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