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
;
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;
45 protected $mDesiredDestName, $mDestName, $mRemoveTempFile, $mSourceType;
46 protected $mTitle = false, $mTitleError = 0;
47 protected $mFilteredName, $mFinalExtension;
48 protected $mLocalFile, $mStashFile, $mFileSize, $mFileProps;
49 protected $mBlackListedExtensions;
50 protected $mJavaDetected, $mSVGNSError;
52 protected static $safeXmlEncodings = [
72 const MIN_LENGTH_PARTNAME
= 4;
73 const ILLEGAL_FILENAME
= 5;
74 const OVERWRITE_EXISTING_FILE
= 7; # Not used anymore; handled by verifyTitlePermissions()
75 const FILETYPE_MISSING
= 8;
76 const FILETYPE_BADTYPE
= 9;
77 const VERIFICATION_ERROR
= 10;
78 const HOOK_ABORTED
= 11;
79 const FILE_TOO_LARGE
= 12;
80 const WINDOWS_NONASCII_FILENAME
= 13;
81 const FILENAME_TOO_LONG
= 14;
87 public function getVerificationErrorCode( $error ) {
89 self
::EMPTY_FILE
=> 'empty-file',
90 self
::FILE_TOO_LARGE
=> 'file-too-large',
91 self
::FILETYPE_MISSING
=> 'filetype-missing',
92 self
::FILETYPE_BADTYPE
=> 'filetype-banned',
93 self
::MIN_LENGTH_PARTNAME
=> 'filename-tooshort',
94 self
::ILLEGAL_FILENAME
=> 'illegal-filename',
95 self
::OVERWRITE_EXISTING_FILE
=> 'overwrite',
96 self
::VERIFICATION_ERROR
=> 'verification-error',
97 self
::HOOK_ABORTED
=> 'hookaborted',
98 self
::WINDOWS_NONASCII_FILENAME
=> 'windows-nonascii-filename',
99 self
::FILENAME_TOO_LONG
=> 'filename-toolong',
101 if ( isset( $code_to_status[$error] ) ) {
102 return $code_to_status[$error];
105 return 'unknown-error';
109 * Returns true if uploads are enabled.
110 * Can be override by subclasses.
113 public static function isEnabled() {
114 global $wgEnableUploads;
116 if ( !$wgEnableUploads ) {
120 # Check php's file_uploads setting
121 return wfIsHHVM() ||
wfIniGetBool( 'file_uploads' );
125 * Returns true if the user can use this upload module or else a string
126 * identifying the missing permission.
127 * Can be overridden by subclasses.
130 * @return bool|string
132 public static function isAllowed( $user ) {
133 foreach ( [ 'upload', 'edit' ] as $permission ) {
134 if ( !$user->isAllowed( $permission ) ) {
143 * Returns true if the user has surpassed the upload rate limit, false otherwise.
148 public static function isThrottled( $user ) {
149 return $user->pingLimiter( 'upload' );
152 // Upload handlers. Should probably just be a global.
153 private static $uploadHandlers = [ 'Stash', 'File', 'Url' ];
156 * Create a form of UploadBase depending on wpSourceType and initializes it
158 * @param WebRequest &$request
159 * @param string|null $type
160 * @return null|UploadBase
162 public static function createFromRequest( &$request, $type = null ) {
163 $type = $type ?
$type : $request->getVal( 'wpSourceType', 'File' );
169 // Get the upload class
170 $type = ucfirst( $type );
172 // Give hooks the chance to handle this request
174 Hooks
::run( 'UploadCreateFromRequest', [ $type, &$className ] );
175 if ( is_null( $className ) ) {
176 $className = 'UploadFrom' . $type;
177 wfDebug( __METHOD__
. ": class name: $className\n" );
178 if ( !in_array( $type, self
::$uploadHandlers ) ) {
183 // Check whether this upload class is enabled
184 if ( !call_user_func( [ $className, 'isEnabled' ] ) ) {
188 // Check whether the request is valid
189 if ( !call_user_func( [ $className, 'isValidRequest' ], $request ) ) {
193 /** @var UploadBase $handler */
194 $handler = new $className;
196 $handler->initializeFromRequest( $request );
202 * Check whether a request if valid for this handler
203 * @param WebRequest $request
206 public static function isValidRequest( $request ) {
210 public function __construct() {
214 * Returns the upload type. Should be overridden by child classes
219 public function getSourceType() {
224 * Initialize the path information
225 * @param string $name The desired destination name
226 * @param string $tempPath The temporary path
227 * @param int $fileSize The file size
228 * @param bool $removeTempFile (false) remove the temporary file?
229 * @throws MWException
231 public function initializePathInfo( $name, $tempPath, $fileSize, $removeTempFile = false ) {
232 $this->mDesiredDestName
= $name;
233 if ( FileBackend
::isStoragePath( $tempPath ) ) {
234 throw new MWException( __METHOD__
. " given storage path `$tempPath`." );
237 $this->setTempFile( $tempPath, $fileSize );
238 $this->mRemoveTempFile
= $removeTempFile;
242 * Initialize from a WebRequest. Override this in a subclass.
244 * @param WebRequest &$request
246 abstract public function initializeFromRequest( &$request );
249 * @param string $tempPath File system path to temporary file containing the upload
250 * @param int $fileSize
252 protected function setTempFile( $tempPath, $fileSize = null ) {
253 $this->mTempPath
= $tempPath;
254 $this->mFileSize
= $fileSize ?
