2017-04-01 09:19:00 +00:00
|
|
|
#include <Dictionaries/FlatDictionary.h>
|
2017-04-28 18:33:31 +00:00
|
|
|
#include <Dictionaries/DictionaryBlockInputStream.h>
|
2016-06-07 21:07:44 +00:00
|
|
|
|
|
|
|
namespace DB
|
|
|
|
{
|
|
|
|
|
|
|
|
namespace ErrorCodes
|
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
extern const int TYPE_MISMATCH;
|
|
|
|
extern const int ARGUMENT_OUT_OF_BOUND;
|
|
|
|
extern const int BAD_ARGUMENTS;
|
|
|
|
extern const int DICTIONARY_IS_EMPTY;
|
|
|
|
extern const int LOGICAL_ERROR;
|
|
|
|
extern const int UNKNOWN_TYPE;
|
2016-06-07 21:07:44 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static const auto initial_array_size = 1024;
|
|
|
|
static const auto max_array_size = 500000;
|
|
|
|
|
|
|
|
|
|
|
|
FlatDictionary::FlatDictionary(const std::string & name, const DictionaryStructure & dict_struct,
|
2018-01-15 12:44:39 +00:00
|
|
|
DictionarySourcePtr source_ptr, const DictionaryLifetime dict_lifetime, bool require_nonempty, BlockPtr saved_block)
|
2017-04-01 07:20:54 +00:00
|
|
|
: name{name}, dict_struct(dict_struct),
|
|
|
|
source_ptr{std::move(source_ptr)}, dict_lifetime(dict_lifetime),
|
|
|
|
require_nonempty(require_nonempty),
|
2018-01-15 12:44:39 +00:00
|
|
|
loaded_ids(initial_array_size, false), saved_block{std::move(saved_block)}
|
2016-06-07 21:07:44 +00:00
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
createAttributes();
|
|
|
|
|
|
|
|
try
|
|
|
|
{
|
|
|
|
loadData();
|
|
|
|
calculateBytesAllocated();
|
|
|
|
}
|
|
|
|
catch (...)
|
|
|
|
{
|
|
|
|
creation_exception = std::current_exception();
|
|
|
|
}
|
|
|
|
|
|
|
|
creation_time = std::chrono::system_clock::now();
|
2016-06-07 21:07:44 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
FlatDictionary::FlatDictionary(const FlatDictionary & other)
|
2018-01-15 12:44:39 +00:00
|
|
|
: FlatDictionary{other.name, other.dict_struct, other.source_ptr->clone(), other.dict_lifetime, other.require_nonempty, other.saved_block}
|
2016-06-07 21:07:44 +00:00
|
|
|
{
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2016-08-07 09:09:18 +00:00
|
|
|
void FlatDictionary::toParent(const PaddedPODArray<Key> & ids, PaddedPODArray<Key> & out) const
|
2016-06-07 21:07:44 +00:00
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
const auto null_value = std::get<UInt64>(hierarchical_attribute->null_values);
|
2016-06-07 21:07:44 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
getItemsNumber<UInt64>(*hierarchical_attribute, ids,
|
2017-07-21 06:35:58 +00:00
|
|
|
[&] (const size_t row, const UInt64 value) { out[row] = value; },
|
|
|
|
[&] (const size_t) { return null_value; });
|
2016-06-07 21:07:44 +00:00
|
|
|
}
|
|
|
|
|
2016-12-12 21:37:57 +00:00
|
|
|
|
|
|
|
/// Allow to use single value in same way as array.
|
|
|
|
static inline FlatDictionary::Key getAt(const PaddedPODArray<FlatDictionary::Key> & arr, const size_t idx) { return arr[idx]; }
|
2017-12-01 20:21:35 +00:00
|
|
|
static inline FlatDictionary::Key getAt(const FlatDictionary::Key & value, const size_t) { return value; }
|
2016-12-12 21:37:57 +00:00
|
|
|
|
|
|
|
template <typename ChildType, typename AncestorType>
|
|
|
|
void FlatDictionary::isInImpl(
|
2017-04-01 07:20:54 +00:00
|
|
|
const ChildType & child_ids,
|
|
|
|
const AncestorType & ancestor_ids,
|
|
|
|
PaddedPODArray<UInt8> & out) const
|
2016-12-12 21:37:57 +00:00
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
const auto null_value = std::get<UInt64>(hierarchical_attribute->null_values);
|
|
|
|
const auto & attr = *std::get<ContainerPtrType<Key>>(hierarchical_attribute->arrays);
|
|
|
|
const auto rows = out.size();
|
2016-12-12 21:37:57 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
size_t loaded_size = attr.size();
|
|
|
|
for (const auto row : ext::range(0, rows))
|
|
|
|
{
|
|
|
|
auto id = getAt(child_ids, row);
|
|
|
|
const auto ancestor_id = getAt(ancestor_ids, row);
|
2016-12-12 21:37:57 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
while (id < loaded_size && id != null_value && id != ancestor_id)
|
|
|
|
id = attr[id];
|
2016-12-12 21:37:57 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
