includes/upload/UploadBase.php

   1 <?php
   2 /**
   3  * @file
   4  * @ingroup upload
   5  *
   6  * UploadBase and subclasses are the backend of MediaWiki's file uploads.
   7  * The frontends are formed by ApiUpload and SpecialUpload.
   8  *
   9  * See also includes/docs/upload.txt
  10  *
  11  * @author Brion Vibber
  12  * @author Bryan Tong Minh
  13  * @author Michael Dale
  14  */
  15
  16 abstract class UploadBase {
  17         protected $mTempPath;
  18         protected $mDesiredDestName, $mDestName, $mRemoveTempFile, $mSourceType;
  19         protected $mTitle = false, $mTitleError = 0;
  20         protected $mFilteredName, $mFinalExtension;
  21         protected $mLocalFile;
  22
  23         const SUCCESS = 0;
  24         const OK = 0;
  25         const EMPTY_FILE = 3;
  26         const MIN_LENGTH_PARTNAME = 4;
  27         const ILLEGAL_FILENAME = 5;
  28         const OVERWRITE_EXISTING_FILE = 7;
  29         const FILETYPE_MISSING = 8;
  30         const FILETYPE_BADTYPE = 9;
  31         const VERIFICATION_ERROR = 10;
  32         const UPLOAD_VERIFICATION_ERROR = 11;
  33         const HOOK_ABORTED = 11;
  34
  35         const SESSION_VERSION = 2;
  36
  37         /**
  38          * Returns true if uploads are enabled.
  39          * Can be override by subclasses.
  40          */
  41         public static function isEnabled() {
  42                 global $wgEnableUploads;
  43                 if ( !$wgEnableUploads ) {
  44                         return false;
  45                 }
  46
  47                 # Check php's file_uploads setting
  48                 if( !wfIniGetBool( 'file_uploads' ) ) {
  49                         return false;
  50                 }
  51                 return true;
  52         }
  53
  54         /**
  55          * Returns true if the user can use this upload module or else a string
  56          * identifying the missing permission.
  57          * Can be overriden by subclasses.
  58          */
  59         public static function isAllowed( $user ) {
  60                 if( !$user->isAllowed( 'upload' ) ) {
  61                         return 'upload';
  62                 }
  63                 return true;
  64         }
  65
  66         // Upload handlers. Should probably just be a global.
  67         static $uploadHandlers = array( 'Stash', 'File', 'Url' );
  68
  69         /**
  70          * Create a form of UploadBase depending on wpSourceType and initializes it
  71          */
  72         public static function createFromRequest( &$request, $type = null ) {
  73                 $type = $type ? $type : $request->getVal( 'wpSourceType', 'File' );
  74
  75                 if( !$type ) {
  76                         return null;
  77                 }
  78
  79                 // Get the upload class
  80                 $type = ucfirst( $type );
  81
  82                 // Give hooks the chance to handle this request
  83                 $className = null;
  84                 wfRunHooks( 'UploadCreateFromRequest', array( $type, &$className ) );
  85                 if ( is_null( $className ) ) {
  86                         $className = 'UploadFrom' . $type;
  87                         wfDebug( __METHOD__ . ": class name: $className\n" );
  88                         if( !in_array( $type, self::$uploadHandlers ) ) {
  89                                 return null;
  90                         }
  91                 }
  92
  93                 // Check whether this upload class is enabled
  94                 if( !call_user_func( array( $className, 'isEnabled' ) ) ) {
  95                         return null;
  96                 }
  97
  98                 // Check whether the request is valid
  99                 if( !call_user_func( array( $className, 'isValidRequest' ), $request ) ) {
 100                         return null;
 101                 }
 102
 103                 $handler = new $className;
 104
 105                 $handler->initializeFromRequest( $request );
 106                 return $handler;
 107         }
 108
 109         /**
 110          * Check whether a request if valid for this handler
 111          */
 112         public static function isValidRequest( $request ) {
 113                 return false;
 114         }
 115
 116         public function __construct() {}
 117
 118         /**
 119          * Initialize the path information
 120          * @param $name string the desired destination name
 121          * @param $tempPath string the temporary path
 122          * @param $fileSize int the file size
 123          * @param $removeTempFile bool (false) remove the temporary file?
 124          * @return null
 125          */
 126         public function initializePathInfo( $name, $tempPath, $fileSize, $removeTempFile = false ) {
 127                 $this->mDesiredDestName = $name;
 128                 $this->mTempPath = $tempPath;
 129                 $this->mFileSize = $fileSize;
 130                 $this->mRemoveTempFile = $removeTempFile;
 131         }
 132
 133         /**
 134          * Initialize from a WebRequest. Override this in a subclass.