: null;
255 if ( strlen( $this->mTempPath
) && file_exists( $this->mTempPath
) ) {
256 $this->tempFileObj
= new TempFSFile( $this->mTempPath
);
258 $this->mFileSize
= filesize( $this->mTempPath
);
261 $this->tempFileObj
= null;
266 * Fetch the file. Usually a no-op
269 public function fetchFile() {
270 return Status
::newGood();
274 * Return true if the file is empty
277 public function isEmptyFile() {
278 return empty( $this->mFileSize
);
282 * Return the file size
285 public function getFileSize() {
286 return $this->mFileSize
;
290 * Get the base 36 SHA1 of the file
293 public function getTempFileSha1Base36() {
294 return FSFile
::getSha1Base36FromPath( $this->mTempPath
);
298 * @param string $srcPath The source path
299 * @return string|bool The real path if it was a virtual URL Returns false on failure
301 public function getRealPath( $srcPath ) {
302 $repo = RepoGroup
::singleton()->getLocalRepo();
303 if ( $repo->isVirtualUrl( $srcPath ) ) {
304 /** @todo Just make uploads work with storage paths UploadFromStash
305 * loads files via virtual URLs.
307 $tmpFile = $repo->getLocalCopy( $srcPath );
309 $tmpFile->bind( $this ); // keep alive with $this
311 $path = $tmpFile ?
$tmpFile->getPath() : false;
320 * Verify whether the upload is sane.
321 * @return mixed Const self::OK or else an array with error information
323 public function verifyUpload() {
325 * If there was no filename or a zero size given, give up quick.
327 if ( $this->isEmptyFile() ) {
328 return [ 'status' => self
::EMPTY_FILE
];
332 * Honor $wgMaxUploadSize
334 $maxSize = self
::getMaxUploadSize( $this->getSourceType() );
335 if ( $this->mFileSize
> $maxSize ) {
337 'status' => self
::FILE_TOO_LARGE
,
343 * Look at the contents of the file; if we can recognize the
344 * type but it's corrupt or data of the wrong type, we should
345 * probably not accept it.
347 $verification = $this->verifyFile();
348 if ( $verification !== true ) {
350 'status' => self
::VERIFICATION_ERROR
,
351 'details' => $verification
356 * Make sure this file can be created
358 $result = $this->validateName();
359 if ( $result !== true ) {
364 if ( !Hooks
::run( 'UploadVerification',
365 [ $this->mDestName
, $this->mTempPath
, &$error ], '1.28' )
367 return [ 'status' => self
::HOOK_ABORTED
, 'error' => $error ];
370 return [ 'status' => self
::OK
];
374 * Verify that the name is valid and, if necessary, that we can overwrite
376 * @return mixed True if valid, otherwise and array with 'status'
379 public function validateName() {
380 $nt = $this->getTitle();
381 if ( is_null( $nt ) ) {
382 $result = [ 'status' => $this->mTitleError
];
383 if ( $this->mTitleError
== self
::ILLEGAL_FILENAME
) {
384 $result['filtered'] = $this->mFilteredName
;
386 if ( $this->mTitleError
== self
::FILETYPE_BADTYPE
) {
387 $result['finalExt'] = $this->mFinalExtension
;
388 if ( count( $this->mBlackListedExtensions
) ) {
389 $result['blacklistedExt'] = $this->mBlackListedExtensions
;
395 $this->mDestName
= $this->getLocalFile()->getName();
401 * Verify the MIME type.
403 * @note Only checks that it is not an evil MIME. The "does it have
404 * correct extension given its MIME type?" check is in verifyFile.
405 * in `verifyFile()` that MIME type and file extension correlate.
406 * @param string $mime Representing the MIME
407 * @return mixed True if the file is verified, an array otherwise
409 protected function verifyMimeType( $mime ) {
410 global $wgVerifyMimeType;
411 if ( $wgVerifyMimeType ) {
412 wfDebug( "mime: <$mime> extension: <{$this->mFinalExtension}>\n" );
413 global $wgMimeTypeBlacklist;
414 if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
415 return [ 'filetype-badmime', $mime ];
418 # Check what Internet Explorer would detect
419 $fp = fopen( $this->mTempPath
, 'rb' );
420 $chunk = fread( $fp, 256 );
423 $magic = MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer();
424 $extMime = $magic->guessTypesForExtension( $this->mFinalExtension
);
425 $ieTypes = $magic->getIEMimeTypes( $this->mTempPath
, $chunk, $extMime );
426 foreach ( $ieTypes as $ieType ) {
427 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
428 return [ 'filetype-bad-ie-mime', $ieType ];
437 * Verifies that it's ok to include the uploaded file
439 * @return mixed True of the file is verified, array otherwise.
441 protected function verifyFile() {
442 global $wgVerifyMimeType, $wgDisableUploadScriptChecks;
444 $status = $this->verifyPartialFile();
445 if ( $status !== true ) {
449 $mwProps = new MWFileProps( MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer() );
450 $this->mFileProps
= $mwProps->getPropsFromPath( $this->mTempPath
, $this->mFinalExtension
);
451 $mime = $this->mFileProps
['mime'];
453 if ( $wgVerifyMimeType ) {
454 # XXX: Missing extension will be caught by validateName() via getTitle()
455 if ( $this->mFinalExtension
!= '' && !$this->verifyExtension( $mime, $this->mFinalExtension
) ) {
456 return [ 'filetype-mime-mismatch', $this->mFinalExtension
, $mime ];
460 # check for htmlish code and javascript
461 if ( !$wgDisableUploadScriptChecks ) {
462 if ( $this->mFinalExtension
== 'svg' ||
$mime == 'image/svg+xml' ) {
463 $svgStatus = $this->detectScriptInSvg( $this->mTempPath
, false );
464 if ( $svgStatus !== false ) {
470 $handler = MediaHandler
::getHandler( $mime );
472 $handlerStatus = $handler->verifyUpload( $this->mTempPath
);
473 if ( !$handlerStatus->isOK() ) {
474 $errors = $handlerStatus->getErrorsArray();
476 return reset( $errors );
481 Hooks
::run( 'UploadVerifyFile', [ $this, $mime, &$error ] );
482 if ( $error !== true ) {
483 if ( !is_array( $error ) ) {
489 wfDebug( __METHOD__
. ": all clear; passing.\n" );
495 * A verification routine suitable for partial files
497 * Runs the blacklist checks, but not any checks that may
498 * assume the entire file is present.
500 * @return mixed True for valid or array with error message key.
