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\Shell\Shell
;
26 * @defgroup Upload Upload related
32 * UploadBase and subclasses are the backend of MediaWiki's file uploads.
33 * The frontends are formed by ApiUpload and SpecialUpload.
35 * @author Brion Vibber
36 * @author Bryan Tong Minh
37 * @author Michael Dale
39 abstract class UploadBase
{
40 /** @var string Local file system path to the file to upload (or a local copy) */
42 /** @var TempFSFile|null Wrapper to handle deleting the temp file */
43 protected $tempFileObj;
44 /** @var string|null */
45 protected $mDesiredDestName;
46 /** @var string|null */
48 /** @var string|null */
49 protected $mRemoveTempFile;
50 /** @var string|null */
51 protected $mSourceType;
52 /** @var Title|bool */
53 protected $mTitle = false;
55 protected $mTitleError = 0;
56 /** @var string|null */
57 protected $mFilteredName;
58 /** @var string|null */
59 protected $mFinalExtension;
61 protected $mLocalFile;
62 /** @var UploadStashFile */
63 protected $mStashFile;
66 /** @var array|null */
67 protected $mFileProps;
69 protected $mBlackListedExtensions;
71 protected $mJavaDetected;
72 /** @var string|null */
73 protected $mSVGNSError;
75 protected static $safeXmlEncodings = [
95 const MIN_LENGTH_PARTNAME
= 4;
96 const ILLEGAL_FILENAME
= 5;
97 const OVERWRITE_EXISTING_FILE
= 7; # Not used anymore; handled by verifyTitlePermissions()
98 const FILETYPE_MISSING
= 8;
99 const FILETYPE_BADTYPE
= 9;
100 const VERIFICATION_ERROR
= 10;
101 const HOOK_ABORTED
= 11;
102 const FILE_TOO_LARGE
= 12;
103 const WINDOWS_NONASCII_FILENAME
= 13;
104 const FILENAME_TOO_LONG
= 14;
110 public function getVerificationErrorCode( $error ) {
112 self
::EMPTY_FILE
=> 'empty-file',
113 self
::FILE_TOO_LARGE
=> 'file-too-large',
114 self
::FILETYPE_MISSING
=> 'filetype-missing',
115 self
::FILETYPE_BADTYPE
=> 'filetype-banned',
116 self
::MIN_LENGTH_PARTNAME
=> 'filename-tooshort',
117 self
::ILLEGAL_FILENAME
=> 'illegal-filename',
118 self
::OVERWRITE_EXISTING_FILE
=> 'overwrite',
119 self
::VERIFICATION_ERROR
=> 'verification-error',
120 self
::HOOK_ABORTED
=> 'hookaborted',
121 self
::WINDOWS_NONASCII_FILENAME
=> 'windows-nonascii-filename',
122 self
::FILENAME_TOO_LONG
=> 'filename-toolong',
124 return $code_to_status[$error] ??
'unknown-error';
128 * Returns true if uploads are enabled.
129 * Can be override by subclasses.
132 public static function isEnabled() {
133 global $wgEnableUploads;
135 if ( !$wgEnableUploads ) {
139 # Check php's file_uploads setting
140 return wfIsHHVM() ||
wfIniGetBool( 'file_uploads' );
144 * Returns true if the user can use this upload module or else a string
145 * identifying the missing permission.
146 * Can be overridden by subclasses.
149 * @return bool|string
151 public static function isAllowed( $user ) {
152 foreach ( [ 'upload', 'edit' ] as $permission ) {
153 if ( !$user->isAllowed( $permission ) ) {
162 * Returns true if the user has surpassed the upload rate limit, false otherwise.
167 public static function isThrottled( $user ) {
168 return $user->pingLimiter( 'upload' );
171 // Upload handlers. Should probably just be a global.
172 private static $uploadHandlers = [ 'Stash', 'File', 'Url' ];
175 * Create a form of UploadBase depending on wpSourceType and initializes it
177 * @param WebRequest &$request
178 * @param string|null $type
179 * @return null|UploadBase
181 public static function createFromRequest( &$request, $type = null ) {
182 $type = $type ?
: $request->getVal( 'wpSourceType', 'File' );
188 // Get the upload class
189 $type = ucfirst( $type );
191 // Give hooks the chance to handle this request
193 Hooks
::run( 'UploadCreateFromRequest', [ $type, &$className ] );
194 if ( is_null( $className ) ) {
195 $className = 'UploadFrom' . $type;
196 wfDebug( __METHOD__
. ": class name: $className\n" );
197 if ( !in_array( $type, self
::$uploadHandlers ) ) {
202 // Check whether this upload class is enabled
203 if ( !call_user_func( [ $className, 'isEnabled' ] ) ) {
207 // Check whether the request is valid
208 if ( !call_user_func( [ $className, 'isValidRequest' ], $request ) ) {
212 /** @var UploadBase $handler */
213 $handler = new $className;
215 $handler->initializeFromRequest( $request );
221 * Check whether a request if valid for this handler
222 * @param WebRequest $request
225 public static function isValidRequest( $request ) {
229 public function __construct() {
233 * Returns the upload type. Should be overridden by child classes
238 public function getSourceType() {
243 * Initialize the path information
244 * @param string $name The desired destination name
245 * @param string $tempPath The temporary path
246 * @param int $fileSize The file size
247 * @param bool $removeTempFile (false) remove the temporary file?
248 * @throws MWException
250 public function initializePathInfo( $name, $tempPath, $fileSize, $removeTempFile = false ) {
251 $this->mDesiredDestName
= $name;
252 if ( FileBackend
::isStoragePath( $tempPath ) ) {
253 throw new MWException( __METHOD__
. " given storage path `$tempPath`." );
256 $this->setTempFile( $tempPath, $fileSize );
257 $this->mRemoveTempFile
= $removeTempFile;
261 * Initialize from a WebRequest. Override this in a subclass.
263 * @param WebRequest &$request
265 abstract public function initializeFromRequest( &$request );
268 * @param string $tempPath File system path to temporary file containing the upload
269 * @param int|null $fileSize
271 protected function setTempFile( $tempPath, $fileSize = null ) {
272 $this->mTempPath
= $tempPath;
273 $this->mFileSize
= $fileSize ?
: null;
274 if ( strlen( $this->mTempPath
) && file_exists( $this->mTempPath
) ) {
275 $this->tempFileObj
= new TempFSFile( $this->mTempPath
);
277 $this->mFileSize
= filesize( $this->mTempPath
);
280 $this->tempFileObj
= null;
285 * Fetch the file. Usually a no-op
288 public function fetchFile() {
289 return Status
::newGood();
293 * Return true if the file is empty
296 public function isEmptyFile() {
297 return empty( $this->mFileSize
);
301 * Return the file size
304 public function getFileSize() {
305 return $this->mFileSize
;
309 * Get the base 36 SHA1 of the file
312 public function getTempFileSha1Base36() {
313 return FSFile
::getSha1Base36FromPath( $this->mTempPath
);
317 * @param string $srcPath The source path
318 * @return string|bool The real path if it was a virtual URL Returns false on failure
320 public function getRealPath( $srcPath ) {
321 $repo = RepoGroup
::singleton()->getLocalRepo();
322 if ( FileRepo
::isVirtualUrl( $srcPath ) ) {
323 /** @todo Just make uploads work with storage paths UploadFromStash
324 * loads files via virtual URLs.
326 $tmpFile = $repo->getLocalCopy( $srcPath );
328 $tmpFile->bind( $this ); // keep alive with $this
330 $path = $tmpFile ?
$tmpFile->getPath() : false;
339 * Verify whether the upload is sane.
340 * @return mixed Const self::OK or else an array with error information
342 public function verifyUpload() {
344 * If there was no filename or a zero size given, give up quick.
346 if ( $this->isEmptyFile() ) {
347 return [ 'status' => self
::EMPTY_FILE
];
351 * Honor $wgMaxUploadSize
353 $maxSize = self
::getMaxUploadSize( $this->getSourceType() );
354 if ( $this->mFileSize
> $maxSize ) {
356 'status' => self
::FILE_TOO_LARGE
,
362 * Look at the contents of the file; if we can recognize the
363 * type but it's corrupt or data of the wrong type, we should
364 * probably not accept it.
