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() {
326 * If there was no filename or a zero size given, give up quick.
328 if ( $this->isEmptyFile() ) {
329 return [ 'status' => self
::EMPTY_FILE
];
333 * Honor $wgMaxUploadSize
335 $maxSize = self
::getMaxUploadSize( $this->getSourceType() );
336 if ( $this->mFileSize
> $maxSize ) {
338 'status' => self
::FILE_TOO_LARGE
,
344 * Look at the contents of the file; if we can recognize the
345 * type but it's corrupt or data of the wrong type, we should
346 * probably not accept it.
348 $verification = $this->verifyFile();
349 if ( $verification !== true ) {
351 'status' => self
::VERIFICATION_ERROR
,
352 'details' => $verification
357 * Make sure this file can be created
359 $result = $this->validateName();
360 if ( $result !== true ) {
365 if ( !Hooks
::run( 'UploadVerification',
366 [ $this->mDestName
, $this->mTempPath
, &$error ], '1.28' )
368 return [ 'status' => self
::HOOK_ABORTED
, 'error' => $error ];
371 return [ 'status' => self
::OK
];
375 * Verify that the name is valid and, if necessary, that we can overwrite
377 * @return mixed True if valid, otherwise and array with 'status'
380 public function validateName() {
381 $nt = $this->getTitle();
382 if ( is_null( $nt ) ) {
383 $result = [ 'status' => $this->mTitleError
];
384 if ( $this->mTitleError
== self
::ILLEGAL_FILENAME
) {
385 $result['filtered'] = $this->mFilteredName
;
387 if ( $this->mTitleError
== self
::FILETYPE_BADTYPE
) {
388 $result['finalExt'] = $this->mFinalExtension
;
389 if ( count( $this->mBlackListedExtensions
) ) {
390 $result['blacklistedExt'] = $this->mBlackListedExtensions
;
396 $this->mDestName
= $this->getLocalFile()->getName();
402 * Verify the MIME type.
404 * @note Only checks that it is not an evil MIME. The "does it have
405 * correct extension given its MIME type?" check is in verifyFile.
406 * in `verifyFile()` that MIME type and file extension correlate.
407 * @param string $mime Representing the MIME
408 * @return mixed True if the file is verified, an array otherwise
410 protected function verifyMimeType( $mime ) {
411 global $wgVerifyMimeType;
412 if ( $wgVerifyMimeType ) {
413 wfDebug( "mime: <$mime> extension: <{$this->mFinalExtension}>\n" );
414 global $wgMimeTypeBlacklist;
415 if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
416 return [ 'filetype-badmime', $mime ];
419 # Check what Internet Explorer would detect
420 $fp = fopen( $this->mTempPath
, 'rb' );
421 $chunk = fread( $fp, 256 );
424 $magic = MimeMagic
::singleton();
425 $extMime = $magic->guessTypesForExtension( $this->mFinalExtension
);
426 $ieTypes = $magic->getIEMimeTypes( $this->mTempPath
, $chunk, $extMime );
427 foreach ( $ieTypes as $ieType ) {
428 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
429 return [ 'filetype-bad-ie-mime', $ieType ];
438 * Verifies that it's ok to include the uploaded file
440 * @return mixed True of the file is verified, array otherwise.
442 protected function verifyFile() {
443 global $wgVerifyMimeType, $wgDisableUploadScriptChecks;
445 $status = $this->verifyPartialFile();
446 if ( $status !== true ) {
450 $mwProps = new MWFileProps( MimeMagic
::singleton() );
451 $this->mFileProps
= $mwProps->getPropsFromPath( $this->mTempPath
, $this->mFinalExtension
);
452 $mime = $this->mFileProps
['mime'];
454 if ( $wgVerifyMimeType ) {
455 # XXX: Missing extension will be caught by validateName() via getTitle()
456 if ( $this->mFinalExtension
!= '' && !$this->verifyExtension( $mime, $this->mFinalExtension
) ) {
457 return [ 'filetype-mime-mismatch', $this->mFinalExtension
, $mime ];
461 # check for htmlish code and javascript
462 if ( !$wgDisableUploadScriptChecks ) {
463 if ( $this->mFinalExtension
== 'svg' ||
$mime == 'image/svg+xml' ) {
464 $svgStatus = $this->detectScriptInSvg( $this->mTempPath
, false );
465 if ( $svgStatus !== false ) {
471 $handler = MediaHandler
::getHandler( $mime );
473 $handlerStatus = $handler->verifyUpload( $this->mTempPath
);
474 if ( !$handlerStatus->isOK() ) {
475 $errors = $handlerStatus->getErrorsArray();
477 return reset( $errors );
482 Hooks
::run( 'UploadVerifyFile', [ $this, $mime, &$error ] );
483 if ( $error !== true ) {
484 if ( !is_array( $error ) ) {
490 wfDebug( __METHOD__
. ": all clear; passing.\n" );
496 * A verification routine suitable for partial files
498 * Runs the blacklist checks, but not any checks that may
499 * assume the entire file is present.
501 * @return mixed True for valid or array with error message key.
503 protected function verifyPartialFile() {
504 global $wgAllowJavaUploads, $wgDisableUploadScriptChecks;
506 # getTitle() sets some internal parameters like $this->mFinalExtension
509 $mwProps = new MWFileProps( MimeMagic
::singleton() );
510 $this->mFileProps
= $mwProps->getPropsFromPath( $this->mTempPath
, $this->mFinalExtension
);
512 # check MIME type, if desired
513 $mime = $this->mFileProps
['file-mime'];
514 $status = $this->verifyMimeType( $mime );
515 if ( $status !== true ) {
519 # check for htmlish code and javascript
520 if ( !$wgDisableUploadScriptChecks ) {
521 if ( self
::detectScript( $this->mTempPath
, $mime, $this->mFinalExtension
) ) {
522 return [ 'uploadscripted' ];
524 if ( $this->mFinalExtension
== 'svg' ||
$mime == 'image/svg+xml' ) {
525 $svgStatus = $this->detectScriptInSvg( $this->mTempPath
, true );
526 if ( $svgStatus !== false ) {
532 # Check for Java applets, which if uploaded can bypass cross-site
534 if ( !$wgAllowJavaUploads ) {
535 $this->mJavaDetected
= false;
536 $zipStatus = ZipDirectoryReader
::read( $this->mTempPath
,
537 [ $this, 'zipEntryCallback' ] );
538 if ( !$zipStatus->isOK() ) {
539 $errors = $zipStatus->getErrorsArray();
540 $error = reset( $errors );
541 if ( $error[0] !== 'zip-wrong-format' ) {
545 if ( $this->mJavaDetected
) {
546 return [ 'uploadjava' ];
550 # Scan the uploaded file for viruses
551 $virus = $this->detectVirus( $this->mTempPath
);
553 return [ 'uploadvirus', $virus ];
560 * Callback for ZipDirectoryReader to detect Java class files.
562 * @param array $entry
564 public function zipEntryCallback( $entry ) {
565 $names = [ $entry['name'] ];
567 // If there is a null character, cut off the name at it, because JDK's
568 // ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name
569 // were constructed which had ".class\0" followed by a string chosen to
570 // make the hash collide with the truncated name, that file could be
571 // returned in response to a request for the .class file.
