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
25 * @defgroup Upload Upload related
31 * UploadBase and subclasses are the backend of MediaWiki's file uploads.
32 * The frontends are formed by ApiUpload and SpecialUpload.
34 * @author Brion Vibber
35 * @author Bryan Tong Minh
36 * @author Michael Dale
38 abstract class UploadBase
{
39 /** @var string Local file system path to the file to upload (or a local copy) */
41 /** @var TempFSFile|null Wrapper to handle deleting the temp file */
42 protected $tempFileObj;
44 protected $mDesiredDestName, $mDestName, $mRemoveTempFile, $mSourceType;
45 protected $mTitle = false, $mTitleError = 0;
46 protected $mFilteredName, $mFinalExtension;
47 protected $mLocalFile, $mFileSize, $mFileProps;
48 protected $mBlackListedExtensions;
49 protected $mJavaDetected, $mSVGNSError;
51 protected static $safeXmlEncodings = [
62 const MIN_LENGTH_PARTNAME
= 4;
63 const ILLEGAL_FILENAME
= 5;
64 const OVERWRITE_EXISTING_FILE
= 7; # Not used anymore; handled by verifyTitlePermissions()
65 const FILETYPE_MISSING
= 8;
66 const FILETYPE_BADTYPE
= 9;
67 const VERIFICATION_ERROR
= 10;
68 const HOOK_ABORTED
= 11;
69 const FILE_TOO_LARGE
= 12;
70 const WINDOWS_NONASCII_FILENAME
= 13;
71 const FILENAME_TOO_LONG
= 14;
77 public function getVerificationErrorCode( $error ) {
79 self
::EMPTY_FILE
=> 'empty-file',
80 self
::FILE_TOO_LARGE
=> 'file-too-large',
81 self
::FILETYPE_MISSING
=> 'filetype-missing',
82 self
::FILETYPE_BADTYPE
=> 'filetype-banned',
83 self
::MIN_LENGTH_PARTNAME
=> 'filename-tooshort',
84 self
::ILLEGAL_FILENAME
=> 'illegal-filename',
85 self
::OVERWRITE_EXISTING_FILE
=> 'overwrite',
86 self
::VERIFICATION_ERROR
=> 'verification-error',
87 self
::HOOK_ABORTED
=> 'hookaborted',
88 self
::WINDOWS_NONASCII_FILENAME
=> 'windows-nonascii-filename',
89 self
::FILENAME_TOO_LONG
=> 'filename-toolong',
91 if ( isset( $code_to_status[$error] ) ) {
92 return $code_to_status[$error];
95 return 'unknown-error';
99 * Returns true if uploads are enabled.
100 * Can be override by subclasses.
103 public static function isEnabled() {
104 global $wgEnableUploads;
106 if ( !$wgEnableUploads ) {
110 # Check php's file_uploads setting
111 return wfIsHHVM() ||
wfIniGetBool( 'file_uploads' );
115 * Returns true if the user can use this upload module or else a string
116 * identifying the missing permission.
117 * Can be overridden by subclasses.
120 * @return bool|string
122 public static function isAllowed( $user ) {
123 foreach ( [ 'upload', 'edit' ] as $permission ) {
124 if ( !$user->isAllowed( $permission ) ) {
133 * Returns true if the user has surpassed the upload rate limit, false otherwise.
138 public static function isThrottled( $user ) {
139 return $user->pingLimiter( 'upload' );
142 // Upload handlers. Should probably just be a global.
143 private static $uploadHandlers = [ 'Stash', 'File', 'Url' ];
146 * Create a form of UploadBase depending on wpSourceType and initializes it
148 * @param WebRequest $request
149 * @param string|null $type
150 * @return null|UploadBase
152 public static function createFromRequest( &$request, $type = null ) {
153 $type = $type ?
$type : $request->getVal( 'wpSourceType', 'File' );
159 // Get the upload class
160 $type = ucfirst( $type );
162 // Give hooks the chance to handle this request
164 Hooks
::run( 'UploadCreateFromRequest', [ $type, &$className ] );
165 if ( is_null( $className ) ) {
166 $className = 'UploadFrom' . $type;
167 wfDebug( __METHOD__
. ": class name: $className\n" );
168 if ( !in_array( $type, self
::$uploadHandlers ) ) {
173 // Check whether this upload class is enabled
174 if ( !call_user_func( [ $className, 'isEnabled' ] ) ) {
178 // Check whether the request is valid
179 if ( !call_user_func( [ $className, 'isValidRequest' ], $request ) ) {
183 /** @var UploadBase $handler */
184 $handler = new $className;
186 $handler->initializeFromRequest( $request );
192 * Check whether a request if valid for this handler
193 * @param WebRequest $request
196 public static function isValidRequest( $request ) {
200 public function __construct() {
204 * Returns the upload type. Should be overridden by child classes
209 public function getSourceType() {
214 * Initialize the path information
215 * @param string $name The desired destination name
216 * @param string $tempPath The temporary path
217 * @param int $fileSize The file size
218 * @param bool $removeTempFile (false) remove the temporary file?
219 * @throws MWException
221 public function initializePathInfo( $name, $tempPath, $fileSize, $removeTempFile = false ) {
222 $this->mDesiredDestName
= $name;
223 if ( FileBackend
::isStoragePath( $tempPath ) ) {
224 throw new MWException( __METHOD__
. " given storage path `$tempPath`." );
227 $this->setTempFile( $tempPath, $fileSize );
228 $this->mRemoveTempFile
= $removeTempFile;
232 * Initialize from a WebRequest. Override this in a subclass.
234 * @param WebRequest $request
236 abstract public function initializeFromRequest( &$request );
239 * @param string $tempPath File system path to temporary file containing the upload
240 * @param integer $fileSize
242 protected function setTempFile( $tempPath, $fileSize = null ) {
243 $this->mTempPath
= $tempPath;
244 $this->mFileSize
= $fileSize ?
: null;
245 if ( strlen( $this->mTempPath
) && file_exists( $this->mTempPath
) ) {
246 $this->tempFileObj
= new TempFSFile( $this->mTempPath
);
248 $this->mFileSize
= filesize( $this->mTempPath
);
251 $this->tempFileObj
= null;
256 * Fetch the file. Usually a no-op
259 public function fetchFile() {
260 return Status
::newGood();
264 * Return true if the file is empty
267 public function isEmptyFile() {
268 return empty( $this->mFileSize
);
272 * Return the file size
275 public function getFileSize() {
276 return $this->mFileSize
;
280 * Get the base 36 SHA1 of the file
283 public function getTempFileSha1Base36() {
284 return FSFile
::getSha1Base36FromPath( $this->mTempPath
);
288 * @param string $srcPath The source path
289 * @return string|bool The real path if it was a virtual URL Returns false on failure
291 function getRealPath( $srcPath ) {
292 $repo = RepoGroup
::singleton()->getLocalRepo();
293 if ( $repo->isVirtualUrl( $srcPath ) ) {
294 /** @todo Just make uploads work with storage paths UploadFromStash
295 * loads files via virtual URLs.
297 $tmpFile = $repo->getLocalCopy( $srcPath );
299 $tmpFile->bind( $this ); // keep alive with $this
301 $path = $tmpFile ?
$tmpFile->getPath() : false;
310 * Verify whether the upload is sane.
311 * @return mixed Const self::OK or else an array with error information
313 public function verifyUpload() {
316 * If there was no filename or a zero size given, give up quick.
