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 integer $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 = MimeMagic
::singleton();
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( MimeMagic
::singleton() );
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( MimeMagic
::singleton() );
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 $dupes = $this->checkAgainstExistingDupes( $hash );
679 $warnings['duplicate'] = $dupes;
682 $archivedDupes = $this->checkAgainstArchiveDupes( $hash );
683 if ( $archivedDupes !== null ) {
684 $warnings['duplicate-archive'] = $archivedDupes;
691 * Check whether the resulting filename is different from the desired one,
692 * but ignore things like ucfirst() and spaces/underscore things
694 * @param string $filename
695 * @param string $desiredFileName
697 * @return string|null String that was determined to be bad or null if the filename is okay
699 private function checkBadFileName( $filename, $desiredFileName ) {
700 $comparableName = str_replace( ' ', '_', $desiredFileName );
701 $comparableName = Title
::capitalize( $comparableName, NS_FILE
);
703 if ( $desiredFileName != $filename && $comparableName != $filename ) {
711 * @param string $fileExtension The file extension to check
713 * @return array|null array with the following keys:
714 * 0 => string The final extension being used
715 * 1 => string[] The extensions that are allowed
716 * 2 => int The number of extensions that are allowed.
718 private function checkUnwantedFileExtensions( $fileExtension ) {
719 global $wgCheckFileExtensions, $wgFileExtensions, $wgLang;
721 if ( $wgCheckFileExtensions ) {
722 $extensions = array_unique( $wgFileExtensions );
723 if ( !$this->checkFileExtension( $fileExtension, $extensions ) ) {
726 $wgLang->commaList( $extensions ),
736 * @param int $fileSize
738 * @return array warnings
740 private function checkFileSize( $fileSize ) {
741 global $wgUploadSizeWarning;
745 if ( $wgUploadSizeWarning && ( $fileSize > $wgUploadSizeWarning ) ) {
746 $warnings['large-file'] = [ $wgUploadSizeWarning, $fileSize ];
749 if ( $fileSize == 0 ) {
750 $warnings['empty-file'] = true;
757 * @param LocalFile $localFile
758 * @param string $hash sha1 hash of the file to check
760 * @return array warnings
762 private function checkLocalFileExists( LocalFile
$localFile, $hash ) {
765 $exists = self
::getExistsWarning( $localFile );
766 if ( $exists !== false ) {
767 $warnings['exists'] = $exists;
769 // check if file is an exact duplicate of current file version
770 if ( $hash === $localFile->getSha1() ) {
771 $warnings['no-change'] = $localFile;
774 // check if file is an exact duplicate of older versions of this file
775 $history = $localFile->getHistory();
776 foreach ( $history as $oldFile ) {
777 if ( $hash === $oldFile->getSha1() ) {
778 $warnings['duplicate-version'][] = $oldFile;
786 private function checkLocalFileWasDeleted( LocalFile
$localFile ) {
787 return $localFile->wasDeleted() && !$localFile->exists();
791 * @param string $hash sha1 hash of the file to check
793 * @return File[] Duplicate files, if found.
795 private function checkAgainstExistingDupes( $hash ) {
796 $dupes = RepoGroup
::singleton()->findBySha1( $hash );
797 $title = $this->getTitle();
798 // Remove all matches against self
799 foreach ( $dupes as $key => $dupe ) {
800 if ( $title->equals( $dupe->getTitle() ) ) {
801 unset( $dupes[$key] );
809 * @param string $hash sha1 hash of the file to check
811 * @return string|null Name of the dupe or empty string if discovered (depending on visibility)
812 * null if the check discovered no dupes.
814 private function checkAgainstArchiveDupes( $hash ) {
815 $archivedFile = new ArchivedFile( null, 0, '', $hash );
816 if ( $archivedFile->getID() > 0 ) {
817 if ( $archivedFile->userCan( File
::DELETED_FILE
) ) {
818 return $archivedFile->getName();
828 * Really perform the upload. Stores the file in the local repo, watches
829 * if necessary and runs the UploadComplete hook.
831 * @param string $comment
832 * @param string $pageText
833 * @param bool $watch Whether the file page should be added to user's watchlist.
834 * (This doesn't check $user's permissions.)
836 * @param string[] $tags Change tags to add to the log entry and page revision.
837 * (This doesn't check $user's permissions.)
838 * @return Status Indicating the whether the upload succeeded.
840 public function performUpload( $comment, $pageText, $watch, $user, $tags = [] ) {
841 $this->getLocalFile()->load( File
::READ_LATEST
);
842 $props = $this->mFileProps
;
845 Hooks
::run( 'UploadVerifyUpload', [ $this, $user, $props, $comment, $pageText, &$error ] );
847 if ( !is_array( $error ) ) {
850 return call_user_func_array( 'Status::newFatal', $error );
853 $status = $this->getLocalFile()->upload(
864 if ( $status->isGood() ) {
866 WatchAction
::doWatch(
867 $this->getLocalFile()->getTitle(),
869 User
::IGNORE_USER_RIGHTS
872 // Avoid PHP 7.1 warning of passing $this by reference
874 Hooks
::run( 'UploadComplete', [ &$uploadBase ] );
876 $this->postProcessUpload();
883 * Perform extra steps after a successful upload.
887 public function postProcessUpload() {
891 * Returns the title of the file to be uploaded. Sets mTitleError in case
892 * the name was illegal.
894 * @return Title|null The title of the file or null in case the name was illegal
896 public function getTitle() {
897 if ( $this->mTitle
!== false ) {
898 return $this->mTitle
;
900 if ( !is_string( $this->mDesiredDestName
) ) {
901 $this->mTitleError
= self
::ILLEGAL_FILENAME
;
902 $this->mTitle
= null;
904 return $this->mTitle
;
906 /* Assume that if a user specified File:Something.jpg, this is an error
907 * and that the namespace prefix needs to be stripped of.
909 $title = Title
::newFromText( $this->mDesiredDestName
);
910 if ( $title && $title->getNamespace() == NS_FILE
) {
911 $this->mFilteredName
= $title->getDBkey();
913 $this->mFilteredName
= $this->mDesiredDestName
;
916 # oi_archive_name is max 255 bytes, which include a timestamp and an
917 # exclamation mark, so restrict file name to 240 bytes.
