includes/MimeMagic.php

   1 <?php
   2 /**
   3  * Module defining helper functions for detecting and dealing with mime types.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  */
  22
  23 /**
  24  * Defines a set of well known mime types
  25  * This is used as a fallback to mime.types files.
  26  * An extensive list of well known mime types is provided by
  27  * the file mime.types in the includes directory.
  28  *
  29  * This list concatenated with mime.types is used to create a mime <-> ext
  30  * map. Each line contains a mime type followed by a space separated list of
  31  * extensions. If multiple extensions for a single mime type exist or if
  32  * multiple mime types exist for a single extension then in most cases
  33  * MediaWiki assumes that the first extension following the mime type is the
  34  * canonical extension, and the first time a mime type appears for a certain
  35  * extension is considered the canonical mime type.
  36  *
  37  * (Note that appending $wgMimeTypeFile to the end of MM_WELL_KNOWN_MIME_TYPES
  38  * sucks because you can't redefine canonical types. This could be fixed by
  39  * appending MM_WELL_KNOWN_MIME_TYPES behind $wgMimeTypeFile, but who knows
  40  * what will break? In practice this probably isn't a problem anyway -- Bryan)
  41  */
  42 define('MM_WELL_KNOWN_MIME_TYPES',<<<END_STRING
  43 application/ogg ogx ogg ogm ogv oga spx
  44 application/pdf pdf
  45 application/vnd.oasis.opendocument.chart odc
  46 application/vnd.oasis.opendocument.chart-template otc
  47 application/vnd.oasis.opendocument.database odb
  48 application/vnd.oasis.opendocument.formula odf
  49 application/vnd.oasis.opendocument.formula-template otf
  50 application/vnd.oasis.opendocument.graphics odg
  51 application/vnd.oasis.opendocument.graphics-template otg
  52 application/vnd.oasis.opendocument.image odi
  53 application/vnd.oasis.opendocument.image-template oti
  54 application/vnd.oasis.opendocument.presentation odp
  55 application/vnd.oasis.opendocument.presentation-template otp
  56 application/vnd.oasis.opendocument.spreadsheet ods
  57 application/vnd.oasis.opendocument.spreadsheet-template ots
  58 application/vnd.oasis.opendocument.text odt
  59 application/vnd.oasis.opendocument.text-master otm
  60 application/vnd.oasis.opendocument.text-template ott
  61 application/vnd.oasis.opendocument.text-web oth
  62 application/x-javascript js
  63 application/x-shockwave-flash swf
  64 audio/midi mid midi kar
  65 audio/mpeg mpga mpa mp2 mp3
  66 audio/x-aiff aif aiff aifc
  67 audio/x-wav wav
  68 audio/ogg oga spx ogg
  69 image/x-bmp bmp
  70 image/gif gif
  71 image/jpeg jpeg jpg jpe
  72 image/png png
  73 image/svg+xml svg
  74 image/svg svg
  75 image/tiff tiff tif
  76 image/vnd.djvu djvu
  77 image/x.djvu djvu
  78 image/x-djvu djvu
  79 image/x-portable-pixmap ppm
  80 image/x-xcf xcf
  81 text/plain txt
  82 text/html html htm
  83 video/ogg ogv ogm ogg
  84 video/mpeg mpg mpeg
  85 END_STRING
  86 );
  87
  88 /**
  89  * Defines a set of well known mime info entries
  90  * This is used as a fallback to mime.info files.
  91  * An extensive list of well known mime types is provided by
  92  * the file mime.info in the includes directory.
  93  */
  94 define('MM_WELL_KNOWN_MIME_INFO', <<<END_STRING
  95 application/pdf [OFFICE]
  96 application/vnd.oasis.opendocument.chart [OFFICE]
  97 application/vnd.oasis.opendocument.chart-template [OFFICE]
  98 application/vnd.oasis.opendocument.database [OFFICE]
  99 application/vnd.oasis.opendocument.formula [OFFICE]
 100 application/vnd.oasis.opendocument.formula-template [OFFICE]
 101 application/vnd.oasis.opendocument.graphics [OFFICE]
 102 application/vnd.oasis.opendocument.graphics-template [OFFICE]
 103 application/vnd.oasis.opendocument.image [OFFICE]
 104 application/vnd.oasis.opendocument.image-template [OFFICE]
 105 application/vnd.oasis.opendocument.presentation [OFFICE]
 106 application/vnd.oasis.opendocument.presentation-template [OFFICE]
 107 application/vnd.oasis.opendocument.spreadsheet [OFFICE]
 108 application/vnd.oasis.opendocument.spreadsheet-template [OFFICE]
 109 application/vnd.oasis.opendocument.text [OFFICE]
 110 application/vnd.oasis.opendocument.text-template [OFFICE]
 111 application/vnd.oasis.opendocument.text-master [OFFICE]
 112 application/vnd.oasis.opendocument.text-web [OFFICE]
 113 text/javascript application/x-javascript [EXECUTABLE]
 114 application/x-shockwave-flash [MULTIMEDIA]
 115 audio/midi [AUDIO]
 116 audio/x-aiff [AUDIO]
 117 audio/x-wav [AUDIO]
 118 audio/mp3 audio/mpeg [AUDIO]
 119 application/ogg audio/ogg video/ogg [MULTIMEDIA]
 120 image/x-bmp image/x-ms-bmp image/bmp [BITMAP]
 121 image/gif [BITMAP]
 122 image/jpeg [BITMAP]
 123 image/png [BITMAP]
 124 image/svg+xml [DRAWING]
 125 image/tiff [BITMAP]
 126 image/vnd.djvu [BITMAP]
 127 image/x-xcf [BITMAP]
 128 image/x-portable-pixmap [BITMAP]
 129 text/plain [TEXT]
 130 text/html [TEXT]
 131 video/ogg [VIDEO]
 132 video/mpeg [VIDEO]
 133 unknown/unknown application/octet-stream application/x-empty [UNKNOWN]
 134 END_STRING
 135 );
 136
 137 /**
 138  * Implements functions related to mime types such as detection and mapping to
 139  * file extension.
