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 string Extra MIME types, set for example by media handling extensions
 168          */
 169         private $mExtraTypes = '';
 170
 171         /** @var string Extra MIME info, set for example by media handling extensions
 172          */
 173         private $mExtraInfo = '';
 174
 175         /** @var MimeMagic The singleton instance
 176          */
 177         private static $instance = null;
 178
 179         /** Initializes the MimeMagic object. This is called by MimeMagic::singleton().
 180          *
 181          * This constructor parses the mime.types and mime.info files and build internal mappings.
 182          */
 183         function __construct() {
 184                 /**
 185                  *   --- load mime.types ---
 186                  */
 187
 188                 global $wgMimeTypeFile, $IP;
 189
 190                 # Allow media handling extensions adding MIME-types and MIME-info
 191                 wfRunHooks( 'MimeMagicInit', array( $this ) );
 192
 193                 $types = MM_WELL_KNOWN_MIME_TYPES;
 194
 195                 if ( $wgMimeTypeFile == 'includes/mime.types' ) {
 196                         $wgMimeTypeFile = "$IP/$wgMimeTypeFile";
 197                 }
 198
 199                 if ( $wgMimeTypeFile ) {
 200                         if ( is_file( $wgMimeTypeFile ) and is_readable( $wgMimeTypeFile ) ) {
 201                                 wfDebug( __METHOD__ . ": loading mime types from $wgMimeTypeFile\n" );
 202                                 $types .= "\n";
 203                                 $types .= file_get_contents( $wgMimeTypeFile );
 204                         } else {
 205                                 wfDebug( __METHOD__ . ": can't load mime types from $wgMimeTypeFile\n" );
 206                         }
 207                 } else {
 208                         wfDebug( __METHOD__ . ": no mime types file defined, using build-ins only.\n" );
 209                 }
 210
 211                 $types .= "\n" . $this->mExtraTypes;
 212
 213                 $types = str_replace( array( "\r\n", "\n\r", "\n\n", "\r\r", "\r" ), "\n", $types );
 214                 $types = str_replace( "\t", " ", $types );
 215
 216                 $this->mMimeToExt = array();
 217                 $this->mExtToMime = array();
 218
 219                 $lines = explode( "\n", $types );
 220                 foreach ( $lines as $s ) {
 221                         $s = trim( $s );
 222                         if ( empty( $s ) ) {
 223                                 continue;
 224                         }
 225                         if ( strpos( $s, '#' ) === 0 ) {
 226                                 continue;
 227                         }
 228
 229                         $s = strtolower( $s );
 230                         $i = strpos( $s, ' ' );
 231
 232                         if ( $i === false ) {
 233                                 continue;
 234                         }
 235
 236                         $mime = substr( $s, 0, $i );
 237                         $ext = trim( substr( $s, $i + 1 ) );
 238
 239                         if ( empty( $ext ) ) {
 240                                 continue;
 241                         }
 242
 243                         if ( !empty( $this->mMimeToExt[$mime] ) ) {
 244                                 $this->mMimeToExt[$mime] .= ' ' . $ext;
 245                         } else {
 246                                 $this->mMimeToExt[$mime] = $ext;
 247                         }
 248
 249                         $extensions = explode( ' ', $ext );
 250
 251                         foreach ( $extensions as $e ) {
 252                                 $e = trim( $e );
 253                                 if ( empty( $e ) ) {
 254                                         continue;
 255                                 }
 256
 257                                 if ( !empty( $this->mExtToMime[$e] ) ) {
 258                                         $this->mExtToMime[$e] .= ' ' . $mime;
 259                                 } else {
 260                                         $this->mExtToMime[$e] = $mime;
 261                                 }
 262                         }
 263                 }
 264
 265                 /**
 266                  *   --- load mime.info ---
 267                  */
 268
 269                 global $wgMimeInfoFile;
 270                 if ( $wgMimeInfoFile == 'includes/mime.info' ) {
 271                         $wgMimeInfoFile = "$IP/$wgMimeInfoFile";
 272                 }
 273
 274                 $info = MM_WELL_KNOWN_MIME_INFO;
 275
 276                 if ( $wgMimeInfoFile ) {
 277                         if ( is_file( $wgMimeInfoFile ) and is_readable( $wgMimeInfoFile ) ) {
 278                                 wfDebug( __METHOD__ . ": loading mime info from $wgMimeInfoFile\n" );
 279                                 $info .= "\n";
 280                                 $info .= file_get_contents( $wgMimeInfoFile );
 281                         } else {
 282                                 wfDebug( __METHOD__ . ": can't load mime info from $wgMimeInfoFile\n" );
 283                         }
 284                 } else {
 285                         wfDebug( __METHOD__ . ": no mime info file defined, using build-ins only.\n" );
 286                 }
 287
 288                 $info .= "\n" . $this->mExtraInfo;
 289
 290                 $info = str_replace( array( "\r\n", "\n\r", "\n\n", "\r\r", "\r" ), "\n", $info );
 291                 $info = str_replace( "\t", " ", $info );
 292
 293                 $this->mMimeTypeAliases = array();
 294                 $this->mMediaTypes = array();
 295
 296                 $lines = explode( "\n", $info );
 297                 foreach ( $lines as $s ) {
 298                         $s = trim( $s );
 299                         if ( empty( $s ) ) {
