9 * UploadBase and subclasses are the backend of MediaWiki's file uploads.
10 * The frontends are formed by ApiUpload and SpecialUpload.
12 * See also includes/docs/upload.txt
14 * @author Brion Vibber
15 * @author Bryan Tong Minh
16 * @author Michael Dale
18 abstract class UploadBase
{
20 protected $mDesiredDestName, $mDestName, $mRemoveTempFile, $mSourceType;
21 protected $mTitle = false, $mTitleError = 0;
22 protected $mFilteredName, $mFinalExtension;
23 protected $mLocalFile, $mFileSize, $mFileProps;
24 protected $mBlackListedExtensions;
25 protected $mJavaDetected;
30 const MIN_LENGTH_PARTNAME
= 4;
31 const ILLEGAL_FILENAME
= 5;
32 const OVERWRITE_EXISTING_FILE
= 7; # Not used anymore; handled by verifyTitlePermissions()
33 const FILETYPE_MISSING
= 8;
34 const FILETYPE_BADTYPE
= 9;
35 const VERIFICATION_ERROR
= 10;
37 # HOOK_ABORTED is the new name of UPLOAD_VERIFICATION_ERROR
38 const UPLOAD_VERIFICATION_ERROR
= 11;
39 const HOOK_ABORTED
= 11;
40 const FILE_TOO_LARGE
= 12;
41 const WINDOWS_NONASCII_FILENAME
= 13;
42 const FILENAME_TOO_LONG
= 14;
44 public function getVerificationErrorCode( $error ) {
45 $code_to_status = array(self
::EMPTY_FILE
=> 'empty-file',
46 self
::FILE_TOO_LARGE
=> 'file-too-large',
47 self
::FILETYPE_MISSING
=> 'filetype-missing',
48 self
::FILETYPE_BADTYPE
=> 'filetype-banned',
49 self
::MIN_LENGTH_PARTNAME
=> 'filename-tooshort',
50 self
::ILLEGAL_FILENAME
=> 'illegal-filename',
51 self
::OVERWRITE_EXISTING_FILE
=> 'overwrite',
52 self
::VERIFICATION_ERROR
=> 'verification-error',
53 self
::HOOK_ABORTED
=> 'hookaborted',
54 self
::WINDOWS_NONASCII_FILENAME
=> 'windows-nonascii-filename',
55 self
::FILENAME_TOO_LONG
=> 'filename-toolong',
57 if( isset( $code_to_status[$error] ) ) {
58 return $code_to_status[$error];
61 return 'unknown-error';
65 * Returns true if uploads are enabled.
66 * Can be override by subclasses.
69 public static function isEnabled() {
70 global $wgEnableUploads;
72 if ( !$wgEnableUploads ) {
76 # Check php's file_uploads setting
77 return wfIsHipHop() ||
wfIniGetBool( 'file_uploads' );
81 * Returns true if the user can use this upload module or else a string
82 * identifying the missing permission.
83 * Can be overriden by subclasses.
88 public static function isAllowed( $user ) {
89 foreach ( array( 'upload', 'edit' ) as $permission ) {
90 if ( !$user->isAllowed( $permission ) ) {
97 // Upload handlers. Should probably just be a global.
98 static $uploadHandlers = array( 'Stash', 'File', 'Url' );
101 * Create a form of UploadBase depending on wpSourceType and initializes it
103 * @param $request WebRequest
107 public static function createFromRequest( &$request, $type = null ) {
108 $type = $type ?
$type : $request->getVal( 'wpSourceType', 'File' );
114 // Get the upload class
115 $type = ucfirst( $type );
117 // Give hooks the chance to handle this request
119 wfRunHooks( 'UploadCreateFromRequest', array( $type, &$className ) );
120 if ( is_null( $className ) ) {
121 $className = 'UploadFrom' . $type;
122 wfDebug( __METHOD__
. ": class name: $className\n" );
123 if( !in_array( $type, self
::$uploadHandlers ) ) {
128 // Check whether this upload class is enabled
129 if( !call_user_func( array( $className, 'isEnabled' ) ) ) {
133 // Check whether the request is valid
134 if( !call_user_func( array( $className, 'isValidRequest' ), $request ) ) {
138 $handler = new $className;
140 $handler->initializeFromRequest( $request );
145 * Check whether a request if valid for this handler
148 public static function isValidRequest( $request ) {
152 public function __construct() {}
155 * Returns the upload type. Should be overridden by child classes
160 public function getSourceType() { return null; }
163 * Initialize the path information
164 * @param $name string the desired destination name
165 * @param $tempPath string the temporary path
166 * @param $fileSize int the file size
167 * @param $removeTempFile bool (false) remove the temporary file?
170 public function initializePathInfo( $name, $tempPath, $fileSize, $removeTempFile = false ) {
171 $this->mDesiredDestName
= $name;
172 if ( FileBackend
::isStoragePath( $tempPath ) ) {
173 throw new MWException( __METHOD__
. " given storage path `$tempPath`." );
175 $this->mTempPath
= $tempPath;
176 $this->mFileSize
= $fileSize;
177 $this->mRemoveTempFile
= $removeTempFile;
181 * Initialize from a WebRequest. Override this in a subclass.
