3 * Base class for the backend of file upload.
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
25 * @defgroup Upload Upload related
31 * UploadBase and subclasses are the backend of MediaWiki's file uploads.
32 * The frontends are formed by ApiUpload and SpecialUpload.
34 * @author Brion Vibber
35 * @author Bryan Tong Minh
36 * @author Michael Dale
38 abstract class UploadBase
{
39 /** @var string Local file system path to the file to upload (or a local copy) */
41 /** @var TempFSFile|null Wrapper to handle deleting the temp file */
42 protected $tempFileObj;
44 protected $mDesiredDestName, $mDestName, $mRemoveTempFile, $mSourceType;
45 protected $mTitle = false, $mTitleError = 0;
46 protected $mFilteredName, $mFinalExtension;
47 protected $mLocalFile, $mStashFile, $mFileSize, $mFileProps;
48 protected $mBlackListedExtensions;
49 protected $mJavaDetected, $mSVGNSError;
51 protected static $safeXmlEncodings = [
71 const MIN_LENGTH_PARTNAME
= 4;
72 const ILLEGAL_FILENAME
= 5;
73 const OVERWRITE_EXISTING_FILE
= 7; # Not used anymore; handled by verifyTitlePermissions()
74 const FILETYPE_MISSING
= 8;
75 const FILETYPE_BADTYPE
= 9;
76 const VERIFICATION_ERROR
= 10;
77 const HOOK_ABORTED
= 11;
78 const FILE_TOO_LARGE
= 12;
79 const WINDOWS_NONASCII_FILENAME
= 13;
80 const FILENAME_TOO_LONG
= 14;
86 public function getVerificationErrorCode( $error ) {
88 self
::EMPTY_FILE
=> 'empty-file',
89 self
::FILE_TOO_LARGE
=> 'file-too-large',
90 self
::FILETYPE_MISSING
=> 'filetype-missing',
91 self
::FILETYPE_BADTYPE
=> 'filetype-banned',
92 self
::MIN_LENGTH_PARTNAME
=> 'filename-tooshort',
93 self
::ILLEGAL_FILENAME
=> 'illegal-filename',
94 self
::OVERWRITE_EXISTING_FILE
=> 'overwrite',
95 self
::VERIFICATION_ERROR
=> 'verification-error',
96 self
::HOOK_ABORTED
=> 'hookaborted',
97 self
::WINDOWS_NONASCII_FILENAME
=> 'windows-nonascii-filename',
98 self
::FILENAME_TOO_LONG
=> 'filename-toolong',
100 if ( isset( $code_to_status[$error] ) ) {
101 return $code_to_status[$error];
104 return 'unknown-error';
108 * Returns true if uploads are enabled.
109 * Can be override by subclasses.
112 public static function isEnabled() {
113 global $wgEnableUploads;
115 if ( !$wgEnableUploads ) {
119 # Check php's file_uploads setting
120 return wfIsHHVM() ||
wfIniGetBool( 'file_uploads' );
124 * Returns true if the user can use this upload module or else a string
125 * identifying the missing permission.
126 * Can be overridden by subclasses.
129 * @return bool|string
131 public static function isAllowed( $user ) {
132 foreach ( [ 'upload', 'edit' ] as $permission ) {
133 if ( !$user->isAllowed( $permission ) ) {
142 * Returns true if the user has surpassed the upload rate limit, false otherwise.
147 public static function isThrottled( $user ) {
148 return $user->pingLimiter( 'upload' );
151 // Upload handlers. Should probably just be a global.
152 private static $uploadHandlers = [ 'Stash', 'File', 'Url' ];
155 * Create a form of UploadBase depending on wpSourceType and initializes it
157 * @param WebRequest $request
158 * @param string|null $type
159 * @return null|UploadBase
161 public static function createFromRequest( &$request, $type = null ) {
162 $type = $type ?
$type : $request->getVal( 'wpSourceType', 'File' );
168 // Get the upload class
169 $type = ucfirst( $type );
171 // Give hooks the chance to handle this request
173 Hooks
::run( 'UploadCreateFromRequest', [ $type, &$className ] );
174 if ( is_null( $className ) ) {
175 $className = 'UploadFrom' . $type;
176 wfDebug( __METHOD__
. ": class name: $className\n" );
177 if ( !in_array( $type, self
::$uploadHandlers ) ) {
182 // Check whether this upload class is enabled
183 if ( !call_user_func( [ $className, 'isEnabled' ] ) ) {
187 // Check whether the request is valid
188 if ( !call_user_func( [ $className, 'isValidRequest' ], $request ) ) {
192 /** @var UploadBase $handler */
193 $handler = new $className;
195 $handler->initializeFromRequest( $request );
201 * Check whether a request if valid for this handler
202 * @param WebRequest $request
205 public static function isValidRequest( $request ) {
209 public function __construct() {
213 * Returns the upload type. Should be overridden by child classes
218 public function getSourceType() {
223 * Initialize the path information
224 * @param string $name The desired destination name
225 * @param string $tempPath The temporary path
226 * @param int $fileSize The file size
227 * @param bool $removeTempFile (false) remove the temporary file?
228 * @throws MWException
230 public function initializePathInfo( $name, $tempPath, $fileSize, $removeTempFile = false ) {
231 $this->mDesiredDestName
= $name;
232 if ( FileBackend
::isStoragePath( $tempPath ) ) {
233 throw new MWException( __METHOD__
. " given storage path `$tempPath`." );
236 $this->setTempFile( $tempPath, $fileSize );
237 $this->mRemoveTempFile
= $removeTempFile;
241 * Initialize from a WebRequest. Override this in a subclass.
243 * @param WebRequest $request
245 abstract public function initializeFromRequest( &$request );
248 * @param string $tempPath File system path to temporary file containing the upload
249 * @param integer $fileSize
251 protected function setTempFile( $tempPath, $fileSize = null ) {
252 $this->mTempPath
= $tempPath;
253 $this->mFileSize
= $fileSize ?
: null;
254 if ( strlen( $this->mTempPath
) && file_exists( $this->mTempPath
) ) {
255 $this->tempFileObj
= new TempFSFile( $this->mTempPath
);
257 $this->mFileSize
= filesize( $this->mTempPath
);
260 $this->tempFileObj
= null;
265 * Fetch the file. Usually a no-op
268 public function fetchFile() {
269 return Status
::newGood();
273 * Return true if the file is empty
276 public function isEmptyFile() {
277 return empty( $this->mFileSize
);
281 * Return the file size
284 public function getFileSize() {
285 return $this->mFileSize
;
289 * Get the base 36 SHA1 of the file
292 public function getTempFileSha1Base36() {
293 return FSFile
::getSha1Base36FromPath( $this->mTempPath
);
297 * @param string $srcPath The source path
298 * @return string|bool The real path if it was a virtual URL Returns false on failure
300 function getRealPath( $srcPath ) {
301 $repo = RepoGroup
::singleton()->getLocalRepo();
302 if ( $repo->isVirtualUrl( $srcPath ) ) {
303 /** @todo Just make uploads work with storage paths UploadFromStash
304 * loads files via virtual URLs.
306 $tmpFile = $repo->getLocalCopy( $srcPath );
308 $tmpFile->bind( $this ); // keep alive with $this
310 $path = $tmpFile ?
$tmpFile->getPath() : false;
319 * Verify whether the upload is sane.
320 * @return mixed Const self::OK or else an array with error information
322 public function verifyUpload() {
325 * If there was no filename or a zero size given, give up quick.
327 if ( $this->isEmptyFile() ) {
328 return [ 'status' => self
::EMPTY_FILE
];
332 * Honor $wgMaxUploadSize
334 $maxSize = self
::getMaxUploadSize( $this->getSourceType() );
335 if ( $this->mFileSize
> $maxSize ) {
337 'status' => self
::FILE_TOO_LARGE
,
343 * Look at the contents of the file; if we can recognize the
344 * type but it's corrupt or data of the wrong type, we should
345 * probably not accept it.
