includes/upload/UploadBase.php

   1 <?php
   2 /**
   3  * @file
   4  * @ingroup upload
   5  *
   6  * UploadBase and subclasses are the backend of MediaWiki's file uploads.
   7  * The frontends are formed by ApiUpload and SpecialUpload.
   8  *
   9  * See also includes/docs/upload.txt
  10  *
  11  * @author Brion Vibber
  12  * @author Bryan Tong Minh
  13  * @author Michael Dale
  14  */
  15
  16 abstract class UploadBase {
  17         protected $mTempPath;
  18         protected $mDesiredDestName, $mDestName, $mRemoveTempFile, $mSourceType;
  19         protected $mTitle = false, $mTitleError = 0;
  20         protected $mFilteredName, $mFinalExtension;
  21         protected $mLocalFile;
  22
  23         const SUCCESS = 0;
  24         const OK = 0;
  25         const EMPTY_FILE = 3;
  26         const MIN_LENGTH_PARTNAME = 4;
  27         const ILLEGAL_FILENAME = 5;
  28         const OVERWRITE_EXISTING_FILE = 7;
  29         const FILETYPE_MISSING = 8;
  30         const FILETYPE_BADTYPE = 9;
  31         const VERIFICATION_ERROR = 10;
  32         # HOOK_ABORTED is the new name of UPLOAD_VERIFICATION_ERROR
  33         const UPLOAD_VERIFICATION_ERROR = 11;
  34         const HOOK_ABORTED = 11;
  35         const FILE_TOO_LARGE = 12;
  36
  37         const SESSION_VERSION = 2;
  38         const SESSION_KEYNAME = 'wsUploadData';
  39
  40         static public function getSessionKeyname() {
  41                 return self::SESSION_KEYNAME;
  42         }
  43
  44         /**
  45          * Returns true if uploads are enabled.
  46          * Can be override by subclasses.
  47          */
  48         public static function isEnabled() {
  49                 global $wgEnableUploads;
  50                 if ( !$wgEnableUploads ) {
  51                         return false;
  52                 }
  53
  54                 # Check php's file_uploads setting
  55                 if( !wfIniGetBool( 'file_uploads' ) ) {
  56                         return false;
  57                 }
  58                 return true;
  59         }
  60
  61         /**
  62          * Returns true if the user can use this upload module or else a string
  63          * identifying the missing permission.
  64          * Can be overriden by subclasses.
  65          */
  66         public static function isAllowed( $user ) {
  67                 if( !$user->isAllowed( 'upload' ) ) {
  68                         return 'upload';
  69                 }
  70                 return true;
  71         }
  72
  73         // Upload handlers. Should probably just be a global.
  74         static $uploadHandlers = array( 'Stash', 'File', 'Url' );
  75
  76         /**
  77          * Create a form of UploadBase depending on wpSourceType and initializes it
  78          */
  79         public static function createFromRequest( &$request, $type = null ) {
  80                 $type = $type ? $type : $request->getVal( 'wpSourceType', 'File' );
  81
  82                 if( !$type ) {
  83                         return null;
  84                 }
  85
  86                 // Get the upload class
  87                 $type = ucfirst( $type );
  88
  89                 // Give hooks the chance to handle this request
  90                 $className = null;
  91                 wfRunHooks( 'UploadCreateFromRequest', array( $type, &$className ) );
  92                 if ( is_null( $className ) ) {
  93                         $className = 'UploadFrom' . $type;
  94                         wfDebug( __METHOD__ . ": class name: $className\n" );
  95                         if( !in_array( $type, self::$uploadHandlers ) ) {
  96                                 return null;
  97                         }
  98                 }
  99
 100                 // Check whether this upload class is enabled
 101                 if( !call_user_func( array( $className, 'isEnabled' ) ) ) {
 102                         return null;
 103                 }
 104
 105                 // Check whether the request is valid
 106                 if( !call_user_func( array( $className, 'isValidRequest' ), $request ) ) {
 107                         return null;
 108                 }
 109
 110                 $handler = new $className;
 111
 112                 $handler->initializeFromRequest( $request );
 113                 return $handler;
 114         }
 115
 116         /**
 117          * Check whether a request if valid for this handler
 118          */
 119         public static function isValidRequest( $request ) {
 120                 return false;
 121         }
 122
 123         public function __construct() {}
 124
 125         /**
 126          * Initialize the path information
 127          * @param $name string the desired destination name
 128          * @param $tempPath string the temporary path
 129          * @param $fileSize int the file size
 130          * @param $removeTempFile bool (false) remove the temporary file?
 131          * @return null
 132          */
 133         public function initializePathInfo( $name, $tempPath, $fileSize, $removeTempFile = false ) {
 134                 $this->mDesiredDestName = $name;
 135                 $this->mTempPath = $tempPath;
 136                 $this->mFileSize = $fileSize;
 137                 $this->mRemoveTempFile = $removeTempFile;
 138         }
 139
 140         /**
 141          * Initialize from a WebRequest. Override this in a subclass.
