clickhouse function
· 5 min read
背景
想要熟悉clickhouse的内容。
实现
## /ClickHouse/src/Functions/isNaN.cpp
## 添加一个struct
struct testtrue
{
static constexpr auto name = "testtrue"; ##函数名
template <typename T>
static bool execute(const T t) ### 执行回调
{
/// Suppression for PVS-Studio.
return true || t;
}
};
### 起别名
using FunctionTestTure = FunctionNumericPredicate<testtrue>;
void registerFunctionIsNaN(FunctionFactory & factory)
{
factory.registerFunction<FunctionIsNaN>();
factory.registerFunction<FunctionTestTure>(); ### 回调注册这个函数
}
重新编译并调用：
SELECT testtrue(2)
Query id: 87e4625d-8b79-4c3a-8153-b333d6a0614f
┌─testtrue(2)─┐
│ 1 │
└─────────────┘
注册路径
(lldb) bt
* thread #1, name = 'clickhouse-serv', stop reason = breakpoint 1.1
* frame #0: 0x00000000148ee2ac clickhouse-server`DB::registerFunctionIsNaN(factory=0x000000002f984a50) at isNaN.cpp:43:5
frame #1: 0x0000000010339e2c clickhouse-server`DB::registerFunctionsMiscellaneous(factory=0x000000002f984a50) at registerFunctionsMiscellaneous.cpp:128:5
frame #2: 0x00000000103391a0 clickhouse-server`DB::registerFunctions() at registerFunctions.cpp:96:5
frame #3: 0x000000000d063d35 clickhouse-server`DB::Server::main(this=0x00007fffffffd9b8, (null)=size=0) at Server.cpp:623:5
frame #4: 0x00000000230abec5 clickhouse-server`Poco::Util::Application::run(this=0x00007fffffffd9b8) at Application.cpp:334:8
frame #5: 0x000000000d06250b clickhouse-server`DB::Server::run(this=0x00007fffffffd9b8) at Server.cpp:461:25
frame #6: 0x00000000230c6c70 clickhouse-server`Poco::Util::ServerApplication::run(this=0x00007fffffffd9b8, argc=1, argv=0x00007ffff70f7038) at ServerApplication.cpp:611:9
frame #7: 0x000000000d05f8e1 clickhouse-server`mainEntryClickHouseServer(argc=1, argv=0x00007ffff70f7038) at Server.cpp:187:20
frame #8: 0x000000000cf7fe63 clickhouse-server`main(argc_=1, argv_=0x00007fffffffdfb8) at main.cpp:409:12
frame #9: 0x00007ffff7d92d90 libc.so.6`__libc_start_call_main(main=(clickhouse-server`main at main.cpp:380), argc=1, argv=0x00007fffffffdfb8) at libc_start_call_main.h:58:16
frame #10: 0x00007ffff7d92e40 libc.so.6`__libc_start_main_impl(main=(clickhouse-server`main at main.cpp:380), argc=1, argv=0x00007fffffffdfb8, init=0x00007ffff7ffd040, fini=<unavailable>, rtld_fini=<unavailable>, stack_end=0x00007fffffffdfa8) at libc-start.c:392:3
frame #11: 0x000000000cf7fb55 clickhouse-server`_start + 37
函数调用路径
lldb 调试
(lldb) b DB::(anonymous namespace)::testtrue::execute
调用堆栈：
(lldb) bt
* thread #3, name = 'TCPHandler', stop reason = breakpoint 3.1
* frame #0: 0x00000000148f3dca clickhouse-server`bool DB::(anonymous namespace)::testtrue::execute<char8_t>(t=0x02 u8'\U00000002') at isNaN.cpp:28:9
frame #1: 0x00000000148f29dd clickhouse-server`COW<DB::IColumn>::immutable_ptr<DB::IColumn> DB::FunctionNumericPredicate<DB::(anonymous namespace)::testtrue>::execute<char8_t>(this=0x00007fff1e44db58, in_untyped=0x00007fff1e447820) const at FunctionNumericPredicate.h:89:31