347 $verification = $this->verifyFile();
348 if ( $verification !== true ) {
350 'status' => self
::VERIFICATION_ERROR
,
351 'details' => $verification
356 * Make sure this file can be created
358 $result = $this->validateName();
359 if ( $result !== true ) {
364 if ( !Hooks
::run( 'UploadVerification',
365 [ $this->mDestName
, $this->mTempPath
, &$error ], '1.28' )
367 return [ 'status' => self
::HOOK_ABORTED
, 'error' => $error ];
370 return [ 'status' => self
::OK
];
374 * Verify that the name is valid and, if necessary, that we can overwrite
376 * @return mixed True if valid, otherwise and array with 'status'
379 public function validateName() {
380 $nt = $this->getTitle();
381 if ( is_null( $nt ) ) {
382 $result = [ 'status' => $this->mTitleError
];
383 if ( $this->mTitleError
== self
::ILLEGAL_FILENAME
) {
384 $result['filtered'] = $this->mFilteredName
;
386 if ( $this->mTitleError
== self
::FILETYPE_BADTYPE
) {
387 $result['finalExt'] = $this->mFinalExtension
;
388 if ( count( $this->mBlackListedExtensions
) ) {
389 $result['blacklistedExt'] = $this->mBlackListedExtensions
;
395 $this->mDestName
= $this->getLocalFile()->getName();
401 * Verify the MIME type.
403 * @note Only checks that it is not an evil MIME. The "does it have
404 * correct extension given its MIME type?" check is in verifyFile.
405 * in `verifyFile()` that MIME type and file extension correlate.
406 * @param string $mime Representing the MIME
407 * @return mixed True if the file is verified, an array otherwise
409 protected function verifyMimeType( $mime ) {
410 global $wgVerifyMimeType;
411 if ( $wgVerifyMimeType ) {
412 wfDebug( "mime: <$mime> extension: <{$this->mFinalExtension}>\n" );
413 global $wgMimeTypeBlacklist;
414 if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
415 return [ 'filetype-badmime', $mime ];
418 # Check what Internet Explorer would detect
419 $fp = fopen( $this->mTempPath
, 'rb' );
420 $chunk = fread( $fp, 256 );
423 $magic = MimeMagic
::singleton();
424 $extMime = $magic->guessTypesForExtension( $this->mFinalExtension
);
425 $ieTypes = $magic->getIEMimeTypes( $this->mTempPath
, $chunk, $extMime );
426 foreach ( $ieTypes as $ieType ) {
427 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
428 return [ 'filetype-bad-ie-mime', $ieType ];
437 * Verifies that it's ok to include the uploaded file
439 * @return mixed True of the file is verified, array otherwise.
441 protected function verifyFile() {
442 global $wgVerifyMimeType, $wgDisableUploadScriptChecks;
444 $status = $this->verifyPartialFile();
445 if ( $status !== true ) {
449 $mwProps = new MWFileProps( MimeMagic
::singleton() );
450 $this->mFileProps
= $mwProps->getPropsFromPath( $this->mTempPath
, $this->mFinalExtension
);
451 $mime = $this->mFileProps
['mime'];
453 if ( $wgVerifyMimeType ) {
454 # XXX: Missing extension will be caught by validateName() via getTitle()
455 if ( $this->mFinalExtension
!= '' && !$this->verifyExtension( $mime, $this->mFinalExtension
) ) {
456 return [ 'filetype-mime-mismatch', $this->mFinalExtension
, $mime ];
460 # check for htmlish code and javascript
461 if ( !$wgDisableUploadScriptChecks ) {
462 if ( $this->mFinalExtension
== 'svg' ||
$mime == 'image/svg+xml' ) {
463 $svgStatus = $this->detectScriptInSvg( $this->mTempPath
, false );
464 if ( $svgStatus !== false ) {
470 $handler = MediaHandler
::getHandler( $mime );
472 $handlerStatus = $handler->verifyUpload( $this->mTempPath
);
473 if ( !$handlerStatus->isOK() ) {
474 $errors = $handlerStatus->getErrorsArray();
476 return reset( $errors );
481 Hooks
::run( 'UploadVerifyFile', [ $this, $mime, &$error ] );
482 if ( $error !== true ) {
483 if ( !is_array( $error ) ) {
489 wfDebug( __METHOD__
. ": all clear; passing.\n" );
495 * A verification routine suitable for partial files
497 * Runs the blacklist checks, but not any checks that may
498 * assume the entire file is present.
500 * @return mixed True for valid or array with error message key.
502 protected function verifyPartialFile() {
503 global $wgAllowJavaUploads, $wgDisableUploadScriptChecks;
505 # getTitle() sets some internal parameters like $this->mFinalExtension
508 $mwProps = new MWFileProps( MimeMagic
::singleton() );
509 $this->mFileProps
= $mwProps->getPropsFromPath( $this->mTempPath
, $this->mFinalExtension
);
511 # check MIME type, if desired
512 $mime = $this->mFileProps
['file-mime'];
513 $status = $this->verifyMimeType( $mime );
514 if ( $status !== true ) {
518 # check for htmlish code and javascript
519 if ( !$wgDisableUploadScriptChecks ) {
520 if ( self
::detectScript( $this->mTempPath
, $mime, $this->mFinalExtension
) ) {
521 return [ 'uploadscripted' ];
523 if ( $this->mFinalExtension
== 'svg' ||
$mime == 'image/svg+xml' ) {
524 $svgStatus = $this->detectScriptInSvg( $this->mTempPath
, true );
525 if ( $svgStatus !== false ) {
531 # Check for Java applets, which if uploaded can bypass cross-site
533 if ( !$wgAllowJavaUploads ) {
534 $this->mJavaDetected
= false;
535 $zipStatus = ZipDirectoryReader
::read( $this->mTempPath
,
536 [ $this, 'zipEntryCallback' ] );
537 if ( !$zipStatus->isOK() ) {
538 $errors = $zipStatus->getErrorsArray();
539 $error = reset( $errors );
540 if ( $error[0] !== 'zip-wrong-format' ) {
544 if ( $this->mJavaDetected
) {
545 return [ 'uploadjava' ];
549 # Scan the uploaded file for viruses
550 $virus = $this->detectVirus( $this->mTempPath
);
552 return [ 'uploadvirus', $virus ];
559 * Callback for ZipDirectoryReader to detect Java class files.
561 * @param array $entry
563 function zipEntryCallback( $entry ) {
564 $names = [ $entry['name'] ];
566 // If there is a null character, cut off the name at it, because JDK's
567 // ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name
568 // were constructed which had ".class\0" followed by a string chosen to
569 // make the hash collide with the truncated name, that file could be
570 // returned in response to a request for the .class file.
571 $nullPos = strpos( $entry['name'], "\000" );
572 if ( $nullPos !== false ) {
573 $names[] = substr( $entry['name'], 0, $nullPos );
576 // If there is a trailing slash in the file name, we have to strip it,
577 // because that's what ZIP_GetEntry() does.
578 if ( preg_grep( '!\.class/?$!', $names ) ) {
579 $this->mJavaDetected
= true;
584 * Alias for verifyTitlePermissions. The function was originally
585 * 'verifyPermissions', but that suggests it's checking the user, when it's
586 * really checking the title + user combination.
588 * @param User $user User object to verify the permissions against
589 * @return mixed An array as returned by getUserPermissionsErrors or true
590 * in case the user has proper permissions.
592 public function verifyPermissions( $user ) {
593 return $this->verifyTitlePermissions( $user );
597 * Check whether the user can edit, upload and create the image. This
598 * checks only against the current title; if it returns errors, it may
599 * very well be that another title will not give errors. Therefore
600 * isAllowed() should be called as well for generic is-user-blocked or
601 * can-user-upload checking.