 142          */
 143         public abstract function initializeFromRequest( &$request );
 144
 145         /**
 146          * Fetch the file. Usually a no-op
 147          */
 148         public function fetchFile() {
 149                 return Status::newGood();
 150         }
 151
 152         /**
 153          * Return true if the file is empty
 154          * @return bool
 155          */
 156         public function isEmptyFile() {
 157                 return empty( $this->mFileSize );
 158         }
 159
 160         /**
 161          * Return the file size
 162          * @return integer
 163          */
 164         public function getFileSize() {
 165                 return $this->mFileSize;
 166         }
 167
 168         /**
 169          * Append a file to the Repo file
 170          *
 171          * @param $srcPath String: path to source file
 172          * @param $toAppendPath String: path to the Repo file that will be appended to.
 173          * @return Status Status
 174          */
 175         protected function appendToUploadFile( $srcPath, $toAppendPath ) {
 176                 $repo = RepoGroup::singleton()->getLocalRepo();
 177                 $status = $repo->append( $srcPath, $toAppendPath );
 178                 return $status;
 179         }
 180
 181         /**
 182          * @param $srcPath String: the source path
 183          * @return the real path if it was a virtual URL
 184          */
 185         function getRealPath( $srcPath ) {
 186                 $repo = RepoGroup::singleton()->getLocalRepo();
 187                 if ( $repo->isVirtualUrl( $srcPath ) ) {
 188                         return $repo->resolveVirtualUrl( $srcPath );
 189                 }
 190                 return $srcPath;
 191         }
 192
 193         /**
 194          * Verify whether the upload is sane.
 195          * @return mixed self::OK or else an array with error information
 196          */
 197         public function verifyUpload( ) {
 198                 /**
 199                  * If there was no filename or a zero size given, give up quick.
 200                  */
 201                 if( $this->isEmptyFile() ) {
 202                         return array( 'status' => self::EMPTY_FILE );
 203                 }
 204
 205                 /**
 206                  * Honor $wgMaxUploadSize
 207                  */
 208                 global $wgMaxUploadSize;
 209                 if( $this->mFileSize > $wgMaxUploadSize ) {
 210                         return array( 'status' => self::FILE_TOO_LARGE );
 211                 }
 212
 213                 /**
 214                  * Look at the contents of the file; if we can recognize the
 215                  * type but it's corrupt or data of the wrong type, we should
 216                  * probably not accept it.
 217                  */
 218                 $verification = $this->verifyFile();
 219                 if( $verification !== true ) {
 220                         if( !is_array( $verification ) ) {
 221                                 $verification = array( $verification );
 222                         }
 223                         return array(
 224                                 'status' => self::VERIFICATION_ERROR,
 225                                 'details' => $verification
 226                         );
 227                 }
 228
 229                 /**
 230                  * Make sure this file can be created
 231                  */
 232                 $result = $this->validateNameAndOverwrite();
 233                 if( $result !== true ) {
 234                         return $result;
 235                 }
 236
 237                 $error = '';
 238                 if( !wfRunHooks( 'UploadVerification',
 239                                 array( $this->mDestName, $this->mTempPath, &$error ) ) ) {
 240                         // @fixme This status needs another name...
 241                         return array( 'status' => self::HOOK_ABORTED, 'error' => $error );
 242                 }
 243
 244                 return array( 'status' => self::OK );
 245         }
 246
 247         /**
 248          * Verify that the name is valid and, if necessary, that we can overwrite
 249          *
 250          * @return mixed true if valid, otherwise and array with 'status'
 251          * and other keys
 252          **/
 253         public function validateNameAndOverwrite() {
 254                 $nt = $this->getTitle();
 255                 if( is_null( $nt ) ) {
 256                         $result = array( 'status' => $this->mTitleError );
 257                         if( $this->mTitleError == self::ILLEGAL_FILENAME ) {
 258                                 $result['filtered'] = $this->mFilteredName;
 259                         }
 260                         if ( $this->mTitleError == self::FILETYPE_BADTYPE ) {
 261                                 $result['finalExt'] = $this->mFinalExtension;
 262                         }
 263                         return $result;
 264                 }
 265                 $this->mDestName = $this->getLocalFile()->getName();
 266
 267                 /**
 268                  * In some cases we may forbid overwriting of existing files.
 269                  */
 270                 $overwrite = $this->checkOverwrite();
 271                 if( $overwrite !== true ) {
 272                         return array(
 273                                 'status' => self::OVERWRITE_EXISTING_FILE,
 274                                 'overwrite' => $overwrite
 275                         );
 276                 }
 277                 return true;
 278         }
 279
 280         /**
 281          * Verifies that it's ok to include the uploaded file
 282          *
 283          * @return mixed true of the file is verified, a string or array otherwise.
