker_cleanup = [has_running_group = running_group == nullptr](ThreadStatus & thread_status)
+    {
+        if (has_running_group)
+            thread_status.detachQuery(false);
+    };
+
+    auto worker_setup = [query_context = std::move(query_context),
+                         running_group = std::move(running_group)](ThreadStatus & thread_status)
+    {
+        /// Save query context if any, because cache implementation needs it.
+        if (query_context)
+            thread_status.attachQueryContext(query_context);
+
+        /// To be able to pass ProfileEvents.
+        if (running_group)
+            thread_status.attachQuery(running_group);
+    };
+
+    return std::make_unique<ParallelReadBuffer>(
+        std::move(factory), &IOThreadPool::get(), download_thread_num, std::move(worker_setup), std::move(worker_cleanup));
+}
+
 String StorageS3Source::getName() const
 {
     return name;
@@ -670,6 +725,7 @@ Pipe StorageS3::read(
         block_for_format = storage_snapshot->metadata->getSampleBlock();
     }
 
+    const size_t max_download_threads = local_context->getSettingsRef().max_download_threads;
     for (size_t i = 0; i < num_streams; ++i)
     {
         pipes.emplace_back(std::make_shared<StorageS3Source>(
@@ -686,7 +742,8 @@ Pipe StorageS3::read(
             compression_method,
             client_auth.client,
             client_auth.uri.bucket,
-            iterator_wrapper));
+            iterator_wrapper,
+            max_download_threads));
     }
     auto pipe = Pipe::unitePipes(std::move(pipes));
 

src/Storages/StorageS3.h

@@ -74,7 +74,8 @@ public:
         String compression_hint_,
         const std::shared_ptr<Aws::S3::S3Client> & client_,
         const String & bucket,
-        std::shared_ptr<IteratorWrapper> file_iterator_);
+        std::shared_ptr<IteratorWrapper> file_iterator_,
+        size_t download_thread_num);
 
     String getName() const override;
 
@@ -101,13 +102,17 @@ private:
     std::unique_ptr<PullingPipelineExecutor> reader;
     /// onCancel and generate can be called concurrently
     std::mutex reader_mutex;
-    bool initialized = false;
     bool with_file_column = false;
     bool with_path_column = false;
     std::shared_ptr<IteratorWrapper> file_iterator;
+    size_t download_thread_num = 1;
+
+    Poco::Logger * log = &Poco::Logger::get("StorageS3Source");
 
     /// Recreate ReadBuffer and BlockInputStream for each file.
     bool initialize();
+
+    std::unique_ptr<ReadBuffer> createS3ReadBuffer(const String & key);
 };
 
 /**

tests/ci/cherry_pick_utils/parser.py

@@ -20,8 +20,6 @@ class Description:
 
     def __init__(self, pull_request):
         self.label_name = str()
-        self.legal = False
-
         self._parse(pull_request["bodyText"])
 
     def _parse(self, text):
@@ -39,12 +37,6 @@ class Description:
                 category = stripped
                 next_category = False
 
-            if (
-                stripped
-                == "I hereby agree to the terms of the CLA available at: https://yandex.ru/legal/cla/?lang=en"
-            ):
-                self.legal = True
-
             category_headers = (
                 "Category (leave one):",
                 "Changelog category (leave one):",

tests/ci/commit_status_helper.py

@@ -59,3 +59,17 @@ def post_commit_status_to_file(file_path, description, state, report_url):
     with open(file_path, "w", encoding="utf-8") as f:
         out = csv.writer(f, delimiter="\t")
         out.writerow([state, report_url, description])
+
+
+def remove_labels(gh, pr_info, labels_names):
+    repo = gh.get_repo(GITHUB_REPOSITORY)
+    pull_request = repo.get_pull(pr_info.number)
+    for label in labels_names:
+        pull_request.remove_from_labels(label)
+
+
+def post_labels(gh, pr_info, labels_names):
+    repo = gh.get_repo(GITHUB_REPOSITORY)
+    pull_request = repo.get_pull(pr_info.number)
+    for label in labels_names:
+        pull_request.add_to_labels(label)

tests/ci/pr_info.py

@@ -236,6 +236,15 @@ class PRInfo:
                 return True
         return False
 
+    def has_changes_in_submodules(self):
+        if self.changed_files is None or not self.changed_files:
+            return True
+
+        for f in self.changed_files:
+            if "contrib" in f:
+                return True
+        return False
+
     def can_skip_builds_and_use_version_from_master(self):
         # TODO: See a broken loop
         if "force tests" in self.labels:

tests/ci/run_check.py

@@ -8,7 +8,7 @@ from github import Github
 from env_helper import GITHUB_RUN_URL, GITHUB_REPOSITORY, GITHUB_SERVER_URL
 from pr_info import PRInfo
 from get_robot_token import get_best_robot_token
-from commit_status_helper import get_commit
+from commit_status_helper import get_commit, post_labels, remove_labels
 
 NAME = "Run Check (actions)"
 
@@ -22,6 +22,7 @@ OK_SKIP_LABELS = {"release", "pr-backport", "pr-cherrypick"}
 CAN_BE_TESTED_LABEL = "can be tested"
 DO_NOT_TEST_LABEL = "do not test"
 FORCE_TESTS_LABEL = "force tests"
+SUBMODULE_CHANGED_LABEL = "submodule changed"
 
 # Individual trusted contirbutors who are not in any trusted organization.
 # Can be changed in runtime: we will append users that we learned to be in
@@ -81,6 +82,25 @@ TRUSTED_CONTRIBUTORS = {
     ]
 }
 