out[row] = id != null_value && id == ancestor_id;
|
|
|
|
}
|
2016-12-12 21:37:57 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
query_count.fetch_add(rows, std::memory_order_relaxed);
|
2016-12-12 21:37:57 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
2016-12-13 21:15:27 +00:00
|
|
|
void FlatDictionary::isInVectorVector(
|
2017-04-01 07:20:54 +00:00
|
|
|
const PaddedPODArray<Key> & child_ids,
|
|
|
|
const PaddedPODArray<Key> & ancestor_ids,
|
|
|
|
PaddedPODArray<UInt8> & out) const
|
2016-12-12 21:37:57 +00:00
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
isInImpl(child_ids, ancestor_ids, out);
|
2016-12-12 21:37:57 +00:00
|
|
|
}
|
|
|
|
|
2016-12-13 21:15:27 +00:00
|
|
|
void FlatDictionary::isInVectorConstant(
|
2017-04-01 07:20:54 +00:00
|
|
|
const PaddedPODArray<Key> & child_ids,
|
|
|
|
const Key ancestor_id,
|
|
|
|
PaddedPODArray<UInt8> & out) const
|
2016-12-12 21:37:57 +00:00
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
isInImpl(child_ids, ancestor_id, out);
|
2016-12-12 21:37:57 +00:00
|
|
|
}
|
|
|
|
|
2016-12-13 21:15:27 +00:00
|
|
|
void FlatDictionary::isInConstantVector(
|
2017-04-01 07:20:54 +00:00
|
|
|
const Key child_id,
|
|
|
|
const PaddedPODArray<Key> & ancestor_ids,
|
|
|
|
PaddedPODArray<UInt8> & out) const
|
2016-12-12 21:37:57 +00:00
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
isInImpl(child_id, ancestor_ids, out);
|
2016-12-12 21:37:57 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
2016-06-07 21:07:44 +00:00
|
|
|
#define DECLARE(TYPE)\
|
2016-08-07 09:09:18 +00:00
|
|
|
void FlatDictionary::get##TYPE(const std::string & attribute_name, const PaddedPODArray<Key> & ids, PaddedPODArray<TYPE> & out) const\
|
2016-06-07 21:07:44 +00:00
|
|
|
{\
|
2017-04-01 07:20:54 +00:00
|
|
|
const auto & attribute = getAttribute(attribute_name);\
|
|
|
|
if (!isAttributeTypeConvertibleTo(attribute.type, AttributeUnderlyingType::TYPE))\
|
|
|
|
throw Exception{\
|
|
|
|
name + ": type mismatch: attribute " + attribute_name + " has type " + toString(attribute.type),\
|
|
|
|
ErrorCodes::TYPE_MISMATCH};\
|
|
|
|
\
|
|
|
|
const auto null_value = std::get<TYPE>(attribute.null_values);\
|
|
|
|
\
|
|
|
|
getItemsNumber<TYPE>(attribute, ids,\
|
2017-07-21 06:35:58 +00:00
|
|
|
[&] (const size_t row, const auto value) { out[row] = value; },\
|
|
|
|
[&] (const size_t) { return null_value; });\
|
2016-06-07 21:07:44 +00:00
|
|
|
}
|
|
|
|
DECLARE(UInt8)
|
|
|
|
DECLARE(UInt16)
|
|
|
|
DECLARE(UInt32)
|
|
|
|
DECLARE(UInt64)
|
2017-11-14 00:08:54 +00:00
|
|
|
DECLARE(UInt128)
|
2016-06-07 21:07:44 +00:00
|
|
|
DECLARE(Int8)
|
|
|
|
DECLARE(Int16)
|
|
|
|
DECLARE(Int32)
|
|
|
|
DECLARE(Int64)
|
|
|
|
DECLARE(Float32)
|
|
|
|
DECLARE(Float64)
|
|
|
|
#undef DECLARE
|
|
|
|
|
2016-08-07 09:09:18 +00:00
|
|
|
void FlatDictionary::getString(const std::string & attribute_name, const PaddedPODArray<Key> & ids, ColumnString * out) const
|
2016-06-07 21:07:44 +00:00
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
const auto & attribute = getAttribute(attribute_name);
|
|
|
|
if (!isAttributeTypeConvertibleTo(attribute.type, AttributeUnderlyingType::String))
|
|
|
|
throw Exception{
|
|
|
|
name + ": type mismatch: attribute " + attribute_name + " has type " + toString(attribute.type),
|
|
|
|
ErrorCodes::TYPE_MISMATCH};
|
2016-06-07 21:07:44 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
const auto & null_value = std::get<StringRef>(attribute.null_values);
|
2016-06-07 21:07:44 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
getItemsImpl<StringRef, StringRef>(attribute, ids,
|
2017-12-01 20:21:35 +00:00
|
|
|
[&] (const size_t, const StringRef value) { out->insertData(value.data, value.size); },
|
2017-07-21 06:35:58 +00:00
|
|
|
[&] (const size_t) { return null_value; });
|
2016-06-07 21:07:44 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
#define DECLARE(TYPE)\
|
|
|
|
void FlatDictionary::get##TYPE(\
|
2017-04-01 07:20:54 +00:00
|
|
|
const std::string & attribute_name, const PaddedPODArray<Key> & ids, const PaddedPODArray<TYPE> & def,\
|
|
|
|
PaddedPODArray<TYPE> & out) const\
|
2016-06-07 21:07:44 +00:00
|
|
|
{\
|
2017-04-01 07:20:54 +00:00
