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
 406                         // XML formats we sure hope we recognize reliably
 407                         'svg',
 408                 );
 409                 return in_array( strtolower( $extension ), $types );
 410         }
 411
 412         /** improves a mime type using the file extension. Some file formats are very generic,
 413         * so their mime type is not very meaningful. A more useful mime type can be derived
 414         * by looking at the file extension. Typically, this method would be called on the
 415         * result of guessMimeType().
 416         *
 417         * Currently, this method does the following:
 418         *
 419         * If $mime is "unknown/unknown" and isRecognizableExtension( $ext ) returns false,
 420         * return the result of guessTypesForExtension($ext).
 421         *
 422         * If $mime is "application/x-opc+zip" and isMatchingExtension( $ext, $mime )
 423         * gives true, return the result of guessTypesForExtension($ext).
 424         *
 425         * @param $mime String: the mime type, typically guessed from a file's content.
 426         * @param $ext String: the file extension, as taken from the file name
 427         *
 428         * @return string the mime type
 429         */
 430         function improveTypeFromExtension( $mime, $ext ) {
 431                 if ( $mime === "unknown/unknown" ) {
 432                         if( $this->isRecognizableExtension( $ext ) ) {
 433                                 wfDebug( __METHOD__. ": refusing to guess mime type for .$ext file, " .
 434                                         "we should have recognized it\n" );
 435                         } else {
 436                                 /* Not something we can detect, so simply
 437                                 * trust the file extension */
 438                                 $mime = $this->guessTypesForExtension( $ext );
 439                         }
 440                 }
 441                 else if ( $mime === "application/x-opc+zip" ) {
 442                         if ( $this->isMatchingExtension( $ext, $mime ) ) {
 443                                 /* A known file extension for an OPC file,
 444                                 * find the proper mime type for that file extension */
 445                                 $mime = $this->guessTypesForExtension( $ext );
 446                         } else {
 447                                 wfDebug( __METHOD__. ": refusing to guess better type for $mime file, " .
 448                                         ".$ext is not a known OPC extension.\n" );
 449                                 $mime = "application/zip";
 450                         }
 451                 }
 452
 453                 if ( isset( $this->mMimeTypeAliases[$mime] ) ) {
 454                         $mime = $this->mMimeTypeAliases[$mime];
 455                 }
 456
 457                 wfDebug(__METHOD__.": improved mime type for .$ext: $mime\n");
 458                 return $mime;
 459         }
 460
 461         /** mime type detection. This uses detectMimeType to detect the mime type of the file,
 462         * but applies additional checks to determine some well known file formats that may be missed
 463         * or misinterpreter by the default mime detection (namely XML based formats like XHTML or SVG,
 464         * as well as ZIP based formats like OPC/ODF files).
 465         *
 466         * @param $file String: the file to check
 467         * @param $ext Mixed: the file extension, or true (default) to extract it from the filename.
 468         *             Set it to false to ignore the extension. DEPRECATED! Set to false, use
 469         *             improveTypeFromExtension($mime, $ext) later to improve mime type.
 470         *
 471         * @return string the mime type of $file
 472         */
 473         function guessMimeType( $file, $ext = true ) {
 474                 if( $ext ) { # TODO: make $ext default to false. Or better, remove it.
 475                         wfDebug( __METHOD__.": WARNING: use of the \$ext parameter is deprecated. " .
 476                                 "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 477                 }
 478
 479                 $mime = $this->doGuessMimeType( $file, $ext );
 480
 481                 if( !$mime ) {
 482                         wfDebug( __METHOD__.": internal type detection failed for $file (.$ext)...\n" );
 483                         $mime = $this->detectMimeType( $file, $ext );
 484                 }
 485
 486                 if ( isset( $this->mMimeTypeAliases[$mime] ) ) {
 487                         $mime = $this->mMimeTypeAliases[$mime];
 488                 }
 489
 490                 wfDebug(__METHOD__.": guessed mime type of $file: $mime\n");
 491                 return $mime;
 492         }
 493
 494         private function doGuessMimeType( $file, $ext ) { # TODO: remove $ext param
 495                 // Read a chunk of the file
 496                 wfSuppressWarnings();
 497                 $f = fopen( $file, "rt" );
 498                 wfRestoreWarnings();
 499                 if( !$f ) return "unknown/unknown";
 500                 $head = fread( $f, 1024 );
 501                 fseek( $f, -65558, SEEK_END );
 502                 $tail = fread( $f, 65558 ); // 65558 = maximum size of a zip EOCDR
 503                 fclose( $f );
 504
 505                 wfDebug( __METHOD__ . ": analyzing head and tail of $file for magic numbers.\n" );
 506
 507                 // Hardcode a few magic number checks...
