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