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                 $frameCount = 0;
  57                 $loopCount = 1;
  58                 $text = array();
  59                 $duration = 0.0;
  60                 $bitDepth = 0;
  61                 $colorType = 'unknown';
  62
  63                 if ( !$filename ) {
  64                         throw new Exception( __METHOD__ . ": No file name specified" );
  65                 } elseif ( !file_exists( $filename ) || is_dir( $filename ) ) {
  66                         throw new Exception( __METHOD__ . ": File $filename does not exist" );
  67                 }
  68
  69                 $fh = fopen( $filename, 'r' );
  70
  71                 if ( !$fh ) {
  72                         throw new Exception( __METHOD__ . ": Unable to open file $filename" );
  73                 }
  74
  75                 // Check for the PNG header
  76                 $buf = fread( $fh, 8 );
  77                 if ( $buf != self::$png_sig ) {
  78                         throw new Exception( __METHOD__ . ": Not a valid PNG file; header: $buf" );
  79                 }
  80
  81                 // Read chunks
  82                 while ( !feof( $fh ) ) {
  83                         $buf = fread( $fh, 4 );
  84                         if ( !$buf ) {
  85                                 throw new Exception( __METHOD__ . ": Read error" );
  86                         }
  87                         $chunk_size = unpack( "N", $buf );
  88                         $chunk_size = $chunk_size[1];
  89
  90                         $chunk_type = fread( $fh, 4 );
  91                         if ( !$chunk_type ) {
  92                                 throw new Exception( __METHOD__ . ": Read error" );
  93                         }
  94
  95                         if ( $chunk_type == "IHDR" ) {
  96                                 $buf = self::read( $fh, $chunk_size );
  97                                 if ( !$buf ) {
  98                                         throw new Exception( __METHOD__ . ": Read error" );
  99                                 }
 100                                 $bitDepth = ord( substr( $buf, 8, 1 ) );
 101                                 // Detect the color type in British English as per the spec
 102                                 // http://www.w3.org/TR/PNG/#11IHDR
 103                                 switch ( ord( substr( $buf, 9, 1 ) ) ) {
 104                                         case 0:
 105                                                 $colorType = 'greyscale';
 106                                                 break;
 107                                         case 2:
 108                                                 $colorType = 'truecolour';
 109                                                 break;
 110                                         case 3:
 111                                                 $colorType = 'index-coloured';
 112                                                 break;
 113                                         case 4:
 114                                                 $colorType = 'greyscale-alpha';
 115                                                 break;
 116                                         case 6:
 117                                                 $colorType = 'truecolour-alpha';
 118                                                 break;
 119                                         default:
 120                                                 $colorType = 'unknown';
 121                                                 break;
 122                                 }
 123                         } elseif ( $chunk_type == "acTL" ) {
 124                                 $buf = fread( $fh, $chunk_size );
 125                                 if( !$buf ) {
 126                                         throw new Exception( __METHOD__ . ": Read error" );
 127                                 }
 128
 129                                 $actl = unpack( "Nframes/Nplays", $buf );
 130                                 $frameCount = $actl['frames'];
 131                                 $loopCount = $actl['plays'];
 132                         } elseif ( $chunk_type == "fcTL" ) {
 133                                 $buf = self::read( $fh, $chunk_size );
 134                                 if ( !$buf ) {
 135                                         throw new Exception( __METHOD__ . ": Read error" );
 136                                 }
 137                                 $buf = substr( $buf, 20 );
 138
 139                                 $fctldur = unpack( "ndelay_num/ndelay_den", $buf );
 140                                 if ( $fctldur['delay_den'] == 0 ) {
 141                                         $fctldur['delay_den'] = 100;
 142                                 }
 143                                 if ( $fctldur['delay_num'] ) {
 144                                         $duration += $fctldur['delay_num'] / $fctldur['delay_den'];
 145                                 }
 146                         } elseif ( $chunk_type == "iTXt" ) {
 147                                 // Extracts iTXt chunks, uncompressing if necessary.
