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          * @param string $mime The mime type, typically guessed from a file's content.
 489          * @param string $ext The file extension, as taken from the file name
 490          *
 491          * @return string The mime type
 492          */
 493         public function improveTypeFromExtension( $mime, $ext ) {
 494                 if ( $mime === 'unknown/unknown' ) {
 495                         if ( $this->isRecognizableExtension( $ext ) ) {
 496                                 wfDebug( __METHOD__ . ': refusing to guess mime type for .' .
 497                                         "$ext file, we should have recognized it\n" );
 498                         } else {
 499                                 // Not something we can detect, so simply
 500                                 // trust the file extension
 501                                 $mime = $this->guessTypesForExtension( $ext );
 502                         }
 503                 } elseif ( $mime === 'application/x-opc+zip' ) {
 504                         if ( $this->isMatchingExtension( $ext, $mime ) ) {
 505                                 // A known file extension for an OPC file,
 506                                 // find the proper mime type for that file extension
 507                                 $mime = $this->guessTypesForExtension( $ext );
 508                         } else {
 509                                 wfDebug( __METHOD__ . ": refusing to guess better type for $mime file, " .
 510                                         ".$ext is not a known OPC extension.\n" );
 511                                 $mime = 'application/zip';
 512                         }
 513                 } elseif ( $mime === 'text/plain' && $this->findMediaType( ".$ext" ) === MEDIATYPE_TEXT ) {
 514                         // Textual types are sometimes not recognized properly.
 515                         // If detected as text/plain, and has an extension which is textual
 516                         // improve to the extension's type. For example, csv and json are often
 517                         // misdetected as text/plain.
 518                         $mime = $this->guessTypesForExtension( $ext );
 519                 }
 520
 521                 if ( isset( $this->mMimeTypeAliases[$mime] ) ) {
 522                         $mime = $this->mMimeTypeAliases[$mime];
 523                 }
 524
 525                 wfDebug( __METHOD__ . ": improved mime type for .$ext: $mime\n" );
 526                 return $mime;
 527         }
 528
 529         /**
 530          * Mime type detection. This uses detectMimeType to detect the mime type
 531          * of the file, but applies additional checks to determine some well known
 532          * file formats that may be missed or misinterpreted by the default mime
 533          * detection (namely XML based formats like XHTML or SVG, as well as ZIP
 534          * based formats like OPC/ODF files).
 535          *
 536          * @param string $file The file to check
 537          * @param string|bool $ext The file extension, or true (default) to extract it from the filename.
 538          *   Set it to false to ignore the extension. DEPRECATED! Set to false, use
 539          *   improveTypeFromExtension($mime, $ext) later to improve mime type.
 540          *
 541          * @return string The mime type of $file
 542          */
 543         public function guessMimeType( $file, $ext = true ) {
 544                 if ( $ext ) { // TODO: make $ext default to false. Or better, remove it.
 545                         wfDebug( __METHOD__ . ": WARNING: use of the \$ext parameter is deprecated. " .
 546                                 "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 547                 }
 548
 549                 $mime = $this->doGuessMimeType( $file, $ext );
 550
 551                 if ( !$mime ) {
 552                         wfDebug( __METHOD__ . ": internal type detection failed for $file (.$ext)...\n" );
 553                         $mime = $this->detectMimeType( $file, $ext );
 554                 }
 555
 556                 if ( isset( $this->mMimeTypeAliases[$mime] ) ) {
 557                         $mime = $this->mMimeTypeAliases[$mime];
 558                 }
 559
 560                 wfDebug( __METHOD__ . ": guessed mime type of $file: $mime\n" );
 561                 return $mime;
 562         }
 563
 564         /**
 565          * Guess the mime type from the file contents.
