includes/MimeMagic.php

   1 <?php
   2 /**
   3  * Module defining helper functions for detecting and dealing with mime types.
   4  *
   5  * @file
   6  */
   7
   8 /**
   9  * Defines a set of well known mime types
  10  * This is used as a fallback to mime.types files.
  11  * An extensive list of well known mime types is provided by
  12  * the file mime.types in the includes directory.
  13  *
  14  * This list concatenated with mime.types is used to create a mime <-> ext
  15  * map. Each line contains a mime type followed by a space separated list of
  16  * extensions. If multiple extensions for a single mime type exist or if
  17  * multiple mime types exist for a single extension then in most cases
  18  * MediaWiki assumes that the first extension following the mime type is the
  19  * canonical extension, and the first time a mime type appears for a certain
  20  * extension is considered the canonical mime type.
  21  *
  22  * (Note that appending $wgMimeTypeFile to the end of MM_WELL_KNOWN_MIME_TYPES
  23  * sucks because you can't redefine canonical types. This could be fixed by
  24  * appending MM_WELL_KNOWN_MIME_TYPES behind $wgMimeTypeFile, but who knows
  25  * what will break? In practice this probably isn't a problem anyway -- Bryan)
  26  */
  27 define('MM_WELL_KNOWN_MIME_TYPES',<<<END_STRING
  28 application/ogg ogx ogg ogm ogv oga spx
  29 application/pdf pdf
  30 application/vnd.oasis.opendocument.chart odc
  31 application/vnd.oasis.opendocument.chart-template otc
  32 application/vnd.oasis.opendocument.formula odf
  33 application/vnd.oasis.opendocument.formula-template otf
  34 application/vnd.oasis.opendocument.graphics odg
  35 application/vnd.oasis.opendocument.graphics-template otg
  36 application/vnd.oasis.opendocument.image odi
  37 application/vnd.oasis.opendocument.image-template oti
  38 application/vnd.oasis.opendocument.presentation odp
  39 application/vnd.oasis.opendocument.presentation-template otp
  40 application/vnd.oasis.opendocument.spreadsheet ods
  41 application/vnd.oasis.opendocument.spreadsheet-template ots
  42 application/vnd.oasis.opendocument.text odt
  43 application/vnd.oasis.opendocument.text-template ott
  44 application/vnd.oasis.opendocument.text-master otm
  45 application/vnd.oasis.opendocument.text-web oth
  46 application/x-javascript js
  47 application/x-shockwave-flash swf
  48 audio/midi mid midi kar
  49 audio/mpeg mpga mpa mp2 mp3
  50 audio/x-aiff aif aiff aifc
  51 audio/x-wav wav
  52 audio/ogg oga spx ogg
  53 image/x-bmp bmp
  54 image/gif gif
  55 image/jpeg jpeg jpg jpe
  56 image/png png
  57 image/svg+xml svg
  58 image/svg svg
  59 image/tiff tiff tif
  60 image/vnd.djvu djvu
  61 image/x.djvu djvu
  62 image/x-djvu djvu
  63 image/x-portable-pixmap ppm
  64 image/x-xcf xcf
  65 text/plain txt
  66 text/html html htm
  67 video/ogg ogv ogm ogg
  68 video/mpeg mpg mpeg
  69 END_STRING
  70 );
  71
  72 /**
  73  * Defines a set of well known mime info entries
  74  * This is used as a fallback to mime.info files.
  75  * An extensive list of well known mime types is provided by
  76  * the file mime.info in the includes directory.
  77  */
  78 define('MM_WELL_KNOWN_MIME_INFO', <<<END_STRING
  79 application/pdf [OFFICE]
  80 application/vnd.oasis.opendocument.chart [OFFICE]
  81 application/vnd.oasis.opendocument.chart-template [OFFICE]
  82 application/vnd.oasis.opendocument.formula [OFFICE]
  83 application/vnd.oasis.opendocument.formula-template [OFFICE]
  84 application/vnd.oasis.opendocument.graphics [OFFICE]
  85 application/vnd.oasis.opendocument.graphics-template [OFFICE]
  86 application/vnd.oasis.opendocument.image [OFFICE]
  87 application/vnd.oasis.opendocument.image-template [OFFICE]
  88 application/vnd.oasis.opendocument.presentation [OFFICE]
  89 application/vnd.oasis.opendocument.presentation-template [OFFICE]
  90 application/vnd.oasis.opendocument.spreadsheet [OFFICE]
  91 application/vnd.oasis.opendocument.spreadsheet-template [OFFICE]
  92 application/vnd.oasis.opendocument.text [OFFICE]
  93 application/vnd.oasis.opendocument.text-template [OFFICE]
  94 application/vnd.oasis.opendocument.text-master [OFFICE]
  95 application/vnd.oasis.opendocument.text-web [OFFICE]
  96 text/javascript application/x-javascript [EXECUTABLE]
  97 application/x-shockwave-flash [MULTIMEDIA]
  98 audio/midi [AUDIO]
  99 audio/x-aiff [AUDIO]
 100 audio/x-wav [AUDIO]
 101 audio/mp3 audio/mpeg [AUDIO]
 102 application/ogg audio/ogg video/ogg [MULTIMEDIA]
 103 image/x-bmp image/x-ms-bmp image/bmp [BITMAP]
 104 image/gif [BITMAP]
 105 image/jpeg [BITMAP]
 106 image/png [BITMAP]
 107 image/svg+xml [DRAWING]
 108 image/tiff [BITMAP]
 109 image/vnd.djvu [BITMAP]
 110 image/x-xcf [BITMAP]
 111 image/x-portable-pixmap [BITMAP]
 112 text/plain [TEXT]
 113 text/html [TEXT]
 114 video/ogg [VIDEO]
 115 video/mpeg [VIDEO]
 116 unknown/unknown application/octet-stream application/x-empty [UNKNOWN]
 117 END_STRING
 118 );
 119
 120 // Note: because this file is possibly included by a function,
 121 // we need to access the global scope explicitely!
