includes/MimeMagic.php

   1 <?php
   2 /**
   3  * Module defining helper functions for detecting and dealing with mime types.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  */
  22
  23 /**
  24  * Defines a set of well known mime types
  25  * This is used as a fallback to mime.types files.
  26  * An extensive list of well known mime types is provided by
  27  * the file mime.types in the includes directory.
  28  *
  29  * This list concatenated with mime.types is used to create a mime <-> ext
  30  * map. Each line contains a mime type followed by a space separated list of
  31  * extensions. If multiple extensions for a single mime type exist or if
  32  * multiple mime types exist for a single extension then in most cases
  33  * MediaWiki assumes that the first extension following the mime type is the
  34  * canonical extension, and the first time a mime type appears for a certain
  35  * extension is considered the canonical mime type.
  36  *
  37  * (Note that appending $wgMimeTypeFile to the end of MM_WELL_KNOWN_MIME_TYPES
  38  * sucks because you can't redefine canonical types. This could be fixed by
  39  * appending MM_WELL_KNOWN_MIME_TYPES behind $wgMimeTypeFile, but who knows
  40  * what will break? In practice this probably isn't a problem anyway -- Bryan)
  41  */
  42 define( 'MM_WELL_KNOWN_MIME_TYPES', <<<END_STRING
  43 application/ogg ogx ogg ogm ogv oga spx
  44 application/pdf pdf
  45 application/vnd.oasis.opendocument.chart odc
  46 application/vnd.oasis.opendocument.chart-template otc
  47 application/vnd.oasis.opendocument.database odb
  48 application/vnd.oasis.opendocument.formula odf
  49 application/vnd.oasis.opendocument.formula-template otf
  50 application/vnd.oasis.opendocument.graphics odg
  51 application/vnd.oasis.opendocument.graphics-template otg
  52 application/vnd.oasis.opendocument.image odi
  53 application/vnd.oasis.opendocument.image-template oti
  54 application/vnd.oasis.opendocument.presentation odp
  55 application/vnd.oasis.opendocument.presentation-template otp
  56 application/vnd.oasis.opendocument.spreadsheet ods
  57 application/vnd.oasis.opendocument.spreadsheet-template ots
  58 application/vnd.oasis.opendocument.text odt
  59 application/vnd.oasis.opendocument.text-master otm
  60 application/vnd.oasis.opendocument.text-template ott
  61 application/vnd.oasis.opendocument.text-web oth
  62 application/x-javascript js
  63 application/x-shockwave-flash swf
  64 audio/midi mid midi kar
  65 audio/mpeg mpga mpa mp2 mp3
  66 audio/x-aiff aif aiff aifc
  67 audio/x-wav wav
  68 audio/ogg oga spx ogg
  69 image/x-bmp bmp
  70 image/gif gif
  71 image/jpeg jpeg jpg jpe
  72 image/png png
  73 image/svg+xml svg
  74 image/svg svg
  75 image/tiff tiff tif
  76 image/vnd.djvu djvu
  77 image/x.djvu djvu
  78 image/x-djvu djvu
  79 image/x-portable-pixmap ppm
  80 image/x-xcf xcf
  81 text/plain txt
  82 text/html html htm
  83 video/ogg ogv ogm ogg
  84 video/mpeg mpg mpeg
  85 END_STRING
  86 );
  87
  88 /**
  89  * Defines a set of well known mime info entries
  90  * This is used as a fallback to mime.info files.
  91  * An extensive list of well known mime types is provided by
  92  * the file mime.info in the includes directory.
  93  */
  94 define( 'MM_WELL_KNOWN_MIME_INFO', <<<END_STRING
  95 application/pdf [OFFICE]
  96 application/vnd.oasis.opendocument.chart [OFFICE]
  97 application/vnd.oasis.opendocument.chart-template [OFFICE]
  98 application/vnd.oasis.opendocument.database [OFFICE]
  99 application/vnd.oasis.opendocument.formula [OFFICE]
 100 application/vnd.oasis.opendocument.formula-template [OFFICE]
 101 application/vnd.oasis.opendocument.graphics [OFFICE]
 102 application/vnd.oasis.opendocument.graphics-template [OFFICE]
 103 application/vnd.oasis.opendocument.image [OFFICE]
 104 application/vnd.oasis.opendocument.image-template [OFFICE]
 105 application/vnd.oasis.opendocument.presentation [OFFICE]
 106 application/vnd.oasis.opendocument.presentation-template [OFFICE]
 107 application/vnd.oasis.opendocument.spreadsheet [OFFICE]
 108 application/vnd.oasis.opendocument.spreadsheet-template [OFFICE]
 109 application/vnd.oasis.opendocument.text [OFFICE]
 110 application/vnd.oasis.opendocument.text-template [OFFICE]
 111 application/vnd.oasis.opendocument.text-master [OFFICE]
 112 application/vnd.oasis.opendocument.text-web [OFFICE]
 113 text/javascript application/x-javascript [EXECUTABLE]
 114 application/x-shockwave-flash [MULTIMEDIA]
 115 audio/midi [AUDIO]
 116 audio/x-aiff [AUDIO]
 117 audio/x-wav [AUDIO]
 118 audio/mp3 audio/mpeg [AUDIO]
 119 application/ogg audio/ogg video/ogg [MULTIMEDIA]
 120 image/x-bmp image/x-ms-bmp image/bmp [BITMAP]
 121 image/gif [BITMAP]
 122 image/jpeg [BITMAP]
 123 image/png [BITMAP]
 124 image/svg+xml [DRAWING]
 125 image/tiff [BITMAP]
 126 image/vnd.djvu [BITMAP]
 127 image/x-xcf [BITMAP]
 128 image/x-portable-pixmap [BITMAP]
 129 text/plain [TEXT]
 130 text/html [TEXT]
 131 video/ogg [VIDEO]
 132 video/mpeg [VIDEO]
 133 unknown/unknown application/octet-stream application/x-empty [UNKNOWN]
 134 END_STRING
 135 );
 136
 137 /**
 138  * Implements functions related to mime types such as detection and mapping to
 139  * file extension.
