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          * @param $mime string
 429          *
 430          * @return bool
 431          */
 432         public function isPHPImageType( $mime ) {
 433                 // As defined by imagegetsize and image_type_to_mime
 434                 static $types = array(
 435                         'image/gif', 'image/jpeg', 'image/png',
 436                         'image/x-bmp', 'image/xbm', 'image/tiff',
 437                         'image/jp2', 'image/jpeg2000', 'image/iff',
 438                         'image/xbm', 'image/x-xbitmap',
 439                         'image/vnd.wap.wbmp', 'image/vnd.xiff',
 440                         'image/x-photoshop',
 441                         'application/x-shockwave-flash',
 442                 );
 443
 444                 return in_array( $mime, $types );
 445         }
 446
 447         /**
 448          * Returns true if the extension represents a type which can
 449          * be reliably detected from its content. Use this to determine
 450          * whether strict content checks should be applied to reject
 451          * invalid uploads; if we can't identify the type we won't
 452          * be able to say if it's invalid.
 453          *
 454          * @todo Be more accurate when using fancy mime detector plugins;
 455          *       right now this is the bare minimum getimagesize() list.
 456          * @return bool
 457          */
 458         function isRecognizableExtension( $extension ) {
 459                 static $types = array(
 460                         // Types recognized by getimagesize()
 461                         'gif', 'jpeg', 'jpg', 'png', 'swf', 'psd',
 462                         'bmp', 'tiff', 'tif', 'jpc', 'jp2',
 463                         'jpx', 'jb2', 'swc', 'iff', 'wbmp',
 464                         'xbm',
 465
 466                         // Formats we recognize magic numbers for
 467                         'djvu', 'ogx', 'ogg', 'ogv', 'oga', 'spx',
 468                         'mid', 'pdf', 'wmf', 'xcf', 'webm', 'mkv', 'mka',
 469                         'webp',
 470
 471                         // XML formats we sure hope we recognize reliably
 472                         'svg',
 473                 );
 474                 return in_array( strtolower( $extension ), $types );
 475         }
 476
 477         /**
 478          * Improves a mime type using the file extension. Some file formats are very generic,
 479          * so their mime type is not very meaningful. A more useful mime type can be derived
 480          * by looking at the file extension. Typically, this method would be called on the
 481          * result of guessMimeType().
 482          *
 483          * Currently, this method does the following:
 484          *
 485          * If $mime is "unknown/unknown" and isRecognizableExtension( $ext ) returns false,
 486          * return the result of guessTypesForExtension($ext).
 487          *
 488          * If $mime is "application/x-opc+zip" and isMatchingExtension( $ext, $mime )
 489          * gives true, return the result of guessTypesForExtension($ext).
 490          *
 491          * @param $mime String: the mime type, typically guessed from a file's content.
 492          * @param $ext String: the file extension, as taken from the file name
 493          *
 494          * @return string the mime type
 495          */
 496         public function improveTypeFromExtension( $mime, $ext ) {
 497                 if ( $mime === 'unknown/unknown' ) {
 498                         if ( $this->isRecognizableExtension( $ext ) ) {
 499                                 wfDebug( __METHOD__. ': refusing to guess mime type for .' .
 500                                         "$ext file, we should have recognized it\n" );
 501                         } else {
 502                                 // Not something we can detect, so simply
 503                                 // trust the file extension
 504                                 $mime = $this->guessTypesForExtension( $ext );
 505                         }
 506                 }
 507                 elseif ( $mime === 'application/x-opc+zip' ) {
 508                         if ( $this->isMatchingExtension( $ext, $mime ) ) {
 509                                 // A known file extension for an OPC file,
 510                                 // find the proper mime type for that file extension
 511                                 $mime = $this->guessTypesForExtension( $ext );
 512                         } else {
 513                                 wfDebug( __METHOD__. ": refusing to guess better type for $mime file, " .
 514                                         ".$ext is not a known OPC extension.\n" );
 515                                 $mime = 'application/zip';
 516                         }
 517                 }
 518
 519                 if ( isset( $this->mMimeTypeAliases[$mime] ) ) {
 520                         $mime = $this->mMimeTypeAliases[$mime];
 521                 }
 522
 523                 wfDebug(__METHOD__.": improved mime type for .$ext: $mime\n");
 524                 return $mime;
 525         }
 526
 527         /**
 528          * Mime type detection. This uses detectMimeType to detect the mime type
 529          * of the file, but applies additional checks to determine some well known
 530          * file formats that may be missed or misinterpreter by the default mime
 531          * detection (namely XML based formats like XHTML or SVG, as well as ZIP
 532          * based formats like OPC/ODF files).
