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 return $code_to_status[$error] ??
'unknown-error';
105 * Returns true if uploads are enabled.
106 * Can be override by subclasses.
109 public static function isEnabled() {
110 global $wgEnableUploads;
112 if ( !$wgEnableUploads ) {
116 # Check php's file_uploads setting
117 return wfIsHHVM() ||
wfIniGetBool( 'file_uploads' );
121 * Returns true if the user can use this upload module or else a string
122 * identifying the missing permission.
123 * Can be overridden by subclasses.
126 * @return bool|string
128 public static function isAllowed( $user ) {
129 foreach ( [ 'upload', 'edit' ] as $permission ) {
130 if ( !$user->isAllowed( $permission ) ) {
139 * Returns true if the user has surpassed the upload rate limit, false otherwise.
144 public static function isThrottled( $user ) {
145 return $user->pingLimiter( 'upload' );
148 // Upload handlers. Should probably just be a global.
149 private static $uploadHandlers = [ 'Stash', 'File', 'Url' ];
152 * Create a form of UploadBase depending on wpSourceType and initializes it
154 * @param WebRequest &$request
155 * @param string|null $type
156 * @return null|UploadBase
158 public static function createFromRequest( &$request, $type = null ) {
159 $type = $type ?
: $request->getVal( 'wpSourceType', 'File' );
165 // Get the upload class
166 $type = ucfirst( $type );
168 // Give hooks the chance to handle this request
170 Hooks
::run( 'UploadCreateFromRequest', [ $type, &$className ] );
171 if ( is_null( $className ) ) {
172 $className = 'UploadFrom' . $type;
173 wfDebug( __METHOD__
. ": class name: $className\n" );
174 if ( !in_array( $type, self
::$uploadHandlers ) ) {
179 // Check whether this upload class is enabled
180 if ( !call_user_func( [ $className, 'isEnabled' ] ) ) {
184 // Check whether the request is valid
185 if ( !call_user_func( [ $className, 'isValidRequest' ], $request ) ) {
189 /** @var UploadBase $handler */
190 $handler = new $className;
192 $handler->initializeFromRequest( $request );
198 * Check whether a request if valid for this handler
199 * @param WebRequest $request
202 public static function isValidRequest( $request ) {
206 public function __construct() {
210 * Returns the upload type. Should be overridden by child classes
215 public function getSourceType() {
220 * Initialize the path information
221 * @param string $name The desired destination name
222 * @param string $tempPath The temporary path
223 * @param int $fileSize The file size
224 * @param bool $removeTempFile (false) remove the temporary file?
225 * @throws MWException
227 public function initializePathInfo( $name, $tempPath, $fileSize, $removeTempFile = false ) {
228 $this->mDesiredDestName
= $name;
229 if ( FileBackend
::isStoragePath( $tempPath ) ) {
230 throw new MWException( __METHOD__
. " given storage path `$tempPath`." );
233 $this->setTempFile( $tempPath, $fileSize );
234 $this->mRemoveTempFile
= $removeTempFile;
238 * Initialize from a WebRequest. Override this in a subclass.
240 * @param WebRequest &$request
242 abstract public function initializeFromRequest( &$request );
245 * @param string $tempPath File system path to temporary file containing the upload
246 * @param int|null $fileSize
248 protected function setTempFile( $tempPath, $fileSize = null ) {
249 $this->mTempPath
= $tempPath;
250 $this->mFileSize
= $fileSize ?
: null;
251 if ( strlen( $this->mTempPath
) && file_exists( $this->mTempPath
) ) {
252 $this->tempFileObj
= new TempFSFile( $this->mTempPath
);
254 $this->mFileSize
= filesize( $this->mTempPath
);
257 $this->tempFileObj
= null;
262 * Fetch the file. Usually a no-op
265 public function fetchFile() {
266 return Status
::newGood();
270 * Return true if the file is empty
273 public function isEmptyFile() {
274 return empty( $this->mFileSize
);
278 * Return the file size
281 public function getFileSize() {
282 return $this->mFileSize
;
286 * Get the base 36 SHA1 of the file
289 public function getTempFileSha1Base36() {
290 return FSFile
::getSha1Base36FromPath( $this->mTempPath
);
294 * @param string $srcPath The source path
295 * @return string|bool The real path if it was a virtual URL Returns false on failure
297 public function getRealPath( $srcPath ) {
298 $repo = RepoGroup
::singleton()->getLocalRepo();
299 if ( $repo->isVirtualUrl( $srcPath ) ) {
300 /** @todo Just make uploads work with storage paths UploadFromStash
301 * loads files via virtual URLs.
303 $tmpFile = $repo->getLocalCopy( $srcPath );
305 $tmpFile->bind( $this ); // keep alive with $this
307 $path = $tmpFile ?
$tmpFile->getPath() : false;
316 * Verify whether the upload is sane.
317 * @return mixed Const self::OK or else an array with error information
319 public function verifyUpload() {
321 * If there was no filename or a zero size given, give up quick.
323 if ( $this->isEmptyFile() ) {
324 return [ 'status' => self
::EMPTY_FILE
];
328 * Honor $wgMaxUploadSize
330 $maxSize = self
::getMaxUploadSize( $this->getSourceType() );
331 if ( $this->mFileSize
> $maxSize ) {
333 'status' => self
::FILE_TOO_LARGE
,
339 * Look at the contents of the file; if we can recognize the
340 * type but it's corrupt or data of the wrong type, we should
341 * probably not accept it.
343 $verification = $this->verifyFile();
344 if ( $verification !== true ) {
346 'status' => self
::VERIFICATION_ERROR
,
347 'details' => $verification
352 * Make sure this file can be created
354 $result = $this->validateName();
355 if ( $result !== true ) {
360 if ( !Hooks
::run( 'UploadVerification',
361 [ $this->mDestName
, $this->mTempPath
, &$error ], '1.28' )
363 return [ 'status' => self
::HOOK_ABORTED
, 'error' => $error ];
366 return [ 'status' => self
::OK
];
370 * Verify that the name is valid and, if necessary, that we can overwrite
372 * @return mixed True if valid, otherwise and array with 'status'
375 public function validateName() {
376 $nt = $this->getTitle();
377 if ( is_null( $nt ) ) {
378 $result = [ 'status' => $this->mTitleError
];
379 if ( $this->mTitleError
== self
::ILLEGAL_FILENAME
) {
380 $result['filtered'] = $this->mFilteredName
;
382 if ( $this->mTitleError
== self
::FILETYPE_BADTYPE
) {
383 $result['finalExt'] = $this->mFinalExtension
;
384 if ( count( $this->mBlackListedExtensions
) ) {
385 $result['blacklistedExt'] = $this->mBlackListedExtensions
;
391 $this->mDestName
= $this->getLocalFile()->getName();
397 * Verify the MIME type.
399 * @note Only checks that it is not an evil MIME. The "does it have
400 * correct extension given its MIME type?" check is in verifyFile.
401 * in `verifyFile()` that MIME type and file extension correlate.
402 * @param string $mime Representing the MIME
403 * @return mixed True if the file is verified, an array otherwise
405 protected function verifyMimeType( $mime ) {
406 global $wgVerifyMimeType;
407 if ( $wgVerifyMimeType ) {
408 wfDebug( "mime: <$mime> extension: <{$this->mFinalExtension}>\n" );
409 global $wgMimeTypeBlacklist;
410 if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
411 return [ 'filetype-badmime', $mime ];
414 # Check what Internet Explorer would detect
415 $fp = fopen( $this->mTempPath
, 'rb' );
416 $chunk = fread( $fp, 256 );
419 $magic = MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer();
420 $extMime = $magic->guessTypesForExtension( $this->mFinalExtension
);
421 $ieTypes = $magic->getIEMimeTypes( $this->mTempPath
, $chunk, $extMime );
422 foreach ( $ieTypes as $ieType ) {
423 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
424 return [ 'filetype-bad-ie-mime', $ieType ];
433 * Verifies that it's ok to include the uploaded file
435 * @return mixed True of the file is verified, array otherwise.
