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.database odb
  33 application/vnd.oasis.opendocument.formula odf
  34 application/vnd.oasis.opendocument.formula-template otf
  35 application/vnd.oasis.opendocument.graphics odg
  36 application/vnd.oasis.opendocument.graphics-template otg
  37 application/vnd.oasis.opendocument.image odi
  38 application/vnd.oasis.opendocument.image-template oti
  39 application/vnd.oasis.opendocument.presentation odp
  40 application/vnd.oasis.opendocument.presentation-template otp
  41 application/vnd.oasis.opendocument.spreadsheet ods
  42 application/vnd.oasis.opendocument.spreadsheet-template ots
  43 application/vnd.oasis.opendocument.text odt
  44 application/vnd.oasis.opendocument.text-master otm
  45 application/vnd.oasis.opendocument.text-template ott
  46 application/vnd.oasis.opendocument.text-web oth
  47 application/x-javascript js
  48 application/x-shockwave-flash swf
  49 audio/midi mid midi kar
  50 audio/mpeg mpga mpa mp2 mp3
  51 audio/x-aiff aif aiff aifc
  52 audio/x-wav wav
  53 audio/ogg oga spx ogg
  54 image/x-bmp bmp
  55 image/gif gif
  56 image/jpeg jpeg jpg jpe
  57 image/png png
  58 image/svg+xml svg
  59 image/svg svg
  60 image/tiff tiff tif
  61 image/vnd.djvu djvu
  62 image/x.djvu djvu
  63 image/x-djvu djvu
  64 image/x-portable-pixmap ppm
  65 image/x-xcf xcf
  66 text/plain txt
  67 text/html html htm
  68 video/ogg ogv ogm ogg
  69 video/mpeg mpg mpeg
  70 END_STRING
  71 );
  72
  73 /**
  74  * Defines a set of well known mime info entries
  75  * This is used as a fallback to mime.info files.
  76  * An extensive list of well known mime types is provided by
  77  * the file mime.info in the includes directory.
  78  */
  79 define('MM_WELL_KNOWN_MIME_INFO', <<<END_STRING
  80 application/pdf [OFFICE]
  81 application/vnd.oasis.opendocument.chart [OFFICE]
  82 application/vnd.oasis.opendocument.chart-template [OFFICE]
  83 application/vnd.oasis.opendocument.database [OFFICE]
  84 application/vnd.oasis.opendocument.formula [OFFICE]
  85 application/vnd.oasis.opendocument.formula-template [OFFICE]
  86 application/vnd.oasis.opendocument.graphics [OFFICE]
  87 application/vnd.oasis.opendocument.graphics-template [OFFICE]
  88 application/vnd.oasis.opendocument.image [OFFICE]
  89 application/vnd.oasis.opendocument.image-template [OFFICE]
  90 application/vnd.oasis.opendocument.presentation [OFFICE]
  91 application/vnd.oasis.opendocument.presentation-template [OFFICE]
  92 application/vnd.oasis.opendocument.spreadsheet [OFFICE]
  93 application/vnd.oasis.opendocument.spreadsheet-template [OFFICE]
  94 application/vnd.oasis.opendocument.text [OFFICE]
  95 application/vnd.oasis.opendocument.text-template [OFFICE]
  96 application/vnd.oasis.opendocument.text-master [OFFICE]
  97 application/vnd.oasis.opendocument.text-web [OFFICE]
  98 text/javascript application/x-javascript [EXECUTABLE]
  99 application/x-shockwave-flash [MULTIMEDIA]
 100 audio/midi [AUDIO]
 101 audio/x-aiff [AUDIO]
 102 audio/x-wav [AUDIO]
 103 audio/mp3 audio/mpeg [AUDIO]
 104 application/ogg audio/ogg video/ogg [MULTIMEDIA]
 105 image/x-bmp image/x-ms-bmp image/bmp [BITMAP]
 106 image/gif [BITMAP]
 107 image/jpeg [BITMAP]
 108 image/png [BITMAP]
 109 image/svg+xml [DRAWING]
 110 image/tiff [BITMAP]
 111 image/vnd.djvu [BITMAP]
 112 image/x-xcf [BITMAP]
 113 image/x-portable-pixmap [BITMAP]
 114 text/plain [TEXT]
 115 text/html [TEXT]
 116 video/ogg [VIDEO]
 117 video/mpeg [VIDEO]
 118 unknown/unknown application/octet-stream application/x-empty [UNKNOWN]
 119 END_STRING
 120 );
 121
 122 /**
 123  * Implements functions related to mime types such as detection and mapping to
 124  * file extension.
 125  *
 126  * Instances of this class are stateles, there only needs to be one global instance
 127  * of MimeMagic. Please use MimeMagic::singleton() to get that instance.
 128  */
 129 class MimeMagic {
 130
 131         /**
 132         * Mapping of media types to arrays of mime types.
 133         * This is used by findMediaType and getMediaType, respectively
 134         */
 135         var $mMediaTypes = null;
 136
 137         /** Map of mime type aliases
 138         */
 139         var $mMimeTypeAliases = null;
 140
 141         /** map of mime types to file extensions (as a space seprarated list)
 142         */
 143         var $mMimeToExt = null;
 144
 145         /** map of file extensions types to mime types (as a space seprarated list)
 146         */
 147         var $mExtToMime = null;
 148
 149         /** IEContentAnalyzer instance
 150          */
 151         var $mIEAnalyzer;
 152
 153         /** The singleton instance
 154          */
 155         private static $instance;
 156
 157         /** True if the fileinfo extension has been loaded
 158          */
 159         private static $extensionLoaded = false;
 160
 161         /** Initializes the MimeMagic object. This is called by MimeMagic::singleton().