502 protected function verifyPartialFile() {
503 global $wgAllowJavaUploads, $wgDisableUploadScriptChecks;
505 # getTitle() sets some internal parameters like $this->mFinalExtension
508 $mwProps = new MWFileProps( MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer() );
509 $this->mFileProps
= $mwProps->getPropsFromPath( $this->mTempPath
, $this->mFinalExtension
);
511 # check MIME type, if desired
512 $mime = $this->mFileProps
['file-mime'];
513 $status = $this->verifyMimeType( $mime );
514 if ( $status !== true ) {
518 # check for htmlish code and javascript
519 if ( !$wgDisableUploadScriptChecks ) {
520 if ( self
::detectScript( $this->mTempPath
, $mime, $this->mFinalExtension
) ) {
521 return [ 'uploadscripted' ];
523 if ( $this->mFinalExtension
== 'svg' ||
$mime == 'image/svg+xml' ) {
524 $svgStatus = $this->detectScriptInSvg( $this->mTempPath
, true );
525 if ( $svgStatus !== false ) {
531 # Check for Java applets, which if uploaded can bypass cross-site
533 if ( !$wgAllowJavaUploads ) {
534 $this->mJavaDetected
= false;
535 $zipStatus = ZipDirectoryReader
::read( $this->mTempPath
,
536 [ $this, 'zipEntryCallback' ] );
537 if ( !$zipStatus->isOK() ) {
538 $errors = $zipStatus->getErrorsArray();
539 $error = reset( $errors );
540 if ( $error[0] !== 'zip-wrong-format' ) {
544 if ( $this->mJavaDetected
) {
545 return [ 'uploadjava' ];
549 # Scan the uploaded file for viruses
550 $virus = $this->detectVirus( $this->mTempPath
);
552 return [ 'uploadvirus', $virus ];
559 * Callback for ZipDirectoryReader to detect Java class files.
561 * @param array $entry
563 public function zipEntryCallback( $entry ) {
564 $names = [ $entry['name'] ];
566 // If there is a null character, cut off the name at it, because JDK's
567 // ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name
568 // were constructed which had ".class\0" followed by a string chosen to
569 // make the hash collide with the truncated name, that file could be
570 // returned in response to a request for the .class file.
571 $nullPos = strpos( $entry['name'], "\000" );
572 if ( $nullPos !== false ) {
573 $names[] = substr( $entry['name'], 0, $nullPos );
576 // If there is a trailing slash in the file name, we have to strip it,
577 // because that's what ZIP_GetEntry() does.
578 if ( preg_grep( '!\.class/?$!', $names ) ) {
579 $this->mJavaDetected
= true;
584 * Alias for verifyTitlePermissions. The function was originally
585 * 'verifyPermissions', but that suggests it's checking the user, when it's
586 * really checking the title + user combination.
588 * @param User $user User object to verify the permissions against
589 * @return mixed An array as returned by getUserPermissionsErrors or true
590 * in case the user has proper permissions.
592 public function verifyPermissions( $user ) {
593 return $this->verifyTitlePermissions( $user );
597 * Check whether the user can edit, upload and create the image. This
598 * checks only against the current title; if it returns errors, it may
599 * very well be that another title will not give errors. Therefore
600 * isAllowed() should be called as well for generic is-user-blocked or
601 * can-user-upload checking.
603 * @param User $user User object to verify the permissions against
604 * @return mixed An array as returned by getUserPermissionsErrors or true
605 * in case the user has proper permissions.
607 public function verifyTitlePermissions( $user ) {
609 * If the image is protected, non-sysop users won't be able
610 * to modify it by uploading a new revision.
612 $nt = $this->getTitle();
613 if ( is_null( $nt ) ) {
616 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
617 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
618 if ( !$nt->exists() ) {
619 $permErrorsCreate = $nt->getUserPermissionsErrors( 'create', $user );
621 $permErrorsCreate = [];
623 if ( $permErrors ||
$permErrorsUpload ||
$permErrorsCreate ) {
624 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
625 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
630 $overwriteError = $this->checkOverwrite( $user );
631 if ( $overwriteError !== true ) {
632 return [ $overwriteError ];
639 * Check for non fatal problems with the file.
641 * This should not assume that mTempPath is set.
643 * @return mixed[] Array of warnings
645 public function checkWarnings() {
648 $localFile = $this->getLocalFile();
649 $localFile->load( File
::READ_LATEST
);
650 $filename = $localFile->getName();
651 $hash = $this->getTempFileSha1Base36();
653 $badFileName = $this->checkBadFileName( $filename, $this->mDesiredDestName
);
654 if ( $badFileName !== null ) {
655 $warnings['badfilename'] = $badFileName;
658 $unwantedFileExtensionDetails = $this->checkUnwantedFileExtensions( $this->mFinalExtension
);
659 if ( $unwantedFileExtensionDetails !== null ) {
660 $warnings['filetype-unwanted-type'] = $unwantedFileExtensionDetails;
663 $fileSizeWarnings = $this->checkFileSize( $this->mFileSize
);
664 if ( $fileSizeWarnings ) {
665 $warnings = array_merge( $warnings, $fileSizeWarnings );
668 $localFileExistsWarnings = $this->checkLocalFileExists( $localFile, $hash );
669 if ( $localFileExistsWarnings ) {
670 $warnings = array_merge( $warnings, $localFileExistsWarnings );
673 if ( $this->checkLocalFileWasDeleted( $localFile ) ) {
674 $warnings['was-deleted'] = $filename;
677 // If a file with the same name exists locally then the local file has already been tested
678 // for duplication of content
679 $ignoreLocalDupes = isset( $warnings[ 'exists '] );
680 $dupes = $this->checkAgainstExistingDupes( $hash, $ignoreLocalDupes );
682 $warnings['duplicate'] = $dupes;
685 $archivedDupes = $this->checkAgainstArchiveDupes( $hash );
686 if ( $archivedDupes !== null ) {
687 $warnings['duplicate-archive'] = $archivedDupes;
694 * Check whether the resulting filename is different from the desired one,
695 * but ignore things like ucfirst() and spaces/underscore things
697 * @param string $filename
698 * @param string $desiredFileName
700 * @return string|null String that was determined to be bad or null if the filename is okay
702 private function checkBadFileName( $filename, $desiredFileName ) {
703 $comparableName = str_replace( ' ', '_', $desiredFileName );
704 $comparableName = Title
::capitalize( $comparableName, NS_FILE
);
706 if ( $desiredFileName != $filename && $comparableName != $filename ) {
714 * @param string $fileExtension The file extension to check
716 * @return array|null array with the following keys:
717 * 0 => string The final extension being used
718 * 1 => string[] The extensions that are allowed
719 * 2 => int The number of extensions that are allowed.