 140  *
 141  * Instances of this class are stateless, there only needs to be one global instance
 142  * of MimeMagic. Please use MimeMagic::singleton() to get that instance.
 143  */
 144 class MimeMagic {
 145         /**
 146          * @var array Mapping of media types to arrays of mime types.
 147          * This is used by findMediaType and getMediaType, respectively
 148          */
 149         protected $mMediaTypes = null;
 150
 151         /** @var array Map of mime type aliases
 152          */
 153         protected $mMimeTypeAliases = null;
 154
 155         /** @var array Map of mime types to file extensions (as a space separated list)
 156          */
 157         protected $mMimeToExt = null;
 158
 159         /** @var array Map of file extensions types to mime types (as a space separated list)
 160          */
 161         public $mExtToMime = null;
 162
 163         /** @var IEContentAnalyzer
 164          */
 165         protected $mIEAnalyzer;
 166
 167         /** @var MimeMagic The singleton instance
 168          */
 169         private static $instance = null;
 170
 171         /** Initializes the MimeMagic object. This is called by MimeMagic::singleton().
 172          *
 173          * This constructor parses the mime.types and mime.info files and build internal mappings.
 174          */
 175         function __construct() {
 176                 /**
 177                  *   --- load mime.types ---
 178                  */
 179
 180                 global $wgMimeTypeFile, $IP;
 181
 182                 $types = MM_WELL_KNOWN_MIME_TYPES;
 183
 184                 if ( $wgMimeTypeFile == 'includes/mime.types' ) {
 185                         $wgMimeTypeFile = "$IP/$wgMimeTypeFile";
 186                 }
 187
 188                 if ( $wgMimeTypeFile ) {
 189                         if ( is_file( $wgMimeTypeFile ) and is_readable( $wgMimeTypeFile ) ) {
 190                                 wfDebug( __METHOD__ . ": loading mime types from $wgMimeTypeFile\n" );
 191                                 $types .= "\n";
 192                                 $types .= file_get_contents( $wgMimeTypeFile );
 193                         } else {
 194                                 wfDebug( __METHOD__ . ": can't load mime types from $wgMimeTypeFile\n" );
 195                         }
 196                 } else {
 197                         wfDebug( __METHOD__ . ": no mime types file defined, using build-ins only.\n" );
 198                 }
 199
 200                 $types = str_replace( array( "\r\n", "\n\r", "\n\n", "\r\r", "\r" ), "\n", $types );
 201                 $types = str_replace( "\t", " ", $types );
 202
 203                 $this->mMimeToExt = array();
 204                 $this->mToMime = array();
 205
 206                 $lines = explode( "\n", $types );
 207                 foreach ( $lines as $s ) {
 208                         $s = trim( $s );
 209                         if ( empty( $s ) ) {
 210                                 continue;
 211                         }
 212                         if ( strpos( $s, '#' ) === 0 ) {
 213                                 continue;
 214                         }
 215
 216                         $s = strtolower( $s );
 217                         $i = strpos( $s, ' ' );
 218
 219                         if ( $i === false ) {
 220                                 continue;
 221                         }
 222
 223                         $mime = substr( $s, 0, $i );
 224                         $ext = trim( substr( $s, $i + 1 ) );
 225
 226                         if ( empty( $ext ) ) {
 227                                 continue;
 228                         }
 229
 230                         if ( !empty( $this->mMimeToExt[$mime] ) ) {
 231                                 $this->mMimeToExt[$mime] .= ' ' . $ext;
 232                         } else {
 233                                 $this->mMimeToExt[$mime] = $ext;
 234                         }
 235
 236                         $extensions = explode( ' ', $ext );
 237
 238                         foreach ( $extensions as $e ) {
 239                                 $e = trim( $e );
 240                                 if ( empty( $e ) ) {
 241                                         continue;
 242                                 }
 243
 244                                 if ( !empty( $this->mExtToMime[$e] ) ) {
 245                                         $this->mExtToMime[$e] .= ' ' . $mime;
 246                                 } else {
 247                                         $this->mExtToMime[$e] = $mime;
 248                                 }
 249                         }
 250                 }
 251
 252                 /**
 253                  *   --- load mime.info ---
 254                  */
 255
 256                 global $wgMimeInfoFile;
 257                 if ( $wgMimeInfoFile == 'includes/mime.info' ) {
 258                         $wgMimeInfoFile = "$IP/$wgMimeInfoFile";
 259                 }
 260
 261                 $info = MM_WELL_KNOWN_MIME_INFO;
 262
 263                 if ( $wgMimeInfoFile ) {
 264                         if ( is_file( $wgMimeInfoFile ) and is_readable( $wgMimeInfoFile ) ) {
 265                                 wfDebug( __METHOD__ . ": loading mime info from $wgMimeInfoFile\n" );
 266                                 $info .= "\n";
 267                                 $info .= file_get_contents( $wgMimeInfoFile );
 268                         } else {
 269                                 wfDebug( __METHOD__ . ": can't load mime info from $wgMimeInfoFile\n" );
 270                         }
 271                 } else {
 272                         wfDebug( __METHOD__ . ": no mime info file defined, using build-ins only.\n" );
 273                 }
 274
 275                 $info = str_replace( array( "\r\n", "\n\r", "\n\n", "\r\r", "\r" ), "\n", $info );
 276                 $info = str_replace( "\t", " ", $info );
 277
 278                 $this->mMimeTypeAliases = array();
 279                 $this->mMediaTypes = array();
 280
 281                 $lines = explode( "\n", $info );
 282                 foreach ( $lines as $s ) {
 283                         $s = trim( $s );
 284                         if ( empty( $s ) ) {
 285                                 continue;
 286                         }
 287                         if ( strpos( $s, '#' ) === 0 ) {
 288                                 continue;
 289                         }
 290
 291                         $s = strtolower( $s );
 292                         $i = strpos( $s, ' ' );
 293
 294                         if ( $i === false ) {
 295                                 continue;