 162          *
 163          * This constructor parses the mime.types and mime.info files and build internal mappings.
 164          */
 165         function __construct() {
 166                 /**
 167                 *   --- load mime.types ---
 168                 */
 169
 170                 global $wgMimeTypeFile, $IP, $wgLoadFileinfoExtension;
 171
 172                 $types = MM_WELL_KNOWN_MIME_TYPES;
 173
 174                 if ( $wgMimeTypeFile == 'includes/mime.types' ) {
 175                         $wgMimeTypeFile = "$IP/$wgMimeTypeFile";
 176                 }
 177
 178                 if ( $wgLoadFileinfoExtension && !self::$extensionLoaded ) {
 179                         self::$extensionLoaded = true;
 180                         wfDl( 'fileinfo' );
 181                 }
 182
 183                 if ( $wgMimeTypeFile ) {
 184                         if ( is_file( $wgMimeTypeFile ) and is_readable( $wgMimeTypeFile ) ) {
 185                                 wfDebug( __METHOD__.": loading mime types from $wgMimeTypeFile\n" );
 186                                 $types .= "\n";
 187                                 $types .= file_get_contents( $wgMimeTypeFile );
 188                         } else {
 189                                 wfDebug( __METHOD__.": can't load mime types from $wgMimeTypeFile\n" );
 190                         }
 191                 } else {
 192                         wfDebug( __METHOD__.": no mime types file defined, using build-ins only.\n" );
 193                 }
 194
 195                 $types = str_replace( array( "\r\n", "\n\r", "\n\n", "\r\r", "\r" ), "\n", $types );
 196                 $types = str_replace( "\t", " ", $types );
 197
 198                 $this->mMimeToExt = array();
 199                 $this->mToMime = array();
 200
 201                 $lines = explode( "\n",$types );
 202                 foreach ( $lines as $s ) {
 203                         $s = trim( $s );
 204                         if ( empty( $s ) ) {
 205                                 continue;
 206                         }
 207                         if ( strpos( $s, '#' ) === 0 ) {
 208                                 continue;
 209                         }
 210
 211                         $s = strtolower( $s );
 212                         $i = strpos( $s, ' ' );
 213
 214                         if ( $i === false ) {
 215                                 continue;
 216                         }
 217
 218                         #print "processing MIME line $s<br>";
 219
 220                         $mime = substr( $s, 0, $i );
 221                         $ext = trim( substr($s, $i+1 ) );
 222
 223                         if ( empty( $ext ) ) {
 224                                 continue;
 225                         }
 226
 227                         if ( !empty( $this->mMimeToExt[$mime] ) ) {
 228                                 $this->mMimeToExt[$mime] .= ' ' . $ext;
 229                         } else {
 230                                 $this->mMimeToExt[$mime] = $ext;
 231                         }
 232
 233                         $extensions = explode( ' ', $ext );
 234
 235                         foreach ( $extensions as $e ) {
 236                                 $e = trim( $e );
 237                                 if ( empty( $e ) ) {
 238                                         continue;
 239                                 }
 240
 241                                 if ( !empty( $this->mExtToMime[$e] ) ) {
 242                                         $this->mExtToMime[$e] .= ' ' . $mime;
 243                                 } else {
 244                                         $this->mExtToMime[$e] = $mime;
 245                                 }
 246                         }
 247                 }
 248
 249                 /**
 250                  *   --- load mime.info ---
 251                  */
 252
 253                 global $wgMimeInfoFile;
 254                 if ( $wgMimeInfoFile == 'includes/mime.info' ) {
 255                         $wgMimeInfoFile = "$IP/$wgMimeInfoFile";
 256                 }
 257
 258                 $info = MM_WELL_KNOWN_MIME_INFO;
 259
 260                 if ( $wgMimeInfoFile ) {
 261                         if ( is_file( $wgMimeInfoFile ) and is_readable( $wgMimeInfoFile ) ) {
 262                                 wfDebug( __METHOD__.": loading mime info from $wgMimeInfoFile\n" );
 263                                 $info .= "\n";
 264                                 $info .= file_get_contents( $wgMimeInfoFile );
 265                         } else {
 266                                 wfDebug(__METHOD__.": can't load mime info from $wgMimeInfoFile\n");
 267                         }
 268                 } else {
 269                         wfDebug(__METHOD__.": no mime info file defined, using build-ins only.\n");
 270                 }
 271
 272                 $info = str_replace( array( "\r\n", "\n\r", "\n\n", "\r\r", "\r" ), "\n", $info);
 273                 $info = str_replace( "\t", " ", $info );
 274
 275                 $this->mMimeTypeAliases = array();
 276                 $this->mMediaTypes = array();
 277
 278                 $lines = explode( "\n", $info );
 279                 foreach ( $lines as $s ) {
 280                         $s = trim( $s );
 281                         if ( empty( $s ) ) {
 282                                 continue;
 283                         }
 284                         if ( strpos( $s, '#' ) === 0 ) {
 285                                 continue;
 286                         }
 287
 288                         $s = strtolower( $s );
 289                         $i = strpos( $s, ' ' );
 290
 291                         if ( $i === false ) {
 292                                 continue;
 293                         }
 294
 295                         #print "processing MIME INFO line $s<br>";
 296
 297                         $match = array();
 298                         if ( preg_match( '!\[\s*(\w+)\s*\]!', $s, $match ) ) {
 299                                 $s = preg_replace( '!\[\s*(\w+)\s*\]!', '', $s );
 300                                 $mtype = trim( strtoupper( $match[1] ) );
 301                         } else {
 302                                 $mtype = MEDIATYPE_UNKNOWN;
 303                         }
 304
 305                         $m = explode( ' ', $s );
 306
 307                         if ( !isset( $this->mMediaTypes[$mtype] ) ) {
 308                                 $this->mMediaTypes[$mtype] = array();
 309                         }
 310
 311                         foreach ( $m as $mime ) {
 312                                 $mime = trim( $mime );
 313                                 if ( empty( $mime ) ) {
 314                                         continue;
 315                                 }
 316
 317                                 $this->mMediaTypes[$mtype][] = $mime;
 318                         }
 319
 320                         if ( sizeof( $m ) > 1 ) {
 321                                 $main = $m[0];
 322                                 for ( $i=1; $i<sizeof($m); $i += 1 ) {
 323                                         $mime = $m[$i];
 324                                         $this->mMimeTypeAliases[$mime] = $main;
 325                                 }
 326                         }
 327                 }
 328
 329         }
 330
 331         /**
 332          * Get an instance of this class
 333          * @return MimeMagic
 334          */
 335         public static function &singleton() {
 336                 if ( !isset( self::$instance ) ) {
 337                         self::$instance = new MimeMagic;
 338                 }
 339                 return self::$instance;
 340         }
 341
 342         /**
 343          * Returns a list of file extensions for a given mime type as a space
 344          * separated string or null if the mime type was unrecognized. Resolves
 345          * mime type aliases.
