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 = isset( $params['initCallback'] )
 199                         ? $params['initCallback']
 200                         : null;
 201                 $this->detectCallback = isset( $params['detectCallback'] )
 202                         ? $params['detectCallback']
 203                         : null;
 204                 $this->guessCallback = isset( $params['guessCallback'] )
 205                         ? $params['guessCallback']
 206                         : null;
 207                 $this->extCallback = isset( $params['extCallback'] )
 208                         ? $params['extCallback']
 209                         : null;
 210                 $this->logger = isset( $params['logger'] )
 211                         ? $params['logger']
 212                         : new \Psr\Log\NullLogger();
 213
 214                 $this->loadFiles();
 215         }
 216
 217         protected function loadFiles() {
 218                 /**
 219                  *   --- load mime.types ---
 220                  */
 221
 222                 # Allow media handling extensions adding MIME-types and MIME-info
 223                 if ( $this->initCallback ) {
 224                         call_user_func( $this->initCallback, $this );
 225                 }
 226
 227                 $types = self::$wellKnownTypes;
 228
 229                 $mimeTypeFile = $this->typeFile;
 230                 if ( $mimeTypeFile ) {
 231                         if ( is_file( $mimeTypeFile ) && is_readable( $mimeTypeFile ) ) {
 232                                 $this->logger->info( __METHOD__ . ": loading mime types from $mimeTypeFile\n" );
 233                                 $types .= "\n";
 234                                 $types .= file_get_contents( $mimeTypeFile );
 235                         } else {
 236                                 $this->logger->info( __METHOD__ . ": can't load mime types from $mimeTypeFile\n" );
 237                         }
 238                 } else {
 239                         $this->logger->info( __METHOD__ .
 240                                 ": no mime types file defined, using built-ins only.\n" );
 241                 }
 242
 243                 $types .= "\n" . $this->extraTypes;
 244
 245                 $types = str_replace( [ "\r\n", "\n\r", "\n\n", "\r\r", "\r" ], "\n", $types );
 246                 $types = str_replace( "\t", " ", $types );
 247
 248                 $this->mimetoExt = [];
 249                 $this->mExtToMime = [];
 250
 251                 $lines = explode( "\n", $types );
 252                 foreach ( $lines as $s ) {
 253                         $s = trim( $s );
 254                         if ( empty( $s ) ) {
 255                                 continue;
 256                         }
 257                         if ( strpos( $s, '#' ) === 0 ) {
 258                                 continue;
 259                         }
 260
 261                         $s = strtolower( $s );
 262                         $i = strpos( $s, ' ' );
 263
 264                         if ( $i === false ) {
 265                                 continue;
 266                         }
 267
 268                         $mime = substr( $s, 0, $i );
 269                         $ext = trim( substr( $s, $i + 1 ) );
 270
 271                         if ( empty( $ext ) ) {
 272                                 continue;
 273                         }
 274
 275                         if ( !empty( $this->mimetoExt[$mime] ) ) {
 276                                 $this->mimetoExt[$mime] .= ' ' . $ext;
 277                         } else {
 278                                 $this->mimetoExt[$mime] = $ext;
 279                         }
 280
 281                         $extensions = explode( ' ', $ext );
 282
 283                         foreach ( $extensions as $e ) {
 284                                 $e = trim( $e );
 285                                 if ( empty( $e ) ) {
 286                                         continue;
 287                                 }
 288
 289                                 if ( !empty( $this->mExtToMime[$e] ) ) {
 290                                         $this->mExtToMime[$e] .= ' ' . $mime;
 291                                 } else {
 292                                         $this->mExtToMime[$e] = $mime;
 293                                 }
 294                         }
 295                 }
 296
 297                 /**
 298                  *   --- load mime.info ---
 299                  */
 300
 301                 $mimeInfoFile = $this->infoFile;
 302
 303                 $info = self::$wellKnownInfo;
 304
 305                 if ( $mimeInfoFile ) {
 306                         if ( is_file( $mimeInfoFile ) && is_readable( $mimeInfoFile ) ) {
 307                                 $this->logger->info( __METHOD__ . ": loading mime info from $mimeInfoFile\n" );
 308                                 $info .= "\n";
 309                                 $info .= file_get_contents( $mimeInfoFile );
 310                         } else {
 311                                 $this->logger->info( __METHOD__ . ": can't load mime info from $mimeInfoFile\n" );
 312                         }
 313                 } else {
 314                         $this->logger->info( __METHOD__ .
