2020-03-26 16:33:00 +00:00
|
|
|
#include <Processors/Formats/Impl/MsgPackRowInputFormat.h>
|
2020-07-10 19:08:18 +00:00
|
|
|
|
|
|
|
#if USE_MSGPACK
|
|
|
|
|
|
|
|
#include <cstdlib>
|
2020-03-26 16:33:00 +00:00
|
|
|
#include <Common/assert_cast.h>
|
|
|
|
#include <IO/ReadHelpers.h>
|
|
|
|
|
|
|
|
#include <DataTypes/DataTypeArray.h>
|
|
|
|
#include <DataTypes/DataTypeDateTime64.h>
|
|
|
|
#include <DataTypes/DataTypeNullable.h>
|
2021-09-16 11:25:45 +00:00
|
|
|
#include <DataTypes/DataTypeMap.h>
|
|
|
|
#include <DataTypes/DataTypeLowCardinality.h>
|
2021-12-15 11:30:57 +00:00
|
|
|
#include <DataTypes/DataTypeString.h>
|
2020-03-26 16:33:00 +00:00
|
|
|
|
|
|
|
#include <Columns/ColumnArray.h>
|
|
|
|
#include <Columns/ColumnNullable.h>
|
|
|
|
#include <Columns/ColumnString.h>
|
|
|
|
#include <Columns/ColumnsNumber.h>
|
2021-09-16 11:25:45 +00:00
|
|
|
#include <Columns/ColumnMap.h>
|
|
|
|
#include <Columns/ColumnLowCardinality.h>
|
2020-03-26 16:33:00 +00:00
|
|
|
|
|
|
|
namespace DB
|
|
|
|
{
|
|
|
|
|
|
|
|
namespace ErrorCodes
|
|
|
|
{
|
|
|
|
extern const int ILLEGAL_COLUMN;
|
|
|
|
extern const int INCORRECT_DATA;
|
2021-12-15 11:30:57 +00:00
|
|
|
extern const int BAD_ARGUMENTS;
|
|
|
|
extern const int UNEXPECTED_END_OF_FILE;
|
2020-03-26 16:33:00 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
MsgPackRowInputFormat::MsgPackRowInputFormat(const Block & header_, ReadBuffer & in_, Params params_)
|
2021-12-10 17:54:08 +00:00
|
|
|
: MsgPackRowInputFormat(header_, std::make_unique<PeekableReadBuffer>(in_), params_) {}
|
|
|
|
|
|
|
|
MsgPackRowInputFormat::MsgPackRowInputFormat(const Block & header_, std::unique_ptr<PeekableReadBuffer> buf_, Params params_)
|
|
|
|
: IRowInputFormat(header_, *buf_, std::move(params_)), buf(std::move(buf_)), parser(visitor), data_types(header_.getDataTypes()) {}
|
2020-04-14 23:08:55 +00:00
|
|
|
|
2020-06-11 00:51:27 +00:00
|
|
|
void MsgPackRowInputFormat::resetParser()
|
|
|
|
{
|
|
|
|
IRowInputFormat::resetParser();
|
2021-12-10 18:37:08 +00:00
|
|
|
buf->reset();
|
2020-06-11 00:51:27 +00:00
|
|
|
visitor.reset();
|
|
|
|
}
|
|
|
|
|
2020-04-18 11:26:57 +00:00
|
|
|
void MsgPackVisitor::set_info(IColumn & column, DataTypePtr type) // NOLINT
|
2020-04-14 23:08:55 +00:00
|
|
|
{
|
2020-04-17 09:35:38 +00:00
|
|
|
while (!info_stack.empty())
|
2020-03-26 16:33:00 +00:00
|
|
|
{
|
2020-04-17 09:35:38 +00:00
|
|
|
info_stack.pop();
|
2020-03-26 16:33:00 +00:00
|
|
|
}
|
2020-04-17 09:35:38 +00:00
|
|
|
info_stack.push(Info{column, type});
|
2020-03-26 16:33:00 +00:00
|
|
|
}
|
|
|
|
|
2020-06-11 00:51:27 +00:00
|
|
|
void MsgPackVisitor::reset()
|
|
|
|
{
|
|
|
|
info_stack = {};
|
|
|
|
}
|
|
|
|
|
2021-09-16 14:05:07 +00:00
|
|
|
template <typename InsertFunc>
|
|
|
|