 346          *
 347          * @param $mime string
 348          * @return string|null
 349          */
 350         public function getExtensionsForType( $mime ) {
 351                 $mime = strtolower( $mime );
 352
 353                 // Check the mime-to-ext map
 354                 if ( isset( $this->mMimeToExt[$mime] ) ) {
 355                         return $this->mMimeToExt[$mime];
 356                 }
 357
 358                 // Resolve the mime type to the canonical type
 359                 if ( isset( $this->mMimeTypeAliases[$mime] ) ) {
 360                         $mime = $this->mMimeTypeAliases[$mime];
 361                         if ( isset( $this->mMimeToExt[$mime] ) ) {
 362                                 return $this->mMimeToExt[$mime];
 363                         }
 364                 }
 365
 366                 return null;
 367         }
 368
 369         /**
 370          * Returns a list of mime types for a given file extension as a space
 371          * separated string or null if the extension was unrecognized.
 372          *
 373          * @param $ext string
 374          * @return string|null
 375          */
 376         public function getTypesForExtension( $ext ) {
 377                 $ext = strtolower( $ext );
 378
 379                 $r = isset( $this->mExtToMime[$ext] ) ? $this->mExtToMime[$ext] : null;
 380                 return $r;
 381         }
 382
 383         /**
 384          * Returns a single mime type for a given file extension or null if unknown.
 385          * This is always the first type from the list returned by getTypesForExtension($ext).
 386          *
 387          * @param $ext string
 388          * @return string|null
 389          */
 390         public function guessTypesForExtension( $ext ) {
 391                 $m = $this->getTypesForExtension( $ext );
 392                 if ( is_null( $m ) ) {
 393                         return null;
 394                 }
 395
 396                 // TODO: Check if this is needed; strtok( $m, ' ' ) should be sufficient
 397                 $m = trim( $m );
 398                 $m = preg_replace( '/\s.*$/', '', $m );
 399
 400                 return $m;
 401         }
 402
 403
 404         /**
 405          * Tests if the extension matches the given mime type. Returns true if a
 406          * match was found, null if the mime type is unknown, and false if the
 407          * mime type is known but no matches where found.
 408          *
 409          * @param $extension string
 410          * @param $mime string
 411          * @return bool|null
 412          */
 413         public function isMatchingExtension( $extension, $mime ) {
 414                 $ext = $this->getExtensionsForType( $mime );
 415
 416                 if ( !$ext ) {
 417                         return null;  // Unknown mime type
 418                 }
 419
 420                 $ext = explode( ' ', $ext );
 421
 422                 $extension = strtolower( $extension );
 423                 return  in_array( $extension, $ext );
 424         }
 425
 426         /**
 427          * Returns true if the mime type is known to represent an image format
 428          * supported by the PHP GD library.
 429          *
 430          * @param $mime string
 431          *
 432          * @return bool
 433          */
 434         public function isPHPImageType( $mime ) {
 435                 // As defined by imagegetsize and image_type_to_mime
 436                 static $types = array(
 437                         'image/gif', 'image/jpeg', 'image/png',
 438                         'image/x-bmp', 'image/xbm', 'image/tiff',
 439                         'image/jp2', 'image/jpeg2000', 'image/iff',
 440                         'image/xbm', 'image/x-xbitmap',
 441                         'image/vnd.wap.wbmp', 'image/vnd.xiff',
 442                         'image/x-photoshop',
 443                         'application/x-shockwave-flash',
 444                 );
 445
 446                 return in_array( $mime, $types );
 447         }
 448
 449         /**
 450          * Returns true if the extension represents a type which can
 451          * be reliably detected from its content. Use this to determine
 452          * whether strict content checks should be applied to reject
 453          * invalid uploads; if we can't identify the type we won't
 454          * be able to say if it's invalid.
 455          *
 456          * @todo Be more accurate when using fancy mime detector plugins;
 457          *       right now this is the bare minimum getimagesize() list.
