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