 315                                 ": no mime info file defined, using built-ins only.\n" );
 316                 }
 317
 318                 $info .= "\n" . $this->extraInfo;
 319
 320                 $info = str_replace( [ "\r\n", "\n\r", "\n\n", "\r\r", "\r" ], "\n", $info );
 321                 $info = str_replace( "\t", " ", $info );
 322
 323                 $this->mimeTypeAliases = [];
 324                 $this->mediaTypes = [];
 325
 326                 $lines = explode( "\n", $info );
 327                 foreach ( $lines as $s ) {
 328                         $s = trim( $s );
 329                         if ( empty( $s ) ) {
 330                                 continue;
 331                         }
 332                         if ( strpos( $s, '#' ) === 0 ) {
 333                                 continue;
 334                         }
 335
 336                         $s = strtolower( $s );
 337                         $i = strpos( $s, ' ' );
 338
 339                         if ( $i === false ) {
 340                                 continue;
 341                         }
 342
 343                         # print "processing MIME INFO line $s<br>";
 344
 345                         $match = [];
 346                         if ( preg_match( '!\[\s*(\w+)\s*\]!', $s, $match ) ) {
 347                                 $s = preg_replace( '!\[\s*(\w+)\s*\]!', '', $s );
 348                                 $mtype = trim( strtoupper( $match[1] ) );
 349                         } else {
 350                                 $mtype = MEDIATYPE_UNKNOWN;
 351                         }
 352
 353                         $m = explode( ' ', $s );
 354
 355                         if ( !isset( $this->mediaTypes[$mtype] ) ) {
 356                                 $this->mediaTypes[$mtype] = [];
 357                         }
 358
 359                         foreach ( $m as $mime ) {
 360                                 $mime = trim( $mime );
 361                                 if ( empty( $mime ) ) {
 362                                         continue;
 363                                 }
 364
 365                                 $this->mediaTypes[$mtype][] = $mime;
 366                         }
 367
 368                         if ( count( $m ) > 1 ) {
 369                                 $main = $m[0];
 370                                 $mCount = count( $m );
 371                                 for ( $i = 1; $i < $mCount; $i += 1 ) {
 372                                         $mime = $m[$i];
 373                                         $this->mimeTypeAliases[$mime] = $main;
 374                                 }
 375                         }
 376                 }
 377         }
 378
 379         public function setLogger( LoggerInterface $logger ) {
 380                 $this->logger = $logger;
 381         }
 382
 383         /**
 384          * Adds to the list mapping MIME to file extensions.
 385          * As an extension author, you are encouraged to submit patches to
 386          * MediaWiki's core to add new MIME types to mime.types.
 387          * @param string $types
 388          */
 389         public function addExtraTypes( $types ) {
 390                 $this->extraTypes .= "\n" . $types;
 391         }
 392
 393         /**
 394          * Adds to the list mapping MIME to media type.
 395          * As an extension author, you are encouraged to submit patches to
 396          * MediaWiki's core to add new MIME info to mime.info.
 397          * @param string $info
 398          */
 399         public function addExtraInfo( $info ) {
 400                 $this->extraInfo .= "\n" . $info;
 401         }
 402
 403         /**
 404          * Returns a list of file extensions for a given MIME type as a space
 405          * separated string or null if the MIME type was unrecognized. Resolves
 406          * MIME type aliases.
 407          *
 408          * @param string $mime
 409          * @return string|null
 410          */
 411         public function getExtensionsForType( $mime ) {
 412                 $mime = strtolower( $mime );
 413
 414                 // Check the mime-to-ext map
 415                 if ( isset( $this->mimetoExt[$mime] ) ) {
 416                         return $this->mimetoExt[$mime];
 417                 }
 418
 419                 // Resolve the MIME type to the canonical type
 420                 if ( isset( $this->mimeTypeAliases[$mime] ) ) {
 421                         $mime = $this->mimeTypeAliases[$mime];
 422                         if ( isset( $this->mimetoExt[$mime] ) ) {
 423                                 return $this->mimetoExt[$mime];
 424                         }
 425                 }
 426
 427                 return null;
 428         }
 429
 430         /**
 431          * Returns a list of MIME types for a given file extension as a space
 432          * separated string or null if the extension was unrecognized.
 433          *
 434          * @param string $ext
 435          * @return string|null
 436          */
 437         public function getTypesForExtension( $ext ) {
 438                 $ext = strtolower( $ext );
 439
 440                 $r = isset( $this->mExtToMime[$ext] ) ? $this->mExtToMime[$ext] : null;
 441                 return $r;
 442         }
 443
 444         /**
 445          * Returns a single MIME type for a given file extension or null if unknown.
 446          * This is always the first type from the list returned by getTypesForExtension($ext).
 447          *
 448          * @param string $ext
 449          * @return string|null
 450          */
 451         public function guessTypesForExtension( $ext ) {
 452                 $m = $this->getTypesForExtension( $ext );
 453                 if ( is_null( $m ) ) {
 454                         return null;
 455                 }
 456
 457                 // TODO: Check if this is needed; strtok( $m, ' ' ) should be sufficient
 458                 $m = trim( $m );
 459                 $m = preg_replace( '/\s.*$/', '', $m );
 460
 461                 return $m;
 462         }
 463
 464         /**
 465          * Tests if the extension matches the given MIME type. Returns true if a
 466          * match was found, null if the MIME type is unknown, and false if the
 467          * MIME type is known but no matches where found.
 468          *
 469          * @param string $extension
 470          * @param string $mime
 471          * @return bool|null
 472          */
 473         public function isMatchingExtension( $extension, $mime ) {
 474                 $ext = $this->getExtensionsForType( $mime );
 475
 476                 if ( !$ext ) {
 477                         return null; // Unknown MIME type
 478                 }
 479
 480                 $ext = explode( ' ', $ext );
 481
 482                 $extension = strtolower( $extension );
 483                 return in_array( $extension, $ext );
 484         }
 485
 486         /**
 487          * Returns true if the MIME type is known to represent an image format
 488          * supported by the PHP GD library.