437 protected function verifyFile() {
438 global $wgVerifyMimeType, $wgDisableUploadScriptChecks;
440 $status = $this->verifyPartialFile();
441 if ( $status !== true ) {
445 $mwProps = new MWFileProps( MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer() );
446 $this->mFileProps
= $mwProps->getPropsFromPath( $this->mTempPath
, $this->mFinalExtension
);
447 $mime = $this->mFileProps
['mime'];
449 if ( $wgVerifyMimeType ) {
450 # XXX: Missing extension will be caught by validateName() via getTitle()
451 if ( $this->mFinalExtension
!= '' && !$this->verifyExtension( $mime, $this->mFinalExtension
) ) {
452 return [ 'filetype-mime-mismatch', $this->mFinalExtension
, $mime ];
456 # check for htmlish code and javascript
457 if ( !$wgDisableUploadScriptChecks ) {
458 if ( $this->mFinalExtension
== 'svg' ||
$mime == 'image/svg+xml' ) {
459 $svgStatus = $this->detectScriptInSvg( $this->mTempPath
, false );
460 if ( $svgStatus !== false ) {
466 $handler = MediaHandler
::getHandler( $mime );
468 $handlerStatus = $handler->verifyUpload( $this->mTempPath
);
469 if ( !$handlerStatus->isOK() ) {
470 $errors = $handlerStatus->getErrorsArray();
472 return reset( $errors );
477 Hooks
::run( 'UploadVerifyFile', [ $this, $mime, &$error ] );
478 if ( $error !== true ) {
479 if ( !is_array( $error ) ) {
485 wfDebug( __METHOD__
. ": all clear; passing.\n" );
491 * A verification routine suitable for partial files
493 * Runs the blacklist checks, but not any checks that may
494 * assume the entire file is present.
496 * @return mixed True for valid or array with error message key.
498 protected function verifyPartialFile() {
499 global $wgAllowJavaUploads, $wgDisableUploadScriptChecks;
501 # getTitle() sets some internal parameters like $this->mFinalExtension
504 $mwProps = new MWFileProps( MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer() );
505 $this->mFileProps
= $mwProps->getPropsFromPath( $this->mTempPath
, $this->mFinalExtension
);
507 # check MIME type, if desired
508 $mime = $this->mFileProps
['file-mime'];
509 $status = $this->verifyMimeType( $mime );
510 if ( $status !== true ) {
514 # check for htmlish code and javascript
515 if ( !$wgDisableUploadScriptChecks ) {
516 if ( self
::detectScript( $this->mTempPath
, $mime, $this->mFinalExtension
) ) {
517 return [ 'uploadscripted' ];
519 if ( $this->mFinalExtension
== 'svg' ||
$mime == 'image/svg+xml' ) {
520 $svgStatus = $this->detectScriptInSvg( $this->mTempPath
, true );
521 if ( $svgStatus !== false ) {
527 # Check for Java applets, which if uploaded can bypass cross-site
529 if ( !$wgAllowJavaUploads ) {
530 $this->mJavaDetected
= false;
531 $zipStatus = ZipDirectoryReader
::read( $this->mTempPath
,
532 [ $this, 'zipEntryCallback' ] );
533 if ( !$zipStatus->isOK() ) {
534 $errors = $zipStatus->getErrorsArray();
535 $error = reset( $errors );
536 if ( $error[0] !== 'zip-wrong-format' ) {
540 if ( $this->mJavaDetected
) {
541 return [ 'uploadjava' ];
545 # Scan the uploaded file for viruses
546 $virus = $this->detectVirus( $this->mTempPath
);
548 return [ 'uploadvirus', $virus ];
555 * Callback for ZipDirectoryReader to detect Java class files.
557 * @param array $entry
559 public function zipEntryCallback( $entry ) {
560 $names = [ $entry['name'] ];
562 // If there is a null character, cut off the name at it, because JDK's
563 // ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name
564 // were constructed which had ".class\0" followed by a string chosen to
565 // make the hash collide with the truncated name, that file could be
566 // returned in response to a request for the .class file.
567 $nullPos = strpos( $entry['name'], "\000" );
568 if ( $nullPos !== false ) {
569 $names[] = substr( $entry['name'], 0, $nullPos );
572 // If there is a trailing slash in the file name, we have to strip it,
573 // because that's what ZIP_GetEntry() does.
574 if ( preg_grep( '!\.class/?$!', $names ) ) {
575 $this->mJavaDetected
= true;
580 * Alias for verifyTitlePermissions. The function was originally
581 * 'verifyPermissions', but that suggests it's checking the user, when it's
582 * really checking the title + user combination.
584 * @param User $user User object to verify the permissions against
585 * @return mixed An array as returned by getUserPermissionsErrors or true
586 * in case the user has proper permissions.
588 public function verifyPermissions( $user ) {
589 return $this->verifyTitlePermissions( $user );
593 * Check whether the user can edit, upload and create the image. This
594 * checks only against the current title; if it returns errors, it may
595 * very well be that another title will not give errors. Therefore
596 * isAllowed() should be called as well for generic is-user-blocked or
597 * can-user-upload checking.
599 * @param User $user User object to verify the permissions against
600 * @return mixed An array as returned by getUserPermissionsErrors or true
601 * in case the user has proper permissions.
603 public function verifyTitlePermissions( $user ) {
605 * If the image is protected, non-sysop users won't be able
606 * to modify it by uploading a new revision.
608 $nt = $this->getTitle();
609 if ( is_null( $nt ) ) {
612 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
613 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
614 if ( !$nt->exists() ) {
615 $permErrorsCreate = $nt->getUserPermissionsErrors( 'create', $user );
617 $permErrorsCreate = [];
619 if ( $permErrors ||
$permErrorsUpload ||
$permErrorsCreate ) {
620 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
621 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
626 $overwriteError = $this->checkOverwrite( $user );
627 if ( $overwriteError !== true ) {
628 return [ $overwriteError ];
635 * Check for non fatal problems with the file.
637 * This should not assume that mTempPath is set.
639 * @return mixed[] Array of warnings
641 public function checkWarnings() {
644 $localFile = $this->getLocalFile();
645 $localFile->load( File
::READ_LATEST
);
646 $filename = $localFile->getName();
647 $hash = $this->getTempFileSha1Base36();
649 $badFileName = $this->checkBadFileName( $filename, $this->mDesiredDestName
);
650 if ( $badFileName !== null ) {
651 $warnings['badfilename'] = $badFileName;
654 $unwantedFileExtensionDetails = $this->checkUnwantedFileExtensions( $this->mFinalExtension
);
655 if ( $unwantedFileExtensionDetails !== null ) {
656 $warnings['filetype-unwanted-type'] = $unwantedFileExtensionDetails;
659 $fileSizeWarnings = $this->checkFileSize( $this->mFileSize
);
660 if ( $fileSizeWarnings ) {
661 $warnings = array_merge( $warnings, $fileSizeWarnings );
664 $localFileExistsWarnings = $this->checkLocalFileExists( $localFile, $hash );
665 if ( $localFileExistsWarnings ) {
666 $warnings = array_merge( $warnings, $localFileExistsWarnings );
669 if ( $this->checkLocalFileWasDeleted( $localFile ) ) {
670 $warnings['was-deleted'] = $filename;
673 // If a file with the same name exists locally then the local file has already been tested
674 // for duplication of content
675 $ignoreLocalDupes = isset( $warnings[ 'exists '] );
676 $dupes = $this->checkAgainstExistingDupes( $hash, $ignoreLocalDupes );
678 $warnings['duplicate'] = $dupes;
681 $archivedDupes = $this->checkAgainstArchiveDupes( $hash );
682 if ( $archivedDupes !== null ) {
683 $warnings['duplicate-archive'] = $archivedDupes;
690 * Check whether the resulting filename is different from the desired one,
691 * but ignore things like ucfirst() and spaces/underscore things
693 * @param string $filename
694 * @param string $desiredFileName
696 * @return string|null String that was determined to be bad or null if the filename is okay
698 private function checkBadFileName( $filename, $desiredFileName ) {
699 $comparableName = str_replace( ' ', '_', $desiredFileName );
700 $comparableName = Title
::capitalize( $comparableName, NS_FILE
);
702 if ( $desiredFileName != $filename && $comparableName != $filename ) {
710 * @param string $fileExtension The file extension to check
712 * @return array|null array with the following keys:
713 * 0 => string The final extension being used
714 * 1 => string[] The extensions that are allowed
715 * 2 => int The number of extensions that are allowed.