 566          *
 567          * @param string $file
 568          * @param mixed $ext
 569          * @return bool|string
 570          * @throws MWException
 571          */
 572         private function doGuessMimeType( $file, $ext ) { // TODO: remove $ext param
 573                 // Read a chunk of the file
 574                 wfSuppressWarnings();
 575                 $f = fopen( $file, 'rb' );
 576                 wfRestoreWarnings();
 577
 578                 if ( !$f ) {
 579                         return 'unknown/unknown';
 580                 }
 581
 582                 $fsize = filesize( $file );
 583                 if ( $fsize === false ) {
 584                         return 'unknown/unknown';
 585                 }
 586
 587                 $head = fread( $f, 1024 );
 588                 $tailLength = min( 65558, $fsize ); // 65558 = maximum size of a zip EOCDR
 589                 if ( fseek( $f, -1 * $tailLength, SEEK_END ) === -1 ) {
 590                         throw new MWException(
 591                                 "Seeking $tailLength bytes from EOF failed in " . __METHOD__ );
 592                 }
 593                 $tail = fread( $f, $tailLength );
 594                 fclose( $f );
 595
 596                 wfDebug( __METHOD__ . ": analyzing head and tail of $file for magic numbers.\n" );
 597
 598                 // Hardcode a few magic number checks...
 599                 $headers = array(
 600                         // Multimedia...
 601                         'MThd'             => 'audio/midi',
 602                         'OggS'             => 'application/ogg',
 603
 604                         // Image formats...
 605                         // Note that WMF may have a bare header, no magic number.
 606                         "\x01\x00\x09\x00" => 'application/x-msmetafile', // Possibly prone to false positives?
 607                         "\xd7\xcd\xc6\x9a" => 'application/x-msmetafile',
 608                         '%PDF'             => 'application/pdf',
 609                         'gimp xcf'         => 'image/x-xcf',
 610
 611                         // Some forbidden fruit...
 612                         'MZ'               => 'application/octet-stream', // DOS/Windows executable
 613                         "\xca\xfe\xba\xbe" => 'application/octet-stream', // Mach-O binary
 614                         "\x7fELF"          => 'application/octet-stream', // ELF binary
 615                 );
 616
 617                 foreach ( $headers as $magic => $candidate ) {
 618                         if ( strncmp( $head, $magic, strlen( $magic ) ) == 0 ) {
 619                                 wfDebug( __METHOD__ . ": magic header in $file recognized as $candidate\n" );
 620                                 return $candidate;
 621                         }
 622                 }
 623
 624                 /* Look for WebM and Matroska files */
 625                 if ( strncmp( $head, pack( "C4", 0x1a, 0x45, 0xdf, 0xa3 ), 4 ) == 0 ) {
 626                         $doctype = strpos( $head, "\x42\x82" );
 627                         if ( $doctype ) {
 628                                 // Next byte is datasize, then data (sizes larger than 1 byte are very stupid muxers)
 629                                 $data = substr( $head, $doctype + 3, 8 );
 630                                 if ( strncmp( $data, "matroska", 8 ) == 0 ) {
 631                                         wfDebug( __METHOD__ . ": recognized file as video/x-matroska\n" );
 632                                         return "video/x-matroska";
 633                                 } elseif ( strncmp( $data, "webm", 4 ) == 0 ) {
 634                                         wfDebug( __METHOD__ . ": recognized file as video/webm\n" );
 635                                         return "video/webm";
 636                                 }
 637                         }
 638                         wfDebug( __METHOD__ . ": unknown EBML file\n" );
 639                         return "unknown/unknown";
 640                 }
 641
 642                 /* Look for WebP */
 643                 if ( strncmp( $head, "RIFF", 4 ) == 0 && strncmp( substr( $head, 8, 8 ), "WEBPVP8 ", 8 ) == 0 ) {
 644                         wfDebug( __METHOD__ . ": recognized file as image/webp\n" );
 645                         return "image/webp";
 646                 }
 647
 648                 /**
 649                  * Look for PHP.  Check for this before HTML/XML...  Warning: this is a
 650                  * heuristic, and won't match a file with a lot of non-PHP before.  It
 651                  * will also match text files which could be PHP. :)
 652                  *
 653                  * @todo FIXME: For this reason, the check is probably useless -- an attacker
 654                  * could almost certainly just pad the file with a lot of nonsense to
 655                  * circumvent the check in any case where it would be a security
 656                  * problem.  On the other hand, it causes harmful false positives (bug
 657                  * 16583).  The heuristic has been cut down to exclude three-character
 658                  * strings like "<? ", but should it be axed completely?