+MAP_CATEGORY_TO_LABEL = {
+    "New Feature": "pr-feature",
+    "Bug Fix": "pr-bugfix",
+    "Bug Fix (user-visible misbehaviour in official stable or prestable release)": "pr-bugfix",
+    "Improvement": "pr-improvement",
+    "Performance Improvement": "pr-performance",
+    "Backward Incompatible Change": "pr-backward-incompatible",
+    "Build/Testing/Packaging Improvement": "pr-build",
+    "Build Improvement": "pr-build",
+    "Build/Testing Improvement": "pr-build",
+    "Build": "pr-build",
+    "Packaging Improvement": "pr-build",
+    "Not for changelog (changelog entry is not required)": "pr-not-for-changelog",
+    "Not for changelog": "pr-not-for-changelog",
+    "Documentation (changelog entry is not required)": "pr-documentation",
+    "Documentation": "pr-documentation",
+    # 'Other': doesn't match anything
+}
+
 
 def pr_is_by_trusted_user(pr_user_login, pr_user_orgs):
     if pr_user_login.lower() in TRUSTED_CONTRIBUTORS:
@@ -168,7 +188,7 @@ def check_pr_description(pr_info):
                     + second_category
                     + "'"
                 )
-                return result_status[:140]
+                return result_status[:140], category
 
         elif re.match(
             r"(?i)^[>*_ ]*(short\s*description|change\s*log\s*entry)", lines[i]
@@ -190,30 +210,57 @@ def check_pr_description(pr_info):
             i += 1
 
     if not category:
-        return "Changelog category is empty"
+        return "Changelog category is empty", category
 
     # Filter out the PR categories that are not for changelog.
     if re.match(
         r"(?i)doc|((non|in|not|un)[-\s]*significant)|(not[ ]*for[ ]*changelog)",
         category,
     ):
-        return ""
+        return "", category
 
     if not entry:
-        return f"Changelog entry required for category '{category}'"
+        return f"Changelog entry required for category '{category}'", category
 
-    return ""
+    return "", category
 
 
 if __name__ == "__main__":
     logging.basicConfig(level=logging.INFO)
 
-    pr_info = PRInfo(need_orgs=True, pr_event_from_api=True)
+    pr_info = PRInfo(need_orgs=True, pr_event_from_api=True, need_changed_files=True)
     can_run, description, labels_state = should_run_checks_for_pr(pr_info)
     gh = Github(get_best_robot_token())
     commit = get_commit(gh, pr_info.sha)
 
-    description_report = check_pr_description(pr_info)[:139]
+    description_report, category = check_pr_description(pr_info)
+    pr_labels_to_add = []
+    pr_labels_to_remove = []
+    if (
+        category in MAP_CATEGORY_TO_LABEL
+        and MAP_CATEGORY_TO_LABEL[category] not in pr_info.labels
+    ):
+        pr_labels_to_add.append(MAP_CATEGORY_TO_LABEL[category])
+
+    for label in pr_info.labels:
+        if (
+            label in MAP_CATEGORY_TO_LABEL.values()
+            and category in MAP_CATEGORY_TO_LABEL
+            and label != MAP_CATEGORY_TO_LABEL[category]
+        ):
+            pr_labels_to_remove.append(label)
+
+    if pr_info.has_changes_in_submodules():
+        pr_labels_to_add.append(SUBMODULE_CHANGED_LABEL)
+    elif SUBMODULE_CHANGED_LABEL in pr_info.labels:
+        pr_labels_to_remove.append(SUBMODULE_CHANGED_LABEL)
+
+    if pr_labels_to_add:
+        post_labels(gh, pr_info, pr_labels_to_add)
+
+    if pr_labels_to_remove:
+        remove_labels(gh, pr_info, pr_labels_to_remove)
+
     if description_report:
         print("::notice ::Cannot run, description does not match the template")
         logging.info(
@@ -225,7 +272,7 @@ if __name__ == "__main__":
         )
         commit.create_status(
             context=NAME,
-            description=description_report,
+            description=description_report[:139],
             state="failure",
             target_url=url,
         )

tests/integration/test_materialized_view_restart_server/test.py

@@ -0,0 +1,25 @@
+import pytest
+from helpers.cluster import ClickHouseCluster
+
+cluster = ClickHouseCluster(__file__)
+node = cluster.add_instance("node", stay_alive=True)
+
+
+@pytest.fixture(scope="module")
+def start_cluster():
+    try:
+        cluster.start()
+        yield cluster
+    finally:
+        cluster.shutdown()
+
+
+def test_materialized_view_with_subquery(start_cluster):
+    node.query("create table test (x UInt32) engine=TinyLog()")
+    node.query(
+        "create materialized view mv engine = TinyLog() as with subquery as (select * from test) select * from subquery"
+    )
+    node.restart_clickhouse(kill=True)
+    node.query("insert into test select 1")
+    result = node.query("select * from mv")
+    assert int(result) == 1

tests/integration/test_storage_s3/test.py

@@ -517,7 +517,7 @@ def test_put_get_with_globs(started_cluster):
         # ("'minio','minio123',",True), Redirect with credentials not working with nginx.
     ],
 )
-def test_multipart_put(started_cluster, maybe_auth, positive):
+def test_multipart(started_cluster, maybe_auth, positive):
     # type: (ClickHouseCluster) -> None
 
     bucket = (
@@ -535,8 +535,9 @@ def test_multipart_put(started_cluster, maybe_auth, positive):
 
     one_line_length = 6  # 3 digits, 2 commas, 1 line separator.
 
+    total_rows = csv_size_bytes // one_line_length
     # Generate data having size more than one part
-    int_data = [[1, 2, 3] for i in range(csv_size_bytes // one_line_length)]
+    int_data = [[1, 2, 3] for i in range(total_rows)]
     csv_data = "".join(["{},{},{}\n".format(x, y, z) for x, y, z in int_data])
 
     assert len(csv_data) > min_part_size_bytes
@@ -573,6 +574,37 @@ def test_multipart_put(started_cluster, maybe_auth, positive):
 
         assert csv_data == get_s3_file_content(started_cluster, bucket, filename)
 
