Merge "Revert "Log the reason why revision->getContent() returns null""
[lhc/web/wiklou.git] / includes / media / JpegMetadataExtractor.php
1 <?php
2 /**
3 * Extraction of JPEG image metadata.
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
19 *
20 * @file
21 * @ingroup Media
22 */
23
24 /**
25 * Class for reading jpegs and extracting metadata.
26 * see also BitmapMetadataHandler.
27 *
28 * Based somewhat on GIFMetadataExtractor.
29 *
30 * @ingroup Media
31 */
32 class JpegMetadataExtractor {
33 const MAX_JPEG_SEGMENTS = 200;
34
35 // the max segment is a sanity check.
36 // A jpeg file should never even remotely have
37 // that many segments. Your average file has about 10.
38
39 /** Function to extract metadata segments of interest from jpeg files
40 * based on GIFMetadataExtractor.
41 *
42 * we can almost use getimagesize to do this
43 * but gis doesn't support having multiple app1 segments
44 * and those can't extract xmp on files containing both exif and xmp data
45 *
46 * @param string $filename Name of jpeg file
47 * @return array Array of interesting segments.
48 * @throws MWException If given invalid file.
49 */
50 static function segmentSplitter( $filename ) {
51 $showXMP = XMPReader::isSupported();
52
53 $segmentCount = 0;
54
55 $segments = [
56 'XMP_ext' => [],
57 'COM' => [],
58 'PSIR' => [],
59 ];
60
61 if ( !$filename ) {
62 throw new MWException( "No filename specified for " . __METHOD__ );
63 }
64 if ( !file_exists( $filename ) || is_dir( $filename ) ) {
65 throw new MWException( "Invalid file $filename passed to " . __METHOD__ );
66 }
67
68 $fh = fopen( $filename, "rb" );
69
70 if ( !$fh ) {
71 throw new MWException( "Could not open file $filename" );
72 }
73
74 $buffer = fread( $fh, 2 );
75 if ( $buffer !== "\xFF\xD8" ) {
76 throw new MWException( "Not a jpeg, no SOI" );
77 }
78 while ( !feof( $fh ) ) {
79 $buffer = fread( $fh, 1 );
80 $segmentCount++;
81 if ( $segmentCount > self::MAX_JPEG_SEGMENTS ) {
82 // this is just a sanity check
83 throw new MWException( 'Too many jpeg segments. Aborting' );
84 }
85 while ( $buffer !== "\xFF" && !feof( $fh ) ) {
86 // In theory JPEG files are not allowed to contain anything between the sections,
87 // but in practice they sometimes do. It's customary to ignore the garbage data.
88 $buffer = fread( $fh, 1 );
89 }
90
91 $buffer = fread( $fh, 1 );
92 while ( $buffer === "\xFF" && !feof( $fh ) ) {
93 // Skip through any 0xFF padding bytes.
94 $buffer = fread( $fh, 1 );
95 }
96 if ( $buffer === "\xFE" ) {
97 // COM section -- file comment
98 // First see if valid utf-8,
99 // if not try to convert it to windows-1252.
100 $com = $oldCom = trim( self::jpegExtractMarker( $fh ) );
101 UtfNormal\Validator::quickIsNFCVerify( $com );
102 // turns $com to valid utf-8.
103 // thus if no change, its utf-8, otherwise its something else.
104 if ( $com !== $oldCom ) {
105 Wikimedia\suppressWarnings();
106 $com = $oldCom = iconv( 'windows-1252', 'UTF-8//IGNORE', $oldCom );
107 Wikimedia\restoreWarnings();
108 }
109 // Try it again, if its still not a valid string, then probably
110 // binary junk or some really weird encoding, so don't extract.
111 UtfNormal\Validator::quickIsNFCVerify( $com );
112 if ( $com === $oldCom ) {
113 $segments["COM"][] = $oldCom;
114 } else {
115 wfDebug( __METHOD__ . " Ignoring JPEG comment as is garbage.\n" );
116 }
117 } elseif ( $buffer === "\xE1" ) {
118 // APP1 section (Exif, XMP, and XMP extended)
119 // only extract if XMP is enabled.
120 $temp = self::jpegExtractMarker( $fh );
121 // check what type of app segment this is.
122 if ( substr( $temp, 0, 29 ) === "http://ns.adobe.com/xap/1.0/\x00" && $showXMP ) {
123 $segments["XMP"] = substr( $temp, 29 );
124 } elseif ( substr( $temp, 0, 35 ) === "http://ns.adobe.com/xmp/extension/\x00" && $showXMP ) {
125 $segments["XMP_ext"][] = substr( $temp, 35 );
126 } elseif ( substr( $temp, 0, 29 ) === "XMP\x00://ns.adobe.com/xap/1.0/\x00" && $showXMP ) {
127 // Some images (especially flickr images) seem to have this.
128 // I really have no idea what the deal is with them, but
129 // whatever...
130 $segments["XMP"] = substr( $temp, 29 );
131 wfDebug( __METHOD__ . ' Found XMP section with wrong app identifier '
132 . "Using anyways.\n" );
133 } elseif ( substr( $temp, 0, 6 ) === "Exif\0\0" ) {
134 // Just need to find out what the byte order is.
135 // because php's exif plugin sucks...
136 // This is a II for little Endian, MM for big. Not a unicode BOM.