|
|
|
const auto & attribute = getAttribute(attribute_name);\
|
|
|
|
if (!isAttributeTypeConvertibleTo(attribute.type, AttributeUnderlyingType::TYPE))\
|
|
|
|
throw Exception{\
|
|
|
|
name + ": type mismatch: attribute " + attribute_name + " has type " + toString(attribute.type),\
|
|
|
|
ErrorCodes::TYPE_MISMATCH};\
|
|
|
|
\
|
|
|
|
getItemsNumber<TYPE>(attribute, ids,\
|
2017-07-21 06:35:58 +00:00
|
|
|
[&] (const size_t row, const auto value) { out[row] = value; },\
|
|
|
|
[&] (const size_t row) { return def[row]; });\
|
2016-06-07 21:07:44 +00:00
|
|
|
}
|
|
|
|
DECLARE(UInt8)
|
|
|
|
DECLARE(UInt16)
|
|
|
|
DECLARE(UInt32)
|
|
|
|
DECLARE(UInt64)
|
2017-11-14 00:08:54 +00:00
|
|
|
DECLARE(UInt128)
|
2016-06-07 21:07:44 +00:00
|
|
|
DECLARE(Int8)
|
|
|
|
DECLARE(Int16)
|
|
|
|
DECLARE(Int32)
|
|
|
|
DECLARE(Int64)
|
|
|
|
DECLARE(Float32)
|
|
|
|
DECLARE(Float64)
|
|
|
|
#undef DECLARE
|
|
|
|
|
|
|
|
void FlatDictionary::getString(
|
2017-04-01 07:20:54 +00:00
|
|
|
const std::string & attribute_name, const PaddedPODArray<Key> & ids, const ColumnString * const def,
|
|
|
|
ColumnString * const out) const
|
2016-06-07 21:07:44 +00:00
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
const auto & attribute = getAttribute(attribute_name);
|
|
|
|
if (!isAttributeTypeConvertibleTo(attribute.type, AttributeUnderlyingType::String))
|
|
|
|
throw Exception{
|
|
|
|
name + ": type mismatch: attribute " + attribute_name + " has type " + toString(attribute.type),
|
|
|
|
ErrorCodes::TYPE_MISMATCH};
|
|
|
|
|
|
|
|
getItemsImpl<StringRef, StringRef>(attribute, ids,
|
2017-12-01 20:21:35 +00:00
|
|
|
[&] (const size_t, const StringRef value) { out->insertData(value.data, value.size); },
|
2017-07-21 06:35:58 +00:00
|
|
|
[&] (const size_t row) { return def->getDataAt(row); });
|
2016-06-07 21:07:44 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
#define DECLARE(TYPE)\
|
|
|
|
void FlatDictionary::get##TYPE(\
|
2017-04-01 07:20:54 +00:00
|
|
|
const std::string & attribute_name, const PaddedPODArray<Key> & ids, const TYPE def,\
|
|
|
|
PaddedPODArray<TYPE> & out) const\
|
2016-06-07 21:07:44 +00:00
|
|
|
{\
|
2017-04-01 07:20:54 +00:00
|
|
|
const auto & attribute = getAttribute(attribute_name);\
|
|
|
|
if (!isAttributeTypeConvertibleTo(attribute.type, AttributeUnderlyingType::TYPE))\
|
|
|
|
throw Exception{\
|
|
|
|
name + ": type mismatch: attribute " + attribute_name + " has type " + toString(attribute.type),\
|
|
|
|
ErrorCodes::TYPE_MISMATCH};\
|
|
|
|
\
|
|
|
|
getItemsNumber<TYPE>(attribute, ids,\
|
2017-07-21 06:35:58 +00:00
|
|
|
[&] (const size_t row, const auto value) { out[row] = value; },\
|
|
|
|
[&] (const size_t) { return def; });\
|
2016-06-07 21:07:44 +00:00
|
|
|
}
|
|
|
|
DECLARE(UInt8)
|
|
|
|
DECLARE(UInt16)
|
|
|
|
DECLARE(UInt32)
|
|
|
|
DECLARE(UInt64)
|
2017-11-14 00:08:54 +00:00
|
|
|
DECLARE(UInt128)
|
2016-06-07 21:07:44 +00:00
|
|
|
DECLARE(Int8)
|
|
|
|
DECLARE(Int16)
|
|
|
|
DECLARE(Int32)
|
|
|
|
DECLARE(Int64)
|
|
|
|
DECLARE(Float32)
|
|
|
|
DECLARE(Float64)
|
|
|
|
#undef DECLARE
|
|
|
|
|
|
|
|
void FlatDictionary::getString(
|
2017-04-01 07:20:54 +00:00
|
|
|
const std::string & attribute_name, const PaddedPODArray<Key> & ids, const String & def,
|
|
|
|
ColumnString * const out) const
|
2016-06-07 21:07:44 +00:00
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
const auto & attribute = getAttribute(attribute_name);
|
|
|
|
if (!isAttributeTypeConvertibleTo(attribute.type, AttributeUnderlyingType::String))
|
|
|
|
throw Exception{
|
|
|
|
name + ": type mismatch: attribute " + attribute_name + " has type " + toString(attribute.type),
|
|
|
|
ErrorCodes::TYPE_MISMATCH};
|
|
|
|
|
|
|
|
FlatDictionary::getItemsImpl<StringRef, StringRef>(attribute, ids,
|
2017-12-01 20:21:35 +00:00
|
|
|
[&] (const size_t, const StringRef value) { out->insertData(value.data, value.size); },
|
2017-07-21 06:35:58 +00:00
|
|
|
[&] (const size_t) { return StringRef{def}; });
|
2016-06-07 21:07:44 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
2016-08-07 09:09:18 +00:00
|
|
|
void FlatDictionary::has(const PaddedPODArray<Key> & ids, PaddedPODArray<UInt8> & out) const
