includes/media/PNGMetadataExtractor.php

   1 <?php
   2 /**
   3  * PNG frame counter and metadata extractor.
   4  * Slightly derived from GIFMetadataExtractor.php
   5  * Deliberately not using MWExceptions to avoid external dependencies, encouraging
   6  * redistribution.
   7  *
   8  * @file
   9  * @ingroup Media
  10  */
  11
  12 /**
  13  * PNG frame counter.
  14  *
  15  * @ingroup Media
  16  */
  17 class PNGMetadataExtractor {
  18         static $png_sig;
  19         static $CRC_size;
  20         static $text_chunks;
  21
  22         const VERSION = 1;
  23         const MAX_CHUNK_SIZE = 3145728; // 3 megabytes
  24
  25         static function getMetadata( $filename ) {
  26                 self::$png_sig = pack( "C8", 137, 80, 78, 71, 13, 10, 26, 10 );
  27                 self::$CRC_size = 4;
  28                 /* based on list at http://owl.phy.queensu.ca/~phil/exiftool/TagNames/PNG.html#TextualData
  29                  * and http://www.w3.org/TR/PNG/#11keywords
  30                  */
  31                 self::$text_chunks = array(
  32                         'xml:com.adobe.xmp' => 'xmp',
  33                         # Artist is unofficial. Author is the recommended
  34                         # keyword in the PNG spec. However some people output
  35                         # Artist so support both.
  36                         'artist'      => 'Artist',
  37                         'model'       => 'Model',
  38                         'make'        => 'Make',
  39                         'author'      => 'Artist',
  40                         'comment'     => 'PNGFileComment',
  41                         'description' => 'ImageDescription',
  42                         'title'       => 'ObjectName',
  43                         'copyright'   => 'Copyright',
  44                         # Source as in original device used to make image
  45                         # not as in who gave you the image
  46                         'source'      => 'Model',
  47                         'software'    => 'Software',
  48                         'disclaimer'  => 'Disclaimer',
  49                         'warning'     => 'ContentWarning',
  50                         'url'         => 'Identifier', # Not sure if this is best mapping. Maybe WebStatement.
  51                         'label'       => 'Label',
  52                         'creation time' => 'DateTimeDigitized',
  53                         /* Other potentially useful things - Document */
  54                 );
  55
  56                 $showXMP = function_exists( 'xml_parser_create_ns' );
  57
  58                 $frameCount = 0;
  59                 $loopCount = 1;
  60                 $text = array();
  61                 $bitDepth = 0;
  62                 $colorType = 'unknown';
  63
  64                 if ( !$filename ) {
  65                         throw new Exception( __METHOD__ . ": No file name specified" );
  66                 } elseif ( !file_exists( $filename ) || is_dir( $filename ) ) {
  67                         throw new Exception( __METHOD__ . ": File $filename does not exist" );
  68                 }
  69
  70                 $fh = fopen( $filename, 'r' );
  71
  72                 if ( !$fh ) {
  73                         throw new Exception( __METHOD__ . ": Unable to open file $filename" );
  74                 }
  75
  76                 // Check for the PNG header
  77                 $buf = fread( $fh, 8 );
  78                 if ( $buf != self::$png_sig ) {
  79                         throw new Exception( __METHOD__ . ": Not a valid PNG file; header: $buf" );
  80                 }
  81
  82                 // Read chunks
  83                 while ( !feof( $fh ) ) {
  84                         $buf = fread( $fh, 4 );
  85                         if ( !$buf ) {
  86                                 throw new Exception( __METHOD__ . ": Read error" );
  87                         }
  88                         $chunk_size = unpack( "N", $buf );
  89                         $chunk_size = $chunk_size[1];
  90
  91                         $chunk_type = fread( $fh, 4 );
  92                         if ( !$chunk_type ) {
  93                                 throw new Exception( __METHOD__ . ": Read error" );
  94                         }
  95
  96                         if ( $chunk_type == "IHDR" ) {
  97                                 $buf = self::read( $fh, $chunk_size );
  98                                 if ( !$buf ) {
  99                                         throw new Exception( __METHOD__ . ": Read error" );
 100                                 }
 101                                 $bitDepth = ord( substr( $buf, 8, 1 ) );
 102                                 // Detect the color type in British English as per the spec
 103                                 // http://www.w3.org/TR/PNG/#11IHDR
 104                                 switch ( ord( substr( $buf, 9, 1 ) ) ) {
 105                                         case 0:
 106                                                 $colorType = 'greyscale';
 107                                                 break;
 108                                         case 2:
 109                                                 $colorType = 'truecolour';
 110                                                 break;
 111                                         case 3:
 112                                                 $colorType = 'index-coloured';
 113                                                 break;
 114                                         case 4:
 115                                                 $colorType = 'greyscale-alpha';
 116                                                 break;
 117                                         case 6:
 118                                                 $colorType = 'truecolour-alpha';
 119                                                 break;
 120                                         default:
 121                                                 $colorType = 'unknown';
 122                                                 break;
 123                                 }
