2015-11-12 16:28:48 +00:00
|
|
|
#pragma once
|
|
|
|
|
2018-12-10 15:25:45 +00:00
|
|
|
#include <atomic>
|
|
|
|
#include <memory>
|
|
|
|
#include <variant>
|
2018-10-08 19:45:17 +00:00
|
|
|
#include <Columns/ColumnDecimal.h>
|
2017-04-01 09:19:00 +00:00
|
|
|
#include <Columns/ColumnString.h>
|
|
|
|
#include <Common/Arena.h>
|
2018-12-10 15:25:45 +00:00
|
|
|
#include <Common/HashTable/HashMap.h>
|
2019-05-17 14:34:25 +00:00
|
|
|
#include <Core/Block.h>
|
2018-12-10 15:25:45 +00:00
|
|
|
#include <common/StringRef.h>
|
2017-06-06 17:18:32 +00:00
|
|
|
#include <ext/range.h>
|
2018-12-10 15:25:45 +00:00
|
|
|
#include "DictionaryStructure.h"
|
|
|
|
#include "IDictionary.h"
|
|
|
|
#include "IDictionarySource.h"
|
2015-11-12 16:28:48 +00:00
|
|
|
|
|
|
|
|
|
|
|
namespace DB
|
|
|
|
{
|
2018-01-15 12:44:39 +00:00
|
|
|
using BlockPtr = std::shared_ptr<Block>;
|
2017-05-26 16:08:56 +00:00
|
|
|
|
2015-11-13 11:21:40 +00:00
|
|
|
class ComplexKeyHashedDictionary final : public IDictionaryBase
|
2015-11-12 16:28:48 +00:00
|
|
|
{
|
|
|
|
public:
|
2017-04-01 07:20:54 +00:00
|
|
|
ComplexKeyHashedDictionary(
|
2019-12-25 23:12:12 +00:00
|
|
|
const std::string & database_,
|
2019-08-03 11:02:40 +00:00
|
|
|
const std::string & name_,
|
|
|
|
const DictionaryStructure & dict_struct_,
|
|
|
|
DictionarySourcePtr source_ptr_,
|
|
|
|
const DictionaryLifetime dict_lifetime_,
|
|
|
|
bool require_nonempty_,
|
|
|
|
BlockPtr saved_block_ = nullptr);
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2018-06-03 20:39:06 +00:00
|
|
|
std::string getKeyDescription() const { return key_description; }
|
2015-11-13 01:44:41 +00:00
|
|
|
|
2019-12-25 23:12:12 +00:00
|
|
|
const std::string & getDatabase() const override { return database; }
|
|
|
|
const std::string & getName() const override { return name; }
|
|
|
|
const std::string & getFullName() const override { return full_name; }
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
std::string getTypeName() const override { return "ComplexKeyHashed"; }
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2017-07-21 06:35:58 +00:00
|
|
|
size_t getBytesAllocated() const override { return bytes_allocated; }
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2017-07-21 06:35:58 +00:00
|
|
|
size_t getQueryCount() const override { return query_count.load(std::memory_order_relaxed); }
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
double getHitRate() const override { return 1.0; }
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2017-07-21 06:35:58 +00:00
|
|
|
size_t getElementCount() const override { return element_count; }
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
double getLoadFactor() const override { return static_cast<double>(element_count) / bucket_count; }
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2019-06-02 12:11:01 +00:00
|
|
|
std::shared_ptr<const IExternalLoadable> clone() const override
|
2019-01-19 23:27:52 +00:00
|
|
|
{
|
2019-12-25 23:12:12 +00:00
|
|
|
return std::make_shared<ComplexKeyHashedDictionary>(database, name, dict_struct, source_ptr->clone(), dict_lifetime, require_nonempty, saved_block);
|
2019-01-19 23:27:52 +00:00
|