717 private function checkUnwantedFileExtensions( $fileExtension ) {
718 global $wgCheckFileExtensions, $wgFileExtensions, $wgLang;
720 if ( $wgCheckFileExtensions ) {
721 $extensions = array_unique( $wgFileExtensions );
722 if ( !$this->checkFileExtension( $fileExtension, $extensions ) ) {
725 $wgLang->commaList( $extensions ),
735 * @param int $fileSize
737 * @return array warnings
739 private function checkFileSize( $fileSize ) {
740 global $wgUploadSizeWarning;
744 if ( $wgUploadSizeWarning && ( $fileSize > $wgUploadSizeWarning ) ) {
745 $warnings['large-file'] = [ $wgUploadSizeWarning, $fileSize ];
748 if ( $fileSize == 0 ) {
749 $warnings['empty-file'] = true;
756 * @param LocalFile $localFile
757 * @param string $hash sha1 hash of the file to check
759 * @return array warnings
761 private function checkLocalFileExists( LocalFile
$localFile, $hash ) {
764 $exists = self
::getExistsWarning( $localFile );
765 if ( $exists !== false ) {
766 $warnings['exists'] = $exists;
768 // check if file is an exact duplicate of current file version
769 if ( $hash === $localFile->getSha1() ) {
770 $warnings['no-change'] = $localFile;
773 // check if file is an exact duplicate of older versions of this file
774 $history = $localFile->getHistory();
775 foreach ( $history as $oldFile ) {
776 if ( $hash === $oldFile->getSha1() ) {
777 $warnings['duplicate-version'][] = $oldFile;
785 private function checkLocalFileWasDeleted( LocalFile
$localFile ) {
786 return $localFile->wasDeleted() && !$localFile->exists();
790 * @param string $hash sha1 hash of the file to check
791 * @param bool $ignoreLocalDupes True to ignore local duplicates
793 * @return File[] Duplicate files, if found.
795 private function checkAgainstExistingDupes( $hash, $ignoreLocalDupes ) {
796 $dupes = RepoGroup
::singleton()->findBySha1( $hash );
797 $title = $this->getTitle();
798 foreach ( $dupes as $key => $dupe ) {
800 ( $dupe instanceof LocalFile
) &&
802 $title->equals( $dupe->getTitle() )
804 unset( $dupes[$key] );
812 * @param string $hash sha1 hash of the file to check
814 * @return string|null Name of the dupe or empty string if discovered (depending on visibility)
815 * null if the check discovered no dupes.
817 private function checkAgainstArchiveDupes( $hash ) {
818 $archivedFile = new ArchivedFile( null, 0, '', $hash );
819 if ( $archivedFile->getID() > 0 ) {
820 if ( $archivedFile->userCan( File
::DELETED_FILE
) ) {
821 return $archivedFile->getName();
831 * Really perform the upload. Stores the file in the local repo, watches
832 * if necessary and runs the UploadComplete hook.
834 * @param string $comment
835 * @param string $pageText
836 * @param bool $watch Whether the file page should be added to user's watchlist.
837 * (This doesn't check $user's permissions.)
839 * @param string[] $tags Change tags to add to the log entry and page revision.
840 * (This doesn't check $user's permissions.)
841 * @return Status Indicating the whether the upload succeeded.
843 public function performUpload( $comment, $pageText, $watch, $user, $tags = [] ) {
844 $this->getLocalFile()->load( File
::READ_LATEST
);
845 $props = $this->mFileProps
;
848 Hooks
::run( 'UploadVerifyUpload', [ $this, $user, $props, $comment, $pageText, &$error ] );
850 if ( !is_array( $error ) ) {
853 return Status
::newFatal( ...$error );
856 $status = $this->getLocalFile()->upload(
867 if ( $status->isGood() ) {
869 WatchAction
::doWatch(
870 $this->getLocalFile()->getTitle(),
872 User
::IGNORE_USER_RIGHTS
875 // Avoid PHP 7.1 warning of passing $this by reference
877 Hooks
::run( 'UploadComplete', [ &$uploadBase ] );
879 $this->postProcessUpload();
886 * Perform extra steps after a successful upload.
890 public function postProcessUpload() {
894 * Returns the title of the file to be uploaded. Sets mTitleError in case
895 * the name was illegal.
897 * @return Title|null The title of the file or null in case the name was illegal
899 public function getTitle() {
900 if ( $this->mTitle
!== false ) {
901 return $this->mTitle
;
903 if ( !is_string( $this->mDesiredDestName
) ) {
904 $this->mTitleError
= self
::ILLEGAL_FILENAME
;
905 $this->mTitle
= null;
907 return $this->mTitle
;
909 /* Assume that if a user specified File:Something.jpg, this is an error
910 * and that the namespace prefix needs to be stripped of.
912 $title = Title
::newFromText( $this->mDesiredDestName
);
913 if ( $title && $title->getNamespace() == NS_FILE
) {
914 $this->mFilteredName
= $title->getDBkey();
916 $this->mFilteredName
= $this->mDesiredDestName
;
919 # oi_archive_name is max 255 bytes, which include a timestamp and an
920 # exclamation mark, so restrict file name to 240 bytes.
921 if ( strlen( $this->mFilteredName
) > 240 ) {
922 $this->mTitleError
= self
::FILENAME_TOO_LONG
;
923 $this->mTitle
= null;
925 return $this->mTitle
;
929 * Chop off any directories in the given filename. Then
930 * filter out illegal characters, and try to make a legible name
931 * out of it. We'll strip some silently that Title would die on.
933 $this->mFilteredName
= wfStripIllegalFilenameChars( $this->mFilteredName
);
934 /* Normalize to title form before we do any further processing */
935 $nt = Title
::makeTitleSafe( NS_FILE
, $this->mFilteredName
);
936 if ( is_null( $nt ) ) {
937 $this->mTitleError
= self
::ILLEGAL_FILENAME
;
938 $this->mTitle
= null;
940 return $this->mTitle
;
942 $this->mFilteredName
= $nt->getDBkey();
945 * We'll want to blacklist against *any* 'extension', and use
946 * only the final one for the whitelist.
948 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName
);
950 if ( count( $ext ) ) {
951 $this->mFinalExtension
= trim( $ext[count( $ext ) - 1] );
953 $this->mFinalExtension
= '';
955 # No extension, try guessing one
956 $magic = MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer();
957 $mime = $magic->guessMimeType( $this->mTempPath
);
958 if ( $mime !== 'unknown/unknown' ) {
959 # Get a space separated list of extensions
960 $extList = $magic->getExtensionsForType( $mime );
962 # Set the extension to the canonical extension
963 $this->mFinalExtension
= strtok( $extList, ' ' );
965 # Fix up the other variables
966 $this->mFilteredName
.= ".{$this->mFinalExtension}";
967 $nt = Title
::makeTitleSafe( NS_FILE
, $this->mFilteredName
);
968 $ext = [ $this->mFinalExtension
];
973 /* Don't allow users to override the blacklist (check file extension) */
974 global $wgCheckFileExtensions, $wgStrictFileExtensions;
975 global $wgFileExtensions, $wgFileBlacklist;
977 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
979 if ( $this->mFinalExtension
== '' ) {
980 $this->mTitleError
= self
::FILETYPE_MISSING
;
981 $this->mTitle
= null;
983 return $this->mTitle
;
984 } elseif ( $blackListedExtensions ||
985 ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
986 !$this->checkFileExtension( $this->mFinalExtension
, $wgFileExtensions ) )
988 $this->mBlackListedExtensions
= $blackListedExtensions;
989 $this->mTitleError
= self
::FILETYPE_BADTYPE
;
990 $this->mTitle
= null;
992 return $this->mTitle
;
995 // Windows may be broken with special characters, see T3780
996 if ( !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() )
997 && !RepoGroup
::singleton()->getLocalRepo()->backendSupportsUnicodePaths()
999 $this->mTitleError
= self
::WINDOWS_NONASCII_FILENAME
;
1000 $this->mTitle
= null;
1002 return $this->mTitle
;
1005 # If there was more than one "extension", reassemble the base
1006 # filename to prevent bogus complaints about length
1007 if ( count( $ext ) > 1 ) {
1008 $iterations = count( $ext ) - 1;
1009 for ( $i = 0; $i < $iterations; $i++
) {
1010 $partname .= '.' . $ext[$i];
1014 if ( strlen( $partname ) < 1 ) {
1015 $this->mTitleError
= self
::MIN_LENGTH_PARTNAME
;
1016 $this->mTitle
= null;
1018 return $this->mTitle
;
1021 $this->mTitle
= $nt;
1023 return $this->mTitle
;
1027 * Return the local file and initializes if necessary.
