MediaWiki  master
BitmapMetadataHandler.php
Go to the documentation of this file.
1 <?php
25 use Wikimedia\XMPReader\Reader as XMPReader;
26 
39  private $metadata = [];
40 
42  private $metaPriority = [
43  20 => [ 'other' ],
44  40 => [ 'native' ],
45  60 => [ 'iptc-good-hash', 'iptc-no-hash' ],
46  70 => [ 'xmp-deprecated' ],
47  80 => [ 'xmp-general' ],
48  90 => [ 'xmp-exif' ],
49  100 => [ 'iptc-bad-hash' ],
50  120 => [ 'exif' ],
51  ];
52 
54  private $iptcType = 'iptc-no-hash';
55 
64  private function doApp13( $app13 ) {
65  try {
66  $this->iptcType = JpegMetadataExtractor::doPSIR( $app13 );
67  } catch ( Exception $e ) {
68  // Error reading the iptc hash information.
69  // This probably means the App13 segment is something other than what we expect.
70  // However, still try to read it, and treat it as if the hash didn't exist.
71  wfDebug( "Error parsing iptc data of file: " . $e->getMessage() . "\n" );
72  $this->iptcType = 'iptc-no-hash';
73  }
74 
75  $iptc = IPTC::parse( $app13 );
76  $this->addMetadata( $iptc, $this->iptcType );
77  }
78 
89  function getExif( $filename, $byteOrder ) {
90  global $wgShowEXIF;
91  if ( file_exists( $filename ) && $wgShowEXIF ) {
92  $exif = new Exif( $filename, $byteOrder );
93  $data = $exif->getFilteredData();
94  if ( $data ) {
95  $this->addMetadata( $data, 'exif' );
96  }
97  }
98  }
99 
106  function addMetadata( $metaArray, $type = 'other' ) {
107  if ( isset( $this->metadata[$type] ) ) {
108  /* merge with old data */
109  $metaArray = $metaArray + $this->metadata[$type];
110  }
111 
112  $this->metadata[$type] = $metaArray;
113  }
114 
124  function getMetadataArray() {
125  // this seems a bit ugly... This is all so its merged in right order
126  // based on the MWG recommendation.
127  $temp = [];
128  krsort( $this->metaPriority );
129  foreach ( $this->metaPriority as $pri ) {
130  foreach ( $pri as $type ) {
131  if ( isset( $this->metadata[$type] ) ) {
132  // Do some special casing for multilingual values.
133  // Don't discard translations if also as a simple value.
134  foreach ( $this->metadata[$type] as $itemName => $item ) {
135  if ( is_array( $item ) && isset( $item['_type'] ) && $item['_type'] === 'lang' &&
136  isset( $temp[$itemName] ) && !is_array( $temp[$itemName] )
137  ) {
138  $default = $temp[$itemName];
139  $temp[$itemName] = $item;
140  $temp[$itemName]['x-default'] = $default;
141  unset( $this->metadata[$type][$itemName] );
142  }
143  }
144 
145  $temp = $temp + $this->metadata[$type];
146  }
147  }
148  }
149 
150  return $temp;
151  }
152 
159  static function Jpeg( $filename ) {
160  $showXMP = XMPReader::isSupported();
161  $meta = new self();
162 
163  $seg = JpegMetadataExtractor::segmentSplitter( $filename );
164 
165  if ( isset( $seg['COM'] ) && isset( $seg['COM'][0] ) ) {
166  $meta->addMetadata( [ 'JPEGFileComment' => $seg['COM'] ], 'native' );
167  }
168  if ( isset( $seg['PSIR'] ) && count( $seg['PSIR'] ) > 0 ) {
169  foreach ( $seg['PSIR'] as $curPSIRValue ) {
170  $meta->doApp13( $curPSIRValue );
171  }
172  }
173  if ( isset( $seg['XMP'] ) && $showXMP ) {
174  $xmp = new XMPReader( LoggerFactory::getInstance( 'XMP' ), $filename );
175  $xmp->parse( $seg['XMP'] );
176  foreach ( $seg['XMP_ext'] as $xmpExt ) {
177  /* Support for extended xmp in jpeg files
178  * is not well tested and a bit fragile.
