Merge "Update docs/hooks.txt for ShowSearchHitTitle"
[mediawiki.git] / includes / upload / UploadBase.php
blob96f8638557d5e33f242f302f5c9f2234dd8a5ae1
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, $mStashFile, $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',
57 'WINDOWS-1250',
58 'WINDOWS-1251',
59 'WINDOWS-1252',
60 'WINDOWS-1253',
61 'WINDOWS-1254',
62 'WINDOWS-1255',
63 'WINDOWS-1256',
64 'WINDOWS-1257',
65 'WINDOWS-1258',
68 const SUCCESS = 0;
69 const OK = 0;
70 const EMPTY_FILE = 3;
71 const MIN_LENGTH_PARTNAME = 4;
72 const ILLEGAL_FILENAME = 5;
73 const OVERWRITE_EXISTING_FILE = 7; # Not used anymore; handled by verifyTitlePermissions()
74 const FILETYPE_MISSING = 8;
75 const FILETYPE_BADTYPE = 9;
76 const VERIFICATION_ERROR = 10;
77 const HOOK_ABORTED = 11;
78 const FILE_TOO_LARGE = 12;
79 const WINDOWS_NONASCII_FILENAME = 13;
80 const FILENAME_TOO_LONG = 14;
82 /**
83 * @param int $error
84 * @return string
86 public function getVerificationErrorCode( $error ) {
87 $code_to_status = [
88 self::EMPTY_FILE => 'empty-file',
89 self::FILE_TOO_LARGE => 'file-too-large',
90 self::FILETYPE_MISSING => 'filetype-missing',
91 self::FILETYPE_BADTYPE => 'filetype-banned',
92 self::MIN_LENGTH_PARTNAME => 'filename-tooshort',
93 self::ILLEGAL_FILENAME => 'illegal-filename',
94 self::OVERWRITE_EXISTING_FILE => 'overwrite',
95 self::VERIFICATION_ERROR => 'verification-error',
96 self::HOOK_ABORTED => 'hookaborted',
97 self::WINDOWS_NONASCII_FILENAME => 'windows-nonascii-filename',
98 self::FILENAME_TOO_LONG => 'filename-toolong',
100 if ( isset( $code_to_status[$error] ) ) {
101 return $code_to_status[$error];
104 return 'unknown-error';
108 * Returns true if uploads are enabled.
109 * Can be override by subclasses.
110 * @return bool
112 public static function isEnabled() {
113 global $wgEnableUploads;
115 if ( !$wgEnableUploads ) {
116 return false;
119 # Check php's file_uploads setting
120 return wfIsHHVM() || wfIniGetBool( 'file_uploads' );
124 * Returns true if the user can use this upload module or else a string
125 * identifying the missing permission.
126 * Can be overridden by subclasses.
128 * @param User $user
129 * @return bool|string
131 public static function isAllowed( $user ) {
132 foreach ( [ 'upload', 'edit' ] as $permission ) {
133 if ( !$user->isAllowed( $permission ) ) {
134 return $permission;
138 return true;
142 * Returns true if the user has surpassed the upload rate limit, false otherwise.
144 * @param User $user
145 * @return bool
147 public static function isThrottled( $user ) {
148 return $user->pingLimiter( 'upload' );
151 // Upload handlers. Should probably just be a global.
152 private static $uploadHandlers = [ 'Stash', 'File', 'Url' ];
155 * Create a form of UploadBase depending on wpSourceType and initializes it
157 * @param WebRequest $request
158 * @param string|null $type
159 * @return null|UploadBase
161 public static function createFromRequest( &$request, $type = null ) {
162 $type = $type ? $type : $request->getVal( 'wpSourceType', 'File' );
164 if ( !$type ) {
165 return null;
168 // Get the upload class
169 $type = ucfirst( $type );
171 // Give hooks the chance to handle this request
172 $className = null;
173 Hooks::run( 'UploadCreateFromRequest', [ $type, &$className ] );
174 if ( is_null( $className ) ) {
175 $className = 'UploadFrom' . $type;
176 wfDebug( __METHOD__ . ": class name: $className\n" );
177 if ( !in_array( $type, self::$uploadHandlers ) ) {
178 return null;
182 // Check whether this upload class is enabled
183 if ( !call_user_func( [ $className, 'isEnabled' ] ) ) {
184 return null;
187 // Check whether the request is valid
188 if ( !call_user_func( [ $className, 'isValidRequest' ], $request ) ) {
189 return null;
192 /** @var UploadBase $handler */
193 $handler = new $className;
195 $handler->initializeFromRequest( $request );
197 return $handler;
201 * Check whether a request if valid for this handler
202 * @param WebRequest $request
203 * @return bool
205 public static function isValidRequest( $request ) {
206 return false;
209 public function __construct() {
213 * Returns the upload type. Should be overridden by child classes
215 * @since 1.18
216 * @return string
218 public function getSourceType() {
219 return null;
223 * Initialize the path information
224 * @param string $name The desired destination name
225 * @param string $tempPath The temporary path
226 * @param int $fileSize The file size
227 * @param bool $removeTempFile (false) remove the temporary file?
228 * @throws MWException
230 public function initializePathInfo( $name, $tempPath, $fileSize, $removeTempFile = false ) {
231 $this->mDesiredDestName = $name;
232 if ( FileBackend::isStoragePath( $tempPath ) ) {
233 throw new MWException( __METHOD__ . " given storage path `$tempPath`." );
236 $this->setTempFile( $tempPath, $fileSize );
237 $this->mRemoveTempFile = $removeTempFile;
241 * Initialize from a WebRequest. Override this in a subclass.
243 * @param WebRequest $request
245 abstract public function initializeFromRequest( &$request );
248 * @param string $tempPath File system path to temporary file containing the upload
249 * @param integer $fileSize
251 protected function setTempFile( $tempPath, $fileSize = null ) {
252 $this->mTempPath = $tempPath;
253 $this->mFileSize = $fileSize ?: null;
254 if ( strlen( $this->mTempPath ) && file_exists( $this->mTempPath ) ) {
255 $this->tempFileObj = new TempFSFile( $this->mTempPath );
256 if ( !$fileSize ) {
257 $this->mFileSize = filesize( $this->mTempPath );
259 } else {
260 $this->tempFileObj = null;
265 * Fetch the file. Usually a no-op
266 * @return Status
268 public function fetchFile() {
269 return Status::newGood();
273 * Return true if the file is empty
274 * @return bool
276 public function isEmptyFile() {
277 return empty( $this->mFileSize );
281 * Return the file size
282 * @return int
284 public function getFileSize() {
285 return $this->mFileSize;
289 * Get the base 36 SHA1 of the file
290 * @return string
292 public function getTempFileSha1Base36() {
293 return FSFile::getSha1Base36FromPath( $this->mTempPath );
297 * @param string $srcPath The source path
298 * @return string|bool The real path if it was a virtual URL Returns false on failure
300 function getRealPath( $srcPath ) {
301 $repo = RepoGroup::singleton()->getLocalRepo();
302 if ( $repo->isVirtualUrl( $srcPath ) ) {
303 /** @todo Just make uploads work with storage paths UploadFromStash
304 * loads files via virtual URLs.
306 $tmpFile = $repo->getLocalCopy( $srcPath );
307 if ( $tmpFile ) {
308 $tmpFile->bind( $this ); // keep alive with $this
310 $path = $tmpFile ? $tmpFile->getPath() : false;
311 } else {
312 $path = $srcPath;
315 return $path;
319 * Verify whether the upload is sane.
320 * @return mixed Const self::OK or else an array with error information
322 public function verifyUpload() {
325 * If there was no filename or a zero size given, give up quick.
327 if ( $this->isEmptyFile() ) {
328 return [ 'status' => self::EMPTY_FILE ];
332 * Honor $wgMaxUploadSize
334 $maxSize = self::getMaxUploadSize( $this->getSourceType() );
335 if ( $this->mFileSize > $maxSize ) {
336 return [
337 'status' => self::FILE_TOO_LARGE,
338 'max' => $maxSize,
343 * Look at the contents of the file; if we can recognize the
344 * type but it's corrupt or data of the wrong type, we should
345 * probably not accept it.
