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
23 use MediaWiki\MediaWikiServices
;
24 use MediaWiki\Shell\Shell
;
27 * @defgroup Upload Upload related
33 * UploadBase and subclasses are the backend of MediaWiki's file uploads.
34 * The frontends are formed by ApiUpload and SpecialUpload.
36 * @author Brion Vibber
37 * @author Bryan Tong Minh
38 * @author Michael Dale
40 abstract class UploadBase
{
41 /** @var string Local file system path to the file to upload (or a local copy) */
43 /** @var TempFSFile|null Wrapper to handle deleting the temp file */
44 protected $tempFileObj;
46 protected $mDesiredDestName, $mDestName, $mRemoveTempFile, $mSourceType;
47 protected $mTitle = false, $mTitleError = 0;
48 protected $mFilteredName, $mFinalExtension;
49 protected $mLocalFile, $mStashFile, $mFileSize, $mFileProps;
50 protected $mBlackListedExtensions;
51 protected $mJavaDetected, $mSVGNSError;
53 protected static $safeXmlEncodings = [
73 const MIN_LENGTH_PARTNAME
= 4;
74 const ILLEGAL_FILENAME
= 5;
75 const OVERWRITE_EXISTING_FILE
= 7; # Not used anymore; handled by verifyTitlePermissions()
76 const FILETYPE_MISSING
= 8;
77 const FILETYPE_BADTYPE
= 9;
78 const VERIFICATION_ERROR
= 10;
79 const HOOK_ABORTED
= 11;
80 const FILE_TOO_LARGE
= 12;
81 const WINDOWS_NONASCII_FILENAME
= 13;
82 const FILENAME_TOO_LONG
= 14;
88 public function getVerificationErrorCode( $error ) {
90 self
::EMPTY_FILE
=> 'empty-file',
91 self
::FILE_TOO_LARGE
=> 'file-too-large',
92 self
::FILETYPE_MISSING
=> 'filetype-missing',
93 self
::FILETYPE_BADTYPE
=> 'filetype-banned',
94 self
::MIN_LENGTH_PARTNAME
=> 'filename-tooshort',
95 self
::ILLEGAL_FILENAME
=> 'illegal-filename',
96 self
::OVERWRITE_EXISTING_FILE
=> 'overwrite',
97 self
::VERIFICATION_ERROR
=> 'verification-error',
98 self
::HOOK_ABORTED
=> 'hookaborted',
99 self
::WINDOWS_NONASCII_FILENAME
=> 'windows-nonascii-filename',
100 self
::FILENAME_TOO_LONG
=> 'filename-toolong',
102 return $code_to_status[$error] ??
'unknown-error';
106 * Returns true if uploads are enabled.
107 * Can be override by subclasses.
110 public static function isEnabled() {
111 global $wgEnableUploads;
113 if ( !$wgEnableUploads ) {
117 # Check php's file_uploads setting
118 return wfIsHHVM() ||
wfIniGetBool( 'file_uploads' );
122 * Returns true if the user can use this upload module or else a string
123 * identifying the missing permission.
124 * Can be overridden by subclasses.
127 * @return bool|string
129 public static function isAllowed( $user ) {
130 foreach ( [ 'upload', 'edit' ] as $permission ) {
131 if ( !$user->isAllowed( $permission ) ) {
140 * Returns true if the user has surpassed the upload rate limit, false otherwise.
145 public static function isThrottled( $user ) {
146 return $user->pingLimiter( 'upload' );
149 // Upload handlers. Should probably just be a global.
150 private static $uploadHandlers = [ 'Stash', 'File', 'Url' ];
153 * Create a form of UploadBase depending on wpSourceType and initializes it
155 * @param WebRequest &$request
156 * @param string|null $type
157 * @return null|UploadBase
159 public static function createFromRequest( &$request, $type = null ) {
160 $type = $type ?
: $request->getVal( 'wpSourceType', 'File' );
166 // Get the upload class
167 $type = ucfirst( $type );
169 // Give hooks the chance to handle this request
171 Hooks
::run( 'UploadCreateFromRequest', [ $type, &$className ] );
172 if ( is_null( $className ) ) {
173 $className = 'UploadFrom' . $type;
174 wfDebug( __METHOD__
. ": class name: $className\n" );
175 if ( !in_array( $type, self
::$uploadHandlers ) ) {
180 // Check whether this upload class is enabled
181 if ( !call_user_func( [ $className, 'isEnabled' ] ) ) {
185 // Check whether the request is valid
186 if ( !call_user_func( [ $className, 'isValidRequest' ], $request ) ) {
190 /** @var UploadBase $handler */
191 $handler = new $className;
193 $handler->initializeFromRequest( $request );
199 * Check whether a request if valid for this handler
200 * @param WebRequest $request
203 public static function isValidRequest( $request ) {
207 public function __construct() {
211 * Returns the upload type. Should be overridden by child classes
216 public function getSourceType() {
221 * Initialize the path information
222 * @param string $name The desired destination name
223 * @param string $tempPath The temporary path
224 * @param int $fileSize The file size
225 * @param bool $removeTempFile (false) remove the temporary file?
226 * @throws MWException
228 public function initializePathInfo( $name, $tempPath, $fileSize, $removeTempFile = false ) {
229 $this->mDesiredDestName
= $name;
230 if ( FileBackend
::isStoragePath( $tempPath ) ) {
231 throw new MWException( __METHOD__
. " given storage path `$tempPath`." );
234 $this->setTempFile( $tempPath, $fileSize );
235 $this->mRemoveTempFile
= $removeTempFile;
239 * Initialize from a WebRequest. Override this in a subclass.
241 * @param WebRequest &$request
243 abstract public function initializeFromRequest( &$request );
246 * @param string $tempPath File system path to temporary file containing the upload
247 * @param int|null $fileSize
249 protected function setTempFile( $tempPath, $fileSize = null ) {
250 $this->mTempPath
= $tempPath;
251 $this->mFileSize
= $fileSize ?
: null;
252 if ( strlen( $this->mTempPath
) && file_exists( $this->mTempPath
) ) {
253 $this->tempFileObj
= new TempFSFile( $this->mTempPath
);
255 $this->mFileSize
= filesize( $this->mTempPath
);
258 $this->tempFileObj
= null;
263 * Fetch the file. Usually a no-op
266 public function fetchFile() {
267 return Status
::newGood();
271 * Return true if the file is empty
274 public function isEmptyFile() {
275 return empty( $this->mFileSize
);
279 * Return the file size
282 public function getFileSize() {
283 return $this->mFileSize
;
287 * Get the base 36 SHA1 of the file
290 public function getTempFileSha1Base36() {
291 return FSFile
::getSha1Base36FromPath( $this->mTempPath
);
295 * @param string $srcPath The source path
296 * @return string|bool The real path if it was a virtual URL Returns false on failure
298 public function getRealPath( $srcPath ) {
299 $repo = RepoGroup
::singleton()->getLocalRepo();
300 if ( FileRepo
::isVirtualUrl( $srcPath ) ) {
301 /** @todo Just make uploads work with storage paths UploadFromStash
302 * loads files via virtual URLs.
304 $tmpFile = $repo->getLocalCopy( $srcPath );
306 $tmpFile->bind( $this ); // keep alive with $this
308 $path = $tmpFile ?
$tmpFile->getPath() : false;
317 * Verify whether the upload is sane.
318 * @return mixed Const self::OK or else an array with error information
320 public function verifyUpload() {
322 * If there was no filename or a zero size given, give up quick.
324 if ( $this->isEmptyFile() ) {
325 return [ 'status' => self
::EMPTY_FILE
];
329 * Honor $wgMaxUploadSize
331 $maxSize = self
::getMaxUploadSize( $this->getSourceType() );
332 if ( $this->mFileSize
> $maxSize ) {
334 'status' => self
::FILE_TOO_LARGE
,
340 * Look at the contents of the file; if we can recognize the
341 * type but it's corrupt or data of the wrong type, we should
342 * probably not accept it.
