d7c19ef5b40f3502fad7a0197d54fb2643e893b2
[lhc/web/wiklou.git] / includes / upload / UploadBase.php
1 <?php
2 /**
3 * Base class for the backend of file upload.
4 *
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.
9 *
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.
14 *
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
19 *
20 * @file
21 * @ingroup Upload
22 */
23
24 /**
25 * @defgroup Upload Upload related
26 */
27
28 /**
29 * @ingroup Upload
30 *
31 * UploadBase and subclasses are the backend of MediaWiki's file uploads.
32 * The frontends are formed by ApiUpload and SpecialUpload.
33 *
34 * @author Brion Vibber
35 * @author Bryan Tong Minh
36 * @author Michael Dale
37 */
38 abstract class UploadBase {
39 protected $mTempPath;
40 protected $mDesiredDestName, $mDestName, $mRemoveTempFile, $mSourceType;
41 protected $mTitle = false, $mTitleError = 0;
42 protected $mFilteredName, $mFinalExtension;
43 protected $mLocalFile, $mFileSize, $mFileProps;
44 protected $mBlackListedExtensions;
45 protected $mJavaDetected, $mSVGNSError;
46
47 protected static $safeXmlEncodings = array(
48 'UTF-8',
49 'ISO-8859-1',
50 'ISO-8859-2',
51 'UTF-16',
52 'UTF-32'
53 );
54
55 const SUCCESS = 0;
56 const OK = 0;
57 const EMPTY_FILE = 3;
58 const MIN_LENGTH_PARTNAME = 4;
59 const ILLEGAL_FILENAME = 5;
60 const OVERWRITE_EXISTING_FILE = 7; # Not used anymore; handled by verifyTitlePermissions()
61 const FILETYPE_MISSING = 8;
62 const FILETYPE_BADTYPE = 9;
63 const VERIFICATION_ERROR = 10;
64
65 # HOOK_ABORTED is the new name of UPLOAD_VERIFICATION_ERROR
66 const UPLOAD_VERIFICATION_ERROR = 11;
67 const HOOK_ABORTED = 11;
68 const FILE_TOO_LARGE = 12;
69 const WINDOWS_NONASCII_FILENAME = 13;
70 const FILENAME_TOO_LONG = 14;
71
72 /**
73 * @param int $error
74 * @return string
75 */
76 public function getVerificationErrorCode( $error ) {
77 $code_to_status = array(
78 self::EMPTY_FILE => 'empty-file',
79 self::FILE_TOO_LARGE => 'file-too-large',
80 self::FILETYPE_MISSING => 'filetype-missing',
81 self::FILETYPE_BADTYPE => 'filetype-banned',
82 self::MIN_LENGTH_PARTNAME => 'filename-tooshort',
83 self::ILLEGAL_FILENAME => 'illegal-filename',
84 self::OVERWRITE_EXISTING_FILE => 'overwrite',
85 self::VERIFICATION_ERROR => 'verification-error',
86 self::HOOK_ABORTED => 'hookaborted',
87 self::WINDOWS_NONASCII_FILENAME => 'windows-nonascii-filename',
88 self::FILENAME_TOO_LONG => 'filename-toolong',
89 );
90 if ( isset( $code_to_status[$error] ) ) {
91 return $code_to_status[$error];
92 }
93
94 return 'unknown-error';
95 }
96
97 /**
98 * Returns true if uploads are enabled.
99 * Can be override by subclasses.
100 * @return bool
101 */
102 public static function isEnabled() {
103 global $wgEnableUploads;
104
105 if ( !$wgEnableUploads ) {
106 return false;
107 }
108
109 # Check php's file_uploads setting
110 return wfIsHHVM() || wfIniGetBool( 'file_uploads' );
111 }
112
113 /**
114 * Returns true if the user can use this upload module or else a string
115 * identifying the missing permission.
116 * Can be overridden by subclasses.
117 *
118 * @param User $user
119 * @return bool|string
120 */
121 public static function isAllowed( $user ) {
122 foreach ( array( 'upload', 'edit' ) as $permission ) {
123 if ( !$user->isAllowed( $permission ) ) {
124 return $permission;
125 }
126 }
127
128 return true;
129 }
130
131 // Upload handlers. Should probably just be a global.
132 private static $uploadHandlers = array( 'Stash', 'File', 'Url' );
133
134 /**
135 * Create a form of UploadBase depending on wpSourceType and initializes it
136 *
137 * @param WebRequest $request
138 * @param string|null $type
139 * @return null|UploadBase
140 */
141 public static function createFromRequest( &$request, $type = null ) {
142 $type = $type ? $type : $request->getVal( 'wpSourceType', 'File' );
143
144 if ( !$type ) {
145 return null;
146 }
147
148 // Get the upload class
149 $type = ucfirst( $type );
150
151 // Give hooks the chance to handle this request
152 $className = null;
153 Hooks::run( 'UploadCreateFromRequest', array( $type, &$className ) );
154 if ( is_null( $className ) ) {
155 $className = 'UploadFrom' . $type;
156 wfDebug( __METHOD__ . ": class name: $className\n" );
157 if ( !in_array( $type, self::$uploadHandlers ) ) {
158 return null;
159 }
160 }
161
162 // Check whether this upload class is enabled
163 if ( !call_user_func( array( $className, 'isEnabled' ) ) ) {
164 return null;
165 }
166
167 // Check whether the request is valid
168 if ( !call_user_func( array( $className, 'isValidRequest' ), $request ) ) {
169 return null;
170 }
171
172 /** @var UploadBase $handler */
173 $handler = new $className;
174
175 $handler->initializeFromRequest( $request );
176
177 return $handler;
178 }
179
180 /**
181 * Check whether a request if valid for this handler
182 * @param WebRequest $request
183 * @return bool
184 */
185 public static function isValidRequest( $request ) {
186 return false;
187 }
188
189 public function __construct() {
190 }
191
192 /**
193 * Returns the upload type. Should be overridden by child classes
194 *
195 * @since 1.18
196 * @return string
197 */
198 public function getSourceType() {
199 return null;
200 }
201
202 /**
203 * Initialize the path information
204 * @param string $name The desired destination name
205 * @param string $tempPath The temporary path
206 * @param int $fileSize The file size
207 * @param bool $removeTempFile (false) remove the temporary file?
208 * @throws MWException
209 */
210 public function initializePathInfo( $name, $tempPath, $fileSize, $removeTempFile = false ) {
211 $this->mDesiredDestName = $name;
212 if ( FileBackend::isStoragePath( $tempPath ) ) {
213 throw new MWException( __METHOD__ . " given storage path `$tempPath`." );
214 }
215 $this->mTempPath = $tempPath;
216 $this->mFileSize = $fileSize;
217 $this->mRemoveTempFile = $removeTempFile;
218 }
219
220 /**
221 * Initialize from a WebRequest. Override this in a subclass.
222 *
223 * @param WebRequest $request
224 */
225 abstract public function initializeFromRequest( &$request );
226
227 /**
228 * Fetch the file. Usually a no-op
229 * @return Status
230 */
231 public function fetchFile() {
232 return Status::newGood();
233 }
234
235 /**
236 * Return true if the file is empty
237 * @return bool
238 */
239 public function isEmptyFile() {
240 return empty( $this->mFileSize );
241 }
242
243 /**
244 * Return the file size
245 * @return int
246 */
247 public function getFileSize() {
248 return $this->mFileSize;
249 }
250
251 /**
252 * Get the base 36 SHA1 of the file
253 * @return string
254 */
255 public function getTempFileSha1Base36() {
256 return FSFile::getSha1Base36FromPath( $this->mTempPath );
257 }
258
259 /**
260 * @param string $srcPath The source path
261 * @return string|bool The real path if it was a virtual URL Returns false on failure
262 */
263 function getRealPath( $srcPath ) {
264 $repo = RepoGroup::singleton()->getLocalRepo();
265 if ( $repo->isVirtualUrl( $srcPath ) ) {
266 /** @todo Just make uploads work with storage paths UploadFromStash
267 * loads files via virtual URLs.
268 */
269 $tmpFile = $repo->getLocalCopy( $srcPath );
270 if ( $tmpFile ) {
271 $tmpFile->bind( $this ); // keep alive with $this
272 }
273 $path = $tmpFile ? $tmpFile->getPath() : false;
274 } else {
275 $path = $srcPath;
276 }
277
278 return $path;
279 }
280
281 /**
282 * Verify whether the upload is sane.
283 * @return mixed Const self::OK or else an array with error information
284 */
285 public function verifyUpload() {
286
287 /**
288 * If there was no filename or a zero size given, give up quick.
