SessionManager: Ignore Session object destruction during global shutdown
[mediawiki.git] / includes / upload / UploadBase.php
blob1185c4da61670d16241b15b7fe2d5e0574c9c618
1 <?php
2 /**
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
20 * @file
21 * @ingroup Upload
24 /**
25 * @defgroup Upload Upload related
28 /**
29 * @ingroup Upload
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) */
40 protected $mTempPath;
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 = [
52 'UTF-8',
53 'ISO-8859-1',
54 'ISO-8859-2',
55 'UTF-16',
56 'UTF-32'
59 const SUCCESS = 0;
60 const OK = 0;
61 const EMPTY_FILE = 3;
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;
73 /**
74 * @param int $error
75 * @return string
77 public function getVerificationErrorCode( $error ) {
78 $code_to_status = [
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';
98 /**
99 * Returns true if uploads are enabled.
100 * Can be override by subclasses.
101 * @return bool
103 public static function isEnabled() {
104 global $wgEnableUploads;
106 if ( !$wgEnableUploads ) {
107 return false;
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.
119 * @param User $user
120 * @return bool|string
122 public static function isAllowed( $user ) {
123 foreach ( [ 'upload', 'edit' ] as $permission ) {
124 if ( !$user->isAllowed( $permission ) ) {
125 return $permission;
129 return true;
133 * Returns true if the user has surpassed the upload rate limit, false otherwise.
135 * @param User $user
136 * @return bool
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 : $request->getVal( 'wpSourceType', 'File' );
155 if ( !$type ) {
156 return null;
159 // Get the upload class
160 $type = ucfirst( $type );
162 // Give hooks the chance to handle this request
163 $className = null;
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 ) ) {
169 return null;
173 // Check whether this upload class is enabled
174 if ( !call_user_func( [ $className, 'isEnabled' ] ) ) {
175 return null;
178 // Check whether the request is valid
179 if ( !call_user_func( [ $className, 'isValidRequest' ], $request ) ) {
180 return null;
183 /** @var UploadBase $handler */
184 $handler = new $className;
186 $handler->initializeFromRequest( $request );
188 return $handler;
192 * Check whether a request if valid for this handler
193 * @param WebRequest $request
194 * @return bool
196 public static function isValidRequest( $request ) {
197 return false;
200 public function __construct() {
204 * Returns the upload type. Should be overridden by child classes
206 * @since 1.18
207 * @return string
209 public function getSourceType() {
210 return null;
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( $this->mTempPath );
247 if ( !$fileSize ) {
248 $this->mFileSize = filesize( $this->mTempPath );
250 } else {
251 $this->tempFileObj = null;
256 * Fetch the file. Usually a no-op
257 * @return Status
259 public function fetchFile() {
260 return Status::newGood();
264 * Return true if the file is empty
265 * @return bool
267 public function isEmptyFile() {
268 return empty( $this->mFileSize );
272 * Return the file size
273 * @return int
275 public function getFileSize() {
276 return $this->mFileSize;
280 * Get the base 36 SHA1 of the file
281 * @return string
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 );
298 if ( $tmpFile ) {
299 $tmpFile->bind( $this ); // keep alive with $this
301 $path = $tmpFile ? $tmpFile->getPath() : false;
302 } else {
303 $path = $srcPath;
306 return $path;
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 ) {
327 return [
328 'status' => self::FILE_TOO_LARGE,
329 'max' => $maxSize,
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 ) {
340 return [
341 'status' => self::VERIFICATION_ERROR,
342 'details' => $verification
347 * Make sure this file can be created
349 $result = $this->validateName();
350 if ( $result !== true ) {
351 return $result;
354 $error = '';
355 if ( !Hooks::run( 'UploadVerification',
356 [ $this->mDestName, $this->mTempPath, &$error ] )
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'
368 * and other keys
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;
384 return $result;
386 $this->mDestName = $this->getLocalFile()->getName();
388 return true;
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 );
412 fclose( $fp );
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 ];
424 return true;
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 ) {
437 return $status;
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 ) {
455 return $svgStatus;
460 $handler = MediaHandler::getHandler( $mime );
461 if ( $handler ) {
462 $handlerStatus = $handler->verifyUpload( $this->mTempPath );
463 if ( !$handlerStatus->isOK() ) {
464 $errors = $handlerStatus->getErrorsArray();
466 return reset( $errors );
470 Hooks::run( 'UploadVerifyFile', [ $this, $mime, &$status ] );
471 if ( $status !== true ) {
472 return $status;
475 wfDebug( __METHOD__ . ": all clear; passing.\n" );
477 return true;
481 * A verification routine suitable for partial files
483 * Runs the blacklist checks, but not any checks that may
484 * assume the entire file is present.
486 * @return mixed True for valid or array with error message key.
488 protected function verifyPartialFile() {
489 global $wgAllowJavaUploads, $wgDisableUploadScriptChecks;
491 # getTitle() sets some internal parameters like $this->mFinalExtension
492 $this->getTitle();
494 $this->mFileProps = FSFile::getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
496 # check MIME type, if desired
497 $mime = $this->mFileProps['file-mime'];
498 $status = $this->verifyMimeType( $mime );
499 if ( $status !== true ) {
500 return $status;
503 # check for htmlish code and javascript
504 if ( !$wgDisableUploadScriptChecks ) {
505 if ( self::detectScript( $this->mTempPath, $mime, $this->mFinalExtension ) ) {
506 return [ 'uploadscripted' ];
508 if ( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
509 $svgStatus = $this->detectScriptInSvg( $this->mTempPath, true );
510 if ( $svgStatus !== false ) {
511 return $svgStatus;
516 # Check for Java applets, which if uploaded can bypass cross-site
517 # restrictions.
518 if ( !$wgAllowJavaUploads ) {
519 $this->mJavaDetected = false;
520 $zipStatus = ZipDirectoryReader::read( $this->mTempPath,
521 [ $this, 'zipEntryCallback' ] );
522 if ( !$zipStatus->isOK() ) {
523 $errors = $zipStatus->getErrorsArray();
524 $error = reset( $errors );
525 if ( $error[0] !== 'zip-wrong-format' ) {
526 return $error;
529 if ( $this->mJavaDetected ) {
530 return [ 'uploadjava' ];
534 # Scan the uploaded file for viruses
535 $virus = $this->detectVirus( $this->mTempPath );
536 if ( $virus ) {
537 return [ 'uploadvirus', $virus ];
540 return true;
544 * Callback for ZipDirectoryReader to detect Java class files.
546 * @param array $entry
548 function zipEntryCallback( $entry ) {
549 $names = [ $entry['name'] ];
551 // If there is a null character, cut off the name at it, because JDK's
552 // ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name
553 // were constructed which had ".class\0" followed by a string chosen to
554 // make the hash collide with the truncated name, that file could be
555 // returned in response to a request for the .class file.
