includes/libs/mime/MimeAnalyzer.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 use Psr\Log\LoggerAwareInterface;
  23 use Psr\Log\LoggerInterface;
  24
  25 /**
  26  * Implements functions related to MIME types such as detection and mapping to file extension
  27  *
  28  * @since 1.28
  29  */
  30 class MimeAnalyzer implements LoggerAwareInterface {
  31         /** @var string */
  32         protected $typeFile;
  33         /** @var string */
  34         protected $infoFile;
  35         /** @var string */
  36         protected $xmlTypes;
  37         /** @var callable */
  38         protected $initCallback;
  39         /** @var callable */
  40         protected $detectCallback;
  41         /** @var callable */
  42         protected $guessCallback;
  43         /** @var callable */
  44         protected $extCallback;
  45         /** @var array Mapping of media types to arrays of MIME types */
  46         protected $mediaTypes = null;
  47         /** @var array Map of MIME type aliases */
  48         protected $mimeTypeAliases = null;
  49         /** @var array Map of MIME types to file extensions (as a space separated list) */
  50         protected $mimetoExt = null;
  51
  52         /** @var array Map of file extensions types to MIME types (as a space separated list) */
  53         public $mExtToMime = null; // legacy name; field accessed by hooks
  54
  55         /** @var IEContentAnalyzer */
  56         protected $IEAnalyzer;
  57
  58         /** @var string Extra MIME types, set for example by media handling extensions */
  59         private $extraTypes = '';
  60         /** @var string Extra MIME info, set for example by media handling extensions */
  61         private $extraInfo = '';
  62
  63         /** @var LoggerInterface */
  64         private $logger;
  65
  66         /**
  67          * Defines a set of well known MIME types
  68          * This is used as a fallback to mime.types files.
  69          * An extensive list of well known MIME types is provided by
  70          * the file mime.types in the includes directory.
  71          *
  72          * This list concatenated with mime.types is used to create a MIME <-> ext
  73          * map. Each line contains a MIME type followed by a space separated list of
  74          * extensions. If multiple extensions for a single MIME type exist or if
  75          * multiple MIME types exist for a single extension then in most cases
  76          * MediaWiki assumes that the first extension following the MIME type is the
  77          * canonical extension, and the first time a MIME type appears for a certain
  78          * extension is considered the canonical MIME type.
  79          *
  80          * (Note that appending the type file list to the end of self::$wellKnownTypes
  81          * sucks because you can't redefine canonical types. This could be fixed by
  82          * appending self::$wellKnownTypes behind type file list, but who knows
  83          * what will break? In practice this probably isn't a problem anyway -- Bryan)
  84          */
  85         protected static $wellKnownTypes = <<<EOT
  86 application/ogg ogx ogg ogm ogv oga spx opus
  87 application/pdf pdf
  88 application/vnd.oasis.opendocument.chart odc
  89 application/vnd.oasis.opendocument.chart-template otc
  90 application/vnd.oasis.opendocument.database odb
  91 application/vnd.oasis.opendocument.formula odf
  92 application/vnd.oasis.opendocument.formula-template otf
  93 application/vnd.oasis.opendocument.graphics odg
  94 application/vnd.oasis.opendocument.graphics-template otg
  95 application/vnd.oasis.opendocument.image odi
  96 application/vnd.oasis.opendocument.image-template oti
  97 application/vnd.oasis.opendocument.presentation odp
  98 application/vnd.oasis.opendocument.presentation-template otp
  99 application/vnd.oasis.opendocument.spreadsheet ods
 100 application/vnd.oasis.opendocument.spreadsheet-template ots
 101 application/vnd.oasis.opendocument.text odt
 102 application/vnd.oasis.opendocument.text-master otm
 103 application/vnd.oasis.opendocument.text-template ott
 104 application/vnd.oasis.opendocument.text-web oth
 105 application/javascript js
 106 application/x-shockwave-flash swf
 107 audio/midi mid midi kar
 108 audio/mpeg mpga mpa mp2 mp3
 109 audio/x-aiff aif aiff aifc
 110 audio/x-wav wav
 111 audio/ogg oga spx ogg opus
 112 audio/opus opus ogg oga ogg spx
 113 image/x-bmp bmp
 114 image/gif gif
 115 image/jpeg jpeg jpg jpe
 116 image/png png
 117 image/svg+xml svg
 118 image/svg svg
 119 image/tiff tiff tif
 120 image/vnd.djvu djvu
 121 image/x.djvu djvu
 122 image/x-djvu djvu
 123 image/x-portable-pixmap ppm
 124 image/x-xcf xcf
 125 text/plain txt
 126 text/html html htm
 127 video/ogg ogv ogm ogg
 128 video/mpeg mpg mpeg
 129 EOT;
 130
 131         /**
 132          * Defines a set of well known MIME info entries
 133          * This is used as a fallback to mime.info files.
 134          * An extensive list of well known MIME types is provided by
 135          * the file mime.info in the includes directory.
 136          */
 137         protected static $wellKnownInfo = <<<EOT
 138 application/pdf [OFFICE]
 139 application/vnd.oasis.opendocument.chart [OFFICE]
 140 application/vnd.oasis.opendocument.chart-template [OFFICE]
 141 application/vnd.oasis.opendocument.database [OFFICE]
 142 application/vnd.oasis.opendocument.formula [OFFICE]
 143 application/vnd.oasis.opendocument.formula-template [OFFICE]
 144 application/vnd.oasis.opendocument.graphics [OFFICE]
 145 application/vnd.oasis.opendocument.graphics-template [OFFICE]
 146 application/vnd.oasis.opendocument.image [OFFICE]
 147 application/vnd.oasis.opendocument.image-template [OFFICE]
 148 application/vnd.oasis.opendocument.presentation [OFFICE]
 149 application/vnd.oasis.opendocument.presentation-template [OFFICE]
 150 application/vnd.oasis.opendocument.spreadsheet [OFFICE]
 151 application/vnd.oasis.opendocument.spreadsheet-template [OFFICE]
 152 application/vnd.oasis.opendocument.text [OFFICE]
 153 application/vnd.oasis.opendocument.text-template [OFFICE]
 154 application/vnd.oasis.opendocument.text-master [OFFICE]
 155 application/vnd.oasis.opendocument.text-web [OFFICE]
 156 application/javascript text/javascript application/x-javascript [EXECUTABLE]
 157 application/x-shockwave-flash [MULTIMEDIA]
 158 audio/midi [AUDIO]
 159 audio/x-aiff [AUDIO]
 160 audio/x-wav [AUDIO]
 161 audio/mp3 audio/mpeg [AUDIO]
 162 application/ogg audio/ogg video/ogg [MULTIMEDIA]
 163 image/x-bmp image/x-ms-bmp image/bmp [BITMAP]
 164 image/gif [BITMAP]
 165 image/jpeg [BITMAP]
 166 image/png [BITMAP]
 167 image/svg+xml [DRAWING]
 168 image/tiff [BITMAP]
 169 image/vnd.djvu [BITMAP]
 170 image/x-xcf [BITMAP]
 171 image/x-portable-pixmap [BITMAP]
 172 text/plain [TEXT]
 173 text/html [TEXT]
 174 video/ogg [VIDEO]
 175 video/mpeg [VIDEO]
 176 unknown/unknown application/octet-stream application/x-empty [UNKNOWN]
 177 EOT;
 178
 179         /**
 180          * @param array $params Configuration map, includes:
 181          *   - typeFile: path to file with the list of known MIME types
 182          *   - infoFile: path to file with the MIME type info
 183          *   - xmlTypes: map of root element names to XML MIME types
 184          *   - initCallback: initialization callback that is passed this object [optional]
 185          *   - detectCallback: alternative to finfo that returns the mime type for a file.
