ClickHouse/dbms/src/Storages/StorageFactory.cpp

276 lines
12 KiB
C++
Raw Normal View History

#include <Poco/Util/Application.h>
#include <Poco/Util/AbstractConfiguration.h>
2012-05-22 19:32:56 +00:00
#include <DB/Parsers/ASTCreateQuery.h>
#include <DB/Parsers/ASTIdentifier.h>
2012-05-30 05:53:09 +00:00
#include <DB/Parsers/ASTLiteral.h>
2012-05-22 19:32:56 +00:00
#include <DB/Interpreters/Context.h>
2011-10-31 17:30:44 +00:00
#include <DB/Storages/StorageLog.h>
2012-06-25 00:17:19 +00:00
#include <DB/Storages/StorageTinyLog.h>
2011-10-31 17:55:06 +00:00
#include <DB/Storages/StorageMemory.h>
2012-05-30 05:53:09 +00:00
#include <DB/Storages/StorageMerge.h>
2012-07-18 19:30:33 +00:00
#include <DB/Storages/StorageMergeTree.h>
2012-05-22 19:32:56 +00:00
#include <DB/Storages/StorageDistributed.h>
2011-10-31 17:30:44 +00:00
#include <DB/Storages/StorageSystemNumbers.h>
#include <DB/Storages/StorageSystemOne.h>
#include <DB/Storages/StorageFactory.h>
#include <DB/Storages/StorageView.h>
#include <DB/Storages/StorageMaterializedView.h>
#include <DB/Storages/StorageChunks.h>
#include <DB/Storages/StorageChunkRef.h>
2013-02-11 09:19:48 +00:00
#include <DB/Storages/StorageChunkMerger.h>
2011-10-31 17:30:44 +00:00
#include <DB/DataTypes/DataTypeArray.h>
#include <DB/DataTypes/DataTypeNested.h>
2011-10-31 17:30:44 +00:00
namespace DB
{
2014-03-18 17:20:44 +00:00
/** Для StorageMergeTree: достать первичный ключ в виде ASTExpressionList.
* Он может быть указан в кортеже: (CounterID, Date),
* или как один столбец: CounterID.
*/
static ASTPtr extractPrimaryKey(const ASTPtr & node, const std::string & storage_name)
{
const ASTFunction * primary_expr_func = dynamic_cast<const ASTFunction *>(&*node);
if (primary_expr_func && primary_expr_func->name == "tuple")
{
/// Первичный ключ указан в кортеже.
return primary_expr_func->children.at(0);
}
else
{
/// Первичный ключ состоит из одного столбца.
ASTExpressionList * res = new ASTExpressionList;
ASTPtr res_ptr = res;
res->children.push_back(node);
return res_ptr;
}
}
2011-10-31 17:30:44 +00:00
StoragePtr StorageFactory::get(
const String & name,
const String & data_path,
const String & table_name,
const String & database_name,
2012-05-22 19:32:56 +00:00
Context & context,
ASTPtr & query,
NamesAndTypesListPtr columns,
bool attach) const
2011-10-31 17:30:44 +00:00
{
columns = DataTypeNested::expandNestedColumns(*columns);
2011-10-31 17:30:44 +00:00
if (name == "Log")
2012-05-22 19:32:56 +00:00
{
return StorageLog::create(data_path, table_name, columns);
2012-06-25 00:17:19 +00:00
}
else if (name == "Chunks")
{
2013-02-08 17:06:29 +00:00
return StorageChunks::create(data_path, table_name, database_name, columns, context, attach);
}
else if (name == "ChunkRef")
{
throw Exception("Table with storage ChunkRef must not be created manually.", ErrorCodes::TABLE_MUST_NOT_BE_CREATED_MANUALLY);
}
else if (name == "View")
{
return StorageView::create(table_name, database_name, context, query, columns);
}
else if (name == "MaterializedView")
{
return StorageMaterializedView::create(table_name, database_name, context, query, columns, attach);
}
2013-02-11 09:19:48 +00:00
else if (name == "ChunkMerger")
{
ASTs & args_func = dynamic_cast<ASTFunction &>(*dynamic_cast<ASTCreateQuery &>(*query).storage).children;
do
{
if (args_func.size() != 1)
break;
ASTs & args = dynamic_cast<ASTExpressionList &>(*args_func.at(0)).children;
if (args.size() < 3 || args.size() > 4)
2013-02-11 09:19:48 +00:00
break;
String source_database = dynamic_cast<ASTIdentifier &>(*args[0]).name;
String source_table_name_regexp = safeGet<const String &>(dynamic_cast<ASTLiteral &>(*args[1]).value);
2013-03-14 11:43:43 +00:00
size_t chunks_to_merge = safeGet<UInt64>(dynamic_cast<ASTLiteral &>(*args[2]).value);
String destination_name_prefix = "group_";
String destination_database = source_database;
if (args.size() > 3)
destination_name_prefix = dynamic_cast<ASTIdentifier &>(*args[3]).name;
2013-02-11 09:19:48 +00:00
return StorageChunkMerger::create(database_name, table_name, columns, source_database, source_table_name_regexp, destination_name_prefix, chunks_to_merge, context);
2013-02-11 09:19:48 +00:00
} while(false);
throw Exception("Storage ChunkMerger requires from 3 to 4 parameters:"
" source database, regexp for source table names, number of chunks to merge, [destination tables name prefix].",
2013-02-11 09:19:48 +00:00
ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH);
}
2012-06-25 00:51:23 +00:00
else if (name == "TinyLog")
2012-06-25 00:17:19 +00:00
{
return StorageTinyLog::create(data_path, table_name, columns, attach);
2012-05-22 19:32:56 +00:00
}
2011-10-31 17:55:06 +00:00
else if (name == "Memory")
2012-05-22 19:32:56 +00:00
{
return StorageMemory::create(table_name, columns);
2012-05-22 19:32:56 +00:00
}
2012-05-30 05:53:09 +00:00
else if (name == "Merge")
{
/** В запросе в качестве аргумента для движка указано имя БД, в которой находятся таблицы-источники,
* а также регексп для имён таблиц-источников.