366 $verification = $this->verifyFile();
367 if ( $verification !== true ) {
369 'status' => self
::VERIFICATION_ERROR
,
370 'details' => $verification
375 * Make sure this file can be created
377 $result = $this->validateName();
378 if ( $result !== true ) {
382 return [ 'status' => self
::OK
];
386 * Verify that the name is valid and, if necessary, that we can overwrite
388 * @return mixed True if valid, otherwise and array with 'status'
391 public function validateName() {
392 $nt = $this->getTitle();
393 if ( is_null( $nt ) ) {
394 $result = [ 'status' => $this->mTitleError
];
395 if ( $this->mTitleError
== self
::ILLEGAL_FILENAME
) {
396 $result['filtered'] = $this->mFilteredName
;
398 if ( $this->mTitleError
== self
::FILETYPE_BADTYPE
) {
399 $result['finalExt'] = $this->mFinalExtension
;
400 if ( count( $this->mBlackListedExtensions
) ) {
401 $result['blacklistedExt'] = $this->mBlackListedExtensions
;
407 $this->mDestName
= $this->getLocalFile()->getName();
413 * Verify the MIME type.
415 * @note Only checks that it is not an evil MIME. The "does it have
416 * correct extension given its MIME type?" check is in verifyFile.
417 * in `verifyFile()` that MIME type and file extension correlate.
418 * @param string $mime Representing the MIME
419 * @return mixed True if the file is verified, an array otherwise
421 protected function verifyMimeType( $mime ) {
422 global $wgVerifyMimeType, $wgVerifyMimeTypeIE;
423 if ( $wgVerifyMimeType ) {
424 wfDebug( "mime: <$mime> extension: <{$this->mFinalExtension}>\n" );
425 global $wgMimeTypeBlacklist;
426 if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
427 return [ 'filetype-badmime', $mime ];
430 if ( $wgVerifyMimeTypeIE ) {
431 # Check what Internet Explorer would detect
432 $fp = fopen( $this->mTempPath
, 'rb' );
433 $chunk = fread( $fp, 256 );
436 $magic = MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer();
437 $extMime = $magic->guessTypesForExtension( $this->mFinalExtension
);
438 $ieTypes = $magic->getIEMimeTypes( $this->mTempPath
, $chunk, $extMime );
439 foreach ( $ieTypes as $ieType ) {
440 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
441 return [ 'filetype-bad-ie-mime', $ieType ];
451 * Verifies that it's ok to include the uploaded file
453 * @return mixed True of the file is verified, array otherwise.
455 protected function verifyFile() {
456 global $wgVerifyMimeType, $wgDisableUploadScriptChecks;
458 $status = $this->verifyPartialFile();
459 if ( $status !== true ) {
463 $mwProps = new MWFileProps( MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer() );
464 $this->mFileProps
= $mwProps->getPropsFromPath( $this->mTempPath
, $this->mFinalExtension
);
465 $mime = $this->mFileProps
['mime'];
467 if ( $wgVerifyMimeType ) {
468 # XXX: Missing extension will be caught by validateName() via getTitle()
469 if ( $this->mFinalExtension
!= '' && !$this->verifyExtension( $mime, $this->mFinalExtension
) ) {
470 return [ 'filetype-mime-mismatch', $this->mFinalExtension
, $mime ];
474 # check for htmlish code and javascript
475 if ( !$wgDisableUploadScriptChecks ) {
476 if ( $this->mFinalExtension
== 'svg' ||
$mime == 'image/svg+xml' ) {
477 $svgStatus = $this->detectScriptInSvg( $this->mTempPath
, false );
478 if ( $svgStatus !== false ) {
484 $handler = MediaHandler
::getHandler( $mime );
486 $handlerStatus = $handler->verifyUpload( $this->mTempPath
);
487 if ( !$handlerStatus->isOK() ) {
488 $errors = $handlerStatus->getErrorsArray();
490 return reset( $errors );
495 Hooks
::run( 'UploadVerifyFile', [ $this, $mime, &$error ] );
496 if ( $error !== true ) {
497 if ( !is_array( $error ) ) {
503 wfDebug( __METHOD__
. ": all clear; passing.\n" );
509 * A verification routine suitable for partial files
511 * Runs the blacklist checks, but not any checks that may
512 * assume the entire file is present.
514 * @return mixed True for valid or array with error message key.
516 protected function verifyPartialFile() {
517 global $wgAllowJavaUploads, $wgDisableUploadScriptChecks;
519 # getTitle() sets some internal parameters like $this->mFinalExtension
522 $mwProps = new MWFileProps( MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer() );
523 $this->mFileProps
= $mwProps->getPropsFromPath( $this->mTempPath
, $this->mFinalExtension
);
525 # check MIME type, if desired
526 $mime = $this->mFileProps
['file-mime'];
527 $status = $this->verifyMimeType( $mime );
528 if ( $status !== true ) {
532 # check for htmlish code and javascript
533 if ( !$wgDisableUploadScriptChecks ) {
534 if ( self
::detectScript( $this->mTempPath
, $mime, $this->mFinalExtension
) ) {
535 return [ 'uploadscripted' ];
537 if ( $this->mFinalExtension
== 'svg' ||
$mime == 'image/svg+xml' ) {
538 $svgStatus = $this->detectScriptInSvg( $this->mTempPath
, true );
539 if ( $svgStatus !== false ) {
545 # Check for Java applets, which if uploaded can bypass cross-site
547 if ( !$wgAllowJavaUploads ) {
548 $this->mJavaDetected
= false;
549 $zipStatus = ZipDirectoryReader
::read( $this->mTempPath
,
550 [ $this, 'zipEntryCallback' ] );
551 if ( !$zipStatus->isOK() ) {
552 $errors = $zipStatus->getErrorsArray();
553 $error = reset( $errors );
554 if ( $error[0] !== 'zip-wrong-format' ) {
558 if ( $this->mJavaDetected
) {
559 return [ 'uploadjava' ];
563 # Scan the uploaded file for viruses
564 $virus = $this->detectVirus( $this->mTempPath
);
566 return [ 'uploadvirus', $virus ];
573 * Callback for ZipDirectoryReader to detect Java class files.
575 * @param array $entry
577 public function zipEntryCallback( $entry ) {
578 $names = [ $entry['name'] ];
580 // If there is a null character, cut off the name at it, because JDK's
581 // ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name
582 // were constructed which had ".class\0" followed by a string chosen to
583 // make the hash collide with the truncated name, that file could be
584 // returned in response to a request for the .class file.
585 $nullPos = strpos( $entry['name'], "\000" );
586 if ( $nullPos !== false ) {
587 $names[] = substr( $entry['name'], 0, $nullPos );
590 // If there is a trailing slash in the file name, we have to strip it,
591 // because that's what ZIP_GetEntry() does.
592 if ( preg_grep( '!\.class/?$!', $names ) ) {
593 $this->mJavaDetected
= true;
598 * Alias for verifyTitlePermissions. The function was originally
599 * 'verifyPermissions', but that suggests it's checking the user, when it's
600 * really checking the title + user combination.
602 * @param User $user User object to verify the permissions against
603 * @return mixed An array as returned by getUserPermissionsErrors or true
604 * in case the user has proper permissions.
606 public function verifyPermissions( $user ) {
607 return $this->verifyTitlePermissions( $user );
611 * Check whether the user can edit, upload and create the image. This
612 * checks only against the current title; if it returns errors, it may
613 * very well be that another title will not give errors. Therefore
614 * isAllowed() should be called as well for generic is-user-blocked or
615 * can-user-upload checking.
617 * @param User $user User object to verify the permissions against
618 * @return mixed An array as returned by getUserPermissionsErrors or true
619 * in case the user has proper permissions.
621 public function verifyTitlePermissions( $user ) {
623 * If the image is protected, non-sysop users won't be able
624 * to modify it by uploading a new revision.
626 $nt = $this->getTitle();
627 if ( is_null( $nt ) ) {
630 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
631 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
632 if ( !$nt->exists() ) {
633 $permErrorsCreate = $nt->getUserPermissionsErrors( 'create', $user );
635 $permErrorsCreate = [];
637 if ( $permErrors ||
$permErrorsUpload ||
$permErrorsCreate ) {
638 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
639 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
644 $overwriteError = $this->checkOverwrite( $user );
645 if ( $overwriteError !== true ) {
646 return [ $overwriteError ];
653 * Check for non fatal problems with the file.
655 * This should not assume that mTempPath is set.