 659                  */
 660                 if ( ( strpos( $head, '<?php' ) !== false ) ||
 661                         ( strpos( $head, "<\x00?\x00p\x00h\x00p" ) !== false ) ||
 662                         ( strpos( $head, "<\x00?\x00 " ) !== false ) ||
 663                         ( strpos( $head, "<\x00?\x00\n" ) !== false ) ||
 664                         ( strpos( $head, "<\x00?\x00\t" ) !== false ) ||
 665                         ( strpos( $head, "<\x00?\x00=" ) !== false ) ) {
 666
 667                         wfDebug( __METHOD__ . ": recognized $file as application/x-php\n" );
 668                         return 'application/x-php';
 669                 }
 670
 671                 /**
 672                  * look for XML formats (XHTML and SVG)
 673                  */
 674                 $xml = new XmlTypeCheck( $file );
 675                 if ( $xml->wellFormed ) {
 676                         global $wgXMLMimeTypes;
 677                         if ( isset( $wgXMLMimeTypes[$xml->getRootElement()] ) ) {
 678                                 return $wgXMLMimeTypes[$xml->getRootElement()];
 679                         } else {
 680                                 return 'application/xml';
 681                         }
 682                 }
 683
 684                 /**
 685                  * look for shell scripts
 686                  */
 687                 $script_type = null;
 688
 689                 # detect by shebang
 690                 if ( substr( $head, 0, 2 ) == "#!" ) {
 691                         $script_type = "ASCII";
 692                 } elseif ( substr( $head, 0, 5 ) == "\xef\xbb\xbf#!" ) {
 693                         $script_type = "UTF-8";
 694                 } elseif ( substr( $head, 0, 7 ) == "\xfe\xff\x00#\x00!" ) {
 695                         $script_type = "UTF-16BE";
 696                 } elseif ( substr( $head, 0, 7 ) == "\xff\xfe#\x00!" ) {
 697                         $script_type = "UTF-16LE";
 698                 }
 699
 700                 if ( $script_type ) {
 701                         if ( $script_type !== "UTF-8" && $script_type !== "ASCII" ) {
 702                                 // Quick and dirty fold down to ASCII!
 703                                 $pack = array( 'UTF-16BE' => 'n*', 'UTF-16LE' => 'v*' );
 704                                 $chars = unpack( $pack[$script_type], substr( $head, 2 ) );
 705                                 $head = '';
 706                                 foreach ( $chars as $codepoint ) {
 707                                         if ( $codepoint < 128 ) {
 708                                                 $head .= chr( $codepoint );
 709                                         } else {
 710                                                 $head .= '?';
 711                                         }
 712                                 }
 713                         }
 714
 715                         $match = array();
 716
 717                         if ( preg_match( '%/?([^\s]+/)(\w+)%', $head, $match ) ) {
 718                                 $mime = "application/x-{$match[2]}";
 719                                 wfDebug( __METHOD__ . ": shell script recognized as $mime\n" );
 720                                 return $mime;
 721                         }
 722                 }
 723
 724                 // Check for ZIP variants (before getimagesize)
 725                 if ( strpos( $tail, "PK\x05\x06" ) !== false ) {
 726                         wfDebug( __METHOD__ . ": ZIP header present in $file\n" );
 727                         return $this->detectZipType( $head, $tail, $ext );
 728                 }
 729
 730                 wfSuppressWarnings();
 731                 $gis = getimagesize( $file );
 732                 wfRestoreWarnings();
 733
 734                 if ( $gis && isset( $gis['mime'] ) ) {
 735                         $mime = $gis['mime'];
 736                         wfDebug( __METHOD__ . ": getimagesize detected $file as $mime\n" );
 737                         return $mime;
 738                 }
 739
 740                 // Also test DjVu
 741                 $deja = new DjVuImage( $file );
 742                 if ( $deja->isValid() ) {
 743                         wfDebug( __METHOD__ . ": detected $file as image/vnd.djvu\n" );
 744                         return 'image/vnd.djvu';
 745                 }
 746
 747                 return false;
 748         }
 749
 750         /**
 751          * Detect application-specific file type of a given ZIP file from its
 752          * header data.  Currently works for OpenDocument and OpenXML types...
