#pragma once #include #include #include #include #include #include #include #include #include #include #include #include namespace DB { /** Функции, использующие словари Яндекс.Метрики * - словари регионов, операционных систем, поисковых систем. * * Подняться по дереву до определенного уровня. * regionToCity, regionToArea, regionToCountry, * OSToRoot, * SEToRoot, * * Преобразовать значения в столбце * regionToName * * Является ли первый идентификатор потомком второго. * regionIn, SEIn, OSIn. * * Получить массив идентификаторов регионов, состоящий из исходного и цепочки родителей. Порядок implementation defined. * regionHierarchy, OSHierarchy, SEHierarchy. * * Функции, использующие подключаемые (внешние) словари. * * Получить значение аттрибута заданного типа. * dictGetType(dictionary, attribute, id), * Type - placeholder для имени типа, в данный момент поддерживаются любые числовые и строковой типы. * Тип должен соответствовать реальному типу аттрибута, с которым он был объявлен в структуре словаря. * * Получить массив идентификаторов, состоящий из исходного и цепочки родителей. * dictGetHierarchy(dictionary, id). * * Является ли первы йидентификатор потомком второго. * dictIsIn(dictionary, child_id, parent_id). */ struct RegionToCityImpl { static UInt32 apply(UInt32 x, const RegionsHierarchy & hierarchy) { return hierarchy.toCity(x); } }; struct RegionToAreaImpl { static UInt32 apply(UInt32 x, const RegionsHierarchy & hierarchy) { return hierarchy.toArea(x); } }; struct RegionToDistrictImpl { static UInt32 apply(UInt32 x, const RegionsHierarchy & hierarchy) { return hierarchy.toDistrict(x); } }; struct RegionToCountryImpl { static UInt32 apply(UInt32 x, const RegionsHierarchy & hierarchy) { return hierarchy.toCountry(x); } }; struct RegionToContinentImpl { static UInt32 apply(UInt32 x, const RegionsHierarchy & hierarchy) { return hierarchy.toContinent(x); } }; struct RegionToPopulationImpl { static UInt32 apply(UInt32 x, const RegionsHierarchy & hierarchy) { return hierarchy.getPopulation(x); } }; struct OSToRootImpl { static UInt8 apply(UInt8 x, const TechDataHierarchy & hierarchy) { return hierarchy.OSToMostAncestor(x); } }; struct SEToRootImpl { static UInt8 apply(UInt8 x, const TechDataHierarchy & hierarchy) { return hierarchy.SEToMostAncestor(x); } }; struct RegionInImpl { static bool apply(UInt32 x, UInt32 y, const RegionsHierarchy & hierarchy) { return hierarchy.in(x, y); } }; struct OSInImpl { static bool apply(UInt32 x, UInt32 y, const TechDataHierarchy & hierarchy) { return hierarchy.isOSIn(x, y); } }; struct SEInImpl { static bool apply(UInt32 x, UInt32 y, const TechDataHierarchy & hierarchy) { return hierarchy.isSEIn(x, y); } }; struct RegionHierarchyImpl { static UInt32 toParent(UInt32 x, const RegionsHierarchy & hierarchy) { return hierarchy.toParent(x); } }; struct OSHierarchyImpl { static UInt8 toParent(UInt8 x, const TechDataHierarchy & hierarchy) { return hierarchy.OSToParent(x); } }; struct SEHierarchyImpl { static UInt8 toParent(UInt8 x, const TechDataHierarchy & hierarchy) { return hierarchy.SEToParent(x); } }; /** Вспомогательная вещь, позволяющая достать из словаря конкретный словарь, соответствующий точке зрения * (ключу словаря, передаваемому в аргументе функции). * Пример: при вызове regionToCountry(x, 'ua'), может быть использован словарь, в котором Крым относится к Украине. */ struct RegionsHierarchyGetter { typedef RegionsHierarchies Src; typedef RegionsHierarchy Dst; static const Dst & get(const Src & src, const std::string & key) { return src.get(key); } }; /** Для словарей без поддержки ключей. Ничего не делает. */ template struct IdentityDictionaryGetter { typedef Dict Src; typedef Dict Dst; static const Dst & get(const Src & src, const std::string & key) { if (key.empty()) return src; else throw Exception("Dictionary doesn't support 'point of view' keys.", ErrorCodes::BAD_ARGUMENTS); } }; /// Преобразует идентификатор, используя словарь. template class FunctionTransformWithDictionary : public IFunction { public: static constexpr auto name = Name::name; using base_type = FunctionTransformWithDictionary; private: const SharedPtr owned_dict; public: FunctionTransformWithDictionary(const SharedPtr & owned_dict_) : owned_dict(owned_dict_) { if (!owned_dict) throw Exception("Dictionaries was not loaded. You need to check configuration file.", ErrorCodes::DICTIONARIES_WAS_NOT_LOADED); } /// Получить имя функции. String getName() const { return name; } /// Получить тип результата по типам аргументов. Если функция неприменима для данных аргументов - кинуть исключение. DataTypePtr getReturnType(const DataTypes & arguments) const { if (arguments.size() != 1 && arguments.size() != 2) throw Exception("Number of arguments for function " + getName() + " doesn't match: passed " + toString(arguments.size()) + ", should be 1 or 2.", ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH); if (arguments[0]->getName() != TypeName::get()) throw Exception("Illegal type " + arguments[0]->getName() + " of argument of function " + getName() + " (must be " + TypeName::get() + ")", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); if (arguments.size() == 2 && arguments[1]->getName() != TypeName::get()) throw Exception("Illegal type " + arguments[1]->getName() + " of the second ('point of view') argument of function " + getName() + " (must be " + TypeName::get() + ")", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); return arguments[0]; } /// Выполнить функцию над блоком. void execute(Block & block, const ColumnNumbers & arguments, size_t result) { /// Ключ словаря, определяющий "точку зрения". std::string dict_key; if (arguments.size() == 2) { const ColumnConstString * key_col = typeid_cast(&*block.getByPosition(arguments[1]).column); if (!key_col) throw Exception("Illegal column " + block.getByPosition(arguments[1]).column->getName() + " of second ('point of view') argument of function " + name + ". Must be constant string.", ErrorCodes::ILLEGAL_COLUMN); dict_key = key_col->getData(); } const typename DictGetter::Dst & dict = DictGetter::get(*owned_dict, dict_key); if (const ColumnVector * col_from = typeid_cast *>(&*block.getByPosition(arguments[0]).column)) { ColumnVector * col_to = new ColumnVector; block.getByPosition(result).column = col_to; const typename ColumnVector::Container_t & vec_from = col_from->getData(); typename ColumnVector::Container_t & vec_to = col_to->getData(); size_t size = vec_from.size(); vec_to.resize(size); for (size_t i = 0; i < size; ++i) vec_to[i] = Transform::apply(vec_from[i], dict); } else if (const ColumnConst * col_from = typeid_cast *>(&*block.getByPosition(arguments[0]).column)) { block.getByPosition(result).column = new ColumnConst(col_from->size(), Transform::apply(col_from->getData(), dict)); } else throw Exception("Illegal column " + block.getByPosition(arguments[0]).column->getName() + " of first argument of function " + name, ErrorCodes::ILLEGAL_COLUMN); } }; /// Проверяет принадлежность, используя словарь. template class FunctionIsInWithDictionary : public IFunction { public: static constexpr auto name = Name::name; using base_type = FunctionIsInWithDictionary; private: const SharedPtr owned_dict; public: FunctionIsInWithDictionary(const SharedPtr & owned_dict_) : owned_dict(owned_dict_) { if (!owned_dict) throw Exception("Dictionaries was not loaded. You need to check configuration file.", ErrorCodes::DICTIONARIES_WAS_NOT_LOADED); } /// Получить имя функции. String getName() const { return name; } /// Получить тип результата по типам аргументов. Если функция неприменима для данных аргументов - кинуть исключение. DataTypePtr getReturnType(const DataTypes & arguments) const { if (arguments.size() != 2 && arguments.size() != 3) throw Exception("Number of arguments for function " + getName() + " doesn't match: passed " + toString(arguments.size()) + ", should be 2 or 3.", ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH); if (arguments[0]->getName() != TypeName::get()) throw Exception("Illegal type " + arguments[0]->getName() + " of first argument of function " + getName() + " (must be " + TypeName::get() + ")", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); if (arguments[1]->getName() != TypeName::get()) throw Exception("Illegal type " + arguments[1]->getName() + " of second argument of function " + getName() + " (must be " + TypeName::get() + ")", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); if (arguments.size() == 3 && arguments[2]->getName() != TypeName::get()) throw Exception("Illegal type " + arguments[2]->getName() + " of the third ('point of view') argument of function " + getName() + " (must be " + TypeName::get() + ")", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); return new DataTypeUInt8; } /// Выполнить функцию над блоком. void execute(Block & block, const ColumnNumbers & arguments, size_t result) { /// Ключ словаря, определяющий "точку зрения". std::string dict_key; if (arguments.size() == 3) { const ColumnConstString * key_col = typeid_cast(&*block.getByPosition(arguments[2]).column); if (!key_col) throw Exception("Illegal column " + block.getByPosition(arguments[2]).column->getName() + " of third ('point of view') argument of function " + name + ". Must be constant string.", ErrorCodes::ILLEGAL_COLUMN); dict_key = key_col->getData(); } const typename DictGetter::Dst & dict = DictGetter::get(*owned_dict, dict_key); const ColumnVector * col_vec1 = typeid_cast *>(&*block.getByPosition(arguments[0]).column); const ColumnVector * col_vec2 = typeid_cast *>(&*block.getByPosition(arguments[1]).column); const ColumnConst * col_const1 = typeid_cast *>(&*block.getByPosition(arguments[0]).column); const ColumnConst * col_const2 = typeid_cast *>(&*block.getByPosition(arguments[1]).column); if (col_vec1 && col_vec2) { ColumnVector * col_to = new ColumnVector; block.getByPosition(result).column = col_to; const typename ColumnVector::Container_t & vec_from1 = col_vec1->getData(); const typename ColumnVector::Container_t & vec_from2 = col_vec2->getData(); typename ColumnVector::Container_t & vec_to = col_to->getData(); size_t size = vec_from1.size(); vec_to.resize(size); for (size_t i = 0; i < size; ++i) vec_to[i] = Transform::apply(vec_from1[i], vec_from2[i], dict); } else if (col_vec1 && col_const2) { ColumnVector * col_to = new ColumnVector; block.getByPosition(result).column = col_to; const typename ColumnVector::Container_t & vec_from1 = col_vec1->getData(); const T const_from2 = col_const2->getData(); typename ColumnVector::Container_t & vec_to = col_to->getData(); size_t size = vec_from1.size(); vec_to.resize(size); for (size_t i = 0; i < size; ++i) vec_to[i] = Transform::apply(vec_from1[i], const_from2, dict); } else if (col_const1 && col_vec2) { ColumnVector * col_to = new ColumnVector; block.getByPosition(result).column = col_to; const T const_from1 = col_const1->getData(); const typename ColumnVector::Container_t & vec_from2 = col_vec2->getData(); typename ColumnVector::Container_t & vec_to = col_to->getData(); size_t