 186          *      For example, the callback can return the output of "file -bi". [optional]
 187          *   - guessCallback: callback to improve the guessed MIME type using the file data.
 188          *      This is intended for fixing mistakes in fileinfo or "detectCallback". [optional]
 189          *   - extCallback: callback to improve the guessed MIME type using the extension. [optional]
 190          *   - logger: PSR-3 logger [optional]
 191          * @note Constructing these instances is expensive due to file reads.
 192          *  A service or singleton pattern should be used to avoid creating instances again and again.
 193          */
 194         public function __construct( array $params ) {
 195                 $this->typeFile = $params['typeFile'];
 196                 $this->infoFile = $params['infoFile'];
 197                 $this->xmlTypes = $params['xmlTypes'];
 198                 $this->initCallback = $params['initCallback'] ?? null;
 199                 $this->detectCallback = $params['detectCallback'] ?? null;
 200                 $this->guessCallback = $params['guessCallback'] ?? null;
 201                 $this->extCallback = $params['extCallback'] ?? null;
 202                 $this->logger = $params['logger'] ?? new \Psr\Log\NullLogger();
 203
 204                 $this->loadFiles();
 205         }
 206
 207         protected function loadFiles() {
 208                 /**
 209                  *   --- load mime.types ---
 210                  */
 211
 212                 # Allow media handling extensions adding MIME-types and MIME-info
 213                 if ( $this->initCallback ) {
 214                         call_user_func( $this->initCallback, $this );
 215                 }
 216
 217                 $types = self::$wellKnownTypes;
 218
 219                 $mimeTypeFile = $this->typeFile;
 220                 if ( $mimeTypeFile ) {
 221                         if ( is_file( $mimeTypeFile ) && is_readable( $mimeTypeFile ) ) {
 222                                 $this->logger->info( __METHOD__ . ": loading mime types from $mimeTypeFile\n" );
 223                                 $types .= "\n";
 224                                 $types .= file_get_contents( $mimeTypeFile );
 225                         } else {
 226                                 $this->logger->info( __METHOD__ . ": can't load mime types from $mimeTypeFile\n" );
 227                         }
 228                 } else {
 229                         $this->logger->info( __METHOD__ .
 230                                 ": no mime types file defined, using built-ins only.\n" );
 231                 }
 232
 233                 $types .= "\n" . $this->extraTypes;
 234
 235                 $types = str_replace( [ "\r\n", "\n\r", "\n\n", "\r\r", "\r" ], "\n", $types );
 236                 $types = str_replace( "\t", " ", $types );
 237
 238                 $this->mimetoExt = [];
 239                 $this->mExtToMime = [];
 240
 241                 $lines = explode( "\n", $types );
 242                 foreach ( $lines as $s ) {
 243                         $s = trim( $s );
 244                         if ( empty( $s ) ) {
 245                                 continue;
 246                         }
 247                         if ( strpos( $s, '#' ) === 0 ) {
 248                                 continue;
 249                         }
 250
 251                         $s = strtolower( $s );
 252                         $i = strpos( $s, ' ' );
 253
 254                         if ( $i === false ) {
 255                                 continue;
 256                         }
 257
 258                         $mime = substr( $s, 0, $i );
 259                         $ext = trim( substr( $s, $i + 1 ) );
 260
 261                         if ( empty( $ext ) ) {
 262                                 continue;
 263                         }
 264
 265                         if ( !empty( $this->mimetoExt[$mime] ) ) {
 266                                 $this->mimetoExt[$mime] .= ' ' . $ext;
 267                         } else {
 268                                 $this->mimetoExt[$mime] = $ext;
 269                         }
 270
 271                         $extensions = explode( ' ', $ext );
 272
 273                         foreach ( $extensions as $e ) {
 274                                 $e = trim( $e );
 275                                 if ( empty( $e ) ) {
 276                                         continue;
 277                                 }
 278
 279                                 if ( !empty( $this->mExtToMime[$e] ) ) {
 280                                         $this->mExtToMime[$e] .= ' ' . $mime;
 281                                 } else {
 282                                         $this->mExtToMime[$e] = $mime;
 283                                 }
 284                         }
 285                 }
 286
 287                 /**
 288                  *   --- load mime.info ---
 289                  */
 290
 291                 $mimeInfoFile = $this->infoFile;
 292
 293                 $info = self::$wellKnownInfo;
 294
 295                 if ( $mimeInfoFile ) {
 296                         if ( is_file( $mimeInfoFile ) && is_readable( $mimeInfoFile ) ) {
 297                                 $this->logger->info( __METHOD__ . ": loading mime info from $mimeInfoFile\n" );
 298                                 $info .= "\n";
 299                                 $info .= file_get_contents( $mimeInfoFile );
 300                         } else {
 301                                 $this->logger->info( __METHOD__ . ": can't load mime info from $mimeInfoFile\n" );
 302                         }
 303                 } else {
 304                         $this->logger->info( __METHOD__ .