 122 global $wgLoadFileinfoExtension;
 123
 124 if ( $wgLoadFileinfoExtension ) {
 125         wfDl( 'fileinfo' );
 126 }
 127
 128 /**
 129  * Implements functions related to mime types such as detection and mapping to
 130  * file extension.
 131  *
 132  * Instances of this class are stateles, there only needs to be one global instance
 133  * of MimeMagic. Please use MimeMagic::singleton() to get that instance.
 134  */
 135 class MimeMagic {
 136
 137         /**
 138         * Mapping of media types to arrays of mime types.
 139         * This is used by findMediaType and getMediaType, respectively
 140         */
 141         var $mMediaTypes = null;
 142
 143         /** Map of mime type aliases
 144         */
 145         var $mMimeTypeAliases = null;
 146
 147         /** map of mime types to file extensions (as a space seprarated list)
 148         */
 149         var $mMimeToExt = null;
 150
 151         /** map of file extensions types to mime types (as a space seprarated list)
 152         */
 153         var $mExtToMime = null;
 154
 155         /** IEContentAnalyzer instance
 156          */
 157         var $mIEAnalyzer;
 158
 159         /** The singleton instance
 160          */
 161         private static $instance;
 162
 163         /** Initializes the MimeMagic object. This is called by MimeMagic::singleton().
 164         *
 165         * This constructor parses the mime.types and mime.info files and build internal mappings.
 166         */
 167         function __construct() {
 168                 /*
 169                 *   --- load mime.types ---
 170                 */
 171
 172                 global $wgMimeTypeFile, $IP;
 173
 174                 $types = MM_WELL_KNOWN_MIME_TYPES;
 175
 176                 if ( $wgMimeTypeFile == 'includes/mime.types' ) {
 177                         $wgMimeTypeFile = "$IP/$wgMimeTypeFile";
 178                 }
 179
 180                 if ( $wgMimeTypeFile ) {
 181                         if ( is_file( $wgMimeTypeFile ) and is_readable( $wgMimeTypeFile ) ) {
 182                                 wfDebug( __METHOD__.": loading mime types from $wgMimeTypeFile\n" );
 183                                 $types .= "\n";
 184                                 $types .= file_get_contents( $wgMimeTypeFile );
 185                         } else {
 186                                 wfDebug( __METHOD__.": can't load mime types from $wgMimeTypeFile\n" );
 187                         }
 188                 } else {
 189                         wfDebug( __METHOD__.": no mime types file defined, using build-ins only.\n" );
 190                 }
 191
 192                 $types = str_replace( array( "\r\n", "\n\r", "\n\n", "\r\r", "\r" ), "\n", $types );
 193                 $types = str_replace( "\t", " ", $types );
 194
 195                 $this->mMimeToExt = array();
 196                 $this->mToMime = array();
 197
 198                 $lines = explode( "\n",$types );
 199                 foreach ( $lines as $s ) {
 200                         $s = trim( $s );
 201                         if ( empty( $s ) ) continue;
 202                         if ( strpos( $s, '#' ) === 0 ) continue;
 203
 204                         $s = strtolower( $s );
 205                         $i = strpos( $s, ' ' );
 206
 207                         if ( $i === false ) continue;
 208
 209                         #print "processing MIME line $s<br>";
 210
 211                         $mime = substr( $s, 0, $i );
 212                         $ext = trim( substr($s, $i+1 ) );
 213
 214                         if ( empty( $ext ) ) continue;
 215
 216                         if ( !empty( $this->mMimeToExt[$mime] ) ) {
 217                                 $this->mMimeToExt[$mime] .= ' ' . $ext;
 218                         } else {
 219                                 $this->mMimeToExt[$mime] = $ext;
 220                         }
 221
 222                         $extensions = explode( ' ', $ext );
 223
 224                         foreach ( $extensions as $e ) {
 225                                 $e = trim( $e );
 226                                 if ( empty( $e ) ) continue;
 227
 228                                 if ( !empty( $this->mExtToMime[$e] ) ) {
 229                                         $this->mExtToMime[$e] .= ' ' . $mime;
 230                                 } else {
 231                                         $this->mExtToMime[$e] = $mime;
 232                                 }
 233                         }
 234                 }
 235
 236                 /*
 237                 *   --- load mime.info ---
 238                 */
 239
 240                 global $wgMimeInfoFile;
 241                 if ( $wgMimeInfoFile == 'includes/mime.info' ) {
 242                         $wgMimeInfoFile = "$IP/$wgMimeInfoFile";
 243                 }
 244
 245                 $info = MM_WELL_KNOWN_MIME_INFO;
 246
 247                 if ( $wgMimeInfoFile ) {
 248                         if ( is_file( $wgMimeInfoFile ) and is_readable( $wgMimeInfoFile ) ) {
 249                                 wfDebug( __METHOD__.": loading mime info from $wgMimeInfoFile\n" );
 250                                 $info .= "\n";
 251                                 $info .= file_get_contents( $wgMimeInfoFile );
 252                         } else {
 253                                 wfDebug(__METHOD__.": can't load mime info from $wgMimeInfoFile\n");
 254                         }
 255                 } else {
 256                         wfDebug(__METHOD__.": no mime info file defined, using build-ins only.\n");
 257                 }
 258
 259                 $info = str_replace( array( "\r\n", "\n\r", "\n\n", "\r\r", "\r" ), "\n", $info);
 260                 $info = str_replace( "\t", " ", $info );
 261
 262                 $this->mMimeTypeAliases = array();
 263                 $this->mMediaTypes = array();
 264
 265                 $lines = explode( "\n", $info );
 266                 foreach ( $lines as $s ) {
 267                         $s = trim( $s );
 268                         if ( empty( $s ) ) continue;
 269                         if ( strpos( $s, '#' ) === 0 ) continue;
 270
 271                         $s = strtolower( $s );
 272                         $i = strpos( $s, ' ' );
 273
 274                         if ( $i === false ) continue;
 275
