2016-10-25 06:49:24 +00:00
|
|
|
#include <city.h>
|
2022-05-08 17:01:47 +00:00
|
|
|
#include <cstring>
|
2016-10-25 06:49:24 +00:00
|
|
|
|
2021-10-02 07:13:14 +00:00
|
|
|
#include <base/types.h>
|
2022-05-16 21:21:20 +00:00
|
|
|
#include <base/unaligned.h>
|
2022-10-07 10:46:45 +00:00
|
|
|
#include <base/defines.h>
|
2016-10-25 06:49:24 +00:00
|
|
|
|
2018-12-20 17:37:02 +00:00
|
|
|
#include <Compression/CompressionFactory.h>
|
2022-05-16 21:21:20 +00:00
|
|
|
#include "CompressedWriteBuffer.h"
|
2020-03-26 12:35:31 +00:00
|
|
|
|
2016-10-25 06:49:24 +00:00
|
|
|
|
|
|
|
namespace DB
|
|
|
|
{
|
|
|
|
|
|
|
|
namespace ErrorCodes
|
|
|
|
{
|
|
|
|
}
|
|
|
|
|
2018-12-20 17:37:02 +00:00
|
|
|
static constexpr auto CHECKSUM_SIZE{sizeof(CityHash_v1_0_2::uint128)};
|
2016-10-25 06:49:24 +00:00
|
|
|
|
|
|
|
void CompressedWriteBuffer::nextImpl()
|
|
|
|
{
|
|
|
|
if (!offset())
|
|
|
|
return;
|
2017-04-01 07:20:54 +00:00
|
|
|
|
2022-10-07 10:46:45 +00:00
|
|
|
chassert(offset() <= INT_MAX);
|
|
|
|
UInt32 decompressed_size = static_cast<UInt32>(offset());
|
2018-12-20 17:37:02 +00:00
|
|
|
UInt32 compressed_reserve_size = codec->getCompressedReserveSize(decompressed_size);
|
2017-04-01 07:20:54 +00:00
|
|
|
|
2022-05-23 10:09:44 +00:00
|
|
|
/** During compression we need buffer with capacity >= compressed_reserve_size + CHECKSUM_SIZE.
|
|
|
|
*
|
|
|
|
* If output buffer has necessary capacity, we can compress data directly in output buffer.
|
|
|
|
* Then we can write checksum at the output buffer begin.
|
|
|
|
*
|
|
|
|
* If output buffer does not have necessary capacity. Compress data in temporary buffer.
|
|
|
|
* Then we can write checksum and temporary buffer in output buffer.
|
|
|
|
*/
|
|
|
|
if (out.available() >= compressed_reserve_size + CHECKSUM_SIZE)
|
2022-05-16 20:22:13 +00:00
|
|
|
{
|
2022-05-16 21:21:20 +00:00
|
|
|
char * out_checksum_ptr = out.position();
|
|
|
|
char * out_compressed_ptr = out.position() + CHECKSUM_SIZE;
|
2022-05-23 10:09:44 +00:00
|
|
|
UInt32 compressed_size = codec->compress(working_buffer.begin(), decompressed_size, out_compressed_ptr);
|
2022-05-16 20:22:13 +00:00
|
|
|
|
|
|
|
CityHash_v1_0_2::uint128 checksum = CityHash_v1_0_2::CityHash128(out_compressed_ptr, compressed_size);
|
2022-05-16 18:44:22 +00:00
|
|
|
memcpy(out_checksum_ptr, reinterpret_cast<const char *>(&checksum), CHECKSUM_SIZE);
|
2022-05-16 20:22:13 +00:00
|
|
|
out.position() += CHECKSUM_SIZE + compressed_size;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
compressed_buffer.resize(compressed_reserve_size);
|
2022-05-23 10:09:44 +00:00
|
|
|
UInt32 compressed_size = codec->compress(working_buffer.begin(), decompressed_size, compressed_buffer.data());
|
2022-05-16 20:22:13 +00:00
|
|
|
|
|
|
|
CityHash_v1_0_2::uint128 checksum = CityHash_v1_0_2::CityHash128(compressed_buffer.data(), compressed_size);
|
|
|
|
out.write(reinterpret_cast<const char *>(&checksum), CHECKSUM_SIZE);
|
|
|
|
out.write(compressed_buffer.data(), compressed_size);
|
|
|
|
}
|
2018-12-20 17:37:02 +00:00
|
|
|
}
|
2016-10-25 06:49:24 +00:00
|
|
|
|
2021-11-10 22:58:56 +00:00
|
|
|
CompressedWriteBuffer::~CompressedWriteBuffer()
|
2021-08-15 08:44:12 +00:00
|
|
|
{
|
2021-11-11 17:27:23 +00:00
|
|
|
finalize();
|
2021-08-15 08:44:12 +00:00
|
|
|
}
|
|
|
|
|
2022-05-16 21:21:20 +00:00
|
|
|
CompressedWriteBuffer::CompressedWriteBuffer(WriteBuffer & out_, CompressionCodecPtr codec_, size_t buf_size)
|
2019-01-21 18:04:08 +00:00
|
|
|
: BufferWithOwnMemory<WriteBuffer>(buf_size), out(out_), codec(std::move(codec_))
|
2018-12-20 17:37:02 +00:00
|
|
|
{
|
2016-10-25 06:49:24 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
}
|