2015-07-13 16:18:28 +00:00
|
|
|
#pragma once
|
|
|
|
|
|
|
|
#include <DB/Dictionaries/IDictionary.h>
|
|
|
|
#include <DB/Dictionaries/IDictionarySource.h>
|
|
|
|
#include <DB/Dictionaries/DictionaryStructure.h>
|
|
|
|
#include <DB/Common/HashTable/HashMap.h>
|
|
|
|
#include <DB/Columns/ColumnString.h>
|
2015-10-05 00:33:43 +00:00
|
|
|
#include <ext/range.hpp>
|
2015-07-13 16:18:28 +00:00
|
|
|
#include <atomic>
|
|
|
|
#include <memory>
|
|
|
|
#include <tuple>
|
|
|
|
|
|
|
|
|
|
|
|
namespace DB
|
|
|
|
{
|
|
|
|
|
2015-07-13 17:10:16 +00:00
|
|
|
class RangeHashedDictionary final : public IDictionaryBase
|
2015-07-13 16:18:28 +00:00
|
|
|
{
|
|
|
|
public:
|
2015-07-17 15:56:23 +00:00
|
|
|
RangeHashedDictionary(
|
|
|
|
const std::string & name, const DictionaryStructure & dict_struct, DictionarySourcePtr source_ptr,
|
2016-06-07 21:07:44 +00:00
|
|
|
const DictionaryLifetime dict_lifetime, bool require_nonempty);
|
2015-07-13 16:18:28 +00:00
|
|
|
|
2016-06-07 21:07:44 +00:00
|
|
|
RangeHashedDictionary(const RangeHashedDictionary & other);
|
2015-07-13 16:18:28 +00:00
|
|
|
|
|
|
|
std::exception_ptr getCreationException() const override { return creation_exception; }
|
|
|
|
|
|
|
|
std::string getName() const override { return name; }
|
|
|
|
|
|
|
|
std::string getTypeName() const override { return "RangeHashed"; }
|
|
|
|
|
|
|
|
std::size_t getBytesAllocated() const override { return bytes_allocated; }
|
|
|
|
|
|
|
|
std::size_t getQueryCount() const override { return query_count.load(std::memory_order_relaxed); }
|
|
|
|
|
|
|
|
double getHitRate() const override { return 1.0; }
|
|
|
|
|
|
|
|
std::size_t getElementCount() const override { return element_count; }
|
|
|
|
|
|
|
|
double getLoadFactor() const override { return static_cast<double>(element_count) / bucket_count; }
|
|
|
|
|
|
|
|
bool isCached() const override { return false; }
|
|
|
|
|
|
|
|
DictionaryPtr clone() const override { return std::make_unique<RangeHashedDictionary>(*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<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;
|
|
|
|
}
|
|
|
|
|
|
|
|
#define DECLARE_MULTIPLE_GETTER(TYPE)\
|
2015-07-14 13:06:25 +00:00
|
|
|
void get##TYPE(\
|
2016-08-07 09:09:18 +00:00
|
|
|
const std::string & attribute_name, const PaddedPODArray<Key> & ids, const PaddedPODArray<UInt16> & dates,\
|
2016-06-07 21:07:44 +00:00
|
|
|
PaddedPODArray<TYPE> & out) const;
|
2015-07-13 16:18:28 +00:00
|
|
|
DECLARE_MULTIPLE_GETTER(UInt8)
|
|
|
|
DECLARE_MULTIPLE_GETTER(UInt16)
|
|
|
|
DECLARE_MULTIPLE_GETTER(UInt32)
|
|
|
|
DECLARE_MULTIPLE_GETTER(UInt64)
|
|
|
|
DECLARE_MULTIPLE_GETTER(Int8)
|
|
|
|
DECLARE_MULTIPLE_GETTER(Int16)
|
|
|
|
DECLARE_MULTIPLE_GETTER(Int32)
|
|