 140  *
 141  * Instances of this class are stateless, there only needs to be one global instance
 142  * of MimeMagic. Please use MimeMagic::singleton() to get that instance.
 143  */
 144 class MimeMagic {
 145
 146         /**
 147         * Mapping of media types to arrays of mime types.
 148         * This is used by findMediaType and getMediaType, respectively
 149         */
 150         var $mMediaTypes = null;
 151
 152         /** Map of mime type aliases
 153         */
 154         var $mMimeTypeAliases = null;
 155
 156         /** map of mime types to file extensions (as a space seprarated list)
 157         */
 158         var $mMimeToExt = null;
 159
 160         /** map of file extensions types to mime types (as a space seprarated list)
 161         */
 162         var $mExtToMime = null;
 163
 164         /** IEContentAnalyzer instance
 165          */
 166         var $mIEAnalyzer;
 167
 168         /** The singleton instance
 169          */
 170         private static $instance;
 171
 172         /** True if the fileinfo extension has been loaded
 173          */
 174         private static $extensionLoaded = false;
 175
 176         /** Initializes the MimeMagic object. This is called by MimeMagic::singleton().
 177          *
 178          * This constructor parses the mime.types and mime.info files and build internal mappings.
 179          */
 180         function __construct() {
 181                 /**
 182                 *   --- load mime.types ---
 183                 */
 184
 185                 global $wgMimeTypeFile, $IP, $wgLoadFileinfoExtension;
 186
 187                 $types = MM_WELL_KNOWN_MIME_TYPES;
 188
 189                 if ( $wgMimeTypeFile == 'includes/mime.types' ) {
 190                         $wgMimeTypeFile = "$IP/$wgMimeTypeFile";
 191                 }
 192
 193                 if ( $wgLoadFileinfoExtension && !self::$extensionLoaded ) {
 194                         self::$extensionLoaded = true;
 195                         wfDl( 'fileinfo' );
 196                 }
 197
 198                 if ( $wgMimeTypeFile ) {
 199                         if ( is_file( $wgMimeTypeFile ) and is_readable( $wgMimeTypeFile ) ) {
 200                                 wfDebug( __METHOD__.": loading mime types from $wgMimeTypeFile\n" );
 201                                 $types .= "\n";
 202                                 $types .= file_get_contents( $wgMimeTypeFile );
 203                         } else {
 204                                 wfDebug( __METHOD__.": can't load mime types from $wgMimeTypeFile\n" );
 205                         }
 206                 } else {
 207                         wfDebug( __METHOD__.": no mime types file defined, using build-ins only.\n" );
 208                 }
 209
 210                 $types = str_replace( array( "\r\n", "\n\r", "\n\n", "\r\r", "\r" ), "\n", $types );
 211                 $types = str_replace( "\t", " ", $types );
 212
 213                 $this->mMimeToExt = array();
 214                 $this->mToMime = array();
 215
 216                 $lines = explode( "\n",$types );
 217                 foreach ( $lines as $s ) {
 218                         $s = trim( $s );
 219                         if ( empty( $s ) ) {
 220                                 continue;
 221                         }
 222                         if ( strpos( $s, '#' ) === 0 ) {
 223                                 continue;
 224                         }
 225
 226                         $s = strtolower( $s );
 227                         $i = strpos( $s, ' ' );
 228
 229                         if ( $i === false ) {
 230                                 continue;
 231                         }
 232
 233                         $mime = substr( $s, 0, $i );
 234                         $ext = trim( substr($s, $i+1 ) );
 235
 236                         if ( empty( $ext ) ) {
 237                                 continue;
 238                         }
 239
 240                         if ( !empty( $this->mMimeToExt[$mime] ) ) {
 241                                 $this->mMimeToExt[$mime] .= ' ' . $ext;
 242                         } else {
 243                                 $this->mMimeToExt[$mime] = $ext;
 244                         }
 245
 246                         $extensions = explode( ' ', $ext );
 247
 248                         foreach ( $extensions as $e ) {
 249                                 $e = trim( $e );
 250                                 if ( empty( $e ) ) {
 251                                         continue;
 252                                 }
 253
 254                                 if ( !empty( $this->mExtToMime[$e] ) ) {
 255                                         $this->mExtToMime[$e] .= ' ' . $mime;
 256                                 } else {
 257                                         $this->mExtToMime[$e] = $mime;
 258                                 }
 259                         }
 260                 }
 261
 262                 /**
 263                  *   --- load mime.info ---
 264                  */
 265
 266                 global $wgMimeInfoFile;
 267                 if ( $wgMimeInfoFile == 'includes/mime.info' ) {
 268                         $wgMimeInfoFile = "$IP/$wgMimeInfoFile";
 269                 }
 270
 271                 $info = MM_WELL_KNOWN_MIME_INFO;
 272
 273                 if ( $wgMimeInfoFile ) {
 274                         if ( is_file( $wgMimeInfoFile ) and is_readable( $wgMimeInfoFile ) ) {
 275                                 wfDebug( __METHOD__.": loading mime info from $wgMimeInfoFile\n" );
 276                                 $info .= "\n";
 277                                 $info .= file_get_contents( $wgMimeInfoFile );
 278                         } else {
 279                                 wfDebug(__METHOD__.": can't load mime info from $wgMimeInfoFile\n");
 280                         }
 281                 } else {
 282                         wfDebug(__METHOD__.": no mime info file defined, using build-ins only.\n");
 283                 }
 284
 285                 $info = str_replace( array( "\r\n", "\n\r", "\n\n", "\r\r", "\r" ), "\n", $info);
 286                 $info = str_replace( "\t", " ", $info );
 287
 288                 $this->mMimeTypeAliases = array();
 289                 $this->mMediaTypes = array();
 290
 291                 $lines = explode( "\n", $info );
 292                 foreach ( $lines as $s ) {
 293                         $s = trim( $s );
 294                         if ( empty( $s ) ) {
 295                                 continue;
 296                         }
 297                         if ( strpos( $s, '#' ) === 0 ) {
 298                                 continue;
 299                         }
 300
 301                         $s = strtolower( $s );
