includes/media/GIFMetadataExtractor.php

   1 <?php
   2 /**
   3  * GIF frame counter.
   4  *
   5  * Originally written in Perl by Steve Sanbeg.
   6  * Ported to PHP by Andrew Garrett
   7  * Deliberately not using MWExceptions to avoid external dependencies, encouraging
   8  * redistribution.
   9  *
  10  * @file
  11  * @ingroup Media
  12  */
  13
  14 /**
  15  * GIF frame counter.
  16  *
  17  * @ingroup Media
  18  */
  19 class GIFMetadataExtractor {
  20         static $gif_frame_sep;
  21         static $gif_extension_sep;
  22         static $gif_term;
  23
  24         const VERSION = 1;
  25
  26         // Each sub-block is less than or equal to 255 bytes.
  27         // Most of the time its 255 bytes, except for in XMP
  28         // blocks, where it's usually between 32-127 bytes each.
  29         const MAX_SUBBLOCKS = 262144; // 5mb divided by 20.
  30
  31         /**
  32          * @throws Exception
  33          * @param $filename string
  34          * @return array
  35          */
  36         static function getMetadata( $filename ) {
  37                 self::$gif_frame_sep = pack( "C", ord("," ) );
  38                 self::$gif_extension_sep = pack( "C", ord("!" ) );
  39                 self::$gif_term = pack( "C", ord(";" ) );
  40
  41                 $frameCount = 0;
  42                 $duration = 0.0;
  43                 $isLooped = false;
  44                 $xmp = "";
  45                 $comment = array();
  46
  47                 if ( !$filename ) {
  48                         throw new Exception( "No file name specified" );
  49                 } elseif ( !file_exists( $filename ) || is_dir( $filename ) ) {
  50                         throw new Exception( "File $filename does not exist" );
  51                 }
  52
  53                 $fh = fopen( $filename, 'r' );
  54
  55                 if ( !$fh ) {
  56                         throw new Exception( "Unable to open file $filename" );
  57                 }
  58
  59                 // Check for the GIF header
  60                 $buf = fread( $fh, 6 );
  61                 if ( !($buf == 'GIF87a' || $buf == 'GIF89a') ) {
  62                         throw new Exception( "Not a valid GIF file; header: $buf" );
  63                 }
  64
  65                 // Skip over width and height.
  66                 fread( $fh, 4 );
  67
  68                 // Read BPP
  69                 $buf = fread( $fh, 1 );
  70                 $bpp = self::decodeBPP( $buf );
  71
  72                 // Skip over background and aspect ratio
  73                 fread( $fh, 2 );
  74
  75                 // Skip over the GCT
  76                 self::readGCT( $fh, $bpp );
  77
  78                 while( !feof( $fh ) ) {
  79                         $buf = fread( $fh, 1 );
  80
  81                         if ($buf == self::$gif_frame_sep) {
  82                                 // Found a frame
  83                                 $frameCount++;
  84
  85                                 ## Skip bounding box
  86                                 fread( $fh, 8 );
  87
  88                                 ## Read BPP
  89                                 $buf = fread( $fh, 1 );
  90                                 $bpp = self::decodeBPP( $buf );
  91
  92                                 ## Read GCT
  93                                 self::readGCT( $fh, $bpp );
  94                                 fread( $fh, 1 );
  95                                 self::skipBlock( $fh );
  96                         } elseif ( $buf == self::$gif_extension_sep ) {
  97                                 $buf = fread( $fh, 1 );
  98                                 $extension_code = unpack( 'C', $buf );
  99                                 $extension_code = $extension_code[1];
 100
 101                                 if ($extension_code == 0xF9) {
 102                                         // Graphics Control Extension.
 103                                         fread( $fh, 1 ); // Block size
 104
 105                                         fread( $fh, 1 ); // Transparency, disposal method, user input
 106
 107                                         $buf = fread( $fh, 2 ); // Delay, in hundredths of seconds.