721 private function checkUnwantedFileExtensions( $fileExtension ) {
722 global $wgCheckFileExtensions, $wgFileExtensions, $wgLang;
724 if ( $wgCheckFileExtensions ) {
725 $extensions = array_unique( $wgFileExtensions );
726 if ( !$this->checkFileExtension( $fileExtension, $extensions ) ) {
729 $wgLang->commaList( $extensions ),
739 * @param int $fileSize
741 * @return array warnings
743 private function checkFileSize( $fileSize ) {
744 global $wgUploadSizeWarning;
748 if ( $wgUploadSizeWarning && ( $fileSize > $wgUploadSizeWarning ) ) {
749 $warnings['large-file'] = [ $wgUploadSizeWarning, $fileSize ];
752 if ( $fileSize == 0 ) {
753 $warnings['empty-file'] = true;
760 * @param LocalFile $localFile
761 * @param string $hash sha1 hash of the file to check
763 * @return array warnings
765 private function checkLocalFileExists( LocalFile
$localFile, $hash ) {
768 $exists = self
::getExistsWarning( $localFile );
769 if ( $exists !== false ) {
770 $warnings['exists'] = $exists;
772 // check if file is an exact duplicate of current file version
773 if ( $hash === $localFile->getSha1() ) {
774 $warnings['no-change'] = $localFile;
777 // check if file is an exact duplicate of older versions of this file
778 $history = $localFile->getHistory();
779 foreach ( $history as $oldFile ) {
780 if ( $hash === $oldFile->getSha1() ) {
781 $warnings['duplicate-version'][] = $oldFile;
789 private function checkLocalFileWasDeleted( LocalFile
$localFile ) {
790 return $localFile->wasDeleted() && !$localFile->exists();
794 * @param string $hash sha1 hash of the file to check
795 * @param bool $ignoreLocalDupes True to ignore local duplicates
797 * @return File[] Duplicate files, if found.
799 private function checkAgainstExistingDupes( $hash, $ignoreLocalDupes ) {
800 $dupes = RepoGroup
::singleton()->findBySha1( $hash );
801 $title = $this->getTitle();
802 foreach ( $dupes as $key => $dupe ) {
804 ( $dupe instanceof LocalFile
) &&
806 $title->equals( $dupe->getTitle() )
808 unset( $dupes[$key] );
816 * @param string $hash sha1 hash of the file to check
818 * @return string|null Name of the dupe or empty string if discovered (depending on visibility)
819 * null if the check discovered no dupes.
821 private function checkAgainstArchiveDupes( $hash ) {
822 $archivedFile = new ArchivedFile( null, 0, '', $hash );
823 if ( $archivedFile->getID() > 0 ) {
824 if ( $archivedFile->userCan( File
::DELETED_FILE
) ) {
825 return $archivedFile->getName();
835 * Really perform the upload. Stores the file in the local repo, watches
836 * if necessary and runs the UploadComplete hook.
838 * @param string $comment
839 * @param string $pageText
840 * @param bool $watch Whether the file page should be added to user's watchlist.
841 * (This doesn't check $user's permissions.)
843 * @param string[] $tags Change tags to add to the log entry and page revision.
844 * (This doesn't check $user's permissions.)
845 * @return Status Indicating the whether the upload succeeded.
847 public function performUpload( $comment, $pageText, $watch, $user, $tags = [] ) {
848 $this->getLocalFile()->load( File
::READ_LATEST
);
849 $props = $this->mFileProps
;
852 Hooks
::run( 'UploadVerifyUpload', [ $this, $user, $props, $comment, $pageText, &$error ] );
854 if ( !is_array( $error ) ) {
857 return call_user_func_array( 'Status::newFatal', $error );
860 $status = $this->getLocalFile()->upload(
871 if ( $status->isGood() ) {
873 WatchAction
::doWatch(
874 $this->getLocalFile()->getTitle(),
876 User
::IGNORE_USER_RIGHTS
879 // Avoid PHP 7.1 warning of passing $this by reference
881 Hooks
::run( 'UploadComplete', [ &$uploadBase ] );
883 $this->postProcessUpload();
890 * Perform extra steps after a successful upload.
894 public function postProcessUpload() {
898 * Returns the title of the file to be uploaded. Sets mTitleError in case
899 * the name was illegal.
901 * @return Title|null The title of the file or null in case the name was illegal
903 public function getTitle() {
904 if ( $this->mTitle
!== false ) {
905 return $this->mTitle
;
907 if ( !is_string( $this->mDesiredDestName
) ) {
908 $this->mTitleError
= self
::ILLEGAL_FILENAME
;
909 $this->mTitle
= null;
911 return $this->mTitle
;
913 /* Assume that if a user specified File:Something.jpg, this is an error
914 * and that the namespace prefix needs to be stripped of.
916 $title = Title
::newFromText( $this->mDesiredDestName
);
917 if ( $title && $title->getNamespace() == NS_FILE
) {
918 $this->mFilteredName
= $title->getDBkey();
920 $this->mFilteredName
= $this->mDesiredDestName
;
923 # oi_archive_name is max 255 bytes, which include a timestamp and an
924 # exclamation mark, so restrict file name to 240 bytes.