603 * @param User $user User object to verify the permissions against
604 * @return mixed An array as returned by getUserPermissionsErrors or true
605 * in case the user has proper permissions.
607 public function verifyTitlePermissions( $user ) {
609 * If the image is protected, non-sysop users won't be able
610 * to modify it by uploading a new revision.
612 $nt = $this->getTitle();
613 if ( is_null( $nt ) ) {
616 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
617 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
618 if ( !$nt->exists() ) {
619 $permErrorsCreate = $nt->getUserPermissionsErrors( 'create', $user );
621 $permErrorsCreate = [];
623 if ( $permErrors ||
$permErrorsUpload ||
$permErrorsCreate ) {
624 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
625 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
630 $overwriteError = $this->checkOverwrite( $user );
631 if ( $overwriteError !== true ) {
632 return [ $overwriteError ];
639 * Check for non fatal problems with the file.
641 * This should not assume that mTempPath is set.
643 * @return array Array of warnings
645 public function checkWarnings() {
650 $localFile = $this->getLocalFile();
651 $localFile->load( File
::READ_LATEST
);
652 $filename = $localFile->getName();
655 * Check whether the resulting filename is different from the desired one,
656 * but ignore things like ucfirst() and spaces/underscore things
658 $comparableName = str_replace( ' ', '_', $this->mDesiredDestName
);
659 $comparableName = Title
::capitalize( $comparableName, NS_FILE
);
661 if ( $this->mDesiredDestName
!= $filename && $comparableName != $filename ) {
662 $warnings['badfilename'] = $filename;
665 // Check whether the file extension is on the unwanted list
666 global $wgCheckFileExtensions, $wgFileExtensions;
667 if ( $wgCheckFileExtensions ) {
668 $extensions = array_unique( $wgFileExtensions );
669 if ( !$this->checkFileExtension( $this->mFinalExtension
, $extensions ) ) {
670 $warnings['filetype-unwanted-type'] = [ $this->mFinalExtension
,
671 $wgLang->commaList( $extensions ), count( $extensions ) ];
675 global $wgUploadSizeWarning;
676 if ( $wgUploadSizeWarning && ( $this->mFileSize
> $wgUploadSizeWarning ) ) {
677 $warnings['large-file'] = [ $wgUploadSizeWarning, $this->mFileSize
];
680 if ( $this->mFileSize
== 0 ) {
681 $warnings['empty-file'] = true;
684 $hash = $this->getTempFileSha1Base36();
685 $exists = self
::getExistsWarning( $localFile );
686 if ( $exists !== false ) {
687 $warnings['exists'] = $exists;
689 // check if file is an exact duplicate of current file version
690 if ( $hash === $localFile->getSha1() ) {
691 $warnings['no-change'] = $localFile;
694 // check if file is an exact duplicate of older versions of this file
695 $history = $localFile->getHistory();
696 foreach ( $history as $oldFile ) {
697 if ( $hash === $oldFile->getSha1() ) {
698 $warnings['duplicate-version'][] = $oldFile;
703 if ( $localFile->wasDeleted() && !$localFile->exists() ) {
704 $warnings['was-deleted'] = $filename;
707 // Check dupes against existing files
708 $dupes = RepoGroup
::singleton()->findBySha1( $hash );
709 $title = $this->getTitle();
710 // Remove all matches against self
711 foreach ( $dupes as $key => $dupe ) {
712 if ( $title->equals( $dupe->getTitle() ) ) {
713 unset( $dupes[$key] );
717 $warnings['duplicate'] = $dupes;
720 // Check dupes against archives
721 $archivedFile = new ArchivedFile( null, 0, '', $hash );
722 if ( $archivedFile->getID() > 0 ) {
723 if ( $archivedFile->userCan( File
::DELETED_FILE
) ) {
724 $warnings['duplicate-archive'] = $archivedFile->getName();
726 $warnings['duplicate-archive'] = '';
734 * Really perform the upload. Stores the file in the local repo, watches
735 * if necessary and runs the UploadComplete hook.
737 * @param string $comment
738 * @param string $pageText
739 * @param bool $watch Whether the file page should be added to user's watchlist.
740 * (This doesn't check $user's permissions.)
742 * @param string[] $tags Change tags to add to the log entry and page revision.
743 * (This doesn't check $user's permissions.)
744 * @return Status Indicating the whether the upload succeeded.
746 public function performUpload( $comment, $pageText, $watch, $user, $tags = [] ) {
747 $this->getLocalFile()->load( File
::READ_LATEST
);
748 $props = $this->mFileProps
;
751 Hooks
::run( 'UploadVerifyUpload', [ $this, $user, $props, $comment, $pageText, &$error ] );
753 if ( !is_array( $error ) ) {
756 return call_user_func_array( 'Status::newFatal', $error );
759 $status = $this->getLocalFile()->upload(
770 if ( $status->isGood() ) {
772 WatchAction
::doWatch(
773 $this->getLocalFile()->getTitle(),
775 User
::IGNORE_USER_RIGHTS
778 Hooks
::run( 'UploadComplete', [ &$this ] );
780 $this->postProcessUpload();
787 * Perform extra steps after a successful upload.
791 public function postProcessUpload() {
795 * Returns the title of the file to be uploaded. Sets mTitleError in case
796 * the name was illegal.
798 * @return Title The title of the file or null in case the name was illegal
800 public function getTitle() {
801 if ( $this->mTitle
!== false ) {
802 return $this->mTitle
;
804 if ( !is_string( $this->mDesiredDestName
) ) {
805 $this->mTitleError
= self
::ILLEGAL_FILENAME
;
806 $this->mTitle
= null;
808 return $this->mTitle
;
810 /* Assume that if a user specified File:Something.jpg, this is an error
811 * and that the namespace prefix needs to be stripped of.
813 $title = Title
::newFromText( $this->mDesiredDestName
);
814 if ( $title && $title->getNamespace() == NS_FILE
) {
815 $this->mFilteredName
= $title->getDBkey();
817 $this->mFilteredName
= $this->mDesiredDestName
;
820 # oi_archive_name is max 255 bytes, which include a timestamp and an
821 # exclamation mark, so restrict file name to 240 bytes.
822 if ( strlen( $this->mFilteredName
) > 240 ) {
823 $this->mTitleError
= self
::FILENAME_TOO_LONG
;
824 $this->mTitle
= null;
826 return $this->mTitle
;
830 * Chop off any directories in the given filename. Then
831 * filter out illegal characters, and try to make a legible name
832 * out of it. We'll strip some silently that Title would die on.
834 $this->mFilteredName
= wfStripIllegalFilenameChars( $this->mFilteredName
);
835 /* Normalize to title form before we do any further processing */
836 $nt = Title
::makeTitleSafe( NS_FILE
, $this->mFilteredName
);
837 if ( is_null( $nt ) ) {
838 $this->mTitleError
= self
::ILLEGAL_FILENAME
;
839 $this->mTitle
= null;
841 return $this->mTitle
;
843 $this->mFilteredName
= $nt->getDBkey();
846 * We'll want to blacklist against *any* 'extension', and use
847 * only the final one for the whitelist.