 300                                 continue;
 301                         }
 302                         if ( strpos( $s, '#' ) === 0 ) {
 303                                 continue;
 304                         }
 305
 306                         $s = strtolower( $s );
 307                         $i = strpos( $s, ' ' );
 308
 309                         if ( $i === false ) {
 310                                 continue;
 311                         }
 312
 313                         #print "processing MIME INFO line $s<br>";
 314
 315                         $match = array();
 316                         if ( preg_match( '!\[\s*(\w+)\s*\]!', $s, $match ) ) {
 317                                 $s = preg_replace( '!\[\s*(\w+)\s*\]!', '', $s );
 318                                 $mtype = trim( strtoupper( $match[1] ) );
 319                         } else {
 320                                 $mtype = MEDIATYPE_UNKNOWN;
 321                         }
 322
 323                         $m = explode( ' ', $s );
 324
 325                         if ( !isset( $this->mMediaTypes[$mtype] ) ) {
 326                                 $this->mMediaTypes[$mtype] = array();
 327                         }
 328
 329                         foreach ( $m as $mime ) {
 330                                 $mime = trim( $mime );
 331                                 if ( empty( $mime ) ) {
 332                                         continue;
 333                                 }
 334
 335                                 $this->mMediaTypes[$mtype][] = $mime;
 336                         }
 337
 338                         if ( count( $m ) > 1 ) {
 339                                 $main = $m[0];
 340                                 $mCount = count( $m );
 341                                 for ( $i = 1; $i < $mCount; $i += 1 ) {
 342                                         $mime = $m[$i];
 343                                         $this->mMimeTypeAliases[$mime] = $main;
 344                                 }
 345                         }
 346                 }
 347         }
 348
 349         /**
 350          * Get an instance of this class
 351          * @return MimeMagic
 352          */
 353         public static function singleton() {
 354                 if ( self::$instance === null ) {
 355                         self::$instance = new MimeMagic;
 356                 }
 357                 return self::$instance;
 358         }
 359
 360         /**
 361          * Adds to the list mapping MIME to file extensions.
 362          * As an extension author, you are encouraged to submit patches to
 363          * MediaWiki's core to add new MIME types to mime.types.
 364          * @param string $types
 365          */
 366         public function addExtraTypes( $types ) {
 367                 $this->mExtraTypes .= "\n" . $types;
 368         }
 369
 370         /**
 371          * Adds to the list mapping MIME to media type.
 372          * As an extension author, you are encouraged to submit patches to
 373          * MediaWiki's core to add new MIME info to mime.info.
 374          * @param string $info
 375          */
 376         public function addExtraInfo( $info ) {
 377                 $this->mExtraInfo .= "\n" . $info;
 378         }
 379
 380         /**
 381          * Returns a list of file extensions for a given mime type as a space
 382          * separated string or null if the mime type was unrecognized. Resolves
 383          * mime type aliases.
 384          *
 385          * @param string $mime
 386          * @return string|null
 387          */
 388         public function getExtensionsForType( $mime ) {
 389                 $mime = strtolower( $mime );
 390
 391                 // Check the mime-to-ext map
 392                 if ( isset( $this->mMimeToExt[$mime] ) ) {
 393                         return $this->mMimeToExt[$mime];
 394                 }
 395
 396                 // Resolve the mime type to the canonical type
 397                 if ( isset( $this->mMimeTypeAliases[$mime] ) ) {
 398                         $mime = $this->mMimeTypeAliases[$mime];
 399                         if ( isset( $this->mMimeToExt[$mime] ) ) {
 400                                 return $this->mMimeToExt[$mime];
 401                         }
 402                 }
 403
 404                 return null;
 405         }
 406
 407         /**
 408          * Returns a list of mime types for a given file extension as a space
 409          * separated string or null if the extension was unrecognized.
 410          *
 411          * @param string $ext
 412          * @return string|null
 413          */
 414         public function getTypesForExtension( $ext ) {
 415                 $ext = strtolower( $ext );
 416
 417                 $r = isset( $this->mExtToMime[$ext] ) ? $this->mExtToMime[$ext] : null;
 418                 return $r;
 419         }
 420
 421         /**
 422          * Returns a single mime type for a given file extension or null if unknown.
 423          * This is always the first type from the list returned by getTypesForExtension($ext).
 424          *
 425          * @param string $ext
 426          * @return string|null
 427          */
 428         public function guessTypesForExtension( $ext ) {
 429                 $m = $this->getTypesForExtension( $ext );
 430                 if ( is_null( $m ) ) {
 431                         return null;
 432                 }
 433
 434                 // TODO: Check if this is needed; strtok( $m, ' ' ) should be sufficient
 435                 $m = trim( $m );
 436                 $m = preg_replace( '/\s.*$/', '', $m );
 437
 438                 return $m;
 439         }
 440
 441         /**
 442          * Tests if the extension matches the given mime type. Returns true if a
 443          * match was found, null if the mime type is unknown, and false if the
 444          * mime type is known but no matches where found.