 305                                 ": no mime info file defined, using built-ins only.\n" );
 306                 }
 307
 308                 $info .= "\n" . $this->extraInfo;
 309
 310                 $info = str_replace( [ "\r\n", "\n\r", "\n\n", "\r\r", "\r" ], "\n", $info );
 311                 $info = str_replace( "\t", " ", $info );
 312
 313                 $this->mimeTypeAliases = [];
 314                 $this->mediaTypes = [];
 315
 316                 $lines = explode( "\n", $info );
 317                 foreach ( $lines as $s ) {
 318                         $s = trim( $s );
 319                         if ( empty( $s ) ) {
 320                                 continue;
 321                         }
 322                         if ( strpos( $s, '#' ) === 0 ) {
 323                                 continue;
 324                         }
 325
 326                         $s = strtolower( $s );
 327                         $i = strpos( $s, ' ' );
 328
 329                         if ( $i === false ) {
 330                                 continue;
 331                         }
 332
 333                         # print "processing MIME INFO line $s<br>";
 334
 335                         $match = [];
 336                         if ( preg_match( '!\[\s*(\w+)\s*\]!', $s, $match ) ) {
 337                                 $s = preg_replace( '!\[\s*(\w+)\s*\]!', '', $s );
 338                                 $mtype = trim( strtoupper( $match[1] ) );
 339                         } else {
 340                                 $mtype = MEDIATYPE_UNKNOWN;
 341                         }
 342
 343                         $m = explode( ' ', $s );
 344
 345                         if ( !isset( $this->mediaTypes[$mtype] ) ) {
 346                                 $this->mediaTypes[$mtype] = [];
 347                         }
 348
 349                         foreach ( $m as $mime ) {
 350                                 $mime = trim( $mime );
 351                                 if ( empty( $mime ) ) {
 352                                         continue;
 353                                 }
 354
 355                                 $this->mediaTypes[$mtype][] = $mime;
 356                         }
 357
 358                         if ( count( $m ) > 1 ) {
 359                                 $main = $m[0];
 360                                 $mCount = count( $m );
 361                                 for ( $i = 1; $i < $mCount; $i += 1 ) {
 362                                         $mime = $m[$i];
 363                                         $this->mimeTypeAliases[$mime] = $main;
 364                                 }
 365                         }
 366                 }
 367         }
 368
 369         public function setLogger( LoggerInterface $logger ) {
 370                 $this->logger = $logger;
 371         }
 372
 373         /**
 374          * Adds to the list mapping MIME to file extensions.
 375          * As an extension author, you are encouraged to submit patches to
 376          * MediaWiki's core to add new MIME types to mime.types.
 377          * @param string $types
 378          */
 379         public function addExtraTypes( $types ) {
 380                 $this->extraTypes .= "\n" . $types;
 381         }
 382
 383         /**
 384          * Adds to the list mapping MIME to media type.
 385          * As an extension author, you are encouraged to submit patches to
 386          * MediaWiki's core to add new MIME info to mime.info.
 387          * @param string $info
 388          */
 389         public function addExtraInfo( $info ) {
 390                 $this->extraInfo .= "\n" . $info;
 391         }
 392
 393         /**
 394          * Returns a list of file extensions for a given MIME type as a space
 395          * separated string or null if the MIME type was unrecognized. Resolves
 396          * MIME type aliases.
 397          *
 398          * @param string $mime
 399          * @return string|null
 400          */
 401         public function getExtensionsForType( $mime ) {
 402                 $mime = strtolower( $mime );
 403
 404                 // Check the mime-to-ext map
 405                 if ( isset( $this->mimetoExt[$mime] ) ) {
 406                         return $this->mimetoExt[$mime];
 407                 }
 408
 409                 // Resolve the MIME type to the canonical type
 410                 if ( isset( $this->mimeTypeAliases[$mime] ) ) {
 411                         $mime = $this->mimeTypeAliases[$mime];
 412                         if ( isset( $this->mimetoExt[$mime] ) ) {
 413                                 return $this->mimetoExt[$mime];
 414                         }
 415                 }
 416
 417                 return null;
 418         }
 419
 420         /**
 421          * Returns a list of MIME types for a given file extension as a space
 422          * separated string or null if the extension was unrecognized.
 423          *
 424          * @param string $ext
 425          * @return string|null
 426          */
 427         public function getTypesForExtension( $ext ) {
 428                 $ext = strtolower( $ext );
 429
 430                 $r = $this->mExtToMime[$ext] ?? null;
 431                 return $r;
 432         }
 433
 434         /**
 435          * Returns a single MIME type for a given file extension or null if unknown.
 436          * This is always the first type from the list returned by getTypesForExtension($ext).
 437          *
 438          * @param string $ext
 439          * @return string|null
 440          */
 441         public function guessTypesForExtension( $ext ) {
 442                 $m = $this->getTypesForExtension( $ext );
 443                 if ( is_null( $m ) ) {
 444                         return null;
 445                 }
 446
 447                 // TODO: Check if this is needed; strtok( $m, ' ' ) should be sufficient
 448                 $m = trim( $m );
 449                 $m = preg_replace( '/\s.*$/', '', $m );
 450
 451                 return $m;
 452         }
 453
 454         /**
 455          * Tests if the extension matches the given MIME type. Returns true if a
 456          * match was found, null if the MIME type is unknown, and false if the
 457          * MIME type is known but no matches where found.
 458          *
 459          * @param string $extension
 460          * @param string $mime
 461          * @return bool|null
 462          */
 463         public function isMatchingExtension( $extension, $mime ) {
 464                 $ext = $this->getExtensionsForType( $mime );
 465
 466                 if ( !$ext ) {
 467                         return null; // Unknown MIME type
 468                 }
 469
 470                 $ext = explode( ' ', $ext );
 471
 472                 $extension = strtolower( $extension );
 473                 return in_array( $extension, $ext );
 474         }
 475
 476         /**
 477          * Returns true if the MIME type is known to represent an image format
 478          * supported by the PHP GD library.
 479          *
 480          * @param string $mime
 481          *
 482          * @return bool
 483          */
 484         public function isPHPImageType( $mime ) {
 485                 // As defined by imagegetsize and image_type_to_mime
 486                 static $types = [
 487                         'image/gif', 'image/jpeg', 'image/png',
 488                         'image/x-bmp', 'image/xbm', 'image/tiff',
 489                         'image/jp2', 'image/jpeg2000', 'image/iff',
 490                         'image/xbm', 'image/x-xbitmap',
 491                         'image/vnd.wap.wbmp', 'image/vnd.xiff',
 492                         'image/x-photoshop',
 493                         'application/x-shockwave-flash',
 494                 ];
 495
 496                 return in_array( $mime, $types );
 497         }
 498
 499         /**
 500          * Returns true if the extension represents a type which can
 501          * be reliably detected from its content. Use this to determine
 502          * whether strict content checks should be applied to reject
 503          * invalid uploads; if we can't identify the type we won't
 504          * be able to say if it's invalid.
