includes/media/BitmapMetadataHandler.php

   1 <?php
   2 /**
   3 Class to deal with reconciling and extracting metadata from bitmap images.
   4 This is meant to comply with http://www.metadataworkinggroup.org/pdf/mwg_guidance.pdf
   5
   6 This sort of acts as an intermediary between MediaHandler::getMetadata
   7 and the various metadata extractors.
   8
   9 @todo other image formats.
  10 */
  11 class BitmapMetadataHandler {
  12
  13         private $metadata = array();
  14         private $metaPriority = array(
  15                 20 => array( 'other' ),
  16                 40 => array( 'native' ),
  17                 60 => array( 'iptc-good-hash', 'iptc-no-hash' ),
  18                 70 => array( 'xmp-deprecated' ),
  19                 80 => array( 'xmp-general' ),
  20                 90 => array( 'xmp-exif' ),
  21                 100 => array( 'iptc-bad-hash' ),
  22                 120 => array( 'exif' ),
  23         );
  24         private $iptcType = 'iptc-no-hash';
  25
  26         /**
  27         * This does the photoshop image resource app13 block
  28         * of interest, IPTC-IIM metadata is stored here.
  29         *
  30         * Mostly just calls doPSIR and doIPTC
  31         *
  32         * @param String $app13 String containing app13 block from jpeg file
  33         */
  34         private function doApp13 ( $app13 ) {
  35                 try {
  36                         $this->iptcType = JpegMetadataExtractor::doPSIR( $app13 );
  37                 } catch ( MWException $e ) {
  38                         // Error reading the iptc hash information.
  39                         // This probably means the App13 segment is something other than what we expect.
  40                         // However, still try to read it, and treat it as if the hash didn't exist.
  41                         wfDebug( "Error parsing iptc data of file: " . $e->getMessage() . "\n" );
  42                         $this->iptcType = 'iptc-no-hash';
  43                 }
  44
  45                 $iptc = IPTC::parse( $app13 );
  46                 $this->addMetadata( $iptc, $this->iptcType );
  47         }
  48
  49
  50         /**
  51          * Get exif info using exif class.
  52          * Basically what used to be in BitmapHandler::getMetadata().
  53          * Just calls stuff in the Exif class.
  54          *
  55          * Parameters are passed to the Exif class.
  56          *
  57          * @param $filename string
  58          * @param $byteOrder string
  59          */
  60         function getExif ( $filename, $byteOrder ) {
  61                 global $wgShowEXIF;
  62                 if ( file_exists( $filename ) && $wgShowEXIF ) {
  63                         $exif = new Exif( $filename, $byteOrder );
  64                         $data = $exif->getFilteredData();
  65                         if ( $data ) {
  66                                 $this->addMetadata( $data, 'exif' );
  67                         }
  68                 }
  69         }
  70         /** Add misc metadata. Warning: atm if the metadata category
  71         * doesn't have a priority, it will be silently discarded.
  72         *
  73         * @param Array $metaArray array of metadata values
  74         * @param string $type type. defaults to other. if two things have the same type they're merged
  75         */
  76         function addMetadata ( $metaArray, $type = 'other' ) {
  77                 if ( isset( $this->metadata[$type] ) ) {
  78                         /* merge with old data */
  79                         $metaArray = $metaArray + $this->metadata[$type];
  80                 }
  81
  82                 $this->metadata[$type] = $metaArray;
  83         }
  84
  85         /**
  86         * Merge together the various types of metadata
  87         * the different types have different priorites,
  88         * and are merged in order.
  89         *
  90         * This function is generally called by the media handlers' getMetadata()
  91         *
  92         * @return Array metadata array
  93         */
  94         function getMetadataArray () {
  95                 // this seems a bit ugly... This is all so its merged in right order
  96                 // based on the MWG recomendation.
  97                 $temp = Array();
  98                 krsort( $this->metaPriority );
  99                 foreach ( $this->metaPriority as $pri ) {
 100                         foreach ( $pri as $type ) {
 101                                 if ( isset( $this->metadata[$type] ) ) {
 102                                         // Do some special casing for multilingual values.
 103                                         // Don't discard translations if also as a simple value.
 104                                         foreach ( $this->metadata[$type] as $itemName => $item ) {
 105                                                 if ( is_array( $item ) && isset( $item['_type'] ) && $item['_type'] === 'lang' ) {
 106                                                         if ( isset( $temp[$itemName] ) && !is_array( $temp[$itemName] ) ) {
 107                                                                 $default = $temp[$itemName];
 108                                                                 $temp[$itemName] = $item;
 109                                                                 $temp[$itemName]['x-default'] = $default;
 110                                                                 unset( $this->metadata[$type][$itemName] );
 111                                                         }
 112                                                 }
 113                                         }
 114
 115                                         $temp = $temp + $this->metadata[$type];
 116                                 }
 117                         }
 118                 }
 119                 return $temp;
 120         }
 121
 122         /** Main entry point for jpeg's.
 123          *
 124          * @param $filename string filename (with full path)
 125          * @return array metadata result array.
 126          * @throws MWException on invalid file.
 127          */
 128         static function Jpeg ( $filename ) {
 129                 $showXMP = function_exists( 'xml_parser_create_ns' );
 130                 $meta = new self();
 131
 132                 $seg = JpegMetadataExtractor::segmentSplitter( $filename );
 133                 if ( isset( $seg['COM'] ) && isset( $seg['COM'][0] ) ) {
 134                         $meta->addMetadata( Array( 'JPEGFileComment' => $seg['COM'] ), 'native' );
 135                 }
 136                 if ( isset( $seg['PSIR'] ) && count( $seg['PSIR'] ) > 0 ) {
 137                         foreach( $seg['PSIR'] as $curPSIRValue ) {
 138                                 $meta->doApp13( $curPSIRValue );
 139                         }
 140                 }
 141                 if ( isset( $seg['XMP'] ) && $showXMP ) {
 142                         $xmp = new XMPReader();
 143                         $xmp->parse( $seg['XMP'] );
 144                         foreach ( $seg['XMP_ext'] as $xmpExt ) {
 145                                 /* Support for extended xmp in jpeg files
 146                                  * is not well tested and a bit fragile.
