3 * Base class for the backend of file upload.
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
23 use MediaWiki\MediaWikiServices
;
24 use MediaWiki\Shell\Shell
;
27 * @defgroup Upload Upload related
33 * UploadBase and subclasses are the backend of MediaWiki's file uploads.
34 * The frontends are formed by ApiUpload and SpecialUpload.
36 * @author Brion Vibber
37 * @author Bryan Tong Minh
38 * @author Michael Dale
40 abstract class UploadBase
{
41 /** @var string Local file system path to the file to upload (or a local copy) */
43 /** @var TempFSFile|null Wrapper to handle deleting the temp file */
44 protected $tempFileObj;
46 protected $mDesiredDestName, $mDestName, $mRemoveTempFile, $mSourceType;
47 protected $mTitle = false, $mTitleError = 0;
48 protected $mFilteredName, $mFinalExtension;
49 protected $mLocalFile, $mStashFile, $mFileSize, $mFileProps;
50 protected $mBlackListedExtensions;
51 protected $mJavaDetected, $mSVGNSError;
53 protected static $safeXmlEncodings = [
73 const MIN_LENGTH_PARTNAME
= 4;
74 const ILLEGAL_FILENAME
= 5;
75 const OVERWRITE_EXISTING_FILE
= 7; # Not used anymore; handled by verifyTitlePermissions()
76 const FILETYPE_MISSING
= 8;
77 const FILETYPE_BADTYPE
= 9;
78 const VERIFICATION_ERROR
= 10;
79 const HOOK_ABORTED
= 11;
80 const FILE_TOO_LARGE
= 12;
81 const WINDOWS_NONASCII_FILENAME
= 13;
82 const FILENAME_TOO_LONG
= 14;
88 public function getVerificationErrorCode( $error ) {
90 self
::EMPTY_FILE
=> 'empty-file',
91 self
::FILE_TOO_LARGE
=> 'file-too-large',
92 self
::FILETYPE_MISSING
=> 'filetype-missing',
93 self
::FILETYPE_BADTYPE
=> 'filetype-banned',
94 self
::MIN_LENGTH_PARTNAME
=> 'filename-tooshort',
95 self
::ILLEGAL_FILENAME
=> 'illegal-filename',
96 self
::OVERWRITE_EXISTING_FILE
=> 'overwrite',
97 self
::VERIFICATION_ERROR
=> 'verification-error',
98 self
::HOOK_ABORTED
=> 'hookaborted',
99 self
::WINDOWS_NONASCII_FILENAME
=> 'windows-nonascii-filename',
100 self
::FILENAME_TOO_LONG
=> 'filename-toolong',
102 return $code_to_status[$error] ??
'unknown-error';
106 * Returns true if uploads are enabled.
107 * Can be override by subclasses.
110 public static function isEnabled() {
111 global $wgEnableUploads;
113 if ( !$wgEnableUploads ) {
117 # Check php's file_uploads setting
118 return wfIsHHVM() ||
wfIniGetBool( 'file_uploads' );
122 * Returns true if the user can use this upload module or else a string
123 * identifying the missing permission.
124 * Can be overridden by subclasses.
127 * @return bool|string
129 public static function isAllowed( $user ) {
130 foreach ( [ 'upload', 'edit' ] as $permission ) {
131 if ( !$user->isAllowed( $permission ) ) {
140 * Returns true if the user has surpassed the upload rate limit, false otherwise.
145 public static function isThrottled( $user ) {
146 return $user->pingLimiter( 'upload' );
149 // Upload handlers. Should probably just be a global.
150 private static $uploadHandlers = [ 'Stash', 'File', 'Url' ];
153 * Create a form of UploadBase depending on wpSourceType and initializes it
155 * @param WebRequest &$request
156 * @param string|null $type
157 * @return null|UploadBase
159 public static function createFromRequest( &$request, $type = null ) {
160 $type = $type ?
: $request->getVal( 'wpSourceType', 'File' );
166 // Get the upload class
167 $type = ucfirst( $type );
169 // Give hooks the chance to handle this request
171 Hooks
::run( 'UploadCreateFromRequest', [ $type, &$className ] );
172 if ( is_null( $className ) ) {
173 $className = 'UploadFrom' . $type;
174 wfDebug( __METHOD__
. ": class name: $className\n" );
175 if ( !in_array( $type, self
::$uploadHandlers ) ) {
180 // Check whether this upload class is enabled
181 if ( !call_user_func( [ $className, 'isEnabled' ] ) ) {
185 // Check whether the request is valid
186 if ( !call_user_func( [ $className, 'isValidRequest' ], $request ) ) {
190 /** @var UploadBase $handler */
191 $handler = new $className;
193 $handler->initializeFromRequest( $request );
199 * Check whether a request if valid for this handler
200 * @param WebRequest $request
203 public static function isValidRequest( $request ) {
207 public function __construct() {
211 * Returns the upload type. Should be overridden by child classes
216 public function getSourceType() {
221 * Initialize the path information
222 * @param string $name The desired destination name
223 * @param string $tempPath The temporary path
224 * @param int $fileSize The file size
225 * @param bool $removeTempFile (false) remove the temporary file?
226 * @throws MWException
228 public function initializePathInfo( $name, $tempPath, $fileSize, $removeTempFile = false ) {
229 $this->mDesiredDestName
= $name;
230 if ( FileBackend
::isStoragePath( $tempPath ) ) {
231 throw new MWException( __METHOD__
. " given storage path `$tempPath`." );
234 $this->setTempFile( $tempPath, $fileSize );
235 $this->mRemoveTempFile
= $removeTempFile;
239 * Initialize from a WebRequest. Override this in a subclass.
241 * @param WebRequest &$request
243 abstract public function initializeFromRequest( &$request );
246 * @param string $tempPath File system path to temporary file containing the upload
247 * @param int|null $fileSize
249 protected function setTempFile( $tempPath, $fileSize = null ) {
250 $this->mTempPath
= $tempPath;
251 $this->mFileSize
= $fileSize ?
: null;
252 if ( strlen( $this->mTempPath
) && file_exists( $this->mTempPath
) ) {
253 $this->tempFileObj
= new TempFSFile( $this->mTempPath
);
255 $this->mFileSize
= filesize( $this->mTempPath
);
258 $this->tempFileObj
= null;
263 * Fetch the file. Usually a no-op
266 public function fetchFile() {
267 return Status
::newGood();
271 * Return true if the file is empty
274 public function isEmptyFile() {
275 return empty( $this->mFileSize
);
279 * Return the file size
282 public function getFileSize() {
283 return $this->mFileSize
;
287 * Get the base 36 SHA1 of the file
290 public function getTempFileSha1Base36() {
291 return FSFile
::getSha1Base36FromPath( $this->mTempPath
);
295 * @param string $srcPath The source path
296 * @return string|bool The real path if it was a virtual URL Returns false on failure
298 public function getRealPath( $srcPath ) {
299 $repo = RepoGroup
::singleton()->getLocalRepo();
300 if ( FileRepo
::isVirtualUrl( $srcPath ) ) {
301 /** @todo Just make uploads work with storage paths UploadFromStash
302 * loads files via virtual URLs.
304 $tmpFile = $repo->getLocalCopy( $srcPath );
306 $tmpFile->bind( $this ); // keep alive with $this
308 $path = $tmpFile ?
$tmpFile->getPath() : false;
317 * Verify whether the upload is sane.
318 * @return mixed Const self::OK or else an array with error information
320 public function verifyUpload() {
322 * If there was no filename or a zero size given, give up quick.
324 if ( $this->isEmptyFile() ) {
325 return [ 'status' => self
::EMPTY_FILE
];
329 * Honor $wgMaxUploadSize
331 $maxSize = self
::getMaxUploadSize( $this->getSourceType() );
332 if ( $this->mFileSize
> $maxSize ) {
334 'status' => self
::FILE_TOO_LARGE
,
340 * Look at the contents of the file; if we can recognize the
341 * type but it's corrupt or data of the wrong type, we should
342 * probably not accept it.
