thevar1able/ClickHouse
1
0
Fork 0
mirror of https://github.com/ClickHouse/ClickHouse.git synced 2024-11-29 19:12:03 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge branch 'master' of https://github.com/yandex/ClickHouse

Browse Source
This commit is contained in:
Odin Hultgren Van Der Horst 2019-01-15 08:13:51 +01:00
commit 8bff8d4e2d
155 changed files with 2126 additions and 1452 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
.gitignore vendored
View File

@ -251,3 +251,6 @@ website/package-lock.json
# cquery cache
/.cquery-cache
# ccls cache
/.ccls-cache

3
.gitmodules vendored
View File

@ -61,3 +61,6 @@
[submodule "contrib/libgsasl"]
path = contrib/libgsasl
url = https://github.com/ClickHouse-Extras/libgsasl.git
[submodule "contrib/cppkafka"]
path = contrib/cppkafka
url = https://github.com/mfontanini/cppkafka.git

15
CMakeLists.txt
View File

@ -90,8 +90,6 @@ if (GLIBC_COMPATIBILITY)
set (USE_INTERNAL_MEMCPY ON)
endif ()
set (COMPILER_FLAGS "${COMPILER_FLAGS}")
string(REGEX MATCH "-?[0-9]+(.[0-9]+)?$" COMPILER_POSTFIX ${CMAKE_CXX_COMPILER})
find_program (LLD_PATH NAMES "lld${COMPILER_POSTFIX}" "lld")
@ -108,10 +106,15 @@ if (LINKER_NAME)
set (CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -fuse-ld=${LINKER_NAME}")
endif ()
option (PIPE "-pipe compiler option [less /tmp usage, more ram usage]" ON)
if (PIPE)
set (COMPILER_FLAGS "${COMPILER_FLAGS} -pipe")
endif ()
cmake_host_system_information(RESULT AVAILABLE_PHYSICAL_MEMORY QUERY AVAILABLE_PHYSICAL_MEMORY) # Not available under freebsd
if(NOT AVAILABLE_PHYSICAL_MEMORY OR AVAILABLE_PHYSICAL_MEMORY GREATER 8000)
option(COMPILER_PIPE "-pipe compiler option [less /tmp usage, more ram usage]" ON)
endif()
if(COMPILER_PIPE)
set(COMPILER_FLAGS "${COMPILER_FLAGS} -pipe")
else()
message(STATUS "Disabling compiler -pipe option (have only ${AVAILABLE_PHYSICAL_MEMORY} mb of memory)")
endif()
include (cmake/test_cpu.cmake)

5
cmake/find_rdkafka.cmake
View File

@ -20,11 +20,13 @@ if (NOT USE_INTERNAL_RDKAFKA_LIBRARY)
if (USE_STATIC_LIBRARIES AND NOT OS_FREEBSD)
find_library (SASL2_LIBRARY sasl2)
endif ()
set (CPPKAFKA_LIBRARY cppkafka) # TODO: try to use unbundled version.
endif ()
if (RDKAFKA_LIB AND RDKAFKA_INCLUDE_DIR)
set (USE_RDKAFKA 1)
set (RDKAFKA_LIBRARY ${RDKAFKA_LIB} ${OPENSSL_LIBRARIES})
set (CPPKAFKA_LIBRARY cppkafka)
if (SASL2_LIBRARY)
list (APPEND RDKAFKA_LIBRARY ${SASL2_LIBRARY})
endif ()
@ -35,9 +37,10 @@ elseif (NOT MISSING_INTERNAL_RDKAFKA_LIBRARY AND NOT ARCH_ARM)
set (USE_INTERNAL_RDKAFKA_LIBRARY 1)
set (RDKAFKA_INCLUDE_DIR "${ClickHouse_SOURCE_DIR}/contrib/librdkafka/src")
set (RDKAFKA_LIBRARY rdkafka)
set (CPPKAFKA_LIBRARY cppkafka)
set (USE_RDKAFKA 1)
endif ()
endif ()
message (STATUS "Using librdkafka=${USE_RDKAFKA}: ${RDKAFKA_INCLUDE_DIR} : ${RDKAFKA_LIBRARY}")
message (STATUS "Using librdkafka=${USE_RDKAFKA}: ${RDKAFKA_INCLUDE_DIR} : ${RDKAFKA_LIBRARY} ${CPPKAFKA_LIBRARY}")

2
cmake/find_ssl.cmake
View File

@ -6,7 +6,7 @@ set (OPENSSL_USE_STATIC_LIBS ${USE_STATIC_LIBRARIES})
if (NOT USE_INTERNAL_SSL_LIBRARY)
if (APPLE)
set (OPENSSL_ROOT_DIR "/usr/local/opt/openssl")
set (OPENSSL_ROOT_DIR "/usr/local/opt/openssl" CACHE INTERNAL "")
# https://rt.openssl.org/Ticket/Display.html?user=guest&pass=guest&id=2232
if (USE_STATIC_LIBRARIES)
message(WARNING "Disable USE_STATIC_LIBRARIES if you have linking problems with OpenSSL on MacOS")

12
cmake/limit_jobs.cmake
View File

@ -4,6 +4,7 @@
# include (cmake/limit_jobs.cmake)
cmake_host_system_information(RESULT AVAILABLE_PHYSICAL_MEMORY QUERY AVAILABLE_PHYSICAL_MEMORY) # Not available under freebsd
cmake_host_system_information(RESULT NUMBER_OF_LOGICAL_CORES QUERY NUMBER_OF_LOGICAL_CORES)
option(PARALLEL_COMPILE_JOBS "Define the maximum number of concurrent compilation jobs" "")
if (NOT PARALLEL_COMPILE_JOBS AND AVAILABLE_PHYSICAL_MEMORY AND MAX_COMPILER_MEMORY)
@ -12,7 +13,7 @@ if (NOT PARALLEL_COMPILE_JOBS AND AVAILABLE_PHYSICAL_MEMORY AND MAX_COMPILER_MEM
set (PARALLEL_COMPILE_JOBS 1)
endif ()
endif ()
if (PARALLEL_COMPILE_JOBS)
if (PARALLEL_COMPILE_JOBS AND (NOT NUMBER_OF_LOGICAL_CORES OR PARALLEL_COMPILE_JOBS LESS NUMBER_OF_LOGICAL_CORES))
set(CMAKE_JOB_POOL_COMPILE compile_job_pool${CMAKE_CURRENT_SOURCE_DIR})
string (REGEX REPLACE "[^a-zA-Z0-9]+" "_" CMAKE_JOB_POOL_COMPILE ${CMAKE_JOB_POOL_COMPILE})
set_property(GLOBAL APPEND PROPERTY JOB_POOLS ${CMAKE_JOB_POOL_COMPILE}=${PARALLEL_COMPILE_JOBS})
@ -25,13 +26,12 @@ if (NOT PARALLEL_LINK_JOBS AND AVAILABLE_PHYSICAL_MEMORY AND MAX_LINKER_MEMORY)
set (PARALLEL_LINK_JOBS 1)
endif ()
endif ()
if (PARALLEL_COMPILE_JOBS OR PARALLEL_LINK_JOBS)
message(STATUS "${CMAKE_CURRENT_SOURCE_DIR}: Have ${AVAILABLE_PHYSICAL_MEMORY} megabytes of memory. Limiting concurrent linkers jobs to ${PARALLEL_LINK_JOBS} and compiler jobs to ${PARALLEL_COMPILE_JOBS}")
endif ()
if (LLVM_PARALLEL_LINK_JOBS)
if (PARALLEL_LINK_JOBS AND (NOT NUMBER_OF_LOGICAL_CORES OR PARALLEL_COMPILE_JOBS LESS NUMBER_OF_LOGICAL_CORES))
set(CMAKE_JOB_POOL_LINK link_job_pool${CMAKE_CURRENT_SOURCE_DIR})
string (REGEX REPLACE "[^a-zA-Z0-9]+" "_" CMAKE_JOB_POOL_LINK ${CMAKE_JOB_POOL_LINK})
set_property(GLOBAL APPEND PROPERTY JOB_POOLS ${CMAKE_JOB_POOL_LINK}=${PARALLEL_LINK_JOBS})
endif ()
if (PARALLEL_COMPILE_JOBS OR PARALLEL_LINK_JOBS)
message(STATUS "${CMAKE_CURRENT_SOURCE_DIR}: Have ${AVAILABLE_PHYSICAL_MEMORY} megabytes of memory. Limiting concurrent linkers jobs to ${PARALLEL_LINK_JOBS} and compiler jobs to ${PARALLEL_COMPILE_JOBS}")
endif ()

5
contrib/CMakeLists.txt vendored
View File

@ -125,6 +125,10 @@ if (USE_INTERNAL_RDKAFKA_LIBRARY)
target_include_directories(rdkafka BEFORE PRIVATE ${OPENSSL_INCLUDE_DIR})
endif ()
if (USE_RDKAFKA)
add_subdirectory (cppkafka-cmake)
endif()
if (ENABLE_ODBC AND USE_INTERNAL_ODBC_LIBRARY)
add_subdirectory (unixodbc-cmake)
endif ()
@ -139,6 +143,7 @@ if (USE_INTERNAL_CAPNP_LIBRARY)
endif ()
if (USE_INTERNAL_POCO_LIBRARY)
set (POCO_VERBOSE_MESSAGES 0 CACHE INTERNAL "")
set (save_CMAKE_CXX_FLAGS ${CMAKE_CXX_FLAGS})
set (save_CMAKE_C_FLAGS ${CMAKE_C_FLAGS})
set (_save ${ENABLE_TESTS})

1
contrib/cppkafka vendored Submodule

@ -0,0 +1 @@
Subproject commit 520465510efef7704346cf8d140967c4abb057c1

31
contrib/cppkafka-cmake/CMakeLists.txt Normal file
View File

@ -0,0 +1,31 @@
set(CPPKAFKA_DIR ${CMAKE_SOURCE_DIR}/contrib/cppkafka)
set(SRCS
${CPPKAFKA_DIR}/src/configuration.cpp
${CPPKAFKA_DIR}/src/topic_configuration.cpp
${CPPKAFKA_DIR}/src/configuration_option.cpp
${CPPKAFKA_DIR}/src/exceptions.cpp
${CPPKAFKA_DIR}/src/topic.cpp
${CPPKAFKA_DIR}/src/buffer.cpp
${CPPKAFKA_DIR}/src/queue.cpp
${CPPKAFKA_DIR}/src/message.cpp
${CPPKAFKA_DIR}/src/message_timestamp.cpp
${CPPKAFKA_DIR}/src/message_internal.cpp
${CPPKAFKA_DIR}/src/topic_partition.cpp
${CPPKAFKA_DIR}/src/topic_partition_list.cpp
${CPPKAFKA_DIR}/src/metadata.cpp
${CPPKAFKA_DIR}/src/group_information.cpp
${CPPKAFKA_DIR}/src/error.cpp
${CPPKAFKA_DIR}/src/event.cpp
${CPPKAFKA_DIR}/src/kafka_handle_base.cpp
${CPPKAFKA_DIR}/src/producer.cpp
${CPPKAFKA_DIR}/src/consumer.cpp
)
add_library(cppkafka ${LINK_MODE} ${SRCS})
target_link_libraries(cppkafka PRIVATE ${RDKAFKA_LIBRARY})
target_include_directories(cppkafka PRIVATE ${CPPKAFKA_DIR}/include/cppkafka)
target_include_directories(cppkafka PRIVATE ${Boost_INCLUDE_DIRS})
target_include_directories(cppkafka SYSTEM PUBLIC ${CPPKAFKA_DIR}/include)

2
contrib/jemalloc vendored

@ -1 +1 @@
Subproject commit cd2931ad9bbd78208565716ab102e86d858c2fff
Subproject commit 41b7372eadee941b9164751b8d4963f915d3ceae

2
contrib/librdkafka vendored

@ -1 +1 @@
Subproject commit 7478b5ef16aadd6543fe38bc6a2deb895c70da98
Subproject commit 363dcad5a23dc29381cc626620e68ae418b3af19

107
contrib/librdkafka-cmake/CMakeLists.txt
View File

@ -1,60 +1,63 @@
set(RDKAFKA_SOURCE_DIR ${CMAKE_SOURCE_DIR}/contrib/librdkafka/src)
set(SRCS
${RDKAFKA_SOURCE_DIR}/crc32c.c
${RDKAFKA_SOURCE_DIR}/rdaddr.c
${RDKAFKA_SOURCE_DIR}/rdavl.c
${RDKAFKA_SOURCE_DIR}/rdbuf.c
${RDKAFKA_SOURCE_DIR}/rdcrc32.c
${RDKAFKA_SOURCE_DIR}/rdkafka.c
${RDKAFKA_SOURCE_DIR}/rdkafka_assignor.c
${RDKAFKA_SOURCE_DIR}/rdkafka_broker.c
${RDKAFKA_SOURCE_DIR}/rdkafka_buf.c
${RDKAFKA_SOURCE_DIR}/rdkafka_cgrp.c
${RDKAFKA_SOURCE_DIR}/rdkafka_conf.c
${RDKAFKA_SOURCE_DIR}/rdkafka_event.c
${RDKAFKA_SOURCE_DIR}/rdkafka_feature.c
${RDKAFKA_SOURCE_DIR}/rdkafka_lz4.c
${RDKAFKA_SOURCE_DIR}/rdkafka_metadata.c
${RDKAFKA_SOURCE_DIR}/rdkafka_metadata_cache.c
${RDKAFKA_SOURCE_DIR}/rdkafka_msg.c
${RDKAFKA_SOURCE_DIR}/rdkafka_msgset_reader.c
${RDKAFKA_SOURCE_DIR}/rdkafka_msgset_writer.c
${RDKAFKA_SOURCE_DIR}/rdkafka_offset.c
${RDKAFKA_SOURCE_DIR}/rdkafka_op.c
${RDKAFKA_SOURCE_DIR}/rdkafka_partition.c
${RDKAFKA_SOURCE_DIR}/rdkafka_pattern.c
${RDKAFKA_SOURCE_DIR}/rdkafka_queue.c
${RDKAFKA_SOURCE_DIR}/rdkafka_range_assignor.c
${RDKAFKA_SOURCE_DIR}/rdkafka_request.c
${RDKAFKA_SOURCE_DIR}/rdkafka_roundrobin_assignor.c
${RDKAFKA_SOURCE_DIR}/rdkafka_sasl.c
${RDKAFKA_SOURCE_DIR}/rdkafka_sasl_plain.c
${RDKAFKA_SOURCE_DIR}/rdkafka_subscription.c
${RDKAFKA_SOURCE_DIR}/rdkafka_timer.c
${RDKAFKA_SOURCE_DIR}/rdkafka_topic.c
${RDKAFKA_SOURCE_DIR}/rdkafka_transport.c
${RDKAFKA_SOURCE_DIR}/rdkafka_interceptor.c
${RDKAFKA_SOURCE_DIR}/rdkafka_header.c
${RDKAFKA_SOURCE_DIR}/rdlist.c
${RDKAFKA_SOURCE_DIR}/rdlog.c
${RDKAFKA_SOURCE_DIR}/rdmurmur2.c
${RDKAFKA_SOURCE_DIR}/rdports.c
${RDKAFKA_SOURCE_DIR}/rdrand.c
${RDKAFKA_SOURCE_DIR}/rdregex.c
${RDKAFKA_SOURCE_DIR}/rdstring.c
${RDKAFKA_SOURCE_DIR}/rdunittest.c
${RDKAFKA_SOURCE_DIR}/rdvarint.c
${RDKAFKA_SOURCE_DIR}/snappy.c
${RDKAFKA_SOURCE_DIR}/tinycthread.c
${RDKAFKA_SOURCE_DIR}/xxhash.c
${RDKAFKA_SOURCE_DIR}/lz4.c
${RDKAFKA_SOURCE_DIR}/lz4frame.c
${RDKAFKA_SOURCE_DIR}/lz4hc.c
${RDKAFKA_SOURCE_DIR}/rdgz.c
${RDKAFKA_SOURCE_DIR}/crc32c.c
${RDKAFKA_SOURCE_DIR}/rdaddr.c
${RDKAFKA_SOURCE_DIR}/rdavl.c
${RDKAFKA_SOURCE_DIR}/rdbuf.c
${RDKAFKA_SOURCE_DIR}/rdcrc32.c
${RDKAFKA_SOURCE_DIR}/rdkafka.c
${RDKAFKA_SOURCE_DIR}/rdkafka_assignor.c
${RDKAFKA_SOURCE_DIR}/rdkafka_background.c
${RDKAFKA_SOURCE_DIR}/rdkafka_broker.c
${RDKAFKA_SOURCE_DIR}/rdkafka_buf.c
${RDKAFKA_SOURCE_DIR}/rdkafka_cgrp.c
${RDKAFKA_SOURCE_DIR}/rdkafka_conf.c
${RDKAFKA_SOURCE_DIR}/rdkafka_event.c
${RDKAFKA_SOURCE_DIR}/rdkafka_feature.c
${RDKAFKA_SOURCE_DIR}/rdkafka_idempotence.c
${RDKAFKA_SOURCE_DIR}/rdkafka_lz4.c
${RDKAFKA_SOURCE_DIR}/rdkafka_metadata.c
${RDKAFKA_SOURCE_DIR}/rdkafka_metadata_cache.c
${RDKAFKA_SOURCE_DIR}/rdkafka_msg.c
${RDKAFKA_SOURCE_DIR}/rdkafka_msgset_reader.c
${RDKAFKA_SOURCE_DIR}/rdkafka_msgset_writer.c
${RDKAFKA_SOURCE_DIR}/rdkafka_offset.c
${RDKAFKA_SOURCE_DIR}/rdkafka_op.c
${RDKAFKA_SOURCE_DIR}/rdkafka_partition.c
${RDKAFKA_SOURCE_DIR}/rdkafka_pattern.c
${RDKAFKA_SOURCE_DIR}/rdkafka_queue.c
${RDKAFKA_SOURCE_DIR}/rdkafka_range_assignor.c
${RDKAFKA_SOURCE_DIR}/rdkafka_request.c
${RDKAFKA_SOURCE_DIR}/rdkafka_roundrobin_assignor.c
${RDKAFKA_SOURCE_DIR}/rdkafka_sasl.c
${RDKAFKA_SOURCE_DIR}/rdkafka_sasl_plain.c
${RDKAFKA_SOURCE_DIR}/rdkafka_subscription.c
${RDKAFKA_SOURCE_DIR}/rdkafka_timer.c
${RDKAFKA_SOURCE_DIR}/rdkafka_topic.c
${RDKAFKA_SOURCE_DIR}/rdkafka_transport.c
${RDKAFKA_SOURCE_DIR}/rdkafka_interceptor.c
${RDKAFKA_SOURCE_DIR}/rdkafka_header.c
${RDKAFKA_SOURCE_DIR}/rdlist.c
${RDKAFKA_SOURCE_DIR}/rdlog.c
${RDKAFKA_SOURCE_DIR}/rdmurmur2.c
${RDKAFKA_SOURCE_DIR}/rdports.c
${RDKAFKA_SOURCE_DIR}/rdrand.c
${RDKAFKA_SOURCE_DIR}/rdregex.c
${RDKAFKA_SOURCE_DIR}/rdstring.c
${RDKAFKA_SOURCE_DIR}/rdunittest.c
${RDKAFKA_SOURCE_DIR}/rdvarint.c
${RDKAFKA_SOURCE_DIR}/snappy.c
${RDKAFKA_SOURCE_DIR}/tinycthread.c
${RDKAFKA_SOURCE_DIR}/tinycthread_extra.c
${RDKAFKA_SOURCE_DIR}/xxhash.c
${RDKAFKA_SOURCE_DIR}/lz4.c
${RDKAFKA_SOURCE_DIR}/lz4frame.c
${RDKAFKA_SOURCE_DIR}/lz4hc.c
${RDKAFKA_SOURCE_DIR}/rdgz.c
)
add_library(rdkafka ${LINK_MODE} ${SRCS})
target_include_directories(rdkafka PRIVATE include)
target_include_directories(rdkafka SYSTEM PUBLIC include)
target_include_directories(rdkafka SYSTEM PUBLIC ${RDKAFKA_SOURCE_DIR})
target_link_libraries(rdkafka PUBLIC ${ZLIB_LIBRARIES} ${OPENSSL_SSL_LIBRARY} ${OPENSSL_CRYPTO_LIBRARY})

4
contrib/librdkafka-cmake/config.h
View File

@ -71,4 +71,8 @@
#define HAVE_PTHREAD_SETNAME_GNU 1
// python
//#define HAVE_PYTHON 1
// C11 threads
#if (__STDC_VERSION__ >= 201112L) && !defined(__STDC_NO_THREADS__)
# define WITH_C11THREADS 1
#endif
#endif /* _CONFIG_H_ */