 284          */
 285         protected function verifyFile() {
 286                 $this->mFileProps = File::getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 287                 $this->checkMacBinary();
 288
 289                 # magically determine mime type
 290                 $magic = MimeMagic::singleton();
 291                 $mime = $magic->guessMimeType( $this->mTempPath, false );
 292
 293                 # check mime type, if desired
 294                 global $wgVerifyMimeType;
 295                 if ( $wgVerifyMimeType ) {
 296                         wfDebug ( "\n\nmime: <$mime> extension: <{$this->mFinalExtension}>\n\n");
 297                         if ( !$this->verifyExtension( $mime, $this->mFinalExtension ) ) {
 298                                 return array( 'filetype-mime-mismatch' );
 299                         }
 300
 301                         global $wgMimeTypeBlacklist;
 302                         if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
 303                                 return array( 'filetype-badmime', $mime );
 304                         }
 305
 306                         # Check IE type
 307                         $fp = fopen( $this->mTempPath, 'rb' );
 308                         $chunk = fread( $fp, 256 );
 309                         fclose( $fp );
 310                         $extMime = $magic->guessTypesForExtension( $this->mFinalExtension );
 311                         $ieTypes = $magic->getIEMimeTypes( $this->mTempPath, $chunk, $extMime );
 312                         foreach ( $ieTypes as $ieType ) {
 313                                 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
 314                                         return array( 'filetype-bad-ie-mime', $ieType );
 315                                 }
 316                         }
 317                 }
 318
 319                 # check for htmlish code and javascript
 320                 if( self::detectScript( $this->mTempPath, $mime, $this->mFinalExtension ) ) {
 321                         return 'uploadscripted';
 322                 }
 323                 if( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
 324                         if( self::detectScriptInSvg( $this->mTempPath ) ) {
 325                                 return 'uploadscripted';
 326                         }
 327                 }
 328
 329                 /**
 330                 * Scan the uploaded file for viruses
 331                 */
 332                 $virus = $this->detectVirus( $this->mTempPath );
 333                 if ( $virus ) {
 334                         return array( 'uploadvirus', $virus );
 335                 }
 336                 wfDebug( __METHOD__ . ": all clear; passing.\n" );
 337                 return true;
 338         }
 339
 340         /**
 341          * Check whether the user can edit, upload and create the image.
 342          *
 343          * @param $user the User object to verify the permissions against
 344          * @return mixed An array as returned by getUserPermissionsErrors or true
 345          *               in case the user has proper permissions.
 346          */
 347         public function verifyPermissions( $user ) {
 348                 /**
 349                  * If the image is protected, non-sysop users won't be able
 350                  * to modify it by uploading a new revision.
 351                  */
 352                 $nt = $this->getTitle();
 353                 if( is_null( $nt ) ) {
 354                         return true;
 355                 }
 356                 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
 357                 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
 358                 $permErrorsCreate = ( $nt->exists() ? array() : $nt->getUserPermissionsErrors( 'create', $user ) );
 359                 if( $permErrors || $permErrorsUpload || $permErrorsCreate ) {
 360                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
 361                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
 362                         return $permErrors;
 363                 }
 364                 return true;
 365         }
 366
 367         /**
 368          * Check for non fatal problems with the file
 369          *
 370          * @return Array of warnings
 371          */
 372         public function checkWarnings() {
 373                 $warnings = array();
 374
 375                 $localFile = $this->getLocalFile();
 376                 $filename = $localFile->getName();
 377                 $n = strrpos( $filename, '.' );
 378                 $partname = $n ? substr( $filename, 0, $n ) : $filename;
 379
 380                 /**
 381                  * Check whether the resulting filename is different from the desired one,
 382                  * but ignore things like ucfirst() and spaces/underscore things
 383                  */
 384                 $comparableName = str_replace( ' ', '_', $this->mDesiredDestName );
 385                 $comparableName = Title::capitalize( $comparableName, NS_FILE );
 386
 387                 if( $this->mDesiredDestName != $filename && $comparableName != $filename ) {
 388                         $warnings['badfilename'] = $filename;
 389                 }
 390
 391                 // Check whether the file extension is on the unwanted list
 392                 global $wgCheckFileExtensions, $wgFileExtensions;
 393                 if ( $wgCheckFileExtensions ) {
 394                         if ( !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) ) {
 395                                 $warnings['filetype-unwanted-type'] = $this->mFinalExtension;
 396                         }
 397                 }
 398
 399                 global $wgUploadSizeWarning;
 400                 if ( $wgUploadSizeWarning && ( $this->mFileSize > $wgUploadSizeWarning ) ) {
 401                         $warnings['large-file'] = $wgUploadSizeWarning;
 402                 }
 403
 404                 if ( $this->mFileSize == 0 ) {
 405                         $warnings['emptyfile'] = true;
 406                 }
 407
 408                 $exists = self::getExistsWarning( $localFile );
 409                 if( $exists !== false ) {
 410                         $warnings['exists'] = $exists;
 411                 }
 412
 413                 // Check dupes against existing files
 414                 $hash = File::sha1Base36( $this->mTempPath );
 415                 $dupes = RepoGroup::singleton()->findBySha1( $hash );
 416                 $title = $this->getTitle();
 417                 // Remove all matches against self
 418                 foreach ( $dupes as $key => $dupe ) {
 419                         if( $title->equals( $dupe->getTitle() ) ) {
 420                                 unset( $dupes[$key] );
 421                         }
 422                 }
 423                 if( $dupes ) {
 424                         $warnings['duplicate'] = $dupes;
 425                 }
 426
 427                 // Check dupes against archives
 428                 $archivedImage = new ArchivedFile( null, 0, "{$hash}.{$this->mFinalExtension}" );
 429                 if ( $archivedImage->getID() > 0 ) {
 430                         $warnings['duplicate-archive'] = $archivedImage->getName();
 431                 }
 432
 433                 return $warnings;
 434         }
 435
 436         /**
 437          * Really perform the upload. Stores the file in the local repo, watches
 438          * if necessary and runs the UploadComplete hook.
 439          *
 440          * @return mixed Status indicating the whether the upload succeeded.
