MediaWiki  1.28.0
JpegMetadataExtractor.php
Go to the documentation of this file.
1 <?php
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 
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" ) {
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 
98  // COM section -- file comment
99  // First see if valid utf-8,
100  // if not try to convert it to windows-1252.
101  $com = $oldCom = trim( self::jpegExtractMarker( $fh ) );
102  UtfNormal\Validator::quickIsNFCVerify( $com );
103  // turns $com to valid utf-8.
104  // thus if no change, its utf-8, otherwise its something else.
105  if ( $com !== $oldCom ) {
106  MediaWiki\suppressWarnings();
107  $com = $oldCom = iconv( 'windows-1252', 'UTF-8//IGNORE', $oldCom );
108  MediaWiki\restoreWarnings();
109  }
110  // Try it again, if its still not a valid string, then probably
111  // binary junk or some really weird encoding, so don't extract.
112  UtfNormal\Validator::quickIsNFCVerify( $com );
113  if ( $com === $oldCom ) {
114  $segments["COM"][] = $oldCom;
115  } else {
116  wfDebug( __METHOD__ . " Ignoring JPEG comment as is garbage.\n" );
117  }
118  } elseif ( $buffer === "\xE1" ) {
119  // APP1 section (Exif, XMP, and XMP extended)
120  // only extract if XMP is enabled.
121  $temp = self::jpegExtractMarker( $fh );
122  // check what type of app segment this is.
123  if ( substr( $temp, 0, 29 ) === "http://ns.adobe.com/xap/1.0/\x00" && $showXMP ) {
124  $segments["XMP"] = substr( $temp, 29 );
125  } elseif ( substr( $temp, 0, 35 ) === "http://ns.adobe.com/xmp/extension/\x00" && $showXMP ) {
126  $segments["XMP_ext"][] = substr( $temp, 35 );
127  } elseif ( substr( $temp, 0, 29 ) === "XMP\x00://ns.adobe.com/xap/1.0/\x00" && $showXMP ) {
128  // Some images (especially flickr images) seem to have this.
129  // I really have no idea what the deal is with them, but
130  // whatever...
131  $segments["XMP"] = substr( $temp, 29 );
132  wfDebug( __METHOD__ . ' Found XMP section with wrong app identifier '
133  . "Using anyways.\n" );
134  } elseif ( substr( $temp, 0, 6 ) === "Exif\0\0" ) {
135  // Just need to find out what the byte order is.
136  // because php's exif plugin sucks...
137  // This is a II for little Endian, MM for big. Not a unicode BOM.
138  $byteOrderMarker = substr( $temp, 6, 2 );
139  if ( $byteOrderMarker === 'MM' ) {
140  $segments['byteOrder'] = 'BE';
141  } elseif ( $byteOrderMarker === 'II' ) {
142  $segments['byteOrder'] = 'LE';
143  } else {
144  wfDebug( __METHOD__ . " Invalid byte ordering?!\n" );
145  }
146  }
147  } elseif ( $buffer === "\xED" ) {
148  // APP13 - PSIR. IPTC and some photoshop stuff
149  $temp = self::jpegExtractMarker( $fh );
150  if ( substr( $temp, 0, 14 ) === "Photoshop 3.0\x00" ) {
151  $segments["PSIR"][] = $temp;
152  }
153  } elseif ( $buffer === "\xD9" || $buffer === "\xDA" ) {
154  // EOI - end of image or SOS - start of scan. either way we're past any interesting segments
155  return $segments;
156  } else {
157  // segment we don't care about, so skip
158  $size = wfUnpack( "nint", fread( $fh, 2 ), 2 );
159  if ( $size['int'] < 2 ) {
160  throw new MWException( "invalid marker size in jpeg" );
161  }
162  fseek( $fh, $size['int'] - 2, SEEK_CUR );
163  }
164  }
165  // shouldn't get here.
166  throw new MWException( "Reached end of jpeg file unexpectedly" );
167  }
168 
175  private static function jpegExtractMarker( &$fh ) {
176  $size = wfUnpack( "nint", fread( $fh, 2 ), 2 );
177  if ( $size['int'] < 2 ) {
178  throw new MWException( "invalid marker size in jpeg" );
179  }
180  if ( $size['int'] === 2 ) {
181  // fread( ..., 0 ) generates a warning
182  return '';
183  }
184  $segment = fread( $fh, $size['int'] - 2 );
185  if ( strlen( $segment ) !== $size['int'] - 2 ) {
186  throw new MWException( "Segment shorter than expected" );
187  }
188 
189  return $segment;
190  }
191 
206  public static function doPSIR( $app13 ) {
207  if ( !$app13 ) {
208  throw new MWException( "No App13 segment given" );
209  }
210  // First compare hash with real thing
211  // 0x404 contains IPTC, 0x425 has hash
212  // This is used to determine if the iptc is newer than
213  // the xmp data, as xmp programs update the hash,
214  // where non-xmp programs don't.