 302                         $i = strpos( $s, ' ' );
 303
 304                         if ( $i === false ) {
 305                                 continue;
 306                         }
 307
 308                         #print "processing MIME INFO line $s<br>";
 309
 310                         $match = array();
 311                         if ( preg_match( '!\[\s*(\w+)\s*\]!', $s, $match ) ) {
 312                                 $s = preg_replace( '!\[\s*(\w+)\s*\]!', '', $s );
 313                                 $mtype = trim( strtoupper( $match[1] ) );
 314                         } else {
 315                                 $mtype = MEDIATYPE_UNKNOWN;
 316                         }
 317
 318                         $m = explode( ' ', $s );
 319
 320                         if ( !isset( $this->mMediaTypes[$mtype] ) ) {
 321                                 $this->mMediaTypes[$mtype] = array();
 322                         }
 323
 324                         foreach ( $m as $mime ) {
 325                                 $mime = trim( $mime );
 326                                 if ( empty( $mime ) ) {
 327                                         continue;
 328                                 }
 329
 330                                 $this->mMediaTypes[$mtype][] = $mime;
 331                         }
 332
 333                         if ( sizeof( $m ) > 1 ) {
 334                                 $main = $m[0];
 335                                 for ( $i=1; $i<sizeof($m); $i += 1 ) {
 336                                         $mime = $m[$i];
 337                                         $this->mMimeTypeAliases[$mime] = $main;
 338                                 }
 339                         }
 340                 }
 341
 342         }
 343
 344         /**
 345          * Get an instance of this class
 346          * @return MimeMagic
 347          */
 348         public static function &singleton() {
 349                 if ( !isset( self::$instance ) ) {
 350                         self::$instance = new MimeMagic;
 351                 }
 352                 return self::$instance;
 353         }
 354
 355         /**
 356          * Returns a list of file extensions for a given mime type as a space
 357          * separated string or null if the mime type was unrecognized. Resolves
 358          * mime type aliases.
 359          *
 360          * @param $mime string
 361          * @return string|null
 362          */
 363         public function getExtensionsForType( $mime ) {
 364                 $mime = strtolower( $mime );
 365
 366                 // Check the mime-to-ext map
 367                 if ( isset( $this->mMimeToExt[$mime] ) ) {
 368                         return $this->mMimeToExt[$mime];
 369                 }
 370
 371                 // Resolve the mime type to the canonical type
 372                 if ( isset( $this->mMimeTypeAliases[$mime] ) ) {
 373                         $mime = $this->mMimeTypeAliases[$mime];
 374                         if ( isset( $this->mMimeToExt[$mime] ) ) {
 375                                 return $this->mMimeToExt[$mime];
 376                         }
 377                 }
 378
 379                 return null;
 380         }
 381
 382         /**
 383          * Returns a list of mime types for a given file extension as a space
 384          * separated string or null if the extension was unrecognized.
 385          *
 386          * @param $ext string
 387          * @return string|null
 388          */
 389         public function getTypesForExtension( $ext ) {
 390                 $ext = strtolower( $ext );
 391
 392                 $r = isset( $this->mExtToMime[$ext] ) ? $this->mExtToMime[$ext] : null;
 393                 return $r;
 394         }
 395
 396         /**
 397          * Returns a single mime type for a given file extension or null if unknown.
 398          * This is always the first type from the list returned by getTypesForExtension($ext).
 399          *
 400          * @param $ext string
 401          * @return string|null
 402          */
 403         public function guessTypesForExtension( $ext ) {
 404                 $m = $this->getTypesForExtension( $ext );
 405                 if ( is_null( $m ) ) {
 406                         return null;
 407                 }
 408
 409                 // TODO: Check if this is needed; strtok( $m, ' ' ) should be sufficient
 410                 $m = trim( $m );
 411                 $m = preg_replace( '/\s.*$/', '', $m );
 412
 413                 return $m;
 414         }
 415
 416
 417         /**
 418          * Tests if the extension matches the given mime type. Returns true if a
 419          * match was found, null if the mime type is unknown, and false if the
 420          * mime type is known but no matches where found.
 421          *
 422          * @param $extension string
 423          * @param $mime string
 424          * @return bool|null
 425          */
 426         public function isMatchingExtension( $extension, $mime ) {
 427                 $ext = $this->getExtensionsForType( $mime );
 428
 429                 if ( !$ext ) {
 430                         return null;  // Unknown mime type
 431                 }
 432
 433                 $ext = explode( ' ', $ext );
 434
 435                 $extension = strtolower( $extension );
 436                 return  in_array( $extension, $ext );
 437         }
 438
 439         /**
 440          * Returns true if the mime type is known to represent an image format
 441          * supported by the PHP GD library.
 442          *
 443          * @param $mime string
 444          *
 445          * @return bool
 446          */
 447         public function isPHPImageType( $mime ) {
 448                 // As defined by imagegetsize and image_type_to_mime
 449                 static $types = array(
 450                         'image/gif', 'image/jpeg', 'image/png',
 451                         'image/x-bmp', 'image/xbm', 'image/tiff',
 452                         'image/jp2', 'image/jpeg2000', 'image/iff',
 453                         'image/xbm', 'image/x-xbitmap',
 454                         'image/vnd.wap.wbmp', 'image/vnd.xiff',
 455                         'image/x-photoshop',
 456                         'application/x-shockwave-flash',
 457                 );
 458
 459                 return in_array( $mime, $types );
 460         }
 461
 462         /**
 463          * Returns true if the extension represents a type which can
 464          * be reliably detected from its content. Use this to determine
 465          * whether strict content checks should be applied to reject
 466          * invalid uploads; if we can't identify the type we won't
 467          * be able to say if it's invalid.
