Mercurial > hg > orthanc
annotate OrthancFramework/Sources/Compression/GzipCompressor.cpp @ 4268:0ae2ca210077
new macro TLOG() to replace VLOG() for trace logs with a category
| author | Sebastien Jodogne <s.jodogne@gmail.com> |
|---|---|
| date | Mon, 02 Nov 2020 14:48:15 +0100 |
| parents | bf7b9edf6b81 |
| children | 3b70a2e6a06c |
| rev | line source |
|---|---|
| 1513 | 1 /** |
| 2 * Orthanc - A Lightweight, RESTful DICOM Store | |
| 1900 | 3 * Copyright (C) 2012-2016 Sebastien Jodogne, Medical Physics |
| 1513 | 4 * Department, University Hospital of Liege, Belgium |
|
3640
94f4a18a79cc
upgrade to year 2020
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
3378
diff
changeset
|
5 * Copyright (C) 2017-2020 Osimis S.A., Belgium |
| 1513 | 6 * |
| 7 * This program is free software: you can redistribute it and/or | |
|
4119
bf7b9edf6b81
re-licensing the OrthancFramework to LGPL, in order to license Stone of Orthanc under LGPL
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
4044
diff
changeset
|
8 * modify it under the terms of the GNU Lesser General Public License |
|
bf7b9edf6b81
re-licensing the OrthancFramework to LGPL, in order to license Stone of Orthanc under LGPL
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
4044
diff
changeset
|
9 * as published by the Free Software Foundation, either version 3 of |
|
bf7b9edf6b81
re-licensing the OrthancFramework to LGPL, in order to license Stone of Orthanc under LGPL
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
4044
diff
changeset
|
10 * the License, or (at your option) any later version. |
| 1513 | 11 * |
| 12 * This program is distributed in the hope that it will be useful, but | |
| 13 * WITHOUT ANY WARRANTY; without even the implied warranty of | |
| 14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
|
4119
bf7b9edf6b81
re-licensing the OrthancFramework to LGPL, in order to license Stone of Orthanc under LGPL
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
4044
diff
changeset
|
15 * Lesser General Public License for more details. |
| 1513 | 16 * |
|
4119
bf7b9edf6b81
re-licensing the OrthancFramework to LGPL, in order to license Stone of Orthanc under LGPL
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
4044
diff
changeset
|
17 * You should have received a copy of the GNU Lesser General Public |
|
bf7b9edf6b81
re-licensing the OrthancFramework to LGPL, in order to license Stone of Orthanc under LGPL
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
4044
diff
changeset
|
18 * License along with this program. If not, see |
|
bf7b9edf6b81
re-licensing the OrthancFramework to LGPL, in order to license Stone of Orthanc under LGPL
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
4044
diff
changeset
|
19 * <http://www.gnu.org/licenses/>. |
| 1513 | 20 **/ |
| 21 | |
| 22 | |
| 23 #include "../PrecompiledHeaders.h" | |
| 24 #include "GzipCompressor.h" | |
| 25 | |
| 26 #include <stdio.h> | |
| 27 #include <string.h> | |
| 28 #include <zlib.h> | |
| 29 | |
| 30 #include "../OrthancException.h" | |
| 31 #include "../Logging.h" | |
| 32 | |
| 33 namespace Orthanc | |
| 34 { | |
| 35 uint64_t GzipCompressor::GuessUncompressedSize(const void* compressed, | |
| 36 size_t compressedSize) | |
| 37 { | |
| 38 /** | |
| 39 * "Is there a way to find out the size of the original file which | |
| 40 * is inside a GZIP file? [...] There is no truly reliable way, | |
| 41 * other than gunzipping the stream. You do not need to save the | |
| 42 * result of the decompression, so you can determine the size by | |
| 43 * simply reading and decoding the entire file without taking up | |
| 44 * space with the decompressed result. | |
| 45 * | |
| 46 * There is an unreliable way to determine the uncompressed size, | |
| 47 * which is to look at the last four bytes of the gzip file, which | |
| 48 * is the uncompressed length of that entry modulo 232 in little | |
| 49 * endian order. | |
| 50 * | |
| 51 * It is unreliable because a) the uncompressed data may be longer | |