215 
216  $offset = 14; // skip past PHOTOSHOP 3.0 identifier. should already be checked.
217  $appLen = strlen( $app13 );
218  $realHash = "";
219  $recordedHash = "";
220 
221  // the +12 is the length of an empty item.
222  while ( $offset + 12 <= $appLen ) {
223  $valid = true;
224  if ( substr( $app13, $offset, 4 ) !== '8BIM' ) {
225  // its supposed to be 8BIM
226  // but apparently sometimes isn't esp. in
227  // really old jpg's
228  $valid = false;
229  }
230  $offset += 4;
231  $id = substr( $app13, $offset, 2 );
232  // id is a 2 byte id number which identifies
233  // the piece of info this record contains.
234 
235  $offset += 2;
236 
237  // some record types can contain a name, which
238  // is a pascal string 0-padded to be an even
239  // number of bytes. Most times (and any time
240  // we care) this is empty, making it two null bytes.
241 
242  $lenName = ord( substr( $app13, $offset, 1 ) ) + 1;
243  // we never use the name so skip it. +1 for length byte
244  if ( $lenName % 2 == 1 ) {
245  $lenName++;
246  } // pad to even.
247  $offset += $lenName;
248 
249  // now length of data (unsigned long big endian)
250  $lenData = wfUnpack( 'Nlen', substr( $app13, $offset, 4 ), 4 );
251  // PHP can take issue with very large unsigned ints and make them negative.
252  // Which should never ever happen, as this has to be inside a segment
253  // which is limited to a 16 bit number.
254  if ( $lenData['len'] < 0 ) {
255  throw new MWException( "Too big PSIR (" . $lenData['len'] . ')' );
256  }
257 
258  $offset += 4; // 4bytes length field;
259 
260  // this should not happen, but check.
261  if ( $lenData['len'] + $offset > $appLen ) {
262  throw new MWException( "PSIR data too long. (item length=" . $lenData['len']
263  . "; offset=$offset; total length=$appLen)" );
264  }
265 
266  if ( $valid ) {
267  switch ( $id ) {
268  case "\x04\x04":
269  // IPTC block
270  $realHash = md5( substr( $app13, $offset, $lenData['len'] ), true );
271  break;
272  case "\x04\x25":
273  $recordedHash = substr( $app13, $offset, $lenData['len'] );
274  break;
275  }
276  }
277 
278  // if odd, add 1 to length to account for
279  // null pad byte.
280  if ( $lenData['len'] % 2 == 1 ) {
281  $lenData['len']++;
282  }
283  $offset += $lenData['len'];
284  }
285 
286  if ( !$realHash || !$recordedHash ) {
287  return 'iptc-no-hash';
288  } elseif ( $realHash === $recordedHash ) {
289  return 'iptc-good-hash';
290  } else { /*$realHash !== $recordedHash */
291  return 'iptc-bad-hash';
292  }
293  }
294 }
static jpegExtractMarker(&$fh)
Helper function for jpegSegmentSplitter.
wfDebug($text, $dest= 'all', array $context=[])
Sends a line to the debug log if enabled or, optionally, to a comment in output.
static isSupported()
Check if this instance supports using this class.
Definition: XMP.php:198
static segmentSplitter($filename)
Function to extract metadata segments of interest from jpeg files based on GIFMetadataExtractor.
$buffer
injection txt This is an overview of how MediaWiki makes use of dependency injection The design described here grew from the discussion of RFC T384 The term dependency this means that anything an object needs to operate should be injected from the the object itself should only know narrow no concrete implementation of the logic it relies on The requirement to inject everything typically results in an architecture that based on two main types of and essentially stateless service objects that use other service objects to operate on the value objects As of the beginning MediaWiki is only starting to use the DI approach Much of the code still relies on global state or direct resulting in a highly cyclical dependency which acts as the top level factory for services in MediaWiki which can be used to gain access to default instances of various services MediaWikiServices however also allows new services to be defined and default services to be redefined Services are defined or redefined by providing a callback the instantiator that will return a new instance of the service When it will create an instance of MediaWikiServices and populate it with the services defined in the files listed by thereby bootstrapping the DI framework Per $wgServiceWiringFiles lists includes ServiceWiring php
Definition: injection.txt:35
wfUnpack($format, $data, $length=false)
Wrapper around php's unpack.
Class for reading jpegs and extracting metadata.
static doPSIR($app13)
This reads the photoshop image resource.