918 if ( strlen( $this->mFilteredName
) > 240 ) {
919 $this->mTitleError
= self
::FILENAME_TOO_LONG
;
920 $this->mTitle
= null;
922 return $this->mTitle
;
926 * Chop off any directories in the given filename. Then
927 * filter out illegal characters, and try to make a legible name
928 * out of it. We'll strip some silently that Title would die on.
930 $this->mFilteredName
= wfStripIllegalFilenameChars( $this->mFilteredName
);
931 /* Normalize to title form before we do any further processing */
932 $nt = Title
::makeTitleSafe( NS_FILE
, $this->mFilteredName
);
933 if ( is_null( $nt ) ) {
934 $this->mTitleError
= self
::ILLEGAL_FILENAME
;
935 $this->mTitle
= null;
937 return $this->mTitle
;
939 $this->mFilteredName
= $nt->getDBkey();
942 * We'll want to blacklist against *any* 'extension', and use
943 * only the final one for the whitelist.
945 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName
);
947 if ( count( $ext ) ) {
948 $this->mFinalExtension
= trim( $ext[count( $ext ) - 1] );
950 $this->mFinalExtension
= '';
952 # No extension, try guessing one
953 $magic = MimeMagic
::singleton();
954 $mime = $magic->guessMimeType( $this->mTempPath
);
955 if ( $mime !== 'unknown/unknown' ) {
956 # Get a space separated list of extensions
957 $extList = $magic->getExtensionsForType( $mime );
959 # Set the extension to the canonical extension
960 $this->mFinalExtension
= strtok( $extList, ' ' );
962 # Fix up the other variables
963 $this->mFilteredName
.= ".{$this->mFinalExtension}";
964 $nt = Title
::makeTitleSafe( NS_FILE
, $this->mFilteredName
);
965 $ext = [ $this->mFinalExtension
];
970 /* Don't allow users to override the blacklist (check file extension) */
971 global $wgCheckFileExtensions, $wgStrictFileExtensions;
972 global $wgFileExtensions, $wgFileBlacklist;
974 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
976 if ( $this->mFinalExtension
== '' ) {
977 $this->mTitleError
= self
::FILETYPE_MISSING
;
978 $this->mTitle
= null;
980 return $this->mTitle
;
981 } elseif ( $blackListedExtensions ||
982 ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
983 !$this->checkFileExtension( $this->mFinalExtension
, $wgFileExtensions ) )
985 $this->mBlackListedExtensions
= $blackListedExtensions;
986 $this->mTitleError
= self
::FILETYPE_BADTYPE
;
987 $this->mTitle
= null;
989 return $this->mTitle
;
992 // Windows may be broken with special characters, see T3780
993 if ( !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() )
994 && !RepoGroup
::singleton()->getLocalRepo()->backendSupportsUnicodePaths()
996 $this->mTitleError
= self
::WINDOWS_NONASCII_FILENAME
;
997 $this->mTitle
= null;
999 return $this->mTitle
;
1002 # If there was more than one "extension", reassemble the base
1003 # filename to prevent bogus complaints about length
1004 if ( count( $ext ) > 1 ) {
1005 $iterations = count( $ext ) - 1;
1006 for ( $i = 0; $i < $iterations; $i++
) {
1007 $partname .= '.' . $ext[$i];
1011 if ( strlen( $partname ) < 1 ) {
1012 $this->mTitleError
= self
::MIN_LENGTH_PARTNAME
;
1013 $this->mTitle
= null;
1015 return $this->mTitle
;
1018 $this->mTitle
= $nt;
1020 return $this->mTitle
;
1024 * Return the local file and initializes if necessary.
1026 * @return LocalFile|null
1028 public function getLocalFile() {
1029 if ( is_null( $this->mLocalFile
) ) {
1030 $nt = $this->getTitle();
1031 $this->mLocalFile
= is_null( $nt ) ?
null : wfLocalFile( $nt );
1034 return $this->mLocalFile
;
1038 * @return UploadStashFile|null
1040 public function getStashFile() {
1041 return $this->mStashFile
;
1045 * Like stashFile(), but respects extensions' wishes to prevent the stashing. verifyUpload() must
1046 * be called before calling this method (unless $isPartial is true).
1048 * Upload stash exceptions are also caught and converted to an error status.
1052 * @param bool $isPartial Pass `true` if this is a part of a chunked upload (not a complete file).
1053 * @return Status If successful, value is an UploadStashFile instance
1055 public function tryStashFile( User
$user, $isPartial = false ) {
1056 if ( !$isPartial ) {
1057 $error = $this->runUploadStashFileHook( $user );
1059 return call_user_func_array( 'Status::newFatal', $error );
1063 $file = $this->doStashFile( $user );
1064 return Status
::newGood( $file );
1065 } catch ( UploadStashException
$e ) {
1066 return Status
::newFatal( 'uploadstash-exception', get_class( $e ), $e->getMessage() );
1072 * @return array|null Error message and parameters, null if there's no error
1074 protected function runUploadStashFileHook( User
$user ) {
1075 $props = $this->mFileProps
;
1077 Hooks
::run( 'UploadStashFile', [ $this, $user, $props, &$error ] );
1079 if ( !is_array( $error ) ) {
1080 $error = [ $error ];
1087 * If the user does not supply all necessary information in the first upload
1088 * form submission (either by accident or by design) then we may want to
1089 * stash the file temporarily, get more information, and publish the file
1092 * This method will stash a file in a temporary directory for later
1093 * processing, and save the necessary descriptive info into the database.
1094 * This method returns the file object, which also has a 'fileKey' property
1095 * which can be passed through a form or API request to find this stashed
1098 * @deprecated since 1.28 Use tryStashFile() instead
1100 * @return UploadStashFile Stashed file
1101 * @throws UploadStashBadPathException
1102 * @throws UploadStashFileException
1103 * @throws UploadStashNotLoggedInException
1105 public function stashFile( User
$user = null ) {
1106 return $this->doStashFile( $user );
1110 * Implementation for stashFile() and tryStashFile().