static bool checkAndInsertNullable(IColumn & column, DataTypePtr type, InsertFunc insert_func)
|
|
|
|
{
|
|
|
|
if (type->isNullable())
|
|
|
|
{
|
|
|
|
auto & nullable_column = assert_cast<ColumnNullable &>(column);
|
|
|
|
auto & nested_column = nullable_column.getNestedColumn();
|
2021-09-17 14:32:23 +00:00
|
|
|
const auto & nested_type = assert_cast<const DataTypeNullable *>(type.get())->getNestedType();
|
2021-09-16 14:05:07 +00:00
|
|
|
insert_func(nested_column, nested_type);
|
|
|
|
nullable_column.getNullMapColumn().insertValue(0);
|
|
|
|
return true;
|
|
|
|
}
|
2021-09-16 11:25:45 +00:00
|
|
|
|
2021-09-16 14:05:07 +00:00
|
|
|
return false;
|
|
|
|
}
|
2021-09-16 11:25:45 +00:00
|
|
|
|
2021-09-16 14:05:07 +00:00
|
|
|
template <typename InsertFunc>
|
|
|
|
static bool checkAndInsertLowCardinality(IColumn & column, DataTypePtr type, InsertFunc insert_func)
|
2020-03-26 16:33:00 +00:00
|
|
|
{
|
2021-09-22 10:47:03 +00:00
|
|
|
if (type->lowCardinality())
|
2021-09-16 14:05:07 +00:00
|
|
|
{
|
|
|
|
auto & lc_column = assert_cast<ColumnLowCardinality &>(column);
|
|
|
|
auto tmp_column = lc_column.getDictionary().getNestedColumn()->cloneEmpty();
|
|
|
|
auto dict_type = assert_cast<const DataTypeLowCardinality *>(type.get())->getDictionaryType();
|
|
|
|
insert_func(*tmp_column, dict_type);
|
|
|
|
lc_column.insertFromFullColumn(*tmp_column, 0);
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void insertInteger(IColumn & column, DataTypePtr type, UInt64 value)
|
|
|
|
{
|
|
|
|
auto insert_func = [&](IColumn & column_, DataTypePtr type_)
|
|
|
|
{
|
|
|
|
insertInteger(column_, type_, value);
|
|
|
|
};
|
|
|
|
|
|
|
|
if (checkAndInsertNullable(column, type, insert_func) || checkAndInsertLowCardinality(column, type, insert_func))
|
|
|
|
return;
|
|
|
|
|
|
|
|
switch (type->getTypeId())
|
2020-03-26 16:33:00 +00:00
|
|
|
{
|
|
|
|
case TypeIndex::UInt8:
|
|
|
|
{
|
2021-09-16 14:05:07 +00:00
|
|
|
assert_cast<ColumnUInt8 &>(column).insertValue(value);
|
2020-04-17 09:35:38 +00:00
|
|
|
break;
|
2020-03-26 16:33:00 +00:00
|
|
|
}
|
|
|
|
case TypeIndex::Date: [[fallthrough]];
|
|
|
|
case TypeIndex::UInt16:
|
|
|
|
{
|
2021-09-16 14:05:07 +00:00
|
|
|
assert_cast<ColumnUInt16 &>(column).insertValue(value);
|
2020-04-17 09:35:38 +00:00
|
|
|
break;
|
2020-03-26 16:33:00 +00:00
|
|
|
}
|
|
|
|
case TypeIndex::DateTime: [[fallthrough]];
|
|
|
|
case TypeIndex::UInt32:
|
|
|
|
{
|
2021-09-16 14:05:07 +00:00
|
|
|
assert_cast<ColumnUInt32 &>(column).insertValue(value);
|
2020-04-17 09:35:38 +00:00
|
|
|
break;
|
2020-03-26 16:33:00 +00:00
|
|
|
}
|
|
|
|
case TypeIndex::UInt64:
|
|
|
|
{
|
2021-09-16 14:05:07 +00:00
|
|
|
assert_cast<ColumnUInt64 &>(column).insertValue(value);
|
2020-04-17 09:35:38 +00:00
|
|
|
break;
|
2020-03-26 16:33:00 +00:00
|
|
|
}
|
|
|
|
case TypeIndex::Int8:
|
|
|
|
{
|