size = vec_from2.size(); vec_to.resize(size); for (size_t i = 0; i < size; ++i) vec_to[i] = Transform::apply(const_from1, vec_from2[i], dict); } else if (col_const1 && col_const2) { block.getByPosition(result).column = new ColumnConst(col_const1->size(), Transform::apply(col_const1->getData(), col_const2->getData(), dict)); } else throw Exception("Illegal columns " + block.getByPosition(arguments[0]).column->getName() + " and " + block.getByPosition(arguments[1]).column->getName() + " of arguments of function " + name, ErrorCodes::ILLEGAL_COLUMN); } }; /// Получает массив идентификаторов, состоящий из исходного и цепочки родителей. template class FunctionHierarchyWithDictionary : public IFunction { public: static constexpr auto name = Name::name; using base_type = FunctionHierarchyWithDictionary; private: const SharedPtr owned_dict; public: FunctionHierarchyWithDictionary(const SharedPtr & owned_dict_) : owned_dict(owned_dict_) { if (!owned_dict) throw Exception("Dictionaries was not loaded. You need to check configuration file.", ErrorCodes::DICTIONARIES_WAS_NOT_LOADED); } /// Получить имя функции. String getName() const { return name; } /// Получить тип результата по типам аргументов. Если функция неприменима для данных аргументов - кинуть исключение. DataTypePtr getReturnType(const DataTypes & arguments) const { if (arguments.size() != 1 && arguments.size() != 2) throw Exception("Number of arguments for function " + getName() + " doesn't match: passed " + toString(arguments.size()) + ", should be 1 or 2.", ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH); if (arguments[0]->getName() != TypeName::get()) throw Exception("Illegal type " + arguments[0]->getName() + " of argument of function " + getName() + " (must be " + TypeName::get() + ")", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); if (arguments.size() == 2 && arguments[1]->getName() != TypeName::get()) throw Exception("Illegal type " + arguments[1]->getName() + " of the second ('point of view') argument of function " + getName() + " (must be " + TypeName::get() + ")", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); return new DataTypeArray(arguments[0]); } /// Выполнить функцию над блоком. void execute(Block & block, const ColumnNumbers & arguments, size_t result) { /// Ключ словаря, определяющий "точку зрения". std::string dict_key; if (arguments.size() == 2) { const ColumnConstString * key_col = typeid_cast(&*block.getByPosition(arguments[1]).column); if (!key_col) throw Exception("Illegal column " + block.getByPosition(arguments[1]).column->getName() + " of second ('point of view') argument of function " + name + ". Must be constant string.", ErrorCodes::ILLEGAL_COLUMN); dict_key = key_col->getData(); } const typename DictGetter::Dst & dict = DictGetter::get(*owned_dict, dict_key); if (const ColumnVector * col_from = typeid_cast *>(&*block.getByPosition(arguments[0]).column)) { ColumnVector * col_values = new ColumnVector; ColumnArray * col_array = new ColumnArray(col_values); block.getByPosition(result).column = col_array; ColumnArray::Offsets_t & res_offsets = col_array->getOffsets(); typename ColumnVector::Container_t & res_values = col_values->getData(); const typename ColumnVector::Container_t & vec_from = col_from->getData(); size_t size = vec_from.size(); res_offsets.resize(size); res_values.reserve(size * 4); for (size_t i = 0; i < size; ++i) { T cur = vec_from[i]; while (cur) { res_values.push_back(cur); cur = Transform::toParent(cur, dict); } res_offsets[i] = res_values.size(); } } else if (const ColumnConst * col_from = typeid_cast *>(&*block.getByPosition(arguments[0]).column)) { Array res; T cur = col_from->getData(); while (cur) { res.push_back(static_cast::Type>(cur)); cur = Transform::toParent(cur, dict); } block.getByPosition(result).column = new ColumnConstArray(col_from->size(), res, new DataTypeArray(new typename DataTypeFromFieldType::Type)); } else throw Exception("Illegal column " + block.getByPosition(arguments[0]).column->getName() + " of first argument of function " + name, ErrorCodes::ILLEGAL_COLUMN); } }; struct NameRegionToCity { static constexpr auto name = "regionToCity"; }; struct NameRegionToArea { static constexpr auto name = "regionToArea"; }; struct NameRegionToDistrict { static constexpr auto name = "regionToDistrict"; }; struct NameRegionToCountry { static constexpr auto name = "regionToCountry"; }; struct NameRegionToContinent { static constexpr auto name = "regionToContient"; }; struct NameRegionToPopulation { static constexpr auto name = "regionToPopulation"; }; struct NameOSToRoot { static constexpr auto name = "OSToRoot"; }; struct NameSEToRoot { static constexpr auto name = "SEToRoot"; }; struct NameRegionIn { static constexpr auto name = "regionIn"; }; struct NameOSIn { static constexpr auto name = "OSIn"; }; struct NameSEIn { static constexpr auto name = "SEIn"; }; struct NameRegionHierarchy { static constexpr auto name = "regionHierarchy"; }; struct NameOSHierarchy { static constexpr auto name = "OSHierarchy"; }; struct NameSEHierarchy { static constexpr auto name = "SEHierarchy"; }; struct FunctionRegionToCity : public FunctionTransformWithDictionary { static IFunction * create(const Context & context) { return new base_type{context.getDictionaries().getRegionsHierarchies()}; } }; struct FunctionRegionToArea : public FunctionTransformWithDictionary { static IFunction * create(const Context & context) { return new base_type{context.getDictionaries().getRegionsHierarchies()}; } }; struct