ClickHouse/dbms/src/DataStreams/NativeBlockInputStream.cpp

181 lines
5.1 KiB
C++
Raw Normal View History

2012-08-26 11:14:52 +00:00
#include <DB/Core/Defines.h>
2011-08-19 19:18:15 +00:00
#include <DB/IO/ReadHelpers.h>
#include <DB/IO/VarInt.h>
#include <DB/IO/CompressedReadBufferFromFile.h>
2011-08-19 19:18:15 +00:00
2012-08-26 11:14:52 +00:00
#include <DB/Columns/ColumnArray.h>
#include <DB/DataTypes/DataTypeArray.h>
#include <DB/DataTypes/DataTypeFactory.h>
2012-08-26 11:14:52 +00:00
2011-08-19 19:18:15 +00:00
#include <DB/DataStreams/NativeBlockInputStream.h>
namespace DB
{
2012-08-26 11:14:52 +00:00
NativeBlockInputStream::NativeBlockInputStream(
ReadBuffer & istr_, UInt64 server_revision_,
const IndexForNativeFormat * index_)
: istr(istr_), server_revision(server_revision_), index(index_)
{
if (index)
{
istr_concrete = typeid_cast<CompressedReadBufferFromFile *>(&istr);
if (!istr_concrete)
throw Exception("When need to use index for NativeBlockInputStream, istr must be CompressedReadBufferFromFile.", ErrorCodes::LOGICAL_ERROR);
index_block_it = index->blocks.begin();
index_column_it = index_block_it->columns.begin();
}
}
void NativeBlockInputStream::readData(const IDataType & type, IColumn & column, ReadBuffer & istr, size_t rows)
2012-08-26 11:14:52 +00:00
{
/** Для массивов требуется сначала десериализовать смещения, а потом значения.
*/
if (const DataTypeArray * type_arr = typeid_cast<const DataTypeArray *>(&type))
2012-08-26 11:14:52 +00:00
{
IColumn & offsets_column = *typeid_cast<ColumnArray &>(column).getOffsetsColumn();
type_arr->getOffsetsType()->deserializeBinary(offsets_column, istr, rows, 0);
2012-08-26 11:14:52 +00:00
if (offsets_column.size() != rows)
throw Exception("Cannot read all data in NativeBlockInputStream.", ErrorCodes::CANNOT_READ_ALL_DATA);
if (rows)
readData(
*type_arr->getNestedType(),
typeid_cast<ColumnArray &>(column).getData(),
2012-08-26 11:14:52 +00:00
istr,
typeid_cast<const ColumnArray &>(column).getOffsets()[rows - 1]);
2012-08-26 11:14:52 +00:00
}
else
type.deserializeBinary(column, istr, rows, 0); /// TODO Использовать avg_value_size_hint.
2012-08-26 11:14:52 +00:00
if (column.size() != rows)
throw Exception("Cannot read all data in NativeBlockInputStream.", ErrorCodes::CANNOT_READ_ALL_DATA);
}
2011-09-04 21:23:19 +00:00
Block NativeBlockInputStream::readImpl()
2011-08-19 19:18:15 +00:00
{
Block res;
const DataTypeFactory & data_type_factory = DataTypeFactory::instance();
if (index && index_block_it == index->blocks.end())
return res;
2011-08-19 19:18:15 +00:00
if (istr.eof())
{
if (index)
throw Exception("Input doesn't contain all data for index.", ErrorCodes::CANNOT_READ_ALL_DATA);
2011-08-19 19:18:15 +00:00
return res;
}
/// Дополнительная информация о блоке.
if (server_revision >= DBMS_MIN_REVISION_WITH_BLOCK_INFO)
res.info.read(istr);
2011-08-19 19:18:15 +00:00
/// Размеры
size_t columns = 0;
size_t rows = 0;
if (!index)
{
readVarUInt(columns, istr);
readVarUInt(rows, istr);
}
else
{
columns = index_block_it->num_columns;
rows = index_block_it->num_rows;
}
2011-08-19 19:18:15 +00:00
for (size_t i = 0; i < columns; ++i)
{
if (index) /// Если текущая позиция какая требуется, то реального seek-а не происходит.
istr_concrete->seek(index_column_it->location.offset_in_compressed_file, index_column_it->location.offset_in_decompressed_block);
ColumnWithTypeAndName column;
2011-08-19 19:18:15 +00:00
/// Имя
readBinary(column.name, istr);
2011-08-19 19:18:15 +00:00
/// Тип
String type_name;
readBinary(type_name, istr);
2011-08-19 19:18:15 +00:00
column.type = data_type_factory.get(type_name);
if (index)
{
/// Индекс позволяет сделать проверки.
if (index_column_it->name != column.name)
throw Exception("Index points to column with wrong name: corrupted index or data", ErrorCodes::INCORRECT_INDEX);
if (index_column_it->type != type_name)
throw Exception("Index points to column with wrong type: corrupted index or data", ErrorCodes::INCORRECT_INDEX);
}
2011-08-19 19:18:15 +00:00
/// Данные
column.column = column.type->createColumn();
2012-08-26 11:14:52 +00:00
readData(*column.type, *column.column, istr, rows);
2011-08-19 19:18:15 +00:00
res.insert(column);
if (index)
++index_column_it;
}
if (index)
{
if (index_column_it != index_block_it->columns.end())
throw Exception("Inconsistent index: not all columns were read", ErrorCodes::INCORRECT_INDEX);
++index_block_it;
2011-08-19 19:18:15 +00:00
}
return res;
}
void IndexForNativeFormat::read(ReadBuffer & istr, const NameSet & required_columns)
{
while (!istr.eof())
{
blocks.emplace_back();
IndexOfBlockForNativeFormat & block = blocks.back();
readVarUInt(block.num_columns, istr);
readVarUInt(block.num_rows, istr);
if (block.num_columns < required_columns.size())
throw Exception("Index contain less than required columns", ErrorCodes::INCORRECT_INDEX);
for (size_t i = 0; i < block.num_columns; ++i)
{
IndexOfOneColumnForNativeFormat column_index;
readBinary(column_index.name, istr);
readBinary(column_index.type, istr);
readBinary(column_index.location.offset_in_compressed_file, istr);
readBinary(column_index.location.offset_in_decompressed_block, istr);
if (required_columns.count(column_index.name))
block.columns.push_back(std::move(column_index));
}
if (block.columns.size() < required_columns.size())
throw Exception("Index contain less than required columns", ErrorCodes::INCORRECT_INDEX);
if (block.columns.size() > required_columns.size())
throw Exception("Index contain duplicate columns", ErrorCodes::INCORRECT_INDEX);
block.num_columns = block.columns.size();
}
}
2011-08-19 19:18:15 +00:00
}