183 public abstract function initializeFromRequest( &$request );
186 * Fetch the file. Usually a no-op
189 public function fetchFile() {
190 return Status
::newGood();
194 * Return true if the file is empty
197 public function isEmptyFile() {
198 return empty( $this->mFileSize
);
202 * Return the file size
205 public function getFileSize() {
206 return $this->mFileSize
;
210 * @param $srcPath String: the source path
211 * @return stringthe real path if it was a virtual URL
213 function getRealPath( $srcPath ) {
214 $repo = RepoGroup
::singleton()->getLocalRepo();
215 if ( $repo->isVirtualUrl( $srcPath ) ) {
216 // @TODO: just make uploads work with storage paths
217 // UploadFromStash loads files via virtuals URLs
218 $tmpFile = $repo->getLocalCopy( $srcPath );
219 $tmpFile->bind( $this ); // keep alive with $thumb
220 return $tmpFile->getPath();
226 * Verify whether the upload is sane.
227 * @return mixed self::OK or else an array with error information
229 public function verifyUpload() {
231 * If there was no filename or a zero size given, give up quick.
233 if( $this->isEmptyFile() ) {
234 return array( 'status' => self
::EMPTY_FILE
);
238 * Honor $wgMaxUploadSize
240 $maxSize = self
::getMaxUploadSize( $this->getSourceType() );
241 if( $this->mFileSize
> $maxSize ) {
243 'status' => self
::FILE_TOO_LARGE
,
249 * Look at the contents of the file; if we can recognize the
250 * type but it's corrupt or data of the wrong type, we should
251 * probably not accept it.
253 $verification = $this->verifyFile();
254 if( $verification !== true ) {
256 'status' => self
::VERIFICATION_ERROR
,
257 'details' => $verification
262 * Make sure this file can be created
264 $result = $this->validateName();
265 if( $result !== true ) {
270 if( !wfRunHooks( 'UploadVerification',
271 array( $this->mDestName
, $this->mTempPath
, &$error ) ) ) {
272 return array( 'status' => self
::HOOK_ABORTED
, 'error' => $error );
275 return array( 'status' => self
::OK
);
279 * Verify that the name is valid and, if necessary, that we can overwrite
281 * @return mixed true if valid, otherwise and array with 'status'
284 protected function validateName() {
285 $nt = $this->getTitle();
286 if( is_null( $nt ) ) {
287 $result = array( 'status' => $this->mTitleError
);
288 if( $this->mTitleError
== self
::ILLEGAL_FILENAME
) {
289 $result['filtered'] = $this->mFilteredName
;
291 if ( $this->mTitleError
== self
::FILETYPE_BADTYPE
) {
292 $result['finalExt'] = $this->mFinalExtension
;
293 if ( count( $this->mBlackListedExtensions
) ) {
294 $result['blacklistedExt'] = $this->mBlackListedExtensions
;
299 $this->mDestName
= $this->getLocalFile()->getName();
305 * Verify the mime type
307 * @param $mime string representing the mime
308 * @return mixed true if the file is verified, an array otherwise
310 protected function verifyMimeType( $mime ) {
311 global $wgVerifyMimeType;
312 if ( $wgVerifyMimeType ) {
313 wfDebug ( "\n\nmime: <$mime> extension: <{$this->mFinalExtension}>\n\n");
314 global $wgMimeTypeBlacklist;
315 if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
316 return array( 'filetype-badmime', $mime );
319 # XXX: Missing extension will be caught by validateName() via getTitle()
320 if ( $this->mFinalExtension
!= '' && !$this->verifyExtension( $mime, $this->mFinalExtension
) ) {
321 return array( 'filetype-mime-mismatch', $this->mFinalExtension
, $mime );
325 $fp = fopen( $this->mTempPath
, 'rb' );
326 $chunk = fread( $fp, 256 );
329 $magic = MimeMagic
::singleton();
330 $extMime = $magic->guessTypesForExtension( $this->mFinalExtension
);
331 $ieTypes = $magic->getIEMimeTypes( $this->mTempPath
, $chunk, $extMime );
332 foreach ( $ieTypes as $ieType ) {
333 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
334 return array( 'filetype-bad-ie-mime', $ieType );
343 * Verifies that it's ok to include the uploaded file
345 * @return mixed true of the file is verified, array otherwise.