 445          *
 446          * @param string $extension
 447          * @param string $mime
 448          * @return bool|null
 449          */
 450         public function isMatchingExtension( $extension, $mime ) {
 451                 $ext = $this->getExtensionsForType( $mime );
 452
 453                 if ( !$ext ) {
 454                         return null; // Unknown mime type
 455                 }
 456
 457                 $ext = explode( ' ', $ext );
 458
 459                 $extension = strtolower( $extension );
 460                 return in_array( $extension, $ext );
 461         }
 462
 463         /**
 464          * Returns true if the mime type is known to represent an image format
 465          * supported by the PHP GD library.
 466          *
 467          * @param string $mime
 468          *
 469          * @return bool
 470          */
 471         public function isPHPImageType( $mime ) {
 472                 // As defined by imagegetsize and image_type_to_mime
 473                 static $types = array(
 474                         'image/gif', 'image/jpeg', 'image/png',
 475                         'image/x-bmp', 'image/xbm', 'image/tiff',
 476                         'image/jp2', 'image/jpeg2000', 'image/iff',
 477                         'image/xbm', 'image/x-xbitmap',
 478                         'image/vnd.wap.wbmp', 'image/vnd.xiff',
 479                         'image/x-photoshop',
 480                         'application/x-shockwave-flash',
 481                 );
 482
 483                 return in_array( $mime, $types );
 484         }
 485
 486         /**
 487          * Returns true if the extension represents a type which can
 488          * be reliably detected from its content. Use this to determine
 489          * whether strict content checks should be applied to reject
 490          * invalid uploads; if we can't identify the type we won't
 491          * be able to say if it's invalid.
 492          *
 493          * @todo Be more accurate when using fancy mime detector plugins;
 494          *       right now this is the bare minimum getimagesize() list.
 495          * @param string $extension
 496          * @return bool
 497          */
 498         function isRecognizableExtension( $extension ) {
 499                 static $types = array(
 500                         // Types recognized by getimagesize()
 501                         'gif', 'jpeg', 'jpg', 'png', 'swf', 'psd',
 502                         'bmp', 'tiff', 'tif', 'jpc', 'jp2',
 503                         'jpx', 'jb2', 'swc', 'iff', 'wbmp',
 504                         'xbm',
 505
 506                         // Formats we recognize magic numbers for
 507                         'djvu', 'ogx', 'ogg', 'ogv', 'oga', 'spx',
 508                         'mid', 'pdf', 'wmf', 'xcf', 'webm', 'mkv', 'mka',
 509                         'webp',
 510
 511                         // XML formats we sure hope we recognize reliably
 512                         'svg',
 513                 );
 514                 return in_array( strtolower( $extension ), $types );
 515         }
 516
 517         /**
 518          * Improves a mime type using the file extension. Some file formats are very generic,
 519          * so their mime type is not very meaningful. A more useful mime type can be derived
 520          * by looking at the file extension. Typically, this method would be called on the
 521          * result of guessMimeType().
 522          *
 523          * @param string $mime The mime type, typically guessed from a file's content.
 524          * @param string $ext The file extension, as taken from the file name
 525          *
 526          * @return string The mime type
 527          */
 528         public function improveTypeFromExtension( $mime, $ext ) {
 529                 if ( $mime === 'unknown/unknown' ) {
 530                         if ( $this->isRecognizableExtension( $ext ) ) {
 531                                 wfDebug( __METHOD__ . ': refusing to guess mime type for .' .
 532                                         "$ext file, we should have recognized it\n" );
 533                         } else {
 534                                 // Not something we can detect, so simply
 535                                 // trust the file extension
 536                                 $mime = $this->guessTypesForExtension( $ext );
 537                         }
 538                 } elseif ( $mime === 'application/x-opc+zip' ) {
 539                         if ( $this->isMatchingExtension( $ext, $mime ) ) {
 540                                 // A known file extension for an OPC file,
 541                                 // find the proper mime type for that file extension
 542                                 $mime = $this->guessTypesForExtension( $ext );
 543                         } else {
 544                                 wfDebug( __METHOD__ . ": refusing to guess better type for $mime file, " .
 545                                         ".$ext is not a known OPC extension.\n" );
 546                                 $mime = 'application/zip';
 547                         }
 548                 } elseif ( $mime === 'text/plain' && $this->findMediaType( ".$ext" ) === MEDIATYPE_TEXT ) {
 549                         // Textual types are sometimes not recognized properly.
 550                         // If detected as text/plain, and has an extension which is textual
 551                         // improve to the extension's type. For example, csv and json are often
 552                         // misdetected as text/plain.
 553                         $mime = $this->guessTypesForExtension( $ext );
 554                 }
 555
 556                 # Media handling extensions can improve the MIME detected
 557                 wfRunHooks( 'MimeMagicImproveFromExtension', array( $this, $ext, &$mime ) );
 558
 559                 if ( isset( $this->mMimeTypeAliases[$mime] ) ) {
 560                         $mime = $this->mMimeTypeAliases[$mime];
 561                 }
 562
 563                 wfDebug( __METHOD__ . ": improved mime type for .$ext: $mime\n" );
 564                 return $mime;
 565         }
 566
 567         /**
 568          * Mime type detection. This uses detectMimeType to detect the mime type
 569          * of the file, but applies additional checks to determine some well known
 570          * file formats that may be missed or misinterpreted by the default mime
 571          * detection (namely XML based formats like XHTML or SVG, as well as ZIP
 572          * based formats like OPC/ODF files).