5
contrib/librdkafka-cmake/include/librdkafka/rdkafka.h Normal file
View File

@ -0,0 +1,5 @@
#if __has_include(<rdkafka.h>) // maybe bundled
# include_next <rdkafka.h> // Y_IGNORE
#else // system
# include_next <librdkafka/rdkafka.h>
#endif

19
dbms/CMakeLists.txt
View File

@ -2,13 +2,17 @@ if (USE_INCLUDE_WHAT_YOU_USE)
set (CMAKE_CXX_INCLUDE_WHAT_YOU_USE ${IWYU_PATH})
endif ()
set (MAX_COMPILER_MEMORY 2500 CACHE INTERNAL "")
if (MAKE_STATIC_LIBRARIES)
set (MAX_LINKER_MEMORY 3500 CACHE INTERNAL "")
if(COMPILER_PIPE)
set(MAX_COMPILER_MEMORY 2500)
else()
set (MAX_LINKER_MEMORY 2500 CACHE INTERNAL "")
endif ()
include (../cmake/limit_jobs.cmake)
set(MAX_COMPILER_MEMORY 1500)
endif()
if(MAKE_STATIC_LIBRARIES)
set(MAX_LINKER_MEMORY 3500)
else()
set(MAX_LINKER_MEMORY 2500)
endif()
include(../cmake/limit_jobs.cmake)
include(cmake/find_vectorclass.cmake)
@ -16,7 +20,7 @@ set (CONFIG_VERSION ${CMAKE_CURRENT_BINARY_DIR}/src/Common/config_version.h)
set (CONFIG_COMMON ${CMAKE_CURRENT_BINARY_DIR}/src/Common/config.h)
include (cmake/version.cmake)
message (STATUS "Will build ${VERSION_FULL}")
message (STATUS "Will build ${VERSION_FULL} revision ${VERSION_REVISION}")
configure_file (src/Common/config.h.in ${CONFIG_COMMON})
configure_file (src/Common/config_version.h.in ${CONFIG_VERSION})
@ -283,6 +287,7 @@ endif ()
if (USE_RDKAFKA)
target_link_libraries (dbms PRIVATE ${RDKAFKA_LIBRARY})
target_link_libraries (dbms PRIVATE ${CPPKAFKA_LIBRARY})
if (NOT USE_INTERNAL_RDKAFKA_LIBRARY)
target_include_directories (dbms SYSTEM BEFORE PRIVATE ${RDKAFKA_INCLUDE_DIR})
endif ()

20
dbms/cmake/version.cmake
View File

@ -1,11 +1,11 @@
# This strings autochanged from release_lib.sh:
set(VERSION_REVISION 54413 CACHE STRING "") # changed manually for tests
set(VERSION_MAJOR 19 CACHE STRING "")
set(VERSION_MINOR 1 CACHE STRING "")
set(VERSION_PATCH 0 CACHE STRING "")
set(VERSION_GITHASH 014e344a36bc19a58621e0add379984cf62b9067 CACHE STRING "")
set(VERSION_DESCRIBE v19.1.0-testing CACHE STRING "")
set(VERSION_STRING 19.1.0 CACHE STRING "")
set(VERSION_REVISION 54413)
set(VERSION_MAJOR 19)
set(VERSION_MINOR 1)
set(VERSION_PATCH 0)
set(VERSION_GITHASH 014e344a36bc19a58621e0add379984cf62b9067)
set(VERSION_DESCRIBE v19.1.0-testing)
set(VERSION_STRING 19.1.0)
# end of autochange
set(VERSION_EXTRA "" CACHE STRING "")
@ -19,8 +19,8 @@ if (VERSION_EXTRA)
string(CONCAT VERSION_STRING ${VERSION_STRING} "." ${VERSION_EXTRA})
endif ()
set (VERSION_NAME "${PROJECT_NAME}" CACHE STRING "")
set (VERSION_FULL "${VERSION_NAME} ${VERSION_STRING}" CACHE STRING "")
set (VERSION_SO "${VERSION_STRING}" CACHE STRING "")
set (VERSION_NAME "${PROJECT_NAME}")
set (VERSION_FULL "${VERSION_NAME} ${VERSION_STRING}")
set (VERSION_SO "${VERSION_STRING}")
math (EXPR VERSION_INTEGER "${VERSION_PATCH} + ${VERSION_MINOR}*1000 + ${VERSION_MAJOR}*1000000")

2
dbms/programs/client/CMakeLists.txt
View File

@ -1,5 +1,5 @@
add_library (clickhouse-client-lib ${LINK_MODE} Client.cpp)
target_link_libraries (clickhouse-client-lib PRIVATE clickhouse_common_io clickhouse_functions clickhouse_aggregate_functions ${LINE_EDITING_LIBS} ${Boost_PROGRAM_OPTIONS_LIBRARY})
target_link_libraries (clickhouse-client-lib PRIVATE clickhouse_common_config clickhouse_functions clickhouse_aggregate_functions clickhouse_common_io ${LINE_EDITING_LIBS} ${Boost_PROGRAM_OPTIONS_LIBRARY})
if (READLINE_INCLUDE_DIR)
target_include_directories (clickhouse-client-lib SYSTEM PRIVATE ${READLINE_INCLUDE_DIR})
endif ()

62
dbms/programs/copier/ClusterCopier.cpp
View File

@ -243,7 +243,7 @@ struct ClusterPartition
UInt64 rows_copied = 0;
UInt64 blocks_copied = 0;
size_t total_tries = 0;
UInt64 total_tries = 0;
};
@ -340,7 +340,7 @@ struct TaskCluster
String default_local_database;
/// Limits number of simultaneous workers
size_t max_workers = 0;
UInt64 max_workers = 0;
/// Base settings for pull and push
Settings settings_common;
@ -773,11 +773,11 @@ public:
}
template <typename T>
decltype(auto) retry(T && func, size_t max_tries = 100)
decltype(auto) retry(T && func, UInt64 max_tries = 100)
{
std::exception_ptr exception;
for (size_t try_number = 1; try_number <= max_tries; ++try_number)
for (UInt64 try_number = 1; try_number <= max_tries; ++try_number)
{
try
{
@ -880,7 +880,7 @@ public:
}
/// Compute set of partitions, assume set of partitions aren't changed during the processing
void discoverTablePartitions(TaskTable & task_table, size_t num_threads = 0)
void discoverTablePartitions(TaskTable & task_table, UInt64 num_threads = 0)
{
/// Fetch partitions list from a shard
{
@ -985,7 +985,7 @@ public:
/// Retry table processing
bool table_is_done = false;
for (size_t num_table_tries = 0; num_table_tries < max_table_tries; ++num_table_tries)
for (UInt64 num_table_tries = 0; num_table_tries < max_table_tries; ++num_table_tries)
{
if (tryProcessTable(task_table))
{
@ -1044,7 +1044,7 @@ protected:
String workers_path = getWorkersPath();
String current_worker_path = getCurrentWorkerNodePath();
size_t num_bad_version_errors = 0;
UInt64 num_bad_version_errors = 0;
while (true)
{
@ -1055,7 +1055,7 @@ protected:
auto version = stat.version;
zookeeper->get(workers_path, &stat);
if (static_cast<size_t>(stat.numChildren) >= task_cluster->max_workers)
if (static_cast<UInt64>(stat.numChildren) >= task_cluster->max_workers)
{
LOG_DEBUG(log, "Too many workers (" << stat.numChildren << ", maximum " << task_cluster->max_workers << ")"
<< ". Postpone processing " << description);
@ -1163,7 +1163,7 @@ protected:
}
// If all task is finished and zxid is not changed then partition could not become dirty again
for (size_t shard_num = 0; shard_num < status_paths.size(); ++shard_num)
for (UInt64 shard_num = 0; shard_num < status_paths.size(); ++shard_num)
{
if (zxid1[shard_num] != zxid2[shard_num])
{
@ -1280,7 +1280,7 @@ protected:
LOG_DEBUG(log, "Execute distributed DROP PARTITION: " << query);
/// Limit number of max executing replicas to 1
size_t num_shards = executeQueryOnCluster(cluster_push, query, nullptr, &settings_push, PoolMode::GET_ONE, 1);
UInt64 num_shards = executeQueryOnCluster(cluster_push, query, nullptr, &settings_push, PoolMode::GET_ONE, 1);
if (num_shards < cluster_push->getShardCount())
{
@ -1299,8 +1299,8 @@ protected:
}
static constexpr size_t max_table_tries = 1000;
static constexpr size_t max_shard_partition_tries = 600;
static constexpr UInt64 max_table_tries = 1000;
static constexpr UInt64 max_shard_partition_tries = 600;
bool tryProcessTable(TaskTable & task_table)
{
@ -1317,7 +1317,7 @@ protected:
Stopwatch watch;
TasksShard expected_shards;
size_t num_failed_shards = 0;
UInt64 num_failed_shards = 0;
++cluster_partition.total_tries;
@ -1368,7 +1368,7 @@ protected:
bool is_unprioritized_task = !previous_shard_is_instantly_finished && shard->priority.is_remote;
PartitionTaskStatus task_status = PartitionTaskStatus::Error;
bool was_error = false;
for (size_t try_num = 0; try_num < max_shard_partition_tries; ++try_num)
for (UInt64 try_num = 0; try_num < max_shard_partition_tries; ++try_num)
{
task_status = tryProcessPartitionTask(partition, is_unprioritized_task);
@ -1434,8 +1434,8 @@ protected:
}
}
size_t required_partitions = task_table.cluster_partitions.size();
size_t finished_partitions = task_table.finished_cluster_partitions.size();
UInt64 required_partitions = task_table.cluster_partitions.size();
UInt64 finished_partitions = task_table.finished_cluster_partitions.size();
bool table_is_done = finished_partitions >= required_partitions;
if (!table_is_done)
@ -1645,7 +1645,7 @@ protected:
String query = queryToString(create_query_push_ast);
LOG_DEBUG(log, "Create destination tables. Query: " << query);
size_t shards = executeQueryOnCluster(task_table.cluster_push, query, create_query_push_ast, &task_cluster->settings_push,
UInt64 shards = executeQueryOnCluster(task_table.cluster_push, query, create_query_push_ast, &task_cluster->settings_push,
PoolMode::GET_MANY);
LOG_DEBUG(log, "Destination tables " << getDatabaseDotTable(task_table.table_push) << " have been created on " << shards
<< " shards of " << task_table.cluster_push->getShardCount());
@ -1699,7 +1699,7 @@ protected:
std::future<Coordination::ExistsResponse> future_is_dirty_checker;
Stopwatch watch(CLOCK_MONOTONIC_COARSE);
constexpr size_t check_period_milliseconds = 500;
constexpr UInt64 check_period_milliseconds = 500;
/// Will asynchronously check that ZooKeeper connection and is_dirty flag appearing while copy data
auto cancel_check = [&] ()
@ -1917,16 +1917,16 @@ protected:
/** Executes simple query (without output streams, for example DDL queries) on each shard of the cluster
* Returns number of shards for which at least one replica executed query successfully
*/
size_t executeQueryOnCluster(
UInt64 executeQueryOnCluster(
const ClusterPtr & cluster,
const String & query,
const ASTPtr & query_ast_ = nullptr,
const Settings * settings = nullptr,
PoolMode pool_mode = PoolMode::GET_ALL,
size_t max_successful_executions_per_shard = 0) const
UInt64 max_successful_executions_per_shard = 0) const
{
auto num_shards = cluster->getShardsInfo().size();
std::vector<size_t> per_shard_num_successful_replicas(num_shards, 0);
std::vector<UInt64> per_shard_num_successful_replicas(num_shards, 0);
ASTPtr query_ast;
if (query_ast_ == nullptr)
@ -1939,10 +1939,10 @@ protected:
/// We need to execute query on one replica at least
auto do_for_shard = [&] (size_t shard_index)
auto do_for_shard = [&] (UInt64 shard_index)
{
const Cluster::ShardInfo & shard = cluster->getShardsInfo().at(shard_index);
size_t & num_successful_executions = per_shard_num_successful_replicas.at(shard_index);
UInt64 & num_successful_executions = per_shard_num_successful_replicas.at(shard_index);
num_successful_executions = 0;
auto increment_and_check_exit = [&] ()
@ -1951,12 +1951,12 @@ protected:
return max_successful_executions_per_shard && num_successful_executions >= max_successful_executions_per_shard;
};
size_t num_replicas = cluster->getShardsAddresses().at(shard_index).size();
size_t num_local_replicas = shard.getLocalNodeCount();
size_t num_remote_replicas = num_replicas - num_local_replicas;
UInt64 num_replicas = cluster->getShardsAddresses().at(shard_index).size();
UInt64 num_local_replicas = shard.getLocalNodeCount();
UInt64 num_remote_replicas = num_replicas - num_local_replicas;
/// In that case we don't have local replicas, but do it just in case
for (size_t i = 0; i < num_local_replicas; ++i)
for (UInt64 i = 0; i < num_local_replicas; ++i)
{
auto interpreter = InterpreterFactory::get(query_ast, context);
interpreter->execute();
@ -1997,16 +1997,16 @@ protected:
};
{
ThreadPool thread_pool(std::min(num_shards, getNumberOfPhysicalCPUCores()));
ThreadPool thread_pool(std::min<UInt64>(num_shards, getNumberOfPhysicalCPUCores()));
for (size_t shard_index = 0; shard_index < num_shards; ++shard_index)
for (UInt64 shard_index = 0; shard_index < num_shards; ++shard_index)
thread_pool.schedule([=] { do_for_shard(shard_index); });
thread_pool.wait();
}
size_t successful_shards = 0;
for (size_t num_replicas : per_shard_num_successful_replicas)
UInt64 successful_shards = 0;
for (UInt64 num_replicas : per_shard_num_successful_replicas)
successful_shards += (num_replicas > 0);
return successful_shards;

4
dbms/programs/obfuscator/Obfuscator.cpp
View File

@ -123,7 +123,7 @@ UInt64 hash(Ts... xs)
UInt64 maskBits(UInt64 x, size_t num_bits)
{
return x & ((1 << num_bits) - 1);
return x & ((1ULL << num_bits) - 1);
}
@ -149,7 +149,7 @@ UInt64 feistelNetwork(UInt64 x, size_t num_bits, UInt64 seed, size_t num_rounds
UInt64 bits = maskBits(x, num_bits);
for (size_t i = 0; i < num_rounds; ++i)
bits = feistelRound(bits, num_bits, seed, i);
return (x & ~((1 << num_bits) - 1)) ^ bits;
return (x & ~((1ULL << num_bits) - 1)) ^ bits;
}

2
dbms/programs/odbc-bridge/CMakeLists.txt
View File

@ -9,7 +9,7 @@ add_library (clickhouse-odbc-bridge-lib ${LINK_MODE}
validateODBCConnectionString.cpp
)
target_link_libraries (clickhouse-odbc-bridge-lib PRIVATE clickhouse_common_io daemon dbms)
target_link_libraries (clickhouse-odbc-bridge-lib PRIVATE clickhouse_dictionaries daemon dbms clickhouse_common_io)
target_include_directories (clickhouse-odbc-bridge-lib PUBLIC ${ClickHouse_SOURCE_DIR}/libs/libdaemon/include)
if (USE_POCO_SQLODBC)

2
dbms/programs/server/Server.cpp
View File

@ -411,7 +411,7 @@ int Server::main(const std::vector<std::string> & /*args*/)
global_context->setMarkCache(mark_cache_size);
#if USE_EMBEDDED_COMPILER
size_t compiled_expression_cache_size = config().getUInt64("compiled_expression_cache_size", std::numeric_limits<UInt64>::max());
size_t compiled_expression_cache_size = config().getUInt64("compiled_expression_cache_size", 500);
if (compiled_expression_cache_size)
global_context->setCompiledExpressionCache(compiled_expression_cache_size);
#endif

2
dbms/src/Columns/ColumnLowCardinality.h
View File

@ -133,7 +133,7 @@ public:
}
bool valuesHaveFixedSize() const override { return getDictionary().valuesHaveFixedSize(); }
bool isFixedAndContiguous() const override { return getDictionary().isFixedAndContiguous(); }
bool isFixedAndContiguous() const override { return false; }
size_t sizeOfValueIfFixed() const override { return getDictionary().sizeOfValueIfFixed(); }
bool isNumeric() const override { return getDictionary().isNumeric(); }
bool lowCardinality() const override { return true; }

3
dbms/src/Common/CurrentThread.cpp
View File

@ -3,7 +3,6 @@
#include "CurrentThread.h"
#include <common/logger_useful.h>
#include <Common/ThreadStatus.h>
#include <Common/ObjectPool.h>
#include <Common/TaskStatsInfoGetter.h>
#include <Interpreters/ProcessList.h>
#include <Interpreters/Context.h>
@ -24,8 +23,6 @@ namespace ErrorCodes
extern const int LOGICAL_ERROR;
}
SimpleObjectPool<TaskStatsInfoGetter> task_stats_info_getter_pool;
// Smoker's implementation to avoid thread_local usage: error: undefined symbol: __cxa_thread_atexit
#if defined(ARCADIA_ROOT)
struct ThreadStatusPtrHolder : ThreadStatusPtr

2
dbms/src/Common/ShellCommand.cpp
View File

@ -36,7 +36,7 @@ namespace
if (0 != pipe2(fds_rw, O_CLOEXEC))
DB::throwFromErrno("Cannot create pipe", DB::ErrorCodes::CANNOT_PIPE);
#else
if (0 != pipe(fds))
if (0 != pipe(fds_rw))
DB::throwFromErrno("Cannot create pipe", DB::ErrorCodes::CANNOT_PIPE);
if (0 != fcntl(fds_rw[0], F_SETFD, FD_CLOEXEC))
DB::throwFromErrno("Cannot create pipe", DB::ErrorCodes::CANNOT_PIPE);

5
dbms/src/Common/ThreadStatus.cpp
View File

@ -21,9 +21,6 @@ namespace ErrorCodes
}
extern SimpleObjectPool<TaskStatsInfoGetter> task_stats_info_getter_pool;
TasksStatsCounters TasksStatsCounters::current()
{
TasksStatsCounters res;
@ -74,7 +71,7 @@ void ThreadStatus::initPerformanceCounters()
if (TaskStatsInfoGetter::checkPermissions())
{
if (!taskstats_getter)
taskstats_getter = task_stats_info_getter_pool.getDefault();
taskstats_getter = std::make_unique<TaskStatsInfoGetter>();
*last_taskstats = TasksStatsCounters::current();
}

4
dbms/src/Common/ThreadStatus.h
View File

@ -2,7 +2,6 @@
#include <Common/ProfileEvents.h>
#include <Common/MemoryTracker.h>
#include <Common/ObjectPool.h>
#include <IO/Progress.h>
@ -175,8 +174,7 @@ protected:
std::unique_ptr<TasksStatsCounters> last_taskstats;
/// Set to non-nullptr only if we have enough capabilities.
/// We use pool because creation and destruction of TaskStatsInfoGetter objects are expensive.
SimpleObjectPool<TaskStatsInfoGetter>::Pointer taskstats_getter;
std::unique_ptr<TaskStatsInfoGetter> taskstats_getter;
};
}

4
dbms/src/Common/ZooKeeper/ZooKeeperImpl.cpp
View File

@ -1039,8 +1039,8 @@ void ZooKeeper::sendThread()
{
/// Wait for the next request in queue. No more than operation timeout. No more than until next heartbeat time.
UInt64 max_wait = std::min(
std::chrono::duration_cast<std::chrono::milliseconds>(next_heartbeat_time - now).count(),
operation_timeout.totalMilliseconds());
UInt64(std::chrono::duration_cast<std::chrono::milliseconds>(next_heartbeat_time - now).count()),
UInt64(operation_timeout.totalMilliseconds()));
RequestInfo info;
if (requests_queue.tryPop(info, max_wait))

1
dbms/src/Common/config.h.in
View File

@ -16,6 +16,7 @@
#cmakedefine01 USE_BASE64
#cmakedefine01 USE_HDFS
#cmakedefine01 USE_XXHASH
#cmakedefine01 USE_INTERNAL_LLVM_LIBRARY
#cmakedefine01 CLICKHOUSE_SPLIT_BINARY
#cmakedefine01 LLVM_HAS_RTTI

2
dbms/src/TableFunctions/parseRemoteDescription.cpp → dbms/src/Common/parseRemoteDescription.cpp
View File

@ -1,4 +1,4 @@
#include <TableFunctions/parseRemoteDescription.h>
#include "parseRemoteDescription.h"
#include <Common/Exception.h>
#include <IO/WriteHelpers.h>