|
|
}
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
const IDictionarySource * getSource() const override { return source_ptr.get(); }
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
const DictionaryLifetime & getLifetime() const override { return dict_lifetime; }
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
const DictionaryStructure & getStructure() const override { return dict_struct; }
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
bool isInjective(const std::string & attribute_name) const override
|
|
|
|
{
|
|
|
|
return dict_struct.attributes[&getAttribute(attribute_name) - attributes.data()].injective;
|
|
|
|
}
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2018-10-08 19:45:17 +00:00
|
|
|
template <typename T>
|
|
|
|
using ResultArrayType = std::conditional_t<IsDecimalNumber<T>, DecimalPaddedPODArray<T>, PaddedPODArray<T>>;
|
|
|
|
|
2018-12-10 15:25:45 +00:00
|
|
|
#define DECLARE(TYPE) \
|
2018-12-10 15:50:58 +00:00
|
|
|
void get##TYPE( \
|
2018-12-10 15:25:45 +00:00
|
|
|
const std::string & attribute_name, const Columns & key_columns, const DataTypes & key_types, ResultArrayType<TYPE> & out) const;
|
2017-04-01 07:20:54 +00:00
|
|
|
DECLARE(UInt8)
|
|
|
|
DECLARE(UInt16)
|
|
|
|
DECLARE(UInt32)
|
|
|
|
DECLARE(UInt64)
|
2017-11-14 00:08:54 +00:00
|
|
|
DECLARE(UInt128)
|
2017-04-01 07:20:54 +00:00
|
|
|
DECLARE(Int8)
|
|
|
|
DECLARE(Int16)
|
|
|
|
DECLARE(Int32)
|
|
|
|
DECLARE(Int64)
|
|
|
|
DECLARE(Float32)
|
|
|
|
DECLARE(Float64)
|
2018-10-08 19:45:17 +00:00
|
|
|
DECLARE(Decimal32)
|
|
|
|
DECLARE(Decimal64)
|
|
|
|
DECLARE(Decimal128)
|
2015-11-20 15:53:23 +00:00
|
|
|
#undef DECLARE
|
2016-06-07 21:07:44 +00:00
|
|
|
|
2018-12-10 15:25:45 +00:00
|
|
|
void getString(const std::string & attribute_name, const Columns & key_columns, const DataTypes & key_types, ColumnString * out) const;
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2018-12-10 15:50:58 +00:00
|
|
|
#define DECLARE(TYPE) \
|
|
|
|
void get##TYPE( \
|
2018-12-10 15:25:45 +00:00
|
|
|
const std::string & attribute_name, \
|
2018-12-10 15:50:58 +00:00
|
|
|
const Columns & key_columns, \
|
|
|
|
const DataTypes & key_types, \
|
|
|
|
const PaddedPODArray<TYPE> & def, \
|
2018-12-10 15:25:45 +00:00
|
|
|
ResultArrayType<TYPE> & out) const;
|
2017-04-01 07:20:54 +00:00
|
|
|
DECLARE(UInt8)
|
|
|
|
DECLARE(UInt16)
|
|
|
|
DECLARE(UInt32)
|
|
|
|
DECLARE(UInt64)
|
2017-11-14 00:08:54 +00:00
|
|
|
DECLARE(UInt128)
|
2017-04-01 07:20:54 +00:00
|
|
|
DECLARE(Int8)
|
|
|
|
DECLARE(Int16)
|
|
|
|
DECLARE(Int32)
|
|
|
|
DECLARE(Int64)
|
|
|
|
DECLARE(Float32)
|
|
|
|
DECLARE(Float64)
|
2018-10-08 19:45:17 +00:00
|
|
|
DECLARE(Decimal32)
|
|
|
|
DECLARE(Decimal64)
|
|
|
|
DECLARE(Decimal128)
|
2015-11-20 15:53:23 +00:00
|
|
|
#undef DECLARE
|
2016-06-07 21:07:44 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
void getString(
|
2018-12-10 15:25:45 +00:00
|
|
|
const std::string & attribute_name,
|
|
|
|
const Columns & key_columns,
|
|
|
|
const DataTypes & key_types,
|
|
|
|
const ColumnString * const def,
|
|
|
|
ColumnString * const out) const;
|
|
|
|
|
2018-12-10 15:50:58 +00:00
|
|
|
#define DECLARE(TYPE) \
