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
24 use MediaWiki\MediaWikiServices
;
25 use MediaWiki\Shell\Shell
;
26 use MediaWiki\User\UserIdentity
;
29 * @defgroup Upload Upload related
35 * UploadBase and subclasses are the backend of MediaWiki's file uploads.
36 * The frontends are formed by ApiUpload and SpecialUpload.
38 * @author Brion Vibber
39 * @author Bryan Tong Minh
40 * @author Michael Dale
42 abstract class UploadBase
{
43 /** @var string Local file system path to the file to upload (or a local copy) */
45 /** @var TempFSFile|null Wrapper to handle deleting the temp file */
46 protected $tempFileObj;
47 /** @var string|null */
48 protected $mDesiredDestName;
49 /** @var string|null */
51 /** @var string|null */
52 protected $mRemoveTempFile;
53 /** @var string|null */
54 protected $mSourceType;
55 /** @var Title|bool */
56 protected $mTitle = false;
58 protected $mTitleError = 0;
59 /** @var string|null */
60 protected $mFilteredName;
61 /** @var string|null */
62 protected $mFinalExtension;
64 protected $mLocalFile;
65 /** @var UploadStashFile */
66 protected $mStashFile;
69 /** @var array|null */
70 protected $mFileProps;
72 protected $mBlackListedExtensions;
74 protected $mJavaDetected;
75 /** @var string|null */
76 protected $mSVGNSError;
78 protected static $safeXmlEncodings = [
98 const MIN_LENGTH_PARTNAME
= 4;
99 const ILLEGAL_FILENAME
= 5;
100 const OVERWRITE_EXISTING_FILE
= 7; # Not used anymore; handled by verifyTitlePermissions()
101 const FILETYPE_MISSING
= 8;
102 const FILETYPE_BADTYPE
= 9;
103 const VERIFICATION_ERROR
= 10;
104 const HOOK_ABORTED
= 11;
105 const FILE_TOO_LARGE
= 12;
106 const WINDOWS_NONASCII_FILENAME
= 13;
107 const FILENAME_TOO_LONG
= 14;
113 public function getVerificationErrorCode( $error ) {
115 self
::EMPTY_FILE
=> 'empty-file',
116 self
::FILE_TOO_LARGE
=> 'file-too-large',
117 self
::FILETYPE_MISSING
=> 'filetype-missing',
118 self
::FILETYPE_BADTYPE
=> 'filetype-banned',
119 self
::MIN_LENGTH_PARTNAME
=> 'filename-tooshort',
120 self
::ILLEGAL_FILENAME
=> 'illegal-filename',
121 self
::OVERWRITE_EXISTING_FILE
=> 'overwrite',
122 self
::VERIFICATION_ERROR
=> 'verification-error',
123 self
::HOOK_ABORTED
=> 'hookaborted',
124 self
::WINDOWS_NONASCII_FILENAME
=> 'windows-nonascii-filename',
125 self
::FILENAME_TOO_LONG
=> 'filename-toolong',
127 return $code_to_status[$error] ??
'unknown-error';
131 * Returns true if uploads are enabled.
132 * Can be override by subclasses.
135 public static function isEnabled() {
136 global $wgEnableUploads;
138 if ( !$wgEnableUploads ) {
142 # Check php's file_uploads setting
143 return wfIsHHVM() ||
wfIniGetBool( 'file_uploads' );
147 * Returns true if the user can use this upload module or else a string
148 * identifying the missing permission.
149 * Can be overridden by subclasses.
151 * @param UserIdentity $user
152 * @return bool|string
154 public static function isAllowed( UserIdentity
$user ) {
155 $permissionManager = MediaWikiServices
::getInstance()->getPermissionManager();
156 foreach ( [ 'upload', 'edit' ] as $permission ) {
157 if ( !$permissionManager->userHasRight( $user, $permission ) ) {
166 * Returns true if the user has surpassed the upload rate limit, false otherwise.
171 public static function isThrottled( $user ) {
172 return $user->pingLimiter( 'upload' );
175 // Upload handlers. Should probably just be a global.
176 private static $uploadHandlers = [ 'Stash', 'File', 'Url' ];
179 * Create a form of UploadBase depending on wpSourceType and initializes it
181 * @param WebRequest &$request
182 * @param string|null $type
183 * @return null|UploadBase
185 public static function createFromRequest( &$request, $type = null ) {
186 $type = $type ?
: $request->getVal( 'wpSourceType', 'File' );
192 // Get the upload class
193 $type = ucfirst( $type );
195 // Give hooks the chance to handle this request
197 Hooks
::run( 'UploadCreateFromRequest', [ $type, &$className ] );
198 if ( is_null( $className ) ) {
199 $className = 'UploadFrom' . $type;
200 wfDebug( __METHOD__
. ": class name: $className\n" );
201 if ( !in_array( $type, self
::$uploadHandlers ) ) {
206 // Check whether this upload class is enabled
207 if ( !call_user_func( [ $className, 'isEnabled' ] ) ) {
211 // Check whether the request is valid
212 if ( !call_user_func( [ $className, 'isValidRequest' ], $request ) ) {
216 /** @var UploadBase $handler */
217 $handler = new $className;
219 $handler->initializeFromRequest( $request );
225 * Check whether a request if valid for this handler
226 * @param WebRequest $request
229 public static function isValidRequest( $request ) {
233 public function __construct() {
237 * Returns the upload type. Should be overridden by child classes
242 public function getSourceType() {
247 * Initialize the path information
248 * @param string $name The desired destination name
249 * @param string $tempPath The temporary path
250 * @param int $fileSize The file size
251 * @param bool $removeTempFile (false) remove the temporary file?
252 * @throws MWException
254 public function initializePathInfo( $name, $tempPath, $fileSize, $removeTempFile = false ) {
255 $this->mDesiredDestName
= $name;
256 if ( FileBackend
::isStoragePath( $tempPath ) ) {
257 throw new MWException( __METHOD__
. " given storage path `$tempPath`." );
260 $this->setTempFile( $tempPath, $fileSize );
261 $this->mRemoveTempFile
= $removeTempFile;
265 * Initialize from a WebRequest. Override this in a subclass.
267 * @param WebRequest &$request
269 abstract public function initializeFromRequest( &$request );
272 * @param string $tempPath File system path to temporary file containing the upload
273 * @param int|null $fileSize
275 protected function setTempFile( $tempPath, $fileSize = null ) {
276 $this->mTempPath
= $tempPath;
277 $this->mFileSize
= $fileSize ?
: null;
278 if ( strlen( $this->mTempPath
) && file_exists( $this->mTempPath
) ) {
279 $this->tempFileObj
= new TempFSFile( $this->mTempPath
);
281 $this->mFileSize
= filesize( $this->mTempPath
);
284 $this->tempFileObj
= null;
289 * Fetch the file. Usually a no-op
292 public function fetchFile() {
293 return Status
::newGood();
297 * Return true if the file is empty
300 public function isEmptyFile() {
301 return empty( $this->mFileSize
);
305 * Return the file size
308 public function getFileSize() {
309 return $this->mFileSize
;
313 * Get the base 36 SHA1 of the file
316 public function getTempFileSha1Base36() {
317 return FSFile
::getSha1Base36FromPath( $this->mTempPath
);
321 * @param string $srcPath The source path
322 * @return string|bool The real path if it was a virtual URL Returns false on failure
324 public function getRealPath( $srcPath ) {
325 $repo = RepoGroup
::singleton()->getLocalRepo();
326 if ( FileRepo
::isVirtualUrl( $srcPath ) ) {
327 /** @todo Just make uploads work with storage paths UploadFromStash
328 * loads files via virtual URLs.
330 $tmpFile = $repo->getLocalCopy( $srcPath );
332 $tmpFile->bind( $this ); // keep alive with $this
334 $path = $tmpFile ?
$tmpFile->getPath() : false;
343 * Verify whether the upload is sane.
344 * @return mixed Const self::OK or else an array with error information
346 public function verifyUpload() {
348 * If there was no filename or a zero size given, give up quick.
350 if ( $this->isEmptyFile() ) {
351 return [ 'status' => self
::EMPTY_FILE
];
355 * Honor $wgMaxUploadSize
357 $maxSize = self
::getMaxUploadSize( $this->getSourceType() );
358 if ( $this->mFileSize
> $maxSize ) {
360 'status' => self
::FILE_TOO_LARGE
,
366 * Look at the contents of the file; if we can recognize the
367 * type but it's corrupt or data of the wrong type, we should
368 * probably not accept it.