2021-09-16 14:05:07 +00:00
|
|
|
assert_cast<ColumnInt8 &>(column).insertValue(value);
|
2020-04-17 09:35:38 +00:00
|
|
|
break;
|
2020-03-26 16:33:00 +00:00
|
|
|
}
|
|
|
|
case TypeIndex::Int16:
|
|
|
|
{
|
2021-09-16 14:05:07 +00:00
|
|
|
assert_cast<ColumnInt16 &>(column).insertValue(value);
|
2020-04-17 09:35:38 +00:00
|
|
|
break;
|
2020-03-26 16:33:00 +00:00
|
|
|
}
|
|
|
|
case TypeIndex::Int32:
|
|
|
|
{
|
2021-09-16 14:05:07 +00:00
|
|
|
assert_cast<ColumnInt32 &>(column).insertValue(value);
|
2020-04-17 09:35:38 +00:00
|
|
|
break;
|
2020-03-26 16:33:00 +00:00
|
|
|
}
|
|
|
|
case TypeIndex::Int64:
|
|
|
|
{
|
2021-09-16 14:05:07 +00:00
|
|
|
assert_cast<ColumnInt64 &>(column).insertValue(value);
|
2020-04-17 09:35:38 +00:00
|
|
|
break;
|
2020-03-26 16:33:00 +00:00
|
|
|
}
|
|
|
|
case TypeIndex::DateTime64:
|
|
|
|
{
|
2021-09-16 14:05:07 +00:00
|
|
|
assert_cast<DataTypeDateTime64::ColumnType &>(column).insertValue(value);
|
2021-09-16 11:25:45 +00:00
|
|
|
break;
|
|
|
|
}
|
2020-03-26 16:33:00 +00:00
|
|
|
default:
|
2021-09-16 14:05:07 +00:00
|
|
|
throw Exception(ErrorCodes::ILLEGAL_COLUMN, "Cannot insert MessagePack integer into column with type {}.", type->getName());
|
2020-03-26 16:33:00 +00:00
|
|
|
}
|
2020-04-17 09:35:38 +00:00
|
|
|
}
|
|
|
|
|
2021-09-16 14:05:07 +00:00
|
|
|
static void insertString(IColumn & column, DataTypePtr type, const char * value, size_t size)
|
|
|
|
{
|
|
|
|
auto insert_func = [&](IColumn & column_, DataTypePtr type_)
|
|
|
|
{
|
|
|
|
insertString(column_, type_, value, size);
|
|
|
|
};
|
|
|
|
|
|
|
|
if (checkAndInsertNullable(column, type, insert_func) || checkAndInsertLowCardinality(column, type, insert_func))
|
|
|
|
return;
|
|
|
|
|
|
|
|
if (!isStringOrFixedString(type))
|
|
|
|
throw Exception(ErrorCodes::ILLEGAL_COLUMN, "Cannot insert MessagePack string into column with type {}.", type->getName());
|
|
|
|
|
|
|
|
column.insertData(value, size);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void insertFloat32(IColumn & column, DataTypePtr type, Float32 value) // NOLINT
|
|
|
|
{
|
|
|
|
auto insert_func = [&](IColumn & column_, DataTypePtr type_)
|
|
|
|
{
|
|
|
|
insertFloat32(column_, type_, value);
|
|
|
|
};
|
|
|
|
|
|
|
|
if (checkAndInsertNullable(column, type, insert_func) || checkAndInsertLowCardinality(column, type, insert_func))
|
|
|
|
return;
|
|
|
|
|
|
|
|
if (!WhichDataType(type).isFloat32())
|
|
|
|
throw Exception(ErrorCodes::ILLEGAL_COLUMN, "Cannot insert MessagePack float32 into column with type {}.", type->getName());
|
|
|
|
|
|
|
|
assert_cast<ColumnFloat32 &>(column).insertValue(value);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void insertFloat64(IColumn & column, DataTypePtr type, Float64 value) // NOLINT
|
|
|
|
{
|
|
|
|
auto insert_func = [&](IColumn & column_, DataTypePtr type_)