572 $nullPos = strpos( $entry['name'], "\000" );
573 if ( $nullPos !== false ) {
574 $names[] = substr( $entry['name'], 0, $nullPos );
577 // If there is a trailing slash in the file name, we have to strip it,
578 // because that's what ZIP_GetEntry() does.
579 if ( preg_grep( '!\.class/?$!', $names ) ) {
580 $this->mJavaDetected
= true;
585 * Alias for verifyTitlePermissions. The function was originally
586 * 'verifyPermissions', but that suggests it's checking the user, when it's
587 * really checking the title + user combination.
589 * @param User $user User object to verify the permissions against
590 * @return mixed An array as returned by getUserPermissionsErrors or true
591 * in case the user has proper permissions.
593 public function verifyPermissions( $user ) {
594 return $this->verifyTitlePermissions( $user );
598 * Check whether the user can edit, upload and create the image. This
599 * checks only against the current title; if it returns errors, it may
600 * very well be that another title will not give errors. Therefore
601 * isAllowed() should be called as well for generic is-user-blocked or
602 * can-user-upload checking.
604 * @param User $user User object to verify the permissions against
605 * @return mixed An array as returned by getUserPermissionsErrors or true
606 * in case the user has proper permissions.
608 public function verifyTitlePermissions( $user ) {
610 * If the image is protected, non-sysop users won't be able
611 * to modify it by uploading a new revision.
613 $nt = $this->getTitle();
614 if ( is_null( $nt ) ) {
617 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
618 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
619 if ( !$nt->exists() ) {
620 $permErrorsCreate = $nt->getUserPermissionsErrors( 'create', $user );
622 $permErrorsCreate = [];
624 if ( $permErrors ||
$permErrorsUpload ||
$permErrorsCreate ) {
625 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
626 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
631 $overwriteError = $this->checkOverwrite( $user );
632 if ( $overwriteError !== true ) {
633 return [ $overwriteError ];
640 * Check for non fatal problems with the file.
642 * This should not assume that mTempPath is set.
644 * @return array Array of warnings
646 public function checkWarnings() {
651 $localFile = $this->getLocalFile();
652 $localFile->load( File
::READ_LATEST
);
653 $filename = $localFile->getName();
656 * Check whether the resulting filename is different from the desired one,
657 * but ignore things like ucfirst() and spaces/underscore things
659 $comparableName = str_replace( ' ', '_', $this->mDesiredDestName
);
660 $comparableName = Title
::capitalize( $comparableName, NS_FILE
);
662 if ( $this->mDesiredDestName
!= $filename && $comparableName != $filename ) {
663 $warnings['badfilename'] = $filename;
666 // Check whether the file extension is on the unwanted list
667 global $wgCheckFileExtensions, $wgFileExtensions;
668 if ( $wgCheckFileExtensions ) {
669 $extensions = array_unique( $wgFileExtensions );
670 if ( !$this->checkFileExtension( $this->mFinalExtension
, $extensions ) ) {
671 $warnings['filetype-unwanted-type'] = [ $this->mFinalExtension
,
672 $wgLang->commaList( $extensions ), count( $extensions ) ];
676 global $wgUploadSizeWarning;
677 if ( $wgUploadSizeWarning && ( $this->mFileSize
> $wgUploadSizeWarning ) ) {
678 $warnings['large-file'] = [ $wgUploadSizeWarning, $this->mFileSize
];
681 if ( $this->mFileSize
== 0 ) {
682 $warnings['empty-file'] = true;
685 $hash = $this->getTempFileSha1Base36();
686 $exists = self
::getExistsWarning( $localFile );
687 if ( $exists !== false ) {
688 $warnings['exists'] = $exists;
690 // check if file is an exact duplicate of current file version
691 if ( $hash === $localFile->getSha1() ) {
692 $warnings['no-change'] = $localFile;
695 // check if file is an exact duplicate of older versions of this file
696 $history = $localFile->getHistory();
697 foreach ( $history as $oldFile ) {
698 if ( $hash === $oldFile->getSha1() ) {
699 $warnings['duplicate-version'][] = $oldFile;
704 if ( $localFile->wasDeleted() && !$localFile->exists() ) {
705 $warnings['was-deleted'] = $filename;
708 // Check dupes against existing files
709 $dupes = RepoGroup
::singleton()->findBySha1( $hash );
710 $title = $this->getTitle();
711 // Remove all matches against self
712 foreach ( $dupes as $key => $dupe ) {
713 if ( $title->equals( $dupe->getTitle() ) ) {
714 unset( $dupes[$key] );
718 $warnings['duplicate'] = $dupes;
721 // Check dupes against archives
722 $archivedFile = new ArchivedFile( null, 0, '', $hash );
723 if ( $archivedFile->getID() > 0 ) {
724 if ( $archivedFile->userCan( File
::DELETED_FILE
) ) {
725 $warnings['duplicate-archive'] = $archivedFile->getName();
727 $warnings['duplicate-archive'] = '';
735 * Really perform the upload. Stores the file in the local repo, watches
736 * if necessary and runs the UploadComplete hook.
738 * @param string $comment
739 * @param string $pageText
740 * @param bool $watch Whether the file page should be added to user's watchlist.
741 * (This doesn't check $user's permissions.)
743 * @param string[] $tags Change tags to add to the log entry and page revision.
744 * (This doesn't check $user's permissions.)
745 * @return Status Indicating the whether the upload succeeded.
747 public function performUpload( $comment, $pageText, $watch, $user, $tags = [] ) {
748 $this->getLocalFile()->load( File
::READ_LATEST
);
749 $props = $this->mFileProps
;
752 Hooks
::run( 'UploadVerifyUpload', [ $this, $user, $props, $comment, $pageText, &$error ] );
754 if ( !is_array( $error ) ) {
757 return call_user_func_array( 'Status::newFatal', $error );
760 $status = $this->getLocalFile()->upload(
771 if ( $status->isGood() ) {
773 WatchAction
::doWatch(
774 $this->getLocalFile()->getTitle(),
776 User
::IGNORE_USER_RIGHTS
779 // Avoid PHP 7.1 warning of passing $this by reference
781 Hooks
::run( 'UploadComplete', [ &$uploadBase ] );
783 $this->postProcessUpload();
790 * Perform extra steps after a successful upload.
794 public function postProcessUpload() {
798 * Returns the title of the file to be uploaded. Sets mTitleError in case
799 * the name was illegal.
801 * @return Title The title of the file or null in case the name was illegal
803 public function getTitle() {
804 if ( $this->mTitle
!== false ) {
805 return $this->mTitle
;
807 if ( !is_string( $this->mDesiredDestName
) ) {
808 $this->mTitleError
= self
::ILLEGAL_FILENAME
;
809 $this->mTitle
= null;
811 return $this->mTitle
;
813 /* Assume that if a user specified File:Something.jpg, this is an error
814 * and that the namespace prefix needs to be stripped of.