 753          * If can't tell, returns 'application/zip'.
 754          *
 755          * @param string $header Some reasonably-sized chunk of file header
 756          * @param string|null $tail The tail of the file
 757          * @param string|bool $ext The file extension, or true to extract it from the filename.
 758          *   Set it to false (default) to ignore the extension. DEPRECATED! Set to false,
 759          *   use improveTypeFromExtension($mime, $ext) later to improve mime type.
 760          *
 761          * @return string
 762          */
 763         function detectZipType( $header, $tail = null, $ext = false ) {
 764                 if ( $ext ) { # TODO: remove $ext param
 765                         wfDebug( __METHOD__ . ": WARNING: use of the \$ext parameter is deprecated. " .
 766                                 "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 767                 }
 768
 769                 $mime = 'application/zip';
 770                 $opendocTypes = array(
 771                         'chart-template',
 772                         'chart',
 773                         'formula-template',
 774                         'formula',
 775                         'graphics-template',
 776                         'graphics',
 777                         'image-template',
 778                         'image',
 779                         'presentation-template',
 780                         'presentation',
 781                         'spreadsheet-template',
 782                         'spreadsheet',
 783                         'text-template',
 784                         'text-master',
 785                         'text-web',
 786                         'text' );
 787
 788                 // http://lists.oasis-open.org/archives/office/200505/msg00006.html
 789                 $types = '(?:' . implode( '|', $opendocTypes ) . ')';
 790                 $opendocRegex = "/^mimetype(application\/vnd\.oasis\.opendocument\.$types)/";
 791
 792                 $openxmlRegex = "/^\[Content_Types\].xml/";
 793
 794                 if ( preg_match( $opendocRegex, substr( $header, 30 ), $matches ) ) {
 795                         $mime = $matches[1];
 796                         wfDebug( __METHOD__ . ": detected $mime from ZIP archive\n" );
 797                 } elseif ( preg_match( $openxmlRegex, substr( $header, 30 ) ) ) {
 798                         $mime = "application/x-opc+zip";
 799                         # TODO: remove the block below, as soon as improveTypeFromExtension is used everywhere
 800                         if ( $ext !== true && $ext !== false ) {
 801                                 /** This is the mode used by getPropsFromPath
 802                                  * These mime's are stored in the database, where we don't really want
 803                                  * x-opc+zip, because we use it only for internal purposes
 804                                  */
 805                                 if ( $this->isMatchingExtension( $ext, $mime ) ) {
 806                                         /* A known file extension for an OPC file,
 807                                          * find the proper mime type for that file extension
 808                                          */
 809                                         $mime = $this->guessTypesForExtension( $ext );
 810                                 } else {
 811                                         $mime = "application/zip";
 812                                 }
 813                         }
 814                         wfDebug( __METHOD__ . ": detected an Open Packaging Conventions archive: $mime\n" );
 815                 } elseif ( substr( $header, 0, 8 ) == "\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" &&
 816                                 ( $headerpos = strpos( $tail, "PK\x03\x04" ) ) !== false &&
 817                                 preg_match( $openxmlRegex, substr( $tail, $headerpos + 30 ) ) ) {
 818                         if ( substr( $header, 512, 4 ) == "\xEC\xA5\xC1\x00" ) {
 819                                 $mime = "application/msword";
 820                         }
 821                         switch ( substr( $header, 512, 6 ) ) {
 822                                 case "\xEC\xA5\xC1\x00\x0E\x00":
 823                                 case "\xEC\xA5\xC1\x00\x1C\x00":
 824                                 case "\xEC\xA5\xC1\x00\x43\x00":
 825                                         $mime = "application/vnd.ms-powerpoint";
 826                                         break;
 827                                 case "\xFD\xFF\xFF\xFF\x10\x00":