 441          */
 442         public function performUpload( $comment, $pageText, $watch, $user ) {
 443                 wfDebug( "\n\n\performUpload: sum:" . $comment . ' c: ' . $pageText . ' w:' . $watch );
 444                 $status = $this->getLocalFile()->upload( $this->mTempPath, $comment, $pageText,
 445                         File::DELETE_SOURCE, $this->mFileProps, false, $user );
 446
 447                 if( $status->isGood() ) {
 448                         if ( $watch ) {
 449                                 $user->addWatch( $this->getLocalFile()->getTitle() );
 450                         }
 451
 452                         wfRunHooks( 'UploadComplete', array( &$this ) );
 453                 }
 454
 455                 return $status;
 456         }
 457
 458         /**
 459          * Returns the title of the file to be uploaded. Sets mTitleError in case
 460          * the name was illegal.
 461          *
 462          * @return Title The title of the file or null in case the name was illegal
 463          */
 464         public function getTitle() {
 465                 if ( $this->mTitle !== false ) {
 466                         return $this->mTitle;
 467                 }
 468
 469                 /**
 470                  * Chop off any directories in the given filename. Then
 471                  * filter out illegal characters, and try to make a legible name
 472                  * out of it. We'll strip some silently that Title would die on.
 473                  */
 474                 $this->mFilteredName = wfStripIllegalFilenameChars( $this->mDesiredDestName );
 475                 /* Normalize to title form before we do any further processing */
 476                 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 477                 if( is_null( $nt ) ) {
 478                         $this->mTitleError = self::ILLEGAL_FILENAME;
 479                         return $this->mTitle = null;
 480                 }
 481                 $this->mFilteredName = $nt->getDBkey();
 482
 483                 /**
 484                  * We'll want to blacklist against *any* 'extension', and use
 485                  * only the final one for the whitelist.
 486                  */
 487                 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName );
 488
 489                 if( count( $ext ) ) {
 490                         $this->mFinalExtension = trim( $ext[count( $ext ) - 1] );
 491                 } else {
 492                         $this->mFinalExtension = '';
 493                 }
 494
 495                 /* Don't allow users to override the blacklist (check file extension) */
 496                 global $wgCheckFileExtensions, $wgStrictFileExtensions;
 497                 global $wgFileExtensions, $wgFileBlacklist;
 498                 if ( $this->mFinalExtension == '' ) {
 499                         $this->mTitleError = self::FILETYPE_MISSING;
 500                         return $this->mTitle = null;
 501                 } elseif ( $this->checkFileExtensionList( $ext, $wgFileBlacklist ) ||
 502                                 ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
 503                                         !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) ) ) {
 504                         $this->mTitleError = self::FILETYPE_BADTYPE;
 505                         return $this->mTitle = null;
 506                 }
 507
 508                 # If there was more than one "extension", reassemble the base
 509                 # filename to prevent bogus complaints about length
 510                 if( count( $ext ) > 1 ) {
 511                         for( $i = 0; $i < count( $ext ) - 1; $i++ ) {
 512                                 $partname .= '.' . $ext[$i];
 513                         }
 514                 }
 515
 516                 if( strlen( $partname ) < 1 ) {
 517                         $this->mTitleError =  self::MIN_LENGTH_PARTNAME;
 518                         return $this->mTitle = null;
 519                 }
 520
 521                 return $this->mTitle = $nt;
 522         }
 523
 524         /**
 525          * Return the local file and initializes if necessary.
 526          */
 527         public function getLocalFile() {
 528                 if( is_null( $this->mLocalFile ) ) {
 529                         $nt = $this->getTitle();
 530                         $this->mLocalFile = is_null( $nt ) ? null : wfLocalFile( $nt );
 531                 }
 532                 return $this->mLocalFile;
 533         }
 534
 535         /**
 536          * Stash a file in a temporary directory for later processing
 537          * after the user has confirmed it.
 538          *
 539          * If the user doesn't explicitly cancel or accept, these files
 540          * can accumulate in the temp directory.
 541          *
 542          * @param $saveName String: the destination filename
 543          * @param $tempSrc String: the source temporary file to save
 544          * @return String: full path the stashed file, or false on failure
 545          */
 546         protected function saveTempUploadedFile( $saveName, $tempSrc ) {
 547                 $repo = RepoGroup::singleton()->getLocalRepo();
 548                 $status = $repo->storeTemp( $saveName, $tempSrc );
 549                 return $status;
 550         }
 551
 552         /**
 553          * Stash a file in a temporary directory for later processing,
 554          * and save the necessary descriptive info into the session.
 555          * Returns a key value which will be passed through a form
 556          * to pick up the path info on a later invocation.
 557          *
 558          * @return Integer: session key
 559          */
 560         public function stashSession() {
 561                 $status = $this->saveTempUploadedFile( $this->mDestName, $this->mTempPath );
 562                 if( !$status->isOK() ) {
 563                         # Couldn't save the file.
