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