Mercurial > hg > orthanc
view OrthancFramework/Sources/Compression/GzipCompressor.cpp @ 5800:2854418a7c0d
fix unit tests
author | Alain Mazy <am@orthanc.team> |
---|---|
date | Fri, 20 Sep 2024 16:05:04 +0200 |
parents | f7adfb22e20e |
children |
line wrap: on
line source
/** * Orthanc - A Lightweight, RESTful DICOM Store * Copyright (C) 2012-2016 Sebastien Jodogne, Medical Physics * Department, University Hospital of Liege, Belgium * Copyright (C) 2017-2023 Osimis S.A., Belgium * Copyright (C) 2024-2024 Orthanc Team SRL, Belgium * Copyright (C) 2021-2024 Sebastien Jodogne, ICTEAM UCLouvain, Belgium * * This program is free software: you can redistribute it and/or * modify it under the terms of the GNU Lesser General Public License * as published by the Free Software Foundation, either version 3 of * the License, or (at your option) any later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU * Lesser General Public License for more details. * * You should have received a copy of the GNU Lesser General Public * License along with this program. If not, see * <http://www.gnu.org/licenses/>. **/ #include "../PrecompiledHeaders.h" #include "GzipCompressor.h" #include <stdio.h> #include <string.h> #include <zlib.h> #include "../OrthancException.h" #include "../Logging.h" namespace Orthanc { uint64_t GzipCompressor::GuessUncompressedSize(const void* compressed, size_t compressedSize) { /** * "Is there a way to find out the size of the original file which * is inside a GZIP file? [...] There is no truly reliable way, * other than gunzipping the stream. You do not need to save the * result of the decompression, so you can determine the size by * simply reading and decoding the entire file without taking up * space with the decompressed result. * * There is an unreliable way to determine the uncompressed size, * which is to look at the last four bytes of the gzip file, which * is the uncompressed length of that entry modulo 232 in little * endian order. * * It is unreliable because a) the uncompressed data may be longer * than 2^32 bytes, and b) the gzip file may consist of multiple * gzip streams, in which case you would find the length of only * the last of those streams. * * If you are in control of the source of the gzip files, you know * that they consist of single gzip streams, and you know that * they are less than 2^32 bytes uncompressed, then and only then * can you use those last four bytes with confidence." * * http://stackoverflow.com/a/9727599/881731 **/ if (compressedSize < 4) { throw OrthancException(ErrorCode_BadFileFormat); } const uint8_t* p = reinterpret_cast<const uint8_t*>(compressed) + compressedSize - 4; return ((static_cast<uint32_t>(p[0]) << 0) + (static_cast<uint32_t>(p[1]) << 8) + (static_cast<uint32_t>(p[2]) << 16) + (static_cast<uint32_t>(p[3]) << 24)); } GzipCompressor::GzipCompressor() { SetPrefixWithUncompressedSize(false); } void GzipCompressor::Compress(std::string& compressed, const void* uncompressed, size_t uncompressedSize) { uLongf compressedSize = compressBound(static_cast<uLong>(uncompressedSize)) + 1024 /* security margin */; if (compressedSize == 0) { compressedSize = 1; } uint8_t* target; if (HasPrefixWithUncompressedSize()) { compressed.resize(compressedSize + sizeof(uint64_t)); target = reinterpret_cast<uint8_t*>(&compressed[0]) + sizeof(uint64_t); } else { compressed.resize(compressedSize); target = reinterpret_cast<uint8_t*>(&compressed[0]); } z_stream stream; memset(&stream, 0, sizeof(stream)); stream.next_in = const_cast<Bytef*>(reinterpret_cast<const Bytef*>(uncompressed)); stream.next_out = reinterpret_cast<Bytef*>(target); stream.avail_in = static_cast<uInt>(uncompressedSize); stream.avail_out = static_cast<uInt>(compressedSize); // Ensure no overflow (if the buffer is too large for the current archicture) if (static_cast<size_t>(stream.avail_in) != uncompressedSize || static_cast<size_t>(stream.avail_out) != compressedSize) { throw OrthancException(ErrorCode_NotEnoughMemory); } // Initialize the compression engine int error = deflateInit2(&stream, GetCompressionLevel(), Z_DEFLATED, MAX_WBITS + 16, // ask for gzip output 8, // default memory level Z_DEFAULT_STRATEGY); if (error != Z_OK) { // Cannot initialize zlib compressed.clear(); throw OrthancException(ErrorCode_InternalError); } // Compress the input buffer error = deflate(&stream, Z_FINISH); if (error != Z_STREAM_END) { deflateEnd(&stream); compressed.clear(); switch (error) { case Z_MEM_ERROR: throw OrthancException(ErrorCode_NotEnoughMemory); default: throw OrthancException(ErrorCode_InternalError); } } size_t size = stream.total_out; if (deflateEnd(&stream) != Z_OK) { throw OrthancException(ErrorCode_InternalError); } // The compression was successful if (HasPrefixWithUncompressedSize()) { uint64_t s = static_cast<uint64_t>(uncompressedSize); memcpy(&compressed[0], &s, sizeof(uint64_t)); compressed.resize(size + sizeof(uint64_t)); } else { compressed.resize(size); } } void GzipCompressor::Uncompress(std::string& uncompressed, const void* compressed, size_t compressedSize) { uint64_t uncompressedSize; const uint8_t* source = reinterpret_cast<const uint8_t*>(compressed); if (HasPrefixWithUncompressedSize()) { uncompressedSize = ReadUncompressedSizePrefix(compressed, compressedSize); source += sizeof(uint64_t); compressedSize -= sizeof(uint64_t); } else { uncompressedSize = GuessUncompressedSize(compressed, compressedSize); } try { uncompressed.resize(static_cast<size_t>(uncompressedSize)); } catch (...) { throw OrthancException(ErrorCode_NotEnoughMemory); } z_stream stream; memset(&stream, 0, sizeof(stream)); char dummy = '\0'; // zlib does not like NULL output buffers (even if the uncompressed data is empty) stream.next_in = const_cast<Bytef*>(source); stream.next_out = reinterpret_cast<Bytef*>(uncompressedSize == 0 ? &dummy : &uncompressed[0]); stream.avail_in = static_cast<uInt>(compressedSize); stream.avail_out = static_cast<uInt>(uncompressedSize); // Ensure no overflow (if the buffer is too large for the current archicture) if (static_cast<size_t>(stream.avail_in) != compressedSize || static_cast<size_t>(stream.avail_out) != uncompressedSize) { throw OrthancException(ErrorCode_NotEnoughMemory); } // Initialize the compression engine int error = inflateInit2(&stream, MAX_WBITS + 16); // this is a gzip input if (error != Z_OK) { // Cannot initialize zlib uncompressed.clear(); throw OrthancException(ErrorCode_InternalError); } // Uncompress the input buffer error = inflate(&stream, Z_FINISH); if (error != Z_STREAM_END) { inflateEnd(&stream); uncompressed.clear(); switch (error) { case Z_MEM_ERROR: throw OrthancException(ErrorCode_NotEnoughMemory); case Z_BUF_ERROR: case Z_NEED_DICT: throw OrthancException(ErrorCode_BadFileFormat); default: throw OrthancException(ErrorCode_InternalError); } } size_t size = stream.total_out; if (inflateEnd(&stream) != Z_OK) { uncompressed.clear(); throw OrthancException(ErrorCode_InternalError); } if (size != uncompressedSize) { uncompressed.clear(); // The uncompressed size was not that properly guess, presumably // because of a file size over 4GB. Should fallback to // stream-based decompression. throw OrthancException(ErrorCode_NotImplemented, "The uncompressed size of a gzip-encoded buffer was not properly guessed"); } } }