1113 * @return UploadStashFile Stashed file
1115 protected function doStashFile( User
$user = null ) {
1116 $stash = RepoGroup
::singleton()->getLocalRepo()->getUploadStash( $user );
1117 $file = $stash->stashFile( $this->mTempPath
, $this->getSourceType() );
1118 $this->mStashFile
= $file;
1124 * Stash a file in a temporary directory, returning a key which can be used
1125 * to find the file again. See stashFile().
1127 * @deprecated since 1.28
1128 * @return string File key
1130 public function stashFileGetKey() {
1131 wfDeprecated( __METHOD__
, '1.28' );
1132 return $this->doStashFile()->getFileKey();
1136 * alias for stashFileGetKey, for backwards compatibility
1138 * @deprecated since 1.28
1139 * @return string File key
1141 public function stashSession() {
1142 wfDeprecated( __METHOD__
, '1.28' );
1143 return $this->doStashFile()->getFileKey();
1147 * If we've modified the upload file we need to manually remove it
1148 * on exit to clean up.
1150 public function cleanupTempFile() {
1151 if ( $this->mRemoveTempFile
&& $this->tempFileObj
) {
1152 // Delete when all relevant TempFSFile handles go out of scope
1153 wfDebug( __METHOD__
. ": Marked temporary file '{$this->mTempPath}' for removal\n" );
1154 $this->tempFileObj
->autocollect();
1158 public function getTempPath() {
1159 return $this->mTempPath
;
1163 * Split a file into a base name and all dot-delimited 'extensions'
1164 * on the end. Some web server configurations will fall back to
1165 * earlier pseudo-'extensions' to determine type and execute
1166 * scripts, so the blacklist needs to check them all.
1168 * @param string $filename
1171 public static function splitExtensions( $filename ) {
1172 $bits = explode( '.', $filename );
1173 $basename = array_shift( $bits );
1175 return [ $basename, $bits ];
1179 * Perform case-insensitive match against a list of file extensions.
1180 * Returns true if the extension is in the list.
1182 * @param string $ext
1183 * @param array $list
1186 public static function checkFileExtension( $ext, $list ) {
1187 return in_array( strtolower( $ext ), $list );
1191 * Perform case-insensitive match against a list of file extensions.
1192 * Returns an array of matching extensions.
1195 * @param array $list
1198 public static function checkFileExtensionList( $ext, $list ) {
1199 return array_intersect( array_map( 'strtolower', $ext ), $list );
1203 * Checks if the MIME type of the uploaded file matches the file extension.
1205 * @param string $mime The MIME type of the uploaded file
1206 * @param string $extension The filename extension that the file is to be served with
1209 public static function verifyExtension( $mime, $extension ) {
1210 $magic = MimeMagic
::singleton();
1212 if ( !$mime ||
$mime == 'unknown' ||
$mime == 'unknown/unknown' ) {
1213 if ( !$magic->isRecognizableExtension( $extension ) ) {
1214 wfDebug( __METHOD__
. ": passing file with unknown detected mime type; " .
1215 "unrecognized extension '$extension', can't verify\n" );
1219 wfDebug( __METHOD__
. ": rejecting file with unknown detected mime type; " .
1220 "recognized extension '$extension', so probably invalid file\n" );
1226 $match = $magic->isMatchingExtension( $extension, $mime );
1228 if ( $match === null ) {
1229 if ( $magic->getTypesForExtension( $extension ) !== null ) {
1230 wfDebug( __METHOD__
. ": No extension known for $mime, but we know a mime for $extension\n" );
1234 wfDebug( __METHOD__
. ": no file extension known for mime type $mime, passing file\n" );
1238 } elseif ( $match === true ) {
1239 wfDebug( __METHOD__
. ": mime type $mime matches extension $extension, passing file\n" );
1241 /** @todo If it's a bitmap, make sure PHP or ImageMagick resp. can handle it! */
1245 . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
1252 * Heuristic for detecting files that *could* contain JavaScript instructions or
1253 * things that may look like HTML to a browser and are thus
1254 * potentially harmful. The present implementation will produce false
1255 * positives in some situations.
1257 * @param string $file Pathname to the temporary upload file
1258 * @param string $mime The MIME type of the file
1259 * @param string $extension The extension of the file
1260 * @return bool True if the file contains something looking like embedded scripts
1262 public static function detectScript( $file, $mime, $extension ) {
1263 global $wgAllowTitlesInSVG;
1265 # ugly hack: for text files, always look at the entire file.
1266 # For binary field, just check the first K.
1268 if ( strpos( $mime, 'text/' ) === 0 ) {
1269 $chunk = file_get_contents( $file );
1271 $fp = fopen( $file, 'rb' );
1272 $chunk = fread( $fp, 1024 );
1276 $chunk = strtolower( $chunk );
1282 # decode from UTF-16 if needed (could be used for obfuscation).
1283 if ( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
1285 } elseif ( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
1292 $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
1295 $chunk = trim( $chunk );
1297 /** @todo FIXME: Convert from UTF-16 if necessary! */
1298 wfDebug( __METHOD__
. ": checking for embedded scripts and HTML stuff\n" );
1300 # check for HTML doctype
1301 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
1305 // Some browsers will interpret obscure xml encodings as UTF-8, while
1306 // PHP/expat will interpret the given encoding in the xml declaration (T49304)
1307 if ( $extension == 'svg' ||
strpos( $mime, 'image/svg' ) === 0 ) {
1308 if ( self
::checkXMLEncodingMissmatch( $file ) ) {
1314 * Internet Explorer for Windows performs some really stupid file type
1315 * autodetection which can cause it to interpret valid image files as HTML
1316 * and potentially execute JavaScript, creating a cross-site scripting
1319 * Apple's Safari browser also performs some unsafe file type autodetection
1320 * which can cause legitimate files to be interpreted as HTML if the
1321 * web server is not correctly configured to send the right content-type
1322 * (or if you're really uploading plain text and octet streams!)
1324 * Returns true if IE is likely to mistake the given file for HTML.
1325 * Also returns true if Safari would mistake the given file for HTML
1326 * when served with a generic content-type.