849 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName
);
851 if ( count( $ext ) ) {
852 $this->mFinalExtension
= trim( $ext[count( $ext ) - 1] );
854 $this->mFinalExtension
= '';
856 # No extension, try guessing one
857 $magic = MimeMagic
::singleton();
858 $mime = $magic->guessMimeType( $this->mTempPath
);
859 if ( $mime !== 'unknown/unknown' ) {
860 # Get a space separated list of extensions
861 $extList = $magic->getExtensionsForType( $mime );
863 # Set the extension to the canonical extension
864 $this->mFinalExtension
= strtok( $extList, ' ' );
866 # Fix up the other variables
867 $this->mFilteredName
.= ".{$this->mFinalExtension}";
868 $nt = Title
::makeTitleSafe( NS_FILE
, $this->mFilteredName
);
869 $ext = [ $this->mFinalExtension
];
874 /* Don't allow users to override the blacklist (check file extension) */
875 global $wgCheckFileExtensions, $wgStrictFileExtensions;
876 global $wgFileExtensions, $wgFileBlacklist;
878 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
880 if ( $this->mFinalExtension
== '' ) {
881 $this->mTitleError
= self
::FILETYPE_MISSING
;
882 $this->mTitle
= null;
884 return $this->mTitle
;
885 } elseif ( $blackListedExtensions ||
886 ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
887 !$this->checkFileExtension( $this->mFinalExtension
, $wgFileExtensions ) )
889 $this->mBlackListedExtensions
= $blackListedExtensions;
890 $this->mTitleError
= self
::FILETYPE_BADTYPE
;
891 $this->mTitle
= null;
893 return $this->mTitle
;
896 // Windows may be broken with special characters, see bug 1780
897 if ( !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() )
898 && !RepoGroup
::singleton()->getLocalRepo()->backendSupportsUnicodePaths()
900 $this->mTitleError
= self
::WINDOWS_NONASCII_FILENAME
;
901 $this->mTitle
= null;
903 return $this->mTitle
;
906 # If there was more than one "extension", reassemble the base
907 # filename to prevent bogus complaints about length
908 if ( count( $ext ) > 1 ) {
909 $iterations = count( $ext ) - 1;
910 for ( $i = 0; $i < $iterations; $i++
) {
911 $partname .= '.' . $ext[$i];
915 if ( strlen( $partname ) < 1 ) {
916 $this->mTitleError
= self
::MIN_LENGTH_PARTNAME
;
917 $this->mTitle
= null;
919 return $this->mTitle
;
924 return $this->mTitle
;
928 * Return the local file and initializes if necessary.
930 * @return LocalFile|null
932 public function getLocalFile() {
933 if ( is_null( $this->mLocalFile
) ) {
934 $nt = $this->getTitle();
935 $this->mLocalFile
= is_null( $nt ) ?
null : wfLocalFile( $nt );
938 return $this->mLocalFile
;
942 * @return UploadStashFile|null
944 public function getStashFile() {
945 return $this->mStashFile
;
949 * Like stashFile(), but respects extensions' wishes to prevent the stashing. verifyUpload() must
950 * be called before calling this method (unless $isPartial is true).
952 * Upload stash exceptions are also caught and converted to an error status.
956 * @param bool $isPartial Pass `true` if this is a part of a chunked upload (not a complete file).
957 * @return Status If successful, value is an UploadStashFile instance
959 public function tryStashFile( User
$user, $isPartial = false ) {
961 $error = $this->runUploadStashFileHook( $user );
963 return call_user_func_array( 'Status::newFatal', $error );
967 $file = $this->doStashFile( $user );
968 return Status
::newGood( $file );
969 } catch ( UploadStashException
$e ) {
970 return Status
::newFatal( 'uploadstash-exception', get_class( $e ), $e->getMessage() );
976 * @return array|null Error message and parameters, null if there's no error
978 protected function runUploadStashFileHook( User
$user ) {
979 $props = $this->mFileProps
;
981 Hooks
::run( 'UploadStashFile', [ $this, $user, $props, &$error ] );
983 if ( !is_array( $error ) ) {
991 * If the user does not supply all necessary information in the first upload
992 * form submission (either by accident or by design) then we may want to
993 * stash the file temporarily, get more information, and publish the file
996 * This method will stash a file in a temporary directory for later
997 * processing, and save the necessary descriptive info into the database.
998 * This method returns the file object, which also has a 'fileKey' property
999 * which can be passed through a form or API request to find this stashed
1002 * @deprecated since 1.28 Use tryStashFile() instead
1004 * @return UploadStashFile Stashed file
1005 * @throws UploadStashBadPathException
1006 * @throws UploadStashFileException
1007 * @throws UploadStashNotLoggedInException
1009 public function stashFile( User
$user = null ) {
1010 return $this->doStashFile( $user );
1014 * Implementation for stashFile() and tryStashFile().
1017 * @return UploadStashFile Stashed file
1019 protected function doStashFile( User
$user = null ) {
1020 $stash = RepoGroup
::singleton()->getLocalRepo()->getUploadStash( $user );
1021 $file = $stash->stashFile( $this->mTempPath
, $this->getSourceType() );
1022 $this->mStashFile
= $file;
1028 * Stash a file in a temporary directory, returning a key which can be used
1029 * to find the file again. See stashFile().
1031 * @deprecated since 1.28
1032 * @return string File key
1034 public function stashFileGetKey() {
1035 wfDeprecated( __METHOD__
, '1.28' );
1036 return $this->doStashFile()->getFileKey();
1040 * alias for stashFileGetKey, for backwards compatibility
1042 * @deprecated since 1.28
1043 * @return string File key
1045 public function stashSession() {
1046 wfDeprecated( __METHOD__
, '1.28' );
1047 return $this->doStashFile()->getFileKey();
1051 * If we've modified the upload file we need to manually remove it
1052 * on exit to clean up.
1054 public function cleanupTempFile() {
1055 if ( $this->mRemoveTempFile
&& $this->tempFileObj
) {
1056 // Delete when all relevant TempFSFile handles go out of scope
1057 wfDebug( __METHOD__
. ": Marked temporary file '{$this->mTempPath}' for removal\n" );
1058 $this->tempFileObj
->autocollect();
1062 public function getTempPath() {
1063 return $this->mTempPath
;
1067 * Split a file into a base name and all dot-delimited 'extensions'
1068 * on the end. Some web server configurations will fall back to
1069 * earlier pseudo-'extensions' to determine type and execute
1070 * scripts, so the blacklist needs to check them all.
1072 * @param string $filename
1075 public static function splitExtensions( $filename ) {
1076 $bits = explode( '.', $filename );
1077 $basename = array_shift( $bits );
1079 return [ $basename, $bits ];
1083 * Perform case-insensitive match against a list of file extensions.
1084 * Returns true if the extension is in the list.
1086 * @param string $ext
1087 * @param array $list
1090 public static function checkFileExtension( $ext, $list ) {
1091 return in_array( strtolower( $ext ), $list );
1095 * Perform case-insensitive match against a list of file extensions.
1096 * Returns an array of matching extensions.
1099 * @param array $list
1102 public static function checkFileExtensionList( $ext, $list ) {
1103 return array_intersect( array_map( 'strtolower', $ext ), $list );
1107 * Checks if the MIME type of the uploaded file matches the file extension.
1109 * @param string $mime The MIME type of the uploaded file
1110 * @param string $extension The filename extension that the file is to be served with
1113 public static function verifyExtension( $mime, $extension ) {
1114 $magic = MimeMagic
::singleton();
1116 if ( !$mime ||
$mime == 'unknown' ||
$mime == 'unknown/unknown' ) {
1117 if ( !$magic->isRecognizableExtension( $extension ) ) {
1118 wfDebug( __METHOD__
. ": passing file with unknown detected mime type; " .
1119 "unrecognized extension '$extension', can't verify\n" );
1123 wfDebug( __METHOD__
. ": rejecting file with unknown detected mime type; " .
1124 "recognized extension '$extension', so probably invalid file\n" );
1130 $match = $magic->isMatchingExtension( $extension, $mime );
1132 if ( $match === null ) {
1133 if ( $magic->getTypesForExtension( $extension ) !== null ) {
1134 wfDebug( __METHOD__
. ": No extension known for $mime, but we know a mime for $extension\n" );
1138 wfDebug( __METHOD__
. ": no file extension known for mime type $mime, passing file\n" );
1142 } elseif ( $match === true ) {
1143 wfDebug( __METHOD__
. ": mime type $mime matches extension $extension, passing file\n" );
1145 /** @todo If it's a bitmap, make sure PHP or ImageMagick resp. can handle it! */
1149 . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
1156 * Heuristic for detecting files that *could* contain JavaScript instructions or
1157 * things that may look like HTML to a browser and are thus
1158 * potentially harmful. The present implementation will produce false
1159 * positives in some situations.