344 $verification = $this->verifyFile();
345 if ( $verification !== true ) {
347 'status' => self
::VERIFICATION_ERROR
,
348 'details' => $verification
353 * Make sure this file can be created
355 $result = $this->validateName();
356 if ( $result !== true ) {
361 if ( !Hooks
::run( 'UploadVerification',
362 [ $this->mDestName
, $this->mTempPath
, &$error ], '1.28' )
364 return [ 'status' => self
::HOOK_ABORTED
, 'error' => $error ];
367 return [ 'status' => self
::OK
];
371 * Verify that the name is valid and, if necessary, that we can overwrite
373 * @return mixed True if valid, otherwise and array with 'status'
376 public function validateName() {
377 $nt = $this->getTitle();
378 if ( is_null( $nt ) ) {
379 $result = [ 'status' => $this->mTitleError
];
380 if ( $this->mTitleError
== self
::ILLEGAL_FILENAME
) {
381 $result['filtered'] = $this->mFilteredName
;
383 if ( $this->mTitleError
== self
::FILETYPE_BADTYPE
) {
384 $result['finalExt'] = $this->mFinalExtension
;
385 if ( count( $this->mBlackListedExtensions
) ) {
386 $result['blacklistedExt'] = $this->mBlackListedExtensions
;
392 $this->mDestName
= $this->getLocalFile()->getName();
398 * Verify the MIME type.
400 * @note Only checks that it is not an evil MIME. The "does it have
401 * correct extension given its MIME type?" check is in verifyFile.
402 * in `verifyFile()` that MIME type and file extension correlate.
403 * @param string $mime Representing the MIME
404 * @return mixed True if the file is verified, an array otherwise
406 protected function verifyMimeType( $mime ) {
407 global $wgVerifyMimeType;
408 if ( $wgVerifyMimeType ) {
409 wfDebug( "mime: <$mime> extension: <{$this->mFinalExtension}>\n" );
410 global $wgMimeTypeBlacklist;
411 if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
412 return [ 'filetype-badmime', $mime ];
415 # Check what Internet Explorer would detect
416 $fp = fopen( $this->mTempPath
, 'rb' );
417 $chunk = fread( $fp, 256 );
420 $magic = MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer();
421 $extMime = $magic->guessTypesForExtension( $this->mFinalExtension
);
422 $ieTypes = $magic->getIEMimeTypes( $this->mTempPath
, $chunk, $extMime );
423 foreach ( $ieTypes as $ieType ) {
424 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
425 return [ 'filetype-bad-ie-mime', $ieType ];
434 * Verifies that it's ok to include the uploaded file
436 * @return mixed True of the file is verified, array otherwise.
438 protected function verifyFile() {
439 global $wgVerifyMimeType, $wgDisableUploadScriptChecks;
441 $status = $this->verifyPartialFile();
442 if ( $status !== true ) {
446 $mwProps = new MWFileProps( MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer() );
447 $this->mFileProps
= $mwProps->getPropsFromPath( $this->mTempPath
, $this->mFinalExtension
);
448 $mime = $this->mFileProps
['mime'];
450 if ( $wgVerifyMimeType ) {
451 # XXX: Missing extension will be caught by validateName() via getTitle()
452 if ( $this->mFinalExtension
!= '' && !$this->verifyExtension( $mime, $this->mFinalExtension
) ) {
453 return [ 'filetype-mime-mismatch', $this->mFinalExtension
, $mime ];
457 # check for htmlish code and javascript
458 if ( !$wgDisableUploadScriptChecks ) {
459 if ( $this->mFinalExtension
== 'svg' ||
$mime == 'image/svg+xml' ) {
460 $svgStatus = $this->detectScriptInSvg( $this->mTempPath
, false );
461 if ( $svgStatus !== false ) {
467 $handler = MediaHandler
::getHandler( $mime );
469 $handlerStatus = $handler->verifyUpload( $this->mTempPath
);
470 if ( !$handlerStatus->isOK() ) {
471 $errors = $handlerStatus->getErrorsArray();
473 return reset( $errors );
478 Hooks
::run( 'UploadVerifyFile', [ $this, $mime, &$error ] );
479 if ( $error !== true ) {
480 if ( !is_array( $error ) ) {
486 wfDebug( __METHOD__
. ": all clear; passing.\n" );
492 * A verification routine suitable for partial files
494 * Runs the blacklist checks, but not any checks that may
495 * assume the entire file is present.
497 * @return mixed True for valid or array with error message key.
499 protected function verifyPartialFile() {
500 global $wgAllowJavaUploads, $wgDisableUploadScriptChecks;
502 # getTitle() sets some internal parameters like $this->mFinalExtension
505 $mwProps = new MWFileProps( MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer() );
506 $this->mFileProps
= $mwProps->getPropsFromPath( $this->mTempPath
, $this->mFinalExtension
);
508 # check MIME type, if desired
509 $mime = $this->mFileProps
['file-mime'];
510 $status = $this->verifyMimeType( $mime );
511 if ( $status !== true ) {
515 # check for htmlish code and javascript
516 if ( !$wgDisableUploadScriptChecks ) {
517 if ( self
::detectScript( $this->mTempPath
, $mime, $this->mFinalExtension
) ) {
518 return [ 'uploadscripted' ];
520 if ( $this->mFinalExtension
== 'svg' ||
$mime == 'image/svg+xml' ) {
521 $svgStatus = $this->detectScriptInSvg( $this->mTempPath
, true );
522 if ( $svgStatus !== false ) {
528 # Check for Java applets, which if uploaded can bypass cross-site
530 if ( !$wgAllowJavaUploads ) {
531 $this->mJavaDetected
= false;
532 $zipStatus = ZipDirectoryReader
::read( $this->mTempPath
,
533 [ $this, 'zipEntryCallback' ] );
534 if ( !$zipStatus->isOK() ) {
535 $errors = $zipStatus->getErrorsArray();
536 $error = reset( $errors );
537 if ( $error[0] !== 'zip-wrong-format' ) {
541 if ( $this->mJavaDetected
) {
542 return [ 'uploadjava' ];
546 # Scan the uploaded file for viruses
547 $virus = $this->detectVirus( $this->mTempPath
);
549 return [ 'uploadvirus', $virus ];
556 * Callback for ZipDirectoryReader to detect Java class files.
558 * @param array $entry
560 public function zipEntryCallback( $entry ) {
561 $names = [ $entry['name'] ];
563 // If there is a null character, cut off the name at it, because JDK's
564 // ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name
565 // were constructed which had ".class\0" followed by a string chosen to
566 // make the hash collide with the truncated name, that file could be
567 // returned in response to a request for the .class file.
568 $nullPos = strpos( $entry['name'], "\000" );
569 if ( $nullPos !== false ) {
570 $names[] = substr( $entry['name'], 0, $nullPos );
573 // If there is a trailing slash in the file name, we have to strip it,
574 // because that's what ZIP_GetEntry() does.
575 if ( preg_grep( '!\.class/?$!', $names ) ) {
576 $this->mJavaDetected
= true;
581 * Alias for verifyTitlePermissions. The function was originally
582 * 'verifyPermissions', but that suggests it's checking the user, when it's
583 * really checking the title + user combination.
585 * @param User $user User object to verify the permissions against
586 * @return mixed An array as returned by getUserPermissionsErrors or true
587 * in case the user has proper permissions.
589 public function verifyPermissions( $user ) {
590 return $this->verifyTitlePermissions( $user );
594 * Check whether the user can edit, upload and create the image. This
595 * checks only against the current title; if it returns errors, it may
596 * very well be that another title will not give errors. Therefore
597 * isAllowed() should be called as well for generic is-user-blocked or
598 * can-user-upload checking.
600 * @param User $user User object to verify the permissions against
601 * @return mixed An array as returned by getUserPermissionsErrors or true
602 * in case the user has proper permissions.
604 public function verifyTitlePermissions( $user ) {
606 * If the image is protected, non-sysop users won't be able
607 * to modify it by uploading a new revision.
609 $nt = $this->getTitle();
610 if ( is_null( $nt ) ) {
613 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
614 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
615 if ( !$nt->exists() ) {
616 $permErrorsCreate = $nt->getUserPermissionsErrors( 'create', $user );
618 $permErrorsCreate = [];
620 if ( $permErrors ||
$permErrorsUpload ||
$permErrorsCreate ) {
621 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
622 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
627 $overwriteError = $this->checkOverwrite( $user );
628 if ( $overwriteError !== true ) {
629 return [ $overwriteError ];
636 * Check for non fatal problems with the file.
638 * This should not assume that mTempPath is set.