1332 '<html', # also in safari
1335 '<script', # also in safari
1339 if ( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
1343 foreach ( $tags as $tag ) {
1344 if ( false !== strpos( $chunk, $tag ) ) {
1345 wfDebug( __METHOD__
. ": found something that may make it be mistaken for html: $tag\n" );
1352 * look for JavaScript
1355 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
1356 $chunk = Sanitizer
::decodeCharReferences( $chunk );
1358 # look for script-types
1359 if ( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
1360 wfDebug( __METHOD__
. ": found script types\n" );
1365 # look for html-style script-urls
1366 if ( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1367 wfDebug( __METHOD__
. ": found html-style script urls\n" );
1372 # look for css-style script-urls
1373 if ( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1374 wfDebug( __METHOD__
. ": found css-style script urls\n" );
1379 wfDebug( __METHOD__
. ": no scripts found\n" );
1385 * Check a whitelist of xml encodings that are known not to be interpreted differently
1386 * by the server's xml parser (expat) and some common browsers.
1388 * @param string $file Pathname to the temporary upload file
1389 * @return bool True if the file contains an encoding that could be misinterpreted
1391 public static function checkXMLEncodingMissmatch( $file ) {
1392 global $wgSVGMetadataCutoff;
1393 $contents = file_get_contents( $file, false, null, -1, $wgSVGMetadataCutoff );
1394 $encodingRegex = '!encoding[ \t\n\r]*=[ \t\n\r]*[\'"](.*?)[\'"]!si';
1396 if ( preg_match( "!<\?xml\b(.*?)\?>!si", $contents, $matches ) ) {
1397 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1398 && !in_array( strtoupper( $encMatch[1] ), self
::$safeXmlEncodings )
1400 wfDebug( __METHOD__
. ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1404 } elseif ( preg_match( "!<\?xml\b!si", $contents ) ) {
1405 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1406 // bytes. There shouldn't be a legitimate reason for this to happen.
1407 wfDebug( __METHOD__
. ": Unmatched XML declaration start\n" );
1410 } elseif ( substr( $contents, 0, 4 ) == "\x4C\x6F\xA7\x94" ) {
1411 // EBCDIC encoded XML
1412 wfDebug( __METHOD__
. ": EBCDIC Encoded XML\n" );
1417 // It's possible the file is encoded with multi-byte encoding, so re-encode attempt to
1418 // detect the encoding in case is specifies an encoding not whitelisted in self::$safeXmlEncodings
1419 $attemptEncodings = [ 'UTF-16', 'UTF-16BE', 'UTF-32', 'UTF-32BE' ];
1420 foreach ( $attemptEncodings as $encoding ) {
1421 MediaWiki\
suppressWarnings();
1422 $str = iconv( $encoding, 'UTF-8', $contents );
1423 MediaWiki\restoreWarnings
();
1424 if ( $str != '' && preg_match( "!<\?xml\b(.*?)\?>!si", $str, $matches ) ) {
1425 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1426 && !in_array( strtoupper( $encMatch[1] ), self
::$safeXmlEncodings )
1428 wfDebug( __METHOD__
. ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1432 } elseif ( $str != '' && preg_match( "!<\?xml\b!si", $str ) ) {
1433 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1434 // bytes. There shouldn't be a legitimate reason for this to happen.
1435 wfDebug( __METHOD__
. ": Unmatched XML declaration start\n" );
1445 * @param string $filename
1446 * @param bool $partial
1447 * @return mixed False of the file is verified (does not contain scripts), array otherwise.
1449 protected function detectScriptInSvg( $filename, $partial ) {
1450 $this->mSVGNSError
= false;
1451 $check = new XmlTypeCheck(
1453 [ $this, 'checkSvgScriptCallback' ],
1456 'processing_instruction_handler' => 'UploadBase::checkSvgPICallback',
1457 'external_dtd_handler' => 'UploadBase::checkSvgExternalDTD',
1460 if ( $check->wellFormed
!== true ) {
1461 // Invalid xml (T60553)
1462 // But only when non-partial (T67724)
1463 return $partial ?
false : [ 'uploadinvalidxml' ];
1464 } elseif ( $check->filterMatch
) {
1465 if ( $this->mSVGNSError
) {
1466 return [ 'uploadscriptednamespace', $this->mSVGNSError
];
1469 return $check->filterMatchType
;
1476 * Callback to filter SVG Processing Instructions.
1477 * @param string $target Processing instruction name
1478 * @param string $data Processing instruction attribute and value
1479 * @return bool (true if the filter identified something bad)
1481 public static function checkSvgPICallback( $target, $data ) {
1482 // Don't allow external stylesheets (T59550)
1483 if ( preg_match( '/xml-stylesheet/i', $target ) ) {
1484 return [ 'upload-scripted-pi-callback' ];
1491 * Verify that DTD urls referenced are only the standard dtds
1493 * Browsers seem to ignore external dtds. However just to be on the
1494 * safe side, only allow dtds from the svg standard.
1496 * @param string $type PUBLIC or SYSTEM
1497 * @param string $publicId The well-known public identifier for the dtd
1498 * @param string $systemId The url for the external dtd
1500 public static function checkSvgExternalDTD( $type, $publicId, $systemId ) {
1501 // This doesn't include the XHTML+MathML+SVG doctype since we don't
1502 // allow XHTML anyways.
1504 'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd',
1505 'http://www.w3.org/TR/2001/REC-SVG-20010904/DTD/svg10.dtd',
1506 'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-basic.dtd',
1507 'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-tiny.dtd',
1508 // https://phabricator.wikimedia.org/T168856
1509 'http://www.w3.org/TR/2001/PR-SVG-20010719/DTD/svg10.dtd',
1511 if ( $type !== 'PUBLIC'
1512 ||
!in_array( $systemId, $allowedDTDs )
1513 ||
strpos( $publicId, "-//W3C//" ) !== 0
1515 return [ 'upload-scripted-dtd' ];
1521 * @todo Replace this with a whitelist filter!
