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