 276                         #print "processing MIME INFO line $s<br>";
 277
 278                         $match = array();
 279                         if ( preg_match( '!\[\s*(\w+)\s*\]!', $s, $match ) ) {
 280                                 $s = preg_replace( '!\[\s*(\w+)\s*\]!', '', $s );
 281                                 $mtype = trim( strtoupper( $match[1] ) );
 282                         } else {
 283                                 $mtype = MEDIATYPE_UNKNOWN;
 284                         }
 285
 286                         $m = explode( ' ', $s );
 287
 288                         if ( !isset( $this->mMediaTypes[$mtype] ) ) {
 289                                 $this->mMediaTypes[$mtype] = array();
 290                         }
 291
 292                         foreach ( $m as $mime ) {
 293                                 $mime = trim( $mime );
 294                                 if ( empty( $mime ) ) continue;
 295
 296                                 $this->mMediaTypes[$mtype][] = $mime;
 297                         }
 298
 299                         if ( sizeof( $m ) > 1 ) {
 300                                 $main = $m[0];
 301                                 for ( $i=1; $i<sizeof($m); $i += 1 ) {
 302                                         $mime = $m[$i];
 303                                         $this->mMimeTypeAliases[$mime] = $main;
 304                                 }
 305                         }
 306                 }
 307
 308         }
 309
 310         /**
 311          * Get an instance of this class
 312          * @return MimeMagic
 313          */
 314         public static function &singleton() {
 315                 if ( !isset( self::$instance ) ) {
 316                         self::$instance = new MimeMagic;
 317                 }
 318                 return self::$instance;
 319         }
 320
 321         /**
 322          * Returns a list of file extensions for a given mime type as a space
 323          * separated string or null if the mime type was unrecognized. Resolves
 324          * mime type aliases.
 325          *
 326          * @param $mime string
 327          * @return string|null
 328          */
 329         public function getExtensionsForType( $mime ) {
 330                 $mime = strtolower( $mime );
 331
 332                 // Check the mime-to-ext map
 333                 if ( isset( $this->mMimeToExt[$mime] ) ) {
 334                         return $this->mMimeToExt[$mime];
 335                 }
 336
 337                 // Resolve the mime type to the canonical type
 338                 if ( isset( $this->mMimeTypeAliases[$mime] ) ) {
 339                         $mime = $this->mMimeTypeAliases[$mime];
 340                         if ( isset( $this->mMimeToExt[$mime] ) ) {
 341                                 return $this->mMimeToExt[$mime];
 342                         }
 343                 }
 344
 345                 return null;
 346         }
 347
 348         /**
 349          * Returns a list of mime types for a given file extension as a space
 350          * separated string or null if the extension was unrecognized.
 351          *
 352          * @param $ext string
 353          * @return string|null
 354          */
 355         public function getTypesForExtension( $ext ) {
 356                 $ext = strtolower( $ext );
 357
 358                 $r = isset( $this->mExtToMime[$ext] ) ? $this->mExtToMime[$ext] : null;
 359                 return $r;
 360         }
 361
 362         /**
 363          * Returns a single mime type for a given file extension or null if unknown.
 364          * This is always the first type from the list returned by getTypesForExtension($ext).
 365          *
 366          * @param $ext string
 367          * @return string|null
 368          */
 369         public function guessTypesForExtension( $ext ) {
 370                 $m = $this->getTypesForExtension( $ext );
 371                 if ( is_null( $m ) ) {
 372                         return null;
 373                 }
 374
 375                 // TODO: Check if this is needed; strtok( $m, ' ' ) should be sufficient
 376                 $m = trim( $m );
 377                 $m = preg_replace( '/\s.*$/', '', $m );
 378
 379                 return $m;
 380         }
 381
 382
 383         /**
 384          * Tests if the extension matches the given mime type. Returns true if a
 385          * match was found, null if the mime type is unknown, and false if the
 386          * mime type is known but no matches where found.
 387          *
 388          * @param $extension string
 389          * @param $mime string
 390          * @return bool|null
 391          */
 392         public function isMatchingExtension( $extension, $mime ) {
 393                 $ext = $this->getExtensionsForType( $mime );
 394
 395                 if ( !$ext ) {
 396                         return null;  // Unknown mime type
 397                 }
 398
 399                 $ext = explode( ' ', $ext );
 400
 401                 $extension = strtolower( $extension );
 402                 return  in_array( $extension, $ext );
 403         }
 404
 405         /**
 406          * Returns true if the mime type is known to represent an image format
 407          * supported by the PHP GD library.
 408          */
 409         public function isPHPImageType( $mime ) {
 410                 // As defined by imagegetsize and image_type_to_mime
 411                 static $types = array(
 412                         'image/gif', 'image/jpeg', 'image/png',
 413                         'image/x-bmp', 'image/xbm', 'image/tiff',
 414                         'image/jp2', 'image/jpeg2000', 'image/iff',
 415                         'image/xbm', 'image/x-xbitmap',
 416                         'image/vnd.wap.wbmp', 'image/vnd.xiff',
 417                         'image/x-photoshop',
 418                         'application/x-shockwave-flash',
 419                 );
 420
 421                 return in_array( $mime, $types );
 422         }
 423
 424         /**
 425          * Returns true if the extension represents a type which can
 426          * be reliably detected from its content. Use this to determine
 427          * whether strict content checks should be applied to reject
 428          * invalid uploads; if we can't identify the type we won't
 429          * be able to say if it's invalid.