816 $title = Title
::newFromText( $this->mDesiredDestName
);
817 if ( $title && $title->getNamespace() == NS_FILE
) {
818 $this->mFilteredName
= $title->getDBkey();
820 $this->mFilteredName
= $this->mDesiredDestName
;
823 # oi_archive_name is max 255 bytes, which include a timestamp and an
824 # exclamation mark, so restrict file name to 240 bytes.
825 if ( strlen( $this->mFilteredName
) > 240 ) {
826 $this->mTitleError
= self
::FILENAME_TOO_LONG
;
827 $this->mTitle
= null;
829 return $this->mTitle
;
833 * Chop off any directories in the given filename. Then
834 * filter out illegal characters, and try to make a legible name
835 * out of it. We'll strip some silently that Title would die on.
837 $this->mFilteredName
= wfStripIllegalFilenameChars( $this->mFilteredName
);
838 /* Normalize to title form before we do any further processing */
839 $nt = Title
::makeTitleSafe( NS_FILE
, $this->mFilteredName
);
840 if ( is_null( $nt ) ) {
841 $this->mTitleError
= self
::ILLEGAL_FILENAME
;
842 $this->mTitle
= null;
844 return $this->mTitle
;
846 $this->mFilteredName
= $nt->getDBkey();
849 * We'll want to blacklist against *any* 'extension', and use
850 * only the final one for the whitelist.
852 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName
);
854 if ( count( $ext ) ) {
855 $this->mFinalExtension
= trim( $ext[count( $ext ) - 1] );
857 $this->mFinalExtension
= '';
859 # No extension, try guessing one
860 $magic = MimeMagic
::singleton();
861 $mime = $magic->guessMimeType( $this->mTempPath
);
862 if ( $mime !== 'unknown/unknown' ) {
863 # Get a space separated list of extensions
864 $extList = $magic->getExtensionsForType( $mime );
866 # Set the extension to the canonical extension
867 $this->mFinalExtension
= strtok( $extList, ' ' );
869 # Fix up the other variables
870 $this->mFilteredName
.= ".{$this->mFinalExtension}";
871 $nt = Title
::makeTitleSafe( NS_FILE
, $this->mFilteredName
);
872 $ext = [ $this->mFinalExtension
];
877 /* Don't allow users to override the blacklist (check file extension) */
878 global $wgCheckFileExtensions, $wgStrictFileExtensions;
879 global $wgFileExtensions, $wgFileBlacklist;
881 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
883 if ( $this->mFinalExtension
== '' ) {
884 $this->mTitleError
= self
::FILETYPE_MISSING
;
885 $this->mTitle
= null;
887 return $this->mTitle
;
888 } elseif ( $blackListedExtensions ||
889 ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
890 !$this->checkFileExtension( $this->mFinalExtension
, $wgFileExtensions ) )
892 $this->mBlackListedExtensions
= $blackListedExtensions;
893 $this->mTitleError
= self
::FILETYPE_BADTYPE
;
894 $this->mTitle
= null;
896 return $this->mTitle
;
899 // Windows may be broken with special characters, see T3780
900 if ( !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() )
901 && !RepoGroup
::singleton()->getLocalRepo()->backendSupportsUnicodePaths()
903 $this->mTitleError
= self
::WINDOWS_NONASCII_FILENAME
;
904 $this->mTitle
= null;
906 return $this->mTitle
;
909 # If there was more than one "extension", reassemble the base
910 # filename to prevent bogus complaints about length
911 if ( count( $ext ) > 1 ) {
912 $iterations = count( $ext ) - 1;
913 for ( $i = 0; $i < $iterations; $i++
) {
914 $partname .= '.' . $ext[$i];
918 if ( strlen( $partname ) < 1 ) {
919 $this->mTitleError
= self
::MIN_LENGTH_PARTNAME
;
920 $this->mTitle
= null;
922 return $this->mTitle
;
927 return $this->mTitle
;
931 * Return the local file and initializes if necessary.
933 * @return LocalFile|null
935 public function getLocalFile() {
936 if ( is_null( $this->mLocalFile
) ) {
937 $nt = $this->getTitle();
938 $this->mLocalFile
= is_null( $nt ) ?
null : wfLocalFile( $nt );
941 return $this->mLocalFile
;
945 * @return UploadStashFile|null
947 public function getStashFile() {
948 return $this->mStashFile
;
952 * Like stashFile(), but respects extensions' wishes to prevent the stashing. verifyUpload() must
953 * be called before calling this method (unless $isPartial is true).
955 * Upload stash exceptions are also caught and converted to an error status.
959 * @param bool $isPartial Pass `true` if this is a part of a chunked upload (not a complete file).
960 * @return Status If successful, value is an UploadStashFile instance
962 public function tryStashFile( User
$user, $isPartial = false ) {
964 $error = $this->runUploadStashFileHook( $user );
966 return call_user_func_array( 'Status::newFatal', $error );
970 $file = $this->doStashFile( $user );
971 return Status
::newGood( $file );
972 } catch ( UploadStashException
$e ) {
973 return Status
::newFatal( 'uploadstash-exception', get_class( $e ), $e->getMessage() );
979 * @return array|null Error message and parameters, null if there's no error
981 protected function runUploadStashFileHook( User
$user ) {
982 $props = $this->mFileProps
;
984 Hooks
::run( 'UploadStashFile', [ $this, $user, $props, &$error ] );
986 if ( !is_array( $error ) ) {
994 * If the user does not supply all necessary information in the first upload
995 * form submission (either by accident or by design) then we may want to
996 * stash the file temporarily, get more information, and publish the file
999 * This method will stash a file in a temporary directory for later
1000 * processing, and save the necessary descriptive info into the database.
1001 * This method returns the file object, which also has a 'fileKey' property
1002 * which can be passed through a form or API request to find this stashed
1005 * @deprecated since 1.28 Use tryStashFile() instead
1007 * @return UploadStashFile Stashed file
1008 * @throws UploadStashBadPathException
1009 * @throws UploadStashFileException
1010 * @throws UploadStashNotLoggedInException
1012 public function stashFile( User
$user = null ) {
1013 return $this->doStashFile( $user );
1017 * Implementation for stashFile() and tryStashFile().
1020 * @return UploadStashFile Stashed file
1022 protected function doStashFile( User
$user = null ) {
1023 $stash = RepoGroup
::singleton()->getLocalRepo()->getUploadStash( $user );
1024 $file = $stash->stashFile( $this->mTempPath
, $this->getSourceType() );
1025 $this->mStashFile
= $file;
1031 * Stash a file in a temporary directory, returning a key which can be used
1032 * to find the file again. See stashFile().
1034 * @deprecated since 1.28
1035 * @return string File key
1037 public function stashFileGetKey() {
1038 wfDeprecated( __METHOD__
, '1.28' );
1039 return $this->doStashFile()->getFileKey();
1043 * alias for stashFileGetKey, for backwards compatibility
1045 * @deprecated since 1.28
1046 * @return string File key
1048 public function stashSession() {
1049 wfDeprecated( __METHOD__
, '1.28' );
1050 return $this->doStashFile()->getFileKey();
1054 * If we've modified the upload file we need to manually remove it
1055 * on exit to clean up.