+    # select uploaded data from many threads
+    select_query = (
+        "select sum(column1), sum(column2), sum(column3) "
+        "from s3('http://{host}:{port}/{bucket}/{filename}', {auth}'CSV', '{table_format}')".format(
+            host=started_cluster.minio_redirect_host,
+            port=started_cluster.minio_redirect_port,
+            bucket=bucket,
+            filename=filename,
+            auth=maybe_auth,
+            table_format=table_format,
+        )
+    )
+    try:
+        select_result = run_query(
+            instance,
+            select_query,
+            settings={
+                "max_download_threads": random.randint(4, 16),
+                "max_download_buffer_size": 1024 * 1024,
+            },
+        )
+    except helpers.client.QueryRuntimeException:
+        if positive:
+            raise
+    else:
+        assert positive
+        assert (
+            select_result
+            == "\t".join(map(str, [total_rows, total_rows * 2, total_rows * 3])) + "\n"
+        )
+
 
 def test_remote_host_filter(started_cluster):
     instance = started_cluster.instances["restricted_dummy"]

tests/queries/0_stateless/02245_format_string_stack_overflow.reference

@@ -0,0 +1 @@
+00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000

utils/github/__init__.py

@@ -1 +0,0 @@
-# -*- coding: utf-8 -*-

utils/github/backport.py

@@ -1,185 +0,0 @@
-# -*- coding: utf-8 -*-
-
-try:
-    from clickhouse.utils.github.cherrypick import CherryPick
-    from clickhouse.utils.github.query import Query as RemoteRepo
-    from clickhouse.utils.github.local import Repository as LocalRepo
-except:
-    from .cherrypick import CherryPick
-    from .query import Query as RemoteRepo
-    from .local import Repository as LocalRepo
-
-import argparse
-import logging
-import re
-import sys
-
-
-class Backport:
-    def __init__(self, token, owner, name, team):
-        self._gh = RemoteRepo(
-            token, owner=owner, name=name, team=team, max_page_size=30, min_page_size=7
-        )
-        self._token = token
-        self.default_branch_name = self._gh.default_branch
-        self.ssh_url = self._gh.ssh_url
-
-    def getPullRequests(self, from_commit):
-        return self._gh.get_pull_requests(from_commit)
-
-    def getBranchesWithRelease(self):
-        branches = set()
-        for pull_request in self._gh.find_pull_requests("release"):
-            branches.add(pull_request["headRefName"])
-        return branches
-
-    def execute(self, repo, upstream, until_commit, run_cherrypick):
-        repo = LocalRepo(repo, upstream, self.default_branch_name)
-        all_branches = repo.get_release_branches()  # [(branch_name, base_commit)]
-
-        release_branches = self.getBranchesWithRelease()
-
-        branches = []
-        # iterate over all branches to preserve their precedence.
-        for branch in all_branches:
-            if branch[0] in release_branches:
-                branches.append(branch)
-
-        if not branches:
-            logging.info("No release branches found!")
-            return
-
-        for branch in branches:
-            logging.info("Found release branch: %s", branch[0])
-
-        if not until_commit:
-            until_commit = branches[0][1]
-        pull_requests = self.getPullRequests(until_commit)
-
-        backport_map = {}
-
-        RE_MUST_BACKPORT = re.compile(r"^v(\d+\.\d+)-must-backport$")
-        RE_NO_BACKPORT = re.compile(r"^v(\d+\.\d+)-no-backport$")
-        RE_BACKPORTED = re.compile(r"^v(\d+\.\d+)-backported$")
-
-        # pull-requests are sorted by ancestry from the most recent.
-        for pr in pull_requests:
-            while repo.comparator(branches[-1][1]) >= repo.comparator(
-                pr["mergeCommit"]["oid"]
-            ):
-                logging.info(
-                    "PR #{} is already inside {}. Dropping this branch for further PRs".format(
-                        pr["number"], branches[-1][0]
-                    )
-                )
-                branches.pop()
-
-            logging.info("Processing PR #{}".format(pr["number"]))
-
-            assert len(branches)
-
-            branch_set = set([branch[0] for branch in branches])
-
-            # First pass. Find all must-backports
-            for label in pr["labels"]["nodes"]:
-                if label["name"] == "pr-must-backport":
-                    backport_map[pr["number"]] = branch_set.copy()
-                    continue
-                matched = RE_MUST_BACKPORT.match(label["name"])
-                if matched:
-                    if pr["number"] not in backport_map:
-                        backport_map[pr["number"]] = set()
-                    backport_map[pr["number"]].add(matched.group(1))
-
-            # Second pass. Find all no-backports
-            for label in pr["labels"]["nodes"]:
-                if label["name"] == "pr-no-backport" and pr["number"] in backport_map:
-                    del backport_map[pr["number"]]
-                    break
-                matched_no_backport = RE_NO_BACKPORT.match(label["name"])
-                matched_backported = RE_BACKPORTED.match(label["name"])
-                if (
-                    matched_no_backport
-                    and pr["number"] in backport_map
-                    and matched_no_backport.group(1) in backport_map[pr["number"]]
-                ):
-                    backport_map[pr["number"]].remove(matched_no_backport.group(1))
-                    logging.info(
-                        "\tskipping %s because of forced no-backport",
-                        matched_no_backport.group(1),
-                    )
-                elif (
-                    matched_backported
-                    and pr["number"] in backport_map
-                    and matched_backported.group(1) in backport_map[pr["number"]]
-                ):
-                    backport_map[pr["number"]].remove(matched_backported.group(1))
-                    logging.info(
-                        "\tskipping %s because it's already backported manually",
-                        matched_backported.group(1),
-                    )
-
-        for pr, branches in list(backport_map.items()):
-            logging.info("PR #%s needs to be backported to:", pr)
-            for branch in branches:
-                logging.info(
-                    "\t%s, and the status is: %s",
-                    branch,
-                    run_cherrypick(self._token, pr, branch),
-                )
-
-        # print API costs
-        logging.info("\nGitHub API total costs per query:")
-        for name, value in list(self._gh.api_costs.items()):
-            logging.info("%s : %s", name, value)
-
-
-if __name__ == "__main__":
-    parser = argparse.ArgumentParser()
-    parser.add_argument(
-        "--token", type=str, required=True, help="token for Github access"
-    )
-    parser.add_argument(
-        "--repo",
-        type=str,
-        required=True,
-        help="path to full repository",
-        metavar="PATH",
-    )
-    parser.add_argument(
-        "--til", type=str, help="check PRs from HEAD til this commit", metavar="COMMIT"
-    )
-    parser.add_argument(
-        "--dry-run",
-        action="store_true",
-        help="do not create or merge any PRs",
-        default=False,
-    )
-    parser.add_argument(
-        "--verbose",
-        "-v",
-        action="store_true",
-        help="more verbose output",
-        default=False,
-    )
-    parser.add_argument(
-        "--upstream",
-        "-u",
-        type=str,
-        help="remote name of upstream in repository",
-        default="origin",
-    )
-    args = parser.parse_args()
-
-    if args.verbose:
-        logging.basicConfig(
-            format="%(message)s", stream=sys.stdout, level=logging.DEBUG
-        )
-    else:
-        logging.basicConfig(format="%(message)s", stream=sys.stdout, level=logging.INFO)
-
-    cherrypick_run = lambda token, pr, branch: CherryPick(
-        token, "ClickHouse", "ClickHouse", "core", pr, branch
-    ).execute(args.repo, args.dry_run)
-    bp = Backport(args.token, "ClickHouse", "ClickHouse", "core")
-    bp.execute(args.repo, args.upstream, args.til, cherrypick_run)