 533          *
 534          * @param $file String: the file to check
 535          * @param $ext Mixed: the file extension, or true (default) to extract it from the filename.
 536          *             Set it to false to ignore the extension. DEPRECATED! Set to false, use
 537          *             improveTypeFromExtension($mime, $ext) later to improve mime type.
 538          *
 539          * @return string the mime type of $file
 540          */
 541         public function guessMimeType( $file, $ext = true ) {
 542                 if ( $ext ) { // TODO: make $ext default to false. Or better, remove it.
 543                         wfDebug( __METHOD__.": WARNING: use of the \$ext parameter is deprecated. " .
 544                                 "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 545                 }
 546
 547                 $mime = $this->doGuessMimeType( $file, $ext );
 548
 549                 if( !$mime ) {
 550                         wfDebug( __METHOD__.": internal type detection failed for $file (.$ext)...\n" );
 551                         $mime = $this->detectMimeType( $file, $ext );
 552                 }
 553
 554                 if ( isset( $this->mMimeTypeAliases[$mime] ) ) {
 555                         $mime = $this->mMimeTypeAliases[$mime];
 556                 }
 557
 558                 wfDebug(__METHOD__.": guessed mime type of $file: $mime\n");
 559                 return $mime;
 560         }
 561
 562         /**
 563          * Guess the mime type from the file contents.
 564          *
 565          * @param string $file
 566          * @param mixed $ext
 567          */
 568         private function doGuessMimeType( $file, $ext ) { // TODO: remove $ext param
 569                 // Read a chunk of the file
 570                 wfSuppressWarnings();
 571                 // @todo FIXME: Shouldn't this be rb?
 572                 $f = fopen( $file, 'rt' );
 573                 wfRestoreWarnings();
 574
 575                 if( !$f ) {
 576                         return 'unknown/unknown';
 577                 }
 578                 $head = fread( $f, 1024 );
 579                 fseek( $f, -65558, SEEK_END );
 580                 $tail = fread( $f, 65558 ); // 65558 = maximum size of a zip EOCDR
 581                 fclose( $f );
 582
 583                 wfDebug( __METHOD__ . ": analyzing head and tail of $file for magic numbers.\n" );
 584
 585                 // Hardcode a few magic number checks...
 586                 $headers = array(
 587                         // Multimedia...
 588                         'MThd'             => 'audio/midi',
 589                         'OggS'             => 'application/ogg',
 590
 591                         // Image formats...
 592                         // Note that WMF may have a bare header, no magic number.
 593                         "\x01\x00\x09\x00" => 'application/x-msmetafile', // Possibly prone to false positives?
 594                         "\xd7\xcd\xc6\x9a" => 'application/x-msmetafile',
 595                         '%PDF'             => 'application/pdf',
 596                         'gimp xcf'         => 'image/x-xcf',
 597
 598                         // Some forbidden fruit...
