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