includes/upload/UploadBase.php

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