|
|
|
|
{
|
|
|
|
insertFloat64(column_, type_, value);
|
|
|
|
};
|
|
|
|
|
|
|
|
if (checkAndInsertNullable(column, type, insert_func) || checkAndInsertLowCardinality(column, type, insert_func))
|
|
|
|
return;
|
|
|
|
|
|
|
|
if (!WhichDataType(type).isFloat64())
|
|
|
|
throw Exception(ErrorCodes::ILLEGAL_COLUMN, "Cannot insert MessagePack float64 into column with type {}.", type->getName());
|
|
|
|
|
|
|
|
assert_cast<ColumnFloat64 &>(column).insertValue(value);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void insertNull(IColumn & column, DataTypePtr type)
|
|
|
|
{
|
|
|
|
auto insert_func = [&](IColumn & column_, DataTypePtr type_)
|
|
|
|
{
|
|
|
|
insertNull(column_, type_);
|
|
|
|
};
|
|
|
|
|
2021-09-17 14:32:23 +00:00
|
|
|
/// LowCardinality(Nullable(...))
|
2021-09-16 14:05:07 +00:00
|
|
|
if (checkAndInsertLowCardinality(column, type, insert_func))
|
|
|
|
return;
|
|
|
|
|
|
|
|
if (!type->isNullable())
|
|
|
|
throw Exception(ErrorCodes::ILLEGAL_COLUMN, "Cannot insert MessagePack null into non-nullable column with type {}.", type->getName());
|
|
|
|
|
|
|
|
assert_cast<ColumnNullable &>(column).insertDefault();
|
|
|
|
}
|
|
|
|
|
2020-04-18 11:26:57 +00:00
|
|
|
bool MsgPackVisitor::visit_positive_integer(UInt64 value) // NOLINT
|
2020-04-17 09:35:38 +00:00
|
|
|
{
|
2021-09-16 14:05:07 +00:00
|
|
|
insertInteger(info_stack.top().column, info_stack.top().type, value);
|
2020-04-17 09:35:38 +00:00
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2020-04-18 11:26:57 +00:00
|
|
|
bool MsgPackVisitor::visit_negative_integer(Int64 value) // NOLINT
|
2020-04-17 09:35:38 +00:00
|
|
|
{
|
2021-09-16 14:05:07 +00:00
|
|
|
insertInteger(info_stack.top().column, info_stack.top().type, value);
|
2020-04-17 09:35:38 +00:00
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2021-09-16 14:05:07 +00:00
|
|
|
bool MsgPackVisitor::visit_str(const char * value, size_t size) // NOLINT
|
2020-04-17 09:35:38 +00:00
|
|
|
{
|
2021-09-16 14:05:07 +00:00
|
|
|
insertString(info_stack.top().column, info_stack.top().type, value, size);
|
2020-04-17 09:35:38 +00:00
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2021-09-17 14:32:23 +00:00
|
|
|
bool MsgPackVisitor::visit_bin(const char * value, size_t size) // NOLINT
|
2020-04-17 09:35:38 +00:00
|
|
|
{
|
2021-09-16 14:05:07 +00:00
|
|
|
insertString(info_stack.top().column, info_stack.top().type, value, size);
|
|
|
|
return true;
|
|
|
|
}
|
2021-09-16 11:25:45 +00:00
|
|
|
|
2021-09-16 14:05:07 +00:00
|
|
|
bool MsgPackVisitor::visit_float32(Float32 value) // NOLINT
|
|
|
|
{
|
|
|
|
insertFloat32(info_stack.top().column, info_stack.top().type, value);
|
2020-04-17 09:35:38 +00:00
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2020-04-18 11:26:57 +00:00
|
|
|
bool MsgPackVisitor::visit_float64(Float64 value) // NOLINT
