includes/upload/UploadBase.php

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