 508                 $headers = array(
 509                         // Multimedia...
 510                         'MThd'             => 'audio/midi',
 511                         'OggS'             => 'application/ogg',
 512
 513                         // Image formats...
 514                         // Note that WMF may have a bare header, no magic number.
 515                         "\x01\x00\x09\x00" => 'application/x-msmetafile', // Possibly prone to false positives?
 516                         "\xd7\xcd\xc6\x9a" => 'application/x-msmetafile',
 517                         '%PDF'             => 'application/pdf',
 518                         'gimp xcf'         => 'image/x-xcf',
 519
 520                         // Some forbidden fruit...
 521                         'MZ'               => 'application/octet-stream', // DOS/Windows executable
 522                         "\xca\xfe\xba\xbe" => 'application/octet-stream', // Mach-O binary
 523                         "\x7fELF"          => 'application/octet-stream', // ELF binary
 524                 );
 525
 526                 foreach( $headers as $magic => $candidate ) {
 527                         if( strncmp( $head, $magic, strlen( $magic ) ) == 0 ) {
 528                                 wfDebug( __METHOD__ . ": magic header in $file recognized as $candidate\n" );
 529                                 return $candidate;
 530                         }
 531                 }
 532
 533                 /* Look for WebM and Matroska files */
 534                 if( strncmp( $head, pack( "C4", 0x1a, 0x45, 0xdf, 0xa3 ), 4 ) == 0 ) {
 535                         $doctype = strpos( $head, "\x42\x82" );
 536                         if( $doctype ) {
 537                                 // Next byte is datasize, then data (sizes larger than 1 byte are very stupid muxers)
 538                                 $data = substr($head, $doctype+3, 8);
 539                                 if( strncmp( $data, "matroska", 8 ) == 0 ) {
 540                                         wfDebug( __METHOD__ . ": recognized file as video/x-matroska\n" );
 541                                         return "video/x-matroska";
 542                                 } else if ( strncmp( $data, "webm", 4 ) == 0 ) {
 543                                         wfDebug( __METHOD__ . ": recognized file as video/webm\n" );
 544                                         return "video/webm";
 545                                 }
 546                         }
 547                         wfDebug( __METHOD__ . ": unknown EBML file\n" );
 548                         return "unknown/unknown";
 549                 }
 550
 551                 /*
 552                  * Look for PHP.  Check for this before HTML/XML...  Warning: this is a
 553                  * heuristic, and won't match a file with a lot of non-PHP before.  It
 554                  * will also match text files which could be PHP. :)
 555                  *
 556                  * FIXME: For this reason, the check is probably useless -- an attacker
 557                  * could almost certainly just pad the file with a lot of nonsense to
 558                  * circumvent the check in any case where it would be a security
 559                  * problem.  On the other hand, it causes harmful false positives (bug
 560                  * 16583).  The heuristic has been cut down to exclude three-character
 561                  * strings like "<? ", but should it be axed completely?
 562                  */
 563                 if( ( strpos( $head, '<?php' ) !== false ) ||
 564
 565                     ( strpos( $head, "<\x00?\x00p\x00h\x00p" ) !== false ) ||
 566                     ( strpos( $head, "<\x00?\x00 " ) !== false ) ||
 567                     ( strpos( $head, "<\x00?\x00\n" ) !== false ) ||
 568                     ( strpos( $head, "<\x00?\x00\t" ) !== false ) ||
 569                     ( strpos( $head, "<\x00?\x00=" ) !== false ) ) {
 570
 571                         wfDebug( __METHOD__ . ": recognized $file as application/x-php\n" );
 572                         return "application/x-php";
 573                 }
 574
 575                 /*
 576                  * look for XML formats (XHTML and SVG)
 577                  */
 578                 $xml = new XmlTypeCheck( $file );
 579                 if( $xml->wellFormed ) {
 580                         global $wgXMLMimeTypes;
 581                         if( isset( $wgXMLMimeTypes[$xml->getRootElement()] ) ) {
 582                                 return $wgXMLMimeTypes[$xml->getRootElement()];
 583                         } else {
 584                                 return 'application/xml';
 585                         }
 586                 }
 587
 588                 /*
 589                  * look for shell scripts
 590                  */
 591                 $script_type = null;
 592
 593                 # detect by shebang
 594                 if ( substr( $head, 0, 2) == "#!" ) {
 595                         $script_type = "ASCII";
 596                 } elseif ( substr( $head, 0, 5) == "\xef\xbb\xbf#!" ) {
 597                         $script_type = "UTF-8";
 598                 } elseif ( substr( $head, 0, 7) == "\xfe\xff\x00#\x00!" ) {
 599                         $script_type = "UTF-16BE";
 600                 } elseif ( substr( $head, 0, 7 ) == "\xff\xfe#\x00!" ) {
 601                         $script_type= "UTF-16LE";
 602                 }
 603
 604                 if ( $script_type ) {
 605                         if ( $script_type !== "UTF-8" && $script_type !== "ASCII") {
 606                                 // Quick and dirty fold down to ASCII!