 108                                         $delay = unpack( 'v', $buf );
 109                                         $delay = $delay[1];
 110                                         $duration += $delay * 0.01;
 111
 112                                         fread( $fh, 1 ); // Transparent colour index
 113
 114                                         $term = fread( $fh, 1 ); // Should be a terminator
 115                                         $term = unpack( 'C', $term );
 116                                         $term = $term[1];
 117                                         if ($term != 0 ) {
 118                                                 throw new Exception( "Malformed Graphics Control Extension block" );
 119                                         }
 120                                 } elseif ($extension_code == 0xFE) {
 121                                         // Comment block(s).
 122                                         $data = self::readBlock( $fh );
 123                                         if ( $data === "" ) {
 124                                                 throw new Exception( 'Read error, zero-length comment block' );
 125                                         }
 126
 127                                         // The standard says this should be ASCII, however its unclear if
 128                                         // thats true in practise. Check to see if its valid utf-8, if so
 129                                         // assume its that, otherwise assume its iso-8859-1
 130                                         $dataCopy = $data;
 131                                         // quickIsNFCVerify has the side effect of replacing any invalid characters
 132                                         UtfNormal::quickIsNFCVerify( $dataCopy );
 133
 134                                         if ( $dataCopy !== $data ) {
 135                                                 wfSuppressWarnings();
 136                                                 $data = iconv( 'ISO-8859-1', 'UTF-8', $data );
 137                                                 wfRestoreWarnings();
 138                                         }
 139
 140                                         $commentCount = count( $comment );
 141                                         if ( $commentCount === 0
 142                                                 || $comment[$commentCount-1] !== $data )
 143                                         {
 144                                                 // Some applications repeat the same comment on each
 145                                                 // frame of an animated GIF image, so if this comment
 146                                                 // is identical to the last, only extract once.
 147                                                 $comment[] = $data;
 148                                         }
 149                                 } elseif ($extension_code == 0xFF) {
 150                                         // Application extension (Netscape info about the animated gif)
 151                                         // or XMP (or theoretically any other type of extension block)
 152                                         $blockLength = fread( $fh, 1 );
 153                                         $blockLength = unpack( 'C', $blockLength );
 154                                         $blockLength = $blockLength[1];
 155                                         $data = fread( $fh, $blockLength );
 156
 157                                         if ($blockLength != 11 ) {
 158                                                 wfDebug( __METHOD__ . ' GIF application block with wrong length' );
 159                                                 fseek( $fh, -($blockLength + 1), SEEK_CUR );
 160                                                 self::skipBlock( $fh );
 161                                                 continue;
 162                                         }
 163
 164                                         // NETSCAPE2.0 (application name for animated gif)
 165                                         if ( $data == 'NETSCAPE2.0' ) {
 166
 167                                                 $data = fread( $fh, 2 ); // Block length and introduction, should be 03 01
 168
 169                                                 if ($data != "\x03\x01") {
 170                                                         throw new Exception( "Expected \x03\x01, got $data" );
 171                                                 }
 172
 173                                                 // Unsigned little-endian integer, loop count or zero for "forever"
 174                                                 $loopData = fread( $fh, 2 );
 175                                                 $loopData = unpack( 'v', $loopData );
 176                                                 $loopCount = $loopData[1];
 177
 178                                                 if ($loopCount != 1) {
 179                                                         $isLooped = true;
 180                                                 }
 181
 182                                                 // Read out terminator byte
 183                                                 fread( $fh, 1 );
 184                                         } elseif ( $data == 'XMP DataXMP' ) {
 185                                                 // application name for XMP data.
 186                                                 // see pg 18 of XMP spec part 3.
 187
 188                                                 $xmp = self::readBlock( $fh, true );
 189
 190                                                 if ( substr( $xmp, -257, 3 ) !== "\x01\xFF\xFE"
 191                                                         || substr( $xmp, -4 ) !== "\x03\x02\x01\x00" )
 192                                                 {
 193                                                         // this is just a sanity check.