 564                         return false;
 565                 }
 566
 567                 $key = $this->getSessionKey();
 568                 $_SESSION[self::SESSION_KEYNAME][$key] = array(
 569                         'mTempPath'       => $status->value,
 570                         'mFileSize'       => $this->mFileSize,
 571                         'mFileProps'      => $this->mFileProps,
 572                         'version'         => self::SESSION_VERSION,
 573                 );
 574                 return $key;
 575         }
 576
 577         /**
 578          * Generate a random session key from stash in cases where we want to start an upload without much information
 579          */
 580         protected function getSessionKey() {
 581                 $key = mt_rand( 0, 0x7fffffff );
 582                 $_SESSION[self::SESSION_KEYNAME][$key] = array();
 583                 return $key;
 584         }
 585
 586         /**
 587          * If we've modified the upload file we need to manually remove it
 588          * on exit to clean up.
 589          */
 590         public function cleanupTempFile() {
 591                 if ( $this->mRemoveTempFile && $this->mTempPath && file_exists( $this->mTempPath ) ) {
 592                         wfDebug( __METHOD__ . ": Removing temporary file {$this->mTempPath}\n" );
 593                         unlink( $this->mTempPath );
 594                 }
 595         }
 596
 597         public function getTempPath() {
 598                 return $this->mTempPath;
 599         }
 600
 601         /**
 602          * Split a file into a base name and all dot-delimited 'extensions'
 603          * on the end. Some web server configurations will fall back to
 604          * earlier pseudo-'extensions' to determine type and execute
 605          * scripts, so the blacklist needs to check them all.
 606          *
 607          * @return array
 608          */
 609         public static function splitExtensions( $filename ) {
 610                 $bits = explode( '.', $filename );
 611                 $basename = array_shift( $bits );
 612                 return array( $basename, $bits );
 613         }
 614
 615         /**
 616          * Perform case-insensitive match against a list of file extensions.
 617          * Returns true if the extension is in the list.
 618          *
 619          * @param $ext String
 620          * @param $list Array
 621          * @return Boolean
 622          */
 623         public static function checkFileExtension( $ext, $list ) {
 624                 return in_array( strtolower( $ext ), $list );
 625         }
 626
 627         /**
 628          * Perform case-insensitive match against a list of file extensions.
 629          * Returns true if any of the extensions are in the list.
 630          *
 631          * @param $ext Array
 632          * @param $list Array
 633          * @return Boolean
 634          */
 635         public static function checkFileExtensionList( $ext, $list ) {
 636                 foreach( $ext as $e ) {
 637                         if( in_array( strtolower( $e ), $list ) ) {
 638                                 return true;
 639                         }
 640                 }
 641                 return false;
 642         }
 643
 644         /**
 645          * Checks if the mime type of the uploaded file matches the file extension.
 646          *
 647          * @param $mime String: the mime type of the uploaded file
 648          * @param $extension String: the filename extension that the file is to be served with
 649          * @return Boolean
 650          */
 651         public static function verifyExtension( $mime, $extension ) {
 652                 $magic = MimeMagic::singleton();
 653
 654                 if ( !$mime || $mime == 'unknown' || $mime == 'unknown/unknown' )
 655                         if ( !$magic->isRecognizableExtension( $extension ) ) {
 656                                 wfDebug( __METHOD__ . ": passing file with unknown detected mime type; " .
 657                                         "unrecognized extension '$extension', can't verify\n" );
 658                                 return true;
 659                         } else {
 660                                 wfDebug( __METHOD__ . ": rejecting file with unknown detected mime type; ".
 661                                         "recognized extension '$extension', so probably invalid file\n" );
 662                                 return false;
 663                         }
 664
 665                 $match = $magic->isMatchingExtension( $extension, $mime );
 666
 667                 if ( $match === null ) {
 668                         wfDebug( __METHOD__ . ": no file extension known for mime type $mime, passing file\n" );
 669                         return true;
 670                 } elseif( $match === true ) {
 671                         wfDebug( __METHOD__ . ": mime type $mime matches extension $extension, passing file\n" );
 672
 673                         #TODO: if it's a bitmap, make sure PHP or ImageMagic resp. can handle it!
 674                         return true;
 675
 676                 } else {
 677                         wfDebug( __METHOD__ . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
 678                         return false;
 679                 }
 680         }
 681
 682         /**
 683          * Heuristic for detecting files that *could* contain JavaScript instructions or
 684          * things that may look like HTML to a browser and are thus
 685          * potentially harmful. The present implementation will produce false
 686          * positives in some situations.
 687          *
 688          * @param $file String: pathname to the temporary upload file
 689          * @param $mime String: the mime type of the file
 690          * @param $extension String: the extension of the file
 691          * @return Boolean: true if the file contains something looking like embedded scripts
 692          */
 693         public static function detectScript( $file, $mime, $extension ) {
 694                 global $wgAllowTitlesInSVG;
 695
 696                 # ugly hack: for text files, always look at the entire file.
 697                 # For binary field, just check the first K.
 698
 699                 if( strpos( $mime,'text/' ) === 0 ) {
 700                         $chunk = file_get_contents( $file );
 701                 } else {
 702                         $fp = fopen( $file, 'rb' );
 703                         $chunk = fread( $fp, 1024 );
 704                         fclose( $fp );
 705                 }
 706
 707                 $chunk = strtolower( $chunk );
 708
 709                 if( !$chunk ) {
 710                         return false;
 711                 }
 712
 713                 # decode from UTF-16 if needed (could be used for obfuscation).