556 $nullPos = strpos( $entry['name'], "\000" );
557 if ( $nullPos !== false ) {
558 $names[] = substr( $entry['name'], 0, $nullPos );
561 // If there is a trailing slash in the file name, we have to strip it,
562 // because that's what ZIP_GetEntry() does.
563 if ( preg_grep( '!\.class/?$!', $names ) ) {
564 $this->mJavaDetected = true;
569 * Alias for verifyTitlePermissions. The function was originally
570 * 'verifyPermissions', but that suggests it's checking the user, when it's
571 * really checking the title + user combination.
573 * @param User $user User object to verify the permissions against
574 * @return mixed An array as returned by getUserPermissionsErrors or true
575 * in case the user has proper permissions.
577 public function verifyPermissions( $user ) {
578 return $this->verifyTitlePermissions( $user );
582 * Check whether the user can edit, upload and create the image. This
583 * checks only against the current title; if it returns errors, it may
584 * very well be that another title will not give errors. Therefore
585 * isAllowed() should be called as well for generic is-user-blocked or
586 * can-user-upload checking.
588 * @param User $user User object to verify the permissions against
589 * @return mixed An array as returned by getUserPermissionsErrors or true
590 * in case the user has proper permissions.
592 public function verifyTitlePermissions( $user ) {
594 * If the image is protected, non-sysop users won't be able
595 * to modify it by uploading a new revision.
597 $nt = $this->getTitle();
598 if ( is_null( $nt ) ) {
599 return true;
601 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
602 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
603 if ( !$nt->exists() ) {
604 $permErrorsCreate = $nt->getUserPermissionsErrors( 'create', $user );
605 } else {
606 $permErrorsCreate = [];
608 if ( $permErrors || $permErrorsUpload || $permErrorsCreate ) {
609 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
610 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
612 return $permErrors;
615 $overwriteError = $this->checkOverwrite( $user );
616 if ( $overwriteError !== true ) {
617 return [ $overwriteError ];
620 return true;
624 * Check for non fatal problems with the file.
626 * This should not assume that mTempPath is set.
628 * @return array Array of warnings
630 public function checkWarnings() {
631 global $wgLang;
633 $warnings = [];
635 $localFile = $this->getLocalFile();
636 $localFile->load( File::READ_LATEST );
637 $filename = $localFile->getName();
640 * Check whether the resulting filename is different from the desired one,
641 * but ignore things like ucfirst() and spaces/underscore things
643 $comparableName = str_replace( ' ', '_', $this->mDesiredDestName );
644 $comparableName = Title::capitalize( $comparableName, NS_FILE );
646 if ( $this->mDesiredDestName != $filename && $comparableName != $filename ) {
647 $warnings['badfilename'] = $filename;
650 // Check whether the file extension is on the unwanted list
651 global $wgCheckFileExtensions, $wgFileExtensions;
652 if ( $wgCheckFileExtensions ) {
653 $extensions = array_unique( $wgFileExtensions );
654 if ( !$this->checkFileExtension( $this->mFinalExtension, $extensions ) ) {
655 $warnings['filetype-unwanted-type'] = [ $this->mFinalExtension,
656 $wgLang->commaList( $extensions ), count( $extensions ) ];
660 global $wgUploadSizeWarning;
661 if ( $wgUploadSizeWarning && ( $this->mFileSize > $wgUploadSizeWarning ) ) {
662 $warnings['large-file'] = [ $wgUploadSizeWarning, $this->mFileSize ];
665 if ( $this->mFileSize == 0 ) {
666 $warnings['empty-file'] = true;
669 $exists = self::getExistsWarning( $localFile );
670 if ( $exists !== false ) {
671 $warnings['exists'] = $exists;
674 if ( $localFile->wasDeleted() && !$localFile->exists() ) {
675 $warnings['was-deleted'] = $filename;
678 // Check dupes against existing files
679 $hash = $this->getTempFileSha1Base36();
680 $dupes = RepoGroup::singleton()->findBySha1( $hash );
681 $title = $this->getTitle();
682 // Remove all matches against self
683 foreach ( $dupes as $key => $dupe ) {
684 if ( $title->equals( $dupe->getTitle() ) ) {
685 unset( $dupes[$key] );
688 if ( $dupes ) {
689 $warnings['duplicate'] = $dupes;
692 // Check dupes against archives
693 $archivedFile = new ArchivedFile( null, 0, '', $hash );
694 if ( $archivedFile->getID() > 0 ) {
695 if ( $archivedFile->userCan( File::DELETED_FILE ) ) {
696 $warnings['duplicate-archive'] = $archivedFile->getName();
697 } else {
698 $warnings['duplicate-archive'] = '';
702 return $warnings;
706 * Really perform the upload. Stores the file in the local repo, watches
707 * if necessary and runs the UploadComplete hook.
709 * @param string $comment
710 * @param string $pageText
711 * @param bool $watch Whether the file page should be added to user's watchlist.
712 * (This doesn't check $user's permissions.)
713 * @param User $user
714 * @param string[] $tags Change tags to add to the log entry and page revision.
715 * (This doesn't check $user's permissions.)
716 * @return Status Indicating the whether the upload succeeded.
718 public function performUpload( $comment, $pageText, $watch, $user, $tags = [] ) {
719 $this->getLocalFile()->load( File::READ_LATEST );
721 $status = $this->getLocalFile()->upload(
722 $this->mTempPath,
723 $comment,
724 $pageText,
725 File::DELETE_SOURCE,
726 $this->mFileProps,
727 false,
728 $user,
729 $tags
732 if ( $status->isGood() ) {
733 if ( $watch ) {
734 WatchAction::doWatch(
735 $this->getLocalFile()->getTitle(),
736 $user,
737 User::IGNORE_USER_RIGHTS
740 Hooks::run( 'UploadComplete', [ &$this ] );
742 $this->postProcessUpload();
745 return $status;
749 * Perform extra steps after a successful upload.
751 * @since 1.25
753 public function postProcessUpload() {
754 global $wgUploadThumbnailRenderMap;
756 $jobs = [];
758 $sizes = $wgUploadThumbnailRenderMap;
759 rsort( $sizes );
761 $file = $this->getLocalFile();
763 foreach ( $sizes as $size ) {
764 if ( $file->isVectorized() || $file->getWidth() > $size ) {
765 $jobs[] = new ThumbnailRenderJob(
766 $file->getTitle(),
767 [ 'transformParams' => [ 'width' => $size ] ]
772 if ( $jobs ) {
773 JobQueueGroup::singleton()->push( $jobs );
778 * Returns the title of the file to be uploaded. Sets mTitleError in case
779 * the name was illegal.
781 * @return Title The title of the file or null in case the name was illegal
783 public function getTitle() {
784 if ( $this->mTitle !== false ) {
785 return $this->mTitle;
787 if ( !is_string( $this->mDesiredDestName ) ) {
788 $this->mTitleError = self::ILLEGAL_FILENAME;
789 $this->mTitle = null;
791 return $this->mTitle;
793 /* Assume that if a user specified File:Something.jpg, this is an error
794 * and that the namespace prefix needs to be stripped of.