657 * @return mixed[] Array of warnings
659 public function checkWarnings() {
662 $localFile = $this->getLocalFile();
663 $localFile->load( File
::READ_LATEST
);
664 $filename = $localFile->getName();
665 $hash = $this->getTempFileSha1Base36();
667 $badFileName = $this->checkBadFileName( $filename, $this->mDesiredDestName
);
668 if ( $badFileName !== null ) {
669 $warnings['badfilename'] = $badFileName;
672 $unwantedFileExtensionDetails = $this->checkUnwantedFileExtensions( $this->mFinalExtension
);
673 if ( $unwantedFileExtensionDetails !== null ) {
674 $warnings['filetype-unwanted-type'] = $unwantedFileExtensionDetails;
677 $fileSizeWarnings = $this->checkFileSize( $this->mFileSize
);
678 if ( $fileSizeWarnings ) {
679 $warnings = array_merge( $warnings, $fileSizeWarnings );
682 $localFileExistsWarnings = $this->checkLocalFileExists( $localFile, $hash );
683 if ( $localFileExistsWarnings ) {
684 $warnings = array_merge( $warnings, $localFileExistsWarnings );
687 if ( $this->checkLocalFileWasDeleted( $localFile ) ) {
688 $warnings['was-deleted'] = $filename;
691 // If a file with the same name exists locally then the local file has already been tested
692 // for duplication of content
693 $ignoreLocalDupes = isset( $warnings[ 'exists '] );
694 $dupes = $this->checkAgainstExistingDupes( $hash, $ignoreLocalDupes );
696 $warnings['duplicate'] = $dupes;
699 $archivedDupes = $this->checkAgainstArchiveDupes( $hash );
700 if ( $archivedDupes !== null ) {
701 $warnings['duplicate-archive'] = $archivedDupes;
708 * Convert the warnings array returned by checkWarnings() to something that
709 * can be serialized. File objects will be converted to an associative array
710 * with the following keys:
712 * - fileName: The name of the file
713 * - timestamp: The upload timestamp
715 * @param mixed[] $warnings
718 public static function makeWarningsSerializable( $warnings ) {
719 array_walk_recursive( $warnings, function ( &$param, $key ) {
720 if ( $param instanceof File
) {
722 'fileName' => $param->getName(),
723 'timestamp' => $param->getTimestamp()
725 } elseif ( is_object( $param ) ) {
726 throw new InvalidArgumentException(
727 'UploadBase::makeWarningsSerializable: ' .
728 'Unexpected object of class ' . get_class( $param ) );
735 * Check whether the resulting filename is different from the desired one,
736 * but ignore things like ucfirst() and spaces/underscore things
738 * @param string $filename
739 * @param string $desiredFileName
741 * @return string|null String that was determined to be bad or null if the filename is okay
743 private function checkBadFileName( $filename, $desiredFileName ) {
744 $comparableName = str_replace( ' ', '_', $desiredFileName );
745 $comparableName = Title
::capitalize( $comparableName, NS_FILE
);
747 if ( $desiredFileName != $filename && $comparableName != $filename ) {
755 * @param string $fileExtension The file extension to check
757 * @return array|null array with the following keys:
758 * 0 => string The final extension being used
759 * 1 => string[] The extensions that are allowed
760 * 2 => int The number of extensions that are allowed.
762 private function checkUnwantedFileExtensions( $fileExtension ) {
763 global $wgCheckFileExtensions, $wgFileExtensions, $wgLang;
765 if ( $wgCheckFileExtensions ) {
766 $extensions = array_unique( $wgFileExtensions );
767 if ( !$this->checkFileExtension( $fileExtension, $extensions ) ) {
770 $wgLang->commaList( $extensions ),
780 * @param int $fileSize
782 * @return array warnings
784 private function checkFileSize( $fileSize ) {
785 global $wgUploadSizeWarning;
789 if ( $wgUploadSizeWarning && ( $fileSize > $wgUploadSizeWarning ) ) {
790 $warnings['large-file'] = [ $wgUploadSizeWarning, $fileSize ];
793 if ( $fileSize == 0 ) {
794 $warnings['empty-file'] = true;
801 * @param LocalFile $localFile
802 * @param string $hash sha1 hash of the file to check
804 * @return array warnings
806 private function checkLocalFileExists( LocalFile
$localFile, $hash ) {
809 $exists = self
::getExistsWarning( $localFile );
810 if ( $exists !== false ) {
811 $warnings['exists'] = $exists;
813 // check if file is an exact duplicate of current file version
814 if ( $hash === $localFile->getSha1() ) {
815 $warnings['no-change'] = $localFile;
818 // check if file is an exact duplicate of older versions of this file
819 $history = $localFile->getHistory();
820 foreach ( $history as $oldFile ) {
821 if ( $hash === $oldFile->getSha1() ) {
822 $warnings['duplicate-version'][] = $oldFile;
830 private function checkLocalFileWasDeleted( LocalFile
$localFile ) {
831 return $localFile->wasDeleted() && !$localFile->exists();
835 * @param string $hash sha1 hash of the file to check
836 * @param bool $ignoreLocalDupes True to ignore local duplicates
838 * @return File[] Duplicate files, if found.
840 private function checkAgainstExistingDupes( $hash, $ignoreLocalDupes ) {
841 $dupes = RepoGroup
::singleton()->findBySha1( $hash );
842 $title = $this->getTitle();
843 foreach ( $dupes as $key => $dupe ) {
845 ( $dupe instanceof LocalFile
) &&
847 $title->equals( $dupe->getTitle() )
849 unset( $dupes[$key] );
857 * @param string $hash sha1 hash of the file to check
859 * @return string|null Name of the dupe or empty string if discovered (depending on visibility)
860 * null if the check discovered no dupes.
862 private function checkAgainstArchiveDupes( $hash ) {
863 $archivedFile = new ArchivedFile( null, 0, '', $hash );
864 if ( $archivedFile->getID() > 0 ) {
865 if ( $archivedFile->userCan( File
::DELETED_FILE
) ) {
866 return $archivedFile->getName();
876 * Really perform the upload. Stores the file in the local repo, watches
877 * if necessary and runs the UploadComplete hook.
879 * @param string $comment
880 * @param string $pageText
881 * @param bool $watch Whether the file page should be added to user's watchlist.
882 * (This doesn't check $user's permissions.)
884 * @param string[] $tags Change tags to add to the log entry and page revision.
885 * (This doesn't check $user's permissions.)
886 * @return Status Indicating the whether the upload succeeded.
888 public function performUpload( $comment, $pageText, $watch, $user, $tags = [] ) {
889 $this->getLocalFile()->load( File
::READ_LATEST
);
890 $props = $this->mFileProps
;
893 Hooks
::run( 'UploadVerifyUpload', [ $this, $user, $props, $comment, $pageText, &$error ] );
895 if ( !is_array( $error ) ) {
898 return Status
::newFatal( ...$error );
901 $status = $this->getLocalFile()->upload(
912 if ( $status->isGood() ) {
914 WatchAction
::doWatch(
915 $this->getLocalFile()->getTitle(),
917 User
::IGNORE_USER_RIGHTS
920 // Avoid PHP 7.1 warning of passing $this by reference
922 Hooks
::run( 'UploadComplete', [ &$uploadBase ] );
924 $this->postProcessUpload();
931 * Perform extra steps after a successful upload.
935 public function postProcessUpload() {
939 * Returns the title of the file to be uploaded. Sets mTitleError in case
940 * the name was illegal.
942 * @return Title|null The title of the file or null in case the name was illegal
944 public function getTitle() {
945 if ( $this->mTitle
!== false ) {
946 return $this->mTitle
;
948 if ( !is_string( $this->mDesiredDestName
) ) {
949 $this->mTitleError
= self
::ILLEGAL_FILENAME
;
950 $this->mTitle
= null;
952 return $this->mTitle
;
954 /* Assume that if a user specified File:Something.jpg, this is an error
955 * and that the namespace prefix needs to be stripped of.
957 $title = Title
::newFromText( $this->mDesiredDestName
);
958 if ( $title && $title->getNamespace() == NS_FILE
) {
959 $this->mFilteredName
= $title->getDBkey();
961 $this->mFilteredName
= $this->mDesiredDestName
;
964 # oi_archive_name is max 255 bytes, which include a timestamp and an
965 # exclamation mark, so restrict file name to 240 bytes.
