MediaWiki
REL1_22
|
00001 <?php 00034 class GIFMetadataExtractor { 00035 static $gif_frame_sep; 00036 static $gif_extension_sep; 00037 static $gif_term; 00038 00039 const VERSION = 1; 00040 00041 // Each sub-block is less than or equal to 255 bytes. 00042 // Most of the time its 255 bytes, except for in XMP 00043 // blocks, where it's usually between 32-127 bytes each. 00044 const MAX_SUBBLOCKS = 262144; // 5mb divided by 20. 00045 00051 static function getMetadata( $filename ) { 00052 self::$gif_frame_sep = pack( "C", ord( "," ) ); 00053 self::$gif_extension_sep = pack( "C", ord( "!" ) ); 00054 self::$gif_term = pack( "C", ord( ";" ) ); 00055 00056 $frameCount = 0; 00057 $duration = 0.0; 00058 $isLooped = false; 00059 $xmp = ""; 00060 $comment = array(); 00061 00062 if ( !$filename ) { 00063 throw new Exception( "No file name specified" ); 00064 } elseif ( !file_exists( $filename ) || is_dir( $filename ) ) { 00065 throw new Exception( "File $filename does not exist" ); 00066 } 00067 00068 $fh = fopen( $filename, 'rb' ); 00069 00070 if ( !$fh ) { 00071 throw new Exception( "Unable to open file $filename" ); 00072 } 00073 00074 // Check for the GIF header 00075 $buf = fread( $fh, 6 ); 00076 if ( !( $buf == 'GIF87a' || $buf == 'GIF89a' ) ) { 00077 throw new Exception( "Not a valid GIF file; header: $buf" ); 00078 } 00079 00080 // Skip over width and height. 00081 fread( $fh, 4 ); 00082 00083 // Read BPP 00084 $buf = fread( $fh, 1 ); 00085 $bpp = self::decodeBPP( $buf ); 00086 00087 // Skip over background and aspect ratio 00088 fread( $fh, 2 ); 00089 00090 // Skip over the GCT 00091 self::readGCT( $fh, $bpp ); 00092 00093 while ( !feof( $fh ) ) { 00094 $buf = fread( $fh, 1 ); 00095 00096 if ( $buf == self::$gif_frame_sep ) { 00097 // Found a frame 00098 $frameCount++; 00099 00100 ## Skip bounding box 00101 fread( $fh, 8 ); 00102 00103 ## Read BPP 00104 $buf = fread( $fh, 1 ); 00105 $bpp = self::decodeBPP( $buf ); 00106 00107 ## Read GCT 00108 self::readGCT( $fh, $bpp ); 00109 fread( $fh, 1 ); 00110 self::skipBlock( $fh ); 00111 } elseif ( $buf == self::$gif_extension_sep ) { 00112 $buf = fread( $fh, 1 ); 00113 if ( strlen( $buf ) < 1 ) { 00114 throw new Exception( "Ran out of input" ); 00115 } 00116 $extension_code = unpack( 'C', $buf ); 00117 $extension_code = $extension_code[1]; 00118 00119 if ( $extension_code == 0xF9 ) { 00120 // Graphics Control Extension. 00121 fread( $fh, 1 ); // Block size 00122 00123 fread( $fh, 1 ); // Transparency, disposal method, user input 00124 00125 $buf = fread( $fh, 2 ); // Delay, in hundredths of seconds. 00126 if ( strlen( $buf ) < 2 ) { 00127 throw new Exception( "Ran out of input" ); 00128 } 00129 $delay = unpack( 'v', $buf ); 00130 $delay = $delay[1]; 00131 $duration += $delay * 0.01; 00132 00133 fread( $fh, 1 ); // Transparent colour index 00134 00135 $term = fread( $fh, 1 ); // Should be a terminator 00136 if ( strlen( $term ) < 1 ) { 00137 throw new Exception( "Ran out of input" ); 00138 } 00139 $term = unpack( 'C', $term ); 00140 $term = $term[1]; 00141 if ( $term != 0 ) { 00142 throw new Exception( "Malformed Graphics Control Extension block" ); 00143 } 00144 } elseif ( $extension_code == 0xFE ) { 00145 // Comment block(s). 00146 $data = self::readBlock( $fh ); 00147 if ( $data === "" ) { 00148 throw new Exception( 'Read error, zero-length comment block' ); 00149 } 00150 00151 // The standard says this should be ASCII, however its unclear if 00152 // thats true in practise. Check to see if its valid utf-8, if so 00153 // assume its that, otherwise assume its windows-1252 (iso-8859-1) 00154 $dataCopy = $data; 00155 // quickIsNFCVerify has the side effect of replacing any invalid characters 00156 UtfNormal::quickIsNFCVerify( $dataCopy ); 00157 00158 if ( $dataCopy !== $data ) { 00159 wfSuppressWarnings(); 00160 $data = iconv( 'windows-1252', 'UTF-8', $data ); 00161 wfRestoreWarnings(); 00162 } 00163 00164 $commentCount = count( $comment ); 00165 if ( $commentCount === 0 00166 || $comment[$commentCount - 1] !