 124                         } elseif ( $chunk_type == "acTL" ) {
 125                                 $buf = fread( $fh, $chunk_size );
 126                                 if( !$buf ) {
 127                                         throw new Exception( __METHOD__ . ": Read error" );
 128                                 }
 129
 130                                 $actl = unpack( "Nframes/Nplays", $buf );
 131                                 $frameCount = $actl['frames'];
 132                                 $loopCount = $actl['plays'];
 133                         } elseif ( $chunk_type == "fcTL" ) {
 134                                 $buf = self::read( $fh, $chunk_size );
 135                                 if ( !$buf ) {
 136                                         throw new Exception( __METHOD__ . ": Read error" );
 137                                 }
 138                                 $buf = substr( $buf, 20 );
 139
 140                                 $fctldur = unpack( "ndelay_num/ndelay_den", $buf );
 141                                 if ( $fctldur['delay_den'] == 0 ) {
 142                                         $fctldur['delay_den'] = 100;
 143                                 }
 144                                 if ( $fctldur['delay_num'] ) {
 145                                         $duration += $fctldur['delay_num'] / $fctldur['delay_den'];
 146                                 }
 147                         } elseif ( $chunk_type == "iTXt" ) {
 148                                 // Extracts iTXt chunks, uncompressing if necessary.
 149                                 $buf = self::read( $fh, $chunk_size );
 150                                 $items = array();
 151                                 if ( preg_match(
 152                                         '/^([^\x00]{1,79})\x00(\x00|\x01)\x00([^\x00]*)(.)[^\x00]*\x00(.*)$/Ds',
 153                                         $buf, $items )
 154                                 ) {
 155                                         /* $items[1] = text chunk name, $items[2] = compressed flag,
 156                                          * $items[3] = lang code (or ""), $items[4]= compression type.
 157                                          * $items[5] = content
 158                                          */
 159
 160                                         // Theoretically should be case-sensitive, but in practise...
 161                                         $items[1] = strtolower( $items[1] );
 162                                         if ( !isset( self::$text_chunks[$items[1]] ) ) {
 163                                                 // Only extract textual chunks on our list.
 164                                                 fseek( $fh, self::$CRC_size, SEEK_CUR );
 165                                                 continue;
 166                                         }
 167
 168                                         $items[3] = strtolower( $items[3] );
 169                                         if ( $items[3] == '' ) {
 170                                                 // if no lang specified use x-default like in xmp.
 171                                                 $items[3] = 'x-default';
 172                                         }
 173
 174                                         // if compressed
 175                                         if ( $items[2] == "\x01" ) {
 176                                                 if ( function_exists( 'gzuncompress' ) && $items[4] === "\x00" ) {
 177                                                         wfSuppressWarnings();
 178                                                         $items[5] = gzuncompress( $items[5] );
 179                                                         wfRestoreWarnings();
 180
 181                                                         if ( $items[5] === false ) {
 182                                                                 // decompression failed
 183                                                                 wfDebug( __METHOD__ . ' Error decompressing iTxt chunk - ' . $items[1] );
 184                                                                 fseek( $fh, self::$CRC_size, SEEK_CUR );
 185                                                                 continue;
 186                                                         }
 187
 188                                                 } else {
 189                                                         wfDebug( __METHOD__ . ' Skipping compressed png iTXt chunk due to lack of zlib,'
 190                                                                 . ' or potentially invalid compression method' );
 191                                                         fseek( $fh, self::$CRC_size, SEEK_CUR );
 192                                                         continue;
 193                                                 }
 194                                         }
 195                                         $finalKeyword = self::$text_chunks[ $items[1] ];
 196                                         $text[ $finalKeyword ][ $items[3] ] = $items[5];
 197                                         $text[ $finalKeyword ]['_type'] = 'lang';
 198
 199                                 } else {
 200                                         // Error reading iTXt chunk
 201                                         throw new Exception( __METHOD__ . ": Read error on iTXt chunk" );
 202                                 }
 203
 204                         } elseif ( $chunk_type == 'tEXt' ) {
 205                                 $buf = self::read( $fh, $chunk_size );
 206                                 $keyword = '';
 207                                 $content = '';
 208
 209                                 list( $keyword, $content ) = explode( "\x00", $buf, 2 );
 210                                 if ( $keyword === '' || $content === '' ) {
 211                                         throw new Exception( __METHOD__ . ": Read error on tEXt chunk" );
 212                                 }
 213
 214                                 // Theoretically should be case-sensitive, but in practise...