347 protected function verifyFile() {
348 global $wgAllowJavaUploads, $wgDisableUploadScriptChecks;
349 # get the title, even though we are doing nothing with it, because
350 # we need to populate mFinalExtension
353 $this->mFileProps
= FSFile
::getPropsFromPath( $this->mTempPath
, $this->mFinalExtension
);
355 # check mime type, if desired
356 $mime = $this->mFileProps
[ 'file-mime' ];
357 $status = $this->verifyMimeType( $mime );
358 if ( $status !== true ) {
362 # check for htmlish code and javascript
363 if ( !$wgDisableUploadScriptChecks ) {
364 if( self
::detectScript( $this->mTempPath
, $mime, $this->mFinalExtension
) ) {
365 return array( 'uploadscripted' );
367 if( $this->mFinalExtension
== 'svg' ||
$mime == 'image/svg+xml' ) {
368 if( $this->detectScriptInSvg( $this->mTempPath
) ) {
369 return array( 'uploadscripted' );
374 # Check for Java applets, which if uploaded can bypass cross-site
376 if ( !$wgAllowJavaUploads ) {
377 $this->mJavaDetected
= false;
378 $zipStatus = ZipDirectoryReader
::read( $this->mTempPath
,
379 array( $this, 'zipEntryCallback' ) );
380 if ( !$zipStatus->isOK() ) {
381 $errors = $zipStatus->getErrorsArray();
382 $error = reset( $errors );
383 if ( $error[0] !== 'zip-wrong-format' ) {
387 if ( $this->mJavaDetected
) {
388 return array( 'uploadjava' );
392 # Scan the uploaded file for viruses
393 $virus = $this->detectVirus( $this->mTempPath
);
395 return array( 'uploadvirus', $virus );
398 $handler = MediaHandler
::getHandler( $mime );
400 $handlerStatus = $handler->verifyUpload( $this->mTempPath
);
401 if ( !$handlerStatus->isOK() ) {
402 $errors = $handlerStatus->getErrorsArray();
403 return reset( $errors );
407 wfRunHooks( 'UploadVerifyFile', array( $this, $mime, &$status ) );
408 if ( $status !== true ) {
412 wfDebug( __METHOD__
. ": all clear; passing.\n" );
417 * Callback for ZipDirectoryReader to detect Java class files.
419 function zipEntryCallback( $entry ) {
420 $names = array( $entry['name'] );
422 // If there is a null character, cut off the name at it, because JDK's
423 // ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name
424 // were constructed which had ".class\0" followed by a string chosen to
425 // make the hash collide with the truncated name, that file could be
426 // returned in response to a request for the .class file.
427 $nullPos = strpos( $entry['name'], "\000" );
428 if ( $nullPos !== false ) {
429 $names[] = substr( $entry['name'], 0, $nullPos );
432 // If there is a trailing slash in the file name, we have to strip it,
433 // because that's what ZIP_GetEntry() does.
434 if ( preg_grep( '!\.class/?$!', $names ) ) {
435 $this->mJavaDetected
= true;
440 * Alias for verifyTitlePermissions. The function was originally 'verifyPermissions'
441 * but that suggests it's checking the user, when it's really checking the title + user combination.
442 * @param $user User object to verify the permissions against
443 * @return mixed An array as returned by getUserPermissionsErrors or true
444 * in case the user has proper permissions.
446 public function verifyPermissions( $user ) {
447 return $this->verifyTitlePermissions( $user );
451 * Check whether the user can edit, upload and create the image. This
452 * checks only against the current title; if it returns errors, it may
453 * very well be that another title will not give errors. Therefore
454 * isAllowed() should be called as well for generic is-user-blocked or
455 * can-user-upload checking.
457 * @param $user User object to verify the permissions against
458 * @return mixed An array as returned by getUserPermissionsErrors or true
459 * in case the user has proper permissions.
461 public function verifyTitlePermissions( $user ) {
463 * If the image is protected, non-sysop users won't be able
464 * to modify it by uploading a new revision.
466 $nt = $this->getTitle();
467 if( is_null( $nt ) ) {
470 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
471 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
472 if ( !$nt->exists() ) {
473 $permErrorsCreate = $nt->getUserPermissionsErrors( 'create', $user );
475 $permErrorsCreate = array();
477 if( $permErrors ||
$permErrorsUpload ||
$permErrorsCreate ) {
478 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
479 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
483 $overwriteError = $this->checkOverwrite( $user );
484 if ( $overwriteError !== true ) {
485 return array( $overwriteError );
492 * Check for non fatal problems with the file
494 * @return Array of warnings
496 public function checkWarnings() {
501 $localFile = $this->getLocalFile();
502 $filename = $localFile->getName();
505 * Check whether the resulting filename is different from the desired one,
506 * but ignore things like ucfirst() and spaces/underscore things
508 $comparableName = str_replace( ' ', '_', $this->mDesiredDestName
);
509 $comparableName = Title
::capitalize( $comparableName, NS_FILE
);
511 if( $this->mDesiredDestName
!= $filename && $comparableName != $filename ) {
512 $warnings['badfilename'] = $filename;
515 // Check whether the file extension is on the unwanted list
516 global $wgCheckFileExtensions, $wgFileExtensions;
517 if ( $wgCheckFileExtensions ) {
518 if ( !$this->checkFileExtension( $this->mFinalExtension
, $wgFileExtensions ) ) {
519 $warnings['filetype-unwanted-type'] = array( $this->mFinalExtension
,
520 $wgLang->commaList( $wgFileExtensions ), count( $wgFileExtensions ) );
524 global $wgUploadSizeWarning;
525 if ( $wgUploadSizeWarning && ( $this->mFileSize
> $wgUploadSizeWarning ) ) {
526 $warnings['large-file'] = $wgUploadSizeWarning;
529 if ( $this->mFileSize
== 0 ) {
530 $warnings['emptyfile'] = true;
533 $exists = self
::getExistsWarning( $localFile );
534 if( $exists !== false ) {
535 $warnings['exists'] = $exists;
538 // Check dupes against existing files
539 $hash = FSFile
::getSha1Base36FromPath( $this->mTempPath
);
540 $dupes = RepoGroup
::singleton()->findBySha1( $hash );
541 $title = $this->getTitle();
542 // Remove all matches against self
543 foreach ( $dupes as $key => $dupe ) {
544 if( $title->equals( $dupe->getTitle() ) ) {
545 unset( $dupes[$key] );
549 $warnings['duplicate'] = $dupes;
552 // Check dupes against archives
553 $archivedImage = new ArchivedFile( null, 0, "{$hash}.{$this->mFinalExtension}" );
554 if ( $archivedImage->getID() > 0 ) {
555 $warnings['duplicate-archive'] = $archivedImage->getName();
562 * Really perform the upload. Stores the file in the local repo, watches
563 * if necessary and runs the UploadComplete hook.