796 $title = Title::newFromText( $this->mDesiredDestName );
797 if ( $title && $title->getNamespace() == NS_FILE ) {
798 $this->mFilteredName = $title->getDBkey();
799 } else {
800 $this->mFilteredName = $this->mDesiredDestName;
803 # oi_archive_name is max 255 bytes, which include a timestamp and an
804 # exclamation mark, so restrict file name to 240 bytes.
805 if ( strlen( $this->mFilteredName ) > 240 ) {
806 $this->mTitleError = self::FILENAME_TOO_LONG;
807 $this->mTitle = null;
809 return $this->mTitle;
813 * Chop off any directories in the given filename. Then
814 * filter out illegal characters, and try to make a legible name
815 * out of it. We'll strip some silently that Title would die on.
817 $this->mFilteredName = wfStripIllegalFilenameChars( $this->mFilteredName );
818 /* Normalize to title form before we do any further processing */
819 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
820 if ( is_null( $nt ) ) {
821 $this->mTitleError = self::ILLEGAL_FILENAME;
822 $this->mTitle = null;
824 return $this->mTitle;
826 $this->mFilteredName = $nt->getDBkey();
829 * We'll want to blacklist against *any* 'extension', and use
830 * only the final one for the whitelist.
832 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName );
834 if ( count( $ext ) ) {
835 $this->mFinalExtension = trim( $ext[count( $ext ) - 1] );
836 } else {
837 $this->mFinalExtension = '';
839 # No extension, try guessing one
840 $magic = MimeMagic::singleton();
841 $mime = $magic->guessMimeType( $this->mTempPath );
842 if ( $mime !== 'unknown/unknown' ) {
843 # Get a space separated list of extensions
844 $extList = $magic->getExtensionsForType( $mime );
845 if ( $extList ) {
846 # Set the extension to the canonical extension
847 $this->mFinalExtension = strtok( $extList, ' ' );
849 # Fix up the other variables
850 $this->mFilteredName .= ".{$this->mFinalExtension}";
851 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
852 $ext = [ $this->mFinalExtension ];
857 /* Don't allow users to override the blacklist (check file extension) */
858 global $wgCheckFileExtensions, $wgStrictFileExtensions;
859 global $wgFileExtensions, $wgFileBlacklist;
861 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
863 if ( $this->mFinalExtension == '' ) {
864 $this->mTitleError = self::FILETYPE_MISSING;
865 $this->mTitle = null;
867 return $this->mTitle;
868 } elseif ( $blackListedExtensions ||
869 ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
870 !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) )
872 $this->mBlackListedExtensions = $blackListedExtensions;
873 $this->mTitleError = self::FILETYPE_BADTYPE;
874 $this->mTitle = null;
876 return $this->mTitle;
879 // Windows may be broken with special characters, see bug 1780
880 if ( !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() )
881 && !RepoGroup::singleton()->getLocalRepo()->backendSupportsUnicodePaths()
883 $this->mTitleError = self::WINDOWS_NONASCII_FILENAME;
884 $this->mTitle = null;
886 return $this->mTitle;
889 # If there was more than one "extension", reassemble the base
890 # filename to prevent bogus complaints about length
891 if ( count( $ext ) > 1 ) {
892 $iterations = count( $ext ) - 1;
893 for ( $i = 0; $i < $iterations; $i++ ) {
894 $partname .= '.' . $ext[$i];
898 if ( strlen( $partname ) < 1 ) {
899 $this->mTitleError = self::MIN_LENGTH_PARTNAME;
900 $this->mTitle = null;
902 return $this->mTitle;
905 $this->mTitle = $nt;
907 return $this->mTitle;
911 * Return the local file and initializes if necessary.
913 * @return LocalFile|UploadStashFile|null
915 public function getLocalFile() {
916 if ( is_null( $this->mLocalFile ) ) {
917 $nt = $this->getTitle();
918 $this->mLocalFile = is_null( $nt ) ? null : wfLocalFile( $nt );
921 return $this->mLocalFile;
925 * If the user does not supply all necessary information in the first upload
926 * form submission (either by accident or by design) then we may want to
927 * stash the file temporarily, get more information, and publish the file
928 * later.
930 * This method will stash a file in a temporary directory for later
931 * processing, and save the necessary descriptive info into the database.
932 * This method returns the file object, which also has a 'fileKey' property
933 * which can be passed through a form or API request to find this stashed
934 * file again.
936 * @param User $user
937 * @return UploadStashFile Stashed file
939 public function stashFile( User $user = null ) {
940 // was stashSessionFile
942 $stash = RepoGroup::singleton()->getLocalRepo()->getUploadStash( $user );
943 $file = $stash->stashFile( $this->mTempPath, $this->getSourceType() );
944 $this->mLocalFile = $file;
946 return $file;
950 * Stash a file in a temporary directory, returning a key which can be used
951 * to find the file again. See stashFile().
953 * @return string File key
955 public function stashFileGetKey() {
956 return $this->stashFile()->getFileKey();
960 * alias for stashFileGetKey, for backwards compatibility
962 * @return string File key
964 public function stashSession() {
965 return $this->stashFileGetKey();
969 * If we've modified the upload file we need to manually remove it
970 * on exit to clean up.
972 public function cleanupTempFile() {
973 if ( $this->mRemoveTempFile && $this->tempFileObj ) {
974 // Delete when all relevant TempFSFile handles go out of scope
975 wfDebug( __METHOD__ . ": Marked temporary file '{$this->mTempPath}' for removal\n" );
976 $this->tempFileObj->autocollect();
980 public function getTempPath() {
981 return $this->mTempPath;
985 * Split a file into a base name and all dot-delimited 'extensions'
986 * on the end. Some web server configurations will fall back to
987 * earlier pseudo-'extensions' to determine type and execute
988 * scripts, so the blacklist needs to check them all.
990 * @param string $filename
991 * @return array
993 public static function splitExtensions( $filename ) {
994 $bits = explode( '.', $filename );
995 $basename = array_shift( $bits );
997 return [ $basename, $bits ];
1001 * Perform case-insensitive match against a list of file extensions.
1002 * Returns true if the extension is in the list.
1004 * @param string $ext
1005 * @param array $list
1006 * @return bool
1008 public static function checkFileExtension( $ext, $list ) {
1009 return in_array( strtolower( $ext ), $list );
1013 * Perform case-insensitive match against a list of file extensions.
1014 * Returns an array of matching extensions.
1016 * @param array $ext
1017 * @param array $list
1018 * @return bool
1020 public static function checkFileExtensionList( $ext, $list ) {
1021 return array_intersect( array_map( 'strtolower', $ext ), $list );
1025 * Checks if the MIME type of the uploaded file matches the file extension.