*/
ASTs & args_func = dynamic_cast<ASTFunction &>(*dynamic_cast<ASTCreateQuery &>(*query).storage).children;
if (args_func.size() != 1)
throw Exception("Storage Merge requires exactly 2 parameters"
" - name of source database and regexp for table names.",
ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH);
ASTs & args = dynamic_cast<ASTExpressionList &>(*args_func.at(0)).children;
if (args.size() != 2)
throw Exception("Storage Merge requires exactly 2 parameters"
" - name of source database and regexp for table names.",
ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH);
String source_database = dynamic_cast<ASTIdentifier &>(*args[0]).name;
String table_name_regexp = safeGet<const String &>(dynamic_cast<ASTLiteral &>(*args[1]).value);
2012-05-30 05:53:09 +00:00
return StorageMerge::create(table_name, columns, source_database, table_name_regexp, context);
2012-05-30 05:53:09 +00:00
}
2012-05-22 19:32:56 +00:00
else if (name == "Distributed")
{
/** В запросе в качестве аргумента для движка указано имя конфигурационной секции,
2012-05-30 05:53:09 +00:00
* в которой задан список удалённых серверов, а также имя удалённой БД и имя удалённой таблицы.
2012-05-22 19:32:56 +00:00
*/
2012-05-22 20:18:45 +00:00
ASTs & args_func = dynamic_cast<ASTFunction &>(*dynamic_cast<ASTCreateQuery &>(*query).storage).children;
if (args_func.size() != 1)
throw Exception("Storage Distributed requires 3 or 4 parameters"
" - name of configuration section with list of remote servers, name of remote database, name of remote table[, sign column name].",
2012-05-22 20:18:45 +00:00
ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH);
ASTs & args = dynamic_cast<ASTExpressionList &>(*args_func.at(0)).children;
2012-05-22 20:00:25 +00:00
if (args.size() != 3 && args.size() != 4)
throw Exception("Storage Distributed requires 3 or 4 parameters"
" - name of configuration section with list of remote servers, name of remote database, name of remote table[, sign column name].",
2012-05-22 19:32:56 +00:00
ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH);
String cluster_name = dynamic_cast<ASTIdentifier &>(*args[0]).name;
2012-05-22 19:32:56 +00:00
String remote_database = dynamic_cast<ASTIdentifier &>(*args[1]).name;
String remote_table = dynamic_cast<ASTIdentifier &>(*args[2]).name;
String sign_column_name = args.size() == 4 ? dynamic_cast<ASTIdentifier &>(*args[3]).name : "";
return StorageDistributed::create(table_name, columns, remote_database, remote_table, cluster_name,
2014-03-10 04:17:17 +00:00
context, sign_column_name);
2012-05-22 19:32:56 +00:00
}
2013-09-30 19:54:25 +00:00
else if (name == "MergeTree" || name == "SummingMergeTree")
2012-07-18 19:30:33 +00:00
{
/** В качестве аргумента для движка должно быть указано:
* - имя столбца с датой;
* - имя столбца для семплирования (запрос с SAMPLE x будет выбирать строки, у которых в этом столбце значение меньше, чем x*UINT32_MAX);
2012-07-18 19:30:33 +00:00
* - выражение для сортировки в скобках;
* - index_granularity.
* Например: ENGINE = MergeTree(EventDate, intHash32(UniqID), (CounterID, EventDate, intHash32(UniqID), EventTime), 8192).
*
2013-09-30 19:54:25 +00:00
* SummingMergeTree - вариант, в котором при слиянии делается суммирование всех числовых столбцов кроме PK
* - для Баннерной Крутилки.