289 */
290 if ( $this->isEmptyFile() ) {
291
292 return array( 'status' => self::EMPTY_FILE );
293 }
294
295 /**
296 * Honor $wgMaxUploadSize
297 */
298 $maxSize = self::getMaxUploadSize( $this->getSourceType() );
299 if ( $this->mFileSize > $maxSize ) {
300
301 return array(
302 'status' => self::FILE_TOO_LARGE,
303 'max' => $maxSize,
304 );
305 }
306
307 /**
308 * Look at the contents of the file; if we can recognize the
309 * type but it's corrupt or data of the wrong type, we should
310 * probably not accept it.
311 */
312 $verification = $this->verifyFile();
313 if ( $verification !== true ) {
314
315 return array(
316 'status' => self::VERIFICATION_ERROR,
317 'details' => $verification
318 );
319 }
320
321 /**
322 * Make sure this file can be created
323 */
324 $result = $this->validateName();
325 if ( $result !== true ) {
326
327 return $result;
328 }
329
330 $error = '';
331 if ( !Hooks::run( 'UploadVerification',
332 array( $this->mDestName, $this->mTempPath, &$error ) )
333 ) {
334
335 return array( 'status' => self::HOOK_ABORTED, 'error' => $error );
336 }
337
338 return array( 'status' => self::OK );
339 }
340
341 /**
342 * Verify that the name is valid and, if necessary, that we can overwrite
343 *
344 * @return mixed True if valid, otherwise and array with 'status'
345 * and other keys
346 */
347 public function validateName() {
348 $nt = $this->getTitle();
349 if ( is_null( $nt ) ) {
350 $result = array( 'status' => $this->mTitleError );
351 if ( $this->mTitleError == self::ILLEGAL_FILENAME ) {
352 $result['filtered'] = $this->mFilteredName;
353 }
354 if ( $this->mTitleError == self::FILETYPE_BADTYPE ) {
355 $result['finalExt'] = $this->mFinalExtension;
356 if ( count( $this->mBlackListedExtensions ) ) {
357 $result['blacklistedExt'] = $this->mBlackListedExtensions;
358 }
359 }
360
361 return $result;
362 }
363 $this->mDestName = $this->getLocalFile()->getName();
364
365 return true;
366 }
367
368 /**
369 * Verify the MIME type.
370 *
371 * @note Only checks that it is not an evil MIME. The "does it have
372 * correct extension given its MIME type?" check is in verifyFile.
373 * in `verifyFile()` that MIME type and file extension correlate.
374 * @param string $mime Representing the MIME
375 * @return mixed True if the file is verified, an array otherwise
376 */
377 protected function verifyMimeType( $mime ) {
378 global $wgVerifyMimeType;
379 if ( $wgVerifyMimeType ) {
380 wfDebug( "mime: <$mime> extension: <{$this->mFinalExtension}>\n" );
381 global $wgMimeTypeBlacklist;
382 if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
383
384 return array( 'filetype-badmime', $mime );
385 }
386
387 # Check what Internet Explorer would detect
388 $fp = fopen( $this->mTempPath, 'rb' );
389 $chunk = fread( $fp, 256 );
390 fclose( $fp );
391
392 $magic = MimeMagic::singleton();
393 $extMime = $magic->guessTypesForExtension( $this->mFinalExtension );
394 $ieTypes = $magic->getIEMimeTypes( $this->mTempPath, $chunk, $extMime );
395 foreach ( $ieTypes as $ieType ) {
396 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
397
398 return array( 'filetype-bad-ie-mime', $ieType );
399 }
400 }
401 }
402
403 return true;
404 }
405
406 /**
407 * Verifies that it's ok to include the uploaded file
408 *
409 * @return mixed True of the file is verified, array otherwise.
410 */
411 protected function verifyFile() {
412 global $wgVerifyMimeType, $wgDisableUploadScriptChecks;
413
414 $status = $this->verifyPartialFile();
415 if ( $status !== true ) {
416
417 return $status;
418 }
419
420 $this->mFileProps = FSFile::getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
421 $mime = $this->mFileProps['mime'];
422
423 if ( $wgVerifyMimeType ) {
424 # XXX: Missing extension will be caught by validateName() via getTitle()
425 if ( $this->mFinalExtension != '' && !$this->verifyExtension( $mime, $this->mFinalExtension ) ) {
426
427 return array( 'filetype-mime-mismatch', $this->mFinalExtension, $mime );
428 }
429 }
430
431 # check for htmlish code and javascript
432 if ( !$wgDisableUploadScriptChecks ) {
433 if ( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
434 $svgStatus = $this->detectScriptInSvg( $this->mTempPath, false );
435 if ( $svgStatus !== false ) {
436
437 return $svgStatus;
438 }
439 }
440 }
441
442 $handler = MediaHandler::getHandler( $mime );
443 if ( $handler ) {
444 $handlerStatus = $handler->verifyUpload( $this->mTempPath );
445 if ( !$handlerStatus->isOK() ) {
446 $errors = $handlerStatus->getErrorsArray();
447
448 return reset( $errors );
449 }
450 }
451
452 Hooks::run( 'UploadVerifyFile', array( $this, $mime, &$status ) );
453 if ( $status !== true ) {
454
455 return $status;
456 }
457
458 wfDebug( __METHOD__ . ": all clear; passing.\n" );
459
460 return true;
461 }
462
463 /**
464 * A verification routine suitable for partial files
465 *
466 * Runs the blacklist checks, but not any checks that may
467 * assume the entire file is present.
468 *
469 * @return mixed True for valid or array with error message key.
470 */
471 protected function verifyPartialFile() {
472 global $wgAllowJavaUploads, $wgDisableUploadScriptChecks;
473
474 # getTitle() sets some internal parameters like $this->mFinalExtension
475 $this->getTitle();
476
477 $this->mFileProps = FSFile::getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
478
479 # check MIME type, if desired
480 $mime = $this->mFileProps['file-mime'];
481 $status = $this->verifyMimeType( $mime );
482 if ( $status !== true ) {
483
484 return $status;
485 }
486
487 # check for htmlish code and javascript
488 if ( !$wgDisableUploadScriptChecks ) {
489 if ( self::detectScript( $this->mTempPath, $mime, $this->mFinalExtension ) ) {
490
491 return array( 'uploadscripted' );
492 }
493 if ( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
494 $svgStatus = $this->detectScriptInSvg( $this->mTempPath, true );
495 if ( $svgStatus !== false ) {
496
497 return $svgStatus;
498 }
499 }
500 }
501
502 # Check for Java applets, which if uploaded can bypass cross-site
503 # restrictions.
504 if ( !$wgAllowJavaUploads ) {
505 $this->mJavaDetected = false;
506 $zipStatus = ZipDirectoryReader::read( $this->mTempPath,
507 array( $this, 'zipEntryCallback' ) );
508 if ( !$zipStatus->isOK() ) {
509 $errors = $zipStatus->getErrorsArray();
510 $error = reset( $errors );
511 if ( $error[0] !== 'zip-wrong-format' ) {
512
513 return $error;
514 }
515 }
516 if ( $this->mJavaDetected ) {
517
518 return array( 'uploadjava' );
519 }
520 }
521
522 # Scan the uploaded file for viruses
523 $virus = $this->detectVirus( $this->mTempPath );
524 if ( $virus ) {
525
526 return array( 'uploadvirus', $virus );
527 }
528
529 return true;
530 }
531
532 /**
533 * Callback for ZipDirectoryReader to detect Java class files.
534 *
535 * @param array $entry
536 */
537 function zipEntryCallback( $entry ) {
538 $names = array( $entry['name'] );
539
540 // If there is a null character, cut off the name at it, because JDK's
541 // ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name
542 // were constructed which had ".class\0" followed by a string chosen to
543 // make the hash collide with the truncated name, that file could be
544 // returned in response to a request for the .class file.
545 $nullPos = strpos( $entry['name'], "\000" );
546 if ( $nullPos !== false ) {
547 $names[] = substr( $entry['name'], 0, $nullPos );
548 }
549
550 // If there is a trailing slash in the file name, we have to strip it,
551 // because that's what ZIP_GetEntry() does.
552 if ( preg_grep( '!\.class/?$!', $names ) ) {
553 $this->mJavaDetected = true;
554 }
555 }
556
557 /**
558 * Alias for verifyTitlePermissions. The function was originally
559 * 'verifyPermissions', but that suggests it's checking the user, when it's
560 * really checking the title + user combination.