utils/github/cherrypick.py

@@ -1,323 +0,0 @@
-# -*- coding: utf-8 -*-
-
-"""
-Backports changes from PR to release branch.
-Requires multiple separate runs as part of the implementation.
-
-First run should do the following:
-1. Merge release branch with a first parent of merge-commit of PR (using 'ours' strategy). (branch: backport/{branch}/{pr})
-2. Create temporary branch over merge-commit to use it for PR creation. (branch: cherrypick/{merge_commit})
-3. Create PR from temporary branch to backport branch (emulating cherry-pick).
-
-Second run checks PR from previous run to be merged or at least being mergeable. If it's not merged then try to merge it.
-
-Third run creates PR from backport branch (with merged previous PR) to release branch.
-"""
-
-try:
-    from clickhouse.utils.github.query import Query as RemoteRepo
-except:
-    from .query import Query as RemoteRepo
-
-import argparse
-from enum import Enum
-import logging
-import os
-import subprocess
-import sys
-
-
-class CherryPick:
-    class Status(Enum):
-        DISCARDED = "discarded"
-        NOT_INITIATED = "not started"
-        FIRST_MERGEABLE = "waiting for 1st stage"
-        FIRST_CONFLICTS = "conflicts on 1st stage"
-        SECOND_MERGEABLE = "waiting for 2nd stage"
-        SECOND_CONFLICTS = "conflicts on 2nd stage"
-        MERGED = "backported"
-
-    def _run(self, args):
-        out = subprocess.check_output(args).rstrip()
-        logging.debug(out)
-        return out
-
-    def __init__(self, token, owner, name, team, pr_number, target_branch):
-        self._gh = RemoteRepo(token, owner=owner, name=name, team=team)
-        self._pr = self._gh.get_pull_request(pr_number)
-
-        self.ssh_url = self._gh.ssh_url
-
-        # TODO: check if pull-request is merged.
-
-        self.merge_commit_oid = self._pr["mergeCommit"]["oid"]
-
-        self.target_branch = target_branch
-        self.backport_branch = "backport/{branch}/{pr}".format(
-            branch=target_branch, pr=pr_number
-        )
-        self.cherrypick_branch = "cherrypick/{branch}/{oid}".format(
-            branch=target_branch, oid=self.merge_commit_oid
-        )
-
-    def getCherryPickPullRequest(self):
-        return self._gh.find_pull_request(
-            base=self.backport_branch, head=self.cherrypick_branch
-        )
-
-    def createCherryPickPullRequest(self, repo_path):
-        DESCRIPTION = (
-            "This pull-request is a first step of an automated backporting.\n"
-            "It contains changes like after calling a local command `git cherry-pick`.\n"
-            "If you intend to continue backporting this changes, then resolve all conflicts if any.\n"
-            "Otherwise, if you do not want to backport them, then just close this pull-request.\n"
-            "\n"
-            "The check results does not matter at this step - you can safely ignore them.\n"
-            "Also this pull-request will be merged automatically as it reaches the mergeable state, but you always can merge it manually.\n"
-        )
-
-        # FIXME: replace with something better than os.system()
-        git_prefix = [
-            "git",
-            "-C",
-            repo_path,
-            "-c",
-            "user.email=robot-clickhouse@yandex-team.ru",
-            "-c",
-            "user.name=robot-clickhouse",
-        ]
-        base_commit_oid = self._pr["mergeCommit"]["parents"]["nodes"][0]["oid"]
-
-        # Create separate branch for backporting, and make it look like real cherry-pick.
-        self._run(git_prefix + ["checkout", "-f", self.target_branch])
-        self._run(git_prefix + ["checkout", "-B", self.backport_branch])
-        self._run(git_prefix + ["merge", "-s", "ours", "--no-edit", base_commit_oid])
-
-        # Create secondary branch to allow pull request with cherry-picked commit.
-        self._run(
-            git_prefix + ["branch", "-f", self.cherrypick_branch, self.merge_commit_oid]
-        )
-
-        self._run(
-            git_prefix
-            + [
-                "push",
-                "-f",
-                "origin",
-                "{branch}:{branch}".format(branch=self.backport_branch),
-            ]
-        )
-        self._run(
-            git_prefix
-            + [
-                "push",
-                "-f",
-                "origin",
-                "{branch}:{branch}".format(branch=self.cherrypick_branch),
-            ]
-        )
-
-        # Create pull-request like a local cherry-pick
-        pr = self._gh.create_pull_request(
-            source=self.cherrypick_branch,
-            target=self.backport_branch,
-            title="Cherry pick #{number} to {target}: {title}".format(
-                number=self._pr["number"],
-                target=self.target_branch,
-                title=self._pr["title"].replace('"', '\\"'),
-            ),
-            description="Original pull-request #{}\n\n{}".format(
-                self._pr["number"], DESCRIPTION
-            ),
-        )
-
-        # FIXME: use `team` to leave a single eligible assignee.
-        self._gh.add_assignee(pr, self._pr["author"])
-        self._gh.add_assignee(pr, self._pr["mergedBy"])
-
-        self._gh.set_label(pr, "do not test")
-        self._gh.set_label(pr, "pr-cherrypick")
-
-        return pr
-
-    def mergeCherryPickPullRequest(self, cherrypick_pr):
-        return self._gh.merge_pull_request(cherrypick_pr["id"])
-
-    def getBackportPullRequest(self):
-        return self._gh.find_pull_request(
-            base=self.target_branch, head=self.backport_branch
-        )
-
-    def createBackportPullRequest(self, cherrypick_pr, repo_path):
-        DESCRIPTION = (
-            "This pull-request is a last step of an automated backporting.\n"
