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