640 * @return mixed[] Array of warnings
642 public function checkWarnings() {
645 $localFile = $this->getLocalFile();
646 $localFile->load( File
::READ_LATEST
);
647 $filename = $localFile->getName();
648 $hash = $this->getTempFileSha1Base36();
650 $badFileName = $this->checkBadFileName( $filename, $this->mDesiredDestName
);
651 if ( $badFileName !== null ) {
652 $warnings['badfilename'] = $badFileName;
655 $unwantedFileExtensionDetails = $this->checkUnwantedFileExtensions( $this->mFinalExtension
);
656 if ( $unwantedFileExtensionDetails !== null ) {
657 $warnings['filetype-unwanted-type'] = $unwantedFileExtensionDetails;
660 $fileSizeWarnings = $this->checkFileSize( $this->mFileSize
);
661 if ( $fileSizeWarnings ) {
662 $warnings = array_merge( $warnings, $fileSizeWarnings );
665 $localFileExistsWarnings = $this->checkLocalFileExists( $localFile, $hash );
666 if ( $localFileExistsWarnings ) {
667 $warnings = array_merge( $warnings, $localFileExistsWarnings );
670 if ( $this->checkLocalFileWasDeleted( $localFile ) ) {
671 $warnings['was-deleted'] = $filename;
674 // If a file with the same name exists locally then the local file has already been tested
675 // for duplication of content
676 $ignoreLocalDupes = isset( $warnings[ 'exists '] );
677 $dupes = $this->checkAgainstExistingDupes( $hash, $ignoreLocalDupes );
679 $warnings['duplicate'] = $dupes;
682 $archivedDupes = $this->checkAgainstArchiveDupes( $hash );
683 if ( $archivedDupes !== null ) {
684 $warnings['duplicate-archive'] = $archivedDupes;
691 * Check whether the resulting filename is different from the desired one,
692 * but ignore things like ucfirst() and spaces/underscore things
694 * @param string $filename
695 * @param string $desiredFileName
697 * @return string|null String that was determined to be bad or null if the filename is okay
699 private function checkBadFileName( $filename, $desiredFileName ) {
700 $comparableName = str_replace( ' ', '_', $desiredFileName );
701 $comparableName = Title
::capitalize( $comparableName, NS_FILE
);
703 if ( $desiredFileName != $filename && $comparableName != $filename ) {
711 * @param string $fileExtension The file extension to check
713 * @return array|null array with the following keys:
714 * 0 => string The final extension being used
715 * 1 => string[] The extensions that are allowed
716 * 2 => int The number of extensions that are allowed.
718 private function checkUnwantedFileExtensions( $fileExtension ) {
719 global $wgCheckFileExtensions, $wgFileExtensions, $wgLang;
721 if ( $wgCheckFileExtensions ) {
722 $extensions = array_unique( $wgFileExtensions );
723 if ( !$this->checkFileExtension( $fileExtension, $extensions ) ) {
726 $wgLang->commaList( $extensions ),
736 * @param int $fileSize
738 * @return array warnings
740 private function checkFileSize( $fileSize ) {
741 global $wgUploadSizeWarning;
745 if ( $wgUploadSizeWarning && ( $fileSize > $wgUploadSizeWarning ) ) {
746 $warnings['large-file'] = [ $wgUploadSizeWarning, $fileSize ];
749 if ( $fileSize == 0 ) {
750 $warnings['empty-file'] = true;
757 * @param LocalFile $localFile
758 * @param string $hash sha1 hash of the file to check
760 * @return array warnings
762 private function checkLocalFileExists( LocalFile
$localFile, $hash ) {
765 $exists = self
::getExistsWarning( $localFile );
766 if ( $exists !== false ) {
767 $warnings['exists'] = $exists;
769 // check if file is an exact duplicate of current file version
770 if ( $hash === $localFile->getSha1() ) {
771 $warnings['no-change'] = $localFile;
774 // check if file is an exact duplicate of older versions of this file
775 $history = $localFile->getHistory();
776 foreach ( $history as $oldFile ) {
777 if ( $hash === $oldFile->getSha1() ) {
778 $warnings['duplicate-version'][] = $oldFile;
786 private function checkLocalFileWasDeleted( LocalFile
$localFile ) {
787 return $localFile->wasDeleted() && !$localFile->exists();
791 * @param string $hash sha1 hash of the file to check
792 * @param bool $ignoreLocalDupes True to ignore local duplicates
794 * @return File[] Duplicate files, if found.
796 private function checkAgainstExistingDupes( $hash, $ignoreLocalDupes ) {
797 $dupes = RepoGroup
::singleton()->findBySha1( $hash );
798 $title = $this->getTitle();
799 foreach ( $dupes as $key => $dupe ) {
801 ( $dupe instanceof LocalFile
) &&
803 $title->equals( $dupe->getTitle() )
805 unset( $dupes[$key] );
813 * @param string $hash sha1 hash of the file to check
815 * @return string|null Name of the dupe or empty string if discovered (depending on visibility)
816 * null if the check discovered no dupes.
818 private function checkAgainstArchiveDupes( $hash ) {
819 $archivedFile = new ArchivedFile( null, 0, '', $hash );
820 if ( $archivedFile->getID() > 0 ) {
821 if ( $archivedFile->userCan( File
::DELETED_FILE
) ) {
822 return $archivedFile->getName();
832 * Really perform the upload. Stores the file in the local repo, watches
833 * if necessary and runs the UploadComplete hook.
835 * @param string $comment
836 * @param string $pageText
837 * @param bool $watch Whether the file page should be added to user's watchlist.
838 * (This doesn't check $user's permissions.)
840 * @param string[] $tags Change tags to add to the log entry and page revision.
841 * (This doesn't check $user's permissions.)
842 * @return Status Indicating the whether the upload succeeded.
844 public function performUpload( $comment, $pageText, $watch, $user, $tags = [] ) {
845 $this->getLocalFile()->load( File
::READ_LATEST
);
846 $props = $this->mFileProps
;
849 Hooks
::run( 'UploadVerifyUpload', [ $this, $user, $props, $comment, $pageText, &$error ] );
851 if ( !is_array( $error ) ) {
854 return Status
::newFatal( ...$error );
857 $status = $this->getLocalFile()->upload(
868 if ( $status->isGood() ) {
870 WatchAction
::doWatch(
871 $this->getLocalFile()->getTitle(),
873 User
::IGNORE_USER_RIGHTS
876 // Avoid PHP 7.1 warning of passing $this by reference
878 Hooks
::run( 'UploadComplete', [ &$uploadBase ] );
880 $this->postProcessUpload();
887 * Perform extra steps after a successful upload.
891 public function postProcessUpload() {
895 * Returns the title of the file to be uploaded. Sets mTitleError in case
896 * the name was illegal.
898 * @return Title|null The title of the file or null in case the name was illegal
900 public function getTitle() {
901 if ( $this->mTitle
!== false ) {
902 return $this->mTitle
;
904 if ( !is_string( $this->mDesiredDestName
) ) {
905 $this->mTitleError
= self
::ILLEGAL_FILENAME
;
906 $this->mTitle
= null;
908 return $this->mTitle
;
910 /* Assume that if a user specified File:Something.jpg, this is an error
911 * and that the namespace prefix needs to be stripped of.
913 $title = Title
::newFromText( $this->mDesiredDestName
);
914 if ( $title && $title->getNamespace() == NS_FILE
) {
915 $this->mFilteredName
= $title->getDBkey();
917 $this->mFilteredName
= $this->mDesiredDestName
;
920 # oi_archive_name is max 255 bytes, which include a timestamp and an
921 # exclamation mark, so restrict file name to 240 bytes.
922 if ( strlen( $this->mFilteredName
) > 240 ) {
923 $this->mTitleError
= self
::FILENAME_TOO_LONG
;
924 $this->mTitle
= null;
926 return $this->mTitle
;
930 * Chop off any directories in the given filename. Then
931 * filter out illegal characters, and try to make a legible name
932 * out of it. We'll strip some silently that Title would die on.
934 $this->mFilteredName
= wfStripIllegalFilenameChars( $this->mFilteredName
);
935 /* Normalize to title form before we do any further processing */
936 $nt = Title
::makeTitleSafe( NS_FILE
, $this->mFilteredName
);
937 if ( is_null( $nt ) ) {
938 $this->mTitleError
= self
::ILLEGAL_FILENAME
;
939 $this->mTitle
= null;
941 return $this->mTitle
;
943 $this->mFilteredName
= $nt->getDBkey();
946 * We'll want to blacklist against *any* 'extension', and use
947 * only the final one for the whitelist.