1027 * @param string $mime The MIME type of the uploaded file
1028 * @param string $extension The filename extension that the file is to be served with
1029 * @return bool
1031 public static function verifyExtension( $mime, $extension ) {
1032 $magic = MimeMagic::singleton();
1034 if ( !$mime || $mime == 'unknown' || $mime == 'unknown/unknown' ) {
1035 if ( !$magic->isRecognizableExtension( $extension ) ) {
1036 wfDebug( __METHOD__ . ": passing file with unknown detected mime type; " .
1037 "unrecognized extension '$extension', can't verify\n" );
1039 return true;
1040 } else {
1041 wfDebug( __METHOD__ . ": rejecting file with unknown detected mime type; " .
1042 "recognized extension '$extension', so probably invalid file\n" );
1044 return false;
1048 $match = $magic->isMatchingExtension( $extension, $mime );
1050 if ( $match === null ) {
1051 if ( $magic->getTypesForExtension( $extension ) !== null ) {
1052 wfDebug( __METHOD__ . ": No extension known for $mime, but we know a mime for $extension\n" );
1054 return false;
1055 } else {
1056 wfDebug( __METHOD__ . ": no file extension known for mime type $mime, passing file\n" );
1058 return true;
1060 } elseif ( $match === true ) {
1061 wfDebug( __METHOD__ . ": mime type $mime matches extension $extension, passing file\n" );
1063 /** @todo If it's a bitmap, make sure PHP or ImageMagick resp. can handle it! */
1064 return true;
1065 } else {
1066 wfDebug( __METHOD__
1067 . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
1069 return false;
1074 * Heuristic for detecting files that *could* contain JavaScript instructions or
1075 * things that may look like HTML to a browser and are thus
1076 * potentially harmful. The present implementation will produce false
1077 * positives in some situations.
1079 * @param string $file Pathname to the temporary upload file
1080 * @param string $mime The MIME type of the file
1081 * @param string $extension The extension of the file
1082 * @return bool True if the file contains something looking like embedded scripts
1084 public static function detectScript( $file, $mime, $extension ) {
1085 global $wgAllowTitlesInSVG;
1087 # ugly hack: for text files, always look at the entire file.
1088 # For binary field, just check the first K.
1090 if ( strpos( $mime, 'text/' ) === 0 ) {
1091 $chunk = file_get_contents( $file );
1092 } else {
1093 $fp = fopen( $file, 'rb' );
1094 $chunk = fread( $fp, 1024 );
1095 fclose( $fp );
1098 $chunk = strtolower( $chunk );
1100 if ( !$chunk ) {
1101 return false;
1104 # decode from UTF-16 if needed (could be used for obfuscation).
1105 if ( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
1106 $enc = 'UTF-16BE';
1107 } elseif ( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
1108 $enc = 'UTF-16LE';
1109 } else {
1110 $enc = null;
1113 if ( $enc ) {
1114 $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
1117 $chunk = trim( $chunk );
1119 /** @todo FIXME: Convert from UTF-16 if necessary! */
1120 wfDebug( __METHOD__ . ": checking for embedded scripts and HTML stuff\n" );
1122 # check for HTML doctype
1123 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
1124 return true;
1127 // Some browsers will interpret obscure xml encodings as UTF-8, while
1128 // PHP/expat will interpret the given encoding in the xml declaration (bug 47304)
1129 if ( $extension == 'svg' || strpos( $mime, 'image/svg' ) === 0 ) {
1130 if ( self::checkXMLEncodingMissmatch( $file ) ) {
1131 return true;
1136 * Internet Explorer for Windows performs some really stupid file type
1137 * autodetection which can cause it to interpret valid image files as HTML
1138 * and potentially execute JavaScript, creating a cross-site scripting
1139 * attack vectors.
1141 * Apple's Safari browser also performs some unsafe file type autodetection
1142 * which can cause legitimate files to be interpreted as HTML if the
1143 * web server is not correctly configured to send the right content-type
1144 * (or if you're really uploading plain text and octet streams!)
1146 * Returns true if IE is likely to mistake the given file for HTML.
1147 * Also returns true if Safari would mistake the given file for HTML
1148 * when served with a generic content-type.
1150 $tags = [
1151 '<a href',
1152 '<body',
1153 '<head',
1154 '<html', # also in safari
1155 '<img',
1156 '<pre',
1157 '<script', # also in safari
1158 '<table'
1161 if ( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
1162 $tags[] = '<title';
1165 foreach ( $tags as $tag ) {
1166 if ( false !== strpos( $chunk, $tag ) ) {
1167 wfDebug( __METHOD__ . ": found something that may make it be mistaken for html: $tag\n" );
1169 return true;
1174 * look for JavaScript
1177 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
1178 $chunk = Sanitizer::decodeCharReferences( $chunk );
1180 # look for script-types
1181 if ( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
1182 wfDebug( __METHOD__ . ": found script types\n" );
1184 return true;
1187 # look for html-style script-urls
1188 if ( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1189 wfDebug( __METHOD__ . ": found html-style script urls\n" );
1191 return true;
1194 # look for css-style script-urls
1195 if ( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1196 wfDebug( __METHOD__ . ": found css-style script urls\n" );
1198 return true;
1201 wfDebug( __METHOD__ . ": no scripts found\n" );
1203 return false;
1207 * Check a whitelist of xml encodings that are known not to be interpreted differently
1208 * by the server's xml parser (expat) and some common browsers.
1210 * @param string $file Pathname to the temporary upload file
1211 * @return bool True if the file contains an encoding that could be misinterpreted
1213 public static function checkXMLEncodingMissmatch( $file ) {
1214 global $wgSVGMetadataCutoff;
1215 $contents = file_get_contents( $file, false, null, -1, $wgSVGMetadataCutoff );
1216 $encodingRegex = '!encoding[ \t\n\r]*=[ \t\n\r]*[\'"](.*?)[\'"]!si';
1218 if ( preg_match( "!<\?xml\b(.*?)\?>!si", $contents, $matches ) ) {
1219 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1220 && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1222 wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1224 return true;
1226 } elseif ( preg_match( "!<\?xml\b!si", $contents ) ) {
1227 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1228 // bytes. There shouldn't be a legitimate reason for this to happen.
1229 wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1231 return true;
1232 } elseif ( substr( $contents, 0, 4 ) == "\x4C\x6F\xA7\x94" ) {
1233 // EBCDIC encoded XML
1234 wfDebug( __METHOD__ . ": EBCDIC Encoded XML\n" );
1236 return true;
1239 // It's possible the file is encoded with multi-byte encoding, so re-encode attempt to
1240 // detect the encoding in case is specifies an encoding not whitelisted in self::$safeXmlEncodings
1241 $attemptEncodings = [ 'UTF-16', 'UTF-16BE', 'UTF-32', 'UTF-32BE' ];
1242 foreach ( $attemptEncodings as $encoding ) {
1243 MediaWiki\suppressWarnings();
1244 $str = iconv( $encoding, 'UTF-8', $contents );
1245 MediaWiki\restoreWarnings();
1246 if ( $str != '' && preg_match( "!<\?xml\b(.*?)\?>!si", $str, $matches ) ) {
1247 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1248 && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1250 wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1252 return true;
1254 } elseif ( $str != '' && preg_match( "!<\?xml\b!si", $str ) ) {
1255 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1256 // bytes. There shouldn't be a legitimate reason for this to happen.