|
2016-06-07 21:07:44 +00:00
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
const auto & attribute = attributes.front();
|
|
|
|
|
|
|
|
switch (attribute.type)
|
|
|
|
{
|
|
|
|
case AttributeUnderlyingType::UInt8: has<UInt8>(attribute, ids, out); break;
|
|
|
|
case AttributeUnderlyingType::UInt16: has<UInt16>(attribute, ids, out); break;
|
|
|
|
case AttributeUnderlyingType::UInt32: has<UInt32>(attribute, ids, out); break;
|
|
|
|
case AttributeUnderlyingType::UInt64: has<UInt64>(attribute, ids, out); break;
|
2017-11-14 00:08:54 +00:00
|
|
|
case AttributeUnderlyingType::UInt128: has<UInt128>(attribute, ids, out); break;
|
2017-04-01 07:20:54 +00:00
|
|
|
case AttributeUnderlyingType::Int8: has<Int8>(attribute, ids, out); break;
|
|
|
|
case AttributeUnderlyingType::Int16: has<Int16>(attribute, ids, out); break;
|
|
|
|
case AttributeUnderlyingType::Int32: has<Int32>(attribute, ids, out); break;
|
|
|
|
case AttributeUnderlyingType::Int64: has<Int64>(attribute, ids, out); break;
|
|
|
|
case AttributeUnderlyingType::Float32: has<Float32>(attribute, ids, out); break;
|
|
|
|
case AttributeUnderlyingType::Float64: has<Float64>(attribute, ids, out); break;
|
|
|
|
case AttributeUnderlyingType::String: has<String>(attribute, ids, out); break;
|
|
|
|
}
|
2016-06-07 21:07:44 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
void FlatDictionary::createAttributes()
|
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
const auto size = dict_struct.attributes.size();
|
|
|
|
attributes.reserve(size);
|
|
|
|
|
|
|
|
for (const auto & attribute : dict_struct.attributes)
|
|
|
|
{
|
|
|
|
attribute_index_by_name.emplace(attribute.name, attributes.size());
|
|
|
|
attributes.push_back(createAttributeWithType(attribute.underlying_type, attribute.null_value));
|
|
|
|
|
|
|
|
if (attribute.hierarchical)
|
|
|
|
{
|
|
|
|
hierarchical_attribute = &attributes.back();
|
|
|
|
|
|
|
|
if (hierarchical_attribute->type != AttributeUnderlyingType::UInt64)
|
|
|
|
throw Exception{
|
|
|
|
name + ": hierarchical attribute must be UInt64.",
|
|
|
|
ErrorCodes::TYPE_MISMATCH};
|
|
|
|
}
|
|
|
|
}
|
2016-06-07 21:07:44 +00:00
|
|
|
}
|
|
|
|
|
2018-02-15 13:08:23 +00:00
|
|
|
void FlatDictionary::blockToAttributes(const Block &block)
|
|
|
|
{
|
|
|
|
const auto & id_column = *block.safeGetByPosition(0).column;
|
|
|
|
element_count += id_column.size();
|
|
|
|
|
|
|
|
for (const auto attribute_idx : ext::range(0, attributes.size()))
|
|
|
|
{
|
|
|
|
const auto &attribute_column = *block.safeGetByPosition(attribute_idx + 1).column;
|
|
|
|
auto &attribute = attributes[attribute_idx];
|
|
|
|
|
|
|
|
for (const auto row_idx : ext::range(0, id_column.size()))
|
|
|
|
setAttributeValue(attribute, id_column[row_idx].get<UInt64>(), attribute_column[row_idx]);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2018-01-15 12:44:39 +00:00
|
|
|
void FlatDictionary::updateData()
|
2016-06-07 21:07:44 +00:00
|
|
|
{
|
2018-02-15 13:08:23 +00:00
|
|
|
if (!saved_block || saved_block->rows() == 0)
|
2018-01-15 12:44:39 +00:00
|
|
|
{
|
2018-02-15 13:08:23 +00:00
|
|
|
auto stream = source_ptr->loadUpdatedAll();
|
2018-01-15 12:44:39 +00:00
|
|
|
stream->readPrefix();
|
2016-06-07 21:07:44 +00:00
|
|
|
|
2018-01-15 12:44:39 +00:00
|
|
|
while (const auto block = stream->read())
|
|
|
|
{
|
2018-02-15 13:08:23 +00:00
|
|
|
/// We are using this to keep saved data if input stream consists of multiple blocks
|
2018-01-15 12:44:39 +00:00
|
|
|
if (!saved_block)
|
|
|
|
saved_block = std::make_shared<DB::Block>(block.cloneEmpty());
|
|
|
|
for (const auto attribute_idx : ext::range(0, attributes.size() + 1))
|
|
|
|
{
|
|
|
|
const IColumn & update_column = *block.getByPosition(attribute_idx).column.get();
|
2018-03-20 10:58:16 +00:00
|
|
|
MutableColumnPtr saved_column = saved_block->getByPosition(attribute_idx).column->assumeMutable();
|
2018-01-15 12:44:39 +00:00
|
|
|
saved_column->insertRangeFrom(update_column, 0, update_column.size());
|
|
|
|
}
|
|
|
|
}
|
|
|
|
stream->readSuffix();
|
|
|
|
}
|
|
|
|
else
|
2017-04-01 07:20:54 +00:00
|
|
|
{
|
2018-02-15 13:08:23 +00:00
|
|
|