318 if ( $this->isEmptyFile() ) {
319 return [ 'status' => self
::EMPTY_FILE
];
323 * Honor $wgMaxUploadSize
325 $maxSize = self
::getMaxUploadSize( $this->getSourceType() );
326 if ( $this->mFileSize
> $maxSize ) {
328 'status' => self
::FILE_TOO_LARGE
,
334 * Look at the contents of the file; if we can recognize the
335 * type but it's corrupt or data of the wrong type, we should
336 * probably not accept it.
338 $verification = $this->verifyFile();
339 if ( $verification !== true ) {
341 'status' => self
::VERIFICATION_ERROR
,
342 'details' => $verification
347 * Make sure this file can be created
349 $result = $this->validateName();
350 if ( $result !== true ) {
355 if ( !Hooks
::run( 'UploadVerification',
356 [ $this->mDestName
, $this->mTempPath
, &$error ], '1.28' )
358 return [ 'status' => self
::HOOK_ABORTED
, 'error' => $error ];
361 return [ 'status' => self
::OK
];
365 * Verify that the name is valid and, if necessary, that we can overwrite
367 * @return mixed True if valid, otherwise and array with 'status'
370 public function validateName() {
371 $nt = $this->getTitle();
372 if ( is_null( $nt ) ) {
373 $result = [ 'status' => $this->mTitleError
];
374 if ( $this->mTitleError
== self
::ILLEGAL_FILENAME
) {
375 $result['filtered'] = $this->mFilteredName
;
377 if ( $this->mTitleError
== self
::FILETYPE_BADTYPE
) {
378 $result['finalExt'] = $this->mFinalExtension
;
379 if ( count( $this->mBlackListedExtensions
) ) {
380 $result['blacklistedExt'] = $this->mBlackListedExtensions
;
386 $this->mDestName
= $this->getLocalFile()->getName();
392 * Verify the MIME type.
394 * @note Only checks that it is not an evil MIME. The "does it have
395 * correct extension given its MIME type?" check is in verifyFile.
396 * in `verifyFile()` that MIME type and file extension correlate.
397 * @param string $mime Representing the MIME
398 * @return mixed True if the file is verified, an array otherwise
400 protected function verifyMimeType( $mime ) {
401 global $wgVerifyMimeType;
402 if ( $wgVerifyMimeType ) {
403 wfDebug( "mime: <$mime> extension: <{$this->mFinalExtension}>\n" );
404 global $wgMimeTypeBlacklist;
405 if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
406 return [ 'filetype-badmime', $mime ];
409 # Check what Internet Explorer would detect
410 $fp = fopen( $this->mTempPath
, 'rb' );
411 $chunk = fread( $fp, 256 );
414 $magic = MimeMagic
::singleton();
415 $extMime = $magic->guessTypesForExtension( $this->mFinalExtension
);
416 $ieTypes = $magic->getIEMimeTypes( $this->mTempPath
, $chunk, $extMime );
417 foreach ( $ieTypes as $ieType ) {
418 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
419 return [ 'filetype-bad-ie-mime', $ieType ];
428 * Verifies that it's ok to include the uploaded file
430 * @return mixed True of the file is verified, array otherwise.
432 protected function verifyFile() {
433 global $wgVerifyMimeType, $wgDisableUploadScriptChecks;
435 $status = $this->verifyPartialFile();
436 if ( $status !== true ) {
440 $this->mFileProps
= FSFile
::getPropsFromPath( $this->mTempPath
, $this->mFinalExtension
);
441 $mime = $this->mFileProps
['mime'];
443 if ( $wgVerifyMimeType ) {
444 # XXX: Missing extension will be caught by validateName() via getTitle()
445 if ( $this->mFinalExtension
!= '' && !$this->verifyExtension( $mime, $this->mFinalExtension
) ) {
446 return [ 'filetype-mime-mismatch', $this->mFinalExtension
, $mime ];
450 # check for htmlish code and javascript
451 if ( !$wgDisableUploadScriptChecks ) {
452 if ( $this->mFinalExtension
== 'svg' ||
$mime == 'image/svg+xml' ) {
453 $svgStatus = $this->detectScriptInSvg( $this->mTempPath
, false );
454 if ( $svgStatus !== false ) {
460 $handler = MediaHandler
::getHandler( $mime );
462 $handlerStatus = $handler->verifyUpload( $this->mTempPath
);
463 if ( !$handlerStatus->isOK() ) {
464 $errors = $handlerStatus->getErrorsArray();
466 return reset( $errors );
471 Hooks
::run( 'UploadVerifyFile', [ $this, $mime, &$error ] );
472 if ( $error !== true ) {
473 if ( !is_array( $error ) ) {
479 wfDebug( __METHOD__
. ": all clear; passing.\n" );
485 * A verification routine suitable for partial files
487 * Runs the blacklist checks, but not any checks that may
488 * assume the entire file is present.
490 * @return mixed True for valid or array with error message key.
492 protected function verifyPartialFile() {
493 global $wgAllowJavaUploads, $wgDisableUploadScriptChecks;
495 # getTitle() sets some internal parameters like $this->mFinalExtension
498 $this->mFileProps
= FSFile
::getPropsFromPath( $this->mTempPath
, $this->mFinalExtension
);
500 # check MIME type, if desired
501 $mime = $this->mFileProps
['file-mime'];
502 $status = $this->verifyMimeType( $mime );
503 if ( $status !== true ) {
507 # check for htmlish code and javascript
508 if ( !$wgDisableUploadScriptChecks ) {
509 if ( self
::detectScript( $this->mTempPath
, $mime, $this->mFinalExtension
) ) {
510 return [ 'uploadscripted' ];
512 if ( $this->mFinalExtension
== 'svg' ||
$mime == 'image/svg+xml' ) {
513 $svgStatus = $this->detectScriptInSvg( $this->mTempPath
, true );
514 if ( $svgStatus !== false ) {
520 # Check for Java applets, which if uploaded can bypass cross-site
522 if ( !$wgAllowJavaUploads ) {
523 $this->mJavaDetected
= false;
524 $zipStatus = ZipDirectoryReader
::read( $this->mTempPath
,
525 [ $this, 'zipEntryCallback' ] );
526 if ( !$zipStatus->isOK() ) {
527 $errors = $zipStatus->getErrorsArray();
528 $error = reset( $errors );
529 if ( $error[0] !== 'zip-wrong-format' ) {
533 if ( $this->mJavaDetected
) {
534 return [ 'uploadjava' ];
538 # Scan the uploaded file for viruses
539 $virus = $this->detectVirus( $this->mTempPath
);
541 return [ 'uploadvirus', $virus ];
548 * Callback for ZipDirectoryReader to detect Java class files.
550 * @param array $entry
552 function zipEntryCallback( $entry ) {
553 $names = [ $entry['name'] ];
555 // If there is a null character, cut off the name at it, because JDK's
556 // ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name
557 // were constructed which had ".class\0" followed by a string chosen to
558 // make the hash collide with the truncated name, that file could be
559 // returned in response to a request for the .class file.
560 $nullPos = strpos( $entry['name'], "\000" );
561 if ( $nullPos !== false ) {
562 $names[] = substr( $entry['name'], 0, $nullPos );
565 // If there is a trailing slash in the file name, we have to strip it,
566 // because that's what ZIP_GetEntry() does.
567 if ( preg_grep( '!\.class/?$!', $names ) ) {
568 $this->mJavaDetected
= true;
573 * Alias for verifyTitlePermissions. The function was originally
574 * 'verifyPermissions', but that suggests it's checking the user, when it's
575 * really checking the title + user combination.
577 * @param User $user User object to verify the permissions against
578 * @return mixed An array as returned by getUserPermissionsErrors or true
579 * in case the user has proper permissions.