|
|
|
|
void get##TYPE( \
|
2018-12-10 15:25:45 +00:00
|
|
|
const std::string & attribute_name, \
|
2018-12-10 15:50:58 +00:00
|
|
|
const Columns & key_columns, \
|
|
|
|
const DataTypes & key_types, \
|
|
|
|
const TYPE def, \
|
2018-12-10 15:25:45 +00:00
|
|
|
ResultArrayType<TYPE> & out) const;
|
2017-04-01 07:20:54 +00:00
|
|
|
DECLARE(UInt8)
|
|
|
|
DECLARE(UInt16)
|
|
|
|
DECLARE(UInt32)
|
|
|
|
DECLARE(UInt64)
|
2017-11-14 00:08:54 +00:00
|
|
|
DECLARE(UInt128)
|
2017-04-01 07:20:54 +00:00
|
|
|
DECLARE(Int8)
|
|
|
|
DECLARE(Int16)
|
|
|
|
DECLARE(Int32)
|
|
|
|
DECLARE(Int64)
|
|
|
|
DECLARE(Float32)
|
|
|
|
DECLARE(Float64)
|
2018-10-08 19:45:17 +00:00
|
|
|
DECLARE(Decimal32)
|
|
|
|
DECLARE(Decimal64)
|
|
|
|
DECLARE(Decimal128)
|
2015-11-20 15:53:23 +00:00
|
|
|
#undef DECLARE
|
2016-06-07 21:07:44 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
void getString(
|
2018-12-10 15:25:45 +00:00
|
|
|
const std::string & attribute_name,
|
|
|
|
const Columns & key_columns,
|
|
|
|
const DataTypes & key_types,
|
|
|
|
const String & def,
|
|
|
|
ColumnString * const out) const;
|
2015-11-20 15:53:23 +00:00
|
|
|
|
2017-05-25 19:52:05 +00:00
|
|
|
void has(const Columns & key_columns, const DataTypes & key_types, PaddedPODArray<UInt8> & out) const;
|
2015-11-19 13:15:02 +00:00
|
|
|
|
2019-02-18 18:51:46 +00:00
|
|
|
BlockInputStreamPtr getBlockInputStream(const Names & column_names, size_t max_block_size) const override;
|
2017-05-04 18:14:23 +00:00
|
|
|
|
2015-11-12 16:28:48 +00:00
|
|
|
private:
|
2018-12-10 15:25:45 +00:00
|
|
|
template <typename Value>
|
|
|
|
using ContainerType = HashMapWithSavedHash<StringRef, Value, StringRefHash>;
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
struct Attribute final
|
|
|
|
{
|
|
|
|
AttributeUnderlyingType type;
|
2018-11-11 19:29:52 +00:00
|
|
|
std::variant<
|
2018-12-10 15:25:45 +00:00
|
|
|
UInt8,
|
|
|
|
UInt16,
|
|
|
|
UInt32,
|
|
|
|
UInt64,
|
2017-11-14 00:08:54 +00:00
|
|
|
UInt128,
|
2018-12-10 15:25:45 +00:00
|
|
|
Int8,
|
|
|
|
Int16,
|
|
|
|
Int32,
|
|
|
|
Int64,
|
|
|
|
Decimal32,
|
|
|
|
Decimal64,
|
|
|
|
Decimal128,
|
|
|
|
Float32,
|
|
|
|
Float64,
|
|
|
|
String>
|
|
|
|
null_values;
|
2018-11-11 19:29:52 +00:00
|
|
|
std::variant<
|
2018-12-10 15:25:45 +00:00
|
|
|
ContainerType<UInt8>,
|
|
|
|
ContainerType<UInt16>,
|
|
|
|
ContainerType<UInt32>,
|
|
|
|
ContainerType<UInt64>,
|
2018-11-13 19:43:17 +00:00
|
|
|
ContainerType<UInt128>,
|
2018-12-10 15:25:45 +00:00
|
|
|
ContainerType<Int8>,
|
|
|
|
ContainerType<Int16>,
|
|
|
|
ContainerType<Int32>,
|
|
|
|
ContainerType<Int64>,
|
|
|
|
ContainerType<Decimal32>,
|
|
|
|
ContainerType<Decimal64>,
|
|
|
|
ContainerType<Decimal128>,
|
|
|
|
ContainerType<Float32>,
|
|
|
|
ContainerType<Float64>,
|
|
|
|
ContainerType<StringRef>>
|
|
|
|
maps;
|
2017-04-01 07:20:54 +00:00
|
|
|
std::unique_ptr<Arena> string_arena;
|
|
|
|
};
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
void createAttributes();
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2018-02-15 13:08:23 +00:00
|
|
|
void blockToAttributes(const Block & block);
|
|
|
|
|
2018-01-15 12:44:39 +00:00
|
|
|
void updateData();
|