 599                         'MZ'               => 'application/octet-stream', // DOS/Windows executable
 600                         "\xca\xfe\xba\xbe" => 'application/octet-stream', // Mach-O binary
 601                         "\x7fELF"          => 'application/octet-stream', // ELF binary
 602                 );
 603
 604                 foreach ( $headers as $magic => $candidate ) {
 605                         if ( strncmp( $head, $magic, strlen( $magic ) ) == 0 ) {
 606                                 wfDebug( __METHOD__ . ": magic header in $file recognized as $candidate\n" );
 607                                 return $candidate;
 608                         }
 609                 }
 610
 611                 /* Look for WebM and Matroska files */
 612                 if ( strncmp( $head, pack( "C4", 0x1a, 0x45, 0xdf, 0xa3 ), 4 ) == 0 ) {
 613                         $doctype = strpos( $head, "\x42\x82" );
 614                         if ( $doctype ) {
 615                                 // Next byte is datasize, then data (sizes larger than 1 byte are very stupid muxers)
 616                                 $data = substr($head, $doctype+3, 8);
 617                                 if ( strncmp( $data, "matroska", 8 ) == 0 ) {
 618                                         wfDebug( __METHOD__ . ": recognized file as video/x-matroska\n" );
 619                                         return "video/x-matroska";
 620                                 } elseif ( strncmp( $data, "webm", 4 ) == 0 ) {
 621                                         wfDebug( __METHOD__ . ": recognized file as video/webm\n" );
 622                                         return "video/webm";
 623                                 }
 624                         }
 625                         wfDebug( __METHOD__ . ": unknown EBML file\n" );
 626                         return "unknown/unknown";
 627                 }
 628
 629                 /* Look for WebP */
 630                 if ( strncmp( $head, "RIFF", 4 ) == 0 && strncmp( substr( $head, 8, 8), "WEBPVP8 ", 8 ) == 0 ) {
 631                         wfDebug( __METHOD__ . ": recognized file as image/webp\n" );
 632                         return "image/webp";
 633                 }
 634
 635                 /**
 636                  * Look for PHP.  Check for this before HTML/XML...  Warning: this is a
 637                  * heuristic, and won't match a file with a lot of non-PHP before.  It
 638                  * will also match text files which could be PHP. :)
 639                  *
 640                  * @todo FIXME: For this reason, the check is probably useless -- an attacker
 641                  * could almost certainly just pad the file with a lot of nonsense to
 642                  * circumvent the check in any case where it would be a security
 643                  * problem.  On the other hand, it causes harmful false positives (bug
 644                  * 16583).  The heuristic has been cut down to exclude three-character
 645                  * strings like "<? ", but should it be axed completely?
 646                  */
 647                 if ( ( strpos( $head, '<?php' ) !== false ) ||
 648
 649                     ( strpos( $head, "<\x00?\x00p\x00h\x00p" ) !== false ) ||
 650                     ( strpos( $head, "<\x00?\x00 " ) !== false ) ||
 651                     ( strpos( $head, "<\x00?\x00\n" ) !== false ) ||
 652                     ( strpos( $head, "<\x00?\x00\t" ) !== false ) ||
 653                     ( strpos( $head, "<\x00?\x00=" ) !== false ) ) {
 654
 655                         wfDebug( __METHOD__ . ": recognized $file as application/x-php\n" );
 656                         return 'application/x-php';
 657                 }
 658
 659                 /**
 660                  * look for XML formats (XHTML and SVG)
 661                  */
 662                 $xml = new XmlTypeCheck( $file );
 663                 if ( $xml->wellFormed ) {
 664                         global $wgXMLMimeTypes;
 665                         if ( isset( $wgXMLMimeTypes[$xml->getRootElement()] ) ) {
 666                                 return $wgXMLMimeTypes[$xml->getRootElement()];
 667                         } else {
 668                                 return 'application/xml';
 669                         }
 670                 }
 671
 672                 /**
 673                  * look for shell scripts
 674                  */
 675                 $script_type = null;
 676
 677                 # detect by shebang
 678                 if ( substr( $head, 0, 2) == "#!" ) {
 679                         $script_type = "ASCII";
 680                 } elseif ( substr( $head, 0, 5) == "\xef\xbb\xbf#!" ) {
 681                         $script_type = "UTF-8";
 682                 } elseif ( substr( $head, 0, 7) == "\xfe\xff\x00#\x00!" ) {
 683                         $script_type = "UTF-16BE";
 684                 } elseif ( substr( $head, 0, 7 ) == "\xff\xfe#\x00!" ) {
 685                         $script_type= "UTF-16LE";
 686                 }
 687
 688                 if ( $script_type ) {
 689                         if ( $script_type !== "UTF-8" && $script_type !== "ASCII") {
 690                                 // Quick and dirty fold down to ASCII!