581 public function verifyPermissions( $user ) {
582 return $this->verifyTitlePermissions( $user );
586 * Check whether the user can edit, upload and create the image. This
587 * checks only against the current title; if it returns errors, it may
588 * very well be that another title will not give errors. Therefore
589 * isAllowed() should be called as well for generic is-user-blocked or
590 * can-user-upload checking.
592 * @param User $user User object to verify the permissions against
593 * @return mixed An array as returned by getUserPermissionsErrors or true
594 * in case the user has proper permissions.
596 public function verifyTitlePermissions( $user ) {
598 * If the image is protected, non-sysop users won't be able
599 * to modify it by uploading a new revision.
601 $nt = $this->getTitle();
602 if ( is_null( $nt ) ) {
605 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
606 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
607 if ( !$nt->exists() ) {
608 $permErrorsCreate = $nt->getUserPermissionsErrors( 'create', $user );
610 $permErrorsCreate = [];
612 if ( $permErrors ||
$permErrorsUpload ||
$permErrorsCreate ) {
613 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
614 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
619 $overwriteError = $this->checkOverwrite( $user );
620 if ( $overwriteError !== true ) {
621 return [ $overwriteError ];
628 * Check for non fatal problems with the file.
630 * This should not assume that mTempPath is set.
632 * @return array Array of warnings
634 public function checkWarnings() {
639 $localFile = $this->getLocalFile();
640 $localFile->load( File
::READ_LATEST
);
641 $filename = $localFile->getName();
644 * Check whether the resulting filename is different from the desired one,
645 * but ignore things like ucfirst() and spaces/underscore things
647 $comparableName = str_replace( ' ', '_', $this->mDesiredDestName
);
648 $comparableName = Title
::capitalize( $comparableName, NS_FILE
);
650 if ( $this->mDesiredDestName
!= $filename && $comparableName != $filename ) {
651 $warnings['badfilename'] = $filename;
654 // Check whether the file extension is on the unwanted list
655 global $wgCheckFileExtensions, $wgFileExtensions;
656 if ( $wgCheckFileExtensions ) {
657 $extensions = array_unique( $wgFileExtensions );
658 if ( !$this->checkFileExtension( $this->mFinalExtension
, $extensions ) ) {
659 $warnings['filetype-unwanted-type'] = [ $this->mFinalExtension
,
660 $wgLang->commaList( $extensions ), count( $extensions ) ];
664 global $wgUploadSizeWarning;
665 if ( $wgUploadSizeWarning && ( $this->mFileSize
> $wgUploadSizeWarning ) ) {
666 $warnings['large-file'] = [ $wgUploadSizeWarning, $this->mFileSize
];
669 if ( $this->mFileSize
== 0 ) {
670 $warnings['empty-file'] = true;
673 $exists = self
::getExistsWarning( $localFile );
674 if ( $exists !== false ) {
675 $warnings['exists'] = $exists;
678 if ( $localFile->wasDeleted() && !$localFile->exists() ) {
679 $warnings['was-deleted'] = $filename;
682 // Check dupes against existing files
683 $hash = $this->getTempFileSha1Base36();
684 $dupes = RepoGroup
::singleton()->findBySha1( $hash );
685 $title = $this->getTitle();
686 // Remove all matches against self
687 foreach ( $dupes as $key => $dupe ) {
688 if ( $title->equals( $dupe->getTitle() ) ) {
689 unset( $dupes[$key] );
693 $warnings['duplicate'] = $dupes;
696 // Check dupes against archives
697 $archivedFile = new ArchivedFile( null, 0, '', $hash );
698 if ( $archivedFile->getID() > 0 ) {
699 if ( $archivedFile->userCan( File
::DELETED_FILE
) ) {
700 $warnings['duplicate-archive'] = $archivedFile->getName();
702 $warnings['duplicate-archive'] = '';
710 * Really perform the upload. Stores the file in the local repo, watches
711 * if necessary and runs the UploadComplete hook.
713 * @param string $comment
714 * @param string $pageText
715 * @param bool $watch Whether the file page should be added to user's watchlist.
716 * (This doesn't check $user's permissions.)
718 * @param string[] $tags Change tags to add to the log entry and page revision.
719 * (This doesn't check $user's permissions.)
720 * @return Status Indicating the whether the upload succeeded.
722 public function performUpload( $comment, $pageText, $watch, $user, $tags = [] ) {
723 $this->getLocalFile()->load( File
::READ_LATEST
);
724 $props = $this->mFileProps
;
727 Hooks
::run( 'UploadVerifyUpload', [ $this, $user, $props, $comment, $pageText, &$error ] );
729 if ( !is_array( $error ) ) {
732 return call_user_func_array( 'Status::newFatal', $error );
735 $status = $this->getLocalFile()->upload(
746 if ( $status->isGood() ) {
748 WatchAction
::doWatch(
749 $this->getLocalFile()->getTitle(),
751 User
::IGNORE_USER_RIGHTS
754 Hooks
::run( 'UploadComplete', [ &$this ] );
756 $this->postProcessUpload();
763 * Perform extra steps after a successful upload.
767 public function postProcessUpload() {
768 global $wgUploadThumbnailRenderMap;
772 $sizes = $wgUploadThumbnailRenderMap;
775 $file = $this->getLocalFile();
777 foreach ( $sizes as $size ) {
778 if ( $file->isVectorized() ||
$file->getWidth() > $size ) {
779 $jobs[] = new ThumbnailRenderJob(
781 [ 'transformParams' => [ 'width' => $size ] ]
787 JobQueueGroup
::singleton()->push( $jobs );
792 * Returns the title of the file to be uploaded. Sets mTitleError in case
793 * the name was illegal.
795 * @return Title The title of the file or null in case the name was illegal
797 public function getTitle() {
798 if ( $this->mTitle
!== false ) {
799 return $this->mTitle
;
801 if ( !is_string( $this->mDesiredDestName
) ) {
802 $this->mTitleError
= self
::ILLEGAL_FILENAME
;
803 $this->mTitle
= null;
805 return $this->mTitle
;
807 /* Assume that if a user specified File:Something.jpg, this is an error
808 * and that the namespace prefix needs to be stripped of.
810 $title = Title
::newFromText( $this->mDesiredDestName
);
811 if ( $title && $title->getNamespace() == NS_FILE
) {
812 $this->mFilteredName
= $title->getDBkey();
814 $this->mFilteredName
= $this->mDesiredDestName
;
817 # oi_archive_name is max 255 bytes, which include a timestamp and an
818 # exclamation mark, so restrict file name to 240 bytes.
819 if ( strlen( $this->mFilteredName
) > 240 ) {
820 $this->mTitleError
= self
::FILENAME_TOO_LONG
;
821 $this->mTitle
= null;
823 return $this->mTitle
;
827 * Chop off any directories in the given filename. Then
828 * filter out illegal characters, and try to make a legible name
829 * out of it. We'll strip some silently that Title would die on.
831 $this->mFilteredName
= wfStripIllegalFilenameChars( $this->mFilteredName
);
832 /* Normalize to title form before we do any further processing */
833 $nt = Title
::makeTitleSafe( NS_FILE
, $this->mFilteredName
);
834 if ( is_null( $nt ) ) {
835 $this->mTitleError
= self
::ILLEGAL_FILENAME
;
836 $this->mTitle
= null;
838 return $this->mTitle
;
840 $this->mFilteredName
= $nt->getDBkey();
843 * We'll want to blacklist against *any* 'extension', and use
844 * only the final one for the whitelist.
