includes/MimeMagic.php

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