561 *
562 * @param User $user User object to verify the permissions against
563 * @return mixed An array as returned by getUserPermissionsErrors or true
564 * in case the user has proper permissions.
565 */
566 public function verifyPermissions( $user ) {
567 return $this->verifyTitlePermissions( $user );
568 }
569
570 /**
571 * Check whether the user can edit, upload and create the image. This
572 * checks only against the current title; if it returns errors, it may
573 * very well be that another title will not give errors. Therefore
574 * isAllowed() should be called as well for generic is-user-blocked or
575 * can-user-upload checking.
576 *
577 * @param User $user User object to verify the permissions against
578 * @return mixed An array as returned by getUserPermissionsErrors or true
579 * in case the user has proper permissions.
580 */
581 public function verifyTitlePermissions( $user ) {
582 /**
583 * If the image is protected, non-sysop users won't be able
584 * to modify it by uploading a new revision.
585 */
586 $nt = $this->getTitle();
587 if ( is_null( $nt ) ) {
588 return true;
589 }
590 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
591 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
592 if ( !$nt->exists() ) {
593 $permErrorsCreate = $nt->getUserPermissionsErrors( 'create', $user );
594 } else {
595 $permErrorsCreate = array();
596 }
597 if ( $permErrors || $permErrorsUpload || $permErrorsCreate ) {
598 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
599 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
600
601 return $permErrors;
602 }
603
604 $overwriteError = $this->checkOverwrite( $user );
605 if ( $overwriteError !== true ) {
606 return array( $overwriteError );
607 }
608
609 return true;
610 }
611
612 /**
613 * Check for non fatal problems with the file.
614 *
615 * This should not assume that mTempPath is set.
616 *
617 * @return array Array of warnings
618 */
619 public function checkWarnings() {
620 global $wgLang;
621
622 $warnings = array();
623
624 $localFile = $this->getLocalFile();
625 $localFile->load( File::READ_LATEST );
626 $filename = $localFile->getName();
627
628 /**
629 * Check whether the resulting filename is different from the desired one,
630 * but ignore things like ucfirst() and spaces/underscore things
631 */
632 $comparableName = str_replace( ' ', '_', $this->mDesiredDestName );
633 $comparableName = Title::capitalize( $comparableName, NS_FILE );
634
635 if ( $this->mDesiredDestName != $filename && $comparableName != $filename ) {
636 $warnings['badfilename'] = $filename;
637 // Debugging for bug 62241
638 wfDebugLog( 'upload', "Filename: '$filename', mDesiredDestName: "
639 . "'$this->mDesiredDestName', comparableName: '$comparableName'" );
640 }
641
642 // Check whether the file extension is on the unwanted list
643 global $wgCheckFileExtensions, $wgFileExtensions;
644 if ( $wgCheckFileExtensions ) {
645 $extensions = array_unique( $wgFileExtensions );
646 if ( !$this->checkFileExtension( $this->mFinalExtension, $extensions ) ) {
647 $warnings['filetype-unwanted-type'] = array( $this->mFinalExtension,
648 $wgLang->commaList( $extensions ), count( $extensions ) );
649 }
650 }
651
652 global $wgUploadSizeWarning;
653 if ( $wgUploadSizeWarning && ( $this->mFileSize > $wgUploadSizeWarning ) ) {
654 $warnings['large-file'] = array( $wgUploadSizeWarning, $this->mFileSize );
655 }
656
657 if ( $this->mFileSize == 0 ) {
658 $warnings['emptyfile'] = true;
659 }
660
661 $exists = self::getExistsWarning( $localFile );
662 if ( $exists !== false ) {
663 $warnings['exists'] = $exists;
664 }
665
666 // Check dupes against existing files
667 $hash = $this->getTempFileSha1Base36();
668 $dupes = RepoGroup::singleton()->findBySha1( $hash );
669 $title = $this->getTitle();
670 // Remove all matches against self
671 foreach ( $dupes as $key => $dupe ) {
672 if ( $title->equals( $dupe->getTitle() ) ) {
673 unset( $dupes[$key] );
674 }
675 }
676 if ( $dupes ) {
677 $warnings['duplicate'] = $dupes;
678 }
679
680 // Check dupes against archives
681 $archivedFile = new ArchivedFile( null, 0, '', $hash );
682 if ( $archivedFile->getID() > 0 ) {
683 if ( $archivedFile->userCan( File::DELETED_FILE ) ) {
684 $warnings['duplicate-archive'] = $archivedFile->getName();
685 } else {
686 $warnings['duplicate-archive'] = '';
687 }
688 }
689
690 return $warnings;
691 }
692
693 /**
694 * Really perform the upload. Stores the file in the local repo, watches
695 * if necessary and runs the UploadComplete hook.
696 *
697 * @param string $comment
698 * @param string $pageText
699 * @param bool $watch
700 * @param User $user
701 *
702 * @return Status Indicating the whether the upload succeeded.
703 */
704 public function performUpload( $comment, $pageText, $watch, $user ) {
705 $this->getLocalFile()->load( File::READ_LATEST );
706
707 $status = $this->getLocalFile()->upload(
708 $this->mTempPath,
709 $comment,
710 $pageText,
711 File::DELETE_SOURCE,
712 $this->mFileProps,
713 false,
714 $user
715 );
716
717 if ( $status->isGood() ) {
718 if ( $watch ) {
719 WatchAction::doWatch(
720 $this->getLocalFile()->getTitle(),
721 $user,
722 WatchedItem::IGNORE_USER_RIGHTS
723 );
724 }
725 Hooks::run( 'UploadComplete', array( &$this ) );
726
727 $this->postProcessUpload();
728 }
729
730 return $status;
731 }
732
733 /**
734 * Perform extra steps after a successful upload.
735 *
736 * @since 1.25
737 */
738 public function postProcessUpload() {
739 global $wgUploadThumbnailRenderMap;
740
741 $jobs = array();
742
743 $sizes = $wgUploadThumbnailRenderMap;
744 rsort( $sizes );
745
746 $file = $this->getLocalFile();
747
748 foreach ( $sizes as $size ) {
749 if ( $file->isVectorized() || $file->getWidth() > $size ) {
750 $jobs[] = new ThumbnailRenderJob(
751 $file->getTitle(),
752 array( 'transformParams' => array( 'width' => $size ) )
753 );
754 }
755 }
756
757 if ( $jobs ) {
758 JobQueueGroup::singleton()->push( $jobs );
759 }
760 }
761
762 /**
763 * Returns the title of the file to be uploaded. Sets mTitleError in case
764 * the name was illegal.
765 *
766 * @return Title The title of the file or null in case the name was illegal
767 */
768 public function getTitle() {
769 if ( $this->mTitle !== false ) {
770 return $this->mTitle;
771 }
772 /* Assume that if a user specified File:Something.jpg, this is an error
773 * and that the namespace prefix needs to be stripped of.
774 */
775 $title = Title::newFromText( $this->mDesiredDestName );
776 if ( $title && $title->getNamespace() == NS_FILE ) {
777 $this->mFilteredName = $title->getDBkey();
778 } else {
779 $this->mFilteredName = $this->mDesiredDestName;
780 }
781
782 # oi_archive_name is max 255 bytes, which include a timestamp and an
783 # exclamation mark, so restrict file name to 240 bytes.
784 if ( strlen( $this->mFilteredName ) > 240 ) {
785 $this->mTitleError = self::FILENAME_TOO_LONG;
786 $this->mTitle = null;
787
788 return $this->mTitle;
789 }
790
791 /**
792 * Chop off any directories in the given filename. Then
793 * filter out illegal characters, and try to make a legible name
794 * out of it. We'll strip some silently that Title would die on.
795 */
796 $this->mFilteredName = wfStripIllegalFilenameChars( $this->mFilteredName );
797 /* Normalize to title form before we do any further processing */
798 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
799 if ( is_null( $nt ) ) {
800 $this->mTitleError = self::ILLEGAL_FILENAME;
801 $this->mTitle = null;
802
803 return $this->mTitle;
804 }
805 $this->mFilteredName = $nt->getDBkey();
806
807 /**
808 * We'll want to blacklist against *any* 'extension', and use
809 * only the final one for the whitelist.