1161 * @param string $file Pathname to the temporary upload file
1162 * @param string $mime The MIME type of the file
1163 * @param string $extension The extension of the file
1164 * @return bool True if the file contains something looking like embedded scripts
1166 public static function detectScript( $file, $mime, $extension ) {
1167 global $wgAllowTitlesInSVG;
1169 # ugly hack: for text files, always look at the entire file.
1170 # For binary field, just check the first K.
1172 if ( strpos( $mime, 'text/' ) === 0 ) {
1173 $chunk = file_get_contents( $file );
1175 $fp = fopen( $file, 'rb' );
1176 $chunk = fread( $fp, 1024 );
1180 $chunk = strtolower( $chunk );
1186 # decode from UTF-16 if needed (could be used for obfuscation).
1187 if ( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
1189 } elseif ( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
1196 $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
1199 $chunk = trim( $chunk );
1201 /** @todo FIXME: Convert from UTF-16 if necessary! */
1202 wfDebug( __METHOD__
. ": checking for embedded scripts and HTML stuff\n" );
1204 # check for HTML doctype
1205 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
1209 // Some browsers will interpret obscure xml encodings as UTF-8, while
1210 // PHP/expat will interpret the given encoding in the xml declaration (bug 47304)
1211 if ( $extension == 'svg' ||
strpos( $mime, 'image/svg' ) === 0 ) {
1212 if ( self
::checkXMLEncodingMissmatch( $file ) ) {
1218 * Internet Explorer for Windows performs some really stupid file type
1219 * autodetection which can cause it to interpret valid image files as HTML
1220 * and potentially execute JavaScript, creating a cross-site scripting
1223 * Apple's Safari browser also performs some unsafe file type autodetection
1224 * which can cause legitimate files to be interpreted as HTML if the
1225 * web server is not correctly configured to send the right content-type
1226 * (or if you're really uploading plain text and octet streams!)
1228 * Returns true if IE is likely to mistake the given file for HTML.
1229 * Also returns true if Safari would mistake the given file for HTML
1230 * when served with a generic content-type.
1236 '<html', # also in safari
1239 '<script', # also in safari
1243 if ( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
1247 foreach ( $tags as $tag ) {
1248 if ( false !== strpos( $chunk, $tag ) ) {
1249 wfDebug( __METHOD__
. ": found something that may make it be mistaken for html: $tag\n" );
1256 * look for JavaScript
1259 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
1260 $chunk = Sanitizer
::decodeCharReferences( $chunk );
1262 # look for script-types
1263 if ( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
1264 wfDebug( __METHOD__
. ": found script types\n" );
1269 # look for html-style script-urls
1270 if ( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1271 wfDebug( __METHOD__
. ": found html-style script urls\n" );
1276 # look for css-style script-urls
1277 if ( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1278 wfDebug( __METHOD__
. ": found css-style script urls\n" );
1283 wfDebug( __METHOD__
. ": no scripts found\n" );
1289 * Check a whitelist of xml encodings that are known not to be interpreted differently
1290 * by the server's xml parser (expat) and some common browsers.
1292 * @param string $file Pathname to the temporary upload file
1293 * @return bool True if the file contains an encoding that could be misinterpreted
1295 public static function checkXMLEncodingMissmatch( $file ) {
1296 global $wgSVGMetadataCutoff;
1297 $contents = file_get_contents( $file, false, null, -1, $wgSVGMetadataCutoff );
1298 $encodingRegex = '!encoding[ \t\n\r]*=[ \t\n\r]*[\'"](.*?)[\'"]!si';
1300 if ( preg_match( "!<\?xml\b(.*?)\?>!si", $contents, $matches ) ) {
1301 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1302 && !in_array( strtoupper( $encMatch[1] ), self
::$safeXmlEncodings )
1304 wfDebug( __METHOD__
. ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1308 } elseif ( preg_match( "!<\?xml\b!si", $contents ) ) {
1309 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1310 // bytes. There shouldn't be a legitimate reason for this to happen.
1311 wfDebug( __METHOD__
. ": Unmatched XML declaration start\n" );
1314 } elseif ( substr( $contents, 0, 4 ) == "\x4C\x6F\xA7\x94" ) {
1315 // EBCDIC encoded XML
1316 wfDebug( __METHOD__
. ": EBCDIC Encoded XML\n" );
1321 // It's possible the file is encoded with multi-byte encoding, so re-encode attempt to
1322 // detect the encoding in case is specifies an encoding not whitelisted in self::$safeXmlEncodings
1323 $attemptEncodings = [ 'UTF-16', 'UTF-16BE', 'UTF-32', 'UTF-32BE' ];
1324 foreach ( $attemptEncodings as $encoding ) {
1325 MediaWiki\
suppressWarnings();
1326 $str = iconv( $encoding, 'UTF-8', $contents );
1327 MediaWiki\restoreWarnings
();
1328 if ( $str != '' && preg_match( "!<\?xml\b(.*?)\?>!si", $str, $matches ) ) {
1329 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1330 && !in_array( strtoupper( $encMatch[1] ), self
::$safeXmlEncodings )
1332 wfDebug( __METHOD__
. ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1336 } elseif ( $str != '' && preg_match( "!<\?xml\b!si", $str ) ) {
1337 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1338 // bytes. There shouldn't be a legitimate reason for this to happen.
1339 wfDebug( __METHOD__
. ": Unmatched XML declaration start\n" );
1349 * @param string $filename
1350 * @param bool $partial
1351 * @return mixed False of the file is verified (does not contain scripts), array otherwise.
1353 protected function detectScriptInSvg( $filename, $partial ) {
1354 $this->mSVGNSError
= false;
1355 $check = new XmlTypeCheck(
1357 [ $this, 'checkSvgScriptCallback' ],
1359 [ 'processing_instruction_handler' => 'UploadBase::checkSvgPICallback' ]
1361 if ( $check->wellFormed
!== true ) {
1362 // Invalid xml (bug 58553)
1363 // But only when non-partial (bug 65724)
1364 return $partial ?
false : [ 'uploadinvalidxml' ];
1365 } elseif ( $check->filterMatch
) {
1366 if ( $this->mSVGNSError
) {
1367 return [ 'uploadscriptednamespace', $this->mSVGNSError
];
1370 return $check->filterMatchType
;
1377 * Callback to filter SVG Processing Instructions.
1378 * @param string $target Processing instruction name
1379 * @param string $data Processing instruction attribute and value
1380 * @return bool (true if the filter identified something bad)
1382 public static function checkSvgPICallback( $target, $data ) {
1383 // Don't allow external stylesheets (bug 57550)
1384 if ( preg_match( '/xml-stylesheet/i', $target ) ) {
1385 return [ 'upload-scripted-pi-callback' ];
1392 * @todo Replace this with a whitelist filter!