| 52 * than 2^32 bytes, and b) the gzip file may consist of multiple | |
| 53 * gzip streams, in which case you would find the length of only | |
| 54 * the last of those streams. | |
| 55 * | |
| 56 * If you are in control of the source of the gzip files, you know | |
| 57 * that they consist of single gzip streams, and you know that | |
| 58 * they are less than 2^32 bytes uncompressed, then and only then | |
| 59 * can you use those last four bytes with confidence." | |
| 60 * | |
| 61 * http://stackoverflow.com/a/9727599/881731 | |
| 62 **/ | |
| 63 | |
| 64 if (compressedSize < 4) | |
| 65 { | |
| 66 throw OrthancException(ErrorCode_BadFileFormat); | |
| 67 } | |
| 68 | |
| 69 const uint8_t* p = reinterpret_cast<const uint8_t*>(compressed) + compressedSize - 4; | |
| 70 | |
| 71 return ((static_cast<uint32_t>(p[0]) << 0) + | |
| 72 (static_cast<uint32_t>(p[1]) << 8) + | |
| 73 (static_cast<uint32_t>(p[2]) << 16) + | |
| 74 (static_cast<uint32_t>(p[3]) << 24)); | |
| 75 } | |
| 76 | |
| 77 | |
| 78 | |
| 79 void GzipCompressor::Compress(std::string& compressed, | |
| 80 const void* uncompressed, | |
| 81 size_t uncompressedSize) | |
| 82 { | |
|
3378
596cfabd72c5
Fixed a couple of truncation warnings
Benjamin Golinvaux <bgo@osimis.io>
parents:
3060
diff
changeset
|
83 uLongf compressedSize = compressBound(static_cast<uLong>(uncompressedSize)) |
|
596cfabd72c5
Fixed a couple of truncation warnings
Benjamin Golinvaux <bgo@osimis.io>
parents:
3060
diff
changeset
|
84 + 1024 /* security margin */; |
|
596cfabd72c5
Fixed a couple of truncation warnings
Benjamin Golinvaux <bgo@osimis.io>
parents:
3060
diff
changeset
|
85 |
| 1513 | 86 if (compressedSize == 0) |
| 87 { | |
| 88 compressedSize = 1; | |
| 89 } | |
| 90 | |
| 91 uint8_t* target; | |
| 92 if (HasPrefixWithUncompressedSize()) | |
| 93 { | |
| 94 compressed.resize(compressedSize + sizeof(uint64_t)); | |
| 95 target = reinterpret_cast<uint8_t*>(&compressed[0]) + sizeof(uint64_t); | |
| 96 } | |
| 97 else | |
| 98 { | |
| 99 compressed.resize(compressedSize); | |
| 100 target = reinterpret_cast<uint8_t*>(&compressed[0]); | |
| 101 } | |
| 102 | |
| 103 z_stream stream; | |
| 104 memset(&stream, 0, sizeof(stream)); | |
| 105 | |
| 106 stream.next_in = const_cast<Bytef*>(reinterpret_cast<const Bytef*>(uncompressed)); | |
| 107 stream.next_out = reinterpret_cast<Bytef*>(target); | |
| 108 | |
| 109 stream.avail_in = static_cast<uInt>(uncompressedSize); | |
| 110 stream.avail_out = static_cast<uInt>(compressedSize); | |
| 111 | |
| 112 // Ensure no overflow (if the buffer is too large for the current archicture) | |
| 113 if (static_cast<size_t>(stream.avail_in) != uncompressedSize || | |
| 114 static_cast<size_t>(stream.avail_out) != compressedSize) | |
| 115 { | |
| 116 throw OrthancException(ErrorCode_NotEnoughMemory); | |
| 117 } | |
| 118 | |
| 119 // Initialize the compression engine | |
| 120 int error = deflateInit2(&stream, | |
| 121 GetCompressionLevel(), | |
| 122 Z_DEFLATED, | |
| 123 MAX_WBITS + 16, // ask for gzip output | |
| 124 8, // default memory level | |
| 125 Z_DEFAULT_STRATEGY); | |
| 126 | |
| 127 if (error != Z_OK) | |
| 128 { | |
| 129 // Cannot initialize zlib | |
| 130 compressed.clear(); | |
| 131 throw OrthancException(ErrorCode_InternalError); | |
| 132 } | |
| 133 | |
| 134 // Compress the input buffer | |
| 135 error = deflate(&stream, Z_FINISH); | |
| 136 | |
| 137 if (error != Z_STREAM_END) | |
| 138 { | |
| 139 deflateEnd(&stream); | |
| 140 compressed.clear(); | |
| 141 | |
| 142 switch (error) | |
| 143 { | |
| 144 case Z_MEM_ERROR: | |
| 145 throw OrthancException(ErrorCode_NotEnoughMemory); | |
| 146 | |
| 147 default: | |
| 148 throw OrthancException(ErrorCode_InternalError); | |
| 149 } | |
| 150 } | |
| 151 | |
| 152 size_t size = stream.total_out; | |
| 153 | |
| 154 if (deflateEnd(&stream) != Z_OK) | |
| 155 { | |
| 156 throw OrthancException(ErrorCode_InternalError); | |
| 157 } | |
| 158 | |
