2017-05-12 21:23:12 +00:00
|
|
|
#pragma once
|
|
|
|
|
2018-11-28 11:37:12 +00:00
|
|
|
#include "IDictionary.h"
|
|
|
|
#include "IDictionarySource.h"
|
|
|
|
#include "DictionaryStructure.h"
|
2017-06-23 20:22:35 +00:00
|
|
|
#include <common/StringRef.h>
|
2017-05-12 21:23:12 +00:00
|
|
|
#include <Common/HashTable/HashMap.h>
|
2018-10-08 19:45:17 +00:00
|
|
|
#include <Columns/ColumnDecimal.h>
|
2017-05-12 21:23:12 +00:00
|
|
|
#include <Columns/ColumnString.h>
|
|
|
|
#include <Common/Arena.h>
|
2017-06-06 17:18:32 +00:00
|
|
|
#include <ext/range.h>
|
2017-05-12 21:23:12 +00:00
|
|
|
#include <atomic>
|
|
|
|
#include <memory>
|
2018-11-11 19:29:52 +00:00
|
|
|
#include <variant>
|
2017-06-05 09:02:05 +00:00
|
|
|
#include <common/logger_useful.h>
|
2017-05-12 21:23:12 +00:00
|
|
|
|
2017-06-23 20:22:35 +00:00
|
|
|
struct btrie_s;
|
|
|
|
typedef struct btrie_s btrie_t;
|
|
|
|
|
2017-05-12 21:23:12 +00:00
|
|
|
namespace DB
|
|
|
|
{
|
|
|
|
|
|
|
|
class TrieDictionary final : public IDictionaryBase
|
|
|
|
{
|
|
|
|
public:
|
|
|
|
TrieDictionary(
|
|
|
|
const std::string & name, const DictionaryStructure & dict_struct, DictionarySourcePtr source_ptr,
|
|
|
|
const DictionaryLifetime dict_lifetime, bool require_nonempty);
|
|
|
|
|
|
|
|
TrieDictionary(const TrieDictionary & other);
|
|
|
|
|
2018-06-03 20:39:06 +00:00
|
|
|
~TrieDictionary() override;
|
2017-05-12 21:23:12 +00:00
|
|
|
|
2018-06-03 20:39:06 +00:00
|
|
|
std::string getKeyDescription() const { return key_description; }
|
2017-05-12 21:23:12 +00:00
|
|
|
|
|
|
|
std::exception_ptr getCreationException() const override { return creation_exception; }
|
|
|
|
|
|
|
|
std::string getName() const override { return name; }
|
|
|
|
|
|
|
|
std::string getTypeName() const override { return "Trie"; }
|
|
|
|
|
2017-07-21 06:35:58 +00:00
|
|
|
size_t getBytesAllocated() const override { return bytes_allocated; }
|
2017-05-12 21:23:12 +00:00
|
|
|
|
2017-07-21 06:35:58 +00:00
|
|
|
size_t getQueryCount() const override { return query_count.load(std::memory_order_relaxed); }
|
2017-05-12 21:23:12 +00:00
|
|
|
|
|
|
|
double getHitRate() const override { return 1.0; }
|
|
|
|
|
2017-07-21 06:35:58 +00:00
|
|
|
size_t getElementCount() const override { return element_count; }
|
2017-05-12 21:23:12 +00:00
|
|
|
|
|
|
|
double getLoadFactor() const override { return static_cast<double>(element_count) / bucket_count; }
|
|
|
|
|
|
|
|
bool isCached() const override { return false; }
|
|
|
|
|
2017-10-26 13:19:58 +00:00
|
|
|
std::unique_ptr<IExternalLoadable> clone() const override { return std::make_unique<TrieDictionary>(*this); }
|
2017-05-12 21:23:12 +00:00
|
|
|
|
|
|
|
const IDictionarySource * getSource() const override { return source_ptr.get(); }
|
|
|
|
|
|
|
|
const DictionaryLifetime & getLifetime() const override { return dict_lifetime; }
|
|
|
|
|
|
|
|
const DictionaryStructure & getStructure() const override { return dict_struct; }
|
|
|
|
|
|
|
|
std::chrono::time_point<std::chrono::system_clock> getCreationTime() const override
|
|
|
|
{
|
|
|
|
return creation_time;
|
|
|
|
}
|
|
|
|
|
|
|
|
bool isInjective(const std::string & attribute_name) const override
|
|
|
|
{
|
|
|
|
return dict_struct.attributes[&getAttribute(attribute_name) - attributes.data()].injective;
|
|
|
|
}
|
|
|
|
|
2018-10-08 19:45:17 +00:00
|
|
|
template <typename T>
|
|
|
|
using ResultArrayType = std::conditional_t<IsDecimalNumber<T>, DecimalPaddedPODArray<T>, PaddedPODArray<T>>;