966 if ( strlen( $this->mFilteredName
) > 240 ) {
967 $this->mTitleError
= self
::FILENAME_TOO_LONG
;
968 $this->mTitle
= null;
970 return $this->mTitle
;
974 * Chop off any directories in the given filename. Then
975 * filter out illegal characters, and try to make a legible name
976 * out of it. We'll strip some silently that Title would die on.
978 $this->mFilteredName
= wfStripIllegalFilenameChars( $this->mFilteredName
);
979 /* Normalize to title form before we do any further processing */
980 $nt = Title
::makeTitleSafe( NS_FILE
, $this->mFilteredName
);
981 if ( is_null( $nt ) ) {
982 $this->mTitleError
= self
::ILLEGAL_FILENAME
;
983 $this->mTitle
= null;
985 return $this->mTitle
;
987 $this->mFilteredName
= $nt->getDBkey();
990 * We'll want to blacklist against *any* 'extension', and use
991 * only the final one for the whitelist.
993 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName
);
996 $this->mFinalExtension
= trim( end( $ext ) );
998 $this->mFinalExtension
= '';
1000 # No extension, try guessing one
1001 $magic = MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer();
1002 $mime = $magic->guessMimeType( $this->mTempPath
);
1003 if ( $mime !== 'unknown/unknown' ) {
1004 # Get a space separated list of extensions
1005 $extList = $magic->getExtensionsForType( $mime );
1007 # Set the extension to the canonical extension
1008 $this->mFinalExtension
= strtok( $extList, ' ' );
1010 # Fix up the other variables
1011 $this->mFilteredName
.= ".{$this->mFinalExtension}";
1012 $nt = Title
::makeTitleSafe( NS_FILE
, $this->mFilteredName
);
1013 $ext = [ $this->mFinalExtension
];
1018 /* Don't allow users to override the blacklist (check file extension) */
1019 global $wgCheckFileExtensions, $wgStrictFileExtensions;
1020 global $wgFileExtensions, $wgFileBlacklist;
1022 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
1024 if ( $this->mFinalExtension
== '' ) {
1025 $this->mTitleError
= self
::FILETYPE_MISSING
;
1026 $this->mTitle
= null;
1028 return $this->mTitle
;
1029 } elseif ( $blackListedExtensions ||
1030 ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
1031 !$this->checkFileExtension( $this->mFinalExtension
, $wgFileExtensions ) )
1033 $this->mBlackListedExtensions
= $blackListedExtensions;
1034 $this->mTitleError
= self
::FILETYPE_BADTYPE
;
1035 $this->mTitle
= null;
1037 return $this->mTitle
;
1040 // Windows may be broken with special characters, see T3780
1041 if ( !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() )
1042 && !RepoGroup
::singleton()->getLocalRepo()->backendSupportsUnicodePaths()
1044 $this->mTitleError
= self
::WINDOWS_NONASCII_FILENAME
;
1045 $this->mTitle
= null;
1047 return $this->mTitle
;
1050 # If there was more than one "extension", reassemble the base
1051 # filename to prevent bogus complaints about length
1052 if ( count( $ext ) > 1 ) {
1053 $iterations = count( $ext ) - 1;
1054 for ( $i = 0; $i < $iterations; $i++
) {
1055 $partname .= '.' . $ext[$i];
1059 if ( strlen( $partname ) < 1 ) {
1060 $this->mTitleError
= self
::MIN_LENGTH_PARTNAME
;
1061 $this->mTitle
= null;
1063 return $this->mTitle
;
1066 $this->mTitle
= $nt;
1068 return $this->mTitle
;
1072 * Return the local file and initializes if necessary.
1074 * @return LocalFile|null
1076 public function getLocalFile() {
1077 if ( is_null( $this->mLocalFile
) ) {
1078 $nt = $this->getTitle();
1079 $this->mLocalFile
= is_null( $nt ) ?
null : wfLocalFile( $nt );
1082 return $this->mLocalFile
;
1086 * @return UploadStashFile|null
1088 public function getStashFile() {
1089 return $this->mStashFile
;
1093 * Like stashFile(), but respects extensions' wishes to prevent the stashing. verifyUpload() must
1094 * be called before calling this method (unless $isPartial is true).
1096 * Upload stash exceptions are also caught and converted to an error status.
1100 * @param bool $isPartial Pass `true` if this is a part of a chunked upload (not a complete file).
1101 * @return Status If successful, value is an UploadStashFile instance
1103 public function tryStashFile( User
$user, $isPartial = false ) {
1104 if ( !$isPartial ) {
1105 $error = $this->runUploadStashFileHook( $user );
1107 return Status
::newFatal( ...$error );
1111 $file = $this->doStashFile( $user );
1112 return Status
::newGood( $file );
1113 } catch ( UploadStashException
$e ) {
1114 return Status
::newFatal( 'uploadstash-exception', get_class( $e ), $e->getMessage() );
1120 * @return array|null Error message and parameters, null if there's no error
1122 protected function runUploadStashFileHook( User
$user ) {
1123 $props = $this->mFileProps
;
1125 Hooks
::run( 'UploadStashFile', [ $this, $user, $props, &$error ] );
1126 if ( $error && !is_array( $error ) ) {
1127 $error = [ $error ];
1133 * If the user does not supply all necessary information in the first upload
1134 * form submission (either by accident or by design) then we may want to
1135 * stash the file temporarily, get more information, and publish the file
1138 * This method will stash a file in a temporary directory for later
1139 * processing, and save the necessary descriptive info into the database.
1140 * This method returns the file object, which also has a 'fileKey' property
1141 * which can be passed through a form or API request to find this stashed
1144 * @deprecated since 1.28 Use tryStashFile() instead
1145 * @param User|null $user
1146 * @return UploadStashFile Stashed file
1147 * @throws UploadStashBadPathException
1148 * @throws UploadStashFileException
1149 * @throws UploadStashNotLoggedInException
1151 public function stashFile( User
$user = null ) {
1152 wfDeprecated( __METHOD__
, '1.28' );
1154 return $this->doStashFile( $user );
1158 * Implementation for stashFile() and tryStashFile().
1160 * @param User|null $user
1161 * @return UploadStashFile Stashed file
1163 protected function doStashFile( User
$user = null ) {
1164 $stash = RepoGroup
::singleton()->getLocalRepo()->getUploadStash( $user );
1165 $file = $stash->stashFile( $this->mTempPath
, $this->getSourceType() );
1166 $this->mStashFile
= $file;
1172 * If we've modified the upload file we need to manually remove it
1173 * on exit to clean up.
1175 public function cleanupTempFile() {
1176 if ( $this->mRemoveTempFile
&& $this->tempFileObj
) {
1177 // Delete when all relevant TempFSFile handles go out of scope
1178 wfDebug( __METHOD__
. ": Marked temporary file '{$this->mTempPath}' for removal\n" );
1179 $this->tempFileObj
->autocollect();
1183 public function getTempPath() {
1184 return $this->mTempPath
;
1188 * Split a file into a base name and all dot-delimited 'extensions'
1189 * on the end. Some web server configurations will fall back to
1190 * earlier pseudo-'extensions' to determine type and execute
1191 * scripts, so the blacklist needs to check them all.
1193 * @param string $filename
1194 * @return array [ string, string[] ]
1196 public static function splitExtensions( $filename ) {
1197 $bits = explode( '.', $filename );
1198 $basename = array_shift( $bits );
1200 return [ $basename, $bits ];
1204 * Perform case-insensitive match against a list of file extensions.
1205 * Returns true if the extension is in the list.
1207 * @param string $ext
1208 * @param array $list
1211 public static function checkFileExtension( $ext, $list ) {
1212 return in_array( strtolower( $ext ), $list );
1216 * Perform case-insensitive match against a list of file extensions.
1217 * Returns an array of matching extensions.
1219 * @param string[] $ext
1220 * @param string[] $list
1223 public static function checkFileExtensionList( $ext, $list ) {
1224 return array_intersect( array_map( 'strtolower', $ext ), $list );
1228 * Checks if the MIME type of the uploaded file matches the file extension.
1230 * @param string $mime The MIME type of the uploaded file
1231 * @param string $extension The filename extension that the file is to be served with
1234 public static function verifyExtension( $mime, $extension ) {
1235 $magic = MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer();
1237 if ( !$mime ||
$mime == 'unknown' ||
$mime == 'unknown/unknown' ) {
1238 if ( !$magic->isRecognizableExtension( $extension ) ) {
1239 wfDebug( __METHOD__
. ": passing file with unknown detected mime type; " .