347 $verification = $this->verifyFile();
348 if ( $verification !== true ) {
349 return [
350 'status' => self::VERIFICATION_ERROR,
351 'details' => $verification
356 * Make sure this file can be created
358 $result = $this->validateName();
359 if ( $result !== true ) {
360 return $result;
363 $error = '';
364 if ( !Hooks::run( 'UploadVerification',
365 [ $this->mDestName, $this->mTempPath, &$error ], '1.28' )
367 return [ 'status' => self::HOOK_ABORTED, 'error' => $error ];
370 return [ 'status' => self::OK ];
374 * Verify that the name is valid and, if necessary, that we can overwrite
376 * @return mixed True if valid, otherwise and array with 'status'
377 * and other keys
379 public function validateName() {
380 $nt = $this->getTitle();
381 if ( is_null( $nt ) ) {
382 $result = [ 'status' => $this->mTitleError ];
383 if ( $this->mTitleError == self::ILLEGAL_FILENAME ) {
384 $result['filtered'] = $this->mFilteredName;
386 if ( $this->mTitleError == self::FILETYPE_BADTYPE ) {
387 $result['finalExt'] = $this->mFinalExtension;
388 if ( count( $this->mBlackListedExtensions ) ) {
389 $result['blacklistedExt'] = $this->mBlackListedExtensions;
393 return $result;
395 $this->mDestName = $this->getLocalFile()->getName();
397 return true;
401 * Verify the MIME type.
403 * @note Only checks that it is not an evil MIME. The "does it have
404 * correct extension given its MIME type?" check is in verifyFile.
405 * in `verifyFile()` that MIME type and file extension correlate.
406 * @param string $mime Representing the MIME
407 * @return mixed True if the file is verified, an array otherwise
409 protected function verifyMimeType( $mime ) {
410 global $wgVerifyMimeType;
411 if ( $wgVerifyMimeType ) {
412 wfDebug( "mime: <$mime> extension: <{$this->mFinalExtension}>\n" );
413 global $wgMimeTypeBlacklist;
414 if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
415 return [ 'filetype-badmime', $mime ];
418 # Check what Internet Explorer would detect
419 $fp = fopen( $this->mTempPath, 'rb' );
420 $chunk = fread( $fp, 256 );
421 fclose( $fp );
423 $magic = MimeMagic::singleton();
424 $extMime = $magic->guessTypesForExtension( $this->mFinalExtension );
425 $ieTypes = $magic->getIEMimeTypes( $this->mTempPath, $chunk, $extMime );
426 foreach ( $ieTypes as $ieType ) {
427 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
428 return [ 'filetype-bad-ie-mime', $ieType ];
433 return true;
437 * Verifies that it's ok to include the uploaded file
439 * @return mixed True of the file is verified, array otherwise.
441 protected function verifyFile() {
442 global $wgVerifyMimeType, $wgDisableUploadScriptChecks;
444 $status = $this->verifyPartialFile();
445 if ( $status !== true ) {
446 return $status;
449 $mwProps = new MWFileProps( MimeMagic::singleton() );
450 $this->mFileProps = $mwProps->getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
451 $mime = $this->mFileProps['mime'];
453 if ( $wgVerifyMimeType ) {
454 # XXX: Missing extension will be caught by validateName() via getTitle()
455 if ( $this->mFinalExtension != '' && !$this->verifyExtension( $mime, $this->mFinalExtension ) ) {
456 return [ 'filetype-mime-mismatch', $this->mFinalExtension, $mime ];
460 # check for htmlish code and javascript
461 if ( !$wgDisableUploadScriptChecks ) {
462 if ( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
463 $svgStatus = $this->detectScriptInSvg( $this->mTempPath, false );
464 if ( $svgStatus !== false ) {
465 return $svgStatus;
470 $handler = MediaHandler::getHandler( $mime );
471 if ( $handler ) {
472 $handlerStatus = $handler->verifyUpload( $this->mTempPath );
473 if ( !$handlerStatus->isOK() ) {
474 $errors = $handlerStatus->getErrorsArray();
476 return reset( $errors );
480 $error = true;
481 Hooks::run( 'UploadVerifyFile', [ $this, $mime, &$error ] );
482 if ( $error !== true ) {
483 if ( !is_array( $error ) ) {
484 $error = [ $error ];
486 return $error;
489 wfDebug( __METHOD__ . ": all clear; passing.\n" );
491 return true;
495 * A verification routine suitable for partial files
497 * Runs the blacklist checks, but not any checks that may
498 * assume the entire file is present.
500 * @return mixed True for valid or array with error message key.
502 protected function verifyPartialFile() {
503 global $wgAllowJavaUploads, $wgDisableUploadScriptChecks;
505 # getTitle() sets some internal parameters like $this->mFinalExtension
506 $this->getTitle();
508 $mwProps = new MWFileProps( MimeMagic::singleton() );
509 $this->mFileProps = $mwProps->getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
511 # check MIME type, if desired
512 $mime = $this->mFileProps['file-mime'];
513 $status = $this->verifyMimeType( $mime );
514 if ( $status !== true ) {
515 return $status;
518 # check for htmlish code and javascript
519 if ( !$wgDisableUploadScriptChecks ) {
520 if ( self::detectScript( $this->mTempPath, $mime, $this->mFinalExtension ) ) {
521 return [ 'uploadscripted' ];
523 if ( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
524 $svgStatus = $this->detectScriptInSvg( $this->mTempPath, true );
525 if ( $svgStatus !== false ) {
526 return $svgStatus;
531 # Check for Java applets, which if uploaded can bypass cross-site
532 # restrictions.
533 if ( !$wgAllowJavaUploads ) {
534 $this->mJavaDetected = false;
535 $zipStatus = ZipDirectoryReader::read( $this->mTempPath,
536 [ $this, 'zipEntryCallback' ] );
537 if ( !$zipStatus->isOK() ) {
538 $errors = $zipStatus->getErrorsArray();
539 $error = reset( $errors );
540 if ( $error[0] !== 'zip-wrong-format' ) {
541 return $error;
544 if ( $this->mJavaDetected ) {
545 return [ 'uploadjava' ];
549 # Scan the uploaded file for viruses
550 $virus = $this->detectVirus( $this->mTempPath );
551 if ( $virus ) {
552 return [ 'uploadvirus', $virus ];
555 return true;
559 * Callback for ZipDirectoryReader to detect Java class files.
561 * @param array $entry
563 function zipEntryCallback( $entry ) {
564 $names = [ $entry['name'] ];
566 // If there is a null character, cut off the name at it, because JDK's
567 // ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name
568 // were constructed which had ".class\0" followed by a string chosen to
569 // make the hash collide with the truncated name, that file could be
570 // returned in response to a request for the .class file.
571 $nullPos = strpos( $entry['name'], "\000" );
572 if ( $nullPos !== false ) {
573 $names[] = substr( $entry['name'], 0, $nullPos );
576 // If there is a trailing slash in the file name, we have to strip it,
577 // because that's what ZIP_GetEntry() does.
578 if ( preg_grep( '!\.class/?$!', $names ) ) {
579 $this->mJavaDetected = true;
584 * Alias for verifyTitlePermissions. The function was originally
585 * 'verifyPermissions', but that suggests it's checking the user, when it's
586 * really checking the title + user combination.
588 * @param User $user User object to verify the permissions against
589 * @return mixed An array as returned by getUserPermissionsErrors or true
590 * in case the user has proper permissions.
592 public function verifyPermissions( $user ) {
593 return $this->verifyTitlePermissions( $user );
597 * Check whether the user can edit, upload and create the image. This
598 * checks only against the current title; if it returns errors, it may
599 * very well be that another title will not give errors. Therefore
600 * isAllowed() should be called as well for generic is-user-blocked or
601 * can-user-upload checking.
603 * @param User $user User object to verify the permissions against
604 * @return mixed An array as returned by getUserPermissionsErrors or true
605 * in case the user has proper permissions.
607 public function verifyTitlePermissions( $user ) {
609 * If the image is protected, non-sysop users won't be able
610 * to modify it by uploading a new revision.
