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