#pragma once #include #include #include #include #include #include #include #include #include #include #include "DictionaryStructure.h" #include "IDictionary.h" #include "IDictionarySource.h" struct btrie_s; typedef struct btrie_s btrie_t; 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); ~TrieDictionary() override; std::string getKeyDescription() const { return key_description; } std::exception_ptr getCreationException() const override { return creation_exception; } std::string getName() const override { return name; } std::string getTypeName() const override { return "Trie"; } size_t getBytesAllocated() const override { return bytes_allocated; } size_t getQueryCount() const override { return query_count.load(std::memory_order_relaxed); } double getHitRate() const override { return 1.0; } size_t getElementCount() const override { return element_count; } double getLoadFactor() const override { return static_cast(element_count) / bucket_count; } bool isCached() const override { return false; } std::unique_ptr clone() const override { return std::make_unique(*this); } 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 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; } template using ResultArrayType = std::conditional_t, DecimalPaddedPODArray, PaddedPODArray>; #define DECLARE(TYPE) \ void get##TYPE( \ const std::string & attribute_name, const Columns & key_columns, const DataTypes & key_types, ResultArrayType & out) const; DECLARE(UInt8) DECLARE(UInt16) DECLARE(UInt32) DECLARE(UInt64) DECLARE(UInt128) DECLARE(Int8) DECLARE(Int16) DECLARE(Int32) DECLARE(Int64) DECLARE(Float32) DECLARE(Float64) DECLARE(Decimal32) DECLARE(Decimal64) DECLARE(Decimal128) #undef DECLARE void getString(const std::string & attribute_name, const Columns & key_columns, const DataTypes & key_types, ColumnString * out) const; #define DECLARE(TYPE) \ void get##TYPE( \ const std::string & attribute_name, \ const Columns & key_columns, \ const DataTypes & key_types, \ const PaddedPODArray & def, \ ResultArrayType & out) const; DECLARE(UInt8) DECLARE(UInt16) DECLARE(UInt32) DECLARE(UInt64) DECLARE(UInt128) DECLARE(Int8) DECLARE(Int16) DECLARE(Int32) DECLARE(Int64) DECLARE(Float32) DECLARE(Float64) DECLARE(Decimal32) DECLARE(Decimal64) DECLARE(Decimal128) #undef DECLARE void getString( const std::string & attribute_name, const Columns & key_columns, const DataTypes & key_types, const ColumnString * const def, ColumnString * const out) const; #define DECLARE(TYPE) \ void get##TYPE( \ const std::string & attribute_name, \ const Columns & key_columns, \ const DataTypes & key_types, \ const TYPE def, \ ResultArrayType & out) const; DECLARE(UInt8) DECLARE(UInt16) DECLARE(UInt32) DECLARE(UInt64) DECLARE(UInt128) DECLARE(Int8) DECLARE(Int16) DECLARE(Int32) DECLARE(Int64) DECLARE(Float32) DECLARE(Float64) DECLARE(Decimal32) DECLARE(Decimal64) DECLARE(Decimal128) #undef DECLARE void getString( const std::string & attribute_name, const Columns & key_columns, const DataTypes & key_types, const String & def, ColumnString * const out) const; void has(const Columns & key_columns, const DataTypes & key_types, PaddedPODArray & out) const; BlockInputStreamPtr getBlockInputStream(const Names & column_names, size_t max_block_size) const override; private: template using ContainerType = std::vector; struct Attribute final { AttributeUnderlyingType type; std::variant< UInt8, UInt16, UInt32, UInt64, UInt128, Int8, Int16, Int32, Int64, Decimal32, Decimal64, Decimal128, Float32, Float64, String> null_values; std::variant< ContainerType, ContainerType, ContainerType, ContainerType, ContainerType, ContainerType, ContainerType, ContainerType, ContainerType, ContainerType, ContainerType, ContainerType, ContainerType, ContainerType, ContainerType> maps; std::unique_ptr string_arena; }; void createAttributes(); void loadData(); template void addAttributeSize(const Attribute & attribute); void calculateBytesAllocated(); void validateKeyTypes(const DataTypes & key_types) const; template void createAttributeImpl(Attribute & attribute, const Field & null_value); Attribute createAttributeWithType(const AttributeUnderlyingType type, const Field & null_value); template void getItemsNumber(const Attribute & attribute, const Columns & key_columns, ValueSetter && set_value, DefaultGetter && get_default) const; template void getItemsImpl(const Attribute & attribute, const Columns & key_columns, ValueSetter && set_value, DefaultGetter && get_default) const; template 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 void has(const Attribute & attribute, const Columns & key_columns, PaddedPODArray & out) const; template void trieTraverse(const btrie_t * trie, Getter && getter) const; Columns getKeyColumns() const; 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()}; btrie_t * trie = nullptr; std::map attribute_index_by_name; std::vector attributes; size_t bytes_allocated = 0; size_t element_count = 0; size_t bucket_count = 0; mutable std::atomic query_count{0}; std::chrono::time_point creation_time; std::exception_ptr creation_exception; Logger * logger; }; }