612 $nt = $this->getTitle();
613 if ( is_null( $nt ) ) {
614 return true;
616 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
617 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
618 if ( !$nt->exists() ) {
619 $permErrorsCreate = $nt->getUserPermissionsErrors( 'create', $user );
620 } else {
621 $permErrorsCreate = [];
623 if ( $permErrors || $permErrorsUpload || $permErrorsCreate ) {
624 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
625 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
627 return $permErrors;
630 $overwriteError = $this->checkOverwrite( $user );
631 if ( $overwriteError !== true ) {
632 return [ $overwriteError ];
635 return true;
639 * Check for non fatal problems with the file.
641 * This should not assume that mTempPath is set.
643 * @return array Array of warnings
645 public function checkWarnings() {
646 global $wgLang;
648 $warnings = [];
650 $localFile = $this->getLocalFile();
651 $localFile->load( File::READ_LATEST );
652 $filename = $localFile->getName();
655 * Check whether the resulting filename is different from the desired one,
656 * but ignore things like ucfirst() and spaces/underscore things
658 $comparableName = str_replace( ' ', '_', $this->mDesiredDestName );
659 $comparableName = Title::capitalize( $comparableName, NS_FILE );
661 if ( $this->mDesiredDestName != $filename && $comparableName != $filename ) {
662 $warnings['badfilename'] = $filename;
665 // Check whether the file extension is on the unwanted list
666 global $wgCheckFileExtensions, $wgFileExtensions;
667 if ( $wgCheckFileExtensions ) {
668 $extensions = array_unique( $wgFileExtensions );
669 if ( !$this->checkFileExtension( $this->mFinalExtension, $extensions ) ) {
670 $warnings['filetype-unwanted-type'] = [ $this->mFinalExtension,
671 $wgLang->commaList( $extensions ), count( $extensions ) ];
675 global $wgUploadSizeWarning;
676 if ( $wgUploadSizeWarning && ( $this->mFileSize > $wgUploadSizeWarning ) ) {
677 $warnings['large-file'] = [ $wgUploadSizeWarning, $this->mFileSize ];
680 if ( $this->mFileSize == 0 ) {
681 $warnings['empty-file'] = true;
684 $hash = $this->getTempFileSha1Base36();
685 $exists = self::getExistsWarning( $localFile );
686 if ( $exists !== false ) {
687 $warnings['exists'] = $exists;
689 // check if file is an exact duplicate of current file version
690 if ( $hash === $localFile->getSha1() ) {
691 $warnings['no-change'] = $localFile;
694 // check if file is an exact duplicate of older versions of this file
695 $history = $localFile->getHistory();
696 foreach ( $history as $oldFile ) {
697 if ( $hash === $oldFile->getSha1() ) {
698 $warnings['duplicate-version'][] = $oldFile;
703 if ( $localFile->wasDeleted() && !$localFile->exists() ) {
704 $warnings['was-deleted'] = $filename;
707 // Check dupes against existing files
708 $dupes = RepoGroup::singleton()->findBySha1( $hash );
709 $title = $this->getTitle();
710 // Remove all matches against self
711 foreach ( $dupes as $key => $dupe ) {
712 if ( $title->equals( $dupe->getTitle() ) ) {
713 unset( $dupes[$key] );
716 if ( $dupes ) {
717 $warnings['duplicate'] = $dupes;
720 // Check dupes against archives
721 $archivedFile = new ArchivedFile( null, 0, '', $hash );
722 if ( $archivedFile->getID() > 0 ) {
723 if ( $archivedFile->userCan( File::DELETED_FILE ) ) {
724 $warnings['duplicate-archive'] = $archivedFile->getName();
725 } else {
726 $warnings['duplicate-archive'] = '';
730 return $warnings;
734 * Really perform the upload. Stores the file in the local repo, watches
735 * if necessary and runs the UploadComplete hook.
737 * @param string $comment
738 * @param string $pageText
739 * @param bool $watch Whether the file page should be added to user's watchlist.
740 * (This doesn't check $user's permissions.)
741 * @param User $user
742 * @param string[] $tags Change tags to add to the log entry and page revision.
743 * (This doesn't check $user's permissions.)
744 * @return Status Indicating the whether the upload succeeded.
746 public function performUpload( $comment, $pageText, $watch, $user, $tags = [] ) {
747 $this->getLocalFile()->load( File::READ_LATEST );
748 $props = $this->mFileProps;
750 $error = null;
751 Hooks::run( 'UploadVerifyUpload', [ $this, $user, $props, $comment, $pageText, &$error ] );
752 if ( $error ) {
753 if ( !is_array( $error ) ) {
754 $error = [ $error ];
756 return call_user_func_array( 'Status::newFatal', $error );
759 $status = $this->getLocalFile()->upload(
760 $this->mTempPath,
761 $comment,
762 $pageText,
763 File::DELETE_SOURCE,
764 $props,
765 false,
766 $user,
767 $tags
770 if ( $status->isGood() ) {
771 if ( $watch ) {
772 WatchAction::doWatch(
773 $this->getLocalFile()->getTitle(),
774 $user,
775 User::IGNORE_USER_RIGHTS
778 // Avoid PHP 7.1 warning of passing $this by reference
779 $uploadBase = $this;
780 Hooks::run( 'UploadComplete', [ &$uploadBase ] );
782 $this->postProcessUpload();
785 return $status;
789 * Perform extra steps after a successful upload.
791 * @since 1.25
793 public function postProcessUpload() {
797 * Returns the title of the file to be uploaded. Sets mTitleError in case
798 * the name was illegal.
800 * @return Title The title of the file or null in case the name was illegal
802 public function getTitle() {
803 if ( $this->mTitle !== false ) {
804 return $this->mTitle;
806 if ( !is_string( $this->mDesiredDestName ) ) {
807 $this->mTitleError = self::ILLEGAL_FILENAME;
808 $this->mTitle = null;
810 return $this->mTitle;
812 /* Assume that if a user specified File:Something.jpg, this is an error
813 * and that the namespace prefix needs to be stripped of.
815 $title = Title::newFromText( $this->mDesiredDestName );
816 if ( $title && $title->getNamespace() == NS_FILE ) {
817 $this->mFilteredName = $title->getDBkey();
818 } else {
819 $this->mFilteredName = $this->mDesiredDestName;
822 # oi_archive_name is max 255 bytes, which include a timestamp and an
823 # exclamation mark, so restrict file name to 240 bytes.
824 if ( strlen( $this->mFilteredName ) > 240 ) {
825 $this->mTitleError = self::FILENAME_TOO_LONG;
826 $this->mTitle = null;
828 return $this->mTitle;
832 * Chop off any directories in the given filename. Then
833 * filter out illegal characters, and try to make a legible name
834 * out of it. We'll strip some silently that Title would die on.
836 $this->mFilteredName = wfStripIllegalFilenameChars( $this->mFilteredName );
837 /* Normalize to title form before we do any further processing */
838 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
839 if ( is_null( $nt ) ) {
840 $this->mTitleError = self::ILLEGAL_FILENAME;
841 $this->mTitle = null;
843 return $this->mTitle;
845 $this->mFilteredName = $nt->getDBkey();
848 * We'll want to blacklist against *any* 'extension', and use
849 * only the final one for the whitelist.