2012-07-18 19:30:33 +00:00
*/
ASTs & args_func = dynamic_cast<ASTFunction &>(*dynamic_cast<ASTCreateQuery &>(*query).storage).children;
if (args_func.size() != 1)
2013-09-30 19:54:25 +00:00
throw Exception("Storage " + name + " requires 3 or 4 parameters"
" - name of column with date, [name of column for sampling], primary key expression, index granularity.",
2012-07-18 19:30:33 +00:00
ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH);
ASTs & args = dynamic_cast<ASTExpressionList &>(*args_func.at(0)).children;
if (args.size() != 3 && args.size() != 4)
2013-09-30 19:54:25 +00:00
throw Exception("Storage " + name + " requires 3 or 4 parameters"
" - name of column with date, [name of column for sampling], primary key expression, index granularity.",
2012-07-18 19:30:33 +00:00
ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH);
size_t arg_offset = args.size() - 3;
2012-07-18 19:30:33 +00:00
String date_column_name = dynamic_cast<ASTIdentifier &>(*args[0]).name;
2012-12-12 15:45:08 +00:00
ASTPtr sampling_expression = arg_offset == 0 ? NULL : args[1];
UInt64 index_granularity = safeGet<UInt64>(dynamic_cast<ASTLiteral &>(*args[arg_offset + 2]).value);
2012-07-18 19:30:33 +00:00
2014-03-18 17:20:44 +00:00
ASTPtr primary_expr_list = extractPrimaryKey(args[arg_offset + 1], name);
2012-07-18 19:30:33 +00:00
2013-09-30 19:54:25 +00:00
return StorageMergeTree::create(
2014-03-18 17:20:44 +00:00
data_path, table_name, columns, context, primary_expr_list, date_column_name, sampling_expression, index_granularity,
2014-03-09 17:36:01 +00:00
name == "SummingMergeTree" ? MergeTreeData::Summing : MergeTreeData::Ordinary);
2012-07-18 19:30:33 +00:00
}
2012-08-16 17:27:40 +00:00
else if (name == "CollapsingMergeTree")
{
/** В качестве аргумента для движка должно быть указано:
* - имя столбца с датой;
* - имя столбца для семплирования (запрос с SAMPLE x будет выбирать строки, у которых в этом столбце значение меньше, чем x*UINT32_MAX);
2012-08-16 17:27:40 +00:00
* - выражение для сортировки в скобках;
* - index_granularity;
* - имя столбца, содержащего тип строчки с изменением "визита" (принимающего значения 1 и -1).
* Например: ENGINE = CollapsingMergeTree(EventDate, (CounterID, EventDate, intHash32(UniqID), VisitID), 8192, Sign).
2012-08-16 17:27:40 +00:00
*/
ASTs & args_func = dynamic_cast<ASTFunction &>(*dynamic_cast<ASTCreateQuery &>(*query).storage).children;
if (args_func.size() != 1)
throw Exception("Storage CollapsingMergeTree requires 4 or 5 parameters"
" - name of column with date, [name of column for sampling], primary key expression, index granularity, sign_column.",
2012-08-16 17:27:40 +00:00
ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH);
ASTs & args = dynamic_cast<ASTExpressionList &>(*args_func.at(0)).children;
if (args.size() != 4 && args.size() != 5)
throw Exception("Storage CollapsingMergeTree requires 4 or 5 parameters"
" - name of column with date, [name of column for sampling], primary key expression, index granularity, sign_column.",
2012-08-16 17:27:40 +00:00
ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH);
size_t arg_offset = args.size() - 4;
2012-08-16 17:27:40 +00:00
String date_column_name = dynamic_cast<ASTIdentifier &>(*args[0]).name;
2012-12-12 15:45:08 +00:00
ASTPtr sampling_expression = arg_offset == 0 ? NULL : args[1];
UInt64 index_granularity = safeGet<UInt64>(dynamic_cast<ASTLiteral &>(*args[arg_offset + 2]).value);
String sign_column_name = dynamic_cast<ASTIdentifier &>(*args[arg_offset + 3]).name;
2012-08-16 17:27:40 +00:00
2014-03-18 17:20:44 +00:00
ASTPtr primary_expr_list = extractPrimaryKey(args[arg_offset + 1], name);
2012-08-16 17:27:40 +00:00
2013-09-30 19:54:25 +00:00
return StorageMergeTree::create(
2014-03-18 17:20:44 +00:00
data_path, table_name, columns, context, primary_expr_list, date_column_name,
2014-03-09 17:36:01 +00:00
sampling_expression, index_granularity, MergeTreeData::Collapsing, sign_column_name);
2012-08-16 17:27:40 +00:00
}
2011-10-31 17:30:44 +00:00
else if (name == "SystemNumbers")
{
if (columns->size() != 1 || columns->begin()->first != "number" || columns->begin()->second->getName() != "UInt64")
throw Exception("Storage SystemNumbers only allows one column with name 'number' and type 'UInt64'",
ErrorCodes::ILLEGAL_COLUMN);
return StorageSystemNumbers::create(table_name);
2011-10-31 17:30:44 +00:00
}
else if (name == "SystemOne")
{
if (columns->size() != 1 || columns->begin()->first != "dummy" || columns->begin()->second->getName() != "UInt8")
throw Exception("Storage SystemOne only allows one column with name 'dummy' and type 'UInt8'",
ErrorCodes::ILLEGAL_COLUMN);
return StorageSystemOne::create(table_name);
2011-10-31 17:30:44 +00:00
}
else
throw Exception("Unknown storage " + name, ErrorCodes::UNKNOWN_STORAGE);
}
}