179  */
180  $xmp->parseExtended( $xmpExt );
181  }
182  $res = $xmp->getResults();
183  foreach ( $res as $type => $array ) {
184  $meta->addMetadata( $array, $type );
185  }
186  }
187 
188  $meta->getExif( $filename, $seg['byteOrder'] ?? 'BE' );
189 
190  return $meta->getMetadataArray();
191  }
192 
201  public static function PNG( $filename ) {
202  $showXMP = XMPReader::isSupported();
203 
204  $meta = new self();
205  $array = PNGMetadataExtractor::getMetadata( $filename );
206  if ( isset( $array['text']['xmp']['x-default'] )
207  && $array['text']['xmp']['x-default'] !== '' && $showXMP
208  ) {
209  $xmp = new XMPReader( LoggerFactory::getInstance( 'XMP' ), $filename );
210  $xmp->parse( $array['text']['xmp']['x-default'] );
211  $xmpRes = $xmp->getResults();
212  foreach ( $xmpRes as $type => $xmpSection ) {
213  $meta->addMetadata( $xmpSection, $type );
214  }
215  }
216  unset( $array['text']['xmp'] );
217  $meta->addMetadata( $array['text'], 'native' );
218  unset( $array['text'] );
219  $array['metadata'] = $meta->getMetadataArray();
220  $array['metadata']['_MW_PNG_VERSION'] = PNGMetadataExtractor::VERSION;
221 
222  return $array;
223  }
224 
233  public static function GIF( $filename ) {
234  $meta = new self();
235  $baseArray = GIFMetadataExtractor::getMetadata( $filename );
236 
237  if ( count( $baseArray['comment'] ) > 0 ) {
238  $meta->addMetadata( [ 'GIFFileComment' => $baseArray['comment'] ], 'native' );
239  }
240 
241  if ( $baseArray['xmp'] !== '' && XMPReader::isSupported() ) {
242  $xmp = new XMPReader( LoggerFactory::getInstance( 'XMP' ), $filename );
243  $xmp->parse( $baseArray['xmp'] );
244  $xmpRes = $xmp->getResults();
245  foreach ( $xmpRes as $type => $xmpSection ) {
246  $meta->addMetadata( $xmpSection, $type );
247  }
248  }
249 
250  unset( $baseArray['comment'] );
251  unset( $baseArray['xmp'] );
252 
253  $baseArray['metadata'] = $meta->getMetadataArray();
254  $baseArray['metadata']['_MW_GIF_VERSION'] = GIFMetadataExtractor::VERSION;
255 
256  return $baseArray;
257  }
258 
272  public static function Tiff( $filename ) {
273  if ( file_exists( $filename ) ) {
274  $byteOrder = self::getTiffByteOrder( $filename );
275  if ( !$byteOrder ) {
276  throw new MWException( "Error determining byte order of $filename" );
277  }
278  $exif = new Exif( $filename, $byteOrder );
279  $data = $exif->getFilteredData();
280  if ( $data ) {
281  $data['MEDIAWIKI_EXIF_VERSION'] = Exif::version();
282 
283  return $data;
284  } else {
285  throw new MWException( "Could not extract data from tiff file $filename" );
286  }
287  } else {
288  throw new MWException( "File doesn't exist - $filename" );
289  }
290  }
291 
299  static function getTiffByteOrder( $filename ) {
300  $fh = fopen( $filename, 'rb' );
301  if ( !$fh ) {
302  return false;
303  }
304  $head = fread( $fh, 2 );
305  fclose( $fh );
306 
307  switch ( $head ) {
308  case 'II':
309  return 'LE'; // II for intel.
310  case 'MM':
311  return 'BE'; // MM for motorla.
312  default:
313  return false; // Something went wrong.
314 
315  }
316  }
317 }
doApp13( $app13)
This does the photoshop image resource app13 block of interest, IPTC-IIM metadata is stored here...
static GIF( $filename)
function for gif images.
static PNG( $filename)
Entry point for png At some point in the future this might merge the png various tEXt chunks to that ...
static Jpeg( $filename)
Main entry point for jpeg&#39;s.
static getTiffByteOrder( $filename)
Read the first 2 bytes of a tiff file to figure out Little Endian or Big Endian.
static version()
#-
Definition: Exif.php:582
static Tiff( $filename)
This doesn&#39;t do much yet, but eventually I plan to add XMP support for Tiff.
static segmentSplitter( $filename)
Function to extract metadata segments of interest from jpeg files based on GIFMetadataExtractor.
static parse( $rawData)
This takes the results of iptcparse() and puts it into a form that can be handled by mediawiki...
Definition: IPTC.php:40
Class to deal with reconciling and extracting metadata from bitmap images.
$wgShowEXIF
Show Exif data, on by default if available.
wfDebug( $text, $dest='all', array $context=[])
Sends a line to the debug log if enabled or, optionally, to a comment in output.
static getMetadata( $filename)
addMetadata( $metaArray, $type='other')
Add misc metadata.
static doPSIR( $app13)
This reads the photoshop image resource.
getMetadataArray()
Merge together the various types of metadata the different types have different priorites, and are merged in order.
Class to extract and validate Exif data from jpeg (and possibly tiff) files.
Definition: Exif.php:32
static getMetadata( $filename)
array $metaPriority
Metadata priority.
getExif( $filename, $byteOrder)
Get exif info using exif class.