1257 wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1259 return true;
1263 return false;
1267 * @param string $filename
1268 * @param bool $partial
1269 * @return mixed False of the file is verified (does not contain scripts), array otherwise.
1271 protected function detectScriptInSvg( $filename, $partial ) {
1272 $this->mSVGNSError = false;
1273 $check = new XmlTypeCheck(
1274 $filename,
1275 [ $this, 'checkSvgScriptCallback' ],
1276 true,
1277 [ 'processing_instruction_handler' => 'UploadBase::checkSvgPICallback' ]
1279 if ( $check->wellFormed !== true ) {
1280 // Invalid xml (bug 58553)
1281 // But only when non-partial (bug 65724)
1282 return $partial ? false : [ 'uploadinvalidxml' ];
1283 } elseif ( $check->filterMatch ) {
1284 if ( $this->mSVGNSError ) {
1285 return [ 'uploadscriptednamespace', $this->mSVGNSError ];
1288 return $check->filterMatchType;
1291 return false;
1295 * Callback to filter SVG Processing Instructions.
1296 * @param string $target Processing instruction name
1297 * @param string $data Processing instruction attribute and value
1298 * @return bool (true if the filter identified something bad)
1300 public static function checkSvgPICallback( $target, $data ) {
1301 // Don't allow external stylesheets (bug 57550)
1302 if ( preg_match( '/xml-stylesheet/i', $target ) ) {
1303 return [ 'upload-scripted-pi-callback' ];
1306 return false;
1310 * @todo Replace this with a whitelist filter!
1311 * @param string $element
1312 * @param array $attribs
1313 * @return bool
1315 public function checkSvgScriptCallback( $element, $attribs, $data = null ) {
1317 list( $namespace, $strippedElement ) = $this->splitXmlNamespace( $element );
1319 // We specifically don't include:
1320 // http://www.w3.org/1999/xhtml (bug 60771)
1321 static $validNamespaces = [
1323 'adobe:ns:meta/',
1324 'http://creativecommons.org/ns#',
1325 'http://inkscape.sourceforge.net/dtd/sodipodi-0.dtd',
1326 'http://ns.adobe.com/adobeillustrator/10.0/',
1327 'http://ns.adobe.com/adobesvgviewerextensions/3.0/',
1328 'http://ns.adobe.com/extensibility/1.0/',
1329 'http://ns.adobe.com/flows/1.0/',
1330 'http://ns.adobe.com/illustrator/1.0/',
1331 'http://ns.adobe.com/imagereplacement/1.0/',
1332 'http://ns.adobe.com/pdf/1.3/',
1333 'http://ns.adobe.com/photoshop/1.0/',
1334 'http://ns.adobe.com/saveforweb/1.0/',
1335 'http://ns.adobe.com/variables/1.0/',
1336 'http://ns.adobe.com/xap/1.0/',
1337 'http://ns.adobe.com/xap/1.0/g/',
1338 'http://ns.adobe.com/xap/1.0/g/img/',
1339 'http://ns.adobe.com/xap/1.0/mm/',
1340 'http://ns.adobe.com/xap/1.0/rights/',
1341 'http://ns.adobe.com/xap/1.0/stype/dimensions#',
1342 'http://ns.adobe.com/xap/1.0/stype/font#',
1343 'http://ns.adobe.com/xap/1.0/stype/manifestitem#',
1344 'http://ns.adobe.com/xap/1.0/stype/resourceevent#',
1345 'http://ns.adobe.com/xap/1.0/stype/resourceref#',
1346 'http://ns.adobe.com/xap/1.0/t/pg/',
1347 'http://purl.org/dc/elements/1.1/',
1348 'http://purl.org/dc/elements/1.1',
1349 'http://schemas.microsoft.com/visio/2003/svgextensions/',
1350 'http://sodipodi.sourceforge.net/dtd/sodipodi-0.dtd',
1351 'http://taptrix.com/inkpad/svg_extensions',
1352 'http://web.resource.org/cc/',
1353 'http://www.freesoftware.fsf.org/bkchem/cdml',
1354 'http://www.inkscape.org/namespaces/inkscape',
1355 'http://www.opengis.net/gml',
1356 'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
1357 'http://www.w3.org/2000/svg',
1358 'http://www.w3.org/tr/rec-rdf-syntax/',
1361 if ( !in_array( $namespace, $validNamespaces ) ) {
1362 wfDebug( __METHOD__ . ": Non-svg namespace '$namespace' in uploaded file.\n" );
1363 /** @todo Return a status object to a closure in XmlTypeCheck, for MW1.21+ */
1364 $this->mSVGNSError = $namespace;
1366 return true;
1370 * check for elements that can contain javascript
1372 if ( $strippedElement == 'script' ) {
1373 wfDebug( __METHOD__ . ": Found script element '$element' in uploaded file.\n" );
1375 return [ 'uploaded-script-svg', $strippedElement ];
1378 # e.g., <svg xmlns="http://www.w3.org/2000/svg">
1379 # <handler xmlns:ev="http://www.w3.org/2001/xml-events" ev:event="load">alert(1)</handler> </svg>
1380 if ( $strippedElement == 'handler' ) {
1381 wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1383 return [ 'uploaded-script-svg', $strippedElement ];
1386 # SVG reported in Feb '12 that used xml:stylesheet to generate javascript block
1387 if ( $strippedElement == 'stylesheet' ) {
1388 wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1390 return [ 'uploaded-script-svg', $strippedElement ];
1393 # Block iframes, in case they pass the namespace check
1394 if ( $strippedElement == 'iframe' ) {
1395 wfDebug( __METHOD__ . ": iframe in uploaded file.\n" );
1397 return [ 'uploaded-script-svg', $strippedElement ];
1400 # Check <style> css
1401 if ( $strippedElement == 'style'
1402 && self::checkCssFragment( Sanitizer::normalizeCss( $data ) )
1404 wfDebug( __METHOD__ . ": hostile css in style element.\n" );
1405 return [ 'uploaded-hostile-svg' ];
1408 foreach ( $attribs as $attrib => $value ) {
1409 $stripped = $this->stripXmlNamespace( $attrib );
1410 $value = strtolower( $value );
1412 if ( substr( $stripped, 0, 2 ) == 'on' ) {
1413 wfDebug( __METHOD__
1414 . ": Found event-handler attribute '$attrib'='$value' in uploaded file.\n" );
1416 return [ 'uploaded-event-handler-on-svg', $attrib, $value ];
1419 # href with non-local target (don't allow http://, javascript:, etc)
1420 if ( $stripped == 'href'
1421 && strpos( $value, 'data:' ) !== 0
1422 && strpos( $value, '#' ) !== 0
1424 if ( !( $strippedElement === 'a'
1425 && preg_match( '!^https?://!im', $value ) )
1427 wfDebug( __METHOD__ . ": Found href attribute <$strippedElement "
1428 . "'$attrib'='$value' in uploaded file.\n" );
1430 return [ 'uploaded-href-attribute-svg', $strippedElement, $attrib, $value ];
1434 # only allow data: targets that should be safe. This prevents vectors like,
1435 # image/svg, text/xml, application/xml, and text/html, which can contain scripts
1436 if ( $stripped == 'href' && strncasecmp( 'data:', $value, 5 ) === 0 ) {
1437 // rfc2397 parameters. This is only slightly slower than (;[\w;]+)*.