344 $verification = $this->verifyFile();
345 if ( $verification !== true ) {
347 'status' => self
::VERIFICATION_ERROR
,
348 'details' => $verification
353 * Make sure this file can be created
355 $result = $this->validateName();
356 if ( $result !== true ) {
361 if ( !Hooks
::run( 'UploadVerification',
362 [ $this->mDestName
, $this->mTempPath
, &$error ], '1.28' )
364 return [ 'status' => self
::HOOK_ABORTED
, 'error' => $error ];
367 return [ 'status' => self
::OK
];
371 * Verify that the name is valid and, if necessary, that we can overwrite
373 * @return mixed True if valid, otherwise and array with 'status'
376 public function validateName() {
377 $nt = $this->getTitle();
378 if ( is_null( $nt ) ) {
379 $result = [ 'status' => $this->mTitleError
];
380 if ( $this->mTitleError
== self
::ILLEGAL_FILENAME
) {
381 $result['filtered'] = $this->mFilteredName
;
383 if ( $this->mTitleError
== self
::FILETYPE_BADTYPE
) {
384 $result['finalExt'] = $this->mFinalExtension
;
385 if ( count( $this->mBlackListedExtensions
) ) {
386 $result['blacklistedExt'] = $this->mBlackListedExtensions
;
392 $this->mDestName
= $this->getLocalFile()->getName();
398 * Verify the MIME type.
400 * @note Only checks that it is not an evil MIME. The "does it have
401 * correct extension given its MIME type?" check is in verifyFile.
402 * in `verifyFile()` that MIME type and file extension correlate.
403 * @param string $mime Representing the MIME
404 * @return mixed True if the file is verified, an array otherwise
406 protected function verifyMimeType( $mime ) {
407 global $wgVerifyMimeType, $wgVerifyMimeTypeIE;
408 if ( $wgVerifyMimeType ) {
409 wfDebug( "mime: <$mime> extension: <{$this->mFinalExtension}>\n" );
410 global $wgMimeTypeBlacklist;
411 if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
412 return [ 'filetype-badmime', $mime ];
415 if ( $wgVerifyMimeTypeIE ) {
416 # Check what Internet Explorer would detect
417 $fp = fopen( $this->mTempPath
, 'rb' );
418 $chunk = fread( $fp, 256 );
421 $magic = MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer();
422 $extMime = $magic->guessTypesForExtension( $this->mFinalExtension
);
423 $ieTypes = $magic->getIEMimeTypes( $this->mTempPath
, $chunk, $extMime );
424 foreach ( $ieTypes as $ieType ) {
425 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
426 return [ 'filetype-bad-ie-mime', $ieType ];
436 * Verifies that it's ok to include the uploaded file
438 * @return mixed True of the file is verified, array otherwise.
440 protected function verifyFile() {
441 global $wgVerifyMimeType, $wgDisableUploadScriptChecks;
443 $status = $this->verifyPartialFile();
444 if ( $status !== true ) {
448 $mwProps = new MWFileProps( MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer() );
449 $this->mFileProps
= $mwProps->getPropsFromPath( $this->mTempPath
, $this->mFinalExtension
);
450 $mime = $this->mFileProps
['mime'];
452 if ( $wgVerifyMimeType ) {
453 # XXX: Missing extension will be caught by validateName() via getTitle()
454 if ( $this->mFinalExtension
!= '' && !$this->verifyExtension( $mime, $this->mFinalExtension
) ) {
455 return [ 'filetype-mime-mismatch', $this->mFinalExtension
, $mime ];
459 # check for htmlish code and javascript
460 if ( !$wgDisableUploadScriptChecks ) {
461 if ( $this->mFinalExtension
== 'svg' ||
$mime == 'image/svg+xml' ) {
462 $svgStatus = $this->detectScriptInSvg( $this->mTempPath
, false );
463 if ( $svgStatus !== false ) {
469 $handler = MediaHandler
::getHandler( $mime );
471 $handlerStatus = $handler->verifyUpload( $this->mTempPath
);
472 if ( !$handlerStatus->isOK() ) {
473 $errors = $handlerStatus->getErrorsArray();
475 return reset( $errors );
480 Hooks
::run( 'UploadVerifyFile', [ $this, $mime, &$error ] );
481 if ( $error !== true ) {
482 if ( !is_array( $error ) ) {
488 wfDebug( __METHOD__
. ": all clear; passing.\n" );
494 * A verification routine suitable for partial files
496 * Runs the blacklist checks, but not any checks that may
497 * assume the entire file is present.
499 * @return mixed True for valid or array with error message key.
501 protected function verifyPartialFile() {
502 global $wgAllowJavaUploads, $wgDisableUploadScriptChecks;
504 # getTitle() sets some internal parameters like $this->mFinalExtension
507 $mwProps = new MWFileProps( MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer() );
508 $this->mFileProps
= $mwProps->getPropsFromPath( $this->mTempPath
, $this->mFinalExtension
);
510 # check MIME type, if desired
511 $mime = $this->mFileProps
['file-mime'];
512 $status = $this->verifyMimeType( $mime );
513 if ( $status !== true ) {
517 # check for htmlish code and javascript
518 if ( !$wgDisableUploadScriptChecks ) {
519 if ( self
::detectScript( $this->mTempPath
, $mime, $this->mFinalExtension
) ) {
520 return [ 'uploadscripted' ];
522 if ( $this->mFinalExtension
== 'svg' ||
$mime == 'image/svg+xml' ) {
523 $svgStatus = $this->detectScriptInSvg( $this->mTempPath
, true );
524 if ( $svgStatus !== false ) {
530 # Check for Java applets, which if uploaded can bypass cross-site
532 if ( !$wgAllowJavaUploads ) {
533 $this->mJavaDetected
= false;
534 $zipStatus = ZipDirectoryReader
::read( $this->mTempPath
,
535 [ $this, 'zipEntryCallback' ] );
536 if ( !$zipStatus->isOK() ) {
537 $errors = $zipStatus->getErrorsArray();
538 $error = reset( $errors );
539 if ( $error[0] !== 'zip-wrong-format' ) {
543 if ( $this->mJavaDetected
) {
544 return [ 'uploadjava' ];
548 # Scan the uploaded file for viruses
549 $virus = $this->detectVirus( $this->mTempPath
);
551 return [ 'uploadvirus', $virus ];
558 * Callback for ZipDirectoryReader to detect Java class files.
560 * @param array $entry
562 public function zipEntryCallback( $entry ) {
563 $names = [ $entry['name'] ];
565 // If there is a null character, cut off the name at it, because JDK's
566 // ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name
567 // were constructed which had ".class\0" followed by a string chosen to
568 // make the hash collide with the truncated name, that file could be
569 // returned in response to a request for the .class file.
570 $nullPos = strpos( $entry['name'], "\000" );
571 if ( $nullPos !== false ) {
572 $names[] = substr( $entry['name'], 0, $nullPos );
575 // If there is a trailing slash in the file name, we have to strip it,
576 // because that's what ZIP_GetEntry() does.
577 if ( preg_grep( '!\.class/?$!', $names ) ) {
578 $this->mJavaDetected
= true;
583 * Alias for verifyTitlePermissions. The function was originally
584 * 'verifyPermissions', but that suggests it's checking the user, when it's
585 * really checking the title + user combination.
587 * @param User $user User object to verify the permissions against
588 * @return mixed An array as returned by getUserPermissionsErrors or true
589 * in case the user has proper permissions.
591 public function verifyPermissions( $user ) {
592 return $this->verifyTitlePermissions( $user );
596 * Check whether the user can edit, upload and create the image. This
597 * checks only against the current title; if it returns errors, it may
598 * very well be that another title will not give errors. Therefore
599 * isAllowed() should be called as well for generic is-user-blocked or
600 * can-user-upload checking.
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 verifyTitlePermissions( $user ) {
608 * If the image is protected, non-sysop users won't be able
609 * to modify it by uploading a new revision.
611 $nt = $this->getTitle();
612 if ( is_null( $nt ) ) {
615 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
616 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
617 if ( !$nt->exists() ) {
618 $permErrorsCreate = $nt->getUserPermissionsErrors( 'create', $user );
620 $permErrorsCreate = [];
622 if ( $permErrors ||
$permErrorsUpload ||
$permErrorsCreate ) {
623 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
624 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
629 $overwriteError = $this->checkOverwrite( $user );
630 if ( $overwriteError !== true ) {
631 return [ $overwriteError ];
638 * Check for non fatal problems with the file.
640 * This should not assume that mTempPath is set.