370 $verification = $this->verifyFile();
371 if ( $verification !== true ) {
373 'status' => self
::VERIFICATION_ERROR
,
374 'details' => $verification
379 * Make sure this file can be created
381 $result = $this->validateName();
382 if ( $result !== true ) {
386 return [ 'status' => self
::OK
];
390 * Verify that the name is valid and, if necessary, that we can overwrite
392 * @return mixed True if valid, otherwise and array with 'status'
395 public function validateName() {
396 $nt = $this->getTitle();
397 if ( is_null( $nt ) ) {
398 $result = [ 'status' => $this->mTitleError
];
399 if ( $this->mTitleError
== self
::ILLEGAL_FILENAME
) {
400 $result['filtered'] = $this->mFilteredName
;
402 if ( $this->mTitleError
== self
::FILETYPE_BADTYPE
) {
403 $result['finalExt'] = $this->mFinalExtension
;
404 if ( count( $this->mBlackListedExtensions
) ) {
405 $result['blacklistedExt'] = $this->mBlackListedExtensions
;
411 $this->mDestName
= $this->getLocalFile()->getName();
417 * Verify the MIME type.
419 * @note Only checks that it is not an evil MIME. The "does it have
420 * correct extension given its MIME type?" check is in verifyFile.
421 * in `verifyFile()` that MIME type and file extension correlate.
422 * @param string $mime Representing the MIME
423 * @return mixed True if the file is verified, an array otherwise
425 protected function verifyMimeType( $mime ) {
426 global $wgVerifyMimeType, $wgVerifyMimeTypeIE;
427 if ( $wgVerifyMimeType ) {
428 wfDebug( "mime: <$mime> extension: <{$this->mFinalExtension}>\n" );
429 global $wgMimeTypeBlacklist;
430 if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
431 return [ 'filetype-badmime', $mime ];
434 if ( $wgVerifyMimeTypeIE ) {
435 # Check what Internet Explorer would detect
436 $fp = fopen( $this->mTempPath
, 'rb' );
437 $chunk = fread( $fp, 256 );
440 $magic = MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer();
441 $extMime = $magic->guessTypesForExtension( $this->mFinalExtension
);
442 $ieTypes = $magic->getIEMimeTypes( $this->mTempPath
, $chunk, $extMime );
443 foreach ( $ieTypes as $ieType ) {
444 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
445 return [ 'filetype-bad-ie-mime', $ieType ];
455 * Verifies that it's ok to include the uploaded file
457 * @return mixed True of the file is verified, array otherwise.
459 protected function verifyFile() {
460 global $wgVerifyMimeType, $wgDisableUploadScriptChecks;
462 $status = $this->verifyPartialFile();
463 if ( $status !== true ) {
467 $mwProps = new MWFileProps( MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer() );
468 $this->mFileProps
= $mwProps->getPropsFromPath( $this->mTempPath
, $this->mFinalExtension
);
469 $mime = $this->mFileProps
['mime'];
471 if ( $wgVerifyMimeType ) {
472 # XXX: Missing extension will be caught by validateName() via getTitle()
473 if ( $this->mFinalExtension
!= '' && !$this->verifyExtension( $mime, $this->mFinalExtension
) ) {
474 return [ 'filetype-mime-mismatch', $this->mFinalExtension
, $mime ];
478 # check for htmlish code and javascript
479 if ( !$wgDisableUploadScriptChecks ) {
480 if ( $this->mFinalExtension
== 'svg' ||
$mime == 'image/svg+xml' ) {
481 $svgStatus = $this->detectScriptInSvg( $this->mTempPath
, false );
482 if ( $svgStatus !== false ) {
488 $handler = MediaHandler
::getHandler( $mime );
490 $handlerStatus = $handler->verifyUpload( $this->mTempPath
);
491 if ( !$handlerStatus->isOK() ) {
492 $errors = $handlerStatus->getErrorsArray();
494 return reset( $errors );
499 Hooks
::run( 'UploadVerifyFile', [ $this, $mime, &$error ] );
500 if ( $error !== true ) {
501 if ( !is_array( $error ) ) {
507 wfDebug( __METHOD__
. ": all clear; passing.\n" );
513 * A verification routine suitable for partial files
515 * Runs the blacklist checks, but not any checks that may
516 * assume the entire file is present.
518 * @return mixed True for valid or array with error message key.
520 protected function verifyPartialFile() {
521 global $wgAllowJavaUploads, $wgDisableUploadScriptChecks;
523 # getTitle() sets some internal parameters like $this->mFinalExtension
526 $mwProps = new MWFileProps( MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer() );
527 $this->mFileProps
= $mwProps->getPropsFromPath( $this->mTempPath
, $this->mFinalExtension
);
529 # check MIME type, if desired
530 $mime = $this->mFileProps
['file-mime'];
531 $status = $this->verifyMimeType( $mime );
532 if ( $status !== true ) {
536 # check for htmlish code and javascript
537 if ( !$wgDisableUploadScriptChecks ) {
538 if ( self
::detectScript( $this->mTempPath
, $mime, $this->mFinalExtension
) ) {
539 return [ 'uploadscripted' ];
541 if ( $this->mFinalExtension
== 'svg' ||
$mime == 'image/svg+xml' ) {
542 $svgStatus = $this->detectScriptInSvg( $this->mTempPath
, true );
543 if ( $svgStatus !== false ) {
549 # Check for Java applets, which if uploaded can bypass cross-site
551 if ( !$wgAllowJavaUploads ) {
552 $this->mJavaDetected
= false;
553 $zipStatus = ZipDirectoryReader
::read( $this->mTempPath
,
554 [ $this, 'zipEntryCallback' ] );
555 if ( !$zipStatus->isOK() ) {
556 $errors = $zipStatus->getErrorsArray();
557 $error = reset( $errors );
558 if ( $error[0] !== 'zip-wrong-format' ) {
562 if ( $this->mJavaDetected
) {
563 return [ 'uploadjava' ];
567 # Scan the uploaded file for viruses
568 $virus = $this->detectVirus( $this->mTempPath
);
570 return [ 'uploadvirus', $virus ];
577 * Callback for ZipDirectoryReader to detect Java class files.
579 * @param array $entry
581 public function zipEntryCallback( $entry ) {
582 $names = [ $entry['name'] ];
584 // If there is a null character, cut off the name at it, because JDK's
585 // ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name
586 // were constructed which had ".class\0" followed by a string chosen to
587 // make the hash collide with the truncated name, that file could be
588 // returned in response to a request for the .class file.
589 $nullPos = strpos( $entry['name'], "\000" );
590 if ( $nullPos !== false ) {
591 $names[] = substr( $entry['name'], 0, $nullPos );
594 // If there is a trailing slash in the file name, we have to strip it,
595 // because that's what ZIP_GetEntry() does.
596 if ( preg_grep( '!\.class/?$!', $names ) ) {
597 $this->mJavaDetected
= true;
602 * Alias for verifyTitlePermissions. The function was originally
603 * 'verifyPermissions', but that suggests it's checking the user, when it's
604 * really checking the title + user combination.
606 * @param User $user User object to verify the permissions against
607 * @return mixed An array as returned by getUserPermissionsErrors or true
608 * in case the user has proper permissions.
610 public function verifyPermissions( $user ) {
611 return $this->verifyTitlePermissions( $user );
615 * Check whether the user can edit, upload and create the image. This
616 * checks only against the current title; if it returns errors, it may
617 * very well be that another title will not give errors. Therefore
618 * isAllowed() should be called as well for generic is-user-blocked or
619 * can-user-upload checking.
621 * @param User $user User object to verify the permissions against
622 * @return mixed An array as returned by getUserPermissionsErrors or true
623 * in case the user has proper permissions.
625 public function verifyTitlePermissions( $user ) {
627 * If the image is protected, non-sysop users won't be able
628 * to modify it by uploading a new revision.
630 $nt = $this->getTitle();
631 if ( is_null( $nt ) ) {
634 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
635 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
636 if ( !$nt->exists() ) {
637 $permErrorsCreate = $nt->getUserPermissionsErrors( 'create', $user );
639 $permErrorsCreate = [];
641 if ( $permErrors ||
$permErrorsUpload ||
$permErrorsCreate ) {
642 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
643 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
648 $overwriteError = $this->checkOverwrite( $user );
649 if ( $overwriteError !== true ) {
650 return [ $overwriteError ];
657 * Check for non fatal problems with the file.