frame #2: 0x00000000148f1e8a clickhouse-server`DB::FunctionNumericPredicate<DB::(anonymous namespace)::testtrue>::executeImpl(this=0x00007fff1e44db58, arguments=size=1, (null)=std::__1::shared_ptr<const DB::IDataType>::element_type @ 0x00007fff1e4899c8 strong=2 weak=2, (null)=1) const at FunctionNumericPredicate.h:61:22
frame #3: 0x00000000103a611c clickhouse-server`DB::IFunction::executeImplDryRun(this=0x00007fff1e44db58, arguments=size=1, result_type=std::__1::shared_ptr<const DB::IDataType>::element_type @ 0x00007fff1e4899c8 strong=2 weak=2, input_rows_count=1) const at IFunction.h:395:16
frame #4: 0x00000000103a484d clickhouse-server`DB::FunctionToExecutableFunctionAdaptor::executeDryRunImpl(this=0x00007fff1e44dba0, arguments=size=1, result_type=std::__1::shared_ptr<const DB::IDataType>::element_type @ 0x00007fff1e4899c8 strong=2 weak=2, input_rows_count=1) const at IFunctionAdaptors.h:26:26
frame #5: 0x000000001a8fa3d9 clickhouse-server`DB::IExecutableFunction::executeWithoutLowCardinalityColumns(this=0x00007fff1e44dba0, args=size=1, result_type=std::__1::shared_ptr<const DB::IDataType>::element_type @ 0x00007fff1e4899c8 strong=2 weak=2, input_rows_count=1, dry_run=true) const at IFunction.cpp:217:15
frame #6: 0x000000001a8fa00a clickhouse-server`DB::IExecutableFunction::defaultImplementationForConstantArguments(this=0x00007fff1e44dba0, args=size=1, result_type=std::__1::shared_ptr<const DB::IDataType>::element_type @ 0x00007fff1e4899c8 strong=2 weak=2, input_rows_count=1, dry_run=true) const at IFunction.cpp:160:31
frame #7: 0x000000001a8fa2c4 clickhouse-server`DB::IExecutableFunction::executeWithoutLowCardinalityColumns(this=0x00007fff1e44dba0, args=size=1, result_type=std::__1::shared_ptr<const DB::IDataType>::element_type @ 0x00007fff1e4899c8 strong=2 weak=2, input_rows_count=1, dry_run=true) const at IFunction.cpp:209:20
frame #8: 0x000000001a8faf55 clickhouse-server`DB::IExecutableFunction::executeWithoutSparseColumns(this=0x00007fff1e44dba0, arguments=size=1, result_type=std::__1::shared_ptr<const DB::IDataType>::element_type @ 0x00007fff1e4899c8 strong=2 weak=2, input_rows_count=1, dry_run=true) const at IFunction.cpp:267:22
frame #9: 0x000000001a8fbe43 clickhouse-server`DB::IExecutableFunction::execute(this=0x00007fff1e44dba0, arguments=size=1, result_type=std::__1::shared_ptr<const DB::IDataType>::element_type @ 0x00007fff1e4899c8 strong=2 weak=2, input_rows_count=1, dry_run=true) const at IFunction.cpp:337:16
frame #10: 0x000000001b02a7b8 clickhouse-server`DB::ActionsDAG::addFunction(this=0x00007fff1e4ac248, function=std::__1::shared_ptr<DB::IFunctionOverloadResolver>::element_type @ 0x00007fff1e44dde0 strong=2 weak=1, children=size=0, result_name="testtrue(2)") at ActionsDAG.cpp:199:37
frame #11: 0x000000001cdad1a1 clickhouse-server`DB::ScopeStack::addFunction(this=0x00007fff2d1e96f0, function=std::__1::shared_ptr<DB::IFunctionOverloadResolver>::element_type @ 0x00007fff1e44dde0 strong=2 weak=1, argument_names=size=1, result_name="") at ActionsVisitor.cpp:598:51