 458          * @return bool
 459          */
 460         function isRecognizableExtension( $extension ) {
 461                 static $types = array(
 462                         // Types recognized by getimagesize()
 463                         'gif', 'jpeg', 'jpg', 'png', 'swf', 'psd',
 464                         'bmp', 'tiff', 'tif', 'jpc', 'jp2',
 465                         'jpx', 'jb2', 'swc', 'iff', 'wbmp',
 466                         'xbm',
 467
 468                         // Formats we recognize magic numbers for
 469                         'djvu', 'ogx', 'ogg', 'ogv', 'oga', 'spx',
 470                         'mid', 'pdf', 'wmf', 'xcf', 'webm', 'mkv', 'mka',
 471                         'webp',
 472
 473                         // XML formats we sure hope we recognize reliably
 474                         'svg',
 475                 );
 476                 return in_array( strtolower( $extension ), $types );
 477         }
 478
 479         /**
 480          * Improves a mime type using the file extension. Some file formats are very generic,
 481          * so their mime type is not very meaningful. A more useful mime type can be derived
 482          * by looking at the file extension. Typically, this method would be called on the
 483          * result of guessMimeType().
 484          *
 485          * Currently, this method does the following:
 486          *
 487          * If $mime is "unknown/unknown" and isRecognizableExtension( $ext ) returns false,
 488          * return the result of guessTypesForExtension($ext).
 489          *
 490          * If $mime is "application/x-opc+zip" and isMatchingExtension( $ext, $mime )
 491          * gives true, return the result of guessTypesForExtension($ext).
 492          *
 493          * @param $mime String: the mime type, typically guessed from a file's content.
 494          * @param $ext String: the file extension, as taken from the file name
 495          *
 496          * @return string the mime type
 497          */
 498         public function improveTypeFromExtension( $mime, $ext ) {
 499                 if ( $mime === 'unknown/unknown' ) {
 500                         if ( $this->isRecognizableExtension( $ext ) ) {
 501                                 wfDebug( __METHOD__. ': refusing to guess mime type for .' .
 502                                         "$ext file, we should have recognized it\n" );
 503                         } else {
 504                                 // Not something we can detect, so simply
 505                                 // trust the file extension
 506                                 $mime = $this->guessTypesForExtension( $ext );
 507                         }
 508                 }
 509                 elseif ( $mime === 'application/x-opc+zip' ) {
 510                         if ( $this->isMatchingExtension( $ext, $mime ) ) {
 511                                 // A known file extension for an OPC file,
 512                                 // find the proper mime type for that file extension
 513                                 $mime = $this->guessTypesForExtension( $ext );
 514                         } else {
 515                                 wfDebug( __METHOD__. ": refusing to guess better type for $mime file, " .
 516                                         ".$ext is not a known OPC extension.\n" );
 517                                 $mime = 'application/zip';
 518                         }
 519                 }
 520
 521                 if ( isset( $this->mMimeTypeAliases[$mime] ) ) {
 522                         $mime = $this->mMimeTypeAliases[$mime];
 523                 }
 524
 525                 wfDebug(__METHOD__.": improved mime type for .$ext: $mime\n");
 526                 return $mime;
 527         }
 528
 529         /**
 530          * Mime type detection. This uses detectMimeType to detect the mime type
 531          * of the file, but applies additional checks to determine some well known
 532          * file formats that may be missed or misinterpreter by the default mime
 533          * detection (namely XML based formats like XHTML or SVG, as well as ZIP
 534          * based formats like OPC/ODF files).
 535          *
 536          * @param $file String: the file to check
 537          * @param $ext Mixed: the file extension, or true (default) to extract it from the filename.
 538          *             Set it to false to ignore the extension. DEPRECATED! Set to false, use
 539          *             improveTypeFromExtension($mime, $ext) later to improve mime type.
 540          *
 541          * @return string the mime type of $file
 542          */
 543         public function guessMimeType( $file, $ext = true ) {
 544                 if ( $ext ) { // TODO: make $ext default to false. Or better, remove it.
 545                         wfDebug( __METHOD__.": WARNING: use of the \$ext parameter is deprecated. " .
 546                                 "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 547                 }
 548
 549                 $mime = $this->doGuessMimeType( $file, $ext );
 550
 551                 if( !$mime ) {
 552                         wfDebug( __METHOD__.": internal type detection failed for $file (.$ext)...\n" );
 553                         $mime = $this->detectMimeType( $file, $ext );
 554                 }
 555
 556                 if ( isset( $this->mMimeTypeAliases[$mime] ) ) {
 557                         $mime = $this->mMimeTypeAliases[$mime];
 558                 }
 559
 560                 wfDebug(__METHOD__.": guessed mime type of $file: $mime\n");
 561                 return $mime;
 562         }
 563
 564         /**
 565          * Guess the mime type from the file contents.
 566          *
 567          * @param string $file
 568          * @param mixed $ext
 569          * @return bool|string
 570          */
 571         private function doGuessMimeType( $file, $ext ) { // TODO: remove $ext param
 572                 // Read a chunk of the file
 573                 wfSuppressWarnings();
 574                 // @todo FIXME: Shouldn't this be rb?
 575                 $f = fopen( $file, 'rt' );
 576                 wfRestoreWarnings();
 577
 578                 if( !$f ) {
 579                         return 'unknown/unknown';
 580                 }
 581                 $head = fread( $f, 1024 );
 582                 fseek( $f, -65558, SEEK_END );
 583                 $tail = fread( $f, 65558 ); // 65558 = maximum size of a zip EOCDR
 584                 fclose( $f );
 585
 586                 wfDebug( __METHOD__ . ": analyzing head and tail of $file for magic numbers.\n" );
 587
 588                 // Hardcode a few magic number checks...
 589                 $headers = array(
 590                         // Multimedia...
 591                         'MThd'             => 'audio/midi',
 592                         'OggS'             => 'application/ogg',
 593
 594                         // Image formats...
 595                         // Note that WMF may have a bare header, no magic number.
