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( $filename, 'rb' );
  70
  71                 if ( !$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 || strlen( $buf ) < 4 ) {
  85                                 throw new Exception( __METHOD__ . ": Read error" );
  86                         }
  87                         $chunk_size = unpack( "N", $buf );
  88                         $chunk_size = $chunk_size[1];
  89
  90                         if ( $chunk_size < 0 ) {
  91                                 throw new Exception( __METHOD__ . ": Chunk size too big for unpack" );
  92                         }
  93
  94                         $chunk_type = fread( $fh, 4 );
  95                         if ( !$chunk_type || strlen( $chunk_type ) < 4 ) {
  96                                 throw new Exception( __METHOD__ . ": Read error" );
  97                         }
  98
  99                         if ( $chunk_type == "IHDR" ) {
 100                                 $buf = self::read( $fh, $chunk_size );
 101                                 if ( !$buf || strlen( $buf ) < $chunk_size ) {
 102                                         throw new Exception( __METHOD__ . ": Read error" );
 103                                 }
 104                                 $bitDepth = ord( substr( $buf, 8, 1 ) );
 105                                 // Detect the color type in British English as per the spec
 106                                 // http://www.w3.org/TR/PNG/#11IHDR
 107                                 switch ( ord( substr( $buf, 9, 1 ) ) ) {
 108                                         case 0:
 109                                                 $colorType = 'greyscale';
 110                                                 break;
 111                                         case 2:
 112                                                 $colorType = 'truecolour';
 113                                                 break;
 114                                         case 3:
 115                                                 $colorType = 'index-coloured';
 116                                                 break;
 117                                         case 4:
 118                                                 $colorType = 'greyscale-alpha';
 119                                                 break;
 120                                         case 6:
 121                                                 $colorType = 'truecolour-alpha';
 122                                                 break;
 123                                         default:
 124                                                 $colorType = 'unknown';
 125                                                 break;
 126                                 }
 127                         } elseif ( $chunk_type == "acTL" ) {
 128                                 $buf = fread( $fh, $chunk_size );
 129                                 if( !$buf || strlen( $buf ) < $chunk_size || $chunk_size < 4 ) {
 130                                         throw new Exception( __METHOD__ . ": Read error" );
 131                                 }
 132
 133                                 $actl = unpack( "Nframes/Nplays", $buf );
 134                                 $frameCount = $actl['frames'];
 135                                 $loopCount = $actl['plays'];
 136                         } elseif ( $chunk_type == "fcTL" ) {
 137                                 $buf = self::read( $fh, $chunk_size );
 138                                 if ( !$buf || strlen( $buf ) < $chunk_size ) {
 139                                         throw new Exception( __METHOD__ . ": Read error" );
 140                                 }
 141                                 $buf = substr( $buf, 20 );
 142                                 if ( strlen( $buf ) < 4 ) {
 143                                         throw new Exception( __METHOD__ . ": Read error" );
 144                                 }
 145
 146                                 $fctldur = unpack( "ndelay_num/ndelay_den", $buf );
 147                                 if ( $fctldur['delay_den'] == 0 ) {
 148                                         $fctldur['delay_den'] = 100;
 149                                 }
 150                                 if ( $fctldur['delay_num'] ) {
 151                                         $duration += $fctldur['delay_num'] / $fctldur['delay_den'];
 152                                 }
 153                         } elseif ( $chunk_type == "iTXt" ) {
 154                                 // Extracts iTXt chunks, uncompressing if necessary.
 155                                 $buf = self::read( $fh, $chunk_size );
 156                                 $items = array();
 157                                 if ( preg_match(
 158                                         '/^([^\x00]{1,79})\x00(\x00|\x01)\x00([^\x00]*)(.)[^\x00]*\x00(.*)$/Ds',
 159                                         $buf, $items )
 160                                 ) {
 161                                         /* $items[1] = text chunk name, $items[2] = compressed flag,
 162                                          * $items[3] = lang code (or ""), $items[4]= compression type.
 163                                          * $items[5] = content
 164                                          */
 165
 166                                         // Theoretically should be case-sensitive, but in practise...
 167                                         $items[1] = strtolower( $items[1] );
 168                                         if ( !isset( self::$text_chunks[$items[1]] ) ) {
 169                                                 // Only extract textual chunks on our list.
 170                                                 fseek( $fh, self::$CRC_size, SEEK_CUR );
 171                                                 continue;
 172                                         }
 173
 174                                         $items[3] = strtolower( $items[3] );
 175                                         if ( $items[3] == '' ) {
 176                                                 // if no lang specified use x-default like in xmp.