 430          *
 431          * @todo Be more accurate when using fancy mime detector plugins;
 432          *       right now this is the bare minimum getimagesize() list.
 433          * @return bool
 434          */
 435         function isRecognizableExtension( $extension ) {
 436                 static $types = array(
 437                         // Types recognized by getimagesize()
 438                         'gif', 'jpeg', 'jpg', 'png', 'swf', 'psd',
 439                         'bmp', 'tiff', 'tif', 'jpc', 'jp2',
 440                         'jpx', 'jb2', 'swc', 'iff', 'wbmp',
 441                         'xbm',
 442
 443                         // Formats we recognize magic numbers for
 444                         'djvu', 'ogx', 'ogg', 'ogv', 'oga', 'spx',
 445                         'mid', 'pdf', 'wmf', 'xcf', 'webm', 'mkv', 'mka',
 446                         'webp',
 447
 448                         // XML formats we sure hope we recognize reliably
 449                         'svg',
 450                 );
 451                 return in_array( strtolower( $extension ), $types );
 452         }
 453
 454         /**
 455          * Improves a mime type using the file extension. Some file formats are very generic,
 456          * so their mime type is not very meaningful. A more useful mime type can be derived
 457          * by looking at the file extension. Typically, this method would be called on the
 458          * result of guessMimeType().
 459          *
 460          * Currently, this method does the following:
 461          *
 462          * If $mime is "unknown/unknown" and isRecognizableExtension( $ext ) returns false,
 463          * return the result of guessTypesForExtension($ext).
 464          *
 465          * If $mime is "application/x-opc+zip" and isMatchingExtension( $ext, $mime )
 466          * gives true, return the result of guessTypesForExtension($ext).
 467          *
 468          * @param $mime String: the mime type, typically guessed from a file's content.
 469          * @param $ext String: the file extension, as taken from the file name
 470          *
 471          * @return string the mime type
 472          */
 473         public function improveTypeFromExtension( $mime, $ext ) {
 474                 if ( $mime === 'unknown/unknown' ) {
 475                         if ( $this->isRecognizableExtension( $ext ) ) {
 476                                 wfDebug( __METHOD__. ': refusing to guess mime type for .' .
 477                                         "$ext file, we should have recognized it\n" );
 478                         } else {
 479                                 // Not something we can detect, so simply
 480                                 // trust the file extension
 481                                 $mime = $this->guessTypesForExtension( $ext );
 482                         }
 483                 }
 484                 elseif ( $mime === 'application/x-opc+zip' ) {
 485                         if ( $this->isMatchingExtension( $ext, $mime ) ) {
 486                                 // A known file extension for an OPC file,
 487                                 // find the proper mime type for that file extension
 488                                 $mime = $this->guessTypesForExtension( $ext );
 489                         } else {
 490                                 wfDebug( __METHOD__. ": refusing to guess better type for $mime file, " .
 491                                         ".$ext is not a known OPC extension.\n" );
 492                                 $mime = 'application/zip';
 493                         }
 494                 }
 495
 496                 if ( isset( $this->mMimeTypeAliases[$mime] ) ) {
 497                         $mime = $this->mMimeTypeAliases[$mime];
 498                 }
 499
 500                 wfDebug(__METHOD__.": improved mime type for .$ext: $mime\n");
 501                 return $mime;
 502         }
 503
 504         /**
 505          * Mime type detection. This uses detectMimeType to detect the mime type
 506          * of the file, but applies additional checks to determine some well known
 507          * file formats that may be missed or misinterpreter by the default mime
 508          * detection (namely XML based formats like XHTML or SVG, as well as ZIP
 509          * based formats like OPC/ODF files).
 510          *
 511          * @param $file String: the file to check
 512          * @param $ext Mixed: the file extension, or true (default) to extract it from the filename.
 513          *             Set it to false to ignore the extension. DEPRECATED! Set to false, use
 514          *             improveTypeFromExtension($mime, $ext) later to improve mime type.
 515          *
 516          * @return string the mime type of $file
 517          */
 518         public function guessMimeType( $file, $ext = true ) {
 519                 if ( $ext ) { // TODO: make $ext default to false. Or better, remove it.
 520                         wfDebug( __METHOD__.": WARNING: use of the \$ext parameter is deprecated. " .
 521                                 "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 522                 }
 523
 524                 $mime = $this->doGuessMimeType( $file, $ext );
 525
 526                 if( !$mime ) {
 527                         wfDebug( __METHOD__.": internal type detection failed for $file (.$ext)...\n" );
 528                         $mime = $this->detectMimeType( $file, $ext );
 529                 }
 530
 531                 if ( isset( $this->mMimeTypeAliases[$mime] ) ) {
 532                         $mime = $this->mMimeTypeAliases[$mime];
 533                 }
 534
 535                 wfDebug(__METHOD__.": guessed mime type of $file: $mime\n");
 536                 return $mime;
 537         }
 538
 539         /**
 540          * Guess the mime type from the file contents.
 541          *
 542          * @param string $file
 543          * @param mixed $ext
 544          */
 545         private function doGuessMimeType( $file, $ext ) { // TODO: remove $ext param
 546                 // Read a chunk of the file
 547                 wfSuppressWarnings();
 548                 $f = fopen( $file, 'rt' ); // FIXME: Shouldn't this be rb?
 549                 wfRestoreWarnings();
 550
 551                 if( !$f ) {
 552                         return 'unknown/unknown';
 553                 }
 554                 $head = fread( $f, 1024 );
 555                 fseek( $f, -65558, SEEK_END );
 556                 $tail = fread( $f, 65558 ); // 65558 = maximum size of a zip EOCDR
 557                 fclose( $f );
 558
 559                 wfDebug( __METHOD__ . ": analyzing head and tail of $file for magic numbers.\n" );
 560
 561                 // Hardcode a few magic number checks...