851 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName );
853 if ( count( $ext ) ) {
854 $this->mFinalExtension = trim( $ext[count( $ext ) - 1] );
855 } else {
856 $this->mFinalExtension = '';
858 # No extension, try guessing one
859 $magic = MimeMagic::singleton();
860 $mime = $magic->guessMimeType( $this->mTempPath );
861 if ( $mime !== 'unknown/unknown' ) {
862 # Get a space separated list of extensions
863 $extList = $magic->getExtensionsForType( $mime );
864 if ( $extList ) {
865 # Set the extension to the canonical extension
866 $this->mFinalExtension = strtok( $extList, ' ' );
868 # Fix up the other variables
869 $this->mFilteredName .= ".{$this->mFinalExtension}";
870 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
871 $ext = [ $this->mFinalExtension ];
876 /* Don't allow users to override the blacklist (check file extension) */
877 global $wgCheckFileExtensions, $wgStrictFileExtensions;
878 global $wgFileExtensions, $wgFileBlacklist;
880 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
882 if ( $this->mFinalExtension == '' ) {
883 $this->mTitleError = self::FILETYPE_MISSING;
884 $this->mTitle = null;
886 return $this->mTitle;
887 } elseif ( $blackListedExtensions ||
888 ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
889 !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) )
891 $this->mBlackListedExtensions = $blackListedExtensions;
892 $this->mTitleError = self::FILETYPE_BADTYPE;
893 $this->mTitle = null;
895 return $this->mTitle;
898 // Windows may be broken with special characters, see bug 1780
899 if ( !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() )
900 && !RepoGroup::singleton()->getLocalRepo()->backendSupportsUnicodePaths()
902 $this->mTitleError = self::WINDOWS_NONASCII_FILENAME;
903 $this->mTitle = null;
905 return $this->mTitle;
908 # If there was more than one "extension", reassemble the base
909 # filename to prevent bogus complaints about length
910 if ( count( $ext ) > 1 ) {
911 $iterations = count( $ext ) - 1;
912 for ( $i = 0; $i < $iterations; $i++ ) {
913 $partname .= '.' . $ext[$i];
917 if ( strlen( $partname ) < 1 ) {
918 $this->mTitleError = self::MIN_LENGTH_PARTNAME;
919 $this->mTitle = null;
921 return $this->mTitle;
924 $this->mTitle = $nt;
926 return $this->mTitle;
930 * Return the local file and initializes if necessary.
932 * @return LocalFile|null
934 public function getLocalFile() {
935 if ( is_null( $this->mLocalFile ) ) {
936 $nt = $this->getTitle();
937 $this->mLocalFile = is_null( $nt ) ? null : wfLocalFile( $nt );
940 return $this->mLocalFile;
944 * @return UploadStashFile|null
946 public function getStashFile() {
947 return $this->mStashFile;
951 * Like stashFile(), but respects extensions' wishes to prevent the stashing. verifyUpload() must
952 * be called before calling this method (unless $isPartial is true).
954 * Upload stash exceptions are also caught and converted to an error status.
956 * @since 1.28
957 * @param User $user
958 * @param bool $isPartial Pass `true` if this is a part of a chunked upload (not a complete file).
959 * @return Status If successful, value is an UploadStashFile instance
961 public function tryStashFile( User $user, $isPartial = false ) {
962 if ( !$isPartial ) {
963 $error = $this->runUploadStashFileHook( $user );
964 if ( $error ) {
965 return call_user_func_array( 'Status::newFatal', $error );
968 try {
969 $file = $this->doStashFile( $user );
970 return Status::newGood( $file );
971 } catch ( UploadStashException $e ) {
972 return Status::newFatal( 'uploadstash-exception', get_class( $e ), $e->getMessage() );
977 * @param User $user
978 * @return array|null Error message and parameters, null if there's no error
980 protected function runUploadStashFileHook( User $user ) {
981 $props = $this->mFileProps;
982 $error = null;
983 Hooks::run( 'UploadStashFile', [ $this, $user, $props, &$error ] );
984 if ( $error ) {
985 if ( !is_array( $error ) ) {
986 $error = [ $error ];
989 return $error;
993 * If the user does not supply all necessary information in the first upload
994 * form submission (either by accident or by design) then we may want to
995 * stash the file temporarily, get more information, and publish the file
996 * later.
998 * This method will stash a file in a temporary directory for later
999 * processing, and save the necessary descriptive info into the database.
1000 * This method returns the file object, which also has a 'fileKey' property
1001 * which can be passed through a form or API request to find this stashed
1002 * file again.
1004 * @deprecated since 1.28 Use tryStashFile() instead
1005 * @param User $user
1006 * @return UploadStashFile Stashed file
1007 * @throws UploadStashBadPathException
1008 * @throws UploadStashFileException
1009 * @throws UploadStashNotLoggedInException
1011 public function stashFile( User $user = null ) {
1012 return $this->doStashFile( $user );
1016 * Implementation for stashFile() and tryStashFile().
1018 * @param User $user
1019 * @return UploadStashFile Stashed file
1021 protected function doStashFile( User $user = null ) {
1022 $stash = RepoGroup::singleton()->getLocalRepo()->getUploadStash( $user );
1023 $file = $stash->stashFile( $this->mTempPath, $this->getSourceType() );
1024 $this->mStashFile = $file;
1026 return $file;
1030 * Stash a file in a temporary directory, returning a key which can be used
1031 * to find the file again. See stashFile().
1033 * @deprecated since 1.28
1034 * @return string File key
1036 public function stashFileGetKey() {
1037 wfDeprecated( __METHOD__, '1.28' );
1038 return $this->doStashFile()->getFileKey();
1042 * alias for stashFileGetKey, for backwards compatibility
1044 * @deprecated since 1.28
1045 * @return string File key
1047 public function stashSession() {
1048 wfDeprecated( __METHOD__, '1.28' );
1049 return $this->doStashFile()->getFileKey();
1053 * If we've modified the upload file we need to manually remove it
1054 * on exit to clean up.
1056 public function cleanupTempFile() {
1057 if ( $this->mRemoveTempFile && $this->tempFileObj ) {
1058 // Delete when all relevant TempFSFile handles go out of scope
1059 wfDebug( __METHOD__ . ": Marked temporary file '{$this->mTempPath}' for removal\n" );
1060 $this->tempFileObj->autocollect();
1064 public function getTempPath() {
1065 return $this->mTempPath;
1069 * Split a file into a base name and all dot-delimited 'extensions'
1070 * on the end. Some web server configurations will fall back to
1071 * earlier pseudo-'extensions' to determine type and execute
1072 * scripts, so the blacklist needs to check them all.
1074 * @param string $filename
1075 * @return array
1077 public static function splitExtensions( $filename ) {
1078 $bits = explode( '.', $filename );
1079 $basename = array_shift( $bits );
1081 return [ $basename, $bits ];
1085 * Perform case-insensitive match against a list of file extensions.
1086 * Returns true if the extension is in the list.
1088 * @param string $ext
1089 * @param array $list
1090 * @return bool
1092 public static function checkFileExtension( $ext, $list ) {
1093 return in_array( strtolower( $ext ), $list );
1097 * Perform case-insensitive match against a list of file extensions.
1098 * Returns an array of matching extensions.
1100 * @param array $ext
1101 * @param array $list
1102 * @return bool
1104 public static function checkFileExtensionList( $ext, $list ) {
1105 return array_intersect( array_map( 'strtolower', $ext ), $list );
1109 * Checks if the MIME type of the uploaded file matches the file extension.
1111 * @param string $mime The MIME type of the uploaded file
1112 * @param string $extension The filename extension that the file is to be served with
1113 * @return bool
1115 public static function verifyExtension( $mime, $extension ) {
1116 $magic = MimeMagic::singleton();
1118 if ( !$mime || $mime == 'unknown' || $mime == 'unknown/unknown' ) {
1119 if ( !$magic->isRecognizableExtension( $extension ) ) {
1120 wfDebug( __METHOD__ . ": passing file with unknown detected mime type; " .
1121 "unrecognized extension '$extension', can't verify\n" );
1123 return true;
1124 } else {
1125 wfDebug( __METHOD__ . ": rejecting file with unknown detected mime type; " .
1126 "recognized extension '$extension', so probably invalid file\n" );
1128 return false;
1132 $match = $magic->isMatchingExtension( $extension, $mime );
1134 if ( $match === null ) {
1135 if ( $magic->getTypesForExtension( $extension ) !== null ) {
1136 wfDebug( __METHOD__ . ": No extension known for $mime, but we know a mime for $extension\n" );
1138 return false;
1139 } else {
1140 wfDebug( __METHOD__ . ": no file extension known for mime type $mime, passing file\n" );
1142 return true;
1144 } elseif ( $match === true ) {
1145 wfDebug( __METHOD__ . ": mime type $mime matches extension $extension, passing file\n" );
1147 /** @todo If it's a bitmap, make sure PHP or ImageMagick resp. can handle it! */
1148 return true;
1149 } else {
1150 wfDebug( __METHOD__
1151 . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
1153 return false;
1158 * Heuristic for detecting files that *could* contain JavaScript instructions or
1159 * things that may look like HTML to a browser and are thus
1160 * potentially harmful. The present implementation will produce false
1161 * positives in some situations.