846 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName
);
848 if ( count( $ext ) ) {
849 $this->mFinalExtension
= trim( $ext[count( $ext ) - 1] );
851 $this->mFinalExtension
= '';
853 # No extension, try guessing one
854 $magic = MimeMagic
::singleton();
855 $mime = $magic->guessMimeType( $this->mTempPath
);
856 if ( $mime !== 'unknown/unknown' ) {
857 # Get a space separated list of extensions
858 $extList = $magic->getExtensionsForType( $mime );
860 # Set the extension to the canonical extension
861 $this->mFinalExtension
= strtok( $extList, ' ' );
863 # Fix up the other variables
864 $this->mFilteredName
.= ".{$this->mFinalExtension}";
865 $nt = Title
::makeTitleSafe( NS_FILE
, $this->mFilteredName
);
866 $ext = [ $this->mFinalExtension
];
871 /* Don't allow users to override the blacklist (check file extension) */
872 global $wgCheckFileExtensions, $wgStrictFileExtensions;
873 global $wgFileExtensions, $wgFileBlacklist;
875 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
877 if ( $this->mFinalExtension
== '' ) {
878 $this->mTitleError
= self
::FILETYPE_MISSING
;
879 $this->mTitle
= null;
881 return $this->mTitle
;
882 } elseif ( $blackListedExtensions ||
883 ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
884 !$this->checkFileExtension( $this->mFinalExtension
, $wgFileExtensions ) )
886 $this->mBlackListedExtensions
= $blackListedExtensions;
887 $this->mTitleError
= self
::FILETYPE_BADTYPE
;
888 $this->mTitle
= null;
890 return $this->mTitle
;
893 // Windows may be broken with special characters, see bug 1780
894 if ( !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() )
895 && !RepoGroup
::singleton()->getLocalRepo()->backendSupportsUnicodePaths()
897 $this->mTitleError
= self
::WINDOWS_NONASCII_FILENAME
;
898 $this->mTitle
= null;
900 return $this->mTitle
;
903 # If there was more than one "extension", reassemble the base
904 # filename to prevent bogus complaints about length
905 if ( count( $ext ) > 1 ) {
906 $iterations = count( $ext ) - 1;
907 for ( $i = 0; $i < $iterations; $i++
) {
908 $partname .= '.' . $ext[$i];
912 if ( strlen( $partname ) < 1 ) {
913 $this->mTitleError
= self
::MIN_LENGTH_PARTNAME
;
914 $this->mTitle
= null;
916 return $this->mTitle
;
921 return $this->mTitle
;
925 * Return the local file and initializes if necessary.
927 * @return LocalFile|UploadStashFile|null
929 public function getLocalFile() {
930 if ( is_null( $this->mLocalFile
) ) {
931 $nt = $this->getTitle();
932 $this->mLocalFile
= is_null( $nt ) ?
null : wfLocalFile( $nt );
935 return $this->mLocalFile
;
939 * If the user does not supply all necessary information in the first upload
940 * form submission (either by accident or by design) then we may want to
941 * stash the file temporarily, get more information, and publish the file
944 * This method will stash a file in a temporary directory for later
945 * processing, and save the necessary descriptive info into the database.
946 * This method returns the file object, which also has a 'fileKey' property
947 * which can be passed through a form or API request to find this stashed
951 * @return UploadStashFile Stashed file
953 public function stashFile( User
$user = null ) {
954 // was stashSessionFile
956 $stash = RepoGroup
::singleton()->getLocalRepo()->getUploadStash( $user );
957 $file = $stash->stashFile( $this->mTempPath
, $this->getSourceType() );
958 $this->mLocalFile
= $file;
964 * Stash a file in a temporary directory, returning a key which can be used
965 * to find the file again. See stashFile().
967 * @return string File key
969 public function stashFileGetKey() {
970 return $this->stashFile()->getFileKey();
974 * alias for stashFileGetKey, for backwards compatibility
976 * @return string File key
978 public function stashSession() {
979 return $this->stashFileGetKey();
983 * If we've modified the upload file we need to manually remove it
984 * on exit to clean up.
986 public function cleanupTempFile() {
987 if ( $this->mRemoveTempFile
&& $this->tempFileObj
) {
988 // Delete when all relevant TempFSFile handles go out of scope
989 wfDebug( __METHOD__
. ": Marked temporary file '{$this->mTempPath}' for removal\n" );
990 $this->tempFileObj
->autocollect();
994 public function getTempPath() {
995 return $this->mTempPath
;
999 * Split a file into a base name and all dot-delimited 'extensions'
1000 * on the end. Some web server configurations will fall back to
1001 * earlier pseudo-'extensions' to determine type and execute
1002 * scripts, so the blacklist needs to check them all.
1004 * @param string $filename
1007 public static function splitExtensions( $filename ) {
1008 $bits = explode( '.', $filename );
1009 $basename = array_shift( $bits );
1011 return [ $basename, $bits ];
1015 * Perform case-insensitive match against a list of file extensions.
1016 * Returns true if the extension is in the list.
1018 * @param string $ext
1019 * @param array $list
1022 public static function checkFileExtension( $ext, $list ) {
1023 return in_array( strtolower( $ext ), $list );
1027 * Perform case-insensitive match against a list of file extensions.
1028 * Returns an array of matching extensions.
1031 * @param array $list
1034 public static function checkFileExtensionList( $ext, $list ) {
1035 return array_intersect( array_map( 'strtolower', $ext ), $list );
1039 * Checks if the MIME type of the uploaded file matches the file extension.
1041 * @param string $mime The MIME type of the uploaded file
1042 * @param string $extension The filename extension that the file is to be served with
1045 public static function verifyExtension( $mime, $extension ) {
1046 $magic = MimeMagic
::singleton();
1048 if ( !$mime ||
$mime == 'unknown' ||
$mime == 'unknown/unknown' ) {
1049 if ( !$magic->isRecognizableExtension( $extension ) ) {
1050 wfDebug( __METHOD__
. ": passing file with unknown detected mime type; " .
1051 "unrecognized extension '$extension', can't verify\n" );
1055 wfDebug( __METHOD__
. ": rejecting file with unknown detected mime type; " .
1056 "recognized extension '$extension', so probably invalid file\n" );
1062 $match = $magic->isMatchingExtension( $extension, $mime );
1064 if ( $match === null ) {
1065 if ( $magic->getTypesForExtension( $extension ) !== null ) {
1066 wfDebug( __METHOD__
. ": No extension known for $mime, but we know a mime for $extension\n" );
1070 wfDebug( __METHOD__
. ": no file extension known for mime type $mime, passing file\n" );
1074 } elseif ( $match === true ) {
1075 wfDebug( __METHOD__
. ": mime type $mime matches extension $extension, passing file\n" );
1077 /** @todo If it's a bitmap, make sure PHP or ImageMagick resp. can handle it! */
1081 . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
1088 * Heuristic for detecting files that *could* contain JavaScript instructions or
1089 * things that may look like HTML to a browser and are thus
1090 * potentially harmful. The present implementation will produce false
1091 * positives in some situations.
1093 * @param string $file Pathname to the temporary upload file
1094 * @param string $mime The MIME type of the file
1095 * @param string $extension The extension of the file
1096 * @return bool True if the file contains something looking like embedded scripts
1098 public static function detectScript( $file, $mime, $extension ) {
1099 global $wgAllowTitlesInSVG;
1101 # ugly hack: for text files, always look at the entire file.
1102 # For binary field, just check the first K.