642 * @return mixed[] Array of warnings
644 public function checkWarnings() {
647 $localFile = $this->getLocalFile();
648 $localFile->load( File
::READ_LATEST
);
649 $filename = $localFile->getName();
650 $hash = $this->getTempFileSha1Base36();
652 $badFileName = $this->checkBadFileName( $filename, $this->mDesiredDestName
);
653 if ( $badFileName !== null ) {
654 $warnings['badfilename'] = $badFileName;
657 $unwantedFileExtensionDetails = $this->checkUnwantedFileExtensions( $this->mFinalExtension
);
658 if ( $unwantedFileExtensionDetails !== null ) {
659 $warnings['filetype-unwanted-type'] = $unwantedFileExtensionDetails;
662 $fileSizeWarnings = $this->checkFileSize( $this->mFileSize
);
663 if ( $fileSizeWarnings ) {
664 $warnings = array_merge( $warnings, $fileSizeWarnings );
667 $localFileExistsWarnings = $this->checkLocalFileExists( $localFile, $hash );
668 if ( $localFileExistsWarnings ) {
669 $warnings = array_merge( $warnings, $localFileExistsWarnings );
672 if ( $this->checkLocalFileWasDeleted( $localFile ) ) {
673 $warnings['was-deleted'] = $filename;
676 // If a file with the same name exists locally then the local file has already been tested
677 // for duplication of content
678 $ignoreLocalDupes = isset( $warnings[ 'exists '] );
679 $dupes = $this->checkAgainstExistingDupes( $hash, $ignoreLocalDupes );
681 $warnings['duplicate'] = $dupes;
684 $archivedDupes = $this->checkAgainstArchiveDupes( $hash );
685 if ( $archivedDupes !== null ) {
686 $warnings['duplicate-archive'] = $archivedDupes;
693 * Check whether the resulting filename is different from the desired one,
694 * but ignore things like ucfirst() and spaces/underscore things
696 * @param string $filename
697 * @param string $desiredFileName
699 * @return string|null String that was determined to be bad or null if the filename is okay
701 private function checkBadFileName( $filename, $desiredFileName ) {
702 $comparableName = str_replace( ' ', '_', $desiredFileName );
703 $comparableName = Title
::capitalize( $comparableName, NS_FILE
);
705 if ( $desiredFileName != $filename && $comparableName != $filename ) {
713 * @param string $fileExtension The file extension to check
715 * @return array|null array with the following keys:
716 * 0 => string The final extension being used
717 * 1 => string[] The extensions that are allowed
718 * 2 => int The number of extensions that are allowed.
720 private function checkUnwantedFileExtensions( $fileExtension ) {
721 global $wgCheckFileExtensions, $wgFileExtensions, $wgLang;
723 if ( $wgCheckFileExtensions ) {
724 $extensions = array_unique( $wgFileExtensions );
725 if ( !$this->checkFileExtension( $fileExtension, $extensions ) ) {
728 $wgLang->commaList( $extensions ),
738 * @param int $fileSize
740 * @return array warnings
742 private function checkFileSize( $fileSize ) {
743 global $wgUploadSizeWarning;
747 if ( $wgUploadSizeWarning && ( $fileSize > $wgUploadSizeWarning ) ) {
748 $warnings['large-file'] = [ $wgUploadSizeWarning, $fileSize ];
751 if ( $fileSize == 0 ) {
752 $warnings['empty-file'] = true;
759 * @param LocalFile $localFile
760 * @param string $hash sha1 hash of the file to check
762 * @return array warnings
764 private function checkLocalFileExists( LocalFile
$localFile, $hash ) {
767 $exists = self
::getExistsWarning( $localFile );
768 if ( $exists !== false ) {
769 $warnings['exists'] = $exists;
771 // check if file is an exact duplicate of current file version
772 if ( $hash === $localFile->getSha1() ) {
773 $warnings['no-change'] = $localFile;
776 // check if file is an exact duplicate of older versions of this file
777 $history = $localFile->getHistory();
778 foreach ( $history as $oldFile ) {
779 if ( $hash === $oldFile->getSha1() ) {
780 $warnings['duplicate-version'][] = $oldFile;
788 private function checkLocalFileWasDeleted( LocalFile
$localFile ) {
789 return $localFile->wasDeleted() && !$localFile->exists();
793 * @param string $hash sha1 hash of the file to check
794 * @param bool $ignoreLocalDupes True to ignore local duplicates
796 * @return File[] Duplicate files, if found.
798 private function checkAgainstExistingDupes( $hash, $ignoreLocalDupes ) {
799 $dupes = RepoGroup
::singleton()->findBySha1( $hash );
800 $title = $this->getTitle();
801 foreach ( $dupes as $key => $dupe ) {
803 ( $dupe instanceof LocalFile
) &&
805 $title->equals( $dupe->getTitle() )
807 unset( $dupes[$key] );
815 * @param string $hash sha1 hash of the file to check
817 * @return string|null Name of the dupe or empty string if discovered (depending on visibility)
818 * null if the check discovered no dupes.
820 private function checkAgainstArchiveDupes( $hash ) {
821 $archivedFile = new ArchivedFile( null, 0, '', $hash );
822 if ( $archivedFile->getID() > 0 ) {
823 if ( $archivedFile->userCan( File
::DELETED_FILE
) ) {
824 return $archivedFile->getName();
834 * Really perform the upload. Stores the file in the local repo, watches
835 * if necessary and runs the UploadComplete hook.
837 * @param string $comment
838 * @param string $pageText
839 * @param bool $watch Whether the file page should be added to user's watchlist.
840 * (This doesn't check $user's permissions.)
842 * @param string[] $tags Change tags to add to the log entry and page revision.
843 * (This doesn't check $user's permissions.)
844 * @return Status Indicating the whether the upload succeeded.
846 public function performUpload( $comment, $pageText, $watch, $user, $tags = [] ) {
847 $this->getLocalFile()->load( File
::READ_LATEST
);
848 $props = $this->mFileProps
;
851 Hooks
::run( 'UploadVerifyUpload', [ $this, $user, $props, $comment, $pageText, &$error ] );
853 if ( !is_array( $error ) ) {
856 return Status
::newFatal( ...$error );
859 $status = $this->getLocalFile()->upload(
870 if ( $status->isGood() ) {
872 WatchAction
::doWatch(
873 $this->getLocalFile()->getTitle(),
875 User
::IGNORE_USER_RIGHTS
878 // Avoid PHP 7.1 warning of passing $this by reference
880 Hooks
::run( 'UploadComplete', [ &$uploadBase ] );
882 $this->postProcessUpload();
889 * Perform extra steps after a successful upload.
893 public function postProcessUpload() {
897 * Returns the title of the file to be uploaded. Sets mTitleError in case
898 * the name was illegal.
900 * @return Title|null The title of the file or null in case the name was illegal
902 public function getTitle() {
903 if ( $this->mTitle
!== false ) {
904 return $this->mTitle
;
906 if ( !is_string( $this->mDesiredDestName
) ) {
907 $this->mTitleError
= self
::ILLEGAL_FILENAME
;
908 $this->mTitle
= null;
910 return $this->mTitle
;
912 /* Assume that if a user specified File:Something.jpg, this is an error
913 * and that the namespace prefix needs to be stripped of.
915 $title = Title
::newFromText( $this->mDesiredDestName
);
916 if ( $title && $title->getNamespace() == NS_FILE
) {
917 $this->mFilteredName
= $title->getDBkey();
919 $this->mFilteredName
= $this->mDesiredDestName
;
922 # oi_archive_name is max 255 bytes, which include a timestamp and an
923 # exclamation mark, so restrict file name to 240 bytes.
924 if ( strlen( $this->mFilteredName
) > 240 ) {
925 $this->mTitleError
= self
::FILENAME_TOO_LONG
;
926 $this->mTitle
= null;
928 return $this->mTitle
;
932 * Chop off any directories in the given filename. Then
933 * filter out illegal characters, and try to make a legible name
934 * out of it. We'll strip some silently that Title would die on.
936 $this->mFilteredName
= wfStripIllegalFilenameChars( $this->mFilteredName
);
937 /* Normalize to title form before we do any further processing */
938 $nt = Title
::makeTitleSafe( NS_FILE
, $this->mFilteredName
);
939 if ( is_null( $nt ) ) {
940 $this->mTitleError
= self
::ILLEGAL_FILENAME
;
941 $this->mTitle
= null;
943 return $this->mTitle
;
945 $this->mFilteredName
= $nt->getDBkey();
948 * We'll want to blacklist against *any* 'extension', and use
949 * only the final one for the whitelist.