-            "Treat it as a standard pull-request: look at the checks and resolve conflicts.\n"
-            "Merge it only if you intend to backport changes to the target branch, otherwise just close it.\n"
-        )
-
-        git_prefix = [
-            "git",
-            "-C",
-            repo_path,
-            "-c",
-            "user.email=robot-clickhouse@clickhouse.com",
-            "-c",
-            "user.name=robot-clickhouse",
-        ]
-
-        pr_title = "Backport #{number} to {target}: {title}".format(
-            number=self._pr["number"],
-            target=self.target_branch,
-            title=self._pr["title"].replace('"', '\\"'),
-        )
-
-        self._run(git_prefix + ["checkout", "-f", self.backport_branch])
-        self._run(git_prefix + ["pull", "--ff-only", "origin", self.backport_branch])
-        self._run(
-            git_prefix
-            + [
-                "reset",
-                "--soft",
-                self._run(
-                    git_prefix
-                    + [
-                        "merge-base",
-                        "origin/" + self.target_branch,
-                        self.backport_branch,
-                    ]
-                ),
-            ]
-        )
-        self._run(git_prefix + ["commit", "-a", "--allow-empty", "-m", pr_title])
-        self._run(
-            git_prefix
-            + [
-                "push",
-                "-f",
-                "origin",
-                "{branch}:{branch}".format(branch=self.backport_branch),
-            ]
-        )
-
-        pr = self._gh.create_pull_request(
-            source=self.backport_branch,
-            target=self.target_branch,
-            title=pr_title,
-            description="Original pull-request #{}\nCherry-pick pull-request #{}\n\n{}".format(
-                self._pr["number"], cherrypick_pr["number"], DESCRIPTION
-            ),
-        )
-
-        # FIXME: use `team` to leave a single eligible assignee.
-        self._gh.add_assignee(pr, self._pr["author"])
-        self._gh.add_assignee(pr, self._pr["mergedBy"])
-
-        self._gh.set_label(pr, "pr-backport")
-
-        return pr
-
-    def execute(self, repo_path, dry_run=False):
-        pr1 = self.getCherryPickPullRequest()
-        if not pr1:
-            if not dry_run:
-                pr1 = self.createCherryPickPullRequest(repo_path)
-                logging.debug(
-                    "Created PR with cherry-pick of %s to %s: %s",
-                    self._pr["number"],
-                    self.target_branch,
-                    pr1["url"],
-                )
-            else:
-                return CherryPick.Status.NOT_INITIATED
-        else:
-            logging.debug(
-                "Found PR with cherry-pick of %s to %s: %s",
-                self._pr["number"],
-                self.target_branch,
-                pr1["url"],
-            )
-
-        if not pr1["merged"] and pr1["mergeable"] == "MERGEABLE" and not pr1["closed"]:
-            if not dry_run:
-                pr1 = self.mergeCherryPickPullRequest(pr1)
-                logging.debug(
-                    "Merged PR with cherry-pick of %s to %s: %s",
-                    self._pr["number"],
-                    self.target_branch,
-                    pr1["url"],
-                )
-
-        if not pr1["merged"]:
-            logging.debug(
-                "Waiting for PR with cherry-pick of %s to %s: %s",
-                self._pr["number"],
-                self.target_branch,
-                pr1["url"],
-            )
-
-            if pr1["closed"]:
-                return CherryPick.Status.DISCARDED
-            elif pr1["mergeable"] == "CONFLICTING":
-                return CherryPick.Status.FIRST_CONFLICTS
-            else:
-                return CherryPick.Status.FIRST_MERGEABLE
-
-        pr2 = self.getBackportPullRequest()
-        if not pr2:
-            if not dry_run:
-                pr2 = self.createBackportPullRequest(pr1, repo_path)
-                logging.debug(
-                    "Created PR with backport of %s to %s: %s",
-                    self._pr["number"],
-                    self.target_branch,
-                    pr2["url"],
-                )
-            else:
-                return CherryPick.Status.FIRST_MERGEABLE
-        else:
-            logging.debug(
-                "Found PR with backport of %s to %s: %s",
-                self._pr["number"],
-                self.target_branch,
-                pr2["url"],
-            )
-
-        if pr2["merged"]:
-            return CherryPick.Status.MERGED
-        elif pr2["closed"]:
-            return CherryPick.Status.DISCARDED
-        elif pr2["mergeable"] == "CONFLICTING":
-            return CherryPick.Status.SECOND_CONFLICTS
-        else:
-            return CherryPick.Status.SECOND_MERGEABLE
-
-
-if __name__ == "__main__":
-    logging.basicConfig(format="%(message)s", stream=sys.stdout, level=logging.DEBUG)
-
-    parser = argparse.ArgumentParser()
-    parser.add_argument(
-        "--token", "-t", type=str, required=True, help="token for Github access"
-    )
-    parser.add_argument("--pr", type=str, required=True, help="PR# to cherry-pick")
-    parser.add_argument(
-        "--branch",
-        "-b",
-        type=str,
-        required=True,
-        help="target branch name for cherry-pick",
-    )
-    parser.add_argument(
-        "--repo",
-        "-r",
-        type=str,
-        required=True,
-        help="path to full repository",
-        metavar="PATH",
-    )
-    args = parser.parse_args()
-
-    cp = CherryPick(
-        args.token, "ClickHouse", "ClickHouse", "core", args.pr, args.branch
-    )
-    cp.execute(args.repo)