 135          */
 136         public abstract function initializeFromRequest( &$request );
 137
 138         /**
 139          * Fetch the file. Usually a no-op
 140          */
 141         public function fetchFile() {
 142                 return Status::newGood();
 143         }
 144
 145         /**
 146          * Return the file size
 147          */
 148         public function isEmptyFile() {
 149                 return empty( $this->mFileSize );
 150         }
 151
 152         /**
 153          * @param string $srcPath the source path
 154          * @returns the real path if it was a virtual URL
 155          */
 156         function getRealPath( $srcPath ) {
 157                 $repo = RepoGroup::singleton()->getLocalRepo();
 158                 if ( $repo->isVirtualUrl( $srcPath ) ) {
 159                         return $repo->resolveVirtualUrl( $srcPath );
 160                 }
 161                 return $srcPath;
 162         }
 163
 164         /**
 165          * Verify whether the upload is sane.
 166          * Returns self::OK or else an array with error information
 167          */
 168         public function verifyUpload() {
 169                 /**
 170                  * If there was no filename or a zero size given, give up quick.
 171                  */
 172                 if( $this->isEmptyFile() ) {
 173                         return array( 'status' => self::EMPTY_FILE );
 174                 }
 175
 176                 /**
 177                  * Look at the contents of the file; if we can recognize the
 178                  * type but it's corrupt or data of the wrong type, we should
 179                  * probably not accept it.
 180                  */
 181                 $verification = $this->verifyFile();
 182                 if( $verification !== true ) {
 183                         if( !is_array( $verification ) ) {
 184                                 $verification = array( $verification );
 185                         }
 186                         return array(
 187                                 'status' => self::VERIFICATION_ERROR,
 188                                 'details' => $verification
 189                         );
 190                 }
 191
 192                 $nt = $this->getTitle();
 193                 if( is_null( $nt ) ) {
 194                         $result = array( 'status' => $this->mTitleError );
 195                         if( $this->mTitleError == self::ILLEGAL_FILENAME ) {
 196                                 $result['filtered'] = $this->mFilteredName;
 197                         }
 198                         if ( $this->mTitleError == self::FILETYPE_BADTYPE ) {
 199                                 $result['finalExt'] = $this->mFinalExtension;
 200                         }
 201                         return $result;
 202                 }
 203                 $this->mDestName = $this->getLocalFile()->getName();
 204
 205                 /**
 206                  * In some cases we may forbid overwriting of existing files.
 207                  */
 208                 $overwrite = $this->checkOverwrite();
 209                 if( $overwrite !== true ) {
 210                         return array(
 211                                 'status' => self::OVERWRITE_EXISTING_FILE,
 212                                 'overwrite' => $overwrite
 213                         );
 214                 }
 215
 216                 $error = '';
 217                 if( !wfRunHooks( 'UploadVerification',
 218                                 array( $this->mDestName, $this->mTempPath, &$error ) ) ) {
 219                         // This status needs another name...
 220                         return array( 'status' => self::HOOK_ABORTED, 'error' => $error );
 221                 }
 222
 223                 return array( 'status' => self::OK );
 224         }
 225
 226         /**
 227          * Verifies that it's ok to include the uploaded file
 228          *
 229          * @return mixed true of the file is verified, a string or array otherwise.
 230          */
 231         protected function verifyFile() {
 232                 $this->mFileProps = File::getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 233                 $this->checkMacBinary();
 234
 235                 # magically determine mime type
 236                 $magic = MimeMagic::singleton();
 237                 $mime = $magic->guessMimeType( $this->mTempPath, false );
 238
 239                 # check mime type, if desired
 240                 global $wgVerifyMimeType;
 241                 if ( $wgVerifyMimeType ) {
 242                         global $wgMimeTypeBlacklist;
 243                         if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
 244                                 return array( 'filetype-badmime', $mime );
 245                         }
 246
 247                         # Check IE type
 248                         $fp = fopen( $this->mTempPath, 'rb' );
 249                         $chunk = fread( $fp, 256 );
 250                         fclose( $fp );
 251                         $extMime = $magic->guessTypesForExtension( $this->mFinalExtension );
 252                         $ieTypes = $magic->getIEMimeTypes( $this->mTempPath, $chunk, $extMime );
 253                         foreach ( $ieTypes as $ieType ) {
 254                                 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
 255                                         return array( 'filetype-bad-ie-mime', $ieType );
 256                                 }
 257                         }
 258                 }
 259
 260                 # check for htmlish code and javascript
 261                 if( self::detectScript( $this->mTempPath, $mime, $this->mFinalExtension ) ) {
 262                         return 'uploadscripted';
 263                 }
 264                 if( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
 265                         if( self::detectScriptInSvg( $this->mTempPath ) ) {
 266                                 return 'uploadscripted';
 267                         }
 268                 }
 269
 270                 /**
 271                 * Scan the uploaded file for viruses
 272                 */
 273                 $virus = $this->detectVirus( $this->mTempPath );
 274                 if ( $virus ) {
 275                         return array( 'uploadvirus', $virus );
 276                 }
 277                 wfDebug( __METHOD__ . ": all clear; passing.\n" );
 278                 return true;
 279         }
 280
 281         /**
 282          * Check whether the user can edit, upload and create the image.
 283          *
 284          * @param User $user the user to verify the permissions against
 285          * @return mixed An array as returned by getUserPermissionsErrors or true
 286          *               in case the user has proper permissions.