 505          *
 506          * @todo Be more accurate when using fancy MIME detector plugins;
 507          *       right now this is the bare minimum getimagesize() list.
 508          * @param string $extension
 509          * @return bool
 510          */
 511         function isRecognizableExtension( $extension ) {
 512                 static $types = [
 513                         // Types recognized by getimagesize()
 514                         'gif', 'jpeg', 'jpg', 'png', 'swf', 'psd',
 515                         'bmp', 'tiff', 'tif', 'jpc', 'jp2',
 516                         'jpx', 'jb2', 'swc', 'iff', 'wbmp',
 517                         'xbm',
 518
 519                         // Formats we recognize magic numbers for
 520                         'djvu', 'ogx', 'ogg', 'ogv', 'oga', 'spx', 'opus',
 521                         'mid', 'pdf', 'wmf', 'xcf', 'webm', 'mkv', 'mka',
 522                         'webp', 'mp3',
 523
 524                         // XML formats we sure hope we recognize reliably
 525                         'svg',
 526
 527                         // 3D formats
 528                         'stl',
 529                 ];
 530                 return in_array( strtolower( $extension ), $types );
 531         }
 532
 533         /**
 534          * Improves a MIME type using the file extension. Some file formats are very generic,
 535          * so their MIME type is not very meaningful. A more useful MIME type can be derived
 536          * by looking at the file extension. Typically, this method would be called on the
 537          * result of guessMimeType().
 538          *
 539          * @param string $mime The MIME type, typically guessed from a file's content.
 540          * @param string $ext The file extension, as taken from the file name
 541          *
 542          * @return string The MIME type
 543          */
 544         public function improveTypeFromExtension( $mime, $ext ) {
 545                 if ( $mime === 'unknown/unknown' ) {
 546                         if ( $this->isRecognizableExtension( $ext ) ) {
 547                                 $this->logger->info( __METHOD__ . ': refusing to guess mime type for .' .
 548                                         "$ext file, we should have recognized it\n" );
 549                         } else {
 550                                 // Not something we can detect, so simply
 551                                 // trust the file extension
 552                                 $mime = $this->guessTypesForExtension( $ext );
 553                         }
 554                 } elseif ( $mime === 'application/x-opc+zip' ) {
 555                         if ( $this->isMatchingExtension( $ext, $mime ) ) {
 556                                 // A known file extension for an OPC file,
 557                                 // find the proper MIME type for that file extension
 558                                 $mime = $this->guessTypesForExtension( $ext );
 559                         } else {
 560                                 $this->logger->info( __METHOD__ .
 561                                         ": refusing to guess better type for $mime file, " .
 562                                         ".$ext is not a known OPC extension.\n" );
 563                                 $mime = 'application/zip';
 564                         }
 565                 } elseif ( $mime === 'text/plain' && $this->findMediaType( ".$ext" ) === MEDIATYPE_TEXT ) {
 566                         // Textual types are sometimes not recognized properly.
 567                         // If detected as text/plain, and has an extension which is textual
 568                         // improve to the extension's type. For example, csv and json are often
 569                         // misdetected as text/plain.
 570                         $mime = $this->guessTypesForExtension( $ext );
 571                 }
 572
 573                 # Media handling extensions can improve the MIME detected
 574                 $callback = $this->extCallback;
 575                 if ( $callback ) {
 576                         $callback( $this, $ext, $mime /* by reference */ );
 577                 }
 578
 579                 if ( isset( $this->mimeTypeAliases[$mime] ) ) {
 580                         $mime = $this->mimeTypeAliases[$mime];
 581                 }
 582
 583                 $this->logger->info( __METHOD__ . ": improved mime type for .$ext: $mime\n" );
 584                 return $mime;
 585         }
 586
 587         /**
 588          * MIME type detection. This uses detectMimeType to detect the MIME type
 589          * of the file, but applies additional checks to determine some well known
 590          * file formats that may be missed or misinterpreted by the default MIME
 591          * detection (namely XML based formats like XHTML or SVG, as well as ZIP
 592          * based formats like OPC/ODF files).
 593          *
 594          * @param string $file The file to check
 595          * @param string|bool $ext The file extension, or true (default) to extract
 596          *   it from the filename. Set it to false to ignore the extension. DEPRECATED!
 597          *   Set to false, use improveTypeFromExtension($mime, $ext) later to improve MIME type.
 598          *
 599          * @return string The MIME type of $file
 600          */
 601         public function guessMimeType( $file, $ext = true ) {
 602                 if ( $ext ) { // TODO: make $ext default to false. Or better, remove it.
 603                         $this->logger->info( __METHOD__ .
 604                                 ": WARNING: use of the \$ext parameter is deprecated. " .
 605                                 "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 606                 }
 607
 608                 $mime = $this->doGuessMimeType( $file, $ext );
 609
 610                 if ( !$mime ) {
 611                         $this->logger->info( __METHOD__ .
 612                                 ": internal type detection failed for $file (.$ext)...\n" );
 613                         $mime = $this->detectMimeType( $file, $ext );
 614                 }
 615
 616                 if ( isset( $this->mimeTypeAliases[$mime] ) ) {
 617                         $mime = $this->mimeTypeAliases[$mime];
 618                 }
 619
 620                 $this->logger->info( __METHOD__ . ": guessed mime type of $file: $mime\n" );
 621                 return $mime;
 622         }
 623
 624         /**
 625          * Guess the MIME type from the file contents.
 626          *
 627          * @todo Remove $ext param
 628          *
 629          * @param string $file
 630          * @param mixed $ext
 631          * @return bool|string
 632          * @throws UnexpectedValueException
 633          */
 634         private function doGuessMimeType( $file, $ext ) {
 635                 // Read a chunk of the file
 636                 Wikimedia\suppressWarnings();
 637                 $f = fopen( $file, 'rb' );
 638                 Wikimedia\restoreWarnings();
 639
 640                 if ( !$f ) {
 641                         return 'unknown/unknown';
 642                 }
 643
 644                 $fsize = filesize( $file );
 645                 if ( $fsize === false ) {
 646                         return 'unknown/unknown';
 647                 }
 648
 649                 $head = fread( $f, 1024 );
 650                 $tailLength = min( 65558, $fsize ); // 65558 = maximum size of a zip EOCDR
 651                 if ( fseek( $f, -1 * $tailLength, SEEK_END ) === -1 ) {
 652                         throw new UnexpectedValueException(
 653                                 "Seeking $tailLength bytes from EOF failed in " . __METHOD__ );
 654                 }
 655                 $tail = $tailLength ? fread( $f, $tailLength ) : '';
 656
 657                 $this->logger->info( __METHOD__ .