 828                                 case "\xFD\xFF\xFF\xFF\x1F\x00":
 829                                 case "\xFD\xFF\xFF\xFF\x22\x00":
 830                                 case "\xFD\xFF\xFF\xFF\x23\x00":
 831                                 case "\xFD\xFF\xFF\xFF\x28\x00":
 832                                 case "\xFD\xFF\xFF\xFF\x29\x00":
 833                                 case "\xFD\xFF\xFF\xFF\x10\x02":
 834                                 case "\xFD\xFF\xFF\xFF\x1F\x02":
 835                                 case "\xFD\xFF\xFF\xFF\x22\x02":
 836                                 case "\xFD\xFF\xFF\xFF\x23\x02":
 837                                 case "\xFD\xFF\xFF\xFF\x28\x02":
 838                                 case "\xFD\xFF\xFF\xFF\x29\x02":
 839                                         $mime = "application/vnd.msexcel";
 840                                         break;
 841                         }
 842
 843                         wfDebug( __METHOD__ . ": detected a MS Office document with OPC trailer\n" );
 844                 } else {
 845                         wfDebug( __METHOD__ . ": unable to identify type of ZIP archive\n" );
 846                 }
 847                 return $mime;
 848         }
 849
 850         /**
 851          * Internal mime type detection. Detection is done using an external
 852          * program, if $wgMimeDetectorCommand is set. Otherwise, the fileinfo
 853          * extension and mime_content_type are tried (in this order), if they
 854          * are available. If the detections fails and $ext is not false, the mime
 855          * type is guessed from the file extension, using guessTypesForExtension.
 856          *
 857          * If the mime type is still unknown, getimagesize is used to detect the
 858          * mime type if the file is an image. If no mime type can be determined,
 859          * this function returns 'unknown/unknown'.
 860          *
 861          * @param string $file The file to check
 862          * @param string|bool $ext The file extension, or true (default) to extract it from the filename.
 863          *   Set it to false to ignore the extension. DEPRECATED! Set to false, use
 864          *   improveTypeFromExtension($mime, $ext) later to improve mime type.
 865          *
 866          * @return string The mime type of $file
 867          */
 868         private function detectMimeType( $file, $ext = true ) {
 869                 global $wgMimeDetectorCommand;
 870
 871                 /** @todo Make $ext default to false. Or better, remove it. */
 872                 if ( $ext ) {
 873                         wfDebug( __METHOD__ . ": WARNING: use of the \$ext parameter is deprecated. "
 874                                 . "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 875                 }
 876
 877                 $m = null;
 878                 if ( $wgMimeDetectorCommand ) {
 879                         $args = wfEscapeShellArg( $file );
 880                         $m = wfShellExec( "$wgMimeDetectorCommand $args" );
 881                 } elseif ( function_exists( "finfo_open" ) && function_exists( "finfo_file" ) ) {
 882
 883                         # This required the fileinfo extension by PECL,
 884                         # see http://pecl.php.net/package/fileinfo
 885                         # This must be compiled into PHP
 886                         #
 887                         # finfo is the official replacement for the deprecated
 888                         # mime_content_type function, see below.
 889                         #
 890                         # If you may need to load the fileinfo extension at runtime, set
 891                         # $wgLoadFileinfoExtension in LocalSettings.php
 892
 893                         $mime_magic_resource = finfo_open( FILEINFO_MIME ); /* return mime type ala mimetype extension */
 894
 895                         if ( $mime_magic_resource ) {
 896                                 $m = finfo_file( $mime_magic_resource, $file );
 897                                 finfo_close( $mime_magic_resource );
 898                         } else {
 899                                 wfDebug( __METHOD__ . ": finfo_open failed on " . FILEINFO_MIME . "!\n" );
 900                         }
 901                 } elseif ( function_exists( "mime_content_type" ) ) {
 902
 903                         # NOTE: this function is available since PHP 4.3.0, but only if
 904                         # PHP was compiled with --with-mime-magic or, before 4.3.2, with
 905                         # --enable-mime-magic.
