includes/MimeMagic.php

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