567 * @return Status indicating the whether the upload succeeded.
569 public function performUpload( $comment, $pageText, $watch, $user ) {
570 $status = $this->getLocalFile()->upload(
580 if( $status->isGood() ) {
582 $user->addWatch( $this->getLocalFile()->getTitle() );
585 wfRunHooks( 'UploadComplete', array( &$this ) );
592 * Returns the title of the file to be uploaded. Sets mTitleError in case
593 * the name was illegal.
595 * @return Title The title of the file or null in case the name was illegal
597 public function getTitle() {
598 if ( $this->mTitle
!== false ) {
599 return $this->mTitle
;
602 /* Assume that if a user specified File:Something.jpg, this is an error
603 * and that the namespace prefix needs to be stripped of.
605 $title = Title
::newFromText( $this->mDesiredDestName
);
606 if ( $title && $title->getNamespace() == NS_FILE
) {
607 $this->mFilteredName
= $title->getDBkey();
609 $this->mFilteredName
= $this->mDesiredDestName
;
612 # oi_archive_name is max 255 bytes, which include a timestamp and an
613 # exclamation mark, so restrict file name to 240 bytes.
614 if ( strlen( $this->mFilteredName
) > 240 ) {
615 $this->mTitleError
= self
::FILENAME_TOO_LONG
;
616 return $this->mTitle
= null;
620 * Chop off any directories in the given filename. Then
621 * filter out illegal characters, and try to make a legible name
622 * out of it. We'll strip some silently that Title would die on.
624 $this->mFilteredName
= wfStripIllegalFilenameChars( $this->mFilteredName
);
625 /* Normalize to title form before we do any further processing */
626 $nt = Title
::makeTitleSafe( NS_FILE
, $this->mFilteredName
);
627 if( is_null( $nt ) ) {
628 $this->mTitleError
= self
::ILLEGAL_FILENAME
;
629 return $this->mTitle
= null;
631 $this->mFilteredName
= $nt->getDBkey();
636 * We'll want to blacklist against *any* 'extension', and use
637 * only the final one for the whitelist.
639 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName
);
641 if( count( $ext ) ) {
642 $this->mFinalExtension
= trim( $ext[count( $ext ) - 1] );
644 $this->mFinalExtension
= '';
646 # No extension, try guessing one
647 $magic = MimeMagic
::singleton();
648 $mime = $magic->guessMimeType( $this->mTempPath
);
649 if ( $mime !== 'unknown/unknown' ) {
650 # Get a space separated list of extensions
651 $extList = $magic->getExtensionsForType( $mime );
653 # Set the extension to the canonical extension
654 $this->mFinalExtension
= strtok( $extList, ' ' );
656 # Fix up the other variables
657 $this->mFilteredName
.= ".{$this->mFinalExtension}";
658 $nt = Title
::makeTitleSafe( NS_FILE
, $this->mFilteredName
);
659 $ext = array( $this->mFinalExtension
);
665 /* Don't allow users to override the blacklist (check file extension) */
666 global $wgCheckFileExtensions, $wgStrictFileExtensions;
667 global $wgFileExtensions, $wgFileBlacklist;
669 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
671 if ( $this->mFinalExtension
== '' ) {
672 $this->mTitleError
= self
::FILETYPE_MISSING
;
673 return $this->mTitle
= null;
674 } elseif ( $blackListedExtensions ||
675 ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
676 !$this->checkFileExtensionList( $ext, $wgFileExtensions ) ) ) {
677 $this->mBlackListedExtensions
= $blackListedExtensions;
678 $this->mTitleError
= self
::FILETYPE_BADTYPE
;
679 return $this->mTitle
= null;
682 // Windows may be broken with special characters, see bug XXX
683 if ( wfIsWindows() && !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() ) ) {
684 $this->mTitleError
= self
::WINDOWS_NONASCII_FILENAME
;
685 return $this->mTitle
= null;
688 # If there was more than one "extension", reassemble the base
689 # filename to prevent bogus complaints about length
690 if( count( $ext ) > 1 ) {
691 for( $i = 0; $i < count( $ext ) - 1; $i++
) {
692 $partname .= '.' . $ext[$i];
696 if( strlen( $partname ) < 1 ) {
697 $this->mTitleError
= self
::MIN_LENGTH_PARTNAME
;
698 return $this->mTitle
= null;
701 return $this->mTitle
= $nt;
705 * Return the local file and initializes if necessary.