1393 * @param string $element
1394 * @param array $attribs
1397 public function checkSvgScriptCallback( $element, $attribs, $data = null ) {
1399 list( $namespace, $strippedElement ) = $this->splitXmlNamespace( $element );
1401 // We specifically don't include:
1402 // http://www.w3.org/1999/xhtml (bug 60771)
1403 static $validNamespaces = [
1406 'http://creativecommons.org/ns#',
1407 'http://inkscape.sourceforge.net/dtd/sodipodi-0.dtd',
1408 'http://ns.adobe.com/adobeillustrator/10.0/',
1409 'http://ns.adobe.com/adobesvgviewerextensions/3.0/',
1410 'http://ns.adobe.com/extensibility/1.0/',
1411 'http://ns.adobe.com/flows/1.0/',
1412 'http://ns.adobe.com/illustrator/1.0/',
1413 'http://ns.adobe.com/imagereplacement/1.0/',
1414 'http://ns.adobe.com/pdf/1.3/',
1415 'http://ns.adobe.com/photoshop/1.0/',
1416 'http://ns.adobe.com/saveforweb/1.0/',
1417 'http://ns.adobe.com/variables/1.0/',
1418 'http://ns.adobe.com/xap/1.0/',
1419 'http://ns.adobe.com/xap/1.0/g/',
1420 'http://ns.adobe.com/xap/1.0/g/img/',
1421 'http://ns.adobe.com/xap/1.0/mm/',
1422 'http://ns.adobe.com/xap/1.0/rights/',
1423 'http://ns.adobe.com/xap/1.0/stype/dimensions#',
1424 'http://ns.adobe.com/xap/1.0/stype/font#',
1425 'http://ns.adobe.com/xap/1.0/stype/manifestitem#',
1426 'http://ns.adobe.com/xap/1.0/stype/resourceevent#',
1427 'http://ns.adobe.com/xap/1.0/stype/resourceref#',
1428 'http://ns.adobe.com/xap/1.0/t/pg/',
1429 'http://purl.org/dc/elements/1.1/',
1430 'http://purl.org/dc/elements/1.1',
1431 'http://schemas.microsoft.com/visio/2003/svgextensions/',
1432 'http://sodipodi.sourceforge.net/dtd/sodipodi-0.dtd',
1433 'http://taptrix.com/inkpad/svg_extensions',
1434 'http://web.resource.org/cc/',
1435 'http://www.freesoftware.fsf.org/bkchem/cdml',
1436 'http://www.inkscape.org/namespaces/inkscape',
1437 'http://www.opengis.net/gml',
1438 'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
1439 'http://www.w3.org/2000/svg',
1440 'http://www.w3.org/tr/rec-rdf-syntax/',
1443 // Inkscape mangles namespace definitions created by Adobe Illustrator.
1444 // This is nasty but harmless. (T144827)
1445 $isBuggyInkscape = preg_match( '/^&(#38;)*ns_[a-z_]+;$/', $namespace );
1447 if ( !( $isBuggyInkscape ||
in_array( $namespace, $validNamespaces ) ) ) {
1448 wfDebug( __METHOD__
. ": Non-svg namespace '$namespace' in uploaded file.\n" );
1449 /** @todo Return a status object to a closure in XmlTypeCheck, for MW1.21+ */
1450 $this->mSVGNSError
= $namespace;
1456 * check for elements that can contain javascript
1458 if ( $strippedElement == 'script' ) {
1459 wfDebug( __METHOD__
. ": Found script element '$element' in uploaded file.\n" );
1461 return [ 'uploaded-script-svg', $strippedElement ];
1464 # e.g., <svg xmlns="http://www.w3.org/2000/svg">
1465 # <handler xmlns:ev="http://www.w3.org/2001/xml-events" ev:event="load">alert(1)</handler> </svg>
1466 if ( $strippedElement == 'handler' ) {
1467 wfDebug( __METHOD__
. ": Found scriptable element '$element' in uploaded file.\n" );
1469 return [ 'uploaded-script-svg', $strippedElement ];
1472 # SVG reported in Feb '12 that used xml:stylesheet to generate javascript block
1473 if ( $strippedElement == 'stylesheet' ) {
1474 wfDebug( __METHOD__
. ": Found scriptable element '$element' in uploaded file.\n" );
1476 return [ 'uploaded-script-svg', $strippedElement ];
1479 # Block iframes, in case they pass the namespace check
1480 if ( $strippedElement == 'iframe' ) {
1481 wfDebug( __METHOD__
. ": iframe in uploaded file.\n" );
1483 return [ 'uploaded-script-svg', $strippedElement ];
1487 if ( $strippedElement == 'style'
1488 && self
::checkCssFragment( Sanitizer
::normalizeCss( $data ) )
1490 wfDebug( __METHOD__
. ": hostile css in style element.\n" );
1491 return [ 'uploaded-hostile-svg' ];
1494 foreach ( $attribs as $attrib => $value ) {
1495 $stripped = $this->stripXmlNamespace( $attrib );
1496 $value = strtolower( $value );
1498 if ( substr( $stripped, 0, 2 ) == 'on' ) {
1500 . ": Found event-handler attribute '$attrib'='$value' in uploaded file.\n" );
1502 return [ 'uploaded-event-handler-on-svg', $attrib, $value ];
1505 # Do not allow relative links, or unsafe url schemas.
1506 # For <a> tags, only data:, http: and https: and same-document
1507 # fragment links are allowed. For all other tags, only data:
1508 # and fragment are allowed.
1509 if ( $stripped == 'href'
1511 && strpos( $value, 'data:' ) !== 0
1512 && strpos( $value, '#' ) !== 0
1514 if ( !( $strippedElement === 'a'
1515 && preg_match( '!^https?://!i', $value ) )
1517 wfDebug( __METHOD__
. ": Found href attribute <$strippedElement "
1518 . "'$attrib'='$value' in uploaded file.\n" );
1520 return [ 'uploaded-href-attribute-svg', $strippedElement, $attrib, $value ];
1524 # only allow data: targets that should be safe. This prevents vectors like,
1525 # image/svg, text/xml, application/xml, and text/html, which can contain scripts
1526 if ( $stripped == 'href' && strncasecmp( 'data:', $value, 5 ) === 0 ) {
1527 // rfc2397 parameters. This is only slightly slower than (;[\w;]+)*.
1528 // @codingStandardsIgnoreStart Generic.Files.LineLength
1529 $parameters = '(?>;[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+=(?>[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+|"(?>[\0-\x0c\x0e-\x21\x23-\x5b\x5d-\x7f]+|\\\\[\0-\x7f])*"))*(?:;base64)?';
1530 // @codingStandardsIgnoreEnd
1532 if ( !preg_match( "!^data:\s*image/(gif|jpeg|jpg|png)$parameters,!i", $value ) ) {
1533 wfDebug( __METHOD__
. ": Found href to unwhitelisted data: uri "
1534 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1535 return [ 'uploaded-href-unsafe-target-svg', $strippedElement, $attrib, $value ];
1539 # Change href with animate from (http://html5sec.org/#137).
1540 if ( $stripped === 'attributename'
1541 && $strippedElement === 'animate'
1542 && $this->stripXmlNamespace( $value ) == 'href'
1544 wfDebug( __METHOD__
. ": Found animate that might be changing href using from "
1545 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1547 return [ 'uploaded-animate-svg', $strippedElement, $attrib, $value ];
1550 # use set/animate to add event-handler attribute to parent
1551 if ( ( $strippedElement == 'set' ||
$strippedElement == 'animate' )
1552 && $stripped == 'attributename'
1553 && substr( $value, 0, 2 ) == 'on'
1555 wfDebug( __METHOD__
. ": Found svg setting event-handler attribute with "
1556 . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1558 return [ 'uploaded-setting-event-handler-svg', $strippedElement, $stripped, $value ];
1561 # use set to add href attribute to parent element
1562 if ( $strippedElement == 'set'
1563 && $stripped == 'attributename'
1564 && strpos( $value, 'href' ) !== false
1566 wfDebug( __METHOD__
. ": Found svg setting href attribute '$value' in uploaded file.\n" );
1568 return [ 'uploaded-setting-href-svg' ];
1571 # use set to add a remote / data / script target to an element
1572 if ( $strippedElement == 'set'
1573 && $stripped == 'to'
1574 && preg_match( '!(http|https|data|script):!sim', $value )
1576 wfDebug( __METHOD__
. ": Found svg setting attribute to '$value' in uploaded file.\n" );
1578 return [ 'uploaded-wrong-setting-svg', $value ];
1581 # use handler attribute with remote / data / script
1582 if ( $stripped == 'handler' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1583 wfDebug( __METHOD__
. ": Found svg setting handler with remote/data/script "
1584 . "'$attrib'='$value' in uploaded file.\n" );
1586 return [ 'uploaded-setting-handler-svg', $attrib, $value ];
1589 # use CSS styles to bring in remote code
1590 if ( $stripped == 'style'
1591 && self
::checkCssFragment( Sanitizer
::normalizeCss( $value ) )
1593 wfDebug( __METHOD__
. ": Found svg setting a style with "
1594 . "remote url '$attrib'='$value' in uploaded file.\n" );
1595 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1598 # Several attributes can include css, css character escaping isn't allowed
1599 $cssAttrs = [ 'font', 'clip-path', 'fill', 'filter', 'marker',
1600 'marker-end', 'marker-mid', 'marker-start', 'mask', 'stroke' ];
1601 if ( in_array( $stripped, $cssAttrs )
1602 && self
::checkCssFragment( $value )
1604 wfDebug( __METHOD__
. ": Found svg setting a style with "
1605 . "remote url '$attrib'='$value' in uploaded file.\n" );
1606 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1609 # image filters can pull in url, which could be svg that executes scripts
1610 if ( $strippedElement == 'image'
1611 && $stripped == 'filter'
1612 && preg_match( '!url\s*\(!sim', $value )
1614 wfDebug( __METHOD__
. ": Found image filter with url: "
1615 . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1617 return [ 'uploaded-image-filter-svg', $strippedElement, $stripped, $value ];
1621 return false; // No scripts detected
1625 * Check a block of CSS or CSS fragment for anything that looks like
1626 * it is bringing in remote code.