 287          */
 288         public function verifyPermissions( $user ) {
 289                 /**
 290                  * If the image is protected, non-sysop users won't be able
 291                  * to modify it by uploading a new revision.
 292                  */
 293                 $nt = $this->getTitle();
 294                 if( is_null( $nt ) ) {
 295                         return true;
 296                 }
 297                 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
 298                 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
 299                 $permErrorsCreate = ( $nt->exists() ? array() : $nt->getUserPermissionsErrors( 'create', $user ) );
 300                 if( $permErrors || $permErrorsUpload || $permErrorsCreate ) {
 301                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
 302                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
 303                         return $permErrors;
 304                 }
 305                 return true;
 306         }
 307
 308         /**
 309          * Check for non fatal problems with the file
 310          *
 311          * @return array Array of warnings
 312          */
 313         public function checkWarnings() {
 314                 $warnings = array();
 315
 316                 $localFile = $this->getLocalFile();
 317                 $filename = $localFile->getName();
 318                 $n = strrpos( $filename, '.' );
 319                 $partname = $n ? substr( $filename, 0, $n ) : $filename;
 320
 321                 /**
 322                  * Check whether the resulting filename is different from the desired one,
 323                  * but ignore things like ucfirst() and spaces/underscore things
 324                  */
 325                 $comparableName = str_replace( ' ', '_', $this->mDesiredDestName );
 326                 $comparableName = Title::capitalize( $comparableName, NS_FILE );
 327
 328                 if( $this->mDesiredDestName != $filename && $comparableName != $filename ) {
 329                         $warnings['badfilename'] = $filename;
 330                 }
 331
 332                 // Check whether the file extension is on the unwanted list
 333                 global $wgCheckFileExtensions, $wgFileExtensions;
 334                 if ( $wgCheckFileExtensions ) {
 335                         if ( !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) ) {
 336                                 $warnings['filetype-unwanted-type'] = $this->mFinalExtension;
 337                         }
 338                 }
 339
 340                 global $wgUploadSizeWarning;
 341                 if ( $wgUploadSizeWarning && ( $this->mFileSize > $wgUploadSizeWarning ) ) {
 342                         $warnings['large-file'] = $wgUploadSizeWarning;
 343                 }
 344
 345                 if ( $this->mFileSize == 0 ) {
 346                         $warnings['emptyfile'] = true;
 347                 }
 348
 349                 $exists = self::getExistsWarning( $localFile );
 350                 if( $exists !== false ) {
 351                         $warnings['exists'] = $exists;
 352                 }
 353
 354                 // Check dupes against existing files
 355                 $hash = File::sha1Base36( $this->mTempPath );
 356                 $dupes = RepoGroup::singleton()->findBySha1( $hash );
 357                 $title = $this->getTitle();
 358                 // Remove all matches against self
 359                 foreach ( $dupes as $key => $dupe ) {
 360                         if( $title->equals( $dupe->getTitle() ) ) {
 361                                 unset( $dupes[$key] );
 362                         }
 363                 }
 364                 if( $dupes ) {
 365                         $warnings['duplicate'] = $dupes;
 366                 }
 367
 368                 // Check dupes against archives
 369                 $archivedImage = new ArchivedFile( null, 0, "{$hash}.{$this->mFinalExtension}" );
 370                 if ( $archivedImage->getID() > 0 ) {
 371                         $warnings['duplicate-archive'] = $archivedImage->getName();
 372                 }
 373
 374                 return $warnings;
 375         }
 376
 377         /**
 378          * Really perform the upload. Stores the file in the local repo, watches
 379          * if necessary and runs the UploadComplete hook.
 380          *
 381          * @return mixed Status indicating the whether the upload succeeded.
 382          */
 383         public function performUpload( $comment, $pageText, $watch, $user ) {
 384                 wfDebug( "\n\n\performUpload: sum:" . $comment . ' c: ' . $pageText . ' w:' . $watch );
 385                 $status = $this->getLocalFile()->upload( $this->mTempPath, $comment, $pageText,
 386                         File::DELETE_SOURCE, $this->mFileProps, false, $user );
 387
 388                 if( $status->isGood() && $watch ) {
 389                         $user->addWatch( $this->getLocalFile()->getTitle() );
 390                 }
 391
 392                 if( $status->isGood() ) {
 393                         wfRunHooks( 'UploadComplete', array( &$this ) );
 394                 }
 395
 396                 return $status;
 397         }
 398
 399         /**
 400          * Returns the title of the file to be uploaded. Sets mTitleError in case
 401          * the name was illegal.
 402          *
 403          * @return Title The title of the file or null in case the name was illegal
 404          */
 405         public function getTitle() {
 406                 if ( $this->mTitle !== false ) {
 407                         return $this->mTitle;
 408                 }
 409
 410                 /**
 411                  * Chop off any directories in the given filename. Then
 412                  * filter out illegal characters, and try to make a legible name
 413                  * out of it. We'll strip some silently that Title would die on.