 658                         ": analyzing head and tail of $file for magic numbers.\n" );
 659
 660                 // Hardcode a few magic number checks...
 661                 $headers = [
 662                         // Multimedia...
 663                         'MThd'             => 'audio/midi',
 664                         'OggS'             => 'application/ogg',
 665                         'ID3'              => 'audio/mpeg',
 666                         "\xff\xfb"         => 'audio/mpeg', // MPEG-1 layer 3
 667                         "\xff\xf3"         => 'audio/mpeg', // MPEG-2 layer 3 (lower sample rates)
 668                         "\xff\xe3"         => 'audio/mpeg', // MPEG-2.5 layer 3 (very low sample rates)
 669
 670                         // Image formats...
 671                         // Note that WMF may have a bare header, no magic number.
 672                         "\x01\x00\x09\x00" => 'application/x-msmetafile', // Possibly prone to false positives?
 673                         "\xd7\xcd\xc6\x9a" => 'application/x-msmetafile',
 674                         '%PDF'             => 'application/pdf',
 675                         'gimp xcf'         => 'image/x-xcf',
 676
 677                         // Some forbidden fruit...
 678                         'MZ'               => 'application/octet-stream', // DOS/Windows executable
 679                         "\xca\xfe\xba\xbe" => 'application/octet-stream', // Mach-O binary
 680                         "\x7fELF"          => 'application/octet-stream', // ELF binary
 681                 ];
 682
 683                 foreach ( $headers as $magic => $candidate ) {
 684                         if ( strncmp( $head, $magic, strlen( $magic ) ) == 0 ) {
 685                                 $this->logger->info( __METHOD__ .
 686                                         ": magic header in $file recognized as $candidate\n" );
 687                                 return $candidate;
 688                         }
 689                 }
 690
 691                 /* Look for WebM and Matroska files */
 692                 if ( strncmp( $head, pack( "C4", 0x1a, 0x45, 0xdf, 0xa3 ), 4 ) == 0 ) {
 693                         $doctype = strpos( $head, "\x42\x82" );
 694                         if ( $doctype ) {
 695                                 // Next byte is datasize, then data (sizes larger than 1 byte are stupid muxers)
 696                                 $data = substr( $head, $doctype + 3, 8 );
 697                                 if ( strncmp( $data, "matroska", 8 ) == 0 ) {
 698                                         $this->logger->info( __METHOD__ . ": recognized file as video/x-matroska\n" );
 699                                         return "video/x-matroska";
 700                                 } elseif ( strncmp( $data, "webm", 4 ) == 0 ) {
 701                                         // XXX HACK look for a video track, if we don't find it, this is an audio file
 702                                         $videotrack = strpos( $head, "\x86\x85V_VP" );
 703
 704                                         if ( $videotrack ) {
 705                                                 // There is a video track, so this is a video file.
 706                                                 $this->logger->info( __METHOD__ . ": recognized file as video/webm\n" );
 707                                                 return "video/webm";
 708                                         }
 709
 710                                         $this->logger->info( __METHOD__ . ": recognized file as audio/webm\n" );
 711                                         return "audio/webm";
 712                                 }
 713                         }
 714                         $this->logger->info( __METHOD__ . ": unknown EBML file\n" );
 715                         return "unknown/unknown";
 716                 }
 717
 718                 /* Look for WebP */
 719                 if ( strncmp( $head, "RIFF", 4 ) == 0 &&
 720                         strncmp( substr( $head, 8, 7 ), "WEBPVP8", 7 ) == 0
 721                 ) {
 722                         $this->logger->info( __METHOD__ . ": recognized file as image/webp\n" );
 723                         return "image/webp";
 724                 }
 725
 726                 /* Look for MS Compound Binary (OLE) files */
 727                 if ( strncmp( $head, "\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1", 8 ) == 0 ) {
 728                         $this->logger->info( __METHOD__ . ': recognized MS CFB (OLE) file' );
 729                         return $this->detectMicrosoftBinaryType( $f );
 730                 }
 731
 732                 /**
 733                  * Look for PHP.  Check for this before HTML/XML...  Warning: this is a
 734                  * heuristic, and won't match a file with a lot of non-PHP before.  It
 735                  * will also match text files which could be PHP. :)
 736                  *
 737                  * @todo FIXME: For this reason, the check is probably useless -- an attacker
 738                  * could almost certainly just pad the file with a lot of nonsense to
 739                  * circumvent the check in any case where it would be a security
 740                  * problem.  On the other hand, it causes harmful false positives (bug
 741                  * 16583).  The heuristic has been cut down to exclude three-character
 742                  * strings like "<? ", but should it be axed completely?
 743                  */
 744                 if ( ( strpos( $head, '<?php' ) !== false ) ||
 745                         ( strpos( $head, "<\x00?\x00p\x00h\x00p" ) !== false ) ||
 746                         ( strpos( $head, "<\x00?\x00 " ) !== false ) ||
 747                         ( strpos( $head, "<\x00?\x00\n" ) !== false ) ||
 748                         ( strpos( $head, "<\x00?\x00\t" ) !== false ) ||
 749                         ( strpos( $head, "<\x00?\x00=" ) !== false )
 750                 ) {
 751                         $this->logger->info( __METHOD__ . ": recognized $file as application/x-php\n" );
 752                         return 'application/x-php';
 753                 }
 754
 755                 /**
 756                  * look for XML formats (XHTML and SVG)
 757                  */
 758                 $xml = new XmlTypeCheck( $file );
 759                 if ( $xml->wellFormed ) {
 760                         $xmlTypes = $this->xmlTypes;
 761                         return $xmlTypes[$xml->getRootElement()] ?? 'application/xml';
 762                 }
 763
 764                 /**
 765                  * look for shell scripts
 766                  */
 767                 $script_type = null;
 768
 769                 # detect by shebang
 770                 if ( substr( $head, 0, 2 ) == "#!" ) {
 771                         $script_type = "ASCII";
 772                 } elseif ( substr( $head, 0, 5 ) == "\xef\xbb\xbf#!" ) {
 773                         $script_type = "UTF-8";
 774                 } elseif ( substr( $head, 0, 7 ) == "\xfe\xff\x00#\x00!" ) {
 775                         $script_type = "UTF-16BE";
 776                 } elseif ( substr( $head, 0, 7 ) == "\xff\xfe#\x00!" ) {
 777                         $script_type = "UTF-16LE";
 778                 }
 779
 780                 if ( $script_type ) {
 781                         if ( $script_type !== "UTF-8" && $script_type !== "ASCII" ) {
 782                                 // Quick and dirty fold down to ASCII!
