#pragma once #include #include #include #include "DictionaryStructure.h" #include "IDictionary.h" #include "IDictionarySource.h" #include #include #include namespace DB { class RangeHashedDictionary final : public IDictionaryBase { public: RangeHashedDictionary( const std::string & database_, const std::string & name_, const DictionaryStructure & dict_struct_, DictionarySourcePtr source_ptr_, const DictionaryLifetime dict_lifetime_, bool require_nonempty_); 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; } std::string getTypeName() const override { return "RangeHashed"; } 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; } std::shared_ptr clone() const override { return std::make_shared(database, name, dict_struct, source_ptr->clone(), dict_lifetime, require_nonempty); } 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; } bool isInjective(const std::string & attribute_name) const override { return dict_struct.attributes[&getAttribute(attribute_name) - attributes.data()].injective; } typedef Int64 RangeStorageType; template using ResultArrayType = std::conditional_t, DecimalPaddedPODArray, PaddedPODArray>; #define DECLARE_MULTIPLE_GETTER(TYPE) \ void get##TYPE( \ const std::string & attribute_name, \ const PaddedPODArray & ids, \ const PaddedPODArray & dates, \ ResultArrayType & out) const; DECLARE_MULTIPLE_GETTER(UInt8) DECLARE_MULTIPLE_GETTER(UInt16) DECLARE_MULTIPLE_GETTER(UInt32) DECLARE_MULTIPLE_GETTER(UInt64) DECLARE_MULTIPLE_GETTER(UInt128) DECLARE_MULTIPLE_GETTER(Int8) DECLARE_MULTIPLE_GETTER(Int16) DECLARE_MULTIPLE_GETTER(Int32) DECLARE_MULTIPLE_GETTER(Int64) DECLARE_MULTIPLE_GETTER(Float32) DECLARE_MULTIPLE_GETTER(Float64) DECLARE_MULTIPLE_GETTER(Decimal32) DECLARE_MULTIPLE_GETTER(Decimal64) DECLARE_MULTIPLE_GETTER(Decimal128) #undef DECLARE_MULTIPLE_GETTER void getString( const std::string & attribute_name, const PaddedPODArray & ids, const PaddedPODArray & dates, ColumnString * out) const; BlockInputStreamPtr getBlockInputStream(const Names & column_names, size_t max_block_size) const override; struct Range { RangeStorageType left; RangeStorageType right; static bool isCorrectDate(const RangeStorageType & date); bool contains(const RangeStorageType & value) const; }; private: template struct Value final { Range range; T value; }; template using Values = std::vector>; template using Collection = HashMap>; template using Ptr = std::unique_ptr>; struct Attribute final { public: AttributeUnderlyingType type; std::variant< UInt8, UInt16, UInt32, UInt64, UInt128, Int8, Int16, Int32, Int64, Decimal32, Decimal64, Decimal128, Float32, Float64, String> null_values; std::variant< Ptr, Ptr, Ptr, Ptr, Ptr, Ptr, Ptr, Ptr, Ptr, Ptr, Ptr, Ptr, Ptr, Ptr, Ptr> maps; std::unique_ptr string_arena; }; void createAttributes(); void loadData(); template void addAttributeSize(const Attribute & attribute); void calculateBytesAllocated(); template void createAttributeImpl(Attribute & attribute, const Field & null_value); Attribute createAttributeWithType(const AttributeUnderlyingType type, const Field & null_value); template void getItems( const Attribute & attribute, const PaddedPODArray & ids, const PaddedPODArray & dates, PaddedPODArray & out) const; template void getItemsImpl( const Attribute & attribute, const PaddedPODArray & ids, const PaddedPODArray & dates, PaddedPODArray & out) const; template void setAttributeValueImpl(Attribute & attribute, const Key id, const Range & range, const T value); void setAttributeValue(Attribute & attribute, const Key id, const Range & range, const Field & value); const Attribute & getAttribute(const std::string & attribute_name) const; const Attribute & getAttributeWithType(const std::string & name, const AttributeUnderlyingType type) const; template void getIdsAndDates(PaddedPODArray & ids, PaddedPODArray & start_dates, PaddedPODArray & end_dates) const; template void getIdsAndDates( const Attribute & attribute, PaddedPODArray & ids, PaddedPODArray & start_dates, PaddedPODArray & end_dates) const; template BlockInputStreamPtr getBlockInputStreamImpl(const Names & column_names, size_t max_block_size) const; friend struct RangeHashedDIctionaryCallGetBlockInputStreamImpl; const std::string database; const std::string name; const std::string full_name; const DictionaryStructure dict_struct; const DictionarySourcePtr source_ptr; const DictionaryLifetime dict_lifetime; const bool require_nonempty; 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}; }; }