 607                                 $pack = array( 'UTF-16BE' => 'n*', 'UTF-16LE' => 'v*' );
 608                                 $chars = unpack( $pack[$script_type], substr( $head, 2 ) );
 609                                 $head = '';
 610                                 foreach( $chars as $codepoint ) {
 611                                         if( $codepoint < 128 ) {
 612                                                 $head .= chr( $codepoint );
 613                                         } else {
 614                                                 $head .= '?';
 615                                         }
 616                                 }
 617                         }
 618
 619                         $match = array();
 620
 621                         if ( preg_match( '%/?([^\s]+/)(\w+)%', $head, $match ) ) {
 622                                 $mime = "application/x-{$match[2]}";
 623                                 wfDebug( __METHOD__.": shell script recognized as $mime\n" );
 624                                 return $mime;
 625                         }
 626                 }
 627
 628                 // Check for ZIP variants (before getimagesize)
 629                 if ( strpos( $tail, "PK\x05\x06" ) !== false ) {
 630                         wfDebug( __METHOD__.": ZIP header present in $file\n" );
 631                         return $this->detectZipType( $head, $tail, $ext );
 632                 }
 633
 634                 wfSuppressWarnings();
 635                 $gis = getimagesize( $file );
 636                 wfRestoreWarnings();
 637
 638                 if( $gis && isset( $gis['mime'] ) ) {
 639                         $mime = $gis['mime'];
 640                         wfDebug( __METHOD__.": getimagesize detected $file as $mime\n" );
 641                         return $mime;
 642                 }
 643
 644                 // Also test DjVu
 645                 $deja = new DjVuImage( $file );
 646                 if( $deja->isValid() ) {
 647                         wfDebug( __METHOD__.": detected $file as image/vnd.djvu\n" );
 648                         return 'image/vnd.djvu';
 649                 }
 650
 651                 return false;
 652         }
 653
 654         /**
 655          * Detect application-specific file type of a given ZIP file from its
 656          * header data.  Currently works for OpenDocument and OpenXML types...
 657          * If can't tell, returns 'application/zip'.
 658          *
 659          * @param $header String: some reasonably-sized chunk of file header
 660          * @param $tail   String: the tail of the file
 661          * @param $ext Mixed: the file extension, or true to extract it from the filename.
 662          *             Set it to false (default) to ignore the extension. DEPRECATED! Set to false,
 663          *             use improveTypeFromExtension($mime, $ext) later to improve mime type.
 664          *
 665          * @return string
 666          */
 667         function detectZipType( $header, $tail = null, $ext = false ) {
 668                 if( $ext ) { # TODO: remove $ext param
 669                         wfDebug( __METHOD__.": WARNING: use of the \$ext parameter is deprecated. " .