 573          *
 574          * @param string $file The file to check
 575          * @param string|bool $ext The file extension, or true (default) to extract it from the filename.
 576          *   Set it to false to ignore the extension. DEPRECATED! Set to false, use
 577          *   improveTypeFromExtension($mime, $ext) later to improve mime type.
 578          *
 579          * @return string The mime type of $file
 580          */
 581         public function guessMimeType( $file, $ext = true ) {
 582                 if ( $ext ) { // TODO: make $ext default to false. Or better, remove it.
 583                         wfDebug( __METHOD__ . ": WARNING: use of the \$ext parameter is deprecated. " .
 584                                 "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 585                 }
 586
 587                 $mime = $this->doGuessMimeType( $file, $ext );
 588
 589                 if ( !$mime ) {
 590                         wfDebug( __METHOD__ . ": internal type detection failed for $file (.$ext)...\n" );
 591                         $mime = $this->detectMimeType( $file, $ext );
 592                 }
 593
 594                 if ( isset( $this->mMimeTypeAliases[$mime] ) ) {
 595                         $mime = $this->mMimeTypeAliases[$mime];
 596                 }
 597
 598                 wfDebug( __METHOD__ . ": guessed mime type of $file: $mime\n" );
 599                 return $mime;
 600         }
 601
 602         /**
 603          * Guess the mime type from the file contents.
 604          *
 605          * @param string $file
 606          * @param mixed $ext
 607          * @return bool|string
 608          * @throws MWException
 609          */
 610         private function doGuessMimeType( $file, $ext ) { // TODO: remove $ext param
 611                 // Read a chunk of the file
 612                 wfSuppressWarnings();
 613                 $f = fopen( $file, 'rb' );
 614                 wfRestoreWarnings();
 615
 616                 if ( !$f ) {
 617                         return 'unknown/unknown';
 618                 }
 619
 620                 $fsize = filesize( $file );
 621                 if ( $fsize === false ) {
 622                         return 'unknown/unknown';
 623                 }
 624
 625                 $head = fread( $f, 1024 );
 626                 $tailLength = min( 65558, $fsize ); // 65558 = maximum size of a zip EOCDR
 627                 if ( fseek( $f, -1 * $tailLength, SEEK_END ) === -1 ) {
 628                         throw new MWException(
 629                                 "Seeking $tailLength bytes from EOF failed in " . __METHOD__ );
 630                 }
 631                 $tail = fread( $f, $tailLength );
 632                 fclose( $f );
 633
 634                 wfDebug( __METHOD__ . ": analyzing head and tail of $file for magic numbers.\n" );
 635
 636                 // Hardcode a few magic number checks...
 637                 $headers = array(
 638                         // Multimedia...
 639                         'MThd'             => 'audio/midi',
 640                         'OggS'             => 'application/ogg',
 641
 642                         // Image formats...
 643                         // Note that WMF may have a bare header, no magic number.
 644                         "\x01\x00\x09\x00" => 'application/x-msmetafile', // Possibly prone to false positives?
 645                         "\xd7\xcd\xc6\x9a" => 'application/x-msmetafile',
 646                         '%PDF'             => 'application/pdf',
 647                         'gimp xcf'         => 'image/x-xcf',
 648
 649                         // Some forbidden fruit...
 650                         'MZ'               => 'application/octet-stream', // DOS/Windows executable
 651                         "\xca\xfe\xba\xbe" => 'application/octet-stream', // Mach-O binary
 652                         "\x7fELF"          => 'application/octet-stream', // ELF binary
 653                 );
 654
 655                 foreach ( $headers as $magic => $candidate ) {
 656                         if ( strncmp( $head, $magic, strlen( $magic ) ) == 0 ) {
 657                                 wfDebug( __METHOD__ . ": magic header in $file recognized as $candidate\n" );
 658                                 return $candidate;
 659                         }
 660                 }
 661
 662                 /* Look for WebM and Matroska files */
 663                 if ( strncmp( $head, pack( "C4", 0x1a, 0x45, 0xdf, 0xa3 ), 4 ) == 0 ) {
 664                         $doctype = strpos( $head, "\x42\x82" );
 665                         if ( $doctype ) {
 666                                 // Next byte is datasize, then data (sizes larger than 1 byte are very stupid muxers)
 667                                 $data = substr( $head, $doctype + 3, 8 );
 668                                 if ( strncmp( $data, "matroska", 8 ) == 0 ) {
 669                                         wfDebug( __METHOD__ . ": recognized file as video/x-matroska\n" );
 670                                         return "video/x-matroska";
 671                                 } elseif ( strncmp( $data, "webm", 4 ) == 0 ) {
 672                                         wfDebug( __METHOD__ . ": recognized file as video/webm\n" );
 673                                         return "video/webm";
 674                                 }
 675                         }
 676                         wfDebug( __METHOD__ . ": unknown EBML file\n" );
 677                         return "unknown/unknown";
 678                 }
 679
 680                 /* Look for WebP */
 681                 if ( strncmp( $head, "RIFF", 4 ) == 0 && strncmp( substr( $head, 8, 8 ), "WEBPVP8 ", 8 ) == 0 ) {
 682                         wfDebug( __METHOD__ . ": recognized file as image/webp\n" );
 683                         return "image/webp";
 684                 }
 685
 686                 /**
 687                  * Look for PHP.  Check for this before HTML/XML...  Warning: this is a
 688                  * heuristic, and won't match a file with a lot of non-PHP before.  It
 689                  * will also match text files which could be PHP. :)
 690                  *
 691                  * @todo FIXME: For this reason, the check is probably useless -- an attacker
 692                  * could almost certainly just pad the file with a lot of nonsense to
 693                  * circumvent the check in any case where it would be a security
 694                  * problem.  On the other hand, it causes harmful false positives (bug
 695                  * 16583).  The heuristic has been cut down to exclude three-character
 696                  * strings like "<? ", but should it be axed completely?