 562                 $headers = array(
 563                         // Multimedia...
 564                         'MThd'             => 'audio/midi',
 565                         'OggS'             => 'application/ogg',
 566
 567                         // Image formats...
 568                         // Note that WMF may have a bare header, no magic number.
 569                         "\x01\x00\x09\x00" => 'application/x-msmetafile', // Possibly prone to false positives?
 570                         "\xd7\xcd\xc6\x9a" => 'application/x-msmetafile',
 571                         '%PDF'             => 'application/pdf',
 572                         'gimp xcf'         => 'image/x-xcf',
 573
 574                         // Some forbidden fruit...
 575                         'MZ'               => 'application/octet-stream', // DOS/Windows executable
 576                         "\xca\xfe\xba\xbe" => 'application/octet-stream', // Mach-O binary
 577                         "\x7fELF"          => 'application/octet-stream', // ELF binary
 578                 );
 579
 580                 foreach ( $headers as $magic => $candidate ) {
 581                         if ( strncmp( $head, $magic, strlen( $magic ) ) == 0 ) {
 582                                 wfDebug( __METHOD__ . ": magic header in $file recognized as $candidate\n" );
 583                                 return $candidate;
 584                         }
 585                 }
 586
 587                 /* Look for WebM and Matroska files */
 588                 if ( strncmp( $head, pack( "C4", 0x1a, 0x45, 0xdf, 0xa3 ), 4 ) == 0 ) {
 589                         $doctype = strpos( $head, "\x42\x82" );
 590                         if ( $doctype ) {
 591                                 // Next byte is datasize, then data (sizes larger than 1 byte are very stupid muxers)
 592                                 $data = substr($head, $doctype+3, 8);
 593                                 if ( strncmp( $data, "matroska", 8 ) == 0 ) {
 594                                         wfDebug( __METHOD__ . ": recognized file as video/x-matroska\n" );
 595                                         return "video/x-matroska";
 596                                 } elseif ( strncmp( $data, "webm", 4 ) == 0 ) {
 597                                         wfDebug( __METHOD__ . ": recognized file as video/webm\n" );
 598                                         return "video/webm";
 599                                 }
 600                         }
 601                         wfDebug( __METHOD__ . ": unknown EBML file\n" );
 602                         return "unknown/unknown";
 603                 }
 604
 605                 /* Look for WebP */
 606                 if ( strncmp( $head, "RIFF", 4 ) == 0 && strncmp( substr( $head, 8, 8), "WEBPVP8 ", 8 ) == 0 ) {
 607                         wfDebug( __METHOD__ . ": recognized file as image/webp\n" );
 608                         return "image/webp";
 609                 }
 610
 611                 /*
 612                  * Look for PHP.  Check for this before HTML/XML...  Warning: this is a
 613                  * heuristic, and won't match a file with a lot of non-PHP before.  It
 614                  * will also match text files which could be PHP. :)
 615                  *
 616                  * FIXME: For this reason, the check is probably useless -- an attacker
 617                  * could almost certainly just pad the file with a lot of nonsense to
 618                  * circumvent the check in any case where it would be a security
 619                  * problem.  On the other hand, it causes harmful false positives (bug
 620                  * 16583).  The heuristic has been cut down to exclude three-character
 621                  * strings like "<? ", but should it be axed completely?
 622                  */
 623                 if ( ( strpos( $head, '<?php' ) !== false ) ||
 624
 625                     ( strpos( $head, "<\x00?\x00p\x00h\x00p" ) !== false ) ||
 626                     ( strpos( $head, "<\x00?\x00 " ) !== false ) ||
 627                     ( strpos( $head, "<\x00?\x00\n" ) !== false ) ||
 628                     ( strpos( $head, "<\x00?\x00\t" ) !== false ) ||
 629                     ( strpos( $head, "<\x00?\x00=" ) !== false ) ) {
 630
 631                         wfDebug( __METHOD__ . ": recognized $file as application/x-php\n" );
 632                         return 'application/x-php';
 633                 }
 634
 635                 /*
 636                  * look for XML formats (XHTML and SVG)
 637                  */
 638                 $xml = new XmlTypeCheck( $file );
 639                 if ( $xml->wellFormed ) {
 640                         global $wgXMLMimeTypes;
 641                         if ( isset( $wgXMLMimeTypes[$xml->getRootElement()] ) ) {
 642                                 return $wgXMLMimeTypes[$xml->getRootElement()];
 643                         } else {
 644                                 return 'application/xml';
 645                         }
 646                 }
 647
 648                 /*
 649                  * look for shell scripts
 650                  */
 651                 $script_type = null;
 652
 653                 # detect by shebang
 654                 if ( substr( $head, 0, 2) == "#!" ) {
 655                         $script_type = "ASCII";
 656                 } elseif ( substr( $head, 0, 5) == "\xef\xbb\xbf#!" ) {
 657                         $script_type = "UTF-8";
 658                 } elseif ( substr( $head, 0, 7) == "\xfe\xff\x00#\x00!" ) {
 659                         $script_type = "UTF-16BE";
 660                 } elseif ( substr( $head, 0, 7 ) == "\xff\xfe#\x00!" ) {
 661                         $script_type= "UTF-16LE";
 662                 }
 663
 664                 if ( $script_type ) {
 665                         if ( $script_type !== "UTF-8" && $script_type !== "ASCII") {
 666                                 // Quick and dirty fold down to ASCII!