 414                  */
 415                 $basename = $this->mDesiredDestName;
 416
 417                 $this->mFilteredName = wfStripIllegalFilenameChars( $basename );
 418                 /* Normalize to title form before we do any further processing */
 419                 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 420                 if( is_null( $nt ) ) {
 421                         $this->mTitleError = self::ILLEGAL_FILENAME;
 422                         return $this->mTitle = null;
 423                 }
 424                 $this->mFilteredName = $nt->getDBkey();
 425
 426                 /**
 427                  * We'll want to blacklist against *any* 'extension', and use
 428                  * only the final one for the whitelist.
 429                  */
 430                 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName );
 431
 432                 if( count( $ext ) ) {
 433                         $this->mFinalExtension = trim( $ext[count( $ext ) - 1] );
 434                 } else {
 435                         $this->mFinalExtension = '';
 436                 }
 437
 438                 /* Don't allow users to override the blacklist (check file extension) */
 439                 global $wgCheckFileExtensions, $wgStrictFileExtensions;
 440                 global $wgFileExtensions, $wgFileBlacklist;
 441                 if ( $this->mFinalExtension == '' ) {
 442                         $this->mTitleError = self::FILETYPE_MISSING;
 443                         return $this->mTitle = null;
 444                 } elseif ( $this->checkFileExtensionList( $ext, $wgFileBlacklist ) ||
 445                                 ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
 446                                         !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) ) ) {
 447                         $this->mTitleError = self::FILETYPE_BADTYPE;
 448                         return $this->mTitle = null;
 449                 }
 450
 451                 # If there was more than one "extension", reassemble the base
 452                 # filename to prevent bogus complaints about length
 453                 if( count( $ext ) > 1 ) {
 454                         for( $i = 0; $i < count( $ext ) - 1; $i++ ) {
 455                                 $partname .= '.' . $ext[$i];
 456                         }
 457                 }
 458
 459                 if( strlen( $partname ) < 1 ) {
 460                         $this->mTitleError =  self::MIN_LENGTH_PARTNAME;
 461                         return $this->mTitle = null;
 462                 }
 463
 464                 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 465                 if( is_null( $nt ) ) {
 466                         $this->mTitleError = self::ILLEGAL_FILENAME;
 467                         return $this->mTitle = null;
 468                 }
 469                 return $this->mTitle = $nt;
 470         }
 471
 472         /**
 473          * Return the local file and initializes if necessary.
 474          */
 475         public function getLocalFile() {
 476                 if( is_null( $this->mLocalFile ) ) {
 477                         $nt = $this->getTitle();
 478                         $this->mLocalFile = is_null( $nt ) ? null : wfLocalFile( $nt );
 479                 }
 480                 return $this->mLocalFile;
 481         }
 482
 483         /**
 484          * Stash a file in a temporary directory for later processing
 485          * after the user has confirmed it.
 486          *
 487          * If the user doesn't explicitly cancel or accept, these files
 488          * can accumulate in the temp directory.
 489          *
 490          * @param string $saveName - the destination filename
 491          * @param string $tempSrc - the source temporary file to save
 492          * @return string - full path the stashed file, or false on failure
 493          */
 494         protected function saveTempUploadedFile( $saveName, $tempSrc ) {
 495                 $repo = RepoGroup::singleton()->getLocalRepo();
 496                 $status = $repo->storeTemp( $saveName, $tempSrc );
 497                 return $status;
 498         }
 499
 500         /**
 501          * Stash a file in a temporary directory for later processing,
 502          * and save the necessary descriptive info into the session.
 503          * Returns a key value which will be passed through a form
 504          * to pick up the path info on a later invocation.
 505          *
 506          * @return int Session key
 507          */
 508         public function stashSession() {
 509                 $status = $this->saveTempUploadedFile( $this->mDestName, $this->mTempPath );
 510                 if( !$status->isOK() ) {
 511                         # Couldn't save the file.
 512                         return false;
 513                 }
 514                 if( !isset( $_SESSION ) ) {
 515                         session_start(); // start up the session (might have been previously closed to prevent php session locking)
 516                 }
 517                 $key = $this->getSessionKey();
 518                 $_SESSION['wsUploadData'][$key] = array(
 519                         'mTempPath'       => $status->value,
 520                         'mFileSize'       => $this->mFileSize,
 521                         'mFileProps'      => $this->mFileProps,
 522                         'version'         => self::SESSION_VERSION,
 523                 );
 524                 return $key;
 525         }
 526
 527         /**
 528          * Generate a random session key from stash in cases where we want to start an upload without much information
 529          */
 530         protected function getSessionKey() {
 531                 $key = mt_rand( 0, 0x7fffffff );
 532                 $_SESSION['wsUploadData'][$key] = array();
 533                 return $key;
 534         }
 535
 536         /**
 537          * If we've modified the upload file we need to manually remove it
 538          * on exit to clean up.