 906                         #
 907                         # On Windows, you must set mime_magic.magicfile in php.ini to point
 908                         # to the mime.magic file bundled with PHP; sometimes, this may even
 909                         # be needed under *nix.
 910                         #
 911                         # Also note that this has been DEPRECATED in favor of the fileinfo
 912                         # extension by PECL, see above.
 913                         # See http://www.php.net/manual/en/ref.mime-magic.php for details.
 914
 915                         $m = mime_content_type( $file );
 916                 } else {
 917                         wfDebug( __METHOD__ . ": no magic mime detector found!\n" );
 918                 }
 919
 920                 if ( $m ) {
 921                         # normalize
 922                         $m = preg_replace( '![;, ].*$!', '', $m ); #strip charset, etc
 923                         $m = trim( $m );
 924                         $m = strtolower( $m );
 925
 926                         if ( strpos( $m, 'unknown' ) !== false ) {
 927                                 $m = null;
 928                         } else {
 929                                 wfDebug( __METHOD__ . ": magic mime type of $file: $m\n" );
 930                                 return $m;
 931                         }
 932                 }
 933
 934                 // If desired, look at extension as a fallback.
 935                 if ( $ext === true ) {
 936                         $i = strrpos( $file, '.' );
 937                         $ext = strtolower( $i ? substr( $file, $i + 1 ) : '' );
 938                 }
 939                 if ( $ext ) {
 940                         if ( $this->isRecognizableExtension( $ext ) ) {
 941                                 wfDebug( __METHOD__ . ": refusing to guess mime type for .$ext file, "
 942                                         . "we should have recognized it\n" );
 943                         } else {
 944                                 $m = $this->guessTypesForExtension( $ext );
 945                                 if ( $m ) {
 946                                         wfDebug( __METHOD__ . ": extension mime type of $file: $m\n" );
 947                                         return $m;
 948                                 }
 949                         }
 950                 }
 951
 952                 // Unknown type
 953                 wfDebug( __METHOD__ . ": failed to guess mime type for $file!\n" );
 954                 return 'unknown/unknown';
 955         }
 956
 957         /**
 958          * Determine the media type code for a file, using its mime type, name and
 959          * possibly its contents.
 960          *
 961          * This function relies on the findMediaType(), mapping extensions and mime
 962          * types to media types.
 963          *
 964          * @todo analyse file if need be
 965          * @todo look at multiple extension, separately and together.
 966          *
 967          * @param string $path Full path to the image file, in case we have to look at the contents
 968          *        (if null, only the mime type is used to determine the media type code).
 969          * @param string $mime Mime type. If null it will be guessed using guessMimeType.
 970          *
 971          * @return string A value to be used with the MEDIATYPE_xxx constants.
 972          */
 973         function getMediaType( $path = null, $mime = null ) {
 974                 if ( !$mime && !$path ) {
 975                         return MEDIATYPE_UNKNOWN;
 976                 }
 977
 978                 // If mime type is unknown, guess it
 979                 if ( !$mime ) {
 980                         $mime = $this->guessMimeType( $path, false );
 981                 }
 982
 983                 // Special code for ogg - detect if it's video (theora),
 984                 // else label it as sound.