|
|
|
|
|
2017-05-12 21:23:12 +00:00
|
|
|
#define DECLARE(TYPE)\
|
|
|
|
void get##TYPE(\
|
2017-05-25 19:52:05 +00:00
|
|
|
const std::string & attribute_name, const Columns & key_columns, const DataTypes & key_types,\
|
2018-10-08 19:45:17 +00:00
|
|
|
ResultArrayType<TYPE> & out) const;
|
2017-05-12 21:23:12 +00:00
|
|
|
DECLARE(UInt8)
|
|
|
|
DECLARE(UInt16)
|
|
|
|
DECLARE(UInt32)
|
|
|
|
DECLARE(UInt64)
|
2017-11-14 00:08:54 +00:00
|
|
|
DECLARE(UInt128)
|
2017-05-12 21:23:12 +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)
|
2017-05-12 21:23:12 +00:00
|
|
|
#undef DECLARE
|
|
|
|
|
|
|
|
void getString(
|
2017-05-25 19:52:05 +00:00
|
|
|
const std::string & attribute_name, const Columns & key_columns, const DataTypes & key_types,
|
2017-05-12 21:23:12 +00:00
|
|
|
ColumnString * out) const;
|
|
|
|
|
|
|
|
#define DECLARE(TYPE)\
|
|
|
|
void get##TYPE(\
|
2017-05-25 19:52:05 +00:00
|
|
|
const std::string & attribute_name, const Columns & key_columns, const DataTypes & key_types,\
|
2018-10-08 19:45:17 +00:00
|
|
|
const PaddedPODArray<TYPE> & def, ResultArrayType<TYPE> & out) const;
|
2017-05-12 21:23:12 +00:00
|
|
|
DECLARE(UInt8)
|
|
|
|
DECLARE(UInt16)
|
|
|
|
DECLARE(UInt32)
|
|
|
|
DECLARE(UInt64)
|
2017-11-14 00:08:54 +00:00
|
|
|
DECLARE(UInt128)
|
2017-05-12 21:23:12 +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)
|
2017-05-12 21:23:12 +00:00
|
|
|
#undef DECLARE
|
|
|
|
|
|
|
|
void getString(
|
2017-05-25 19:52:05 +00:00
|
|
|
const std::string & attribute_name, const Columns & key_columns, const DataTypes & key_types,
|
2017-05-12 21:23:12 +00:00
|
|
|
const ColumnString * const def, ColumnString * const out) const;
|
|
|
|
|
|
|
|
#define DECLARE(TYPE)\
|
|
|
|
void get##TYPE(\
|
2017-05-25 19:52:05 +00:00
|
|
|
const std::string & attribute_name, const Columns & key_columns, const DataTypes & key_types,\
|
2018-10-08 19:45:17 +00:00
|
|
|
const TYPE def, ResultArrayType<TYPE> & out) const;
|
2017-05-12 21:23:12 +00:00
|
|
|
DECLARE(UInt8)
|
|
|
|
DECLARE(UInt16)
|
|
|
|
DECLARE(UInt32)
|
|
|
|
DECLARE(UInt64)
|
2017-11-14 00:08:54 +00:00
|
|
|
DECLARE(UInt128)
|
2017-05-12 21:23:12 +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)
|
2017-05-12 21:23:12 +00:00
|
|
|
#undef DECLARE
|
|
|
|
|
|
|
|
void getString(
|
2017-05-25 19:52:05 +00:00
|
|
|
const std::string & attribute_name, const Columns & key_columns, const DataTypes & key_types,
|
2017-05-12 21:23:12 +00:00
|
|
|
const String & def, ColumnString * const out) const;
|
|
|
|
|
2017-05-25 19:52:05 +00:00
|
|
|
void has(const Columns & key_columns, const DataTypes & key_types, PaddedPODArray<UInt8> & out) const;
|
2017-05-12 21:23:12 +00:00
|
|
|
|
2017-06-05 09:02:05 +00:00
|
|
|
BlockInputStreamPtr getBlockInputStream(const Names & column_names, size_t max_block_size) const override;
|
2017-05-26 16:08:56 +00:00
|
|
|
|
2017-05-12 21:23:12 +00:00
|
|
|
private:
|
|
|
|
template <typename Value> using ContainerType = std::vector<Value>;
|
|
|
|
|
|
|
|
struct Attribute final
|
|
|
|
{
|
|
|
|
AttributeUnderlyingType type;
|
2018-11-11 19:29:52 +00:00
|
|
|
std::variant<
|
2017-05-12 21:23:12 +00:00
|
|
|
UInt8, UInt16, UInt32, UInt64,
|
2017-11-14 00:08:54 +00:00
|
|
|
UInt128,
|
2017-05-12 21:23:12 +00:00
|
|
|
Int8, Int16, Int32, Int64,
|
2018-10-08 19:45:17 +00:00
|
|
|
Decimal32, Decimal64, Decimal128,
|
2017-05-12 21:23:12 +00:00
|
|
|
Float32, Float64,
|
|