1057 public function cleanupTempFile() {
1058 if ( $this->mRemoveTempFile
&& $this->tempFileObj
) {
1059 // Delete when all relevant TempFSFile handles go out of scope
1060 wfDebug( __METHOD__
. ": Marked temporary file '{$this->mTempPath}' for removal\n" );
1061 $this->tempFileObj
->autocollect();
1065 public function getTempPath() {
1066 return $this->mTempPath
;
1070 * Split a file into a base name and all dot-delimited 'extensions'
1071 * on the end. Some web server configurations will fall back to
1072 * earlier pseudo-'extensions' to determine type and execute
1073 * scripts, so the blacklist needs to check them all.
1075 * @param string $filename
1078 public static function splitExtensions( $filename ) {
1079 $bits = explode( '.', $filename );
1080 $basename = array_shift( $bits );
1082 return [ $basename, $bits ];
1086 * Perform case-insensitive match against a list of file extensions.
1087 * Returns true if the extension is in the list.
1089 * @param string $ext
1090 * @param array $list
1093 public static function checkFileExtension( $ext, $list ) {
1094 return in_array( strtolower( $ext ), $list );
1098 * Perform case-insensitive match against a list of file extensions.
1099 * Returns an array of matching extensions.
1102 * @param array $list
1105 public static function checkFileExtensionList( $ext, $list ) {
1106 return array_intersect( array_map( 'strtolower', $ext ), $list );
1110 * Checks if the MIME type of the uploaded file matches the file extension.
1112 * @param string $mime The MIME type of the uploaded file
1113 * @param string $extension The filename extension that the file is to be served with
1116 public static function verifyExtension( $mime, $extension ) {
1117 $magic = MimeMagic
::singleton();
1119 if ( !$mime ||
$mime == 'unknown' ||
$mime == 'unknown/unknown' ) {
1120 if ( !$magic->isRecognizableExtension( $extension ) ) {
1121 wfDebug( __METHOD__
. ": passing file with unknown detected mime type; " .
1122 "unrecognized extension '$extension', can't verify\n" );
1126 wfDebug( __METHOD__
. ": rejecting file with unknown detected mime type; " .
1127 "recognized extension '$extension', so probably invalid file\n" );
1133 $match = $magic->isMatchingExtension( $extension, $mime );
1135 if ( $match === null ) {
1136 if ( $magic->getTypesForExtension( $extension ) !== null ) {
1137 wfDebug( __METHOD__
. ": No extension known for $mime, but we know a mime for $extension\n" );
1141 wfDebug( __METHOD__
. ": no file extension known for mime type $mime, passing file\n" );
1145 } elseif ( $match === true ) {
1146 wfDebug( __METHOD__
. ": mime type $mime matches extension $extension, passing file\n" );
1148 /** @todo If it's a bitmap, make sure PHP or ImageMagick resp. can handle it! */
1152 . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
1159 * Heuristic for detecting files that *could* contain JavaScript instructions or
1160 * things that may look like HTML to a browser and are thus
1161 * potentially harmful. The present implementation will produce false
1162 * positives in some situations.
1164 * @param string $file Pathname to the temporary upload file
1165 * @param string $mime The MIME type of the file
1166 * @param string $extension The extension of the file
1167 * @return bool True if the file contains something looking like embedded scripts
1169 public static function detectScript( $file, $mime, $extension ) {
1170 global $wgAllowTitlesInSVG;
1172 # ugly hack: for text files, always look at the entire file.
1173 # For binary field, just check the first K.
1175 if ( strpos( $mime, 'text/' ) === 0 ) {
1176 $chunk = file_get_contents( $file );
1178 $fp = fopen( $file, 'rb' );
1179 $chunk = fread( $fp, 1024 );
1183 $chunk = strtolower( $chunk );
1189 # decode from UTF-16 if needed (could be used for obfuscation).
1190 if ( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
1192 } elseif ( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
1199 $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
1202 $chunk = trim( $chunk );
1204 /** @todo FIXME: Convert from UTF-16 if necessary! */
1205 wfDebug( __METHOD__
. ": checking for embedded scripts and HTML stuff\n" );
1207 # check for HTML doctype
1208 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
1212 // Some browsers will interpret obscure xml encodings as UTF-8, while
1213 // PHP/expat will interpret the given encoding in the xml declaration (T49304)
1214 if ( $extension == 'svg' ||
strpos( $mime, 'image/svg' ) === 0 ) {
1215 if ( self
::checkXMLEncodingMissmatch( $file ) ) {
1221 * Internet Explorer for Windows performs some really stupid file type
1222 * autodetection which can cause it to interpret valid image files as HTML
1223 * and potentially execute JavaScript, creating a cross-site scripting
1226 * Apple's Safari browser also performs some unsafe file type autodetection
1227 * which can cause legitimate files to be interpreted as HTML if the
1228 * web server is not correctly configured to send the right content-type
1229 * (or if you're really uploading plain text and octet streams!)
1231 * Returns true if IE is likely to mistake the given file for HTML.
1232 * Also returns true if Safari would mistake the given file for HTML
1233 * when served with a generic content-type.
1239 '<html', # also in safari
1242 '<script', # also in safari
1246 if ( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
1250 foreach ( $tags as $tag ) {
1251 if ( false !== strpos( $chunk, $tag ) ) {
1252 wfDebug( __METHOD__
. ": found something that may make it be mistaken for html: $tag\n" );
1259 * look for JavaScript
1262 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
1263 $chunk = Sanitizer
::decodeCharReferences( $chunk );
1265 # look for script-types
1266 if ( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
1267 wfDebug( __METHOD__
. ": found script types\n" );
1272 # look for html-style script-urls
1273 if ( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1274 wfDebug( __METHOD__
. ": found html-style script urls\n" );
1279 # look for css-style script-urls
1280 if ( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1281 wfDebug( __METHOD__
. ": found css-style script urls\n" );
1286 wfDebug( __METHOD__
. ": no scripts found\n" );
1292 * Check a whitelist of xml encodings that are known not to be interpreted differently
1293 * by the server's xml parser (expat) and some common browsers.
1295 * @param string $file Pathname to the temporary upload file
1296 * @return bool True if the file contains an encoding that could be misinterpreted
1298 public static function checkXMLEncodingMissmatch( $file ) {
1299 global $wgSVGMetadataCutoff;
1300 $contents = file_get_contents( $file, false, null, -1, $wgSVGMetadataCutoff );
1301 $encodingRegex = '!encoding[ \t\n\r]*=[ \t\n\r]*[\'"](.*?)[\'"]!si';
1303 if ( preg_match( "!<\?xml\b(.*?)\?>!si", $contents, $matches ) ) {
1304 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1305 && !in_array( strtoupper( $encMatch[1] ), self
::$safeXmlEncodings )
1307 wfDebug( __METHOD__
. ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1311 } elseif ( preg_match( "!<\?xml\b!si", $contents ) ) {
1312 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1313 // bytes. There shouldn't be a legitimate reason for this to happen.