utils/github/local.py

@@ -1,108 +0,0 @@
-# -*- coding: utf-8 -*-
-
-import functools
-import logging
-import os
-import re
-
-
-class RepositoryBase:
-    def __init__(self, repo_path):
-        import git
-
-        self._repo = git.Repo(repo_path, search_parent_directories=(not repo_path))
-
-        # comparator of commits
-        def cmp(x, y):
-            if str(x) == str(y):
-                return 0
-            if self._repo.is_ancestor(x, y):
-                return -1
-            else:
-                return 1
-
-        self.comparator = functools.cmp_to_key(cmp)
-
-    def get_head_commit(self):
-        return self._repo.commit(self._default)
-
-    def iterate(self, begin, end):
-        rev_range = "{}...{}".format(begin, end)
-        for commit in self._repo.iter_commits(rev_range, first_parent=True):
-            yield commit
-
-
-class Repository(RepositoryBase):
-    def __init__(self, repo_path, remote_name, default_branch_name):
-        super(Repository, self).__init__(repo_path)
-        self._remote = self._repo.remotes[remote_name]
-        self._remote.fetch()
-        self._default = self._remote.refs[default_branch_name]
-
-    def get_release_branches(self):
-        """
-        Returns sorted list of tuples:
-         * remote branch (git.refs.remote.RemoteReference),
-         * base commit (git.Commit),
-         * head (git.Commit)).
-        List is sorted by commits in ascending order.
-        """
-        release_branches = []
-
-        RE_RELEASE_BRANCH_REF = re.compile(r"^refs/remotes/.+/\d+\.\d+$")
-
-        for branch in [
-            r for r in self._remote.refs if RE_RELEASE_BRANCH_REF.match(r.path)
-        ]:
-            base = self._repo.merge_base(self._default, self._repo.commit(branch))
-            if not base:
-                logging.info(
-                    "Branch %s is not based on branch %s. Ignoring.",
-                    branch.path,
-                    self._default,
-                )
-            elif len(base) > 1:
-                logging.info(
-                    "Branch %s has more than one base commit. Ignoring.", branch.path
-                )
-            else:
-                release_branches.append((os.path.basename(branch.name), base[0]))
-
-        return sorted(release_branches, key=lambda x: self.comparator(x[1]))
-
-
-class BareRepository(RepositoryBase):
-    def __init__(self, repo_path, default_branch_name):
-        super(BareRepository, self).__init__(repo_path)
-        self._default = self._repo.branches[default_branch_name]
-
-    def get_release_branches(self):
-        """
-        Returns sorted list of tuples:
-         * branch (git.refs.head?),
-         * base commit (git.Commit),
-         * head (git.Commit)).
-        List is sorted by commits in ascending order.
-        """
-        release_branches = []
-
-        RE_RELEASE_BRANCH_REF = re.compile(r"^refs/heads/\d+\.\d+$")
-
-        for branch in [
-            r for r in self._repo.branches if RE_RELEASE_BRANCH_REF.match(r.path)
-        ]:
-            base = self._repo.merge_base(self._default, self._repo.commit(branch))
-            if not base:
-                logging.info(
-                    "Branch %s is not based on branch %s. Ignoring.",
-                    branch.path,
-                    self._default,
-                )
-            elif len(base) > 1:
-                logging.info(
-                    "Branch %s has more than one base commit. Ignoring.", branch.path
-                )
-            else:
-                release_branches.append((os.path.basename(branch.name), base[0]))
-
-        return sorted(release_branches, key=lambda x: self.comparator(x[1]))

utils/github/parser.py

@@ -1,64 +0,0 @@
-# -*- coding: utf-8 -*-
-
-
-class Description:
-    """Parsed description representation"""
-
-    MAP_CATEGORY_TO_LABEL = {
-        "New Feature": "pr-feature",
-        "Bug Fix": "pr-bugfix",
-        "Improvement": "pr-improvement",
-        "Performance Improvement": "pr-performance",
-        # 'Backward Incompatible Change': doesn't match anything
-        "Build/Testing/Packaging Improvement": "pr-build",
-        "Non-significant (changelog entry is not needed)": "pr-non-significant",
-        "Non-significant (changelog entry is not required)": "pr-non-significant",
-        "Non-significant": "pr-non-significant",
-        "Documentation (changelog entry is not required)": "pr-documentation",
-        # 'Other': doesn't match anything
-    }
-
-    def __init__(self, pull_request):
-        self.label_name = str()
-        self.legal = False
-
-        self._parse(pull_request["bodyText"])
-
-    def _parse(self, text):
-        lines = text.splitlines()
-        next_category = False
-        category = str()
-
-        for line in lines:
-            stripped = line.strip()
-
-            if not stripped:
-                continue
-
-            if next_category:
-                category = stripped
-                next_category = False
-
-            if (
-                stripped
-                == "I hereby agree to the terms of the CLA available at: https://yandex.ru/legal/cla/?lang=en"
-            ):
-                self.legal = True
-
-            category_headers = (
-                "Category (leave one):",
-                "Changelog category (leave one):",
-                "Changelog category:",
-                "Category:",
-            )
-
-            if stripped in category_headers:
-                next_category = True
-
-        if category in Description.MAP_CATEGORY_TO_LABEL:
-            self.label_name = Description.MAP_CATEGORY_TO_LABEL[category]
-        else:
-            if not category:
-                print("Cannot find category in pr description")
-            else:
-                print(("Unknown category: " + category))