1029 * @return LocalFile|null
1031 public function getLocalFile() {
1032 if ( is_null( $this->mLocalFile
) ) {
1033 $nt = $this->getTitle();
1034 $this->mLocalFile
= is_null( $nt ) ?
null : wfLocalFile( $nt );
1037 return $this->mLocalFile
;
1041 * @return UploadStashFile|null
1043 public function getStashFile() {
1044 return $this->mStashFile
;
1048 * Like stashFile(), but respects extensions' wishes to prevent the stashing. verifyUpload() must
1049 * be called before calling this method (unless $isPartial is true).
1051 * Upload stash exceptions are also caught and converted to an error status.
1055 * @param bool $isPartial Pass `true` if this is a part of a chunked upload (not a complete file).
1056 * @return Status If successful, value is an UploadStashFile instance
1058 public function tryStashFile( User
$user, $isPartial = false ) {
1059 if ( !$isPartial ) {
1060 $error = $this->runUploadStashFileHook( $user );
1062 return Status
::newFatal( ...$error );
1066 $file = $this->doStashFile( $user );
1067 return Status
::newGood( $file );
1068 } catch ( UploadStashException
$e ) {
1069 return Status
::newFatal( 'uploadstash-exception', get_class( $e ), $e->getMessage() );
1075 * @return array|null Error message and parameters, null if there's no error
1077 protected function runUploadStashFileHook( User
$user ) {
1078 $props = $this->mFileProps
;
1080 Hooks
::run( 'UploadStashFile', [ $this, $user, $props, &$error ] );
1082 if ( !is_array( $error ) ) {
1083 $error = [ $error ];
1090 * If the user does not supply all necessary information in the first upload
1091 * form submission (either by accident or by design) then we may want to
1092 * stash the file temporarily, get more information, and publish the file
1095 * This method will stash a file in a temporary directory for later
1096 * processing, and save the necessary descriptive info into the database.
1097 * This method returns the file object, which also has a 'fileKey' property
1098 * which can be passed through a form or API request to find this stashed
1101 * @deprecated since 1.28 Use tryStashFile() instead
1102 * @param User|null $user
1103 * @return UploadStashFile Stashed file
1104 * @throws UploadStashBadPathException
1105 * @throws UploadStashFileException
1106 * @throws UploadStashNotLoggedInException
1108 public function stashFile( User
$user = null ) {
1109 return $this->doStashFile( $user );
1113 * Implementation for stashFile() and tryStashFile().
1115 * @param User|null $user
1116 * @return UploadStashFile Stashed file
1118 protected function doStashFile( User
$user = null ) {
1119 $stash = RepoGroup
::singleton()->getLocalRepo()->getUploadStash( $user );
1120 $file = $stash->stashFile( $this->mTempPath
, $this->getSourceType() );
1121 $this->mStashFile
= $file;
1127 * Stash a file in a temporary directory, returning a key which can be used
1128 * to find the file again. See stashFile().
1130 * @deprecated since 1.28
1131 * @return string File key
1133 public function stashFileGetKey() {
1134 wfDeprecated( __METHOD__
, '1.28' );
1135 return $this->doStashFile()->getFileKey();
1139 * alias for stashFileGetKey, for backwards compatibility
1141 * @deprecated since 1.28
1142 * @return string File key
1144 public function stashSession() {
1145 wfDeprecated( __METHOD__
, '1.28' );
1146 return $this->doStashFile()->getFileKey();
1150 * If we've modified the upload file we need to manually remove it
1151 * on exit to clean up.
1153 public function cleanupTempFile() {
1154 if ( $this->mRemoveTempFile
&& $this->tempFileObj
) {
1155 // Delete when all relevant TempFSFile handles go out of scope
1156 wfDebug( __METHOD__
. ": Marked temporary file '{$this->mTempPath}' for removal\n" );
1157 $this->tempFileObj
->autocollect();
1161 public function getTempPath() {
1162 return $this->mTempPath
;
1166 * Split a file into a base name and all dot-delimited 'extensions'
1167 * on the end. Some web server configurations will fall back to
1168 * earlier pseudo-'extensions' to determine type and execute
1169 * scripts, so the blacklist needs to check them all.
1171 * @param string $filename
1174 public static function splitExtensions( $filename ) {
1175 $bits = explode( '.', $filename );
1176 $basename = array_shift( $bits );
1178 return [ $basename, $bits ];
1182 * Perform case-insensitive match against a list of file extensions.
1183 * Returns true if the extension is in the list.
1185 * @param string $ext
1186 * @param array $list
1189 public static function checkFileExtension( $ext, $list ) {
1190 return in_array( strtolower( $ext ), $list );
1194 * Perform case-insensitive match against a list of file extensions.
1195 * Returns an array of matching extensions.
1198 * @param array $list
1201 public static function checkFileExtensionList( $ext, $list ) {
1202 return array_intersect( array_map( 'strtolower', $ext ), $list );
1206 * Checks if the MIME type of the uploaded file matches the file extension.
1208 * @param string $mime The MIME type of the uploaded file
1209 * @param string $extension The filename extension that the file is to be served with
1212 public static function verifyExtension( $mime, $extension ) {
1213 $magic = MediaWiki\MediaWikiServices
::getInstance()->getMimeAnalyzer();
1215 if ( !$mime ||
$mime == 'unknown' ||
$mime == 'unknown/unknown' ) {
1216 if ( !$magic->isRecognizableExtension( $extension ) ) {
1217 wfDebug( __METHOD__
. ": passing file with unknown detected mime type; " .
1218 "unrecognized extension '$extension', can't verify\n" );
1222 wfDebug( __METHOD__
. ": rejecting file with unknown detected mime type; " .
1223 "recognized extension '$extension', so probably invalid file\n" );
1229 $match = $magic->isMatchingExtension( $extension, $mime );
1231 if ( $match === null ) {
1232 if ( $magic->getTypesForExtension( $extension ) !== null ) {
1233 wfDebug( __METHOD__
. ": No extension known for $mime, but we know a mime for $extension\n" );
1237 wfDebug( __METHOD__
. ": no file extension known for mime type $mime, passing file\n" );
1241 } elseif ( $match === true ) {
1242 wfDebug( __METHOD__
. ": mime type $mime matches extension $extension, passing file\n" );
1244 /** @todo If it's a bitmap, make sure PHP or ImageMagick resp. can handle it! */
1248 . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
1255 * Heuristic for detecting files that *could* contain JavaScript instructions or
1256 * things that may look like HTML to a browser and are thus
1257 * potentially harmful. The present implementation will produce false
1258 * positives in some situations.
1260 * @param string $file Pathname to the temporary upload file
1261 * @param string $mime The MIME type of the file
1262 * @param string $extension The extension of the file
1263 * @return bool True if the file contains something looking like embedded scripts
1265 public static function detectScript( $file, $mime, $extension ) {
1266 global $wgAllowTitlesInSVG;
1268 # ugly hack: for text files, always look at the entire file.