 296                         }
 297
 298                         #print "processing MIME INFO line $s<br>";
 299
 300                         $match = array();
 301                         if ( preg_match( '!\[\s*(\w+)\s*\]!', $s, $match ) ) {
 302                                 $s = preg_replace( '!\[\s*(\w+)\s*\]!', '', $s );
 303                                 $mtype = trim( strtoupper( $match[1] ) );
 304                         } else {
 305                                 $mtype = MEDIATYPE_UNKNOWN;
 306                         }
 307
 308                         $m = explode( ' ', $s );
 309
 310                         if ( !isset( $this->mMediaTypes[$mtype] ) ) {
 311                                 $this->mMediaTypes[$mtype] = array();
 312                         }
 313
 314                         foreach ( $m as $mime ) {
 315                                 $mime = trim( $mime );
 316                                 if ( empty( $mime ) ) {
 317                                         continue;
 318                                 }
 319
 320                                 $this->mMediaTypes[$mtype][] = $mime;
 321                         }
 322
 323                         if ( count( $m ) > 1 ) {
 324                                 $main = $m[0];
 325                                 $mCount = count( $m );
 326                                 for ( $i = 1; $i < $mCount; $i += 1 ) {
 327                                         $mime = $m[$i];
 328                                         $this->mMimeTypeAliases[$mime] = $main;
 329                                 }
 330                         }
 331                 }
 332         }
 333
 334         /**
 335          * Get an instance of this class
 336          * @return MimeMagic
 337          */
 338         public static function singleton() {
 339                 if ( self::$instance === null ) {
 340                         self::$instance = new MimeMagic;
 341                 }
 342                 return self::$instance;
 343         }
 344
 345         /**
 346          * Returns a list of file extensions for a given mime type as a space
 347          * separated string or null if the mime type was unrecognized. Resolves
 348          * mime type aliases.
 349          *
 350          * @param string $mime
 351          * @return string|null
 352          */
 353         public function getExtensionsForType( $mime ) {
 354                 $mime = strtolower( $mime );
 355
 356                 // Check the mime-to-ext map
 357                 if ( isset( $this->mMimeToExt[$mime] ) ) {
 358                         return $this->mMimeToExt[$mime];
 359                 }
 360
 361                 // Resolve the mime type to the canonical type
 362                 if ( isset( $this->mMimeTypeAliases[$mime] ) ) {
 363                         $mime = $this->mMimeTypeAliases[$mime];
 364                         if ( isset( $this->mMimeToExt[$mime] ) ) {
 365                                 return $this->mMimeToExt[$mime];
 366                         }
 367                 }
 368
 369                 return null;
 370         }
 371
 372         /**
 373          * Returns a list of mime types for a given file extension as a space
 374          * separated string or null if the extension was unrecognized.
 375          *
 376          * @param string $ext
 377          * @return string|null
 378          */
 379         public function getTypesForExtension( $ext ) {
 380                 $ext = strtolower( $ext );
 381
 382                 $r = isset( $this->mExtToMime[$ext] ) ? $this->mExtToMime[$ext] : null;
 383                 return $r;
 384         }
 385
 386         /**
 387          * Returns a single mime type for a given file extension or null if unknown.
 388          * This is always the first type from the list returned by getTypesForExtension($ext).
 389          *
 390          * @param string $ext
 391          * @return string|null
 392          */
 393         public function guessTypesForExtension( $ext ) {
 394                 $m = $this->getTypesForExtension( $ext );
 395                 if ( is_null( $m ) ) {
 396                         return null;
 397                 }
 398
 399                 // TODO: Check if this is needed; strtok( $m, ' ' ) should be sufficient
 400                 $m = trim( $m );
 401                 $m = preg_replace( '/\s.*$/', '', $m );
 402
 403                 return $m;
 404         }
 405
 406         /**
 407          * Tests if the extension matches the given mime type. Returns true if a
 408          * match was found, null if the mime type is unknown, and false if the
 409          * mime type is known but no matches where found.
 410          *
 411          * @param string $extension
 412          * @param string $mime
 413          * @return bool|null
 414          */
 415         public function isMatchingExtension( $extension, $mime ) {
 416                 $ext = $this->getExtensionsForType( $mime );
 417
 418                 if ( !$ext ) {
 419                         return null; // Unknown mime type
 420                 }
 421
 422                 $ext = explode( ' ', $ext );
 423
 424                 $extension = strtolower( $extension );
 425                 return in_array( $extension, $ext );
 426         }
 427
 428         /**
 429          * Returns true if the mime type is known to represent an image format
 430          * supported by the PHP GD library.
 431          *
 432          * @param string $mime
 433          *
 434          * @return bool
 435          */
 436         public function isPHPImageType( $mime ) {
 437                 // As defined by imagegetsize and image_type_to_mime
 438                 static $types = array(
 439                         'image/gif', 'image/jpeg', 'image/png',
 440                         'image/x-bmp', 'image/xbm', 'image/tiff',
 441                         'image/jp2', 'image/jpeg2000', 'image/iff',
 442                         'image/xbm', 'image/x-xbitmap',
 443                         'image/vnd.wap.wbmp', 'image/vnd.xiff',
 444                         'image/x-photoshop',
 445                         'application/x-shockwave-flash',
 446                 );
 447
 448                 return in_array( $mime, $types );
 449         }
 450
 451         /**
 452          * Returns true if the extension represents a type which can
 453          * be reliably detected from its content. Use this to determine
 454          * whether strict content checks should be applied to reject
 455          * invalid uploads; if we can't identify the type we won't
 456          * be able to say if it's invalid.