951 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName
);
954 $this->mFinalExtension
= trim( end( $ext ) );
956 $this->mFinalExtension
= '';
958 # No extension, try guessing one
959 $magic = MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer();
960 $mime = $magic->guessMimeType( $this->mTempPath
);
961 if ( $mime !== 'unknown/unknown' ) {
962 # Get a space separated list of extensions
963 $extList = $magic->getExtensionsForType( $mime );
965 # Set the extension to the canonical extension
966 $this->mFinalExtension
= strtok( $extList, ' ' );
968 # Fix up the other variables
969 $this->mFilteredName
.= ".{$this->mFinalExtension}";
970 $nt = Title
::makeTitleSafe( NS_FILE
, $this->mFilteredName
);
971 $ext = [ $this->mFinalExtension
];
976 /* Don't allow users to override the blacklist (check file extension) */
977 global $wgCheckFileExtensions, $wgStrictFileExtensions;
978 global $wgFileExtensions, $wgFileBlacklist;
980 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
982 if ( $this->mFinalExtension
== '' ) {
983 $this->mTitleError
= self
::FILETYPE_MISSING
;
984 $this->mTitle
= null;
986 return $this->mTitle
;
987 } elseif ( $blackListedExtensions ||
988 ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
989 !$this->checkFileExtension( $this->mFinalExtension
, $wgFileExtensions ) )
991 $this->mBlackListedExtensions
= $blackListedExtensions;
992 $this->mTitleError
= self
::FILETYPE_BADTYPE
;
993 $this->mTitle
= null;
995 return $this->mTitle
;
998 // Windows may be broken with special characters, see T3780
999 if ( !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() )
1000 && !RepoGroup
::singleton()->getLocalRepo()->backendSupportsUnicodePaths()
1002 $this->mTitleError
= self
::WINDOWS_NONASCII_FILENAME
;
1003 $this->mTitle
= null;
1005 return $this->mTitle
;
1008 # If there was more than one "extension", reassemble the base
1009 # filename to prevent bogus complaints about length
1010 if ( count( $ext ) > 1 ) {
1011 $iterations = count( $ext ) - 1;
1012 for ( $i = 0; $i < $iterations; $i++
) {
1013 $partname .= '.' . $ext[$i];
1017 if ( strlen( $partname ) < 1 ) {
1018 $this->mTitleError
= self
::MIN_LENGTH_PARTNAME
;
1019 $this->mTitle
= null;
1021 return $this->mTitle
;
1024 $this->mTitle
= $nt;
1026 return $this->mTitle
;
1030 * Return the local file and initializes if necessary.
1032 * @return LocalFile|null
1034 public function getLocalFile() {
1035 if ( is_null( $this->mLocalFile
) ) {
1036 $nt = $this->getTitle();
1037 $this->mLocalFile
= is_null( $nt ) ?
null : wfLocalFile( $nt );
1040 return $this->mLocalFile
;
1044 * @return UploadStashFile|null
1046 public function getStashFile() {
1047 return $this->mStashFile
;
1051 * Like stashFile(), but respects extensions' wishes to prevent the stashing. verifyUpload() must
1052 * be called before calling this method (unless $isPartial is true).
1054 * Upload stash exceptions are also caught and converted to an error status.
1058 * @param bool $isPartial Pass `true` if this is a part of a chunked upload (not a complete file).
1059 * @return Status If successful, value is an UploadStashFile instance
1061 public function tryStashFile( User
$user, $isPartial = false ) {
1062 if ( !$isPartial ) {
1063 $error = $this->runUploadStashFileHook( $user );
1065 return Status
::newFatal( ...$error );
1069 $file = $this->doStashFile( $user );
1070 return Status
::newGood( $file );
1071 } catch ( UploadStashException
$e ) {
1072 return Status
::newFatal( 'uploadstash-exception', get_class( $e ), $e->getMessage() );
1078 * @return array|null Error message and parameters, null if there's no error
1080 protected function runUploadStashFileHook( User
$user ) {
1081 $props = $this->mFileProps
;
1083 Hooks
::run( 'UploadStashFile', [ $this, $user, $props, &$error ] );
1084 if ( $error && !is_array( $error ) ) {
1085 $error = [ $error ];
1091 * If the user does not supply all necessary information in the first upload
1092 * form submission (either by accident or by design) then we may want to
1093 * stash the file temporarily, get more information, and publish the file
1096 * This method will stash a file in a temporary directory for later
1097 * processing, and save the necessary descriptive info into the database.
1098 * This method returns the file object, which also has a 'fileKey' property
1099 * which can be passed through a form or API request to find this stashed
1102 * @deprecated since 1.28 Use tryStashFile() instead
1103 * @param User|null $user
1104 * @return UploadStashFile Stashed file
1105 * @throws UploadStashBadPathException
1106 * @throws UploadStashFileException
1107 * @throws UploadStashNotLoggedInException
1109 public function stashFile( User
$user = null ) {
1110 return $this->doStashFile( $user );
1114 * Implementation for stashFile() and tryStashFile().
1116 * @param User|null $user
1117 * @return UploadStashFile Stashed file
1119 protected function doStashFile( User
$user = null ) {
1120 $stash = RepoGroup
::singleton()->getLocalRepo()->getUploadStash( $user );
1121 $file = $stash->stashFile( $this->mTempPath
, $this->getSourceType() );
1122 $this->mStashFile
= $file;
1128 * Stash a file in a temporary directory, returning a key which can be used
1129 * to find the file again. See stashFile().
1131 * @deprecated since 1.28
1132 * @return string File key
1134 public function stashFileGetKey() {
1135 wfDeprecated( __METHOD__
, '1.28' );
1136 return $this->doStashFile()->getFileKey();
1140 * alias for stashFileGetKey, for backwards compatibility
1142 * @deprecated since 1.28
1143 * @return string File key
1145 public function stashSession() {
1146 wfDeprecated( __METHOD__
, '1.28' );
1147 return $this->doStashFile()->getFileKey();
1151 * If we've modified the upload file we need to manually remove it
1152 * on exit to clean up.
1154 public function cleanupTempFile() {
1155 if ( $this->mRemoveTempFile
&& $this->tempFileObj
) {
1156 // Delete when all relevant TempFSFile handles go out of scope
1157 wfDebug( __METHOD__
. ": Marked temporary file '{$this->mTempPath}' for removal\n" );
1158 $this->tempFileObj
->autocollect();
1162 public function getTempPath() {
1163 return $this->mTempPath
;
1167 * Split a file into a base name and all dot-delimited 'extensions'
1168 * on the end. Some web server configurations will fall back to
1169 * earlier pseudo-'extensions' to determine type and execute
1170 * scripts, so the blacklist needs to check them all.
1172 * @param string $filename
1173 * @return array [ string, string[] ]
1175 public static function splitExtensions( $filename ) {
1176 $bits = explode( '.', $filename );
1177 $basename = array_shift( $bits );
1179 return [ $basename, $bits ];
1183 * Perform case-insensitive match against a list of file extensions.
1184 * Returns true if the extension is in the list.
1186 * @param string $ext
1187 * @param array $list
1190 public static function checkFileExtension( $ext, $list ) {
1191 return in_array( strtolower( $ext ), $list );
1195 * Perform case-insensitive match against a list of file extensions.
1196 * Returns an array of matching extensions.
1198 * @param string[] $ext
1199 * @param string[] $list
1202 public static function checkFileExtensionList( $ext, $list ) {
1203 return array_intersect( array_map( 'strtolower', $ext ), $list );
1207 * Checks if the MIME type of the uploaded file matches the file extension.
1209 * @param string $mime The MIME type of the uploaded file
1210 * @param string $extension The filename extension that the file is to be served with
1213 public static function verifyExtension( $mime, $extension ) {
1214 $magic = MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer();
1216 if ( !$mime ||
$mime == 'unknown' ||
$mime == 'unknown/unknown' ) {
1217 if ( !$magic->isRecognizableExtension( $extension ) ) {
1218 wfDebug( __METHOD__
. ": passing file with unknown detected mime type; " .