utils/github/query.py

@@ -1,492 +0,0 @@
-# -*- coding: utf-8 -*-
-
-import requests
-
-
-class Query:
-    """
-    Implements queries to the Github API using GraphQL
-    """
-
-    _PULL_REQUEST = """
-        author {{
-            ... on User {{
-                id
-                login
-            }}
-        }}
-
-        baseRepository {{
-            nameWithOwner
-        }}
-
-        mergeCommit {{
-            oid
-            parents(first: {min_page_size}) {{
-                totalCount
-                nodes {{
-                    oid
-                }}
-            }}
-        }}
-
-        mergedBy {{
-            ... on User {{
-                id
-                login
-            }}
-        }}
-
-        baseRefName
-        closed
-        headRefName
-        id
-        mergeable
-        merged
-        number
-        title
-        url
-    """
-
-    def __init__(self, token, owner, name, team, max_page_size=100, min_page_size=10):
-        self._PULL_REQUEST = Query._PULL_REQUEST.format(min_page_size=min_page_size)
-
-        self._token = token
-        self._owner = owner
-        self._name = name
-        self._team = team
-
-        self._max_page_size = max_page_size
-        self._min_page_size = min_page_size
-
-        self.api_costs = {}
-
-        repo = self.get_repository()
-        self._id = repo["id"]
-        self.ssh_url = repo["sshUrl"]
-        self.default_branch = repo["defaultBranchRef"]["name"]
-
-        self.members = set(self.get_members())
-
-    def get_repository(self):
-        _QUERY = """
-            repository(owner: "{owner}" name: "{name}") {{
-                defaultBranchRef {{
-                    name
-                }}
-                id
-                sshUrl
-            }}
-        """
-
-        query = _QUERY.format(owner=self._owner, name=self._name)
-        return self._run(query)["repository"]
-
-    def get_members(self):
-        """Get all team members for organization
-
-        Returns:
-            members: a map of members' logins to ids
-        """
-
-        _QUERY = """
-            organization(login: "{organization}") {{
-                team(slug: "{team}") {{
-                    members(first: {max_page_size} {next}) {{
-                        pageInfo {{
-                            hasNextPage
-                            endCursor
-                        }}
-                        nodes {{
-                            id
-                            login
-                        }}
-                    }}
-                }}
-            }}
-        """
-
-        members = {}
-        not_end = True
-        query = _QUERY.format(
-            organization=self._owner,
-            team=self._team,
-            max_page_size=self._max_page_size,
-            next="",
-        )
-
-        while not_end:
-            result = self._run(query)["organization"]["team"]
-            if result is None:
-                break
-            result = result["members"]
-            not_end = result["pageInfo"]["hasNextPage"]
-            query = _QUERY.format(
-                organization=self._owner,
-                team=self._team,
-                max_page_size=self._max_page_size,
-                next='after: "{}"'.format(result["pageInfo"]["endCursor"]),
-            )
-
-            members += dict([(node["login"], node["id"]) for node in result["nodes"]])
-
-        return members
-
-    def get_pull_request(self, number):
-        _QUERY = """
-            repository(owner: "{owner}" name: "{name}") {{
-                pullRequest(number: {number}) {{
-                    {pull_request_data}
-                }}
-            }}
-        """
-
-        query = _QUERY.format(
-            owner=self._owner,
-            name=self._name,
-            number=number,
-            pull_request_data=self._PULL_REQUEST,
-            min_page_size=self._min_page_size,
-        )
-        return self._run(query)["repository"]["pullRequest"]
-
-    def find_pull_request(self, base, head):
-        _QUERY = """
-            repository(owner: "{owner}" name: "{name}") {{
-                pullRequests(first: {min_page_size} baseRefName: "{base}" headRefName: "{head}") {{
-                    nodes {{
-                        {pull_request_data}
-                    }}
-                    totalCount
-                }}
-            }}
-        """
-
-        query = _QUERY.format(
-            owner=self._owner,
-            name=self._name,
-            base=base,
-            head=head,
-            pull_request_data=self._PULL_REQUEST,
-            min_page_size=self._min_page_size,
-        )
-        result = self._run(query)["repository"]["pullRequests"]
-        if result["totalCount"] > 0:
-            return result["nodes"][0]
-        else:
-            return {}
-
-    def find_pull_requests(self, label_name):
-        """
-        Get all pull-requests filtered by label name
-        """
-        _QUERY = """
-            repository(owner: "{owner}" name: "{name}") {{
-                pullRequests(first: {min_page_size} labels: "{label_name}" states: OPEN) {{
-                    nodes {{
-                        {pull_request_data}
-                    }}
-                }}
-            }}
-        """
-
-        query = _QUERY.format(
-            owner=self._owner,
-            name=self._name,
-            label_name=label_name,
-            pull_request_data=self._PULL_REQUEST,
-            min_page_size=self._min_page_size,
-        )
-        return self._run(query)["repository"]["pullRequests"]["nodes"]
-
-    def get_pull_requests(self, before_commit):
-        """
-        Get all merged pull-requests from the HEAD of default branch to the last commit (excluding)
-        """
-
-        _QUERY = """
-            repository(owner: "{owner}" name: "{name}") {{
-                defaultBranchRef {{
-                    target {{
-                        ... on Commit {{
-                            history(first: {max_page_size} {next}) {{
-                                pageInfo {{
-                                    hasNextPage
-                                    endCursor
-                                }}
-                                nodes {{
-                                    oid
-                                    associatedPullRequests(first: {min_page_size}) {{
-                                        totalCount
-                                        nodes {{
-                                            ... on PullRequest {{
-                                                {pull_request_data}
-
-                                                labels(first: {min_page_size}) {{
-                                                    totalCount
-                                                    pageInfo {{
-                                                        hasNextPage
-                                                        endCursor
-                                                    }}
-                                                    nodes {{
-                                                        name
-                                                        color
-                                                    }}
-                                                }}
-                                            }}
-                                        }}
-                                    }}
-                                }}
-                            }}
-                        }}
-                    }}
-                }}
-            }}
-        """
-
-        pull_requests = []
-        not_end = True
-        query = _QUERY.format(
-            owner=self._owner,
-            name=self._name,
-            max_page_size=self._max_page_size,
-            min_page_size=self._min_page_size,
-            pull_request_data=self._PULL_REQUEST,
-            next="",
-        )
-