 457          *
 458          * @todo Be more accurate when using fancy mime detector plugins;
 459          *       right now this is the bare minimum getimagesize() list.
 460          * @param string $extension
 461          * @return bool
 462          */
 463         function isRecognizableExtension( $extension ) {
 464                 static $types = array(
 465                         // Types recognized by getimagesize()
 466                         'gif', 'jpeg', 'jpg', 'png', 'swf', 'psd',
 467                         'bmp', 'tiff', 'tif', 'jpc', 'jp2',
 468                         'jpx', 'jb2', 'swc', 'iff', 'wbmp',
 469                         'xbm',
 470
 471                         // Formats we recognize magic numbers for
 472                         'djvu', 'ogx', 'ogg', 'ogv', 'oga', 'spx',
 473                         'mid', 'pdf', 'wmf', 'xcf', 'webm', 'mkv', 'mka',
 474                         'webp',
 475
 476                         // XML formats we sure hope we recognize reliably
 477                         'svg',
 478                 );
 479                 return in_array( strtolower( $extension ), $types );
 480         }
 481
 482         /**
 483          * Improves a mime type using the file extension. Some file formats are very generic,
 484          * so their mime type is not very meaningful. A more useful mime type can be derived
 485          * by looking at the file extension. Typically, this method would be called on the
 486          * result of guessMimeType().
 487          *
 488          * Currently, this method does the following:
 489          *
 490          * If $mime is "unknown/unknown" and isRecognizableExtension( $ext ) returns false,
 491          * return the result of guessTypesForExtension($ext).
 492          *
 493          * If $mime is "application/x-opc+zip" and isMatchingExtension( $ext, $mime )
 494          * gives true, return the result of guessTypesForExtension($ext).
 495          *
 496          * @param string $mime The mime type, typically guessed from a file's content.
 497          * @param string $ext The file extension, as taken from the file name
 498          *
 499          * @return string The mime type
 500          */
 501         public function improveTypeFromExtension( $mime, $ext ) {
 502                 if ( $mime === 'unknown/unknown' ) {
 503                         if ( $this->isRecognizableExtension( $ext ) ) {
 504                                 wfDebug( __METHOD__ . ': refusing to guess mime type for .' .
 505                                         "$ext file, we should have recognized it\n" );
 506                         } else {
 507                                 // Not something we can detect, so simply
 508                                 // trust the file extension
 509                                 $mime = $this->guessTypesForExtension( $ext );
 510                         }
 511                 } elseif ( $mime === 'application/x-opc+zip' ) {
 512                         if ( $this->isMatchingExtension( $ext, $mime ) ) {
 513                                 // A known file extension for an OPC file,
 514                                 // find the proper mime type for that file extension
 515                                 $mime = $this->guessTypesForExtension( $ext );
 516                         } else {
 517                                 wfDebug( __METHOD__ . ": refusing to guess better type for $mime file, " .
 518                                         ".$ext is not a known OPC extension.\n" );
 519                                 $mime = 'application/zip';
 520                         }
 521                 }
 522
 523                 if ( isset( $this->mMimeTypeAliases[$mime] ) ) {
 524                         $mime = $this->mMimeTypeAliases[$mime];
 525                 }
 526
 527                 wfDebug( __METHOD__ . ": improved mime type for .$ext: $mime\n" );
 528                 return $mime;
 529         }
 530
 531         /**
 532          * Mime type detection. This uses detectMimeType to detect the mime type
 533          * of the file, but applies additional checks to determine some well known
 534          * file formats that may be missed or misinterpreted by the default mime
 535          * detection (namely XML based formats like XHTML or SVG, as well as ZIP
 536          * based formats like OPC/ODF files).
 537          *
 538          * @param string $file The file to check
 539          * @param string|bool $ext The file extension, or true (default) to extract it from the filename.
 540          *   Set it to false to ignore the extension. DEPRECATED! Set to false, use
 541          *   improveTypeFromExtension($mime, $ext) later to improve mime type.
 542          *
 543          * @return string The mime type of $file
 544          */
 545         public function guessMimeType( $file, $ext = true ) {
 546                 if ( $ext ) { // TODO: make $ext default to false. Or better, remove it.
 547                         wfDebug( __METHOD__ . ": WARNING: use of the \$ext parameter is deprecated. " .
 548                                 "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 549                 }
 550
 551                 $mime = $this->doGuessMimeType( $file, $ext );
 552
 553                 if ( !$mime ) {
 554                         wfDebug( __METHOD__ . ": internal type detection failed for $file (.$ext)...\n" );
 555                         $mime = $this->detectMimeType( $file, $ext );
 556                 }
 557
 558                 if ( isset( $this->mMimeTypeAliases[$mime] ) ) {
 559                         $mime = $this->mMimeTypeAliases[$mime];
 560                 }
 561
 562                 wfDebug( __METHOD__ . ": guessed mime type of $file: $mime\n" );
 563                 return $mime;
 564         }
 565
 566         /**
 567          * Guess the mime type from the file contents.
 568          *
 569          * @param string $file
 570          * @param mixed $ext
 571          * @return bool|string
 572          */
 573         private function doGuessMimeType( $file, $ext ) { // TODO: remove $ext param
 574                 // Read a chunk of the file
 575                 wfSuppressWarnings();
 576                 // @todo FIXME: Shouldn't this be rb?
 577                 $f = fopen( $file, 'rt' );
 578                 wfRestoreWarnings();
 579
 580                 if ( !$f ) {
 581                         return 'unknown/unknown';
 582                 }
 583                 $head = fread( $f, 1024 );
 584                 fseek( $f, -65558, SEEK_END );
 585                 $tail = fread( $f, 65558 ); // 65558 = maximum size of a zip EOCDR
 586                 fclose( $f );
 587
 588                 wfDebug( __METHOD__ . ": analyzing head and tail of $file for magic numbers.\n" );
 589
 590                 // Hardcode a few magic number checks...