 783                                 $pack = [ 'UTF-16BE' => 'n*', 'UTF-16LE' => 'v*' ];
 784                                 $chars = unpack( $pack[$script_type], substr( $head, 2 ) );
 785                                 $head = '';
 786                                 foreach ( $chars as $codepoint ) {
 787                                         if ( $codepoint < 128 ) {
 788                                                 $head .= chr( $codepoint );
 789                                         } else {
 790                                                 $head .= '?';
 791                                         }
 792                                 }
 793                         }
 794
 795                         $match = [];
 796
 797                         if ( preg_match( '%/?([^\s]+/)(\w+)%', $head, $match ) ) {
 798                                 $mime = "application/x-{$match[2]}";
 799                                 $this->logger->info( __METHOD__ . ": shell script recognized as $mime\n" );
 800                                 return $mime;
 801                         }
 802                 }
 803
 804                 // Check for ZIP variants (before getimagesize)
 805                 $eocdrPos = strpos( $tail, "PK\x05\x06" );
 806                 if ( $eocdrPos !== false ) {
 807                         $this->logger->info( __METHOD__ . ": ZIP signature present in $file\n" );
 808                         // Check if it really is a ZIP file, make sure the EOCDR is at the end (T40432)
 809                         // FIXME: unpack()'s third argument was added in PHP 7.1
 810                         // @phan-suppress-next-line PhanParamTooManyInternal
 811                         $commentLength = unpack( "n", $tail, $eocdrPos + 20 )[0];
 812                         if ( $eocdrPos + 22 + $commentLength !== strlen( $tail ) ) {
 813                                 $this->logger->info( __METHOD__ . ": ZIP EOCDR not at end. Not a ZIP file." );
 814                         } else {
 815                                 return $this->detectZipType( $head, $tail, $ext );
 816                         }
 817                 }
 818
 819                 // Check for STL (3D) files
 820                 // @see https://en.wikipedia.org/wiki/STL_(file_format)
 821                 if ( $fsize >= 15 &&
 822                         stripos( $head, 'SOLID ' ) === 0 &&
 823                         preg_match( '/\RENDSOLID .*$/i', $tail ) ) {
 824                         // ASCII STL file
 825                         return 'application/sla';
 826                 } elseif ( $fsize > 84 ) {
 827                         // binary STL file
 828                         $triangles = substr( $head, 80, 4 );
 829                         $triangles = unpack( 'V', $triangles );
 830                         $triangles = reset( $triangles );
 831                         if ( $triangles !== false && $fsize === 84 + ( $triangles * 50 ) ) {
 832                                 return 'application/sla';
 833                         }
 834                 }
 835
 836                 Wikimedia\suppressWarnings();
 837                 $gis = getimagesize( $file );
 838                 Wikimedia\restoreWarnings();
 839
 840                 if ( $gis && isset( $gis['mime'] ) ) {
 841                         $mime = $gis['mime'];
 842                         $this->logger->info( __METHOD__ . ": getimagesize detected $file as $mime\n" );
 843                         return $mime;
 844                 }
 845
 846                 # Media handling extensions can guess the MIME by content
 847                 # It's intentionally here so that if core is wrong about a type (false positive),
 848                 # people will hopefully nag and submit patches :)
 849                 $mime = false;
 850                 # Some strings by reference for performance - assuming well-behaved hooks
 851                 $callback = $this->guessCallback;
 852                 if ( $callback ) {
 853                         $callback( $this, $head, $tail, $file, $mime /* by reference */ );
 854                 };
 855
 856                 return $mime;
 857         }
 858
 859         /**
 860          * Detect application-specific file type of a given ZIP file from its
 861          * header data.  Currently works for OpenDocument and OpenXML types...
 862          * If can't tell, returns 'application/zip'.
 863          *
 864          * @param string $header Some reasonably-sized chunk of file header
 865          * @param string|null $tail The tail of the file
 866          * @param string|bool $ext The file extension, or true to extract it from the filename.
 867          *   Set it to false (default) to ignore the extension. DEPRECATED! Set to false,
 868          *   use improveTypeFromExtension($mime, $ext) later to improve MIME type.
 869          *
 870          * @return string
 871          */
 872         function detectZipType( $header, $tail = null, $ext = false ) {
 873                 if ( $ext ) { # TODO: remove $ext param
 874                         $this->logger->info( __METHOD__ .
 875                                 ": WARNING: use of the \$ext parameter is deprecated. " .
 876                                 "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 877                 }
 878
 879                 $mime = 'application/zip';
 880                 $opendocTypes = [
 881                         'chart-template',
 882                         'chart',
 883                         'formula-template',
 884                         'formula',
 885                         'graphics-template',
 886                         'graphics',
 887                         'image-template',
 888                         'image',
 889                         'presentation-template',
 890                         'presentation',
 891                         'spreadsheet-template',
 892                         'spreadsheet',
 893                         'text-template',
 894                         'text-master',
 895                         'text-web',
 896                         'text' ];
 897
 898                 // https://lists.oasis-open.org/archives/office/200505/msg00006.html
 899                 $types = '(?:' . implode( '|', $opendocTypes ) . ')';
 900                 $opendocRegex = "/^mimetype(application\/vnd\.oasis\.opendocument\.$types)/";
 901
 902                 $openxmlRegex = "/^\[Content_Types\].xml/";
 903
 904                 if ( preg_match( $opendocRegex, substr( $header, 30 ), $matches ) ) {
 905                         $mime = $matches[1];
 906                         $this->logger->info( __METHOD__ . ": detected $mime from ZIP archive\n" );
 907                 } elseif ( preg_match( $openxmlRegex, substr( $header, 30 ) ) ) {
 908                         $mime = "application/x-opc+zip";
 909                         # TODO: remove the block below, as soon as improveTypeFromExtension is used everywhere
 910                         if ( $ext !== true && $ext !== false ) {
 911                                 /** This is the mode used by getPropsFromPath
 912                                  * These MIME's are stored in the database, where we don't really want
 913                                  * x-opc+zip, because we use it only for internal purposes
 914                                  */
 915                                 if ( $this->isMatchingExtension( $ext, $mime ) ) {
 916                                         /* A known file extension for an OPC file,
 917                                          * find the proper mime type for that file extension
 918                                          */
 919                                         $mime = $this->guessTypesForExtension( $ext );
 920                                 } else {
 921                                         $mime = "application/zip";
 922                                 }
 923                         }
 924                         $this->logger->info( __METHOD__ .