 468          *
 469          * @todo Be more accurate when using fancy mime detector plugins;
 470          *       right now this is the bare minimum getimagesize() list.
 471          * @return bool
 472          */
 473         function isRecognizableExtension( $extension ) {
 474                 static $types = array(
 475                         // Types recognized by getimagesize()
 476                         'gif', 'jpeg', 'jpg', 'png', 'swf', 'psd',
 477                         'bmp', 'tiff', 'tif', 'jpc', 'jp2',
 478                         'jpx', 'jb2', 'swc', 'iff', 'wbmp',
 479                         'xbm',
 480
 481                         // Formats we recognize magic numbers for
 482                         'djvu', 'ogx', 'ogg', 'ogv', 'oga', 'spx',
 483                         'mid', 'pdf', 'wmf', 'xcf', 'webm', 'mkv', 'mka',
 484                         'webp',
 485
 486                         // XML formats we sure hope we recognize reliably
 487                         'svg',
 488                 );
 489                 return in_array( strtolower( $extension ), $types );
 490         }
 491
 492         /**
 493          * Improves a mime type using the file extension. Some file formats are very generic,
 494          * so their mime type is not very meaningful. A more useful mime type can be derived
 495          * by looking at the file extension. Typically, this method would be called on the
 496          * result of guessMimeType().
 497          *
 498          * Currently, this method does the following:
 499          *
 500          * If $mime is "unknown/unknown" and isRecognizableExtension( $ext ) returns false,
 501          * return the result of guessTypesForExtension($ext).
 502          *
 503          * If $mime is "application/x-opc+zip" and isMatchingExtension( $ext, $mime )
 504          * gives true, return the result of guessTypesForExtension($ext).
 505          *
 506          * @param $mime String: the mime type, typically guessed from a file's content.
 507          * @param $ext String: the file extension, as taken from the file name
 508          *
 509          * @return string the mime type
 510          */
 511         public function improveTypeFromExtension( $mime, $ext ) {
 512                 if ( $mime === 'unknown/unknown' ) {
 513                         if ( $this->isRecognizableExtension( $ext ) ) {
 514                                 wfDebug( __METHOD__. ': refusing to guess mime type for .' .
 515                                         "$ext file, we should have recognized it\n" );
 516                         } else {
 517                                 // Not something we can detect, so simply
 518                                 // trust the file extension
 519                                 $mime = $this->guessTypesForExtension( $ext );
 520                         }
 521                 }
 522                 elseif ( $mime === 'application/x-opc+zip' ) {
 523                         if ( $this->isMatchingExtension( $ext, $mime ) ) {
 524                                 // A known file extension for an OPC file,
 525                                 // find the proper mime type for that file extension
 526                                 $mime = $this->guessTypesForExtension( $ext );
 527                         } else {
 528                                 wfDebug( __METHOD__. ": refusing to guess better type for $mime file, " .
 529                                         ".$ext is not a known OPC extension.\n" );
 530                                 $mime = 'application/zip';
 531                         }
 532                 }
 533
 534                 if ( isset( $this->mMimeTypeAliases[$mime] ) ) {
 535                         $mime = $this->mMimeTypeAliases[$mime];
 536                 }
 537
 538                 wfDebug(__METHOD__.": improved mime type for .$ext: $mime\n");
 539                 return $mime;
 540         }
 541
 542         /**
 543          * Mime type detection. This uses detectMimeType to detect the mime type
 544          * of the file, but applies additional checks to determine some well known
 545          * file formats that may be missed or misinterpreter by the default mime
 546          * detection (namely XML based formats like XHTML or SVG, as well as ZIP
 547          * based formats like OPC/ODF files).
 548          *
 549          * @param $file String: the file to check
 550          * @param $ext Mixed: the file extension, or true (default) to extract it from the filename.
 551          *             Set it to false to ignore the extension. DEPRECATED! Set to false, use
 552          *             improveTypeFromExtension($mime, $ext) later to improve mime type.
 553          *
 554          * @return string the mime type of $file
 555          */
 556         public function guessMimeType( $file, $ext = true ) {
 557                 if ( $ext ) { // TODO: make $ext default to false. Or better, remove it.
 558                         wfDebug( __METHOD__.": WARNING: use of the \$ext parameter is deprecated. " .
 559                                 "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 560                 }
 561
 562                 $mime = $this->doGuessMimeType( $file, $ext );
 563
 564                 if( !$mime ) {
 565                         wfDebug( __METHOD__.": internal type detection failed for $file (.$ext)...\n" );
 566                         $mime = $this->detectMimeType( $file, $ext );
 567                 }
 568
 569                 if ( isset( $this->mMimeTypeAliases[$mime] ) ) {
 570                         $mime = $this->mMimeTypeAliases[$mime];
 571                 }
 572
 573                 wfDebug(__METHOD__.": guessed mime type of $file: $mime\n");
 574                 return $mime;
 575         }
 576
 577         /**
 578          * Guess the mime type from the file contents.
 579          *
 580          * @param string $file
 581          * @param mixed $ext
 582          * @return bool|string
 583          */
 584         private function doGuessMimeType( $file, $ext ) { // TODO: remove $ext param
 585                 // Read a chunk of the file
 586                 wfSuppressWarnings();
 587                 // @todo FIXME: Shouldn't this be rb?