 714                 if( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
 715                         $enc = 'UTF-16BE';
 716                 } elseif( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
 717                         $enc = 'UTF-16LE';
 718                 } else {
 719                         $enc = null;
 720                 }
 721
 722                 if( $enc ) {
 723                         $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
 724                 }
 725
 726                 $chunk = trim( $chunk );
 727
 728                 # FIXME: convert from UTF-16 if necessarry!
 729                 wfDebug( __METHOD__ . ": checking for embedded scripts and HTML stuff\n" );
 730
 731                 # check for HTML doctype
 732                 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
 733                         return true;
 734                 }
 735
 736                 /**
 737                  * Internet Explorer for Windows performs some really stupid file type
 738                  * autodetection which can cause it to interpret valid image files as HTML
 739                  * and potentially execute JavaScript, creating a cross-site scripting
 740                  * attack vectors.
 741                  *
 742                  * Apple's Safari browser also performs some unsafe file type autodetection
 743                  * which can cause legitimate files to be interpreted as HTML if the
 744                  * web server is not correctly configured to send the right content-type
 745                  * (or if you're really uploading plain text and octet streams!)
 746                  *
 747                  * Returns true if IE is likely to mistake the given file for HTML.
 748                  * Also returns true if Safari would mistake the given file for HTML
 749                  * when served with a generic content-type.
 750                  */
 751                 $tags = array(
 752                         '<a href',
 753                         '<body',
 754                         '<head',
 755                         '<html',   #also in safari
 756                         '<img',
 757                         '<pre',
 758                         '<script', #also in safari
 759                         '<table'
 760                 );
 761
 762                 if( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
 763                         $tags[] = '<title';
 764                 }
 765
 766                 foreach( $tags as $tag ) {
 767                         if( false !== strpos( $chunk, $tag ) ) {
 768                                 return true;
 769                         }
 770                 }
 771
 772                 /*
 773                  * look for JavaScript
 774                  */
 775
 776                 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
 777                 $chunk = Sanitizer::decodeCharReferences( $chunk );
 778
 779                 # look for script-types
 780                 if( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
 781                         return true;
 782                 }
 783
 784                 # look for html-style script-urls
 785                 if( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
 786                         return true;
 787                 }
 788
 789                 # look for css-style script-urls
 790                 if( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
 791                         return true;
 792                 }
 793
 794                 wfDebug( __METHOD__ . ": no scripts found\n" );
 795                 return false;
 796         }
 797
 798         protected function detectScriptInSvg( $filename ) {
 799                 $check = new XmlTypeCheck( $filename, array( $this, 'checkSvgScriptCallback' ) );
 800                 return $check->filterMatch;
 801         }
 802
 803         /**
 804          * @todo Replace this with a whitelist filter!
 805          */
 806         public function checkSvgScriptCallback( $element, $attribs ) {
 807                 $stripped = $this->stripXmlNamespace( $element );
 808
 809                 if( $stripped == 'script' ) {
 810                         wfDebug( __METHOD__ . ": Found script element '$element' in uploaded file.\n" );
 811                         return true;
 812                 }
 813
 814                 foreach( $attribs as $attrib => $value ) {
 815                         $stripped = $this->stripXmlNamespace( $attrib );
 816                         if( substr( $stripped, 0, 2 ) == 'on' ) {
 817                                 wfDebug( __METHOD__ . ": Found script attribute '$attrib'='value' in uploaded file.\n" );
 818                                 return true;
 819                         }
 820                         if( $stripped == 'href' && strpos( strtolower( $value ), 'javascript:' ) !== false ) {
 821                                 wfDebug( __METHOD__ . ": Found script href attribute '$attrib'='$value' in uploaded file.\n" );
 822                                 return true;
 823                         }
 824                 }
 825         }
 826
 827         private function stripXmlNamespace( $name ) {
 828                 // 'http://www.w3.org/2000/svg:script' -> 'script'
 829                 $parts = explode( ':', strtolower( $name ) );
 830                 return array_pop( $parts );
 831         }
 832
 833         /**
 834          * Generic wrapper function for a virus scanner program.
 835          * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
 836          * $wgAntivirusRequired may be used to deny upload if the scan fails.
 837          *
 838          * @param $file String: pathname to the temporary upload file
 839          * @return mixed false if not virus is found, NULL if the scan fails or is disabled,
 840          *         or a string containing feedback from the virus scanner if a virus was found.
 841          *         If textual feedback is missing but a virus was found, this function returns true.
 842          */
 843         public static function detectVirus( $file ) {
 844                 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
 845
 846                 if ( !$wgAntivirus ) {
 847                         wfDebug( __METHOD__ . ": virus scanner disabled\n" );
 848                         return null;
 849                 }
 850
 851                 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
 852                         wfDebug( __METHOD__ . ": unknown virus scanner: $wgAntivirus\n" );
 853                         $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1</div>", array( 'virus-badscanner', $wgAntivirus ) );
 854                         return wfMsg( 'virus-unknownscanner' ) . " $wgAntivirus";
 855                 }
 856
 857                 # look up scanner configuration
 858                 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
 859                 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
 860                 $msgPattern = isset( $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ) ?