1163 * @param string $file Pathname to the temporary upload file
1164 * @param string $mime The MIME type of the file
1165 * @param string $extension The extension of the file
1166 * @return bool True if the file contains something looking like embedded scripts
1168 public static function detectScript( $file, $mime, $extension ) {
1169 global $wgAllowTitlesInSVG;
1171 # ugly hack: for text files, always look at the entire file.
1172 # For binary field, just check the first K.
1174 if ( strpos( $mime, 'text/' ) === 0 ) {
1175 $chunk = file_get_contents( $file );
1176 } else {
1177 $fp = fopen( $file, 'rb' );
1178 $chunk = fread( $fp, 1024 );
1179 fclose( $fp );
1182 $chunk = strtolower( $chunk );
1184 if ( !$chunk ) {
1185 return false;
1188 # decode from UTF-16 if needed (could be used for obfuscation).
1189 if ( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
1190 $enc = 'UTF-16BE';
1191 } elseif ( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
1192 $enc = 'UTF-16LE';
1193 } else {
1194 $enc = null;
1197 if ( $enc ) {
1198 $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
1201 $chunk = trim( $chunk );
1203 /** @todo FIXME: Convert from UTF-16 if necessary! */
1204 wfDebug( __METHOD__ . ": checking for embedded scripts and HTML stuff\n" );
1206 # check for HTML doctype
1207 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
1208 return true;
1211 // Some browsers will interpret obscure xml encodings as UTF-8, while
1212 // PHP/expat will interpret the given encoding in the xml declaration (bug 47304)
1213 if ( $extension == 'svg' || strpos( $mime, 'image/svg' ) === 0 ) {
1214 if ( self::checkXMLEncodingMissmatch( $file ) ) {
1215 return true;
1220 * Internet Explorer for Windows performs some really stupid file type
1221 * autodetection which can cause it to interpret valid image files as HTML
1222 * and potentially execute JavaScript, creating a cross-site scripting
1223 * attack vectors.
1225 * Apple's Safari browser also performs some unsafe file type autodetection
1226 * which can cause legitimate files to be interpreted as HTML if the
1227 * web server is not correctly configured to send the right content-type
1228 * (or if you're really uploading plain text and octet streams!)
1230 * Returns true if IE is likely to mistake the given file for HTML.
1231 * Also returns true if Safari would mistake the given file for HTML
1232 * when served with a generic content-type.
1234 $tags = [
1235 '<a href',
1236 '<body',
1237 '<head',
1238 '<html', # also in safari
1239 '<img',
1240 '<pre',
1241 '<script', # also in safari
1242 '<table'
1245 if ( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
1246 $tags[] = '<title';
1249 foreach ( $tags as $tag ) {
1250 if ( false !== strpos( $chunk, $tag ) ) {
1251 wfDebug( __METHOD__ . ": found something that may make it be mistaken for html: $tag\n" );
1253 return true;
1258 * look for JavaScript
1261 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
1262 $chunk = Sanitizer::decodeCharReferences( $chunk );
1264 # look for script-types
1265 if ( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
1266 wfDebug( __METHOD__ . ": found script types\n" );
1268 return true;
1271 # look for html-style script-urls
1272 if ( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1273 wfDebug( __METHOD__ . ": found html-style script urls\n" );
1275 return true;
1278 # look for css-style script-urls
1279 if ( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1280 wfDebug( __METHOD__ . ": found css-style script urls\n" );
1282 return true;
1285 wfDebug( __METHOD__ . ": no scripts found\n" );
1287 return false;
1291 * Check a whitelist of xml encodings that are known not to be interpreted differently
1292 * by the server's xml parser (expat) and some common browsers.
1294 * @param string $file Pathname to the temporary upload file
1295 * @return bool True if the file contains an encoding that could be misinterpreted
1297 public static function checkXMLEncodingMissmatch( $file ) {
1298 global $wgSVGMetadataCutoff;
1299 $contents = file_get_contents( $file, false, null, -1, $wgSVGMetadataCutoff );
1300 $encodingRegex = '!encoding[ \t\n\r]*=[ \t\n\r]*[\'"](.*?)[\'"]!si';
1302 if ( preg_match( "!<\?xml\b(.*?)\?>!si", $contents, $matches ) ) {
1303 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1304 && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1306 wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1308 return true;
1310 } elseif ( preg_match( "!<\?xml\b!si", $contents ) ) {
1311 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1312 // bytes. There shouldn't be a legitimate reason for this to happen.
1313 wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1315 return true;
1316 } elseif ( substr( $contents, 0, 4 ) == "\x4C\x6F\xA7\x94" ) {
1317 // EBCDIC encoded XML
1318 wfDebug( __METHOD__ . ": EBCDIC Encoded XML\n" );
1320 return true;
1323 // It's possible the file is encoded with multi-byte encoding, so re-encode attempt to
1324 // detect the encoding in case is specifies an encoding not whitelisted in self::$safeXmlEncodings
1325 $attemptEncodings = [ 'UTF-16', 'UTF-16BE', 'UTF-32', 'UTF-32BE' ];
1326 foreach ( $attemptEncodings as $encoding ) {
1327 MediaWiki\suppressWarnings();
1328 $str = iconv( $encoding, 'UTF-8', $contents );
1329 MediaWiki\restoreWarnings();
1330 if ( $str != '' && preg_match( "!<\?xml\b(.*?)\?>!si", $str, $matches ) ) {
1331 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1332 && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1334 wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1336 return true;
1338 } elseif ( $str != '' && preg_match( "!<\?xml\b!si", $str ) ) {
1339 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1340 // bytes. There shouldn't be a legitimate reason for this to happen.
1341 wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1343 return true;
1347 return false;
1351 * @param string $filename
1352 * @param bool $partial
1353 * @return mixed False of the file is verified (does not contain scripts), array otherwise.
1355 protected function detectScriptInSvg( $filename, $partial ) {
1356 $this->mSVGNSError = false;
1357 $check = new XmlTypeCheck(
1358 $filename,
1359 [ $this, 'checkSvgScriptCallback' ],
1360 true,
1361 [ 'processing_instruction_handler' => 'UploadBase::checkSvgPICallback' ]
1363 if ( $check->wellFormed !== true ) {
1364 // Invalid xml (bug 58553)
1365 // But only when non-partial (bug 65724)
1366 return $partial ? false : [ 'uploadinvalidxml' ];
1367 } elseif ( $check->filterMatch ) {
1368 if ( $this->mSVGNSError ) {
1369 return [ 'uploadscriptednamespace', $this->mSVGNSError ];
1372 return $check->filterMatchType;
1375 return false;
1379 * Callback to filter SVG Processing Instructions.
1380 * @param string $target Processing instruction name
1381 * @param string $data Processing instruction attribute and value
1382 * @return bool (true if the filter identified something bad)
1384 public static function checkSvgPICallback( $target, $data ) {
1385 // Don't allow external stylesheets (bug 57550)
1386 if ( preg_match( '/xml-stylesheet/i', $target ) ) {
1387 return [ 'upload-scripted-pi-callback' ];
1390 return false;
1394 * @todo Replace this with a whitelist filter!