|
|
DECLARE_MULTIPLE_GETTER(Int64)
|
|
|
|
DECLARE_MULTIPLE_GETTER(Float32)
|
|
|
|
DECLARE_MULTIPLE_GETTER(Float64)
|
|
|
|
#undef DECLARE_MULTIPLE_GETTER
|
|
|
|
|
2015-07-14 13:06:25 +00:00
|
|
|
void getString(
|
2016-08-07 09:09:18 +00:00
|
|
|
const std::string & attribute_name, const PaddedPODArray<Key> & ids, const PaddedPODArray<UInt16> & dates,
|
2016-06-07 21:07:44 +00:00
|
|
|
ColumnString * out) const;
|
2015-07-13 16:18:28 +00:00
|
|
|
|
|
|
|
private:
|
2016-08-07 09:09:18 +00:00
|
|
|
struct Range : std::pair<UInt16, UInt16>
|
2015-07-13 16:18:28 +00:00
|
|
|
{
|
|
|
|
using std::pair<UInt16, UInt16>::pair;
|
|
|
|
|
|
|
|
bool contains(const UInt16 date) const
|
|
|
|
{
|
|
|
|
const auto & left = first;
|
|
|
|
const auto & right = second;
|
|
|
|
|
|
|
|
if (left <= date && date <= right)
|
|
|
|
return true;
|
|
|
|
|
|
|
|
const auto has_left_bound = 0 < left && left <= DATE_LUT_MAX_DAY_NUM;
|
|
|
|
const auto has_right_bound = 0 < right && right <= DATE_LUT_MAX_DAY_NUM;
|
|
|
|
|
|
|
|
if ((!has_left_bound || left <= date) && (!has_right_bound || date <= right))
|
|
|
|
return true;
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
};
|
|
|
|
|
|
|
|
template <typename T>
|
2016-08-07 09:09:18 +00:00
|
|
|
struct Value final
|
2015-07-13 16:18:28 +00:00
|
|
|
{
|
2016-08-07 09:09:18 +00:00
|
|
|
Range range;
|
2015-07-13 16:18:28 +00:00
|
|
|
T value;
|
|
|
|
};
|
|
|
|
|
2016-08-07 09:09:18 +00:00
|
|
|
template <typename T> using Values = std::vector<Value<T>>;
|
|
|
|
template <typename T> using Collection = HashMap<UInt64, Values<T>>;
|
|
|
|
template <typename T> using Ptr = std::unique_ptr<Collection<T>>;
|
2015-07-13 16:18:28 +00:00
|
|
|
|
2016-08-07 09:09:18 +00:00
|
|
|
struct Attribute final
|
2015-07-13 16:18:28 +00:00
|
|
|
{
|
|
|
|
public:
|
|
|
|
AttributeUnderlyingType type;
|
|
|
|
std::tuple<UInt8, UInt16, UInt32, UInt64,
|
|
|
|
Int8, Int16, Int32, Int64,
|
|
|
|
Float32, Float64,
|
|
|
|
String> null_values;
|
2016-08-07 09:09:18 +00:00
|
|
|
std::tuple<Ptr<UInt8>, Ptr<UInt16>, Ptr<UInt32>, Ptr<UInt64>,
|
|
|
|
Ptr<Int8>, Ptr<Int16>, Ptr<Int32>, Ptr<Int64>,
|
|
|
|
Ptr<Float32>, Ptr<Float64>, Ptr<StringRef>> maps;
|
2015-07-13 16:18:28 +00:00
|
|
|
std::unique_ptr<Arena> string_arena;
|
|
|
|
};
|
|
|
|
|
2016-06-07 21:07:44 +00:00
|
|
|
void createAttributes();
|
2015-07-13 16:18:28 +00:00
|
|
|
|
2016-06-07 21:07:44 +00:00
|
|
|
void loadData();
|
2015-07-13 16:18:28 +00:00
|
|
|
|
|
|
|
template <typename T>
|
2016-08-07 09:09:18 +00:00
|
|
|
void addAttributeSize(const Attribute & attribute);
|
2015-07-13 16:18:28 +00:00
|
|
|
|
2016-06-07 21:07:44 +00:00
|
|
|
void calculateBytesAllocated();
|
2015-07-13 16:18:28 +00:00
|
|
|
|
|
|
|