659 * This should not assume that mTempPath is set.
661 * @return mixed[] Array of warnings
663 public function checkWarnings() {
666 $localFile = $this->getLocalFile();
667 $localFile->load( File
::READ_LATEST
);
668 $filename = $localFile->getName();
669 $hash = $this->getTempFileSha1Base36();
671 $badFileName = $this->checkBadFileName( $filename, $this->mDesiredDestName
);
672 if ( $badFileName !== null ) {
673 $warnings['badfilename'] = $badFileName;
676 $unwantedFileExtensionDetails = $this->checkUnwantedFileExtensions( $this->mFinalExtension
);
677 if ( $unwantedFileExtensionDetails !== null ) {
678 $warnings['filetype-unwanted-type'] = $unwantedFileExtensionDetails;
681 $fileSizeWarnings = $this->checkFileSize( $this->mFileSize
);
682 if ( $fileSizeWarnings ) {
683 $warnings = array_merge( $warnings, $fileSizeWarnings );
686 $localFileExistsWarnings = $this->checkLocalFileExists( $localFile, $hash );
687 if ( $localFileExistsWarnings ) {
688 $warnings = array_merge( $warnings, $localFileExistsWarnings );
691 if ( $this->checkLocalFileWasDeleted( $localFile ) ) {
692 $warnings['was-deleted'] = $filename;
695 // If a file with the same name exists locally then the local file has already been tested
696 // for duplication of content
697 $ignoreLocalDupes = isset( $warnings[ 'exists '] );
698 $dupes = $this->checkAgainstExistingDupes( $hash, $ignoreLocalDupes );
700 $warnings['duplicate'] = $dupes;
703 $archivedDupes = $this->checkAgainstArchiveDupes( $hash );
704 if ( $archivedDupes !== null ) {
705 $warnings['duplicate-archive'] = $archivedDupes;
712 * Convert the warnings array returned by checkWarnings() to something that
713 * can be serialized. File objects will be converted to an associative array
714 * with the following keys:
716 * - fileName: The name of the file
717 * - timestamp: The upload timestamp
719 * @param mixed[] $warnings
722 public static function makeWarningsSerializable( $warnings ) {
723 array_walk_recursive( $warnings, function ( &$param, $key ) {
724 if ( $param instanceof File
) {
726 'fileName' => $param->getName(),
727 'timestamp' => $param->getTimestamp()
729 } elseif ( is_object( $param ) ) {
730 throw new InvalidArgumentException(
731 'UploadBase::makeWarningsSerializable: ' .
732 'Unexpected object of class ' . get_class( $param ) );
739 * Check whether the resulting filename is different from the desired one,
740 * but ignore things like ucfirst() and spaces/underscore things
742 * @param string $filename
743 * @param string $desiredFileName
745 * @return string|null String that was determined to be bad or null if the filename is okay
747 private function checkBadFileName( $filename, $desiredFileName ) {
748 $comparableName = str_replace( ' ', '_', $desiredFileName );
749 $comparableName = Title
::capitalize( $comparableName, NS_FILE
);
751 if ( $desiredFileName != $filename && $comparableName != $filename ) {
759 * @param string $fileExtension The file extension to check
761 * @return array|null array with the following keys:
762 * 0 => string The final extension being used
763 * 1 => string[] The extensions that are allowed
764 * 2 => int The number of extensions that are allowed.
766 private function checkUnwantedFileExtensions( $fileExtension ) {
767 global $wgCheckFileExtensions, $wgFileExtensions, $wgLang;
769 if ( $wgCheckFileExtensions ) {
770 $extensions = array_unique( $wgFileExtensions );
771 if ( !$this->checkFileExtension( $fileExtension, $extensions ) ) {
774 $wgLang->commaList( $extensions ),
784 * @param int $fileSize
786 * @return array warnings
788 private function checkFileSize( $fileSize ) {
789 global $wgUploadSizeWarning;
793 if ( $wgUploadSizeWarning && ( $fileSize > $wgUploadSizeWarning ) ) {
794 $warnings['large-file'] = [ $wgUploadSizeWarning, $fileSize ];
797 if ( $fileSize == 0 ) {
798 $warnings['empty-file'] = true;
805 * @param LocalFile $localFile
806 * @param string $hash sha1 hash of the file to check
808 * @return array warnings
810 private function checkLocalFileExists( LocalFile
$localFile, $hash ) {
813 $exists = self
::getExistsWarning( $localFile );
814 if ( $exists !== false ) {
815 $warnings['exists'] = $exists;
817 // check if file is an exact duplicate of current file version
818 if ( $hash === $localFile->getSha1() ) {
819 $warnings['no-change'] = $localFile;
822 // check if file is an exact duplicate of older versions of this file
823 $history = $localFile->getHistory();
824 foreach ( $history as $oldFile ) {
825 if ( $hash === $oldFile->getSha1() ) {
826 $warnings['duplicate-version'][] = $oldFile;
834 private function checkLocalFileWasDeleted( LocalFile
$localFile ) {
835 return $localFile->wasDeleted() && !$localFile->exists();
839 * @param string $hash sha1 hash of the file to check
840 * @param bool $ignoreLocalDupes True to ignore local duplicates
842 * @return File[] Duplicate files, if found.
844 private function checkAgainstExistingDupes( $hash, $ignoreLocalDupes ) {
845 $dupes = RepoGroup
::singleton()->findBySha1( $hash );
846 $title = $this->getTitle();
847 foreach ( $dupes as $key => $dupe ) {
849 ( $dupe instanceof LocalFile
) &&
851 $title->equals( $dupe->getTitle() )
853 unset( $dupes[$key] );
861 * @param string $hash sha1 hash of the file to check
863 * @return string|null Name of the dupe or empty string if discovered (depending on visibility)
864 * null if the check discovered no dupes.
866 private function checkAgainstArchiveDupes( $hash ) {
867 $archivedFile = new ArchivedFile( null, 0, '', $hash );
868 if ( $archivedFile->getID() > 0 ) {
869 if ( $archivedFile->userCan( File
::DELETED_FILE
) ) {
870 return $archivedFile->getName();
880 * Really perform the upload. Stores the file in the local repo, watches
881 * if necessary and runs the UploadComplete hook.
883 * @param string $comment
884 * @param string $pageText
885 * @param bool $watch Whether the file page should be added to user's watchlist.
886 * (This doesn't check $user's permissions.)
888 * @param string[] $tags Change tags to add to the log entry and page revision.
889 * (This doesn't check $user's permissions.)
890 * @return Status Indicating the whether the upload succeeded.
892 public function performUpload( $comment, $pageText, $watch, $user, $tags = [] ) {
893 $this->getLocalFile()->load( File
::READ_LATEST
);
894 $props = $this->mFileProps
;
897 Hooks
::run( 'UploadVerifyUpload', [ $this, $user, $props, $comment, $pageText, &$error ] );
899 if ( !is_array( $error ) ) {
902 return Status
::newFatal( ...$error );
905 $status = $this->getLocalFile()->upload(
916 if ( $status->isGood() ) {
918 WatchAction
::doWatch(
919 $this->getLocalFile()->getTitle(),
921 User
::IGNORE_USER_RIGHTS
924 // Avoid PHP 7.1 warning of passing $this by reference
926 Hooks
::run( 'UploadComplete', [ &$uploadBase ] );
928 $this->postProcessUpload();
935 * Perform extra steps after a successful upload.
939 public function postProcessUpload() {
943 * Returns the title of the file to be uploaded. Sets mTitleError in case
944 * the name was illegal.
946 * @return Title|null The title of the file or null in case the name was illegal
948 public function getTitle() {
949 if ( $this->mTitle
!== false ) {
950 return $this->mTitle
;
952 if ( !is_string( $this->mDesiredDestName
) ) {
953 $this->mTitleError
= self
::ILLEGAL_FILENAME
;
954 $this->mTitle
= null;
956 return $this->mTitle
;
958 /* Assume that if a user specified File:Something.jpg, this is an error
959 * and that the namespace prefix needs to be stripped of.
961 $title = Title
::newFromText( $this->mDesiredDestName
);
962 if ( $title && $title->getNamespace() == NS_FILE
) {
963 $this->mFilteredName
= $title->getDBkey();
965 $this->mFilteredName
= $this->mDesiredDestName
;
968 # oi_archive_name is max 255 bytes, which include a timestamp and an
969 # exclamation mark, so restrict file name to 240 bytes.
