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