1522 * @param string $element
1523 * @param array $attribs
1526 public function checkSvgScriptCallback( $element, $attribs, $data = null ) {
1527 list( $namespace, $strippedElement ) = $this->splitXmlNamespace( $element );
1529 // We specifically don't include:
1530 // http://www.w3.org/1999/xhtml (T62771)
1531 static $validNamespaces = [
1534 'http://creativecommons.org/ns#',
1535 'http://inkscape.sourceforge.net/dtd/sodipodi-0.dtd',
1536 'http://ns.adobe.com/adobeillustrator/10.0/',
1537 'http://ns.adobe.com/adobesvgviewerextensions/3.0/',
1538 'http://ns.adobe.com/extensibility/1.0/',
1539 'http://ns.adobe.com/flows/1.0/',
1540 'http://ns.adobe.com/illustrator/1.0/',
1541 'http://ns.adobe.com/imagereplacement/1.0/',
1542 'http://ns.adobe.com/pdf/1.3/',
1543 'http://ns.adobe.com/photoshop/1.0/',
1544 'http://ns.adobe.com/saveforweb/1.0/',
1545 'http://ns.adobe.com/variables/1.0/',
1546 'http://ns.adobe.com/xap/1.0/',
1547 'http://ns.adobe.com/xap/1.0/g/',
1548 'http://ns.adobe.com/xap/1.0/g/img/',
1549 'http://ns.adobe.com/xap/1.0/mm/',
1550 'http://ns.adobe.com/xap/1.0/rights/',
1551 'http://ns.adobe.com/xap/1.0/stype/dimensions#',
1552 'http://ns.adobe.com/xap/1.0/stype/font#',
1553 'http://ns.adobe.com/xap/1.0/stype/manifestitem#',
1554 'http://ns.adobe.com/xap/1.0/stype/resourceevent#',
1555 'http://ns.adobe.com/xap/1.0/stype/resourceref#',
1556 'http://ns.adobe.com/xap/1.0/t/pg/',
1557 'http://purl.org/dc/elements/1.1/',
1558 'http://purl.org/dc/elements/1.1',
1559 'http://schemas.microsoft.com/visio/2003/svgextensions/',
1560 'http://sodipodi.sourceforge.net/dtd/sodipodi-0.dtd',
1561 'http://taptrix.com/inkpad/svg_extensions',
1562 'http://web.resource.org/cc/',
1563 'http://www.freesoftware.fsf.org/bkchem/cdml',
1564 'http://www.inkscape.org/namespaces/inkscape',
1565 'http://www.opengis.net/gml',
1566 'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
1567 'http://www.w3.org/2000/svg',
1568 'http://www.w3.org/tr/rec-rdf-syntax/',
1569 'http://www.w3.org/2000/01/rdf-schema#',
1572 // Inkscape mangles namespace definitions created by Adobe Illustrator.
1573 // This is nasty but harmless. (T144827)
1574 $isBuggyInkscape = preg_match( '/^&(#38;)*ns_[a-z_]+;$/', $namespace );
1576 if ( !( $isBuggyInkscape ||
in_array( $namespace, $validNamespaces ) ) ) {
1577 wfDebug( __METHOD__
. ": Non-svg namespace '$namespace' in uploaded file.\n" );
1578 /** @todo Return a status object to a closure in XmlTypeCheck, for MW1.21+ */
1579 $this->mSVGNSError
= $namespace;
1585 * check for elements that can contain javascript
1587 if ( $strippedElement == 'script' ) {
1588 wfDebug( __METHOD__
. ": Found script element '$element' in uploaded file.\n" );
1590 return [ 'uploaded-script-svg', $strippedElement ];
1593 # e.g., <svg xmlns="http://www.w3.org/2000/svg">
1594 # <handler xmlns:ev="http://www.w3.org/2001/xml-events" ev:event="load">alert(1)</handler> </svg>
1595 if ( $strippedElement == 'handler' ) {
1596 wfDebug( __METHOD__
. ": Found scriptable element '$element' in uploaded file.\n" );
1598 return [ 'uploaded-script-svg', $strippedElement ];
1601 # SVG reported in Feb '12 that used xml:stylesheet to generate javascript block
1602 if ( $strippedElement == 'stylesheet' ) {
1603 wfDebug( __METHOD__
. ": Found scriptable element '$element' in uploaded file.\n" );
1605 return [ 'uploaded-script-svg', $strippedElement ];
1608 # Block iframes, in case they pass the namespace check
1609 if ( $strippedElement == 'iframe' ) {
1610 wfDebug( __METHOD__
. ": iframe in uploaded file.\n" );
1612 return [ 'uploaded-script-svg', $strippedElement ];
1616 if ( $strippedElement == 'style'
1617 && self
::checkCssFragment( Sanitizer
::normalizeCss( $data ) )
1619 wfDebug( __METHOD__
. ": hostile css in style element.\n" );
1620 return [ 'uploaded-hostile-svg' ];
1623 foreach ( $attribs as $attrib => $value ) {
1624 $stripped = $this->stripXmlNamespace( $attrib );
1625 $value = strtolower( $value );
1627 if ( substr( $stripped, 0, 2 ) == 'on' ) {
1629 . ": Found event-handler attribute '$attrib'='$value' in uploaded file.\n" );
1631 return [ 'uploaded-event-handler-on-svg', $attrib, $value ];
1634 # Do not allow relative links, or unsafe url schemas.
1635 # For <a> tags, only data:, http: and https: and same-document
1636 # fragment links are allowed. For all other tags, only data:
1637 # and fragment are allowed.
1638 if ( $stripped == 'href'
1640 && strpos( $value, 'data:' ) !== 0
1641 && strpos( $value, '#' ) !== 0
1643 if ( !( $strippedElement === 'a'
1644 && preg_match( '!^https?://!i', $value ) )
1646 wfDebug( __METHOD__
. ": Found href attribute <$strippedElement "
1647 . "'$attrib'='$value' in uploaded file.\n" );
1649 return [ 'uploaded-href-attribute-svg', $strippedElement, $attrib, $value ];
1653 # only allow data: targets that should be safe. This prevents vectors like,
1654 # image/svg, text/xml, application/xml, and text/html, which can contain scripts
1655 if ( $stripped == 'href' && strncasecmp( 'data:', $value, 5 ) === 0 ) {
1656 // rfc2397 parameters. This is only slightly slower than (;[\w;]+)*.