1627 * @param string $value a string of CSS
1628 * @param bool $propOnly only check css properties (start regex with :)
1629 * @return bool true if the CSS contains an illegal string, false if otherwise
1631 private static function checkCssFragment( $value ) {
1633 # Forbid external stylesheets, for both reliability and to protect viewer's privacy
1634 if ( stripos( $value, '@import' ) !== false ) {
1638 # We allow @font-face to embed fonts with data: urls, so we snip the string
1639 # 'url' out so this case won't match when we check for urls below
1640 $pattern = '!(@font-face\s*{[^}]*src:)url(\("data:;base64,)!im';
1641 $value = preg_replace( $pattern, '$1$2', $value );
1643 # Check for remote and executable CSS. Unlike in Sanitizer::checkCss, the CSS
1644 # properties filter and accelerator don't seem to be useful for xss in SVG files.
1645 # Expression and -o-link don't seem to work either, but filtering them here in case.
1646 # Additionally, we catch remote urls like url("http:..., url('http:..., url(http:...,
1647 # but not local ones such as url("#..., url('#..., url(#....
1648 if ( preg_match( '!expression
1650 | -o-link-source\s*:
1651 | -o-replace\s*:!imx', $value ) ) {
1655 if ( preg_match_all(
1656 "!(\s*(url|image|image-set)\s*\(\s*[\"']?\s*[^#]+.*?\))!sim",
1661 # TODO: redo this in one regex. Until then, url("#whatever") matches the first
1662 foreach ( $matches[1] as $match ) {
1663 if ( !preg_match( "!\s*(url|image|image-set)\s*\(\s*(#|'#|\"#)!im", $match ) ) {
1669 if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ) {
1677 * Divide the element name passed by the xml parser to the callback into URI and prifix.
1678 * @param string $element
1679 * @return array Containing the namespace URI and prefix
1681 private static function splitXmlNamespace( $element ) {
1682 // 'http://www.w3.org/2000/svg:script' -> [ 'http://www.w3.org/2000/svg', 'script' ]
1683 $parts = explode( ':', strtolower( $element ) );
1684 $name = array_pop( $parts );
1685 $ns = implode( ':', $parts );
1687 return [ $ns, $name ];
1691 * @param string $name
1694 private function stripXmlNamespace( $name ) {
1695 // 'http://www.w3.org/2000/svg:script' -> 'script'
1696 $parts = explode( ':', strtolower( $name ) );
1698 return array_pop( $parts );
1702 * Generic wrapper function for a virus scanner program.
1703 * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
1704 * $wgAntivirusRequired may be used to deny upload if the scan fails.
1706 * @param string $file Pathname to the temporary upload file
1707 * @return mixed False if not virus is found, null if the scan fails or is disabled,
1708 * or a string containing feedback from the virus scanner if a virus was found.
1709 * If textual feedback is missing but a virus was found, this function returns true.
1711 public static function detectVirus( $file ) {
1712 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
1714 if ( !$wgAntivirus ) {
1715 wfDebug( __METHOD__
. ": virus scanner disabled\n" );
1720 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
1721 wfDebug( __METHOD__
. ": unknown virus scanner: $wgAntivirus\n" );
1722 $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
1723 [ 'virus-badscanner', $wgAntivirus ] );
1725 return wfMessage( 'virus-unknownscanner' )->text() . " $wgAntivirus";
1728 # look up scanner configuration
1729 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
1730 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
1731 $msgPattern = isset( $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ) ?
1732 $wgAntivirusSetup[$wgAntivirus]['messagepattern'] : null;
1734 if ( strpos( $command, "%f" ) === false ) {
1735 # simple pattern: append file to scan
1736 $command .= " " . wfEscapeShellArg( $file );
1738 # complex pattern: replace "%f" with file to scan
1739 $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
1742 wfDebug( __METHOD__
. ": running virus scan: $command \n" );
1744 # execute virus scanner
1747 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1748 # that does not seem to be worth the pain.
1749 # Ask me (Duesentrieb) about it if it's ever needed.
1750 $output = wfShellExecWithStderr( $command, $exitCode );
1752 # map exit code to AV_xxx constants.
1753 $mappedCode = $exitCode;
1754 if ( $exitCodeMap ) {
1755 if ( isset( $exitCodeMap[$exitCode] ) ) {
1756 $mappedCode = $exitCodeMap[$exitCode];
1757 } elseif ( isset( $exitCodeMap["*"] ) ) {
1758 $mappedCode = $exitCodeMap["*"];
1762 /* NB: AV_NO_VIRUS is 0 but AV_SCAN_FAILED is false,
1763 * so we need the strict equalities === and thus can't use a switch here
1765 if ( $mappedCode === AV_SCAN_FAILED
) {
1766 # scan failed (code was mapped to false by $exitCodeMap)
1767 wfDebug( __METHOD__
. ": failed to scan $file (code $exitCode).\n" );
1769 $output = $wgAntivirusRequired
1770 ?
wfMessage( 'virus-scanfailed', [ $exitCode ] )->text()
1772 } elseif ( $mappedCode === AV_SCAN_ABORTED
) {
1773 # scan failed because filetype is unknown (probably imune)
1774 wfDebug( __METHOD__
. ": unsupported file type $file (code $exitCode).\n" );
1776 } elseif ( $mappedCode === AV_NO_VIRUS
) {
1778 wfDebug( __METHOD__
. ": file passed virus scan.\n" );
1781 $output = trim( $output );
1784 $output = true; # if there's no output, return true
1785 } elseif ( $msgPattern ) {
1787 if ( preg_match( $msgPattern, $output, $groups ) ) {
1789 $output = $groups[1];
1794 wfDebug( __METHOD__
. ": FOUND VIRUS! scanner feedback: $output \n" );
1801 * Check if there's an overwrite conflict and, if so, if restrictions
1802 * forbid this user from performing the upload.
1806 * @return mixed True on success, array on failure
1808 private function checkOverwrite( $user ) {
1809 // First check whether the local file can be overwritten
1810 $file = $this->getLocalFile();
1811 $file->load( File
::READ_LATEST
);
1812 if ( $file->exists() ) {
1813 if ( !self
::userCanReUpload( $user, $file ) ) {
1814 return [ 'fileexists-forbidden', $file->getName() ];
1820 /* Check shared conflicts: if the local file does not exist, but
1821 * wfFindFile finds a file, it exists in a shared repository.