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
void loadData();
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
template <typename T>
|
|
|
|
void addAttributeSize(const Attribute & attribute);
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
void calculateBytesAllocated();
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
template <typename T>
|
|
|
|
void createAttributeImpl(Attribute & attribute, const Field & null_value);
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
Attribute createAttributeWithType(const AttributeUnderlyingType type, const Field & null_value);
|
2015-11-13 01:44:41 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
template <typename AttributeType, typename OutputType, typename ValueSetter, typename DefaultGetter>
|
2018-12-10 15:25:45 +00:00
|
|
|
void
|
|
|
|
getItemsImpl(const Attribute & attribute, const Columns & key_columns, ValueSetter && set_value, DefaultGetter && get_default) const;
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
template <typename T>
|
|
|
|
bool setAttributeValueImpl(Attribute & attribute, const StringRef key, const T value);
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
bool setAttributeValue(Attribute & attribute, const StringRef key, const Field & value);
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
const Attribute & getAttribute(const std::string & attribute_name) const;
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2018-12-10 15:25:45 +00:00
|
|
|
static StringRef placeKeysInPool(const size_t row, const Columns & key_columns, StringRefs & keys, Arena & pool);
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
template <typename T>
|
2017-05-25 19:52:05 +00:00
|
|
|
void has(const Attribute & attribute, const Columns & key_columns, PaddedPODArray<UInt8> & out) const;
|
2015-11-19 13:15:02 +00:00
|
|
|
|
2017-05-04 18:14:23 +00:00
|
|
|
std::vector<StringRef> getKeys() const;
|
|
|
|
|
|
|
|
template <typename T>
|
2017-05-30 15:02:44 +00:00
|
|
|
std::vector<StringRef> getKeys(const Attribute & attribute) const;
|
2017-05-04 18:14:23 +00:00
|
|
|
|
2019-12-25 23:12:12 +00:00
|
|
|
const std::string database;
|
2017-04-01 07:20:54 +00:00
|
|
|
const std::string name;
|
2019-12-25 23:12:12 +00:00
|
|
|
const std::string full_name;
|
2017-04-01 07:20:54 +00:00
|
|
|
const DictionaryStructure dict_struct;
|
|
|
|
const DictionarySourcePtr source_ptr;
|
|
|
|
const DictionaryLifetime dict_lifetime;
|
|
|
|
const bool require_nonempty;
|
|
|
|
const std::string key_description{dict_struct.getKeyDescription()};
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2017-07-21 06:35:58 +00:00
|
|
|
std::map<std::string, size_t> attribute_index_by_name;
|
2017-04-01 07:20:54 +00:00
|
|
|
std::vector<Attribute> attributes;
|
|
|
|
Arena keys_pool;
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2017-07-21 06:35:58 +00:00
|
|
|
size_t bytes_allocated = 0;
|
|
|
|
size_t element_count = 0;
|
|
|
|
size_t bucket_count = 0;
|
|
|
|
mutable std::atomic<size_t> query_count{0};
|
2015-11-12 16:28:48 +00:00
|
|
|
|
2018-01-15 12:44:39 +00:00
|
|
|
BlockPtr saved_block;
|
2015-11-12 16:28:48 +00:00
|
|
|
};
|
|
|
|
|
|
|
|
}
|