 925                                 ": detected an Open Packaging Conventions archive: $mime\n" );
 926                 } elseif ( substr( $header, 0, 8 ) == "\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" &&
 927                         ( $headerpos = strpos( $tail, "PK\x03\x04" ) ) !== false &&
 928                         preg_match( $openxmlRegex, substr( $tail, $headerpos + 30 ) ) ) {
 929                         if ( substr( $header, 512, 4 ) == "\xEC\xA5\xC1\x00" ) {
 930                                 $mime = "application/msword";
 931                         }
 932                         switch ( substr( $header, 512, 6 ) ) {
 933                                 case "\xEC\xA5\xC1\x00\x0E\x00":
 934                                 case "\xEC\xA5\xC1\x00\x1C\x00":
 935                                 case "\xEC\xA5\xC1\x00\x43\x00":
 936                                         $mime = "application/vnd.ms-powerpoint";
 937                                         break;
 938                                 case "\xFD\xFF\xFF\xFF\x10\x00":
 939                                 case "\xFD\xFF\xFF\xFF\x1F\x00":
 940                                 case "\xFD\xFF\xFF\xFF\x22\x00":
 941                                 case "\xFD\xFF\xFF\xFF\x23\x00":
 942                                 case "\xFD\xFF\xFF\xFF\x28\x00":
 943                                 case "\xFD\xFF\xFF\xFF\x29\x00":
 944                                 case "\xFD\xFF\xFF\xFF\x10\x02":
 945                                 case "\xFD\xFF\xFF\xFF\x1F\x02":
 946                                 case "\xFD\xFF\xFF\xFF\x22\x02":
 947                                 case "\xFD\xFF\xFF\xFF\x23\x02":
 948                                 case "\xFD\xFF\xFF\xFF\x28\x02":
 949                                 case "\xFD\xFF\xFF\xFF\x29\x02":
 950                                         $mime = "application/vnd.msexcel";
 951                                         break;
 952                         }
 953
 954                         $this->logger->info( __METHOD__ .
 955                                 ": detected a MS Office document with OPC trailer\n" );
 956                 } else {
 957                         $this->logger->info( __METHOD__ . ": unable to identify type of ZIP archive\n" );
 958                 }
 959                 return $mime;
 960         }
 961
 962         /**
 963          * Detect the type of a Microsoft Compound Binary a.k.a. OLE file.
 964          * These are old style pre-ODF files such as .doc and .xls
 965          *
 966          * @param resource $handle An opened seekable file handle
 967          * @return string The detected MIME type
 968          */
 969         function detectMicrosoftBinaryType( $handle ) {
 970                 $info = MSCompoundFileReader::readHandle( $handle );
 971                 if ( !$info['valid'] ) {
 972                         $this->logger->info( __METHOD__ . ': invalid file format' );
 973                         return 'unknown/unknown';
 974                 }
 975                 if ( !$info['mime'] ) {
 976                         $this->logger->info( __METHOD__ . ": unrecognised document subtype" );
 977                         return 'unknown/unknown';
 978                 }
 979                 return $info['mime'];
 980         }
 981
 982         /**
 983          * Internal MIME type detection. Detection is done using the fileinfo
 984          * extension if it is available. It can be overriden by callback, which could
 985          * use an external program, for example. If detection fails and $ext is not false,
 986          * the MIME type is guessed from the file extension, using guessTypesForExtension.
 987          *
 988          * If the MIME type is still unknown, getimagesize is used to detect the
 989          * MIME type if the file is an image. If no MIME type can be determined,
 990          * this function returns 'unknown/unknown'.
 991          *
 992          * @param string $file The file to check
 993          * @param string|bool $ext The file extension, or true (default) to extract it from the filename.
 994          *   Set it to false to ignore the extension. DEPRECATED! Set to false, use
 995          *   improveTypeFromExtension($mime, $ext) later to improve MIME type.
 996          *
 997          * @return string The MIME type of $file
 998          */
 999         private function detectMimeType( $file, $ext = true ) {
1000                 /** @todo Make $ext default to false. Or better, remove it. */
1001                 if ( $ext ) {
1002                         $this->logger->info( __METHOD__ .
1003                                 ": WARNING: use of the \$ext parameter is deprecated. "
1004                                 . "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
1005                 }
1006
1007                 $callback = $this->detectCallback;
1008                 $m = null;
1009                 if ( $callback ) {
1010                         $m = $callback( $file );
1011                 } else {
1012                         $m = mime_content_type( $file );
1013                 }
1014
1015                 if ( $m ) {
1016                         # normalize
1017                         $m = preg_replace( '![;, ].*$!', '', $m ); # strip charset, etc
1018                         $m = trim( $m );
1019                         $m = strtolower( $m );
1020
1021                         if ( strpos( $m, 'unknown' ) !== false ) {
1022                                 $m = null;
1023                         } else {
1024                                 $this->logger->info( __METHOD__ . ": magic mime type of $file: $m\n" );
1025                                 return $m;
1026                         }
1027                 }
1028
1029                 // If desired, look at extension as a fallback.
1030                 if ( $ext === true ) {
1031                         $i = strrpos( $file, '.' );
1032                         $ext = strtolower( $i ? substr( $file, $i + 1 ) : '' );
1033                 }
1034                 if ( $ext ) {
1035                         if ( $this->isRecognizableExtension( $ext ) ) {
1036                                 $this->logger->info( __METHOD__ . ": refusing to guess mime type for .$ext file, "
1037                                         . "we should have recognized it\n" );
1038                         } else {
1039                                 $m = $this->guessTypesForExtension( $ext );
1040                                 if ( $m ) {
1041                                         $this->logger->info( __METHOD__ . ": extension mime type of $file: $m\n" );
1042                                         return $m;
1043                                 }
1044                         }
1045                 }
1046
1047                 // Unknown type
1048                 $this->logger->info( __METHOD__ . ": failed to guess mime type for $file!\n" );
1049                 return 'unknown/unknown';
1050         }
1051
1052         /**
1053          * Determine the media type code for a file, using its MIME type, name and
1054          * possibly its contents.