|
2020-04-17 09:35:38 +00:00
|
|
|
{
|
2021-09-16 14:05:07 +00:00
|
|
|
insertFloat64(info_stack.top().column, info_stack.top().type, value);
|
2020-04-17 09:35:38 +00:00
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2021-09-17 14:32:23 +00:00
|
|
|
bool MsgPackVisitor::visit_boolean(bool value)
|
|
|
|
{
|
|
|
|
insertInteger(info_stack.top().column, info_stack.top().type, UInt64(value));
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2020-04-18 11:26:57 +00:00
|
|
|
bool MsgPackVisitor::start_array(size_t size) // NOLINT
|
2020-04-17 09:35:38 +00:00
|
|
|
{
|
2021-09-16 11:25:45 +00:00
|
|
|
if (!isArray(info_stack.top().type))
|
|
|
|
throw Exception(ErrorCodes::ILLEGAL_COLUMN, "Cannot insert MessagePack array into column with type {}.", info_stack.top().type->getName());
|
|
|
|
|
2020-04-17 09:35:38 +00:00
|
|
|
auto nested_type = assert_cast<const DataTypeArray &>(*info_stack.top().type).getNestedType();
|
|
|
|
ColumnArray & column_array = assert_cast<ColumnArray &>(info_stack.top().column);
|
|
|
|
ColumnArray::Offsets & offsets = column_array.getOffsets();
|
|
|
|
IColumn & nested_column = column_array.getData();
|
|
|
|
offsets.push_back(offsets.back() + size);
|
|
|
|
info_stack.push(Info{nested_column, nested_type});
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2020-04-18 11:26:57 +00:00
|
|
|
bool MsgPackVisitor::end_array() // NOLINT
|
2020-04-17 09:35:38 +00:00
|
|
|
{
|
|
|
|
info_stack.pop();
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2021-09-16 11:25:45 +00:00
|
|
|
bool MsgPackVisitor::start_map(uint32_t size) // NOLINT
|
|
|
|
{
|
|
|
|
if (!isMap(info_stack.top().type))
|
|
|
|
throw Exception(ErrorCodes::ILLEGAL_COLUMN, "Cannot insert MessagePack map into column with type {}.", info_stack.top().type->getName());
|
|
|
|
ColumnArray & column_array = assert_cast<ColumnMap &>(info_stack.top().column).getNestedColumn();
|
|
|
|
ColumnArray::Offsets & offsets = column_array.getOffsets();
|
|
|
|
offsets.push_back(offsets.back() + size);
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
bool MsgPackVisitor::start_map_key() // NOLINT
|
|
|
|
{
|
|
|
|
auto key_column = assert_cast<ColumnMap &>(info_stack.top().column).getNestedData().getColumns()[0];
|
2021-09-16 14:05:07 +00:00
|
|
|
auto key_type = assert_cast<const DataTypeMap &>(*info_stack.top().type).getKeyType();
|
2021-09-16 11:25:45 +00:00
|
|
|
info_stack.push(Info{*key_column, key_type});
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
bool MsgPackVisitor::end_map_key() // NOLINT
|
|
|
|
{
|
|
|
|
info_stack.pop();
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
bool MsgPackVisitor::start_map_value() // NOLINT
|
|
|
|
{
|
|
|
|
auto value_column = assert_cast<ColumnMap &>(info_stack.top().column).getNestedData().getColumns()[1];