810 */
811 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName );
812
813 if ( count( $ext ) ) {
814 $this->mFinalExtension = trim( $ext[count( $ext ) - 1] );
815 } else {
816 $this->mFinalExtension = '';
817
818 # No extension, try guessing one
819 $magic = MimeMagic::singleton();
820 $mime = $magic->guessMimeType( $this->mTempPath );
821 if ( $mime !== 'unknown/unknown' ) {
822 # Get a space separated list of extensions
823 $extList = $magic->getExtensionsForType( $mime );
824 if ( $extList ) {
825 # Set the extension to the canonical extension
826 $this->mFinalExtension = strtok( $extList, ' ' );
827
828 # Fix up the other variables
829 $this->mFilteredName .= ".{$this->mFinalExtension}";
830 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
831 $ext = array( $this->mFinalExtension );
832 }
833 }
834 }
835
836 /* Don't allow users to override the blacklist (check file extension) */
837 global $wgCheckFileExtensions, $wgStrictFileExtensions;
838 global $wgFileExtensions, $wgFileBlacklist;
839
840 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
841
842 if ( $this->mFinalExtension == '' ) {
843 $this->mTitleError = self::FILETYPE_MISSING;
844 $this->mTitle = null;
845
846 return $this->mTitle;
847 } elseif ( $blackListedExtensions ||
848 ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
849 !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) )
850 ) {
851 $this->mBlackListedExtensions = $blackListedExtensions;
852 $this->mTitleError = self::FILETYPE_BADTYPE;
853 $this->mTitle = null;
854
855 return $this->mTitle;
856 }
857
858 // Windows may be broken with special characters, see bug 1780
859 if ( !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() )
860 && !RepoGroup::singleton()->getLocalRepo()->backendSupportsUnicodePaths()
861 ) {
862 $this->mTitleError = self::WINDOWS_NONASCII_FILENAME;
863 $this->mTitle = null;
864
865 return $this->mTitle;
866 }
867
868 # If there was more than one "extension", reassemble the base
869 # filename to prevent bogus complaints about length
870 if ( count( $ext ) > 1 ) {
871 $iterations = count( $ext ) - 1;
872 for ( $i = 0; $i < $iterations; $i++ ) {
873 $partname .= '.' . $ext[$i];
874 }
875 }
876
877 if ( strlen( $partname ) < 1 ) {
878 $this->mTitleError = self::MIN_LENGTH_PARTNAME;
879 $this->mTitle = null;
880
881 return $this->mTitle;
882 }
883
884 $this->mTitle = $nt;
885
886 return $this->mTitle;
887 }
888
889 /**
890 * Return the local file and initializes if necessary.
891 *
892 * @return LocalFile|UploadStashFile|null
893 */
894 public function getLocalFile() {
895 if ( is_null( $this->mLocalFile ) ) {
896 $nt = $this->getTitle();
897 $this->mLocalFile = is_null( $nt ) ? null : wfLocalFile( $nt );
898 }
899
900 return $this->mLocalFile;
901 }
902
903 /**
904 * If the user does not supply all necessary information in the first upload
905 * form submission (either by accident or by design) then we may want to
906 * stash the file temporarily, get more information, and publish the file
907 * later.
908 *
909 * This method will stash a file in a temporary directory for later
910 * processing, and save the necessary descriptive info into the database.
911 * This method returns the file object, which also has a 'fileKey' property
912 * which can be passed through a form or API request to find this stashed
913 * file again.
914 *
915 * @param User $user
916 * @return UploadStashFile Stashed file
917 */
918 public function stashFile( User $user = null ) {
919 // was stashSessionFile
920
921 $stash = RepoGroup::singleton()->getLocalRepo()->getUploadStash( $user );
922 $file = $stash->stashFile( $this->mTempPath, $this->getSourceType() );
923 $this->mLocalFile = $file;
924
925 return $file;
926 }
927
928 /**
929 * Stash a file in a temporary directory, returning a key which can be used
930 * to find the file again. See stashFile().
931 *
932 * @return string File key
933 */
934 public function stashFileGetKey() {
935 return $this->stashFile()->getFileKey();
936 }
937
938 /**
939 * alias for stashFileGetKey, for backwards compatibility
940 *
941 * @return string File key
942 */
943 public function stashSession() {
944 return $this->stashFileGetKey();
945 }
946
947 /**
948 * If we've modified the upload file we need to manually remove it
949 * on exit to clean up.
950 */
951 public function cleanupTempFile() {
952 if ( $this->mRemoveTempFile && $this->mTempPath && file_exists( $this->mTempPath ) ) {
953 wfDebug( __METHOD__ . ": Removing temporary file {$this->mTempPath}\n" );
954 unlink( $this->mTempPath );
955 }
956 }
957
958 public function getTempPath() {
959 return $this->mTempPath;
960 }
961
962 /**
963 * Split a file into a base name and all dot-delimited 'extensions'
964 * on the end. Some web server configurations will fall back to
965 * earlier pseudo-'extensions' to determine type and execute
966 * scripts, so the blacklist needs to check them all.
967 *
968 * @param string $filename
969 * @return array
970 */
971 public static function splitExtensions( $filename ) {
972 $bits = explode( '.', $filename );
973 $basename = array_shift( $bits );
974
975 return array( $basename, $bits );
976 }
977
978 /**
979 * Perform case-insensitive match against a list of file extensions.
980 * Returns true if the extension is in the list.
981 *
982 * @param string $ext
983 * @param array $list
984 * @return bool
985 */
986 public static function checkFileExtension( $ext, $list ) {
987 return in_array( strtolower( $ext ), $list );
988 }
989
990 /**
991 * Perform case-insensitive match against a list of file extensions.
992 * Returns an array of matching extensions.
993 *
994 * @param array $ext
995 * @param array $list
996 * @return bool
997 */
998 public static function checkFileExtensionList( $ext, $list ) {
999 return array_intersect( array_map( 'strtolower', $ext ), $list );
1000 }
1001
1002 /**
1003 * Checks if the MIME type of the uploaded file matches the file extension.
1004 *
1005 * @param string $mime The MIME type of the uploaded file
1006 * @param string $extension The filename extension that the file is to be served with
1007 * @return bool
1008 */
1009 public static function verifyExtension( $mime, $extension ) {
1010 $magic = MimeMagic::singleton();
1011
1012 if ( !$mime || $mime == 'unknown' || $mime == 'unknown/unknown' ) {
1013 if ( !$magic->isRecognizableExtension( $extension ) ) {
1014 wfDebug( __METHOD__ . ": passing file with unknown detected mime type; " .
1015 "unrecognized extension '$extension', can't verify\n" );
1016
1017 return true;
1018 } else {
1019 wfDebug( __METHOD__ . ": rejecting file with unknown detected mime type; " .
1020 "recognized extension '$extension', so probably invalid file\n" );
1021
1022 return false;
1023 }
1024 }
1025
1026 $match = $magic->isMatchingExtension( $extension, $mime );
1027
1028 if ( $match === null ) {
1029 if ( $magic->getTypesForExtension( $extension ) !== null ) {
1030 wfDebug( __METHOD__ . ": No extension known for $mime, but we know a mime for $extension\n" );
1031
1032 return false;
1033 } else {
1034 wfDebug( __METHOD__ . ": no file extension known for mime type $mime, passing file\n" );
1035
1036 return true;
1037 }
1038 } elseif ( $match === true ) {
1039 wfDebug( __METHOD__ . ": mime type $mime matches extension $extension, passing file\n" );
1040
1041 /** @todo If it's a bitmap, make sure PHP or ImageMagick resp. can handle it! */
1042 return true;
1043 } else {
1044 wfDebug( __METHOD__
1045 . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
1046
1047 return false;
1048 }
1049 }
1050
1051 /**
1052 * Heuristic for detecting files that *could* contain JavaScript instructions or
1053 * things that may look like HTML to a browser and are thus
1054 * potentially harmful. The present implementation will produce false
1055 * positives in some situations.
1056 *
1057 * @param string $file Pathname to the temporary upload file
1058 * @param string $mime The MIME type of the file
1059 * @param string $extension The extension of the file
1060 * @return bool True if the file contains something looking like embedded scripts
1061 */
1062 public static function detectScript( $file, $mime, $extension ) {
1063 global $wgAllowTitlesInSVG;
1064
1065 # ugly hack: for text files, always look at the entire file.
1066 # For binary field, just check the first K.
1067
1068 if ( strpos( $mime, 'text/' ) === 0 ) {
1069 $chunk = file_get_contents( $file );
1070 } else {
1071 $fp = fopen( $file, 'rb' );
1072 $chunk = fread( $fp, 1024 );
1073 fclose( $fp );
1074 }
1075
1076 $chunk = strtolower( $chunk );
1077
1078 if ( !$chunk ) {
1079
1080 return false;
1081 }
1082
1083 # decode from UTF-16 if needed (could be used for obfuscation).