 691                                 $pack = array( 'UTF-16BE' => 'n*', 'UTF-16LE' => 'v*' );
 692                                 $chars = unpack( $pack[$script_type], substr( $head, 2 ) );
 693                                 $head = '';
 694                                 foreach( $chars as $codepoint ) {
 695                                         if( $codepoint < 128 ) {
 696                                                 $head .= chr( $codepoint );
 697                                         } else {
 698                                                 $head .= '?';
 699                                         }
 700                                 }
 701                         }
 702
 703                         $match = array();
 704
 705                         if ( preg_match( '%/?([^\s]+/)(\w+)%', $head, $match ) ) {
 706                                 $mime = "application/x-{$match[2]}";
 707                                 wfDebug( __METHOD__.": shell script recognized as $mime\n" );
 708                                 return $mime;
 709                         }
 710                 }
 711
 712                 // Check for ZIP variants (before getimagesize)
 713                 if ( strpos( $tail, "PK\x05\x06" ) !== false ) {
 714                         wfDebug( __METHOD__.": ZIP header present in $file\n" );
 715                         return $this->detectZipType( $head, $tail, $ext );
 716                 }
 717
 718                 wfSuppressWarnings();
 719                 $gis = getimagesize( $file );
 720                 wfRestoreWarnings();
 721
 722                 if( $gis && isset( $gis['mime'] ) ) {
 723                         $mime = $gis['mime'];
 724                         wfDebug( __METHOD__.": getimagesize detected $file as $mime\n" );
 725                         return $mime;
 726                 }
 727
 728                 // Also test DjVu
 729                 $deja = new DjVuImage( $file );
 730                 if( $deja->isValid() ) {
 731                         wfDebug( __METHOD__.": detected $file as image/vnd.djvu\n" );
 732                         return 'image/vnd.djvu';
 733                 }
 734
 735                 return false;
 736         }
 737
 738         /**
 739          * Detect application-specific file type of a given ZIP file from its
 740          * header data.  Currently works for OpenDocument and OpenXML types...
 741          * If can't tell, returns 'application/zip'.
 742          *
 743          * @param $header String: some reasonably-sized chunk of file header
 744          * @param $tail   String: the tail of the file
 745          * @param $ext Mixed: the file extension, or true to extract it from the filename.
 746          *             Set it to false (default) to ignore the extension. DEPRECATED! Set to false,
 747          *             use improveTypeFromExtension($mime, $ext) later to improve mime type.
 748          *
 749          * @return string
 750          */
 751         function detectZipType( $header, $tail = null, $ext = false ) {
 752                 if( $ext ) { # TODO: remove $ext param
 753                         wfDebug( __METHOD__.": WARNING: use of the \$ext parameter is deprecated. " .
 754                                 "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 755                 }
 756
 757                 $mime = 'application/zip';
 758                 $opendocTypes = array(
 759                         'chart-template',
 760                         'chart',
 761                         'formula-template',
 762                         'formula',
 763                         'graphics-template',
 764                         'graphics',
 765                         'image-template',
 766                         'image',
 767                         'presentation-template',
 768                         'presentation',
 769                         'spreadsheet-template',
 770                         'spreadsheet',
 771                         'text-template',
 772                         'text-master',
 773                         'text-web',
 774                         'text' );
 775
 776                 // http://lists.oasis-open.org/archives/office/200505/msg00006.html
 777                 $types = '(?:' . implode( '|', $opendocTypes ) . ')';
 778                 $opendocRegex = "/^mimetype(application\/vnd\.oasis\.opendocument\.$types)/";
 779
 780                 $openxmlRegex = "/^\[Content_Types\].xml/";
 781
 782                 if ( preg_match( $opendocRegex, substr( $header, 30 ), $matches ) ) {
 783                         $mime = $matches[1];
 784                         wfDebug( __METHOD__.": detected $mime from ZIP archive\n" );
 785                 } elseif ( preg_match( $openxmlRegex, substr( $header, 30 ) ) ) {
 786                         $mime = "application/x-opc+zip";
 787                         # TODO: remove the block below, as soon as improveTypeFromExtension is used everywhere
 788                         if ( $ext !== true && $ext !== false ) {
 789                                 /** This is the mode used by getPropsFromPath
 790                                 * These mime's are stored in the database, where we don't really want