949 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName
);
952 $this->mFinalExtension
= trim( end( $ext ) );
954 $this->mFinalExtension
= '';
956 # No extension, try guessing one
957 $magic = MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer();
958 $mime = $magic->guessMimeType( $this->mTempPath
);
959 if ( $mime !== 'unknown/unknown' ) {
960 # Get a space separated list of extensions
961 $extList = $magic->getExtensionsForType( $mime );
963 # Set the extension to the canonical extension
964 $this->mFinalExtension
= strtok( $extList, ' ' );
966 # Fix up the other variables
967 $this->mFilteredName
.= ".{$this->mFinalExtension}";
968 $nt = Title
::makeTitleSafe( NS_FILE
, $this->mFilteredName
);
969 $ext = [ $this->mFinalExtension
];
974 /* Don't allow users to override the blacklist (check file extension) */
975 global $wgCheckFileExtensions, $wgStrictFileExtensions;
976 global $wgFileExtensions, $wgFileBlacklist;
978 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
980 if ( $this->mFinalExtension
== '' ) {
981 $this->mTitleError
= self
::FILETYPE_MISSING
;
982 $this->mTitle
= null;
984 return $this->mTitle
;
985 } elseif ( $blackListedExtensions ||
986 ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
987 !$this->checkFileExtension( $this->mFinalExtension
, $wgFileExtensions ) )
989 $this->mBlackListedExtensions
= $blackListedExtensions;
990 $this->mTitleError
= self
::FILETYPE_BADTYPE
;
991 $this->mTitle
= null;
993 return $this->mTitle
;
996 // Windows may be broken with special characters, see T3780
997 if ( !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() )
998 && !RepoGroup
::singleton()->getLocalRepo()->backendSupportsUnicodePaths()
1000 $this->mTitleError
= self
::WINDOWS_NONASCII_FILENAME
;
1001 $this->mTitle
= null;
1003 return $this->mTitle
;
1006 # If there was more than one "extension", reassemble the base
1007 # filename to prevent bogus complaints about length
1008 if ( count( $ext ) > 1 ) {
1009 $iterations = count( $ext ) - 1;
1010 for ( $i = 0; $i < $iterations; $i++
) {
1011 $partname .= '.' . $ext[$i];
1015 if ( strlen( $partname ) < 1 ) {
1016 $this->mTitleError
= self
::MIN_LENGTH_PARTNAME
;
1017 $this->mTitle
= null;
1019 return $this->mTitle
;
1022 $this->mTitle
= $nt;
1024 return $this->mTitle
;
1028 * Return the local file and initializes if necessary.
1030 * @return LocalFile|null
1032 public function getLocalFile() {
1033 if ( is_null( $this->mLocalFile
) ) {
1034 $nt = $this->getTitle();
1035 $this->mLocalFile
= is_null( $nt ) ?
null : wfLocalFile( $nt );
1038 return $this->mLocalFile
;
1042 * @return UploadStashFile|null
1044 public function getStashFile() {
1045 return $this->mStashFile
;
1049 * Like stashFile(), but respects extensions' wishes to prevent the stashing. verifyUpload() must
1050 * be called before calling this method (unless $isPartial is true).
1052 * Upload stash exceptions are also caught and converted to an error status.
1056 * @param bool $isPartial Pass `true` if this is a part of a chunked upload (not a complete file).
1057 * @return Status If successful, value is an UploadStashFile instance
1059 public function tryStashFile( User
$user, $isPartial = false ) {
1060 if ( !$isPartial ) {
1061 $error = $this->runUploadStashFileHook( $user );
1063 return Status
::newFatal( ...$error );
1067 $file = $this->doStashFile( $user );
1068 return Status
::newGood( $file );
1069 } catch ( UploadStashException
$e ) {
1070 return Status
::newFatal( 'uploadstash-exception', get_class( $e ), $e->getMessage() );
1076 * @return array|null Error message and parameters, null if there's no error
1078 protected function runUploadStashFileHook( User
$user ) {
1079 $props = $this->mFileProps
;
1081 Hooks
::run( 'UploadStashFile', [ $this, $user, $props, &$error ] );
1082 if ( $error && !is_array( $error ) ) {
1083 $error = [ $error ];
1089 * If the user does not supply all necessary information in the first upload
1090 * form submission (either by accident or by design) then we may want to
1091 * stash the file temporarily, get more information, and publish the file
1094 * This method will stash a file in a temporary directory for later
1095 * processing, and save the necessary descriptive info into the database.
1096 * This method returns the file object, which also has a 'fileKey' property
1097 * which can be passed through a form or API request to find this stashed
1100 * @deprecated since 1.28 Use tryStashFile() instead
1101 * @param User|null $user
1102 * @return UploadStashFile Stashed file
1103 * @throws UploadStashBadPathException
1104 * @throws UploadStashFileException
1105 * @throws UploadStashNotLoggedInException
1107 public function stashFile( User
$user = null ) {
1108 return $this->doStashFile( $user );
1112 * Implementation for stashFile() and tryStashFile().
1114 * @param User|null $user
1115 * @return UploadStashFile Stashed file
1117 protected function doStashFile( User
$user = null ) {
1118 $stash = RepoGroup
::singleton()->getLocalRepo()->getUploadStash( $user );
1119 $file = $stash->stashFile( $this->mTempPath
, $this->getSourceType() );
1120 $this->mStashFile
= $file;
1126 * Stash a file in a temporary directory, returning a key which can be used
1127 * to find the file again. See stashFile().
1129 * @deprecated since 1.28
1130 * @return string File key
1132 public function stashFileGetKey() {
1133 wfDeprecated( __METHOD__
, '1.28' );
1134 return $this->doStashFile()->getFileKey();
1138 * alias for stashFileGetKey, for backwards compatibility
1140 * @deprecated since 1.28
1141 * @return string File key
1143 public function stashSession() {
1144 wfDeprecated( __METHOD__
, '1.28' );
1145 return $this->doStashFile()->getFileKey();
1149 * If we've modified the upload file we need to manually remove it
1150 * on exit to clean up.
1152 public function cleanupTempFile() {
1153 if ( $this->mRemoveTempFile
&& $this->tempFileObj
) {
1154 // Delete when all relevant TempFSFile handles go out of scope
1155 wfDebug( __METHOD__
. ": Marked temporary file '{$this->mTempPath}' for removal\n" );
1156 $this->tempFileObj
->autocollect();
1160 public function getTempPath() {
1161 return $this->mTempPath
;
1165 * Split a file into a base name and all dot-delimited 'extensions'
1166 * on the end. Some web server configurations will fall back to
1167 * earlier pseudo-'extensions' to determine type and execute
1168 * scripts, so the blacklist needs to check them all.
1170 * @param string $filename
1171 * @return array [ string, string[] ]
1173 public static function splitExtensions( $filename ) {
1174 $bits = explode( '.', $filename );
1175 $basename = array_shift( $bits );
1177 return [ $basename, $bits ];
1181 * Perform case-insensitive match against a list of file extensions.
1182 * Returns true if the extension is in the list.
1184 * @param string $ext
1185 * @param array $list
1188 public static function checkFileExtension( $ext, $list ) {
1189 return in_array( strtolower( $ext ), $list );
1193 * Perform case-insensitive match against a list of file extensions.
1194 * Returns an array of matching extensions.
1196 * @param string[] $ext
1197 * @param string[] $list
1200 public static function checkFileExtensionList( $ext, $list ) {
1201 return array_intersect( array_map( 'strtolower', $ext ), $list );
1205 * Checks if the MIME type of the uploaded file matches the file extension.
1207 * @param string $mime The MIME type of the uploaded file
1208 * @param string $extension The filename extension that the file is to be served with
1211 public static function verifyExtension( $mime, $extension ) {
1212 $magic = MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer();
1214 if ( !$mime ||
$mime == 'unknown' ||
$mime == 'unknown/unknown' ) {
1215 if ( !$magic->isRecognizableExtension( $extension ) ) {
1216 wfDebug( __METHOD__
. ": passing file with unknown detected mime type; " .
1217 "unrecognized extension '$extension', can't verify\n" );
1221 wfDebug( __METHOD__
. ": rejecting file with unknown detected mime type; " .