0
dbms/src/TableFunctions/parseRemoteDescription.h → dbms/src/Common/parseRemoteDescription.h
View File

2
dbms/src/Compression/CompressionCodecZSTD.cpp
View File

@ -63,7 +63,7 @@ CompressionCodecZSTD::CompressionCodecZSTD(int level_)
void registerCodecZSTD(CompressionCodecFactory & factory)
{
UInt8 method_code = static_cast<char>(CompressionMethodByte::ZSTD);
UInt8 method_code = UInt8(CompressionMethodByte::ZSTD);
factory.registerCompressionCodec("ZSTD", method_code, [&](const ASTPtr & arguments) -> CompressionCodecPtr
{
int level = CompressionCodecZSTD::ZSTD_DEFAULT_LEVEL;

2
dbms/src/DataStreams/MergeSortingBlockInputStream.cpp
View File

@ -78,7 +78,7 @@ Block MergeSortingBlockInputStream::readImpl()
if (max_bytes_before_external_sort && sum_bytes_in_blocks > max_bytes_before_external_sort)
{
Poco::File(tmp_path).createDirectories();
temporary_files.emplace_back(new Poco::TemporaryFile(tmp_path));
temporary_files.emplace_back(std::make_unique<Poco::TemporaryFile>(tmp_path));
const std::string & path = temporary_files.back()->path();
WriteBufferFromFile file_buf(path);
CompressedWriteBuffer compressed_buf(file_buf);

4
dbms/src/DataTypes/DataTypeAggregateFunction.cpp
View File

@ -317,9 +317,9 @@ static DataTypePtr create(const ASTPtr & arguments)
params_row[i] = lit->value;
}
}
else if (const ASTIdentifier * identifier = typeid_cast<ASTIdentifier *>(arguments->children[0].get()))
else if (auto opt_name = getIdentifierName(arguments->children[0]))
{
function_name = identifier->name;
function_name = *opt_name;
}
else if (typeid_cast<ASTLiteral *>(arguments->children[0].get()))
{

6
dbms/src/DataTypes/DataTypeLowCardinality.cpp
View File

@ -508,6 +508,10 @@ void DataTypeLowCardinality::serializeBinaryBulkWithMultipleStreams(
size_t max_limit = column.size() - offset;
limit = limit ? std::min(limit, max_limit) : max_limit;
/// Do not write anything for empty column. (May happen while writing empty arrays.)
if (limit == 0)
return;
auto sub_column = low_cardinality_column.cutAndCompact(offset, limit);
ColumnPtr positions = sub_column->getIndexesPtr();
ColumnPtr keys = sub_column->getDictionary().getNestedColumn();
@ -709,7 +713,7 @@ void DataTypeLowCardinality::deserializeBinaryBulkWithMultipleStreams(
readIntBinary(low_cardinality_state->num_pending_rows, *indexes_stream);
}
size_t num_rows_to_read = std::min(limit, low_cardinality_state->num_pending_rows);
size_t num_rows_to_read = std::min<UInt64>(limit, low_cardinality_state->num_pending_rows);
readIndexes(num_rows_to_read);
limit -= num_rows_to_read;
low_cardinality_state->num_pending_rows -= num_rows_to_read;

1
dbms/src/Databases/DatabaseOrdinary.cpp
View File

@ -7,6 +7,7 @@
#include <Databases/DatabaseOrdinary.h>
#include <Databases/DatabaseMemory.h>
#include <Databases/DatabasesCommon.h>
#include <Common/typeid_cast.h>
#include <Common/escapeForFileName.h>
#include <Common/StringUtils/StringUtils.h>
#include <Common/Stopwatch.h>

1
dbms/src/Databases/DatabasesCommon.cpp
View File

@ -1,5 +1,6 @@
#include <sstream>
#include <Common/typeid_cast.h>
#include <Parsers/parseQuery.h>
#include <Parsers/ParserCreateQuery.h>
#include <Parsers/ASTCreateQuery.h>

8
dbms/src/Dictionaries/CMakeLists.txt
View File

@ -20,9 +20,7 @@ endif()
if(USE_POCO_SQLODBC)
target_link_libraries(clickhouse_dictionaries PRIVATE ${Poco_SQLODBC_LIBRARY} ${Poco_SQL_LIBRARY})
if (NOT USE_INTERNAL_POCO_LIBRARY)
target_include_directories(clickhouse_dictionaries SYSTEM PRIVATE ${ODBC_INCLUDE_DIRECTORIES} ${Poco_SQLODBC_INCLUDE_DIR} ${Poco_SQL_INCLUDE_DIR})
endif()
target_include_directories(clickhouse_dictionaries SYSTEM PRIVATE ${ODBC_INCLUDE_DIRECTORIES} ${Poco_SQLODBC_INCLUDE_DIR} ${Poco_SQL_INCLUDE_DIR})
endif()
if(Poco_Data_FOUND)
@ -31,9 +29,7 @@ endif()
if(USE_POCO_DATAODBC)
target_link_libraries(clickhouse_dictionaries PRIVATE ${Poco_DataODBC_LIBRARY} ${Poco_Data_LIBRARY})
if (NOT USE_INTERNAL_POCO_LIBRARY)
target_include_directories(clickhouse_dictionaries SYSTEM PRIVATE ${ODBC_INCLUDE_DIRECTORIES} ${Poco_DataODBC_INCLUDE_DIR})
endif()
target_include_directories(clickhouse_dictionaries SYSTEM PRIVATE ${ODBC_INCLUDE_DIRECTORIES} ${Poco_DataODBC_INCLUDE_DIR})
endif()
if(USE_POCO_MONGODB)

2
dbms/src/Dictionaries/Embedded/RegionsNames.h
View File

@ -73,7 +73,7 @@ public:
{
size_t language_id = static_cast<size_t>(language);
if (region_id > names_refs[language_id].size())
if (region_id >= names_refs[language_id].size())
return StringRef("", 0);
StringRef ref = names_refs[language_id][region_id];

4
dbms/src/Formats/PrettyBlockOutputStream.cpp
View File

@ -61,7 +61,7 @@ void PrettyBlockOutputStream::calculateWidths(
elem.type->serializeText(*elem.column, j, out, format_settings);
}
widths[i][j] = std::min(format_settings.pretty.max_column_pad_width,
widths[i][j] = std::min<UInt64>(format_settings.pretty.max_column_pad_width,
UTF8::computeWidth(reinterpret_cast<const UInt8 *>(serialized_value.data()), serialized_value.size(), prefix));
max_widths[i] = std::max(max_widths[i], widths[i][j]);
}
@ -69,7 +69,7 @@ void PrettyBlockOutputStream::calculateWidths(
/// And also calculate widths for names of columns.
{
// name string doesn't contain Tab, no need to pass `prefix`
name_widths[i] = std::min(format_settings.pretty.max_column_pad_width,
name_widths[i] = std::min<UInt64>(format_settings.pretty.max_column_pad_width,
UTF8::computeWidth(reinterpret_cast<const UInt8 *>(elem.name.data()), elem.name.size()));
max_widths[i] = std::max(max_widths[i], name_widths[i]);
}

8
dbms/src/Functions/CMakeLists.txt
View File

@ -47,9 +47,11 @@ if (ENABLE_TESTS)
endif ()
if (USE_EMBEDDED_COMPILER)
target_include_directories (clickhouse_functions SYSTEM BEFORE PUBLIC ${LLVM_INCLUDE_DIRS})
llvm_libs_all(REQUIRED_LLVM_LIBRARIES)
target_link_libraries(clickhouse_functions PRIVATE ${REQUIRED_LLVM_LIBRARIES})
target_include_directories(clickhouse_functions SYSTEM BEFORE PUBLIC ${LLVM_INCLUDE_DIRS})
endif ()
if (USE_BASE64)
target_include_directories (clickhouse_functions SYSTEM PRIVATE ${BASE64_INCLUDE_DIR})
if(USE_BASE64)
target_include_directories(clickhouse_functions SYSTEM PRIVATE ${BASE64_INCLUDE_DIR})
endif()

4
dbms/src/Functions/IFunction.cpp
View File

@ -512,8 +512,8 @@ static std::optional<DataTypes> removeNullables(const DataTypes & types)
if (!typeid_cast<const DataTypeNullable *>(type.get()))
continue;
DataTypes filtered;
for (const auto & type : types)
filtered.emplace_back(removeNullable(type));
for (const auto & sub_type : types)
filtered.emplace_back(removeNullable(sub_type));
return filtered;
}
return {};

2
dbms/src/Functions/hasColumnInTable.cpp
View File

@ -132,7 +132,7 @@ void FunctionHasColumnInTable::executeImpl(Block & block, const ColumnNumbers &
has_column = remote_columns.hasPhysical(column_name);
}
block.getByPosition(result).column = DataTypeUInt8().createColumnConst(input_rows_count, has_column);
block.getByPosition(result).column = DataTypeUInt8().createColumnConst(input_rows_count, Field(has_column));
}

1
dbms/src/IO/Progress.h
View File

@ -1,6 +1,7 @@
#pragma once
#include <atomic>
#include <cstddef>
#include <common/Types.h>
#include <Core/Defines.h>

6
dbms/src/Interpreters/AnalyzedJoin.cpp
View File

@ -7,7 +7,6 @@
#include <Parsers/ASTExpressionList.h>
#include <Parsers/ASTTablesInSelectQuery.h>
#include <Parsers/ASTSelectQuery.h>
#include <Parsers/ASTIdentifier.h>
#include <Storages/IStorage.h>
@ -48,7 +47,7 @@ ExpressionActionsPtr AnalyzedJoin::createJoinedBlockActions(
source_column_names.emplace_back(column.name_and_type);
ASTPtr query = expression_list;
auto syntax_result = SyntaxAnalyzer(context, {}).analyze(query, source_column_names, required_columns);
auto syntax_result = SyntaxAnalyzer(context).analyze(query, source_column_names, required_columns);
ExpressionAnalyzer analyzer(query, syntax_result, context, {}, required_columns);
auto joined_block_actions = analyzer.getActions(false);
@ -118,8 +117,7 @@ NamesAndTypesList getNamesAndTypeListFromTableExpression(const ASTTableExpressio
}
else if (table_expression.database_and_table_name)
{
const auto & identifier = static_cast<const ASTIdentifier &>(*table_expression.database_and_table_name);
DatabaseAndTableWithAlias database_table(identifier);
DatabaseAndTableWithAlias database_table(table_expression.database_and_table_name);
const auto & table = context.getTable(database_table.database, database_table.table);
names_and_type_list = table->getSampleBlockNonMaterialized().getNamesAndTypesList();
}

2
dbms/src/Interpreters/ArrayJoinedColumnsVisitor.h
View File

@ -58,7 +58,7 @@ private:
NameToNameMap & array_join_alias_to_name = data.array_join_alias_to_name;
NameToNameMap & array_join_result_to_source = data.array_join_result_to_source;
if (!node.general())
if (!getColumnIdentifierName(node))
return;
auto splitted = Nested::splitName(node.name); /// ParsedParams, Key1

3
dbms/src/Interpreters/AsynchronousMetrics.cpp
View File

@ -137,10 +137,7 @@ void AsynchronousMetrics::update()
#if USE_EMBEDDED_COMPILER
{
if (auto compiled_expression_cache = context.getCompiledExpressionCache())
{
set("CompiledExpressionCacheBytes", compiled_expression_cache->weight());
set("CompiledExpressionCacheCount", compiled_expression_cache->count());
}
}
#endif

21
dbms/src/Interpreters/ClusterProxy/executeQuery.cpp
View File

@ -14,14 +14,8 @@ namespace DB
namespace ClusterProxy
{
BlockInputStreams executeQuery(
IStreamFactory & stream_factory, const ClusterPtr & cluster,
const ASTPtr & query_ast, const Context & context, const Settings & settings)
Context removeUserRestrictionsFromSettings(const Context & context, const Settings & settings)
{
BlockInputStreams res;
const std::string query = queryToString(query_ast);
Settings new_settings = settings;
new_settings.queue_max_wait_ms = Cluster::saturate(new_settings.queue_max_wait_ms, settings.max_execution_time);
@ -39,6 +33,19 @@ BlockInputStreams executeQuery(
Context new_context(context);
new_context.setSettings(new_settings);
return new_context;
}
BlockInputStreams executeQuery(
IStreamFactory & stream_factory, const ClusterPtr & cluster,
const ASTPtr & query_ast, const Context & context, const Settings & settings)
{
BlockInputStreams res;
const std::string query = queryToString(query_ast);
Context new_context = removeUserRestrictionsFromSettings(context, settings);
ThrottlerPtr user_level_throttler;
if (auto process_list_element = context.getProcessListElement())
user_level_throttler = process_list_element->getUserNetworkThrottler();

4
dbms/src/Interpreters/ClusterProxy/executeQuery.h
View File

@ -16,6 +16,10 @@ namespace ClusterProxy
class IStreamFactory;
/// removes different restrictions (like max_concurrent_queries_for_user, max_memory_usage_for_user, etc.)
/// from settings and creates new context with them
Context removeUserRestrictionsFromSettings(const Context & context, const Settings & settings);
/// Execute a distributed query, creating a vector of BlockInputStreams, from which the result can be read.
/// `stream_factory` object encapsulates the logic of creating streams for a different type of query
/// (currently SELECT, DESCRIBE).

2
dbms/src/Interpreters/ColumnNamesContext.cpp
View File

@ -31,7 +31,7 @@ bool ColumnNamesContext::addColumnAliasIfAny(const IAST & ast, bool is_public)
void ColumnNamesContext::addColumnIdentifier(const ASTIdentifier & node, bool is_public)
{
if (!node.general())
if (!getColumnIdentifierName(node))
return;
required_names.insert(node.name);

24
dbms/src/Interpreters/ColumnNamesContext.h
View File

@ -16,18 +16,21 @@ struct ColumnNamesContext
{
struct JoinedTable
{
const ASTTableExpression * expr;
const ASTTableJoin * join;
const ASTTableExpression * expr = nullptr;
const ASTTableJoin * join = nullptr;
std::optional<String> alias() const
{
String alias;
if (expr->database_and_table_name)
alias = expr->database_and_table_name->tryGetAlias();
else if (expr->table_function)
alias = expr->table_function->tryGetAlias();
else if (expr->subquery)
alias = expr->subquery->tryGetAlias();
if (expr)
{
if (expr->database_and_table_name)
alias = expr->database_and_table_name->tryGetAlias();
else if (expr->table_function)
alias = expr->table_function->tryGetAlias();
else if (expr->subquery)
alias = expr->subquery->tryGetAlias();
}
if (!alias.empty())
return alias;
return {};
@ -35,9 +38,8 @@ struct ColumnNamesContext
std::optional<String> name() const
{
if (auto * node = expr->database_and_table_name.get())
if (auto * identifier = typeid_cast<const ASTIdentifier *>(node))
return identifier->name;
if (expr)
return getIdentifierName(expr->database_and_table_name);
return {};
}

44
dbms/src/Interpreters/DatabaseAndTableWithAlias.cpp
View File

@ -54,8 +54,6 @@ size_t getNumComponentsToStripInOrderToTranslateQualifiedName(const ASTIdentifie
{
size_t num_qualifiers_to_strip = 0;
auto get_identifier_name = [](const ASTPtr & ast) { return static_cast<const ASTIdentifier &>(*ast).name; };
/// It is compound identifier
if (!identifier.children.empty())
{
@ -64,16 +62,16 @@ size_t getNumComponentsToStripInOrderToTranslateQualifiedName(const ASTIdentifie
/// database.table.column
if (num_components >= 3
&& !names.database.empty()
&& get_identifier_name(identifier.children[0]) == names.database
&& get_identifier_name(identifier.children[1]) == names.table)
&& *getIdentifierName(identifier.children[0]) == names.database
&& *getIdentifierName(identifier.children[1]) == names.table)
{
num_qualifiers_to_strip = 2;
}
/// table.column or alias.column. If num_components > 2, it is like table.nested.column.
if (num_components >= 2
&& ((!names.table.empty() && get_identifier_name(identifier.children[0]) == names.table)
|| (!names.alias.empty() && get_identifier_name(identifier.children[0]) == names.alias)))
&& ((!names.table.empty() && *getIdentifierName(identifier.children[0]) == names.table)
|| (!names.alias.empty() && *getIdentifierName(identifier.children[0]) == names.alias)))
{
num_qualifiers_to_strip = 1;
}
@ -94,26 +92,24 @@ DatabaseAndTableWithAlias::DatabaseAndTableWithAlias(const ASTIdentifier & ident
if (identifier.children.size() != 2)
throw Exception("Logical error: number of components in table expression not equal to two", ErrorCodes::LOGICAL_ERROR);
const ASTIdentifier * db_identifier = typeid_cast<const ASTIdentifier *>(identifier.children[0].get());
const ASTIdentifier * table_identifier = typeid_cast<const ASTIdentifier *>(identifier.children[1].get());
if (!db_identifier || !table_identifier)
throw Exception("Logical error: identifiers expected", ErrorCodes::LOGICAL_ERROR);
database = db_identifier->name;
table = table_identifier->name;
getIdentifierName(identifier.children[0], database);
getIdentifierName(identifier.children[1], table);
}
}
DatabaseAndTableWithAlias::DatabaseAndTableWithAlias(const ASTPtr & node, const String & current_database)
{
const auto * identifier = typeid_cast<const ASTIdentifier *>(node.get());
if (!identifier)
throw Exception("Logical error: identifier expected", ErrorCodes::LOGICAL_ERROR);
*this = DatabaseAndTableWithAlias(*identifier, current_database);
}
DatabaseAndTableWithAlias::DatabaseAndTableWithAlias(const ASTTableExpression & table_expression, const String & current_database)
{
if (table_expression.database_and_table_name)
{
const auto * identifier = typeid_cast<const ASTIdentifier *>(table_expression.database_and_table_name.get());
if (!identifier)
throw Exception("Logical error: identifier expected", ErrorCodes::LOGICAL_ERROR);
*this = DatabaseAndTableWithAlias(*identifier, current_database);
}
*this = DatabaseAndTableWithAlias(table_expression.database_and_table_name, current_database);
else if (table_expression.table_function)
alias = table_expression.table_function->tryGetAlias();
else if (table_expression.subquery)
@ -207,14 +203,10 @@ std::optional<DatabaseAndTableWithAlias> getDatabaseAndTable(const ASTSelectQuer
return {};
ASTPtr database_and_table_name = table_expression->database_and_table_name;
if (!database_and_table_name)
if (!database_and_table_name || !isIdentifier(database_and_table_name))
return {};
const ASTIdentifier * identifier = typeid_cast<const ASTIdentifier *>(database_and_table_name.get());
if (!identifier)
return {};
return *identifier;
return DatabaseAndTableWithAlias(database_and_table_name);
}
ASTPtr getTableFunctionOrSubquery(const ASTSelectQuery & select, size_t table_number)

2
dbms/src/Interpreters/DatabaseAndTableWithAlias.h
View File

@ -23,6 +23,8 @@ struct DatabaseAndTableWithAlias
String table;
String alias;
DatabaseAndTableWithAlias() = default;
DatabaseAndTableWithAlias(const ASTPtr & identifier_node, const String & current_database = "");
DatabaseAndTableWithAlias(const ASTIdentifier & identifier, const String & current_database = "");
DatabaseAndTableWithAlias(const ASTTableExpression & table_expression, const String & current_database);

10
dbms/src/Interpreters/ExecuteScalarSubqueriesVisitor.cpp
View File

@ -37,7 +37,7 @@ static ASTPtr addTypeConversion(std::unique_ptr<ASTLiteral> && ast, const String
return res;
}
bool ExecuteScalarSubqueriesMatcher::needChildVisit(ASTPtr & node, const ASTPtr &)
bool ExecuteScalarSubqueriesMatcher::needChildVisit(ASTPtr & node, const ASTPtr & child)
{
/// Processed
if (typeid_cast<ASTSubquery *>(node.get()) ||
@ -48,6 +48,14 @@ bool ExecuteScalarSubqueriesMatcher::needChildVisit(ASTPtr & node, const ASTPtr
if (typeid_cast<ASTTableExpression *>(node.get()))
return false;
if (typeid_cast<ASTSelectQuery *>(node.get()))
{
/// Do not go to FROM, JOIN, UNION.
if (typeid_cast<ASTTableExpression *>(child.get()) ||
typeid_cast<ASTSelectQuery *>(child.get()))
return false;
}
return true;
}