 667                                 $pack = array( 'UTF-16BE' => 'n*', 'UTF-16LE' => 'v*' );
 668                                 $chars = unpack( $pack[$script_type], substr( $head, 2 ) );
 669                                 $head = '';
 670                                 foreach( $chars as $codepoint ) {
 671                                         if( $codepoint < 128 ) {
 672                                                 $head .= chr( $codepoint );
 673                                         } else {
 674                                                 $head .= '?';
 675                                         }
 676                                 }
 677                         }
 678
 679                         $match = array();
 680
 681                         if ( preg_match( '%/?([^\s]+/)(\w+)%', $head, $match ) ) {
 682                                 $mime = "application/x-{$match[2]}";
 683                                 wfDebug( __METHOD__.": shell script recognized as $mime\n" );
 684                                 return $mime;
 685                         }
 686                 }
 687
 688                 // Check for ZIP variants (before getimagesize)
 689                 if ( strpos( $tail, "PK\x05\x06" ) !== false ) {
 690                         wfDebug( __METHOD__.": ZIP header present in $file\n" );
 691                         return $this->detectZipType( $head, $tail, $ext );
 692                 }
 693
 694                 wfSuppressWarnings();
 695                 $gis = getimagesize( $file );
 696                 wfRestoreWarnings();
 697
 698                 if( $gis && isset( $gis['mime'] ) ) {
 699                         $mime = $gis['mime'];
 700                         wfDebug( __METHOD__.": getimagesize detected $file as $mime\n" );
 701                         return $mime;
 702                 }
 703
 704                 // Also test DjVu
 705                 $deja = new DjVuImage( $file );
 706                 if( $deja->isValid() ) {
 707                         wfDebug( __METHOD__.": detected $file as image/vnd.djvu\n" );
 708                         return 'image/vnd.djvu';
 709                 }
 710
 711                 return false;
 712         }
 713
 714         /**
 715          * Detect application-specific file type of a given ZIP file from its
 716          * header data.  Currently works for OpenDocument and OpenXML types...
 717          * If can't tell, returns 'application/zip'.
 718          *
 719          * @param $header String: some reasonably-sized chunk of file header
 720          * @param $tail   String: the tail of the file
 721          * @param $ext Mixed: the file extension, or true to extract it from the filename.
 722          *             Set it to false (default) to ignore the extension. DEPRECATED! Set to false,
 723          *             use improveTypeFromExtension($mime, $ext) later to improve mime type.
 724          *
 725          * @return string
 726          */
 727         function detectZipType( $header, $tail = null, $ext = false ) {
 728                 if( $ext ) { # TODO: remove $ext param
 729                         wfDebug( __METHOD__.": WARNING: use of the \$ext parameter is deprecated. " .
 730                                 "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 731                 }
 732
 733                 $mime = 'application/zip';
 734                 $opendocTypes = array(
 735                         'chart-template',
 736                         'chart',
 737                         'formula-template',
 738                         'formula',
 739                         'graphics-template',
 740                         'graphics',
 741                         'image-template',
 742                         'image',
 743                         'presentation-template',
 744                         'presentation',
 745                         'spreadsheet-template',
 746                         'spreadsheet',
 747                         'text-template',
 748                         'text-master',
 749                         'text-web',
 750                         'text' );
 751
 752                 // http://lists.oasis-open.org/archives/office/200505/msg00006.html
 753                 $types = '(?:' . implode( '|', $opendocTypes ) . ')';
 754                 $opendocRegex = "/^mimetype(application\/vnd\.oasis\.opendocument\.$types)/";
 755
 756                 $openxmlRegex = "/^\[Content_Types\].xml/";
 757
 758                 if ( preg_match( $opendocRegex, substr( $header, 30 ), $matches ) ) {
 759                         $mime = $matches[1];
 760                         wfDebug( __METHOD__.": detected $mime from ZIP archive\n" );
 761                 } elseif ( preg_match( $openxmlRegex, substr( $header, 30 ) ) ) {
 762                         $mime = "application/x-opc+zip";
 763                         # TODO: remove the block below, as soon as improveTypeFromExtension is used everywhere
 764                         if ( $ext !== true && $ext !== false ) {
 765                                 /** This is the mode used by getPropsFromPath
 766                                 * These mime's are stored in the database, where we don't really want
 767                                 * x-opc+zip, because we use it only for internal purposes
 768                                 */
 769                                 if ( $this->isMatchingExtension( $ext, $mime) ) {
 770                                         /* A known file extension for an OPC file,
 771                                         * find the proper mime type for that file extension */
 772                                         $mime = $this->guessTypesForExtension( $ext );
 773                                 } else {
 774                                         $mime = "application/zip";
 775                                 }
 776                         }
 777                         wfDebug( __METHOD__.": detected an Open Packaging Conventions archive: $mime\n" );
 778                 } elseif ( substr( $header, 0, 8 ) == "\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" &&
 779                                 ($headerpos = strpos( $tail, "PK\x03\x04" ) ) !== false &&
 780                                 preg_match( $openxmlRegex, substr( $tail, $headerpos + 30 ) ) ) {
 781                         if ( substr( $header, 512, 4) == "\xEC\xA5\xC1\x00" ) {
 782                                 $mime = "application/msword";
 783                         }
 784                         switch( substr( $header, 512, 6) ) {
 785                                 case "\xEC\xA5\xC1\x00\x0E\x00":
 786                                 case "\xEC\xA5\xC1\x00\x1C\x00":
 787                                 case "\xEC\xA5\xC1\x00\x43\x00":
 788                                         $mime = "application/vnd.ms-powerpoint";
 789                                         break;
 790                                 case "\xFD\xFF\xFF\xFF\x10\x00":
 791                                 case "\xFD\xFF\xFF\xFF\x1F\x00":
 792                                 case "\xFD\xFF\xFF\xFF\x22\x00":
 793                                 case "\xFD\xFF\xFF\xFF\x23\x00":
 794                                 case "\xFD\xFF\xFF\xFF\x28\x00":
 795                                 case "\xFD\xFF\xFF\xFF\x29\x00":
 796                                 case "\xFD\xFF\xFF\xFF\x10\x02":
 797                                 case "\xFD\xFF\xFF\xFF\x1F\x02":
 798                                 case "\xFD\xFF\xFF\xFF\x22\x02":
 799                                 case "\xFD\xFF\xFF\xFF\x23\x02":
 800                                 case "\xFD\xFF\xFF\xFF\x28\x02":
 801                                 case "\xFD\xFF\xFF\xFF\x29\x02":
 802                                         $mime = "application/vnd.msexcel";
 803                                         break;
 804                         }
 805
 806                         wfDebug( __METHOD__.": detected a MS Office document with OPC trailer\n");
 807                 } else {
 808                         wfDebug( __METHOD__.": unable to identify type of ZIP archive\n" );
 809                 }
 810                 return $mime;
 811         }
 812
 813         /**
 814          * Internal mime type detection. Detection is done using an external
 815          * program, if $wgMimeDetectorCommand is set. Otherwise, the fileinfo
 816          * extension and mime_content_type are tried (in this order), if they
 817          * are available. If the dections fails and $ext is not false, the mime
 818          * type is guessed from the file extension, using guessTypesForExtension.