1269 # For binary field, just check the first K.
1271 if ( strpos( $mime, 'text/' ) === 0 ) {
1272 $chunk = file_get_contents( $file );
1274 $fp = fopen( $file, 'rb' );
1275 $chunk = fread( $fp, 1024 );
1279 $chunk = strtolower( $chunk );
1285 # decode from UTF-16 if needed (could be used for obfuscation).
1286 if ( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
1288 } elseif ( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
1295 $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
1298 $chunk = trim( $chunk );
1300 /** @todo FIXME: Convert from UTF-16 if necessary! */
1301 wfDebug( __METHOD__
. ": checking for embedded scripts and HTML stuff\n" );
1303 # check for HTML doctype
1304 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
1308 // Some browsers will interpret obscure xml encodings as UTF-8, while
1309 // PHP/expat will interpret the given encoding in the xml declaration (T49304)
1310 if ( $extension == 'svg' ||
strpos( $mime, 'image/svg' ) === 0 ) {
1311 if ( self
::checkXMLEncodingMissmatch( $file ) ) {
1317 * Internet Explorer for Windows performs some really stupid file type
1318 * autodetection which can cause it to interpret valid image files as HTML
1319 * and potentially execute JavaScript, creating a cross-site scripting
1322 * Apple's Safari browser also performs some unsafe file type autodetection
1323 * which can cause legitimate files to be interpreted as HTML if the
1324 * web server is not correctly configured to send the right content-type
1325 * (or if you're really uploading plain text and octet streams!)
1327 * Returns true if IE is likely to mistake the given file for HTML.
1328 * Also returns true if Safari would mistake the given file for HTML
1329 * when served with a generic content-type.
1335 '<html', # also in safari
1338 '<script', # also in safari
1342 if ( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
1346 foreach ( $tags as $tag ) {
1347 if ( strpos( $chunk, $tag ) !== false ) {
1348 wfDebug( __METHOD__
. ": found something that may make it be mistaken for html: $tag\n" );
1355 * look for JavaScript
1358 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
1359 $chunk = Sanitizer
::decodeCharReferences( $chunk );
1361 # look for script-types
1362 if ( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
1363 wfDebug( __METHOD__
. ": found script types\n" );
1368 # look for html-style script-urls
1369 if ( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1370 wfDebug( __METHOD__
. ": found html-style script urls\n" );
1375 # look for css-style script-urls
1376 if ( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1377 wfDebug( __METHOD__
. ": found css-style script urls\n" );
1382 wfDebug( __METHOD__
. ": no scripts found\n" );
1388 * Check a whitelist of xml encodings that are known not to be interpreted differently
1389 * by the server's xml parser (expat) and some common browsers.
1391 * @param string $file Pathname to the temporary upload file
1392 * @return bool True if the file contains an encoding that could be misinterpreted
1394 public static function checkXMLEncodingMissmatch( $file ) {
1395 global $wgSVGMetadataCutoff;
1396 $contents = file_get_contents( $file, false, null, 0, $wgSVGMetadataCutoff );
1397 $encodingRegex = '!encoding[ \t\n\r]*=[ \t\n\r]*[\'"](.*?)[\'"]!si';
1399 if ( preg_match( "!<\?xml\b(.*?)\?>!si", $contents, $matches ) ) {
1400 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1401 && !in_array( strtoupper( $encMatch[1] ), self
::$safeXmlEncodings )
1403 wfDebug( __METHOD__
. ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1407 } elseif ( preg_match( "!<\?xml\b!si", $contents ) ) {
1408 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1409 // bytes. There shouldn't be a legitimate reason for this to happen.
1410 wfDebug( __METHOD__
. ": Unmatched XML declaration start\n" );
1413 } elseif ( substr( $contents, 0, 4 ) == "\x4C\x6F\xA7\x94" ) {
1414 // EBCDIC encoded XML
1415 wfDebug( __METHOD__
. ": EBCDIC Encoded XML\n" );
1420 // It's possible the file is encoded with multi-byte encoding, so re-encode attempt to
1421 // detect the encoding in case is specifies an encoding not whitelisted in self::$safeXmlEncodings
1422 $attemptEncodings = [ 'UTF-16', 'UTF-16BE', 'UTF-32', 'UTF-32BE' ];
1423 foreach ( $attemptEncodings as $encoding ) {
1424 Wikimedia\
suppressWarnings();
1425 $str = iconv( $encoding, 'UTF-8', $contents );
1426 Wikimedia\restoreWarnings
();
1427 if ( $str != '' && preg_match( "!<\?xml\b(.*?)\?>!si", $str, $matches ) ) {
1428 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1429 && !in_array( strtoupper( $encMatch[1] ), self
::$safeXmlEncodings )
1431 wfDebug( __METHOD__
. ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1435 } elseif ( $str != '' && preg_match( "!<\?xml\b!si", $str ) ) {
1436 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1437 // bytes. There shouldn't be a legitimate reason for this to happen.
1438 wfDebug( __METHOD__
. ": Unmatched XML declaration start\n" );
1448 * @param string $filename
1449 * @param bool $partial
1450 * @return mixed False of the file is verified (does not contain scripts), array otherwise.
1452 protected function detectScriptInSvg( $filename, $partial ) {
1453 $this->mSVGNSError
= false;
1454 $check = new XmlTypeCheck(
1456 [ $this, 'checkSvgScriptCallback' ],
1459 'processing_instruction_handler' => 'UploadBase::checkSvgPICallback',
1460 'external_dtd_handler' => 'UploadBase::checkSvgExternalDTD',
1463 if ( $check->wellFormed
!== true ) {
1464 // Invalid xml (T60553)
1465 // But only when non-partial (T67724)
1466 return $partial ?
false : [ 'uploadinvalidxml' ];
1467 } elseif ( $check->filterMatch
) {
1468 if ( $this->mSVGNSError
) {
1469 return [ 'uploadscriptednamespace', $this->mSVGNSError
];
1472 return $check->filterMatchType
;
1479 * Callback to filter SVG Processing Instructions.
1480 * @param string $target Processing instruction name
1481 * @param string $data Processing instruction attribute and value
1482 * @return bool (true if the filter identified something bad)
1484 public static function checkSvgPICallback( $target, $data ) {
1485 // Don't allow external stylesheets (T59550)
1486 if ( preg_match( '/xml-stylesheet/i', $target ) ) {
1487 return [ 'upload-scripted-pi-callback' ];
1494 * Verify that DTD urls referenced are only the standard dtds
1496 * Browsers seem to ignore external dtds. However just to be on the
1497 * safe side, only allow dtds from the svg standard.
1499 * @param string $type PUBLIC or SYSTEM
1500 * @param string $publicId The well-known public identifier for the dtd
1501 * @param string $systemId The url for the external dtd
1502 * @return bool|array
1504 public static function checkSvgExternalDTD( $type, $publicId, $systemId ) {
1505 // This doesn't include the XHTML+MathML+SVG doctype since we don't
1506 // allow XHTML anyways.
1508 'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd',
1509 'http://www.w3.org/TR/2001/REC-SVG-20010904/DTD/svg10.dtd',
1510 'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-basic.dtd',
1511 'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-tiny.dtd',
1512 // https://phabricator.wikimedia.org/T168856
1513 'http://www.w3.org/TR/2001/PR-SVG-20010719/DTD/svg10.dtd',
1515 if ( $type !== 'PUBLIC'
1516 ||
!in_array( $systemId, $allowedDTDs )
1517 ||
strpos( $publicId, "-//W3C//" ) !== 0
1519 return [ 'upload-scripted-dtd' ];
1525 * @todo Replace this with a whitelist filter!