1240 "unrecognized extension '$extension', can't verify\n" );
1244 wfDebug( __METHOD__
. ": rejecting file with unknown detected mime type; " .
1245 "recognized extension '$extension', so probably invalid file\n" );
1251 $match = $magic->isMatchingExtension( $extension, $mime );
1253 if ( $match === null ) {
1254 if ( $magic->getTypesForExtension( $extension ) !== null ) {
1255 wfDebug( __METHOD__
. ": No extension known for $mime, but we know a mime for $extension\n" );
1259 wfDebug( __METHOD__
. ": no file extension known for mime type $mime, passing file\n" );
1263 } elseif ( $match === true ) {
1264 wfDebug( __METHOD__
. ": mime type $mime matches extension $extension, passing file\n" );
1266 /** @todo If it's a bitmap, make sure PHP or ImageMagick resp. can handle it! */
1270 . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
1277 * Heuristic for detecting files that *could* contain JavaScript instructions or
1278 * things that may look like HTML to a browser and are thus
1279 * potentially harmful. The present implementation will produce false
1280 * positives in some situations.
1282 * @param string $file Pathname to the temporary upload file
1283 * @param string $mime The MIME type of the file
1284 * @param string $extension The extension of the file
1285 * @return bool True if the file contains something looking like embedded scripts
1287 public static function detectScript( $file, $mime, $extension ) {
1288 # ugly hack: for text files, always look at the entire file.
1289 # For binary field, just check the first K.
1291 $isText = strpos( $mime, 'text/' ) === 0;
1293 $chunk = file_get_contents( $file );
1295 $fp = fopen( $file, 'rb' );
1296 $chunk = fread( $fp, 1024 );
1300 $chunk = strtolower( $chunk );
1306 # decode from UTF-16 if needed (could be used for obfuscation).
1307 if ( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
1309 } elseif ( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
1316 $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
1319 $chunk = trim( $chunk );
1321 /** @todo FIXME: Convert from UTF-16 if necessary! */
1322 wfDebug( __METHOD__
. ": checking for embedded scripts and HTML stuff\n" );
1324 # check for HTML doctype
1325 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
1329 // Some browsers will interpret obscure xml encodings as UTF-8, while
1330 // PHP/expat will interpret the given encoding in the xml declaration (T49304)
1331 if ( $extension == 'svg' ||
strpos( $mime, 'image/svg' ) === 0 ) {
1332 if ( self
::checkXMLEncodingMissmatch( $file ) ) {
1337 // Quick check for HTML heuristics in old IE and Safari.
1339 // The exact heuristics IE uses are checked separately via verifyMimeType(), so we
1340 // don't need them all here as it can cause many false positives.
1342 // Check for `<script` and such still to forbid script tags and embedded HTML in SVG:
1346 '<html', # also in safari
1347 '<script', # also in safari
1350 foreach ( $tags as $tag ) {
1351 if ( strpos( $chunk, $tag ) !== false ) {
1352 wfDebug( __METHOD__
. ": found something that may make it be mistaken for html: $tag\n" );
1359 * look for JavaScript
1362 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
1363 $chunk = Sanitizer
::decodeCharReferences( $chunk );
1365 # look for script-types
1366 if ( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
1367 wfDebug( __METHOD__
. ": found script types\n" );
1372 # look for html-style script-urls
1373 if ( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1374 wfDebug( __METHOD__
. ": found html-style script urls\n" );
1379 # look for css-style script-urls
1380 if ( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1381 wfDebug( __METHOD__
. ": found css-style script urls\n" );
1386 wfDebug( __METHOD__
. ": no scripts found\n" );
1392 * Check a whitelist of xml encodings that are known not to be interpreted differently
1393 * by the server's xml parser (expat) and some common browsers.
1395 * @param string $file Pathname to the temporary upload file
1396 * @return bool True if the file contains an encoding that could be misinterpreted
1398 public static function checkXMLEncodingMissmatch( $file ) {
1399 global $wgSVGMetadataCutoff;
1400 $contents = file_get_contents( $file, false, null, 0, $wgSVGMetadataCutoff );
1401 $encodingRegex = '!encoding[ \t\n\r]*=[ \t\n\r]*[\'"](.*?)[\'"]!si';
1403 if ( preg_match( "!<\?xml\b(.*?)\?>!si", $contents, $matches ) ) {
1404 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1405 && !in_array( strtoupper( $encMatch[1] ), self
::$safeXmlEncodings )
1407 wfDebug( __METHOD__
. ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1411 } elseif ( preg_match( "!<\?xml\b!si", $contents ) ) {
1412 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1413 // bytes. There shouldn't be a legitimate reason for this to happen.
1414 wfDebug( __METHOD__
. ": Unmatched XML declaration start\n" );
1417 } elseif ( substr( $contents, 0, 4 ) == "\x4C\x6F\xA7\x94" ) {
1418 // EBCDIC encoded XML
1419 wfDebug( __METHOD__
. ": EBCDIC Encoded XML\n" );
1424 // It's possible the file is encoded with multi-byte encoding, so re-encode attempt to
1425 // detect the encoding in case is specifies an encoding not whitelisted in self::$safeXmlEncodings
1426 $attemptEncodings = [ 'UTF-16', 'UTF-16BE', 'UTF-32', 'UTF-32BE' ];
1427 foreach ( $attemptEncodings as $encoding ) {
1428 Wikimedia\
suppressWarnings();
1429 $str = iconv( $encoding, 'UTF-8', $contents );
1430 Wikimedia\restoreWarnings
();
1431 if ( $str != '' && preg_match( "!<\?xml\b(.*?)\?>!si", $str, $matches ) ) {
1432 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1433 && !in_array( strtoupper( $encMatch[1] ), self
::$safeXmlEncodings )
1435 wfDebug( __METHOD__
. ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1439 } elseif ( $str != '' && preg_match( "!<\?xml\b!si", $str ) ) {
1440 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1441 // bytes. There shouldn't be a legitimate reason for this to happen.
1442 wfDebug( __METHOD__
. ": Unmatched XML declaration start\n" );
1452 * @param string $filename
1453 * @param bool $partial
1454 * @return bool|array
1456 protected function detectScriptInSvg( $filename, $partial ) {
1457 $this->mSVGNSError
= false;
1458 $check = new XmlTypeCheck(
1460 [ $this, 'checkSvgScriptCallback' ],
1463 'processing_instruction_handler' => 'UploadBase::checkSvgPICallback',
1464 'external_dtd_handler' => 'UploadBase::checkSvgExternalDTD',
1467 if ( $check->wellFormed
!== true ) {
1468 // Invalid xml (T60553)
1469 // But only when non-partial (T67724)
1470 return $partial ?
false : [ 'uploadinvalidxml' ];
1471 } elseif ( $check->filterMatch
) {
1472 if ( $this->mSVGNSError
) {
1473 return [ 'uploadscriptednamespace', $this->mSVGNSError
];
1476 return $check->filterMatchType
;
1483 * Callback to filter SVG Processing Instructions.
1484 * @param string $target Processing instruction name
1485 * @param string $data Processing instruction attribute and value
1486 * @return bool|array
1488 public static function checkSvgPICallback( $target, $data ) {
1489 // Don't allow external stylesheets (T59550)
1490 if ( preg_match( '/xml-stylesheet/i', $target ) ) {
1491 return [ 'upload-scripted-pi-callback' ];
1498 * Verify that DTD urls referenced are only the standard dtds
1500 * Browsers seem to ignore external dtds. However just to be on the
1501 * safe side, only allow dtds from the svg standard.
1503 * @param string $type PUBLIC or SYSTEM
1504 * @param string $publicId The well-known public identifier for the dtd
1505 * @param string $systemId The url for the external dtd
1506 * @return bool|array
1508 public static function checkSvgExternalDTD( $type, $publicId, $systemId ) {
1509 // This doesn't include the XHTML+MathML+SVG doctype since we don't
1510 // allow XHTML anyways.