|
|
String> null_values;
|
2018-11-11 19:29:52 +00:00
|
|
|
std::variant<
|
2018-11-13 19:43:17 +00:00
|
|
|
ContainerType<UInt8>, ContainerType<UInt16>, ContainerType<UInt32>, ContainerType<UInt64>,
|
|
|
|
ContainerType<UInt128>,
|
|
|
|
ContainerType<Int8>, ContainerType<Int16>, ContainerType<Int32>, ContainerType<Int64>,
|
|
|
|
ContainerType<Decimal32>, ContainerType<Decimal64>, ContainerType<Decimal128>,
|
|
|
|
ContainerType<Float32>, ContainerType<Float64>,
|
|
|
|
ContainerType<StringRef>> maps;
|
2017-05-12 21:23:12 +00:00
|
|
|
std::unique_ptr<Arena> string_arena;
|
|
|
|
};
|
|
|
|
|
|
|
|
void createAttributes();
|
|
|
|
|
|
|
|
void loadData();
|
|
|
|
|
|
|
|
template <typename T>
|
|
|
|
void addAttributeSize(const Attribute & attribute);
|
|
|
|
|
|
|
|
void calculateBytesAllocated();
|
|
|
|
|
|
|
|
void validateKeyTypes(const DataTypes & key_types) const;
|
|
|
|
|
|
|
|
template <typename T>
|
|
|
|
void createAttributeImpl(Attribute & attribute, const Field & null_value);
|
|
|
|
|
|
|
|
Attribute createAttributeWithType(const AttributeUnderlyingType type, const Field & null_value);
|
|
|
|
|
|
|
|
|
|
|
|
template <typename OutputType, typename ValueSetter, typename DefaultGetter>
|
|
|
|
void getItemsNumber(
|
|
|
|
const Attribute & attribute,
|
2017-05-25 19:52:05 +00:00
|
|
|
const Columns & key_columns,
|
2017-05-12 21:23:12 +00:00
|
|
|
ValueSetter && set_value,
|
|
|
|
DefaultGetter && get_default) const;
|
|
|
|
|
|
|
|
template <typename AttributeType, typename OutputType, typename ValueSetter, typename DefaultGetter>
|
|
|
|
void getItemsImpl(
|
|
|
|
const Attribute & attribute,
|
2017-05-25 19:52:05 +00:00
|
|
|
const Columns & key_columns,
|
2017-05-12 21:23:12 +00:00
|
|
|
ValueSetter && set_value,
|
|
|
|
DefaultGetter && get_default) const;
|
|
|
|
|
|
|
|
|
|
|
|
template <typename T>
|
|
|
|
bool setAttributeValueImpl(Attribute & attribute, const StringRef key, const T value);
|
|
|
|
|
|
|
|
bool setAttributeValue(Attribute & attribute, const StringRef key, const Field & value);
|
|
|
|
|
|
|
|
const Attribute & getAttribute(const std::string & attribute_name) const;
|
|
|
|
|
|
|
|
template <typename T>
|
2017-05-25 19:52:05 +00:00
|
|
|
void has(const Attribute & attribute, const Columns & key_columns, PaddedPODArray<UInt8> & out) const;
|
2017-05-12 21:23:12 +00:00
|
|
|
|
2017-06-05 09:02:05 +00:00
|
|
|
template <typename Getter, typename KeyType>
|
|
|
|
void trieTraverse(const btrie_t * trie, Getter && getter) const;
|
|
|
|
|
|
|
|
Columns getKeyColumns() const;
|
|
|
|
|
2017-05-12 21:23:12 +00:00
|
|
|
const std::string name;
|
|
|
|
const DictionaryStructure dict_struct;
|
|
|
|
const DictionarySourcePtr source_ptr;
|
|
|
|
const DictionaryLifetime dict_lifetime;
|
|
|
|
const bool require_nonempty;
|
|
|
|
const std::string key_description{dict_struct.getKeyDescription()};
|
|
|
|
|
|
|
|
|
2017-07-12 11:46:52 +00:00
|
|
|
btrie_t * trie = nullptr;
|
2017-07-21 06:35:58 +00:00
|
|
|
std::map<std::string, size_t> attribute_index_by_name;
|
2017-05-12 21:23:12 +00:00
|
|
|
std::vector<Attribute> attributes;
|
|
|
|
|
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};
|
2017-05-12 21:23:12 +00:00
|
|
|
|
|
|
|
std::chrono::time_point<std::chrono::system_clock> creation_time;
|
|
|
|
|
|
|
|
std::exception_ptr creation_exception;
|
2017-06-05 09:02:05 +00:00
|
|
|
|
|
|
|
Logger * logger;
|
2017-05-12 21:23:12 +00:00
|
|
|
};
|
|
|
|
|
|
|
|
}
|