 539          */
 540         public function cleanupTempFile() {
 541                 if ( $this->mRemoveTempFile && $this->mTempPath && file_exists( $this->mTempPath ) ) {
 542                         wfDebug( __METHOD__ . ": Removing temporary file {$this->mTempPath}\n" );
 543                         unlink( $this->mTempPath );
 544                 }
 545         }
 546
 547         public function getTempPath() {
 548                 return $this->mTempPath;
 549         }
 550
 551         /**
 552          * Split a file into a base name and all dot-delimited 'extensions'
 553          * on the end. Some web server configurations will fall back to
 554          * earlier pseudo-'extensions' to determine type and execute
 555          * scripts, so the blacklist needs to check them all.
 556          *
 557          * @return array
 558          */
 559         public static function splitExtensions( $filename ) {
 560                 $bits = explode( '.', $filename );
 561                 $basename = array_shift( $bits );
 562                 return array( $basename, $bits );
 563         }
 564
 565         /**
 566          * Perform case-insensitive match against a list of file extensions.
 567          * Returns true if the extension is in the list.
 568          *
 569          * @param string $ext
 570          * @param array $list
 571          * @return bool
 572          */
 573         public static function checkFileExtension( $ext, $list ) {
 574                 return in_array( strtolower( $ext ), $list );
 575         }
 576
 577         /**
 578          * Perform case-insensitive match against a list of file extensions.
 579          * Returns true if any of the extensions are in the list.
 580          *
 581          * @param array $ext
 582          * @param array $list
 583          * @return bool
 584          */
 585         public static function checkFileExtensionList( $ext, $list ) {
 586                 foreach( $ext as $e ) {
 587                         if( in_array( strtolower( $e ), $list ) ) {
 588                                 return true;
 589                         }
 590                 }
 591                 return false;
 592         }
 593
 594         /**
 595          * Checks if the mime type of the uploaded file matches the file extension.
 596          *
 597          * @param string $mime the mime type of the uploaded file
 598          * @param string $extension The filename extension that the file is to be served with
 599          * @return bool
 600          */
 601         public static function verifyExtension( $mime, $extension ) {
 602                 $magic = MimeMagic::singleton();
 603
 604                 if ( !$mime || $mime == 'unknown' || $mime == 'unknown/unknown' )
 605                         if ( !$magic->isRecognizableExtension( $extension ) ) {
 606                                 wfDebug( __METHOD__ . ": passing file with unknown detected mime type; " .
 607                                         "unrecognized extension '$extension', can't verify\n" );
 608                                 return true;
 609                         } else {
 610                                 wfDebug( __METHOD__ . ": rejecting file with unknown detected mime type; ".
 611                                         "recognized extension '$extension', so probably invalid file\n" );
 612                                 return false;
 613                         }
 614
 615                 $match = $magic->isMatchingExtension( $extension, $mime );
 616
 617                 if ( $match === null ) {
 618                         wfDebug( __METHOD__ . ": no file extension known for mime type $mime, passing file\n" );
 619                         return true;
 620                 } elseif( $match === true ) {
 621                         wfDebug( __METHOD__ . ": mime type $mime matches extension $extension, passing file\n" );
 622
 623                         #TODO: if it's a bitmap, make sure PHP or ImageMagic resp. can handle it!
 624                         return true;
 625
 626                 } else {
 627                         wfDebug( __METHOD__ . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
 628                         return false;
 629                 }
 630         }
 631
 632         /**
 633          * Heuristic for detecting files that *could* contain JavaScript instructions or
 634          * things that may look like HTML to a browser and are thus
 635          * potentially harmful. The present implementation will produce false
 636          * positives in some situations.
 637          *
 638          * @param string $file Pathname to the temporary upload file
 639          * @param string $mime The mime type of the file
 640          * @param string $extension The extension of the file
 641          * @return bool true if the file contains something looking like embedded scripts
 642          */
 643         public static function detectScript( $file, $mime, $extension ) {
 644                 global $wgAllowTitlesInSVG;
 645
 646                 # ugly hack: for text files, always look at the entire file.
 647                 # For binary field, just check the first K.
 648
 649                 if( strpos( $mime,'text/' ) === 0 ) {
 650                         $chunk = file_get_contents( $file );
 651                 } else {
 652                         $fp = fopen( $file, 'rb' );
 653                         $chunk = fread( $fp, 1024 );
 654                         fclose( $fp );
 655                 }
 656
 657                 $chunk = strtolower( $chunk );
 658
 659                 if( !$chunk ) {
 660                         return false;
 661                 }
 662
 663                 # decode from UTF-16 if needed (could be used for obfuscation).
 664                 if( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
 665                         $enc = 'UTF-16BE';
 666                 } elseif( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
 667                         $enc = 'UTF-16LE';
 668                 } else {
 669                         $enc = null;
 670                 }
 671
 672                 if( $enc ) {
 673                         $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
 674                 }
 675
 676                 $chunk = trim( $chunk );
 677
 678                 # FIXME: convert from UTF-16 if necessarry!
 679                 wfDebug( __METHOD__ . ": checking for embedded scripts and HTML stuff\n" );
 680
 681                 # check for HTML doctype
 682                 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
 683                         return true;
 684                 }
 685
 686                 /**
 687                  * Internet Explorer for Windows performs some really stupid file type
 688                  * autodetection which can cause it to interpret valid image files as HTML
 689                  * and potentially execute JavaScript, creating a cross-site scripting
 690                  * attack vectors.
