ClickHouse/dbms/DataStreams/ConvertingBlockInputStream.cpp

121 lines
4.1 KiB
C++
Raw Normal View History

#include <DataStreams/ConvertingBlockInputStream.h>
#include <Interpreters/castColumn.h>
#include <Columns/ColumnConst.h>
#include <Common/assert_cast.h>
#include <Common/quoteString.h>
#include <Parsers/IAST.h>
2020-01-23 15:53:58 +00:00
namespace DB
{
namespace ErrorCodes
{
extern const int THERE_IS_NO_COLUMN;
extern const int BLOCKS_HAVE_DIFFERENT_STRUCTURE;
2018-02-23 01:00:47 +00:00
extern const int NUMBER_OF_COLUMNS_DOESNT_MATCH;
}
2018-08-20 16:10:58 +00:00
static ColumnPtr castColumnWithDiagnostic(const ColumnWithTypeAndName & src_elem, const ColumnWithTypeAndName & res_elem, const Context & context)
{
try
{
return castColumn(src_elem, res_elem.type, context);
}
catch (Exception & e)
{
e.addMessage("while converting source column " + backQuoteIfNeed(src_elem.name) + " to destination column " + backQuoteIfNeed(res_elem.name));
throw;
}
}
ConvertingBlockInputStream::ConvertingBlockInputStream(
const Context & context_,
const BlockInputStreamPtr & input,
2018-02-23 01:00:47 +00:00
const Block & result_header,
2020-01-23 15:53:58 +00:00
MatchColumnsMode mode)
: context(context_), header(result_header), conversion(header.columns())
{
children.emplace_back(input);
Block input_header = input->getHeader();
2018-02-23 01:00:47 +00:00
size_t num_input_columns = input_header.columns();
2018-02-23 01:00:47 +00:00
size_t num_result_columns = result_header.columns();
if (mode == MatchColumnsMode::Position && num_input_columns != num_result_columns)
throw Exception("Number of columns doesn't match", ErrorCodes::NUMBER_OF_COLUMNS_DOESNT_MATCH);
2018-02-23 01:00:47 +00:00
for (size_t result_col_num = 0; result_col_num < num_result_columns; ++result_col_num)
{
const auto & res_elem = result_header.getByPosition(result_col_num);
2018-02-23 01:00:47 +00:00
switch (mode)
{
case MatchColumnsMode::Position:
conversion[result_col_num] = result_col_num;
break;
case MatchColumnsMode::Name:
if (input_header.has(res_elem.name))
conversion[result_col_num] = input_header.getPositionByName(res_elem.name);
else
throw Exception("Cannot find column " + backQuote(res_elem.name) + " in source stream",
2018-02-23 01:00:47 +00:00
ErrorCodes::THERE_IS_NO_COLUMN);
break;
}
const auto & src_elem = input_header.getByPosition(conversion[result_col_num]);
/// Check constants.
if (isColumnConst(*res_elem.column))
{
if (!isColumnConst(*src_elem.column))
throw Exception("Cannot convert column " + backQuoteIfNeed(res_elem.name)
+ " because it is non constant in source stream but must be constant in result",
ErrorCodes::BLOCKS_HAVE_DIFFERENT_STRUCTURE);
else if (assert_cast<const ColumnConst &>(*src_elem.column).getField() != assert_cast<const ColumnConst &>(*res_elem.column).getField())
throw Exception("Cannot convert column " + backQuoteIfNeed(res_elem.name)
+ " because it is constant but values of constants are different in source and result",
ErrorCodes::BLOCKS_HAVE_DIFFERENT_STRUCTURE);
}
/// Check conversion by dry run CAST function.
2018-08-20 16:10:58 +00:00
castColumnWithDiagnostic(src_elem, res_elem, context);
}
}
Block ConvertingBlockInputStream::readImpl()
{
Block src = children.back()->read();
if (!src)
return src;
Block res = header.cloneEmpty();
2020-02-21 19:17:09 +00:00
/// This is important because header.cloneEmpty() doesn't copy info about aggregation bucket.
/// Otherwise information in buckets may be lost (and aggregation will return wrong result).
res.info = src.info;
for (size_t res_pos = 0, size = conversion.size(); res_pos < size; ++res_pos)
{
const auto & src_elem = src.getByPosition(conversion[res_pos]);
2020-01-23 15:53:58 +00:00
auto & res_elem = res.getByPosition(res_pos);
2018-08-20 16:10:58 +00:00
ColumnPtr converted = castColumnWithDiagnostic(src_elem, res_elem, context);
if (isColumnConst(*src_elem.column) && !isColumnConst(*res_elem.column))
converted = converted->convertToFullColumnIfConst();
res_elem.column = std::move(converted);
}
return res;
}
}