frame #12: 0x000000001cdb7485 clickhouse-server`DB::ActionsMatcher::Data::addFunction(this=0x00007fff2d1e9698, function=std::__1::shared_ptr<DB::IFunctionOverloadResolver>::element_type @ 0x00007fff1e44dde0 strong=2 weak=1, argument_names=size=1, result_name=<unavailable>) at ActionsVisitor.h:140:27
frame #13: 0x000000001cdb0a2b clickhouse-server`DB::ActionsMatcher::visit(node=0x00007fff2a45c9b8, ast=std::__1::shared_ptr<DB::IAST>::element_type @ 0x00007fff2a45c9b8 strong=1 weak=2, data=0x00007fff2d1e9698) at ActionsVisitor.cpp:1093:14
frame #14: 0x000000001cdad64d clickhouse-server`DB::ActionsMatcher::visit(ast=std::__1::shared_ptr<DB::IAST>::element_type @ 0x00007fff2a45c9b8 strong=1 weak=2, data=0x00007fff2d1e9698) at ActionsVisitor.cpp:655:9
frame #15: 0x000000001cdb125c clickhouse-server`DB::ActionsMatcher::visit(expression_list=0x00007fff1e4895b8, (null)=std::__1::shared_ptr<DB::IAST>::element_type @ 0x00007fff1e4895b8 strong=2 weak=2, data=0x00007fff2d1e9698) at ActionsVisitor.cpp:763:17
frame #16: 0x000000001cdad6b9 clickhouse-server`DB::ActionsMatcher::visit(ast=std::__1::shared_ptr<DB::IAST>::element_type @ 0x00007fff1e4895b8 strong=2 weak=2, data=0x00007fff2d1e9698) at ActionsVisitor.cpp:659:9
frame #17: 0x000000001b1ebce5 clickhouse-server`DB::InDepthNodeVisitor<DB::ActionsMatcher, true, false, std::__1::shared_ptr<DB::IAST> const>::visit(this=0x00007fff2d1e9638, ast=std::__1::shared_ptr<DB::IAST>::element_type @ 0x00007fff1e4895b8 strong=2 weak=2) at InDepthNodeVisitor.h:34:13
frame #18: 0x000000001b1dc0ea clickhouse-server`DB::ExpressionAnalyzer::getRootActions(this=0x00007fff1e47f780, ast=std::__1::shared_ptr<DB::IAST>::element_type @ 0x00007fff1e4895b8 strong=2 weak=2, no_makeset_for_subqueries=false, actions=nullptr, only_consts=false) at ExpressionAnalyzer.cpp:587:48
frame #19: 0x000000001b1e2f9e clickhouse-server`DB::SelectQueryExpressionAnalyzer::appendSelect(this=0x00007fff1e47f780, chain=0x00007fff2d1ea5a0, only_types=false) at ExpressionAnalyzer.cpp:1383:5
frame #20: 0x000000001b1e6da8 clickhouse-server`DB::ExpressionAnalysisResult::ExpressionAnalysisResult(this=0x00007fff2d1eaa88, query_analyzer=0x00007fff1e47f780, metadata_snapshot=std::__1::shared_ptr<const DB::StorageInMemoryMetadata>::element_type @ 0x00007ffff705a800 strong=4 weak=1, first_stage_=true, second_stage_=true, only_types=false, filter_info_=nullptr, source_header=0x00007fff1e4c9550) at ExpressionAnalyzer.cpp:1830:24
frame #21: 0x000000001b57ab9d clickhouse-server`DB::InterpreterSelectQuery::getSampleBlockImpl(this=0x00007fff1e4c9000) at InterpreterSelectQuery.cpp:692:23
frame #22: 0x000000001b5747f9 clickhouse-server`DB::InterpreterSelectQuery::InterpreterSelectQuery(this=0x00007fff2d1ec148, try_move_to_prewhere=true)::$_1::operator()(bool) const at InterpreterSelectQuery.cpp:552:25
frame #23: 0x000000001b5709f6 clickhouse-server`DB::InterpreterSelectQuery::InterpreterSelectQuery(this=0x00007fff1e4c9000, query_ptr_=std::__1::shared_ptr<DB::IAST>::element_type @ 0x00007fff1e48e198 strong=1 weak=2, context_=std::__1::shared_ptr<const DB::Context>::element_type @ 0x00007fff1e495000 strong=3 weak=2, input_pipe_= Has Value=false , storage_=nullptr, options_=0x00007fff1e461770, required_result_column_names=size=0, metadata_snapshot_=nullptr, subquery_for_sets_=size=0, prepared_sets_=size=0) at InterpreterSelectQuery.cpp:555:5