970 if ( strlen( $this->mFilteredName
) > 240 ) {
971 $this->mTitleError
= self
::FILENAME_TOO_LONG
;
972 $this->mTitle
= null;
974 return $this->mTitle
;
978 * Chop off any directories in the given filename. Then
979 * filter out illegal characters, and try to make a legible name
980 * out of it. We'll strip some silently that Title would die on.
982 $this->mFilteredName
= wfStripIllegalFilenameChars( $this->mFilteredName
);
983 /* Normalize to title form before we do any further processing */
984 $nt = Title
::makeTitleSafe( NS_FILE
, $this->mFilteredName
);
985 if ( is_null( $nt ) ) {
986 $this->mTitleError
= self
::ILLEGAL_FILENAME
;
987 $this->mTitle
= null;
989 return $this->mTitle
;
991 $this->mFilteredName
= $nt->getDBkey();
994 * We'll want to blacklist against *any* 'extension', and use
995 * only the final one for the whitelist.
997 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName
);
1000 $this->mFinalExtension
= trim( end( $ext ) );
1002 $this->mFinalExtension
= '';
1004 # No extension, try guessing one
1005 $magic = MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer();
1006 $mime = $magic->guessMimeType( $this->mTempPath
);
1007 if ( $mime !== 'unknown/unknown' ) {
1008 # Get a space separated list of extensions
1009 $extList = $magic->getExtensionsForType( $mime );
1011 # Set the extension to the canonical extension
1012 $this->mFinalExtension
= strtok( $extList, ' ' );
1014 # Fix up the other variables
1015 $this->mFilteredName
.= ".{$this->mFinalExtension}";
1016 $nt = Title
::makeTitleSafe( NS_FILE
, $this->mFilteredName
);
1017 $ext = [ $this->mFinalExtension
];
1022 /* Don't allow users to override the blacklist (check file extension) */
1023 global $wgCheckFileExtensions, $wgStrictFileExtensions;
1024 global $wgFileExtensions, $wgFileBlacklist;
1026 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
1028 if ( $this->mFinalExtension
== '' ) {
1029 $this->mTitleError
= self
::FILETYPE_MISSING
;
1030 $this->mTitle
= null;
1032 return $this->mTitle
;
1033 } elseif ( $blackListedExtensions ||
1034 ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
1035 !$this->checkFileExtension( $this->mFinalExtension
, $wgFileExtensions ) )
1037 $this->mBlackListedExtensions
= $blackListedExtensions;
1038 $this->mTitleError
= self
::FILETYPE_BADTYPE
;
1039 $this->mTitle
= null;
1041 return $this->mTitle
;
1044 // Windows may be broken with special characters, see T3780
1045 if ( !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() )
1046 && !RepoGroup
::singleton()->getLocalRepo()->backendSupportsUnicodePaths()
1048 $this->mTitleError
= self
::WINDOWS_NONASCII_FILENAME
;
1049 $this->mTitle
= null;
1051 return $this->mTitle
;
1054 # If there was more than one "extension", reassemble the base
1055 # filename to prevent bogus complaints about length
1056 if ( count( $ext ) > 1 ) {
1057 $iterations = count( $ext ) - 1;
1058 for ( $i = 0; $i < $iterations; $i++
) {
1059 $partname .= '.' . $ext[$i];
1063 if ( strlen( $partname ) < 1 ) {
1064 $this->mTitleError
= self
::MIN_LENGTH_PARTNAME
;
1065 $this->mTitle
= null;
1067 return $this->mTitle
;
1070 $this->mTitle
= $nt;
1072 return $this->mTitle
;
1076 * Return the local file and initializes if necessary.
1078 * @return LocalFile|null
1080 public function getLocalFile() {
1081 if ( is_null( $this->mLocalFile
) ) {
1082 $nt = $this->getTitle();
1083 $this->mLocalFile
= is_null( $nt ) ?
null : wfLocalFile( $nt );
1086 return $this->mLocalFile
;
1090 * @return UploadStashFile|null
1092 public function getStashFile() {
1093 return $this->mStashFile
;
1097 * Like stashFile(), but respects extensions' wishes to prevent the stashing. verifyUpload() must
1098 * be called before calling this method (unless $isPartial is true).
1100 * Upload stash exceptions are also caught and converted to an error status.
1104 * @param bool $isPartial Pass `true` if this is a part of a chunked upload (not a complete file).
1105 * @return Status If successful, value is an UploadStashFile instance
1107 public function tryStashFile( User
$user, $isPartial = false ) {
1108 if ( !$isPartial ) {
1109 $error = $this->runUploadStashFileHook( $user );
1111 return Status
::newFatal( ...$error );
1115 $file = $this->doStashFile( $user );
1116 return Status
::newGood( $file );
1117 } catch ( UploadStashException
$e ) {
1118 return Status
::newFatal( 'uploadstash-exception', get_class( $e ), $e->getMessage() );
1124 * @return array|null Error message and parameters, null if there's no error
1126 protected function runUploadStashFileHook( User
$user ) {
1127 $props = $this->mFileProps
;
1129 Hooks
::run( 'UploadStashFile', [ $this, $user, $props, &$error ] );
1130 if ( $error && !is_array( $error ) ) {
1131 $error = [ $error ];
1137 * If the user does not supply all necessary information in the first upload
1138 * form submission (either by accident or by design) then we may want to
1139 * stash the file temporarily, get more information, and publish the file
1142 * This method will stash a file in a temporary directory for later
1143 * processing, and save the necessary descriptive info into the database.
1144 * This method returns the file object, which also has a 'fileKey' property
1145 * which can be passed through a form or API request to find this stashed
1148 * @deprecated since 1.28 Use tryStashFile() instead
1149 * @param User|null $user
1150 * @return UploadStashFile Stashed file
1151 * @throws UploadStashBadPathException
1152 * @throws UploadStashFileException
1153 * @throws UploadStashNotLoggedInException
1155 public function stashFile( User
$user = null ) {
1156 wfDeprecated( __METHOD__
, '1.28' );
1158 return $this->doStashFile( $user );
1162 * Implementation for stashFile() and tryStashFile().
1164 * @param User|null $user
1165 * @return UploadStashFile Stashed file
1167 protected function doStashFile( User
$user = null ) {
1168 $stash = RepoGroup
::singleton()->getLocalRepo()->getUploadStash( $user );
1169 $file = $stash->stashFile( $this->mTempPath
, $this->getSourceType() );
1170 $this->mStashFile
= $file;
1176 * If we've modified the upload file we need to manually remove it
1177 * on exit to clean up.
1179 public function cleanupTempFile() {
1180 if ( $this->mRemoveTempFile
&& $this->tempFileObj
) {
1181 // Delete when all relevant TempFSFile handles go out of scope
1182 wfDebug( __METHOD__
. ": Marked temporary file '{$this->mTempPath}' for removal\n" );
1183 $this->tempFileObj
->autocollect();
1187 public function getTempPath() {
1188 return $this->mTempPath
;
1192 * Split a file into a base name and all dot-delimited 'extensions'
1193 * on the end. Some web server configurations will fall back to
1194 * earlier pseudo-'extensions' to determine type and execute
1195 * scripts, so the blacklist needs to check them all.
1197 * @param string $filename
1198 * @return array [ string, string[] ]
1200 public static function splitExtensions( $filename ) {
1201 $bits = explode( '.', $filename );
1202 $basename = array_shift( $bits );
1204 return [ $basename, $bits ];
1208 * Perform case-insensitive match against a list of file extensions.
1209 * Returns true if the extension is in the list.
1211 * @param string $ext
1212 * @param array $list
1215 public static function checkFileExtension( $ext, $list ) {
1216 return in_array( strtolower( $ext ), $list );
1220 * Perform case-insensitive match against a list of file extensions.
1221 * Returns an array of matching extensions.
1223 * @param string[] $ext
1224 * @param string[] $list
1227 public static function checkFileExtensionList( $ext, $list ) {
1228 return array_intersect( array_map( 'strtolower', $ext ), $list );
1232 * Checks if the MIME type of the uploaded file matches the file extension.
1234 * @param string $mime The MIME type of the uploaded file
1235 * @param string $extension The filename extension that the file is to be served with
1238 public static function verifyExtension( $mime, $extension ) {
1239 $magic = MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer();
1241 if ( !$mime ||
$mime == 'unknown' ||
$mime == 'unknown/unknown' ) {
1242 if ( !$magic->isRecognizableExtension( $extension ) ) {
1243 wfDebug( __METHOD__
. ": passing file with unknown detected mime type; " .