 177                                                 $items[3] = 'x-default';
 178                                         }
 179
 180                                         // if compressed
 181                                         if ( $items[2] == "\x01" ) {
 182                                                 if ( function_exists( 'gzuncompress' ) && $items[4] === "\x00" ) {
 183                                                         wfSuppressWarnings();
 184                                                         $items[5] = gzuncompress( $items[5] );
 185                                                         wfRestoreWarnings();
 186
 187                                                         if ( $items[5] === false ) {
 188                                                                 // decompression failed
 189                                                                 wfDebug( __METHOD__ . ' Error decompressing iTxt chunk - ' . $items[1] );
 190                                                                 fseek( $fh, self::$CRC_size, SEEK_CUR );
 191                                                                 continue;
 192                                                         }
 193
 194                                                 } else {
 195                                                         wfDebug( __METHOD__ . ' Skipping compressed png iTXt chunk due to lack of zlib,'
 196                                                                 . ' or potentially invalid compression method' );
 197                                                         fseek( $fh, self::$CRC_size, SEEK_CUR );
 198                                                         continue;
 199                                                 }
 200                                         }
 201                                         $finalKeyword = self::$text_chunks[ $items[1] ];
 202                                         $text[ $finalKeyword ][ $items[3] ] = $items[5];
 203                                         $text[ $finalKeyword ]['_type'] = 'lang';
 204
 205                                 } else {
 206                                         // Error reading iTXt chunk
 207                                         throw new Exception( __METHOD__ . ": Read error on iTXt chunk" );
 208                                 }
 209
 210                         } elseif ( $chunk_type == 'tEXt' ) {
 211                                 $buf = self::read( $fh, $chunk_size );
 212
 213                                 // In case there is no \x00 which will make explode fail.
 214                                 if ( strpos( $buf, "\x00" ) === false ) {
 215                                         throw new Exception( __METHOD__ . ": Read error on tEXt chunk" );
 216                                 }
 217
 218                                 list( $keyword, $content ) = explode( "\x00", $buf, 2 );
 219                                 if ( $keyword === '' || $content === '' ) {
 220                                         throw new Exception( __METHOD__ . ": Read error on tEXt chunk" );
 221                                 }
 222
 223                                 // Theoretically should be case-sensitive, but in practise...
 224                                 $keyword = strtolower( $keyword );
 225                                 if ( !isset( self::$text_chunks[ $keyword ] ) ) {
 226                                         // Don't recognize chunk, so skip.
 227                                         fseek( $fh, self::$CRC_size, SEEK_CUR );
 228                                         continue;
 229                                 }
 230                                 wfSuppressWarnings();
 231                                 $content = iconv( 'ISO-8859-1', 'UTF-8', $content );
 232                                 wfRestoreWarnings();
 233
 234                                 if ( $content === false ) {
 235                                         throw new Exception( __METHOD__ . ": Read error (error with iconv)" );
 236                                 }
 237
 238                                 $finalKeyword = self::$text_chunks[ $keyword ];
 239                                 $text[ $finalKeyword ][ 'x-default' ] = $content;
 240                                 $text[ $finalKeyword ]['_type'] = 'lang';
 241
 242                         } elseif ( $chunk_type == 'zTXt' ) {
 243                                 if ( function_exists( 'gzuncompress' ) ) {
 244                                         $buf = self::read( $fh, $chunk_size );
 245
 246                                         // In case there is no \x00 which will make explode fail.
 247                                         if ( strpos( $buf, "\x00" ) === false ) {
 248                                                 throw new Exception( __METHOD__ . ": Read error on zTXt chunk" );
 249                                         }
 250
 251                                         list( $keyword, $postKeyword ) = explode( "\x00", $buf, 2 );
 252                                         if ( $keyword === '' || $postKeyword === '' ) {
 253                                                 throw new Exception( __METHOD__ . ": Read error on zTXt chunk" );
 254                                         }
 255                                         // Theoretically should be case-sensitive, but in practise...
 256                                         $keyword = strtolower( $keyword );
 257
 258                                         if ( !isset( self::$text_chunks[ $keyword ] ) ) {
 259                                                 // Don't recognize chunk, so skip.
 260                                                 fseek( $fh, self::$CRC_size, SEEK_CUR );
 261                                                 continue;
 262                                         }
 263                                         $compression = substr( $postKeyword, 0, 1 );
 264                                         $content = substr( $postKeyword, 1 );
 265                                         if ( $compression !== "\x00" ) {
 266                                                 wfDebug( __METHOD__ . " Unrecognized compression method in zTXt ($keyword). Skipping." );
 267                                                 fseek( $fh, self::$CRC_size, SEEK_CUR );
 268                                                 continue;
 269                                         }
 270
 271                                         wfSuppressWarnings();
 272                                         $content = gzuncompress( $content );
 273                                         wfRestoreWarnings();
 274
 275                                         if ( $content === false ) {
 276                                                 // decompression failed
 277                                                 wfDebug( __METHOD__ . ' Error decompressing zTXt chunk - ' . $keyword );
 278                                                 fseek( $fh, self::$CRC_size, SEEK_CUR );
 279                                                 continue;
 280                                         }
 281
 282                                         wfSuppressWarnings();
 283                                         $content = iconv( 'ISO-8859-1', 'UTF-8', $content );
 284                                         wfRestoreWarnings();
 285
 286                                         if ( $content === false ) {
 287                                                 throw new Exception( __METHOD__ . ": Read error (error with iconv)" );
 288                                         }
 289
 290                                         $finalKeyword = self::$text_chunks[ $keyword ];
 291                                         $text[ $finalKeyword ][ 'x-default' ] = $content;
 292                                         $text[ $finalKeyword ]['_type'] = 'lang';
 293
 294                                 } else {
 295                                         wfDebug( __METHOD__ . " Cannot decompress zTXt chunk due to lack of zlib. Skipping." );
 296                                         fseek( $fh, $chunk_size, SEEK_CUR );
 297                                 }
 298                         } elseif ( $chunk_type == 'tIME' ) {
 299                                 // last mod timestamp.