709 public function getLocalFile() {
710 if( is_null( $this->mLocalFile
) ) {
711 $nt = $this->getTitle();
712 $this->mLocalFile
= is_null( $nt ) ?
null : wfLocalFile( $nt );
714 return $this->mLocalFile
;
718 * If the user does not supply all necessary information in the first upload form submission (either by accident or
719 * by design) then we may want to stash the file temporarily, get more information, and publish the file later.
721 * This method will stash a file in a temporary directory for later processing, and save the necessary descriptive info
723 * This method returns the file object, which also has a 'fileKey' property which can be passed through a form or
724 * API request to find this stashed file again.
726 * @return UploadStashFile stashed file
728 public function stashFile() {
729 // was stashSessionFile
730 $stash = RepoGroup
::singleton()->getLocalRepo()->getUploadStash();
731 $file = $stash->stashFile( $this->mTempPath
, $this->getSourceType() );
732 $this->mLocalFile
= $file;
737 * Stash a file in a temporary directory, returning a key which can be used to find the file again. See stashFile().
739 * @return String: file key
741 public function stashFileGetKey() {
742 return $this->stashFile()->getFileKey();
746 * alias for stashFileGetKey, for backwards compatibility
748 * @return String: file key
750 public function stashSession() {
751 return $this->stashFileGetKey();
755 * If we've modified the upload file we need to manually remove it
756 * on exit to clean up.
758 public function cleanupTempFile() {
759 if ( $this->mRemoveTempFile
&& $this->mTempPath
&& file_exists( $this->mTempPath
) ) {
760 wfDebug( __METHOD__
. ": Removing temporary file {$this->mTempPath}\n" );
761 unlink( $this->mTempPath
);
765 public function getTempPath() {
766 return $this->mTempPath
;
770 * Split a file into a base name and all dot-delimited 'extensions'
771 * on the end. Some web server configurations will fall back to
772 * earlier pseudo-'extensions' to determine type and execute
773 * scripts, so the blacklist needs to check them all.
777 public static function splitExtensions( $filename ) {
778 $bits = explode( '.', $filename );
779 $basename = array_shift( $bits );
780 return array( $basename, $bits );
784 * Perform case-insensitive match against a list of file extensions.
785 * Returns true if the extension is in the list.
791 public static function checkFileExtension( $ext, $list ) {
792 return in_array( strtolower( $ext ), $list );
796 * Perform case-insensitive match against a list of file extensions.
797 * Returns an array of matching extensions.
803 public static function checkFileExtensionList( $ext, $list ) {
804 return array_intersect( array_map( 'strtolower', $ext ), $list );
808 * Checks if the mime type of the uploaded file matches the file extension.
810 * @param $mime String: the mime type of the uploaded file
811 * @param $extension String: the filename extension that the file is to be served with
814 public static function verifyExtension( $mime, $extension ) {
815 $magic = MimeMagic
::singleton();
817 if ( !$mime ||
$mime == 'unknown' ||
$mime == 'unknown/unknown' )
818 if ( !$magic->isRecognizableExtension( $extension ) ) {
819 wfDebug( __METHOD__
. ": passing file with unknown detected mime type; " .
820 "unrecognized extension '$extension', can't verify\n" );
823 wfDebug( __METHOD__
. ": rejecting file with unknown detected mime type; ".
824 "recognized extension '$extension', so probably invalid file\n" );
828 $match = $magic->isMatchingExtension( $extension, $mime );
830 if ( $match === null ) {
831 wfDebug( __METHOD__
. ": no file extension known for mime type $mime, passing file\n" );
833 } elseif( $match === true ) {
834 wfDebug( __METHOD__
. ": mime type $mime matches extension $extension, passing file\n" );
836 #TODO: if it's a bitmap, make sure PHP or ImageMagic resp. can handle it!
840 wfDebug( __METHOD__
. ": mime type $mime mismatches file extension $extension, rejecting file\n" );
846 * Heuristic for detecting files that *could* contain JavaScript instructions or
847 * things that may look like HTML to a browser and are thus
848 * potentially harmful. The present implementation will produce false
849 * positives in some situations.
851 * @param $file String: pathname to the temporary upload file
852 * @param $mime String: the mime type of the file
853 * @param $extension String: the extension of the file
854 * @return Boolean: true if the file contains something looking like embedded scripts
856 public static function detectScript( $file, $mime, $extension ) {
857 global $wgAllowTitlesInSVG;
859 # ugly hack: for text files, always look at the entire file.
860 # For binary field, just check the first K.
862 if( strpos( $mime,'text/' ) === 0 ) {
863 $chunk = file_get_contents( $file );
865 $fp = fopen( $file, 'rb' );
866 $chunk = fread( $fp, 1024 );
870 $chunk = strtolower( $chunk );
876 # decode from UTF-16 if needed (could be used for obfuscation).
877 if( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
879 } elseif( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
886 $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
889 $chunk = trim( $chunk );
891 # @todo FIXME: Convert from UTF-16 if necessarry!
892 wfDebug( __METHOD__
. ": checking for embedded scripts and HTML stuff\n" );
894 # check for HTML doctype
895 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
900 * Internet Explorer for Windows performs some really stupid file type
901 * autodetection which can cause it to interpret valid image files as HTML
902 * and potentially execute JavaScript, creating a cross-site scripting
905 * Apple's Safari browser also performs some unsafe file type autodetection
906 * which can cause legitimate files to be interpreted as HTML if the
907 * web server is not correctly configured to send the right content-type
908 * (or if you're really uploading plain text and octet streams!)