1244 "unrecognized extension '$extension', can't verify\n" );
1248 wfDebug( __METHOD__
. ": rejecting file with unknown detected mime type; " .
1249 "recognized extension '$extension', so probably invalid file\n" );
1255 $match = $magic->isMatchingExtension( $extension, $mime );
1257 if ( $match === null ) {
1258 if ( $magic->getTypesForExtension( $extension ) !== null ) {
1259 wfDebug( __METHOD__
. ": No extension known for $mime, but we know a mime for $extension\n" );
1263 wfDebug( __METHOD__
. ": no file extension known for mime type $mime, passing file\n" );
1267 } elseif ( $match === true ) {
1268 wfDebug( __METHOD__
. ": mime type $mime matches extension $extension, passing file\n" );
1270 /** @todo If it's a bitmap, make sure PHP or ImageMagick resp. can handle it! */
1274 . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
1281 * Heuristic for detecting files that *could* contain JavaScript instructions or
1282 * things that may look like HTML to a browser and are thus
1283 * potentially harmful. The present implementation will produce false
1284 * positives in some situations.
1286 * @param string $file Pathname to the temporary upload file
1287 * @param string $mime The MIME type of the file
1288 * @param string $extension The extension of the file
1289 * @return bool True if the file contains something looking like embedded scripts
1291 public static function detectScript( $file, $mime, $extension ) {
1292 # ugly hack: for text files, always look at the entire file.
1293 # For binary field, just check the first K.
1295 $isText = strpos( $mime, 'text/' ) === 0;
1297 $chunk = file_get_contents( $file );
1299 $fp = fopen( $file, 'rb' );
1300 $chunk = fread( $fp, 1024 );
1304 $chunk = strtolower( $chunk );
1310 # decode from UTF-16 if needed (could be used for obfuscation).
1311 if ( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
1313 } elseif ( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
1319 if ( $enc !== null ) {
1320 $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
1323 $chunk = trim( $chunk );
1325 /** @todo FIXME: Convert from UTF-16 if necessary! */
1326 wfDebug( __METHOD__
. ": checking for embedded scripts and HTML stuff\n" );
1328 # check for HTML doctype
1329 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
1333 // Some browsers will interpret obscure xml encodings as UTF-8, while
1334 // PHP/expat will interpret the given encoding in the xml declaration (T49304)
1335 if ( $extension == 'svg' ||
strpos( $mime, 'image/svg' ) === 0 ) {
1336 if ( self
::checkXMLEncodingMissmatch( $file ) ) {
1341 // Quick check for HTML heuristics in old IE and Safari.
1343 // The exact heuristics IE uses are checked separately via verifyMimeType(), so we
1344 // don't need them all here as it can cause many false positives.
1346 // Check for `<script` and such still to forbid script tags and embedded HTML in SVG:
1350 '<html', # also in safari
1351 '<script', # also in safari
1354 foreach ( $tags as $tag ) {
1355 if ( strpos( $chunk, $tag ) !== false ) {
1356 wfDebug( __METHOD__
. ": found something that may make it be mistaken for html: $tag\n" );
1363 * look for JavaScript
1366 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
1367 $chunk = Sanitizer
::decodeCharReferences( $chunk );
1369 # look for script-types
1370 if ( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
1371 wfDebug( __METHOD__
. ": found script types\n" );
1376 # look for html-style script-urls
1377 if ( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1378 wfDebug( __METHOD__
. ": found html-style script urls\n" );
1383 # look for css-style script-urls
1384 if ( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1385 wfDebug( __METHOD__
. ": found css-style script urls\n" );
1390 wfDebug( __METHOD__
. ": no scripts found\n" );
1396 * Check a whitelist of xml encodings that are known not to be interpreted differently
1397 * by the server's xml parser (expat) and some common browsers.
1399 * @param string $file Pathname to the temporary upload file
1400 * @return bool True if the file contains an encoding that could be misinterpreted
1402 public static function checkXMLEncodingMissmatch( $file ) {
1403 global $wgSVGMetadataCutoff;
1404 $contents = file_get_contents( $file, false, null, 0, $wgSVGMetadataCutoff );
1405 $encodingRegex = '!encoding[ \t\n\r]*=[ \t\n\r]*[\'"](.*?)[\'"]!si';
1407 if ( preg_match( "!<\?xml\b(.*?)\?>!si", $contents, $matches ) ) {
1408 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1409 && !in_array( strtoupper( $encMatch[1] ), self
::$safeXmlEncodings )
1411 wfDebug( __METHOD__
. ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1415 } elseif ( preg_match( "!<\?xml\b!si", $contents ) ) {
1416 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1417 // bytes. There shouldn't be a legitimate reason for this to happen.
1418 wfDebug( __METHOD__
. ": Unmatched XML declaration start\n" );
1421 } elseif ( substr( $contents, 0, 4 ) == "\x4C\x6F\xA7\x94" ) {
1422 // EBCDIC encoded XML
1423 wfDebug( __METHOD__
. ": EBCDIC Encoded XML\n" );
1428 // It's possible the file is encoded with multi-byte encoding, so re-encode attempt to
1429 // detect the encoding in case is specifies an encoding not whitelisted in self::$safeXmlEncodings
1430 $attemptEncodings = [ 'UTF-16', 'UTF-16BE', 'UTF-32', 'UTF-32BE' ];
1431 foreach ( $attemptEncodings as $encoding ) {
1432 Wikimedia\
suppressWarnings();
1433 $str = iconv( $encoding, 'UTF-8', $contents );
1434 Wikimedia\restoreWarnings
();
1435 if ( $str != '' && preg_match( "!<\?xml\b(.*?)\?>!si", $str, $matches ) ) {
1436 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1437 && !in_array( strtoupper( $encMatch[1] ), self
::$safeXmlEncodings )
1439 wfDebug( __METHOD__
. ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1443 } elseif ( $str != '' && preg_match( "!<\?xml\b!si", $str ) ) {
1444 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1445 // bytes. There shouldn't be a legitimate reason for this to happen.
1446 wfDebug( __METHOD__
. ": Unmatched XML declaration start\n" );
1456 * @param string $filename
1457 * @param bool $partial
1458 * @return bool|array
1460 protected function detectScriptInSvg( $filename, $partial ) {
1461 $this->mSVGNSError
= false;
1462 $check = new XmlTypeCheck(
1464 [ $this, 'checkSvgScriptCallback' ],
1467 'processing_instruction_handler' => 'UploadBase::checkSvgPICallback',
1468 'external_dtd_handler' => 'UploadBase::checkSvgExternalDTD',
1471 if ( $check->wellFormed
!== true ) {
1472 // Invalid xml (T60553)
1473 // But only when non-partial (T67724)
1474 return $partial ?
false : [ 'uploadinvalidxml' ];
1475 } elseif ( $check->filterMatch
) {
1476 if ( $this->mSVGNSError
) {
1477 return [ 'uploadscriptednamespace', $this->mSVGNSError
];
1480 return $check->filterMatchType
;
1487 * Callback to filter SVG Processing Instructions.
1488 * @param string $target Processing instruction name
1489 * @param string $data Processing instruction attribute and value
1490 * @return bool|array
1492 public static function checkSvgPICallback( $target, $data ) {
1493 // Don't allow external stylesheets (T59550)
1494 if ( preg_match( '/xml-stylesheet/i', $target ) ) {
1495 return [ 'upload-scripted-pi-callback' ];
1502 * Verify that DTD urls referenced are only the standard dtds
1504 * Browsers seem to ignore external dtds. However just to be on the
1505 * safe side, only allow dtds from the svg standard.
1507 * @param string $type PUBLIC or SYSTEM
1508 * @param string $publicId The well-known public identifier for the dtd
1509 * @param string $systemId The url for the external dtd
1510 * @return bool|array
1512 public static function checkSvgExternalDTD( $type, $publicId, $systemId ) {
1513 // This doesn't include the XHTML+MathML+SVG doctype since we don't
1514 // allow XHTML anyways.
1516 'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd',
1517 'http://www.w3.org/TR/2001/REC-SVG-20010904/DTD/svg10.dtd',
1518 'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-basic.dtd',
1519 'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-tiny.dtd',
1520 // https://phabricator.wikimedia.org/T168856
1521 'http://www.w3.org/TR/2001/PR-SVG-20010719/DTD/svg10.dtd',
1523 if ( $type !== 'PUBLIC'
1524 ||
!in_array( $systemId, $allowedDTDs )
1525 ||
strpos( $publicId, "-//W3C//" ) !== 0
1527 return [ 'upload-scripted-dtd' ];
1533 * @todo Replace this with a whitelist filter!