|
2021-09-16 14:05:07 +00:00
|
|
|
auto value_type = assert_cast<const DataTypeMap &>(*info_stack.top().type).getValueType();
|
2021-09-16 11:25:45 +00:00
|
|
|
info_stack.push(Info{*value_column, value_type});
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
bool MsgPackVisitor::end_map_value() // NOLINT
|
|
|
|
{
|
|
|
|
info_stack.pop();
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
bool MsgPackVisitor::visit_nil()
|
|
|
|
{
|
2021-09-16 14:05:07 +00:00
|
|
|
insertNull(info_stack.top().column, info_stack.top().type);
|
|
|
|
return true;
|
2021-09-16 11:25:45 +00:00
|
|
|
}
|
|
|
|
|
2020-04-18 11:26:57 +00:00
|
|
|
void MsgPackVisitor::parse_error(size_t, size_t) // NOLINT
|
2020-04-17 09:35:38 +00:00
|
|
|
{
|
|
|
|
throw Exception("Error occurred while parsing msgpack data.", ErrorCodes::INCORRECT_DATA);
|
|
|
|
}
|
|
|
|
|
|
|
|
bool MsgPackRowInputFormat::readObject()
|
|
|
|
{
|
2021-12-10 17:54:08 +00:00
|
|
|
if (buf->eof())
|
2020-04-17 09:35:38 +00:00
|
|
|
return false;
|
|
|
|
|
2021-12-10 17:54:08 +00:00
|
|
|
PeekableReadBufferCheckpoint checkpoint{*buf};
|
2020-04-17 09:35:38 +00:00
|
|
|
size_t offset = 0;
|
2021-12-10 17:54:08 +00:00
|
|
|
while (!parser.execute(buf->position(), buf->available(), offset))
|
2020-04-17 09:35:38 +00:00
|
|
|
{
|
2021-12-10 17:54:08 +00:00
|
|
|
buf->position() = buf->buffer().end();
|
|
|
|
if (buf->eof())
|
2020-04-17 09:35:38 +00:00
|
|
|
throw Exception("Unexpected end of file while parsing msgpack object.", ErrorCodes::INCORRECT_DATA);
|
2021-12-10 17:54:08 +00:00
|
|
|
buf->position() = buf->buffer().end();
|
|
|
|
buf->makeContinuousMemoryFromCheckpointToPos();
|
|
|
|
buf->rollbackToCheckpoint();
|
2020-04-17 09:35:38 +00:00
|
|
|
}
|
2021-12-10 17:54:08 +00:00
|
|
|
buf->position() += offset;
|
2020-04-17 09:35:38 +00:00
|
|
|
return true;
|
2020-03-26 16:33:00 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
bool MsgPackRowInputFormat::readRow(MutableColumns & columns, RowReadExtension &)
|
|
|
|
{
|
|
|
|
size_t column_index = 0;
|
|
|
|
bool has_more_data = true;
|
|
|
|
for (; column_index != columns.size(); ++column_index)
|
|
|
|
{
|
2020-04-17 09:35:38 +00:00
|
|
|
visitor.set_info(*columns[column_index], data_types[column_index]);
|
2020-03-26 16:33:00 +00:00
|
|
|
has_more_data = readObject();
|
|
|
|
if (!has_more_data)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
if (!has_more_data)
|
|
|
|
{
|
|
|
|
if (column_index != 0)
|
|
|
|
throw Exception("Not enough values to complete the row.", ErrorCodes::INCORRECT_DATA);
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2021-12-10 17:54:08 +00:00
|
|
|
void MsgPackRowInputFormat::setReadBuffer(ReadBuffer & in_)
|
|
|
|
{
|
|
|
|
buf = std::make_unique<PeekableReadBuffer>(in_);
|
2021-12-15 11:30:57 +00:00
|
|
|