1526 * @param string $element
1527 * @param array $attribs
1528 * @param array|null $data
1531 public function checkSvgScriptCallback( $element, $attribs, $data = null ) {
1532 list( $namespace, $strippedElement ) = $this->splitXmlNamespace( $element );
1534 // We specifically don't include:
1535 // http://www.w3.org/1999/xhtml (T62771)
1536 static $validNamespaces = [
1539 'http://creativecommons.org/ns#',
1540 'http://inkscape.sourceforge.net/dtd/sodipodi-0.dtd',
1541 'http://ns.adobe.com/adobeillustrator/10.0/',
1542 'http://ns.adobe.com/adobesvgviewerextensions/3.0/',
1543 'http://ns.adobe.com/extensibility/1.0/',
1544 'http://ns.adobe.com/flows/1.0/',
1545 'http://ns.adobe.com/illustrator/1.0/',
1546 'http://ns.adobe.com/imagereplacement/1.0/',
1547 'http://ns.adobe.com/pdf/1.3/',
1548 'http://ns.adobe.com/photoshop/1.0/',
1549 'http://ns.adobe.com/saveforweb/1.0/',
1550 'http://ns.adobe.com/variables/1.0/',
1551 'http://ns.adobe.com/xap/1.0/',
1552 'http://ns.adobe.com/xap/1.0/g/',
1553 'http://ns.adobe.com/xap/1.0/g/img/',
1554 'http://ns.adobe.com/xap/1.0/mm/',
1555 'http://ns.adobe.com/xap/1.0/rights/',
1556 'http://ns.adobe.com/xap/1.0/stype/dimensions#',
1557 'http://ns.adobe.com/xap/1.0/stype/font#',
1558 'http://ns.adobe.com/xap/1.0/stype/manifestitem#',
1559 'http://ns.adobe.com/xap/1.0/stype/resourceevent#',
1560 'http://ns.adobe.com/xap/1.0/stype/resourceref#',
1561 'http://ns.adobe.com/xap/1.0/t/pg/',
1562 'http://purl.org/dc/elements/1.1/',
1563 'http://purl.org/dc/elements/1.1',
1564 'http://schemas.microsoft.com/visio/2003/svgextensions/',
1565 'http://sodipodi.sourceforge.net/dtd/sodipodi-0.dtd',
1566 'http://taptrix.com/inkpad/svg_extensions',
1567 'http://web.resource.org/cc/',
1568 'http://www.freesoftware.fsf.org/bkchem/cdml',
1569 'http://www.inkscape.org/namespaces/inkscape',
1570 'http://www.opengis.net/gml',
1571 'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
1572 'http://www.w3.org/2000/svg',
1573 'http://www.w3.org/tr/rec-rdf-syntax/',
1574 'http://www.w3.org/2000/01/rdf-schema#',
1577 // Inkscape mangles namespace definitions created by Adobe Illustrator.
1578 // This is nasty but harmless. (T144827)
1579 $isBuggyInkscape = preg_match( '/^&(#38;)*ns_[a-z_]+;$/', $namespace );
1581 if ( !( $isBuggyInkscape ||
in_array( $namespace, $validNamespaces ) ) ) {
1582 wfDebug( __METHOD__
. ": Non-svg namespace '$namespace' in uploaded file.\n" );
1583 /** @todo Return a status object to a closure in XmlTypeCheck, for MW1.21+ */
1584 $this->mSVGNSError
= $namespace;
1590 * check for elements that can contain javascript
1592 if ( $strippedElement == 'script' ) {
1593 wfDebug( __METHOD__
. ": Found script element '$element' in uploaded file.\n" );
1595 return [ 'uploaded-script-svg', $strippedElement ];
1598 # e.g., <svg xmlns="http://www.w3.org/2000/svg">
1599 # <handler xmlns:ev="http://www.w3.org/2001/xml-events" ev:event="load">alert(1)</handler> </svg>
1600 if ( $strippedElement == 'handler' ) {
1601 wfDebug( __METHOD__
. ": Found scriptable element '$element' in uploaded file.\n" );
1603 return [ 'uploaded-script-svg', $strippedElement ];
1606 # SVG reported in Feb '12 that used xml:stylesheet to generate javascript block
1607 if ( $strippedElement == 'stylesheet' ) {
1608 wfDebug( __METHOD__
. ": Found scriptable element '$element' in uploaded file.\n" );
1610 return [ 'uploaded-script-svg', $strippedElement ];
1613 # Block iframes, in case they pass the namespace check
1614 if ( $strippedElement == 'iframe' ) {
1615 wfDebug( __METHOD__
. ": iframe in uploaded file.\n" );
1617 return [ 'uploaded-script-svg', $strippedElement ];
1621 if ( $strippedElement == 'style'
1622 && self
::checkCssFragment( Sanitizer
::normalizeCss( $data ) )
1624 wfDebug( __METHOD__
. ": hostile css in style element.\n" );
1625 return [ 'uploaded-hostile-svg' ];
1628 foreach ( $attribs as $attrib => $value ) {
1629 $stripped = $this->stripXmlNamespace( $attrib );
1630 $value = strtolower( $value );
1632 if ( substr( $stripped, 0, 2 ) == 'on' ) {
1634 . ": Found event-handler attribute '$attrib'='$value' in uploaded file.\n" );
1636 return [ 'uploaded-event-handler-on-svg', $attrib, $value ];
1639 # Do not allow relative links, or unsafe url schemas.
1640 # For <a> tags, only data:, http: and https: and same-document
1641 # fragment links are allowed. For all other tags, only data:
1642 # and fragment are allowed.
1643 if ( $stripped == 'href'
1645 && strpos( $value, 'data:' ) !== 0
1646 && strpos( $value, '#' ) !== 0
1648 if ( !( $strippedElement === 'a'
1649 && preg_match( '!^https?://!i', $value ) )
1651 wfDebug( __METHOD__
. ": Found href attribute <$strippedElement "
1652 . "'$attrib'='$value' in uploaded file.\n" );
1654 return [ 'uploaded-href-attribute-svg', $strippedElement, $attrib, $value ];
1658 # only allow data: targets that should be safe. This prevents vectors like,
1659 # image/svg, text/xml, application/xml, and text/html, which can contain scripts
1660 if ( $stripped == 'href' && strncasecmp( 'data:', $value, 5 ) === 0 ) {
1661 // rfc2397 parameters. This is only slightly slower than (;[\w;]+)*.
1662 // phpcs:ignore Generic.Files.LineLength
1663 $parameters = '(?>;[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+=(?>[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+|"(?>[\0-\x0c\x0e-\x21\x23-\x5b\x5d-\x7f]+|\\\\[\0-\x7f])*"))*(?:;base64)?';
1665 if ( !preg_match( "!^data:\s*image/(gif|jpeg|jpg|png)$parameters,!i", $value ) ) {
1666 wfDebug( __METHOD__
. ": Found href to unwhitelisted data: uri "
1667 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1668 return [ 'uploaded-href-unsafe-target-svg', $strippedElement, $attrib, $value ];
1672 # Change href with animate from (http://html5sec.org/#137).