 588                 $f = fopen( $file, 'rt' );
 589                 wfRestoreWarnings();
 590
 591                 if( !$f ) {
 592                         return 'unknown/unknown';
 593                 }
 594                 $head = fread( $f, 1024 );
 595                 fseek( $f, -65558, SEEK_END );
 596                 $tail = fread( $f, 65558 ); // 65558 = maximum size of a zip EOCDR
 597                 fclose( $f );
 598
 599                 wfDebug( __METHOD__ . ": analyzing head and tail of $file for magic numbers.\n" );
 600
 601                 // Hardcode a few magic number checks...
 602                 $headers = array(
 603                         // Multimedia...
 604                         'MThd'             => 'audio/midi',
 605                         'OggS'             => 'application/ogg',
 606
 607                         // Image formats...
 608                         // Note that WMF may have a bare header, no magic number.
 609                         "\x01\x00\x09\x00" => 'application/x-msmetafile', // Possibly prone to false positives?
 610                         "\xd7\xcd\xc6\x9a" => 'application/x-msmetafile',
 611                         '%PDF'             => 'application/pdf',
 612                         'gimp xcf'         => 'image/x-xcf',
 613
 614                         // Some forbidden fruit...
 615                         'MZ'               => 'application/octet-stream', // DOS/Windows executable
 616                         "\xca\xfe\xba\xbe" => 'application/octet-stream', // Mach-O binary
 617                         "\x7fELF"          => 'application/octet-stream', // ELF binary
 618                 );
 619
 620                 foreach ( $headers as $magic => $candidate ) {
 621                         if ( strncmp( $head, $magic, strlen( $magic ) ) == 0 ) {
 622                                 wfDebug( __METHOD__ . ": magic header in $file recognized as $candidate\n" );
 623                                 return $candidate;
 624                         }
 625                 }
 626
 627                 /* Look for WebM and Matroska files */
 628                 if ( strncmp( $head, pack( "C4", 0x1a, 0x45, 0xdf, 0xa3 ), 4 ) == 0 ) {
 629                         $doctype = strpos( $head, "\x42\x82" );
 630                         if ( $doctype ) {
 631                                 // Next byte is datasize, then data (sizes larger than 1 byte are very stupid muxers)
 632                                 $data = substr($head, $doctype+3, 8);
 633                                 if ( strncmp( $data, "matroska", 8 ) == 0 ) {
 634                                         wfDebug( __METHOD__ . ": recognized file as video/x-matroska\n" );
 635                                         return "video/x-matroska";
 636                                 } elseif ( strncmp( $data, "webm", 4 ) == 0 ) {
 637                                         wfDebug( __METHOD__ . ": recognized file as video/webm\n" );
 638                                         return "video/webm";
 639                                 }
 640                         }
 641                         wfDebug( __METHOD__ . ": unknown EBML file\n" );
 642                         return "unknown/unknown";
 643                 }
 644
 645                 /* Look for WebP */
 646                 if ( strncmp( $head, "RIFF", 4 ) == 0 && strncmp( substr( $head, 8, 8), "WEBPVP8 ", 8 ) == 0 ) {
 647                         wfDebug( __METHOD__ . ": recognized file as image/webp\n" );
 648                         return "image/webp";
 649                 }
 650
 651                 /**
 652                  * Look for PHP.  Check for this before HTML/XML...  Warning: this is a
 653                  * heuristic, and won't match a file with a lot of non-PHP before.  It
 654                  * will also match text files which could be PHP. :)
 655                  *
 656                  * @todo FIXME: For this reason, the check is probably useless -- an attacker
 657                  * could almost certainly just pad the file with a lot of nonsense to
 658                  * circumvent the check in any case where it would be a security
 659                  * problem.  On the other hand, it causes harmful false positives (bug
 660                  * 16583).  The heuristic has been cut down to exclude three-character
 661                  * strings like "<? ", but should it be axed completely?
 662                  */
 663                 if ( ( strpos( $head, '<?php' ) !== false ) ||
 664
 665                     ( strpos( $head, "<\x00?\x00p\x00h\x00p" ) !== false ) ||
 666                     ( strpos( $head, "<\x00?\x00 " ) !== false ) ||
 667                     ( strpos( $head, "<\x00?\x00\n" ) !== false ) ||
 668                     ( strpos( $head, "<\x00?\x00\t" ) !== false ) ||
 669                     ( strpos( $head, "<\x00?\x00=" ) !== false ) ) {
 670
 671                         wfDebug( __METHOD__ . ": recognized $file as application/x-php\n" );
 672                         return 'application/x-php';
 673                 }
 674
 675                 /**
 676                  * look for XML formats (XHTML and SVG)
 677                  */
 678                 $xml = new XmlTypeCheck( $file );
 679                 if ( $xml->wellFormed ) {
 680                         global $wgXMLMimeTypes;
 681                         if ( isset( $wgXMLMimeTypes[$xml->getRootElement()] ) ) {
 682                                 return $wgXMLMimeTypes[$xml->getRootElement()];
 683                         } else {
 684                                 return 'application/xml';
 685                         }
 686                 }
 687
 688                 /**
 689                  * look for shell scripts
 690                  */
 691                 $script_type = null;
 692
 693                 # detect by shebang
 694                 if ( substr( $head, 0, 2) == "#!" ) {
 695                         $script_type = "ASCII";
 696                 } elseif ( substr( $head, 0, 5) == "\xef\xbb\xbf#!" ) {
 697                         $script_type = "UTF-8";
 698                 } elseif ( substr( $head, 0, 7) == "\xfe\xff\x00#\x00!" ) {
 699                         $script_type = "UTF-16BE";
 700                 } elseif ( substr( $head, 0, 7 ) == "\xff\xfe#\x00!" ) {
 701                         $script_type= "UTF-16LE";
 702                 }
 703
 704                 if ( $script_type ) {
 705                         if ( $script_type !== "UTF-8" && $script_type !== "ASCII") {
 706                                 // Quick and dirty fold down to ASCII!