auto stream = source_ptr->loadUpdatedAll();
|
2018-01-15 12:44:39 +00:00
|
|
|
stream->readPrefix();
|
|
|
|
|
|
|
|
while (const auto block = stream->read())
|
|
|
|
{
|
2018-02-15 13:08:23 +00:00
|
|
|
const auto &saved_id_column = *saved_block->safeGetByPosition(0).column;
|
2018-01-15 12:44:39 +00:00
|
|
|
const auto &update_id_column = *block.safeGetByPosition(0).column;
|
|
|
|
|
2018-02-15 13:08:23 +00:00
|
|
|
std::unordered_map<Key, std::vector<size_t>> update_ids;
|
|
|
|
for (size_t row = 0; row < update_id_column.size(); ++row)
|
2018-01-15 12:44:39 +00:00
|
|
|
{
|
2018-02-15 13:08:23 +00:00
|
|
|
const auto id = update_id_column.get64(row);
|
|
|
|
update_ids[id].push_back(row);
|
2018-01-15 12:44:39 +00:00
|
|
|
}
|
|
|
|
|
2018-02-15 13:08:23 +00:00
|
|
|
const size_t saved_rows = saved_id_column.size();
|
|
|
|
IColumn::Filter filter(saved_rows);
|
|
|
|
std::unordered_map<Key, std::vector<size_t>>::iterator it;
|
|
|
|
|
|
|
|
for (size_t row = 0; row < saved_id_column.size(); ++row)
|
|
|
|
{
|
|
|
|
auto id = saved_id_column.get64(row);
|
|
|
|
it = update_ids.find(id);
|
2018-01-15 12:44:39 +00:00
|
|
|
|
2018-02-15 13:08:23 +00:00
|
|
|
if (it != update_ids.end())
|
|
|
|
filter[row] = 0;
|
2018-01-15 12:44:39 +00:00
|
|
|
else
|
2018-02-15 13:08:23 +00:00
|
|
|
filter[row] = 1;
|
2018-01-15 12:44:39 +00:00
|
|
|
}
|
2018-02-15 13:08:23 +00:00
|
|
|
|
|
|
|
auto block_columns = block.mutateColumns();
|
|
|
|
for (const auto attribute_idx : ext::range(0, attributes.size() + 1))
|
2018-01-15 12:44:39 +00:00
|
|
|
{
|
2018-02-15 13:08:23 +00:00
|
|
|
auto & column = saved_block->safeGetByPosition(attribute_idx).column;
|
|
|
|
const auto & filtered_column = column->filter(filter, -1);
|
|
|
|
|
|
|
|
block_columns[attribute_idx]->insertRangeFrom(*filtered_column.get(), 0, filtered_column->size());
|
2018-01-15 12:44:39 +00:00
|
|
|
}
|
2018-02-15 13:08:23 +00:00
|
|
|
|
|
|
|
saved_block->setColumns(std::move(block_columns));
|
2018-01-15 12:44:39 +00:00
|
|
|
}
|
|
|
|
stream->readSuffix();
|
|
|
|
}
|
2016-06-07 21:07:44 +00:00
|
|
|
|
2018-01-15 12:44:39 +00:00
|
|
|
if (saved_block)
|
2018-02-15 13:08:23 +00:00
|
|
|
blockToAttributes(*saved_block.get());
|
2018-01-15 12:44:39 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
void FlatDictionary::loadData()
|
|
|
|
{
|
2018-02-15 13:08:23 +00:00
|
|
|
if (!source_ptr->hasUpdateField()) {
|
2018-01-15 12:44:39 +00:00
|
|
|
auto stream = source_ptr->loadAll();
|
|
|
|
stream->readPrefix();
|
2016-06-07 21:07:44 +00:00
|
|
|
|
2018-02-15 13:08:23 +00:00
|
|
|
while (const auto block = stream->read())
|
|
|
|
blockToAttributes(block);
|
2018-01-15 12:44:39 +00:00
|
|
|
|
|
|
|
stream->readSuffix();
|
|
|
|
}
|
|
|
|
else
|
|
|
|
updateData();
|
2016-06-07 21:07:44 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
if (require_nonempty && 0 == element_count)
|
|
|
|
throw Exception{
|
|
|
|
name + ": dictionary source is empty and 'require_nonempty' property is set.",
|
|
|
|
ErrorCodes::DICTIONARY_IS_EMPTY};
|
2016-06-07 21:07:44 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
template <typename T>
|
2016-08-07 09:09:18 +00:00
|
|
|
void FlatDictionary::addAttributeSize(const Attribute & attribute)
|
2016-06-07 21:07:44 +00:00
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
const auto & array_ref = std::get<ContainerPtrType<T>>(attribute.arrays);
|
2017-07-13 16:49:09 +00:00
|
|
|
bytes_allocated += sizeof(PaddedPODArray<T>) + array_ref->allocated_bytes();
|
2017-04-01 07:20:54 +00:00
|
|
|
bucket_count = array_ref->capacity();
|
2016-06-07 21:07:44 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
void FlatDictionary::calculateBytesAllocated()
|
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
bytes_allocated += attributes.size() * sizeof(attributes.front());
|
|
|
|
|
|
|
|
for (const auto & attribute : attributes)
|
|
|
|
{
|
|
|
|
switch (attribute.type)
|
|
|
|
{
|
|
|
|
case AttributeUnderlyingType::UInt8: addAttributeSize<UInt8>(attribute); break;
|
|
|
|
case AttributeUnderlyingType::UInt16: addAttributeSize<UInt16>(attribute); break;
|
|
|
|
case AttributeUnderlyingType::UInt32: addAttributeSize<UInt32>(attribute); break;