1314 wfDebug( __METHOD__
. ": Unmatched XML declaration start\n" );
1317 } elseif ( substr( $contents, 0, 4 ) == "\x4C\x6F\xA7\x94" ) {
1318 // EBCDIC encoded XML
1319 wfDebug( __METHOD__
. ": EBCDIC Encoded XML\n" );
1324 // It's possible the file is encoded with multi-byte encoding, so re-encode attempt to
1325 // detect the encoding in case is specifies an encoding not whitelisted in self::$safeXmlEncodings
1326 $attemptEncodings = [ 'UTF-16', 'UTF-16BE', 'UTF-32', 'UTF-32BE' ];
1327 foreach ( $attemptEncodings as $encoding ) {
1328 MediaWiki\
suppressWarnings();
1329 $str = iconv( $encoding, 'UTF-8', $contents );
1330 MediaWiki\restoreWarnings
();
1331 if ( $str != '' && preg_match( "!<\?xml\b(.*?)\?>!si", $str, $matches ) ) {
1332 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1333 && !in_array( strtoupper( $encMatch[1] ), self
::$safeXmlEncodings )
1335 wfDebug( __METHOD__
. ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1339 } elseif ( $str != '' && preg_match( "!<\?xml\b!si", $str ) ) {
1340 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1341 // bytes. There shouldn't be a legitimate reason for this to happen.
1342 wfDebug( __METHOD__
. ": Unmatched XML declaration start\n" );
1352 * @param string $filename
1353 * @param bool $partial
1354 * @return mixed False of the file is verified (does not contain scripts), array otherwise.
1356 protected function detectScriptInSvg( $filename, $partial ) {
1357 $this->mSVGNSError
= false;
1358 $check = new XmlTypeCheck(
1360 [ $this, 'checkSvgScriptCallback' ],
1363 'processing_instruction_handler' => 'UploadBase::checkSvgPICallback',
1364 'external_dtd_handler' => 'UploadBase::checkSvgExternalDTD',
1367 if ( $check->wellFormed
!== true ) {
1368 // Invalid xml (T60553)
1369 // But only when non-partial (T67724)
1370 return $partial ?
false : [ 'uploadinvalidxml' ];
1371 } elseif ( $check->filterMatch
) {
1372 if ( $this->mSVGNSError
) {
1373 return [ 'uploadscriptednamespace', $this->mSVGNSError
];
1376 return $check->filterMatchType
;
1383 * Callback to filter SVG Processing Instructions.
1384 * @param string $target Processing instruction name
1385 * @param string $data Processing instruction attribute and value
1386 * @return bool (true if the filter identified something bad)
1388 public static function checkSvgPICallback( $target, $data ) {
1389 // Don't allow external stylesheets (T59550)
1390 if ( preg_match( '/xml-stylesheet/i', $target ) ) {
1391 return [ 'upload-scripted-pi-callback' ];
1398 * Verify that DTD urls referenced are only the standard dtds
1400 * Browsers seem to ignore external dtds. However just to be on the
1401 * safe side, only allow dtds from the svg standard.
1403 * @param string $type PUBLIC or SYSTEM
1404 * @param string $publicId The well-known public identifier for the dtd
1405 * @param string $systemId The url for the external dtd
1407 public static function checkSvgExternalDTD( $type, $publicId, $systemId ) {
1408 // This doesn't include the XHTML+MathML+SVG doctype since we don't
1409 // allow XHTML anyways.
1411 'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd',
1412 'http://www.w3.org/TR/2001/REC-SVG-20010904/DTD/svg10.dtd',
1413 'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-basic.dtd',
1414 'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-tiny.dtd'
1416 if ( $type !== 'PUBLIC'
1417 ||
!in_array( $systemId, $allowedDTDs )
1418 ||
strpos( $publicId, "-//W3C//" ) !== 0
1420 return [ 'upload-scripted-dtd' ];
1426 * @todo Replace this with a whitelist filter!
1427 * @param string $element
1428 * @param array $attribs
1431 public function checkSvgScriptCallback( $element, $attribs, $data = null ) {
1433 list( $namespace, $strippedElement ) = $this->splitXmlNamespace( $element );
1435 // We specifically don't include:
1436 // http://www.w3.org/1999/xhtml (T62771)
1437 static $validNamespaces = [
1440 'http://creativecommons.org/ns#',
1441 'http://inkscape.sourceforge.net/dtd/sodipodi-0.dtd',
1442 'http://ns.adobe.com/adobeillustrator/10.0/',
1443 'http://ns.adobe.com/adobesvgviewerextensions/3.0/',
1444 'http://ns.adobe.com/extensibility/1.0/',
1445 'http://ns.adobe.com/flows/1.0/',
1446 'http://ns.adobe.com/illustrator/1.0/',
1447 'http://ns.adobe.com/imagereplacement/1.0/',
1448 'http://ns.adobe.com/pdf/1.3/',
1449 'http://ns.adobe.com/photoshop/1.0/',
1450 'http://ns.adobe.com/saveforweb/1.0/',
1451 'http://ns.adobe.com/variables/1.0/',
1452 'http://ns.adobe.com/xap/1.0/',
1453 'http://ns.adobe.com/xap/1.0/g/',
1454 'http://ns.adobe.com/xap/1.0/g/img/',
1455 'http://ns.adobe.com/xap/1.0/mm/',
1456 'http://ns.adobe.com/xap/1.0/rights/',
1457 'http://ns.adobe.com/xap/1.0/stype/dimensions#',
1458 'http://ns.adobe.com/xap/1.0/stype/font#',
1459 'http://ns.adobe.com/xap/1.0/stype/manifestitem#',
1460 'http://ns.adobe.com/xap/1.0/stype/resourceevent#',
1461 'http://ns.adobe.com/xap/1.0/stype/resourceref#',
1462 'http://ns.adobe.com/xap/1.0/t/pg/',
1463 'http://purl.org/dc/elements/1.1/',
1464 'http://purl.org/dc/elements/1.1',
1465 'http://schemas.microsoft.com/visio/2003/svgextensions/',
1466 'http://sodipodi.sourceforge.net/dtd/sodipodi-0.dtd',
1467 'http://taptrix.com/inkpad/svg_extensions',
1468 'http://web.resource.org/cc/',
1469 'http://www.freesoftware.fsf.org/bkchem/cdml',
1470 'http://www.inkscape.org/namespaces/inkscape',
1471 'http://www.opengis.net/gml',
1472 'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
1473 'http://www.w3.org/2000/svg',
1474 'http://www.w3.org/tr/rec-rdf-syntax/',
1475 'http://www.w3.org/2000/01/rdf-schema#',
1478 // Inkscape mangles namespace definitions created by Adobe Illustrator.