6
dbms/src/Interpreters/ExpressionAnalyzer.cpp
View File

@ -78,9 +78,6 @@ namespace ErrorCodes
extern const int EXPECTED_ALL_OR_ANY;
}
/// From SyntaxAnalyzer.cpp
extern void removeDuplicateColumns(NamesAndTypesList & columns);
ExpressionAnalyzer::ExpressionAnalyzer(
const ASTPtr & query_,
const SyntaxAnalyzerResultPtr & syntax_analyzer_result_,
@ -551,8 +548,7 @@ bool ExpressionAnalyzer::appendJoin(ExpressionActionsChain & chain, bool only_ty
/// TODO This syntax does not support specifying a database name.
if (table_to_join.database_and_table_name)
{
const auto & identifier = static_cast<const ASTIdentifier &>(*table_to_join.database_and_table_name);
DatabaseAndTableWithAlias database_table(identifier);
DatabaseAndTableWithAlias database_table(table_to_join.database_and_table_name);
StoragePtr table = context.tryGetTable(database_table.database, database_table.table);
if (table)

165
dbms/src/Interpreters/ExpressionJIT.cpp
View File

@ -161,21 +161,21 @@ auto wrapJITSymbolResolver(llvm::JITSymbolResolver & jsr)
// Actually this should work for 7.0.0 but now we have OLDER 7.0.0svn in contrib
auto flags = [&](const llvm::orc::SymbolNameSet & symbols)
{
llvm::orc::SymbolFlagsMap flags;
llvm::orc::SymbolFlagsMap flags_map;
for (const auto & symbol : symbols)
{
auto resolved = jsr.lookupFlags({*symbol});
if (resolved && resolved->size())
flags.emplace(symbol, resolved->begin()->second);
flags_map.emplace(symbol, resolved->begin()->second);
}
return flags;
return flags_map;
};
#endif
auto symbols = [&](std::shared_ptr<llvm::orc::AsynchronousSymbolQuery> query, llvm::orc::SymbolNameSet symbols)
auto symbols = [&](std::shared_ptr<llvm::orc::AsynchronousSymbolQuery> query, llvm::orc::SymbolNameSet symbols_set)
{
llvm::orc::SymbolNameSet missing;
for (const auto & symbol : symbols)
for (const auto & symbol : symbols_set)
{
auto resolved = jsr.lookup({*symbol});
if (resolved && resolved->size())
@ -189,70 +189,36 @@ auto wrapJITSymbolResolver(llvm::JITSymbolResolver & jsr)
}
#endif
#if LLVM_VERSION_MAJOR >= 6
struct CountingMMapper final : public llvm::SectionMemoryManager::MemoryMapper
{
MemoryTracker memory_tracker{VariableContext::Global};
llvm::sys::MemoryBlock allocateMappedMemory(llvm::SectionMemoryManager::AllocationPurpose /*purpose*/,
size_t num_bytes,
const llvm::sys::MemoryBlock * const near_block,
unsigned flags,
std::error_code & error_code) override
{
memory_tracker.alloc(num_bytes);
return llvm::sys::Memory::allocateMappedMemory(num_bytes, near_block, flags, error_code);
}
std::error_code protectMappedMemory(const llvm::sys::MemoryBlock & block, unsigned flags) override
{
return llvm::sys::Memory::protectMappedMemory(block, flags);
}
std::error_code releaseMappedMemory(llvm::sys::MemoryBlock & block) override
{
memory_tracker.free(block.size());
return llvm::sys::Memory::releaseMappedMemory(block);
}
};
#if LLVM_VERSION_MAJOR >= 7
using ModulePtr = std::unique_ptr<llvm::Module>;
#else
using ModulePtr = std::shared_ptr<llvm::Module>;
#endif
struct LLVMContext
{
static inline std::atomic<size_t> id_counter{0};
llvm::LLVMContext context;
std::shared_ptr<llvm::LLVMContext> context;
#if LLVM_VERSION_MAJOR >= 7
llvm::orc::ExecutionSession execution_session;
std::unique_ptr<llvm::Module> module;
#else
std::shared_ptr<llvm::Module> module;
#endif
ModulePtr module;
std::unique_ptr<llvm::TargetMachine> machine;
#if LLVM_VERSION_MAJOR >= 6
std::unique_ptr<CountingMMapper> memory_mapper;
#endif
std::shared_ptr<llvm::SectionMemoryManager> memory_manager;
llvm::orc::RTDyldObjectLinkingLayer object_layer;
llvm::orc::IRCompileLayer<decltype(object_layer), llvm::orc::SimpleCompiler> compile_layer;
llvm::DataLayout layout;
llvm::IRBuilder<> builder;
std::unordered_map<std::string, void *> symbols;
size_t id;
LLVMContext()
: context(std::make_shared<llvm::LLVMContext>())
#if LLVM_VERSION_MAJOR >= 7
: module(std::make_unique<llvm::Module>("jit", context))
, module(std::make_unique<llvm::Module>("jit", *context))
#else
: module(std::make_shared<llvm::Module>("jit", context))
, module(std::make_shared<llvm::Module>("jit", *context))
#endif
, machine(getNativeMachine())
#if LLVM_VERSION_MAJOR >= 6
, memory_mapper(std::make_unique<CountingMMapper>())
, memory_manager(std::make_shared<llvm::SectionMemoryManager>(memory_mapper.get()))
#else
, memory_manager(std::make_shared<llvm::SectionMemoryManager>())
#endif
#if LLVM_VERSION_MAJOR >= 7
, object_layer(execution_session, [this](llvm::orc::VModuleKey)
{
@ -263,32 +229,31 @@ struct LLVMContext
#endif
, compile_layer(object_layer, llvm::orc::SimpleCompiler(*machine))
, layout(machine->createDataLayout())
, builder(context)
, id(id_counter++)
, builder(*context)
{
module->setDataLayout(layout);
module->setTargetTriple(machine->getTargetTriple().getTriple());
}
/// returns used memory
size_t compileAllFunctionsToNativeCode()
void compileAllFunctionsToNativeCode()
{
if (!module->size())
return 0;
llvm::PassManagerBuilder builder;
return;
llvm::PassManagerBuilder pass_manager_builder;
llvm::legacy::PassManager mpm;
llvm::legacy::FunctionPassManager fpm(module.get());
builder.OptLevel = 3;
builder.SLPVectorize = true;
builder.LoopVectorize = true;
builder.RerollLoops = true;
builder.VerifyInput = true;
builder.VerifyOutput = true;
machine->adjustPassManager(builder);
pass_manager_builder.OptLevel = 3;
pass_manager_builder.SLPVectorize = true;
pass_manager_builder.LoopVectorize = true;
pass_manager_builder.RerollLoops = true;
pass_manager_builder.VerifyInput = true;
pass_manager_builder.VerifyOutput = true;
machine->adjustPassManager(pass_manager_builder);
fpm.add(llvm::createTargetTransformInfoWrapperPass(machine->getTargetIRAnalysis()));
mpm.add(llvm::createTargetTransformInfoWrapperPass(machine->getTargetIRAnalysis()));
builder.populateFunctionPassManager(fpm);
builder.populateModulePassManager(mpm);
pass_manager_builder.populateFunctionPassManager(fpm);
pass_manager_builder.populateModulePassManager(mpm);
fpm.doInitialization();
for (auto & function : *module)
fpm.run(function);
@ -323,26 +288,20 @@ struct LLVMContext
throw Exception("Function " + name + " failed to link", ErrorCodes::CANNOT_COMPILE_CODE);
symbols[name] = reinterpret_cast<void *>(*address);
}
#if LLVM_VERSION_MAJOR >= 6
return memory_mapper->memory_tracker.get();
#else
return 0;
#endif
}
};
class LLVMPreparedFunction : public PreparedFunctionImpl
{
std::string name;
std::shared_ptr<LLVMContext> context;
void * function;
public:
LLVMPreparedFunction(std::string name_, std::shared_ptr<LLVMContext> context)
: name(std::move(name_)), context(context)
LLVMPreparedFunction(const std::string & name_, const std::unordered_map<std::string, void *> & symbols)
: name(name_)
{
auto it = context->symbols.find(name);
if (context->symbols.end() == it)
auto it = symbols.find(name);
if (symbols.end() == it)
throw Exception("Cannot find symbol " + name + " in LLVMContext", ErrorCodes::LOGICAL_ERROR);
function = it->second;
}
@ -373,16 +332,16 @@ public:
}
};
static void compileFunctionToLLVMByteCode(std::shared_ptr<LLVMContext> & context, const IFunctionBase & f)
static void compileFunctionToLLVMByteCode(LLVMContext & context, const IFunctionBase & f)
{
ProfileEvents::increment(ProfileEvents::CompileFunction);
auto & arg_types = f.getArgumentTypes();
auto & b = context->builder;
auto & b = context.builder;
auto * size_type = b.getIntNTy(sizeof(size_t) * 8);
auto * data_type = llvm::StructType::get(b.getInt8PtrTy(), b.getInt8PtrTy(), size_type);
auto * func_type = llvm::FunctionType::get(b.getVoidTy(), { size_type, data_type->getPointerTo() }, /*isVarArg=*/false);
auto * func = llvm::Function::Create(func_type, llvm::Function::ExternalLinkage, f.getName(), context->module.get());
auto * func = llvm::Function::Create(func_type, llvm::Function::ExternalLinkage, f.getName(), context.module.get());
auto args = func->args().begin();
llvm::Value * counter_arg = &*args++;
llvm::Value * columns_arg = &*args++;
@ -504,12 +463,21 @@ static CompilableExpression subexpression(const IFunctionBase & f, std::vector<C
};
}
LLVMFunction::LLVMFunction(const ExpressionActions::Actions & actions, std::shared_ptr<LLVMContext> context, const Block & sample_block)
: name(actions.back().result_name), context(context)
struct LLVMModuleState
{
std::unordered_map<std::string, void *> symbols;
std::shared_ptr<llvm::LLVMContext> major_context;
std::shared_ptr<llvm::SectionMemoryManager> memory_manager;
};
LLVMFunction::LLVMFunction(const ExpressionActions::Actions & actions, const Block & sample_block)
: name(actions.back().result_name)
, module_state(std::make_unique<LLVMModuleState>())
{
LLVMContext context;
for (const auto & c : sample_block)
/// TODO: implement `getNativeValue` for all types & replace the check with `c.column && toNativeType(...)`
if (c.column && getNativeValue(toNativeType(context->builder, c.type), *c.column, 0))
if (c.column && getNativeValue(toNativeType(context.builder, c.type), *c.column, 0))
subexpressions[c.name] = subexpression(c.column, c.type);
for (const auto & action : actions)
{
@ -530,6 +498,11 @@ LLVMFunction::LLVMFunction(const ExpressionActions::Actions & actions, std::shar
originals.push_back(action.function_base);
}
compileFunctionToLLVMByteCode(context, *this);
context.compileAllFunctionsToNativeCode();
module_state->symbols = context.symbols;
module_state->major_context = context.context;
module_state->memory_manager = context.memory_manager;
}
llvm::Value * LLVMFunction::compile(llvm::IRBuilderBase & builder, ValuePlaceholders values) const
@ -540,8 +513,7 @@ llvm::Value * LLVMFunction::compile(llvm::IRBuilderBase & builder, ValuePlacehol
return it->second(builder, values);
}
PreparedFunctionPtr LLVMFunction::prepare(const Block &, const ColumnNumbers &, size_t) const { return std::make_shared<LLVMPreparedFunction>(name, context); }
PreparedFunctionPtr LLVMFunction::prepare(const Block &, const ColumnNumbers &, size_t) const { return std::make_shared<LLVMPreparedFunction>(name, module_state->symbols); }
bool LLVMFunction::isDeterministic() const
{
@ -622,28 +594,6 @@ static bool isCompilable(const IFunctionBase & function)
return function.isCompilable();
}
size_t CompiledExpressionCache::weight() const
{
#if LLVM_VERSION_MAJOR >= 6
std::lock_guard lock(mutex);
size_t result{0};
std::unordered_set<size_t> seen;
for (const auto & cell : cells)
{
auto function_context = cell.second.value->getContext();
if (!seen.count(function_context->id))
{
result += function_context->memory_mapper->memory_tracker.get();
seen.insert(function_context->id);
}
}
return result;
#else
return Base::weight();
#endif
}
std::vector<std::unordered_set<std::optional<size_t>>> getActionsDependents(const ExpressionActions::Actions & actions, const Names & output_columns)
{
/// an empty optional is a poisoned value prohibiting the column's producer from being removed
@ -748,21 +698,16 @@ void compileFunctions(ExpressionActions::Actions & actions, const Names & output
std::tie(fn, std::ignore) = compilation_cache->getOrSet(hash_key, [&inlined_func=std::as_const(fused[i]), &sample_block] ()
{
Stopwatch watch;
std::shared_ptr<LLVMContext> context = std::make_shared<LLVMContext>();
auto result_fn = std::make_shared<LLVMFunction>(inlined_func, context, sample_block);
size_t used_memory = context->compileAllFunctionsToNativeCode();
ProfileEvents::increment(ProfileEvents::CompileExpressionsBytes, used_memory);
std::shared_ptr<LLVMFunction> result_fn;
result_fn = std::make_shared<LLVMFunction>(inlined_func, sample_block);
ProfileEvents::increment(ProfileEvents::CompileExpressionsMicroseconds, watch.elapsedMicroseconds());
return result_fn;
});
}
else
{
std::shared_ptr<LLVMContext> context = std::make_shared<LLVMContext>();
Stopwatch watch;
fn = std::make_shared<LLVMFunction>(fused[i], context, sample_block);
size_t used_memory = context->compileAllFunctionsToNativeCode();
ProfileEvents::increment(ProfileEvents::CompileExpressionsBytes, used_memory);
fn = std::make_shared<LLVMFunction>(fused[i], sample_block);
ProfileEvents::increment(ProfileEvents::CompileExpressionsMicroseconds, watch.elapsedMicroseconds());
}

19
dbms/src/Interpreters/ExpressionJIT.h
View File

@ -14,19 +14,23 @@
namespace DB
{
struct LLVMContext;
using CompilableExpression = std::function<llvm::Value * (llvm::IRBuilderBase &, const ValuePlaceholders &)>;
struct LLVMModuleState;
class LLVMFunction : public IFunctionBase
{
std::string name;
Names arg_names;
DataTypes arg_types;
std::shared_ptr<LLVMContext> context;
std::vector<FunctionBasePtr> originals;
std::unordered_map<StringRef, CompilableExpression> subexpressions;
std::unique_ptr<LLVMModuleState> module_state;
public:
LLVMFunction(const ExpressionActions::Actions & actions, std::shared_ptr<LLVMContext> context, const Block & sample_block);
LLVMFunction(const ExpressionActions::Actions & actions, const Block & sample_block);
bool isCompilable() const override { return true; }
@ -54,8 +58,7 @@ public:
Monotonicity getMonotonicityForRange(const IDataType & type, const Field & left, const Field & right) const override;
std::shared_ptr<LLVMContext> getContext() const { return context; }
const LLVMModuleState * getLLVMModuleState() const { return module_state.get(); }
};
/** This child of LRUCache breaks one of it's invariants: total weight may be changed after insertion.
@ -63,13 +66,9 @@ public:
*/
class CompiledExpressionCache : public LRUCache<UInt128, LLVMFunction, UInt128Hash>
{
private:
using Base = LRUCache<UInt128, LLVMFunction, UInt128Hash>;
public:
using Base = LRUCache<UInt128, LLVMFunction, UInt128Hash>;
using Base::Base;
size_t weight() const;
};
/// For each APPLY_FUNCTION action, try to compile the function to native code; if the only uses of a compilable

6
dbms/src/Interpreters/ExternalTablesVisitor.h
View File

@ -33,9 +33,9 @@ public:
private:
static std::vector<ASTPtr *> visit(const ASTIdentifier & node, ASTPtr &, Data & data)
{
if (node.special())
if (StoragePtr external_storage = data.context.tryGetExternalTable(node.name))
data.external_tables[node.name] = external_storage;
if (auto opt_name = getTableIdentifierName(node))
if (StoragePtr external_storage = data.context.tryGetExternalTable(*opt_name))
data.external_tables[*opt_name] = external_storage;
return {};
}
};

8
dbms/src/Interpreters/InJoinSubqueriesPreprocessor.cpp
View File

@ -5,7 +5,6 @@
#include <Parsers/ASTSelectQuery.h>
#include <Parsers/ASTTablesInSelectQuery.h>
#include <Parsers/ASTFunction.h>
#include <Parsers/ASTIdentifier.h>
#include <Common/typeid_cast.h>
@ -82,12 +81,7 @@ void forEachTable(IAST * node, F && f)
StoragePtr tryGetTable(const ASTPtr & database_and_table, const Context & context)
{
const ASTIdentifier * id = typeid_cast<const ASTIdentifier *>(database_and_table.get());
if (!id)
throw Exception("Logical error: identifier expected", ErrorCodes::LOGICAL_ERROR);
DatabaseAndTableWithAlias db_and_table(*id);
DatabaseAndTableWithAlias db_and_table(database_and_table);
return context.tryGetTable(db_and_table.database, db_and_table.table);
}

2
dbms/src/Interpreters/InterpreterCreateQuery.cpp
View File

@ -242,7 +242,7 @@ static ColumnsDeclarationAndModifiers parseColumns(const ASTExpressionList & col
/// set missing types and wrap default_expression's in a conversion-function if necessary
if (!defaulted_columns.empty())
{
auto syntax_analyzer_result = SyntaxAnalyzer(context, {}).analyze(default_expr_list, columns);
auto syntax_analyzer_result = SyntaxAnalyzer(context).analyze(default_expr_list, columns);
const auto actions = ExpressionAnalyzer(default_expr_list, syntax_analyzer_result, context).getActions(true);
const auto block = actions->getSampleBlock();

8
dbms/src/Interpreters/InterpreterDescribeQuery.cpp
View File

@ -95,13 +95,11 @@ BlockInputStreamPtr InterpreterDescribeQuery::executeImpl()
auto database_ptr = identifier->children[0];
auto table_ptr = identifier->children[1];
if (database_ptr)
database_name = typeid_cast<ASTIdentifier &>(*database_ptr).name;
if (table_ptr)
table_name = typeid_cast<ASTIdentifier &>(*table_ptr).name;
getIdentifierName(database_ptr, database_name);
getIdentifierName(table_ptr, table_name);
}
else
table_name = typeid_cast<ASTIdentifier &>(*identifier).name;
getIdentifierName(identifier, table_name);
table = context.getTable(database_name, table_name);
}

1
dbms/src/Interpreters/InterpreterInsertQuery.cpp
View File

@ -99,7 +99,6 @@ BlockIO InterpreterInsertQuery::execute()
out = std::make_shared<PushingToViewsBlockOutputStream>(query.database, query.table, table, context, query_ptr, query.no_destination);
/// Do not squash blocks if it is a sync INSERT into Distributed, since it lead to double bufferization on client and server side.
/// Client-side bufferization might cause excessive timeouts (especially in case of big blocks).
if (!(context.getSettingsRef().insert_distributed_sync && table->isRemote()))

8
dbms/src/Interpreters/InterpreterSelectQuery.cpp
View File

@ -184,8 +184,8 @@ InterpreterSelectQuery::InterpreterSelectQuery(
if (storage)
table_lock = storage->lockStructure(false);
syntax_analyzer_result = SyntaxAnalyzer(context, storage)
.analyze(query_ptr, source_header.getNamesAndTypesList(), required_result_column_names, subquery_depth);
syntax_analyzer_result = SyntaxAnalyzer(context, subquery_depth).analyze(
query_ptr, source_header.getNamesAndTypesList(), required_result_column_names, storage);
query_analyzer = std::make_unique<ExpressionAnalyzer>(
query_ptr, syntax_analyzer_result, context, NamesAndTypesList(), required_result_column_names, subquery_depth, !only_analyze);
@ -792,7 +792,7 @@ void InterpreterSelectQuery::executeFetchColumns(
}
auto additional_source_columns_set = ext::map<NameSet>(additional_source_columns, [] (const auto & it) { return it.name; });
auto syntax_result = SyntaxAnalyzer(context, storage).analyze(required_columns_expr_list, additional_source_columns);
auto syntax_result = SyntaxAnalyzer(context).analyze(required_columns_expr_list, additional_source_columns, {}, storage);
alias_actions = ExpressionAnalyzer(required_columns_expr_list, syntax_result, context).getActions(true);
/// The set of required columns could be added as a result of adding an action to calculate ALIAS.
@ -829,7 +829,7 @@ void InterpreterSelectQuery::executeFetchColumns(
}
prewhere_info->prewhere_actions = std::move(new_actions);
auto analyzed_result = SyntaxAnalyzer(context, {}).analyze(required_prewhere_columns_expr_list, storage->getColumns().getAllPhysical());
auto analyzed_result = SyntaxAnalyzer(context).analyze(required_prewhere_columns_expr_list, storage->getColumns().getAllPhysical());
prewhere_info->alias_actions =
ExpressionAnalyzer(required_prewhere_columns_expr_list, analyzed_result, context)
.getActions(true, false);