frame #24: 0x000000001b56edf3 clickhouse-server`DB::InterpreterSelectQuery::InterpreterSelectQuery(this=0x00007fff1e4c9000, query_ptr_=std::__1::shared_ptr<DB::IAST>::element_type @ 0x00007fff1e48e198 strong=1 weak=2, context_=std::__1::shared_ptr<const DB::Context>::element_type @ 0x00007fff1e495000 strong=3 weak=2, options_=0x00007fff1e461770, required_result_column_names_=size=0) at InterpreterSelectQuery.cpp:165:7
frame #25: 0x000000001b5f4ce5 clickhouse-server`std::__1::__unique_if<DB::InterpreterSelectQuery>::__unique_single std::__1::make_unique<DB::InterpreterSelectQuery, std::__1::shared_ptr<DB::IAST> const&, std::__1::shared_ptr<DB::Context>&, DB::SelectQueryOptions&, std::__1::vector<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> >, std::__1::allocator<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> > > > const&>(__args=std::__1::shared_ptr<DB::IAST>::element_type @ 0x00007fff1e48e198 strong=1 weak=2, __args=std::__1::shared_ptr<DB::Context>::element_type @ 0x00007fff1e495000 strong=3 weak=2, __args=0x00007fff1e461770, __args=size=0) at unique_ptr.h:725:32
frame #26: 0x000000001b5f2d09 clickhouse-server`DB::InterpreterSelectWithUnionQuery::buildCurrentChildInterpreter(this=0x00007fff1e461700, ast_ptr_=std::__1::shared_ptr<DB::IAST>::element_type @ 0x00007fff1e48e198 strong=1 weak=2, current_required_result_column_names=size=0) at InterpreterSelectWithUnionQuery.cpp:223:16
frame #27: 0x000000001b5f23ed clickhouse-server`DB::InterpreterSelectWithUnionQuery::InterpreterSelectWithUnionQuery(this=0x00007fff1e461700, query_ptr_=std::__1::shared_ptr<DB::IAST>::element_type @ 0x00007fff1e48f218 strong=2 weak=2, context_=std::__1::shared_ptr<const DB::Context>::element_type @ 0x00007fff1e492800 strong=5 weak=9, options_=0x00007fff2d1ef588, required_result_column_names=size=0) at InterpreterSelectWithUnionQuery.cpp:140:13
frame #28: 0x000000001b53b830 clickhouse-server`std::__1::__unique_if<DB::InterpreterSelectWithUnionQuery>::__unique_single std::__1::make_unique<DB::InterpreterSelectWithUnionQuery, std::__1::shared_ptr<DB::IAST>&, std::__1::shared_ptr<DB::Context>&, DB::SelectQueryOptions const&>(__args=std::__1::shared_ptr<DB::IAST>::element_type @ 0x00007fff1e48f218 strong=2 weak=2, __args=std::__1::shared_ptr<DB::Context>::element_type @ 0x00007fff1e492800 strong=5 weak=9, __args=0x00007fff2d1ef588) at unique_ptr.h:725:32
frame #29: 0x000000001b539e5d clickhouse-server`DB::InterpreterFactory::get(query=std::__1::shared_ptr<DB::IAST>::element_type @ 0x00007fff1e48f218 strong=2 weak=2, context=std::__1::shared_ptr<DB::Context>::element_type @ 0x00007fff1e492800 strong=5 weak=9, options=0x00007fff2d1ef588) at InterpreterFactory.cpp:122:16
