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