3
dbms/src/Interpreters/InterpreterSystemQuery.cpp
View File

@ -21,6 +21,7 @@
#include <Parsers/ASTDropQuery.h>
#include <Parsers/ASTCreateQuery.h>
#include <csignal>
#include <algorithm>
namespace DB
@ -289,7 +290,7 @@ void InterpreterSystemQuery::restartReplicas(Context & system_context)
if (replica_names.empty())
return;
ThreadPool pool(std::min(getNumberOfPhysicalCPUCores(), replica_names.size()));
ThreadPool pool(std::min(size_t(getNumberOfPhysicalCPUCores()), replica_names.size()));
for (auto & table : replica_names)
pool.schedule([&] () { tryRestartReplica(table.first, table.second, system_context); });
pool.wait();

4
dbms/src/Interpreters/JoinToSubqueryTransformVisitor.cpp
View File

@ -83,8 +83,8 @@ static void appendTableNameAndAlias(std::vector<String> & hidden, const ASTPtr &
if (!alias.empty())
hidden.push_back(alias);
if (auto * identifier = typeid_cast<const ASTIdentifier *>(table_expression->children[0].get()))
hidden.push_back(identifier->name);
if (auto opt_name = getIdentifierName(table_expression->children[0]))
hidden.push_back(*opt_name);
else if (alias.empty())
throw Exception("Expected Identifier or subquery with alias", ErrorCodes::LOGICAL_ERROR);
}

7
dbms/src/Interpreters/MutationsInterpreter.cpp
View File

@ -194,7 +194,7 @@ void MutationsInterpreter::prepare(bool dry_run)
if (col_default.kind == ColumnDefaultKind::Materialized)
{
auto query = col_default.expression->clone();
auto syntax_result = SyntaxAnalyzer(context, {}).analyze(query, all_columns);
auto syntax_result = SyntaxAnalyzer(context).analyze(query, all_columns);
ExpressionAnalyzer analyzer(query, syntax_result, context);
for (const String & dependency : analyzer.getRequiredSourceColumns())
{
@ -203,10 +203,9 @@ void MutationsInterpreter::prepare(bool dry_run)
}
}
}
}
if (!updated_columns.empty())
validateUpdateColumns(storage, updated_columns, column_to_affected_materialized);
}
/// First, break a sequence of commands into stages.
stages.emplace_back(context);
@ -301,7 +300,7 @@ void MutationsInterpreter::prepare(bool dry_run)
for (const String & column : stage.output_columns)
all_asts->children.push_back(std::make_shared<ASTIdentifier>(column));
auto syntax_result = SyntaxAnalyzer(context, {}).analyze(all_asts, all_columns);
auto syntax_result = SyntaxAnalyzer(context).analyze(all_asts, all_columns);
stage.analyzer = std::make_unique<ExpressionAnalyzer>(all_asts, syntax_result, context);
ExpressionActionsChain & actions_chain = stage.expressions_chain;

108
dbms/src/Interpreters/OptimizeIfWithConstantConditionVisitor.cpp Normal file
View File

@ -0,0 +1,108 @@
#include <Common/typeid_cast.h>
#include <Parsers/ASTLiteral.h>
#include <Parsers/ASTFunction.h>
#include <Parsers/ASTExpressionList.h>
#include <Interpreters/OptimizeIfWithConstantConditionVisitor.h>
#include <IO/WriteHelpers.h>
namespace DB
{
namespace ErrorCodes
{
extern const int NUMBER_OF_ARGUMENTS_DOESNT_MATCH;
}
static bool tryExtractConstValueFromCondition(const ASTPtr & condition, bool & value)
{
/// numeric constant in condition
if (const ASTLiteral * literal = typeid_cast<ASTLiteral *>(condition.get()))
{
if (literal->value.getType() == Field::Types::Int64 ||
literal->value.getType() == Field::Types::UInt64)
{
value = literal->value.get<Int64>();
return true;
}
}
/// cast of numeric constant in condition to UInt8
if (const ASTFunction * function = typeid_cast<ASTFunction * >(condition.get()))
{
if (function->name == "CAST")
{
if (ASTExpressionList * expr_list = typeid_cast<ASTExpressionList *>(function->arguments.get()))
{
const ASTPtr & type_ast = expr_list->children.at(1);
if (const ASTLiteral * type_literal = typeid_cast<ASTLiteral *>(type_ast.get()))
{
if (type_literal->value.getType() == Field::Types::String &&
type_literal->value.get<std::string>() == "UInt8")
return tryExtractConstValueFromCondition(expr_list->children.at(0), value);
}
}
}
}
return false;
}
void OptimizeIfWithConstantConditionVisitor::visit(ASTPtr & current_ast)
{
if (!current_ast)
return;
for (ASTPtr & child : current_ast->children)
{
auto * function_node = typeid_cast<ASTFunction *>(child.get());
if (!function_node || function_node->name != "if")
{
visit(child);
continue;
}
visit(function_node->arguments);
auto * args = typeid_cast<ASTExpressionList *>(function_node->arguments.get());
if (args->children.size() != 3)
throw Exception("Wrong number of arguments for function 'if' (" + toString(args->children.size()) + " instead of 3)",
ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH);
ASTPtr condition_expr = args->children[0];
ASTPtr then_expr = args->children[1];
ASTPtr else_expr = args->children[2];
bool condition;
if (tryExtractConstValueFromCondition(condition_expr, condition))
{
ASTPtr replace_ast = condition ? then_expr : else_expr;
ASTPtr child_copy = child;
String replace_alias = replace_ast->tryGetAlias();
String if_alias = child->tryGetAlias();
if (replace_alias.empty())
{
replace_ast->setAlias(if_alias);
child = replace_ast;
}
else
{
/// Only copy of one node is required here.
/// But IAST has only method for deep copy of subtree.
/// This can be a reason of performance degradation in case of deep queries.
ASTPtr replace_ast_deep_copy = replace_ast->clone();
replace_ast_deep_copy->setAlias(if_alias);
child = replace_ast_deep_copy;
}
if (!if_alias.empty())
{
auto alias_it = aliases.find(if_alias);
if (alias_it != aliases.end() && alias_it->second.get() == child_copy.get())
alias_it->second = child;
}
}
}
}
}

27
dbms/src/Interpreters/OptimizeIfWithConstantConditionVisitor.h Normal file
View File

@ -0,0 +1,27 @@
#pragma once
#include <unordered_map>
#include <Parsers/IAST.h>
namespace DB
{
/// It removes Function_if node from AST if condition is constant.
/// TODO: rewrite with InDepthNodeVisitor
class OptimizeIfWithConstantConditionVisitor
{
public:
using Aliases = std::unordered_map<String, ASTPtr>;
OptimizeIfWithConstantConditionVisitor(Aliases & aliases_)
: aliases(aliases_)
{}
void visit(ASTPtr & ast);
private:
Aliases & aliases;
};
}

13
dbms/src/Interpreters/PredicateExpressionsOptimizer.cpp
View File

@ -14,6 +14,12 @@
namespace DB
{
namespace ErrorCodes
{
extern const int LOGICAL_ERROR;
extern const int UNKNOWN_ELEMENT_IN_AST;
}
static constexpr auto and_function_name = "and";
PredicateExpressionsOptimizer::PredicateExpressionsOptimizer(
@ -313,9 +319,12 @@ ASTs PredicateExpressionsOptimizer::getSelectQueryProjectionColumns(ASTPtr & ast
TranslateQualifiedNamesVisitor::Data qn_visitor_data{{}, tables};
TranslateQualifiedNamesVisitor(qn_visitor_data).visit(ast);
QueryAliasesVisitor::Data query_aliases_data{aliases};
QueryAliasesVisitor(query_aliases_data).visit(ast);
QueryNormalizer(ast, aliases, settings, {}, {}).perform();
QueryNormalizer::Data normalizer_data(aliases, settings);
QueryNormalizer(normalizer_data).visit(ast);
for (const auto & projection_column : select_query->select_expression_list->children)
{
@ -400,6 +409,8 @@ ASTs PredicateExpressionsOptimizer::evaluateAsterisk(ASTSelectQuery * select_que
DatabaseAndTableWithAlias database_and_table_name(*database_and_table_ast);
storage = context.getTable(database_and_table_name.database, database_and_table_name.table);
}
else
throw Exception("Logical error: unexpected table expression", ErrorCodes::LOGICAL_ERROR);
const auto block = storage->getSampleBlock();
for (size_t idx = 0; idx < block.columns(); idx++)

6
dbms/src/Interpreters/PredicateExpressionsOptimizer.h
View File

@ -14,12 +14,6 @@
namespace DB
{
namespace ErrorCodes
{
extern const int LOGICAL_ERROR;
extern const int NUMBER_OF_COLUMNS_DOESNT_MATCH;
}
using PredicateExpressions = std::vector<ASTPtr>;
using ProjectionWithAlias = std::pair<ASTPtr, String>;
using ProjectionsWithAliases = std::vector<ProjectionWithAlias>;

463
dbms/src/Interpreters/QueryNormalizer.cpp
View File

@ -20,39 +20,253 @@ namespace ErrorCodes
extern const int CYCLIC_ALIASES;
}
QueryNormalizer::QueryNormalizer(ASTPtr & query, const QueryNormalizer::Aliases & aliases,
ExtractedSettings && settings_, const Names & all_column_names,
const TableNamesAndColumnNames & table_names_and_column_names)
: query(query), aliases(aliases), settings(settings_), all_column_names(all_column_names),
table_names_and_column_names(table_names_and_column_names)
class CheckASTDepth
{
}
void QueryNormalizer::perform()
{
SetOfASTs tmp_set;
MapOfASTs tmp_map;
performImpl(query, tmp_map, tmp_set, "", 0);
try
public:
CheckASTDepth(QueryNormalizer::Data & data_)
: data(data_)
{
query->checkSize(settings.max_expanded_ast_elements);
if (data.level > data.settings.max_ast_depth)
throw Exception("Normalized AST is too deep. Maximum: " + toString(data.settings.max_ast_depth), ErrorCodes::TOO_DEEP_AST);
++data.level;
}
catch (Exception & e)
~CheckASTDepth()
{
e.addMessage("(after expansion of aliases)");
throw;
--data.level;
}
private:
QueryNormalizer::Data & data;
};
class RestoreAliasOnExitScope
{
public:
RestoreAliasOnExitScope(String & alias_)
: alias(alias_)
, copy(alias_)
{}
~RestoreAliasOnExitScope()
{
alias = copy;
}
private:
String & alias;
const String copy;
};
void QueryNormalizer::visit(ASTFunction & node, const ASTPtr &, Data & data)
{
auto & aliases = data.aliases;
String & func_name = node.name;
ASTPtr & func_arguments = node.arguments;
/// `IN t` can be specified, where t is a table, which is equivalent to `IN (SELECT * FROM t)`.
if (functionIsInOrGlobalInOperator(func_name))
{
auto & ast = func_arguments->children.at(1);
if (auto opt_name = getIdentifierName(ast))
if (!aliases.count(*opt_name))
setIdentifierSpecial(ast);
}
/// Special cases for count function.
String func_name_lowercase = Poco::toLower(func_name);
if (startsWith(func_name_lowercase, "count"))
{
/// Select implementation of countDistinct based on settings.
/// Important that it is done as query rewrite. It means rewritten query
/// will be sent to remote servers during distributed query execution,
/// and on all remote servers, function implementation will be same.
if (endsWith(func_name, "Distinct") && func_name_lowercase == "countdistinct")
func_name = data.settings.count_distinct_implementation;
/// As special case, treat count(*) as count(), not as count(list of all columns).
if (func_name_lowercase == "count" && func_arguments->children.size() == 1
&& typeid_cast<const ASTAsterisk *>(func_arguments->children[0].get()))
{
func_arguments->children.clear();
}
}
}
/// finished_asts - already processed vertices (and by what they replaced)
/// current_asts - vertices in the current call stack of this method
/// current_alias - the alias referencing to the ancestor of ast (the deepest ancestor with aliases)
void QueryNormalizer::performImpl(ASTPtr & ast, MapOfASTs & finished_asts, SetOfASTs & current_asts, std::string current_alias, size_t level)
void QueryNormalizer::visit(ASTIdentifier & node, ASTPtr & ast, Data & data)
{
if (level > settings.max_ast_depth)
throw Exception("Normalized AST is too deep. Maximum: " + toString(settings.max_ast_depth), ErrorCodes::TOO_DEEP_AST);
auto & current_asts = data.current_asts;
String & current_alias = data.current_alias;
if (!getColumnIdentifierName(node))
return;
/// If it is an alias, but not a parent alias (for constructs like "SELECT column + 1 AS column").
auto it_alias = data.aliases.find(node.name);
if (it_alias != data.aliases.end() && current_alias != node.name)
{
auto & alias_node = it_alias->second;
/// Let's replace it with the corresponding tree node.
if (current_asts.count(alias_node.get()))
throw Exception("Cyclic aliases", ErrorCodes::CYCLIC_ALIASES);
String my_alias = ast->tryGetAlias();
if (!my_alias.empty() && my_alias != alias_node->getAliasOrColumnName())
{
/// Avoid infinite recursion here
auto opt_name = getColumnIdentifierName(alias_node);
bool is_cycle = opt_name && *opt_name == node.name;
if (!is_cycle)
{
/// In a construct like "a AS b", where a is an alias, you must set alias b to the result of substituting alias a.
ast = alias_node->clone();
ast->setAlias(my_alias);
}
}
else
ast = alias_node;
}
}
/// Replace *, alias.*, database.table.* with a list of columns.
void QueryNormalizer::visit(ASTExpressionList & node, const ASTPtr &, Data & data)
{
auto & tables_with_columns = data.tables_with_columns;
ASTs old_children;
if (data.processAsterisks())
{
bool has_asterisk = false;
for (const auto & child : node.children)
{
if (typeid_cast<const ASTAsterisk *>(child.get()) ||
typeid_cast<const ASTQualifiedAsterisk *>(child.get()))
{
has_asterisk = true;
break;
}
}
if (has_asterisk)
{
old_children.swap(node.children);
node.children.reserve(old_children.size());
}
}
for (const auto & child : old_children)
{
if (typeid_cast<const ASTAsterisk *>(child.get()))
{
for (const auto & pr : tables_with_columns)
for (const auto & column_name : pr.second)
node.children.emplace_back(std::make_shared<ASTIdentifier>(column_name));
}
else if (const auto * qualified_asterisk = typeid_cast<const ASTQualifiedAsterisk *>(child.get()))
{
const ASTIdentifier * identifier = typeid_cast<const ASTIdentifier *>(qualified_asterisk->children[0].get());
size_t num_components = identifier->children.size();
for (const auto & [table_name, table_columns] : tables_with_columns)
{
if ((num_components == 2 /// database.table.*
&& !table_name.database.empty() /// This is normal (not a temporary) table.
&& static_cast<const ASTIdentifier &>(*identifier->children[0]).name == table_name.database
&& static_cast<const ASTIdentifier &>(*identifier->children[1]).name == table_name.table)
|| (num_components == 0 /// t.*
&& ((!table_name.table.empty() && identifier->name == table_name.table) /// table.*
|| (!table_name.alias.empty() && identifier->name == table_name.alias)))) /// alias.*
{
for (const auto & column_name : table_columns)
node.children.emplace_back(std::make_shared<ASTIdentifier>(column_name));
break;
}
}
}
else
node.children.emplace_back(child);
}
}
/// mark table identifiers as 'not columns'
void QueryNormalizer::visit(ASTTablesInSelectQueryElement & node, const ASTPtr &, Data &)
{
if (node.table_expression)
{
auto & expr = static_cast<ASTTableExpression &>(*node.table_expression);
setIdentifierSpecial(expr.database_and_table_name);
}
}
/// special visitChildren() for ASTSelectQuery
void QueryNormalizer::visit(ASTSelectQuery & select, const ASTPtr & ast, Data & data)
{
for (auto & child : ast->children)
{
if (typeid_cast<const ASTSelectQuery *>(child.get()) ||
typeid_cast<const ASTTableExpression *>(child.get()))
continue;
visit(child, data);
}
/// If the WHERE clause or HAVING consists of a single alias, the reference must be replaced not only in children,
/// but also in where_expression and having_expression.
if (select.prewhere_expression)
visit(select.prewhere_expression, data);
if (select.where_expression)
visit(select.where_expression, data);
if (select.having_expression)
visit(select.having_expression, data);
}
/// Don't go into subqueries.
/// Don't go into components of compound identifiers.
/// Don't go into select query. It processes children itself.
/// Do not go to the left argument of lambda expressions, so as not to replace the formal parameters
/// on aliases in expressions of the form 123 AS x, arrayMap(x -> 1, [2]).
void QueryNormalizer::visitChildren(const ASTPtr & node, Data & data)
{
ASTFunction * func_node = typeid_cast<ASTFunction *>(node.get());
if (func_node && func_node->name == "lambda")
{
/// We skip the first argument. We also assume that the lambda function can not have parameters.
for (size_t i = 1, size = func_node->arguments->children.size(); i < size; ++i)
{
auto & child = func_node->arguments->children[i];
if (typeid_cast<const ASTSelectQuery *>(child.get()) ||
typeid_cast<const ASTTableExpression *>(child.get()))
continue;
visit(child, data);
}
}
else if (!typeid_cast<ASTIdentifier *>(node.get()) &&
!typeid_cast<ASTSelectQuery *>(node.get()))
{
for (auto & child : node->children)
{
if (typeid_cast<const ASTSelectQuery *>(child.get()) ||
typeid_cast<const ASTTableExpression *>(child.get()))
continue;
visit(child, data);
}
}
}
void QueryNormalizer::visit(ASTPtr & ast, Data & data)
{
CheckASTDepth scope1(data);
RestoreAliasOnExitScope scope2(data.current_alias);
auto & finished_asts = data.finished_asts;
auto & current_asts = data.current_asts;
if (finished_asts.count(ast))
{
@ -63,185 +277,46 @@ void QueryNormalizer::performImpl(ASTPtr & ast, MapOfASTs & finished_asts, SetOf
ASTPtr initial_ast = ast;
current_asts.insert(initial_ast.get());
String my_alias = ast->tryGetAlias();
if (!my_alias.empty())
current_alias = my_alias;
/// rewrite rules that act when you go from top to bottom.
bool replaced = false;
ASTIdentifier * identifier_node = nullptr;
ASTFunction * func_node = nullptr;
if ((func_node = typeid_cast<ASTFunction *>(ast.get())))
{
/// `IN t` can be specified, where t is a table, which is equivalent to `IN (SELECT * FROM t)`.
if (functionIsInOrGlobalInOperator(func_node->name))
if (ASTIdentifier * right = typeid_cast<ASTIdentifier *>(func_node->arguments->children.at(1).get()))
if (!aliases.count(right->name))
right->setSpecial();
/// Special cases for count function.
String func_name_lowercase = Poco::toLower(func_node->name);
if (startsWith(func_name_lowercase, "count"))
{
/// Select implementation of countDistinct based on settings.
/// Important that it is done as query rewrite. It means rewritten query
/// will be sent to remote servers during distributed query execution,
/// and on all remote servers, function implementation will be same.
if (endsWith(func_node->name, "Distinct") && func_name_lowercase == "countdistinct")
func_node->name = settings.count_distinct_implementation;
/// As special case, treat count(*) as count(), not as count(list of all columns).
if (func_name_lowercase == "count" && func_node->arguments->children.size() == 1
&& typeid_cast<const ASTAsterisk *>(func_node->arguments->children[0].get()))
{
func_node->arguments->children.clear();
}
}
String my_alias = ast->tryGetAlias();
if (!my_alias.empty())
data.current_alias = my_alias;
}
else if ((identifier_node = typeid_cast<ASTIdentifier *>(ast.get())))
{
if (identifier_node->general())
{
/// If it is an alias, but not a parent alias (for constructs like "SELECT column + 1 AS column").
auto it_alias = aliases.find(identifier_node->name);
if (it_alias != aliases.end() && current_alias != identifier_node->name)
{
/// Let's replace it with the corresponding tree node.
if (current_asts.count(it_alias->second.get()))
throw Exception("Cyclic aliases", ErrorCodes::CYCLIC_ALIASES);
if (!my_alias.empty() && my_alias != it_alias->second->getAliasOrColumnName())
{
/// Avoid infinite recursion here
auto replace_to_identifier = typeid_cast<ASTIdentifier *>(it_alias->second.get());
bool is_cycle = replace_to_identifier && replace_to_identifier->general()
&& replace_to_identifier->name == identifier_node->name;
if (!is_cycle)
{
/// In a construct like "a AS b", where a is an alias, you must set alias b to the result of substituting alias a.
ast = it_alias->second->clone();
ast->setAlias(my_alias);
replaced = true;
}
}
else
{
ast = it_alias->second;
replaced = true;
}
}
}
}
else if (ASTExpressionList * expr_list = typeid_cast<ASTExpressionList *>(ast.get()))
{
/// Replace *, alias.*, database.table.* with a list of columns.
ASTs & asts = expr_list->children;
for (ssize_t expr_idx = asts.size() - 1; expr_idx >= 0; --expr_idx)
{
if (typeid_cast<const ASTAsterisk *>(asts[expr_idx].get()) && !all_column_names.empty())
{
asts.erase(asts.begin() + expr_idx);
for (size_t column_idx = 0; column_idx < all_column_names.size(); ++column_idx)
asts.insert(asts.begin() + column_idx + expr_idx, std::make_shared<ASTIdentifier>(all_column_names[column_idx]));
}
else if (typeid_cast<const ASTQualifiedAsterisk *>(asts[expr_idx].get()) && !table_names_and_column_names.empty())
{
const ASTQualifiedAsterisk * qualified_asterisk = static_cast<const ASTQualifiedAsterisk *>(asts[expr_idx].get());
const ASTIdentifier * identifier = typeid_cast<const ASTIdentifier *>(qualified_asterisk->children[0].get());
size_t num_components = identifier->children.size();
for (const auto & [table_name, table_all_column_names] : table_names_and_column_names)
{
if ((num_components == 2 /// database.table.*
&& !table_name.database.empty() /// This is normal (not a temporary) table.
&& static_cast<const ASTIdentifier &>(*identifier->children[0]).name == table_name.database
&& static_cast<const ASTIdentifier &>(*identifier->children[1]).name == table_name.table)
|| (num_components == 0 /// t.*
&& ((!table_name.table.empty() && identifier->name == table_name.table) /// table.*
|| (!table_name.alias.empty() && identifier->name == table_name.alias)))) /// alias.*
{
asts.erase(asts.begin() + expr_idx);
for (size_t column_idx = 0; column_idx < table_all_column_names.size(); ++column_idx)
asts.insert(asts.begin() + column_idx + expr_idx, std::make_shared<ASTIdentifier>(table_all_column_names[column_idx]));
break;
}
}
}
}
}
else if (ASTTablesInSelectQueryElement * tables_elem = typeid_cast<ASTTablesInSelectQueryElement *>(ast.get()))
{
if (tables_elem->table_expression)
{
auto & database_and_table_name = static_cast<ASTTableExpression &>(*tables_elem->table_expression).database_and_table_name;
if (database_and_table_name)
{
if (ASTIdentifier * right = typeid_cast<ASTIdentifier *>(database_and_table_name.get()))
right->setSpecial();
}
}
}
if (auto * node = typeid_cast<ASTFunction *>(ast.get()))
visit(*node, ast, data);
if (auto * node = typeid_cast<ASTIdentifier *>(ast.get()))
visit(*node, ast, data);
if (auto * node = typeid_cast<ASTExpressionList *>(ast.get()))
visit(*node, ast, data);
if (auto * node = typeid_cast<ASTTablesInSelectQueryElement *>(ast.get()))
visit(*node, ast, data);
if (auto * node = typeid_cast<ASTSelectQuery *>(ast.get()))
visit(*node, ast, data);
/// If we replace the root of the subtree, we will be called again for the new root, in case the alias is replaced by an alias.
if (replaced)
{
performImpl(ast, finished_asts, current_asts, current_alias, level + 1);
current_asts.erase(initial_ast.get());
current_asts.erase(ast.get());
finished_asts[initial_ast] = ast;
return;
}
/// Recurring calls. Don't go into subqueries. Don't go into components of compound identifiers.
/// We also do not go to the left argument of lambda expressions, so as not to replace the formal parameters
/// on aliases in expressions of the form 123 AS x, arrayMap(x -> 1, [2]).
if (func_node && func_node->name == "lambda")
{
/// We skip the first argument. We also assume that the lambda function can not have parameters.
for (size_t i = 1, size = func_node->arguments->children.size(); i < size; ++i)
{
auto & child = func_node->arguments->children[i];
if (typeid_cast<const ASTSelectQuery *>(child.get()) || typeid_cast<const ASTTableExpression *>(child.get()))
continue;
performImpl(child, finished_asts, current_asts, current_alias, level + 1);
}
}
else if (identifier_node)
{
}
if (ast.get() != initial_ast.get())
visit(ast, data);
else
{
for (auto & child : ast->children)
{
if (typeid_cast<const ASTSelectQuery *>(child.get()) || typeid_cast<const ASTTableExpression *>(child.get()))
continue;
performImpl(child, finished_asts, current_asts, current_alias, level + 1);
}
}
/// If the WHERE clause or HAVING consists of a single alias, the reference must be replaced not only in children, but also in where_expression and having_expression.
if (ASTSelectQuery * select = typeid_cast<ASTSelectQuery *>(ast.get()))
{
if (select->prewhere_expression)
performImpl(select->prewhere_expression, finished_asts, current_asts, current_alias, level + 1);
if (select->where_expression)
performImpl(select->where_expression, finished_asts, current_asts, current_alias, level + 1);
if (select->having_expression)
performImpl(select->having_expression, finished_asts, current_asts, current_alias, level + 1);
}
visitChildren(ast, data);
current_asts.erase(initial_ast.get());
current_asts.erase(ast.get());
finished_asts[initial_ast] = ast;
/// @note can not place it in CheckASTDepth dtor cause of exception.
if (data.level == 1)
{
try
{
ast->checkSize(data.settings.max_expanded_ast_elements);
}
catch (Exception & e)
{
e.addMessage("(after expansion of aliases)");
throw;
}
}
}
}