 691                  *
 692                  * Apple's Safari browser also performs some unsafe file type autodetection
 693                  * which can cause legitimate files to be interpreted as HTML if the
 694                  * web server is not correctly configured to send the right content-type
 695                  * (or if you're really uploading plain text and octet streams!)
 696                  *
 697                  * Returns true if IE is likely to mistake the given file for HTML.
 698                  * Also returns true if Safari would mistake the given file for HTML
 699                  * when served with a generic content-type.
 700                  */
 701                 $tags = array(
 702                         '<a href',
 703                         '<body',
 704                         '<head',
 705                         '<html',   #also in safari
 706                         '<img',
 707                         '<pre',
 708                         '<script', #also in safari
 709                         '<table'
 710                 );
 711
 712                 if( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
 713                         $tags[] = '<title';
 714                 }
 715
 716                 foreach( $tags as $tag ) {
 717                         if( false !== strpos( $chunk, $tag ) ) {
 718                                 return true;
 719                         }
 720                 }
 721
 722                 /*
 723                  * look for JavaScript
 724                  */
 725
 726                 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
 727                 $chunk = Sanitizer::decodeCharReferences( $chunk );
 728
 729                 # look for script-types
 730                 if( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
 731                         return true;
 732                 }
 733
 734                 # look for html-style script-urls
 735                 if( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
 736                         return true;
 737                 }
 738
 739                 # look for css-style script-urls
 740                 if( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
 741                         return true;
 742                 }
 743
 744                 wfDebug( __METHOD__ . ": no scripts found\n" );
 745                 return false;
 746         }
 747
 748         protected function detectScriptInSvg( $filename ) {
 749                 $check = new XmlTypeCheck( $filename, array( $this, 'checkSvgScriptCallback' ) );
 750                 return $check->filterMatch;
 751         }
 752
 753         /**
 754          * @todo Replace this with a whitelist filter!
 755          */
 756         public function checkSvgScriptCallback( $element, $attribs ) {
 757                 $stripped = $this->stripXmlNamespace( $element );
 758
 759                 if( $stripped == 'script' ) {
 760                         wfDebug( __METHOD__ . ": Found script element '$element' in uploaded file.\n" );
 761                         return true;
 762                 }
 763
 764                 foreach( $attribs as $attrib => $value ) {
 765                         $stripped = $this->stripXmlNamespace( $attrib );
 766                         if( substr( $stripped, 0, 2 ) == 'on' ) {
 767                                 wfDebug( __METHOD__ . ": Found script attribute '$attrib'='value' in uploaded file.\n" );
 768                                 return true;
 769                         }
 770                         if( $stripped == 'href' && strpos( strtolower( $value ), 'javascript:' ) !== false ) {
 771                                 wfDebug( __METHOD__ . ": Found script href attribute '$attrib'='$value' in uploaded file.\n" );
 772                                 return true;
 773                         }
 774                 }
 775         }
 776
 777         private function stripXmlNamespace( $name ) {
 778                 // 'http://www.w3.org/2000/svg:script' -> 'script'
 779                 $parts = explode( ':', strtolower( $name ) );
 780                 return array_pop( $parts );
 781         }
 782
 783         /**
 784          * Generic wrapper function for a virus scanner program.
 785          * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
 786          * $wgAntivirusRequired may be used to deny upload if the scan fails.
 787          *
 788          * @param string $file Pathname to the temporary upload file
 789          * @return mixed false if not virus is found, NULL if the scan fails or is disabled,
 790          *         or a string containing feedback from the virus scanner if a virus was found.
 791          *         If textual feedback is missing but a virus was found, this function returns true.
 792          */
 793         public static function detectVirus( $file ) {
 794                 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
 795
 796                 if ( !$wgAntivirus ) {
 797                         wfDebug( __METHOD__ . ": virus scanner disabled\n" );
 798                         return null;
 799                 }
 800
 801                 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
 802                         wfDebug( __METHOD__ . ": unknown virus scanner: $wgAntivirus\n" );
 803                         $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1</div>", array( 'virus-badscanner', $wgAntivirus ) );
 804                         return wfMsg( 'virus-unknownscanner' ) . " $wgAntivirus";
 805                 }
 806
 807                 # look up scanner configuration
 808                 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
 809                 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
 810                 $msgPattern = isset( $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ) ?
 811                         $wgAntivirusSetup[$wgAntivirus]['messagepattern'] : null;
 812
 813                 if ( strpos( $command, "%f" ) === false ) {
 814                         # simple pattern: append file to scan
 815                         $command .= " " . wfEscapeShellArg( $file );
 816                 } else {
 817                         # complex pattern: replace "%f" with file to scan
 818                         $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
 819                 }
 820
 821                 wfDebug( __METHOD__ . ": running virus scan: $command \n" );
 822
 823                 # execute virus scanner
 824                 $exitCode = false;
 825
 826                 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
 827                 #      that does not seem to be worth the pain.
 828                 #      Ask me (Duesentrieb) about it if it's ever needed.