 670                                 "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 671                 }
 672
 673                 $mime = 'application/zip';
 674                 $opendocTypes = array(
 675                         'chart-template',
 676                         'chart',
 677                         'formula-template',
 678                         'formula',
 679                         'graphics-template',
 680                         'graphics',
 681                         'image-template',
 682                         'image',
 683                         'presentation-template',
 684                         'presentation',
 685                         'spreadsheet-template',
 686                         'spreadsheet',
 687                         'text-template',
 688                         'text-master',
 689                         'text-web',
 690                         'text' );
 691
 692                 // http://lists.oasis-open.org/archives/office/200505/msg00006.html
 693                 $types = '(?:' . implode( '|', $opendocTypes ) . ')';
 694                 $opendocRegex = "/^mimetype(application\/vnd\.oasis\.opendocument\.$types)/";
 695
 696                 $openxmlRegex = "/^\[Content_Types\].xml/";
 697
 698                 if( preg_match( $opendocRegex, substr( $header, 30 ), $matches ) ) {
 699                         $mime = $matches[1];
 700                         wfDebug( __METHOD__.": detected $mime from ZIP archive\n" );
 701                 } elseif( preg_match( $openxmlRegex, substr( $header, 30 ) ) ) {
 702                         $mime = "application/x-opc+zip";
 703                         # TODO: remove the block below, as soon as improveTypeFromExtension is used everywhere
 704                         if( $ext !== true && $ext !== false ) {
 705                                 /** This is the mode used by getPropsFromPath
 706                                 * These mime's are stored in the database, where we don't really want
 707                                 * x-opc+zip, because we use it only for internal purposes
 708                                 */
 709                                 if( $this->isMatchingExtension( $ext, $mime) ) {
 710                                         /* A known file extension for an OPC file,
 711                                         * find the proper mime type for that file extension */
 712                                         $mime = $this->guessTypesForExtension( $ext );
 713                                 } else {
 714                                         $mime = "application/zip";
 715                                 }
 716                         }
 717                         wfDebug( __METHOD__.": detected an Open Packaging Conventions archive: $mime\n" );
 718                 } else if( substr( $header, 0, 8 ) == "\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" &&
 719                                 ($headerpos = strpos( $tail, "PK\x03\x04" ) ) !== false &&
 720                                 preg_match( $openxmlRegex, substr( $tail, $headerpos + 30 ) ) ) {
 721                         if( substr( $header, 512, 4) == "\xEC\xA5\xC1\x00" ) {
 722                                 $mime = "application/msword";
 723                         }
 724                         switch( substr( $header, 512, 6) ) {
 725                                 case "\xEC\xA5\xC1\x00\x0E\x00":
 726                                 case "\xEC\xA5\xC1\x00\x1C\x00":
 727                                 case "\xEC\xA5\xC1\x00\x43\x00":
 728                                         $mime = "application/vnd.ms-powerpoint";
 729                                         break;
 730                                 case "\xFD\xFF\xFF\xFF\x10\x00":
 731                                 case "\xFD\xFF\xFF\xFF\x1F\x00":
 732                                 case "\xFD\xFF\xFF\xFF\x22\x00":
 733                                 case "\xFD\xFF\xFF\xFF\x23\x00":
 734                                 case "\xFD\xFF\xFF\xFF\x28\x00":
 735                                 case "\xFD\xFF\xFF\xFF\x29\x00":
 736                                 case "\xFD\xFF\xFF\xFF\x10\x02":
 737                                 case "\xFD\xFF\xFF\xFF\x1F\x02":
 738                                 case "\xFD\xFF\xFF\xFF\x22\x02":
 739                                 case "\xFD\xFF\xFF\xFF\x23\x02":
 740                                 case "\xFD\xFF\xFF\xFF\x28\x02":
 741                                 case "\xFD\xFF\xFF\xFF\x29\x02":
 742                                         $mime = "application/vnd.msexcel";
 743                                         break;
 744                         }
 745
 746                         wfDebug( __METHOD__.": detected a MS Office document with OPC trailer\n");
 747                 } else {
 748                         wfDebug( __METHOD__.": unable to identify type of ZIP archive\n" );
 749                 }
 750                 return $mime;
 751         }
 752
 753         /** Internal mime type detection, please use guessMimeType() for application code instead.
 754         * Detection is done using an external program, if $wgMimeDetectorCommand is set.
 755         * Otherwise, the fileinfo extension and mime_content_type are tried (in this order), if they are available.
 756         * If the dections fails and $ext is not false, the mime type is guessed from the file extension, using
 757         * guessTypesForExtension.
 758         * If the mime type is still unknown, getimagesize is used to detect the mime type if the file is an image.
 759         * If no mime type can be determined, this function returns "unknown/unknown".
 760         *
 761         * @param $file String: the file to check
 762         * @param $ext Mixed: the file extension, or true (default) to extract it from the filename.
 763         *             Set it to false to ignore the extension. DEPRECATED! Set to false, use
 764         *             improveTypeFromExtension($mime, $ext) later to improve mime type.
 765         *
 766         * @return string the mime type of $file
 767         * @access private
 768         */
 769         private function detectMimeType( $file, $ext = true ) {
 770                 global $wgMimeDetectorCommand;
 771
 772                 if( $ext ) { # TODO:  make $ext default to false. Or better, remove it.