 707                                 $pack = array( 'UTF-16BE' => 'n*', 'UTF-16LE' => 'v*' );
 708                                 $chars = unpack( $pack[$script_type], substr( $head, 2 ) );
 709                                 $head = '';
 710                                 foreach( $chars as $codepoint ) {
 711                                         if( $codepoint < 128 ) {
 712                                                 $head .= chr( $codepoint );
 713                                         } else {
 714                                                 $head .= '?';
 715                                         }
 716                                 }
 717                         }
 718
 719                         $match = array();
 720
 721                         if ( preg_match( '%/?([^\s]+/)(\w+)%', $head, $match ) ) {
 722                                 $mime = "application/x-{$match[2]}";
 723                                 wfDebug( __METHOD__.": shell script recognized as $mime\n" );
 724                                 return $mime;
 725                         }
 726                 }
 727
 728                 // Check for ZIP variants (before getimagesize)
 729                 if ( strpos( $tail, "PK\x05\x06" ) !== false ) {
 730                         wfDebug( __METHOD__.": ZIP header present in $file\n" );
 731                         return $this->detectZipType( $head, $tail, $ext );
 732                 }
 733
 734                 wfSuppressWarnings();
 735                 $gis = getimagesize( $file );
 736                 wfRestoreWarnings();
 737
 738                 if( $gis && isset( $gis['mime'] ) ) {
 739                         $mime = $gis['mime'];
 740                         wfDebug( __METHOD__.": getimagesize detected $file as $mime\n" );
 741                         return $mime;
 742                 }
 743
 744                 // Also test DjVu
 745                 $deja = new DjVuImage( $file );
 746                 if( $deja->isValid() ) {
 747                         wfDebug( __METHOD__.": detected $file as image/vnd.djvu\n" );
 748                         return 'image/vnd.djvu';
 749                 }
 750
 751                 return false;
 752         }
 753
 754         /**
 755          * Detect application-specific file type of a given ZIP file from its
 756          * header data.  Currently works for OpenDocument and OpenXML types...
 757          * If can't tell, returns 'application/zip'.
 758          *
 759          * @param $header String: some reasonably-sized chunk of file header
 760          * @param $tail   String: the tail of the file
 761          * @param $ext Mixed: the file extension, or true to extract it from the filename.
 762          *             Set it to false (default) to ignore the extension. DEPRECATED! Set to false,
 763          *             use improveTypeFromExtension($mime, $ext) later to improve mime type.
 764          *
 765          * @return string
 766          */
 767         function detectZipType( $header, $tail = null, $ext = false ) {
 768                 if( $ext ) { # TODO: remove $ext param
 769                         wfDebug( __METHOD__.": WARNING: use of the \$ext parameter is deprecated. " .
 770                                 "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 771                 }
 772
 773                 $mime = 'application/zip';
 774                 $opendocTypes = array(
 775                         'chart-template',
 776                         'chart',
 777                         'formula-template',
 778                         'formula',
 779                         'graphics-template',
 780                         'graphics',
 781                         'image-template',
 782                         'image',
 783                         'presentation-template',
 784                         'presentation',
 785                         'spreadsheet-template',
 786                         'spreadsheet',
 787                         'text-template',
 788                         'text-master',
 789                         'text-web',
 790                         'text' );
 791
 792                 // http://lists.oasis-open.org/archives/office/200505/msg00006.html
 793                 $types = '(?:' . implode( '|', $opendocTypes ) . ')';
 794                 $opendocRegex = "/^mimetype(application\/vnd\.oasis\.opendocument\.$types)/";
 795
 796                 $openxmlRegex = "/^\[Content_Types\].xml/";
 797
 798                 if ( preg_match( $opendocRegex, substr( $header, 30 ), $matches ) ) {
 799                         $mime = $matches[1];
 800                         wfDebug( __METHOD__.": detected $mime from ZIP archive\n" );
 801                 } elseif ( preg_match( $openxmlRegex, substr( $header, 30 ) ) ) {
 802                         $mime = "application/x-opc+zip";
 803                         # TODO: remove the block below, as soon as improveTypeFromExtension is used everywhere
 804                         if ( $ext !== true && $ext !== false ) {
 805                                 /** This is the mode used by getPropsFromPath
 806                                 * These mime's are stored in the database, where we don't really want
 807                                 * x-opc+zip, because we use it only for internal purposes
 808                                 */
 809                                 if ( $this->isMatchingExtension( $ext, $mime) ) {
 810                                         /* A known file extension for an OPC file,
 811                                         * find the proper mime type for that file extension */
 812                                         $mime = $this->guessTypesForExtension( $ext );
 813                                 } else {
 814                                         $mime = "application/zip";
 815                                 }
 816                         }
 817                         wfDebug( __METHOD__.": detected an Open Packaging Conventions archive: $mime\n" );
 818                 } elseif ( substr( $header, 0, 8 ) == "\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" &&
 819                                 ($headerpos = strpos( $tail, "PK\x03\x04" ) ) !== false &&
 820                                 preg_match( $openxmlRegex, substr( $tail, $headerpos + 30 ) ) ) {
 821                         if ( substr( $header, 512, 4) == "\xEC\xA5\xC1\x00" ) {
 822                                 $mime = "application/msword";
 823                         }
 824                         switch( substr( $header, 512, 6) ) {
 825                                 case "\xEC\xA5\xC1\x00\x0E\x00":
 826                                 case "\xEC\xA5\xC1\x00\x1C\x00":
 827                                 case "\xEC\xA5\xC1\x00\x43\x00":
 828                                         $mime = "application/vnd.ms-powerpoint";
 829                                         break;
 830                                 case "\xFD\xFF\xFF\xFF\x10\x00":
 831                                 case "\xFD\xFF\xFF\xFF\x1F\x00":
 832                                 case "\xFD\xFF\xFF\xFF\x22\x00":
 833                                 case "\xFD\xFF\xFF\xFF\x23\x00":
 834                                 case "\xFD\xFF\xFF\xFF\x28\x00":
 835                                 case "\xFD\xFF\xFF\xFF\x29\x00":
 836                                 case "\xFD\xFF\xFF\xFF\x10\x02":
 837                                 case "\xFD\xFF\xFF\xFF\x1F\x02":
 838                                 case "\xFD\xFF\xFF\xFF\x22\x02":
 839                                 case "\xFD\xFF\xFF\xFF\x23\x02":
 840                                 case "\xFD\xFF\xFF\xFF\x28\x02":
 841                                 case "\xFD\xFF\xFF\xFF\x29\x02":
 842                                         $mime = "application/vnd.msexcel";
 843                                         break;
 844                         }
 845
 846                         wfDebug( __METHOD__.": detected a MS Office document with OPC trailer\n");
 847                 } else {
 848                         wfDebug( __METHOD__.": unable to identify type of ZIP archive\n" );
 849                 }
 850                 return $mime;
 851         }
 852
 853         /**
 854          * Internal mime type detection. Detection is done using an external
 855          * program, if $wgMimeDetectorCommand is set. Otherwise, the fileinfo
 856          * extension and mime_content_type are tried (in this order), if they
 857          * are available. If the dections fails and $ext is not false, the mime
 858          * type is guessed from the file extension, using guessTypesForExtension.