1104 if ( strpos( $mime, 'text/' ) === 0 ) {
1105 $chunk = file_get_contents( $file );
1107 $fp = fopen( $file, 'rb' );
1108 $chunk = fread( $fp, 1024 );
1112 $chunk = strtolower( $chunk );
1118 # decode from UTF-16 if needed (could be used for obfuscation).
1119 if ( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
1121 } elseif ( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
1128 $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
1131 $chunk = trim( $chunk );
1133 /** @todo FIXME: Convert from UTF-16 if necessary! */
1134 wfDebug( __METHOD__
. ": checking for embedded scripts and HTML stuff\n" );
1136 # check for HTML doctype
1137 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
1141 // Some browsers will interpret obscure xml encodings as UTF-8, while
1142 // PHP/expat will interpret the given encoding in the xml declaration (bug 47304)
1143 if ( $extension == 'svg' ||
strpos( $mime, 'image/svg' ) === 0 ) {
1144 if ( self
::checkXMLEncodingMissmatch( $file ) ) {
1150 * Internet Explorer for Windows performs some really stupid file type
1151 * autodetection which can cause it to interpret valid image files as HTML
1152 * and potentially execute JavaScript, creating a cross-site scripting
1155 * Apple's Safari browser also performs some unsafe file type autodetection
1156 * which can cause legitimate files to be interpreted as HTML if the
1157 * web server is not correctly configured to send the right content-type
1158 * (or if you're really uploading plain text and octet streams!)
1160 * Returns true if IE is likely to mistake the given file for HTML.
1161 * Also returns true if Safari would mistake the given file for HTML
1162 * when served with a generic content-type.
1168 '<html', # also in safari
1171 '<script', # also in safari
1175 if ( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
1179 foreach ( $tags as $tag ) {
1180 if ( false !== strpos( $chunk, $tag ) ) {
1181 wfDebug( __METHOD__
. ": found something that may make it be mistaken for html: $tag\n" );
1188 * look for JavaScript
1191 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
1192 $chunk = Sanitizer
::decodeCharReferences( $chunk );
1194 # look for script-types
1195 if ( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
1196 wfDebug( __METHOD__
. ": found script types\n" );
1201 # look for html-style script-urls
1202 if ( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1203 wfDebug( __METHOD__
. ": found html-style script urls\n" );
1208 # look for css-style script-urls
1209 if ( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1210 wfDebug( __METHOD__
. ": found css-style script urls\n" );
1215 wfDebug( __METHOD__
. ": no scripts found\n" );
1221 * Check a whitelist of xml encodings that are known not to be interpreted differently
1222 * by the server's xml parser (expat) and some common browsers.
1224 * @param string $file Pathname to the temporary upload file
1225 * @return bool True if the file contains an encoding that could be misinterpreted
1227 public static function checkXMLEncodingMissmatch( $file ) {
1228 global $wgSVGMetadataCutoff;
1229 $contents = file_get_contents( $file, false, null, -1, $wgSVGMetadataCutoff );
1230 $encodingRegex = '!encoding[ \t\n\r]*=[ \t\n\r]*[\'"](.*?)[\'"]!si';
1232 if ( preg_match( "!<\?xml\b(.*?)\?>!si", $contents, $matches ) ) {
1233 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1234 && !in_array( strtoupper( $encMatch[1] ), self
::$safeXmlEncodings )
1236 wfDebug( __METHOD__
. ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1240 } elseif ( preg_match( "!<\?xml\b!si", $contents ) ) {
1241 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1242 // bytes. There shouldn't be a legitimate reason for this to happen.
1243 wfDebug( __METHOD__
. ": Unmatched XML declaration start\n" );
1246 } elseif ( substr( $contents, 0, 4 ) == "\x4C\x6F\xA7\x94" ) {
1247 // EBCDIC encoded XML
1248 wfDebug( __METHOD__
. ": EBCDIC Encoded XML\n" );
1253 // It's possible the file is encoded with multi-byte encoding, so re-encode attempt to
1254 // detect the encoding in case is specifies an encoding not whitelisted in self::$safeXmlEncodings
1255 $attemptEncodings = [ 'UTF-16', 'UTF-16BE', 'UTF-32', 'UTF-32BE' ];
1256 foreach ( $attemptEncodings as $encoding ) {
1257 MediaWiki\
suppressWarnings();
1258 $str = iconv( $encoding, 'UTF-8', $contents );
1259 MediaWiki\restoreWarnings
();
1260 if ( $str != '' && preg_match( "!<\?xml\b(.*?)\?>!si", $str, $matches ) ) {
1261 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1262 && !in_array( strtoupper( $encMatch[1] ), self
::$safeXmlEncodings )
1264 wfDebug( __METHOD__
. ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1268 } elseif ( $str != '' && preg_match( "!<\?xml\b!si", $str ) ) {
1269 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1270 // bytes. There shouldn't be a legitimate reason for this to happen.
1271 wfDebug( __METHOD__
. ": Unmatched XML declaration start\n" );
1281 * @param string $filename
1282 * @param bool $partial
1283 * @return mixed False of the file is verified (does not contain scripts), array otherwise.
1285 protected function detectScriptInSvg( $filename, $partial ) {
1286 $this->mSVGNSError
= false;
1287 $check = new XmlTypeCheck(
1289 [ $this, 'checkSvgScriptCallback' ],
1291 [ 'processing_instruction_handler' => 'UploadBase::checkSvgPICallback' ]
1293 if ( $check->wellFormed
!== true ) {
1294 // Invalid xml (bug 58553)
1295 // But only when non-partial (bug 65724)
1296 return $partial ?
false : [ 'uploadinvalidxml' ];
1297 } elseif ( $check->filterMatch
) {
1298 if ( $this->mSVGNSError
) {
1299 return [ 'uploadscriptednamespace', $this->mSVGNSError
];
1302 return $check->filterMatchType
;
1309 * Callback to filter SVG Processing Instructions.
1310 * @param string $target Processing instruction name
1311 * @param string $data Processing instruction attribute and value
1312 * @return bool (true if the filter identified something bad)
1314 public static function checkSvgPICallback( $target, $data ) {
1315 // Don't allow external stylesheets (bug 57550)
1316 if ( preg_match( '/xml-stylesheet/i', $target ) ) {
1317 return [ 'upload-scripted-pi-callback' ];
1324 * @todo Replace this with a whitelist filter!