1534 * @param string $element
1535 * @param array $attribs
1536 * @param string|null $data
1537 * @return bool|array
1539 public function checkSvgScriptCallback( $element, $attribs, $data = null ) {
1540 list( $namespace, $strippedElement ) = $this->splitXmlNamespace( $element );
1542 // We specifically don't include:
1543 // http://www.w3.org/1999/xhtml (T62771)
1544 static $validNamespaces = [
1547 'http://creativecommons.org/ns#',
1548 'http://inkscape.sourceforge.net/dtd/sodipodi-0.dtd',
1549 'http://ns.adobe.com/adobeillustrator/10.0/',
1550 'http://ns.adobe.com/adobesvgviewerextensions/3.0/',
1551 'http://ns.adobe.com/extensibility/1.0/',
1552 'http://ns.adobe.com/flows/1.0/',
1553 'http://ns.adobe.com/illustrator/1.0/',
1554 'http://ns.adobe.com/imagereplacement/1.0/',
1555 'http://ns.adobe.com/pdf/1.3/',
1556 'http://ns.adobe.com/photoshop/1.0/',
1557 'http://ns.adobe.com/saveforweb/1.0/',
1558 'http://ns.adobe.com/variables/1.0/',
1559 'http://ns.adobe.com/xap/1.0/',
1560 'http://ns.adobe.com/xap/1.0/g/',
1561 'http://ns.adobe.com/xap/1.0/g/img/',
1562 'http://ns.adobe.com/xap/1.0/mm/',
1563 'http://ns.adobe.com/xap/1.0/rights/',
1564 'http://ns.adobe.com/xap/1.0/stype/dimensions#',
1565 'http://ns.adobe.com/xap/1.0/stype/font#',
1566 'http://ns.adobe.com/xap/1.0/stype/manifestitem#',
1567 'http://ns.adobe.com/xap/1.0/stype/resourceevent#',
1568 'http://ns.adobe.com/xap/1.0/stype/resourceref#',
1569 'http://ns.adobe.com/xap/1.0/t/pg/',
1570 'http://purl.org/dc/elements/1.1/',
1571 'http://purl.org/dc/elements/1.1',
1572 'http://schemas.microsoft.com/visio/2003/svgextensions/',
1573 'http://sodipodi.sourceforge.net/dtd/sodipodi-0.dtd',
1574 'http://taptrix.com/inkpad/svg_extensions',
1575 'http://web.resource.org/cc/',
1576 'http://www.freesoftware.fsf.org/bkchem/cdml',
1577 'http://www.inkscape.org/namespaces/inkscape',
1578 'http://www.opengis.net/gml',
1579 'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
1580 'http://www.w3.org/2000/svg',
1581 'http://www.w3.org/tr/rec-rdf-syntax/',
1582 'http://www.w3.org/2000/01/rdf-schema#',
1585 // Inkscape mangles namespace definitions created by Adobe Illustrator.
1586 // This is nasty but harmless. (T144827)
1587 $isBuggyInkscape = preg_match( '/^&(#38;)*ns_[a-z_]+;$/', $namespace );
1589 if ( !( $isBuggyInkscape ||
in_array( $namespace, $validNamespaces ) ) ) {
1590 wfDebug( __METHOD__
. ": Non-svg namespace '$namespace' in uploaded file.\n" );
1591 /** @todo Return a status object to a closure in XmlTypeCheck, for MW1.21+ */
1592 $this->mSVGNSError
= $namespace;
1598 * check for elements that can contain javascript
1600 if ( $strippedElement == 'script' ) {
1601 wfDebug( __METHOD__
. ": Found script element '$element' in uploaded file.\n" );
1603 return [ 'uploaded-script-svg', $strippedElement ];
1606 # e.g., <svg xmlns="http://www.w3.org/2000/svg">
1607 # <handler xmlns:ev="http://www.w3.org/2001/xml-events" ev:event="load">alert(1)</handler> </svg>
1608 if ( $strippedElement == 'handler' ) {
1609 wfDebug( __METHOD__
. ": Found scriptable element '$element' in uploaded file.\n" );
1611 return [ 'uploaded-script-svg', $strippedElement ];
1614 # SVG reported in Feb '12 that used xml:stylesheet to generate javascript block
1615 if ( $strippedElement == 'stylesheet' ) {
1616 wfDebug( __METHOD__
. ": Found scriptable element '$element' in uploaded file.\n" );
1618 return [ 'uploaded-script-svg', $strippedElement ];
1621 # Block iframes, in case they pass the namespace check
1622 if ( $strippedElement == 'iframe' ) {
1623 wfDebug( __METHOD__
. ": iframe in uploaded file.\n" );
1625 return [ 'uploaded-script-svg', $strippedElement ];
1629 if ( $strippedElement == 'style'
1630 && self
::checkCssFragment( Sanitizer
::normalizeCss( $data ) )
1632 wfDebug( __METHOD__
. ": hostile css in style element.\n" );
1633 return [ 'uploaded-hostile-svg' ];
1636 foreach ( $attribs as $attrib => $value ) {
1637 $stripped = $this->stripXmlNamespace( $attrib );
1638 $value = strtolower( $value );
1640 if ( substr( $stripped, 0, 2 ) == 'on' ) {
1642 . ": Found event-handler attribute '$attrib'='$value' in uploaded file.\n" );
1644 return [ 'uploaded-event-handler-on-svg', $attrib, $value ];
1647 # Do not allow relative links, or unsafe url schemas.
1648 # For <a> tags, only data:, http: and https: and same-document
1649 # fragment links are allowed. For all other tags, only data:
1650 # and fragment are allowed.
1651 if ( $stripped == 'href'
1653 && strpos( $value, 'data:' ) !== 0
1654 && strpos( $value, '#' ) !== 0
1656 if ( !( $strippedElement === 'a'
1657 && preg_match( '!^https?://!i', $value ) )
1659 wfDebug( __METHOD__
. ": Found href attribute <$strippedElement "
1660 . "'$attrib'='$value' in uploaded file.\n" );
1662 return [ 'uploaded-href-attribute-svg', $strippedElement, $attrib, $value ];
1666 # only allow data: targets that should be safe. This prevents vectors like,
1667 # image/svg, text/xml, application/xml, and text/html, which can contain scripts
1668 if ( $stripped == 'href' && strncasecmp( 'data:', $value, 5 ) === 0 ) {
1669 // rfc2397 parameters. This is only slightly slower than (;[\w;]+)*.
1670 // phpcs:ignore Generic.Files.LineLength
1671 $parameters = '(?>;[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+=(?>[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+|"(?>[\0-\x0c\x0e-\x21\x23-\x5b\x5d-\x7f]+|\\\\[\0-\x7f])*"))*(?:;base64)?';
1673 if ( !preg_match( "!^data:\s*image/(gif|jpeg|jpg|png)$parameters,!i", $value ) ) {
1674 wfDebug( __METHOD__
. ": Found href to unwhitelisted data: uri "
1675 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1676 return [ 'uploaded-href-unsafe-target-svg', $strippedElement, $attrib, $value ];
1680 # Change href with animate from (http://html5sec.org/#137).