1084 if ( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
1085 $enc = 'UTF-16BE';
1086 } elseif ( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
1087 $enc = 'UTF-16LE';
1088 } else {
1089 $enc = null;
1090 }
1091
1092 if ( $enc ) {
1093 $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
1094 }
1095
1096 $chunk = trim( $chunk );
1097
1098 /** @todo FIXME: Convert from UTF-16 if necessary! */
1099 wfDebug( __METHOD__ . ": checking for embedded scripts and HTML stuff\n" );
1100
1101 # check for HTML doctype
1102 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
1103
1104 return true;
1105 }
1106
1107 // Some browsers will interpret obscure xml encodings as UTF-8, while
1108 // PHP/expat will interpret the given encoding in the xml declaration (bug 47304)
1109 if ( $extension == 'svg' || strpos( $mime, 'image/svg' ) === 0 ) {
1110 if ( self::checkXMLEncodingMissmatch( $file ) ) {
1111
1112 return true;
1113 }
1114 }
1115
1116 /**
1117 * Internet Explorer for Windows performs some really stupid file type
1118 * autodetection which can cause it to interpret valid image files as HTML
1119 * and potentially execute JavaScript, creating a cross-site scripting
1120 * attack vectors.
1121 *
1122 * Apple's Safari browser also performs some unsafe file type autodetection
1123 * which can cause legitimate files to be interpreted as HTML if the
1124 * web server is not correctly configured to send the right content-type
1125 * (or if you're really uploading plain text and octet streams!)
1126 *
1127 * Returns true if IE is likely to mistake the given file for HTML.
1128 * Also returns true if Safari would mistake the given file for HTML
1129 * when served with a generic content-type.
1130 */
1131 $tags = array(
1132 '<a href',
1133 '<body',
1134 '<head',
1135 '<html', #also in safari
1136 '<img',
1137 '<pre',
1138 '<script', #also in safari
1139 '<table'
1140 );
1141
1142 if ( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
1143 $tags[] = '<title';
1144 }
1145
1146 foreach ( $tags as $tag ) {
1147 if ( false !== strpos( $chunk, $tag ) ) {
1148 wfDebug( __METHOD__ . ": found something that may make it be mistaken for html: $tag\n" );
1149
1150 return true;
1151 }
1152 }
1153
1154 /*
1155 * look for JavaScript
1156 */
1157
1158 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
1159 $chunk = Sanitizer::decodeCharReferences( $chunk );
1160
1161 # look for script-types
1162 if ( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
1163 wfDebug( __METHOD__ . ": found script types\n" );
1164
1165 return true;
1166 }
1167
1168 # look for html-style script-urls
1169 if ( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1170 wfDebug( __METHOD__ . ": found html-style script urls\n" );
1171
1172 return true;
1173 }
1174
1175 # look for css-style script-urls
1176 if ( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1177 wfDebug( __METHOD__ . ": found css-style script urls\n" );
1178
1179 return true;
1180 }
1181
1182 wfDebug( __METHOD__ . ": no scripts found\n" );
1183
1184 return false;
1185 }
1186
1187 /**
1188 * Check a whitelist of xml encodings that are known not to be interpreted differently
1189 * by the server's xml parser (expat) and some common browsers.
1190 *
1191 * @param string $file Pathname to the temporary upload file
1192 * @return bool True if the file contains an encoding that could be misinterpreted
1193 */
1194 public static function checkXMLEncodingMissmatch( $file ) {
1195 global $wgSVGMetadataCutoff;
1196 $contents = file_get_contents( $file, false, null, -1, $wgSVGMetadataCutoff );
1197 $encodingRegex = '!encoding[ \t\n\r]*=[ \t\n\r]*[\'"](.*?)[\'"]!si';
1198
1199 if ( preg_match( "!<\?xml\b(.*?)\?>!si", $contents, $matches ) ) {
1200 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1201 && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1202 ) {
1203 wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1204
1205 return true;
1206 }
1207 } elseif ( preg_match( "!<\?xml\b!si", $contents ) ) {
1208 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1209 // bytes. There shouldn't be a legitimate reason for this to happen.
1210 wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1211
1212 return true;
1213 } elseif ( substr( $contents, 0, 4 ) == "\x4C\x6F\xA7\x94" ) {
1214 // EBCDIC encoded XML
1215 wfDebug( __METHOD__ . ": EBCDIC Encoded XML\n" );
1216
1217 return true;
1218 }
1219
1220 // It's possible the file is encoded with multi-byte encoding, so re-encode attempt to
1221 // detect the encoding in case is specifies an encoding not whitelisted in self::$safeXmlEncodings
1222 $attemptEncodings = array( 'UTF-16', 'UTF-16BE', 'UTF-32', 'UTF-32BE' );
1223 foreach ( $attemptEncodings as $encoding ) {
1224 wfSuppressWarnings();
1225 $str = iconv( $encoding, 'UTF-8', $contents );
1226 wfRestoreWarnings();
1227 if ( $str != '' && preg_match( "!<\?xml\b(.*?)\?>!si", $str, $matches ) ) {
1228 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1229 && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1230 ) {
1231 wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1232
1233 return true;
1234 }
1235 } elseif ( $str != '' && preg_match( "!<\?xml\b!si", $str ) ) {
1236 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1237 // bytes. There shouldn't be a legitimate reason for this to happen.
1238 wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1239
1240 return true;
1241 }
1242 }
1243
1244 return false;
1245 }
1246
1247 /**
1248 * @param string $filename
1249 * @param bool $partial
1250 * @return mixed False of the file is verified (does not contain scripts), array otherwise.
1251 */
1252 protected function detectScriptInSvg( $filename, $partial ) {
1253 $this->mSVGNSError = false;
1254 $check = new XmlTypeCheck(
1255 $filename,
1256 array( $this, 'checkSvgScriptCallback' ),
1257 true,
1258 array( 'processing_instruction_handler' => 'UploadBase::checkSvgPICallback' )
1259 );
1260 if ( $check->wellFormed !== true ) {
1261 // Invalid xml (bug 58553)
1262 // But only when non-partial (bug 65724)
1263 return $partial ? false : array( 'uploadinvalidxml' );
1264 } elseif ( $check->filterMatch ) {
1265 if ( $this->mSVGNSError ) {
1266 return array( 'uploadscriptednamespace', $this->mSVGNSError );
1267 }
1268
1269 return $check->filterMatchType;
1270 }
1271
1272 return false;
1273 }
1274
1275 /**
1276 * Callback to filter SVG Processing Instructions.
1277 * @param string $target Processing instruction name
1278 * @param string $data Processing instruction attribute and value
1279 * @return bool (true if the filter identified something bad)
1280 */
1281 public static function checkSvgPICallback( $target, $data ) {
1282 // Don't allow external stylesheets (bug 57550)
1283 if ( preg_match( '/xml-stylesheet/i', $target ) ) {
1284 return array( 'upload-scripted-pi-callback' );
1285 }
1286
1287 return false;
1288 }
1289
1290 /**
1291 * @todo Replace this with a whitelist filter!