137 $byteOrderMarker = substr( $temp, 6, 2 );
138 if ( $byteOrderMarker === 'MM' ) {
139 $segments['byteOrder'] = 'BE';
140 } elseif ( $byteOrderMarker === 'II' ) {
141 $segments['byteOrder'] = 'LE';
142 } else {
143 wfDebug( __METHOD__ . " Invalid byte ordering?!\n" );
144 }
145 }
146 } elseif ( $buffer === "\xED" ) {
147 // APP13 - PSIR. IPTC and some photoshop stuff
148 $temp = self::jpegExtractMarker( $fh );
149 if ( substr( $temp, 0, 14 ) === "Photoshop 3.0\x00" ) {
150 $segments["PSIR"][] = $temp;
151 }
152 } elseif ( $buffer === "\xD9" || $buffer === "\xDA" ) {
153 // EOI - end of image or SOS - start of scan. either way we're past any interesting segments
154 return $segments;
155 } else {
156 // segment we don't care about, so skip
157 $size = wfUnpack( "nint", fread( $fh, 2 ), 2 );
158 if ( $size['int'] < 2 ) {
159 throw new MWException( "invalid marker size in jpeg" );
160 }
161 // Note it's possible to seek beyond end of file if truncated.
162 // fseek doesn't report a failure in this case.
163 fseek( $fh, $size['int'] - 2, SEEK_CUR );
164 }
165 }
166 // shouldn't get here.
167 throw new MWException( "Reached end of jpeg file unexpectedly" );
168 }
169
170 /**
171 * Helper function for jpegSegmentSplitter
172 * @param resource &$fh File handle for JPEG file
173 * @throws MWException
174 * @return string Data content of segment.
175 */
176 private static function jpegExtractMarker( &$fh ) {
177 $size = wfUnpack( "nint", fread( $fh, 2 ), 2 );
178 if ( $size['int'] < 2 ) {
179 throw new MWException( "invalid marker size in jpeg" );
180 }
181 if ( $size['int'] === 2 ) {
182 // fread( ..., 0 ) generates a warning
183 return '';
184 }
185 $segment = fread( $fh, $size['int'] - 2 );
186 if ( strlen( $segment ) !== $size['int'] - 2 ) {
187 throw new MWException( "Segment shorter than expected" );
188 }
189
190 return $segment;
191 }
192
193 /**
194 * This reads the photoshop image resource.
195 * Currently it only compares the iptc/iim hash
196 * with the stored hash, which is used to determine the precedence
197 * of the iptc data. In future it may extract some other info, like
198 * url of copyright license.
199 *
200 * This should generally be called by BitmapMetadataHandler::doApp13()
201 *
202 * @param string $app13 Photoshop psir app13 block from jpg.
203 * @throws MWException (It gets caught next level up though)
204 * @return string If the iptc hash is good or not. One of 'iptc-no-hash',
205 * 'iptc-good-hash', 'iptc-bad-hash'.
206 */
207 public static function doPSIR( $app13 ) {
208 if ( !$app13 ) {
209 throw new MWException( "No App13 segment given" );
210 }
211 // First compare hash with real thing
212 // 0x404 contains IPTC, 0x425 has hash
213 // This is used to determine if the iptc is newer than
214 // the xmp data, as xmp programs update the hash,
215 // where non-xmp programs don't.
216
217 $offset = 14; // skip past PHOTOSHOP 3.0 identifier. should already be checked.
218 $appLen = strlen( $app13 );
219 $realHash = "";
220 $recordedHash = "";
221
222 // the +12 is the length of an empty item.
223 while ( $offset + 12 <= $appLen ) {
224 $valid = true;
225 if ( substr( $app13, $offset, 4 ) !== '8BIM' ) {
226 // its supposed to be 8BIM
227 // but apparently sometimes isn't esp. in
228 // really old jpg's
229 $valid = false;
230 }
231 $offset += 4;
232 $id = substr( $app13, $offset, 2 );
233 // id is a 2 byte id number which identifies
234 // the piece of info this record contains.
235
236 $offset += 2;
237
238 // some record types can contain a name, which
239 // is a pascal string 0-padded to be an even
240 // number of bytes. Most times (and any time
241 // we care) this is empty, making it two null bytes.
242
243 $lenName = ord( substr( $app13, $offset, 1 ) ) + 1;
244 // we never use the name so skip it. +1 for length byte
245 if ( $lenName % 2 == 1 ) {
246 $lenName++;
247 } // pad to even.
248 $offset += $lenName;
249
250 // now length of data (unsigned long big endian)
251 $lenData = wfUnpack( 'Nlen', substr( $app13, $offset, 4 ), 4 );
252 // PHP can take issue with very large unsigned ints and make them negative.
253 // Which should never ever happen, as this has to be inside a segment
254 // which is limited to a 16 bit number.
255 if ( $lenData['len'] < 0 ) {
256 throw new MWException( "Too big PSIR (" . $lenData['len'] . ')' );
257 }
258
259 $offset += 4; // 4bytes length field;
260
261 // this should not happen, but check.
262 if ( $lenData['len'] + $offset > $appLen ) {
263 throw new MWException( "PSIR data too long. (item length=" . $lenData['len']
264 . "; offset=$offset; total length=$appLen)" );
265 }
266
267 if ( $valid ) {
268 switch ( $id ) {
269 case "\x04\x04":
270 // IPTC block
271 $realHash = md5( substr( $app13, $offset, $lenData['len'] ), true );
272 break;
273 case "\x04\x25":
274 $recordedHash = substr( $app13, $offset, $lenData['len'] );
275 break;
276 }
277 }
278
279 // if odd, add 1 to length to account for
280 // null pad byte.
281 if ( $lenData['len'] % 2 == 1 ) {
282 $lenData['len']++;
283 }
284 $offset += $lenData['len'];
285 }
286
287 if ( !$realHash || !$recordedHash ) {
288 return 'iptc-no-hash';
289 } elseif ( $realHash === $recordedHash ) {
290 return 'iptc-good-hash';
291 } else { /*$realHash !== $recordedHash */
292 return 'iptc-bad-hash';
293 }
294 }
295 }