925 if ( strlen( $this->mFilteredName
) > 240 ) {
926 $this->mTitleError
= self
::FILENAME_TOO_LONG
;
927 $this->mTitle
= null;
929 return $this->mTitle
;
933 * Chop off any directories in the given filename. Then
934 * filter out illegal characters, and try to make a legible name
935 * out of it. We'll strip some silently that Title would die on.
937 $this->mFilteredName
= wfStripIllegalFilenameChars( $this->mFilteredName
);
938 /* Normalize to title form before we do any further processing */
939 $nt = Title
::makeTitleSafe( NS_FILE
, $this->mFilteredName
);
940 if ( is_null( $nt ) ) {
941 $this->mTitleError
= self
::ILLEGAL_FILENAME
;
942 $this->mTitle
= null;
944 return $this->mTitle
;
946 $this->mFilteredName
= $nt->getDBkey();
949 * We'll want to blacklist against *any* 'extension', and use
950 * only the final one for the whitelist.
952 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName
);
954 if ( count( $ext ) ) {
955 $this->mFinalExtension
= trim( $ext[count( $ext ) - 1] );
957 $this->mFinalExtension
= '';
959 # No extension, try guessing one
960 $magic = MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer();
961 $mime = $magic->guessMimeType( $this->mTempPath
);
962 if ( $mime !== 'unknown/unknown' ) {
963 # Get a space separated list of extensions
964 $extList = $magic->getExtensionsForType( $mime );
966 # Set the extension to the canonical extension
967 $this->mFinalExtension
= strtok( $extList, ' ' );
969 # Fix up the other variables
970 $this->mFilteredName
.= ".{$this->mFinalExtension}";
971 $nt = Title
::makeTitleSafe( NS_FILE
, $this->mFilteredName
);
972 $ext = [ $this->mFinalExtension
];
977 /* Don't allow users to override the blacklist (check file extension) */
978 global $wgCheckFileExtensions, $wgStrictFileExtensions;
979 global $wgFileExtensions, $wgFileBlacklist;
981 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
983 if ( $this->mFinalExtension
== '' ) {
984 $this->mTitleError
= self
::FILETYPE_MISSING
;
985 $this->mTitle
= null;
987 return $this->mTitle
;
988 } elseif ( $blackListedExtensions ||
989 ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
990 !$this->checkFileExtension( $this->mFinalExtension
, $wgFileExtensions ) )
992 $this->mBlackListedExtensions
= $blackListedExtensions;
993 $this->mTitleError
= self
::FILETYPE_BADTYPE
;
994 $this->mTitle
= null;
996 return $this->mTitle
;
999 // Windows may be broken with special characters, see T3780
1000 if ( !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() )
1001 && !RepoGroup
::singleton()->getLocalRepo()->backendSupportsUnicodePaths()
1003 $this->mTitleError
= self
::WINDOWS_NONASCII_FILENAME
;
1004 $this->mTitle
= null;
1006 return $this->mTitle
;
1009 # If there was more than one "extension", reassemble the base
1010 # filename to prevent bogus complaints about length
1011 if ( count( $ext ) > 1 ) {
1012 $iterations = count( $ext ) - 1;
1013 for ( $i = 0; $i < $iterations; $i++
) {
1014 $partname .= '.' . $ext[$i];
1018 if ( strlen( $partname ) < 1 ) {
1019 $this->mTitleError
= self
::MIN_LENGTH_PARTNAME
;
1020 $this->mTitle
= null;
1022 return $this->mTitle
;
1025 $this->mTitle
= $nt;
1027 return $this->mTitle
;
1031 * Return the local file and initializes if necessary.
1033 * @return LocalFile|null
1035 public function getLocalFile() {
1036 if ( is_null( $this->mLocalFile
) ) {
1037 $nt = $this->getTitle();
1038 $this->mLocalFile
= is_null( $nt ) ?
null : wfLocalFile( $nt );
1041 return $this->mLocalFile
;
1045 * @return UploadStashFile|null
1047 public function getStashFile() {
1048 return $this->mStashFile
;
1052 * Like stashFile(), but respects extensions' wishes to prevent the stashing. verifyUpload() must
1053 * be called before calling this method (unless $isPartial is true).
1055 * Upload stash exceptions are also caught and converted to an error status.
1059 * @param bool $isPartial Pass `true` if this is a part of a chunked upload (not a complete file).
1060 * @return Status If successful, value is an UploadStashFile instance
1062 public function tryStashFile( User
$user, $isPartial = false ) {
1063 if ( !$isPartial ) {
1064 $error = $this->runUploadStashFileHook( $user );
1066 return call_user_func_array( 'Status::newFatal', $error );
1070 $file = $this->doStashFile( $user );
1071 return Status
::newGood( $file );
1072 } catch ( UploadStashException
$e ) {
1073 return Status
::newFatal( 'uploadstash-exception', get_class( $e ), $e->getMessage() );
1079 * @return array|null Error message and parameters, null if there's no error
1081 protected function runUploadStashFileHook( User
$user ) {
1082 $props = $this->mFileProps
;
1084 Hooks
::run( 'UploadStashFile', [ $this, $user, $props, &$error ] );
1086 if ( !is_array( $error ) ) {
1087 $error = [ $error ];
1094 * If the user does not supply all necessary information in the first upload
1095 * form submission (either by accident or by design) then we may want to
1096 * stash the file temporarily, get more information, and publish the file
1099 * This method will stash a file in a temporary directory for later
1100 * processing, and save the necessary descriptive info into the database.
1101 * This method returns the file object, which also has a 'fileKey' property
1102 * which can be passed through a form or API request to find this stashed
1105 * @deprecated since 1.28 Use tryStashFile() instead
1107 * @return UploadStashFile Stashed file
1108 * @throws UploadStashBadPathException
1109 * @throws UploadStashFileException
1110 * @throws UploadStashNotLoggedInException
1112 public function stashFile( User
$user = null ) {
1113 return $this->doStashFile( $user );
1117 * Implementation for stashFile() and tryStashFile().