59
dbms/src/Interpreters/QueryNormalizer.h
View File

@ -18,8 +18,10 @@ inline bool functionIsInOrGlobalInOperator(const String & name)
}
using TableNameAndColumnNames = std::pair<DatabaseAndTableWithAlias, Names>;
using TableNamesAndColumnNames = std::vector<TableNameAndColumnNames>;
class ASTFunction;
class ASTIdentifier;
class ASTExpressionList;
struct ASTTablesInSelectQueryElement;
class QueryNormalizer
@ -41,23 +43,54 @@ class QueryNormalizer
public:
using Aliases = std::unordered_map<String, ASTPtr>;
using TableWithColumnNames = std::pair<DatabaseAndTableWithAlias, Names>;
QueryNormalizer(ASTPtr & query, const Aliases & aliases, ExtractedSettings && settings, const Names & all_columns_name,
const TableNamesAndColumnNames & table_names_and_column_names);
struct Data
{
using SetOfASTs = std::set<const IAST *>;
using MapOfASTs = std::map<ASTPtr, ASTPtr>;
void perform();
const Aliases & aliases;
const ExtractedSettings settings;
const std::vector<TableWithColumnNames> tables_with_columns;
/// tmp data
size_t level;
MapOfASTs finished_asts; /// already processed vertices (and by what they replaced)
SetOfASTs current_asts; /// vertices in the current call stack of this method
std::string current_alias; /// the alias referencing to the ancestor of ast (the deepest ancestor with aliases)
Data(const Aliases & aliases_, ExtractedSettings && settings_, std::vector<TableWithColumnNames> && tables_with_columns_ = {})
: aliases(aliases_)
, settings(settings_)
, tables_with_columns(tables_with_columns_)
, level(0)
{}
bool processAsterisks() const { return !tables_with_columns.empty(); }
};
QueryNormalizer(Data & data)
: visitor_data(data)
{}
void visit(ASTPtr & ast)
{
visit(ast, visitor_data);
}
private:
using SetOfASTs = std::set<const IAST *>;
using MapOfASTs = std::map<ASTPtr, ASTPtr>;
Data & visitor_data;
ASTPtr & query;
const Aliases & aliases;
const ExtractedSettings settings;
const Names & all_column_names;
const TableNamesAndColumnNames & table_names_and_column_names;
static void visit(ASTPtr & query, Data & data);
void performImpl(ASTPtr & ast, MapOfASTs & finished_asts, SetOfASTs & current_asts, std::string current_alias, size_t level);
static void visit(ASTIdentifier &, ASTPtr &, Data &);
static void visit(ASTFunction &, const ASTPtr &, Data &);
static void visit(ASTExpressionList &, const ASTPtr &, Data &);
static void visit(ASTTablesInSelectQueryElement &, const ASTPtr &, Data &);
static void visit(ASTSelectQuery &, const ASTPtr &, Data &);
static void visitChildren(const ASTPtr &, Data & data);
};
}

2
dbms/src/Interpreters/SetVariants.cpp
View File

@ -132,7 +132,7 @@ typename SetVariantsTemplate<Variant>::Type SetVariantsTemplate<Variant>::choose
}
/// If there is one numeric key that fits into 64 bits
if (keys_size == 1 && nested_key_columns[0]->isNumeric())
if (keys_size == 1 && nested_key_columns[0]->isNumeric() && !nested_key_columns[0]->lowCardinality())
{
size_t size_of_field = nested_key_columns[0]->sizeOfValueIfFixed();
if (size_of_field == 1)