FunctionRegionToDistrict : public FunctionTransformWithDictionary { static IFunction * create(const Context & context) { return new base_type{context.getDictionaries().getRegionsHierarchies()}; } }; struct FunctionRegionToCountry : public FunctionTransformWithDictionary { static IFunction * create(const Context & context) { return new base_type{context.getDictionaries().getRegionsHierarchies()}; } }; struct FunctionRegionToContinent : public FunctionTransformWithDictionary { static IFunction * create(const Context & context) { return new base_type{context.getDictionaries().getRegionsHierarchies()}; } }; struct FunctionRegionToPopulation : public FunctionTransformWithDictionary { static IFunction * create(const Context & context) { return new base_type{context.getDictionaries().getRegionsHierarchies()}; } }; struct FunctionOSToRoot : public FunctionTransformWithDictionary, NameOSToRoot> { static IFunction * create(const Context & context) { return new base_type{context.getDictionaries().getTechDataHierarchy()}; } }; struct FunctionSEToRoot : public FunctionTransformWithDictionary, NameSEToRoot> { static IFunction * create(const Context & context) { return new base_type{context.getDictionaries().getTechDataHierarchy()}; } }; struct FunctionRegionIn : public FunctionIsInWithDictionary { static IFunction * create(const Context & context) { return new base_type{context.getDictionaries().getRegionsHierarchies()}; } }; struct FunctionOSIn : public FunctionIsInWithDictionary, NameOSIn> { static IFunction * create(const Context & context) { return new base_type{context.getDictionaries().getTechDataHierarchy()}; } }; struct FunctionSEIn : public FunctionIsInWithDictionary, NameSEIn> { static IFunction * create(const Context & context) { return new base_type{context.getDictionaries().getTechDataHierarchy()}; } }; struct FunctionRegionHierarchy : public FunctionHierarchyWithDictionary { static IFunction * create(const Context & context) { return new base_type{context.getDictionaries().getRegionsHierarchies()}; } }; struct FunctionOSHierarchy : public FunctionHierarchyWithDictionary, NameOSHierarchy> { static IFunction * create(const Context & context) { return new base_type{context.getDictionaries().getTechDataHierarchy()}; } }; struct FunctionSEHierarchy : public FunctionHierarchyWithDictionary, NameSEHierarchy> { static IFunction * create(const Context & context) { return new base_type{context.getDictionaries().getTechDataHierarchy()}; } }; /// Преобразует числовой идентификатор региона в имя на заданном языке, используя словарь. class FunctionRegionToName : public IFunction { public: static constexpr auto name = "regionToName"; static IFunction * create(const Context & context) { return new FunctionRegionToName(context.getDictionaries().getRegionsNames()); } private: const SharedPtr owned_dict; public: FunctionRegionToName(const SharedPtr & owned_dict_) : owned_dict(owned_dict_) { if (!owned_dict) throw Exception("Dictionaries was not loaded. You need to check configuration file.", ErrorCodes::DICTIONARIES_WAS_NOT_LOADED); } /// Получить имя функции. String getName() const { return name; } /// Получить тип результата по типам аргументов. Если функция неприменима для данных аргументов - кинуть исключение. DataTypePtr getReturnType(const DataTypes & arguments) const { if (arguments.size() != 1 && arguments.size() != 2) throw Exception("Number of arguments for function " + getName() + " doesn't match: passed " + toString(arguments.size()) + ", should be 1 or 2.", ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH); if (arguments[0]->getName() != TypeName::get()) throw Exception("Illegal type " + arguments[0]->getName() + " of the first argument of function " + getName() + " (must be " + TypeName::get() + ")", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); if (arguments.size() == 2 && arguments[1]->getName() != TypeName::get()) throw Exception("Illegal type " + arguments[0]->getName() + " of the second argument of function " + getName() + " (must be " + TypeName::get() + ")", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); return new DataTypeString; } /// Выполнить функцию над блоком. void execute(Block & block, const ColumnNumbers & arguments, size_t result) { RegionsNames::Language language = RegionsNames::Language::RU; /// Если указан язык результата if (arguments.size() == 2) { if (const ColumnConstString * col_language = typeid_cast(&*block.getByPosition(arguments[1]).column)) language = RegionsNames::getLanguageEnum(col_language->getData()); else throw Exception("Illegal column " + block.getByPosition(arguments[1]).column->getName() + " of the second argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); } const RegionsNames & dict = *owned_dict; if (const ColumnVector * col_from = typeid_cast *>(&*block.getByPosition(arguments[0]).column)) { ColumnString * col_to = new ColumnString; block.getByPosition(result).column = col_to; const ColumnVector::Container_t & region_ids = col_from->getData(); for (size_t i = 0; i < region_ids.size(); ++i) { const StringRef & name_ref = dict.getRegionName(region_ids[i], language); col_to->insertDataWithTerminatingZero(name_ref.data, name_ref.size + 1); } } else if (const ColumnConst * col_from = typeid_cast *>(&*block.getByPosition(arguments[0]).column)) { UInt32 region_id = col_from->getData(); const StringRef & name_ref = dict.getRegionName(region_id, language); block.getByPosition(result).column = new ColumnConstString(col_from->size(), name_ref.toString()); } else