 791                                 * x-opc+zip, because we use it only for internal purposes
 792                                 */
 793                                 if ( $this->isMatchingExtension( $ext, $mime) ) {
 794                                         /* A known file extension for an OPC file,
 795                                         * find the proper mime type for that file extension */
 796                                         $mime = $this->guessTypesForExtension( $ext );
 797                                 } else {
 798                                         $mime = "application/zip";
 799                                 }
 800                         }
 801                         wfDebug( __METHOD__.": detected an Open Packaging Conventions archive: $mime\n" );
 802                 } elseif ( substr( $header, 0, 8 ) == "\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" &&
 803                                 ($headerpos = strpos( $tail, "PK\x03\x04" ) ) !== false &&
 804                                 preg_match( $openxmlRegex, substr( $tail, $headerpos + 30 ) ) ) {
 805                         if ( substr( $header, 512, 4) == "\xEC\xA5\xC1\x00" ) {
 806                                 $mime = "application/msword";
 807                         }
 808                         switch( substr( $header, 512, 6) ) {
 809                                 case "\xEC\xA5\xC1\x00\x0E\x00":
 810                                 case "\xEC\xA5\xC1\x00\x1C\x00":
 811                                 case "\xEC\xA5\xC1\x00\x43\x00":
 812                                         $mime = "application/vnd.ms-powerpoint";
 813                                         break;
 814                                 case "\xFD\xFF\xFF\xFF\x10\x00":
 815                                 case "\xFD\xFF\xFF\xFF\x1F\x00":
 816                                 case "\xFD\xFF\xFF\xFF\x22\x00":
 817                                 case "\xFD\xFF\xFF\xFF\x23\x00":
 818                                 case "\xFD\xFF\xFF\xFF\x28\x00":
 819                                 case "\xFD\xFF\xFF\xFF\x29\x00":
 820                                 case "\xFD\xFF\xFF\xFF\x10\x02":
 821                                 case "\xFD\xFF\xFF\xFF\x1F\x02":
 822                                 case "\xFD\xFF\xFF\xFF\x22\x02":
 823                                 case "\xFD\xFF\xFF\xFF\x23\x02":
 824                                 case "\xFD\xFF\xFF\xFF\x28\x02":
 825                                 case "\xFD\xFF\xFF\xFF\x29\x02":
 826                                         $mime = "application/vnd.msexcel";
 827                                         break;
 828                         }
 829
 830                         wfDebug( __METHOD__.": detected a MS Office document with OPC trailer\n");
 831                 } else {
 832                         wfDebug( __METHOD__.": unable to identify type of ZIP archive\n" );
 833                 }
 834                 return $mime;
 835         }
 836
 837         /**
 838          * Internal mime type detection. Detection is done using an external
 839          * program, if $wgMimeDetectorCommand is set. Otherwise, the fileinfo
 840          * extension and mime_content_type are tried (in this order), if they
 841          * are available. If the dections fails and $ext is not false, the mime
 842          * type is guessed from the file extension, using guessTypesForExtension.
 843          *
 844          * If the mime type is still unknown, getimagesize is used to detect the
 845          * mime type if the file is an image. If no mime type can be determined,
 846          * this function returns 'unknown/unknown'.
 847          *
 848          * @param $file String: the file to check
 849          * @param $ext Mixed: the file extension, or true (default) to extract it from the filename.
 850          *             Set it to false to ignore the extension. DEPRECATED! Set to false, use
 851          *             improveTypeFromExtension($mime, $ext) later to improve mime type.
 852          *
 853          * @return string the mime type of $file
 854          */
 855         private function detectMimeType( $file, $ext = true ) {
 856                 global $wgMimeDetectorCommand;
 857
 858                 if ( $ext ) { # TODO:  make $ext default to false. Or better, remove it.
 859                         wfDebug( __METHOD__.": WARNING: use of the \$ext parameter is deprecated. Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 860                 }
 861
 862                 $m = null;
 863                 if ( $wgMimeDetectorCommand ) {
 864                         // @todo FIXME: Use wfShellExec
 865                         $fn = wfEscapeShellArg( $file );
 866                         $m = `$wgMimeDetectorCommand $fn`;
 867                 } elseif ( function_exists( "finfo_open" ) && function_exists( "finfo_file" ) ) {
 868
 869                         # This required the fileinfo extension by PECL,
 870                         # see http://pecl.php.net/package/fileinfo
 871                         # This must be compiled into PHP
 872                         #
 873                         # finfo is the official replacement for the deprecated
 874                         # mime_content_type function, see below.