 489          *
 490          * @param string $mime
 491          *
 492          * @return bool
 493          */
 494         public function isPHPImageType( $mime ) {
 495                 // As defined by imagegetsize and image_type_to_mime
 496                 static $types = [
 497                         'image/gif', 'image/jpeg', 'image/png',
 498                         'image/x-bmp', 'image/xbm', 'image/tiff',
 499                         'image/jp2', 'image/jpeg2000', 'image/iff',
 500                         'image/xbm', 'image/x-xbitmap',
 501                         'image/vnd.wap.wbmp', 'image/vnd.xiff',
 502                         'image/x-photoshop',
 503                         'application/x-shockwave-flash',
 504                 ];
 505
 506                 return in_array( $mime, $types );
 507         }
 508
 509         /**
 510          * Returns true if the extension represents a type which can
 511          * be reliably detected from its content. Use this to determine
 512          * whether strict content checks should be applied to reject
 513          * invalid uploads; if we can't identify the type we won't
 514          * be able to say if it's invalid.
 515          *
 516          * @todo Be more accurate when using fancy MIME detector plugins;
 517          *       right now this is the bare minimum getimagesize() list.
 518          * @param string $extension
 519          * @return bool
 520          */
 521         function isRecognizableExtension( $extension ) {
 522                 static $types = [
 523                         // Types recognized by getimagesize()
 524                         'gif', 'jpeg', 'jpg', 'png', 'swf', 'psd',
 525                         'bmp', 'tiff', 'tif', 'jpc', 'jp2',
 526                         'jpx', 'jb2', 'swc', 'iff', 'wbmp',
 527                         'xbm',
 528
 529                         // Formats we recognize magic numbers for
 530                         'djvu', 'ogx', 'ogg', 'ogv', 'oga', 'spx', 'opus',
 531                         'mid', 'pdf', 'wmf', 'xcf', 'webm', 'mkv', 'mka',
 532                         'webp',
 533
 534                         // XML formats we sure hope we recognize reliably
 535                         'svg',
 536                 ];
 537                 return in_array( strtolower( $extension ), $types );
 538         }
 539
 540         /**
 541          * Improves a MIME type using the file extension. Some file formats are very generic,
 542          * so their MIME type is not very meaningful. A more useful MIME type can be derived
 543          * by looking at the file extension. Typically, this method would be called on the
 544          * result of guessMimeType().
 545          *
 546          * @param string $mime The MIME type, typically guessed from a file's content.
 547          * @param string $ext The file extension, as taken from the file name
 548          *
 549          * @return string The MIME type
 550          */
 551         public function improveTypeFromExtension( $mime, $ext ) {
 552                 if ( $mime === 'unknown/unknown' ) {
 553                         if ( $this->isRecognizableExtension( $ext ) ) {
 554                                 $this->logger->info( __METHOD__ . ': refusing to guess mime type for .' .
 555                                         "$ext file, we should have recognized it\n" );
 556                         } else {
 557                                 // Not something we can detect, so simply
 558                                 // trust the file extension
 559                                 $mime = $this->guessTypesForExtension( $ext );
 560                         }
 561                 } elseif ( $mime === 'application/x-opc+zip' ) {
 562                         if ( $this->isMatchingExtension( $ext, $mime ) ) {
 563                                 // A known file extension for an OPC file,
 564                                 // find the proper MIME type for that file extension
 565                                 $mime = $this->guessTypesForExtension( $ext );
 566                         } else {
 567                                 $this->logger->info( __METHOD__ .
 568                                         ": refusing to guess better type for $mime file, " .
 569                                         ".$ext is not a known OPC extension.\n" );
 570                                 $mime = 'application/zip';
 571                         }
 572                 } elseif ( $mime === 'text/plain' && $this->findMediaType( ".$ext" ) === MEDIATYPE_TEXT ) {
 573                         // Textual types are sometimes not recognized properly.
 574                         // If detected as text/plain, and has an extension which is textual
 575                         // improve to the extension's type. For example, csv and json are often
 576                         // misdetected as text/plain.
 577                         $mime = $this->guessTypesForExtension( $ext );
 578                 }
 579
 580                 # Media handling extensions can improve the MIME detected
 581                 $callback = $this->extCallback;
 582                 if ( $callback ) {
 583                         $callback( $this, $ext, $mime /* by reference */ );
 584                 }
 585
 586                 if ( isset( $this->mimeTypeAliases[$mime] ) ) {
 587                         $mime = $this->mimeTypeAliases[$mime];
 588                 }
 589
 590                 $this->logger->info( __METHOD__ . ": improved mime type for .$ext: $mime\n" );
 591                 return $mime;
 592         }
 593
 594         /**
 595          * MIME type detection. This uses detectMimeType to detect the MIME type
 596          * of the file, but applies additional checks to determine some well known
 597          * file formats that may be missed or misinterpreted by the default MIME
 598          * detection (namely XML based formats like XHTML or SVG, as well as ZIP
 599          * based formats like OPC/ODF files).