 773                         wfDebug( __METHOD__.": WARNING: use of the \$ext parameter is deprecated. Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 774                 }
 775
 776                 $m = null;
 777                 if ( $wgMimeDetectorCommand ) {
 778                         $fn = wfEscapeShellArg( $file );
 779                         $m = `$wgMimeDetectorCommand $fn`;
 780                 } elseif ( function_exists( "finfo_open" ) && function_exists( "finfo_file" ) ) {
 781
 782                         # This required the fileinfo extension by PECL,
 783                         # see http://pecl.php.net/package/fileinfo
 784                         # This must be compiled into PHP
 785                         #
 786                         # finfo is the official replacement for the deprecated
 787                         # mime_content_type function, see below.
 788                         #
 789                         # If you may need to load the fileinfo extension at runtime, set
 790                         # $wgLoadFileinfoExtension in LocalSettings.php
 791
 792                         $mime_magic_resource = finfo_open(FILEINFO_MIME); /* return mime type ala mimetype extension */
 793
 794                         if ($mime_magic_resource) {
 795                                 $m = finfo_file( $mime_magic_resource, $file );
 796                                 finfo_close( $mime_magic_resource );
 797                         } else {
 798                                 wfDebug( __METHOD__.": finfo_open failed on ".FILEINFO_MIME."!\n" );
 799                         }
 800                 } elseif ( function_exists( "mime_content_type" ) ) {
 801
 802                         # NOTE: this function is available since PHP 4.3.0, but only if
 803                         # PHP was compiled with --with-mime-magic or, before 4.3.2, with --enable-mime-magic.
 804                         #
 805                         # On Windows, you must set mime_magic.magicfile in php.ini to point to the mime.magic file bundeled with PHP;
 806                         # sometimes, this may even be needed under linus/unix.
 807                         #
 808                         # Also note that this has been DEPRECATED in favor of the fileinfo extension by PECL, see above.
 809                         # see http://www.php.net/manual/en/ref.mime-magic.php for details.
 810
 811                         $m = mime_content_type($file);
 812                 } else {
 813                         wfDebug( __METHOD__.": no magic mime detector found!\n" );
 814                 }
 815
 816                 if ( $m ) {
 817                         # normalize
 818                         $m = preg_replace( '![;, ].*$!', '', $m ); #strip charset, etc
 819                         $m = trim( $m );
 820                         $m = strtolower( $m );
 821
 822                         if ( strpos( $m, 'unknown' ) !== false ) {
 823                                 $m = null;
 824                         } else {
 825                                 wfDebug( __METHOD__.": magic mime type of $file: $m\n" );
 826                                 return $m;
 827                         }
 828                 }
 829
 830                 # if desired, look at extension as a fallback.
 831                 if ( $ext === true ) {
 832                         $i = strrpos( $file, '.' );
 833                         $ext = strtolower( $i ? substr( $file, $i + 1 ) : '' );
 834                 }
 835                 if ( $ext ) {
 836                         if( $this->isRecognizableExtension( $ext ) ) {
 837                                 wfDebug( __METHOD__. ": refusing to guess mime type for .$ext file, we should have recognized it\n" );
 838                         } else {
 839                                 $m = $this->guessTypesForExtension( $ext );
 840                                 if ( $m ) {
 841                                         wfDebug( __METHOD__.": extension mime type of $file: $m\n" );
 842                                         return $m;
 843                                 }
 844                         }
 845                 }
 846
 847                 #unknown type
 848                 wfDebug( __METHOD__.": failed to guess mime type for $file!\n" );
 849                 return "unknown/unknown";
 850         }
 851
 852         /**
 853         * Determine the media type code for a file, using its mime type, name and possibly
 854         * its contents.
 855         *
 856         * This function relies on the findMediaType(), mapping extensions and mime
 857         * types to media types.
 858         *
 859         * @todo analyse file if need be
 860         * @todo look at multiple extension, separately and together.
 861         *
 862         * @param $path String: full path to the image file, in case we have to look at the contents
 863         *        (if null, only the mime type is used to determine the media type code).
 864         * @param $mime String: mime type. If null it will be guessed using guessMimeType.
 865         *
 866         * @return (int?string?) a value to be used with the MEDIATYPE_xxx constants.
 867         */
 868         function getMediaType( $path = null, $mime = null ) {
 869                 if( !$mime && !$path ) return MEDIATYPE_UNKNOWN;
 870
 871                 # If mime type is unknown, guess it
 872                 if( !$mime ) $mime = $this->guessMimeType( $path, false );
 873
 874                 # Special code for ogg - detect if it's video (theora),
 875                 # else label it as sound.