1823 $file = wfFindFile( $this->getTitle(), [ 'latest' => true ] );
1824 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1825 return [ 'fileexists-shared-forbidden', $file->getName() ];
1832 * Check if a user is the last uploader
1838 public static function userCanReUpload( User
$user, File
$img ) {
1839 if ( $user->isAllowed( 'reupload' ) ) {
1840 return true; // non-conditional
1841 } elseif ( !$user->isAllowed( 'reupload-own' ) ) {
1845 if ( !( $img instanceof LocalFile
) ) {
1851 return $user->getId() == $img->getUser( 'id' );
1855 * Helper function that does various existence checks for a file.
1856 * The following checks are performed:
1858 * - Article with the same name as the file exists
1859 * - File exists with normalized extension
1860 * - The file looks like a thumbnail and the original exists
1862 * @param File $file The File object to check
1863 * @return mixed False if the file does not exists, else an array
1865 public static function getExistsWarning( $file ) {
1866 if ( $file->exists() ) {
1867 return [ 'warning' => 'exists', 'file' => $file ];
1870 if ( $file->getTitle()->getArticleID() ) {
1871 return [ 'warning' => 'page-exists', 'file' => $file ];
1874 if ( strpos( $file->getName(), '.' ) == false ) {
1875 $partname = $file->getName();
1878 $n = strrpos( $file->getName(), '.' );
1879 $extension = substr( $file->getName(), $n +
1 );
1880 $partname = substr( $file->getName(), 0, $n );
1882 $normalizedExtension = File
::normalizeExtension( $extension );
1884 if ( $normalizedExtension != $extension ) {
1885 // We're not using the normalized form of the extension.
1886 // Normal form is lowercase, using most common of alternate
1887 // extensions (eg 'jpg' rather than 'JPEG').
1889 // Check for another file using the normalized form...
1890 $nt_lc = Title
::makeTitle( NS_FILE
, "{$partname}.{$normalizedExtension}" );
1891 $file_lc = wfLocalFile( $nt_lc );
1893 if ( $file_lc->exists() ) {
1895 'warning' => 'exists-normalized',
1897 'normalizedFile' => $file_lc
1902 // Check for files with the same name but a different extension
1903 $similarFiles = RepoGroup
::singleton()->getLocalRepo()->findFilesByPrefix(
1904 "{$partname}.", 1 );
1905 if ( count( $similarFiles ) ) {
1907 'warning' => 'exists-normalized',
1909 'normalizedFile' => $similarFiles[0],
1913 if ( self
::isThumbName( $file->getName() ) ) {
1914 # Check for filenames like 50px- or 180px-, these are mostly thumbnails
1915 $nt_thb = Title
::newFromText(
1916 substr( $partname, strpos( $partname, '-' ) +
1 ) . '.' . $extension,
1919 $file_thb = wfLocalFile( $nt_thb );
1920 if ( $file_thb->exists() ) {
1922 'warning' => 'thumb',
1924 'thumbFile' => $file_thb
1927 // File does not exist, but we just don't like the name
1929 'warning' => 'thumb-name',
1931 'thumbFile' => $file_thb
1936 foreach ( self
::getFilenamePrefixBlacklist() as $prefix ) {
1937 if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
1939 'warning' => 'bad-prefix',
1950 * Helper function that checks whether the filename looks like a thumbnail
1951 * @param string $filename
1954 public static function isThumbName( $filename ) {
1955 $n = strrpos( $filename, '.' );
1956 $partname = $n ?
substr( $filename, 0, $n ) : $filename;
1959 substr( $partname, 3, 3 ) == 'px-' ||
1960 substr( $partname, 2, 3 ) == 'px-'
1962 preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) );
1966 * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
1968 * @return array List of prefixes
1970 public static function getFilenamePrefixBlacklist() {
1972 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
1973 if ( !$message->isDisabled() ) {
1974 $lines = explode( "\n", $message->plain() );
1975 foreach ( $lines as $line ) {
1976 // Remove comment lines
1977 $comment = substr( trim( $line ), 0, 1 );
1978 if ( $comment == '#' ||
$comment == '' ) {
1981 // Remove additional comments after a prefix
1982 $comment = strpos( $line, '#' );
1983 if ( $comment > 0 ) {
1984 $line = substr( $line, 0, $comment - 1 );
1986 $blacklist[] = trim( $line );
1994 * Gets image info about the file just uploaded.
1996 * Also has the effect of setting metadata to be an 'indexed tag name' in
1997 * returned API result if 'metadata' was requested. Oddly, we have to pass
1998 * the "result" object down just so it can do that with the appropriate
1999 * format, presumably.
2001 * @param ApiResult $result
2002 * @return array Image info
2004 public function getImageInfo( $result ) {
2005 $localFile = $this->getLocalFile();
2006 $stashFile = $this->getStashFile();
2007 // Calling a different API module depending on whether the file was stashed is less than optimal.
2008 // In fact, calling API modules here at all is less than optimal. Maybe it should be refactored.
2010 $imParam = ApiQueryStashImageInfo
::getPropertyNames();
2011 $info = ApiQueryStashImageInfo
::getInfo( $stashFile, array_flip( $imParam ), $result );
2013 $imParam = ApiQueryImageInfo
::getPropertyNames();
2014 $info = ApiQueryImageInfo
::getInfo( $localFile, array_flip( $imParam ), $result );
2021 * @param array $error
2024 public function convertVerifyErrorToStatus( $error ) {
2025 $code = $error['status'];
2026 unset( $code['status'] );
2028 return Status
::newFatal( $this->getVerificationErrorCode( $code ), $error );
2032 * Get the MediaWiki maximum uploaded file size for given type of upload, based on
2035 * @param null|string $forType
2038 public static function getMaxUploadSize( $forType = null ) {
2039 global $wgMaxUploadSize;
2041 if ( is_array( $wgMaxUploadSize ) ) {
2042 if ( !is_null( $forType ) && isset( $wgMaxUploadSize[$forType] ) ) {
2043 return $wgMaxUploadSize[$forType];
2045 return $wgMaxUploadSize['*'];
2048 return intval( $wgMaxUploadSize );
2053 * Get the PHP maximum uploaded file size, based on ini settings. If there is no limit or the
2054 * limit can't be guessed, returns a very large number (PHP_INT_MAX).
2059 public static function getMaxPhpUploadSize() {
2060 $phpMaxFileSize = wfShorthandToInteger(
2061 ini_get( 'upload_max_filesize' ) ?
: ini_get( 'hhvm.server.upload.upload_max_file_size' ),
2064 $phpMaxPostSize = wfShorthandToInteger(
2065 ini_get( 'post_max_size' ) ?
: ini_get( 'hhvm.server.max_post_size' ),
2068 return min( $phpMaxFileSize, $phpMaxPostSize );
2072 * Get the current status of a chunked upload (used for polling)
2074 * The value will be read from cache.
2077 * @param string $statusKey
2078 * @return Status[]|bool
2080 public static function getSessionStatus( User
$user, $statusKey ) {
2081 $key = wfMemcKey( 'uploadstatus', $user->getId() ?
: md5( $user->getName() ), $statusKey );
2083 return ObjectCache
::getMainStashInstance()->get( $key );
2087 * Set the current status of a chunked upload (used for polling)
2089 * The value will be set in cache for 1 day
2092 * @param string $statusKey
2093 * @param array|bool $value
2096 public static function setSessionStatus( User
$user, $statusKey, $value ) {
2097 $key = wfMemcKey( 'uploadstatus', $user->getId() ?
: md5( $user->getName() ), $statusKey );
2099 $cache = ObjectCache
::getMainStashInstance();
2100 if ( $value === false ) {
2101 $cache->delete( $key );
2103 $cache->set( $key, $value, $cache::TTL_DAY
);