1512 'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd',
1513 'http://www.w3.org/TR/2001/REC-SVG-20010904/DTD/svg10.dtd',
1514 'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-basic.dtd',
1515 'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-tiny.dtd',
1516 // https://phabricator.wikimedia.org/T168856
1517 'http://www.w3.org/TR/2001/PR-SVG-20010719/DTD/svg10.dtd',
1519 if ( $type !== 'PUBLIC'
1520 ||
!in_array( $systemId, $allowedDTDs )
1521 ||
strpos( $publicId, "-//W3C//" ) !== 0
1523 return [ 'upload-scripted-dtd' ];
1529 * @todo Replace this with a whitelist filter!
1530 * @param string $element
1531 * @param array $attribs
1532 * @param string|null $data
1533 * @return bool|array
1535 public function checkSvgScriptCallback( $element, $attribs, $data = null ) {
1536 list( $namespace, $strippedElement ) = $this->splitXmlNamespace( $element );
1538 // We specifically don't include:
1539 // http://www.w3.org/1999/xhtml (T62771)
1540 static $validNamespaces = [
1543 'http://creativecommons.org/ns#',
1544 'http://inkscape.sourceforge.net/dtd/sodipodi-0.dtd',
1545 'http://ns.adobe.com/adobeillustrator/10.0/',
1546 'http://ns.adobe.com/adobesvgviewerextensions/3.0/',
1547 'http://ns.adobe.com/extensibility/1.0/',
1548 'http://ns.adobe.com/flows/1.0/',
1549 'http://ns.adobe.com/illustrator/1.0/',
1550 'http://ns.adobe.com/imagereplacement/1.0/',
1551 'http://ns.adobe.com/pdf/1.3/',
1552 'http://ns.adobe.com/photoshop/1.0/',
1553 'http://ns.adobe.com/saveforweb/1.0/',
1554 'http://ns.adobe.com/variables/1.0/',
1555 'http://ns.adobe.com/xap/1.0/',
1556 'http://ns.adobe.com/xap/1.0/g/',
1557 'http://ns.adobe.com/xap/1.0/g/img/',
1558 'http://ns.adobe.com/xap/1.0/mm/',
1559 'http://ns.adobe.com/xap/1.0/rights/',
1560 'http://ns.adobe.com/xap/1.0/stype/dimensions#',
1561 'http://ns.adobe.com/xap/1.0/stype/font#',
1562 'http://ns.adobe.com/xap/1.0/stype/manifestitem#',
1563 'http://ns.adobe.com/xap/1.0/stype/resourceevent#',
1564 'http://ns.adobe.com/xap/1.0/stype/resourceref#',
1565 'http://ns.adobe.com/xap/1.0/t/pg/',
1566 'http://purl.org/dc/elements/1.1/',
1567 'http://purl.org/dc/elements/1.1',
1568 'http://schemas.microsoft.com/visio/2003/svgextensions/',
1569 'http://sodipodi.sourceforge.net/dtd/sodipodi-0.dtd',
1570 'http://taptrix.com/inkpad/svg_extensions',
1571 'http://web.resource.org/cc/',
1572 'http://www.freesoftware.fsf.org/bkchem/cdml',
1573 'http://www.inkscape.org/namespaces/inkscape',
1574 'http://www.opengis.net/gml',
1575 'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
1576 'http://www.w3.org/2000/svg',
1577 'http://www.w3.org/tr/rec-rdf-syntax/',
1578 'http://www.w3.org/2000/01/rdf-schema#',
1581 // Inkscape mangles namespace definitions created by Adobe Illustrator.
1582 // This is nasty but harmless. (T144827)
1583 $isBuggyInkscape = preg_match( '/^&(#38;)*ns_[a-z_]+;$/', $namespace );
1585 if ( !( $isBuggyInkscape ||
in_array( $namespace, $validNamespaces ) ) ) {
1586 wfDebug( __METHOD__
. ": Non-svg namespace '$namespace' in uploaded file.\n" );
1587 /** @todo Return a status object to a closure in XmlTypeCheck, for MW1.21+ */
1588 $this->mSVGNSError
= $namespace;
1594 * check for elements that can contain javascript
1596 if ( $strippedElement == 'script' ) {
1597 wfDebug( __METHOD__
. ": Found script element '$element' in uploaded file.\n" );
1599 return [ 'uploaded-script-svg', $strippedElement ];
1602 # e.g., <svg xmlns="http://www.w3.org/2000/svg">
1603 # <handler xmlns:ev="http://www.w3.org/2001/xml-events" ev:event="load">alert(1)</handler> </svg>
1604 if ( $strippedElement == 'handler' ) {
1605 wfDebug( __METHOD__
. ": Found scriptable element '$element' in uploaded file.\n" );
1607 return [ 'uploaded-script-svg', $strippedElement ];
1610 # SVG reported in Feb '12 that used xml:stylesheet to generate javascript block
1611 if ( $strippedElement == 'stylesheet' ) {
1612 wfDebug( __METHOD__
. ": Found scriptable element '$element' in uploaded file.\n" );
1614 return [ 'uploaded-script-svg', $strippedElement ];
1617 # Block iframes, in case they pass the namespace check
1618 if ( $strippedElement == 'iframe' ) {
1619 wfDebug( __METHOD__
. ": iframe in uploaded file.\n" );
1621 return [ 'uploaded-script-svg', $strippedElement ];
1625 if ( $strippedElement == 'style'
1626 && self
::checkCssFragment( Sanitizer
::normalizeCss( $data ) )
1628 wfDebug( __METHOD__
. ": hostile css in style element.\n" );
1629 return [ 'uploaded-hostile-svg' ];
1632 foreach ( $attribs as $attrib => $value ) {
1633 $stripped = $this->stripXmlNamespace( $attrib );
1634 $value = strtolower( $value );
1636 if ( substr( $stripped, 0, 2 ) == 'on' ) {
1638 . ": Found event-handler attribute '$attrib'='$value' in uploaded file.\n" );
1640 return [ 'uploaded-event-handler-on-svg', $attrib, $value ];
1643 # Do not allow relative links, or unsafe url schemas.
1644 # For <a> tags, only data:, http: and https: and same-document
1645 # fragment links are allowed. For all other tags, only data:
1646 # and fragment are allowed.
1647 if ( $stripped == 'href'
1649 && strpos( $value, 'data:' ) !== 0
1650 && strpos( $value, '#' ) !== 0
1652 if ( !( $strippedElement === 'a'
1653 && preg_match( '!^https?://!i', $value ) )
1655 wfDebug( __METHOD__
. ": Found href attribute <$strippedElement "
1656 . "'$attrib'='$value' in uploaded file.\n" );
1658 return [ 'uploaded-href-attribute-svg', $strippedElement, $attrib, $value ];
1662 # only allow data: targets that should be safe. This prevents vectors like,
1663 # image/svg, text/xml, application/xml, and text/html, which can contain scripts
1664 if ( $stripped == 'href' && strncasecmp( 'data:', $value, 5 ) === 0 ) {
1665 // rfc2397 parameters. This is only slightly slower than (;[\w;]+)*.
1666 // phpcs:ignore Generic.Files.LineLength
1667 $parameters = '(?>;[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+=(?>[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+|"(?>[\0-\x0c\x0e-\x21\x23-\x5b\x5d-\x7f]+|\\\\[\0-\x7f])*"))*(?:;base64)?';
1669 if ( !preg_match( "!^data:\s*image/(gif|jpeg|jpg|png)$parameters,!i", $value ) ) {
1670 wfDebug( __METHOD__
. ": Found href to unwhitelisted data: uri "
1671 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1672 return [ 'uploaded-href-unsafe-target-svg', $strippedElement, $attrib, $value ];
1676 # Change href with animate from (http://html5sec.org/#137).