430
dbms/src/Interpreters/SyntaxAnalyzer.cpp
View File

@ -11,6 +11,7 @@
#include <Interpreters/ExecuteScalarSubqueriesVisitor.h>
#include <Interpreters/PredicateExpressionsOptimizer.h>
#include <Interpreters/ExternalDictionaries.h>
#include <Interpreters/OptimizeIfWithConstantConditionVisitor.h>
#include <Parsers/ASTSelectQuery.h>
#include <Parsers/ASTLiteral.h>
@ -34,7 +35,6 @@ namespace DB
namespace ErrorCodes
{
extern const int NUMBER_OF_ARGUMENTS_DOESNT_MATCH;
extern const int ALIAS_REQUIRED;
extern const int MULTIPLE_EXPRESSIONS_FOR_ALIAS;
extern const int EMPTY_NESTED_TABLE;
@ -42,144 +42,9 @@ namespace ErrorCodes
extern const int INVALID_JOIN_ON_EXPRESSION;
}
namespace
NameSet removeDuplicateColumns(NamesAndTypesList & columns)
{
using LogAST = DebugASTLog<false>; /// set to true to enable logs
using Aliases = SyntaxAnalyzerResult::Aliases;
/// Add columns from storage to source_columns list.
void collectSourceColumns(ASTSelectQuery * select_query, const Context & context,
StoragePtr & storage, NamesAndTypesList & source_columns);
/// Translate qualified names such as db.table.column, table.column, table_alias.column to unqualified names.
void translateQualifiedNames(ASTPtr & query, ASTSelectQuery * select_query,
const NameSet & source_columns, const Context & context);
/// For star nodes(`*`), expand them to a list of all columns. For literal nodes, substitute aliases.
void normalizeTree(
ASTPtr & query,
SyntaxAnalyzerResult & result,
const Names & source_columns,
const NameSet & source_columns_set,
const StoragePtr & storage,
const Context & context,
const ASTSelectQuery * select_query,
bool asterisk_left_columns_only);
/// Sometimes we have to calculate more columns in SELECT clause than will be returned from query.
/// This is the case when we have DISTINCT or arrayJoin: we require more columns in SELECT even if we need less columns in result.
void removeUnneededColumnsFromSelectClause(const ASTSelectQuery * select_query, const Names & required_result_columns);
/// Replacing scalar subqueries with constant values.
void executeScalarSubqueries(ASTPtr & query, const ASTSelectQuery * select_query,
const Context & context, size_t subquery_depth);
/// Remove Function_if AST if condition is constant.
void optimizeIfWithConstantCondition(ASTPtr & current_ast, Aliases & aliases);
/// Eliminates injective function calls and constant expressions from group by statement.
void optimizeGroupBy(ASTSelectQuery * select_query, const NameSet & source_columns, const Context & context);
/// Remove duplicate items from ORDER BY.
void optimizeOrderBy(const ASTSelectQuery * select_query);
/// Remove duplicate items from LIMIT BY.
void optimizeLimitBy(const ASTSelectQuery * select_query);
/// Remove duplicated columns from USING(...).
void optimizeUsing(const ASTSelectQuery * select_query);
void getArrayJoinedColumns(ASTPtr & query, SyntaxAnalyzerResult & result, const ASTSelectQuery * select_query,
const Names & source_columns, const NameSet & source_columns_set);
/// Parse JOIN ON expression and collect ASTs for joined columns.
void collectJoinedColumnsFromJoinOnExpr(AnalyzedJoin & analyzed_join, const ASTSelectQuery * select_query,
const NameSet & source_columns, const Context & context);
/// Find the columns that are obtained by JOIN.
void collectJoinedColumns(AnalyzedJoin & analyzed_join, const ASTSelectQuery * select_query,
const NameSet & source_columns, const Context & context);
}
SyntaxAnalyzerResultPtr SyntaxAnalyzer::analyze(
ASTPtr & query,
const NamesAndTypesList & source_columns_,
const Names & required_result_columns,
size_t subquery_depth) const
{
SyntaxAnalyzerResult result;
result.storage = storage;
result.source_columns = source_columns_;
auto * select_query = typeid_cast<ASTSelectQuery *>(query.get());
collectSourceColumns(select_query, context, result.storage, result.source_columns);
const auto & settings = context.getSettingsRef();
Names source_columns_list;
source_columns_list.reserve(result.source_columns.size());
for (const auto & type_name : result.source_columns)
source_columns_list.emplace_back(type_name.name);
NameSet source_columns_set(source_columns_list.begin(), source_columns_list.end());
translateQualifiedNames(query, select_query, source_columns_set, context);
/// Depending on the user's profile, check for the execution rights
/// distributed subqueries inside the IN or JOIN sections and process these subqueries.
InJoinSubqueriesPreprocessor(context).process(select_query);
/// Optimizes logical expressions.
LogicalExpressionsOptimizer(select_query, settings.optimize_min_equality_disjunction_chain_length.value).perform();
/// Creates a dictionary `aliases`: alias -> ASTPtr
{
LogAST log;
QueryAliasesVisitor::Data query_aliases_data{result.aliases};
QueryAliasesVisitor(query_aliases_data, log.stream()).visit(query);
}
/// Common subexpression elimination. Rewrite rules.
normalizeTree(query, result, source_columns_list, source_columns_set, result.storage,
context, select_query, settings.asterisk_left_columns_only != 0);
/// Remove unneeded columns according to 'required_result_columns'.
/// Leave all selected columns in case of DISTINCT; columns that contain arrayJoin function inside.
/// Must be after 'normalizeTree' (after expanding aliases, for aliases not get lost)
/// and before 'executeScalarSubqueries', 'analyzeAggregation', etc. to avoid excessive calculations.
removeUnneededColumnsFromSelectClause(select_query, required_result_columns);
/// Executing scalar subqueries - replacing them with constant values.
executeScalarSubqueries(query, select_query, context, subquery_depth);
/// Optimize if with constant condition after constants was substituted instead of sclalar subqueries.
optimizeIfWithConstantCondition(query, result.aliases);
/// GROUP BY injective function elimination.
optimizeGroupBy(select_query, source_columns_set, context);
/// Remove duplicate items from ORDER BY.
optimizeOrderBy(select_query);
// Remove duplicated elements from LIMIT BY clause.
optimizeLimitBy(select_query);
/// Remove duplicated columns from USING(...).
optimizeUsing(select_query);
/// array_join_alias_to_name, array_join_result_to_source.
getArrayJoinedColumns(query, result, select_query, source_columns_list, source_columns_set);
/// Push the predicate expression down to the subqueries.
result.rewrite_subqueries = PredicateExpressionsOptimizer(select_query, settings, context).optimize();
collectJoinedColumns(result.analyzed_join, select_query, source_columns_set, context);
return std::make_shared<const SyntaxAnalyzerResult>(result);
}
void removeDuplicateColumns(NamesAndTypesList & columns)
{
std::set<String> names;
NameSet names;
for (auto it = columns.begin(); it != columns.end();)
{
if (names.emplace(it->name).second)
@ -187,20 +52,18 @@ void removeDuplicateColumns(NamesAndTypesList & columns)
else
columns.erase(it++);
}
return names;
}
namespace
{
void collectSourceColumns(ASTSelectQuery * select_query, const Context & context,
StoragePtr & storage, NamesAndTypesList & source_columns)
{
if (!storage && select_query)
{
if (auto db_and_table = getDatabaseAndTable(*select_query, 0))
storage = context.tryGetTable(db_and_table->database, db_and_table->table);
}
using LogAST = DebugASTLog<false>; /// set to true to enable logs
/// Add columns from storage to source_columns list.
void collectSourceColumns(ASTSelectQuery * select_query, StoragePtr storage, NamesAndTypesList & source_columns)
{
if (storage)
{
auto physical_columns = storage->getColumns().getAllPhysical();
@ -215,14 +78,13 @@ void collectSourceColumns(ASTSelectQuery * select_query, const Context & context
source_columns.insert(source_columns.end(), storage_aliases.begin(), storage_aliases.end());
}
}
removeDuplicateColumns(source_columns);
}
/// Translate qualified names such as db.table.column, table.column, table_alias.column to unqualified names.
void translateQualifiedNames(ASTPtr & query, ASTSelectQuery * select_query,
const NameSet & source_columns, const Context & context)
{
if (!select_query || !select_query->tables || select_query->tables->children.empty())
if (!select_query->tables || select_query->tables->children.empty())
return;
std::vector<DatabaseAndTableWithAlias> tables = getDatabaseAndTables(*select_query, context.getCurrentDatabase());
@ -233,17 +95,17 @@ void translateQualifiedNames(ASTPtr & query, ASTSelectQuery * select_query,
visitor.visit(query);
}
/// For star nodes(`*`), expand them to a list of all columns. For literal nodes, substitute aliases.
void normalizeTree(
ASTPtr & query,
SyntaxAnalyzerResult & result,
const Names & source_columns,
const NameSet & source_columns_set,
const StoragePtr & storage,
const Context & context,
const ASTSelectQuery * select_query,
bool asterisk_left_columns_only)
{
Names all_columns_name = storage ? storage->getColumns().ordinary.getNames() : source_columns;
Names all_columns_name = source_columns;
if (!asterisk_left_columns_only)
{
@ -255,17 +117,20 @@ void normalizeTree(
if (all_columns_name.empty())
throw Exception("An asterisk cannot be replaced with empty columns.", ErrorCodes::LOGICAL_ERROR);
TableNamesAndColumnNames table_names_and_column_names;
std::vector<QueryNormalizer::TableWithColumnNames> table_with_columns;
if (select_query && select_query->tables && !select_query->tables->children.empty())
{
std::vector<const ASTTableExpression *> tables_expression = getSelectTablesExpression(*select_query);
bool first = true;
String current_database = context.getCurrentDatabase();
for (const auto * table_expression : tables_expression)
{
DatabaseAndTableWithAlias table_name(*table_expression, context.getCurrentDatabase());
DatabaseAndTableWithAlias table_name(*table_expression, current_database);
NamesAndTypesList names_and_types = getNamesAndTypeListFromTableExpression(*table_expression, context);
removeDuplicateColumns(names_and_types);
if (!first)
{
/// For joined tables qualify duplicating names.
@ -276,12 +141,14 @@ void normalizeTree(
first = false;
table_names_and_column_names.emplace_back(std::pair(table_name, names_and_types.getNames()));
table_with_columns.emplace_back(std::move(table_name), names_and_types.getNames());
}
}
else
table_with_columns.emplace_back(DatabaseAndTableWithAlias{}, std::move(all_columns_name));
auto & settings = context.getSettingsRef();
QueryNormalizer(query, result.aliases, settings, all_columns_name, table_names_and_column_names).perform();
QueryNormalizer::Data normalizer_data(result.aliases, context.getSettingsRef(), std::move(table_with_columns));
QueryNormalizer(normalizer_data).visit(query);
}
bool hasArrayJoin(const ASTPtr & ast)
@ -297,11 +164,10 @@ bool hasArrayJoin(const ASTPtr & ast)
return false;
}
/// Sometimes we have to calculate more columns in SELECT clause than will be returned from query.
/// This is the case when we have DISTINCT or arrayJoin: we require more columns in SELECT even if we need less columns in result.
void removeUnneededColumnsFromSelectClause(const ASTSelectQuery * select_query, const Names & required_result_columns)
{
if (!select_query)
return;
if (required_result_columns.empty())
return;
@ -335,121 +201,12 @@ void removeUnneededColumnsFromSelectClause(const ASTSelectQuery * select_query,
elements = std::move(new_elements);
}
void executeScalarSubqueries(ASTPtr & query, const ASTSelectQuery * select_query,
const Context & context, size_t subquery_depth)
/// Replacing scalar subqueries with constant values.
void executeScalarSubqueries(ASTPtr & query, const Context & context, size_t subquery_depth)
{
LogAST log;
if (!select_query)
{
ExecuteScalarSubqueriesVisitor::Data visitor_data{context, subquery_depth};
ExecuteScalarSubqueriesVisitor(visitor_data, log.stream()).visit(query);
}
else
{
for (auto & child : query->children)
{
/// Do not go to FROM, JOIN, UNION.
if (!typeid_cast<const ASTTableExpression *>(child.get())
&& !typeid_cast<const ASTSelectQuery *>(child.get()))
{
ExecuteScalarSubqueriesVisitor::Data visitor_data{context, subquery_depth};
ExecuteScalarSubqueriesVisitor(visitor_data, log.stream()).visit(child);
}
}
}
}
bool tryExtractConstValueFromCondition(const ASTPtr & condition, bool & value)
{
/// numeric constant in condition
if (const ASTLiteral * literal = typeid_cast<ASTLiteral *>(condition.get()))
{
if (literal->value.getType() == Field::Types::Int64 ||
literal->value.getType() == Field::Types::UInt64)
{
value = literal->value.get<Int64>();
return true;
}
}
/// cast of numeric constant in condition to UInt8
if (const ASTFunction * function = typeid_cast<ASTFunction * >(condition.get()))
{
if (function->name == "CAST")
{
if (ASTExpressionList * expr_list = typeid_cast<ASTExpressionList *>(function->arguments.get()))
{
const ASTPtr & type_ast = expr_list->children.at(1);
if (const ASTLiteral * type_literal = typeid_cast<ASTLiteral *>(type_ast.get()))
{
if (type_literal->value.getType() == Field::Types::String &&
type_literal->value.get<std::string>() == "UInt8")
return tryExtractConstValueFromCondition(expr_list->children.at(0), value);
}
}
}
}
return false;
}
void optimizeIfWithConstantCondition(ASTPtr & current_ast, Aliases & aliases)
{
if (!current_ast)
return;
for (ASTPtr & child : current_ast->children)
{
auto * function_node = typeid_cast<ASTFunction *>(child.get());
if (!function_node || function_node->name != "if")
{
optimizeIfWithConstantCondition(child, aliases);
continue;
}
optimizeIfWithConstantCondition(function_node->arguments, aliases);
auto * args = typeid_cast<ASTExpressionList *>(function_node->arguments.get());
if (args->children.size() != 3)
throw Exception("Wrong number of arguments for function 'if' (" + toString(args->children.size()) + " instead of 3)",
ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH);
ASTPtr condition_expr = args->children[0];
ASTPtr then_expr = args->children[1];
ASTPtr else_expr = args->children[2];
bool condition;
if (tryExtractConstValueFromCondition(condition_expr, condition))
{
ASTPtr replace_ast = condition ? then_expr : else_expr;
ASTPtr child_copy = child;
String replace_alias = replace_ast->tryGetAlias();
String if_alias = child->tryGetAlias();
if (replace_alias.empty())
{
replace_ast->setAlias(if_alias);
child = replace_ast;
}
else
{
/// Only copy of one node is required here.
/// But IAST has only method for deep copy of subtree.
/// This can be a reason of performance degradation in case of deep queries.
ASTPtr replace_ast_deep_copy = replace_ast->clone();
replace_ast_deep_copy->setAlias(if_alias);
child = replace_ast_deep_copy;
}
if (!if_alias.empty())
{
auto alias_it = aliases.find(if_alias);
if (alias_it != aliases.end() && alias_it->second.get() == child_copy.get())
alias_it->second = child;
}
}
}
ExecuteScalarSubqueriesVisitor::Data visitor_data{context, subquery_depth};
ExecuteScalarSubqueriesVisitor(visitor_data, log.stream()).visit(query);
}
/** Calls to these functions in the GROUP BY statement would be
@ -491,9 +248,10 @@ const std::unordered_set<String> possibly_injective_function_names
"dictGetDateTime"
};
/// Eliminates injective function calls and constant expressions from group by statement.
void optimizeGroupBy(ASTSelectQuery * select_query, const NameSet & source_columns, const Context & context)
{
if (!(select_query && select_query->group_expression_list))
if (!select_query->group_expression_list)
return;
const auto is_literal = [] (const ASTPtr & ast)
@ -594,9 +352,10 @@ void optimizeGroupBy(ASTSelectQuery * select_query, const NameSet & source_colum
}
}
/// Remove duplicate items from ORDER BY.
void optimizeOrderBy(const ASTSelectQuery * select_query)
{
if (!(select_query && select_query->order_expression_list))
if (!select_query->order_expression_list)
return;
/// Make unique sorting conditions.
@ -620,9 +379,10 @@ void optimizeOrderBy(const ASTSelectQuery * select_query)
elems = unique_elems;
}
/// Remove duplicate items from LIMIT BY.
void optimizeLimitBy(const ASTSelectQuery * select_query)
{
if (!(select_query && select_query->limit_by_expression_list))
if (!select_query->limit_by_expression_list)
return;
std::set<String> elems_set;
@ -641,11 +401,9 @@ void optimizeLimitBy(const ASTSelectQuery * select_query)
elems = unique_elems;
}
/// Remove duplicated columns from USING(...).
void optimizeUsing(const ASTSelectQuery * select_query)
{
if (!select_query)
return;
auto node = const_cast<ASTTablesInSelectQueryElement *>(select_query->join());
if (!node)
return;
@ -676,9 +434,6 @@ void optimizeUsing(const ASTSelectQuery * select_query)
void getArrayJoinedColumns(ASTPtr & query, SyntaxAnalyzerResult & result, const ASTSelectQuery * select_query,
const Names & source_columns, const NameSet & source_columns_set)
{
if (!select_query)
return;
ASTPtr array_join_expression_list = select_query->array_join_expression_list();
if (array_join_expression_list)
{
@ -688,7 +443,7 @@ void getArrayJoinedColumns(ASTPtr & query, SyntaxAnalyzerResult & result, const
const String nested_table_name = ast->getColumnName();
const String nested_table_alias = ast->getAliasOrColumnName();
if (nested_table_alias == nested_table_name && !typeid_cast<const ASTIdentifier *>(ast.get()))
if (nested_table_alias == nested_table_name && !isIdentifier(ast))
throw Exception("No alias for non-trivial value in ARRAY JOIN: " + nested_table_name,
ErrorCodes::ALIAS_REQUIRED);
@ -716,7 +471,7 @@ void getArrayJoinedColumns(ASTPtr & query, SyntaxAnalyzerResult & result, const
String result_name = expr->getAliasOrColumnName();
/// This is an array.
if (!typeid_cast<ASTIdentifier *>(expr.get()) || source_columns_set.count(source_name))
if (!isIdentifier(expr) || source_columns_set.count(source_name))
{
result.array_join_result_to_source[result_name] = source_name;
}
@ -740,6 +495,7 @@ void getArrayJoinedColumns(ASTPtr & query, SyntaxAnalyzerResult & result, const
}
}
/// Parse JOIN ON expression and collect ASTs for joined columns.
void collectJoinedColumnsFromJoinOnExpr(AnalyzedJoin & analyzed_join, const ASTSelectQuery * select_query,
const NameSet & source_columns, const Context & context)
{
@ -772,10 +528,10 @@ void collectJoinedColumnsFromJoinOnExpr(AnalyzedJoin & analyzed_join, const ASTS
std::function<TableBelonging(const ASTPtr &)> get_table_belonging;
get_table_belonging = [&](const ASTPtr & ast) -> TableBelonging
{
auto * identifier = typeid_cast<const ASTIdentifier *>(ast.get());
if (identifier)
if (getColumnIdentifierName(ast))
{
if (identifier->general())
auto * identifier = typeid_cast<const ASTIdentifier *>(ast.get());
{
auto left_num_components = getNumComponentsToStripInOrderToTranslateQualifiedName(*identifier, left_source_names);
auto right_num_components = getNumComponentsToStripInOrderToTranslateQualifiedName(*identifier, right_source_names);
@ -811,9 +567,10 @@ void collectJoinedColumnsFromJoinOnExpr(AnalyzedJoin & analyzed_join, const ASTS
std::function<void(ASTPtr &, const DatabaseAndTableWithAlias &, bool)> translate_qualified_names;
translate_qualified_names = [&](ASTPtr & ast, const DatabaseAndTableWithAlias & source_names, bool right_table)
{
if (auto * identifier = typeid_cast<const ASTIdentifier *>(ast.get()))
if (getColumnIdentifierName(ast))
{
if (identifier->general())
auto * identifier = typeid_cast<const ASTIdentifier *>(ast.get());
{
auto num_components = getNumComponentsToStripInOrderToTranslateQualifiedName(*identifier, source_names);
stripIdentifier(ast, num_components);
@ -899,12 +656,10 @@ void collectJoinedColumnsFromJoinOnExpr(AnalyzedJoin & analyzed_join, const ASTS
add_columns_from_equals_expr(table_join.on_expression);
}
/// Find the columns that are obtained by JOIN.
void collectJoinedColumns(AnalyzedJoin & analyzed_join, const ASTSelectQuery * select_query,
const NameSet & source_columns, const Context & context)
{
if (!select_query)
return;
const ASTTablesInSelectQueryElement * node = select_query->join();
if (!node)
@ -969,4 +724,97 @@ void collectJoinedColumns(AnalyzedJoin & analyzed_join, const ASTSelectQuery * s
}
SyntaxAnalyzerResultPtr SyntaxAnalyzer::analyze(
ASTPtr & query,
const NamesAndTypesList & source_columns_,
const Names & required_result_columns,
StoragePtr storage) const
{
auto * select_query = typeid_cast<ASTSelectQuery *>(query.get());
if (!storage && select_query)
{
if (auto db_and_table = getDatabaseAndTable(*select_query, 0))
storage = context.tryGetTable(db_and_table->database, db_and_table->table);
}
SyntaxAnalyzerResult result;
result.storage = storage;
result.source_columns = source_columns_;
collectSourceColumns(select_query, result.storage, result.source_columns);
NameSet source_columns_set = removeDuplicateColumns(result.source_columns);
const auto & settings = context.getSettingsRef();
Names source_columns_list;
source_columns_list.reserve(result.source_columns.size());
for (const auto & type_name : result.source_columns)
source_columns_list.emplace_back(type_name.name);
if (source_columns_set.size() != source_columns_list.size())
throw Exception("Unexpected duplicates in source columns list.", ErrorCodes::LOGICAL_ERROR);
if (select_query)
{
translateQualifiedNames(query, select_query, source_columns_set, context);
/// Depending on the user's profile, check for the execution rights
/// distributed subqueries inside the IN or JOIN sections and process these subqueries.
InJoinSubqueriesPreprocessor(context).process(select_query);
/// Optimizes logical expressions.
LogicalExpressionsOptimizer(select_query, settings.optimize_min_equality_disjunction_chain_length.value).perform();
}
/// Creates a dictionary `aliases`: alias -> ASTPtr
{
LogAST log;
QueryAliasesVisitor::Data query_aliases_data{result.aliases};
QueryAliasesVisitor(query_aliases_data, log.stream()).visit(query);
}
/// Common subexpression elimination. Rewrite rules.
normalizeTree(query, result, (storage ? storage->getColumns().ordinary.getNames() : source_columns_list), source_columns_set,
context, select_query, settings.asterisk_left_columns_only != 0);
/// Remove unneeded columns according to 'required_result_columns'.
/// Leave all selected columns in case of DISTINCT; columns that contain arrayJoin function inside.
/// Must be after 'normalizeTree' (after expanding aliases, for aliases not get lost)
/// and before 'executeScalarSubqueries', 'analyzeAggregation', etc. to avoid excessive calculations.
if (select_query)
removeUnneededColumnsFromSelectClause(select_query, required_result_columns);
/// Executing scalar subqueries - replacing them with constant values.
executeScalarSubqueries(query, context, subquery_depth);
/// Optimize if with constant condition after constants was substituted instead of sclalar subqueries.
OptimizeIfWithConstantConditionVisitor(result.aliases).visit(query);
if (select_query)
{
/// GROUP BY injective function elimination.
optimizeGroupBy(select_query, source_columns_set, context);
/// Remove duplicate items from ORDER BY.
optimizeOrderBy(select_query);
/// Remove duplicated elements from LIMIT BY clause.
optimizeLimitBy(select_query);
/// Remove duplicated columns from USING(...).
optimizeUsing(select_query);
/// array_join_alias_to_name, array_join_result_to_source.
getArrayJoinedColumns(query, result, select_query, source_columns_list, source_columns_set);
/// Push the predicate expression down to the subqueries.
result.rewrite_subqueries = PredicateExpressionsOptimizer(select_query, settings, context).optimize();
collectJoinedColumns(result.analyzed_join, select_query, source_columns_set, context);
}
return std::make_shared<const SyntaxAnalyzerResult>(result);
}
}

12
dbms/src/Interpreters/SyntaxAnalyzer.h
View File

@ -8,6 +8,8 @@ namespace DB
class IStorage;
using StoragePtr = std::shared_ptr<IStorage>;
NameSet removeDuplicateColumns(NamesAndTypesList & columns);
struct SyntaxAnalyzerResult
{
StoragePtr storage;
@ -54,16 +56,20 @@ using SyntaxAnalyzerResultPtr = std::shared_ptr<const SyntaxAnalyzerResult>;
class SyntaxAnalyzer
{
public:
SyntaxAnalyzer(const Context & context, StoragePtr storage) : context(context), storage(std::move(storage)) {}
SyntaxAnalyzer(const Context & context_, size_t subquery_depth_ = 0)
: context(context_)
, subquery_depth(subquery_depth_)
{}
SyntaxAnalyzerResultPtr analyze(
ASTPtr & query,
const NamesAndTypesList & source_columns_,
const Names & required_result_columns = {},
size_t subquery_depth = 0) const;
StoragePtr storage = {}) const;
private:
const Context & context;
StoragePtr storage;
size_t subquery_depth;
};
}

2
dbms/src/Interpreters/TranslateQualifiedNamesVisitor.cpp
View File

@ -55,7 +55,7 @@ std::vector<ASTPtr *> TranslateQualifiedNamesMatcher::visit(const ASTIdentifier
const NameSet & source_columns = data.source_columns;
const std::vector<DatabaseAndTableWithAlias> & tables = data.tables;
if (identifier.general())
if (getColumnIdentifierName(identifier))
{
/// Select first table name with max number of qualifiers which can be stripped.
size_t max_num_qualifiers_to_strip = 0;

2
dbms/src/Interpreters/evaluateConstantExpression.cpp
View File

@ -31,7 +31,7 @@ std::pair<Field, std::shared_ptr<const IDataType>> evaluateConstantExpression(co
{
NamesAndTypesList source_columns = {{ "_dummy", std::make_shared<DataTypeUInt8>() }};
auto ast = node->clone();
auto syntax_result = SyntaxAnalyzer(context, {}).analyze(ast, source_columns);
auto syntax_result = SyntaxAnalyzer(context).analyze(ast, source_columns);
ExpressionActionsPtr expr_for_constant_folding = ExpressionAnalyzer(ast, syntax_result, context).getConstActions();
/// There must be at least one column in the block so that it knows the number of rows.

4
dbms/src/Interpreters/evaluateMissingDefaults.cpp
View File

@ -48,7 +48,7 @@ void evaluateMissingDefaults(Block & block,
if (!save_unneeded_columns)
{
auto syntax_result = SyntaxAnalyzer(context, {}).analyze(default_expr_list, block.getNamesAndTypesList());
auto syntax_result = SyntaxAnalyzer(context).analyze(default_expr_list, block.getNamesAndTypesList());
ExpressionAnalyzer{default_expr_list, syntax_result, context}.getActions(true)->execute(block);
return;
}
@ -57,7 +57,7 @@ void evaluateMissingDefaults(Block & block,
* we are going to operate on a copy instead of the original block */
Block copy_block{block};
auto syntax_result = SyntaxAnalyzer(context, {}).analyze(default_expr_list, block.getNamesAndTypesList());
auto syntax_result = SyntaxAnalyzer(context).analyze(default_expr_list, block.getNamesAndTypesList());
ExpressionAnalyzer{default_expr_list, syntax_result, context}.getActions(true)->execute(copy_block);
/// move evaluated columns to the original block, materializing them at the same time

2
dbms/src/Interpreters/executeQuery.cpp
View File

@ -499,7 +499,7 @@ void executeQuery(
}
String format_name = ast_query_with_output && (ast_query_with_output->format != nullptr)
? typeid_cast<const ASTIdentifier &>(*ast_query_with_output->format).name
? *getIdentifierName(ast_query_with_output->format)
: context.getDefaultFormat();
BlockOutputStreamPtr out = context.getOutputFormat(format_name, *out_buf, streams.in->getHeader());

1
dbms/src/Interpreters/loadMetadata.cpp
View File

@ -22,6 +22,7 @@
#include <Common/escapeForFileName.h>
#include <Common/Stopwatch.h>
#include <Common/typeid_cast.h>
namespace DB

29
dbms/src/Interpreters/tests/hash_map.cpp
View File

@ -107,13 +107,13 @@ int main(int argc, char ** argv)
AggregateFunctionPtr func_avg = factory.get("avg", data_types_uint64);
AggregateFunctionPtr func_uniq = factory.get("uniq", data_types_uint64);
#define INIT \
{ \
value.resize(3); \
\
value[0] = func_count.get();\
value[1] = func_avg.get(); \
value[2] = func_uniq.get(); \
#define INIT \
{ \
value.resize(3); \
\
value[0] = func_count.get(); \
value[1] = func_avg.get(); \
value[2] = func_uniq.get(); \
}
INIT
@ -162,7 +162,8 @@ int main(int argc, char ** argv)
map.emplace(data[i], it, inserted);
if (inserted)
{
new(&it->second) Value(std::move(value));
new(&it->second) Value;
std::swap(it->second, value);
INIT
}
}
@ -192,7 +193,8 @@ int main(int argc, char ** argv)
map.emplace(data[i], it, inserted);
if (inserted)
{
new(&it->second) Value(std::move(value));
new(&it->second) Value;
std::swap(it->second, value);
INIT
}
}
@ -223,7 +225,8 @@ int main(int argc, char ** argv)
map.emplace(data[i], it, inserted);
if (inserted)
{
new(&it->second) Value(std::move(value));
new(&it->second) Value;
std::swap(it->second, value);
INIT
}
}
@ -248,7 +251,7 @@ int main(int argc, char ** argv)
std::unordered_map<Key, Value, DefaultHash<Key>>::iterator it;
for (size_t i = 0; i < n; ++i)
{
it = map.insert(std::make_pair(data[i], std::move(value))).first;
it = map.insert(std::make_pair(data[i], value)).first;
INIT
}
@ -269,7 +272,7 @@ int main(int argc, char ** argv)
map.set_empty_key(-1ULL);
for (size_t i = 0; i < n; ++i)
{
it = map.insert(std::make_pair(data[i], std::move(value))).first;
it = map.insert(std::make_pair(data[i], value)).first;
INIT
}
@ -289,7 +292,7 @@ int main(int argc, char ** argv)
GOOGLE_NAMESPACE::sparse_hash_map<Key, Value, DefaultHash<Key>>::iterator it;
for (size_t i = 0; i < n; ++i)
{
map.insert(std::make_pair(data[i], std::move(value)));
map.insert(std::make_pair(data[i], value));
INIT
}