1219 "unrecognized extension '$extension', can't verify\n" );
1223 wfDebug( __METHOD__
. ": rejecting file with unknown detected mime type; " .
1224 "recognized extension '$extension', so probably invalid file\n" );
1230 $match = $magic->isMatchingExtension( $extension, $mime );
1232 if ( $match === null ) {
1233 if ( $magic->getTypesForExtension( $extension ) !== null ) {
1234 wfDebug( __METHOD__
. ": No extension known for $mime, but we know a mime for $extension\n" );
1238 wfDebug( __METHOD__
. ": no file extension known for mime type $mime, passing file\n" );
1242 } elseif ( $match === true ) {
1243 wfDebug( __METHOD__
. ": mime type $mime matches extension $extension, passing file\n" );
1245 /** @todo If it's a bitmap, make sure PHP or ImageMagick resp. can handle it! */
1249 . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
1256 * Heuristic for detecting files that *could* contain JavaScript instructions or
1257 * things that may look like HTML to a browser and are thus
1258 * potentially harmful. The present implementation will produce false
1259 * positives in some situations.
1261 * @param string $file Pathname to the temporary upload file
1262 * @param string $mime The MIME type of the file
1263 * @param string $extension The extension of the file
1264 * @return bool True if the file contains something looking like embedded scripts
1266 public static function detectScript( $file, $mime, $extension ) {
1267 # ugly hack: for text files, always look at the entire file.
1268 # For binary field, just check the first K.
1270 $isText = strpos( $mime, 'text/' ) === 0;
1272 $chunk = file_get_contents( $file );
1274 $fp = fopen( $file, 'rb' );
1275 $chunk = fread( $fp, 1024 );
1279 $chunk = strtolower( $chunk );
1285 # decode from UTF-16 if needed (could be used for obfuscation).
1286 if ( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
1288 } elseif ( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
1295 $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
1298 $chunk = trim( $chunk );
1300 /** @todo FIXME: Convert from UTF-16 if necessary! */
1301 wfDebug( __METHOD__
. ": checking for embedded scripts and HTML stuff\n" );
1303 # check for HTML doctype
1304 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
1308 // Some browsers will interpret obscure xml encodings as UTF-8, while
1309 // PHP/expat will interpret the given encoding in the xml declaration (T49304)
1310 if ( $extension == 'svg' ||
strpos( $mime, 'image/svg' ) === 0 ) {
1311 if ( self
::checkXMLEncodingMissmatch( $file ) ) {
1316 // Quick check for HTML heuristics in old IE and Safari.
1318 // The exact heuristics IE uses are checked separately via verifyMimeType(), so we
1319 // don't need them all here as it can cause many false positives.
1321 // Check for `<script` and such still to forbid script tags and embedded HTML in SVG:
1325 '<html', # also in safari
1326 '<script', # also in safari
1329 foreach ( $tags as $tag ) {
1330 if ( strpos( $chunk, $tag ) !== false ) {
1331 wfDebug( __METHOD__
. ": found something that may make it be mistaken for html: $tag\n" );
1338 * look for JavaScript
1341 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
1342 $chunk = Sanitizer
::decodeCharReferences( $chunk );
1344 # look for script-types
1345 if ( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
1346 wfDebug( __METHOD__
. ": found script types\n" );
1351 # look for html-style script-urls
1352 if ( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1353 wfDebug( __METHOD__
. ": found html-style script urls\n" );
1358 # look for css-style script-urls
1359 if ( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1360 wfDebug( __METHOD__
. ": found css-style script urls\n" );
1365 wfDebug( __METHOD__
. ": no scripts found\n" );
1371 * Check a whitelist of xml encodings that are known not to be interpreted differently
1372 * by the server's xml parser (expat) and some common browsers.
1374 * @param string $file Pathname to the temporary upload file
1375 * @return bool True if the file contains an encoding that could be misinterpreted
1377 public static function checkXMLEncodingMissmatch( $file ) {
1378 global $wgSVGMetadataCutoff;
1379 $contents = file_get_contents( $file, false, null, 0, $wgSVGMetadataCutoff );
1380 $encodingRegex = '!encoding[ \t\n\r]*=[ \t\n\r]*[\'"](.*?)[\'"]!si';
1382 if ( preg_match( "!<\?xml\b(.*?)\?>!si", $contents, $matches ) ) {
1383 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1384 && !in_array( strtoupper( $encMatch[1] ), self
::$safeXmlEncodings )
1386 wfDebug( __METHOD__
. ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1390 } elseif ( preg_match( "!<\?xml\b!si", $contents ) ) {
1391 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1392 // bytes. There shouldn't be a legitimate reason for this to happen.
1393 wfDebug( __METHOD__
. ": Unmatched XML declaration start\n" );
1396 } elseif ( substr( $contents, 0, 4 ) == "\x4C\x6F\xA7\x94" ) {
1397 // EBCDIC encoded XML
1398 wfDebug( __METHOD__
. ": EBCDIC Encoded XML\n" );
1403 // It's possible the file is encoded with multi-byte encoding, so re-encode attempt to
1404 // detect the encoding in case is specifies an encoding not whitelisted in self::$safeXmlEncodings
1405 $attemptEncodings = [ 'UTF-16', 'UTF-16BE', 'UTF-32', 'UTF-32BE' ];
1406 foreach ( $attemptEncodings as $encoding ) {
1407 Wikimedia\
suppressWarnings();
1408 $str = iconv( $encoding, 'UTF-8', $contents );
1409 Wikimedia\restoreWarnings
();
1410 if ( $str != '' && preg_match( "!<\?xml\b(.*?)\?>!si", $str, $matches ) ) {
1411 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1412 && !in_array( strtoupper( $encMatch[1] ), self
::$safeXmlEncodings )
1414 wfDebug( __METHOD__
. ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1418 } elseif ( $str != '' && preg_match( "!<\?xml\b!si", $str ) ) {
1419 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1420 // bytes. There shouldn't be a legitimate reason for this to happen.
1421 wfDebug( __METHOD__
. ": Unmatched XML declaration start\n" );
1431 * @param string $filename
1432 * @param bool $partial
1433 * @return bool|array
1435 protected function detectScriptInSvg( $filename, $partial ) {
1436 $this->mSVGNSError
= false;
1437 $check = new XmlTypeCheck(
1439 [ $this, 'checkSvgScriptCallback' ],
1442 'processing_instruction_handler' => 'UploadBase::checkSvgPICallback',
1443 'external_dtd_handler' => 'UploadBase::checkSvgExternalDTD',
1446 if ( $check->wellFormed
!== true ) {
1447 // Invalid xml (T60553)
1448 // But only when non-partial (T67724)
1449 return $partial ?
false : [ 'uploadinvalidxml' ];
1450 } elseif ( $check->filterMatch
) {
1451 if ( $this->mSVGNSError
) {
1452 return [ 'uploadscriptednamespace', $this->mSVGNSError
];
1455 return $check->filterMatchType
;
1462 * Callback to filter SVG Processing Instructions.
1463 * @param string $target Processing instruction name
1464 * @param string $data Processing instruction attribute and value
1465 * @return bool|array
1467 public static function checkSvgPICallback( $target, $data ) {
1468 // Don't allow external stylesheets (T59550)
1469 if ( preg_match( '/xml-stylesheet/i', $target ) ) {
1470 return [ 'upload-scripted-pi-callback' ];
1477 * Verify that DTD urls referenced are only the standard dtds
1479 * Browsers seem to ignore external dtds. However just to be on the
1480 * safe side, only allow dtds from the svg standard.
1482 * @param string $type PUBLIC or SYSTEM
1483 * @param string $publicId The well-known public identifier for the dtd
1484 * @param string $systemId The url for the external dtd
1485 * @return bool|array
1487 public static function checkSvgExternalDTD( $type, $publicId, $systemId ) {
1488 // This doesn't include the XHTML+MathML+SVG doctype since we don't
1489 // allow XHTML anyways.
1491 'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd',
1492 'http://www.w3.org/TR/2001/REC-SVG-20010904/DTD/svg10.dtd',
1493 'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-basic.dtd',
1494 'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-tiny.dtd',
1495 // https://phabricator.wikimedia.org/T168856
1496 'http://www.w3.org/TR/2001/PR-SVG-20010719/DTD/svg10.dtd',
1498 if ( $type !== 'PUBLIC'
1499 ||
!in_array( $systemId, $allowedDTDs )
1500 ||
strpos( $publicId, "-//W3C//" ) !== 0
1502 return [ 'upload-scripted-dtd' ];
1508 * @todo Replace this with a whitelist filter!