| 159 // The compression was successful | |
| 160 if (HasPrefixWithUncompressedSize()) | |
| 161 { | |
| 162 uint64_t s = static_cast<uint64_t>(uncompressedSize); | |
| 163 memcpy(&compressed[0], &s, sizeof(uint64_t)); | |
| 164 compressed.resize(size + sizeof(uint64_t)); | |
| 165 } | |
| 166 else | |
| 167 { | |
| 168 compressed.resize(size); | |
| 169 } | |
| 170 } | |
| 171 | |
| 172 | |
| 173 void GzipCompressor::Uncompress(std::string& uncompressed, | |
| 174 const void* compressed, | |
| 175 size_t compressedSize) | |
| 176 { | |
| 177 uint64_t uncompressedSize; | |
| 178 const uint8_t* source = reinterpret_cast<const uint8_t*>(compressed); | |
| 179 | |
| 180 if (HasPrefixWithUncompressedSize()) | |
| 181 { | |
| 182 uncompressedSize = ReadUncompressedSizePrefix(compressed, compressedSize); | |
| 183 source += sizeof(uint64_t); | |
| 184 compressedSize -= sizeof(uint64_t); | |
| 185 } | |
| 186 else | |
| 187 { | |
| 188 uncompressedSize = GuessUncompressedSize(compressed, compressedSize); | |
| 189 } | |
| 190 | |
| 191 try | |
| 192 { | |
| 1545 | 193 uncompressed.resize(static_cast<size_t>(uncompressedSize)); |
| 1513 | 194 } |
| 195 catch (...) | |
| 196 { | |
| 197 throw OrthancException(ErrorCode_NotEnoughMemory); | |
| 198 } | |
| 199 | |
| 200 z_stream stream; | |
| 201 memset(&stream, 0, sizeof(stream)); | |
| 202 | |
| 203 char dummy = '\0'; // zlib does not like NULL output buffers (even if the uncompressed data is empty) | |
| 204 stream.next_in = const_cast<Bytef*>(source); | |
| 205 stream.next_out = reinterpret_cast<Bytef*>(uncompressedSize == 0 ? &dummy : &uncompressed[0]); | |
| 206 | |
| 207 stream.avail_in = static_cast<uInt>(compressedSize); | |
| 208 stream.avail_out = static_cast<uInt>(uncompressedSize); | |
| 209 | |
| 210 // Ensure no overflow (if the buffer is too large for the current archicture) | |
| 211 if (static_cast<size_t>(stream.avail_in) != compressedSize || | |
| 212 static_cast<size_t>(stream.avail_out) != uncompressedSize) | |
| 213 { | |
| 214 throw OrthancException(ErrorCode_NotEnoughMemory); | |
| 215 } | |
| 216 | |
| 217 // Initialize the compression engine | |
| 218 int error = inflateInit2(&stream, | |
| 219 MAX_WBITS + 16); // this is a gzip input | |
| 220 | |
| 221 if (error != Z_OK) | |
| 222 { | |
| 223 // Cannot initialize zlib | |
| 224 uncompressed.clear(); | |
| 225 throw OrthancException(ErrorCode_InternalError); | |
| 226 } | |
| 227 | |
| 228 // Uncompress the input buffer | |
| 229 error = inflate(&stream, Z_FINISH); | |
| 230 | |
| 231 if (error != Z_STREAM_END) | |
| 232 { | |
| 233 inflateEnd(&stream); | |
| 234 uncompressed.clear(); | |
| 235 | |
| 236 switch (error) | |
| 237 { | |
| 238 case Z_MEM_ERROR: | |
| 239 throw OrthancException(ErrorCode_NotEnoughMemory); | |
| 240 | |
| 241 case Z_BUF_ERROR: | |
| 242 case Z_NEED_DICT: | |
| 243 throw OrthancException(ErrorCode_BadFileFormat); | |
| 244 | |
| 245 default: | |
| 246 throw OrthancException(ErrorCode_InternalError); | |
| 247 } | |
| 248 } | |
| 249 | |
| 250 size_t size = stream.total_out; | |
| 251 | |
| 252 if (inflateEnd(&stream) != Z_OK) | |
| 253 { | |
| 254 uncompressed.clear(); | |
| 255 throw OrthancException(ErrorCode_InternalError); | |
| 256 } | |
| 257 | |
| 258 if (size != uncompressedSize) | |
| 259 { | |
| 260 uncompressed.clear(); | |
| 261 | |
| 262 // The uncompressed size was not that properly guess, presumably | |
| 263 // because of a file size over 4GB. Should fallback to | |
| 264 // stream-based decompression. | |
|
2954
d924f9bb61cc
taking advantage of details in OrthancException
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
2447
diff
changeset
|
265 throw OrthancException(ErrorCode_NotImplemented, |
|
d924f9bb61cc
taking advantage of details in OrthancException
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
2447
diff
changeset
|
266 "The uncompressed size of a gzip-encoded buffer was not properly guessed"); |
| 1513 | 267 } |
| 268 } | |
| 269 } |