1395 * @param string $element
1396 * @param array $attribs
1397 * @return bool
1399 public function checkSvgScriptCallback( $element, $attribs, $data = null ) {
1401 list( $namespace, $strippedElement ) = $this->splitXmlNamespace( $element );
1403 // We specifically don't include:
1404 // http://www.w3.org/1999/xhtml (bug 60771)
1405 static $validNamespaces = [
1407 'adobe:ns:meta/',
1408 'http://creativecommons.org/ns#',
1409 'http://inkscape.sourceforge.net/dtd/sodipodi-0.dtd',
1410 'http://ns.adobe.com/adobeillustrator/10.0/',
1411 'http://ns.adobe.com/adobesvgviewerextensions/3.0/',
1412 'http://ns.adobe.com/extensibility/1.0/',
1413 'http://ns.adobe.com/flows/1.0/',
1414 'http://ns.adobe.com/illustrator/1.0/',
1415 'http://ns.adobe.com/imagereplacement/1.0/',
1416 'http://ns.adobe.com/pdf/1.3/',
1417 'http://ns.adobe.com/photoshop/1.0/',
1418 'http://ns.adobe.com/saveforweb/1.0/',
1419 'http://ns.adobe.com/variables/1.0/',
1420 'http://ns.adobe.com/xap/1.0/',
1421 'http://ns.adobe.com/xap/1.0/g/',
1422 'http://ns.adobe.com/xap/1.0/g/img/',
1423 'http://ns.adobe.com/xap/1.0/mm/',
1424 'http://ns.adobe.com/xap/1.0/rights/',
1425 'http://ns.adobe.com/xap/1.0/stype/dimensions#',
1426 'http://ns.adobe.com/xap/1.0/stype/font#',
1427 'http://ns.adobe.com/xap/1.0/stype/manifestitem#',
1428 'http://ns.adobe.com/xap/1.0/stype/resourceevent#',
1429 'http://ns.adobe.com/xap/1.0/stype/resourceref#',
1430 'http://ns.adobe.com/xap/1.0/t/pg/',
1431 'http://purl.org/dc/elements/1.1/',
1432 'http://purl.org/dc/elements/1.1',
1433 'http://schemas.microsoft.com/visio/2003/svgextensions/',
1434 'http://sodipodi.sourceforge.net/dtd/sodipodi-0.dtd',
1435 'http://taptrix.com/inkpad/svg_extensions',
1436 'http://web.resource.org/cc/',
1437 'http://www.freesoftware.fsf.org/bkchem/cdml',
1438 'http://www.inkscape.org/namespaces/inkscape',
1439 'http://www.opengis.net/gml',
1440 'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
1441 'http://www.w3.org/2000/svg',
1442 'http://www.w3.org/tr/rec-rdf-syntax/',
1445 // Inkscape mangles namespace definitions created by Adobe Illustrator.
1446 // This is nasty but harmless. (T144827)
1447 $isBuggyInkscape = preg_match( '/^&(#38;)*ns_[a-z_]+;$/', $namespace );
1449 if ( !( $isBuggyInkscape || in_array( $namespace, $validNamespaces ) ) ) {
1450 wfDebug( __METHOD__ . ": Non-svg namespace '$namespace' in uploaded file.\n" );
1451 /** @todo Return a status object to a closure in XmlTypeCheck, for MW1.21+ */
1452 $this->mSVGNSError = $namespace;
1454 return true;
1458 * check for elements that can contain javascript
1460 if ( $strippedElement == 'script' ) {
1461 wfDebug( __METHOD__ . ": Found script element '$element' in uploaded file.\n" );
1463 return [ 'uploaded-script-svg', $strippedElement ];
1466 # e.g., <svg xmlns="http://www.w3.org/2000/svg">
1467 # <handler xmlns:ev="http://www.w3.org/2001/xml-events" ev:event="load">alert(1)</handler> </svg>
1468 if ( $strippedElement == 'handler' ) {
1469 wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1471 return [ 'uploaded-script-svg', $strippedElement ];
1474 # SVG reported in Feb '12 that used xml:stylesheet to generate javascript block
1475 if ( $strippedElement == 'stylesheet' ) {
1476 wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1478 return [ 'uploaded-script-svg', $strippedElement ];
1481 # Block iframes, in case they pass the namespace check
1482 if ( $strippedElement == 'iframe' ) {
1483 wfDebug( __METHOD__ . ": iframe in uploaded file.\n" );
1485 return [ 'uploaded-script-svg', $strippedElement ];
1488 # Check <style> css
1489 if ( $strippedElement == 'style'
1490 && self::checkCssFragment( Sanitizer::normalizeCss( $data ) )
1492 wfDebug( __METHOD__ . ": hostile css in style element.\n" );
1493 return [ 'uploaded-hostile-svg' ];
1496 foreach ( $attribs as $attrib => $value ) {
1497 $stripped = $this->stripXmlNamespace( $attrib );
1498 $value = strtolower( $value );
1500 if ( substr( $stripped, 0, 2 ) == 'on' ) {
1501 wfDebug( __METHOD__
1502 . ": Found event-handler attribute '$attrib'='$value' in uploaded file.\n" );
1504 return [ 'uploaded-event-handler-on-svg', $attrib, $value ];
1507 # Do not allow relative links, or unsafe url schemas.
1508 # For <a> tags, only data:, http: and https: and same-document
1509 # fragment links are allowed. For all other tags, only data:
1510 # and fragment are allowed.
1511 if ( $stripped == 'href'
1512 && $value !== ''
1513 && strpos( $value, 'data:' ) !== 0
1514 && strpos( $value, '#' ) !== 0
1516 if ( !( $strippedElement === 'a'
1517 && preg_match( '!^https?://!i', $value ) )
1519 wfDebug( __METHOD__ . ": Found href attribute <$strippedElement "
1520 . "'$attrib'='$value' in uploaded file.\n" );
1522 return [ 'uploaded-href-attribute-svg', $strippedElement, $attrib, $value ];
1526 # only allow data: targets that should be safe. This prevents vectors like,
1527 # image/svg, text/xml, application/xml, and text/html, which can contain scripts
1528 if ( $stripped == 'href' && strncasecmp( 'data:', $value, 5 ) === 0 ) {
1529 // rfc2397 parameters. This is only slightly slower than (;[\w;]+)*.
1530 // @codingStandardsIgnoreStart Generic.Files.LineLength
1531 $parameters = '(?>;[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+=(?>[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+|"(?>[\0-\x0c\x0e-\x21\x23-\x5b\x5d-\x7f]+|\\\\[\0-\x7f])*"))*(?:;base64)?';
1532 // @codingStandardsIgnoreEnd
1534 if ( !preg_match( "!^data:\s*image/(gif|jpeg|jpg|png)$parameters,!i", $value ) ) {
1535 wfDebug( __METHOD__ . ": Found href to unwhitelisted data: uri "
1536 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1537 return [ 'uploaded-href-unsafe-target-svg', $strippedElement, $attrib, $value ];
1541 # Change href with animate from (http://html5sec.org/#137).
1542 if ( $stripped === 'attributename'
1543 && $strippedElement === 'animate'
1544 && $this->stripXmlNamespace( $value ) == 'href'
1546 wfDebug( __METHOD__ . ": Found animate that might be changing href using from "
1547 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1549 return [ 'uploaded-animate-svg', $strippedElement, $attrib, $value ];
1552 # use set/animate to add event-handler attribute to parent
1553 if ( ( $strippedElement == 'set' || $strippedElement == 'animate' )
1554 && $stripped == 'attributename'
1555 && substr( $value, 0, 2 ) == 'on'
1557 wfDebug( __METHOD__ . ": Found svg setting event-handler attribute with "
1558 . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1560 return [ 'uploaded-setting-event-handler-svg', $strippedElement, $stripped, $value ];
1563 # use set to add href attribute to parent element
1564 if ( $strippedElement == 'set'
1565 && $stripped == 'attributename'
1566 && strpos( $value, 'href' ) !== false
1568 wfDebug( __METHOD__ . ": Found svg setting href attribute '$value' in uploaded file.\n" );
1570 return [ 'uploaded-setting-href-svg' ];
1573 # use set to add a remote / data / script target to an element
1574 if ( $strippedElement == 'set'
1575 && $stripped == 'to'
1576 && preg_match( '!(http|https|data|script):!sim', $value )
1578 wfDebug( __METHOD__ . ": Found svg setting attribute to '$value' in uploaded file.\n" );
1580 return [ 'uploaded-wrong-setting-svg', $value ];
1583 # use handler attribute with remote / data / script
1584 if ( $stripped == 'handler' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1585 wfDebug( __METHOD__ . ": Found svg setting handler with remote/data/script "
1586 . "'$attrib'='$value' in uploaded file.\n" );
1588 return [ 'uploaded-setting-handler-svg', $attrib, $value ];
1591 # use CSS styles to bring in remote code
1592 if ( $stripped == 'style'
1593 && self::checkCssFragment( Sanitizer::normalizeCss( $value ) )
1595 wfDebug( __METHOD__ . ": Found svg setting a style with "
1596 . "remote url '$attrib'='$value' in uploaded file.\n" );
1597 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1600 # Several attributes can include css, css character escaping isn't allowed
1601 $cssAttrs = [ 'font', 'clip-path', 'fill', 'filter', 'marker',
1602 'marker-end', 'marker-mid', 'marker-start', 'mask', 'stroke' ];
1603 if ( in_array( $stripped, $cssAttrs )
1604 && self::checkCssFragment( $value )
1606 wfDebug( __METHOD__ . ": Found svg setting a style with "
1607 . "remote url '$attrib'='$value' in uploaded file.\n" );
1608 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1611 # image filters can pull in url, which could be svg that executes scripts
1612 if ( $strippedElement == 'image'
1613 && $stripped == 'filter'
1614 && preg_match( '!url\s*\(!sim', $value )
1616 wfDebug( __METHOD__ . ": Found image filter with url: "
1617 . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1619 return [ 'uploaded-image-filter-svg', $strippedElement, $stripped, $value ];
1623 return false; // No scripts detected
1627 * Check a block of CSS or CSS fragment for anything that looks like
1628 * it is bringing in remote code.