1509 * @param string $element
1510 * @param array $attribs
1511 * @param array|null $data
1512 * @return bool|array
1514 public function checkSvgScriptCallback( $element, $attribs, $data = null ) {
1515 list( $namespace, $strippedElement ) = $this->splitXmlNamespace( $element );
1517 // We specifically don't include:
1518 // http://www.w3.org/1999/xhtml (T62771)
1519 static $validNamespaces = [
1522 'http://creativecommons.org/ns#',
1523 'http://inkscape.sourceforge.net/dtd/sodipodi-0.dtd',
1524 'http://ns.adobe.com/adobeillustrator/10.0/',
1525 'http://ns.adobe.com/adobesvgviewerextensions/3.0/',
1526 'http://ns.adobe.com/extensibility/1.0/',
1527 'http://ns.adobe.com/flows/1.0/',
1528 'http://ns.adobe.com/illustrator/1.0/',
1529 'http://ns.adobe.com/imagereplacement/1.0/',
1530 'http://ns.adobe.com/pdf/1.3/',
1531 'http://ns.adobe.com/photoshop/1.0/',
1532 'http://ns.adobe.com/saveforweb/1.0/',
1533 'http://ns.adobe.com/variables/1.0/',
1534 'http://ns.adobe.com/xap/1.0/',
1535 'http://ns.adobe.com/xap/1.0/g/',
1536 'http://ns.adobe.com/xap/1.0/g/img/',
1537 'http://ns.adobe.com/xap/1.0/mm/',
1538 'http://ns.adobe.com/xap/1.0/rights/',
1539 'http://ns.adobe.com/xap/1.0/stype/dimensions#',
1540 'http://ns.adobe.com/xap/1.0/stype/font#',
1541 'http://ns.adobe.com/xap/1.0/stype/manifestitem#',
1542 'http://ns.adobe.com/xap/1.0/stype/resourceevent#',
1543 'http://ns.adobe.com/xap/1.0/stype/resourceref#',
1544 'http://ns.adobe.com/xap/1.0/t/pg/',
1545 'http://purl.org/dc/elements/1.1/',
1546 'http://purl.org/dc/elements/1.1',
1547 'http://schemas.microsoft.com/visio/2003/svgextensions/',
1548 'http://sodipodi.sourceforge.net/dtd/sodipodi-0.dtd',
1549 'http://taptrix.com/inkpad/svg_extensions',
1550 'http://web.resource.org/cc/',
1551 'http://www.freesoftware.fsf.org/bkchem/cdml',
1552 'http://www.inkscape.org/namespaces/inkscape',
1553 'http://www.opengis.net/gml',
1554 'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
1555 'http://www.w3.org/2000/svg',
1556 'http://www.w3.org/tr/rec-rdf-syntax/',
1557 'http://www.w3.org/2000/01/rdf-schema#',
1560 // Inkscape mangles namespace definitions created by Adobe Illustrator.
1561 // This is nasty but harmless. (T144827)
1562 $isBuggyInkscape = preg_match( '/^&(#38;)*ns_[a-z_]+;$/', $namespace );
1564 if ( !( $isBuggyInkscape ||
in_array( $namespace, $validNamespaces ) ) ) {
1565 wfDebug( __METHOD__
. ": Non-svg namespace '$namespace' in uploaded file.\n" );
1566 /** @todo Return a status object to a closure in XmlTypeCheck, for MW1.21+ */
1567 $this->mSVGNSError
= $namespace;
1573 * check for elements that can contain javascript
1575 if ( $strippedElement == 'script' ) {
1576 wfDebug( __METHOD__
. ": Found script element '$element' in uploaded file.\n" );
1578 return [ 'uploaded-script-svg', $strippedElement ];
1581 # e.g., <svg xmlns="http://www.w3.org/2000/svg">
1582 # <handler xmlns:ev="http://www.w3.org/2001/xml-events" ev:event="load">alert(1)</handler> </svg>
1583 if ( $strippedElement == 'handler' ) {
1584 wfDebug( __METHOD__
. ": Found scriptable element '$element' in uploaded file.\n" );
1586 return [ 'uploaded-script-svg', $strippedElement ];
1589 # SVG reported in Feb '12 that used xml:stylesheet to generate javascript block
1590 if ( $strippedElement == 'stylesheet' ) {
1591 wfDebug( __METHOD__
. ": Found scriptable element '$element' in uploaded file.\n" );
1593 return [ 'uploaded-script-svg', $strippedElement ];
1596 # Block iframes, in case they pass the namespace check
1597 if ( $strippedElement == 'iframe' ) {
1598 wfDebug( __METHOD__
. ": iframe in uploaded file.\n" );
1600 return [ 'uploaded-script-svg', $strippedElement ];
1604 if ( $strippedElement == 'style'
1605 && self
::checkCssFragment( Sanitizer
::normalizeCss( $data ) )
1607 wfDebug( __METHOD__
. ": hostile css in style element.\n" );
1608 return [ 'uploaded-hostile-svg' ];
1611 foreach ( $attribs as $attrib => $value ) {
1612 $stripped = $this->stripXmlNamespace( $attrib );
1613 $value = strtolower( $value );
1615 if ( substr( $stripped, 0, 2 ) == 'on' ) {
1617 . ": Found event-handler attribute '$attrib'='$value' in uploaded file.\n" );
1619 return [ 'uploaded-event-handler-on-svg', $attrib, $value ];
1622 # Do not allow relative links, or unsafe url schemas.
1623 # For <a> tags, only data:, http: and https: and same-document
1624 # fragment links are allowed. For all other tags, only data:
1625 # and fragment are allowed.
1626 if ( $stripped == 'href'
1628 && strpos( $value, 'data:' ) !== 0
1629 && strpos( $value, '#' ) !== 0
1631 if ( !( $strippedElement === 'a'
1632 && preg_match( '!^https?://!i', $value ) )
1634 wfDebug( __METHOD__
. ": Found href attribute <$strippedElement "
1635 . "'$attrib'='$value' in uploaded file.\n" );
1637 return [ 'uploaded-href-attribute-svg', $strippedElement, $attrib, $value ];
1641 # only allow data: targets that should be safe. This prevents vectors like,
1642 # image/svg, text/xml, application/xml, and text/html, which can contain scripts
1643 if ( $stripped == 'href' && strncasecmp( 'data:', $value, 5 ) === 0 ) {
1644 // rfc2397 parameters. This is only slightly slower than (;[\w;]+)*.
1645 // phpcs:ignore Generic.Files.LineLength
1646 $parameters = '(?>;[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+=(?>[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+|"(?>[\0-\x0c\x0e-\x21\x23-\x5b\x5d-\x7f]+|\\\\[\0-\x7f])*"))*(?:;base64)?';
1648 if ( !preg_match( "!^data:\s*image/(gif|jpeg|jpg|png)$parameters,!i", $value ) ) {
1649 wfDebug( __METHOD__
. ": Found href to unwhitelisted data: uri "
1650 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1651 return [ 'uploaded-href-unsafe-target-svg', $strippedElement, $attrib, $value ];
1655 # Change href with animate from (http://html5sec.org/#137).