1120 * @return UploadStashFile Stashed file
1122 protected function doStashFile( User
$user = null ) {
1123 $stash = RepoGroup
::singleton()->getLocalRepo()->getUploadStash( $user );
1124 $file = $stash->stashFile( $this->mTempPath
, $this->getSourceType() );
1125 $this->mStashFile
= $file;
1131 * Stash a file in a temporary directory, returning a key which can be used
1132 * to find the file again. See stashFile().
1134 * @deprecated since 1.28
1135 * @return string File key
1137 public function stashFileGetKey() {
1138 wfDeprecated( __METHOD__
, '1.28' );
1139 return $this->doStashFile()->getFileKey();
1143 * alias for stashFileGetKey, for backwards compatibility
1145 * @deprecated since 1.28
1146 * @return string File key
1148 public function stashSession() {
1149 wfDeprecated( __METHOD__
, '1.28' );
1150 return $this->doStashFile()->getFileKey();
1154 * If we've modified the upload file we need to manually remove it
1155 * on exit to clean up.
1157 public function cleanupTempFile() {
1158 if ( $this->mRemoveTempFile
&& $this->tempFileObj
) {
1159 // Delete when all relevant TempFSFile handles go out of scope
1160 wfDebug( __METHOD__
. ": Marked temporary file '{$this->mTempPath}' for removal\n" );
1161 $this->tempFileObj
->autocollect();
1165 public function getTempPath() {
1166 return $this->mTempPath
;
1170 * Split a file into a base name and all dot-delimited 'extensions'
1171 * on the end. Some web server configurations will fall back to
1172 * earlier pseudo-'extensions' to determine type and execute
1173 * scripts, so the blacklist needs to check them all.
1175 * @param string $filename
1178 public static function splitExtensions( $filename ) {
1179 $bits = explode( '.', $filename );
1180 $basename = array_shift( $bits );
1182 return [ $basename, $bits ];
1186 * Perform case-insensitive match against a list of file extensions.
1187 * Returns true if the extension is in the list.
1189 * @param string $ext
1190 * @param array $list
1193 public static function checkFileExtension( $ext, $list ) {
1194 return in_array( strtolower( $ext ), $list );
1198 * Perform case-insensitive match against a list of file extensions.
1199 * Returns an array of matching extensions.
1202 * @param array $list
1205 public static function checkFileExtensionList( $ext, $list ) {
1206 return array_intersect( array_map( 'strtolower', $ext ), $list );
1210 * Checks if the MIME type of the uploaded file matches the file extension.
1212 * @param string $mime The MIME type of the uploaded file
1213 * @param string $extension The filename extension that the file is to be served with
1216 public static function verifyExtension( $mime, $extension ) {
1217 $magic = MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer();
1219 if ( !$mime ||
$mime == 'unknown' ||
$mime == 'unknown/unknown' ) {
1220 if ( !$magic->isRecognizableExtension( $extension ) ) {
1221 wfDebug( __METHOD__
. ": passing file with unknown detected mime type; " .
1222 "unrecognized extension '$extension', can't verify\n" );
1226 wfDebug( __METHOD__
. ": rejecting file with unknown detected mime type; " .
1227 "recognized extension '$extension', so probably invalid file\n" );
1233 $match = $magic->isMatchingExtension( $extension, $mime );
1235 if ( $match === null ) {
1236 if ( $magic->getTypesForExtension( $extension ) !== null ) {
1237 wfDebug( __METHOD__
. ": No extension known for $mime, but we know a mime for $extension\n" );
1241 wfDebug( __METHOD__
. ": no file extension known for mime type $mime, passing file\n" );
1245 } elseif ( $match === true ) {
1246 wfDebug( __METHOD__
. ": mime type $mime matches extension $extension, passing file\n" );
1248 /** @todo If it's a bitmap, make sure PHP or ImageMagick resp. can handle it! */
1252 . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
1259 * Heuristic for detecting files that *could* contain JavaScript instructions or
1260 * things that may look like HTML to a browser and are thus
1261 * potentially harmful. The present implementation will produce false
1262 * positives in some situations.
1264 * @param string $file Pathname to the temporary upload file
1265 * @param string $mime The MIME type of the file
1266 * @param string $extension The extension of the file
1267 * @return bool True if the file contains something looking like embedded scripts
1269 public static function detectScript( $file, $mime, $extension ) {
1270 global $wgAllowTitlesInSVG;
1272 # ugly hack: for text files, always look at the entire file.
1273 # For binary field, just check the first K.
1275 if ( strpos( $mime, 'text/' ) === 0 ) {
1276 $chunk = file_get_contents( $file );
1278 $fp = fopen( $file, 'rb' );
1279 $chunk = fread( $fp, 1024 );
1283 $chunk = strtolower( $chunk );
1289 # decode from UTF-16 if needed (could be used for obfuscation).
1290 if ( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
1292 } elseif ( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
1299 $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
1302 $chunk = trim( $chunk );
1304 /** @todo FIXME: Convert from UTF-16 if necessary! */
1305 wfDebug( __METHOD__
. ": checking for embedded scripts and HTML stuff\n" );
1307 # check for HTML doctype
1308 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
1312 // Some browsers will interpret obscure xml encodings as UTF-8, while
1313 // PHP/expat will interpret the given encoding in the xml declaration (T49304)
1314 if ( $extension == 'svg' ||
strpos( $mime, 'image/svg' ) === 0 ) {
1315 if ( self
::checkXMLEncodingMissmatch( $file ) ) {
1321 * Internet Explorer for Windows performs some really stupid file type
1322 * autodetection which can cause it to interpret valid image files as HTML
1323 * and potentially execute JavaScript, creating a cross-site scripting
1326 * Apple's Safari browser also performs some unsafe file type autodetection
1327 * which can cause legitimate files to be interpreted as HTML if the
1328 * web server is not correctly configured to send the right content-type
1329 * (or if you're really uploading plain text and octet streams!)
1331 * Returns true if IE is likely to mistake the given file for HTML.
1332 * Also returns true if Safari would mistake the given file for HTML
1333 * when served with a generic content-type.