 600          *
 601          * @param string $file The file to check
 602          * @param string|bool $ext The file extension, or true (default) to extract
 603          *   it from the filename. Set it to false to ignore the extension. DEPRECATED!
 604          *   Set to false, use improveTypeFromExtension($mime, $ext) later to improve MIME type.
 605          *
 606          * @return string The MIME type of $file
 607          */
 608         public function guessMimeType( $file, $ext = true ) {
 609                 if ( $ext ) { // TODO: make $ext default to false. Or better, remove it.
 610                         $this->logger->info( __METHOD__ .
 611                                 ": WARNING: use of the \$ext parameter is deprecated. " .
 612                                 "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 613                 }
 614
 615                 $mime = $this->doGuessMimeType( $file, $ext );
 616
 617                 if ( !$mime ) {
 618                         $this->logger->info( __METHOD__ .
 619                                 ": internal type detection failed for $file (.$ext)...\n" );
 620                         $mime = $this->detectMimeType( $file, $ext );
 621                 }
 622
 623                 if ( isset( $this->mimeTypeAliases[$mime] ) ) {
 624                         $mime = $this->mimeTypeAliases[$mime];
 625                 }
 626
 627                 $this->logger->info( __METHOD__ . ": guessed mime type of $file: $mime\n" );
 628                 return $mime;
 629         }
 630
 631         /**
 632          * Guess the MIME type from the file contents.
 633          *
 634          * @todo Remove $ext param
 635          *
 636          * @param string $file
 637          * @param mixed $ext
 638          * @return bool|string
 639          * @throws UnexpectedValueException
 640          */
 641         private function doGuessMimeType( $file, $ext ) {
 642                 // Read a chunk of the file
 643                 MediaWiki\suppressWarnings();
 644                 $f = fopen( $file, 'rb' );
 645                 MediaWiki\restoreWarnings();
 646
 647                 if ( !$f ) {
 648                         return 'unknown/unknown';
 649                 }
 650
 651                 $fsize = filesize( $file );
 652                 if ( $fsize === false ) {
 653                         return 'unknown/unknown';
 654                 }
 655
 656                 $head = fread( $f, 1024 );
 657                 $tailLength = min( 65558, $fsize ); // 65558 = maximum size of a zip EOCDR
 658                 if ( fseek( $f, -1 * $tailLength, SEEK_END ) === -1 ) {
 659                         throw new UnexpectedValueException(
 660                                 "Seeking $tailLength bytes from EOF failed in " . __METHOD__ );
 661                 }
 662                 $tail = $tailLength ? fread( $f, $tailLength ) : '';
 663                 fclose( $f );
 664
 665                 $this->logger->info( __METHOD__ .
 666                         ": analyzing head and tail of $file for magic numbers.\n" );
 667
 668                 // Hardcode a few magic number checks...
 669                 $headers = [
 670                         // Multimedia...
 671                         'MThd'             => 'audio/midi',
 672                         'OggS'             => 'application/ogg',
 673
 674                         // Image formats...
 675                         // Note that WMF may have a bare header, no magic number.
 676                         "\x01\x00\x09\x00" => 'application/x-msmetafile', // Possibly prone to false positives?
 677                         "\xd7\xcd\xc6\x9a" => 'application/x-msmetafile',
 678                         '%PDF'             => 'application/pdf',
 679                         'gimp xcf'         => 'image/x-xcf',
 680
 681                         // Some forbidden fruit...
 682                         'MZ'               => 'application/octet-stream', // DOS/Windows executable
 683                         "\xca\xfe\xba\xbe" => 'application/octet-stream', // Mach-O binary
 684                         "\x7fELF"          => 'application/octet-stream', // ELF binary
 685                 ];
 686
 687                 foreach ( $headers as $magic => $candidate ) {
 688                         if ( strncmp( $head, $magic, strlen( $magic ) ) == 0 ) {
 689                                 $this->logger->info( __METHOD__ .