throw Exception("Illegal column " + block.getByPosition(arguments[0]).column->getName() + " of the first argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); } }; class FunctionDictGetString : public IFunction { public: static constexpr auto name = "dictGetString"; static IFunction * create(const Context & context) { return new FunctionDictGetString{context.getExternalDictionaries()}; }; FunctionDictGetString(const ExternalDictionaries & dictionaries) : dictionaries(dictionaries) {} String getName() const override { return name; } private: DataTypePtr getReturnType(const DataTypes & arguments) const override { if (arguments.size() != 3) throw Exception{ "Number of arguments for function " + getName() + " doesn't match: passed " + toString(arguments.size()) + ", should be 3.", ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH }; if (!typeid_cast(arguments[0].get())) { throw Exception{ "Illegal type " + arguments[0]->getName() + " of first argument of function " + getName() + ", expected a string.", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT }; } if (!typeid_cast(arguments[1].get())) { throw Exception{ "Illegal type " + arguments[1]->getName() + " of second argument of function " + getName() + ", expected a string.", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT }; } const auto id_arg = arguments[2].get(); if (!typeid_cast(id_arg) && !typeid_cast(id_arg) && !typeid_cast(id_arg) && !typeid_cast(id_arg) && !typeid_cast(id_arg) && !typeid_cast(id_arg) && !typeid_cast(id_arg) && !typeid_cast(id_arg)) { throw Exception{ "Illegal type " + arguments[2]->getName() + " of third argument of function " + getName() + ", expected an integer.", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT }; } return new DataTypeString; } void execute(Block & block, const ColumnNumbers & arguments, const size_t result) { const auto dict_name_col = typeid_cast *>(block.getByPosition(arguments[0]).column.get()); if (!dict_name_col) throw Exception{ "First argument of function " + getName() + " must be a constant string", ErrorCodes::ILLEGAL_COLUMN }; auto dict = dictionaries.getDictionary(dict_name_col->getData()); const auto dict_ptr = dict.get(); if (!executeDispatch(block, arguments, result, dict_ptr) && !executeDispatch(block, arguments, result, dict_ptr) && !executeDispatch(block, arguments, result, dict_ptr)) throw Exception{ "Unsupported dictionary type " + dict_ptr->getTypeName(), ErrorCodes::UNKNOWN_TYPE }; } template bool executeDispatch(Block & block, const ColumnNumbers & arguments, const size_t result, const IDictionary * const dictionary) { const auto dict = typeid_cast(dictionary); if (!dict) return false; const auto attr_name_col = typeid_cast *>(block.getByPosition(arguments[1]).column.get()); if (!attr_name_col) throw Exception{ "Second argument of function " + getName() + " must be a constant string", ErrorCodes::ILLEGAL_COLUMN }; const auto & attr_name = attr_name_col->getData(); const auto id_col = block.getByPosition(arguments[2]).column.get(); if (!execute(block, result, dict, attr_name, id_col) && !execute(block, result, dict, attr_name, id_col) && !execute(block, result, dict, attr_name, id_col) && !execute(block, result, dict, attr_name, id_col) && !execute(block, result, dict, attr_name, id_col) && !execute(block, result, dict, attr_name, id_col) && !execute(block, result, dict, attr_name, id_col) && !execute(block, result, dict, attr_name, id_col)) { throw Exception{ "Third argument of function " + getName() + " must be integral", ErrorCodes::ILLEGAL_COLUMN }; } return true; } template bool execute(Block & block, const size_t result, const DictionaryType * const dictionary, const std::string & attr_name, const IColumn * const id_col_untyped) { if (const auto id_col = typeid_cast *>(id_col_untyped)) { const auto out = new ColumnString; block.getByPosition(result).column = out; const auto attribute_idx = dictionary->getAttributeIndex(attr_name); if (!dictionary->isString(attribute_idx)) throw Exception{ "Type mismatch: attribute " + attr_name + " has type different from String", ErrorCodes::TYPE_MISMATCH }; for (const auto & id : id_col->getData()) { const auto string_ref = dictionary->getStringUnsafe(attribute_idx, id); out->insertData(string_ref.data, string_ref.size); } return true; } else if (const auto id_col = typeid_cast *>(id_col_untyped)) { block.getByPosition(result).column = new ColumnConst{ id_col->size(), dictionary->getString(attr_name, id_col->getData()).toString() }; return true; }; return false; } const ExternalDictionaries & dictionaries; }; template struct DictGetTraits; #define DECLARE_DICT_GET_TRAITS(TYPE, DATA_TYPE) \ template <> struct DictGetTraits\ {\ static TYPE get(const IDictionary * const dict, const std::string & name, const IDictionary::id_t id)\ {\ return dict->get##TYPE(name, id);\ }\ static bool is(const IDictionary * const dict, const std::size_t idx) { return dict->is##TYPE(idx); } \ static TYPE get(const IDictionary * const dict, const std::size_t idx, const IDictionary::id_t id)\ {\ return dict->get##TYPE##Unsafe(idx, id);\ }\ }; DECLARE_DICT_GET_TRAITS(UInt8, DataTypeUInt8) DECLARE_DICT_GET_TRAITS(UInt16, DataTypeUInt16) DECLARE_DICT_GET_TRAITS(UInt32, DataTypeUInt32) DECLARE_DICT_GET_TRAITS(UInt64, DataTypeUInt64) DECLARE_DICT_GET_TRAITS(Int8, DataTypeInt8) DECLARE_DICT_GET_TRAITS(Int16, DataTypeInt16) DECLARE_DICT_GET_TRAITS(Int32, DataTypeInt32) DECLARE_DICT_GET_TRAITS(Int64, DataTypeInt64) DECLARE_DICT_GET_TRAITS(Float32, DataTypeFloat32) DECLARE_DICT_GET_TRAITS(Float64, DataTypeFloat64) #undef