1479 // This is nasty but harmless. (T144827)
1480 $isBuggyInkscape = preg_match( '/^&(#38;)*ns_[a-z_]+;$/', $namespace );
1482 if ( !( $isBuggyInkscape ||
in_array( $namespace, $validNamespaces ) ) ) {
1483 wfDebug( __METHOD__
. ": Non-svg namespace '$namespace' in uploaded file.\n" );
1484 /** @todo Return a status object to a closure in XmlTypeCheck, for MW1.21+ */
1485 $this->mSVGNSError
= $namespace;
1491 * check for elements that can contain javascript
1493 if ( $strippedElement == 'script' ) {
1494 wfDebug( __METHOD__
. ": Found script element '$element' in uploaded file.\n" );
1496 return [ 'uploaded-script-svg', $strippedElement ];
1499 # e.g., <svg xmlns="http://www.w3.org/2000/svg">
1500 # <handler xmlns:ev="http://www.w3.org/2001/xml-events" ev:event="load">alert(1)</handler> </svg>
1501 if ( $strippedElement == 'handler' ) {
1502 wfDebug( __METHOD__
. ": Found scriptable element '$element' in uploaded file.\n" );
1504 return [ 'uploaded-script-svg', $strippedElement ];
1507 # SVG reported in Feb '12 that used xml:stylesheet to generate javascript block
1508 if ( $strippedElement == 'stylesheet' ) {
1509 wfDebug( __METHOD__
. ": Found scriptable element '$element' in uploaded file.\n" );
1511 return [ 'uploaded-script-svg', $strippedElement ];
1514 # Block iframes, in case they pass the namespace check
1515 if ( $strippedElement == 'iframe' ) {
1516 wfDebug( __METHOD__
. ": iframe in uploaded file.\n" );
1518 return [ 'uploaded-script-svg', $strippedElement ];
1522 if ( $strippedElement == 'style'
1523 && self
::checkCssFragment( Sanitizer
::normalizeCss( $data ) )
1525 wfDebug( __METHOD__
. ": hostile css in style element.\n" );
1526 return [ 'uploaded-hostile-svg' ];
1529 foreach ( $attribs as $attrib => $value ) {
1530 $stripped = $this->stripXmlNamespace( $attrib );
1531 $value = strtolower( $value );
1533 if ( substr( $stripped, 0, 2 ) == 'on' ) {
1535 . ": Found event-handler attribute '$attrib'='$value' in uploaded file.\n" );
1537 return [ 'uploaded-event-handler-on-svg', $attrib, $value ];
1540 # Do not allow relative links, or unsafe url schemas.
1541 # For <a> tags, only data:, http: and https: and same-document
1542 # fragment links are allowed. For all other tags, only data:
1543 # and fragment are allowed.
1544 if ( $stripped == 'href'
1546 && strpos( $value, 'data:' ) !== 0
1547 && strpos( $value, '#' ) !== 0
1549 if ( !( $strippedElement === 'a'
1550 && preg_match( '!^https?://!i', $value ) )
1552 wfDebug( __METHOD__
. ": Found href attribute <$strippedElement "
1553 . "'$attrib'='$value' in uploaded file.\n" );
1555 return [ 'uploaded-href-attribute-svg', $strippedElement, $attrib, $value ];
1559 # only allow data: targets that should be safe. This prevents vectors like,
1560 # image/svg, text/xml, application/xml, and text/html, which can contain scripts
1561 if ( $stripped == 'href' && strncasecmp( 'data:', $value, 5 ) === 0 ) {
1562 // rfc2397 parameters. This is only slightly slower than (;[\w;]+)*.
1563 // @codingStandardsIgnoreStart Generic.Files.LineLength
1564 $parameters = '(?>;[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+=(?>[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+|"(?>[\0-\x0c\x0e-\x21\x23-\x5b\x5d-\x7f]+|\\\\[\0-\x7f])*"))*(?:;base64)?';
1565 // @codingStandardsIgnoreEnd
1567 if ( !preg_match( "!^data:\s*image/(gif|jpeg|jpg|png)$parameters,!i", $value ) ) {
1568 wfDebug( __METHOD__
. ": Found href to unwhitelisted data: uri "
1569 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1570 return [ 'uploaded-href-unsafe-target-svg', $strippedElement, $attrib, $value ];
1574 # Change href with animate from (http://html5sec.org/#137).
1575 if ( $stripped === 'attributename'
1576 && $strippedElement === 'animate'
1577 && $this->stripXmlNamespace( $value ) == 'href'
1579 wfDebug( __METHOD__
. ": Found animate that might be changing href using from "
1580 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1582 return [ 'uploaded-animate-svg', $strippedElement, $attrib, $value ];
1585 # use set/animate to add event-handler attribute to parent
1586 if ( ( $strippedElement == 'set' ||
$strippedElement == 'animate' )
1587 && $stripped == 'attributename'
1588 && substr( $value, 0, 2 ) == 'on'
1590 wfDebug( __METHOD__
. ": Found svg setting event-handler attribute with "
1591 . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1593 return [ 'uploaded-setting-event-handler-svg', $strippedElement, $stripped, $value ];
1596 # use set to add href attribute to parent element
1597 if ( $strippedElement == 'set'
1598 && $stripped == 'attributename'
1599 && strpos( $value, 'href' ) !== false
1601 wfDebug( __METHOD__
. ": Found svg setting href attribute '$value' in uploaded file.\n" );
1603 return [ 'uploaded-setting-href-svg' ];
1606 # use set to add a remote / data / script target to an element
1607 if ( $strippedElement == 'set'
1608 && $stripped == 'to'
1609 && preg_match( '!(http|https|data|script):!sim', $value )
1611 wfDebug( __METHOD__
. ": Found svg setting attribute to '$value' in uploaded file.\n" );
1613 return [ 'uploaded-wrong-setting-svg', $value ];
1616 # use handler attribute with remote / data / script
1617 if ( $stripped == 'handler' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1618 wfDebug( __METHOD__
. ": Found svg setting handler with remote/data/script "
1619 . "'$attrib'='$value' in uploaded file.\n" );
1621 return [ 'uploaded-setting-handler-svg', $attrib, $value ];
1624 # use CSS styles to bring in remote code
1625 if ( $stripped == 'style'
1626 && self
::checkCssFragment( Sanitizer
::normalizeCss( $value ) )
1628 wfDebug( __METHOD__
. ": Found svg setting a style with "
1629 . "remote url '$attrib'='$value' in uploaded file.\n" );
1630 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1633 # Several attributes can include css, css character escaping isn't allowed
1634 $cssAttrs = [ 'font', 'clip-path', 'fill', 'filter', 'marker',
1635 'marker-end', 'marker-mid', 'marker-start', 'mask', 'stroke' ];
1636 if ( in_array( $stripped, $cssAttrs )
1637 && self
::checkCssFragment( $value )
1639 wfDebug( __METHOD__
. ": Found svg setting a style with "
1640 . "remote url '$attrib'='$value' in uploaded file.\n" );
1641 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1644 # image filters can pull in url, which could be svg that executes scripts
1645 if ( $strippedElement == 'image'
1646 && $stripped == 'filter'
1647 && preg_match( '!url\s*\(!sim', $value )
1649 wfDebug( __METHOD__
. ": Found image filter with url: "
1650 . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1652 return [ 'uploaded-image-filter-svg', $strippedElement, $stripped, $value ];
1656 return false; // No scripts detected
1660 * Check a block of CSS or CSS fragment for anything that looks like
1661 * it is bringing in remote code.