 596                         "\x01\x00\x09\x00" => 'application/x-msmetafile', // Possibly prone to false positives?
 597                         "\xd7\xcd\xc6\x9a" => 'application/x-msmetafile',
 598                         '%PDF'             => 'application/pdf',
 599                         'gimp xcf'         => 'image/x-xcf',
 600
 601                         // Some forbidden fruit...
 602                         'MZ'               => 'application/octet-stream', // DOS/Windows executable
 603                         "\xca\xfe\xba\xbe" => 'application/octet-stream', // Mach-O binary
 604                         "\x7fELF"          => 'application/octet-stream', // ELF binary
 605                 );
 606
 607                 foreach ( $headers as $magic => $candidate ) {
 608                         if ( strncmp( $head, $magic, strlen( $magic ) ) == 0 ) {
 609                                 wfDebug( __METHOD__ . ": magic header in $file recognized as $candidate\n" );
 610                                 return $candidate;
 611                         }
 612                 }
 613
 614                 /* Look for WebM and Matroska files */
 615                 if ( strncmp( $head, pack( "C4", 0x1a, 0x45, 0xdf, 0xa3 ), 4 ) == 0 ) {
 616                         $doctype = strpos( $head, "\x42\x82" );
 617                         if ( $doctype ) {
 618                                 // Next byte is datasize, then data (sizes larger than 1 byte are very stupid muxers)
 619                                 $data = substr($head, $doctype+3, 8);
 620                                 if ( strncmp( $data, "matroska", 8 ) == 0 ) {
 621                                         wfDebug( __METHOD__ . ": recognized file as video/x-matroska\n" );
 622                                         return "video/x-matroska";
 623                                 } elseif ( strncmp( $data, "webm", 4 ) == 0 ) {
 624                                         wfDebug( __METHOD__ . ": recognized file as video/webm\n" );
 625                                         return "video/webm";
 626                                 }
 627                         }
 628                         wfDebug( __METHOD__ . ": unknown EBML file\n" );
 629                         return "unknown/unknown";
 630                 }
 631
 632                 /* Look for WebP */
 633                 if ( strncmp( $head, "RIFF", 4 ) == 0 && strncmp( substr( $head, 8, 8), "WEBPVP8 ", 8 ) == 0 ) {
 634                         wfDebug( __METHOD__ . ": recognized file as image/webp\n" );
 635                         return "image/webp";
 636                 }
 637
 638                 /**
 639                  * Look for PHP.  Check for this before HTML/XML...  Warning: this is a
 640                  * heuristic, and won't match a file with a lot of non-PHP before.  It
 641                  * will also match text files which could be PHP. :)
 642                  *
 643                  * @todo FIXME: For this reason, the check is probably useless -- an attacker
 644                  * could almost certainly just pad the file with a lot of nonsense to
 645                  * circumvent the check in any case where it would be a security
 646                  * problem.  On the other hand, it causes harmful false positives (bug
 647                  * 16583).  The heuristic has been cut down to exclude three-character
 648                  * strings like "<? ", but should it be axed completely?
 649                  */
 650                 if ( ( strpos( $head, '<?php' ) !== false ) ||
 651
 652                     ( strpos( $head, "<\x00?\x00p\x00h\x00p" ) !== false ) ||
 653                     ( strpos( $head, "<\x00?\x00 " ) !== false ) ||
 654                     ( strpos( $head, "<\x00?\x00\n" ) !== false ) ||
 655                     ( strpos( $head, "<\x00?\x00\t" ) !== false ) ||
 656                     ( strpos( $head, "<\x00?\x00=" ) !== false ) ) {
 657
 658                         wfDebug( __METHOD__ . ": recognized $file as application/x-php\n" );
 659                         return 'application/x-php';
 660                 }
 661
 662                 /**
 663                  * look for XML formats (XHTML and SVG)
 664                  */
 665                 $xml = new XmlTypeCheck( $file );
 666                 if ( $xml->wellFormed ) {
 667                         global $wgXMLMimeTypes;
 668                         if ( isset( $wgXMLMimeTypes[$xml->getRootElement()] ) ) {
 669                                 return $wgXMLMimeTypes[$xml->getRootElement()];
 670                         } else {
 671                                 return 'application/xml';
 672                         }
 673                 }
 674
 675                 /**
 676                  * look for shell scripts
 677                  */
 678                 $script_type = null;
 679
 680                 # detect by shebang
 681                 if ( substr( $head, 0, 2) == "#!" ) {
 682                         $script_type = "ASCII";
 683                 } elseif ( substr( $head, 0, 5) == "\xef\xbb\xbf#!" ) {
 684                         $script_type = "UTF-8";
 685                 } elseif ( substr( $head, 0, 7) == "\xfe\xff\x00#\x00!" ) {
 686                         $script_type = "UTF-16BE";
 687                 } elseif ( substr( $head, 0, 7 ) == "\xff\xfe#\x00!" ) {
 688                         $script_type= "UTF-16LE";
 689                 }
 690
 691                 if ( $script_type ) {
 692                         if ( $script_type !== "UTF-8" && $script_type !== "ASCII") {
 693                                 // Quick and dirty fold down to ASCII!