DECLARE_DICT_GET_TRAITS template class FunctionDictGet final : public IFunction { using Type = typename DataType::FieldType; public: static const std::string name; static IFunction * create(const Context & context) { return new FunctionDictGet{context.getExternalDictionaries()}; }; FunctionDictGet(const ExternalDictionaries & dictionaries) : dictionaries(dictionaries) {} String getName() const override { return name; } private: DataTypePtr getReturnType(const DataTypes & arguments) const override { if (arguments.size() != 3) throw Exception{ "Number of arguments for function " + getName() + " doesn't match: passed " + toString(arguments.size()) + ", should be 3.", ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH }; if (!typeid_cast(arguments[0].get())) { throw Exception{ "Illegal type " + arguments[0]->getName() + " of first argument of function " + getName() + ", expected a string.", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT }; } if (!typeid_cast(arguments[1].get())) { throw Exception{ "Illegal type " + arguments[1]->getName() + " of second argument of function " + getName() + ", expected a string.", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT }; } const auto id_arg = arguments[2].get(); if (!typeid_cast(id_arg) && !typeid_cast(id_arg) && !typeid_cast(id_arg) && !typeid_cast(id_arg) && !typeid_cast(id_arg) && !typeid_cast(id_arg) && !typeid_cast(id_arg) && !typeid_cast(id_arg)) { throw Exception{ "Illegal type " + id_arg->getName() + " of third argument of function " + getName() + ", expected an integer.", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT }; } return new DataType; } void execute(Block & block, const ColumnNumbers & arguments, const size_t result) override { const auto dict_name_col = typeid_cast *>(block.getByPosition(arguments[0]).column.get()); if (!dict_name_col) throw Exception{ "First argument of function " + getName() + " must be a constant string", ErrorCodes::ILLEGAL_COLUMN }; auto dict = dictionaries.getDictionary(dict_name_col->getData()); const auto dict_ptr = dict.get(); if (!executeDispatch(block, arguments, result, dict_ptr) && !executeDispatch(block, arguments, result, dict_ptr) && !executeDispatch(block, arguments, result, dict_ptr)) throw Exception{ "Unsupported dictionary type " + dict_ptr->getTypeName(), ErrorCodes::UNKNOWN_TYPE }; } template bool executeDispatch(Block & block, const ColumnNumbers & arguments, const size_t result, const IDictionary * const dictionary) { const auto dict = typeid_cast(dictionary); if (!dict) return false; const auto attr_name_col = typeid_cast *>(block.getByPosition(arguments[1]).column.get()); if (!attr_name_col) throw Exception{ "Second argument of function " + getName() + " must be a constant string", ErrorCodes::ILLEGAL_COLUMN }; const auto & attr_name = attr_name_col->getData(); const auto id_col = block.getByPosition(arguments[2]).column.get(); if (!execute(block, result, dict, attr_name, id_col) && !execute(block, result, dict, attr_name, id_col) && !execute(block, result, dict, attr_name, id_col) && !execute(block, result, dict, attr_name, id_col) && !execute(block, result, dict, attr_name, id_col) && !execute(block, result, dict, attr_name, id_col) && !execute(block, result, dict, attr_name, id_col) && !execute(block, result, dict, attr_name, id_col)) { throw Exception{ "Third argument of function " + getName() + " must be integral", ErrorCodes::ILLEGAL_COLUMN }; } return true; } template bool execute(Block & block, const size_t result, const DictionaryType * const dictionary, const std::string & attr_name, const IColumn * const id_col_untyped) { if (const auto id_col = typeid_cast *>(id_col_untyped)) { const auto attribute_idx = dictionary->getAttributeIndex(attr_name); if (!DictGetTraits::is(dictionary, attribute_idx)) throw Exception{ "Type mismatch: attribute " + attr_name + " has type different from UInt64", ErrorCodes::TYPE_MISMATCH }; const auto out = new ColumnVector; block.getByPosition(result).column = out; const auto & ids = id_col->getData(); auto & data = out->getData(); const auto size = ids.size(); data.resize(size); for (const auto idx : ext::range(0, size)) data[idx] = DictGetTraits::get(dictionary, attribute_idx, ids[idx]); return true; } else if (const auto id_col = typeid_cast *>(id_col_untyped)) { block.getByPosition(result).column = new ColumnConst{ id_col->size(), DictGetTraits::get(dictionary, attr_name, id_col->getData()) }; return true; }; return false; } const ExternalDictionaries & dictionaries; }; template const std::string FunctionDictGet::name = "dictGet" + TypeName::get(); using FunctionDictGetUInt8 = FunctionDictGet; using FunctionDictGetUInt16 = FunctionDictGet; using FunctionDictGetUInt32 = FunctionDictGet; using FunctionDictGetUInt64 = FunctionDictGet; using FunctionDictGetInt8 = FunctionDictGet; using FunctionDictGetInt16 = FunctionDictGet; using FunctionDictGetInt32 = FunctionDictGet; using FunctionDictGetInt64 = FunctionDictGet; using FunctionDictGetFloat32 = FunctionDictGet; using FunctionDictGetFloat64 = FunctionDictGet; class FunctionDictGetHierarchy final : public IFunction { public: static constexpr auto name = "dictGetHierarchy"; static IFunction * create(const Context & context) { return new FunctionDictGetHierarchy{context.getExternalDictionaries()}; }; FunctionDictGetHierarchy(const ExternalDictionaries & dictionaries) : dictionaries(dictionaries) {} String