1662 * @param string $value a string of CSS
1663 * @param bool $propOnly only check css properties (start regex with :)
1664 * @return bool true if the CSS contains an illegal string, false if otherwise
1666 private static function checkCssFragment( $value ) {
1668 # Forbid external stylesheets, for both reliability and to protect viewer's privacy
1669 if ( stripos( $value, '@import' ) !== false ) {
1673 # We allow @font-face to embed fonts with data: urls, so we snip the string
1674 # 'url' out so this case won't match when we check for urls below
1675 $pattern = '!(@font-face\s*{[^}]*src:)url(\("data:;base64,)!im';
1676 $value = preg_replace( $pattern, '$1$2', $value );
1678 # Check for remote and executable CSS. Unlike in Sanitizer::checkCss, the CSS
1679 # properties filter and accelerator don't seem to be useful for xss in SVG files.
1680 # Expression and -o-link don't seem to work either, but filtering them here in case.
1681 # Additionally, we catch remote urls like url("http:..., url('http:..., url(http:...,
1682 # but not local ones such as url("#..., url('#..., url(#....
1683 if ( preg_match( '!expression
1685 | -o-link-source\s*:
1686 | -o-replace\s*:!imx', $value ) ) {
1690 if ( preg_match_all(
1691 "!(\s*(url|image|image-set)\s*\(\s*[\"']?\s*[^#]+.*?\))!sim",
1696 # TODO: redo this in one regex. Until then, url("#whatever") matches the first
1697 foreach ( $matches[1] as $match ) {
1698 if ( !preg_match( "!\s*(url|image|image-set)\s*\(\s*(#|'#|\"#)!im", $match ) ) {
1704 if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ) {
1712 * Divide the element name passed by the xml parser to the callback into URI and prifix.
1713 * @param string $element
1714 * @return array Containing the namespace URI and prefix
1716 private static function splitXmlNamespace( $element ) {
1717 // 'http://www.w3.org/2000/svg:script' -> [ 'http://www.w3.org/2000/svg', 'script' ]
1718 $parts = explode( ':', strtolower( $element ) );
1719 $name = array_pop( $parts );
1720 $ns = implode( ':', $parts );
1722 return [ $ns, $name ];
1726 * @param string $name
1729 private function stripXmlNamespace( $name ) {
1730 // 'http://www.w3.org/2000/svg:script' -> 'script'
1731 $parts = explode( ':', strtolower( $name ) );
1733 return array_pop( $parts );
1737 * Generic wrapper function for a virus scanner program.
1738 * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
1739 * $wgAntivirusRequired may be used to deny upload if the scan fails.
1741 * @param string $file Pathname to the temporary upload file
1742 * @return mixed False if not virus is found, null if the scan fails or is disabled,
1743 * or a string containing feedback from the virus scanner if a virus was found.
1744 * If textual feedback is missing but a virus was found, this function returns true.
1746 public static function detectVirus( $file ) {
1747 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
1749 if ( !$wgAntivirus ) {
1750 wfDebug( __METHOD__
. ": virus scanner disabled\n" );
1755 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
1756 wfDebug( __METHOD__
. ": unknown virus scanner: $wgAntivirus\n" );
1757 $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
1758 [ 'virus-badscanner', $wgAntivirus ] );
1760 return wfMessage( 'virus-unknownscanner' )->text() . " $wgAntivirus";
1763 # look up scanner configuration
1764 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
1765 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
1766 $msgPattern = isset( $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ) ?
1767 $wgAntivirusSetup[$wgAntivirus]['messagepattern'] : null;
1769 if ( strpos( $command, "%f" ) === false ) {
1770 # simple pattern: append file to scan
1771 $command .= " " . wfEscapeShellArg( $file );
1773 # complex pattern: replace "%f" with file to scan
1774 $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
1777 wfDebug( __METHOD__
. ": running virus scan: $command \n" );
1779 # execute virus scanner
1782 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1783 # that does not seem to be worth the pain.
1784 # Ask me (Duesentrieb) about it if it's ever needed.
1785 $output = wfShellExecWithStderr( $command, $exitCode );
1787 # map exit code to AV_xxx constants.
1788 $mappedCode = $exitCode;
1789 if ( $exitCodeMap ) {
1790 if ( isset( $exitCodeMap[$exitCode] ) ) {
1791 $mappedCode = $exitCodeMap[$exitCode];
1792 } elseif ( isset( $exitCodeMap["*"] ) ) {
1793 $mappedCode = $exitCodeMap["*"];
1797 /* NB: AV_NO_VIRUS is 0 but AV_SCAN_FAILED is false,
1798 * so we need the strict equalities === and thus can't use a switch here
1800 if ( $mappedCode === AV_SCAN_FAILED
) {
1801 # scan failed (code was mapped to false by $exitCodeMap)
1802 wfDebug( __METHOD__
. ": failed to scan $file (code $exitCode).\n" );
1804 $output = $wgAntivirusRequired
1805 ?
wfMessage( 'virus-scanfailed', [ $exitCode ] )->text()
1807 } elseif ( $mappedCode === AV_SCAN_ABORTED
) {
1808 # scan failed because filetype is unknown (probably imune)
1809 wfDebug( __METHOD__
. ": unsupported file type $file (code $exitCode).\n" );
1811 } elseif ( $mappedCode === AV_NO_VIRUS
) {
1813 wfDebug( __METHOD__
. ": file passed virus scan.\n" );
1816 $output = trim( $output );
1819 $output = true; # if there's no output, return true
1820 } elseif ( $msgPattern ) {
1822 if ( preg_match( $msgPattern, $output, $groups ) ) {
1824 $output = $groups[1];
1829 wfDebug( __METHOD__
. ": FOUND VIRUS! scanner feedback: $output \n" );
1836 * Check if there's an overwrite conflict and, if so, if restrictions
1837 * forbid this user from performing the upload.
1841 * @return mixed True on success, array on failure
1843 private function checkOverwrite( $user ) {
1844 // First check whether the local file can be overwritten
1845 $file = $this->getLocalFile();
1846 $file->load( File
::READ_LATEST
);
1847 if ( $file->exists() ) {
1848 if ( !self
::userCanReUpload( $user, $file ) ) {
1849 return [ 'fileexists-forbidden', $file->getName() ];
1855 /* Check shared conflicts: if the local file does not exist, but
1856 * wfFindFile finds a file, it exists in a shared repository.
1858 $file = wfFindFile( $this->getTitle(), [ 'latest' => true ] );
1859 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1860 return [ 'fileexists-shared-forbidden', $file->getName() ];
1867 * Check if a user is the last uploader
1873 public static function userCanReUpload( User
$user, File
$img ) {
1874 if ( $user->isAllowed( 'reupload' ) ) {
1875 return true; // non-conditional
1876 } elseif ( !$user->isAllowed( 'reupload-own' ) ) {
1880 if ( !( $img instanceof LocalFile
) ) {
1886 return $user->getId() == $img->getUser( 'id' );
1890 * Helper function that does various existence checks for a file.