1438 // @codingStandardsIgnoreStart Generic.Files.LineLength
1439 $parameters = '(?>;[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+=(?>[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+|"(?>[\0-\x0c\x0e-\x21\x23-\x5b\x5d-\x7f]+|\\\\[\0-\x7f])*"))*(?:;base64)?';
1440 // @codingStandardsIgnoreEnd
1442 if ( !preg_match( "!^data:\s*image/(gif|jpeg|jpg|png)$parameters,!i", $value ) ) {
1443 wfDebug( __METHOD__ . ": Found href to unwhitelisted data: uri "
1444 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1445 return [ 'uploaded-href-unsafe-target-svg', $strippedElement, $attrib, $value ];
1449 # Change href with animate from (http://html5sec.org/#137).
1450 if ( $stripped === 'attributename'
1451 && $strippedElement === 'animate'
1452 && $this->stripXmlNamespace( $value ) == 'href'
1454 wfDebug( __METHOD__ . ": Found animate that might be changing href using from "
1455 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1457 return [ 'uploaded-animate-svg', $strippedElement, $attrib, $value ];
1460 # use set/animate to add event-handler attribute to parent
1461 if ( ( $strippedElement == 'set' || $strippedElement == 'animate' )
1462 && $stripped == 'attributename'
1463 && substr( $value, 0, 2 ) == 'on'
1465 wfDebug( __METHOD__ . ": Found svg setting event-handler attribute with "
1466 . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1468 return [ 'uploaded-setting-event-handler-svg', $strippedElement, $stripped, $value ];
1471 # use set to add href attribute to parent element
1472 if ( $strippedElement == 'set'
1473 && $stripped == 'attributename'
1474 && strpos( $value, 'href' ) !== false
1476 wfDebug( __METHOD__ . ": Found svg setting href attribute '$value' in uploaded file.\n" );
1478 return [ 'uploaded-setting-href-svg' ];
1481 # use set to add a remote / data / script target to an element
1482 if ( $strippedElement == 'set'
1483 && $stripped == 'to'
1484 && preg_match( '!(http|https|data|script):!sim', $value )
1486 wfDebug( __METHOD__ . ": Found svg setting attribute to '$value' in uploaded file.\n" );
1488 return [ 'uploaded-wrong-setting-svg', $value ];
1491 # use handler attribute with remote / data / script
1492 if ( $stripped == 'handler' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1493 wfDebug( __METHOD__ . ": Found svg setting handler with remote/data/script "
1494 . "'$attrib'='$value' in uploaded file.\n" );
1496 return [ 'uploaded-setting-handler-svg', $attrib, $value ];
1499 # use CSS styles to bring in remote code
1500 if ( $stripped == 'style'
1501 && self::checkCssFragment( Sanitizer::normalizeCss( $value ) )
1503 wfDebug( __METHOD__ . ": Found svg setting a style with "
1504 . "remote url '$attrib'='$value' in uploaded file.\n" );
1505 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1508 # Several attributes can include css, css character escaping isn't allowed
1509 $cssAttrs = [ 'font', 'clip-path', 'fill', 'filter', 'marker',
1510 'marker-end', 'marker-mid', 'marker-start', 'mask', 'stroke' ];
1511 if ( in_array( $stripped, $cssAttrs )
1512 && self::checkCssFragment( $value )
1514 wfDebug( __METHOD__ . ": Found svg setting a style with "
1515 . "remote url '$attrib'='$value' in uploaded file.\n" );
1516 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1519 # image filters can pull in url, which could be svg that executes scripts
1520 if ( $strippedElement == 'image'
1521 && $stripped == 'filter'
1522 && preg_match( '!url\s*\(!sim', $value )
1524 wfDebug( __METHOD__ . ": Found image filter with url: "
1525 . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1527 return [ 'uploaded-image-filter-svg', $strippedElement, $stripped, $value ];
1531 return false; // No scripts detected
1535 * Check a block of CSS or CSS fragment for anything that looks like
1536 * it is bringing in remote code.
1537 * @param string $value a string of CSS
1538 * @param bool $propOnly only check css properties (start regex with :)
1539 * @return bool true if the CSS contains an illegal string, false if otherwise
1541 private static function checkCssFragment( $value ) {
1543 # Forbid external stylesheets, for both reliability and to protect viewer's privacy
1544 if ( stripos( $value, '@import' ) !== false ) {
1545 return true;
1548 # We allow @font-face to embed fonts with data: urls, so we snip the string
1549 # 'url' out so this case won't match when we check for urls below
1550 $pattern = '!(@font-face\s*{[^}]*src:)url(\("data:;base64,)!im';
1551 $value = preg_replace( $pattern, '$1$2', $value );
1553 # Check for remote and executable CSS. Unlike in Sanitizer::checkCss, the CSS
1554 # properties filter and accelerator don't seem to be useful for xss in SVG files.
1555 # Expression and -o-link don't seem to work either, but filtering them here in case.
1556 # Additionally, we catch remote urls like url("http:..., url('http:..., url(http:...,
1557 # but not local ones such as url("#..., url('#..., url(#....
1558 if ( preg_match( '!expression
1559 | -o-link\s*:
1560 | -o-link-source\s*:
1561 | -o-replace\s*:!imx', $value ) ) {
1562 return true;
1565 if ( preg_match_all(
1566 "!(\s*(url|image|image-set)\s*\(\s*[\"']?\s*[^#]+.*?\))!sim",
1567 $value,
1568 $matches
1569 ) !== 0
1571 # TODO: redo this in one regex. Until then, url("#whatever") matches the first
1572 foreach ( $matches[1] as $match ) {
1573 if ( !preg_match( "!\s*(url|image|image-set)\s*\(\s*(#|'#|\"#)!im", $match ) ) {
1574 return true;
1579 if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ) {
1580 return true;
1583 return false;
1587 * Divide the element name passed by the xml parser to the callback into URI and prifix.