== $data ) 00167 { 00168 // Some applications repeat the same comment on each 00169 // frame of an animated GIF image, so if this comment 00170 // is identical to the last, only extract once. 00171 $comment[] = $data; 00172 } 00173 } elseif ( $extension_code == 0xFF ) { 00174 // Application extension (Netscape info about the animated gif) 00175 // or XMP (or theoretically any other type of extension block) 00176 $blockLength = fread( $fh, 1 ); 00177 if ( strlen( $blockLength ) < 1 ) { 00178 throw new Exception( "Ran out of input" ); 00179 } 00180 $blockLength = unpack( 'C', $blockLength ); 00181 $blockLength = $blockLength[1]; 00182 $data = fread( $fh, $blockLength ); 00183 00184 if ( $blockLength != 11 ) { 00185 wfDebug( __METHOD__ . " GIF application block with wrong length\n" ); 00186 fseek( $fh, -( $blockLength + 1 ), SEEK_CUR ); 00187 self::skipBlock( $fh ); 00188 continue; 00189 } 00190 00191 // NETSCAPE2.0 (application name for animated gif) 00192 if ( $data == 'NETSCAPE2.0' ) { 00193 $data = fread( $fh, 2 ); // Block length and introduction, should be 03 01 00194 00195 if ( $data != "\x03\x01" ) { 00196 throw new Exception( "Expected \x03\x01, got $data" ); 00197 } 00198 00199 // Unsigned little-endian integer, loop count or zero for "forever" 00200 $loopData = fread( $fh, 2 ); 00201 if ( strlen( $loopData ) < 2 ) { 00202 throw new Exception( "Ran out of input" ); 00203 } 00204 $loopData = unpack( 'v', $loopData ); 00205 $loopCount = $loopData[1]; 00206 00207 if ( $loopCount != 1 ) { 00208 $isLooped = true; 00209 } 00210 00211 // Read out terminator byte 00212 fread( $fh, 1 ); 00213 } elseif ( $data == 'XMP DataXMP' ) { 00214 // application name for XMP data. 00215 // see pg 18 of XMP spec part 3. 00216 00217 $xmp = self::readBlock( $fh, true ); 00218 00219 if ( substr( $xmp, -257, 3 ) !== "\x01\xFF\xFE" 00220 || substr( $xmp, -4 ) !== "\x03\x02\x01\x00" ) 00221 { 00222 // this is just a sanity check. 00223 throw new Exception( "XMP does not have magic trailer!" ); 00224 } 00225 00226 // strip out trailer. 00227 $xmp = substr( $xmp, 0, -257 ); 00228 00229 } else { 00230 // unrecognized extension block 00231 fseek( $fh, -( $blockLength + 1 ), SEEK_CUR ); 00232 self::skipBlock( $fh ); 00233 continue; 00234 } 00235 } else { 00236 self::skipBlock( $fh ); 00237 } 00238 } elseif ( $buf == self::$gif_term ) { 00239 break; 00240 } else { 00241 if ( strlen( $buf ) < 1 ) { 00242 throw new Exception( "Ran out of input" ); 00243 } 00244 $byte = unpack( 'C', $buf ); 00245 $byte = $byte[1]; 00246 throw new Exception( "At position: " . ftell( $fh ) . ", Unknown byte " . $byte ); 00247 } 00248 } 00249 00250 return array( 00251 'frameCount' => $frameCount, 00252 'looped' => $isLooped, 00253 'duration' => $duration, 00254 'xmp' => $xmp, 00255 'comment' => $comment, 00256 ); 00257 } 00258 00264 static function readGCT( $fh, $bpp ) { 00265 if ( $bpp > 0 ) { 00266 for ( $i = 1; $i <= pow( 2, $bpp ); ++$i ) { 00267 fread( $fh, 3 ); 00268 } 00269 } 00270 } 00271 00277 static function decodeBPP( $data ) { 00278 if ( strlen( $data ) < 1 ) { 00279 throw new Exception( "Ran out of input" ); 00280 } 00281 $buf = unpack( 'C', $data ); 00282 $buf = $buf[1]; 00283 $bpp = ( $buf & 7 ) + 1; 00284 $buf >>= 7; 00285 00286 $have_map = $buf & 1; 00287 00288 return $have_map ? $bpp : 0; 00289 } 00290 00295 static function skipBlock( $fh ) { 00296 while ( !feof( $fh ) ) { 00297 $buf = fread( $fh, 1 ); 00298 if ( strlen( $buf ) < 1 ) { 00299 throw new Exception( "Ran out of input" ); 00300 } 00301 $block_len = unpack( 'C', $buf ); 00302 $block_len = $block_len[1]; 00303 if ( $block_len == 0 ) { 00304 return; 00305 } 00306 fread( $fh, $block_len ); 00307 } 00308 } 00309 00324 static function readBlock( $fh, $includeLengths = false ) { 00325 $data = ''; 00326 $subLength = fread( $fh, 1 ); 00327 $blocks = 0; 00328 00329 while ( $subLength !== "\0" ) { 00330 $blocks++; 00331 if ( $blocks > self::MAX_SUBBLOCKS ) { 00332 throw new Exception( "MAX_SUBBLOCKS exceeded (over $blocks sub-blocks)" ); 00333 } 00334 if ( feof( $fh ) ) { 00335 throw new Exception( "Read error: Unexpected EOF." ); 00336 } 00337 if ( $includeLengths ) { 00338 $data .= $subLength; 00339 } 00340 00341 $data .= fread( $fh, ord( $subLength ) ); 00342 $subLength = fread( $fh, 1 ); 00343 } 00344 return $data; 00345 } 00346 00347 }