66
dbms/src/Parsers/ASTIdentifier.cpp
View File

@ -1,4 +1,5 @@
#include <Parsers/ASTIdentifier.h>
#include <Common/typeid_cast.h>
#include <IO/WriteBufferFromOStream.h>
#include <IO/WriteHelpers.h>
@ -38,4 +39,69 @@ void ASTIdentifier::appendColumnNameImpl(WriteBuffer & ostr) const
writeString(name, ostr);
}
bool isIdentifier(const IAST * const ast)
{
if (ast)
return typeid_cast<const ASTIdentifier *>(ast);
return false;
}
std::optional<String> getIdentifierName(const IAST * const ast)
{
if (ast)
if (auto node = typeid_cast<const ASTIdentifier *>(ast))
return node->name;
return {};
}
bool getIdentifierName(const ASTPtr & ast, String & name)
{
if (ast)
if (auto node = typeid_cast<const ASTIdentifier *>(ast.get()))
{
name = node->name;
return true;
}
return false;
}
std::optional<String> getColumnIdentifierName(const ASTIdentifier & node)
{
if (!node.special())
return node.name;
return {};
}
std::optional<String> getColumnIdentifierName(const ASTPtr & ast)
{
if (ast)
if (auto id = typeid_cast<const ASTIdentifier *>(ast.get()))
if (!id->special())
return id->name;
return {};
}
std::optional<String> getTableIdentifierName(const ASTIdentifier & node)
{
if (node.special())
return node.name;
return {};
}
std::optional<String> getTableIdentifierName(const ASTPtr & ast)
{
if (ast)
if (auto id = typeid_cast<const ASTIdentifier *>(ast.get()))
if (id->special())
return id->name;
return {};
}
void setIdentifierSpecial(ASTPtr & ast)
{
if (ast)
if (ASTIdentifier * id = typeid_cast<ASTIdentifier *>(ast.get()))
id->setSpecial();
}
}

38
dbms/src/Parsers/ASTIdentifier.h
View File

@ -1,13 +1,14 @@
#pragma once
#include <optional>
#include <Parsers/ASTWithAlias.h>
namespace DB
{
/** Identifier (column or alias)
*/
/// Identifier (column, table or alias)
class ASTIdentifier : public ASTWithAlias
{
enum Kind /// TODO This is semantic, not syntax. Remove it.
@ -33,10 +34,6 @@ public:
set.insert(name);
}
void setSpecial() { kind = Special; }
bool general() const { return kind == General; }
bool special() const { return kind == Special; }
static std::shared_ptr<ASTIdentifier> createSpecial(const String & name_)
{
return std::make_shared<ASTIdentifier>(name_, ASTIdentifier::Special);
@ -48,6 +45,35 @@ protected:
private:
Kind kind;
void setSpecial() { kind = Special; }
bool special() const { return kind == Special; }
friend void setIdentifierSpecial(ASTPtr &);
friend std::optional<String> getColumnIdentifierName(const ASTIdentifier & node);
friend std::optional<String> getColumnIdentifierName(const ASTPtr & ast);
friend std::optional<String> getTableIdentifierName(const ASTIdentifier & node);
friend std::optional<String> getTableIdentifierName(const ASTPtr & ast);
};
/// ASTIdentifier Helpers: hide casts and semantic.
bool isIdentifier(const IAST * const ast);
inline bool isIdentifier(const ASTPtr & ast) { return isIdentifier(ast.get()); }
std::optional<String> getIdentifierName(const IAST * const ast);
inline std::optional<String> getIdentifierName(const ASTPtr & ast) { return getIdentifierName(ast.get()); }
bool getIdentifierName(const ASTPtr & ast, String & name);
/// @returns name for column identifiers
std::optional<String> getColumnIdentifierName(const ASTIdentifier & node);
std::optional<String> getColumnIdentifierName(const ASTPtr & ast);
/// @returns name for 'not a column' identifiers
std::optional<String> getTableIdentifierName(const ASTIdentifier & node);
std::optional<String> getTableIdentifierName(const ASTPtr & ast);
void setIdentifierSpecial(ASTPtr & ast);
}

17
dbms/src/Parsers/ExpressionElementParsers.cpp
View File

@ -5,6 +5,7 @@
#include <IO/ReadHelpers.h>
#include <IO/ReadBufferFromMemory.h>
#include <Common/typeid_cast.h>
#include <Parsers/DumpASTNode.h>
#include <Parsers/IAST.h>
@ -173,7 +174,7 @@ bool ParserCompoundIdentifier::parseImpl(Pos & pos, ASTPtr & node, Expected & ex
{
if (!name.empty())
name += '.';
name += static_cast<const ASTIdentifier &>(*child.get()).name;
name += *getIdentifierName(child);
}
node = std::make_shared<ASTIdentifier>(name);
@ -222,7 +223,7 @@ bool ParserFunction::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
* If you do not report that the first option is an error, then the argument will be interpreted as 2014 - 01 - 01 - some number,
* and the query silently returns an unexpected result.
*/
if (typeid_cast<const ASTIdentifier &>(*identifier).name == "toDate"
if (*getIdentifierName(identifier) == "toDate"
&& contents_end - contents_begin == strlen("2014-01-01")
&& contents_begin[0] >= '2' && contents_begin[0] <= '3'
&& contents_begin[1] >= '0' && contents_begin[1] <= '9'
@ -264,7 +265,7 @@ bool ParserFunction::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
}
auto function_node = std::make_shared<ASTFunction>();
function_node->name = typeid_cast<ASTIdentifier &>(*identifier).name;
getIdentifierName(identifier, function_node->name);
/// func(DISTINCT ...) is equivalent to funcDistinct(...)
if (has_distinct_modifier)
@ -1157,7 +1158,7 @@ bool ParserAlias::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
* and in the query "SELECT x FRO FROM t", the word FRO was considered an alias.
*/
const String & name = static_cast<const ASTIdentifier &>(*node.get()).name;
const String name = *getIdentifierName(node);
for (const char ** keyword = restricted_keywords; *keyword != nullptr; ++keyword)
if (0 == strcasecmp(name.data(), *keyword))
@ -1249,18 +1250,16 @@ bool ParserWithOptionalAlias::parseImpl(Pos & pos, ASTPtr & node, Expected & exp
*/
bool allow_alias_without_as_keyword_now = allow_alias_without_as_keyword;
if (allow_alias_without_as_keyword)
if (const ASTIdentifier * id = typeid_cast<const ASTIdentifier *>(node.get()))
if (0 == strcasecmp(id->name.data(), "FROM"))
if (auto opt_id = getIdentifierName(node))
if (0 == strcasecmp(opt_id->data(), "FROM"))
allow_alias_without_as_keyword_now = false;
ASTPtr alias_node;
if (ParserAlias(allow_alias_without_as_keyword_now).parse(pos, alias_node, expected))
{
String alias_name = typeid_cast<const ASTIdentifier &>(*alias_node).name;
if (ASTWithAlias * ast_with_alias = dynamic_cast<ASTWithAlias *>(node.get()))
{
ast_with_alias->alias = alias_name;
getIdentifierName(alias_node, ast_with_alias->alias);
ast_with_alias->prefer_alias_to_column_name = prefer_alias_to_column_name;
}
else

3
dbms/src/Parsers/ExpressionListParsers.cpp
View File

@ -1,10 +1,9 @@
#include <Common/typeid_cast.h>
#include <Parsers/IAST.h>
#include <Parsers/ASTExpressionList.h>
#include <Parsers/ASTFunction.h>
#include <Parsers/CommonParsers.h>
#include <Parsers/ExpressionElementParsers.h>
#include <Parsers/ExpressionListParsers.h>
#include <Parsers/ParserCreateQuery.h>

3
dbms/src/Parsers/ParserAlterQuery.cpp
View File

@ -1,3 +1,4 @@
#include <Common/typeid_cast.h>
#include <Parsers/ParserAlterQuery.h>
#include <Parsers/CommonParsers.h>
#include <Parsers/ExpressionElementParsers.h>
@ -312,7 +313,7 @@ bool ParserAssignment::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
if (!p_expression.parse(pos, assignment->expression, expected))
return false;
assignment->column_name = typeid_cast<const ASTIdentifier &>(*column).name;
getIdentifierName(column, assignment->column_name);
if (assignment->expression)
assignment->children.push_back(assignment->expression);

8
dbms/src/Parsers/ParserCheckQuery.cpp
View File

@ -4,8 +4,6 @@
#include <Parsers/ExpressionElementParsers.h>
#include <Parsers/ASTCheckQuery.h>
#include <Common/typeid_cast.h>
namespace DB
{
@ -31,15 +29,15 @@ bool ParserCheckQuery::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
return false;
auto query = std::make_shared<ASTCheckQuery>();
query->database = typeid_cast<const ASTIdentifier &>(*database).name;
query->table = typeid_cast<const ASTIdentifier &>(*table).name;
getIdentifierName(database, query->database);
getIdentifierName(table, query->table);
node = query;
}
else
{
table = database;
auto query = std::make_shared<ASTCheckQuery>();
query->table = typeid_cast<const ASTIdentifier &>(*table).name;
getIdentifierName(table, query->table);
node = query;
}

29
dbms/src/Parsers/ParserCreateQuery.cpp
View File

@ -35,7 +35,7 @@ bool ParserNestedTable::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
return false;
auto func = std::make_shared<ASTFunction>();
func->name = typeid_cast<ASTIdentifier &>(*name).name;
getIdentifierName(name, func->name);
func->arguments = columns;
func->children.push_back(columns);
node = func;
@ -70,7 +70,7 @@ bool ParserIdentifierWithOptionalParameters::parseImpl(Pos & pos, ASTPtr & node,
if (non_parametric.parse(pos, ident, expected))
{
auto func = std::make_shared<ASTFunction>();
func->name = typeid_cast<ASTIdentifier &>(*ident).name;
getIdentifierName(ident, func->name);
node = func;
return true;
}
@ -257,10 +257,8 @@ bool ParserCreateQuery::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
query->if_not_exists = if_not_exists;
query->cluster = cluster_str;
if (database)
query->database = typeid_cast<ASTIdentifier &>(*database).name;
if (table)
query->table = typeid_cast<ASTIdentifier &>(*table).name;
getIdentifierName(database, query->database);
getIdentifierName(table, query->table);
return true;
}
@ -405,23 +403,18 @@ bool ParserCreateQuery::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
query->is_populate = is_populate;
query->temporary = is_temporary;
if (database)
query->database = typeid_cast<ASTIdentifier &>(*database).name;
if (table)
query->table = typeid_cast<ASTIdentifier &>(*table).name;
getIdentifierName(database, query->database);
getIdentifierName(table, query->table);
query->cluster = cluster_str;
if (to_database)
query->to_database = typeid_cast<ASTIdentifier &>(*to_database).name;
if (to_table)
query->to_table = typeid_cast<ASTIdentifier &>(*to_table).name;
getIdentifierName(to_database, query->to_database);
getIdentifierName(to_table, query->to_table);
query->set(query->columns, columns);
query->set(query->storage, storage);
if (as_database)
query->as_database = typeid_cast<ASTIdentifier &>(*as_database).name;
if (as_table)
query->as_table = typeid_cast<ASTIdentifier &>(*as_table).name;
getIdentifierName(as_database, query->as_database);
getIdentifierName(as_table, query->as_table);
query->set(query->select, select);
return true;

5
dbms/src/Parsers/ParserCreateQuery.h
View File

@ -8,7 +8,6 @@
#include <Parsers/ASTIdentifier.h>
#include <Parsers/ASTLiteral.h>
#include <Parsers/CommonParsers.h>
#include <Common/typeid_cast.h>
#include <Poco/String.h>
@ -74,7 +73,7 @@ bool IParserNameTypePair<NameParser>::parseImpl(Pos & pos, ASTPtr & node, Expect
&& type_parser.parse(pos, type, expected))
{
auto name_type_pair = std::make_shared<ASTNameTypePair>();
name_type_pair->name = typeid_cast<const ASTIdentifier &>(*name).name;
getIdentifierName(name, name_type_pair->name);
name_type_pair->type = type;
name_type_pair->children.push_back(type);
node = name_type_pair;
@ -181,7 +180,7 @@ bool IParserColumnDeclaration<NameParser>::parseImpl(Pos & pos, ASTPtr & node, E
const auto column_declaration = std::make_shared<ASTColumnDeclaration>();
node = column_declaration;
column_declaration->name = typeid_cast<ASTIdentifier &>(*name).name;
getIdentifierName(name, column_declaration->name);
if (type)
{

10
dbms/src/Parsers/ParserDropQuery.cpp
View File

@ -4,8 +4,6 @@
#include <Parsers/CommonParsers.h>
#include <Parsers/ParserDropQuery.h>
#include <Common/typeid_cast.h>
namespace DB
{
@ -117,10 +115,10 @@ bool ParserDropQuery::parseDropQuery(Pos & pos, ASTPtr & node, Expected & expect
query->kind = ASTDropQuery::Kind::Drop;
query->if_exists = if_exists;
query->temporary = temporary;
if (database)
query->database = typeid_cast<ASTIdentifier &>(*database).name;
if (table)
query->table = typeid_cast<ASTIdentifier &>(*table).name;
getIdentifierName(database, query->database);
getIdentifierName(table, query->table);
query->cluster = cluster_str;
return true;

11
dbms/src/Parsers/ParserInsertQuery.cpp
View File

@ -9,8 +9,6 @@
#include <Parsers/ParserInsertQuery.h>
#include <Parsers/ASTFunction.h>
#include <Common/typeid_cast.h>
namespace DB
{
@ -136,14 +134,11 @@ bool ParserInsertQuery::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
}
else
{
if (database)
query->database = typeid_cast<ASTIdentifier &>(*database).name;
query->table = typeid_cast<ASTIdentifier &>(*table).name;
getIdentifierName(database, query->database);
getIdentifierName(table, query->table);
}
if (format)
query->format = typeid_cast<ASTIdentifier &>(*format).name;
getIdentifierName(format, query->format);
query->columns = columns;
query->select = select;

8
dbms/src/Parsers/ParserOptimizeQuery.cpp
View File

@ -5,8 +5,6 @@
#include <Parsers/ASTOptimizeQuery.h>
#include <Parsers/ASTIdentifier.h>
#include <Common/typeid_cast.h>
namespace DB
{
@ -60,10 +58,8 @@ bool ParserOptimizeQuery::parseImpl(Pos & pos, ASTPtr & node, Expected & expecte
auto query = std::make_shared<ASTOptimizeQuery>();
node = query;
if (database)
query->database = typeid_cast<const ASTIdentifier &>(*database).name;
if (table)
query->table = typeid_cast<const ASTIdentifier &>(*table).name;
getIdentifierName(database, query->database);
getIdentifierName(table, query->table);
query->cluster = cluster_str;
query->partition = partition;

2
dbms/src/Parsers/ParserQueryWithOutput.cpp
View File

@ -76,7 +76,7 @@ bool ParserQueryWithOutput::parseImpl(Pos & pos, ASTPtr & node, Expected & expec
if (!format_p.parse(pos, query_with_output.format, expected))
return false;
typeid_cast<ASTIdentifier &>(*(query_with_output.format)).setSpecial();
setIdentifierSpecial(query_with_output.format);
query_with_output.children.push_back(query_with_output.format);
}

7
dbms/src/Parsers/ParserRenameQuery.cpp
View File

@ -4,8 +4,6 @@
#include <Parsers/CommonParsers.h>
#include <Parsers/ParserRenameQuery.h>
#include <Common/typeid_cast.h>
namespace DB
{
@ -31,8 +29,9 @@ static bool parseDatabaseAndTable(
return false;
}
db_and_table.database = database ? typeid_cast<const ASTIdentifier &>(*database).name : "";
db_and_table.table = typeid_cast<const ASTIdentifier &>(*table).name;
db_and_table.database.clear();
getIdentifierName(database, db_and_table.database);
getIdentifierName(table, db_and_table.table);
return true;
}

2
dbms/src/Parsers/ParserSetQuery.cpp
View File

@ -31,7 +31,7 @@ static bool parseNameValuePair(ASTSetQuery::Change & change, IParser::Pos & pos,
if (!value_p.parse(pos, value, expected))
return false;
change.name = typeid_cast<const ASTIdentifier &>(*name).name;
getIdentifierName(name, change.name);
change.value = typeid_cast<const ASTLiteral &>(*value).value;
return true;

3
dbms/src/Parsers/ParserShowTablesQuery.cpp
View File

@ -65,8 +65,7 @@ bool ParserShowTablesQuery::parseImpl(Pos & pos, ASTPtr & node, Expected & expec
return false;
}
if (database)
query->from = typeid_cast<ASTIdentifier &>(*database).name;
getIdentifierName(database, query->from);
if (like)
query->like = safeGet<const String &>(typeid_cast<ASTLiteral &>(*like).value);

6
dbms/src/Parsers/ParserTablePropertiesQuery.cpp
View File

@ -75,10 +75,8 @@ bool ParserTablePropertiesQuery::parseImpl(Pos & pos, ASTPtr & node, Expected &
}
}
if (database)
query->database = typeid_cast<ASTIdentifier &>(*database).name;
if (table)
query->table = typeid_cast<ASTIdentifier &>(*table).name;
getIdentifierName(database, query->database);
getIdentifierName(table, query->table);
node = query;

5
dbms/src/Parsers/ParserUseQuery.cpp
View File

@ -15,16 +15,15 @@ bool ParserUseQuery::parseImpl(Pos & pos, ASTPtr & node, Expected & expected)
ParserKeyword s_use("USE");
ParserIdentifier name_p;
ASTPtr database;
if (!s_use.ignore(pos, expected))
return false;
ASTPtr database;
if (!name_p.parse(pos, database, expected))
return false;
auto query = std::make_shared<ASTUseQuery>();
query->database = typeid_cast<ASTIdentifier &>(*database).name;
getIdentifierName(database, query->database);
node = query;
return true;

7
dbms/src/Parsers/parseDatabaseAndTableName.cpp
View File

@ -2,7 +2,6 @@
#include <Parsers/ExpressionElementParsers.h>
#include <Parsers/ASTIdentifier.h>
#include <Parsers/CommonParsers.h>
#include <Common/typeid_cast.h>
namespace DB
@ -30,13 +29,13 @@ bool parseDatabaseAndTableName(IParser::Pos & pos, Expected & expected, String &
return false;
}
database_str = typeid_cast<ASTIdentifier &>(*database).name;
table_str = typeid_cast<ASTIdentifier &>(*table).name;
getIdentifierName(database, database_str);
getIdentifierName(table, table_str);
}
else
{
database_str = "";
table_str = typeid_cast<ASTIdentifier &>(*database).name;
getIdentifierName(database, table_str);
}
return true;

2
dbms/src/Parsers/parseIdentifierOrStringLiteral.cpp
View File

@ -20,7 +20,7 @@ bool parseIdentifierOrStringLiteral(IParser::Pos & pos, Expected & expected, Str
result = typeid_cast<const ASTLiteral &>(*res).value.safeGet<String>();
}
else
result = typeid_cast<const ASTIdentifier &>(*res).name;
result = *getIdentifierName(res);
return true;
}

11
dbms/src/Storages/AlterCommands.cpp
View File

@ -50,7 +50,7 @@ std::optional<AlterCommand> AlterCommand::parse(const ASTAlterCommand * command_
}
if (command_ast->column)
command.after_column = typeid_cast<const ASTIdentifier &>(*command_ast->column).name;
command.after_column = *getIdentifierName(command_ast->column);
command.if_not_exists = command_ast->if_not_exists;
@ -63,7 +63,7 @@ std::optional<AlterCommand> AlterCommand::parse(const ASTAlterCommand * command_
AlterCommand command;
command.type = AlterCommand::DROP_COLUMN;
command.column_name = typeid_cast<const ASTIdentifier &>(*(command_ast->column)).name;
command.column_name = *getIdentifierName(command_ast->column);
command.if_exists = command_ast->if_exists;
return command;
}
@ -99,8 +99,7 @@ std::optional<AlterCommand> AlterCommand::parse(const ASTAlterCommand * command_
{
AlterCommand command;
command.type = COMMENT_COLUMN;
const auto & ast_identifier = typeid_cast<ASTIdentifier &>(*command_ast->column);
command.column_name = ast_identifier.name;
command.column_name = *getIdentifierName(command_ast->column);
const auto & ast_comment = typeid_cast<ASTLiteral &>(*command_ast->comment);
command.comment = ast_comment.value.get<String>();
command.if_exists = command_ast->if_exists;
@ -398,7 +397,7 @@ void AlterCommands::validate(const IStorage & table, const Context & context)
{
const auto & default_expression = default_column.second.expression;
ASTPtr query = default_expression;
auto syntax_result = SyntaxAnalyzer(context, {}).analyze(query, all_columns);
auto syntax_result = SyntaxAnalyzer(context).analyze(query, all_columns);
const auto actions = ExpressionAnalyzer(query, syntax_result, context).getActions(true);
const auto required_columns = actions->getRequiredColumns();
@ -473,7 +472,7 @@ void AlterCommands::validate(const IStorage & table, const Context & context)
}
ASTPtr query = default_expr_list;
auto syntax_result = SyntaxAnalyzer(context, {}).analyze(query, all_columns);
auto syntax_result = SyntaxAnalyzer(context).analyze(query, all_columns);
const auto actions = ExpressionAnalyzer(query, syntax_result, context).getActions(true);
const auto block = actions->getSampleBlock();

Some files were not shown because too many files have changed in this diff Show More