 148                                 $buf = self::read( $fh, $chunk_size );
 149                                 $items = array();
 150                                 if ( preg_match(
 151                                         '/^([^\x00]{1,79})\x00(\x00|\x01)\x00([^\x00]*)(.)[^\x00]*\x00(.*)$/Ds',
 152                                         $buf, $items )
 153                                 ) {
 154                                         /* $items[1] = text chunk name, $items[2] = compressed flag,
 155                                          * $items[3] = lang code (or ""), $items[4]= compression type.
 156                                          * $items[5] = content
 157                                          */
 158
 159                                         // Theoretically should be case-sensitive, but in practise...
 160                                         $items[1] = strtolower( $items[1] );
 161                                         if ( !isset( self::$text_chunks[$items[1]] ) ) {
 162                                                 // Only extract textual chunks on our list.
 163                                                 fseek( $fh, self::$CRC_size, SEEK_CUR );
 164                                                 continue;
 165                                         }
 166
 167                                         $items[3] = strtolower( $items[3] );
 168                                         if ( $items[3] == '' ) {
 169                                                 // if no lang specified use x-default like in xmp.
 170                                                 $items[3] = 'x-default';
 171                                         }
 172
 173                                         // if compressed
 174                                         if ( $items[2] == "\x01" ) {
 175                                                 if ( function_exists( 'gzuncompress' ) && $items[4] === "\x00" ) {
 176                                                         wfSuppressWarnings();
 177                                                         $items[5] = gzuncompress( $items[5] );
 178                                                         wfRestoreWarnings();
 179
 180                                                         if ( $items[5] === false ) {
 181                                                                 // decompression failed
 182                                                                 wfDebug( __METHOD__ . ' Error decompressing iTxt chunk - ' . $items[1] );
 183                                                                 fseek( $fh, self::$CRC_size, SEEK_CUR );
 184                                                                 continue;
 185                                                         }
 186
 187                                                 } else {
 188                                                         wfDebug( __METHOD__ . ' Skipping compressed png iTXt chunk due to lack of zlib,'
 189                                                                 . ' or potentially invalid compression method' );
 190                                                         fseek( $fh, self::$CRC_size, SEEK_CUR );
 191                                                         continue;
 192                                                 }
 193                                         }
 194                                         $finalKeyword = self::$text_chunks[ $items[1] ];
 195                                         $text[ $finalKeyword ][ $items[3] ] = $items[5];
 196                                         $text[ $finalKeyword ]['_type'] = 'lang';
 197
 198                                 } else {
 199                                         // Error reading iTXt chunk
 200                                         throw new Exception( __METHOD__ . ": Read error on iTXt chunk" );
 201                                 }
 202
 203                         } elseif ( $chunk_type == 'tEXt' ) {
 204                                 $buf = self::read( $fh, $chunk_size );
 205
 206                                 // In case there is no \x00 which will make explode fail.
 207                                 if ( strpos( $buf, "\x00" ) === false ) {
 208                                         throw new Exception( __METHOD__ . ": Read error on tEXt chunk" );
 209                                 }
 210
 211                                 list( $keyword, $content ) = explode( "\x00", $buf, 2 );
 212                                 if ( $keyword === '' || $content === '' ) {
 213                                         throw new Exception( __METHOD__ . ": Read error on tEXt chunk" );
 214                                 }
 215
 216                                 // Theoretically should be case-sensitive, but in practise...
 217                                 $keyword = strtolower( $keyword );
 218                                 if ( !isset( self::$text_chunks[ $keyword ] ) ) {
 219                                         // Don't recognize chunk, so skip.