-        while not_end:
-            result = self._run(query)["repository"]["defaultBranchRef"]["target"][
-                "history"
-            ]
-            not_end = result["pageInfo"]["hasNextPage"]
-            query = _QUERY.format(
-                owner=self._owner,
-                name=self._name,
-                max_page_size=self._max_page_size,
-                min_page_size=self._min_page_size,
-                pull_request_data=self._PULL_REQUEST,
-                next='after: "{}"'.format(result["pageInfo"]["endCursor"]),
-            )
-
-            for commit in result["nodes"]:
-                # FIXME: maybe include `before_commit`?
-                if str(commit["oid"]) == str(before_commit):
-                    not_end = False
-                    break
-
-                # TODO: fetch all pull-requests that were merged in a single commit.
-                assert (
-                    commit["associatedPullRequests"]["totalCount"]
-                    <= self._min_page_size
-                )
-
-                for pull_request in commit["associatedPullRequests"]["nodes"]:
-                    if (
-                        pull_request["baseRepository"]["nameWithOwner"]
-                        == "{}/{}".format(self._owner, self._name)
-                        and pull_request["baseRefName"] == self.default_branch
-                        and pull_request["mergeCommit"]["oid"] == commit["oid"]
-                    ):
-                        pull_requests.append(pull_request)
-
-        return pull_requests
-
-    def create_pull_request(
-        self, source, target, title, description="", draft=False, can_modify=True
-    ):
-        _QUERY = """
-            createPullRequest(input: {{
-                baseRefName: "{target}",
-                headRefName: "{source}",
-                repositoryId: "{id}",
-                title: "{title}",
-                body: "{body}",
-                draft: {draft},
-                maintainerCanModify: {modify}
-            }}) {{
-                pullRequest {{
-                    {pull_request_data}
-                }}
-            }}
-        """
-
-        query = _QUERY.format(
-            target=target,
-            source=source,
-            id=self._id,
-            title=title,
-            body=description,
-            draft="true" if draft else "false",
-            modify="true" if can_modify else "false",
-            pull_request_data=self._PULL_REQUEST,
-        )
-        return self._run(query, is_mutation=True)["createPullRequest"]["pullRequest"]
-
-    def merge_pull_request(self, id):
-        _QUERY = """
-            mergePullRequest(input: {{
-                pullRequestId: "{id}"
-            }}) {{
-                pullRequest {{
-                    {pull_request_data}
-                }}
-            }}
-        """
-
-        query = _QUERY.format(id=id, pull_request_data=self._PULL_REQUEST)
-        return self._run(query, is_mutation=True)["mergePullRequest"]["pullRequest"]
-
-    # FIXME: figure out how to add more assignees at once
-    def add_assignee(self, pr, assignee):
-        _QUERY = """
-            addAssigneesToAssignable(input: {{
-                assignableId: "{id1}",
-                assigneeIds: "{id2}"
-            }}) {{
-                clientMutationId
-            }}
-        """
-
-        query = _QUERY.format(id1=pr["id"], id2=assignee["id"])
-        self._run(query, is_mutation=True)
-
-    def set_label(self, pull_request, label_name):
-        """
-        Set label by name to the pull request
-
-        Args:
-            pull_request: JSON object returned by `get_pull_requests()`
-            label_name (string): label name
-        """
-
-        _GET_LABEL = """
-            repository(owner: "{owner}" name: "{name}") {{
-                labels(first: {max_page_size} {next} query: "{label_name}") {{
-                    pageInfo {{
-                        hasNextPage
-                        endCursor
-                    }}
-                    nodes {{
-                        id
-                        name
-                        color
-                    }}
-                }}
-            }}
-        """
-
-        _SET_LABEL = """
-            addLabelsToLabelable(input: {{
-                labelableId: "{pr_id}",
-                labelIds: "{label_id}"
-            }}) {{
-                clientMutationId
-            }}
-        """
-
-        labels = []
-        not_end = True
-        query = _GET_LABEL.format(
-            owner=self._owner,
-            name=self._name,
-            label_name=label_name,
-            max_page_size=self._max_page_size,
-            next="",
-        )
-
-        while not_end:
-            result = self._run(query)["repository"]["labels"]
-            not_end = result["pageInfo"]["hasNextPage"]
-            query = _GET_LABEL.format(
-                owner=self._owner,
-                name=self._name,
-                label_name=label_name,
-                max_page_size=self._max_page_size,
-                next='after: "{}"'.format(result["pageInfo"]["endCursor"]),
-            )
-
-            labels += [label for label in result["nodes"]]
-
-        if not labels:
-            return
-
-        query = _SET_LABEL.format(pr_id=pull_request["id"], label_id=labels[0]["id"])
-        self._run(query, is_mutation=True)
-
-    def _run(self, query, is_mutation=False):
-        from requests.adapters import HTTPAdapter
-        from urllib3.util.retry import Retry
-
-        def requests_retry_session(
-            retries=3,
-            backoff_factor=0.3,
-            status_forcelist=(500, 502, 504),
-            session=None,
-        ):
-            session = session or requests.Session()
-            retry = Retry(
-                total=retries,
-                read=retries,
-                connect=retries,
-                backoff_factor=backoff_factor,
-                status_forcelist=status_forcelist,
-            )
-            adapter = HTTPAdapter(max_retries=retry)
-            session.mount("http://", adapter)
-            session.mount("https://", adapter)
-            return session
-
-        headers = {"Authorization": "bearer {}".format(self._token)}
-        if is_mutation:
-            query = """
-            mutation {{
-                {query}
-            }}
-            """.format(
-                query=query
-            )
-        else:
-            query = """
-            query {{
-                {query}
-                rateLimit {{
-                    cost
-                    remaining
-                }}
-            }}
-            """.format(
-                query=query
-            )
-
-        while True:
-            request = requests_retry_session().post(
-                "https://api.github.com/graphql", json={"query": query}, headers=headers
-            )
-            if request.status_code == 200:
-                result = request.json()
-                if "errors" in result:
-                    raise Exception(
-                        "Errors occurred: {}\nOriginal query: {}".format(
-                            result["errors"], query
-                        )
-                    )
-
-                if not is_mutation:
-                    import inspect
-
-                    caller = inspect.getouterframes(inspect.currentframe(), 2)[1][3]
-                    if caller not in list(self.api_costs.keys()):
-                        self.api_costs[caller] = 0
-                    self.api_costs[caller] += result["data"]["rateLimit"]["cost"]
-
-                return result["data"]
-            else:
-                import json
-
-                raise Exception(
-                    "Query failed with code {code}:\n{json}".format(
-                        code=request.status_code,
-                        json=json.dumps(request.json(), indent=4),
-                    )
-                )