 697                  */
 698                 if ( ( strpos( $head, '<?php' ) !== false ) ||
 699                         ( strpos( $head, "<\x00?\x00p\x00h\x00p" ) !== false ) ||
 700                         ( strpos( $head, "<\x00?\x00 " ) !== false ) ||
 701                         ( strpos( $head, "<\x00?\x00\n" ) !== false ) ||
 702                         ( strpos( $head, "<\x00?\x00\t" ) !== false ) ||
 703                         ( strpos( $head, "<\x00?\x00=" ) !== false ) ) {
 704
 705                         wfDebug( __METHOD__ . ": recognized $file as application/x-php\n" );
 706                         return 'application/x-php';
 707                 }
 708
 709                 /**
 710                  * look for XML formats (XHTML and SVG)
 711                  */
 712                 $xml = new XmlTypeCheck( $file );
 713                 if ( $xml->wellFormed ) {
 714                         global $wgXMLMimeTypes;
 715                         if ( isset( $wgXMLMimeTypes[$xml->getRootElement()] ) ) {
 716                                 return $wgXMLMimeTypes[$xml->getRootElement()];
 717                         } else {
 718                                 return 'application/xml';
 719                         }
 720                 }
 721
 722                 /**
 723                  * look for shell scripts
 724                  */
 725                 $script_type = null;
 726
 727                 # detect by shebang
 728                 if ( substr( $head, 0, 2 ) == "#!" ) {
 729                         $script_type = "ASCII";
 730                 } elseif ( substr( $head, 0, 5 ) == "\xef\xbb\xbf#!" ) {
 731                         $script_type = "UTF-8";
 732                 } elseif ( substr( $head, 0, 7 ) == "\xfe\xff\x00#\x00!" ) {
 733                         $script_type = "UTF-16BE";
 734                 } elseif ( substr( $head, 0, 7 ) == "\xff\xfe#\x00!" ) {
 735                         $script_type = "UTF-16LE";
 736                 }
 737
 738                 if ( $script_type ) {
 739                         if ( $script_type !== "UTF-8" && $script_type !== "ASCII" ) {
 740                                 // Quick and dirty fold down to ASCII!
 741                                 $pack = array( 'UTF-16BE' => 'n*', 'UTF-16LE' => 'v*' );
 742                                 $chars = unpack( $pack[$script_type], substr( $head, 2 ) );
 743                                 $head = '';
 744                                 foreach ( $chars as $codepoint ) {
 745                                         if ( $codepoint < 128 ) {
 746                                                 $head .= chr( $codepoint );
 747                                         } else {
 748                                                 $head .= '?';
 749                                         }
 750                                 }
 751                         }
 752
 753                         $match = array();
 754
 755                         if ( preg_match( '%/?([^\s]+/)(\w+)%', $head, $match ) ) {
 756                                 $mime = "application/x-{$match[2]}";
 757                                 wfDebug( __METHOD__ . ": shell script recognized as $mime\n" );
 758                                 return $mime;
 759                         }
 760                 }
 761
 762                 // Check for ZIP variants (before getimagesize)
 763                 if ( strpos( $tail, "PK\x05\x06" ) !== false ) {
 764                         wfDebug( __METHOD__ . ": ZIP header present in $file\n" );
 765                         return $this->detectZipType( $head, $tail, $ext );
 766                 }
 767
 768                 wfSuppressWarnings();
 769                 $gis = getimagesize( $file );
 770                 wfRestoreWarnings();
 771
 772                 if ( $gis && isset( $gis['mime'] ) ) {
 773                         $mime = $gis['mime'];
 774                         wfDebug( __METHOD__ . ": getimagesize detected $file as $mime\n" );
 775                         return $mime;
 776                 }
 777
 778                 // Also test DjVu
 779                 $deja = new DjVuImage( $file );
 780                 if ( $deja->isValid() ) {
 781                         wfDebug( __METHOD__ . ": detected $file as image/vnd.djvu\n" );
 782                         return 'image/vnd.djvu';
 783                 }
 784
 785                 # Media handling extensions can guess the MIME by content
 786                 # It's intentionally here so that if core is wrong about a type (false positive),
 787                 # people will hopefully nag and submit patches :)
 788                 $mime = false;
 789                 # Some strings by reference for performance - assuming well-behaved hooks
 790                 wfRunHooks(
 791                         'MimeMagicGuessFromContent',
 792                         array( $this, &$head, &$tail, $file, &$mime )
 793                 );
 794
 795                 return $mime;
 796         }
 797
 798         /**
 799          * Detect application-specific file type of a given ZIP file from its
 800          * header data.  Currently works for OpenDocument and OpenXML types...