|
|
|
|
case AttributeUnderlyingType::UInt64: addAttributeSize<UInt64>(attribute); break;
|
2017-11-14 00:08:54 +00:00
|
|
|
case AttributeUnderlyingType::UInt128: addAttributeSize<UInt128>(attribute); break;
|
2017-04-01 07:20:54 +00:00
|
|
|
case AttributeUnderlyingType::Int8: addAttributeSize<Int8>(attribute); break;
|
|
|
|
case AttributeUnderlyingType::Int16: addAttributeSize<Int16>(attribute); break;
|
|
|
|
case AttributeUnderlyingType::Int32: addAttributeSize<Int32>(attribute); break;
|
|
|
|
case AttributeUnderlyingType::Int64: addAttributeSize<Int64>(attribute); break;
|
|
|
|
case AttributeUnderlyingType::Float32: addAttributeSize<Float32>(attribute); break;
|
|
|
|
case AttributeUnderlyingType::Float64: addAttributeSize<Float64>(attribute); break;
|
|
|
|
case AttributeUnderlyingType::String:
|
|
|
|
{
|
|
|
|
addAttributeSize<StringRef>(attribute);
|
|
|
|
bytes_allocated += sizeof(Arena) + attribute.string_arena->size();
|
|
|
|
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
2016-06-07 21:07:44 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
template <typename T>
|
2016-08-07 09:09:18 +00:00
|
|
|
void FlatDictionary::createAttributeImpl(Attribute & attribute, const Field & null_value)
|
2016-06-07 21:07:44 +00:00
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
const auto & null_value_ref = std::get<T>(attribute.null_values) =
|
|
|
|
null_value.get<typename NearestFieldType<T>::Type>();
|
|
|
|
std::get<ContainerPtrType<T>>(attribute.arrays) =
|
|
|
|
std::make_unique<ContainerType<T>>(initial_array_size, null_value_ref);
|
2016-06-07 21:07:44 +00:00
|
|
|
}
|
|
|
|
|
2016-08-24 22:21:48 +00:00
|
|
|
template <>
|
|
|
|
void FlatDictionary::createAttributeImpl<String>(Attribute & attribute, const Field & null_value)
|
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
attribute.string_arena = std::make_unique<Arena>();
|
|
|
|
auto & null_value_ref = std::get<StringRef>(attribute.null_values);
|
2017-09-08 03:47:27 +00:00
|
|
|
const String & string = null_value.get<typename NearestFieldType<String>::Type>();
|
2017-04-01 07:20:54 +00:00
|
|
|
const auto string_in_arena = attribute.string_arena->insert(string.data(), string.size());
|
|
|
|
null_value_ref = StringRef{string_in_arena, string.size()};
|
|
|
|
std::get<ContainerPtrType<StringRef>>(attribute.arrays) =
|
|
|
|
std::make_unique<ContainerType<StringRef>>(initial_array_size, null_value_ref);
|
2016-08-24 22:21:48 +00:00
|
|
|
}
|
|
|
|
|
2016-06-07 21:07:44 +00:00
|
|
|
|
2016-08-07 09:09:18 +00:00
|
|
|
FlatDictionary::Attribute FlatDictionary::createAttributeWithType(const AttributeUnderlyingType type, const Field & null_value)
|
2016-06-07 21:07:44 +00:00
|
|
|
{
|
2017-12-01 20:21:35 +00:00
|
|
|
Attribute attr{type, {}, {}, {}};
|
2017-04-01 07:20:54 +00:00
|
|
|
|
|
|
|
switch (type)
|
|
|
|
{
|
|
|
|
case AttributeUnderlyingType::UInt8: createAttributeImpl<UInt8>(attr, null_value); break;
|
|
|
|
case AttributeUnderlyingType::UInt16: createAttributeImpl<UInt16>(attr, null_value); break;
|
|
|
|
case AttributeUnderlyingType::UInt32: createAttributeImpl<UInt32>(attr, null_value); break;
|
|
|
|
case AttributeUnderlyingType::UInt64: createAttributeImpl<UInt64>(attr, null_value); break;
|
2017-11-14 00:08:54 +00:00
|
|
|
case AttributeUnderlyingType::UInt128: createAttributeImpl<UInt128>(attr, null_value); break;
|
2017-04-01 07:20:54 +00:00
|
|
|
case AttributeUnderlyingType::Int8: createAttributeImpl<Int8>(attr, null_value); break;
|
|
|
|
case AttributeUnderlyingType::Int16: createAttributeImpl<Int16>(attr, null_value); break;
|
|
|
|
case AttributeUnderlyingType::Int32: createAttributeImpl<Int32>(attr, null_value); break;
|
|
|
|
case AttributeUnderlyingType::Int64: createAttributeImpl<Int64>(attr, null_value); break;
|
|
|
|
case AttributeUnderlyingType::Float32: createAttributeImpl<Float32>(attr, null_value); break;
|
|
|
|
case AttributeUnderlyingType::Float64: createAttributeImpl<Float64>(attr, null_value); break;
|
|
|
|
case AttributeUnderlyingType::String: createAttributeImpl<String>(attr, null_value); break;
|
|
|
|
}
|
|
|
|
|
|
|
|
return attr;
|