 694                                 $pack = array( 'UTF-16BE' => 'n*', 'UTF-16LE' => 'v*' );
 695                                 $chars = unpack( $pack[$script_type], substr( $head, 2 ) );
 696                                 $head = '';
 697                                 foreach( $chars as $codepoint ) {
 698                                         if( $codepoint < 128 ) {
 699                                                 $head .= chr( $codepoint );
 700                                         } else {
 701                                                 $head .= '?';
 702                                         }
 703                                 }
 704                         }
 705
 706                         $match = array();
 707
 708                         if ( preg_match( '%/?([^\s]+/)(\w+)%', $head, $match ) ) {
 709                                 $mime = "application/x-{$match[2]}";
 710                                 wfDebug( __METHOD__.": shell script recognized as $mime\n" );
 711                                 return $mime;
 712                         }
 713                 }
 714
 715                 // Check for ZIP variants (before getimagesize)
 716                 if ( strpos( $tail, "PK\x05\x06" ) !== false ) {
 717                         wfDebug( __METHOD__.": ZIP header present in $file\n" );
 718                         return $this->detectZipType( $head, $tail, $ext );
 719                 }
 720
 721                 wfSuppressWarnings();
 722                 $gis = getimagesize( $file );
 723                 wfRestoreWarnings();
 724
 725                 if( $gis && isset( $gis['mime'] ) ) {
 726                         $mime = $gis['mime'];
 727                         wfDebug( __METHOD__.": getimagesize detected $file as $mime\n" );
 728                         return $mime;
 729                 }
 730
 731                 // Also test DjVu
 732                 $deja = new DjVuImage( $file );
 733                 if( $deja->isValid() ) {
 734                         wfDebug( __METHOD__.": detected $file as image/vnd.djvu\n" );
 735                         return 'image/vnd.djvu';
 736                 }
 737
 738                 return false;
 739         }
 740
 741         /**
 742          * Detect application-specific file type of a given ZIP file from its
 743          * header data.  Currently works for OpenDocument and OpenXML types...
 744          * If can't tell, returns 'application/zip'.
 745          *
 746          * @param $header String: some reasonably-sized chunk of file header
 747          * @param $tail   String: the tail of the file
 748          * @param $ext Mixed: the file extension, or true to extract it from the filename.
 749          *             Set it to false (default) to ignore the extension. DEPRECATED! Set to false,
 750          *             use improveTypeFromExtension($mime, $ext) later to improve mime type.
 751          *
 752          * @return string
 753          */
 754         function detectZipType( $header, $tail = null, $ext = false ) {
 755                 if( $ext ) { # TODO: remove $ext param
 756                         wfDebug( __METHOD__.": WARNING: use of the \$ext parameter is deprecated. " .
 757                                 "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 758                 }
 759
 760                 $mime = 'application/zip';
 761                 $opendocTypes = array(
 762                         'chart-template',
 763                         'chart',
 764                         'formula-template',
 765                         'formula',
 766                         'graphics-template',
 767                         'graphics',
 768                         'image-template',
 769                         'image',
 770                         'presentation-template',
 771                         'presentation',
 772                         'spreadsheet-template',
 773                         'spreadsheet',
 774                         'text-template',
 775                         'text-master',
 776                         'text-web',
 777                         'text' );
 778
 779                 // http://lists.oasis-open.org/archives/office/200505/msg00006.html
 780                 $types = '(?:' . implode( '|', $opendocTypes ) . ')';
 781                 $opendocRegex = "/^mimetype(application\/vnd\.oasis\.opendocument\.$types)/";
 782
 783                 $openxmlRegex = "/^\[Content_Types\].xml/";
 784
 785                 if ( preg_match( $opendocRegex, substr( $header, 30 ), $matches ) ) {
 786                         $mime = $matches[1];
 787                         wfDebug( __METHOD__.": detected $mime from ZIP archive\n" );
 788                 } elseif ( preg_match( $openxmlRegex, substr( $header, 30 ) ) ) {
 789                         $mime = "application/x-opc+zip";
 790                         # TODO: remove the block below, as soon as improveTypeFromExtension is used everywhere
 791                         if ( $ext !== true && $ext !== false ) {
 792                                 /** This is the mode used by getPropsFromPath
 793                                 * These mime's are stored in the database, where we don't really want
 794                                 * x-opc+zip, because we use it only for internal purposes
 795                                 */
 796                                 if ( $this->isMatchingExtension( $ext, $mime) ) {
 797                                         /* A known file extension for an OPC file,
 798                                         * find the proper mime type for that file extension */
 799                                         $mime = $this->guessTypesForExtension( $ext );
 800                                 } else {
 801                                         $mime = "application/zip";
 802                                 }
 803                         }
 804                         wfDebug( __METHOD__.": detected an Open Packaging Conventions archive: $mime\n" );
 805                 } elseif ( substr( $header, 0, 8 ) == "\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" &&
 806                                 ($headerpos = strpos( $tail, "PK\x03\x04" ) ) !== false &&
 807                                 preg_match( $openxmlRegex, substr( $tail, $headerpos + 30 ) ) ) {
 808                         if ( substr( $header, 512, 4) == "\xEC\xA5\xC1\x00" ) {
 809                                 $mime = "application/msword";
 810                         }
 811                         switch( substr( $header, 512, 6) ) {
 812                                 case "\xEC\xA5\xC1\x00\x0E\x00":
 813                                 case "\xEC\xA5\xC1\x00\x1C\x00":
 814                                 case "\xEC\xA5\xC1\x00\x43\x00":
 815                                         $mime = "application/vnd.ms-powerpoint";
 816                                         break;
 817                                 case "\xFD\xFF\xFF\xFF\x10\x00":
 818                                 case "\xFD\xFF\xFF\xFF\x1F\x00":
 819                                 case "\xFD\xFF\xFF\xFF\x22\x00":
 820                                 case "\xFD\xFF\xFF\xFF\x23\x00":
 821                                 case "\xFD\xFF\xFF\xFF\x28\x00":
 822                                 case "\xFD\xFF\xFF\xFF\x29\x00":
 823                                 case "\xFD\xFF\xFF\xFF\x10\x02":
 824                                 case "\xFD\xFF\xFF\xFF\x1F\x02":
 825                                 case "\xFD\xFF\xFF\xFF\x22\x02":
 826                                 case "\xFD\xFF\xFF\xFF\x23\x02":
 827                                 case "\xFD\xFF\xFF\xFF\x28\x02":
 828                                 case "\xFD\xFF\xFF\xFF\x29\x02":
 829                                         $mime = "application/vnd.msexcel";
 830                                         break;
 831                         }
 832
 833                         wfDebug( __METHOD__.": detected a MS Office document with OPC trailer\n");
 834                 } else {
 835                         wfDebug( __METHOD__.": unable to identify type of ZIP archive\n" );
 836                 }
 837                 return $mime;
 838         }
 839
 840         /**
 841          * Internal mime type detection. Detection is done using an external
 842          * program, if $wgMimeDetectorCommand is set. Otherwise, the fileinfo
 843          * extension and mime_content_type are tried (in this order), if they
 844          * are available. If the dections fails and $ext is not false, the mime
 845          * type is guessed from the file extension, using guessTypesForExtension.