1222 "recognized extension '$extension', so probably invalid file\n" );
1228 $match = $magic->isMatchingExtension( $extension, $mime );
1230 if ( $match === null ) {
1231 if ( $magic->getTypesForExtension( $extension ) !== null ) {
1232 wfDebug( __METHOD__
. ": No extension known for $mime, but we know a mime for $extension\n" );
1236 wfDebug( __METHOD__
. ": no file extension known for mime type $mime, passing file\n" );
1240 } elseif ( $match === true ) {
1241 wfDebug( __METHOD__
. ": mime type $mime matches extension $extension, passing file\n" );
1243 /** @todo If it's a bitmap, make sure PHP or ImageMagick resp. can handle it! */
1247 . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
1254 * Heuristic for detecting files that *could* contain JavaScript instructions or
1255 * things that may look like HTML to a browser and are thus
1256 * potentially harmful. The present implementation will produce false
1257 * positives in some situations.
1259 * @param string $file Pathname to the temporary upload file
1260 * @param string $mime The MIME type of the file
1261 * @param string $extension The extension of the file
1262 * @return bool True if the file contains something looking like embedded scripts
1264 public static function detectScript( $file, $mime, $extension ) {
1265 global $wgAllowTitlesInSVG;
1267 # ugly hack: for text files, always look at the entire file.
1268 # For binary field, just check the first K.
1270 if ( strpos( $mime, 'text/' ) === 0 ) {
1271 $chunk = file_get_contents( $file );
1273 $fp = fopen( $file, 'rb' );
1274 $chunk = fread( $fp, 1024 );
1278 $chunk = strtolower( $chunk );
1284 # decode from UTF-16 if needed (could be used for obfuscation).
1285 if ( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
1287 } elseif ( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
1294 $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
1297 $chunk = trim( $chunk );
1299 /** @todo FIXME: Convert from UTF-16 if necessary! */
1300 wfDebug( __METHOD__
. ": checking for embedded scripts and HTML stuff\n" );
1302 # check for HTML doctype
1303 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
1307 // Some browsers will interpret obscure xml encodings as UTF-8, while
1308 // PHP/expat will interpret the given encoding in the xml declaration (T49304)
1309 if ( $extension == 'svg' ||
strpos( $mime, 'image/svg' ) === 0 ) {
1310 if ( self
::checkXMLEncodingMissmatch( $file ) ) {
1316 * Internet Explorer for Windows performs some really stupid file type
1317 * autodetection which can cause it to interpret valid image files as HTML
1318 * and potentially execute JavaScript, creating a cross-site scripting
1321 * Apple's Safari browser also performs some unsafe file type autodetection
1322 * which can cause legitimate files to be interpreted as HTML if the
1323 * web server is not correctly configured to send the right content-type
1324 * (or if you're really uploading plain text and octet streams!)
1326 * Returns true if IE is likely to mistake the given file for HTML.
1327 * Also returns true if Safari would mistake the given file for HTML
1328 * when served with a generic content-type.
1334 '<html', # also in safari
1337 '<script', # also in safari
1341 if ( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
1345 foreach ( $tags as $tag ) {
1346 if ( strpos( $chunk, $tag ) !== false ) {
1347 wfDebug( __METHOD__
. ": found something that may make it be mistaken for html: $tag\n" );
1354 * look for JavaScript
1357 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
1358 $chunk = Sanitizer
::decodeCharReferences( $chunk );
1360 # look for script-types
1361 if ( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
1362 wfDebug( __METHOD__
. ": found script types\n" );
1367 # look for html-style script-urls
1368 if ( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1369 wfDebug( __METHOD__
. ": found html-style script urls\n" );
1374 # look for css-style script-urls
1375 if ( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1376 wfDebug( __METHOD__
. ": found css-style script urls\n" );
1381 wfDebug( __METHOD__
. ": no scripts found\n" );
1387 * Check a whitelist of xml encodings that are known not to be interpreted differently
1388 * by the server's xml parser (expat) and some common browsers.
1390 * @param string $file Pathname to the temporary upload file
1391 * @return bool True if the file contains an encoding that could be misinterpreted
1393 public static function checkXMLEncodingMissmatch( $file ) {
1394 global $wgSVGMetadataCutoff;
1395 $contents = file_get_contents( $file, false, null, 0, $wgSVGMetadataCutoff );
1396 $encodingRegex = '!encoding[ \t\n\r]*=[ \t\n\r]*[\'"](.*?)[\'"]!si';
1398 if ( preg_match( "!<\?xml\b(.*?)\?>!si", $contents, $matches ) ) {
1399 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1400 && !in_array( strtoupper( $encMatch[1] ), self
::$safeXmlEncodings )
1402 wfDebug( __METHOD__
. ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1406 } elseif ( preg_match( "!<\?xml\b!si", $contents ) ) {
1407 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1408 // bytes. There shouldn't be a legitimate reason for this to happen.
1409 wfDebug( __METHOD__
. ": Unmatched XML declaration start\n" );
1412 } elseif ( substr( $contents, 0, 4 ) == "\x4C\x6F\xA7\x94" ) {
1413 // EBCDIC encoded XML
1414 wfDebug( __METHOD__
. ": EBCDIC Encoded XML\n" );
1419 // It's possible the file is encoded with multi-byte encoding, so re-encode attempt to
1420 // detect the encoding in case is specifies an encoding not whitelisted in self::$safeXmlEncodings
1421 $attemptEncodings = [ 'UTF-16', 'UTF-16BE', 'UTF-32', 'UTF-32BE' ];
1422 foreach ( $attemptEncodings as $encoding ) {
1423 Wikimedia\
suppressWarnings();
1424 $str = iconv( $encoding, 'UTF-8', $contents );
1425 Wikimedia\restoreWarnings
();
1426 if ( $str != '' && preg_match( "!<\?xml\b(.*?)\?>!si", $str, $matches ) ) {
1427 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1428 && !in_array( strtoupper( $encMatch[1] ), self
::$safeXmlEncodings )
1430 wfDebug( __METHOD__
. ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1434 } elseif ( $str != '' && preg_match( "!<\?xml\b!si", $str ) ) {
1435 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1436 // bytes. There shouldn't be a legitimate reason for this to happen.
1437 wfDebug( __METHOD__
. ": Unmatched XML declaration start\n" );
1447 * @param string $filename
1448 * @param bool $partial
1449 * @return bool|array
1451 protected function detectScriptInSvg( $filename, $partial ) {
1452 $this->mSVGNSError
= false;
1453 $check = new XmlTypeCheck(
1455 [ $this, 'checkSvgScriptCallback' ],
1458 'processing_instruction_handler' => 'UploadBase::checkSvgPICallback',
1459 'external_dtd_handler' => 'UploadBase::checkSvgExternalDTD',
1462 if ( $check->wellFormed
!== true ) {
1463 // Invalid xml (T60553)
1464 // But only when non-partial (T67724)
1465 return $partial ?
false : [ 'uploadinvalidxml' ];
1466 } elseif ( $check->filterMatch
) {
1467 if ( $this->mSVGNSError
) {
1468 return [ 'uploadscriptednamespace', $this->mSVGNSError
];
1471 return $check->filterMatchType
;
1478 * Callback to filter SVG Processing Instructions.
1479 * @param string $target Processing instruction name
1480 * @param string $data Processing instruction attribute and value
1481 * @return bool|array
1483 public static function checkSvgPICallback( $target, $data ) {
1484 // Don't allow external stylesheets (T59550)
1485 if ( preg_match( '/xml-stylesheet/i', $target ) ) {
1486 return [ 'upload-scripted-pi-callback' ];
1493 * Verify that DTD urls referenced are only the standard dtds
1495 * Browsers seem to ignore external dtds. However just to be on the
1496 * safe side, only allow dtds from the svg standard.
1498 * @param string $type PUBLIC or SYSTEM
1499 * @param string $publicId The well-known public identifier for the dtd
1500 * @param string $systemId The url for the external dtd
1501 * @return bool|array
1503 public static function checkSvgExternalDTD( $type, $publicId, $systemId ) {
1504 // This doesn't include the XHTML+MathML+SVG doctype since we don't
1505 // allow XHTML anyways.