 215                                 $keyword = strtolower( $keyword );
 216                                 if ( !isset( self::$text_chunks[ $keyword ] ) ) {
 217                                         // Don't recognize chunk, so skip.
 218                                         fseek( $fh, self::$CRC_size, SEEK_CUR );
 219                                         continue;
 220                                 }
 221                                 wfSuppressWarnings();
 222                                 $content = iconv( 'ISO-8859-1', 'UTF-8', $content );
 223                                 wfRestoreWarnings();
 224
 225                                 if ( $content === false ) {
 226                                         throw new Exception( __METHOD__ . ": Read error (error with iconv)" );
 227                                 }
 228
 229                                 $finalKeyword = self::$text_chunks[ $keyword ];
 230                                 $text[ $finalKeyword ][ 'x-default' ] = $content;
 231                                 $text[ $finalKeyword ]['_type'] = 'lang';
 232
 233                         } elseif ( $chunk_type == 'zTXt' ) {
 234                                 if ( function_exists( 'gzuncompress' ) ) {
 235                                         $buf = self::read( $fh, $chunk_size );
 236                                         $keyword = '';
 237                                         $postKeyword = '';
 238
 239                                         list( $keyword, $postKeyword ) = explode( "\x00", $buf, 2 );
 240                                         if ( $keyword === '' || $postKeyword === '' ) {
 241                                                 throw new Exception( __METHOD__ . ": Read error on zTXt chunk" );
 242                                         }
 243                                         // Theoretically should be case-sensitive, but in practise...
 244                                         $keyword = strtolower( $keyword );
 245
 246                                         if ( !isset( self::$text_chunks[ $keyword ] ) ) {
 247                                                 // Don't recognize chunk, so skip.
 248                                                 fseek( $fh, self::$CRC_size, SEEK_CUR );
 249                                                 continue;
 250                                         }
 251                                         $compression = substr( $postKeyword, 0, 1 );
 252                                         $content = substr( $postKeyword, 1 );
 253                                         if ( $compression !== "\x00" ) {
 254                                                 wfDebug( __METHOD__ . " Unrecognized compression method in zTXt ($keyword). Skipping." );
 255                                                 fseek( $fh, self::$CRC_size, SEEK_CUR );
 256                                                 continue;
 257                                         }
 258
 259                                         wfSuppressWarnings();
 260                                         $content = gzuncompress( $content );
 261                                         wfRestoreWarnings();
 262
 263                                         if ( $content === false ) {
 264                                                 // decompression failed
 265                                                 wfDebug( __METHOD__ . ' Error decompressing zTXt chunk - ' . $keyword );
 266                                                 fseek( $fh, self::$CRC_size, SEEK_CUR );
 267                                                 continue;
 268                                         }
 269
 270                                         wfSuppressWarnings();
 271                                         $content = iconv( 'ISO-8859-1', 'UTF-8', $content );
 272                                         wfRestoreWarnings();
 273
 274                                         if ( $content === false ) {
 275                                                 throw new Exception( __METHOD__ . ": Read error (error with iconv)" );
 276                                         }
 277
 278                                         $finalKeyword = self::$text_chunks[ $keyword ];
 279                                         $text[ $finalKeyword ][ 'x-default' ] = $content;
 280                                         $text[ $finalKeyword ]['_type'] = 'lang';
 281
 282                                 } else {
 283                                         wfDebug( __METHOD__ . " Cannot decompress zTXt chunk due to lack of zlib. Skipping." );
 284                                         fseek( $fh, $chunk_size, SEEK_CUR );
 285                                 }
 286                         } elseif ( $chunk_type == 'tIME' ) {
 287                                 // last mod timestamp.