1325 * @param string $element
1326 * @param array $attribs
1329 public function checkSvgScriptCallback( $element, $attribs, $data = null ) {
1331 list( $namespace, $strippedElement ) = $this->splitXmlNamespace( $element );
1333 // We specifically don't include:
1334 // http://www.w3.org/1999/xhtml (bug 60771)
1335 static $validNamespaces = [
1338 'http://creativecommons.org/ns#',
1339 'http://inkscape.sourceforge.net/dtd/sodipodi-0.dtd',
1340 'http://ns.adobe.com/adobeillustrator/10.0/',
1341 'http://ns.adobe.com/adobesvgviewerextensions/3.0/',
1342 'http://ns.adobe.com/extensibility/1.0/',
1343 'http://ns.adobe.com/flows/1.0/',
1344 'http://ns.adobe.com/illustrator/1.0/',
1345 'http://ns.adobe.com/imagereplacement/1.0/',
1346 'http://ns.adobe.com/pdf/1.3/',
1347 'http://ns.adobe.com/photoshop/1.0/',
1348 'http://ns.adobe.com/saveforweb/1.0/',
1349 'http://ns.adobe.com/variables/1.0/',
1350 'http://ns.adobe.com/xap/1.0/',
1351 'http://ns.adobe.com/xap/1.0/g/',
1352 'http://ns.adobe.com/xap/1.0/g/img/',
1353 'http://ns.adobe.com/xap/1.0/mm/',
1354 'http://ns.adobe.com/xap/1.0/rights/',
1355 'http://ns.adobe.com/xap/1.0/stype/dimensions#',
1356 'http://ns.adobe.com/xap/1.0/stype/font#',
1357 'http://ns.adobe.com/xap/1.0/stype/manifestitem#',
1358 'http://ns.adobe.com/xap/1.0/stype/resourceevent#',
1359 'http://ns.adobe.com/xap/1.0/stype/resourceref#',
1360 'http://ns.adobe.com/xap/1.0/t/pg/',
1361 'http://purl.org/dc/elements/1.1/',
1362 'http://purl.org/dc/elements/1.1',
1363 'http://schemas.microsoft.com/visio/2003/svgextensions/',
1364 'http://sodipodi.sourceforge.net/dtd/sodipodi-0.dtd',
1365 'http://taptrix.com/inkpad/svg_extensions',
1366 'http://web.resource.org/cc/',
1367 'http://www.freesoftware.fsf.org/bkchem/cdml',
1368 'http://www.inkscape.org/namespaces/inkscape',
1369 'http://www.opengis.net/gml',
1370 'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
1371 'http://www.w3.org/2000/svg',
1372 'http://www.w3.org/tr/rec-rdf-syntax/',
1375 if ( !in_array( $namespace, $validNamespaces ) ) {
1376 wfDebug( __METHOD__
. ": Non-svg namespace '$namespace' in uploaded file.\n" );
1377 /** @todo Return a status object to a closure in XmlTypeCheck, for MW1.21+ */
1378 $this->mSVGNSError
= $namespace;
1384 * check for elements that can contain javascript
1386 if ( $strippedElement == 'script' ) {
1387 wfDebug( __METHOD__
. ": Found script element '$element' in uploaded file.\n" );
1389 return [ 'uploaded-script-svg', $strippedElement ];
1392 # e.g., <svg xmlns="http://www.w3.org/2000/svg">
1393 # <handler xmlns:ev="http://www.w3.org/2001/xml-events" ev:event="load">alert(1)</handler> </svg>
1394 if ( $strippedElement == 'handler' ) {
1395 wfDebug( __METHOD__
. ": Found scriptable element '$element' in uploaded file.\n" );
1397 return [ 'uploaded-script-svg', $strippedElement ];
1400 # SVG reported in Feb '12 that used xml:stylesheet to generate javascript block
1401 if ( $strippedElement == 'stylesheet' ) {
1402 wfDebug( __METHOD__
. ": Found scriptable element '$element' in uploaded file.\n" );
1404 return [ 'uploaded-script-svg', $strippedElement ];
1407 # Block iframes, in case they pass the namespace check
1408 if ( $strippedElement == 'iframe' ) {
1409 wfDebug( __METHOD__
. ": iframe in uploaded file.\n" );
1411 return [ 'uploaded-script-svg', $strippedElement ];
1415 if ( $strippedElement == 'style'
1416 && self
::checkCssFragment( Sanitizer
::normalizeCss( $data ) )
1418 wfDebug( __METHOD__
. ": hostile css in style element.\n" );
1419 return [ 'uploaded-hostile-svg' ];
1422 foreach ( $attribs as $attrib => $value ) {
1423 $stripped = $this->stripXmlNamespace( $attrib );
1424 $value = strtolower( $value );
1426 if ( substr( $stripped, 0, 2 ) == 'on' ) {
1428 . ": Found event-handler attribute '$attrib'='$value' in uploaded file.\n" );
1430 return [ 'uploaded-event-handler-on-svg', $attrib, $value ];
1433 # Do not allow relative links, or unsafe url schemas.
1434 # For <a> tags, only data:, http: and https: and same-document
1435 # fragment links are allowed. For all other tags, only data:
1436 # and fragment are allowed.
1437 if ( $stripped == 'href'
1438 && strpos( $value, 'data:' ) !== 0
1439 && strpos( $value, '#' ) !== 0
1441 if ( !( $strippedElement === 'a'
1442 && preg_match( '!^https?://!i', $value ) )
1444 wfDebug( __METHOD__
. ": Found href attribute <$strippedElement "
1445 . "'$attrib'='$value' in uploaded file.\n" );
1447 return [ 'uploaded-href-attribute-svg', $strippedElement, $attrib, $value ];
1451 # only allow data: targets that should be safe. This prevents vectors like,
1452 # image/svg, text/xml, application/xml, and text/html, which can contain scripts
1453 if ( $stripped == 'href' && strncasecmp( 'data:', $value, 5 ) === 0 ) {
1454 // rfc2397 parameters. This is only slightly slower than (;[\w;]+)*.
1455 // @codingStandardsIgnoreStart Generic.Files.LineLength
1456 $parameters = '(?>;[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+=(?>[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+|"(?>[\0-\x0c\x0e-\x21\x23-\x5b\x5d-\x7f]+|\\\\[\0-\x7f])*"))*(?:;base64)?';
1457 // @codingStandardsIgnoreEnd
1459 if ( !preg_match( "!^data:\s*image/(gif|jpeg|jpg|png)$parameters,!i", $value ) ) {
1460 wfDebug( __METHOD__
. ": Found href to unwhitelisted data: uri "
1461 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1462 return [ 'uploaded-href-unsafe-target-svg', $strippedElement, $attrib, $value ];
1466 # Change href with animate from (http://html5sec.org/#137).
1467 if ( $stripped === 'attributename'
1468 && $strippedElement === 'animate'
1469 && $this->stripXmlNamespace( $value ) == 'href'
1471 wfDebug( __METHOD__
. ": Found animate that might be changing href using from "
1472 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1474 return [ 'uploaded-animate-svg', $strippedElement, $attrib, $value ];
1477 # use set/animate to add event-handler attribute to parent
1478 if ( ( $strippedElement == 'set' ||
$strippedElement == 'animate' )
1479 && $stripped == 'attributename'
1480 && substr( $value, 0, 2 ) == 'on'
1482 wfDebug( __METHOD__
. ": Found svg setting event-handler attribute with "
1483 . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1485 return [ 'uploaded-setting-event-handler-svg', $strippedElement, $stripped, $value ];
1488 # use set to add href attribute to parent element
1489 if ( $strippedElement == 'set'
1490 && $stripped == 'attributename'
1491 && strpos( $value, 'href' ) !== false
1493 wfDebug( __METHOD__
. ": Found svg setting href attribute '$value' in uploaded file.\n" );
1495 return [ 'uploaded-setting-href-svg' ];
1498 # use set to add a remote / data / script target to an element
1499 if ( $strippedElement == 'set'
1500 && $stripped == 'to'
1501 && preg_match( '!(http|https|data|script):!sim', $value )
1503 wfDebug( __METHOD__
. ": Found svg setting attribute to '$value' in uploaded file.\n" );
1505 return [ 'uploaded-wrong-setting-svg', $value ];
1508 # use handler attribute with remote / data / script
1509 if ( $stripped == 'handler' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1510 wfDebug( __METHOD__
. ": Found svg setting handler with remote/data/script "
1511 . "'$attrib'='$value' in uploaded file.\n" );
1513 return [ 'uploaded-setting-handler-svg', $attrib, $value ];
1516 # use CSS styles to bring in remote code
1517 if ( $stripped == 'style'
1518 && self
::checkCssFragment( Sanitizer
::normalizeCss( $value ) )
1520 wfDebug( __METHOD__
. ": Found svg setting a style with "
1521 . "remote url '$attrib'='$value' in uploaded file.\n" );
1522 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1525 # Several attributes can include css, css character escaping isn't allowed
1526 $cssAttrs = [ 'font', 'clip-path', 'fill', 'filter', 'marker',
1527 'marker-end', 'marker-mid', 'marker-start', 'mask', 'stroke' ];
1528 if ( in_array( $stripped, $cssAttrs )
1529 && self
::checkCssFragment( $value )
1531 wfDebug( __METHOD__
. ": Found svg setting a style with "
1532 . "remote url '$attrib'='$value' in uploaded file.\n" );
1533 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1536 # image filters can pull in url, which could be svg that executes scripts
1537 if ( $strippedElement == 'image'
1538 && $stripped == 'filter'
1539 && preg_match( '!url\s*\(!sim', $value )
1541 wfDebug( __METHOD__
. ": Found image filter with url: "
1542 . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1544 return [ 'uploaded-image-filter-svg', $strippedElement, $stripped, $value ];
1548 return false; // No scripts detected
1552 * Check a block of CSS or CSS fragment for anything that looks like
1553 * it is bringing in remote code.
