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