 801          * If can't tell, returns 'application/zip'.
 802          *
 803          * @param string $header Some reasonably-sized chunk of file header
 804          * @param string|null $tail The tail of the file
 805          * @param string|bool $ext The file extension, or true to extract it from the filename.
 806          *   Set it to false (default) to ignore the extension. DEPRECATED! Set to false,
 807          *   use improveTypeFromExtension($mime, $ext) later to improve mime type.
 808          *
 809          * @return string
 810          */
 811         function detectZipType( $header, $tail = null, $ext = false ) {
 812                 if ( $ext ) { # TODO: remove $ext param
 813                         wfDebug( __METHOD__ . ": WARNING: use of the \$ext parameter is deprecated. " .
 814                                 "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 815                 }
 816
 817                 $mime = 'application/zip';
 818                 $opendocTypes = array(
 819                         'chart-template',
 820                         'chart',
 821                         'formula-template',
 822                         'formula',
 823                         'graphics-template',
 824                         'graphics',
 825                         'image-template',
 826                         'image',
 827                         'presentation-template',
 828                         'presentation',
 829                         'spreadsheet-template',
 830                         'spreadsheet',
 831                         'text-template',
 832                         'text-master',
 833                         'text-web',
 834                         'text' );
 835
 836                 // http://lists.oasis-open.org/archives/office/200505/msg00006.html
 837                 $types = '(?:' . implode( '|', $opendocTypes ) . ')';
 838                 $opendocRegex = "/^mimetype(application\/vnd\.oasis\.opendocument\.$types)/";
 839
 840                 $openxmlRegex = "/^\[Content_Types\].xml/";
 841
 842                 if ( preg_match( $opendocRegex, substr( $header, 30 ), $matches ) ) {
 843                         $mime = $matches[1];
 844                         wfDebug( __METHOD__ . ": detected $mime from ZIP archive\n" );
 845                 } elseif ( preg_match( $openxmlRegex, substr( $header, 30 ) ) ) {
 846                         $mime = "application/x-opc+zip";
 847                         # TODO: remove the block below, as soon as improveTypeFromExtension is used everywhere
 848                         if ( $ext !== true && $ext !== false ) {
 849                                 /** This is the mode used by getPropsFromPath
 850                                  * These mime's are stored in the database, where we don't really want
 851                                  * x-opc+zip, because we use it only for internal purposes
 852                                  */
 853                                 if ( $this->isMatchingExtension( $ext, $mime ) ) {
 854                                         /* A known file extension for an OPC file,
 855                                          * find the proper mime type for that file extension
 856                                          */
 857                                         $mime = $this->guessTypesForExtension( $ext );
 858                                 } else {
 859                                         $mime = "application/zip";
 860                                 }
 861                         }
 862                         wfDebug( __METHOD__ . ": detected an Open Packaging Conventions archive: $mime\n" );
 863                 } elseif ( substr( $header, 0, 8 ) == "\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" &&
 864                                 ( $headerpos = strpos( $tail, "PK\x03\x04" ) ) !== false &&
 865                                 preg_match( $openxmlRegex, substr( $tail, $headerpos + 30 ) ) ) {
 866                         if ( substr( $header, 512, 4 ) == "\xEC\xA5\xC1\x00" ) {
 867                                 $mime = "application/msword";
 868                         }
 869                         switch ( substr( $header, 512, 6 ) ) {
 870                                 case "\xEC\xA5\xC1\x00\x0E\x00":
 871                                 case "\xEC\xA5\xC1\x00\x1C\x00":
 872                                 case "\xEC\xA5\xC1\x00\x43\x00":
 873                                         $mime = "application/vnd.ms-powerpoint";
 874                                         break;
 875                                 case "\xFD\xFF\xFF\xFF\x10\x00":
 876                                 case "\xFD\xFF\xFF\xFF\x1F\x00":
 877                                 case "\xFD\xFF\xFF\xFF\x22\x00":
 878                                 case "\xFD\xFF\xFF\xFF\x23\x00":
 879                                 case "\xFD\xFF\xFF\xFF\x28\x00":
 880                                 case "\xFD\xFF\xFF\xFF\x29\x00":
 881                                 case "\xFD\xFF\xFF\xFF\x10\x02":
 882                                 case "\xFD\xFF\xFF\xFF\x1F\x02":
 883                                 case "\xFD\xFF\xFF\xFF\x22\x02":
 884                                 case "\xFD\xFF\xFF\xFF\x23\x02":
 885                                 case "\xFD\xFF\xFF\xFF\x28\x02":
 886                                 case "\xFD\xFF\xFF\xFF\x29\x02":
 887                                         $mime = "application/vnd.msexcel";
 888                                         break;
 889                         }
 890
 891                         wfDebug( __METHOD__ . ": detected a MS Office document with OPC trailer\n" );
 892                 } else {
 893                         wfDebug( __METHOD__ . ": unable to identify type of ZIP archive\n" );
 894                 }
 895                 return $mime;
 896         }
 897
 898         /**
 899          * Internal mime type detection. Detection is done using an external
 900          * program, if $wgMimeDetectorCommand is set. Otherwise, the fileinfo
 901          * extension and mime_content_type are tried (in this order), if they
 902          * are available. If the detections fails and $ext is not false, the mime
 903          * type is guessed from the file extension, using guessTypesForExtension.