1339 '<html', # also in safari
1342 '<script', # also in safari
1346 if ( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
1350 foreach ( $tags as $tag ) {
1351 if ( false !== strpos( $chunk, $tag ) ) {
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, -1, $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 MediaWiki\
suppressWarnings();
1429 $str = iconv( $encoding, 'UTF-8', $contents );
1430 MediaWiki\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 mixed False of the file is verified (does not contain scripts), array otherwise.
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 (true if the filter identified something bad)
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 array $data
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 if ( $strippedElement == 'image'
1748 && $stripped == 'filter'
1749 && preg_match( '!url\s*\(!sim', $value )
1751 wfDebug( __METHOD__
. ": Found image filter with url: "
1752 . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1754 return [ 'uploaded-image-filter-svg', $strippedElement, $stripped, $value ];
1758 return false; // No scripts detected
1762 * Check a block of CSS or CSS fragment for anything that looks like
1763 * it is bringing in remote code.
1764 * @param string $value a string of CSS
1765 * @param bool $propOnly only check css properties (start regex with :)
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 mixed 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 = isset( $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ) ?
1868 $wgAntivirusSetup[$wgAntivirus]['messagepattern'] : null;
1870 if ( strpos( $command, "%f" ) === false ) {
1871 # simple pattern: append file to scan
1872 $command .= " " . wfEscapeShellArg( $file );
1874 # complex pattern: replace "%f" with file to scan
1875 $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
1878 wfDebug( __METHOD__
. ": running virus scan: $command \n" );
1880 # execute virus scanner
1883 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1884 # that does not seem to be worth the pain.
1885 # Ask me (Duesentrieb) about it if it's ever needed.
1886 $output = wfShellExecWithStderr( $command, $exitCode );
1888 # map exit code to AV_xxx constants.
1889 $mappedCode = $exitCode;
1890 if ( $exitCodeMap ) {
1891 if ( isset( $exitCodeMap[$exitCode] ) ) {
1892 $mappedCode = $exitCodeMap[$exitCode];
1893 } elseif ( isset( $exitCodeMap["*"] ) ) {
1894 $mappedCode = $exitCodeMap["*"];
1898 /* NB: AV_NO_VIRUS is 0 but AV_SCAN_FAILED is false,
1899 * so we need the strict equalities === and thus can't use a switch here
1901 if ( $mappedCode === AV_SCAN_FAILED
) {
1902 # scan failed (code was mapped to false by $exitCodeMap)
1903 wfDebug( __METHOD__
. ": failed to scan $file (code $exitCode).\n" );
1905 $output = $wgAntivirusRequired
1906 ?
wfMessage( 'virus-scanfailed', [ $exitCode ] )->text()
1908 } elseif ( $mappedCode === AV_SCAN_ABORTED
) {
1909 # scan failed because filetype is unknown (probably imune)
1910 wfDebug( __METHOD__
. ": unsupported file type $file (code $exitCode).\n" );
1912 } elseif ( $mappedCode === AV_NO_VIRUS
) {
1914 wfDebug( __METHOD__
. ": file passed virus scan.\n" );
1917 $output = trim( $output );
1920 $output = true; # if there's no output, return true
1921 } elseif ( $msgPattern ) {
1923 if ( preg_match( $msgPattern, $output, $groups ) ) {
1925 $output = $groups[1];
1930 wfDebug( __METHOD__
. ": FOUND VIRUS! scanner feedback: $output \n" );
1937 * Check if there's an overwrite conflict and, if so, if restrictions
1938 * forbid this user from performing the upload.
1942 * @return mixed True on success, array on failure
1944 private function checkOverwrite( $user ) {
1945 // First check whether the local file can be overwritten
1946 $file = $this->getLocalFile();
1947 $file->load( File
::READ_LATEST
);
1948 if ( $file->exists() ) {
1949 if ( !self
::userCanReUpload( $user, $file ) ) {
1950 return [ 'fileexists-forbidden', $file->getName() ];
1956 /* Check shared conflicts: if the local file does not exist, but
1957 * wfFindFile finds a file, it exists in a shared repository.
1959 $file = wfFindFile( $this->getTitle(), [ 'latest' => true ] );
1960 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1961 return [ 'fileexists-shared-forbidden', $file->getName() ];
1968 * Check if a user is the last uploader
1974 public static function userCanReUpload( User
$user, File
$img ) {
1975 if ( $user->isAllowed( 'reupload' ) ) {
1976 return true; // non-conditional
1977 } elseif ( !$user->isAllowed( 'reupload-own' ) ) {
1981 if ( !( $img instanceof LocalFile
) ) {
1987 return $user->getId() == $img->getUser( 'id' );
1991 * Helper function that does various existence checks for a file.
1992 * The following checks are performed:
1994 * - Article with the same name as the file exists
1995 * - File exists with normalized extension
1996 * - The file looks like a thumbnail and the original exists
1998 * @param File $file The File object to check
1999 * @return mixed False if the file does not exists, else an array
2001 public static function getExistsWarning( $file ) {
2002 if ( $file->exists() ) {
2003 return [ 'warning' => 'exists', 'file' => $file ];
2006 if ( $file->getTitle()->getArticleID() ) {
2007 return [ 'warning' => 'page-exists', 'file' => $file ];
2010 if ( strpos( $file->getName(), '.' ) == false ) {
2011 $partname = $file->getName();
2014 $n = strrpos( $file->getName(), '.' );
2015 $extension = substr( $file->getName(), $n +
1 );
2016 $partname = substr( $file->getName(), 0, $n );
2018 $normalizedExtension = File
::normalizeExtension( $extension );
2020 if ( $normalizedExtension != $extension ) {
2021 // We're not using the normalized form of the extension.
2022 // Normal form is lowercase, using most common of alternate
2023 // extensions (eg 'jpg' rather than 'JPEG').
2025 // Check for another file using the normalized form...