1681 if ( $stripped === 'attributename'
1682 && $strippedElement === 'animate'
1683 && $this->stripXmlNamespace( $value ) == 'href'
1685 wfDebug( __METHOD__
. ": Found animate that might be changing href using from "
1686 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1688 return [ 'uploaded-animate-svg', $strippedElement, $attrib, $value ];
1691 # use set/animate to add event-handler attribute to parent
1692 if ( ( $strippedElement == 'set' ||
$strippedElement == 'animate' )
1693 && $stripped == 'attributename'
1694 && substr( $value, 0, 2 ) == 'on'
1696 wfDebug( __METHOD__
. ": Found svg setting event-handler attribute with "
1697 . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1699 return [ 'uploaded-setting-event-handler-svg', $strippedElement, $stripped, $value ];
1702 # use set to add href attribute to parent element
1703 if ( $strippedElement == 'set'
1704 && $stripped == 'attributename'
1705 && strpos( $value, 'href' ) !== false
1707 wfDebug( __METHOD__
. ": Found svg setting href attribute '$value' in uploaded file.\n" );
1709 return [ 'uploaded-setting-href-svg' ];
1712 # use set to add a remote / data / script target to an element
1713 if ( $strippedElement == 'set'
1714 && $stripped == 'to'
1715 && preg_match( '!(http|https|data|script):!sim', $value )
1717 wfDebug( __METHOD__
. ": Found svg setting attribute to '$value' in uploaded file.\n" );
1719 return [ 'uploaded-wrong-setting-svg', $value ];
1722 # use handler attribute with remote / data / script
1723 if ( $stripped == 'handler' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1724 wfDebug( __METHOD__
. ": Found svg setting handler with remote/data/script "
1725 . "'$attrib'='$value' in uploaded file.\n" );
1727 return [ 'uploaded-setting-handler-svg', $attrib, $value ];
1730 # use CSS styles to bring in remote code
1731 if ( $stripped == 'style'
1732 && self
::checkCssFragment( Sanitizer
::normalizeCss( $value ) )
1734 wfDebug( __METHOD__
. ": Found svg setting a style with "
1735 . "remote url '$attrib'='$value' in uploaded file.\n" );
1736 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1739 # Several attributes can include css, css character escaping isn't allowed
1740 $cssAttrs = [ 'font', 'clip-path', 'fill', 'filter', 'marker',
1741 'marker-end', 'marker-mid', 'marker-start', 'mask', 'stroke' ];
1742 if ( in_array( $stripped, $cssAttrs )
1743 && self
::checkCssFragment( $value )
1745 wfDebug( __METHOD__
. ": Found svg setting a style with "
1746 . "remote url '$attrib'='$value' in uploaded file.\n" );
1747 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1750 # image filters can pull in url, which could be svg that executes scripts
1751 # Only allow url( "#foo" ). Do not allow url( http://example.com )
1752 if ( $strippedElement == 'image'
1753 && $stripped == 'filter'
1754 && preg_match( '!url\s*\(\s*["\']?[^#]!sim', $value )
1756 wfDebug( __METHOD__
. ": Found image filter with url: "
1757 . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1759 return [ 'uploaded-image-filter-svg', $strippedElement, $stripped, $value ];
1763 return false; // No scripts detected
1767 * Check a block of CSS or CSS fragment for anything that looks like
1768 * it is bringing in remote code.
1769 * @param string $value a string of CSS
1770 * @return bool true if the CSS contains an illegal string, false if otherwise
1772 private static function checkCssFragment( $value ) {
1773 # Forbid external stylesheets, for both reliability and to protect viewer's privacy
1774 if ( stripos( $value, '@import' ) !== false ) {
1778 # We allow @font-face to embed fonts with data: urls, so we snip the string
1779 # 'url' out so this case won't match when we check for urls below
1780 $pattern = '!(@font-face\s*{[^}]*src:)url(\("data:;base64,)!im';
1781 $value = preg_replace( $pattern, '$1$2', $value );
1783 # Check for remote and executable CSS. Unlike in Sanitizer::checkCss, the CSS
1784 # properties filter and accelerator don't seem to be useful for xss in SVG files.
1785 # Expression and -o-link don't seem to work either, but filtering them here in case.
1786 # Additionally, we catch remote urls like url("http:..., url('http:..., url(http:...,
1787 # but not local ones such as url("#..., url('#..., url(#....
1788 if ( preg_match( '!expression
1790 | -o-link-source\s*:
1791 | -o-replace\s*:!imx', $value ) ) {
1795 if ( preg_match_all(
1796 "!(\s*(url|image|image-set)\s*\(\s*[\"']?\s*[^#]+.*?\))!sim",
1801 # TODO: redo this in one regex. Until then, url("#whatever") matches the first
1802 foreach ( $matches[1] as $match ) {
1803 if ( !preg_match( "!\s*(url|image|image-set)\s*\(\s*(#|'#|\"#)!im", $match ) ) {
1809 if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ) {
1817 * Divide the element name passed by the xml parser to the callback into URI and prifix.
1818 * @param string $element
1819 * @return array Containing the namespace URI and prefix
1821 private static function splitXmlNamespace( $element ) {
1822 // 'http://www.w3.org/2000/svg:script' -> [ 'http://www.w3.org/2000/svg', 'script' ]
1823 $parts = explode( ':', strtolower( $element ) );
1824 $name = array_pop( $parts );
1825 $ns = implode( ':', $parts );
1827 return [ $ns, $name ];
1831 * @param string $name
1834 private function stripXmlNamespace( $name ) {
1835 // 'http://www.w3.org/2000/svg:script' -> 'script'
1836 $parts = explode( ':', strtolower( $name ) );
1838 return array_pop( $parts );
1842 * Generic wrapper function for a virus scanner program.
1843 * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
1844 * $wgAntivirusRequired may be used to deny upload if the scan fails.
1846 * @param string $file Pathname to the temporary upload file
1847 * @return bool|null|string False if not virus is found, null if the scan fails or is disabled,
1848 * or a string containing feedback from the virus scanner if a virus was found.
1849 * If textual feedback is missing but a virus was found, this function returns true.
1851 public static function detectVirus( $file ) {
1852 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
1854 if ( !$wgAntivirus ) {
1855 wfDebug( __METHOD__
. ": virus scanner disabled\n" );
1860 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
1861 wfDebug( __METHOD__
. ": unknown virus scanner: $wgAntivirus\n" );
1862 $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
1863 [ 'virus-badscanner', $wgAntivirus ] );
1865 return wfMessage( 'virus-unknownscanner' )->text() . " $wgAntivirus";
1868 # look up scanner configuration
1869 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
1870 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
1871 $msgPattern = $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ??
null;
1873 if ( strpos( $command, "%f" ) === false ) {
1874 # simple pattern: append file to scan
1875 $command .= " " . Shell
::escape( $file );
1877 # complex pattern: replace "%f" with file to scan
1878 $command = str_replace( "%f", Shell
::escape( $file ), $command );
1881 wfDebug( __METHOD__
. ": running virus scan: $command \n" );
1883 # execute virus scanner
1886 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1887 # that does not seem to be worth the pain.
1888 # Ask me (Duesentrieb) about it if it's ever needed.
1889 $output = wfShellExecWithStderr( $command, $exitCode );
1891 # map exit code to AV_xxx constants.
1892 $mappedCode = $exitCode;
1893 if ( $exitCodeMap ) {
1894 if ( isset( $exitCodeMap[$exitCode] ) ) {
1895 $mappedCode = $exitCodeMap[$exitCode];
1896 } elseif ( isset( $exitCodeMap["*"] ) ) {
1897 $mappedCode = $exitCodeMap["*"];
1901 /* NB: AV_NO_VIRUS is 0 but AV_SCAN_FAILED is false,
1902 * so we need the strict equalities === and thus can't use a switch here
1904 if ( $mappedCode === AV_SCAN_FAILED
) {
1905 # scan failed (code was mapped to false by $exitCodeMap)
1906 wfDebug( __METHOD__
. ": failed to scan $file (code $exitCode).\n" );
1908 $output = $wgAntivirusRequired
1909 ?
wfMessage( 'virus-scanfailed', [ $exitCode ] )->text()
1911 } elseif ( $mappedCode === AV_SCAN_ABORTED
) {
1912 # scan failed because filetype is unknown (probably imune)
1913 wfDebug( __METHOD__
. ": unsupported file type $file (code $exitCode).\n" );
1915 } elseif ( $mappedCode === AV_NO_VIRUS
) {
1917 wfDebug( __METHOD__
. ": file passed virus scan.\n" );
1920 $output = trim( $output );
1923 $output = true; # if there's no output, return true
1924 } elseif ( $msgPattern ) {
1926 if ( preg_match( $msgPattern, $output, $groups ) && $groups[1] ) {
1927 $output = $groups[1];
1931 wfDebug( __METHOD__
. ": FOUND VIRUS! scanner feedback: $output \n" );
1938 * Check if there's an overwrite conflict and, if so, if restrictions
1939 * forbid this user from performing the upload.
1943 * @return bool|array
1945 private function checkOverwrite( $user ) {
1946 // First check whether the local file can be overwritten
1947 $file = $this->getLocalFile();
1948 $file->load( File
::READ_LATEST
);
1949 if ( $file->exists() ) {
1950 if ( !self
::userCanReUpload( $user, $file ) ) {
1951 return [ 'fileexists-forbidden', $file->getName() ];
1957 /* Check shared conflicts: if the local file does not exist, but
1958 * wfFindFile finds a file, it exists in a shared repository.