1554 * @param string $value a string of CSS
1555 * @param bool $propOnly only check css properties (start regex with :)
1556 * @return bool true if the CSS contains an illegal string, false if otherwise
1558 private static function checkCssFragment( $value ) {
1560 # Forbid external stylesheets, for both reliability and to protect viewer's privacy
1561 if ( stripos( $value, '@import' ) !== false ) {
1565 # We allow @font-face to embed fonts with data: urls, so we snip the string
1566 # 'url' out so this case won't match when we check for urls below
1567 $pattern = '!(@font-face\s*{[^}]*src:)url(\("data:;base64,)!im';
1568 $value = preg_replace( $pattern, '$1$2', $value );
1570 # Check for remote and executable CSS. Unlike in Sanitizer::checkCss, the CSS
1571 # properties filter and accelerator don't seem to be useful for xss in SVG files.
1572 # Expression and -o-link don't seem to work either, but filtering them here in case.
1573 # Additionally, we catch remote urls like url("http:..., url('http:..., url(http:...,
1574 # but not local ones such as url("#..., url('#..., url(#....
1575 if ( preg_match( '!expression
1577 | -o-link-source\s*:
1578 | -o-replace\s*:!imx', $value ) ) {
1582 if ( preg_match_all(
1583 "!(\s*(url|image|image-set)\s*\(\s*[\"']?\s*[^#]+.*?\))!sim",
1588 # TODO: redo this in one regex. Until then, url("#whatever") matches the first
1589 foreach ( $matches[1] as $match ) {
1590 if ( !preg_match( "!\s*(url|image|image-set)\s*\(\s*(#|'#|\"#)!im", $match ) ) {
1596 if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ) {
1604 * Divide the element name passed by the xml parser to the callback into URI and prifix.
1605 * @param string $element
1606 * @return array Containing the namespace URI and prefix
1608 private static function splitXmlNamespace( $element ) {
1609 // 'http://www.w3.org/2000/svg:script' -> array( 'http://www.w3.org/2000/svg', 'script' )
1610 $parts = explode( ':', strtolower( $element ) );
1611 $name = array_pop( $parts );
1612 $ns = implode( ':', $parts );
1614 return [ $ns, $name ];
1618 * @param string $name
1621 private function stripXmlNamespace( $name ) {
1622 // 'http://www.w3.org/2000/svg:script' -> 'script'
1623 $parts = explode( ':', strtolower( $name ) );
1625 return array_pop( $parts );
1629 * Generic wrapper function for a virus scanner program.
1630 * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
1631 * $wgAntivirusRequired may be used to deny upload if the scan fails.
1633 * @param string $file Pathname to the temporary upload file
1634 * @return mixed False if not virus is found, null if the scan fails or is disabled,
1635 * or a string containing feedback from the virus scanner if a virus was found.
1636 * If textual feedback is missing but a virus was found, this function returns true.
1638 public static function detectVirus( $file ) {
1639 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
1641 if ( !$wgAntivirus ) {
1642 wfDebug( __METHOD__
. ": virus scanner disabled\n" );
1647 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
1648 wfDebug( __METHOD__
. ": unknown virus scanner: $wgAntivirus\n" );
1649 $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
1650 [ 'virus-badscanner', $wgAntivirus ] );
1652 return wfMessage( 'virus-unknownscanner' )->text() . " $wgAntivirus";
1655 # look up scanner configuration
1656 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
1657 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
1658 $msgPattern = isset( $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ) ?
1659 $wgAntivirusSetup[$wgAntivirus]['messagepattern'] : null;
1661 if ( strpos( $command, "%f" ) === false ) {
1662 # simple pattern: append file to scan
1663 $command .= " " . wfEscapeShellArg( $file );
1665 # complex pattern: replace "%f" with file to scan
1666 $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
1669 wfDebug( __METHOD__
. ": running virus scan: $command \n" );
1671 # execute virus scanner
1674 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1675 # that does not seem to be worth the pain.
1676 # Ask me (Duesentrieb) about it if it's ever needed.
1677 $output = wfShellExecWithStderr( $command, $exitCode );
1679 # map exit code to AV_xxx constants.
1680 $mappedCode = $exitCode;
1681 if ( $exitCodeMap ) {
1682 if ( isset( $exitCodeMap[$exitCode] ) ) {
1683 $mappedCode = $exitCodeMap[$exitCode];
1684 } elseif ( isset( $exitCodeMap["*"] ) ) {
1685 $mappedCode = $exitCodeMap["*"];
1689 /* NB: AV_NO_VIRUS is 0 but AV_SCAN_FAILED is false,
1690 * so we need the strict equalities === and thus can't use a switch here
1692 if ( $mappedCode === AV_SCAN_FAILED
) {
1693 # scan failed (code was mapped to false by $exitCodeMap)
1694 wfDebug( __METHOD__
. ": failed to scan $file (code $exitCode).\n" );
1696 $output = $wgAntivirusRequired
1697 ?
wfMessage( 'virus-scanfailed', [ $exitCode ] )->text()
1699 } elseif ( $mappedCode === AV_SCAN_ABORTED
) {
1700 # scan failed because filetype is unknown (probably imune)
1701 wfDebug( __METHOD__
. ": unsupported file type $file (code $exitCode).\n" );
1703 } elseif ( $mappedCode === AV_NO_VIRUS
) {
1705 wfDebug( __METHOD__
. ": file passed virus scan.\n" );
1708 $output = trim( $output );
1711 $output = true; # if there's no output, return true
1712 } elseif ( $msgPattern ) {
1714 if ( preg_match( $msgPattern, $output, $groups ) ) {
1716 $output = $groups[1];
1721 wfDebug( __METHOD__
. ": FOUND VIRUS! scanner feedback: $output \n" );
1728 * Check if there's an overwrite conflict and, if so, if restrictions
1729 * forbid this user from performing the upload.
1733 * @return mixed True on success, array on failure
1735 private function checkOverwrite( $user ) {
1736 // First check whether the local file can be overwritten
1737 $file = $this->getLocalFile();
1738 $file->load( File
::READ_LATEST
);
1739 if ( $file->exists() ) {
1740 if ( !self
::userCanReUpload( $user, $file ) ) {
1741 return [ 'fileexists-forbidden', $file->getName() ];
1747 /* Check shared conflicts: if the local file does not exist, but
1748 * wfFindFile finds a file, it exists in a shared repository.
