#pragma once #include #include #ifdef USE_QUICKLZ #include #endif #include #include #include #include #include #include #include #include #include #include namespace ProfileEvents { extern const Event ReadCompressedBytes; extern const Event CompressedReadBufferBlocks; extern const Event CompressedReadBufferBytes; } namespace DB { namespace ErrorCodes { extern const int UNKNOWN_COMPRESSION_METHOD; extern const int TOO_LARGE_SIZE_COMPRESSED; extern const int CHECKSUM_DOESNT_MATCH; extern const int CANNOT_DECOMPRESS; } /** Basic functionality for implementation of * CompressedReadBuffer, CompressedReadBufferFromFile and CachedCompressedReadBuffer. */ class CompressedReadBufferBase { protected: ReadBuffer * compressed_in; /// If 'compressed_in' buffer has whole compressed block - than use it. Otherwise copy parts of data to 'own_compressed_buffer'. PODArray own_compressed_buffer{COMPRESSED_BLOCK_HEADER_SIZE}; /// Points to memory, holding compressed block. char * compressed_buffer = nullptr; #ifdef USE_QUICKLZ std::unique_ptr qlz_state; #else void * fixed_size_padding = nullptr; /// ABI compatibility for USE_QUICKLZ #endif /// Don't checksum on decompressing. bool disable_checksum = false; /// Read compressed data into compressed_buffer. Get size of decompressed data from block header. Checksum if need. /// Returns number of compressed bytes read. size_t readCompressedData(size_t & size_decompressed, size_t & size_compressed_without_checksum) { if (compressed_in->eof()) return 0; uint128 checksum; compressed_in->readStrict(reinterpret_cast(&checksum), sizeof(checksum)); own_compressed_buffer.resize(COMPRESSED_BLOCK_HEADER_SIZE); compressed_in->readStrict(&own_compressed_buffer[0], COMPRESSED_BLOCK_HEADER_SIZE); UInt8 method = own_compressed_buffer[0]; /// See CompressedWriteBuffer.h size_t & size_compressed = size_compressed_without_checksum; if (method < 0x80) { #ifdef USE_QUICKLZ size_compressed = qlz_size_compressed(&own_compressed_buffer[0]); size_decompressed = qlz_size_decompressed(&own_compressed_buffer[0]); #else throw Exception("QuickLZ compression method is disabled", ErrorCodes::UNKNOWN_COMPRESSION_METHOD); #endif } else if (method == static_cast(CompressionMethodByte::LZ4) || method == static_cast(CompressionMethodByte::ZSTD)) { size_compressed = unalignedLoad(&own_compressed_buffer[1]); size_decompressed = unalignedLoad(&own_compressed_buffer[5]); } else throw Exception("Unknown compression method: " + toString(method), ErrorCodes::UNKNOWN_COMPRESSION_METHOD); if (size_compressed > DBMS_MAX_COMPRESSED_SIZE) throw Exception("Too large size_compressed. Most likely corrupted data.", ErrorCodes::TOO_LARGE_SIZE_COMPRESSED); ProfileEvents::increment(ProfileEvents::ReadCompressedBytes, size_compressed + sizeof(checksum)); /// Is whole compressed block located in 'compressed_in' buffer? if (compressed_in->offset() >= COMPRESSED_BLOCK_HEADER_SIZE && compressed_in->position() + size_compressed - COMPRESSED_BLOCK_HEADER_SIZE <= compressed_in->buffer().end()) { compressed_in->position() -= COMPRESSED_BLOCK_HEADER_SIZE; compressed_buffer = compressed_in->position(); compressed_in->position() += size_compressed; } else { own_compressed_buffer.resize(size_compressed); compressed_buffer = &own_compressed_buffer[0]; compressed_in->readStrict(&compressed_buffer[COMPRESSED_BLOCK_HEADER_SIZE], size_compressed - COMPRESSED_BLOCK_HEADER_SIZE); } if (!disable_checksum && checksum != CityHash128(&compressed_buffer[0], size_compressed)) throw Exception("Checksum doesn't match: corrupted data.", ErrorCodes::CHECKSUM_DOESNT_MATCH); return size_compressed + sizeof(checksum); } void decompress(char * to, size_t size_decompressed, size_t size_compressed_without_checksum) { ProfileEvents::increment(ProfileEvents::CompressedReadBufferBlocks); ProfileEvents::increment(ProfileEvents::CompressedReadBufferBytes, size_decompressed); UInt8 method = compressed_buffer[0]; /// See CompressedWriteBuffer.h if (method < 0x80) { #ifdef USE_QUICKLZ if (!qlz_state) qlz_state = std::make_unique(); qlz_decompress(&compressed_buffer[0], to, qlz_state.get()); #else throw Exception("QuickLZ compression method is disabled", ErrorCodes::UNKNOWN_COMPRESSION_METHOD); #endif } else if (method == static_cast(CompressionMethodByte::LZ4)) { if (LZ4_decompress_fast(&compressed_buffer[COMPRESSED_BLOCK_HEADER_SIZE], to, size_decompressed) < 0) throw Exception("Cannot LZ4_decompress_fast", ErrorCodes::CANNOT_DECOMPRESS); } else if (method == static_cast(CompressionMethodByte::ZSTD)) { size_t res = ZSTD_decompress( to, size_decompressed, &compressed_buffer[COMPRESSED_BLOCK_HEADER_SIZE], size_compressed_without_checksum - COMPRESSED_BLOCK_HEADER_SIZE); if (ZSTD_isError(res)) throw Exception("Cannot ZSTD_decompress: " + std::string(ZSTD_getErrorName(res)), ErrorCodes::CANNOT_DECOMPRESS); } else throw Exception("Unknown compression method: " + toString(method), ErrorCodes::UNKNOWN_COMPRESSION_METHOD); } public: /// 'compressed_in' could be initialized lazily, but before first call of 'readCompressedData'. CompressedReadBufferBase(ReadBuffer * in = nullptr) : compressed_in(in) { } /** Disable checksums. * For example, may be used when * compressed data is generated by client, that cannot calculate checksums, and fill checksums with zeros instead. */ void disableChecksumming() { disable_checksum = true; } }; }