1292 * @param string $element
1293 * @param array $attribs
1294 * @return bool
1295 */
1296 public function checkSvgScriptCallback( $element, $attribs, $data = null ) {
1297
1298 list( $namespace, $strippedElement ) = $this->splitXmlNamespace( $element );
1299
1300 // We specifically don't include:
1301 // http://www.w3.org/1999/xhtml (bug 60771)
1302 static $validNamespaces = array(
1303 '',
1304 'adobe:ns:meta/',
1305 'http://creativecommons.org/ns#',
1306 'http://inkscape.sourceforge.net/dtd/sodipodi-0.dtd',
1307 'http://ns.adobe.com/adobeillustrator/10.0/',
1308 'http://ns.adobe.com/adobesvgviewerextensions/3.0/',
1309 'http://ns.adobe.com/extensibility/1.0/',
1310 'http://ns.adobe.com/flows/1.0/',
1311 'http://ns.adobe.com/illustrator/1.0/',
1312 'http://ns.adobe.com/imagereplacement/1.0/',
1313 'http://ns.adobe.com/pdf/1.3/',
1314 'http://ns.adobe.com/photoshop/1.0/',
1315 'http://ns.adobe.com/saveforweb/1.0/',
1316 'http://ns.adobe.com/variables/1.0/',
1317 'http://ns.adobe.com/xap/1.0/',
1318 'http://ns.adobe.com/xap/1.0/g/',
1319 'http://ns.adobe.com/xap/1.0/g/img/',
1320 'http://ns.adobe.com/xap/1.0/mm/',
1321 'http://ns.adobe.com/xap/1.0/rights/',
1322 'http://ns.adobe.com/xap/1.0/stype/dimensions#',
1323 'http://ns.adobe.com/xap/1.0/stype/font#',
1324 'http://ns.adobe.com/xap/1.0/stype/manifestitem#',
1325 'http://ns.adobe.com/xap/1.0/stype/resourceevent#',
1326 'http://ns.adobe.com/xap/1.0/stype/resourceref#',
1327 'http://ns.adobe.com/xap/1.0/t/pg/',
1328 'http://purl.org/dc/elements/1.1/',
1329 'http://purl.org/dc/elements/1.1',
1330 'http://schemas.microsoft.com/visio/2003/svgextensions/',
1331 'http://sodipodi.sourceforge.net/dtd/sodipodi-0.dtd',
1332 'http://taptrix.com/inkpad/svg_extensions',
1333 'http://web.resource.org/cc/',
1334 'http://www.freesoftware.fsf.org/bkchem/cdml',
1335 'http://www.inkscape.org/namespaces/inkscape',
1336 'http://www.opengis.net/gml',
1337 'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
1338 'http://www.w3.org/2000/svg',
1339 'http://www.w3.org/tr/rec-rdf-syntax/',
1340 );
1341
1342 if ( !in_array( $namespace, $validNamespaces ) ) {
1343 wfDebug( __METHOD__ . ": Non-svg namespace '$namespace' in uploaded file.\n" );
1344 /** @todo Return a status object to a closure in XmlTypeCheck, for MW1.21+ */
1345 $this->mSVGNSError = $namespace;
1346
1347 return true;
1348 }
1349
1350 /*
1351 * check for elements that can contain javascript
1352 */
1353 if ( $strippedElement == 'script' ) {
1354 wfDebug( __METHOD__ . ": Found script element '$element' in uploaded file.\n" );
1355
1356 return array( 'uploaded-script-svg', $strippedElement );
1357 }
1358
1359 # e.g., <svg xmlns="http://www.w3.org/2000/svg">
1360 # <handler xmlns:ev="http://www.w3.org/2001/xml-events" ev:event="load">alert(1)</handler> </svg>
1361 if ( $strippedElement == 'handler' ) {
1362 wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1363
1364 return array( 'uploaded-script-svg', $strippedElement );
1365 }
1366
1367 # SVG reported in Feb '12 that used xml:stylesheet to generate javascript block
1368 if ( $strippedElement == 'stylesheet' ) {
1369 wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1370
1371 return array( 'uploaded-script-svg', $strippedElement );
1372 }
1373
1374 # Block iframes, in case they pass the namespace check
1375 if ( $strippedElement == 'iframe' ) {
1376 wfDebug( __METHOD__ . ": iframe in uploaded file.\n" );
1377
1378 return array( 'uploaded-script-svg', $strippedElement );
1379 }
1380
1381 # Check <style> css
1382 if ( $strippedElement == 'style'
1383 && self::checkCssFragment( Sanitizer::normalizeCss( $data ) )
1384 ) {
1385 wfDebug( __METHOD__ . ": hostile css in style element.\n" );
1386 return array( 'uploaded-hostile-svg' );
1387 }
1388
1389 foreach ( $attribs as $attrib => $value ) {
1390 $stripped = $this->stripXmlNamespace( $attrib );
1391 $value = strtolower( $value );
1392
1393 if ( substr( $stripped, 0, 2 ) == 'on' ) {
1394 wfDebug( __METHOD__
1395 . ": Found event-handler attribute '$attrib'='$value' in uploaded file.\n" );
1396
1397 return array( 'uploaded-event-handler-on-svg', $attrib, $value );
1398 }
1399
1400 # href with non-local target (don't allow http://, javascript:, etc)
1401 if ( $stripped == 'href'
1402 && strpos( $value, 'data:' ) !== 0
1403 && strpos( $value, '#' ) !== 0
1404 ) {
1405 if ( !( $strippedElement === 'a'
1406 && preg_match( '!^https?://!im', $value ) )
1407 ) {
1408 wfDebug( __METHOD__ . ": Found href attribute <$strippedElement "
1409 . "'$attrib'='$value' in uploaded file.\n" );
1410
1411 return array( 'uploaded-href-attribute-svg', $strippedElement, $attrib, $value );
1412 }
1413 }
1414
1415 # only allow data: targets that should be safe. This prevents vectors like,
1416 # image/svg, text/xml, application/xml, and text/html, which can contain scripts
1417 if ( $stripped == 'href' && strncasecmp( 'data:', $value, 5 ) === 0 ) {
1418 // rfc2397 parameters. This is only slightly slower than (;[\w;]+)*.
1419 $parameters = '(?>;[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+=(?>[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+|"(?>[\0-\x0c\x0e-\x21\x23-\x5b\x5d-\x7f]+|\\\\[\0-\x7f])*"))*(?:;base64)?';
1420 if ( !preg_match( "!^data:\s*image/(gif|jpeg|jpg|png)$parameters,!i", $value ) ) {
1421 wfDebug( __METHOD__ . ": Found href to unwhitelisted data: uri "
1422 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1423 return array( 'uploaded-href-unsafe-target-svg', $strippedElement, $attrib, $value );
1424 }
1425 }
1426
1427 # Change href with animate from (http://html5sec.org/#137).
1428 if ( $stripped === 'attributename'
1429 && $strippedElement === 'animate'
1430 && $this->stripXmlNamespace( $value ) == 'href'
1431 ) {
1432 wfDebug( __METHOD__ . ": Found animate that might be changing href using from "
1433 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1434
1435 return array( 'uploaded-animate-svg', $strippedElement, $attrib, $value );
1436 }
1437
1438 # use set/animate to add event-handler attribute to parent
1439 if ( ( $strippedElement == 'set' || $strippedElement == 'animate' )
1440 && $stripped == 'attributename'
1441 && substr( $value, 0, 2 ) == 'on'
1442 ) {
1443 wfDebug( __METHOD__ . ": Found svg setting event-handler attribute with "
1444 . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1445
1446 return array( 'uploaded-setting-event-handler-svg', $strippedElement, $stripped, $value );
1447 }
1448
1449 # use set to add href attribute to parent element
1450 if ( $strippedElement == 'set'
1451 && $stripped == 'attributename'
1452 && strpos( $value, 'href' ) !== false
1453 ) {
1454 wfDebug( __METHOD__ . ": Found svg setting href attribute '$value' in uploaded file.\n" );
1455
1456 return array( 'uploaded-setting-href-svg' );
1457 }
1458
1459 # use set to add a remote / data / script target to an element
1460 if ( $strippedElement == 'set'
1461 && $stripped == 'to'
1462 && preg_match( '!(http|https|data|script):!sim', $value )
1463 ) {
1464 wfDebug( __METHOD__ . ": Found svg setting attribute to '$value' in uploaded file.\n" );
1465
1466 return array( 'uploaded-wrong-setting-svg', $value );
1467 }
1468
1469 # use handler attribute with remote / data / script
1470 if ( $stripped == 'handler' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1471 wfDebug( __METHOD__ . ": Found svg setting handler with remote/data/script "
1472 . "'$attrib'='$value' in uploaded file.\n" );
1473
1474 return array( 'uploaded-setting-handler-svg', $attrib, $value );
1475 }
1476
1477 # use CSS styles to bring in remote code
1478 if ( $stripped == 'style'
1479 && self::checkCssFragment( Sanitizer::normalizeCss( $value ) )
1480 ) {
1481 wfDebug( __METHOD__ . ": Found svg setting a style with "
1482 . "remote url '$attrib'='$value' in uploaded file.\n" );
1483 return array( 'uploaded-remote-url-svg', $attrib, $value );
1484 }
1485
1486 # Several attributes can include css, css character escaping isn't allowed
1487 $cssAttrs = array( 'font', 'clip-path', 'fill', 'filter', 'marker',
1488 'marker-end', 'marker-mid', 'marker-start', 'mask', 'stroke' );
1489 if ( in_array( $stripped, $cssAttrs )
1490 && self::checkCssFragment( $value )
1491 ) {
1492 wfDebug( __METHOD__ . ": Found svg setting a style with "
1493 . "remote url '$attrib'='$value' in uploaded file.\n" );
1494 return array( 'uploaded-remote-url-svg', $attrib, $value );
1495 }
1496
1497 # image filters can pull in url, which could be svg that executes scripts
1498 if ( $strippedElement == 'image'
1499 && $stripped == 'filter'
1500 && preg_match( '!url\s*\(!sim', $value )
1501 ) {
1502 wfDebug( __METHOD__ . ": Found image filter with url: "
1503 . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1504
1505 return array( 'uploaded-image-filter-svg', $strippedElement, $stripped, $value );
1506 }
1507 }
1508
1509 return false; //No scripts detected
1510 }
1511
1512 /**
1513 * Check a block of CSS or CSS fragment for anything that looks like
1514 * it is bringing in remote code.