1673 if ( $stripped === 'attributename'
1674 && $strippedElement === 'animate'
1675 && $this->stripXmlNamespace( $value ) == 'href'
1677 wfDebug( __METHOD__
. ": Found animate that might be changing href using from "
1678 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1680 return [ 'uploaded-animate-svg', $strippedElement, $attrib, $value ];
1683 # use set/animate to add event-handler attribute to parent
1684 if ( ( $strippedElement == 'set' ||
$strippedElement == 'animate' )
1685 && $stripped == 'attributename'
1686 && substr( $value, 0, 2 ) == 'on'
1688 wfDebug( __METHOD__
. ": Found svg setting event-handler attribute with "
1689 . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1691 return [ 'uploaded-setting-event-handler-svg', $strippedElement, $stripped, $value ];
1694 # use set to add href attribute to parent element
1695 if ( $strippedElement == 'set'
1696 && $stripped == 'attributename'
1697 && strpos( $value, 'href' ) !== false
1699 wfDebug( __METHOD__
. ": Found svg setting href attribute '$value' in uploaded file.\n" );
1701 return [ 'uploaded-setting-href-svg' ];
1704 # use set to add a remote / data / script target to an element
1705 if ( $strippedElement == 'set'
1706 && $stripped == 'to'
1707 && preg_match( '!(http|https|data|script):!sim', $value )
1709 wfDebug( __METHOD__
. ": Found svg setting attribute to '$value' in uploaded file.\n" );
1711 return [ 'uploaded-wrong-setting-svg', $value ];
1714 # use handler attribute with remote / data / script
1715 if ( $stripped == 'handler' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1716 wfDebug( __METHOD__
. ": Found svg setting handler with remote/data/script "
1717 . "'$attrib'='$value' in uploaded file.\n" );
1719 return [ 'uploaded-setting-handler-svg', $attrib, $value ];
1722 # use CSS styles to bring in remote code
1723 if ( $stripped == 'style'
1724 && self
::checkCssFragment( Sanitizer
::normalizeCss( $value ) )
1726 wfDebug( __METHOD__
. ": Found svg setting a style with "
1727 . "remote url '$attrib'='$value' in uploaded file.\n" );
1728 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1731 # Several attributes can include css, css character escaping isn't allowed
1732 $cssAttrs = [ 'font', 'clip-path', 'fill', 'filter', 'marker',
1733 'marker-end', 'marker-mid', 'marker-start', 'mask', 'stroke' ];
1734 if ( in_array( $stripped, $cssAttrs )
1735 && self
::checkCssFragment( $value )
1737 wfDebug( __METHOD__
. ": Found svg setting a style with "
1738 . "remote url '$attrib'='$value' in uploaded file.\n" );
1739 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1742 # image filters can pull in url, which could be svg that executes scripts
1743 # Only allow url( "#foo" ). Do not allow url( http://example.com )
1744 if ( $strippedElement == 'image'
1745 && $stripped == 'filter'
1746 && preg_match( '!url\s*\(\s*["\']?[^#]!sim', $value )
1748 wfDebug( __METHOD__
. ": Found image filter with url: "
1749 . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1751 return [ 'uploaded-image-filter-svg', $strippedElement, $stripped, $value ];
1755 return false; // No scripts detected
1759 * Check a block of CSS or CSS fragment for anything that looks like
1760 * it is bringing in remote code.
1761 * @param string $value a string of CSS
1762 * @param bool $propOnly only check css properties (start regex with :)
1763 * @return bool true if the CSS contains an illegal string, false if otherwise
1765 private static function checkCssFragment( $value ) {
1766 # Forbid external stylesheets, for both reliability and to protect viewer's privacy
1767 if ( stripos( $value, '@import' ) !== false ) {
1771 # We allow @font-face to embed fonts with data: urls, so we snip the string
1772 # 'url' out so this case won't match when we check for urls below
1773 $pattern = '!(@font-face\s*{[^}]*src:)url(\("data:;base64,)!im';
1774 $value = preg_replace( $pattern, '$1$2', $value );
1776 # Check for remote and executable CSS. Unlike in Sanitizer::checkCss, the CSS
1777 # properties filter and accelerator don't seem to be useful for xss in SVG files.
1778 # Expression and -o-link don't seem to work either, but filtering them here in case.
1779 # Additionally, we catch remote urls like url("http:..., url('http:..., url(http:...,
1780 # but not local ones such as url("#..., url('#..., url(#....
1781 if ( preg_match( '!expression
1783 | -o-link-source\s*:
1784 | -o-replace\s*:!imx', $value ) ) {
1788 if ( preg_match_all(
1789 "!(\s*(url|image|image-set)\s*\(\s*[\"']?\s*[^#]+.*?\))!sim",
1794 # TODO: redo this in one regex. Until then, url("#whatever") matches the first
1795 foreach ( $matches[1] as $match ) {
1796 if ( !preg_match( "!\s*(url|image|image-set)\s*\(\s*(#|'#|\"#)!im", $match ) ) {
1802 if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ) {
1810 * Divide the element name passed by the xml parser to the callback into URI and prifix.
1811 * @param string $element
1812 * @return array Containing the namespace URI and prefix
1814 private static function splitXmlNamespace( $element ) {
1815 // 'http://www.w3.org/2000/svg:script' -> [ 'http://www.w3.org/2000/svg', 'script' ]
1816 $parts = explode( ':', strtolower( $element ) );
1817 $name = array_pop( $parts );
1818 $ns = implode( ':', $parts );
1820 return [ $ns, $name ];
1824 * @param string $name
1827 private function stripXmlNamespace( $name ) {
1828 // 'http://www.w3.org/2000/svg:script' -> 'script'
1829 $parts = explode( ':', strtolower( $name ) );
1831 return array_pop( $parts );
1835 * Generic wrapper function for a virus scanner program.
1836 * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
1837 * $wgAntivirusRequired may be used to deny upload if the scan fails.
1839 * @param string $file Pathname to the temporary upload file
1840 * @return mixed False if not virus is found, null if the scan fails or is disabled,
1841 * or a string containing feedback from the virus scanner if a virus was found.
1842 * If textual feedback is missing but a virus was found, this function returns true.
1844 public static function detectVirus( $file ) {
1845 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
1847 if ( !$wgAntivirus ) {
1848 wfDebug( __METHOD__
. ": virus scanner disabled\n" );
1853 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
1854 wfDebug( __METHOD__
. ": unknown virus scanner: $wgAntivirus\n" );
1855 $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
1856 [ 'virus-badscanner', $wgAntivirus ] );
1858 return wfMessage( 'virus-unknownscanner' )->text() . " $wgAntivirus";
1861 # look up scanner configuration
1862 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
1863 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
1864 $msgPattern = $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ??
null;
1866 if ( strpos( $command, "%f" ) === false ) {
1867 # simple pattern: append file to scan
1868 $command .= " " . wfEscapeShellArg( $file );
1870 # complex pattern: replace "%f" with file to scan
1871 $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
1874 wfDebug( __METHOD__
. ": running virus scan: $command \n" );
1876 # execute virus scanner
1879 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1880 # that does not seem to be worth the pain.
1881 # Ask me (Duesentrieb) about it if it's ever needed.
1882 $output = wfShellExecWithStderr( $command, $exitCode );
1884 # map exit code to AV_xxx constants.
1885 $mappedCode = $exitCode;
1886 if ( $exitCodeMap ) {
1887 if ( isset( $exitCodeMap[$exitCode] ) ) {
1888 $mappedCode = $exitCodeMap[$exitCode];
1889 } elseif ( isset( $exitCodeMap["*"] ) ) {
1890 $mappedCode = $exitCodeMap["*"];
1894 /* NB: AV_NO_VIRUS is 0 but AV_SCAN_FAILED is false,
1895 * so we need the strict equalities === and thus can't use a switch here
1897 if ( $mappedCode === AV_SCAN_FAILED
) {
1898 # scan failed (code was mapped to false by $exitCodeMap)
1899 wfDebug( __METHOD__
. ": failed to scan $file (code $exitCode).\n" );
1901 $output = $wgAntivirusRequired
1902 ?
wfMessage( 'virus-scanfailed', [ $exitCode ] )->text()
1904 } elseif ( $mappedCode === AV_SCAN_ABORTED
) {
1905 # scan failed because filetype is unknown (probably imune)
1906 wfDebug( __METHOD__
. ": unsupported file type $file (code $exitCode).\n" );
1908 } elseif ( $mappedCode === AV_NO_VIRUS
) {
1910 wfDebug( __METHOD__
. ": file passed virus scan.\n" );
1913 $output = trim( $output );
1916 $output = true; # if there's no output, return true
1917 } elseif ( $msgPattern ) {
1919 if ( preg_match( $msgPattern, $output, $groups ) ) {
1921 $output = $groups[1];
1926 wfDebug( __METHOD__
. ": FOUND VIRUS! scanner feedback: $output \n" );
1933 * Check if there's an overwrite conflict and, if so, if restrictions
1934 * forbid this user from performing the upload.
1938 * @return mixed True on success, array on failure
1940 private function checkOverwrite( $user ) {
1941 // First check whether the local file can be overwritten
1942 $file = $this->getLocalFile();
1943 $file->load( File
::READ_LATEST
);
1944 if ( $file->exists() ) {
1945 if ( !self
::userCanReUpload( $user, $file ) ) {
1946 return [ 'fileexists-forbidden', $file->getName() ];
1952 /* Check shared conflicts: if the local file does not exist, but
1953 * wfFindFile finds a file, it exists in a shared repository.