frame #30: 0x000000001b961577 clickhouse-server`DB::executeQueryImpl(begin="select testtrue(2);", end="", context=std::__1::shared_ptr<DB::Context>::element_type @ 0x00007fff1e492800 strong=5 weak=9, internal=false, stage=Complete, istr=0x0000000000000000) at executeQuery.cpp:658:27
frame #31: 0x000000001b95ee64 clickhouse-server`DB::executeQuery(query="select testtrue(2);", context=std::__1::shared_ptr<DB::Context>::element_type @ 0x00007fff1e492800 strong=5 weak=9, internal=false, stage=Complete) at executeQuery.cpp:1067:30
frame #32: 0x000000001c590f25 clickhouse-server`DB::TCPHandler::runImpl(this=0x00007fff1e46e000) at TCPHandler.cpp:332:24
frame #33: 0x000000001c59f9e5 clickhouse-server`DB::TCPHandler::run(this=0x00007fff1e46e000) at TCPHandler.cpp:1781:9
frame #34: 0x0000000023091f79 clickhouse-server`Poco::Net::TCPServerConnection::start(this=0x00007fff1e46e000) at TCPServerConnection.cpp:43:3
frame #35: 0x0000000023092786 clickhouse-server`Poco::Net::TCPServerDispatcher::run(this=0x00007fff26648600) at TCPServerDispatcher.cpp:115:20
frame #36: 0x00000000232d2a94 clickhouse-server`Poco::PooledThread::run(this=0x00007ffff702df80) at ThreadPool.cpp:199:14
frame #37: 0x00000000232cf5ba clickhouse-server`Poco::(anonymous namespace)::RunnableHolder::run(this=0x00007ffff7001330) at Thread.cpp:55:11
frame #38: 0x00000000232ce39e clickhouse-server`Poco::ThreadImpl::runnableEntry(pThread=0x00007ffff702dfb8) at Thread_POSIX.cpp:345:27
frame #39: 0x00007ffff7dfdb43 libc.so.6`start_thread(arg=<unavailable>) at pthread_create.c:442:8
frame #40: 0x00007ffff7e8fa00 libc.so.6`__clone3 at clone3.S:81
function的实现
所有函数都是继承IFunction
## ClickHouse/src/Functions/IFunction.h
class IFunction
{
public:
virtual ~IFunction() = default;
virtual String getName() const = 0;
virtual ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const = 0;
virtual ColumnPtr executeImplDryRun(const ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const
{
return executeImpl(arguments, result_type, input_rows_count);
}
...
};
using FunctionPtr = std::shared_ptr<IFunction>;
核心是virtual方法executeImpl
看看实现的模板类
## ClickHouse/src/Functions/FunctionNumericPredicate.h
ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override
{
const auto * in = arguments.front().column.get();
ColumnPtr res;
if (!((res = execute<UInt8>(in))
|| (res = execute<UInt16>(in))
|| (res = execute<UInt32>(in))
|| (res = execute<UInt64>(in))
|| (res = execute<Int8>(in))
|| (res = execute<Int16>(in))
|| (res = execute<Int32>(in))
|| (res = execute<Int64>(in))
|| (res = execute<Float32>(in))
|| (res = execute<Float64>(in))))
throw Exception{"Illegal column " + in->getName() + " of first argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN};
return res;
}
template <typename T>
ColumnPtr execute(const IColumn * in_untyped) const
{
if (const auto in = checkAndGetColumn<ColumnVector<T>>(in_untyped))
{
const auto size = in->size();
auto out = ColumnUInt8::create(size);
const auto & in_data = in->getData();
auto & out_data = out->getData();
for (const auto i : collections::range(0, size))
out_data[i] = Impl::execute(in_data[i]);
return out;
}
return nullptr;
}