1656 if ( $stripped === 'attributename'
1657 && $strippedElement === 'animate'
1658 && $this->stripXmlNamespace( $value ) == 'href'
1660 wfDebug( __METHOD__
. ": Found animate that might be changing href using from "
1661 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1663 return [ 'uploaded-animate-svg', $strippedElement, $attrib, $value ];
1666 # use set/animate to add event-handler attribute to parent
1667 if ( ( $strippedElement == 'set' ||
$strippedElement == 'animate' )
1668 && $stripped == 'attributename'
1669 && substr( $value, 0, 2 ) == 'on'
1671 wfDebug( __METHOD__
. ": Found svg setting event-handler attribute with "
1672 . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1674 return [ 'uploaded-setting-event-handler-svg', $strippedElement, $stripped, $value ];
1677 # use set to add href attribute to parent element
1678 if ( $strippedElement == 'set'
1679 && $stripped == 'attributename'
1680 && strpos( $value, 'href' ) !== false
1682 wfDebug( __METHOD__
. ": Found svg setting href attribute '$value' in uploaded file.\n" );
1684 return [ 'uploaded-setting-href-svg' ];
1687 # use set to add a remote / data / script target to an element
1688 if ( $strippedElement == 'set'
1689 && $stripped == 'to'
1690 && preg_match( '!(http|https|data|script):!sim', $value )
1692 wfDebug( __METHOD__
. ": Found svg setting attribute to '$value' in uploaded file.\n" );
1694 return [ 'uploaded-wrong-setting-svg', $value ];
1697 # use handler attribute with remote / data / script
1698 if ( $stripped == 'handler' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1699 wfDebug( __METHOD__
. ": Found svg setting handler with remote/data/script "
1700 . "'$attrib'='$value' in uploaded file.\n" );
1702 return [ 'uploaded-setting-handler-svg', $attrib, $value ];
1705 # use CSS styles to bring in remote code
1706 if ( $stripped == 'style'
1707 && self
::checkCssFragment( Sanitizer
::normalizeCss( $value ) )
1709 wfDebug( __METHOD__
. ": Found svg setting a style with "
1710 . "remote url '$attrib'='$value' in uploaded file.\n" );
1711 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1714 # Several attributes can include css, css character escaping isn't allowed
1715 $cssAttrs = [ 'font', 'clip-path', 'fill', 'filter', 'marker',
1716 'marker-end', 'marker-mid', 'marker-start', 'mask', 'stroke' ];
1717 if ( in_array( $stripped, $cssAttrs )
1718 && self
::checkCssFragment( $value )
1720 wfDebug( __METHOD__
. ": Found svg setting a style with "
1721 . "remote url '$attrib'='$value' in uploaded file.\n" );
1722 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1725 # image filters can pull in url, which could be svg that executes scripts
1726 # Only allow url( "#foo" ). Do not allow url( http://example.com )
1727 if ( $strippedElement == 'image'
1728 && $stripped == 'filter'
1729 && preg_match( '!url\s*\(\s*["\']?[^#]!sim', $value )
1731 wfDebug( __METHOD__
. ": Found image filter with url: "
1732 . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1734 return [ 'uploaded-image-filter-svg', $strippedElement, $stripped, $value ];
1738 return false; // No scripts detected
1742 * Check a block of CSS or CSS fragment for anything that looks like
1743 * it is bringing in remote code.
1744 * @param string $value a string of CSS
1745 * @param bool $propOnly only check css properties (start regex with :)
1746 * @return bool true if the CSS contains an illegal string, false if otherwise
1748 private static function checkCssFragment( $value ) {
1749 # Forbid external stylesheets, for both reliability and to protect viewer's privacy
1750 if ( stripos( $value, '@import' ) !== false ) {
1754 # We allow @font-face to embed fonts with data: urls, so we snip the string
1755 # 'url' out so this case won't match when we check for urls below
1756 $pattern = '!(@font-face\s*{[^}]*src:)url(\("data:;base64,)!im';
1757 $value = preg_replace( $pattern, '$1$2', $value );
1759 # Check for remote and executable CSS. Unlike in Sanitizer::checkCss, the CSS
1760 # properties filter and accelerator don't seem to be useful for xss in SVG files.
1761 # Expression and -o-link don't seem to work either, but filtering them here in case.
1762 # Additionally, we catch remote urls like url("http:..., url('http:..., url(http:...,
1763 # but not local ones such as url("#..., url('#..., url(#....
1764 if ( preg_match( '!expression
1766 | -o-link-source\s*:
1767 | -o-replace\s*:!imx', $value ) ) {
1771 if ( preg_match_all(
1772 "!(\s*(url|image|image-set)\s*\(\s*[\"']?\s*[^#]+.*?\))!sim",
1777 # TODO: redo this in one regex. Until then, url("#whatever") matches the first
1778 foreach ( $matches[1] as $match ) {
1779 if ( !preg_match( "!\s*(url|image|image-set)\s*\(\s*(#|'#|\"#)!im", $match ) ) {
1785 if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ) {
1793 * Divide the element name passed by the xml parser to the callback into URI and prifix.
1794 * @param string $element
1795 * @return array Containing the namespace URI and prefix
1797 private static function splitXmlNamespace( $element ) {
1798 // 'http://www.w3.org/2000/svg:script' -> [ 'http://www.w3.org/2000/svg', 'script' ]
1799 $parts = explode( ':', strtolower( $element ) );
1800 $name = array_pop( $parts );
1801 $ns = implode( ':', $parts );
1803 return [ $ns, $name ];
1807 * @param string $name
1810 private function stripXmlNamespace( $name ) {
1811 // 'http://www.w3.org/2000/svg:script' -> 'script'
1812 $parts = explode( ':', strtolower( $name ) );
1814 return array_pop( $parts );
1818 * Generic wrapper function for a virus scanner program.
1819 * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
1820 * $wgAntivirusRequired may be used to deny upload if the scan fails.
1822 * @param string $file Pathname to the temporary upload file
1823 * @return bool|null|string False if not virus is found, null if the scan fails or is disabled,
1824 * or a string containing feedback from the virus scanner if a virus was found.
1825 * If textual feedback is missing but a virus was found, this function returns true.
1827 public static function detectVirus( $file ) {
1828 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
1830 if ( !$wgAntivirus ) {
1831 wfDebug( __METHOD__
. ": virus scanner disabled\n" );
1836 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
1837 wfDebug( __METHOD__
. ": unknown virus scanner: $wgAntivirus\n" );
1838 $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
1839 [ 'virus-badscanner', $wgAntivirus ] );
1841 return wfMessage( 'virus-unknownscanner' )->text() . " $wgAntivirus";
1844 # look up scanner configuration
1845 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
1846 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
1847 $msgPattern = $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ??
null;
1849 if ( strpos( $command, "%f" ) === false ) {
1850 # simple pattern: append file to scan
1851 $command .= " " . Shell
::escape( $file );
1853 # complex pattern: replace "%f" with file to scan
1854 $command = str_replace( "%f", Shell
::escape( $file ), $command );
1857 wfDebug( __METHOD__
. ": running virus scan: $command \n" );
1859 # execute virus scanner
1862 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1863 # that does not seem to be worth the pain.
1864 # Ask me (Duesentrieb) about it if it's ever needed.
1865 $output = wfShellExecWithStderr( $command, $exitCode );
1867 # map exit code to AV_xxx constants.
1868 $mappedCode = $exitCode;
1869 if ( $exitCodeMap ) {
1870 if ( isset( $exitCodeMap[$exitCode] ) ) {
1871 $mappedCode = $exitCodeMap[$exitCode];
1872 } elseif ( isset( $exitCodeMap["*"] ) ) {
1873 $mappedCode = $exitCodeMap["*"];
1877 /* NB: AV_NO_VIRUS is 0 but AV_SCAN_FAILED is false,
1878 * so we need the strict equalities === and thus can't use a switch here
1880 if ( $mappedCode === AV_SCAN_FAILED
) {
1881 # scan failed (code was mapped to false by $exitCodeMap)
1882 wfDebug( __METHOD__
. ": failed to scan $file (code $exitCode).\n" );
1884 $output = $wgAntivirusRequired
1885 ?
wfMessage( 'virus-scanfailed', [ $exitCode ] )->text()
1887 } elseif ( $mappedCode === AV_SCAN_ABORTED
) {
1888 # scan failed because filetype is unknown (probably imune)
1889 wfDebug( __METHOD__
. ": unsupported file type $file (code $exitCode).\n" );
1891 } elseif ( $mappedCode === AV_NO_VIRUS
) {
1893 wfDebug( __METHOD__
. ": file passed virus scan.\n" );
1896 $output = trim( $output );
1899 $output = true; # if there's no output, return true
1900 } elseif ( $msgPattern ) {
1902 if ( preg_match( $msgPattern, $output, $groups ) && $groups[1] ) {
1903 $output = $groups[1];
1907 wfDebug( __METHOD__
. ": FOUND VIRUS! scanner feedback: $output \n" );
1914 * Check if there's an overwrite conflict and, if so, if restrictions
1915 * forbid this user from performing the upload.