 591                 $headers = array(
 592                         // Multimedia...
 593                         'MThd'             => 'audio/midi',
 594                         'OggS'             => 'application/ogg',
 595
 596                         // Image formats...
 597                         // Note that WMF may have a bare header, no magic number.
 598                         "\x01\x00\x09\x00" => 'application/x-msmetafile', // Possibly prone to false positives?
 599                         "\xd7\xcd\xc6\x9a" => 'application/x-msmetafile',
 600                         '%PDF'             => 'application/pdf',
 601                         'gimp xcf'         => 'image/x-xcf',
 602
 603                         // Some forbidden fruit...
 604                         'MZ'               => 'application/octet-stream', // DOS/Windows executable
 605                         "\xca\xfe\xba\xbe" => 'application/octet-stream', // Mach-O binary
 606                         "\x7fELF"          => 'application/octet-stream', // ELF binary
 607                 );
 608
 609                 foreach ( $headers as $magic => $candidate ) {
 610                         if ( strncmp( $head, $magic, strlen( $magic ) ) == 0 ) {
 611                                 wfDebug( __METHOD__ . ": magic header in $file recognized as $candidate\n" );
 612                                 return $candidate;
 613                         }
 614                 }
 615
 616                 /* Look for WebM and Matroska files */
 617                 if ( strncmp( $head, pack( "C4", 0x1a, 0x45, 0xdf, 0xa3 ), 4 ) == 0 ) {
 618                         $doctype = strpos( $head, "\x42\x82" );
 619                         if ( $doctype ) {
 620                                 // Next byte is datasize, then data (sizes larger than 1 byte are very stupid muxers)
 621                                 $data = substr( $head, $doctype + 3, 8 );
 622                                 if ( strncmp( $data, "matroska", 8 ) == 0 ) {
 623                                         wfDebug( __METHOD__ . ": recognized file as video/x-matroska\n" );
 624                                         return "video/x-matroska";
 625                                 } elseif ( strncmp( $data, "webm", 4 ) == 0 ) {
 626                                         wfDebug( __METHOD__ . ": recognized file as video/webm\n" );
 627                                         return "video/webm";
 628                                 }
 629                         }
 630                         wfDebug( __METHOD__ . ": unknown EBML file\n" );
 631                         return "unknown/unknown";
 632                 }
 633
 634                 /* Look for WebP */
 635                 if ( strncmp( $head, "RIFF", 4 ) == 0 && strncmp( substr( $head, 8, 8 ), "WEBPVP8 ", 8 ) == 0 ) {
 636                         wfDebug( __METHOD__ . ": recognized file as image/webp\n" );
 637                         return "image/webp";
 638                 }
 639
 640                 /**
 641                  * Look for PHP.  Check for this before HTML/XML...  Warning: this is a
 642                  * heuristic, and won't match a file with a lot of non-PHP before.  It
 643                  * will also match text files which could be PHP. :)
 644                  *
 645                  * @todo FIXME: For this reason, the check is probably useless -- an attacker
 646                  * could almost certainly just pad the file with a lot of nonsense to
 647                  * circumvent the check in any case where it would be a security
 648                  * problem.  On the other hand, it causes harmful false positives (bug
 649                  * 16583).  The heuristic has been cut down to exclude three-character
 650                  * strings like "<? ", but should it be axed completely?
 651                  */
 652                 if ( ( strpos( $head, '<?php' ) !== false ) ||
 653                         ( strpos( $head, "<\x00?\x00p\x00h\x00p" ) !== false ) ||
 654                         ( strpos( $head, "<\x00?\x00 " ) !== false ) ||
 655                         ( strpos( $head, "<\x00?\x00\n" ) !== false ) ||
 656                         ( strpos( $head, "<\x00?\x00\t" ) !== false ) ||
 657                         ( strpos( $head, "<\x00?\x00=" ) !== false ) ) {
 658
 659                         wfDebug( __METHOD__ . ": recognized $file as application/x-php\n" );
 660                         return 'application/x-php';
 661                 }
 662
 663                 /**
 664                  * look for XML formats (XHTML and SVG)
 665                  */
 666                 $xml = new XmlTypeCheck( $file );
 667                 if ( $xml->wellFormed ) {
 668                         global $wgXMLMimeTypes;
 669                         if ( isset( $wgXMLMimeTypes[$xml->getRootElement()] ) ) {
 670                                 return $wgXMLMimeTypes[$xml->getRootElement()];
 671                         } else {
 672                                 return 'application/xml';
 673                         }
 674                 }
 675
 676                 /**
 677                  * look for shell scripts
 678                  */
 679                 $script_type = null;
 680
 681                 # detect by shebang
 682                 if ( substr( $head, 0, 2 ) == "#!" ) {
 683                         $script_type = "ASCII";
 684                 } elseif ( substr( $head, 0, 5 ) == "\xef\xbb\xbf#!" ) {
 685                         $script_type = "UTF-8";
 686                 } elseif ( substr( $head, 0, 7 ) == "\xfe\xff\x00#\x00!" ) {
 687                         $script_type = "UTF-16BE";
 688                 } elseif ( substr( $head, 0, 7 ) == "\xff\xfe#\x00!" ) {
 689                         $script_type = "UTF-16LE";
 690                 }
 691
 692                 if ( $script_type ) {
 693                         if ( $script_type !== "UTF-8" && $script_type !== "ASCII" ) {
 694                                 // Quick and dirty fold down to ASCII!