 829                 $output = wfShellExec( "$command 2>&1", $exitCode );
 830
 831                 # map exit code to AV_xxx constants.
 832                 $mappedCode = $exitCode;
 833                 if ( $exitCodeMap ) {
 834                         if ( isset( $exitCodeMap[$exitCode] ) ) {
 835                                 $mappedCode = $exitCodeMap[$exitCode];
 836                         } elseif ( isset( $exitCodeMap["*"] ) ) {
 837                                 $mappedCode = $exitCodeMap["*"];
 838                         }
 839                 }
 840
 841                 if ( $mappedCode === AV_SCAN_FAILED ) {
 842                         # scan failed (code was mapped to false by $exitCodeMap)
 843                         wfDebug( __METHOD__ . ": failed to scan $file (code $exitCode).\n" );
 844
 845                         if ( $wgAntivirusRequired ) {
 846                                 return wfMsg( 'virus-scanfailed', array( $exitCode ) );
 847                         } else {
 848                                 return null;
 849                         }
 850                 } elseif ( $mappedCode === AV_SCAN_ABORTED ) {
 851                         # scan failed because filetype is unknown (probably imune)
 852                         wfDebug( __METHOD__ . ": unsupported file type $file (code $exitCode).\n" );
 853                         return null;
 854                 } elseif ( $mappedCode === AV_NO_VIRUS ) {
 855                         # no virus found
 856                         wfDebug( __METHOD__ . ": file passed virus scan.\n" );
 857                         return false;
 858                 } else {
 859                         $output = trim( $output );
 860
 861                         if ( !$output ) {
 862                                 $output = true; #if there's no output, return true
 863                         } elseif ( $msgPattern ) {
 864                                 $groups = array();
 865                                 if ( preg_match( $msgPattern, $output, $groups ) ) {
 866                                         if ( $groups[1] ) {
 867                                                 $output = $groups[1];
 868                                         }
 869                                 }
 870                         }
 871
 872                         wfDebug( __METHOD__ . ": FOUND VIRUS! scanner feedback: $output \n" );
 873                         return $output;
 874                 }
 875         }
 876
 877         /**
 878          * Check if the temporary file is MacBinary-encoded, as some uploads
 879          * from Internet Explorer on Mac OS Classic and Mac OS X will be.
 880          * If so, the data fork will be extracted to a second temporary file,
 881          * which will then be checked for validity and either kept or discarded.
 882          */
 883         private function checkMacBinary() {
 884                 $macbin = new MacBinary( $this->mTempPath );
 885                 if( $macbin->isValid() ) {
 886                         $dataFile = tempnam( wfTempDir(), 'WikiMacBinary' );
 887                         $dataHandle = fopen( $dataFile, 'wb' );
 888
 889                         wfDebug( __METHOD__ . ": Extracting MacBinary data fork to $dataFile\n" );
 890                         $macbin->extractData( $dataHandle );
 891
 892                         $this->mTempPath = $dataFile;
 893                         $this->mFileSize = $macbin->dataForkLength();
 894
 895                         // We'll have to manually remove the new file if it's not kept.
 896                         $this->mRemoveTempFile = true;
 897                 }
 898                 $macbin->close();
 899         }
 900
 901         /**
 902          * Check if there's an overwrite conflict and, if so, if restrictions
 903          * forbid this user from performing the upload.
 904          *
 905          * @return mixed true on success, error string on failure
 906          */
 907         private function checkOverwrite() {
 908                 global $wgUser;
 909                 // First check whether the local file can be overwritten
 910                 $file = $this->getLocalFile();
 911                 if( $file->exists() ) {
 912                         if( !self::userCanReUpload( $wgUser, $file ) ) {
 913                                 return 'fileexists-forbidden';
 914                         } else {
 915                                 return true;
 916                         }
 917                 }
 918
 919                 /* Check shared conflicts: if the local file does not exist, but
 920                  * wfFindFile finds a file, it exists in a shared repository.
 921                  */
 922                 $file = wfFindFile( $this->getTitle() );
 923                 if ( $file && !$wgUser->isAllowed( 'reupload-shared' ) ) {
 924                         return 'fileexists-shared-forbidden';
 925                 }
 926
 927                 return true;
 928         }
 929
 930         /**
 931          * Check if a user is the last uploader
 932          *
 933          * @param User $user
 934          * @param string $img, image name
 935          * @return bool
 936          */
 937         public static function userCanReUpload( User $user, $img ) {
 938                 if( $user->isAllowed( 'reupload' ) ) {
 939                         return true; // non-conditional
 940                 }
 941                 if( !$user->isAllowed( 'reupload-own' ) ) {
 942                         return false;
 943                 }
 944                 if( is_string( $img ) ) {
 945                         $img = wfLocalFile( $img );
 946                 }
 947                 if ( !( $img instanceof LocalFile ) ) {
 948                         return false;
 949                 }
 950
 951                 return $user->getId() == $img->getUser( 'id' );
 952         }
 953
 954         /**
 955          * Helper function that does various existence checks for a file.