 875                         #
 876                         # If you may need to load the fileinfo extension at runtime, set
 877                         # $wgLoadFileinfoExtension in LocalSettings.php
 878
 879                         $mime_magic_resource = finfo_open( FILEINFO_MIME ); /* return mime type ala mimetype extension */
 880
 881                         if ( $mime_magic_resource ) {
 882                                 $m = finfo_file( $mime_magic_resource, $file );
 883                                 finfo_close( $mime_magic_resource );
 884                         } else {
 885                                 wfDebug( __METHOD__.": finfo_open failed on ".FILEINFO_MIME."!\n" );
 886                         }
 887                 } elseif ( function_exists( "mime_content_type" ) ) {
 888
 889                         # NOTE: this function is available since PHP 4.3.0, but only if
 890                         # PHP was compiled with --with-mime-magic or, before 4.3.2, with --enable-mime-magic.
 891                         #
 892                         # On Windows, you must set mime_magic.magicfile in php.ini to point to the mime.magic file bundeled with PHP;
 893                         # sometimes, this may even be needed under linus/unix.
 894                         #
 895                         # Also note that this has been DEPRECATED in favor of the fileinfo extension by PECL, see above.
 896                         # see http://www.php.net/manual/en/ref.mime-magic.php for details.
 897
 898                         $m = mime_content_type($file);
 899                 } else {
 900                         wfDebug( __METHOD__.": no magic mime detector found!\n" );
 901                 }
 902
 903                 if ( $m ) {
 904                         # normalize
 905                         $m = preg_replace( '![;, ].*$!', '', $m ); #strip charset, etc
 906                         $m = trim( $m );
 907                         $m = strtolower( $m );
 908
 909                         if ( strpos( $m, 'unknown' ) !== false ) {
 910                                 $m = null;
 911                         } else {
 912                                 wfDebug( __METHOD__.": magic mime type of $file: $m\n" );
 913                                 return $m;
 914                         }
 915                 }
 916
 917                 // If desired, look at extension as a fallback.
 918                 if ( $ext === true ) {
 919                         $i = strrpos( $file, '.' );
 920                         $ext = strtolower( $i ? substr( $file, $i + 1 ) : '' );
 921                 }
 922                 if ( $ext ) {
 923                         if( $this->isRecognizableExtension( $ext ) ) {
 924                                 wfDebug( __METHOD__. ": refusing to guess mime type for .$ext file, we should have recognized it\n" );
 925                         } else {
 926                                 $m = $this->guessTypesForExtension( $ext );
 927                                 if ( $m ) {
 928                                         wfDebug( __METHOD__.": extension mime type of $file: $m\n" );
 929                                         return $m;
 930                                 }
 931                         }
 932                 }
 933
 934                 // Unknown type
 935                 wfDebug( __METHOD__ . ": failed to guess mime type for $file!\n" );
 936                 return 'unknown/unknown';
 937         }
 938
 939         /**
 940          * Determine the media type code for a file, using its mime type, name and
 941          * possibly its contents.
 942          *
 943          * This function relies on the findMediaType(), mapping extensions and mime
 944          * types to media types.
 945          *
 946          * @todo analyse file if need be
 947          * @todo look at multiple extension, separately and together.
 948          *
 949          * @param $path String: full path to the image file, in case we have to look at the contents
 950          *        (if null, only the mime type is used to determine the media type code).
 951          * @param $mime String: mime type. If null it will be guessed using guessMimeType.
 952          *
 953          * @return (int?string?) a value to be used with the MEDIATYPE_xxx constants.
 954          */
 955         function getMediaType( $path = null, $mime = null ) {
 956                 if( !$mime && !$path ) {
 957                         return MEDIATYPE_UNKNOWN;
 958                 }
 959
 960                 // If mime type is unknown, guess it
 961                 if( !$mime ) {
 962                         $mime = $this->guessMimeType( $path, false );
 963                 }
 964
 965                 // Special code for ogg - detect if it's video (theora),
 966                 // else label it as sound.