 194                                                         throw new Exception( "XMP does not have magic trailer!" );
 195                                                 }
 196
 197                                                 // strip out trailer.
 198                                                 $xmp = substr( $xmp, 0, -257 );
 199
 200                                         } else {
 201                                                 // unrecognized extension block
 202                                                 fseek( $fh, -($blockLength + 1), SEEK_CUR );
 203                                                 self::skipBlock( $fh );
 204                                                 continue;
 205                                         }
 206                                 } else {
 207                                         self::skipBlock( $fh );
 208                                 }
 209                         } elseif ( $buf == self::$gif_term ) {
 210                                 break;
 211                         } else {
 212                                 $byte = unpack( 'C', $buf );
 213                                 $byte = $byte[1];
 214                                 throw new Exception( "At position: ".ftell($fh). ", Unknown byte ".$byte );
 215                         }
 216                 }
 217
 218                 return array(
 219                         'frameCount' => $frameCount,
 220                         'looped' => $isLooped,
 221                         'duration' => $duration,
 222                         'xmp' => $xmp,
 223                         'comment' => $comment,
 224                 );
 225         }
 226
 227         /**
 228          * @param $fh
 229          * @param $bpp
 230          * @return void
 231          */
 232         static function readGCT( $fh, $bpp ) {
 233                 if ( $bpp > 0 ) {
 234                         for( $i=1; $i<=pow( 2, $bpp ); ++$i ) {
 235                                 fread( $fh, 3 );
 236                         }
 237                 }
 238         }
 239
 240         /**
 241          * @param $data
 242          * @return int
 243          */
 244         static function decodeBPP( $data ) {
 245                 $buf = unpack( 'C', $data );
 246                 $buf = $buf[1];
 247                 $bpp = ( $buf & 7 ) + 1;
 248                 $buf >>= 7;
 249
 250                 $have_map = $buf & 1;
 251
 252                 return $have_map ? $bpp : 0;
 253         }
 254
 255         /**
 256          * @param $fh
 257          * @return
 258          */
 259         static function skipBlock( $fh ) {
 260                 while ( !feof( $fh ) ) {
 261                         $buf = fread( $fh, 1 );
 262                         $block_len = unpack( 'C', $buf );
 263                         $block_len = $block_len[1];
 264                         if ($block_len == 0) {
 265                                 return;
 266                         }
 267                         fread( $fh, $block_len );
 268                 }
 269         }
 270         /**
 271          * Read a block. In the GIF format, a block is made up of
 272          * several sub-blocks. Each sub block starts with one byte
 273          * saying how long the sub-block is, followed by the sub-block.
 274          * The entire block is terminated by a sub-block of length
 275          * 0.
 276          * @param $fh FileHandle
 277          * @param $includeLengths Boolean Include the length bytes of the
 278          *  sub-blocks in the returned value. Normally this is false,
 279          *  except XMP is weird and does a hack where you need to keep
 280          *  these length bytes.
 281          * @return The data.
 282          */
 283         static function readBlock( $fh, $includeLengths = false ) {
 284                 $data = '';
 285                 $subLength = fread( $fh, 1 );
 286                 $blocks = 0;
 287
 288                 while( $subLength !== "\0" ) {
 289                         $blocks++;
 290                         if ( $blocks > self::MAX_SUBBLOCKS ) {
 291                                 throw new Exception( "MAX_SUBBLOCKS exceeded (over $blocks sub-blocks)" );
 292                         }
 293                         if ( feof( $fh ) ) {
 294                                 throw new Exception( "Read error: Unexpected EOF." );
 295                         }
 296                         if ( $includeLengths ) {
 297                                 $data .= $subLength;
 298                         }
 299
 300                         $data .= fread( $fh, ord( $subLength ) );
 301                         $subLength = fread( $fh, 1 );
 302                 }
 303                 return $data;
 304         }
 305
 306 }