1657 // @codingStandardsIgnoreStart Generic.Files.LineLength
1658 $parameters = '(?>;[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+=(?>[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+|"(?>[\0-\x0c\x0e-\x21\x23-\x5b\x5d-\x7f]+|\\\\[\0-\x7f])*"))*(?:;base64)?';
1659 // @codingStandardsIgnoreEnd
1661 if ( !preg_match( "!^data:\s*image/(gif|jpeg|jpg|png)$parameters,!i", $value ) ) {
1662 wfDebug( __METHOD__
. ": Found href to unwhitelisted data: uri "
1663 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1664 return [ 'uploaded-href-unsafe-target-svg', $strippedElement, $attrib, $value ];
1668 # Change href with animate from (http://html5sec.org/#137).
1669 if ( $stripped === 'attributename'
1670 && $strippedElement === 'animate'
1671 && $this->stripXmlNamespace( $value ) == 'href'
1673 wfDebug( __METHOD__
. ": Found animate that might be changing href using from "
1674 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1676 return [ 'uploaded-animate-svg', $strippedElement, $attrib, $value ];
1679 # use set/animate to add event-handler attribute to parent
1680 if ( ( $strippedElement == 'set' ||
$strippedElement == 'animate' )
1681 && $stripped == 'attributename'
1682 && substr( $value, 0, 2 ) == 'on'
1684 wfDebug( __METHOD__
. ": Found svg setting event-handler attribute with "
1685 . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1687 return [ 'uploaded-setting-event-handler-svg', $strippedElement, $stripped, $value ];
1690 # use set to add href attribute to parent element
1691 if ( $strippedElement == 'set'
1692 && $stripped == 'attributename'
1693 && strpos( $value, 'href' ) !== false
1695 wfDebug( __METHOD__
. ": Found svg setting href attribute '$value' in uploaded file.\n" );
1697 return [ 'uploaded-setting-href-svg' ];
1700 # use set to add a remote / data / script target to an element
1701 if ( $strippedElement == 'set'
1702 && $stripped == 'to'
1703 && preg_match( '!(http|https|data|script):!sim', $value )
1705 wfDebug( __METHOD__
. ": Found svg setting attribute to '$value' in uploaded file.\n" );
1707 return [ 'uploaded-wrong-setting-svg', $value ];
1710 # use handler attribute with remote / data / script
1711 if ( $stripped == 'handler' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1712 wfDebug( __METHOD__
. ": Found svg setting handler with remote/data/script "
1713 . "'$attrib'='$value' in uploaded file.\n" );
1715 return [ 'uploaded-setting-handler-svg', $attrib, $value ];
1718 # use CSS styles to bring in remote code
1719 if ( $stripped == 'style'
1720 && self
::checkCssFragment( Sanitizer
::normalizeCss( $value ) )
1722 wfDebug( __METHOD__
. ": Found svg setting a style with "
1723 . "remote url '$attrib'='$value' in uploaded file.\n" );
1724 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1727 # Several attributes can include css, css character escaping isn't allowed
1728 $cssAttrs = [ 'font', 'clip-path', 'fill', 'filter', 'marker',
1729 'marker-end', 'marker-mid', 'marker-start', 'mask', 'stroke' ];
1730 if ( in_array( $stripped, $cssAttrs )
1731 && self
::checkCssFragment( $value )
1733 wfDebug( __METHOD__
. ": Found svg setting a style with "
1734 . "remote url '$attrib'='$value' in uploaded file.\n" );
1735 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1738 # image filters can pull in url, which could be svg that executes scripts
1739 if ( $strippedElement == 'image'
1740 && $stripped == 'filter'
1741 && preg_match( '!url\s*\(!sim', $value )
1743 wfDebug( __METHOD__
. ": Found image filter with url: "
1744 . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1746 return [ 'uploaded-image-filter-svg', $strippedElement, $stripped, $value ];
1750 return false; // No scripts detected
1754 * Check a block of CSS or CSS fragment for anything that looks like
1755 * it is bringing in remote code.
1756 * @param string $value a string of CSS
1757 * @param bool $propOnly only check css properties (start regex with :)
1758 * @return bool true if the CSS contains an illegal string, false if otherwise
1760 private static function checkCssFragment( $value ) {
1761 # Forbid external stylesheets, for both reliability and to protect viewer's privacy
1762 if ( stripos( $value, '@import' ) !== false ) {
1766 # We allow @font-face to embed fonts with data: urls, so we snip the string
1767 # 'url' out so this case won't match when we check for urls below
1768 $pattern = '!(@font-face\s*{[^}]*src:)url(\("data:;base64,)!im';
1769 $value = preg_replace( $pattern, '$1$2', $value );
1771 # Check for remote and executable CSS. Unlike in Sanitizer::checkCss, the CSS
1772 # properties filter and accelerator don't seem to be useful for xss in SVG files.
1773 # Expression and -o-link don't seem to work either, but filtering them here in case.
1774 # Additionally, we catch remote urls like url("http:..., url('http:..., url(http:...,
1775 # but not local ones such as url("#..., url('#..., url(#....
1776 if ( preg_match( '!expression
1778 | -o-link-source\s*:
1779 | -o-replace\s*:!imx', $value ) ) {
1783 if ( preg_match_all(
1784 "!(\s*(url|image|image-set)\s*\(\s*[\"']?\s*[^#]+.*?\))!sim",
1789 # TODO: redo this in one regex. Until then, url("#whatever") matches the first
1790 foreach ( $matches[1] as $match ) {
1791 if ( !preg_match( "!\s*(url|image|image-set)\s*\(\s*(#|'#|\"#)!im", $match ) ) {
1797 if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ) {
1805 * Divide the element name passed by the xml parser to the callback into URI and prifix.