1629 * @param string $value a string of CSS
1630 * @param bool $propOnly only check css properties (start regex with :)
1631 * @return bool true if the CSS contains an illegal string, false if otherwise
1633 private static function checkCssFragment( $value ) {
1635 # Forbid external stylesheets, for both reliability and to protect viewer's privacy
1636 if ( stripos( $value, '@import' ) !== false ) {
1637 return true;
1640 # We allow @font-face to embed fonts with data: urls, so we snip the string
1641 # 'url' out so this case won't match when we check for urls below
1642 $pattern = '!(@font-face\s*{[^}]*src:)url(\("data:;base64,)!im';
1643 $value = preg_replace( $pattern, '$1$2', $value );
1645 # Check for remote and executable CSS. Unlike in Sanitizer::checkCss, the CSS
1646 # properties filter and accelerator don't seem to be useful for xss in SVG files.
1647 # Expression and -o-link don't seem to work either, but filtering them here in case.
1648 # Additionally, we catch remote urls like url("http:..., url('http:..., url(http:...,
1649 # but not local ones such as url("#..., url('#..., url(#....
1650 if ( preg_match( '!expression
1651 | -o-link\s*:
1652 | -o-link-source\s*:
1653 | -o-replace\s*:!imx', $value ) ) {
1654 return true;
1657 if ( preg_match_all(
1658 "!(\s*(url|image|image-set)\s*\(\s*[\"']?\s*[^#]+.*?\))!sim",
1659 $value,
1660 $matches
1661 ) !== 0
1663 # TODO: redo this in one regex. Until then, url("#whatever") matches the first
1664 foreach ( $matches[1] as $match ) {
1665 if ( !preg_match( "!\s*(url|image|image-set)\s*\(\s*(#|'#|\"#)!im", $match ) ) {
1666 return true;
1671 if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ) {
1672 return true;
1675 return false;
1679 * Divide the element name passed by the xml parser to the callback into URI and prifix.
1680 * @param string $element
1681 * @return array Containing the namespace URI and prefix
1683 private static function splitXmlNamespace( $element ) {
1684 // 'http://www.w3.org/2000/svg:script' -> [ 'http://www.w3.org/2000/svg', 'script' ]
1685 $parts = explode( ':', strtolower( $element ) );
1686 $name = array_pop( $parts );
1687 $ns = implode( ':', $parts );
1689 return [ $ns, $name ];
1693 * @param string $name
1694 * @return string
1696 private function stripXmlNamespace( $name ) {
1697 // 'http://www.w3.org/2000/svg:script' -> 'script'
1698 $parts = explode( ':', strtolower( $name ) );
1700 return array_pop( $parts );
1704 * Generic wrapper function for a virus scanner program.
1705 * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
1706 * $wgAntivirusRequired may be used to deny upload if the scan fails.
1708 * @param string $file Pathname to the temporary upload file
1709 * @return mixed False if not virus is found, null if the scan fails or is disabled,
1710 * or a string containing feedback from the virus scanner if a virus was found.
1711 * If textual feedback is missing but a virus was found, this function returns true.
1713 public static function detectVirus( $file ) {
1714 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
1716 if ( !$wgAntivirus ) {
1717 wfDebug( __METHOD__ . ": virus scanner disabled\n" );
1719 return null;
1722 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
1723 wfDebug( __METHOD__ . ": unknown virus scanner: $wgAntivirus\n" );
1724 $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
1725 [ 'virus-badscanner', $wgAntivirus ] );
1727 return wfMessage( 'virus-unknownscanner' )->text() . " $wgAntivirus";
1730 # look up scanner configuration
1731 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
1732 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
1733 $msgPattern = isset( $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ) ?
1734 $wgAntivirusSetup[$wgAntivirus]['messagepattern'] : null;
1736 if ( strpos( $command, "%f" ) === false ) {
1737 # simple pattern: append file to scan
1738 $command .= " " . wfEscapeShellArg( $file );
1739 } else {
1740 # complex pattern: replace "%f" with file to scan
1741 $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
1744 wfDebug( __METHOD__ . ": running virus scan: $command \n" );
1746 # execute virus scanner
1747 $exitCode = false;
1749 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1750 # that does not seem to be worth the pain.
1751 # Ask me (Duesentrieb) about it if it's ever needed.
1752 $output = wfShellExecWithStderr( $command, $exitCode );
1754 # map exit code to AV_xxx constants.
1755 $mappedCode = $exitCode;
1756 if ( $exitCodeMap ) {
1757 if ( isset( $exitCodeMap[$exitCode] ) ) {
1758 $mappedCode = $exitCodeMap[$exitCode];
1759 } elseif ( isset( $exitCodeMap["*"] ) ) {
1760 $mappedCode = $exitCodeMap["*"];
1764 /* NB: AV_NO_VIRUS is 0 but AV_SCAN_FAILED is false,
1765 * so we need the strict equalities === and thus can't use a switch here
1767 if ( $mappedCode === AV_SCAN_FAILED ) {
1768 # scan failed (code was mapped to false by $exitCodeMap)
1769 wfDebug( __METHOD__ . ": failed to scan $file (code $exitCode).\n" );
1771 $output = $wgAntivirusRequired
1772 ? wfMessage( 'virus-scanfailed', [ $exitCode ] )->text()
1773 : null;
1774 } elseif ( $mappedCode === AV_SCAN_ABORTED ) {
1775 # scan failed because filetype is unknown (probably imune)
1776 wfDebug( __METHOD__ . ": unsupported file type $file (code $exitCode).\n" );
1777 $output = null;
1778 } elseif ( $mappedCode === AV_NO_VIRUS ) {
1779 # no virus found
1780 wfDebug( __METHOD__ . ": file passed virus scan.\n" );
1781 $output = false;
1782 } else {
1783 $output = trim( $output );
1785 if ( !$output ) {
1786 $output = true; # if there's no output, return true
1787 } elseif ( $msgPattern ) {
1788 $groups = [];
1789 if ( preg_match( $msgPattern, $output, $groups ) ) {
1790 if ( $groups[1] ) {
1791 $output = $groups[1];
1796 wfDebug( __METHOD__ . ": FOUND VIRUS! scanner feedback: $output \n" );
1799 return $output;
1803 * Check if there's an overwrite conflict and, if so, if restrictions
1804 * forbid this user from performing the upload.
1806 * @param User $user
1808 * @return mixed True on success, array on failure
1810 private function checkOverwrite( $user ) {
1811 // First check whether the local file can be overwritten
1812 $file = $this->getLocalFile();
1813 $file->load( File::READ_LATEST );
1814 if ( $file->exists() ) {
1815 if ( !self::userCanReUpload( $user, $file ) ) {
1816 return [ 'fileexists-forbidden', $file->getName() ];
1817 } else {
1818 return true;
1822 /* Check shared conflicts: if the local file does not exist, but
1823 * wfFindFile finds a file, it exists in a shared repository.
1825 $file = wfFindFile( $this->getTitle(), [ 'latest' => true ] );
1826 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1827 return [ 'fileexists-shared-forbidden', $file->getName() ];
1830 return true;
1834 * Check if a user is the last uploader
1836 * @param User $user
1837 * @param File $img
1838 * @return bool
1840 public static function userCanReUpload( User $user, File $img ) {
1841 if ( $user->isAllowed( 'reupload' ) ) {
1842 return true; // non-conditional
1843 } elseif ( !$user->isAllowed( 'reupload-own' ) ) {
1844 return false;
1847 if ( !( $img instanceof LocalFile ) ) {
1848 return false;
1851 $img->load();
1853 return $user->getId() == $img->getUser( 'id' );
1857 * Helper function that does various existence checks for a file.