2026 $nt_lc = Title
::makeTitle( NS_FILE
, "{$partname}.{$normalizedExtension}" );
2027 $file_lc = wfLocalFile( $nt_lc );
2029 if ( $file_lc->exists() ) {
2031 'warning' => 'exists-normalized',
2033 'normalizedFile' => $file_lc
2038 // Check for files with the same name but a different extension
2039 $similarFiles = RepoGroup
::singleton()->getLocalRepo()->findFilesByPrefix(
2040 "{$partname}.", 1 );
2041 if ( count( $similarFiles ) ) {
2043 'warning' => 'exists-normalized',
2045 'normalizedFile' => $similarFiles[0],
2049 if ( self
::isThumbName( $file->getName() ) ) {
2050 # Check for filenames like 50px- or 180px-, these are mostly thumbnails
2051 $nt_thb = Title
::newFromText(
2052 substr( $partname, strpos( $partname, '-' ) +
1 ) . '.' . $extension,
2055 $file_thb = wfLocalFile( $nt_thb );
2056 if ( $file_thb->exists() ) {
2058 'warning' => 'thumb',
2060 'thumbFile' => $file_thb
2063 // File does not exist, but we just don't like the name
2065 'warning' => 'thumb-name',
2067 'thumbFile' => $file_thb
2072 foreach ( self
::getFilenamePrefixBlacklist() as $prefix ) {
2073 if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
2075 'warning' => 'bad-prefix',
2086 * Helper function that checks whether the filename looks like a thumbnail
2087 * @param string $filename
2090 public static function isThumbName( $filename ) {
2091 $n = strrpos( $filename, '.' );
2092 $partname = $n ?
substr( $filename, 0, $n ) : $filename;
2095 substr( $partname, 3, 3 ) == 'px-' ||
2096 substr( $partname, 2, 3 ) == 'px-'
2098 preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) );
2102 * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
2104 * @return array List of prefixes
2106 public static function getFilenamePrefixBlacklist() {
2108 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
2109 if ( !$message->isDisabled() ) {
2110 $lines = explode( "\n", $message->plain() );
2111 foreach ( $lines as $line ) {
2112 // Remove comment lines
2113 $comment = substr( trim( $line ), 0, 1 );
2114 if ( $comment == '#' ||
$comment == '' ) {
2117 // Remove additional comments after a prefix
2118 $comment = strpos( $line, '#' );
2119 if ( $comment > 0 ) {
2120 $line = substr( $line, 0, $comment - 1 );
2122 $blacklist[] = trim( $line );
2130 * Gets image info about the file just uploaded.
2132 * Also has the effect of setting metadata to be an 'indexed tag name' in
2133 * returned API result if 'metadata' was requested. Oddly, we have to pass
2134 * the "result" object down just so it can do that with the appropriate
2135 * format, presumably.
2137 * @param ApiResult $result
2138 * @return array Image info
2140 public function getImageInfo( $result ) {
2141 $localFile = $this->getLocalFile();
2142 $stashFile = $this->getStashFile();
2143 // Calling a different API module depending on whether the file was stashed is less than optimal.
2144 // In fact, calling API modules here at all is less than optimal. Maybe it should be refactored.
2146 $imParam = ApiQueryStashImageInfo
::getPropertyNames();
2147 $info = ApiQueryStashImageInfo
::getInfo( $stashFile, array_flip( $imParam ), $result );
2149 $imParam = ApiQueryImageInfo
::getPropertyNames();
2150 $info = ApiQueryImageInfo
::getInfo( $localFile, array_flip( $imParam ), $result );
2157 * @param array $error
2160 public function convertVerifyErrorToStatus( $error ) {
2161 $code = $error['status'];
2162 unset( $code['status'] );
2164 return Status
::newFatal( $this->getVerificationErrorCode( $code ), $error );
2168 * Get the MediaWiki maximum uploaded file size for given type of upload, based on
2171 * @param null|string $forType
2174 public static function getMaxUploadSize( $forType = null ) {
2175 global $wgMaxUploadSize;
2177 if ( is_array( $wgMaxUploadSize ) ) {
2178 if ( !is_null( $forType ) && isset( $wgMaxUploadSize[$forType] ) ) {
2179 return $wgMaxUploadSize[$forType];
2181 return $wgMaxUploadSize['*'];
2184 return intval( $wgMaxUploadSize );
2189 * Get the PHP maximum uploaded file size, based on ini settings. If there is no limit or the
2190 * limit can't be guessed, returns a very large number (PHP_INT_MAX).
2195 public static function getMaxPhpUploadSize() {
2196 $phpMaxFileSize = wfShorthandToInteger(
2197 ini_get( 'upload_max_filesize' ) ?
: ini_get( 'hhvm.server.upload.upload_max_file_size' ),
2200 $phpMaxPostSize = wfShorthandToInteger(
2201 ini_get( 'post_max_size' ) ?
: ini_get( 'hhvm.server.max_post_size' ),
2204 return min( $phpMaxFileSize, $phpMaxPostSize );
2208 * Get the current status of a chunked upload (used for polling)
2210 * The value will be read from cache.
2213 * @param string $statusKey
2214 * @return Status[]|bool
2216 public static function getSessionStatus( User
$user, $statusKey ) {
2217 $cache = MediaWikiServices
::getInstance()->getMainObjectStash();
2218 $key = $cache->makeKey( 'uploadstatus', $user->getId() ?
: md5( $user->getName() ), $statusKey );
2220 return $cache->get( $key );
2224 * Set the current status of a chunked upload (used for polling)
2226 * The value will be set in cache for 1 day
2229 * @param string $statusKey
2230 * @param array|bool $value
2233 public static function setSessionStatus( User
$user, $statusKey, $value ) {
2234 $cache = MediaWikiServices
::getInstance()->getMainObjectStash();
2235 $key = $cache->makeKey( 'uploadstatus', $user->getId() ?
: md5( $user->getName() ), $statusKey );
2237 if ( $value === false ) {
2238 $cache->delete( $key );
2240 $cache->set( $key, $value, $cache::TTL_DAY
);