 690                                         ": magic header in $file recognized as $candidate\n" );
 691                                 return $candidate;
 692                         }
 693                 }
 694
 695                 /* Look for WebM and Matroska files */
 696                 if ( strncmp( $head, pack( "C4", 0x1a, 0x45, 0xdf, 0xa3 ), 4 ) == 0 ) {
 697                         $doctype = strpos( $head, "\x42\x82" );
 698                         if ( $doctype ) {
 699                                 // Next byte is datasize, then data (sizes larger than 1 byte are stupid muxers)
 700                                 $data = substr( $head, $doctype + 3, 8 );
 701                                 if ( strncmp( $data, "matroska", 8 ) == 0 ) {
 702                                         $this->logger->info( __METHOD__ . ": recognized file as video/x-matroska\n" );
 703                                         return "video/x-matroska";
 704                                 } elseif ( strncmp( $data, "webm", 4 ) == 0 ) {
 705                                         $this->logger->info( __METHOD__ . ": recognized file as video/webm\n" );
 706                                         return "video/webm";
 707                                 }
 708                         }
 709                         $this->logger->info( __METHOD__ . ": unknown EBML file\n" );
 710                         return "unknown/unknown";
 711                 }
 712
 713                 /* Look for WebP */
 714                 if ( strncmp( $head, "RIFF", 4 ) == 0 &&
 715                         strncmp( substr( $head, 8, 7 ), "WEBPVP8", 7 ) == 0
 716                 ) {
 717                         $this->logger->info( __METHOD__ . ": recognized file as image/webp\n" );
 718                         return "image/webp";
 719                 }
 720
 721                 /**
 722                  * Look for PHP.  Check for this before HTML/XML...  Warning: this is a
 723                  * heuristic, and won't match a file with a lot of non-PHP before.  It
 724                  * will also match text files which could be PHP. :)
 725                  *
 726                  * @todo FIXME: For this reason, the check is probably useless -- an attacker
 727                  * could almost certainly just pad the file with a lot of nonsense to
 728                  * circumvent the check in any case where it would be a security
 729                  * problem.  On the other hand, it causes harmful false positives (bug
 730                  * 16583).  The heuristic has been cut down to exclude three-character
 731                  * strings like "<? ", but should it be axed completely?
 732                  */
 733                 if ( ( strpos( $head, '<?php' ) !== false ) ||
 734                         ( strpos( $head, "<\x00?\x00p\x00h\x00p" ) !== false ) ||
 735                         ( strpos( $head, "<\x00?\x00 " ) !== false ) ||
 736                         ( strpos( $head, "<\x00?\x00\n" ) !== false ) ||
 737                         ( strpos( $head, "<\x00?\x00\t" ) !== false ) ||
 738                         ( strpos( $head, "<\x00?\x00=" ) !== false ) ) {
 739
 740                         $this->logger->info( __METHOD__ . ": recognized $file as application/x-php\n" );
 741                         return 'application/x-php';
 742                 }
 743
 744                 /**
 745                  * look for XML formats (XHTML and SVG)
 746                  */
 747                 $xml = new XmlTypeCheck( $file );
 748                 if ( $xml->wellFormed ) {
 749                         $xmlTypes = $this->xmlTypes;
 750                         if ( isset( $xmlTypes[$xml->getRootElement()] ) ) {
 751                                 return $xmlTypes[$xml->getRootElement()];
 752                         } else {
 753                                 return 'application/xml';
 754                         }
 755                 }
 756
 757                 /**
 758                  * look for shell scripts
 759                  */
 760                 $script_type = null;
 761
 762                 # detect by shebang
 763                 if ( substr( $head, 0, 2 ) == "#!" ) {
 764                         $script_type = "ASCII";
 765                 } elseif ( substr( $head, 0, 5 ) == "\xef\xbb\xbf#!" ) {
 766                         $script_type = "UTF-8";
 767                 } elseif ( substr( $head, 0, 7 ) == "\xfe\xff\x00#\x00!" ) {
 768                         $script_type = "UTF-16BE";
 769                 } elseif ( substr( $head, 0, 7 ) == "\xff\xfe#\x00!" ) {
 770                         $script_type = "UTF-16LE";
 771                 }
 772
 773                 if ( $script_type ) {
 774                         if ( $script_type !== "UTF-8" && $script_type !== "ASCII" ) {
 775                                 // Quick and dirty fold down to ASCII!
 776                                 $pack = [ 'UTF-16BE' => 'n*', 'UTF-16LE' => 'v*' ];
 777                                 $chars = unpack( $pack[$script_type], substr( $head, 2 ) );
 778                                 $head = '';
 779                                 foreach ( $chars as $codepoint ) {
 780                                         if ( $codepoint < 128 ) {
 781                                                 $head .= chr( $codepoint );
 782                                         } else {
 783                                                 $head .= '?';
 784                                         }
 785                                 }
 786                         }
 787
 788                         $match = [];
 789
 790                         if ( preg_match( '%/?([^\s]+/)(\w+)%', $head, $match ) ) {
 791                                 $mime = "application/x-{$match[2]}";
 792                                 $this->logger->info( __METHOD__ . ": shell script recognized as $mime\n" );
 793                                 return $mime;
 794                         }
 795                 }
 796
 797                 // Check for ZIP variants (before getimagesize)
 798                 if ( strpos( $tail, "PK\x05\x06" ) !== false ) {
 799                         $this->logger->info( __METHOD__ . ": ZIP header present in $file\n" );
 800                         return $this->detectZipType( $head, $tail, $ext );
 801                 }
 802
 803                 MediaWiki\suppressWarnings();
 804                 $gis = getimagesize( $file );
 805                 MediaWiki\restoreWarnings();
 806
 807                 if ( $gis && isset( $gis['mime'] ) ) {
 808                         $mime = $gis['mime'];
 809                         $this->logger->info( __METHOD__ . ": getimagesize detected $file as $mime\n" );
 810                         return $mime;
 811                 }
 812
 813                 # Media handling extensions can guess the MIME by content
 814                 # It's intentionally here so that if core is wrong about a type (false positive),
 815                 # people will hopefully nag and submit patches :)
 816                 $mime = false;
 817                 # Some strings by reference for performance - assuming well-behaved hooks
 818                 $callback = $this->guessCallback;
 819                 if ( $callback ) {
 820                         $callback( $this, $head, $tail, $file, $mime /* by reference */ );
 821                 };
 822
 823                 return $mime;
 824         }
 825
 826         /**
 827          * Detect application-specific file type of a given ZIP file from its
 828          * header data.  Currently works for OpenDocument and OpenXML types...