 819          *
 820          * If the mime type is still unknown, getimagesize is used to detect the
 821          * mime type if the file is an image. If no mime type can be determined,
 822          * this function returns 'unknown/unknown'.
 823          *
 824          * @param $file String: the file to check
 825          * @param $ext Mixed: the file extension, or true (default) to extract it from the filename.
 826          *             Set it to false to ignore the extension. DEPRECATED! Set to false, use
 827          *             improveTypeFromExtension($mime, $ext) later to improve mime type.
 828          *
 829          * @return string the mime type of $file
 830          */
 831         private function detectMimeType( $file, $ext = true ) {
 832                 global $wgMimeDetectorCommand;
 833
 834                 if ( $ext ) { # TODO:  make $ext default to false. Or better, remove it.
 835                         wfDebug( __METHOD__.": WARNING: use of the \$ext parameter is deprecated. Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 836                 }
 837
 838                 $m = null;
 839                 if ( $wgMimeDetectorCommand ) {
 840                         // FIXME: Use wfShellExec
 841                         $fn = wfEscapeShellArg( $file );
 842                         $m = `$wgMimeDetectorCommand $fn`;
 843                 } elseif ( function_exists( "finfo_open" ) && function_exists( "finfo_file" ) ) {
 844
 845                         # This required the fileinfo extension by PECL,
 846                         # see http://pecl.php.net/package/fileinfo
 847                         # This must be compiled into PHP
 848                         #
 849                         # finfo is the official replacement for the deprecated
 850                         # mime_content_type function, see below.
 851                         #
 852                         # If you may need to load the fileinfo extension at runtime, set
 853                         # $wgLoadFileinfoExtension in LocalSettings.php
 854
 855                         $mime_magic_resource = finfo_open( FILEINFO_MIME ); /* return mime type ala mimetype extension */
 856
 857                         if ( $mime_magic_resource ) {
 858                                 $m = finfo_file( $mime_magic_resource, $file );
 859                                 finfo_close( $mime_magic_resource );
 860                         } else {
 861                                 wfDebug( __METHOD__.": finfo_open failed on ".FILEINFO_MIME."!\n" );
 862                         }
 863                 } elseif ( function_exists( "mime_content_type" ) ) {
 864
 865                         # NOTE: this function is available since PHP 4.3.0, but only if
 866                         # PHP was compiled with --with-mime-magic or, before 4.3.2, with --enable-mime-magic.
 867                         #
 868                         # On Windows, you must set mime_magic.magicfile in php.ini to point to the mime.magic file bundeled with PHP;
 869                         # sometimes, this may even be needed under linus/unix.
 870                         #
 871                         # Also note that this has been DEPRECATED in favor of the fileinfo extension by PECL, see above.
 872                         # see http://www.php.net/manual/en/ref.mime-magic.php for details.
 873
 874                         $m = mime_content_type($file);
 875                 } else {
 876                         wfDebug( __METHOD__.": no magic mime detector found!\n" );
 877                 }
 878
 879                 if ( $m ) {
 880                         # normalize
 881                         $m = preg_replace( '![;, ].*$!', '', $m ); #strip charset, etc
 882                         $m = trim( $m );
 883                         $m = strtolower( $m );
 884
 885                         if ( strpos( $m, 'unknown' ) !== false ) {
 886                                 $m = null;
 887                         } else {
 888                                 wfDebug( __METHOD__.": magic mime type of $file: $m\n" );
 889                                 return $m;
 890                         }
 891                 }
 892
 893                 // If desired, look at extension as a fallback.
 894                 if ( $ext === true ) {
 895                         $i = strrpos( $file, '.' );
 896                         $ext = strtolower( $i ? substr( $file, $i + 1 ) : '' );
 897                 }
 898                 if ( $ext ) {
 899                         if( $this->isRecognizableExtension( $ext ) ) {
 900                                 wfDebug( __METHOD__. ": refusing to guess mime type for .$ext file, we should have recognized it\n" );
 901                         } else {
 902                                 $m = $this->guessTypesForExtension( $ext );
 903                                 if ( $m ) {
 904                                         wfDebug( __METHOD__.": extension mime type of $file: $m\n" );
 905                                         return $m;
 906                                 }
 907                         }
 908                 }
 909
 910                 // Unknown type
 911                 wfDebug( __METHOD__ . ": failed to guess mime type for $file!\n" );
 912                 return 'unknown/unknown';
 913         }
 914
 915         /**
 916          * Determine the media type code for a file, using its mime type, name and
 917          * possibly its contents.