template <typename T>
|
2016-08-07 09:09:18 +00:00
|
|
|
void createAttributeImpl(Attribute & attribute, const Field & null_value);
|
2015-07-13 16:18:28 +00:00
|
|
|
|
2016-08-07 09:09:18 +00:00
|
|
|
Attribute createAttributeWithType(const AttributeUnderlyingType type, const Field & null_value);
|
2015-07-13 16:18:28 +00:00
|
|
|
|
2016-06-07 19:11:04 +00:00
|
|
|
|
|
|
|
template <typename OutputType>
|
2015-07-14 13:06:25 +00:00
|
|
|
void getItems(
|
2016-08-07 09:09:18 +00:00
|
|
|
const Attribute & attribute,
|
|
|
|
const PaddedPODArray<Key> & ids,
|
2016-06-07 19:11:04 +00:00
|
|
|
const PaddedPODArray<UInt16> & dates,
|
2016-06-07 21:07:44 +00:00
|
|
|
PaddedPODArray<OutputType> & out) const;
|
2016-06-07 19:11:04 +00:00
|
|
|
|
|
|
|
template <typename AttributeType, typename OutputType>
|
|
|
|
void getItemsImpl(
|
2016-08-07 09:09:18 +00:00
|
|
|
const Attribute & attribute,
|
|
|
|
const PaddedPODArray<Key> & ids,
|
2016-06-07 19:11:04 +00:00
|
|
|
const PaddedPODArray<UInt16> & dates,
|
2016-06-07 21:07:44 +00:00
|
|
|
PaddedPODArray<OutputType> & out) const;
|
2015-07-13 16:18:28 +00:00
|
|
|
|
2016-06-07 19:11:04 +00:00
|
|
|
|
2015-07-13 16:18:28 +00:00
|
|
|
template <typename T>
|
2016-08-07 09:09:18 +00:00
|
|
|
void setAttributeValueImpl(Attribute & attribute, const Key id, const Range & range, const T value);
|
2015-07-13 16:18:28 +00:00
|
|
|
|
2016-08-07 09:09:18 +00:00
|
|
|
void setAttributeValue(Attribute & attribute, const Key id, const Range & range, const Field & value);
|
2015-07-13 16:18:28 +00:00
|
|
|
|
2016-08-07 09:09:18 +00:00
|
|
|
const Attribute & getAttribute(const std::string & attribute_name) const;
|
2015-07-13 16:18:28 +00:00
|
|
|
|
2016-08-07 09:09:18 +00:00
|
|
|
const Attribute & getAttributeWithType(const std::string & name, const AttributeUnderlyingType type) const;
|
2015-07-14 13:06:25 +00:00
|
|
|
|
2015-07-13 16:18:28 +00:00
|
|
|
const std::string name;
|
|
|
|
const DictionaryStructure dict_struct;
|
|
|
|
const DictionarySourcePtr source_ptr;
|
|
|
|
const DictionaryLifetime dict_lifetime;
|
2015-08-11 21:32:27 +00:00
|
|
|
const bool require_nonempty;
|
2015-07-13 16:18:28 +00:00
|
|
|
|
|
|
|
std::map<std::string, std::size_t> attribute_index_by_name;
|
2016-08-07 09:09:18 +00:00
|
|
|
std::vector<Attribute> attributes;
|
2015-07-13 16:18:28 +00:00
|
|
|
|
|
|
|
std::size_t bytes_allocated = 0;
|
|
|
|
std::size_t element_count = 0;
|
|
|
|
std::size_t bucket_count = 0;
|
2015-08-12 04:21:10 +00:00
|
|
|
mutable std::atomic<std::size_t> query_count{0};
|
2015-07-13 16:18:28 +00:00
|
|
|
|
|
|
|
std::chrono::time_point<std::chrono::system_clock> creation_time;
|
|
|
|
|
|
|
|
std::exception_ptr creation_exception;
|
|
|
|
};
|
|
|
|
|
|
|
|
}
|