 859          *
 860          * If the mime type is still unknown, getimagesize is used to detect the
 861          * mime type if the file is an image. If no mime type can be determined,
 862          * this function returns 'unknown/unknown'.
 863          *
 864          * @param $file String: the file to check
 865          * @param $ext Mixed: the file extension, or true (default) to extract it from the filename.
 866          *             Set it to false to ignore the extension. DEPRECATED! Set to false, use
 867          *             improveTypeFromExtension($mime, $ext) later to improve mime type.
 868          *
 869          * @return string the mime type of $file
 870          */
 871         private function detectMimeType( $file, $ext = true ) {
 872                 global $wgMimeDetectorCommand;
 873
 874                 if ( $ext ) { # TODO:  make $ext default to false. Or better, remove it.
 875                         wfDebug( __METHOD__.": WARNING: use of the \$ext parameter is deprecated. Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 876                 }
 877
 878                 $m = null;
 879                 if ( $wgMimeDetectorCommand ) {
 880                         // @todo FIXME: Use wfShellExec
 881                         $fn = wfEscapeShellArg( $file );
 882                         $m = `$wgMimeDetectorCommand $fn`;
 883                 } elseif ( function_exists( "finfo_open" ) && function_exists( "finfo_file" ) ) {
 884
 885                         # This required the fileinfo extension by PECL,
 886                         # see http://pecl.php.net/package/fileinfo
 887                         # This must be compiled into PHP
 888                         #
 889                         # finfo is the official replacement for the deprecated
 890                         # mime_content_type function, see below.
 891                         #
 892                         # If you may need to load the fileinfo extension at runtime, set
 893                         # $wgLoadFileinfoExtension in LocalSettings.php
 894
 895                         $mime_magic_resource = finfo_open( FILEINFO_MIME ); /* return mime type ala mimetype extension */
 896
 897                         if ( $mime_magic_resource ) {
 898                                 $m = finfo_file( $mime_magic_resource, $file );
 899                                 finfo_close( $mime_magic_resource );
 900                         } else {
 901                                 wfDebug( __METHOD__.": finfo_open failed on ".FILEINFO_MIME."!\n" );
 902                         }
 903                 } elseif ( function_exists( "mime_content_type" ) ) {
 904
 905                         # NOTE: this function is available since PHP 4.3.0, but only if
 906                         # PHP was compiled with --with-mime-magic or, before 4.3.2, with --enable-mime-magic.
 907                         #
 908                         # On Windows, you must set mime_magic.magicfile in php.ini to point to the mime.magic file bundeled with PHP;
 909                         # sometimes, this may even be needed under linus/unix.
 910                         #
 911                         # Also note that this has been DEPRECATED in favor of the fileinfo extension by PECL, see above.
 912                         # see http://www.php.net/manual/en/ref.mime-magic.php for details.
 913
 914                         $m = mime_content_type($file);
 915                 } else {
 916                         wfDebug( __METHOD__.": no magic mime detector found!\n" );
 917                 }
 918
 919                 if ( $m ) {
 920                         # normalize
 921                         $m = preg_replace( '![;, ].*$!', '', $m ); #strip charset, etc
 922                         $m = trim( $m );
 923                         $m = strtolower( $m );
 924
 925                         if ( strpos( $m, 'unknown' ) !== false ) {
 926                                 $m = null;
 927                         } else {
 928                                 wfDebug( __METHOD__.": magic mime type of $file: $m\n" );
 929                                 return $m;
 930                         }
 931                 }
 932
 933                 // If desired, look at extension as a fallback.
 934                 if ( $ext === true ) {
 935                         $i = strrpos( $file, '.' );
 936                         $ext = strtolower( $i ? substr( $file, $i + 1 ) : '' );
 937                 }
 938                 if ( $ext ) {
 939                         if( $this->isRecognizableExtension( $ext ) ) {
 940                                 wfDebug( __METHOD__. ": refusing to guess mime type for .$ext file, we should have recognized it\n" );
 941                         } else {
 942                                 $m = $this->guessTypesForExtension( $ext );
 943                                 if ( $m ) {
 944                                         wfDebug( __METHOD__.": extension mime type of $file: $m\n" );
 945                                         return $m;
 946                                 }
 947                         }
 948                 }
 949
 950                 // Unknown type
 951                 wfDebug( __METHOD__ . ": failed to guess mime type for $file!\n" );
 952                 return 'unknown/unknown';
 953         }
 954
 955         /**
 956          * Determine the media type code for a file, using its mime type, name and
 957          * possibly its contents.