 695                                 $pack = array( 'UTF-16BE' => 'n*', 'UTF-16LE' => 'v*' );
 696                                 $chars = unpack( $pack[$script_type], substr( $head, 2 ) );
 697                                 $head = '';
 698                                 foreach ( $chars as $codepoint ) {
 699                                         if ( $codepoint < 128 ) {
 700                                                 $head .= chr( $codepoint );
 701                                         } else {
 702                                                 $head .= '?';
 703                                         }
 704                                 }
 705                         }
 706
 707                         $match = array();
 708
 709                         if ( preg_match( '%/?([^\s]+/)(\w+)%', $head, $match ) ) {
 710                                 $mime = "application/x-{$match[2]}";
 711                                 wfDebug( __METHOD__ . ": shell script recognized as $mime\n" );
 712                                 return $mime;
 713                         }
 714                 }
 715
 716                 // Check for ZIP variants (before getimagesize)
 717                 if ( strpos( $tail, "PK\x05\x06" ) !== false ) {
 718                         wfDebug( __METHOD__ . ": ZIP header present in $file\n" );
 719                         return $this->detectZipType( $head, $tail, $ext );
 720                 }
 721
 722                 wfSuppressWarnings();
 723                 $gis = getimagesize( $file );
 724                 wfRestoreWarnings();
 725
 726                 if ( $gis && isset( $gis['mime'] ) ) {
 727                         $mime = $gis['mime'];
 728                         wfDebug( __METHOD__ . ": getimagesize detected $file as $mime\n" );
 729                         return $mime;
 730                 }
 731
 732                 // Also test DjVu
 733                 $deja = new DjVuImage( $file );
 734                 if ( $deja->isValid() ) {
 735                         wfDebug( __METHOD__ . ": detected $file as image/vnd.djvu\n" );
 736                         return 'image/vnd.djvu';
 737                 }
 738
 739                 return false;
 740         }
 741
 742         /**
 743          * Detect application-specific file type of a given ZIP file from its
 744          * header data.  Currently works for OpenDocument and OpenXML types...
 745          * If can't tell, returns 'application/zip'.
 746          *
 747          * @param string $header Some reasonably-sized chunk of file header
 748          * @param string|null $tail The tail of the file
 749          * @param string|bool $ext The file extension, or true to extract it from the filename.
 750          *   Set it to false (default) to ignore the extension. DEPRECATED! Set to false,
 751          *   use improveTypeFromExtension($mime, $ext) later to improve mime type.
 752          *
 753          * @return string
 754          */
 755         function detectZipType( $header, $tail = null, $ext = false ) {
 756                 if ( $ext ) { # TODO: remove $ext param
 757                         wfDebug( __METHOD__ . ": WARNING: use of the \$ext parameter is deprecated. " .
 758                                 "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 759                 }
 760
 761                 $mime = 'application/zip';
 762                 $opendocTypes = array(
 763                         'chart-template',
 764                         'chart',
 765                         'formula-template',
 766                         'formula',
 767                         'graphics-template',
 768                         'graphics',
 769                         'image-template',
 770                         'image',
 771                         'presentation-template',
 772                         'presentation',
 773                         'spreadsheet-template',
 774                         'spreadsheet',
 775                         'text-template',
 776                         'text-master',
 777                         'text-web',
 778                         'text' );
 779
 780                 // http://lists.oasis-open.org/archives/office/200505/msg00006.html
 781                 $types = '(?:' . implode( '|', $opendocTypes ) . ')';
 782                 $opendocRegex = "/^mimetype(application\/vnd\.oasis\.opendocument\.$types)/";
 783
 784                 $openxmlRegex = "/^\[Content_Types\].xml/";
 785
 786                 if ( preg_match( $opendocRegex, substr( $header, 30 ), $matches ) ) {
 787                         $mime = $matches[1];
 788                         wfDebug( __METHOD__ . ": detected $mime from ZIP archive\n" );
 789                 } elseif ( preg_match( $openxmlRegex, substr( $header, 30 ) ) ) {
 790                         $mime = "application/x-opc+zip";
 791                         # TODO: remove the block below, as soon as improveTypeFromExtension is used everywhere
 792                         if ( $ext !== true && $ext !== false ) {
 793                                 /** This is the mode used by getPropsFromPath
 794                                  * These mime's are stored in the database, where we don't really want
 795                                  * x-opc+zip, because we use it only for internal purposes
 796                                  */
 797                                 if ( $this->isMatchingExtension( $ext, $mime ) ) {
 798                                         /* A known file extension for an OPC file,
 799                                          * find the proper mime type for that file extension
 800                                          */
 801                                         $mime = $this->guessTypesForExtension( $ext );
 802                                 } else {
 803                                         $mime = "application/zip";
 804                                 }
 805                         }
 806                         wfDebug( __METHOD__ . ": detected an Open Packaging Conventions archive: $mime\n" );
 807                 } elseif ( substr( $header, 0, 8 ) == "\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" &&
 808                                 ( $headerpos = strpos( $tail, "PK\x03\x04" ) ) !== false &&
 809                                 preg_match( $openxmlRegex, substr( $tail, $headerpos + 30 ) ) ) {
 810                         if ( substr( $header, 512, 4 ) == "\xEC\xA5\xC1\x00" ) {
 811                                 $mime = "application/msword";
 812                         }
 813                         switch ( substr( $header, 512, 6 ) ) {
 814                                 case "\xEC\xA5\xC1\x00\x0E\x00":
 815                                 case "\xEC\xA5\xC1\x00\x1C\x00":
 816                                 case "\xEC\xA5\xC1\x00\x43\x00":
 817                                         $mime = "application/vnd.ms-powerpoint";
 818                                         break;
 819                                 case "\xFD\xFF\xFF\xFF\x10\x00":
 820                                 case "\xFD\xFF\xFF\xFF\x1F\x00":
 821                                 case "\xFD\xFF\xFF\xFF\x22\x00":
 822                                 case "\xFD\xFF\xFF\xFF\x23\x00":
 823                                 case "\xFD\xFF\xFF\xFF\x28\x00":
 824                                 case "\xFD\xFF\xFF\xFF\x29\x00":
 825                                 case "\xFD\xFF\xFF\xFF\x10\x02":
 826                                 case "\xFD\xFF\xFF\xFF\x1F\x02":
 827                                 case "\xFD\xFF\xFF\xFF\x22\x02":
 828                                 case "\xFD\xFF\xFF\xFF\x23\x02":
 829                                 case "\xFD\xFF\xFF\xFF\x28\x02":
 830                                 case "\xFD\xFF\xFF\xFF\x29\x02":
 831                                         $mime = "application/vnd.msexcel";
 832                                         break;
 833                         }
 834
 835                         wfDebug( __METHOD__ . ": detected a MS Office document with OPC trailer\n" );
 836                 } else {
 837                         wfDebug( __METHOD__ . ": unable to identify type of ZIP archive\n" );
 838                 }
 839                 return $mime;
 840         }
 841
 842         /**
 843          * Internal mime type detection. Detection is done using an external
 844          * program, if $wgMimeDetectorCommand is set. Otherwise, the fileinfo
 845          * extension and mime_content_type are tried (in this order), if they
 846          * are available. If the detections fails and $ext is not false, the mime
 847          * type is guessed from the file extension, using guessTypesForExtension.