910 * Returns true if IE is likely to mistake the given file for HTML.
911 * Also returns true if Safari would mistake the given file for HTML
912 * when served with a generic content-type.
918 '<html', #also in safari
921 '<script', #also in safari
925 if( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
929 foreach( $tags as $tag ) {
930 if( false !== strpos( $chunk, $tag ) ) {
931 wfDebug( __METHOD__
. ": found something that may make it be mistaken for html: $tag\n" );
937 * look for JavaScript
940 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
941 $chunk = Sanitizer
::decodeCharReferences( $chunk );
943 # look for script-types
944 if( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
945 wfDebug( __METHOD__
. ": found script types\n" );
949 # look for html-style script-urls
950 if( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
951 wfDebug( __METHOD__
. ": found html-style script urls\n" );
955 # look for css-style script-urls
956 if( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
957 wfDebug( __METHOD__
. ": found css-style script urls\n" );
961 wfDebug( __METHOD__
. ": no scripts found\n" );
965 protected function detectScriptInSvg( $filename ) {
966 $check = new XmlTypeCheck( $filename, array( $this, 'checkSvgScriptCallback' ) );
967 return $check->filterMatch
;
971 * @todo Replace this with a whitelist filter!
974 public function checkSvgScriptCallback( $element, $attribs ) {
975 $strippedElement = $this->stripXmlNamespace( $element );
978 * check for elements that can contain javascript
980 if( $strippedElement == 'script' ) {
981 wfDebug( __METHOD__
. ": Found script element '$element' in uploaded file.\n" );
985 # e.g., <svg xmlns="http://www.w3.org/2000/svg"> <handler xmlns:ev="http://www.w3.org/2001/xml-events" ev:event="load">alert(1)</handler> </svg>
986 if( $strippedElement == 'handler' ) {
987 wfDebug( __METHOD__
. ": Found scriptable element '$element' in uploaded file.\n" );
991 # SVG reported in Feb '12 that used xml:stylesheet to generate javascript block
992 if( $strippedElement == 'stylesheet' ) {
993 wfDebug( __METHOD__
. ": Found scriptable element '$element' in uploaded file.\n" );
997 foreach( $attribs as $attrib => $value ) {
998 $stripped = $this->stripXmlNamespace( $attrib );
999 $value = strtolower($value);
1001 if( substr( $stripped, 0, 2 ) == 'on' ) {
1002 wfDebug( __METHOD__
. ": Found event-handler attribute '$attrib'='$value' in uploaded file.\n" );
1006 # href with javascript target
1007 if( $stripped == 'href' && strpos( strtolower( $value ), 'javascript:' ) !== false ) {
1008 wfDebug( __METHOD__
. ": Found script in href attribute '$attrib'='$value' in uploaded file.\n" );
1012 # href with embeded svg as target
1013 if( $stripped == 'href' && preg_match( '!data:[^,]*image/svg[^,]*,!sim', $value ) ) {
1014 wfDebug( __METHOD__
. ": Found href to embedded svg \"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1018 # href with embeded (text/xml) svg as target
1019 if( $stripped == 'href' && preg_match( '!data:[^,]*text/xml[^,]*,!sim', $value ) ) {
1020 wfDebug( __METHOD__
. ": Found href to embedded svg \"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1024 # use set/animate to add event-handler attribute to parent
1025 if( ( $strippedElement == 'set' ||
$strippedElement == 'animate' ) && $stripped == 'attributename' && substr( $value, 0, 2 ) == 'on' ) {
1026 wfDebug( __METHOD__
. ": Found svg setting event-handler attribute with \"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1030 # use set to add href attribute to parent element
1031 if( $strippedElement == 'set' && $stripped == 'attributename' && strpos( $value, 'href' ) !== false ) {
1032 wfDebug( __METHOD__
. ": Found svg setting href attibute '$value' in uploaded file.\n" );
1036 # use set to add a remote / data / script target to an element
1037 if( $strippedElement == 'set' && $stripped == 'to' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1038 wfDebug( __METHOD__
. ": Found svg setting attibute to '$value' in uploaded file.\n" );
1043 # use handler attribute with remote / data / script
1044 if( $stripped == 'handler' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1045 wfDebug( __METHOD__
. ": Found svg setting handler with remote/data/script '$attrib'='$value' in uploaded file.\n" );
1049 # use CSS styles to bring in remote code
1050 # catch url("http:..., url('http:..., url(http:..., but not url("#..., url('#..., url(#....