getName() const override { return name; } private: DataTypePtr getReturnType(const DataTypes & arguments) const override { if (arguments.size() != 2) throw Exception{ "Number of arguments for function " + getName() + " doesn't match: passed " + toString(arguments.size()) + ", should be 2.", ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH }; if (!typeid_cast(arguments[0].get())) { throw Exception{ "Illegal type " + arguments[0]->getName() + " of first argument of function " + getName() + ", expected a string.", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT }; } const auto id_arg = arguments[1].get(); if (!typeid_cast(id_arg) && !typeid_cast(id_arg) && !typeid_cast(id_arg) && !typeid_cast(id_arg) && !typeid_cast(id_arg) && !typeid_cast(id_arg) && !typeid_cast(id_arg) && !typeid_cast(id_arg)) { throw Exception{ "Illegal type " + id_arg->getName() + " of second argument of function " + getName() + ", expected an integer.", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT }; } return new DataTypeArray{new DataTypeUInt64}; }; void execute(Block & block, const ColumnNumbers & arguments, const size_t result) override { const auto dict_name_col = typeid_cast *>(block.getByPosition(arguments[0]).column.get()); if (!dict_name_col) throw Exception{ "First argument of function " + getName() + " must be a constant string", ErrorCodes::ILLEGAL_COLUMN }; auto dict = dictionaries.getDictionary(dict_name_col->getData()); const auto dict_ptr = dict.get(); if (!dict->hasHierarchy()) throw Exception{ "Dictionary does not have a hierarchy", ErrorCodes::UNSUPPORTED_METHOD }; if (!executeDispatch(block, arguments, result, dict_ptr) && !executeDispatch(block, arguments, result, dict_ptr) && !executeDispatch(block, arguments, result, dict_ptr)) throw Exception{ "Unsupported dictionary type " + dict_ptr->getTypeName(), ErrorCodes::UNKNOWN_TYPE }; } template bool executeDispatch(Block & block, const ColumnNumbers & arguments, const size_t result, const IDictionary * const dictionary) { const auto dict = typeid_cast(dictionary); if (!dict) return false; const auto id_col = block.getByPosition(arguments[1]).column.get(); if (!execute(block, result, dict, id_col) && !execute(block, result, dict, id_col) && !execute(block, result, dict, id_col) && !execute(block, result, dict, id_col) && !execute(block, result, dict, id_col) && !execute(block, result, dict, id_col) && !execute(block, result, dict, id_col) && !execute(block, result, dict, id_col)) { throw Exception{ "Second argument of function " + getName() + " must be integral", ErrorCodes::ILLEGAL_COLUMN }; } return true; } template bool execute(Block & block, const size_t result, const DictionaryType * const dictionary, const IColumn * const id_col_untyped) { if (const auto id_col = typeid_cast *>(id_col_untyped)) { const auto backend = new ColumnVector; const auto array = new ColumnArray{backend}; block.getByPosition(result).column = array; const auto & in = id_col->getData(); const auto size = in.size(); auto & out = backend->getData(); auto & offsets = array->getOffsets(); offsets.resize(size); out.reserve(size * 4); for (const auto idx : ext::range(0, size)) { IDictionary::id_t cur = in[idx]; while (cur) { out.push_back(cur); cur = dictionary->toParent(cur); } offsets[idx] = out.size(); }; return true; } else if (const auto id_col = typeid_cast *>(id_col_untyped)) { Array res; IDictionary::id_t cur = id_col->getData(); while (cur) { res.push_back(static_cast::Type>(cur)); cur = dictionary->toParent(cur); } block.getByPosition(result).column = new ColumnConstArray{ id_col->size(), res, new DataTypeArray{new DataTypeUInt64} }; return true; }; return false; } const ExternalDictionaries & dictionaries; }; class FunctionDictIsIn final : public IFunction { public: static constexpr auto name = "dictIsIn"; static IFunction * create(const Context & context) { return new FunctionDictIsIn{context.getExternalDictionaries()}; }; FunctionDictIsIn(const ExternalDictionaries & dictionaries) : dictionaries(dictionaries) {} String getName() const override { return name; } private: DataTypePtr getReturnType(const DataTypes & arguments) const override { if (arguments.size() != 3) throw Exception{ "Number of arguments for function " + getName() + " doesn't match: passed " + toString(arguments.size()) + ", should be 3.", ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH }; if (!typeid_cast(arguments[0].get())) { throw Exception{ "Illegal type " + arguments[0]->getName() + " of first argument of function " + getName() + ", expected a string.", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT }; } const auto child_id_arg = arguments[1].get(); if (!typeid_cast(child_id_arg) && !typeid_cast(child_id_arg) && !typeid_cast(child_id_arg) && !typeid_cast(child_id_arg) && !typeid_cast(child_id_arg) && !typeid_cast(child_id_arg) && !typeid_cast(child_id_arg) && !typeid_cast(child_id_arg)) { throw Exception{ "Illegal type " + child_id_arg->getName() + " of second argument of function " + getName() + ", expected an integer.", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT }; } const auto ancestor_id_arg = arguments[2].get(); if (!typeid_cast(ancestor_id_arg) && !typeid_cast(ancestor_id_arg) && !typeid_cast(ancestor_id_arg) && !typeid_cast(ancestor_id_arg) && !typeid_cast(ancestor_id_arg) && !typeid_cast(ancestor_id_arg) && !typeid_cast(ancestor_id_arg) && !typeid_cast(ancestor_id_arg)) { throw Exception{ "Illegal type " + ancestor_id_arg->getName() + " of argument of third function " + getName() + ", expected an integer.", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT }; } return new DataTypeUInt8; } void execute(Block & block, const ColumnNumbers & arguments, const size_t result) override { const auto dict_name_col = typeid_cast *>(block.getByPosition(arguments[0]).column.get()); if (!dict_name_col) throw Exception{ "First argument of function " + getName() + " must be a constant string", ErrorCodes::ILLEGAL_COLUMN }; auto dict = dictionaries.getDictionary(dict_name_col->getData()); const auto dict_ptr = dict.get(); if (!dict->hasHierarchy()) throw Exception{ "Dictionary does not have a hierarchy", ErrorCodes::UNSUPPORTED_METHOD }; if (!executeDispatch(block, arguments, result, dict_ptr) && !executeDispatch(block, arguments, result, dict_ptr) && !executeDispatch(block, arguments, result, dict_ptr)) throw Exception{ "Unsupported dictionary type " + dict_ptr->getTypeName(), ErrorCodes::UNKNOWN_TYPE }; } template bool executeDispatch(Block & block, const ColumnNumbers & arguments, const size_t result, const IDictionary * const dictionary) { const auto dict = typeid_cast(dictionary); if (!dict) return false; const auto child_id_col = block.getByPosition(arguments[1]).column.get(); const auto ancestor_id_col = block.getByPosition(arguments[2]).column.get(); if (!execute(block, result, dict, child_id_col, ancestor_id_col) && !execute(block, result, dict, child_id_col, ancestor_id_col) && !execute(block, result, dict, child_id_col, ancestor_id_col) && !execute(block, result, dict, child_id_col, ancestor_id_col) && !execute(block, result, dict, child_id_col, ancestor_id_col) && !execute(block, result, dict, child_id_col, ancestor_id_col) && !execute(block, result, dict, child_id_col, ancestor_id_col) && !execute(block, result, dict, child_id_col, ancestor_id_col)) { throw Exception{ "Illegal column " + child_id_col->getName() + " of second argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN }; } return true; } template bool execute(Block & block, const size_t result, const DictionaryType * const dictionary, const IColumn * const child_id_col_untyped, const IColumn * const ancestor_id_col_untyped) { if (execute>(block, result, dictionary, child_id_col_untyped, ancestor_id_col_untyped) || execute>(block, result, dictionary, child_id_col_untyped, ancestor_id_col_untyped)) return true; return false; } template bool execute(Block & block, const size_t result, const DictionaryType * dictionary, const IColumn * const child_id_col_untyped, const IColumn * const ancestor_id_col_untyped) { if (const auto child_id_col = typeid_cast(child_id_col_untyped)) { if (execute(block, result, dictionary, child_id_col, ancestor_id_col_untyped) || execute(block, result, dictionary, child_id_col, ancestor_id_col_untyped) || execute(block, result, dictionary, child_id_col, ancestor_id_col_untyped) || execute(block, result, dictionary, child_id_col, ancestor_id_col_untyped) || execute(block, result, dictionary, child_id_col, ancestor_id_col_untyped) || execute(block, result, dictionary, child_id_col, ancestor_id_col_untyped) || execute(block, result, dictionary, child_id_col, ancestor_id_col_untyped) || execute(block, result, dictionary, child_id_col, ancestor_id_col_untyped)) return true; else throw Exception{ "Illegal column " + ancestor_id_col_untyped->getName() + " of third argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN }; } return false; } template bool execute(Block & block, const size_t result, const DictionaryType * const dictionary, const ColumnVector * const child_id_col, const IColumn * const ancestor_id_col_untyped) { if (const auto ancestor_id_col = typeid_cast *>(ancestor_id_col_untyped)) { const auto out = new ColumnVector; block.getByPosition(result).column = out; const auto & child_ids = child_id_col->getData(); const auto & ancestor_ids = ancestor_id_col->getData(); auto & data = out->getData(); const auto size = child_id_col->size(); data.resize(size); for (const auto idx : ext::range(0, size)) data[idx] = dictionary->in(child_ids[idx], ancestor_ids[idx]); return true; } else if (const auto ancestor_id_col = typeid_cast *>(ancestor_id_col_untyped)) { const auto out = new ColumnVector; block.getByPosition(result).column = out; const auto & child_ids = child_id_col->getData(); const auto ancestor_id = ancestor_id_col->getData(); auto & data = out->getData(); const auto size = child_id_col->size(); data.resize(size); for (const auto idx : ext::range(0, size)) data[idx] = dictionary->in(child_ids[idx], ancestor_id); return true; } return false; } template bool execute(Block & block, const size_t result, const DictionaryType * const dictionary, const ColumnConst * const child_id_col, const IColumn * const ancestor_id_col_untyped) { if (const auto ancestor_id_col = typeid_cast *>(ancestor_id_col_untyped)) { const auto out = new ColumnVector; block.getByPosition(result).column = out; const auto child_id = child_id_col->getData(); const auto & ancestor_ids = ancestor_id_col->getData(); auto & data = out->getData(); const auto size = child_id_col->size(); data.resize(size); for (const auto idx : ext::range(0, size)) data[idx] = dictionary->in(child_id, ancestor_ids[idx]); return true; } else if (const auto ancestor_id_col = typeid_cast *>(ancestor_id_col_untyped)) { block.getByPosition(result).column = new ColumnConst{ child_id_col->size(), dictionary->in(child_id_col->getData(), ancestor_id_col->getData()) }; return true; } return false; } const ExternalDictionaries & dictionaries; }; }