 861                         $wgAntivirusSetup[$wgAntivirus]['messagepattern'] : null;
 862
 863                 if ( strpos( $command, "%f" ) === false ) {
 864                         # simple pattern: append file to scan
 865                         $command .= " " . wfEscapeShellArg( $file );
 866                 } else {
 867                         # complex pattern: replace "%f" with file to scan
 868                         $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
 869                 }
 870
 871                 wfDebug( __METHOD__ . ": running virus scan: $command \n" );
 872
 873                 # execute virus scanner
 874                 $exitCode = false;
 875
 876                 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
 877                 #      that does not seem to be worth the pain.
 878                 #      Ask me (Duesentrieb) about it if it's ever needed.
 879                 $output = wfShellExec( "$command 2>&1", $exitCode );
 880
 881                 # map exit code to AV_xxx constants.
 882                 $mappedCode = $exitCode;
 883                 if ( $exitCodeMap ) {
 884                         if ( isset( $exitCodeMap[$exitCode] ) ) {
 885                                 $mappedCode = $exitCodeMap[$exitCode];
 886                         } elseif ( isset( $exitCodeMap["*"] ) ) {
 887                                 $mappedCode = $exitCodeMap["*"];
 888                         }
 889                 }
 890
 891                 if ( $mappedCode === AV_SCAN_FAILED ) {
 892                         # scan failed (code was mapped to false by $exitCodeMap)
 893                         wfDebug( __METHOD__ . ": failed to scan $file (code $exitCode).\n" );
 894
 895                         if ( $wgAntivirusRequired ) {
 896                                 return wfMsg( 'virus-scanfailed', array( $exitCode ) );
 897                         } else {
 898                                 return null;
 899                         }
 900                 } elseif ( $mappedCode === AV_SCAN_ABORTED ) {
 901                         # scan failed because filetype is unknown (probably imune)
 902                         wfDebug( __METHOD__ . ": unsupported file type $file (code $exitCode).\n" );
 903                         return null;
 904                 } elseif ( $mappedCode === AV_NO_VIRUS ) {
 905                         # no virus found
 906                         wfDebug( __METHOD__ . ": file passed virus scan.\n" );
 907                         return false;
 908                 } else {
 909                         $output = trim( $output );
 910
 911                         if ( !$output ) {
 912                                 $output = true; #if there's no output, return true
 913                         } elseif ( $msgPattern ) {
 914                                 $groups = array();
 915                                 if ( preg_match( $msgPattern, $output, $groups ) ) {
 916                                         if ( $groups[1] ) {
 917                                                 $output = $groups[1];
 918                                         }
 919                                 }
 920                         }
 921
 922                         wfDebug( __METHOD__ . ": FOUND VIRUS! scanner feedback: $output \n" );
 923                         return $output;
 924                 }
 925         }
 926
 927         /**
 928          * Check if the temporary file is MacBinary-encoded, as some uploads
 929          * from Internet Explorer on Mac OS Classic and Mac OS X will be.
 930          * If so, the data fork will be extracted to a second temporary file,
 931          * which will then be checked for validity and either kept or discarded.
 932          */
 933         private function checkMacBinary() {
 934                 $macbin = new MacBinary( $this->mTempPath );
 935                 if( $macbin->isValid() ) {
 936                         $dataFile = tempnam( wfTempDir(), 'WikiMacBinary' );
 937                         $dataHandle = fopen( $dataFile, 'wb' );
 938
 939                         wfDebug( __METHOD__ . ": Extracting MacBinary data fork to $dataFile\n" );
 940                         $macbin->extractData( $dataHandle );
 941
 942                         $this->mTempPath = $dataFile;
 943                         $this->mFileSize = $macbin->dataForkLength();
 944
 945                         // We'll have to manually remove the new file if it's not kept.
 946                         $this->mRemoveTempFile = true;
 947                 }
 948                 $macbin->close();
 949         }
 950
 951         /**
 952          * Check if there's an overwrite conflict and, if so, if restrictions
 953          * forbid this user from performing the upload.
 954          *
 955          * @return mixed true on success, error string on failure
 956          */
 957         private function checkOverwrite() {
 958                 global $wgUser;
 959                 // First check whether the local file can be overwritten
 960                 $file = $this->getLocalFile();
 961                 if( $file->exists() ) {
 962                         if( !self::userCanReUpload( $wgUser, $file ) ) {
 963                                 return 'fileexists-forbidden';
 964                         } else {
 965                                 return true;
 966                         }
 967                 }
 968
 969                 /* Check shared conflicts: if the local file does not exist, but
 970                  * wfFindFile finds a file, it exists in a shared repository.
 971                  */
 972                 $file = wfFindFile( $this->getTitle() );
 973                 if ( $file && !$wgUser->isAllowed( 'reupload-shared' ) ) {
 974                         return 'fileexists-shared-forbidden';
 975                 }
 976
 977                 return true;
 978         }
 979
 980         /**
 981          * Check if a user is the last uploader
 982          *
 983          * @param $user User object
 984          * @param $img String: image name
 985          * @return Boolean
 986          */
 987         public static function userCanReUpload( User $user, $img ) {
 988                 if( $user->isAllowed( 'reupload' ) ) {
 989                         return true; // non-conditional
 990                 }
 991                 if( !$user->isAllowed( 'reupload-own' ) ) {
 992                         return false;
 993                 }
 994                 if( is_string( $img ) ) {
 995                         $img = wfLocalFile( $img );
 996                 }
 997                 if ( !( $img instanceof LocalFile ) ) {
 998                         return false;
 999                 }
1000
1001                 return $user->getId() == $img->getUser( 'id' );
1002         }
1003
1004         /**
1005          * Helper function that does various existence checks for a file.