1515 * @param string $value a string of CSS
1516 * @param bool $propOnly only check css properties (start regex with :)
1517 * @return bool true if the CSS contains an illegal string, false if otherwise
1518 */
1519 private static function checkCssFragment( $value ) {
1520
1521 # Forbid external stylesheets, for both reliability and to protect viewer's privacy
1522 if ( stripos( $value, '@import' ) !== false ) {
1523 return true;
1524 }
1525
1526 # We allow @font-face to embed fonts with data: urls, so we snip the string
1527 # 'url' out so this case won't match when we check for urls below
1528 $pattern = '!(@font-face\s*{[^}]*src:)url(\("data:;base64,)!im';
1529 $value = preg_replace( $pattern, '$1$2', $value );
1530
1531 # Check for remote and executable CSS. Unlike in Sanitizer::checkCss, the CSS
1532 # properties filter and accelerator don't seem to be useful for xss in SVG files.
1533 # Expression and -o-link don't seem to work either, but filtering them here in case.
1534 # Additionally, we catch remote urls like url("http:..., url('http:..., url(http:...,
1535 # but not local ones such as url("#..., url('#..., url(#....
1536 if ( preg_match( '!expression
1537 | -o-link\s*:
1538 | -o-link-source\s*:
1539 | -o-replace\s*:!imx', $value ) ) {
1540 return true;
1541 }
1542
1543 if ( preg_match_all(
1544 "!(\s*(url|image|image-set)\s*\(\s*[\"']?\s*[^#]+.*?\))!sim",
1545 $value,
1546 $matches
1547 ) !== 0
1548 ) {
1549 # TODO: redo this in one regex. Until then, url("#whatever") matches the first
1550 foreach ( $matches[1] as $match ) {
1551 if ( !preg_match( "!\s*(url|image|image-set)\s*\(\s*(#|'#|\"#)!im", $match ) ) {
1552 return true;
1553 }
1554 }
1555 }
1556
1557 if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ) {
1558 return true;
1559 }
1560
1561 return false;
1562 }
1563
1564 /**
1565 * Divide the element name passed by the xml parser to the callback into URI and prifix.
1566 * @param string $element
1567 * @return array Containing the namespace URI and prefix
1568 */
1569 private static function splitXmlNamespace( $element ) {
1570 // 'http://www.w3.org/2000/svg:script' -> array( 'http://www.w3.org/2000/svg', 'script' )
1571 $parts = explode( ':', strtolower( $element ) );
1572 $name = array_pop( $parts );
1573 $ns = implode( ':', $parts );
1574
1575 return array( $ns, $name );
1576 }
1577
1578 /**
1579 * @param string $name
1580 * @return string
1581 */
1582 private function stripXmlNamespace( $name ) {
1583 // 'http://www.w3.org/2000/svg:script' -> 'script'
1584 $parts = explode( ':', strtolower( $name ) );
1585
1586 return array_pop( $parts );
1587 }
1588
1589 /**
1590 * Generic wrapper function for a virus scanner program.
1591 * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
1592 * $wgAntivirusRequired may be used to deny upload if the scan fails.
1593 *
1594 * @param string $file Pathname to the temporary upload file
1595 * @return mixed False if not virus is found, null if the scan fails or is disabled,
1596 * or a string containing feedback from the virus scanner if a virus was found.
1597 * If textual feedback is missing but a virus was found, this function returns true.
1598 */
1599 public static function detectVirus( $file ) {
1600 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
1601
1602 if ( !$wgAntivirus ) {
1603 wfDebug( __METHOD__ . ": virus scanner disabled\n" );
1604
1605 return null;
1606 }
1607
1608 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
1609 wfDebug( __METHOD__ . ": unknown virus scanner: $wgAntivirus\n" );
1610 $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
1611 array( 'virus-badscanner', $wgAntivirus ) );
1612
1613 return wfMessage( 'virus-unknownscanner' )->text() . " $wgAntivirus";
1614 }
1615
1616 # look up scanner configuration
1617 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
1618 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
1619 $msgPattern = isset( $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ) ?
1620 $wgAntivirusSetup[$wgAntivirus]['messagepattern'] : null;
1621
1622 if ( strpos( $command, "%f" ) === false ) {
1623 # simple pattern: append file to scan
1624 $command .= " " . wfEscapeShellArg( $file );
1625 } else {
1626 # complex pattern: replace "%f" with file to scan
1627 $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
1628 }
1629
1630 wfDebug( __METHOD__ . ": running virus scan: $command \n" );
1631
1632 # execute virus scanner
1633 $exitCode = false;
1634
1635 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1636 # that does not seem to be worth the pain.
1637 # Ask me (Duesentrieb) about it if it's ever needed.
1638 $output = wfShellExecWithStderr( $command, $exitCode );
1639
1640 # map exit code to AV_xxx constants.
1641 $mappedCode = $exitCode;
1642 if ( $exitCodeMap ) {
1643 if ( isset( $exitCodeMap[$exitCode] ) ) {
1644 $mappedCode = $exitCodeMap[$exitCode];
1645 } elseif ( isset( $exitCodeMap["*"] ) ) {
1646 $mappedCode = $exitCodeMap["*"];
1647 }
1648 }
1649
1650 /* NB: AV_NO_VIRUS is 0 but AV_SCAN_FAILED is false,
1651 * so we need the strict equalities === and thus can't use a switch here
1652 */
1653 if ( $mappedCode === AV_SCAN_FAILED ) {
1654 # scan failed (code was mapped to false by $exitCodeMap)
1655 wfDebug( __METHOD__ . ": failed to scan $file (code $exitCode).\n" );
1656
1657 $output = $wgAntivirusRequired
1658 ? wfMessage( 'virus-scanfailed', array( $exitCode ) )->text()
1659 : null;
1660 } elseif ( $mappedCode === AV_SCAN_ABORTED ) {
1661 # scan failed because filetype is unknown (probably imune)
1662 wfDebug( __METHOD__ . ": unsupported file type $file (code $exitCode).\n" );
1663 $output = null;
1664 } elseif ( $mappedCode === AV_NO_VIRUS ) {
1665 # no virus found
1666 wfDebug( __METHOD__ . ": file passed virus scan.\n" );
1667 $output = false;
1668 } else {
1669 $output = trim( $output );
1670
1671 if ( !$output ) {
1672 $output = true; #if there's no output, return true
1673 } elseif ( $msgPattern ) {
1674 $groups = array();
1675 if ( preg_match( $msgPattern, $output, $groups ) ) {
1676 if ( $groups[1] ) {
1677 $output = $groups[1];
1678 }
1679 }
1680 }
1681
1682 wfDebug( __METHOD__ . ": FOUND VIRUS! scanner feedback: $output \n" );
1683 }
1684
1685 return $output;
1686 }
1687
1688 /**
1689 * Check if there's an overwrite conflict and, if so, if restrictions
1690 * forbid this user from performing the upload.
1691 *
1692 * @param User $user
1693 *
1694 * @return mixed True on success, array on failure
1695 */
1696 private function checkOverwrite( $user ) {
1697 // First check whether the local file can be overwritten
1698 $file = $this->getLocalFile();
1699 $file->load( File::READ_LATEST );
1700 if ( $file->exists() ) {
1701 if ( !self::userCanReUpload( $user, $file ) ) {
1702 return array( 'fileexists-forbidden', $file->getName() );
1703 } else {
1704 return true;
1705 }
1706 }
1707
1708 /* Check shared conflicts: if the local file does not exist, but
1709 * wfFindFile finds a file, it exists in a shared repository.
1710 */
1711 $file = wfFindFile( $this->getTitle(), array( 'latest' => true ) );
1712 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1713 return array( 'fileexists-shared-forbidden', $file->getName() );
1714 }
1715
1716 return true;
1717 }
1718
1719 /**
1720 * Check if a user is the last uploader
1721 *
1722 * @param User $user
1723 * @param string $img Image name
1724 * @return bool
1725 */
1726 public static function userCanReUpload( User $user, $img ) {
1727 if ( $user->isAllowed( 'reupload' ) ) {
1728 return true; // non-conditional
1729 }
1730 if ( !$user->isAllowed( 'reupload-own' ) ) {
1731 return false;
1732 }
1733 if ( is_string( $img ) ) {
1734 $img = wfLocalFile( $img );
1735 }
1736 if ( !( $img instanceof LocalFile ) ) {
1737 return false;
1738 }
1739
1740 $img->load( File::READ_LATEST );
1741
1742 return $user->getId() == $img->getUser( 'id' );
1743 }
1744
1745 /**
1746 * Helper function that does various existence checks for a file.