 876                 if( $mime == "application/ogg" && file_exists( $path ) ) {
 877
 878                         // Read a chunk of the file
 879                         $f = fopen( $path, "rt" );
 880                         if ( !$f ) return MEDIATYPE_UNKNOWN;
 881                         $head = fread( $f, 256 );
 882                         fclose( $f );
 883
 884                         $head = strtolower( $head );
 885
 886                         # This is an UGLY HACK, file should be parsed correctly
 887                         if ( strpos( $head, 'theora' ) !== false ) return MEDIATYPE_VIDEO;
 888                         elseif ( strpos( $head, 'vorbis' ) !== false ) return MEDIATYPE_AUDIO;
 889                         elseif ( strpos( $head, 'flac' ) !== false ) return MEDIATYPE_AUDIO;
 890                         elseif ( strpos( $head, 'speex' ) !== false ) return MEDIATYPE_AUDIO;
 891                         else return MEDIATYPE_MULTIMEDIA;
 892                 }
 893
 894                 # check for entry for full mime type
 895                 if( $mime ) {
 896                         $type = $this->findMediaType( $mime );
 897                         if( $type !== MEDIATYPE_UNKNOWN ) return $type;
 898                 }
 899
 900                 # Check for entry for file extension
 901                 $e = null;
 902                 if ( $path ) {
 903                         $i = strrpos( $path, '.' );
 904                         $e = strtolower( $i ? substr( $path, $i + 1 ) : '' );
 905
 906                         # TODO: look at multi-extension if this fails, parse from full path
 907
 908                         $type = $this->findMediaType( '.' . $e );
 909                         if ( $type !== MEDIATYPE_UNKNOWN ) return $type;
 910                 }
 911
 912                 # Check major mime type
 913                 if( $mime ) {
 914                         $i = strpos( $mime, '/' );
 915                         if( $i !== false ) {
 916                                 $major = substr( $mime, 0, $i );
 917                                 $type = $this->findMediaType( $major );
 918                                 if( $type !== MEDIATYPE_UNKNOWN ) return $type;
 919                         }
 920                 }
 921
 922                 if( !$type ) $type = MEDIATYPE_UNKNOWN;
 923
 924                 return $type;
 925         }
 926
 927         /** returns a media code matching the given mime type or file extension.
 928         * File extensions are represented by a string starting with a dot (.) to
 929         * distinguish them from mime types.
 930         *
 931         * This funktion relies on the mapping defined by $this->mMediaTypes
 932         * @access private
 933         */
 934         function findMediaType( $extMime ) {
 935                 if ( strpos( $extMime, '.' ) === 0 ) { #if it's an extension, look up the mime types
 936                         $m = $this->getTypesForExtension( substr( $extMime, 1 ) );
 937                         if ( !$m ) return MEDIATYPE_UNKNOWN;
 938
 939                         $m = explode( ' ', $m );
 940                 } else {
 941                         # Normalize mime type
 942                         if ( isset( $this->mMimeTypeAliases[$extMime] ) ) {
 943                                 $extMime = $this->mMimeTypeAliases[$extMime];
 944                         }
 945
 946                         $m = array($extMime);
 947                 }
 948
 949                 foreach ( $m as $mime ) {
 950                         foreach ( $this->mMediaTypes as $type => $codes ) {
 951                                 if ( in_array($mime, $codes, true ) ) {
 952                                         return $type;
 953                                 }
 954                         }
 955                 }
 956
 957                 return MEDIATYPE_UNKNOWN;
 958         }
 959
 960         /**
 961          * Get the MIME types that various versions of Internet Explorer would
 962          * detect from a chunk of the content.
 963          *
 964          * @param $fileName String: the file name (unused at present)
 965          * @param $chunk String: the first 256 bytes of the file
 966          * @param $proposed String: the MIME type proposed by the server
 967          */
 968         public function getIEMimeTypes( $fileName, $chunk, $proposed ) {
 969                 $ca = $this->getIEContentAnalyzer();
 970                 return $ca->getRealMimesFromData( $fileName, $chunk, $proposed );
 971         }
 972
 973         /**
 974          * Get a cached instance of IEContentAnalyzer
 975          */
 976         protected function getIEContentAnalyzer() {
 977                 if ( is_null( $this->mIEAnalyzer ) ) {
 978                         $this->mIEAnalyzer = new IEContentAnalyzer;
 979                 }
 980                 return $this->mIEAnalyzer;
 981         }
 982 }