 848          *
 849          * If the mime type is still unknown, getimagesize is used to detect the
 850          * mime type if the file is an image. If no mime type can be determined,
 851          * this function returns 'unknown/unknown'.
 852          *
 853          * @param string $file The file to check
 854          * @param string|bool $ext The file extension, or true (default) to extract it from the filename.
 855          *   Set it to false to ignore the extension. DEPRECATED! Set to false, use
 856          *   improveTypeFromExtension($mime, $ext) later to improve mime type.
 857          *
 858          * @return string The mime type of $file
 859          */
 860         private function detectMimeType( $file, $ext = true ) {
 861                 global $wgMimeDetectorCommand;
 862
 863                 /** @todo Make $ext default to false. Or better, remove it. */
 864                 if ( $ext ) {
 865                         wfDebug( __METHOD__ . ": WARNING: use of the \$ext parameter is deprecated. "
 866                                 . "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 867                 }
 868
 869                 $m = null;
 870                 if ( $wgMimeDetectorCommand ) {
 871                         $args = wfEscapeShellArg( $file );
 872                         $m = wfShellExec( "$wgMimeDetectorCommand $args" );
 873                 } elseif ( function_exists( "finfo_open" ) && function_exists( "finfo_file" ) ) {
 874
 875                         # This required the fileinfo extension by PECL,
 876                         # see http://pecl.php.net/package/fileinfo
 877                         # This must be compiled into PHP
 878                         #
 879                         # finfo is the official replacement for the deprecated
 880                         # mime_content_type function, see below.
 881                         #
 882                         # If you may need to load the fileinfo extension at runtime, set
 883                         # $wgLoadFileinfoExtension in LocalSettings.php
 884
 885                         $mime_magic_resource = finfo_open( FILEINFO_MIME ); /* return mime type ala mimetype extension */
 886
 887                         if ( $mime_magic_resource ) {
 888                                 $m = finfo_file( $mime_magic_resource, $file );
 889                                 finfo_close( $mime_magic_resource );
 890                         } else {
 891                                 wfDebug( __METHOD__ . ": finfo_open failed on " . FILEINFO_MIME . "!\n" );
 892                         }
 893                 } elseif ( function_exists( "mime_content_type" ) ) {
 894
 895                         # NOTE: this function is available since PHP 4.3.0, but only if
 896                         # PHP was compiled with --with-mime-magic or, before 4.3.2, with
 897                         # --enable-mime-magic.
 898                         #
 899                         # On Windows, you must set mime_magic.magicfile in php.ini to point
 900                         # to the mime.magic file bundled with PHP; sometimes, this may even
 901                         # be needed under *nix.
 902                         #
 903                         # Also note that this has been DEPRECATED in favor of the fileinfo
 904                         # extension by PECL, see above.
 905                         # See http://www.php.net/manual/en/ref.mime-magic.php for details.
 906
 907                         $m = mime_content_type( $file );
 908                 } else {
 909                         wfDebug( __METHOD__ . ": no magic mime detector found!\n" );
 910                 }
 911
 912                 if ( $m ) {
 913                         # normalize
 914                         $m = preg_replace( '![;, ].*$!', '', $m ); #strip charset, etc
 915                         $m = trim( $m );
 916                         $m = strtolower( $m );
 917
 918                         if ( strpos( $m, 'unknown' ) !== false ) {
 919                                 $m = null;
 920                         } else {
 921                                 wfDebug( __METHOD__ . ": magic mime type of $file: $m\n" );
 922                                 return $m;
 923                         }
 924                 }
 925
 926                 // If desired, look at extension as a fallback.
 927                 if ( $ext === true ) {
 928                         $i = strrpos( $file, '.' );
 929                         $ext = strtolower( $i ? substr( $file, $i + 1 ) : '' );
 930                 }
 931                 if ( $ext ) {
 932                         if ( $this->isRecognizableExtension( $ext ) ) {
 933                                 wfDebug( __METHOD__ . ": refusing to guess mime type for .$ext file, "
 934                                         . "we should have recognized it\n" );
 935                         } else {
 936                                 $m = $this->guessTypesForExtension( $ext );
 937                                 if ( $m ) {
 938                                         wfDebug( __METHOD__ . ": extension mime type of $file: $m\n" );
 939                                         return $m;
 940                                 }
 941                         }
 942                 }
 943
 944                 // Unknown type
 945                 wfDebug( __METHOD__ . ": failed to guess mime type for $file!\n" );
 946                 return 'unknown/unknown';
 947         }
 948
 949         /**
 950          * Determine the media type code for a file, using its mime type, name and
 951          * possibly its contents.