 956          * The following checks are performed:
 957          * - The file exists
 958          * - Article with the same name as the file exists
 959          * - File exists with normalized extension
 960          * - The file looks like a thumbnail and the original exists
 961          *
 962          * @param File $file The file to check
 963          * @return mixed False if the file does not exists, else an array
 964          */
 965         public static function getExistsWarning( $file ) {
 966                 if( $file->exists() ) {
 967                         return array( 'warning' => 'exists', 'file' => $file );
 968                 }
 969
 970                 if( $file->getTitle()->getArticleID() ) {
 971                         return array( 'warning' => 'page-exists', 'file' => $file );
 972                 }
 973
 974                 if ( $file->wasDeleted() && !$file->exists() ) {
 975                         return array( 'warning' => 'was-deleted', 'file' => $file );
 976                 }
 977
 978                 if( strpos( $file->getName(), '.' ) == false ) {
 979                         $partname = $file->getName();
 980                         $extension = '';
 981                 } else {
 982                         $n = strrpos( $file->getName(), '.' );
 983                         $extension = substr( $file->getName(), $n + 1 );
 984                         $partname = substr( $file->getName(), 0, $n );
 985                 }
 986                 $normalizedExtension = File::normalizeExtension( $extension );
 987
 988                 if ( $normalizedExtension != $extension ) {
 989                         // We're not using the normalized form of the extension.
 990                         // Normal form is lowercase, using most common of alternate
 991                         // extensions (eg 'jpg' rather than 'JPEG').
 992                         //
 993                         // Check for another file using the normalized form...
 994                         $nt_lc = Title::makeTitle( NS_FILE, "{$partname}.{$normalizedExtension}" );
 995                         $file_lc = wfLocalFile( $nt_lc );
 996
 997                         if( $file_lc->exists() ) {
 998                                 return array(
 999                                         'warning' => 'exists-normalized',
1000                                         'file' => $file,
1001                                         'normalizedFile' => $file_lc
1002                                 );
1003                         }
1004                 }
1005
1006                 if ( self::isThumbName( $file->getName() ) ) {
1007                         # Check for filenames like 50px- or 180px-, these are mostly thumbnails
1008                         $nt_thb = Title::newFromText( substr( $partname , strpos( $partname , '-' ) +1 ) . '.' . $extension, NS_FILE );
1009                         $file_thb = wfLocalFile( $nt_thb );
1010                         if( $file_thb->exists() ) {
1011                                 return array(
1012                                         'warning' => 'thumb',
1013                                         'file' => $file,
1014                                         'thumbFile' => $file_thb
1015                                 );
1016                         } else {
1017                                 // File does not exist, but we just don't like the name
1018                                 return array(
1019                                         'warning' => 'thumb-name',
1020                                         'file' => $file,
1021                                         'thumbFile' => $file_thb
1022                                 );
1023                         }
1024                 }
1025
1026
1027                 foreach( self::getFilenamePrefixBlacklist() as $prefix ) {
1028                         if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
1029                                 return array(
1030                                         'warning' => 'bad-prefix',
1031                                         'file' => $file,
1032                                         'prefix' => $prefix
1033                                 );
1034                         }
1035                 }
1036
1037                 return false;
1038         }
1039
1040         /**
1041          * Helper function that checks whether the filename looks like a thumbnail
1042          */
1043         public static function isThumbName( $filename ) {
1044                 $n = strrpos( $filename, '.' );
1045                 $partname = $n ? substr( $filename, 0, $n ) : $filename;
1046                 return (
1047                                         substr( $partname , 3, 3 ) == 'px-' ||
1048                                         substr( $partname , 2, 3 ) == 'px-'
1049                                 ) &&
1050                                 preg_match( "/[0-9]{2}/" , substr( $partname , 0, 2 ) );
1051         }
1052
1053         /**
1054          * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
1055          *
1056          * @return array list of prefixes
1057          */
1058         public static function getFilenamePrefixBlacklist() {
1059                 $blacklist = array();
1060                 $message = wfMsgForContent( 'filename-prefix-blacklist' );
1061                 if( $message && !( wfEmptyMsg( 'filename-prefix-blacklist', $message ) || $message == '-' ) ) {
1062                         $lines = explode( "\n", $message );
1063                         foreach( $lines as $line ) {
1064                                 // Remove comment lines
1065                                 $comment = substr( trim( $line ), 0, 1 );
1066                                 if ( $comment == '#' || $comment == '' ) {
1067                                         continue;
1068                                 }
1069                                 // Remove additional comments after a prefix
1070                                 $comment = strpos( $line, '#' );
1071                                 if ( $comment > 0 ) {
1072                                         $line = substr( $line, 0, $comment-1 );
1073                                 }
1074                                 $blacklist[] = trim( $line );
1075                         }
1076                 }
1077                 return $blacklist;
1078         }
1079
1080         public function getImageInfo( $result ) {
1081                 $file = $this->getLocalFile();
1082                 $imParam = ApiQueryImageInfo::getPropertyNames();
1083                 return ApiQueryImageInfo::getInfo( $file, array_flip( $imParam ), $result );
1084         }
1085
1086 }