 846          *
 847          * If the mime type is still unknown, getimagesize is used to detect the
 848          * mime type if the file is an image. If no mime type can be determined,
 849          * this function returns 'unknown/unknown'.
 850          *
 851          * @param $file String: the file to check
 852          * @param $ext Mixed: the file extension, or true (default) to extract it from the filename.
 853          *             Set it to false to ignore the extension. DEPRECATED! Set to false, use
 854          *             improveTypeFromExtension($mime, $ext) later to improve mime type.
 855          *
 856          * @return string the mime type of $file
 857          */
 858         private function detectMimeType( $file, $ext = true ) {
 859                 global $wgMimeDetectorCommand;
 860
 861                 if ( $ext ) { # TODO:  make $ext default to false. Or better, remove it.
 862                         wfDebug( __METHOD__.": WARNING: use of the \$ext parameter is deprecated. Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 863                 }
 864
 865                 $m = null;
 866                 if ( $wgMimeDetectorCommand ) {
 867                         // @todo FIXME: Use wfShellExec
 868                         $fn = wfEscapeShellArg( $file );
 869                         $m = `$wgMimeDetectorCommand $fn`;
 870                 } elseif ( function_exists( "finfo_open" ) && function_exists( "finfo_file" ) ) {
 871
 872                         # This required the fileinfo extension by PECL,
 873                         # see http://pecl.php.net/package/fileinfo
 874                         # This must be compiled into PHP
 875                         #
 876                         # finfo is the official replacement for the deprecated
 877                         # mime_content_type function, see below.
 878                         #
 879                         # If you may need to load the fileinfo extension at runtime, set
 880                         # $wgLoadFileinfoExtension in LocalSettings.php
 881
 882                         $mime_magic_resource = finfo_open( FILEINFO_MIME ); /* return mime type ala mimetype extension */
 883
 884                         if ( $mime_magic_resource ) {
 885                                 $m = finfo_file( $mime_magic_resource, $file );
 886                                 finfo_close( $mime_magic_resource );
 887                         } else {
 888                                 wfDebug( __METHOD__.": finfo_open failed on ".FILEINFO_MIME."!\n" );
 889                         }
 890                 } elseif ( function_exists( "mime_content_type" ) ) {
 891
 892                         # NOTE: this function is available since PHP 4.3.0, but only if
 893                         # PHP was compiled with --with-mime-magic or, before 4.3.2, with --enable-mime-magic.
 894                         #
 895                         # On Windows, you must set mime_magic.magicfile in php.ini to point to the mime.magic file bundeled with PHP;
 896                         # sometimes, this may even be needed under linus/unix.
 897                         #
 898                         # Also note that this has been DEPRECATED in favor of the fileinfo extension by PECL, see above.
 899                         # see http://www.php.net/manual/en/ref.mime-magic.php for details.
 900
 901                         $m = mime_content_type($file);
 902                 } else {
 903                         wfDebug( __METHOD__.": no magic mime detector found!\n" );
 904                 }
 905
 906                 if ( $m ) {
 907                         # normalize
 908                         $m = preg_replace( '![;, ].*$!', '', $m ); #strip charset, etc
 909                         $m = trim( $m );
 910                         $m = strtolower( $m );
 911
 912                         if ( strpos( $m, 'unknown' ) !== false ) {
 913                                 $m = null;
 914                         } else {
 915                                 wfDebug( __METHOD__.": magic mime type of $file: $m\n" );
 916                                 return $m;
 917                         }
 918                 }
 919
 920                 // If desired, look at extension as a fallback.
 921                 if ( $ext === true ) {
 922                         $i = strrpos( $file, '.' );
 923                         $ext = strtolower( $i ? substr( $file, $i + 1 ) : '' );
 924                 }
 925                 if ( $ext ) {
 926                         if( $this->isRecognizableExtension( $ext ) ) {
 927                                 wfDebug( __METHOD__. ": refusing to guess mime type for .$ext file, we should have recognized it\n" );
 928                         } else {
 929                                 $m = $this->guessTypesForExtension( $ext );
 930                                 if ( $m ) {
 931                                         wfDebug( __METHOD__.": extension mime type of $file: $m\n" );
 932                                         return $m;
 933                                 }
 934                         }
 935                 }
 936
 937                 // Unknown type
 938                 wfDebug( __METHOD__ . ": failed to guess mime type for $file!\n" );
 939                 return 'unknown/unknown';
 940         }
 941
 942         /**
 943          * Determine the media type code for a file, using its mime type, name and
 944          * possibly its contents.