1588 * @param string $element
1589 * @return array Containing the namespace URI and prefix
1591 private static function splitXmlNamespace( $element ) {
1592 // 'http://www.w3.org/2000/svg:script' -> array( 'http://www.w3.org/2000/svg', 'script' )
1593 $parts = explode( ':', strtolower( $element ) );
1594 $name = array_pop( $parts );
1595 $ns = implode( ':', $parts );
1597 return [ $ns, $name ];
1601 * @param string $name
1602 * @return string
1604 private function stripXmlNamespace( $name ) {
1605 // 'http://www.w3.org/2000/svg:script' -> 'script'
1606 $parts = explode( ':', strtolower( $name ) );
1608 return array_pop( $parts );
1612 * Generic wrapper function for a virus scanner program.
1613 * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
1614 * $wgAntivirusRequired may be used to deny upload if the scan fails.
1616 * @param string $file Pathname to the temporary upload file
1617 * @return mixed False if not virus is found, null if the scan fails or is disabled,
1618 * or a string containing feedback from the virus scanner if a virus was found.
1619 * If textual feedback is missing but a virus was found, this function returns true.
1621 public static function detectVirus( $file ) {
1622 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
1624 if ( !$wgAntivirus ) {
1625 wfDebug( __METHOD__ . ": virus scanner disabled\n" );
1627 return null;
1630 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
1631 wfDebug( __METHOD__ . ": unknown virus scanner: $wgAntivirus\n" );
1632 $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
1633 [ 'virus-badscanner', $wgAntivirus ] );
1635 return wfMessage( 'virus-unknownscanner' )->text() . " $wgAntivirus";
1638 # look up scanner configuration
1639 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
1640 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
1641 $msgPattern = isset( $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ) ?
1642 $wgAntivirusSetup[$wgAntivirus]['messagepattern'] : null;
1644 if ( strpos( $command, "%f" ) === false ) {
1645 # simple pattern: append file to scan
1646 $command .= " " . wfEscapeShellArg( $file );
1647 } else {
1648 # complex pattern: replace "%f" with file to scan
1649 $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
1652 wfDebug( __METHOD__ . ": running virus scan: $command \n" );
1654 # execute virus scanner
1655 $exitCode = false;
1657 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1658 # that does not seem to be worth the pain.
1659 # Ask me (Duesentrieb) about it if it's ever needed.
1660 $output = wfShellExecWithStderr( $command, $exitCode );
1662 # map exit code to AV_xxx constants.
1663 $mappedCode = $exitCode;
1664 if ( $exitCodeMap ) {
1665 if ( isset( $exitCodeMap[$exitCode] ) ) {
1666 $mappedCode = $exitCodeMap[$exitCode];
1667 } elseif ( isset( $exitCodeMap["*"] ) ) {
1668 $mappedCode = $exitCodeMap["*"];
1672 /* NB: AV_NO_VIRUS is 0 but AV_SCAN_FAILED is false,
1673 * so we need the strict equalities === and thus can't use a switch here
1675 if ( $mappedCode === AV_SCAN_FAILED ) {
1676 # scan failed (code was mapped to false by $exitCodeMap)
1677 wfDebug( __METHOD__ . ": failed to scan $file (code $exitCode).\n" );
1679 $output = $wgAntivirusRequired
1680 ? wfMessage( 'virus-scanfailed', [ $exitCode ] )->text()
1681 : null;
1682 } elseif ( $mappedCode === AV_SCAN_ABORTED ) {
1683 # scan failed because filetype is unknown (probably imune)
1684 wfDebug( __METHOD__ . ": unsupported file type $file (code $exitCode).\n" );
1685 $output = null;
1686 } elseif ( $mappedCode === AV_NO_VIRUS ) {
1687 # no virus found
1688 wfDebug( __METHOD__ . ": file passed virus scan.\n" );
1689 $output = false;
1690 } else {
1691 $output = trim( $output );
1693 if ( !$output ) {
1694 $output = true; # if there's no output, return true
1695 } elseif ( $msgPattern ) {
1696 $groups = [];
1697 if ( preg_match( $msgPattern, $output, $groups ) ) {
1698 if ( $groups[1] ) {
1699 $output = $groups[1];
1704 wfDebug( __METHOD__ . ": FOUND VIRUS! scanner feedback: $output \n" );
1707 return $output;
1711 * Check if there's an overwrite conflict and, if so, if restrictions
1712 * forbid this user from performing the upload.
1714 * @param User $user
1716 * @return mixed True on success, array on failure
1718 private function checkOverwrite( $user ) {
1719 // First check whether the local file can be overwritten
1720 $file = $this->getLocalFile();
1721 $file->load( File::READ_LATEST );
1722 if ( $file->exists() ) {
1723 if ( !self::userCanReUpload( $user, $file ) ) {
1724 return [ 'fileexists-forbidden', $file->getName() ];
1725 } else {
1726 return true;
1730 /* Check shared conflicts: if the local file does not exist, but
1731 * wfFindFile finds a file, it exists in a shared repository.
1733 $file = wfFindFile( $this->getTitle(), [ 'latest' => true ] );
1734 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1735 return [ 'fileexists-shared-forbidden', $file->getName() ];
1738 return true;
1742 * Check if a user is the last uploader
1744 * @param User $user
1745 * @param File $img
1746 * @return bool
1748 public static function userCanReUpload( User $user, File $img ) {
1749 if ( $user->isAllowed( 'reupload' ) ) {
1750 return true; // non-conditional
1751 } elseif ( !$user->isAllowed( 'reupload-own' ) ) {
1752 return false;
1755 if ( !( $img instanceof LocalFile ) ) {
1756 return false;
1759 $img->load();
1761 return $user->getId() == $img->getUser( 'id' );
1765 * Helper function that does various existence checks for a file.
1766 * The following checks are performed:
1767 * - The file exists
1768 * - Article with the same name as the file exists
1769 * - File exists with normalized extension
1770 * - The file looks like a thumbnail and the original exists
1772 * @param File $file The File object to check
1773 * @return mixed False if the file does not exists, else an array
1775 public static function getExistsWarning( $file ) {
1776 if ( $file->exists() ) {
1777 return [ 'warning' => 'exists', 'file' => $file ];
1780 if ( $file->getTitle()->getArticleID() ) {
1781 return [ 'warning' => 'page-exists', 'file' => $file ];
1784 if ( strpos( $file->getName(), '.' ) == false ) {
1785 $partname = $file->getName();
1786 $extension = '';
1787 } else {
1788 $n = strrpos( $file->getName(), '.' );
1789 $extension = substr( $file->getName(), $n + 1 );
1790 $partname = substr( $file->getName(), 0, $n );
1792 $normalizedExtension = File::normalizeExtension( $extension );
1794 if ( $normalizedExtension != $extension ) {
1795 // We're not using the normalized form of the extension.
1796 // Normal form is lowercase, using most common of alternate
1797 // extensions (eg 'jpg' rather than 'JPEG').