1806 * @param string $element
1807 * @return array Containing the namespace URI and prefix
1809 private static function splitXmlNamespace( $element ) {
1810 // 'http://www.w3.org/2000/svg:script' -> [ 'http://www.w3.org/2000/svg', 'script' ]
1811 $parts = explode( ':', strtolower( $element ) );
1812 $name = array_pop( $parts );
1813 $ns = implode( ':', $parts );
1815 return [ $ns, $name ];
1819 * @param string $name
1822 private function stripXmlNamespace( $name ) {
1823 // 'http://www.w3.org/2000/svg:script' -> 'script'
1824 $parts = explode( ':', strtolower( $name ) );
1826 return array_pop( $parts );
1830 * Generic wrapper function for a virus scanner program.
1831 * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
1832 * $wgAntivirusRequired may be used to deny upload if the scan fails.
1834 * @param string $file Pathname to the temporary upload file
1835 * @return mixed False if not virus is found, null if the scan fails or is disabled,
1836 * or a string containing feedback from the virus scanner if a virus was found.
1837 * If textual feedback is missing but a virus was found, this function returns true.
1839 public static function detectVirus( $file ) {
1840 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
1842 if ( !$wgAntivirus ) {
1843 wfDebug( __METHOD__
. ": virus scanner disabled\n" );
1848 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
1849 wfDebug( __METHOD__
. ": unknown virus scanner: $wgAntivirus\n" );
1850 $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
1851 [ 'virus-badscanner', $wgAntivirus ] );
1853 return wfMessage( 'virus-unknownscanner' )->text() . " $wgAntivirus";
1856 # look up scanner configuration
1857 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
1858 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
1859 $msgPattern = isset( $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ) ?
1860 $wgAntivirusSetup[$wgAntivirus]['messagepattern'] : null;
1862 if ( strpos( $command, "%f" ) === false ) {
1863 # simple pattern: append file to scan
1864 $command .= " " . wfEscapeShellArg( $file );
1866 # complex pattern: replace "%f" with file to scan
1867 $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
1870 wfDebug( __METHOD__
. ": running virus scan: $command \n" );
1872 # execute virus scanner
1875 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1876 # that does not seem to be worth the pain.
1877 # Ask me (Duesentrieb) about it if it's ever needed.
1878 $output = wfShellExecWithStderr( $command, $exitCode );
1880 # map exit code to AV_xxx constants.
1881 $mappedCode = $exitCode;
1882 if ( $exitCodeMap ) {
1883 if ( isset( $exitCodeMap[$exitCode] ) ) {
1884 $mappedCode = $exitCodeMap[$exitCode];
1885 } elseif ( isset( $exitCodeMap["*"] ) ) {
1886 $mappedCode = $exitCodeMap["*"];
1890 /* NB: AV_NO_VIRUS is 0 but AV_SCAN_FAILED is false,
1891 * so we need the strict equalities === and thus can't use a switch here
1893 if ( $mappedCode === AV_SCAN_FAILED
) {
1894 # scan failed (code was mapped to false by $exitCodeMap)
1895 wfDebug( __METHOD__
. ": failed to scan $file (code $exitCode).\n" );
1897 $output = $wgAntivirusRequired
1898 ?
wfMessage( 'virus-scanfailed', [ $exitCode ] )->text()
1900 } elseif ( $mappedCode === AV_SCAN_ABORTED
) {
1901 # scan failed because filetype is unknown (probably imune)
1902 wfDebug( __METHOD__
. ": unsupported file type $file (code $exitCode).\n" );
1904 } elseif ( $mappedCode === AV_NO_VIRUS
) {
1906 wfDebug( __METHOD__
. ": file passed virus scan.\n" );
1909 $output = trim( $output );
1912 $output = true; # if there's no output, return true
1913 } elseif ( $msgPattern ) {
1915 if ( preg_match( $msgPattern, $output, $groups ) ) {
1917 $output = $groups[1];
1922 wfDebug( __METHOD__
. ": FOUND VIRUS! scanner feedback: $output \n" );
1929 * Check if there's an overwrite conflict and, if so, if restrictions
1930 * forbid this user from performing the upload.
1934 * @return mixed True on success, array on failure
1936 private function checkOverwrite( $user ) {
1937 // First check whether the local file can be overwritten
1938 $file = $this->getLocalFile();
1939 $file->load( File
::READ_LATEST
);
1940 if ( $file->exists() ) {
1941 if ( !self
::userCanReUpload( $user, $file ) ) {
1942 return [ 'fileexists-forbidden', $file->getName() ];
1948 /* Check shared conflicts: if the local file does not exist, but
1949 * wfFindFile finds a file, it exists in a shared repository.
1951 $file = wfFindFile( $this->getTitle(), [ 'latest' => true ] );
1952 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1953 return [ 'fileexists-shared-forbidden', $file->getName() ];
1960 * Check if a user is the last uploader
1966 public static function userCanReUpload( User
$user, File
$img ) {
1967 if ( $user->isAllowed( 'reupload' ) ) {
1968 return true; // non-conditional
1969 } elseif ( !$user->isAllowed( 'reupload-own' ) ) {
1973 if ( !( $img instanceof LocalFile
) ) {
1979 return $user->getId() == $img->getUser( 'id' );
1983 * Helper function that does various existence checks for a file.
1984 * The following checks are performed:
1986 * - Article with the same name as the file exists
1987 * - File exists with normalized extension
1988 * - The file looks like a thumbnail and the original exists
1990 * @param File $file The File object to check
1991 * @return mixed False if the file does not exists, else an array
1993 public static function getExistsWarning( $file ) {
1994 if ( $file->exists() ) {
1995 return [ 'warning' => 'exists', 'file' => $file ];
1998 if ( $file->getTitle()->getArticleID() ) {
1999 return [ 'warning' => 'page-exists', 'file' => $file ];
2002 if ( strpos( $file->getName(), '.' ) == false ) {
2003 $partname = $file->getName();
2006 $n = strrpos( $file->getName(), '.' );
2007 $extension = substr( $file->getName(), $n +
1 );
2008 $partname = substr( $file->getName(), 0, $n );
2010 $normalizedExtension = File
::normalizeExtension( $extension );
2012 if ( $normalizedExtension != $extension ) {
2013 // We're not using the normalized form of the extension.
2014 // Normal form is lowercase, using most common of alternate
2015 // extensions (eg 'jpg' rather than 'JPEG').