1507 'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd',
1508 'http://www.w3.org/TR/2001/REC-SVG-20010904/DTD/svg10.dtd',
1509 'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-basic.dtd',
1510 'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-tiny.dtd',
1511 // https://phabricator.wikimedia.org/T168856
1512 'http://www.w3.org/TR/2001/PR-SVG-20010719/DTD/svg10.dtd',
1514 if ( $type !== 'PUBLIC'
1515 ||
!in_array( $systemId, $allowedDTDs )
1516 ||
strpos( $publicId, "-//W3C//" ) !== 0
1518 return [ 'upload-scripted-dtd' ];
1524 * @todo Replace this with a whitelist filter!
1525 * @param string $element
1526 * @param array $attribs
1527 * @param array|null $data
1528 * @return bool|array
1530 public function checkSvgScriptCallback( $element, $attribs, $data = null ) {
1531 list( $namespace, $strippedElement ) = $this->splitXmlNamespace( $element );
1533 // We specifically don't include:
1534 // http://www.w3.org/1999/xhtml (T62771)
1535 static $validNamespaces = [
1538 'http://creativecommons.org/ns#',
1539 'http://inkscape.sourceforge.net/dtd/sodipodi-0.dtd',
1540 'http://ns.adobe.com/adobeillustrator/10.0/',
1541 'http://ns.adobe.com/adobesvgviewerextensions/3.0/',
1542 'http://ns.adobe.com/extensibility/1.0/',
1543 'http://ns.adobe.com/flows/1.0/',
1544 'http://ns.adobe.com/illustrator/1.0/',
1545 'http://ns.adobe.com/imagereplacement/1.0/',
1546 'http://ns.adobe.com/pdf/1.3/',
1547 'http://ns.adobe.com/photoshop/1.0/',
1548 'http://ns.adobe.com/saveforweb/1.0/',
1549 'http://ns.adobe.com/variables/1.0/',
1550 'http://ns.adobe.com/xap/1.0/',
1551 'http://ns.adobe.com/xap/1.0/g/',
1552 'http://ns.adobe.com/xap/1.0/g/img/',
1553 'http://ns.adobe.com/xap/1.0/mm/',
1554 'http://ns.adobe.com/xap/1.0/rights/',
1555 'http://ns.adobe.com/xap/1.0/stype/dimensions#',
1556 'http://ns.adobe.com/xap/1.0/stype/font#',
1557 'http://ns.adobe.com/xap/1.0/stype/manifestitem#',
1558 'http://ns.adobe.com/xap/1.0/stype/resourceevent#',
1559 'http://ns.adobe.com/xap/1.0/stype/resourceref#',
1560 'http://ns.adobe.com/xap/1.0/t/pg/',
1561 'http://purl.org/dc/elements/1.1/',
1562 'http://purl.org/dc/elements/1.1',
1563 'http://schemas.microsoft.com/visio/2003/svgextensions/',
1564 'http://sodipodi.sourceforge.net/dtd/sodipodi-0.dtd',
1565 'http://taptrix.com/inkpad/svg_extensions',
1566 'http://web.resource.org/cc/',
1567 'http://www.freesoftware.fsf.org/bkchem/cdml',
1568 'http://www.inkscape.org/namespaces/inkscape',
1569 'http://www.opengis.net/gml',
1570 'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
1571 'http://www.w3.org/2000/svg',
1572 'http://www.w3.org/tr/rec-rdf-syntax/',
1573 'http://www.w3.org/2000/01/rdf-schema#',
1576 // Inkscape mangles namespace definitions created by Adobe Illustrator.
1577 // This is nasty but harmless. (T144827)
1578 $isBuggyInkscape = preg_match( '/^&(#38;)*ns_[a-z_]+;$/', $namespace );
1580 if ( !( $isBuggyInkscape ||
in_array( $namespace, $validNamespaces ) ) ) {
1581 wfDebug( __METHOD__
. ": Non-svg namespace '$namespace' in uploaded file.\n" );
1582 /** @todo Return a status object to a closure in XmlTypeCheck, for MW1.21+ */
1583 $this->mSVGNSError
= $namespace;
1589 * check for elements that can contain javascript
1591 if ( $strippedElement == 'script' ) {
1592 wfDebug( __METHOD__
. ": Found script element '$element' in uploaded file.\n" );
1594 return [ 'uploaded-script-svg', $strippedElement ];
1597 # e.g., <svg xmlns="http://www.w3.org/2000/svg">
1598 # <handler xmlns:ev="http://www.w3.org/2001/xml-events" ev:event="load">alert(1)</handler> </svg>
1599 if ( $strippedElement == 'handler' ) {
1600 wfDebug( __METHOD__
. ": Found scriptable element '$element' in uploaded file.\n" );
1602 return [ 'uploaded-script-svg', $strippedElement ];
1605 # SVG reported in Feb '12 that used xml:stylesheet to generate javascript block
1606 if ( $strippedElement == 'stylesheet' ) {
1607 wfDebug( __METHOD__
. ": Found scriptable element '$element' in uploaded file.\n" );
1609 return [ 'uploaded-script-svg', $strippedElement ];
1612 # Block iframes, in case they pass the namespace check
1613 if ( $strippedElement == 'iframe' ) {
1614 wfDebug( __METHOD__
. ": iframe in uploaded file.\n" );
1616 return [ 'uploaded-script-svg', $strippedElement ];
1620 if ( $strippedElement == 'style'
1621 && self
::checkCssFragment( Sanitizer
::normalizeCss( $data ) )
1623 wfDebug( __METHOD__
. ": hostile css in style element.\n" );
1624 return [ 'uploaded-hostile-svg' ];
1627 foreach ( $attribs as $attrib => $value ) {
1628 $stripped = $this->stripXmlNamespace( $attrib );
1629 $value = strtolower( $value );
1631 if ( substr( $stripped, 0, 2 ) == 'on' ) {
1633 . ": Found event-handler attribute '$attrib'='$value' in uploaded file.\n" );
1635 return [ 'uploaded-event-handler-on-svg', $attrib, $value ];
1638 # Do not allow relative links, or unsafe url schemas.
1639 # For <a> tags, only data:, http: and https: and same-document
1640 # fragment links are allowed. For all other tags, only data:
1641 # and fragment are allowed.
1642 if ( $stripped == 'href'
1644 && strpos( $value, 'data:' ) !== 0
1645 && strpos( $value, '#' ) !== 0
1647 if ( !( $strippedElement === 'a'
1648 && preg_match( '!^https?://!i', $value ) )
1650 wfDebug( __METHOD__
. ": Found href attribute <$strippedElement "
1651 . "'$attrib'='$value' in uploaded file.\n" );
1653 return [ 'uploaded-href-attribute-svg', $strippedElement, $attrib, $value ];
1657 # only allow data: targets that should be safe. This prevents vectors like,
1658 # image/svg, text/xml, application/xml, and text/html, which can contain scripts
1659 if ( $stripped == 'href' && strncasecmp( 'data:', $value, 5 ) === 0 ) {
1660 // rfc2397 parameters. This is only slightly slower than (;[\w;]+)*.
1661 // phpcs:ignore Generic.Files.LineLength
1662 $parameters = '(?>;[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+=(?>[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+|"(?>[\0-\x0c\x0e-\x21\x23-\x5b\x5d-\x7f]+|\\\\[\0-\x7f])*"))*(?:;base64)?';
1664 if ( !preg_match( "!^data:\s*image/(gif|jpeg|jpg|png)$parameters,!i", $value ) ) {
1665 wfDebug( __METHOD__
. ": Found href to unwhitelisted data: uri "
1666 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1667 return [ 'uploaded-href-unsafe-target-svg', $strippedElement, $attrib, $value ];
1671 # Change href with animate from (http://html5sec.org/#137).