1799 // Check for another file using the normalized form...
1800 $nt_lc = Title::makeTitle( NS_FILE, "{$partname}.{$normalizedExtension}" );
1801 $file_lc = wfLocalFile( $nt_lc );
1803 if ( $file_lc->exists() ) {
1804 return [
1805 'warning' => 'exists-normalized',
1806 'file' => $file,
1807 'normalizedFile' => $file_lc
1812 // Check for files with the same name but a different extension
1813 $similarFiles = RepoGroup::singleton()->getLocalRepo()->findFilesByPrefix(
1814 "{$partname}.", 1 );
1815 if ( count( $similarFiles ) ) {
1816 return [
1817 'warning' => 'exists-normalized',
1818 'file' => $file,
1819 'normalizedFile' => $similarFiles[0],
1823 if ( self::isThumbName( $file->getName() ) ) {
1824 # Check for filenames like 50px- or 180px-, these are mostly thumbnails
1825 $nt_thb = Title::newFromText(
1826 substr( $partname, strpos( $partname, '-' ) + 1 ) . '.' . $extension,
1827 NS_FILE
1829 $file_thb = wfLocalFile( $nt_thb );
1830 if ( $file_thb->exists() ) {
1831 return [
1832 'warning' => 'thumb',
1833 'file' => $file,
1834 'thumbFile' => $file_thb
1836 } else {
1837 // File does not exist, but we just don't like the name
1838 return [
1839 'warning' => 'thumb-name',
1840 'file' => $file,
1841 'thumbFile' => $file_thb
1846 foreach ( self::getFilenamePrefixBlacklist() as $prefix ) {
1847 if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
1848 return [
1849 'warning' => 'bad-prefix',
1850 'file' => $file,
1851 'prefix' => $prefix
1856 return false;
1860 * Helper function that checks whether the filename looks like a thumbnail
1861 * @param string $filename
1862 * @return bool
1864 public static function isThumbName( $filename ) {
1865 $n = strrpos( $filename, '.' );
1866 $partname = $n ? substr( $filename, 0, $n ) : $filename;
1868 return (
1869 substr( $partname, 3, 3 ) == 'px-' ||
1870 substr( $partname, 2, 3 ) == 'px-'
1871 ) &&
1872 preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) );
1876 * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
1878 * @return array List of prefixes
1880 public static function getFilenamePrefixBlacklist() {
1881 $blacklist = [];
1882 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
1883 if ( !$message->isDisabled() ) {
1884 $lines = explode( "\n", $message->plain() );
1885 foreach ( $lines as $line ) {
1886 // Remove comment lines
1887 $comment = substr( trim( $line ), 0, 1 );
1888 if ( $comment == '#' || $comment == '' ) {
1889 continue;
1891 // Remove additional comments after a prefix
1892 $comment = strpos( $line, '#' );
1893 if ( $comment > 0 ) {
1894 $line = substr( $line, 0, $comment - 1 );
1896 $blacklist[] = trim( $line );
1900 return $blacklist;
1904 * Gets image info about the file just uploaded.
1906 * Also has the effect of setting metadata to be an 'indexed tag name' in
1907 * returned API result if 'metadata' was requested. Oddly, we have to pass
1908 * the "result" object down just so it can do that with the appropriate
1909 * format, presumably.
1911 * @param ApiResult $result
1912 * @return array Image info
1914 public function getImageInfo( $result ) {
1915 $file = $this->getLocalFile();
1916 /** @todo This cries out for refactoring.
1917 * We really want to say $file->getAllInfo(); here.
1918 * Perhaps "info" methods should be moved into files, and the API should
1919 * just wrap them in queries.
1921 if ( $file instanceof UploadStashFile ) {
1922 $imParam = ApiQueryStashImageInfo::getPropertyNames();
1923 $info = ApiQueryStashImageInfo::getInfo( $file, array_flip( $imParam ), $result );
1924 } else {
1925 $imParam = ApiQueryImageInfo::getPropertyNames();
1926 $info = ApiQueryImageInfo::getInfo( $file, array_flip( $imParam ), $result );
1929 return $info;
1933 * @param array $error
1934 * @return Status
1936 public function convertVerifyErrorToStatus( $error ) {
1937 $code = $error['status'];
1938 unset( $code['status'] );
1940 return Status::newFatal( $this->getVerificationErrorCode( $code ), $error );
1944 * Get the MediaWiki maximum uploaded file size for given type of upload, based on
1945 * $wgMaxUploadSize.
1947 * @param null|string $forType
1948 * @return int
1950 public static function getMaxUploadSize( $forType = null ) {
1951 global $wgMaxUploadSize;
1953 if ( is_array( $wgMaxUploadSize ) ) {
1954 if ( !is_null( $forType ) && isset( $wgMaxUploadSize[$forType] ) ) {
1955 return $wgMaxUploadSize[$forType];
1956 } else {
1957 return $wgMaxUploadSize['*'];
1959 } else {
1960 return intval( $wgMaxUploadSize );
1965 * Get the PHP maximum uploaded file size, based on ini settings. If there is no limit or the
1966 * limit can't be guessed, returns a very large number (PHP_INT_MAX).
1968 * @since 1.27
1969 * @return int
1971 public static function getMaxPhpUploadSize() {
1972 $phpMaxFileSize = wfShorthandToInteger(
1973 ini_get( 'upload_max_filesize' ) ?: ini_get( 'hhvm.server.upload.upload_max_file_size' ),
1974 PHP_INT_MAX
1976 $phpMaxPostSize = wfShorthandToInteger(
1977 ini_get( 'post_max_size' ) ?: ini_get( 'hhvm.server.max_post_size' ),
1978 PHP_INT_MAX
1979 ) ?: PHP_INT_MAX;
1980 return min( $phpMaxFileSize, $phpMaxPostSize );
1984 * Get the current status of a chunked upload (used for polling)
1986 * The value will be read from cache.
1988 * @param User $user
1989 * @param string $statusKey
1990 * @return Status[]|bool
1992 public static function getSessionStatus( User $user, $statusKey ) {
1993 $key = wfMemcKey( 'uploadstatus', $user->getId() ?: md5( $user->getName() ), $statusKey );
1995 return ObjectCache::getMainStashInstance()->get( $key );
1999 * Set the current status of a chunked upload (used for polling)
2001 * The value will be set in cache for 1 day
2003 * @param User $user
2004 * @param string $statusKey
2005 * @param array|bool $value
2006 * @return void
2008 public static function setSessionStatus( User $user, $statusKey, $value ) {
2009 $key = wfMemcKey( 'uploadstatus', $user->getId() ?: md5( $user->getName() ), $statusKey );
2011 $cache = ObjectCache::getMainStashInstance();
2012 if ( $value === false ) {
2013 $cache->delete( $key );
2014 } else {
2015 $cache->set( $key, $value, $cache::TTL_DAY );