 945          *
 946          * This function relies on the findMediaType(), mapping extensions and mime
 947          * types to media types.
 948          *
 949          * @todo analyse file if need be
 950          * @todo look at multiple extension, separately and together.
 951          *
 952          * @param $path String: full path to the image file, in case we have to look at the contents
 953          *        (if null, only the mime type is used to determine the media type code).
 954          * @param $mime String: mime type. If null it will be guessed using guessMimeType.
 955          *
 956          * @return (int?string?) a value to be used with the MEDIATYPE_xxx constants.
 957          */
 958         function getMediaType( $path = null, $mime = null ) {
 959                 if( !$mime && !$path ) {
 960                         return MEDIATYPE_UNKNOWN;
 961                 }
 962
 963                 // If mime type is unknown, guess it
 964                 if( !$mime ) {
 965                         $mime = $this->guessMimeType( $path, false );
 966                 }
 967
 968                 // Special code for ogg - detect if it's video (theora),
 969                 // else label it as sound.
 970                 if ( $mime == 'application/ogg' && file_exists( $path ) ) {
 971
 972                         // Read a chunk of the file
 973                         $f = fopen( $path, "rt" );
 974                         if ( !$f ) return MEDIATYPE_UNKNOWN;
 975                         $head = fread( $f, 256 );
 976                         fclose( $f );
 977
 978                         $head = strtolower( $head );
 979
 980                         // This is an UGLY HACK, file should be parsed correctly
 981                         if ( strpos( $head, 'theora' ) !== false ) return MEDIATYPE_VIDEO;
 982                         elseif ( strpos( $head, 'vorbis' ) !== false ) return MEDIATYPE_AUDIO;
 983                         elseif ( strpos( $head, 'flac' ) !== false ) return MEDIATYPE_AUDIO;
 984                         elseif ( strpos( $head, 'speex' ) !== false ) return MEDIATYPE_AUDIO;
 985                         else return MEDIATYPE_MULTIMEDIA;
 986                 }
 987
 988                 // Check for entry for full mime type
 989                 if( $mime ) {
 990                         $type = $this->findMediaType( $mime );
 991                         if ( $type !== MEDIATYPE_UNKNOWN ) {
 992                                 return $type;
 993                         }
 994                 }
 995
 996                 // Check for entry for file extension
 997                 if ( $path ) {
 998                         $i = strrpos( $path, '.' );
 999                         $e = strtolower( $i ? substr( $path, $i + 1 ) : '' );
1000
1001                         // TODO: look at multi-extension if this fails, parse from full path
1002                         $type = $this->findMediaType( '.' . $e );
1003                         if ( $type !== MEDIATYPE_UNKNOWN ) {
1004                                 return $type;
1005                         }
1006                 }
1007
1008                 // Check major mime type
1009                 if ( $mime ) {
1010                         $i = strpos( $mime, '/' );
1011                         if ( $i !== false ) {
1012                                 $major = substr( $mime, 0, $i );
1013                                 $type = $this->findMediaType( $major );
1014                                 if ( $type !== MEDIATYPE_UNKNOWN ) {
1015                                         return $type;
1016                                 }
1017                         }
1018                 }
1019
1020                 if( !$type ) {
1021                         $type = MEDIATYPE_UNKNOWN;
1022                 }
1023
1024                 return $type;
1025         }
1026
1027         /**
1028          * Returns a media code matching the given mime type or file extension.
1029          * File extensions are represented by a string starting with a dot (.) to
1030          * distinguish them from mime types.
1031          *
1032          * This funktion relies on the mapping defined by $this->mMediaTypes
1033          * @access private
1034          * @return int|string
1035          */
1036         function findMediaType( $extMime ) {
1037                 if ( strpos( $extMime, '.' ) === 0 ) {
1038                         // If it's an extension, look up the mime types
1039                         $m = $this->getTypesForExtension( substr( $extMime, 1 ) );
1040                         if ( !$m ) {
1041                                 return MEDIATYPE_UNKNOWN;
1042                         }
1043
1044                         $m = explode( ' ', $m );
1045                 } else {
1046                         // Normalize mime type
1047                         if ( isset( $this->mMimeTypeAliases[$extMime] ) ) {
1048                                 $extMime = $this->mMimeTypeAliases[$extMime];
1049                         }
1050
1051                         $m = array( $extMime );
1052                 }
1053
1054                 foreach ( $m as $mime ) {
1055                         foreach ( $this->mMediaTypes as $type => $codes ) {
1056                                 if ( in_array($mime, $codes, true ) ) {
1057                                         return $type;
1058                                 }
1059                         }
1060                 }
1061
1062                 return MEDIATYPE_UNKNOWN;
1063         }
1064
1065         /**
1066          * Get the MIME types that various versions of Internet Explorer would
1067          * detect from a chunk of the content.
1068          *
1069          * @param $fileName String: the file name (unused at present)
1070          * @param $chunk String: the first 256 bytes of the file
1071          * @param $proposed String: the MIME type proposed by the server
1072          * @return Array
1073          */
1074         public function getIEMimeTypes( $fileName, $chunk, $proposed ) {
1075                 $ca = $this->getIEContentAnalyzer();
1076                 return $ca->getRealMimesFromData( $fileName, $chunk, $proposed );
1077         }
1078
1079         /**
1080          * Get a cached instance of IEContentAnalyzer
1081          *
1082          * @return IEContentAnalyzer
1083          */
1084         protected function getIEContentAnalyzer() {
1085                 if ( is_null( $this->mIEAnalyzer ) ) {
1086                         $this->mIEAnalyzer = new IEContentAnalyzer;
1087                 }
1088                 return $this->mIEAnalyzer;
1089         }
1090 }