 829          * If can't tell, returns 'application/zip'.
 830          *
 831          * @param string $header Some reasonably-sized chunk of file header
 832          * @param string|null $tail The tail of the file
 833          * @param string|bool $ext The file extension, or true to extract it from the filename.
 834          *   Set it to false (default) to ignore the extension. DEPRECATED! Set to false,
 835          *   use improveTypeFromExtension($mime, $ext) later to improve MIME type.
 836          *
 837          * @return string
 838          */
 839         function detectZipType( $header, $tail = null, $ext = false ) {
 840                 if ( $ext ) { # TODO: remove $ext param
 841                         $this->logger->info( __METHOD__ .
 842                                 ": WARNING: use of the \$ext parameter is deprecated. " .
 843                                 "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 844                 }
 845
 846                 $mime = 'application/zip';
 847                 $opendocTypes = [
 848                         'chart-template',
 849                         'chart',
 850                         'formula-template',
 851                         'formula',
 852                         'graphics-template',
 853                         'graphics',
 854                         'image-template',
 855                         'image',
 856                         'presentation-template',
 857                         'presentation',
 858                         'spreadsheet-template',
 859                         'spreadsheet',
 860                         'text-template',
 861                         'text-master',
 862                         'text-web',
 863                         'text' ];
 864
 865                 // https://lists.oasis-open.org/archives/office/200505/msg00006.html
 866                 $types = '(?:' . implode( '|', $opendocTypes ) . ')';
 867                 $opendocRegex = "/^mimetype(application\/vnd\.oasis\.opendocument\.$types)/";
 868
 869                 $openxmlRegex = "/^\[Content_Types\].xml/";
 870
 871                 if ( preg_match( $opendocRegex, substr( $header, 30 ), $matches ) ) {
 872                         $mime = $matches[1];
 873                         $this->logger->info( __METHOD__ . ": detected $mime from ZIP archive\n" );
 874                 } elseif ( preg_match( $openxmlRegex, substr( $header, 30 ) ) ) {
 875                         $mime = "application/x-opc+zip";
 876                         # TODO: remove the block below, as soon as improveTypeFromExtension is used everywhere
 877                         if ( $ext !== true && $ext !== false ) {
 878                                 /** This is the mode used by getPropsFromPath
 879                                  * These MIME's are stored in the database, where we don't really want
 880                                  * x-opc+zip, because we use it only for internal purposes
 881                                  */
 882                                 if ( $this->isMatchingExtension( $ext, $mime ) ) {
 883                                         /* A known file extension for an OPC file,
 884                                          * find the proper mime type for that file extension
 885                                          */
 886                                         $mime = $this->guessTypesForExtension( $ext );
 887                                 } else {
 888                                         $mime = "application/zip";
 889                                 }
 890                         }
 891                         $this->logger->info( __METHOD__ .
 892                                 ": detected an Open Packaging Conventions archive: $mime\n" );
 893                 } elseif ( substr( $header, 0, 8 ) == "\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" &&
 894                         ( $headerpos = strpos( $tail, "PK\x03\x04" ) ) !== false &&
 895                         preg_match( $openxmlRegex, substr( $tail, $headerpos + 30 ) ) ) {
 896                         if ( substr( $header, 512, 4 ) == "\xEC\xA5\xC1\x00" ) {
 897                                 $mime = "application/msword";
 898                         }
 899                         switch ( substr( $header, 512, 6 ) ) {
 900                                 case "\xEC\xA5\xC1\x00\x0E\x00":
 901                                 case "\xEC\xA5\xC1\x00\x1C\x00":
 902                                 case "\xEC\xA5\xC1\x00\x43\x00":
 903                                         $mime = "application/vnd.ms-powerpoint";
 904                                         break;
 905                                 case "\xFD\xFF\xFF\xFF\x10\x00":
 906                                 case "\xFD\xFF\xFF\xFF\x1F\x00":
 907                                 case "\xFD\xFF\xFF\xFF\x22\x00":
 908                                 case "\xFD\xFF\xFF\xFF\x23\x00":
 909                                 case "\xFD\xFF\xFF\xFF\x28\x00":
 910                                 case "\xFD\xFF\xFF\xFF\x29\x00":
 911                                 case "\xFD\xFF\xFF\xFF\x10\x02":
 912                                 case "\xFD\xFF\xFF\xFF\x1F\x02":
 913                                 case "\xFD\xFF\xFF\xFF\x22\x02":
 914                                 case "\xFD\xFF\xFF\xFF\x23\x02":
 915                                 case "\xFD\xFF\xFF\xFF\x28\x02":
 916                                 case "\xFD\xFF\xFF\xFF\x29\x02":
 917                                         $mime = "application/vnd.msexcel";
 918                                         break;
 919                         }
 920
 921                         $this->logger->info( __METHOD__ .