1677 if ( $stripped === 'attributename'
1678 && $strippedElement === 'animate'
1679 && $this->stripXmlNamespace( $value ) == 'href'
1681 wfDebug( __METHOD__
. ": Found animate that might be changing href using from "
1682 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1684 return [ 'uploaded-animate-svg', $strippedElement, $attrib, $value ];
1687 # use set/animate to add event-handler attribute to parent
1688 if ( ( $strippedElement == 'set' ||
$strippedElement == 'animate' )
1689 && $stripped == 'attributename'
1690 && substr( $value, 0, 2 ) == 'on'
1692 wfDebug( __METHOD__
. ": Found svg setting event-handler attribute with "
1693 . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1695 return [ 'uploaded-setting-event-handler-svg', $strippedElement, $stripped, $value ];
1698 # use set to add href attribute to parent element
1699 if ( $strippedElement == 'set'
1700 && $stripped == 'attributename'
1701 && strpos( $value, 'href' ) !== false
1703 wfDebug( __METHOD__
. ": Found svg setting href attribute '$value' in uploaded file.\n" );
1705 return [ 'uploaded-setting-href-svg' ];
1708 # use set to add a remote / data / script target to an element
1709 if ( $strippedElement == 'set'
1710 && $stripped == 'to'
1711 && preg_match( '!(http|https|data|script):!sim', $value )
1713 wfDebug( __METHOD__
. ": Found svg setting attribute to '$value' in uploaded file.\n" );
1715 return [ 'uploaded-wrong-setting-svg', $value ];
1718 # use handler attribute with remote / data / script
1719 if ( $stripped == 'handler' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1720 wfDebug( __METHOD__
. ": Found svg setting handler with remote/data/script "
1721 . "'$attrib'='$value' in uploaded file.\n" );
1723 return [ 'uploaded-setting-handler-svg', $attrib, $value ];
1726 # use CSS styles to bring in remote code
1727 if ( $stripped == 'style'
1728 && self
::checkCssFragment( Sanitizer
::normalizeCss( $value ) )
1730 wfDebug( __METHOD__
. ": Found svg setting a style with "
1731 . "remote url '$attrib'='$value' in uploaded file.\n" );
1732 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1735 # Several attributes can include css, css character escaping isn't allowed
1736 $cssAttrs = [ 'font', 'clip-path', 'fill', 'filter', 'marker',
1737 'marker-end', 'marker-mid', 'marker-start', 'mask', 'stroke' ];
1738 if ( in_array( $stripped, $cssAttrs )
1739 && self
::checkCssFragment( $value )
1741 wfDebug( __METHOD__
. ": Found svg setting a style with "
1742 . "remote url '$attrib'='$value' in uploaded file.\n" );
1743 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1746 # image filters can pull in url, which could be svg that executes scripts
1747 # Only allow url( "#foo" ). Do not allow url( http://example.com )
1748 if ( $strippedElement == 'image'
1749 && $stripped == 'filter'
1750 && preg_match( '!url\s*\(\s*["\']?[^#]!sim', $value )
1752 wfDebug( __METHOD__
. ": Found image filter with url: "
1753 . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1755 return [ 'uploaded-image-filter-svg', $strippedElement, $stripped, $value ];
1759 return false; // No scripts detected
1763 * Check a block of CSS or CSS fragment for anything that looks like
1764 * it is bringing in remote code.
1765 * @param string $value a string of CSS
1766 * @return bool true if the CSS contains an illegal string, false if otherwise
1768 private static function checkCssFragment( $value ) {
1769 # Forbid external stylesheets, for both reliability and to protect viewer's privacy
1770 if ( stripos( $value, '@import' ) !== false ) {
1774 # We allow @font-face to embed fonts with data: urls, so we snip the string
1775 # 'url' out so this case won't match when we check for urls below
1776 $pattern = '!(@font-face\s*{[^}]*src:)url(\("data:;base64,)!im';
1777 $value = preg_replace( $pattern, '$1$2', $value );
1779 # Check for remote and executable CSS. Unlike in Sanitizer::checkCss, the CSS
1780 # properties filter and accelerator don't seem to be useful for xss in SVG files.
1781 # Expression and -o-link don't seem to work either, but filtering them here in case.
1782 # Additionally, we catch remote urls like url("http:..., url('http:..., url(http:...,
1783 # but not local ones such as url("#..., url('#..., url(#....
1784 if ( preg_match( '!expression
1786 | -o-link-source\s*:
1787 | -o-replace\s*:!imx', $value ) ) {
1791 if ( preg_match_all(
1792 "!(\s*(url|image|image-set)\s*\(\s*[\"']?\s*[^#]+.*?\))!sim",
1797 # TODO: redo this in one regex. Until then, url("#whatever") matches the first
1798 foreach ( $matches[1] as $match ) {
1799 if ( !preg_match( "!\s*(url|image|image-set)\s*\(\s*(#|'#|\"#)!im", $match ) ) {
1805 if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ) {
1813 * Divide the element name passed by the xml parser to the callback into URI and prifix.
1814 * @param string $element
1815 * @return array Containing the namespace URI and prefix
1817 private static function splitXmlNamespace( $element ) {
1818 // 'http://www.w3.org/2000/svg:script' -> [ 'http://www.w3.org/2000/svg', 'script' ]
1819 $parts = explode( ':', strtolower( $element ) );
1820 $name = array_pop( $parts );
1821 $ns = implode( ':', $parts );
1823 return [ $ns, $name ];
1827 * @param string $name
1830 private function stripXmlNamespace( $name ) {
1831 // 'http://www.w3.org/2000/svg:script' -> 'script'
1832 $parts = explode( ':', strtolower( $name ) );
1834 return array_pop( $parts );
1838 * Generic wrapper function for a virus scanner program.
1839 * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
1840 * $wgAntivirusRequired may be used to deny upload if the scan fails.
1842 * @param string $file Pathname to the temporary upload file
1843 * @return bool|null|string False if not virus is found, null if the scan fails or is disabled,
1844 * or a string containing feedback from the virus scanner if a virus was found.
1845 * If textual feedback is missing but a virus was found, this function returns true.
1847 public static function detectVirus( $file ) {
1848 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
1850 if ( !$wgAntivirus ) {
1851 wfDebug( __METHOD__
. ": virus scanner disabled\n" );
1856 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
1857 wfDebug( __METHOD__
. ": unknown virus scanner: $wgAntivirus\n" );
1858 $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
1859 [ 'virus-badscanner', $wgAntivirus ] );
1861 return wfMessage( 'virus-unknownscanner' )->text() . " $wgAntivirus";
1864 # look up scanner configuration
1865 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
1866 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
1867 $msgPattern = $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ??
null;
1869 if ( strpos( $command, "%f" ) === false ) {
1870 # simple pattern: append file to scan
1871 $command .= " " . Shell
::escape( $file );
1873 # complex pattern: replace "%f" with file to scan
1874 $command = str_replace( "%f", Shell
::escape( $file ), $command );
1877 wfDebug( __METHOD__
. ": running virus scan: $command \n" );
1879 # execute virus scanner
1882 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1883 # that does not seem to be worth the pain.
1884 # Ask me (Duesentrieb) about it if it's ever needed.
1885 $output = wfShellExecWithStderr( $command, $exitCode );
1887 # map exit code to AV_xxx constants.
1888 $mappedCode = $exitCode;
1889 if ( $exitCodeMap ) {
1890 if ( isset( $exitCodeMap[$exitCode] ) ) {
1891 $mappedCode = $exitCodeMap[$exitCode];
1892 } elseif ( isset( $exitCodeMap["*"] ) ) {
1893 $mappedCode = $exitCodeMap["*"];
1897 /* NB: AV_NO_VIRUS is 0 but AV_SCAN_FAILED is false,
1898 * so we need the strict equalities === and thus can't use a switch here
1900 if ( $mappedCode === AV_SCAN_FAILED
) {
1901 # scan failed (code was mapped to false by $exitCodeMap)
1902 wfDebug( __METHOD__
. ": failed to scan $file (code $exitCode).\n" );
1904 $output = $wgAntivirusRequired
1905 ?
wfMessage( 'virus-scanfailed', [ $exitCode ] )->text()
1907 } elseif ( $mappedCode === AV_SCAN_ABORTED
) {
1908 # scan failed because filetype is unknown (probably imune)
1909 wfDebug( __METHOD__
. ": unsupported file type $file (code $exitCode).\n" );
1911 } elseif ( $mappedCode === AV_NO_VIRUS
) {
1913 wfDebug( __METHOD__
. ": file passed virus scan.\n" );
1916 $output = trim( $output );
1919 $output = true; # if there's no output, return true
1920 } elseif ( $msgPattern ) {
1922 if ( preg_match( $msgPattern, $output, $groups ) && $groups[1] ) {
1923 $output = $groups[1];
1927 wfDebug( __METHOD__
. ": FOUND VIRUS! scanner feedback: $output \n" );
1934 * Check if there's an overwrite conflict and, if so, if restrictions
1935 * forbid this user from performing the upload.