 918          *
 919          * This function relies on the findMediaType(), mapping extensions and mime
 920          * types to media types.
 921          *
 922          * @todo analyse file if need be
 923          * @todo look at multiple extension, separately and together.
 924          *
 925          * @param $path String: full path to the image file, in case we have to look at the contents
 926          *        (if null, only the mime type is used to determine the media type code).
 927          * @param $mime String: mime type. If null it will be guessed using guessMimeType.
 928          *
 929          * @return (int?string?) a value to be used with the MEDIATYPE_xxx constants.
 930          */
 931         function getMediaType( $path = null, $mime = null ) {
 932                 if( !$mime && !$path ) {
 933                         return MEDIATYPE_UNKNOWN;
 934                 }
 935
 936                 // If mime type is unknown, guess it
 937                 if( !$mime ) {
 938                         $mime = $this->guessMimeType( $path, false );
 939                 }
 940
 941                 // Special code for ogg - detect if it's video (theora),
 942                 // else label it as sound.
 943                 if ( $mime == 'application/ogg' && file_exists( $path ) ) {
 944
 945                         // Read a chunk of the file
 946                         $f = fopen( $path, "rt" );
 947                         if ( !$f ) return MEDIATYPE_UNKNOWN;
 948                         $head = fread( $f, 256 );
 949                         fclose( $f );
 950
 951                         $head = strtolower( $head );
 952
 953                         // This is an UGLY HACK, file should be parsed correctly
 954                         if ( strpos( $head, 'theora' ) !== false ) return MEDIATYPE_VIDEO;
 955                         elseif ( strpos( $head, 'vorbis' ) !== false ) return MEDIATYPE_AUDIO;
 956                         elseif ( strpos( $head, 'flac' ) !== false ) return MEDIATYPE_AUDIO;
 957                         elseif ( strpos( $head, 'speex' ) !== false ) return MEDIATYPE_AUDIO;
 958                         else return MEDIATYPE_MULTIMEDIA;
 959                 }
 960
 961                 // Check for entry for full mime type
 962                 if( $mime ) {
 963                         $type = $this->findMediaType( $mime );
 964                         if ( $type !== MEDIATYPE_UNKNOWN ) {
 965                                 return $type;
 966                         }
 967                 }
 968
 969                 // Check for entry for file extension
 970                 if ( $path ) {
 971                         $i = strrpos( $path, '.' );
 972                         $e = strtolower( $i ? substr( $path, $i + 1 ) : '' );
 973
 974                         // TODO: look at multi-extension if this fails, parse from full path
 975                         $type = $this->findMediaType( '.' . $e );
 976                         if ( $type !== MEDIATYPE_UNKNOWN ) {
 977                                 return $type;
 978                         }
 979                 }
 980
 981                 // Check major mime type
 982                 if ( $mime ) {
 983                         $i = strpos( $mime, '/' );
 984                         if ( $i !== false ) {
 985                                 $major = substr( $mime, 0, $i );
 986                                 $type = $this->findMediaType( $major );
 987                                 if ( $type !== MEDIATYPE_UNKNOWN ) {
 988                                         return $type;
 989                                 }
 990                         }
 991                 }
 992
 993                 if( !$type ) {
 994                         $type = MEDIATYPE_UNKNOWN;
 995                 }
 996
 997                 return $type;
 998         }
 999
1000         /**
1001          * Returns a media code matching the given mime type or file extension.
1002          * File extensions are represented by a string starting with a dot (.) to
1003          * distinguish them from mime types.
1004          *
1005          * This funktion relies on the mapping defined by $this->mMediaTypes
1006          * @access private
1007          */
1008         function findMediaType( $extMime ) {
1009                 if ( strpos( $extMime, '.' ) === 0 ) {
1010                         // If it's an extension, look up the mime types
1011                         $m = $this->getTypesForExtension( substr( $extMime, 1 ) );
1012                         if ( !$m ) {
1013                                 return MEDIATYPE_UNKNOWN;
1014                         }
1015
1016                         $m = explode( ' ', $m );
1017                 } else {
1018                         // Normalize mime type
1019                         if ( isset( $this->mMimeTypeAliases[$extMime] ) ) {
1020                                 $extMime = $this->mMimeTypeAliases[$extMime];
1021                         }
1022
1023                         $m = array( $extMime );
1024                 }
1025
1026                 foreach ( $m as $mime ) {
1027                         foreach ( $this->mMediaTypes as $type => $codes ) {
1028                                 if ( in_array($mime, $codes, true ) ) {
1029                                         return $type;
1030                                 }
1031                         }
1032                 }
1033
1034                 return MEDIATYPE_UNKNOWN;
1035         }
1036
1037         /**
1038          * Get the MIME types that various versions of Internet Explorer would
1039          * detect from a chunk of the content.
1040          *
1041          * @param $fileName String: the file name (unused at present)
1042          * @param $chunk String: the first 256 bytes of the file
1043          * @param $proposed String: the MIME type proposed by the server
1044          */
1045         public function getIEMimeTypes( $fileName, $chunk, $proposed ) {
1046                 $ca = $this->getIEContentAnalyzer();
1047                 return $ca->getRealMimesFromData( $fileName, $chunk, $proposed );
1048         }
1049
1050         /**
1051          * Get a cached instance of IEContentAnalyzer
1052          */
1053         protected function getIEContentAnalyzer() {
1054                 if ( is_null( $this->mIEAnalyzer ) ) {
1055                         $this->mIEAnalyzer = new IEContentAnalyzer;
1056                 }
1057                 return $this->mIEAnalyzer;
1058         }
1059 }