includes/upload/UploadBase.php

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