 904          *
 905          * If the mime type is still unknown, getimagesize is used to detect the
 906          * mime type if the file is an image. If no mime type can be determined,
 907          * this function returns 'unknown/unknown'.
 908          *
 909          * @param string $file The file to check
 910          * @param string|bool $ext The file extension, or true (default) to extract it from the filename.
 911          *   Set it to false to ignore the extension. DEPRECATED! Set to false, use
 912          *   improveTypeFromExtension($mime, $ext) later to improve mime type.
 913          *
 914          * @return string The mime type of $file
 915          */
 916         private function detectMimeType( $file, $ext = true ) {
 917                 global $wgMimeDetectorCommand;
 918
 919                 /** @todo Make $ext default to false. Or better, remove it. */
 920                 if ( $ext ) {
 921                         wfDebug( __METHOD__ . ": WARNING: use of the \$ext parameter is deprecated. "
 922                                 . "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 923                 }
 924
 925                 $m = null;
 926                 if ( $wgMimeDetectorCommand ) {
 927                         $args = wfEscapeShellArg( $file );
 928                         $m = wfShellExec( "$wgMimeDetectorCommand $args" );
 929                 } elseif ( function_exists( "finfo_open" ) && function_exists( "finfo_file" ) ) {
 930
 931                         # This required the fileinfo extension by PECL,
 932                         # see http://pecl.php.net/package/fileinfo
 933                         # This must be compiled into PHP
 934                         #
 935                         # finfo is the official replacement for the deprecated
 936                         # mime_content_type function, see below.
 937                         #
 938                         # If you may need to load the fileinfo extension at runtime, set
 939                         # $wgLoadFileinfoExtension in LocalSettings.php
 940
 941                         $mime_magic_resource = finfo_open( FILEINFO_MIME ); /* return mime type ala mimetype extension */
 942
 943                         if ( $mime_magic_resource ) {
 944                                 $m = finfo_file( $mime_magic_resource, $file );
 945                                 finfo_close( $mime_magic_resource );
 946                         } else {
 947                                 wfDebug( __METHOD__ . ": finfo_open failed on " . FILEINFO_MIME . "!\n" );
 948                         }
 949                 } elseif ( function_exists( "mime_content_type" ) ) {
 950
 951                         # NOTE: this function is available since PHP 4.3.0, but only if
 952                         # PHP was compiled with --with-mime-magic or, before 4.3.2, with
 953                         # --enable-mime-magic.
 954                         #
 955                         # On Windows, you must set mime_magic.magicfile in php.ini to point
 956                         # to the mime.magic file bundled with PHP; sometimes, this may even
 957                         # be needed under *nix.
 958                         #
 959                         # Also note that this has been DEPRECATED in favor of the fileinfo
 960                         # extension by PECL, see above.
 961                         # See http://www.php.net/manual/en/ref.mime-magic.php for details.
 962
 963                         $m = mime_content_type( $file );
 964                 } else {
 965                         wfDebug( __METHOD__ . ": no magic mime detector found!\n" );
 966                 }
 967
 968                 if ( $m ) {
 969                         # normalize
 970                         $m = preg_replace( '![;, ].*$!', '', $m ); #strip charset, etc
 971                         $m = trim( $m );
 972                         $m = strtolower( $m );
 973
 974                         if ( strpos( $m, 'unknown' ) !== false ) {
 975                                 $m = null;
 976                         } else {
 977                                 wfDebug( __METHOD__ . ": magic mime type of $file: $m\n" );
 978                                 return $m;
 979                         }
 980                 }
 981
 982                 // If desired, look at extension as a fallback.
 983                 if ( $ext === true ) {
 984                         $i = strrpos( $file, '.' );
 985                         $ext = strtolower( $i ? substr( $file, $i + 1 ) : '' );
 986                 }
 987                 if ( $ext ) {
 988                         if ( $this->isRecognizableExtension( $ext ) ) {
 989                                 wfDebug( __METHOD__ . ": refusing to guess mime type for .$ext file, "
 990                                         . "we should have recognized it\n" );
 991                         } else {
 992                                 $m = $this->guessTypesForExtension( $ext );
 993                                 if ( $m ) {
 994                                         wfDebug( __METHOD__ . ": extension mime type of $file: $m\n" );
 995                                         return $m;
 996                                 }
 997                         }
 998                 }
 999
1000                 // Unknown type
1001                 wfDebug( __METHOD__ . ": failed to guess mime type for $file!\n" );
1002                 return 'unknown/unknown';
1003         }
1004
1005         /**
1006          * Determine the media type code for a file, using its mime type, name and
1007          * possibly its contents.
1008          *
1009          * This function relies on the findMediaType(), mapping extensions and mime
1010          * types to media types.
1011          *
1012          * @todo analyse file if need be
1013          * @todo look at multiple extension, separately and together.
1014          *
1015          * @param string $path Full path to the image file, in case we have to look at the contents
1016          *        (if null, only the mime type is used to determine the media type code).
1017          * @param string $mime Mime type. If null it will be guessed using guessMimeType.
1018          *
1019          * @return string A value to be used with the MEDIATYPE_xxx constants.