1858 * The following checks are performed:
1859 * - The file exists
1860 * - Article with the same name as the file exists
1861 * - File exists with normalized extension
1862 * - The file looks like a thumbnail and the original exists
1864 * @param File $file The File object to check
1865 * @return mixed False if the file does not exists, else an array
1867 public static function getExistsWarning( $file ) {
1868 if ( $file->exists() ) {
1869 return [ 'warning' => 'exists', 'file' => $file ];
1872 if ( $file->getTitle()->getArticleID() ) {
1873 return [ 'warning' => 'page-exists', 'file' => $file ];
1876 if ( strpos( $file->getName(), '.' ) == false ) {
1877 $partname = $file->getName();
1878 $extension = '';
1879 } else {
1880 $n = strrpos( $file->getName(), '.' );
1881 $extension = substr( $file->getName(), $n + 1 );
1882 $partname = substr( $file->getName(), 0, $n );
1884 $normalizedExtension = File::normalizeExtension( $extension );
1886 if ( $normalizedExtension != $extension ) {
1887 // We're not using the normalized form of the extension.
1888 // Normal form is lowercase, using most common of alternate
1889 // extensions (eg 'jpg' rather than 'JPEG').
1891 // Check for another file using the normalized form...
1892 $nt_lc = Title::makeTitle( NS_FILE, "{$partname}.{$normalizedExtension}" );
1893 $file_lc = wfLocalFile( $nt_lc );
1895 if ( $file_lc->exists() ) {
1896 return [
1897 'warning' => 'exists-normalized',
1898 'file' => $file,
1899 'normalizedFile' => $file_lc
1904 // Check for files with the same name but a different extension
1905 $similarFiles = RepoGroup::singleton()->getLocalRepo()->findFilesByPrefix(
1906 "{$partname}.", 1 );
1907 if ( count( $similarFiles ) ) {
1908 return [
1909 'warning' => 'exists-normalized',
1910 'file' => $file,
1911 'normalizedFile' => $similarFiles[0],
1915 if ( self::isThumbName( $file->getName() ) ) {
1916 # Check for filenames like 50px- or 180px-, these are mostly thumbnails
1917 $nt_thb = Title::newFromText(
1918 substr( $partname, strpos( $partname, '-' ) + 1 ) . '.' . $extension,
1919 NS_FILE
1921 $file_thb = wfLocalFile( $nt_thb );
1922 if ( $file_thb->exists() ) {
1923 return [
1924 'warning' => 'thumb',
1925 'file' => $file,
1926 'thumbFile' => $file_thb
1928 } else {
1929 // File does not exist, but we just don't like the name
1930 return [
1931 'warning' => 'thumb-name',
1932 'file' => $file,
1933 'thumbFile' => $file_thb
1938 foreach ( self::getFilenamePrefixBlacklist() as $prefix ) {
1939 if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
1940 return [
1941 'warning' => 'bad-prefix',
1942 'file' => $file,
1943 'prefix' => $prefix
1948 return false;
1952 * Helper function that checks whether the filename looks like a thumbnail
1953 * @param string $filename
1954 * @return bool
1956 public static function isThumbName( $filename ) {
1957 $n = strrpos( $filename, '.' );
1958 $partname = $n ? substr( $filename, 0, $n ) : $filename;
1960 return (
1961 substr( $partname, 3, 3 ) == 'px-' ||
1962 substr( $partname, 2, 3 ) == 'px-'
1963 ) &&
1964 preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) );
1968 * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
1970 * @return array List of prefixes
1972 public static function getFilenamePrefixBlacklist() {
1973 $blacklist = [];
1974 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
1975 if ( !$message->isDisabled() ) {
1976 $lines = explode( "\n", $message->plain() );
1977 foreach ( $lines as $line ) {
1978 // Remove comment lines
1979 $comment = substr( trim( $line ), 0, 1 );
1980 if ( $comment == '#' || $comment == '' ) {
1981 continue;
1983 // Remove additional comments after a prefix
1984 $comment = strpos( $line, '#' );
1985 if ( $comment > 0 ) {
1986 $line = substr( $line, 0, $comment - 1 );
1988 $blacklist[] = trim( $line );
1992 return $blacklist;
1996 * Gets image info about the file just uploaded.
1998 * Also has the effect of setting metadata to be an 'indexed tag name' in
1999 * returned API result if 'metadata' was requested. Oddly, we have to pass
2000 * the "result" object down just so it can do that with the appropriate
2001 * format, presumably.
2003 * @param ApiResult $result
2004 * @return array Image info
2006 public function getImageInfo( $result ) {
2007 $localFile = $this->getLocalFile();
2008 $stashFile = $this->getStashFile();
2009 // Calling a different API module depending on whether the file was stashed is less than optimal.
2010 // In fact, calling API modules here at all is less than optimal. Maybe it should be refactored.
2011 if ( $stashFile ) {
2012 $imParam = ApiQueryStashImageInfo::getPropertyNames();
2013 $info = ApiQueryStashImageInfo::getInfo( $stashFile, array_flip( $imParam ), $result );
2014 } else {
2015 $imParam = ApiQueryImageInfo::getPropertyNames();
2016 $info = ApiQueryImageInfo::getInfo( $localFile, array_flip( $imParam ), $result );
2019 return $info;
2023 * @param array $error
2024 * @return Status
2026 public function convertVerifyErrorToStatus( $error ) {
2027 $code = $error['status'];
2028 unset( $code['status'] );
2030 return Status::newFatal( $this->getVerificationErrorCode( $code ), $error );
2034 * Get the MediaWiki maximum uploaded file size for given type of upload, based on
2035 * $wgMaxUploadSize.
2037 * @param null|string $forType
2038 * @return int
2040 public static function getMaxUploadSize( $forType = null ) {
2041 global $wgMaxUploadSize;
2043 if ( is_array( $wgMaxUploadSize ) ) {
2044 if ( !is_null( $forType ) && isset( $wgMaxUploadSize[$forType] ) ) {
2045 return $wgMaxUploadSize[$forType];
2046 } else {
2047 return $wgMaxUploadSize['*'];
2049 } else {
2050 return intval( $wgMaxUploadSize );
2055 * Get the PHP maximum uploaded file size, based on ini settings. If there is no limit or the
2056 * limit can't be guessed, returns a very large number (PHP_INT_MAX).
2058 * @since 1.27
2059 * @return int
2061 public static function getMaxPhpUploadSize() {
2062 $phpMaxFileSize = wfShorthandToInteger(
2063 ini_get( 'upload_max_filesize' ) ?: ini_get( 'hhvm.server.upload.upload_max_file_size' ),
2064 PHP_INT_MAX
2066 $phpMaxPostSize = wfShorthandToInteger(
2067 ini_get( 'post_max_size' ) ?: ini_get( 'hhvm.server.max_post_size' ),
2068 PHP_INT_MAX
2069 ) ?: PHP_INT_MAX;
2070 return min( $phpMaxFileSize, $phpMaxPostSize );
2074 * Get the current status of a chunked upload (used for polling)
2076 * The value will be read from cache.
2078 * @param User $user
2079 * @param string $statusKey
2080 * @return Status[]|bool
2082 public static function getSessionStatus( User $user, $statusKey ) {
2083 $key = wfMemcKey( 'uploadstatus', $user->getId() ?: md5( $user->getName() ), $statusKey );
2085 return ObjectCache::getMainStashInstance()->get( $key );
2089 * Set the current status of a chunked upload (used for polling)
2091 * The value will be set in cache for 1 day
2093 * @param User $user
2094 * @param string $statusKey
2095 * @param array|bool $value
2096 * @return void
2098 public static function setSessionStatus( User $user, $statusKey, $value ) {
2099 $key = wfMemcKey( 'uploadstatus', $user->getId() ?: md5( $user->getName() ), $statusKey );
2101 $cache = ObjectCache::getMainStashInstance();
2102 if ( $value === false ) {
2103 $cache->delete( $key );
2104 } else {
2105 $cache->set( $key, $value, $cache::TTL_DAY );