 967                 if ( $mime == 'application/ogg' && file_exists( $path ) ) {
 968
 969                         // Read a chunk of the file
 970                         $f = fopen( $path, "rt" );
 971                         if ( !$f ) return MEDIATYPE_UNKNOWN;
 972                         $head = fread( $f, 256 );
 973                         fclose( $f );
 974
 975                         $head = strtolower( $head );
 976
 977                         // This is an UGLY HACK, file should be parsed correctly
 978                         if ( strpos( $head, 'theora' ) !== false ) return MEDIATYPE_VIDEO;
 979                         elseif ( strpos( $head, 'vorbis' ) !== false ) return MEDIATYPE_AUDIO;
 980                         elseif ( strpos( $head, 'flac' ) !== false ) return MEDIATYPE_AUDIO;
 981                         elseif ( strpos( $head, 'speex' ) !== false ) return MEDIATYPE_AUDIO;
 982                         else return MEDIATYPE_MULTIMEDIA;
 983                 }
 984
 985                 // Check for entry for full mime type
 986                 if( $mime ) {
 987                         $type = $this->findMediaType( $mime );
 988                         if ( $type !== MEDIATYPE_UNKNOWN ) {
 989                                 return $type;
 990                         }
 991                 }
 992
 993                 // Check for entry for file extension
 994                 if ( $path ) {
 995                         $i = strrpos( $path, '.' );
 996                         $e = strtolower( $i ? substr( $path, $i + 1 ) : '' );
 997
 998                         // TODO: look at multi-extension if this fails, parse from full path
 999                         $type = $this->findMediaType( '.' . $e );
1000                         if ( $type !== MEDIATYPE_UNKNOWN ) {
1001                                 return $type;
1002                         }
1003                 }
1004
1005                 // Check major mime type
1006                 if ( $mime ) {
1007                         $i = strpos( $mime, '/' );
1008                         if ( $i !== false ) {
1009                                 $major = substr( $mime, 0, $i );
1010                                 $type = $this->findMediaType( $major );
1011                                 if ( $type !== MEDIATYPE_UNKNOWN ) {
1012                                         return $type;
1013                                 }
1014                         }
1015                 }
1016
1017                 if( !$type ) {
1018                         $type = MEDIATYPE_UNKNOWN;
1019                 }
1020
1021                 return $type;
1022         }
1023
1024         /**
1025          * Returns a media code matching the given mime type or file extension.
1026          * File extensions are represented by a string starting with a dot (.) to
1027          * distinguish them from mime types.
1028          *
1029          * This funktion relies on the mapping defined by $this->mMediaTypes
1030          * @access private
1031          */
1032         function findMediaType( $extMime ) {
1033                 if ( strpos( $extMime, '.' ) === 0 ) {
1034                         // If it's an extension, look up the mime types
1035                         $m = $this->getTypesForExtension( substr( $extMime, 1 ) );
1036                         if ( !$m ) {
1037                                 return MEDIATYPE_UNKNOWN;
1038                         }
1039
1040                         $m = explode( ' ', $m );
1041                 } else {
1042                         // Normalize mime type
1043                         if ( isset( $this->mMimeTypeAliases[$extMime] ) ) {
1044                                 $extMime = $this->mMimeTypeAliases[$extMime];
1045                         }
1046
1047                         $m = array( $extMime );
1048                 }
1049
1050                 foreach ( $m as $mime ) {
1051                         foreach ( $this->mMediaTypes as $type => $codes ) {
1052                                 if ( in_array($mime, $codes, true ) ) {
1053                                         return $type;
1054                                 }
1055                         }
1056                 }
1057
1058                 return MEDIATYPE_UNKNOWN;
1059         }
1060
1061         /**
1062          * Get the MIME types that various versions of Internet Explorer would
1063          * detect from a chunk of the content.
1064          *
1065          * @param $fileName String: the file name (unused at present)
1066          * @param $chunk String: the first 256 bytes of the file
1067          * @param $proposed String: the MIME type proposed by the server
1068          */
1069         public function getIEMimeTypes( $fileName, $chunk, $proposed ) {
1070                 $ca = $this->getIEContentAnalyzer();
1071                 return $ca->getRealMimesFromData( $fileName, $chunk, $proposed );
1072         }
1073
1074         /**
1075          * Get a cached instance of IEContentAnalyzer
1076          *
1077          * @return IEContentAnalyzer
1078          */
1079         protected function getIEContentAnalyzer() {
1080                 if ( is_null( $this->mIEAnalyzer ) ) {
1081                         $this->mIEAnalyzer = new IEContentAnalyzer;
1082                 }
1083                 return $this->mIEAnalyzer;
1084         }
1085 }