1960 $file = wfFindFile( $this->getTitle(), [ 'latest' => true ] );
1961 if ( $file && !MediaWikiServices
::getInstance()
1962 ->getPermissionManager()
1963 ->userHasRight( $user, 'reupload-shared' )
1965 return [ 'fileexists-shared-forbidden', $file->getName() ];
1972 * Check if a user is the last uploader
1978 public static function userCanReUpload( User
$user, File
$img ) {
1979 $permissionManager = MediaWikiServices
::getInstance()->getPermissionManager();
1980 if ( $permissionManager->userHasRight( $user, 'reupload' ) ) {
1981 return true; // non-conditional
1982 } elseif ( !$permissionManager->userHasRight( $user, 'reupload-own' ) ) {
1986 if ( !( $img instanceof LocalFile
) ) {
1992 return $user->getId() == $img->getUser( 'id' );
1996 * Helper function that does various existence checks for a file.
1997 * The following checks are performed:
1999 * - Article with the same name as the file exists
2000 * - File exists with normalized extension
2001 * - The file looks like a thumbnail and the original exists
2003 * @param File $file The File object to check
2004 * @return mixed False if the file does not exists, else an array
2006 public static function getExistsWarning( $file ) {
2007 if ( $file->exists() ) {
2008 return [ 'warning' => 'exists', 'file' => $file ];
2011 if ( $file->getTitle()->getArticleID() ) {
2012 return [ 'warning' => 'page-exists', 'file' => $file ];
2015 if ( strpos( $file->getName(), '.' ) == false ) {
2016 $partname = $file->getName();
2019 $n = strrpos( $file->getName(), '.' );
2020 $extension = substr( $file->getName(), $n +
1 );
2021 $partname = substr( $file->getName(), 0, $n );
2023 $normalizedExtension = File
::normalizeExtension( $extension );
2025 if ( $normalizedExtension != $extension ) {
2026 // We're not using the normalized form of the extension.
2027 // Normal form is lowercase, using most common of alternate
2028 // extensions (eg 'jpg' rather than 'JPEG').
2030 // Check for another file using the normalized form...
2031 $nt_lc = Title
::makeTitle( NS_FILE
, "{$partname}.{$normalizedExtension}" );
2032 $file_lc = wfLocalFile( $nt_lc );
2034 if ( $file_lc->exists() ) {
2036 'warning' => 'exists-normalized',
2038 'normalizedFile' => $file_lc
2043 // Check for files with the same name but a different extension
2044 $similarFiles = RepoGroup
::singleton()->getLocalRepo()->findFilesByPrefix(
2045 "{$partname}.", 1 );
2046 if ( count( $similarFiles ) ) {
2048 'warning' => 'exists-normalized',
2050 'normalizedFile' => $similarFiles[0],
2054 if ( self
::isThumbName( $file->getName() ) ) {
2055 # Check for filenames like 50px- or 180px-, these are mostly thumbnails
2056 $nt_thb = Title
::newFromText(
2057 substr( $partname, strpos( $partname, '-' ) +
1 ) . '.' . $extension,
2060 $file_thb = wfLocalFile( $nt_thb );
2061 if ( $file_thb->exists() ) {
2063 'warning' => 'thumb',
2065 'thumbFile' => $file_thb
2068 // File does not exist, but we just don't like the name
2070 'warning' => 'thumb-name',
2072 'thumbFile' => $file_thb
2077 foreach ( self
::getFilenamePrefixBlacklist() as $prefix ) {
2078 if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
2080 'warning' => 'bad-prefix',
2091 * Helper function that checks whether the filename looks like a thumbnail
2092 * @param string $filename
2095 public static function isThumbName( $filename ) {
2096 $n = strrpos( $filename, '.' );
2097 $partname = $n ?
substr( $filename, 0, $n ) : $filename;
2100 substr( $partname, 3, 3 ) == 'px-' ||
2101 substr( $partname, 2, 3 ) == 'px-'
2103 preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) );
2107 * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
2109 * @return array List of prefixes
2111 public static function getFilenamePrefixBlacklist() {
2113 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
2114 if ( !$message->isDisabled() ) {
2115 $lines = explode( "\n", $message->plain() );
2116 foreach ( $lines as $line ) {
2117 // Remove comment lines
2118 $comment = substr( trim( $line ), 0, 1 );
2119 if ( $comment == '#' ||
$comment == '' ) {
2122 // Remove additional comments after a prefix
2123 $comment = strpos( $line, '#' );
2124 if ( $comment > 0 ) {
2125 $line = substr( $line, 0, $comment - 1 );
2127 $blacklist[] = trim( $line );
2135 * Gets image info about the file just uploaded.
2137 * Also has the effect of setting metadata to be an 'indexed tag name' in
2138 * returned API result if 'metadata' was requested. Oddly, we have to pass
2139 * the "result" object down just so it can do that with the appropriate
2140 * format, presumably.
2142 * @param ApiResult $result
2143 * @return array Image info
2145 public function getImageInfo( $result ) {
2146 $localFile = $this->getLocalFile();
2147 $stashFile = $this->getStashFile();
2148 // Calling a different API module depending on whether the file was stashed is less than optimal.
2149 // In fact, calling API modules here at all is less than optimal. Maybe it should be refactored.
2151 $imParam = ApiQueryStashImageInfo
::getPropertyNames();
2152 $info = ApiQueryStashImageInfo
::getInfo( $stashFile, array_flip( $imParam ), $result );
2154 $imParam = ApiQueryImageInfo
::getPropertyNames();
2155 $info = ApiQueryImageInfo
::getInfo( $localFile, array_flip( $imParam ), $result );
2162 * @param array $error
2165 public function convertVerifyErrorToStatus( $error ) {
2166 $code = $error['status'];
2167 unset( $code['status'] );
2169 return Status
::newFatal( $this->getVerificationErrorCode( $code ), $error );
2173 * Get the MediaWiki maximum uploaded file size for given type of upload, based on
2176 * @param null|string $forType
2179 public static function getMaxUploadSize( $forType = null ) {
2180 global $wgMaxUploadSize;
2182 if ( is_array( $wgMaxUploadSize ) ) {
2183 if ( !is_null( $forType ) && isset( $wgMaxUploadSize[$forType] ) ) {
2184 return $wgMaxUploadSize[$forType];
2186 return $wgMaxUploadSize['*'];
2189 return intval( $wgMaxUploadSize );
2194 * Get the PHP maximum uploaded file size, based on ini settings. If there is no limit or the
2195 * limit can't be guessed, returns a very large number (PHP_INT_MAX).
2200 public static function getMaxPhpUploadSize() {
2201 $phpMaxFileSize = wfShorthandToInteger(
2202 ini_get( 'upload_max_filesize' ) ?
: ini_get( 'hhvm.server.upload.upload_max_file_size' ),
2205 $phpMaxPostSize = wfShorthandToInteger(
2206 ini_get( 'post_max_size' ) ?
: ini_get( 'hhvm.server.max_post_size' ),
2209 return min( $phpMaxFileSize, $phpMaxPostSize );
2213 * Get the current status of a chunked upload (used for polling)
2215 * The value will be read from cache.
2218 * @param string $statusKey
2219 * @return Status[]|bool
2221 public static function getSessionStatus( User
$user, $statusKey ) {
2222 $store = self
::getUploadSessionStore();
2223 $key = self
::getUploadSessionKey( $store, $user, $statusKey );
2225 return $store->get( $key );
2229 * Set the current status of a chunked upload (used for polling)
2231 * The value will be set in cache for 1 day
2233 * Avoid triggering this method on HTTP GET/HEAD requests
2236 * @param string $statusKey
2237 * @param array|bool $value
2240 public static function setSessionStatus( User
$user, $statusKey, $value ) {
2241 $store = self
::getUploadSessionStore();
2242 $key = self
::getUploadSessionKey( $store, $user, $statusKey );
2244 if ( $value === false ) {
2245 $store->delete( $key );
2247 $store->set( $key, $value, $store::TTL_DAY
);
2252 * @param BagOStuff $store
2254 * @param string $statusKey
2257 private static function getUploadSessionKey( BagOStuff
$store, User
$user, $statusKey ) {
2258 return $store->makeKey(
2260 $user->getId() ?
: md5( $user->getName() ),
2268 private static function getUploadSessionStore() {
2269 return ObjectCache
::getInstance( 'db-replicated' );