1955 $file = wfFindFile( $this->getTitle(), [ 'latest' => true ] );
1956 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1957 return [ 'fileexists-shared-forbidden', $file->getName() ];
1964 * Check if a user is the last uploader
1970 public static function userCanReUpload( User
$user, File
$img ) {
1971 if ( $user->isAllowed( 'reupload' ) ) {
1972 return true; // non-conditional
1973 } elseif ( !$user->isAllowed( 'reupload-own' ) ) {
1977 if ( !( $img instanceof LocalFile
) ) {
1983 return $user->getId() == $img->getUser( 'id' );
1987 * Helper function that does various existence checks for a file.
1988 * The following checks are performed:
1990 * - Article with the same name as the file exists
1991 * - File exists with normalized extension
1992 * - The file looks like a thumbnail and the original exists
1994 * @param File $file The File object to check
1995 * @return mixed False if the file does not exists, else an array
1997 public static function getExistsWarning( $file ) {
1998 if ( $file->exists() ) {
1999 return [ 'warning' => 'exists', 'file' => $file ];
2002 if ( $file->getTitle()->getArticleID() ) {
2003 return [ 'warning' => 'page-exists', 'file' => $file ];
2006 if ( strpos( $file->getName(), '.' ) == false ) {
2007 $partname = $file->getName();
2010 $n = strrpos( $file->getName(), '.' );
2011 $extension = substr( $file->getName(), $n +
1 );
2012 $partname = substr( $file->getName(), 0, $n );
2014 $normalizedExtension = File
::normalizeExtension( $extension );
2016 if ( $normalizedExtension != $extension ) {
2017 // We're not using the normalized form of the extension.
2018 // Normal form is lowercase, using most common of alternate
2019 // extensions (eg 'jpg' rather than 'JPEG').
2021 // Check for another file using the normalized form...
2022 $nt_lc = Title
::makeTitle( NS_FILE
, "{$partname}.{$normalizedExtension}" );
2023 $file_lc = wfLocalFile( $nt_lc );
2025 if ( $file_lc->exists() ) {
2027 'warning' => 'exists-normalized',
2029 'normalizedFile' => $file_lc
2034 // Check for files with the same name but a different extension
2035 $similarFiles = RepoGroup
::singleton()->getLocalRepo()->findFilesByPrefix(
2036 "{$partname}.", 1 );
2037 if ( count( $similarFiles ) ) {
2039 'warning' => 'exists-normalized',
2041 'normalizedFile' => $similarFiles[0],
2045 if ( self
::isThumbName( $file->getName() ) ) {
2046 # Check for filenames like 50px- or 180px-, these are mostly thumbnails
2047 $nt_thb = Title
::newFromText(
2048 substr( $partname, strpos( $partname, '-' ) +
1 ) . '.' . $extension,
2051 $file_thb = wfLocalFile( $nt_thb );
2052 if ( $file_thb->exists() ) {
2054 'warning' => 'thumb',
2056 'thumbFile' => $file_thb
2059 // File does not exist, but we just don't like the name
2061 'warning' => 'thumb-name',
2063 'thumbFile' => $file_thb
2068 foreach ( self
::getFilenamePrefixBlacklist() as $prefix ) {
2069 if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
2071 'warning' => 'bad-prefix',
2082 * Helper function that checks whether the filename looks like a thumbnail
2083 * @param string $filename
2086 public static function isThumbName( $filename ) {
2087 $n = strrpos( $filename, '.' );
2088 $partname = $n ?
substr( $filename, 0, $n ) : $filename;
2091 substr( $partname, 3, 3 ) == 'px-' ||
2092 substr( $partname, 2, 3 ) == 'px-'
2094 preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) );
2098 * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
2100 * @return array List of prefixes
2102 public static function getFilenamePrefixBlacklist() {
2104 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
2105 if ( !$message->isDisabled() ) {
2106 $lines = explode( "\n", $message->plain() );
2107 foreach ( $lines as $line ) {
2108 // Remove comment lines
2109 $comment = substr( trim( $line ), 0, 1 );
2110 if ( $comment == '#' ||
$comment == '' ) {
2113 // Remove additional comments after a prefix
2114 $comment = strpos( $line, '#' );
2115 if ( $comment > 0 ) {
2116 $line = substr( $line, 0, $comment - 1 );
2118 $blacklist[] = trim( $line );
2126 * Gets image info about the file just uploaded.
2128 * Also has the effect of setting metadata to be an 'indexed tag name' in
2129 * returned API result if 'metadata' was requested. Oddly, we have to pass
2130 * the "result" object down just so it can do that with the appropriate
2131 * format, presumably.
2133 * @param ApiResult $result
2134 * @return array Image info
2136 public function getImageInfo( $result ) {
2137 $localFile = $this->getLocalFile();
2138 $stashFile = $this->getStashFile();
2139 // Calling a different API module depending on whether the file was stashed is less than optimal.
2140 // In fact, calling API modules here at all is less than optimal. Maybe it should be refactored.
2142 $imParam = ApiQueryStashImageInfo
::getPropertyNames();
2143 $info = ApiQueryStashImageInfo
::getInfo( $stashFile, array_flip( $imParam ), $result );
2145 $imParam = ApiQueryImageInfo
::getPropertyNames();
2146 $info = ApiQueryImageInfo
::getInfo( $localFile, array_flip( $imParam ), $result );
2153 * @param array $error
2156 public function convertVerifyErrorToStatus( $error ) {
2157 $code = $error['status'];
2158 unset( $code['status'] );
2160 return Status
::newFatal( $this->getVerificationErrorCode( $code ), $error );
2164 * Get the MediaWiki maximum uploaded file size for given type of upload, based on
2167 * @param null|string $forType
2170 public static function getMaxUploadSize( $forType = null ) {
2171 global $wgMaxUploadSize;
2173 if ( is_array( $wgMaxUploadSize ) ) {
2174 if ( !is_null( $forType ) && isset( $wgMaxUploadSize[$forType] ) ) {
2175 return $wgMaxUploadSize[$forType];
2177 return $wgMaxUploadSize['*'];
2180 return intval( $wgMaxUploadSize );
2185 * Get the PHP maximum uploaded file size, based on ini settings. If there is no limit or the
2186 * limit can't be guessed, returns a very large number (PHP_INT_MAX).
2191 public static function getMaxPhpUploadSize() {
2192 $phpMaxFileSize = wfShorthandToInteger(
2193 ini_get( 'upload_max_filesize' ) ?
: ini_get( 'hhvm.server.upload.upload_max_file_size' ),
2196 $phpMaxPostSize = wfShorthandToInteger(
2197 ini_get( 'post_max_size' ) ?
: ini_get( 'hhvm.server.max_post_size' ),
2200 return min( $phpMaxFileSize, $phpMaxPostSize );
2204 * Get the current status of a chunked upload (used for polling)
2206 * The value will be read from cache.
2209 * @param string $statusKey
2210 * @return Status[]|bool
2212 public static function getSessionStatus( User
$user, $statusKey ) {
2213 $cache = MediaWikiServices
::getInstance()->getMainObjectStash();
2214 $key = $cache->makeKey( 'uploadstatus', $user->getId() ?
: md5( $user->getName() ), $statusKey );
2216 return $cache->get( $key );
2220 * Set the current status of a chunked upload (used for polling)
2222 * The value will be set in cache for 1 day
2225 * @param string $statusKey
2226 * @param array|bool $value
2229 public static function setSessionStatus( User
$user, $statusKey, $value ) {
2230 $cache = MediaWikiServices
::getInstance()->getMainObjectStash();
2231 $key = $cache->makeKey( 'uploadstatus', $user->getId() ?
: md5( $user->getName() ), $statusKey );
2233 if ( $value === false ) {
2234 $cache->delete( $key );
2236 $cache->set( $key, $value, $cache::TTL_DAY
);