1672 if ( $stripped === 'attributename'
1673 && $strippedElement === 'animate'
1674 && $this->stripXmlNamespace( $value ) == 'href'
1676 wfDebug( __METHOD__
. ": Found animate that might be changing href using from "
1677 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1679 return [ 'uploaded-animate-svg', $strippedElement, $attrib, $value ];
1682 # use set/animate to add event-handler attribute to parent
1683 if ( ( $strippedElement == 'set' ||
$strippedElement == 'animate' )
1684 && $stripped == 'attributename'
1685 && substr( $value, 0, 2 ) == 'on'
1687 wfDebug( __METHOD__
. ": Found svg setting event-handler attribute with "
1688 . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1690 return [ 'uploaded-setting-event-handler-svg', $strippedElement, $stripped, $value ];
1693 # use set to add href attribute to parent element
1694 if ( $strippedElement == 'set'
1695 && $stripped == 'attributename'
1696 && strpos( $value, 'href' ) !== false
1698 wfDebug( __METHOD__
. ": Found svg setting href attribute '$value' in uploaded file.\n" );
1700 return [ 'uploaded-setting-href-svg' ];
1703 # use set to add a remote / data / script target to an element
1704 if ( $strippedElement == 'set'
1705 && $stripped == 'to'
1706 && preg_match( '!(http|https|data|script):!sim', $value )
1708 wfDebug( __METHOD__
. ": Found svg setting attribute to '$value' in uploaded file.\n" );
1710 return [ 'uploaded-wrong-setting-svg', $value ];
1713 # use handler attribute with remote / data / script
1714 if ( $stripped == 'handler' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1715 wfDebug( __METHOD__
. ": Found svg setting handler with remote/data/script "
1716 . "'$attrib'='$value' in uploaded file.\n" );
1718 return [ 'uploaded-setting-handler-svg', $attrib, $value ];
1721 # use CSS styles to bring in remote code
1722 if ( $stripped == 'style'
1723 && self
::checkCssFragment( Sanitizer
::normalizeCss( $value ) )
1725 wfDebug( __METHOD__
. ": Found svg setting a style with "
1726 . "remote url '$attrib'='$value' in uploaded file.\n" );
1727 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1730 # Several attributes can include css, css character escaping isn't allowed
1731 $cssAttrs = [ 'font', 'clip-path', 'fill', 'filter', 'marker',
1732 'marker-end', 'marker-mid', 'marker-start', 'mask', 'stroke' ];
1733 if ( in_array( $stripped, $cssAttrs )
1734 && self
::checkCssFragment( $value )
1736 wfDebug( __METHOD__
. ": Found svg setting a style with "
1737 . "remote url '$attrib'='$value' in uploaded file.\n" );
1738 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1741 # image filters can pull in url, which could be svg that executes scripts
1742 # Only allow url( "#foo" ). Do not allow url( http://example.com )
1743 if ( $strippedElement == 'image'
1744 && $stripped == 'filter'
1745 && preg_match( '!url\s*\(\s*["\']?[^#]!sim', $value )
1747 wfDebug( __METHOD__
. ": Found image filter with url: "
1748 . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1750 return [ 'uploaded-image-filter-svg', $strippedElement, $stripped, $value ];
1754 return false; // No scripts detected
1758 * Check a block of CSS or CSS fragment for anything that looks like
1759 * it is bringing in remote code.
1760 * @param string $value a string of CSS
1761 * @param bool $propOnly only check css properties (start regex with :)
1762 * @return bool true if the CSS contains an illegal string, false if otherwise
1764 private static function checkCssFragment( $value ) {
1765 # Forbid external stylesheets, for both reliability and to protect viewer's privacy
1766 if ( stripos( $value, '@import' ) !== false ) {
1770 # We allow @font-face to embed fonts with data: urls, so we snip the string
1771 # 'url' out so this case won't match when we check for urls below
1772 $pattern = '!(@font-face\s*{[^}]*src:)url(\("data:;base64,)!im';
1773 $value = preg_replace( $pattern, '$1$2', $value );
1775 # Check for remote and executable CSS. Unlike in Sanitizer::checkCss, the CSS
1776 # properties filter and accelerator don't seem to be useful for xss in SVG files.
1777 # Expression and -o-link don't seem to work either, but filtering them here in case.
1778 # Additionally, we catch remote urls like url("http:..., url('http:..., url(http:...,
1779 # but not local ones such as url("#..., url('#..., url(#....
1780 if ( preg_match( '!expression
1782 | -o-link-source\s*:
1783 | -o-replace\s*:!imx', $value ) ) {
1787 if ( preg_match_all(
1788 "!(\s*(url|image|image-set)\s*\(\s*[\"']?\s*[^#]+.*?\))!sim",
1793 # TODO: redo this in one regex. Until then, url("#whatever") matches the first
1794 foreach ( $matches[1] as $match ) {
1795 if ( !preg_match( "!\s*(url|image|image-set)\s*\(\s*(#|'#|\"#)!im", $match ) ) {
1801 if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ) {
1809 * Divide the element name passed by the xml parser to the callback into URI and prifix.
1810 * @param string $element
1811 * @return array Containing the namespace URI and prefix
1813 private static function splitXmlNamespace( $element ) {
1814 // 'http://www.w3.org/2000/svg:script' -> [ 'http://www.w3.org/2000/svg', 'script' ]
1815 $parts = explode( ':', strtolower( $element ) );
1816 $name = array_pop( $parts );
1817 $ns = implode( ':', $parts );
1819 return [ $ns, $name ];
1823 * @param string $name
1826 private function stripXmlNamespace( $name ) {
1827 // 'http://www.w3.org/2000/svg:script' -> 'script'
1828 $parts = explode( ':', strtolower( $name ) );
1830 return array_pop( $parts );
1834 * Generic wrapper function for a virus scanner program.
1835 * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
1836 * $wgAntivirusRequired may be used to deny upload if the scan fails.
1838 * @param string $file Pathname to the temporary upload file
1839 * @return bool|null|string False if not virus is found, null if the scan fails or is disabled,
1840 * or a string containing feedback from the virus scanner if a virus was found.
1841 * If textual feedback is missing but a virus was found, this function returns true.
1843 public static function detectVirus( $file ) {
1844 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
1846 if ( !$wgAntivirus ) {
1847 wfDebug( __METHOD__
. ": virus scanner disabled\n" );
1852 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
1853 wfDebug( __METHOD__
. ": unknown virus scanner: $wgAntivirus\n" );
1854 $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
1855 [ 'virus-badscanner', $wgAntivirus ] );
1857 return wfMessage( 'virus-unknownscanner' )->text() . " $wgAntivirus";
1860 # look up scanner configuration
1861 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
1862 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
1863 $msgPattern = $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ??
null;
1865 if ( strpos( $command, "%f" ) === false ) {
1866 # simple pattern: append file to scan
1867 $command .= " " . Shell
::escape( $file );
1869 # complex pattern: replace "%f" with file to scan
1870 $command = str_replace( "%f", Shell
::escape( $file ), $command );
1873 wfDebug( __METHOD__
. ": running virus scan: $command \n" );
1875 # execute virus scanner
1878 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1879 # that does not seem to be worth the pain.
1880 # Ask me (Duesentrieb) about it if it's ever needed.
1881 $output = wfShellExecWithStderr( $command, $exitCode );
1883 # map exit code to AV_xxx constants.
1884 $mappedCode = $exitCode;
1885 if ( $exitCodeMap ) {
1886 if ( isset( $exitCodeMap[$exitCode] ) ) {
1887 $mappedCode = $exitCodeMap[$exitCode];
1888 } elseif ( isset( $exitCodeMap["*"] ) ) {
1889 $mappedCode = $exitCodeMap["*"];
1893 /* NB: AV_NO_VIRUS is 0 but AV_SCAN_FAILED is false,
1894 * so we need the strict equalities === and thus can't use a switch here
1896 if ( $mappedCode === AV_SCAN_FAILED
) {
1897 # scan failed (code was mapped to false by $exitCodeMap)
1898 wfDebug( __METHOD__
. ": failed to scan $file (code $exitCode).\n" );
1900 $output = $wgAntivirusRequired
1901 ?
wfMessage( 'virus-scanfailed', [ $exitCode ] )->text()
1903 } elseif ( $mappedCode === AV_SCAN_ABORTED
) {
1904 # scan failed because filetype is unknown (probably imune)
1905 wfDebug( __METHOD__
. ": unsupported file type $file (code $exitCode).\n" );
1907 } elseif ( $mappedCode === AV_NO_VIRUS
) {
1909 wfDebug( __METHOD__
. ": file passed virus scan.\n" );
1912 $output = trim( $output );
1915 $output = true; # if there's no output, return true
1916 } elseif ( $msgPattern ) {
1918 if ( preg_match( $msgPattern, $output, $groups ) && $groups[1] ) {
1919 $output = $groups[1];
1923 wfDebug( __METHOD__
. ": FOUND VIRUS! scanner feedback: $output \n" );
1930 * Check if there's an overwrite conflict and, if so, if restrictions
1931 * forbid this user from performing the upload.