1891 * The following checks are performed:
1893 * - Article with the same name as the file exists
1894 * - File exists with normalized extension
1895 * - The file looks like a thumbnail and the original exists
1897 * @param File $file The File object to check
1898 * @return mixed False if the file does not exists, else an array
1900 public static function getExistsWarning( $file ) {
1901 if ( $file->exists() ) {
1902 return [ 'warning' => 'exists', 'file' => $file ];
1905 if ( $file->getTitle()->getArticleID() ) {
1906 return [ 'warning' => 'page-exists', 'file' => $file ];
1909 if ( strpos( $file->getName(), '.' ) == false ) {
1910 $partname = $file->getName();
1913 $n = strrpos( $file->getName(), '.' );
1914 $extension = substr( $file->getName(), $n +
1 );
1915 $partname = substr( $file->getName(), 0, $n );
1917 $normalizedExtension = File
::normalizeExtension( $extension );
1919 if ( $normalizedExtension != $extension ) {
1920 // We're not using the normalized form of the extension.
1921 // Normal form is lowercase, using most common of alternate
1922 // extensions (eg 'jpg' rather than 'JPEG').
1924 // Check for another file using the normalized form...
1925 $nt_lc = Title
::makeTitle( NS_FILE
, "{$partname}.{$normalizedExtension}" );
1926 $file_lc = wfLocalFile( $nt_lc );
1928 if ( $file_lc->exists() ) {
1930 'warning' => 'exists-normalized',
1932 'normalizedFile' => $file_lc
1937 // Check for files with the same name but a different extension
1938 $similarFiles = RepoGroup
::singleton()->getLocalRepo()->findFilesByPrefix(
1939 "{$partname}.", 1 );
1940 if ( count( $similarFiles ) ) {
1942 'warning' => 'exists-normalized',
1944 'normalizedFile' => $similarFiles[0],
1948 if ( self
::isThumbName( $file->getName() ) ) {
1949 # Check for filenames like 50px- or 180px-, these are mostly thumbnails
1950 $nt_thb = Title
::newFromText(
1951 substr( $partname, strpos( $partname, '-' ) +
1 ) . '.' . $extension,
1954 $file_thb = wfLocalFile( $nt_thb );
1955 if ( $file_thb->exists() ) {
1957 'warning' => 'thumb',
1959 'thumbFile' => $file_thb
1962 // File does not exist, but we just don't like the name
1964 'warning' => 'thumb-name',
1966 'thumbFile' => $file_thb
1971 foreach ( self
::getFilenamePrefixBlacklist() as $prefix ) {
1972 if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
1974 'warning' => 'bad-prefix',
1985 * Helper function that checks whether the filename looks like a thumbnail
1986 * @param string $filename
1989 public static function isThumbName( $filename ) {
1990 $n = strrpos( $filename, '.' );
1991 $partname = $n ?
substr( $filename, 0, $n ) : $filename;
1994 substr( $partname, 3, 3 ) == 'px-' ||
1995 substr( $partname, 2, 3 ) == 'px-'
1997 preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) );
2001 * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
2003 * @return array List of prefixes
2005 public static function getFilenamePrefixBlacklist() {
2007 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
2008 if ( !$message->isDisabled() ) {
2009 $lines = explode( "\n", $message->plain() );
2010 foreach ( $lines as $line ) {
2011 // Remove comment lines
2012 $comment = substr( trim( $line ), 0, 1 );
2013 if ( $comment == '#' ||
$comment == '' ) {
2016 // Remove additional comments after a prefix
2017 $comment = strpos( $line, '#' );
2018 if ( $comment > 0 ) {
2019 $line = substr( $line, 0, $comment - 1 );
2021 $blacklist[] = trim( $line );
2029 * Gets image info about the file just uploaded.
2031 * Also has the effect of setting metadata to be an 'indexed tag name' in
2032 * returned API result if 'metadata' was requested. Oddly, we have to pass
2033 * the "result" object down just so it can do that with the appropriate
2034 * format, presumably.
2036 * @param ApiResult $result
2037 * @return array Image info
2039 public function getImageInfo( $result ) {
2040 $localFile = $this->getLocalFile();
2041 $stashFile = $this->getStashFile();
2042 // Calling a different API module depending on whether the file was stashed is less than optimal.
2043 // In fact, calling API modules here at all is less than optimal. Maybe it should be refactored.
2045 $imParam = ApiQueryStashImageInfo
::getPropertyNames();
2046 $info = ApiQueryStashImageInfo
::getInfo( $stashFile, array_flip( $imParam ), $result );
2048 $imParam = ApiQueryImageInfo
::getPropertyNames();
2049 $info = ApiQueryImageInfo
::getInfo( $localFile, array_flip( $imParam ), $result );
2056 * @param array $error
2059 public function convertVerifyErrorToStatus( $error ) {
2060 $code = $error['status'];
2061 unset( $code['status'] );
2063 return Status
::newFatal( $this->getVerificationErrorCode( $code ), $error );
2067 * Get the MediaWiki maximum uploaded file size for given type of upload, based on
2070 * @param null|string $forType
2073 public static function getMaxUploadSize( $forType = null ) {
2074 global $wgMaxUploadSize;
2076 if ( is_array( $wgMaxUploadSize ) ) {
2077 if ( !is_null( $forType ) && isset( $wgMaxUploadSize[$forType] ) ) {
2078 return $wgMaxUploadSize[$forType];
2080 return $wgMaxUploadSize['*'];
2083 return intval( $wgMaxUploadSize );
2088 * Get the PHP maximum uploaded file size, based on ini settings. If there is no limit or the
2089 * limit can't be guessed, returns a very large number (PHP_INT_MAX).
2094 public static function getMaxPhpUploadSize() {
2095 $phpMaxFileSize = wfShorthandToInteger(
2096 ini_get( 'upload_max_filesize' ) ?
: ini_get( 'hhvm.server.upload.upload_max_file_size' ),
2099 $phpMaxPostSize = wfShorthandToInteger(
2100 ini_get( 'post_max_size' ) ?
: ini_get( 'hhvm.server.max_post_size' ),
2103 return min( $phpMaxFileSize, $phpMaxPostSize );
2107 * Get the current status of a chunked upload (used for polling)
2109 * The value will be read from cache.
2112 * @param string $statusKey
2113 * @return Status[]|bool
2115 public static function getSessionStatus( User
$user, $statusKey ) {
2116 $key = wfMemcKey( 'uploadstatus', $user->getId() ?
: md5( $user->getName() ), $statusKey );
2118 return MediaWikiServices
::getInstance()->getMainObjectStash()->get( $key );
2122 * Set the current status of a chunked upload (used for polling)
2124 * The value will be set in cache for 1 day
2127 * @param string $statusKey
2128 * @param array|bool $value
2131 public static function setSessionStatus( User
$user, $statusKey, $value ) {
2132 $key = wfMemcKey( 'uploadstatus', $user->getId() ?
: md5( $user->getName() ), $statusKey );
2134 $cache = MediaWikiServices
::getInstance()->getMainObjectStash();
2135 if ( $value === false ) {
2136 $cache->delete( $key );
2138 $cache->set( $key, $value, $cache::TTL_DAY
);