 958          *
 959          * This function relies on the findMediaType(), mapping extensions and mime
 960          * types to media types.
 961          *
 962          * @todo analyse file if need be
 963          * @todo look at multiple extension, separately and together.
 964          *
 965          * @param $path String: full path to the image file, in case we have to look at the contents
 966          *        (if null, only the mime type is used to determine the media type code).
 967          * @param $mime String: mime type. If null it will be guessed using guessMimeType.
 968          *
 969          * @return (int?string?) a value to be used with the MEDIATYPE_xxx constants.
 970          */
 971         function getMediaType( $path = null, $mime = null ) {
 972                 if( !$mime && !$path ) {
 973                         return MEDIATYPE_UNKNOWN;
 974                 }
 975
 976                 // If mime type is unknown, guess it
 977                 if( !$mime ) {
 978                         $mime = $this->guessMimeType( $path, false );
 979                 }
 980
 981                 // Special code for ogg - detect if it's video (theora),
 982                 // else label it as sound.
 983                 if ( $mime == 'application/ogg' && file_exists( $path ) ) {
 984
 985                         // Read a chunk of the file
 986                         $f = fopen( $path, "rt" );
 987                         if ( !$f ) return MEDIATYPE_UNKNOWN;
 988                         $head = fread( $f, 256 );
 989                         fclose( $f );
 990
 991                         $head = strtolower( $head );
 992
 993                         // This is an UGLY HACK, file should be parsed correctly
 994                         if ( strpos( $head, 'theora' ) !== false ) return MEDIATYPE_VIDEO;
 995                         elseif ( strpos( $head, 'vorbis' ) !== false ) return MEDIATYPE_AUDIO;
 996                         elseif ( strpos( $head, 'flac' ) !== false ) return MEDIATYPE_AUDIO;
 997                         elseif ( strpos( $head, 'speex' ) !== false ) return MEDIATYPE_AUDIO;
 998                         else return MEDIATYPE_MULTIMEDIA;
 999                 }
1000
1001                 // Check for entry for full mime type
1002                 if( $mime ) {
1003                         $type = $this->findMediaType( $mime );
1004                         if ( $type !== MEDIATYPE_UNKNOWN ) {
1005                                 return $type;
1006                         }
1007                 }
1008
1009                 // Check for entry for file extension
1010                 if ( $path ) {
1011                         $i = strrpos( $path, '.' );
1012                         $e = strtolower( $i ? substr( $path, $i + 1 ) : '' );
1013
1014                         // TODO: look at multi-extension if this fails, parse from full path
1015                         $type = $this->findMediaType( '.' . $e );
1016                         if ( $type !== MEDIATYPE_UNKNOWN ) {
1017                                 return $type;
1018                         }
1019                 }
1020
1021                 // Check major mime type
1022                 if ( $mime ) {
1023                         $i = strpos( $mime, '/' );
1024                         if ( $i !== false ) {
1025                                 $major = substr( $mime, 0, $i );
1026                                 $type = $this->findMediaType( $major );
1027                                 if ( $type !== MEDIATYPE_UNKNOWN ) {
1028                                         return $type;
1029                                 }
1030                         }
1031                 }
1032
1033                 if( !$type ) {
1034                         $type = MEDIATYPE_UNKNOWN;
1035                 }
1036
1037                 return $type;
1038         }
1039
1040         /**
1041          * Returns a media code matching the given mime type or file extension.
1042          * File extensions are represented by a string starting with a dot (.) to
1043          * distinguish them from mime types.
1044          *
1045          * This funktion relies on the mapping defined by $this->mMediaTypes
1046          * @access private
1047          * @return int|string
1048          */
1049         function findMediaType( $extMime ) {
1050                 if ( strpos( $extMime, '.' ) === 0 ) {
1051                         // If it's an extension, look up the mime types
1052                         $m = $this->getTypesForExtension( substr( $extMime, 1 ) );
1053                         if ( !$m ) {
1054                                 return MEDIATYPE_UNKNOWN;
1055                         }
1056
1057                         $m = explode( ' ', $m );
1058                 } else {
1059                         // Normalize mime type
1060                         if ( isset( $this->mMimeTypeAliases[$extMime] ) ) {
1061                                 $extMime = $this->mMimeTypeAliases[$extMime];
1062                         }
1063
1064                         $m = array( $extMime );
1065                 }
1066
1067                 foreach ( $m as $mime ) {
1068                         foreach ( $this->mMediaTypes as $type => $codes ) {
1069                                 if ( in_array($mime, $codes, true ) ) {
1070                                         return $type;
1071                                 }
1072                         }
1073                 }
1074
1075                 return MEDIATYPE_UNKNOWN;
1076         }
1077
1078         /**
1079          * Get the MIME types that various versions of Internet Explorer would
1080          * detect from a chunk of the content.
1081          *
1082          * @param $fileName String: the file name (unused at present)
1083          * @param $chunk String: the first 256 bytes of the file
1084          * @param $proposed String: the MIME type proposed by the server
1085          * @return Array
1086          */
1087         public function getIEMimeTypes( $fileName, $chunk, $proposed ) {
1088                 $ca = $this->getIEContentAnalyzer();
1089                 return $ca->getRealMimesFromData( $fileName, $chunk, $proposed );
1090         }
1091
1092         /**
1093          * Get a cached instance of IEContentAnalyzer
1094          *
1095          * @return IEContentAnalyzer
1096          */
1097         protected function getIEContentAnalyzer() {
1098                 if ( is_null( $this->mIEAnalyzer ) ) {
1099                         $this->mIEAnalyzer = new IEContentAnalyzer;
1100                 }
1101                 return $this->mIEAnalyzer;
1102         }
1103 }