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