2016-06-07 21:07:44 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
template <typename OutputType, typename ValueSetter, typename DefaultGetter>
|
|
|
|
void FlatDictionary::getItemsNumber(
|
2017-04-01 07:20:54 +00:00
|
|
|
const Attribute & attribute,
|
|
|
|
const PaddedPODArray<Key> & ids,
|
|
|
|
ValueSetter && set_value,
|
|
|
|
DefaultGetter && get_default) const
|
2016-06-07 21:07:44 +00:00
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
if (false) {}
|
2016-06-07 21:07:44 +00:00
|
|
|
#define DISPATCH(TYPE) \
|
2017-04-01 07:20:54 +00:00
|
|
|
else if (attribute.type == AttributeUnderlyingType::TYPE) \
|
|
|
|
getItemsImpl<TYPE, OutputType>(attribute, ids, std::forward<ValueSetter>(set_value), std::forward<DefaultGetter>(get_default));
|
|
|
|
DISPATCH(UInt8)
|
|
|
|
DISPATCH(UInt16)
|
|
|
|
DISPATCH(UInt32)
|
|
|
|
DISPATCH(UInt64)
|
2017-11-14 00:08:54 +00:00
|
|
|
DISPATCH(UInt128)
|
2017-04-01 07:20:54 +00:00
|
|
|
DISPATCH(Int8)
|
|
|
|
DISPATCH(Int16)
|
|
|
|
DISPATCH(Int32)
|
|
|
|
DISPATCH(Int64)
|
|
|
|
DISPATCH(Float32)
|
|
|
|
DISPATCH(Float64)
|
2016-06-07 21:07:44 +00:00
|
|
|
#undef DISPATCH
|
2017-04-01 07:20:54 +00:00
|
|
|
else
|
|
|
|
throw Exception("Unexpected type of attribute: " + toString(attribute.type), ErrorCodes::LOGICAL_ERROR);
|
2016-06-07 21:07:44 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
template <typename AttributeType, typename OutputType, typename ValueSetter, typename DefaultGetter>
|
|
|
|
void FlatDictionary::getItemsImpl(
|
2017-04-01 07:20:54 +00:00
|
|
|
const Attribute & attribute,
|
|
|
|
const PaddedPODArray<Key> & ids,
|
|
|
|
ValueSetter && set_value,
|
|
|
|
DefaultGetter && get_default) const
|
2016-06-07 21:07:44 +00:00
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
const auto & attr = *std::get<ContainerPtrType<AttributeType>>(attribute.arrays);
|
|
|
|
const auto rows = ext::size(ids);
|
2016-06-07 21:07:44 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
for (const auto row : ext::range(0, rows))
|
|
|
|
{
|
|
|
|
const auto id = ids[row];
|
2017-11-14 00:08:54 +00:00
|
|
|
set_value(row, id < ext::size(attr) && loaded_ids[id] ? static_cast<OutputType>(attr[id]) : get_default(row));
|
2017-04-01 07:20:54 +00:00
|
|
|
}
|
2016-06-07 21:07:44 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
query_count.fetch_add(rows, std::memory_order_relaxed);
|
2016-06-07 21:07:44 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
template <typename T>
|
2016-08-24 22:21:48 +00:00
|
|
|
void FlatDictionary::resize(Attribute & attribute, const Key id)
|
2016-06-07 21:07:44 +00:00
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
if (id >= max_array_size)
|
|
|
|
throw Exception{
|
|
|
|
name + ": identifier should be less than " + toString(max_array_size),
|
|
|
|
ErrorCodes::ARGUMENT_OUT_OF_BOUND};
|
|
|
|
|
|
|
|
auto & array = *std::get<ContainerPtrType<T>>(attribute.arrays);
|
|
|
|
if (id >= array.size())
|
|
|
|
{
|
|
|
|
const size_t elements_count = id + 1; //id=0 -> elements_count=1
|
|
|
|
loaded_ids.resize(elements_count, false);
|
|
|
|
array.resize_fill(elements_count, std::get<T>(attribute.null_values));
|
|
|
|
}
|
2016-08-24 22:21:48 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
template <typename T>
|
|
|
|
void FlatDictionary::setAttributeValueImpl(Attribute & attribute, const Key id, const T & value)
|
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
resize<T>(attribute, id);
|
|
|
|
auto & array = *std::get<ContainerPtrType<T>>(attribute.arrays);
|
|
|
|
array[id] = value;
|
|
|
|
loaded_ids[id] = true;
|
2016-06-07 21:07:44 +00:00
|
|
|
}
|
|
|
|
|
2016-08-24 22:21:48 +00:00
|
|
|
template <>
|
|
|
|
void FlatDictionary::setAttributeValueImpl<String>(Attribute & attribute, const Key id, const String & string)
|
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
resize<StringRef>(attribute, id);
|
|
|
|
const auto string_in_arena = attribute.string_arena->insert(string.data(), string.size());
|
|
|
|
auto & array = *std::get<ContainerPtrType<StringRef>>(attribute.arrays);
|
|
|
|
array[id] = StringRef{string_in_arena, string.size()};
|
|
|
|
loaded_ids[id] = true;
|
2016-08-24 22:21:48 +00:00