1006          * The following checks are performed:
1007          * - The file exists
1008          * - Article with the same name as the file exists
1009          * - File exists with normalized extension
1010          * - The file looks like a thumbnail and the original exists
1011          *
1012          * @param $file The File object to check
1013          * @return mixed False if the file does not exists, else an array
1014          */
1015         public static function getExistsWarning( $file ) {
1016                 if( $file->exists() ) {
1017                         return array( 'warning' => 'exists', 'file' => $file );
1018                 }
1019
1020                 if( $file->getTitle()->getArticleID() ) {
1021                         return array( 'warning' => 'page-exists', 'file' => $file );
1022                 }
1023
1024                 if ( $file->wasDeleted() && !$file->exists() ) {
1025                         return array( 'warning' => 'was-deleted', 'file' => $file );
1026                 }
1027
1028                 if( strpos( $file->getName(), '.' ) == false ) {
1029                         $partname = $file->getName();
1030                         $extension = '';
1031                 } else {
1032                         $n = strrpos( $file->getName(), '.' );
1033                         $extension = substr( $file->getName(), $n + 1 );
1034                         $partname = substr( $file->getName(), 0, $n );
1035                 }
1036                 $normalizedExtension = File::normalizeExtension( $extension );
1037
1038                 if ( $normalizedExtension != $extension ) {
1039                         // We're not using the normalized form of the extension.
1040                         // Normal form is lowercase, using most common of alternate
1041                         // extensions (eg 'jpg' rather than 'JPEG').
1042                         //
1043                         // Check for another file using the normalized form...
1044                         $nt_lc = Title::makeTitle( NS_FILE, "{$partname}.{$normalizedExtension}" );
1045                         $file_lc = wfLocalFile( $nt_lc );
1046
1047                         if( $file_lc->exists() ) {
1048                                 return array(
1049                                         'warning' => 'exists-normalized',
1050                                         'file' => $file,
1051                                         'normalizedFile' => $file_lc
1052                                 );
1053                         }
1054                 }
1055
1056                 if ( self::isThumbName( $file->getName() ) ) {
1057                         # Check for filenames like 50px- or 180px-, these are mostly thumbnails
1058                         $nt_thb = Title::newFromText( substr( $partname , strpos( $partname , '-' ) +1 ) . '.' . $extension, NS_FILE );
1059                         $file_thb = wfLocalFile( $nt_thb );
1060                         if( $file_thb->exists() ) {
1061                                 return array(
1062                                         'warning' => 'thumb',
1063                                         'file' => $file,
1064                                         'thumbFile' => $file_thb
1065                                 );
1066                         } else {
1067                                 // File does not exist, but we just don't like the name
1068                                 return array(
1069                                         'warning' => 'thumb-name',
1070                                         'file' => $file,
1071                                         'thumbFile' => $file_thb
1072                                 );
1073                         }
1074                 }
1075
1076
1077                 foreach( self::getFilenamePrefixBlacklist() as $prefix ) {
1078                         if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
1079                                 return array(
1080                                         'warning' => 'bad-prefix',
1081                                         'file' => $file,
1082                                         'prefix' => $prefix
1083                                 );
1084                         }
1085                 }
1086
1087                 return false;
1088         }
1089
1090         /**
1091          * Helper function that checks whether the filename looks like a thumbnail
1092          */
1093         public static function isThumbName( $filename ) {
1094                 $n = strrpos( $filename, '.' );
1095                 $partname = $n ? substr( $filename, 0, $n ) : $filename;
1096                 return (
1097                                         substr( $partname , 3, 3 ) == 'px-' ||
1098                                         substr( $partname , 2, 3 ) == 'px-'
1099                                 ) &&
1100                                 preg_match( "/[0-9]{2}/" , substr( $partname , 0, 2 ) );
1101         }
1102
1103         /**
1104          * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
1105          *
1106          * @return array list of prefixes
1107          */
1108         public static function getFilenamePrefixBlacklist() {
1109                 $blacklist = array();
1110                 $message = wfMsgForContent( 'filename-prefix-blacklist' );
1111                 if( $message && !( wfEmptyMsg( 'filename-prefix-blacklist', $message ) || $message == '-' ) ) {
1112                         $lines = explode( "\n", $message );
1113                         foreach( $lines as $line ) {
1114                                 // Remove comment lines
1115                                 $comment = substr( trim( $line ), 0, 1 );
1116                                 if ( $comment == '#' || $comment == '' ) {
1117                                         continue;
1118                                 }
1119                                 // Remove additional comments after a prefix
1120                                 $comment = strpos( $line, '#' );
1121                                 if ( $comment > 0 ) {
1122                                         $line = substr( $line, 0, $comment-1 );
1123                                 }
1124                                 $blacklist[] = trim( $line );
1125                         }
1126                 }
1127                 return $blacklist;
1128         }
1129
1130         public function getImageInfo( $result ) {
1131                 $file = $this->getLocalFile();
1132                 $imParam = ApiQueryImageInfo::getPropertyNames();
1133                 return ApiQueryImageInfo::getInfo( $file, array_flip( $imParam ), $result );
1134         }
1135
1136 }