 220                                         fseek( $fh, self::$CRC_size, SEEK_CUR );
 221                                         continue;
 222                                 }
 223                                 wfSuppressWarnings();
 224                                 $content = iconv( 'ISO-8859-1', 'UTF-8', $content );
 225                                 wfRestoreWarnings();
 226
 227                                 if ( $content === false ) {
 228                                         throw new Exception( __METHOD__ . ": Read error (error with iconv)" );
 229                                 }
 230
 231                                 $finalKeyword = self::$text_chunks[ $keyword ];
 232                                 $text[ $finalKeyword ][ 'x-default' ] = $content;
 233                                 $text[ $finalKeyword ]['_type'] = 'lang';
 234
 235                         } elseif ( $chunk_type == 'zTXt' ) {
 236                                 if ( function_exists( 'gzuncompress' ) ) {
 237                                         $buf = self::read( $fh, $chunk_size );
 238
 239                                         // In case there is no \x00 which will make explode fail.
 240                                         if ( strpos( $buf, "\x00" ) === false ) {
 241                                                 throw new Exception( __METHOD__ . ": Read error on zTXt chunk" );
 242                                         }
 243
 244                                         list( $keyword, $postKeyword ) = explode( "\x00", $buf, 2 );
 245                                         if ( $keyword === '' || $postKeyword === '' ) {
 246                                                 throw new Exception( __METHOD__ . ": Read error on zTXt chunk" );
 247                                         }
 248                                         // Theoretically should be case-sensitive, but in practise...
 249                                         $keyword = strtolower( $keyword );
 250
 251                                         if ( !isset( self::$text_chunks[ $keyword ] ) ) {
 252                                                 // Don't recognize chunk, so skip.
 253                                                 fseek( $fh, self::$CRC_size, SEEK_CUR );
 254                                                 continue;
 255                                         }
 256                                         $compression = substr( $postKeyword, 0, 1 );
 257                                         $content = substr( $postKeyword, 1 );
 258                                         if ( $compression !== "\x00" ) {
 259                                                 wfDebug( __METHOD__ . " Unrecognized compression method in zTXt ($keyword). Skipping." );
 260                                                 fseek( $fh, self::$CRC_size, SEEK_CUR );
 261                                                 continue;
 262                                         }
 263
 264                                         wfSuppressWarnings();
 265                                         $content = gzuncompress( $content );
 266                                         wfRestoreWarnings();
 267
 268                                         if ( $content === false ) {
 269                                                 // decompression failed
 270                                                 wfDebug( __METHOD__ . ' Error decompressing zTXt chunk - ' . $keyword );
 271                                                 fseek( $fh, self::$CRC_size, SEEK_CUR );
 272                                                 continue;
 273                                         }
 274
 275                                         wfSuppressWarnings();
 276                                         $content = iconv( 'ISO-8859-1', 'UTF-8', $content );
 277                                         wfRestoreWarnings();
 278
 279                                         if ( $content === false ) {
 280                                                 throw new Exception( __METHOD__ . ": Read error (error with iconv)" );
 281                                         }
 282
 283                                         $finalKeyword = self::$text_chunks[ $keyword ];
 284                                         $text[ $finalKeyword ][ 'x-default' ] = $content;
 285                                         $text[ $finalKeyword ]['_type'] = 'lang';
 286
 287                                 } else {
 288                                         wfDebug( __METHOD__ . " Cannot decompress zTXt chunk due to lack of zlib. Skipping." );
 289                                         fseek( $fh, $chunk_size, SEEK_CUR );
 290                                 }
 291                         } elseif ( $chunk_type == 'tIME' ) {
 292                                 // last mod timestamp.
 293                                 if ( $chunk_size !== 7 ) {
 294                                         throw new Exception( __METHOD__ . ": tIME wrong size" );
 295                                 }
 296                                 $buf = self::read( $fh, $chunk_size );
 297                                 if ( !$buf ) {
 298                                         throw new Exception( __METHOD__ . ": Read error" );
 299                                 }
 300
 301                                 // Note: spec says this should be UTC.