1939 * @return bool|array
1941 private function checkOverwrite( $user ) {
1942 // First check whether the local file can be overwritten
1943 $file = $this->getLocalFile();
1944 $file->load( File
::READ_LATEST
);
1945 if ( $file->exists() ) {
1946 if ( !self
::userCanReUpload( $user, $file ) ) {
1947 return [ 'fileexists-forbidden', $file->getName() ];
1953 /* Check shared conflicts: if the local file does not exist, but
1954 * wfFindFile finds a file, it exists in a shared repository.
1956 $file = wfFindFile( $this->getTitle(), [ 'latest' => true ] );
1957 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1958 return [ 'fileexists-shared-forbidden', $file->getName() ];
1965 * Check if a user is the last uploader
1971 public static function userCanReUpload( User
$user, File
$img ) {
1972 if ( $user->isAllowed( 'reupload' ) ) {
1973 return true; // non-conditional
1974 } elseif ( !$user->isAllowed( 'reupload-own' ) ) {
1978 if ( !( $img instanceof LocalFile
) ) {
1984 return $user->getId() == $img->getUser( 'id' );
1988 * Helper function that does various existence checks for a file.
1989 * The following checks are performed:
1991 * - Article with the same name as the file exists
1992 * - File exists with normalized extension
1993 * - The file looks like a thumbnail and the original exists
1995 * @param File $file The File object to check
1996 * @return mixed False if the file does not exists, else an array
1998 public static function getExistsWarning( $file ) {
1999 if ( $file->exists() ) {
2000 return [ 'warning' => 'exists', 'file' => $file ];
2003 if ( $file->getTitle()->getArticleID() ) {
2004 return [ 'warning' => 'page-exists', 'file' => $file ];
2007 if ( strpos( $file->getName(), '.' ) == false ) {
2008 $partname = $file->getName();
2011 $n = strrpos( $file->getName(), '.' );
2012 $extension = substr( $file->getName(), $n +
1 );
2013 $partname = substr( $file->getName(), 0, $n );
2015 $normalizedExtension = File
::normalizeExtension( $extension );
2017 if ( $normalizedExtension != $extension ) {
2018 // We're not using the normalized form of the extension.
2019 // Normal form is lowercase, using most common of alternate
2020 // extensions (eg 'jpg' rather than 'JPEG').
2022 // Check for another file using the normalized form...
2023 $nt_lc = Title
::makeTitle( NS_FILE
, "{$partname}.{$normalizedExtension}" );
2024 $file_lc = wfLocalFile( $nt_lc );
2026 if ( $file_lc->exists() ) {
2028 'warning' => 'exists-normalized',
2030 'normalizedFile' => $file_lc
2035 // Check for files with the same name but a different extension
2036 $similarFiles = RepoGroup
::singleton()->getLocalRepo()->findFilesByPrefix(
2037 "{$partname}.", 1 );
2038 if ( count( $similarFiles ) ) {
2040 'warning' => 'exists-normalized',
2042 'normalizedFile' => $similarFiles[0],
2046 if ( self
::isThumbName( $file->getName() ) ) {
2047 # Check for filenames like 50px- or 180px-, these are mostly thumbnails
2048 $nt_thb = Title
::newFromText(
2049 substr( $partname, strpos( $partname, '-' ) +
1 ) . '.' . $extension,
2052 $file_thb = wfLocalFile( $nt_thb );
2053 if ( $file_thb->exists() ) {
2055 'warning' => 'thumb',
2057 'thumbFile' => $file_thb
2060 // File does not exist, but we just don't like the name
2062 'warning' => 'thumb-name',
2064 'thumbFile' => $file_thb
2069 foreach ( self
::getFilenamePrefixBlacklist() as $prefix ) {
2070 if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
2072 'warning' => 'bad-prefix',
2083 * Helper function that checks whether the filename looks like a thumbnail
2084 * @param string $filename
2087 public static function isThumbName( $filename ) {
2088 $n = strrpos( $filename, '.' );
2089 $partname = $n ?
substr( $filename, 0, $n ) : $filename;
2092 substr( $partname, 3, 3 ) == 'px-' ||
2093 substr( $partname, 2, 3 ) == 'px-'
2095 preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) );
2099 * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
2101 * @return array List of prefixes
2103 public static function getFilenamePrefixBlacklist() {
2105 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
2106 if ( !$message->isDisabled() ) {
2107 $lines = explode( "\n", $message->plain() );
2108 foreach ( $lines as $line ) {
2109 // Remove comment lines
2110 $comment = substr( trim( $line ), 0, 1 );
2111 if ( $comment == '#' ||
$comment == '' ) {
2114 // Remove additional comments after a prefix
2115 $comment = strpos( $line, '#' );
2116 if ( $comment > 0 ) {
2117 $line = substr( $line, 0, $comment - 1 );
2119 $blacklist[] = trim( $line );
2127 * Gets image info about the file just uploaded.
2129 * Also has the effect of setting metadata to be an 'indexed tag name' in
2130 * returned API result if 'metadata' was requested. Oddly, we have to pass
2131 * the "result" object down just so it can do that with the appropriate
2132 * format, presumably.
2134 * @param ApiResult $result
2135 * @return array Image info
2137 public function getImageInfo( $result ) {
2138 $localFile = $this->getLocalFile();
2139 $stashFile = $this->getStashFile();
2140 // Calling a different API module depending on whether the file was stashed is less than optimal.
2141 // In fact, calling API modules here at all is less than optimal. Maybe it should be refactored.
2143 $imParam = ApiQueryStashImageInfo
::getPropertyNames();
2144 $info = ApiQueryStashImageInfo
::getInfo( $stashFile, array_flip( $imParam ), $result );
2146 $imParam = ApiQueryImageInfo
::getPropertyNames();
2147 $info = ApiQueryImageInfo
::getInfo( $localFile, array_flip( $imParam ), $result );
2154 * @param array $error
2157 public function convertVerifyErrorToStatus( $error ) {
2158 $code = $error['status'];
2159 unset( $code['status'] );
2161 return Status
::newFatal( $this->getVerificationErrorCode( $code ), $error );
2165 * Get the MediaWiki maximum uploaded file size for given type of upload, based on
2168 * @param null|string $forType
2171 public static function getMaxUploadSize( $forType = null ) {
2172 global $wgMaxUploadSize;
2174 if ( is_array( $wgMaxUploadSize ) ) {
2175 if ( !is_null( $forType ) && isset( $wgMaxUploadSize[$forType] ) ) {
2176 return $wgMaxUploadSize[$forType];
2178 return $wgMaxUploadSize['*'];
2181 return intval( $wgMaxUploadSize );
2186 * Get the PHP maximum uploaded file size, based on ini settings. If there is no limit or the
2187 * limit can't be guessed, returns a very large number (PHP_INT_MAX).
2192 public static function getMaxPhpUploadSize() {
2193 $phpMaxFileSize = wfShorthandToInteger(
2194 ini_get( 'upload_max_filesize' ) ?
: ini_get( 'hhvm.server.upload.upload_max_file_size' ),
2197 $phpMaxPostSize = wfShorthandToInteger(
2198 ini_get( 'post_max_size' ) ?
: ini_get( 'hhvm.server.max_post_size' ),
2201 return min( $phpMaxFileSize, $phpMaxPostSize );
2205 * Get the current status of a chunked upload (used for polling)
2207 * The value will be read from cache.
2210 * @param string $statusKey
2211 * @return Status[]|bool
2213 public static function getSessionStatus( User
$user, $statusKey ) {
2214 $store = self
::getUploadSessionStore();
2215 $key = self
::getUploadSessionKey( $store, $user, $statusKey );
2217 return $store->get( $key );
2221 * Set the current status of a chunked upload (used for polling)
2223 * The value will be set in cache for 1 day
2225 * Avoid triggering this method on HTTP GET/HEAD requests
2228 * @param string $statusKey
2229 * @param array|bool $value
2232 public static function setSessionStatus( User
$user, $statusKey, $value ) {
2233 $store = self
::getUploadSessionStore();
2234 $key = self
::getUploadSessionKey( $store, $user, $statusKey );
2236 if ( $value === false ) {
2237 $store->delete( $key );
2239 $store->set( $key, $value, $store::TTL_DAY
);
2244 * @param BagOStuff $store
2246 * @param string $statusKey
2249 private static function getUploadSessionKey( BagOStuff
$store, User
$user, $statusKey ) {
2250 return $store->makeKey(
2252 $user->getId() ?
: md5( $user->getName() ),
2260 private static function getUploadSessionStore() {
2261 return ObjectCache
::getInstance( 'db-replicated' );