IInputFormat::setReadBuffer(in_);
|
|
|
|
}
|
|
|
|
|
2021-12-28 06:40:36 +00:00
|
|
|
MsgPackSchemaReader::MsgPackSchemaReader(ReadBuffer & in_, const FormatSettings & format_settings_)
|
|
|
|
: IRowSchemaReader(buf, format_settings_.max_rows_to_read_for_schema_inference), buf(in_), number_of_columns(format_settings_.msgpack.number_of_columns)
|
2021-12-15 11:30:57 +00:00
|
|
|
{
|
|
|
|
if (!number_of_columns)
|
|
|
|
throw Exception(ErrorCodes::BAD_ARGUMENTS, "You must specify setting input_format_msgpack_number_of_columns to extract table schema from MsgPack data");
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
msgpack::object_handle MsgPackSchemaReader::readObject()
|
|
|
|
{
|
|
|
|
if (buf.eof())
|
|
|
|
throw Exception(ErrorCodes::UNEXPECTED_END_OF_FILE, "Unexpected eof while parsing msgpack object");
|
|
|
|
|
|
|
|
PeekableReadBufferCheckpoint checkpoint{buf};
|
|
|
|
size_t offset = 0;
|
|
|
|
bool need_more_data = true;
|
|
|
|
msgpack::object_handle object_handle;
|
|
|
|
while (need_more_data)
|
|
|
|
{
|
|
|
|
offset = 0;
|
|
|
|
try
|
|
|
|
{
|
|
|
|
object_handle = msgpack::unpack(buf.position(), buf.buffer().end() - buf.position(), offset);
|
|
|
|
need_more_data = false;
|
|
|
|
}
|
|
|
|
catch (msgpack::insufficient_bytes &)
|
|
|
|
{
|
|
|
|
buf.position() = buf.buffer().end();
|
|
|
|
if (buf.eof())
|
|
|
|
throw Exception("Unexpected end of file while parsing msgpack object", ErrorCodes::UNEXPECTED_END_OF_FILE);
|
|
|
|
buf.position() = buf.buffer().end();
|
|
|
|
buf.makeContinuousMemoryFromCheckpointToPos();
|
|
|
|
buf.rollbackToCheckpoint();
|
|
|
|
}
|
|
|
|
}
|
|
|
|
buf.position() += offset;
|
|
|
|
return object_handle;
|
|
|
|
}
|
|
|
|
|
|
|
|
DataTypePtr MsgPackSchemaReader::getDataType(const msgpack::object & object)
|
|
|
|
{
|
|
|
|
switch (object.type)
|
|
|
|
{
|
2021-12-17 15:34:13 +00:00
|
|
|
case msgpack::type::object_type::POSITIVE_INTEGER: [[fallthrough]];
|
2021-12-15 11:30:57 +00:00
|
|
|
case msgpack::type::object_type::NEGATIVE_INTEGER:
|
|
|
|
return makeNullable(std::make_shared<DataTypeInt64>());
|
|
|
|
case msgpack::type::object_type::FLOAT32:
|
|
|
|
return makeNullable(std::make_shared<DataTypeFloat32>());
|
|
|
|
case msgpack::type::object_type::FLOAT64:
|
|
|
|
return makeNullable(std::make_shared<DataTypeFloat64>());
|
|
|
|
case msgpack::type::object_type::BOOLEAN:
|
|
|
|
return makeNullable(std::make_shared<DataTypeUInt8>());
|
|
|
|
case msgpack::type::object_type::BIN: [[fallthrough]];
|
|
|
|
case msgpack::type::object_type::STR:
|
|
|
|
return makeNullable(std::make_shared<DataTypeString>());
|
|
|
|
case msgpack::type::object_type::ARRAY:
|
|
|
|
{
|
|
|
|
msgpack::object_array object_array = object.via.array;
|
|
|
|
if (object_array.size)
|
|
|
|
{
|
|
|
|
auto nested_type = getDataType(object_array.ptr[0]);