 302                                 $t = unpack( "ny/Cm/Cd/Ch/Cmin/Cs", $buf );
 303                                 $strTime = sprintf( "%04d%02d%02d%02d%02d%02d",
 304                                         $t['y'], $t['m'], $t['d'], $t['h'],
 305                                         $t['min'], $t['s'] );
 306
 307                                 $exifTime = wfTimestamp( TS_EXIF, $strTime );
 308
 309                                 if ( $exifTime ) {
 310                                         $text['DateTime'] = $exifTime;
 311                                 }
 312
 313                         } elseif ( $chunk_type == 'pHYs' ) {
 314                                 // how big pixels are (dots per meter).
 315                                 if ( $chunk_size !== 9 ) {
 316                                         throw new Exception( __METHOD__ . ": pHYs wrong size" );
 317                                 }
 318
 319                                 $buf = self::read( $fh, $chunk_size );
 320                                 if ( !$buf ) {
 321                                         throw new Exception( __METHOD__ . ": Read error" );
 322                                 }
 323
 324                                 $dim = unpack( "Nwidth/Nheight/Cunit", $buf );
 325                                 if ( $dim['unit'] == 1 ) {
 326                                         // unit is meters
 327                                         // (as opposed to 0 = undefined )
 328                                         $text['XResolution'] = $dim['width']
 329                                                 . '/100';
 330                                         $text['YResolution'] = $dim['height']
 331                                                 . '/100';
 332                                         $text['ResolutionUnit'] = 3;
 333                                         // 3 = dots per cm (from Exif).
 334                                 }
 335
 336                         } elseif ( $chunk_type == "IEND" ) {
 337                                 break;
 338                         } else {
 339                                 fseek( $fh, $chunk_size, SEEK_CUR );
 340                         }
 341                         fseek( $fh, self::$CRC_size, SEEK_CUR );
 342                 }
 343                 fclose( $fh );
 344
 345                 if ( $loopCount > 1 ) {
 346                         $duration *= $loopCount;
 347                 }
 348
 349                 if ( isset( $text['DateTimeDigitized'] ) ) {
 350                         // Convert date format from rfc2822 to exif.
 351                         foreach ( $text['DateTimeDigitized'] as $name => &$value ) {
 352                                 if ( $name === '_type' ) {
 353                                         continue;
 354                                 }
 355
 356                                 // fixme: currently timezones are ignored.
 357                                 // possibly should be wfTimestamp's
 358                                 // responsibility. (at least for numeric TZ)
 359                                 $formatted = wfTimestamp( TS_EXIF, $value );
 360                                 if ( $formatted ) {
 361                                         // Only change if we could convert the
 362                                         // date.
 363                                         // The png standard says it should be
 364                                         // in rfc2822 format, but not required.
 365                                         // In general for the exif stuff we
 366                                         // prettify the date if we can, but we
 367                                         // display as-is if we cannot or if
 368                                         // it is invalid.
 369                                         // So do the same here.
 370
 371                                         $value = $formatted;
 372                                 }
 373                         }
 374                 }
 375                 return array(
 376                         'frameCount' => $frameCount,
 377                         'loopCount' => $loopCount,
 378                         'duration' => $duration,
 379                         'text' => $text,
 380                         'bitDepth' => $bitDepth,
 381                         'colorType' => $colorType,
 382                 );
 383
 384         }
 385         /**
 386          * Read a chunk, checking to make sure its not too big.
 387          *
 388          * @param $fh resource The file handle
 389          * @param $size Integer size in bytes.
 390          * @throws Exception if too big.
 391          * @return String The chunk.
 392          */
 393         static private function read( $fh, $size ) {
 394                 if ( $size > self::MAX_CHUNK_SIZE ) {
 395                         throw new Exception( __METHOD__ . ': Chunk size of ' . $size .
 396                                 ' too big. Max size is: ' . self::MAX_CHUNK_SIZE );
 397                 }
 398                 return fread( $fh, $size );
 399         }
 400 }