1935 * @return bool|array
1937 private function checkOverwrite( $user ) {
1938 // First check whether the local file can be overwritten
1939 $file = $this->getLocalFile();
1940 $file->load( File
::READ_LATEST
);
1941 if ( $file->exists() ) {
1942 if ( !self
::userCanReUpload( $user, $file ) ) {
1943 return [ 'fileexists-forbidden', $file->getName() ];
1949 /* Check shared conflicts: if the local file does not exist, but
1950 * wfFindFile finds a file, it exists in a shared repository.
1952 $file = wfFindFile( $this->getTitle(), [ 'latest' => true ] );
1953 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1954 return [ 'fileexists-shared-forbidden', $file->getName() ];
1961 * Check if a user is the last uploader
1967 public static function userCanReUpload( User
$user, File
$img ) {
1968 if ( $user->isAllowed( 'reupload' ) ) {
1969 return true; // non-conditional
1970 } elseif ( !$user->isAllowed( 'reupload-own' ) ) {
1974 if ( !( $img instanceof LocalFile
) ) {
1980 return $user->getId() == $img->getUser( 'id' );
1984 * Helper function that does various existence checks for a file.
1985 * The following checks are performed:
1987 * - Article with the same name as the file exists
1988 * - File exists with normalized extension
1989 * - The file looks like a thumbnail and the original exists
1991 * @param File $file The File object to check
1992 * @return mixed False if the file does not exists, else an array
1994 public static function getExistsWarning( $file ) {
1995 if ( $file->exists() ) {
1996 return [ 'warning' => 'exists', 'file' => $file ];
1999 if ( $file->getTitle()->getArticleID() ) {
2000 return [ 'warning' => 'page-exists', 'file' => $file ];
2003 if ( strpos( $file->getName(), '.' ) == false ) {
2004 $partname = $file->getName();
2007 $n = strrpos( $file->getName(), '.' );
2008 $extension = substr( $file->getName(), $n +
1 );
2009 $partname = substr( $file->getName(), 0, $n );
2011 $normalizedExtension = File
::normalizeExtension( $extension );
2013 if ( $normalizedExtension != $extension ) {
2014 // We're not using the normalized form of the extension.
2015 // Normal form is lowercase, using most common of alternate
2016 // extensions (eg 'jpg' rather than 'JPEG').
2018 // Check for another file using the normalized form...
2019 $nt_lc = Title
::makeTitle( NS_FILE
, "{$partname}.{$normalizedExtension}" );
2020 $file_lc = wfLocalFile( $nt_lc );
2022 if ( $file_lc->exists() ) {
2024 'warning' => 'exists-normalized',
2026 'normalizedFile' => $file_lc
2031 // Check for files with the same name but a different extension
2032 $similarFiles = RepoGroup
::singleton()->getLocalRepo()->findFilesByPrefix(
2033 "{$partname}.", 1 );
2034 if ( count( $similarFiles ) ) {
2036 'warning' => 'exists-normalized',
2038 'normalizedFile' => $similarFiles[0],
2042 if ( self
::isThumbName( $file->getName() ) ) {
2043 # Check for filenames like 50px- or 180px-, these are mostly thumbnails
2044 $nt_thb = Title
::newFromText(
2045 substr( $partname, strpos( $partname, '-' ) +
1 ) . '.' . $extension,
2048 $file_thb = wfLocalFile( $nt_thb );
2049 if ( $file_thb->exists() ) {
2051 'warning' => 'thumb',
2053 'thumbFile' => $file_thb
2056 // File does not exist, but we just don't like the name
2058 'warning' => 'thumb-name',
2060 'thumbFile' => $file_thb
2065 foreach ( self
::getFilenamePrefixBlacklist() as $prefix ) {
2066 if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
2068 'warning' => 'bad-prefix',
2079 * Helper function that checks whether the filename looks like a thumbnail
2080 * @param string $filename
2083 public static function isThumbName( $filename ) {
2084 $n = strrpos( $filename, '.' );
2085 $partname = $n ?
substr( $filename, 0, $n ) : $filename;
2088 substr( $partname, 3, 3 ) == 'px-' ||
2089 substr( $partname, 2, 3 ) == 'px-'
2091 preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) );
2095 * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
2097 * @return array List of prefixes
2099 public static function getFilenamePrefixBlacklist() {
2101 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
2102 if ( !$message->isDisabled() ) {
2103 $lines = explode( "\n", $message->plain() );
2104 foreach ( $lines as $line ) {
2105 // Remove comment lines
2106 $comment = substr( trim( $line ), 0, 1 );
2107 if ( $comment == '#' ||
$comment == '' ) {
2110 // Remove additional comments after a prefix
2111 $comment = strpos( $line, '#' );
2112 if ( $comment > 0 ) {
2113 $line = substr( $line, 0, $comment - 1 );
2115 $blacklist[] = trim( $line );
2123 * Gets image info about the file just uploaded.
2125 * Also has the effect of setting metadata to be an 'indexed tag name' in
2126 * returned API result if 'metadata' was requested. Oddly, we have to pass
2127 * the "result" object down just so it can do that with the appropriate
2128 * format, presumably.
2130 * @param ApiResult $result
2131 * @return array Image info
2133 public function getImageInfo( $result ) {
2134 $localFile = $this->getLocalFile();
2135 $stashFile = $this->getStashFile();
2136 // Calling a different API module depending on whether the file was stashed is less than optimal.
2137 // In fact, calling API modules here at all is less than optimal. Maybe it should be refactored.
2139 $imParam = ApiQueryStashImageInfo
::getPropertyNames();
2140 $info = ApiQueryStashImageInfo
::getInfo( $stashFile, array_flip( $imParam ), $result );
2142 $imParam = ApiQueryImageInfo
::getPropertyNames();
2143 $info = ApiQueryImageInfo
::getInfo( $localFile, array_flip( $imParam ), $result );
2150 * @param array $error
2153 public function convertVerifyErrorToStatus( $error ) {
2154 $code = $error['status'];
2155 unset( $code['status'] );
2157 return Status
::newFatal( $this->getVerificationErrorCode( $code ), $error );
2161 * Get the MediaWiki maximum uploaded file size for given type of upload, based on
2164 * @param null|string $forType
2167 public static function getMaxUploadSize( $forType = null ) {
2168 global $wgMaxUploadSize;
2170 if ( is_array( $wgMaxUploadSize ) ) {
2171 if ( !is_null( $forType ) && isset( $wgMaxUploadSize[$forType] ) ) {
2172 return $wgMaxUploadSize[$forType];
2174 return $wgMaxUploadSize['*'];
2177 return intval( $wgMaxUploadSize );
2182 * Get the PHP maximum uploaded file size, based on ini settings. If there is no limit or the
2183 * limit can't be guessed, returns a very large number (PHP_INT_MAX).
2188 public static function getMaxPhpUploadSize() {
2189 $phpMaxFileSize = wfShorthandToInteger(
2190 ini_get( 'upload_max_filesize' ) ?
: ini_get( 'hhvm.server.upload.upload_max_file_size' ),
2193 $phpMaxPostSize = wfShorthandToInteger(
2194 ini_get( 'post_max_size' ) ?
: ini_get( 'hhvm.server.max_post_size' ),
2197 return min( $phpMaxFileSize, $phpMaxPostSize );
2201 * Get the current status of a chunked upload (used for polling)
2203 * The value will be read from cache.
2206 * @param string $statusKey
2207 * @return Status[]|bool
2209 public static function getSessionStatus( User
$user, $statusKey ) {
2210 $cache = MediaWikiServices
::getInstance()->getMainObjectStash();
2211 $key = $cache->makeKey( 'uploadstatus', $user->getId() ?
: md5( $user->getName() ), $statusKey );
2213 return $cache->get( $key );
2217 * Set the current status of a chunked upload (used for polling)
2219 * The value will be set in cache for 1 day
2222 * @param string $statusKey
2223 * @param array|bool $value
2226 public static function setSessionStatus( User
$user, $statusKey, $value ) {
2227 $cache = MediaWikiServices
::getInstance()->getMainObjectStash();
2228 $key = $cache->makeKey( 'uploadstatus', $user->getId() ?
: md5( $user->getName() ), $statusKey );
2230 if ( $value === false ) {
2231 $cache->delete( $key );
2233 $cache->set( $key, $value, $cache::TTL_DAY
);