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