1055          *
1056          * This function relies on the findMediaType(), mapping extensions and MIME
1057          * types to media types.
1058          *
1059          * @todo analyse file if need be
1060          * @todo look at multiple extension, separately and together.
1061          *
1062          * @param string|null $path Full path to the image file, in case we have to look at the contents
1063          *        (if null, only the MIME type is used to determine the media type code).
1064          * @param string|null $mime MIME type. If null it will be guessed using guessMimeType.
1065          *
1066          * @return string A value to be used with the MEDIATYPE_xxx constants.
1067          */
1068         function getMediaType( $path = null, $mime = null ) {
1069                 if ( !$mime && !$path ) {
1070                         return MEDIATYPE_UNKNOWN;
1071                 }
1072
1073                 // If MIME type is unknown, guess it
1074                 if ( !$mime ) {
1075                         $mime = $this->guessMimeType( $path, false );
1076                 }
1077
1078                 // Special code for ogg - detect if it's video (theora),
1079                 // else label it as sound.
1080                 if ( $mime == 'application/ogg' && file_exists( $path ) ) {
1081                         // Read a chunk of the file
1082                         $f = fopen( $path, "rt" );
1083                         if ( !$f ) {
1084                                 return MEDIATYPE_UNKNOWN;
1085                         }
1086                         $head = fread( $f, 256 );
1087                         fclose( $f );
1088
1089                         $head = str_replace( 'ffmpeg2theora', '', strtolower( $head ) );
1090
1091                         // This is an UGLY HACK, file should be parsed correctly
1092                         if ( strpos( $head, 'theora' ) !== false ) {
1093                                 return MEDIATYPE_VIDEO;
1094                         } elseif ( strpos( $head, 'vorbis' ) !== false ) {
1095                                 return MEDIATYPE_AUDIO;
1096                         } elseif ( strpos( $head, 'flac' ) !== false ) {
1097                                 return MEDIATYPE_AUDIO;
1098                         } elseif ( strpos( $head, 'speex' ) !== false ) {
1099                                 return MEDIATYPE_AUDIO;
1100                         } elseif ( strpos( $head, 'opus' ) !== false ) {
1101                                 return MEDIATYPE_AUDIO;
1102                         } else {
1103                                 return MEDIATYPE_MULTIMEDIA;
1104                         }
1105                 }
1106
1107                 $type = null;
1108                 // Check for entry for full MIME type
1109                 if ( $mime ) {
1110                         $type = $this->findMediaType( $mime );
1111                         if ( $type !== MEDIATYPE_UNKNOWN ) {
1112                                 return $type;
1113                         }
1114                 }
1115
1116                 // Check for entry for file extension
1117                 if ( $path ) {
1118                         $i = strrpos( $path, '.' );
1119                         $e = strtolower( $i ? substr( $path, $i + 1 ) : '' );
1120
1121                         // TODO: look at multi-extension if this fails, parse from full path
1122                         $type = $this->findMediaType( '.' . $e );
1123                         if ( $type !== MEDIATYPE_UNKNOWN ) {
1124                                 return $type;
1125                         }
1126                 }
1127
1128                 // Check major MIME type
1129                 if ( $mime ) {
1130                         $i = strpos( $mime, '/' );
1131                         if ( $i !== false ) {
1132                                 $major = substr( $mime, 0, $i );
1133                                 $type = $this->findMediaType( $major );
1134                                 if ( $type !== MEDIATYPE_UNKNOWN ) {
1135                                         return $type;
1136                                 }
1137                         }
1138                 }
1139
1140                 if ( !$type ) {
1141                         $type = MEDIATYPE_UNKNOWN;
1142                 }
1143
1144                 return $type;
1145         }
1146
1147         /**
1148          * Returns a media code matching the given MIME type or file extension.
1149          * File extensions are represented by a string starting with a dot (.) to
1150          * distinguish them from MIME types.
1151          *
1152          * This function relies on the mapping defined by $this->mMediaTypes
1153          * @private
1154          * @param string $extMime
1155          * @return int|string
1156          */
1157         function findMediaType( $extMime ) {
1158                 if ( strpos( $extMime, '.' ) === 0 ) {
1159                         // If it's an extension, look up the MIME types
1160                         $m = $this->getTypesForExtension( substr( $extMime, 1 ) );
1161                         if ( !$m ) {
1162                                 return MEDIATYPE_UNKNOWN;
1163                         }
1164
1165                         $m = explode( ' ', $m );
1166                 } else {
1167                         // Normalize MIME type
1168                         if ( isset( $this->mimeTypeAliases[$extMime] ) ) {
1169                                 $extMime = $this->mimeTypeAliases[$extMime];
1170                         }
1171
1172                         $m = [ $extMime ];
1173                 }
1174
1175                 foreach ( $m as $mime ) {
1176                         foreach ( $this->mediaTypes as $type => $codes ) {
1177                                 if ( in_array( $mime, $codes, true ) ) {
1178                                         return $type;
1179                                 }
1180                         }
1181                 }
1182
1183                 return MEDIATYPE_UNKNOWN;
1184         }
1185
1186         /**
1187          * Returns an array of media types (MEDIATYPE_xxx constants)
1188          *
1189          * @return array
1190          */
1191         public function getMediaTypes() {
1192                 return array_keys( $this->mediaTypes );
1193         }
1194
1195         /**
1196          * Get the MIME types that various versions of Internet Explorer would
1197          * detect from a chunk of the content.
1198          *
1199          * @param string $fileName The file name (unused at present)
1200          * @param string $chunk The first 256 bytes of the file
1201          * @param string $proposed The MIME type proposed by the server
1202          * @return array
1203          */
1204         public function getIEMimeTypes( $fileName, $chunk, $proposed ) {
1205                 $ca = $this->getIEContentAnalyzer();
1206                 return $ca->getRealMimesFromData( $fileName, $chunk, $proposed );
1207         }
1208
1209         /**
1210          * Get a cached instance of IEContentAnalyzer
1211          *
1212          * @return IEContentAnalyzer
1213          */
1214         protected function getIEContentAnalyzer() {
1215                 if ( is_null( $this->IEAnalyzer ) ) {
1216                         $this->IEAnalyzer = new IEContentAnalyzer;
1217                 }
1218                 return $this->IEAnalyzer;
1219         }
1220 }