 952          *
 953          * This function relies on the findMediaType(), mapping extensions and mime
 954          * types to media types.
 955          *
 956          * @todo analyse file if need be
 957          * @todo look at multiple extension, separately and together.
 958          *
 959          * @param string $path Full path to the image file, in case we have to look at the contents
 960          *        (if null, only the mime type is used to determine the media type code).
 961          * @param string $mime Mime type. If null it will be guessed using guessMimeType.
 962          *
 963          * @return string A value to be used with the MEDIATYPE_xxx constants.
 964          */
 965         function getMediaType( $path = null, $mime = null ) {
 966                 if ( !$mime && !$path ) {
 967                         return MEDIATYPE_UNKNOWN;
 968                 }
 969
 970                 // If mime type is unknown, guess it
 971                 if ( !$mime ) {
 972                         $mime = $this->guessMimeType( $path, false );
 973                 }
 974
 975                 // Special code for ogg - detect if it's video (theora),
 976                 // else label it as sound.
 977                 if ( $mime == 'application/ogg' && file_exists( $path ) ) {
 978
 979                         // Read a chunk of the file
 980                         $f = fopen( $path, "rt" );
 981                         if ( !$f ) {
 982                                 return MEDIATYPE_UNKNOWN;
 983                         }
 984                         $head = fread( $f, 256 );
 985                         fclose( $f );
 986
 987                         $head = strtolower( $head );
 988
 989                         // This is an UGLY HACK, file should be parsed correctly
 990                         if ( strpos( $head, 'theora' ) !== false ) {
 991                                 return MEDIATYPE_VIDEO;
 992                         } elseif ( strpos( $head, 'vorbis' ) !== false ) {
 993                                 return MEDIATYPE_AUDIO;
 994                         } elseif ( strpos( $head, 'flac' ) !== false ) {
 995                                 return MEDIATYPE_AUDIO;
 996                         } elseif ( strpos( $head, 'speex' ) !== false ) {
 997                                 return MEDIATYPE_AUDIO;
 998                         } else {
 999                                 return MEDIATYPE_MULTIMEDIA;
1000                         }
1001                 }
1002
1003                 // Check for entry for full mime type
1004                 if ( $mime ) {
1005                         $type = $this->findMediaType( $mime );
1006                         if ( $type !== MEDIATYPE_UNKNOWN ) {
1007                                 return $type;
1008                         }
1009                 }
1010
1011                 // Check for entry for file extension
1012                 if ( $path ) {
1013                         $i = strrpos( $path, '.' );
1014                         $e = strtolower( $i ? substr( $path, $i + 1 ) : '' );
1015
1016                         // TODO: look at multi-extension if this fails, parse from full path
1017                         $type = $this->findMediaType( '.' . $e );
1018                         if ( $type !== MEDIATYPE_UNKNOWN ) {
1019                                 return $type;
1020                         }
1021                 }
1022
1023                 // Check major mime type
1024                 if ( $mime ) {
1025                         $i = strpos( $mime, '/' );
1026                         if ( $i !== false ) {
1027                                 $major = substr( $mime, 0, $i );
1028                                 $type = $this->findMediaType( $major );
1029                                 if ( $type !== MEDIATYPE_UNKNOWN ) {
1030                                         return $type;
1031                                 }
1032                         }
1033                 }
1034
1035                 if ( !$type ) {
1036                         $type = MEDIATYPE_UNKNOWN;
1037                 }
1038
1039                 return $type;
1040         }
1041
1042         /**
1043          * Returns a media code matching the given mime type or file extension.
1044          * File extensions are represented by a string starting with a dot (.) to
1045          * distinguish them from mime types.
1046          *
1047          * This function relies on the mapping defined by $this->mMediaTypes
1048          * @access private
1049          * @param string $extMime
1050          * @return int|string
1051          */
1052         function findMediaType( $extMime ) {
1053                 if ( strpos( $extMime, '.' ) === 0 ) {
1054                         // If it's an extension, look up the mime types
1055                         $m = $this->getTypesForExtension( substr( $extMime, 1 ) );
1056                         if ( !$m ) {
1057                                 return MEDIATYPE_UNKNOWN;
1058                         }
1059
1060                         $m = explode( ' ', $m );
1061                 } else {
1062                         // Normalize mime type
1063                         if ( isset( $this->mMimeTypeAliases[$extMime] ) ) {
1064                                 $extMime = $this->mMimeTypeAliases[$extMime];
1065                         }
1066
1067                         $m = array( $extMime );
1068                 }
1069
1070                 foreach ( $m as $mime ) {
1071                         foreach ( $this->mMediaTypes as $type => $codes ) {
1072                                 if ( in_array( $mime, $codes, true ) ) {
1073                                         return $type;
1074                                 }
1075                         }
1076                 }
1077
1078                 return MEDIATYPE_UNKNOWN;
1079         }
1080
1081         /**
1082          * Get the MIME types that various versions of Internet Explorer would
1083          * detect from a chunk of the content.
1084          *
1085          * @param string $fileName The file name (unused at present)
1086          * @param string $chunk The first 256 bytes of the file
1087          * @param string $proposed The MIME type proposed by the server
1088          * @return array
1089          */
1090         public function getIEMimeTypes( $fileName, $chunk, $proposed ) {
1091                 $ca = $this->getIEContentAnalyzer();
1092                 return $ca->getRealMimesFromData( $fileName, $chunk, $proposed );
1093         }
1094
1095         /**
1096          * Get a cached instance of IEContentAnalyzer
1097          *
1098          * @return IEContentAnalyzer
1099          */
1100         protected function getIEContentAnalyzer() {
1101                 if ( is_null( $this->mIEAnalyzer ) ) {
1102                         $this->mIEAnalyzer = new IEContentAnalyzer;
1103                 }
1104                 return $this->mIEAnalyzer;
1105         }
1106 }