|
|
|
|
if (nested_type)
|
|
|
|
return std::make_shared<DataTypeArray>(getDataType(object_array.ptr[0]));
|
|
|
|
}
|
|
|
|
return nullptr;
|
|
|
|
}
|
|
|
|
case msgpack::type::object_type::MAP:
|
|
|
|
{
|
|
|
|
msgpack::object_map object_map = object.via.map;
|
|
|
|
if (object_map.size)
|
|
|
|
{
|
2021-12-17 15:34:13 +00:00
|
|
|
auto key_type = removeNullable(getDataType(object_map.ptr[0].key));
|
|
|
|
auto value_type = getDataType(object_map.ptr[0].val);
|
2021-12-15 11:30:57 +00:00
|
|
|
if (key_type && value_type)
|
|
|
|
return std::make_shared<DataTypeMap>(key_type, value_type);
|
|
|
|
}
|
|
|
|
return nullptr;
|
|
|
|
}
|
|
|
|
case msgpack::type::object_type::NIL:
|
|
|
|
return nullptr;
|
|
|
|
default:
|
|
|
|
throw Exception(ErrorCodes::BAD_ARGUMENTS, "Msgpack type is not supported");
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
DataTypes MsgPackSchemaReader::readRowAndGetDataTypes()
|
|
|
|
{
|
|
|
|
if (buf.eof())
|
|
|
|
return {};
|
|
|
|
|
|
|
|
DataTypes data_types;
|
|
|
|
data_types.reserve(number_of_columns);
|
|
|
|
for (size_t i = 0; i != number_of_columns; ++i)
|
|
|
|
{
|
|
|
|
auto object_handle = readObject();
|
|
|
|
data_types.push_back(getDataType(object_handle.get()));
|
|
|
|
}
|
|
|
|
|
|
|
|
return data_types;
|
2021-12-10 17:54:08 +00:00
|
|
|
}
|
|
|
|
|
2021-10-11 16:11:50 +00:00
|
|
|
void registerInputFormatMsgPack(FormatFactory & factory)
|
2020-03-26 21:11:33 +00:00
|
|
|
{
|
2021-10-11 16:11:50 +00:00
|
|
|
factory.registerInputFormat("MsgPack", [](
|
2020-04-03 20:44:13 +00:00
|
|
|
ReadBuffer & buf,
|
|
|
|
const Block & sample,
|
|
|
|
const RowInputFormatParams & params,
|
2020-03-26 21:11:33 +00:00
|
|
|
const FormatSettings &)
|
|
|
|
{
|
2020-03-26 16:33:00 +00:00
|
|
|
return std::make_shared<MsgPackRowInputFormat>(sample, buf, params);
|
|
|
|
});
|
2022-01-12 15:28:13 +00:00
|
|
|
factory.registerFileExtension("messagepack", "MsgPack");
|
2020-03-26 16:33:00 +00:00
|
|
|
}
|
|
|
|
|
2021-12-15 11:30:57 +00:00
|
|
|
void registerMsgPackSchemaReader(FormatFactory & factory)
|
|
|
|
{
|
|
|
|
factory.registerSchemaReader("MsgPack", [](ReadBuffer & buf, const FormatSettings & settings, ContextPtr)
|
|
|
|
{
|
|
|
|
return std::make_shared<MsgPackSchemaReader>(buf, settings);
|
|
|
|
});
|
|
|
|
}
|
|
|
|
|
2020-03-26 21:11:33 +00:00
|
|
|
}
|
2020-07-10 19:08:18 +00:00
|
|
|
|
2020-07-10 19:27:12 +00:00
|
|
|
#else
|
|
|
|
|
|
|
|
namespace DB
|
|
|
|
{
|
|
|
|
class FormatFactory;
|
2021-10-11 16:11:50 +00:00
|
|
|
void registerInputFormatMsgPack(FormatFactory &)
|
2020-07-10 19:27:12 +00:00
|
|
|
{
|
|
|
|
}
|
2021-12-15 11:30:57 +00:00
|
|
|
|
|
|
|
void registerMsgPackSchemaReader(FormatFactory &)
|
|
|
|
{
|
|
|
|
}
|
2020-07-10 19:27:12 +00:00
|
|
|
}
|
|
|
|
|
2020-07-10 19:08:18 +00:00
|
|
|
#endif
|