 985                 if ( $mime == 'application/ogg' && file_exists( $path ) ) {
 986
 987                         // Read a chunk of the file
 988                         $f = fopen( $path, "rt" );
 989                         if ( !$f ) {
 990                                 return MEDIATYPE_UNKNOWN;
 991                         }
 992                         $head = fread( $f, 256 );
 993                         fclose( $f );
 994
 995                         $head = str_replace( 'ffmpeg2theora', '', strtolower( $head ) );
 996
 997                         // This is an UGLY HACK, file should be parsed correctly
 998                         if ( strpos( $head, 'theora' ) !== false ) {
 999                                 return MEDIATYPE_VIDEO;
1000                         } elseif ( strpos( $head, 'vorbis' ) !== false ) {
1001                                 return MEDIATYPE_AUDIO;
1002                         } elseif ( strpos( $head, 'flac' ) !== false ) {
1003                                 return MEDIATYPE_AUDIO;
1004                         } elseif ( strpos( $head, 'speex' ) !== false ) {
1005                                 return MEDIATYPE_AUDIO;
1006                         } else {
1007                                 return MEDIATYPE_MULTIMEDIA;
1008                         }
1009                 }
1010
1011                 // Check for entry for full mime type
1012                 if ( $mime ) {
1013                         $type = $this->findMediaType( $mime );
1014                         if ( $type !== MEDIATYPE_UNKNOWN ) {
1015                                 return $type;
1016                         }
1017                 }
1018
1019                 // Check for entry for file extension
1020                 if ( $path ) {
1021                         $i = strrpos( $path, '.' );
1022                         $e = strtolower( $i ? substr( $path, $i + 1 ) : '' );
1023
1024                         // TODO: look at multi-extension if this fails, parse from full path
1025                         $type = $this->findMediaType( '.' . $e );
1026                         if ( $type !== MEDIATYPE_UNKNOWN ) {
1027                                 return $type;
1028                         }
1029                 }
1030
1031                 // Check major mime type
1032                 if ( $mime ) {
1033                         $i = strpos( $mime, '/' );
1034                         if ( $i !== false ) {
1035                                 $major = substr( $mime, 0, $i );
1036                                 $type = $this->findMediaType( $major );
1037                                 if ( $type !== MEDIATYPE_UNKNOWN ) {
1038                                         return $type;
1039                                 }
1040                         }
1041                 }
1042
1043                 if ( !$type ) {
1044                         $type = MEDIATYPE_UNKNOWN;
1045                 }
1046
1047                 return $type;
1048         }
1049
1050         /**
1051          * Returns a media code matching the given mime type or file extension.
1052          * File extensions are represented by a string starting with a dot (.) to
1053          * distinguish them from mime types.
1054          *
1055          * This function relies on the mapping defined by $this->mMediaTypes
1056          * @access private
1057          * @param string $extMime
1058          * @return int|string
1059          */
1060         function findMediaType( $extMime ) {
1061                 if ( strpos( $extMime, '.' ) === 0 ) {
1062                         // If it's an extension, look up the mime types
1063                         $m = $this->getTypesForExtension( substr( $extMime, 1 ) );
1064                         if ( !$m ) {
1065                                 return MEDIATYPE_UNKNOWN;
1066                         }
1067
1068                         $m = explode( ' ', $m );
1069                 } else {
1070                         // Normalize mime type
1071                         if ( isset( $this->mMimeTypeAliases[$extMime] ) ) {
1072                                 $extMime = $this->mMimeTypeAliases[$extMime];
1073                         }
1074
1075                         $m = array( $extMime );
1076                 }
1077
1078                 foreach ( $m as $mime ) {
1079                         foreach ( $this->mMediaTypes as $type => $codes ) {
1080                                 if ( in_array( $mime, $codes, true ) ) {
1081                                         return $type;
1082                                 }
1083                         }
1084                 }
1085
1086                 return MEDIATYPE_UNKNOWN;
1087         }
1088
1089         /**
1090          * Get the MIME types that various versions of Internet Explorer would
1091          * detect from a chunk of the content.
1092          *
1093          * @param string $fileName The file name (unused at present)
1094          * @param string $chunk The first 256 bytes of the file
1095          * @param string $proposed The MIME type proposed by the server
1096          * @return array
1097          */
1098         public function getIEMimeTypes( $fileName, $chunk, $proposed ) {
1099                 $ca = $this->getIEContentAnalyzer();
1100                 return $ca->getRealMimesFromData( $fileName, $chunk, $proposed );
1101         }
1102
1103         /**
1104          * Get a cached instance of IEContentAnalyzer
1105          *
1106          * @return IEContentAnalyzer
1107          */
1108         protected function getIEContentAnalyzer() {
1109                 if ( is_null( $this->mIEAnalyzer ) ) {
1110                         $this->mIEAnalyzer = new IEContentAnalyzer;
1111                 }
1112                 return $this->mIEAnalyzer;
1113         }
1114 }