1747 * The following checks are performed:
1748 * - The file exists
1749 * - Article with the same name as the file exists
1750 * - File exists with normalized extension
1751 * - The file looks like a thumbnail and the original exists
1752 *
1753 * @param File $file The File object to check
1754 * @return mixed False if the file does not exists, else an array
1755 */
1756 public static function getExistsWarning( $file ) {
1757 if ( $file->exists() ) {
1758 return array( 'warning' => 'exists', 'file' => $file );
1759 }
1760
1761 if ( $file->getTitle()->getArticleID() ) {
1762 return array( 'warning' => 'page-exists', 'file' => $file );
1763 }
1764
1765 if ( $file->wasDeleted() && !$file->exists() ) {
1766 return array( 'warning' => 'was-deleted', 'file' => $file );
1767 }
1768
1769 if ( strpos( $file->getName(), '.' ) == false ) {
1770 $partname = $file->getName();
1771 $extension = '';
1772 } else {
1773 $n = strrpos( $file->getName(), '.' );
1774 $extension = substr( $file->getName(), $n + 1 );
1775 $partname = substr( $file->getName(), 0, $n );
1776 }
1777 $normalizedExtension = File::normalizeExtension( $extension );
1778
1779 if ( $normalizedExtension != $extension ) {
1780 // We're not using the normalized form of the extension.
1781 // Normal form is lowercase, using most common of alternate
1782 // extensions (eg 'jpg' rather than 'JPEG').
1783 //
1784 // Check for another file using the normalized form...
1785 $nt_lc = Title::makeTitle( NS_FILE, "{$partname}.{$normalizedExtension}" );
1786 $file_lc = wfLocalFile( $nt_lc );
1787
1788 if ( $file_lc->exists() ) {
1789 return array(
1790 'warning' => 'exists-normalized',
1791 'file' => $file,
1792 'normalizedFile' => $file_lc
1793 );
1794 }
1795 }
1796
1797 // Check for files with the same name but a different extension
1798 $similarFiles = RepoGroup::singleton()->getLocalRepo()->findFilesByPrefix(
1799 "{$partname}.", 1 );
1800 if ( count( $similarFiles ) ) {
1801 return array(
1802 'warning' => 'exists-normalized',
1803 'file' => $file,
1804 'normalizedFile' => $similarFiles[0],
1805 );
1806 }
1807
1808 if ( self::isThumbName( $file->getName() ) ) {
1809 # Check for filenames like 50px- or 180px-, these are mostly thumbnails
1810 $nt_thb = Title::newFromText(
1811 substr( $partname, strpos( $partname, '-' ) + 1 ) . '.' . $extension,
1812 NS_FILE
1813 );
1814 $file_thb = wfLocalFile( $nt_thb );
1815 if ( $file_thb->exists() ) {
1816 return array(
1817 'warning' => 'thumb',
1818 'file' => $file,
1819 'thumbFile' => $file_thb
1820 );
1821 } else {
1822 // File does not exist, but we just don't like the name
1823 return array(
1824 'warning' => 'thumb-name',
1825 'file' => $file,
1826 'thumbFile' => $file_thb
1827 );
1828 }
1829 }
1830
1831 foreach ( self::getFilenamePrefixBlacklist() as $prefix ) {
1832 if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
1833 return array(
1834 'warning' => 'bad-prefix',
1835 'file' => $file,
1836 'prefix' => $prefix
1837 );
1838 }
1839 }
1840
1841 return false;
1842 }
1843
1844 /**
1845 * Helper function that checks whether the filename looks like a thumbnail
1846 * @param string $filename
1847 * @return bool
1848 */
1849 public static function isThumbName( $filename ) {
1850 $n = strrpos( $filename, '.' );
1851 $partname = $n ? substr( $filename, 0, $n ) : $filename;
1852
1853 return (
1854 substr( $partname, 3, 3 ) == 'px-' ||
1855 substr( $partname, 2, 3 ) == 'px-'
1856 ) &&
1857 preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) );
1858 }
1859
1860 /**
1861 * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
1862 *
1863 * @return array List of prefixes
1864 */
1865 public static function getFilenamePrefixBlacklist() {
1866 $blacklist = array();
1867 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
1868 if ( !$message->isDisabled() ) {
1869 $lines = explode( "\n", $message->plain() );
1870 foreach ( $lines as $line ) {
1871 // Remove comment lines
1872 $comment = substr( trim( $line ), 0, 1 );
1873 if ( $comment == '#' || $comment == '' ) {
1874 continue;
1875 }
1876 // Remove additional comments after a prefix
1877 $comment = strpos( $line, '#' );
1878 if ( $comment > 0 ) {
1879 $line = substr( $line, 0, $comment - 1 );
1880 }
1881 $blacklist[] = trim( $line );
1882 }
1883 }
1884
1885 return $blacklist;
1886 }
1887
1888 /**
1889 * Gets image info about the file just uploaded.
1890 *
1891 * Also has the effect of setting metadata to be an 'indexed tag name' in
1892 * returned API result if 'metadata' was requested. Oddly, we have to pass
1893 * the "result" object down just so it can do that with the appropriate
1894 * format, presumably.
1895 *
1896 * @param ApiResult $result
1897 * @return array Image info
1898 */
1899 public function getImageInfo( $result ) {
1900 $file = $this->getLocalFile();
1901 /** @todo This cries out for refactoring.
1902 * We really want to say $file->getAllInfo(); here.
1903 * Perhaps "info" methods should be moved into files, and the API should
1904 * just wrap them in queries.
1905 */
1906 if ( $file instanceof UploadStashFile ) {
1907 $imParam = ApiQueryStashImageInfo::getPropertyNames();
1908 $info = ApiQueryStashImageInfo::getInfo( $file, array_flip( $imParam ), $result );
1909 } else {
1910 $imParam = ApiQueryImageInfo::getPropertyNames();
1911 $info = ApiQueryImageInfo::getInfo( $file, array_flip( $imParam ), $result );
1912 }
1913
1914 return $info;
1915 }
1916
1917 /**
1918 * @param array $error
1919 * @return Status
1920 */
1921 public function convertVerifyErrorToStatus( $error ) {
1922 $code = $error['status'];
1923 unset( $code['status'] );
1924
1925 return Status::newFatal( $this->getVerificationErrorCode( $code ), $error );
1926 }
1927
1928 /**
1929 * @param null|string $forType
1930 * @return int
1931 */
1932 public static function getMaxUploadSize( $forType = null ) {
1933 global $wgMaxUploadSize;
1934
1935 if ( is_array( $wgMaxUploadSize ) ) {
1936 if ( !is_null( $forType ) && isset( $wgMaxUploadSize[$forType] ) ) {
1937 return $wgMaxUploadSize[$forType];
1938 } else {
1939 return $wgMaxUploadSize['*'];
1940 }
1941 } else {
1942 return intval( $wgMaxUploadSize );
1943 }
1944 }
1945
1946 /**
1947 * Get the current status of a chunked upload (used for polling)
1948 *
1949 * The value will be read from cache.
1950 *
1951 * @param User $user
1952 * @param string $statusKey
1953 * @return Status[]|bool
1954 */
1955 public static function getSessionStatus( User $user, $statusKey ) {
1956 $key = wfMemcKey( 'uploadstatus', $user->getId() ?: md5( $user->getName() ), $statusKey );
1957
1958 return wfGetCache( CACHE_ANYTHING )->get( $key );
1959 }
1960
1961 /**
1962 * Set the current status of a chunked upload (used for polling)
1963 *
1964 * The value will be set in cache for 1 day
1965 *
1966 * @param User $user
1967 * @param string $statusKey
1968 * @param array|bool $value
1969 * @return void
1970 */
1971 public static function setSessionStatus( User $user, $statusKey, $value ) {
1972 $key = wfMemcKey( 'uploadstatus', $user->getId() ?: md5( $user->getName() ), $statusKey );
1973
1974 $cache = wfGetCache( CACHE_ANYTHING );
1975 if ( $value === false ) {
1976 $cache->delete( $key );
1977 } else {
1978 $cache->set( $key, $value, 86400 );
1979 }
1980 }
1981 }