1051 if( $stripped == 'style' && preg_match_all( '!((?:font|clip-path|fill|filter|marker|marker-end|marker-mid|marker-start|mask|stroke)\s*:\s*url\s*\(\s*["\']?\s*[^#]+.*?\))!sim', $value, $matches ) ) {
1052 foreach ($matches[1] as $match) {
1053 if (!preg_match( '!(?:font|clip-path|fill|filter|marker|marker-end|marker-mid|marker-start|mask|stroke)\s*:\s*url\s*\(\s*(#|\'#|"#)!sim', $match ) ) {
1054 wfDebug( __METHOD__
. ": Found svg setting a style with remote url '$attrib'='$value' in uploaded file.\n" );
1060 # image filters can pull in url, which could be svg that executes scripts
1061 if( $strippedElement == 'image' && $stripped == 'filter' && preg_match( '!url\s*\(!sim', $value ) ) {
1062 wfDebug( __METHOD__
. ": Found image filter with url: \"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1068 return false; //No scripts detected
1071 private function stripXmlNamespace( $name ) {
1072 // 'http://www.w3.org/2000/svg:script' -> 'script'
1073 $parts = explode( ':', strtolower( $name ) );
1074 return array_pop( $parts );
1078 * Generic wrapper function for a virus scanner program.
1079 * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
1080 * $wgAntivirusRequired may be used to deny upload if the scan fails.
1082 * @param $file String: pathname to the temporary upload file
1083 * @return mixed false if not virus is found, NULL if the scan fails or is disabled,
1084 * or a string containing feedback from the virus scanner if a virus was found.
1085 * If textual feedback is missing but a virus was found, this function returns true.
1087 public static function detectVirus( $file ) {
1088 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
1090 if ( !$wgAntivirus ) {
1091 wfDebug( __METHOD__
. ": virus scanner disabled\n" );
1095 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
1096 wfDebug( __METHOD__
. ": unknown virus scanner: $wgAntivirus\n" );
1097 $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
1098 array( 'virus-badscanner', $wgAntivirus ) );
1099 return wfMsg( 'virus-unknownscanner' ) . " $wgAntivirus";
1102 # look up scanner configuration
1103 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
1104 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
1105 $msgPattern = isset( $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ) ?
1106 $wgAntivirusSetup[$wgAntivirus]['messagepattern'] : null;
1108 if ( strpos( $command, "%f" ) === false ) {
1109 # simple pattern: append file to scan
1110 $command .= " " . wfEscapeShellArg( $file );
1112 # complex pattern: replace "%f" with file to scan
1113 $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
1116 wfDebug( __METHOD__
. ": running virus scan: $command \n" );
1118 # execute virus scanner
1121 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1122 # that does not seem to be worth the pain.
1123 # Ask me (Duesentrieb) about it if it's ever needed.
1124 $output = wfShellExec( "$command 2>&1", $exitCode );
1126 # map exit code to AV_xxx constants.
1127 $mappedCode = $exitCode;
1128 if ( $exitCodeMap ) {
1129 if ( isset( $exitCodeMap[$exitCode] ) ) {
1130 $mappedCode = $exitCodeMap[$exitCode];
1131 } elseif ( isset( $exitCodeMap["*"] ) ) {
1132 $mappedCode = $exitCodeMap["*"];
1136 if ( $mappedCode === AV_SCAN_FAILED
) {
1137 # scan failed (code was mapped to false by $exitCodeMap)
1138 wfDebug( __METHOD__
. ": failed to scan $file (code $exitCode).\n" );
1140 if ( $wgAntivirusRequired ) {
1141 return wfMsg( 'virus-scanfailed', array( $exitCode ) );
1145 } elseif ( $mappedCode === AV_SCAN_ABORTED
) {
1146 # scan failed because filetype is unknown (probably imune)
1147 wfDebug( __METHOD__
. ": unsupported file type $file (code $exitCode).\n" );
1149 } elseif ( $mappedCode === AV_NO_VIRUS
) {
1151 wfDebug( __METHOD__
. ": file passed virus scan.\n" );
1154 $output = trim( $output );
1157 $output = true; #if there's no output, return true
1158 } elseif ( $msgPattern ) {
1160 if ( preg_match( $msgPattern, $output, $groups ) ) {
1162 $output = $groups[1];
1167 wfDebug( __METHOD__
. ": FOUND VIRUS! scanner feedback: $output \n" );
1173 * Check if there's an overwrite conflict and, if so, if restrictions
1174 * forbid this user from performing the upload.
1178 * @return mixed true on success, array on failure
1180 private function checkOverwrite( $user ) {
1181 // First check whether the local file can be overwritten
1182 $file = $this->getLocalFile();
1183 if( $file->exists() ) {
1184 if( !self
::userCanReUpload( $user, $file ) ) {
1185 return array( 'fileexists-forbidden', $file->getName() );
1191 /* Check shared conflicts: if the local file does not exist, but
1192 * wfFindFile finds a file, it exists in a shared repository.
1194 $file = wfFindFile( $this->getTitle() );
1195 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1196 return array( 'fileexists-shared-forbidden', $file->getName() );
1203 * Check if a user is the last uploader
1205 * @param $user User object
1206 * @param $img String: image name
1209 public static function userCanReUpload( User
$user, $img ) {
1210 if( $user->isAllowed( 'reupload' ) ) {
1211 return true; // non-conditional
1213 if( !$user->isAllowed( 'reupload-own' ) ) {
1216 if( is_string( $img ) ) {
1217 $img = wfLocalFile( $img );
1219 if ( !( $img instanceof LocalFile
) ) {
1223 return $user->getId() == $img->getUser( 'id' );
1227 * Helper function that does various existence checks for a file.