 288                                 if ( $chunk_size !== 7 ) {
 289                                         throw new Exception( __METHOD__ . ": tIME wrong size" );
 290                                 }
 291                                 $buf = self::read( $fh, $chunk_size );
 292                                 if ( !$buf ) {
 293                                         throw new Exception( __METHOD__ . ": Read error" );
 294                                 }
 295
 296                                 // Note: spec says this should be UTC.
 297                                 $t = unpack( "ny/Cm/Cd/Ch/Cmin/Cs", $buf );
 298                                 $strTime = sprintf( "%04d%02d%02d%02d%02d%02d",
 299                                         $t['y'], $t['m'], $t['d'], $t['h'],
 300                                         $t['min'], $t['s'] );
 301
 302                                 $exifTime = wfTimestamp( TS_EXIF, $strTime );
 303
 304                                 if ( $exifTime ) {
 305                                         $text['DateTime'] = $exifTime;
 306                                 }
 307
 308                         } elseif ( $chunk_type == 'pHYs' ) {
 309                                 // how big pixels are (dots per meter).
 310                                 if ( $chunk_size !== 9 ) {
 311                                         throw new Exception( __METHOD__ . ": pHYs wrong size" );
 312                                 }
 313
 314                                 $buf = self::read( $fh, $chunk_size );
 315                                 if ( !$buf ) {
 316                                         throw new Exception( __METHOD__ . ": Read error" );
 317                                 }
 318
 319                                 $dim = unpack( "Nwidth/Nheight/Cunit", $buf );
 320                                 if ( $dim['unit'] == 1 ) {
 321                                         // unit is meters
 322                                         // (as opposed to 0 = undefined )
 323                                         $text['XResolution'] = $dim['width']
 324                                                 . '/100';
 325                                         $text['YResolution'] = $dim['height']
 326                                                 . '/100';
 327                                         $text['ResolutionUnit'] = 3;
 328                                         // 3 = dots per cm (from Exif).
 329                                 }
 330
 331                         } elseif ( $chunk_type == "IEND" ) {
 332                                 break;
 333                         } else {
 334                                 fseek( $fh, $chunk_size, SEEK_CUR );
 335                         }
 336                         fseek( $fh, self::$CRC_size, SEEK_CUR );
 337                 }
 338                 fclose( $fh );
 339
 340                 if ( $loopCount > 1 ) {
 341                         $duration *= $loopCount;
 342                 }
 343
 344                 if ( isset( $text['DateTimeDigitized'] ) ) {
 345                         // Convert date format from rfc2822 to exif.
 346                         foreach ( $text['DateTimeDigitized'] as $name => &$value ) {
 347                                 if ( $name === '_type' ) {
 348                                         continue;
 349                                 }
 350
 351                                 // fixme: currently timezones are ignored.
 352                                 // possibly should be wfTimestamp's
 353                                 // responsibility. (at least for numeric TZ)
 354                                 $formatted = wfTimestamp( TS_EXIF, $value );
 355                                 if ( $formatted ) {
 356                                         // Only change if we could convert the
 357                                         // date.
 358                                         // The png standard says it should be
 359                                         // in rfc2822 format, but not required.
 360                                         // In general for the exif stuff we
 361                                         // prettify the date if we can, but we
 362                                         // display as-is if we cannot or if
 363                                         // it is invalid.
 364                                         // So do the same here.
 365
 366                                         $value = $formatted;
 367                                 }
 368                         }
 369                 }
 370                 return array(
 371                         'frameCount' => $frameCount,
 372                         'loopCount' => $loopCount,
 373                         'duration' => $duration,
 374                         'text' => $text,
 375                         'duration' => $duration,
 376                         'bitDepth' => $bitDepth,
 377                         'colorType' => $colorType,
 378                 );
 379
 380         }
 381         /**
 382          * Read a chunk, checking to make sure its not too big.
 383          *
 384          * @param $fh resource The file handle
 385          * @param $size Integer size in bytes.
 386          * @throws Exception if too big.
 387          * @return String The chunk.
 388          */
 389         static private function read( $fh, $size ) {
 390                 if ( $size > self::MAX_CHUNK_SIZE ) {
 391                         throw new Exception( __METHOD__ . ': Chunk size of ' . $size .
 392                                 ' too big. Max size is: ' . self::MAX_CHUNK_SIZE );
 393                 }
 394                 return fread( $fh, $size );
 395         }
 396 }