1919 * @return bool|array
1921 private function checkOverwrite( $user ) {
1922 // First check whether the local file can be overwritten
1923 $file = $this->getLocalFile();
1924 $file->load( File
::READ_LATEST
);
1925 if ( $file->exists() ) {
1926 if ( !self
::userCanReUpload( $user, $file ) ) {
1927 return [ 'fileexists-forbidden', $file->getName() ];
1933 /* Check shared conflicts: if the local file does not exist, but
1934 * wfFindFile finds a file, it exists in a shared repository.
1936 $file = wfFindFile( $this->getTitle(), [ 'latest' => true ] );
1937 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1938 return [ 'fileexists-shared-forbidden', $file->getName() ];
1945 * Check if a user is the last uploader
1951 public static function userCanReUpload( User
$user, File
$img ) {
1952 if ( $user->isAllowed( 'reupload' ) ) {
1953 return true; // non-conditional
1954 } elseif ( !$user->isAllowed( 'reupload-own' ) ) {
1958 if ( !( $img instanceof LocalFile
) ) {
1964 return $user->getId() == $img->getUser( 'id' );
1968 * Helper function that does various existence checks for a file.
1969 * The following checks are performed:
1971 * - Article with the same name as the file exists
1972 * - File exists with normalized extension
1973 * - The file looks like a thumbnail and the original exists
1975 * @param File $file The File object to check
1976 * @return mixed False if the file does not exists, else an array
1978 public static function getExistsWarning( $file ) {
1979 if ( $file->exists() ) {
1980 return [ 'warning' => 'exists', 'file' => $file ];
1983 if ( $file->getTitle()->getArticleID() ) {
1984 return [ 'warning' => 'page-exists', 'file' => $file ];
1987 if ( strpos( $file->getName(), '.' ) == false ) {
1988 $partname = $file->getName();
1991 $n = strrpos( $file->getName(), '.' );
1992 $extension = substr( $file->getName(), $n +
1 );
1993 $partname = substr( $file->getName(), 0, $n );
1995 $normalizedExtension = File
::normalizeExtension( $extension );
1997 if ( $normalizedExtension != $extension ) {
1998 // We're not using the normalized form of the extension.
1999 // Normal form is lowercase, using most common of alternate
2000 // extensions (eg 'jpg' rather than 'JPEG').
2002 // Check for another file using the normalized form...
2003 $nt_lc = Title
::makeTitle( NS_FILE
, "{$partname}.{$normalizedExtension}" );
2004 $file_lc = wfLocalFile( $nt_lc );
2006 if ( $file_lc->exists() ) {
2008 'warning' => 'exists-normalized',
2010 'normalizedFile' => $file_lc
2015 // Check for files with the same name but a different extension
2016 $similarFiles = RepoGroup
::singleton()->getLocalRepo()->findFilesByPrefix(
2017 "{$partname}.", 1 );
2018 if ( count( $similarFiles ) ) {
2020 'warning' => 'exists-normalized',
2022 'normalizedFile' => $similarFiles[0],
2026 if ( self
::isThumbName( $file->getName() ) ) {
2027 # Check for filenames like 50px- or 180px-, these are mostly thumbnails
2028 $nt_thb = Title
::newFromText(
2029 substr( $partname, strpos( $partname, '-' ) +
1 ) . '.' . $extension,
2032 $file_thb = wfLocalFile( $nt_thb );
2033 if ( $file_thb->exists() ) {
2035 'warning' => 'thumb',
2037 'thumbFile' => $file_thb
2040 // File does not exist, but we just don't like the name
2042 'warning' => 'thumb-name',
2044 'thumbFile' => $file_thb
2049 foreach ( self
::getFilenamePrefixBlacklist() as $prefix ) {
2050 if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
2052 'warning' => 'bad-prefix',
2063 * Helper function that checks whether the filename looks like a thumbnail
2064 * @param string $filename
2067 public static function isThumbName( $filename ) {
2068 $n = strrpos( $filename, '.' );
2069 $partname = $n ?
substr( $filename, 0, $n ) : $filename;
2072 substr( $partname, 3, 3 ) == 'px-' ||
2073 substr( $partname, 2, 3 ) == 'px-'
2075 preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) );
2079 * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
2081 * @return array List of prefixes
2083 public static function getFilenamePrefixBlacklist() {
2085 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
2086 if ( !$message->isDisabled() ) {
2087 $lines = explode( "\n", $message->plain() );
2088 foreach ( $lines as $line ) {
2089 // Remove comment lines
2090 $comment = substr( trim( $line ), 0, 1 );
2091 if ( $comment == '#' ||
$comment == '' ) {
2094 // Remove additional comments after a prefix
2095 $comment = strpos( $line, '#' );
2096 if ( $comment > 0 ) {
2097 $line = substr( $line, 0, $comment - 1 );
2099 $blacklist[] = trim( $line );
2107 * Gets image info about the file just uploaded.
2109 * Also has the effect of setting metadata to be an 'indexed tag name' in
2110 * returned API result if 'metadata' was requested. Oddly, we have to pass
2111 * the "result" object down just so it can do that with the appropriate
2112 * format, presumably.
2114 * @param ApiResult $result
2115 * @return array Image info
2117 public function getImageInfo( $result ) {
2118 $localFile = $this->getLocalFile();
2119 $stashFile = $this->getStashFile();
2120 // Calling a different API module depending on whether the file was stashed is less than optimal.
2121 // In fact, calling API modules here at all is less than optimal. Maybe it should be refactored.
2123 $imParam = ApiQueryStashImageInfo
::getPropertyNames();
2124 $info = ApiQueryStashImageInfo
::getInfo( $stashFile, array_flip( $imParam ), $result );
2126 $imParam = ApiQueryImageInfo
::getPropertyNames();
2127 $info = ApiQueryImageInfo
::getInfo( $localFile, array_flip( $imParam ), $result );
2134 * @param array $error
2137 public function convertVerifyErrorToStatus( $error ) {
2138 $code = $error['status'];
2139 unset( $code['status'] );
2141 return Status
::newFatal( $this->getVerificationErrorCode( $code ), $error );
2145 * Get the MediaWiki maximum uploaded file size for given type of upload, based on
2148 * @param null|string $forType
2151 public static function getMaxUploadSize( $forType = null ) {
2152 global $wgMaxUploadSize;
2154 if ( is_array( $wgMaxUploadSize ) ) {
2155 if ( !is_null( $forType ) && isset( $wgMaxUploadSize[$forType] ) ) {
2156 return $wgMaxUploadSize[$forType];
2158 return $wgMaxUploadSize['*'];
2161 return intval( $wgMaxUploadSize );
2166 * Get the PHP maximum uploaded file size, based on ini settings. If there is no limit or the
2167 * limit can't be guessed, returns a very large number (PHP_INT_MAX).
2172 public static function getMaxPhpUploadSize() {
2173 $phpMaxFileSize = wfShorthandToInteger(
2174 ini_get( 'upload_max_filesize' ) ?
: ini_get( 'hhvm.server.upload.upload_max_file_size' ),
2177 $phpMaxPostSize = wfShorthandToInteger(
2178 ini_get( 'post_max_size' ) ?
: ini_get( 'hhvm.server.max_post_size' ),
2181 return min( $phpMaxFileSize, $phpMaxPostSize );
2185 * Get the current status of a chunked upload (used for polling)
2187 * The value will be read from cache.
2190 * @param string $statusKey
2191 * @return Status[]|bool
2193 public static function getSessionStatus( User
$user, $statusKey ) {
2194 $cache = MediaWikiServices
::getInstance()->getMainObjectStash();
2195 $key = $cache->makeKey( 'uploadstatus', $user->getId() ?
: md5( $user->getName() ), $statusKey );
2197 return $cache->get( $key );
2201 * Set the current status of a chunked upload (used for polling)
2203 * The value will be set in cache for 1 day
2206 * @param string $statusKey
2207 * @param array|bool $value
2210 public static function setSessionStatus( User
$user, $statusKey, $value ) {
2211 $cache = MediaWikiServices
::getInstance()->getMainObjectStash();
2212 $key = $cache->makeKey( 'uploadstatus', $user->getId() ?
: md5( $user->getName() ), $statusKey );
2214 if ( $value === false ) {
2215 $cache->delete( $key );
2217 $cache->set( $key, $value, $cache::TTL_DAY
);