1228 * The following checks are performed:
1230 * - Article with the same name as the file exists
1231 * - File exists with normalized extension
1232 * - The file looks like a thumbnail and the original exists
1234 * @param $file File The File object to check
1235 * @return mixed False if the file does not exists, else an array
1237 public static function getExistsWarning( $file ) {
1238 if( $file->exists() ) {
1239 return array( 'warning' => 'exists', 'file' => $file );
1242 if( $file->getTitle()->getArticleID() ) {
1243 return array( 'warning' => 'page-exists', 'file' => $file );
1246 if ( $file->wasDeleted() && !$file->exists() ) {
1247 return array( 'warning' => 'was-deleted', 'file' => $file );
1250 if( strpos( $file->getName(), '.' ) == false ) {
1251 $partname = $file->getName();
1254 $n = strrpos( $file->getName(), '.' );
1255 $extension = substr( $file->getName(), $n +
1 );
1256 $partname = substr( $file->getName(), 0, $n );
1258 $normalizedExtension = File
::normalizeExtension( $extension );
1260 if ( $normalizedExtension != $extension ) {
1261 // We're not using the normalized form of the extension.
1262 // Normal form is lowercase, using most common of alternate
1263 // extensions (eg 'jpg' rather than 'JPEG').
1265 // Check for another file using the normalized form...
1266 $nt_lc = Title
::makeTitle( NS_FILE
, "{$partname}.{$normalizedExtension}" );
1267 $file_lc = wfLocalFile( $nt_lc );
1269 if( $file_lc->exists() ) {
1271 'warning' => 'exists-normalized',
1273 'normalizedFile' => $file_lc
1278 if ( self
::isThumbName( $file->getName() ) ) {
1279 # Check for filenames like 50px- or 180px-, these are mostly thumbnails
1280 $nt_thb = Title
::newFromText( substr( $partname , strpos( $partname , '-' ) +
1 ) . '.' . $extension, NS_FILE
);
1281 $file_thb = wfLocalFile( $nt_thb );
1282 if( $file_thb->exists() ) {
1284 'warning' => 'thumb',
1286 'thumbFile' => $file_thb
1289 // File does not exist, but we just don't like the name
1291 'warning' => 'thumb-name',
1293 'thumbFile' => $file_thb
1299 foreach( self
::getFilenamePrefixBlacklist() as $prefix ) {
1300 if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
1302 'warning' => 'bad-prefix',
1313 * Helper function that checks whether the filename looks like a thumbnail
1316 public static function isThumbName( $filename ) {
1317 $n = strrpos( $filename, '.' );
1318 $partname = $n ?
substr( $filename, 0, $n ) : $filename;
1320 substr( $partname , 3, 3 ) == 'px-' ||
1321 substr( $partname , 2, 3 ) == 'px-'
1323 preg_match( "/[0-9]{2}/" , substr( $partname , 0, 2 ) );
1327 * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
1329 * @return array list of prefixes
1331 public static function getFilenamePrefixBlacklist() {
1332 $blacklist = array();
1333 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
1334 if( !$message->isDisabled() ) {
1335 $lines = explode( "\n", $message->plain() );
1336 foreach( $lines as $line ) {
1337 // Remove comment lines
1338 $comment = substr( trim( $line ), 0, 1 );
1339 if ( $comment == '#' ||
$comment == '' ) {
1342 // Remove additional comments after a prefix
1343 $comment = strpos( $line, '#' );
1344 if ( $comment > 0 ) {
1345 $line = substr( $line, 0, $comment-1 );
1347 $blacklist[] = trim( $line );
1354 * Gets image info about the file just uploaded.
1356 * Also has the effect of setting metadata to be an 'indexed tag name' in returned API result if
1357 * 'metadata' was requested. Oddly, we have to pass the "result" object down just so it can do that
1358 * with the appropriate format, presumably.
1360 * @param $result ApiResult:
1361 * @return Array: image info
1363 public function getImageInfo( $result ) {
1364 $file = $this->getLocalFile();
1365 // TODO This cries out for refactoring. We really want to say $file->getAllInfo(); here.
1366 // Perhaps "info" methods should be moved into files, and the API should just wrap them in queries.
1367 if ( $file instanceof UploadStashFile
) {
1368 $imParam = ApiQueryStashImageInfo
::getPropertyNames();
1369 $info = ApiQueryStashImageInfo
::getInfo( $file, array_flip( $imParam ), $result );
1371 $imParam = ApiQueryImageInfo
::getPropertyNames();
1372 $info = ApiQueryImageInfo
::getInfo( $file, array_flip( $imParam ), $result );
1378 public function convertVerifyErrorToStatus( $error ) {
1379 $code = $error['status'];
1380 unset( $code['status'] );
1381 return Status
::newFatal( $this->getVerificationErrorCode( $code ), $error );
1384 public static function getMaxUploadSize( $forType = null ) {
1385 global $wgMaxUploadSize;
1387 if ( is_array( $wgMaxUploadSize ) ) {
1388 if ( !is_null( $forType ) && isset( $wgMaxUploadSize[$forType] ) ) {
1389 return $wgMaxUploadSize[$forType];
1391 return $wgMaxUploadSize['*'];
1394 return intval( $wgMaxUploadSize );