 147                                  */
 148                                 $xmp->parseExtended( $xmpExt );
 149
 150                         }
 151                         $res = $xmp->getResults();
 152                         foreach ( $res as $type => $array ) {
 153                                 $meta->addMetadata( $array, $type );
 154                         }
 155                 }
 156                 if ( isset( $seg['byteOrder'] ) ) {
 157                         $meta->getExif( $filename, $seg['byteOrder'] );
 158                 }
 159                 return $meta->getMetadataArray();
 160         }
 161
 162         /** Entry point for png
 163          * At some point in the future this might
 164          * merge the png various tEXt chunks to that
 165          * are interesting, but for now it only does XMP
 166          *
 167          * @param $filename String full path to file
 168          * @return Array Array for storage in img_metadata.
 169          */
 170         static public function PNG ( $filename ) {
 171                 $showXMP = function_exists( 'xml_parser_create_ns' );
 172
 173                 $meta = new self();
 174                 $array = PNGMetadataExtractor::getMetadata( $filename );
 175                 if ( isset( $array['text']['xmp']['x-default'] ) && $array['text']['xmp']['x-default'] !== '' && $showXMP ) {
 176                         $xmp = new XMPReader();
 177                         $xmp->parse( $array['text']['xmp']['x-default'] );
 178                         $xmpRes = $xmp->getResults();
 179                         foreach ( $xmpRes as $type => $xmpSection ) {
 180                                 $meta->addMetadata( $xmpSection, $type );
 181                         }
 182                 }
 183                 unset( $array['text']['xmp'] );
 184                 $meta->addMetadata( $array['text'], 'native' );
 185                 unset( $array['text'] );
 186                 $array['metadata'] = $meta->getMetadataArray();
 187                 $array['metadata']['_MW_PNG_VERSION'] = PNGMetadataExtractor::VERSION;
 188                 return $array;
 189         }
 190
 191         /** function for gif images.
 192          *
 193          * They don't really have native metadata, so just merges together
 194          * XMP and image comment.
 195          *
 196          * @param $filename string full path to file
 197          * @return Array metadata array
 198          */
 199         static public function GIF ( $filename ) {
 200
 201                 $meta = new self();
 202                 $baseArray = GIFMetadataExtractor::getMetadata( $filename );
 203
 204                 if ( count( $baseArray['comment'] ) > 0 ) {
 205                         $meta->addMetadata( array( 'GIFFileComment' => $baseArray['comment'] ), 'native' );
 206                 }
 207
 208                 if ( $baseArray['xmp'] !== '' && function_exists( 'xml_parser_create_ns' ) ) {
 209                         $xmp = new XMPReader();
 210                         $xmp->parse( $baseArray['xmp'] );
 211                         $xmpRes = $xmp->getResults();
 212                         foreach ( $xmpRes as $type => $xmpSection ) {
 213                                 $meta->addMetadata( $xmpSection, $type );
 214                         }
 215
 216                 }
 217
 218                 unset( $baseArray['comment'] );
 219                 unset( $baseArray['xmp'] );
 220
 221                 $baseArray['metadata'] = $meta->getMetadataArray();
 222                 $baseArray['metadata']['_MW_GIF_VERSION'] = GIFMetadataExtractor::VERSION;
 223                 return $baseArray;
 224         }
 225
 226         /**
 227          * This doesn't do much yet, but eventually I plan to add
 228          * XMP support for Tiff. (PHP's exif support already extracts
 229          * but needs some further processing because PHP's exif support
 230          * is stupid...)
 231          *
 232          * @todo Add XMP support, so this function actually makes
 233          * sense to put here.
 234          *
 235          * The various exceptions this throws are caught later.
 236          * @param $filename String
 237          * @return Array The metadata.
 238          */
 239         static public function Tiff ( $filename ) {
 240                 if ( file_exists( $filename ) ) {
 241                         $byteOrder = self::getTiffByteOrder( $filename );
 242                         if ( !$byteOrder ) {
 243                                 throw new MWException( "Error determining byte order of $filename" );
 244                         }
 245                         $exif = new Exif( $filename, $byteOrder );
 246                         $data = $exif->getFilteredData();
 247                         if ( $data ) {
 248                                 $data['MEDIAWIKI_EXIF_VERSION'] = Exif::version();
 249                                 return $data;
 250                         } else {
 251                                 throw new MWException( "Could not extract data from tiff file $filename" );
 252                         }
 253                 } else {
 254                         throw new MWException( "File doesn't exist - $filename" );
 255                 }
 256         }
 257         /**
 258          * Read the first 2 bytes of a tiff file to figure out
 259          * Little Endian or Big Endian. Needed for exif stuff.
 260          *
 261          * @param $filename String The filename
 262          * @return String 'BE' or 'LE' or false
 263          */
 264         static function getTiffByteOrder( $filename ) {
 265                 $fh = fopen( $filename, 'rb' );
 266                 if ( !$fh ) return false;
 267                 $head = fread( $fh, 2 );
 268                 fclose( $fh );
 269
 270                 switch( $head ) {
 271                         case 'II':
 272                                 return 'LE'; // II for intel.
 273                         case 'MM':
 274                                 return 'BE'; // MM for motorla.
 275                         default:
 276                                 return false; // Something went wrong.
 277
 278                 }
 279         }
 280
 281
 282 }