 922                                 ": detected a MS Office document with OPC trailer\n" );
 923                 } else {
 924                         $this->logger->info( __METHOD__ . ": unable to identify type of ZIP archive\n" );
 925                 }
 926                 return $mime;
 927         }
 928
 929         /**
 930          * Internal MIME type detection. Detection is done using the fileinfo
 931          * extension if it is available. It can be overriden by callback, which could
 932          * use an external program, for example. If detection fails and $ext is not false,
 933          * the MIME type is guessed from the file extension, using guessTypesForExtension.
 934          *
 935          * If the MIME type is still unknown, getimagesize is used to detect the
 936          * MIME type if the file is an image. If no MIME type can be determined,
 937          * this function returns 'unknown/unknown'.
 938          *
 939          * @param string $file The file to check
 940          * @param string|bool $ext The file extension, or true (default) to extract it from the filename.
 941          *   Set it to false to ignore the extension. DEPRECATED! Set to false, use
 942          *   improveTypeFromExtension($mime, $ext) later to improve MIME type.
 943          *
 944          * @return string The MIME type of $file
 945          */
 946         private function detectMimeType( $file, $ext = true ) {
 947                 /** @todo Make $ext default to false. Or better, remove it. */
 948                 if ( $ext ) {
 949                         $this->logger->info( __METHOD__ .
 950                                 ": WARNING: use of the \$ext parameter is deprecated. "
 951                                 . "Use improveTypeFromExtension(\$mime, \$ext) instead.\n" );
 952                 }
 953
 954                 $callback = $this->detectCallback;
 955                 $m = null;
 956                 if ( $callback ) {
 957                         $m = $callback( $file );
 958                 } elseif ( function_exists( "finfo_open" ) && function_exists( "finfo_file" ) ) {
 959                         $mime_magic_resource = finfo_open( FILEINFO_MIME );
 960
 961                         if ( $mime_magic_resource ) {
 962                                 $m = finfo_file( $mime_magic_resource, $file );
 963                                 finfo_close( $mime_magic_resource );
 964                         } else {
 965                                 $this->logger->info( __METHOD__ .
 966                                         ": finfo_open failed on " . FILEINFO_MIME . "!\n" );
 967                         }
 968                 } else {
 969                         $this->logger->info( __METHOD__ . ": no magic mime detector found!\n" );
 970                 }
 971
 972                 if ( $m ) {
 973                         # normalize
 974                         $m = preg_replace( '![;, ].*$!', '', $m ); # strip charset, etc
 975                         $m = trim( $m );
 976                         $m = strtolower( $m );
 977
 978                         if ( strpos( $m, 'unknown' ) !== false ) {
 979                                 $m = null;
 980                         } else {
 981                                 $this->logger->info( __METHOD__ . ": magic mime type of $file: $m\n" );
 982                                 return $m;
 983                         }
 984                 }
 985
 986                 // If desired, look at extension as a fallback.
 987                 if ( $ext === true ) {
 988                         $i = strrpos( $file, '.' );
 989                         $ext = strtolower( $i ? substr( $file, $i + 1 ) : '' );
 990                 }
 991                 if ( $ext ) {
 992                         if ( $this->isRecognizableExtension( $ext ) ) {
 993                                 $this->logger->info( __METHOD__ . ": refusing to guess mime type for .$ext file, "
 994                                         . "we should have recognized it\n" );
 995                         } else {
 996                                 $m = $this->guessTypesForExtension( $ext );
 997                                 if ( $m ) {
 998                                         $this->logger->info( __METHOD__ . ": extension mime type of $file: $m\n" );
 999                                         return $m;
1000                                 }
1001                         }
1002                 }
1003
1004                 // Unknown type
1005                 $this->logger->info( __METHOD__ . ": failed to guess mime type for $file!\n" );
1006                 return 'unknown/unknown';
1007         }
1008
1009         /**
1010          * Determine the media type code for a file, using its MIME type, name and
1011          * possibly its contents.
1012          *
1013          * This function relies on the findMediaType(), mapping extensions and MIME
1014          * types to media types.
1015          *
1016          * @todo analyse file if need be
1017          * @todo look at multiple extension, separately and together.
1018          *
1019          * @param string $path Full path to the image file, in case we have to look at the contents
1020          *        (if null, only the MIME type is used to determine the media type code).
1021          * @param string $mime MIME type. If null it will be guessed using guessMimeType.
1022          *
1023          * @return string A value to be used with the MEDIATYPE_xxx constants.
1024          */
1025         function getMediaType( $path = null, $mime = null ) {
1026                 if ( !$mime && !$path ) {
1027                         return MEDIATYPE_UNKNOWN;
1028                 }
1029
1030                 // If MIME type is unknown, guess it
1031                 if ( !$mime ) {
1032                         $mime = $this->guessMimeType( $path, false );
1033                 }
1034
1035                 // Special code for ogg - detect if it's video (theora),
1036                 // else label it as sound.
