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