 300                                 if ( $chunk_size !== 7 ) {
 301                                         throw new Exception( __METHOD__ . ": tIME wrong size" );
 302                                 }
 303                                 $buf = self::read( $fh, $chunk_size );
 304                                 if ( !$buf || strlen( $buf ) < $chunk_size ) {
 305                                         throw new Exception( __METHOD__ . ": Read error" );
 306                                 }
 307
 308                                 // Note: spec says this should be UTC.
 309                                 $t = unpack( "ny/Cm/Cd/Ch/Cmin/Cs", $buf );
 310                                 $strTime = sprintf( "%04d%02d%02d%02d%02d%02d",
 311                                         $t['y'], $t['m'], $t['d'], $t['h'],
 312                                         $t['min'], $t['s'] );
 313
 314                                 $exifTime = wfTimestamp( TS_EXIF, $strTime );
 315
 316                                 if ( $exifTime ) {
 317                                         $text['DateTime'] = $exifTime;
 318                                 }
 319
 320                         } elseif ( $chunk_type == 'pHYs' ) {
 321                                 // how big pixels are (dots per meter).
 322                                 if ( $chunk_size !== 9 ) {
 323                                         throw new Exception( __METHOD__ . ": pHYs wrong size" );
 324                                 }
 325
 326                                 $buf = self::read( $fh, $chunk_size );
 327                                 if ( !$buf || strlen( $buf ) < $chunk_size ) {
 328                                         throw new Exception( __METHOD__ . ": Read error" );
 329                                 }
 330
 331                                 $dim = unpack( "Nwidth/Nheight/Cunit", $buf );
 332                                 if ( $dim['unit'] == 1 ) {
 333                                         // Need to check for negative because php
 334                                         // doesn't deal with super-large unsigned 32-bit ints well
 335                                         if ( $dim['width'] > 0 && $dim['height'] > 0 ) {
 336                                                 // unit is meters
 337                                                 // (as opposed to 0 = undefined )
 338                                                 $text['XResolution'] = $dim['width']
 339                                                         . '/100';
 340                                                 $text['YResolution'] = $dim['height']
 341                                                         . '/100';
 342                                                 $text['ResolutionUnit'] = 3;
 343                                                 // 3 = dots per cm (from Exif).
 344                                         }
 345                                 }
 346
 347                         } elseif ( $chunk_type == "IEND" ) {
 348                                 break;
 349                         } else {
 350                                 fseek( $fh, $chunk_size, SEEK_CUR );
 351                         }
 352                         fseek( $fh, self::$CRC_size, SEEK_CUR );
 353                 }
 354                 fclose( $fh );
 355
 356                 if ( $loopCount > 1 ) {
 357                         $duration *= $loopCount;
 358                 }
 359
 360                 if ( isset( $text['DateTimeDigitized'] ) ) {
 361                         // Convert date format from rfc2822 to exif.
 362                         foreach ( $text['DateTimeDigitized'] as $name => &$value ) {
 363                                 if ( $name === '_type' ) {
 364                                         continue;
 365                                 }
 366
 367                                 // @todo FIXME: Currently timezones are ignored.
 368                                 // possibly should be wfTimestamp's
 369                                 // responsibility. (at least for numeric TZ)
 370                                 $formatted = wfTimestamp( TS_EXIF, $value );
 371                                 if ( $formatted ) {
 372                                         // Only change if we could convert the
 373                                         // date.
 374                                         // The png standard says it should be
 375                                         // in rfc2822 format, but not required.
 376                                         // In general for the exif stuff we
 377                                         // prettify the date if we can, but we
 378                                         // display as-is if we cannot or if
 379                                         // it is invalid.
 380                                         // So do the same here.
 381
 382                                         $value = $formatted;
 383                                 }
 384                         }
 385                 }
 386                 return array(
 387                         'frameCount' => $frameCount,
 388                         'loopCount' => $loopCount,
 389                         'duration' => $duration,
 390                         'text' => $text,
 391                         'bitDepth' => $bitDepth,
 392                         'colorType' => $colorType,
 393                 );
 394
 395         }
 396         /**
 397          * Read a chunk, checking to make sure its not too big.
 398          *
 399          * @param $fh resource The file handle
 400          * @param $size Integer size in bytes.
 401          * @throws Exception if too big.
 402          * @return String The chunk.
 403          */
 404         static private function read( $fh, $size ) {
 405                 if ( $size > self::MAX_CHUNK_SIZE ) {
 406                         throw new Exception( __METHOD__ . ': Chunk size of ' . $size .
 407                                 ' too big. Max size is: ' . self::MAX_CHUNK_SIZE );
 408                 }
 409                 return fread( $fh, $size );
 410         }
 411 }