2017 // Check for another file using the normalized form...
2018 $nt_lc = Title
::makeTitle( NS_FILE
, "{$partname}.{$normalizedExtension}" );
2019 $file_lc = wfLocalFile( $nt_lc );
2021 if ( $file_lc->exists() ) {
2023 'warning' => 'exists-normalized',
2025 'normalizedFile' => $file_lc
2030 // Check for files with the same name but a different extension
2031 $similarFiles = RepoGroup
::singleton()->getLocalRepo()->findFilesByPrefix(
2032 "{$partname}.", 1 );
2033 if ( count( $similarFiles ) ) {
2035 'warning' => 'exists-normalized',
2037 'normalizedFile' => $similarFiles[0],
2041 if ( self
::isThumbName( $file->getName() ) ) {
2042 # Check for filenames like 50px- or 180px-, these are mostly thumbnails
2043 $nt_thb = Title
::newFromText(
2044 substr( $partname, strpos( $partname, '-' ) +
1 ) . '.' . $extension,
2047 $file_thb = wfLocalFile( $nt_thb );
2048 if ( $file_thb->exists() ) {
2050 'warning' => 'thumb',
2052 'thumbFile' => $file_thb
2055 // File does not exist, but we just don't like the name
2057 'warning' => 'thumb-name',
2059 'thumbFile' => $file_thb
2064 foreach ( self
::getFilenamePrefixBlacklist() as $prefix ) {
2065 if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
2067 'warning' => 'bad-prefix',
2078 * Helper function that checks whether the filename looks like a thumbnail
2079 * @param string $filename
2082 public static function isThumbName( $filename ) {
2083 $n = strrpos( $filename, '.' );
2084 $partname = $n ?
substr( $filename, 0, $n ) : $filename;
2087 substr( $partname, 3, 3 ) == 'px-' ||
2088 substr( $partname, 2, 3 ) == 'px-'
2090 preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) );
2094 * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
2096 * @return array List of prefixes
2098 public static function getFilenamePrefixBlacklist() {
2100 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
2101 if ( !$message->isDisabled() ) {
2102 $lines = explode( "\n", $message->plain() );
2103 foreach ( $lines as $line ) {
2104 // Remove comment lines
2105 $comment = substr( trim( $line ), 0, 1 );
2106 if ( $comment == '#' ||
$comment == '' ) {
2109 // Remove additional comments after a prefix
2110 $comment = strpos( $line, '#' );
2111 if ( $comment > 0 ) {
2112 $line = substr( $line, 0, $comment - 1 );
2114 $blacklist[] = trim( $line );
2122 * Gets image info about the file just uploaded.
2124 * Also has the effect of setting metadata to be an 'indexed tag name' in
2125 * returned API result if 'metadata' was requested. Oddly, we have to pass
2126 * the "result" object down just so it can do that with the appropriate
2127 * format, presumably.
2129 * @param ApiResult $result
2130 * @return array Image info
2132 public function getImageInfo( $result ) {
2133 $localFile = $this->getLocalFile();
2134 $stashFile = $this->getStashFile();
2135 // Calling a different API module depending on whether the file was stashed is less than optimal.
2136 // In fact, calling API modules here at all is less than optimal. Maybe it should be refactored.
2138 $imParam = ApiQueryStashImageInfo
::getPropertyNames();
2139 $info = ApiQueryStashImageInfo
::getInfo( $stashFile, array_flip( $imParam ), $result );
2141 $imParam = ApiQueryImageInfo
::getPropertyNames();
2142 $info = ApiQueryImageInfo
::getInfo( $localFile, array_flip( $imParam ), $result );
2149 * @param array $error
2152 public function convertVerifyErrorToStatus( $error ) {
2153 $code = $error['status'];
2154 unset( $code['status'] );
2156 return Status
::newFatal( $this->getVerificationErrorCode( $code ), $error );
2160 * Get the MediaWiki maximum uploaded file size for given type of upload, based on
2163 * @param null|string $forType
2166 public static function getMaxUploadSize( $forType = null ) {
2167 global $wgMaxUploadSize;
2169 if ( is_array( $wgMaxUploadSize ) ) {
2170 if ( !is_null( $forType ) && isset( $wgMaxUploadSize[$forType] ) ) {
2171 return $wgMaxUploadSize[$forType];
2173 return $wgMaxUploadSize['*'];
2176 return intval( $wgMaxUploadSize );
2181 * Get the PHP maximum uploaded file size, based on ini settings. If there is no limit or the
2182 * limit can't be guessed, returns a very large number (PHP_INT_MAX).
2187 public static function getMaxPhpUploadSize() {
2188 $phpMaxFileSize = wfShorthandToInteger(
2189 ini_get( 'upload_max_filesize' ) ?
: ini_get( 'hhvm.server.upload.upload_max_file_size' ),
2192 $phpMaxPostSize = wfShorthandToInteger(
2193 ini_get( 'post_max_size' ) ?
: ini_get( 'hhvm.server.max_post_size' ),
2196 return min( $phpMaxFileSize, $phpMaxPostSize );
2200 * Get the current status of a chunked upload (used for polling)
2202 * The value will be read from cache.
2205 * @param string $statusKey
2206 * @return Status[]|bool
2208 public static function getSessionStatus( User
$user, $statusKey ) {
2209 $cache = MediaWikiServices
::getInstance()->getMainObjectStash();
2210 $key = $cache->makeKey( 'uploadstatus', $user->getId() ?
: md5( $user->getName() ), $statusKey );
2212 return $cache->get( $key );
2216 * Set the current status of a chunked upload (used for polling)
2218 * The value will be set in cache for 1 day
2221 * @param string $statusKey
2222 * @param array|bool $value
2225 public static function setSessionStatus( User
$user, $statusKey, $value ) {
2226 $cache = MediaWikiServices
::getInstance()->getMainObjectStash();
2227 $key = $cache->makeKey( 'uploadstatus', $user->getId() ?
: md5( $user->getName() ), $statusKey );
2229 if ( $value === false ) {
2230 $cache->delete( $key );
2232 $cache->set( $key, $value, $cache::TTL_DAY
);