|
|
|
}
|
2016-06-07 21:07:44 +00:00
|
|
|
|
2016-08-07 09:09:18 +00:00
|
|
|
void FlatDictionary::setAttributeValue(Attribute & attribute, const Key id, const Field & value)
|
2016-06-07 21:07:44 +00:00
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
switch (attribute.type)
|
|
|
|
{
|
|
|
|
case AttributeUnderlyingType::UInt8: setAttributeValueImpl<UInt8>(attribute, id, value.get<UInt64>()); break;
|
|
|
|
case AttributeUnderlyingType::UInt16: setAttributeValueImpl<UInt16>(attribute, id, value.get<UInt64>()); break;
|
|
|
|
case AttributeUnderlyingType::UInt32: setAttributeValueImpl<UInt32>(attribute, id, value.get<UInt64>()); break;
|
|
|
|
case AttributeUnderlyingType::UInt64: setAttributeValueImpl<UInt64>(attribute, id, value.get<UInt64>()); break;
|
2017-11-14 00:08:54 +00:00
|
|
|
case AttributeUnderlyingType::UInt128: setAttributeValueImpl<UInt128>(attribute, id, value.get<UInt128>()); break;
|
2017-04-01 07:20:54 +00:00
|
|
|
case AttributeUnderlyingType::Int8: setAttributeValueImpl<Int8>(attribute, id, value.get<Int64>()); break;
|
|
|
|
case AttributeUnderlyingType::Int16: setAttributeValueImpl<Int16>(attribute, id, value.get<Int64>()); break;
|
|
|
|
case AttributeUnderlyingType::Int32: setAttributeValueImpl<Int32>(attribute, id, value.get<Int64>()); break;
|
|
|
|
case AttributeUnderlyingType::Int64: setAttributeValueImpl<Int64>(attribute, id, value.get<Int64>()); break;
|
|
|
|
case AttributeUnderlyingType::Float32: setAttributeValueImpl<Float32>(attribute, id, value.get<Float64>()); break;
|
|
|
|
case AttributeUnderlyingType::Float64: setAttributeValueImpl<Float64>(attribute, id, value.get<Float64>()); break;
|
|
|
|
case AttributeUnderlyingType::String: setAttributeValueImpl<String>(attribute, id, value.get<String>()); break;
|
|
|
|
}
|
2016-06-07 21:07:44 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
2016-08-07 09:09:18 +00:00
|
|
|
const FlatDictionary::Attribute & FlatDictionary::getAttribute(const std::string & attribute_name) const
|
2016-06-07 21:07:44 +00:00
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
const auto it = attribute_index_by_name.find(attribute_name);
|
|
|
|
if (it == std::end(attribute_index_by_name))
|
|
|
|
throw Exception{
|
|
|
|
name + ": no such attribute '" + attribute_name + "'",
|
|
|
|
ErrorCodes::BAD_ARGUMENTS};
|
2016-06-07 21:07:44 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
return attributes[it->second];
|
2016-06-07 21:07:44 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
template <typename T>
|
2017-12-01 20:21:35 +00:00
|
|
|
void FlatDictionary::has(const Attribute &, const PaddedPODArray<Key> & ids, PaddedPODArray<UInt8> & out) const
|
2016-06-07 21:07:44 +00:00
|
|
|
{
|
2017-04-01 07:20:54 +00:00
|
|
|
const auto ids_count = ext::size(ids);
|
2016-06-07 21:07:44 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
for (const auto i : ext::range(0, ids_count))
|
|
|
|
{
|
|
|
|
const auto id = ids[i];
|
|
|
|
out[i] = id < loaded_ids.size() && loaded_ids[id];
|
|
|
|
}
|
2016-06-07 21:07:44 +00:00
|
|
|
|
2017-04-01 07:20:54 +00:00
|
|
|
query_count.fetch_add(ids_count, std::memory_order_relaxed);
|
2016-06-07 21:07:44 +00:00
|
|
|
}
|
|
|
|
|
2017-04-28 18:33:31 +00:00
|
|
|
|
|
|
|
PaddedPODArray<FlatDictionary::Key> FlatDictionary::getIds() const
|
|
|
|
{
|
|
|
|
const auto ids_count = ext::size(loaded_ids);
|
|
|
|
|
|
|
|
PaddedPODArray<Key> ids;
|
|
|
|
for (auto idx : ext::range(0, ids_count))
|
2018-01-10 00:04:08 +00:00
|
|
|
if (loaded_ids[idx])
|
2017-04-28 18:33:31 +00:00
|
|
|
ids.push_back(idx);
|
|
|
|
return ids;
|
|
|
|
}
|
|
|
|
|
2017-05-29 17:26:45 +00:00
|
|
|
BlockInputStreamPtr FlatDictionary::getBlockInputStream(const Names & column_names, size_t max_block_size) const
|
2017-04-28 18:33:31 +00:00
|
|
|
{
|
2017-05-04 18:14:23 +00:00
|
|
|
using BlockInputStreamType = DictionaryBlockInputStream<FlatDictionary, Key>;
|
2017-05-29 17:26:45 +00:00
|
|
|
return std::make_shared<BlockInputStreamType>(shared_from_this(), max_block_size, getIds() ,column_names);
|
2017-04-28 18:33:31 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
2016-06-07 21:07:44 +00:00
|
|
|
}
|