1020          */
1021         function getMediaType( $path = null, $mime = null ) {
1022                 if ( !$mime && !$path ) {
1023                         return MEDIATYPE_UNKNOWN;
1024                 }
1025
1026                 // If mime type is unknown, guess it
1027                 if ( !$mime ) {
1028                         $mime = $this->guessMimeType( $path, false );
1029                 }
1030
1031                 // Special code for ogg - detect if it's video (theora),
1032                 // else label it as sound.
1033                 if ( $mime == 'application/ogg' && file_exists( $path ) ) {
1034
1035                         // Read a chunk of the file
1036                         $f = fopen( $path, "rt" );
1037                         if ( !$f ) {
1038                                 return MEDIATYPE_UNKNOWN;
1039                         }
1040                         $head = fread( $f, 256 );
1041                         fclose( $f );
1042
1043                         $head = str_replace( 'ffmpeg2theora', '', strtolower( $head ) );
1044
1045                         // This is an UGLY HACK, file should be parsed correctly
1046                         if ( strpos( $head, 'theora' ) !== false ) {
1047                                 return MEDIATYPE_VIDEO;
1048                         } elseif ( strpos( $head, 'vorbis' ) !== false ) {
1049                                 return MEDIATYPE_AUDIO;
1050                         } elseif ( strpos( $head, 'flac' ) !== false ) {
1051                                 return MEDIATYPE_AUDIO;
1052                         } elseif ( strpos( $head, 'speex' ) !== false ) {
1053                                 return MEDIATYPE_AUDIO;
1054                         } else {
1055                                 return MEDIATYPE_MULTIMEDIA;
1056                         }
1057                 }
1058
1059                 // Check for entry for full mime type
1060                 if ( $mime ) {
1061                         $type = $this->findMediaType( $mime );
1062                         if ( $type !== MEDIATYPE_UNKNOWN ) {
1063                                 return $type;
1064                         }
1065                 }
1066
1067                 // Check for entry for file extension
1068                 if ( $path ) {
1069                         $i = strrpos( $path, '.' );
1070                         $e = strtolower( $i ? substr( $path, $i + 1 ) : '' );
1071
1072                         // TODO: look at multi-extension if this fails, parse from full path
1073                         $type = $this->findMediaType( '.' . $e );
1074                         if ( $type !== MEDIATYPE_UNKNOWN ) {
1075                                 return $type;
1076                         }
1077                 }
1078
1079                 // Check major mime type
1080                 if ( $mime ) {
1081                         $i = strpos( $mime, '/' );
1082                         if ( $i !== false ) {
1083                                 $major = substr( $mime, 0, $i );
1084                                 $type = $this->findMediaType( $major );
1085                                 if ( $type !== MEDIATYPE_UNKNOWN ) {
1086                                         return $type;
1087                                 }
1088                         }
1089                 }
1090
1091                 if ( !$type ) {
1092                         $type = MEDIATYPE_UNKNOWN;
1093                 }
1094
1095                 return $type;
1096         }
1097
1098         /**
1099          * Returns a media code matching the given mime type or file extension.
1100          * File extensions are represented by a string starting with a dot (.) to
1101          * distinguish them from mime types.
1102          *
1103          * This function relies on the mapping defined by $this->mMediaTypes
1104          * @access private
1105          * @param string $extMime
1106          * @return int|string
1107          */
1108         function findMediaType( $extMime ) {
1109                 if ( strpos( $extMime, '.' ) === 0 ) {
1110                         // If it's an extension, look up the mime types
1111                         $m = $this->getTypesForExtension( substr( $extMime, 1 ) );
1112                         if ( !$m ) {
1113                                 return MEDIATYPE_UNKNOWN;
1114                         }
1115
1116                         $m = explode( ' ', $m );
1117                 } else {
1118                         // Normalize mime type
1119                         if ( isset( $this->mMimeTypeAliases[$extMime] ) ) {
1120                                 $extMime = $this->mMimeTypeAliases[$extMime];
1121                         }
1122
1123                         $m = array( $extMime );
1124                 }
1125
1126                 foreach ( $m as $mime ) {
1127                         foreach ( $this->mMediaTypes as $type => $codes ) {
1128                                 if ( in_array( $mime, $codes, true ) ) {
1129                                         return $type;
1130                                 }
1131                         }
1132                 }
1133
1134                 return MEDIATYPE_UNKNOWN;
1135         }
1136
1137         /**
1138          * Get the MIME types that various versions of Internet Explorer would
1139          * detect from a chunk of the content.
1140          *
1141          * @param string $fileName The file name (unused at present)
1142          * @param string $chunk The first 256 bytes of the file
1143          * @param string $proposed The MIME type proposed by the server
1144          * @return array
1145          */
1146         public function getIEMimeTypes( $fileName, $chunk, $proposed ) {
1147                 $ca = $this->getIEContentAnalyzer();
1148                 return $ca->getRealMimesFromData( $fileName, $chunk, $proposed );
1149         }
1150
1151         /**
1152          * Get a cached instance of IEContentAnalyzer
1153          *
1154          * @return IEContentAnalyzer
1155          */
1156         protected function getIEContentAnalyzer() {
1157                 if ( is_null( $this->mIEAnalyzer ) ) {
1158                         $this->mIEAnalyzer = new IEContentAnalyzer;
1159                 }
1160                 return $this->mIEAnalyzer;
1161         }
1162 }