1037                 if ( $mime == 'application/ogg' && file_exists( $path ) ) {
1038
1039                         // Read a chunk of the file
1040                         $f = fopen( $path, "rt" );
1041                         if ( !$f ) {
1042                                 return MEDIATYPE_UNKNOWN;
1043                         }
1044                         $head = fread( $f, 256 );
1045                         fclose( $f );
1046
1047                         $head = str_replace( 'ffmpeg2theora', '', strtolower( $head ) );
1048
1049                         // This is an UGLY HACK, file should be parsed correctly
1050                         if ( strpos( $head, 'theora' ) !== false ) {
1051                                 return MEDIATYPE_VIDEO;
1052                         } elseif ( strpos( $head, 'vorbis' ) !== false ) {
1053                                 return MEDIATYPE_AUDIO;
1054                         } elseif ( strpos( $head, 'flac' ) !== false ) {
1055                                 return MEDIATYPE_AUDIO;
1056                         } elseif ( strpos( $head, 'speex' ) !== false ) {
1057                                 return MEDIATYPE_AUDIO;
1058                         } elseif ( strpos( $head, 'opus' ) !== false ) {
1059                                 return MEDIATYPE_AUDIO;
1060                         } else {
1061                                 return MEDIATYPE_MULTIMEDIA;
1062                         }
1063                 }
1064
1065                 $type = null;
1066                 // Check for entry for full MIME type
1067                 if ( $mime ) {
1068                         $type = $this->findMediaType( $mime );
1069                         if ( $type !== MEDIATYPE_UNKNOWN ) {
1070                                 return $type;
1071                         }
1072                 }
1073
1074                 // Check for entry for file extension
1075                 if ( $path ) {
1076                         $i = strrpos( $path, '.' );
1077                         $e = strtolower( $i ? substr( $path, $i + 1 ) : '' );
1078
1079                         // TODO: look at multi-extension if this fails, parse from full path
1080                         $type = $this->findMediaType( '.' . $e );
1081                         if ( $type !== MEDIATYPE_UNKNOWN ) {
1082                                 return $type;
1083                         }
1084                 }
1085
1086                 // Check major MIME type
1087                 if ( $mime ) {
1088                         $i = strpos( $mime, '/' );
1089                         if ( $i !== false ) {
1090                                 $major = substr( $mime, 0, $i );
1091                                 $type = $this->findMediaType( $major );
1092                                 if ( $type !== MEDIATYPE_UNKNOWN ) {
1093                                         return $type;
1094                                 }
1095                         }
1096                 }
1097
1098                 if ( !$type ) {
1099                         $type = MEDIATYPE_UNKNOWN;
1100                 }
1101
1102                 return $type;
1103         }
1104
1105         /**
1106          * Returns a media code matching the given MIME type or file extension.
1107          * File extensions are represented by a string starting with a dot (.) to
1108          * distinguish them from MIME types.
1109          *
1110          * This function relies on the mapping defined by $this->mMediaTypes
1111          * @access private
1112          * @param string $extMime
1113          * @return int|string
1114          */
1115         function findMediaType( $extMime ) {
1116                 if ( strpos( $extMime, '.' ) === 0 ) {
1117                         // If it's an extension, look up the MIME types
1118                         $m = $this->getTypesForExtension( substr( $extMime, 1 ) );
1119                         if ( !$m ) {
1120                                 return MEDIATYPE_UNKNOWN;
1121                         }
1122
1123                         $m = explode( ' ', $m );
1124                 } else {
1125                         // Normalize MIME type
1126                         if ( isset( $this->mimeTypeAliases[$extMime] ) ) {
1127                                 $extMime = $this->mimeTypeAliases[$extMime];
1128                         }
1129
1130                         $m = [ $extMime ];
1131                 }
1132
1133                 foreach ( $m as $mime ) {
1134                         foreach ( $this->mediaTypes as $type => $codes ) {
1135                                 if ( in_array( $mime, $codes, true ) ) {
1136                                         return $type;
1137                                 }
1138                         }
1139                 }
1140
1141                 return MEDIATYPE_UNKNOWN;
1142         }
1143
1144         /**
1145          * Get the MIME types that various versions of Internet Explorer would
1146          * detect from a chunk of the content.
1147          *
1148          * @param string $fileName The file name (unused at present)
1149          * @param string $chunk The first 256 bytes of the file
1150          * @param string $proposed The MIME type proposed by the server
1151          * @return array
1152          */
1153         public function getIEMimeTypes( $fileName, $chunk, $proposed ) {
1154                 $ca = $this->getIEContentAnalyzer();
1155                 return $ca->getRealMimesFromData( $fileName, $chunk, $proposed );
1156         }
1157
1158         /**
1159          * Get a cached instance of IEContentAnalyzer
1160          *
1161          * @return IEContentAnalyzer
1162          */
1163         protected function getIEContentAnalyzer() {
1164                 if ( is_null( $this->IEAnalyzer ) ) {
1165                         $this->IEAnalyzer = new IEContentAnalyzer;
1166                 }
1167                 return $this->IEAnalyzer;
1168         }
1169 }