1750 $file = wfFindFile( $this->getTitle(), [ 'latest' => true ] );
1751 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1752 return [ 'fileexists-shared-forbidden', $file->getName() ];
1759 * Check if a user is the last uploader
1765 public static function userCanReUpload( User
$user, File
$img ) {
1766 if ( $user->isAllowed( 'reupload' ) ) {
1767 return true; // non-conditional
1768 } elseif ( !$user->isAllowed( 'reupload-own' ) ) {
1772 if ( !( $img instanceof LocalFile
) ) {
1778 return $user->getId() == $img->getUser( 'id' );
1782 * Helper function that does various existence checks for a file.
1783 * The following checks are performed:
1785 * - Article with the same name as the file exists
1786 * - File exists with normalized extension
1787 * - The file looks like a thumbnail and the original exists
1789 * @param File $file The File object to check
1790 * @return mixed False if the file does not exists, else an array
1792 public static function getExistsWarning( $file ) {
1793 if ( $file->exists() ) {
1794 return [ 'warning' => 'exists', 'file' => $file ];
1797 if ( $file->getTitle()->getArticleID() ) {
1798 return [ 'warning' => 'page-exists', 'file' => $file ];
1801 if ( strpos( $file->getName(), '.' ) == false ) {
1802 $partname = $file->getName();
1805 $n = strrpos( $file->getName(), '.' );
1806 $extension = substr( $file->getName(), $n +
1 );
1807 $partname = substr( $file->getName(), 0, $n );
1809 $normalizedExtension = File
::normalizeExtension( $extension );
1811 if ( $normalizedExtension != $extension ) {
1812 // We're not using the normalized form of the extension.
1813 // Normal form is lowercase, using most common of alternate
1814 // extensions (eg 'jpg' rather than 'JPEG').
1816 // Check for another file using the normalized form...
1817 $nt_lc = Title
::makeTitle( NS_FILE
, "{$partname}.{$normalizedExtension}" );
1818 $file_lc = wfLocalFile( $nt_lc );
1820 if ( $file_lc->exists() ) {
1822 'warning' => 'exists-normalized',
1824 'normalizedFile' => $file_lc
1829 // Check for files with the same name but a different extension
1830 $similarFiles = RepoGroup
::singleton()->getLocalRepo()->findFilesByPrefix(
1831 "{$partname}.", 1 );
1832 if ( count( $similarFiles ) ) {
1834 'warning' => 'exists-normalized',
1836 'normalizedFile' => $similarFiles[0],
1840 if ( self
::isThumbName( $file->getName() ) ) {
1841 # Check for filenames like 50px- or 180px-, these are mostly thumbnails
1842 $nt_thb = Title
::newFromText(
1843 substr( $partname, strpos( $partname, '-' ) +
1 ) . '.' . $extension,
1846 $file_thb = wfLocalFile( $nt_thb );
1847 if ( $file_thb->exists() ) {
1849 'warning' => 'thumb',
1851 'thumbFile' => $file_thb
1854 // File does not exist, but we just don't like the name
1856 'warning' => 'thumb-name',
1858 'thumbFile' => $file_thb
1863 foreach ( self
::getFilenamePrefixBlacklist() as $prefix ) {
1864 if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
1866 'warning' => 'bad-prefix',
1877 * Helper function that checks whether the filename looks like a thumbnail
1878 * @param string $filename
1881 public static function isThumbName( $filename ) {
1882 $n = strrpos( $filename, '.' );
1883 $partname = $n ?
substr( $filename, 0, $n ) : $filename;
1886 substr( $partname, 3, 3 ) == 'px-' ||
1887 substr( $partname, 2, 3 ) == 'px-'
1889 preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) );
1893 * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
1895 * @return array List of prefixes
1897 public static function getFilenamePrefixBlacklist() {
1899 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
1900 if ( !$message->isDisabled() ) {
1901 $lines = explode( "\n", $message->plain() );
1902 foreach ( $lines as $line ) {
1903 // Remove comment lines
1904 $comment = substr( trim( $line ), 0, 1 );
1905 if ( $comment == '#' ||
$comment == '' ) {
1908 // Remove additional comments after a prefix
1909 $comment = strpos( $line, '#' );
1910 if ( $comment > 0 ) {
1911 $line = substr( $line, 0, $comment - 1 );
1913 $blacklist[] = trim( $line );
1921 * Gets image info about the file just uploaded.
1923 * Also has the effect of setting metadata to be an 'indexed tag name' in
1924 * returned API result if 'metadata' was requested. Oddly, we have to pass
1925 * the "result" object down just so it can do that with the appropriate
1926 * format, presumably.
1928 * @param ApiResult $result
1929 * @return array Image info
1931 public function getImageInfo( $result ) {
1932 $file = $this->getLocalFile();
1933 /** @todo This cries out for refactoring.
1934 * We really want to say $file->getAllInfo(); here.
1935 * Perhaps "info" methods should be moved into files, and the API should
1936 * just wrap them in queries.
1938 if ( $file instanceof UploadStashFile
) {
1939 $imParam = ApiQueryStashImageInfo
::getPropertyNames();
1940 $info = ApiQueryStashImageInfo
::getInfo( $file, array_flip( $imParam ), $result );
1942 $imParam = ApiQueryImageInfo
::getPropertyNames();
1943 $info = ApiQueryImageInfo
::getInfo( $file, array_flip( $imParam ), $result );
1950 * @param array $error
1953 public function convertVerifyErrorToStatus( $error ) {
1954 $code = $error['status'];
1955 unset( $code['status'] );
1957 return Status
::newFatal( $this->getVerificationErrorCode( $code ), $error );
1961 * Get the MediaWiki maximum uploaded file size for given type of upload, based on
1964 * @param null|string $forType
1967 public static function getMaxUploadSize( $forType = null ) {
1968 global $wgMaxUploadSize;
1970 if ( is_array( $wgMaxUploadSize ) ) {
1971 if ( !is_null( $forType ) && isset( $wgMaxUploadSize[$forType] ) ) {
1972 return $wgMaxUploadSize[$forType];
1974 return $wgMaxUploadSize['*'];
1977 return intval( $wgMaxUploadSize );
1982 * Get the PHP maximum uploaded file size, based on ini settings. If there is no limit or the
1983 * limit can't be guessed, returns a very large number (PHP_INT_MAX).
1988 public static function getMaxPhpUploadSize() {
1989 $phpMaxFileSize = wfShorthandToInteger(
1990 ini_get( 'upload_max_filesize' ) ?
: ini_get( 'hhvm.server.upload.upload_max_file_size' ),
1993 $phpMaxPostSize = wfShorthandToInteger(
1994 ini_get( 'post_max_size' ) ?
: ini_get( 'hhvm.server.max_post_size' ),
1997 return min( $phpMaxFileSize, $phpMaxPostSize );
2001 * Get the current status of a chunked upload (used for polling)
2003 * The value will be read from cache.
2006 * @param string $statusKey
2007 * @return Status[]|bool
2009 public static function getSessionStatus( User
$user, $statusKey ) {
2010 $key = wfMemcKey( 'uploadstatus', $user->getId() ?
: md5( $user->getName() ), $statusKey );
2012 return ObjectCache
::getMainStashInstance()->get( $key );
2016 * Set the current status of a chunked upload (used for polling)
2018 * The value will be set in cache for 1 day
2021 * @param string $statusKey
2022 * @param array|bool $value
2025 public static function setSessionStatus( User
$user, $statusKey, $value ) {
2026 $key = wfMemcKey( 'uploadstatus', $user->getId() ?
: md5( $user->getName() ), $statusKey );
2028 $cache = ObjectCache
::getMainStashInstance();
2029 if ( $value === false ) {
2030 $cache->delete( $key );
2032 $cache->set( $key, $value, $cache::TTL_DAY
);