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