Merge "PrefixSearch: Add unit tests for StringPrefixSearch"
[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 wfRunHooks( '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 wfProfileIn( __METHOD__ );
265 $repo = RepoGroup::singleton()->getLocalRepo();
266 if ( $repo->isVirtualUrl( $srcPath ) ) {
267 /** @todo Just make uploads work with storage paths UploadFromStash
268 * loads files via virtual URLs.
269 */
270 $tmpFile = $repo->getLocalCopy( $srcPath );
271 if ( $tmpFile ) {
272 $tmpFile->bind( $this ); // keep alive with $this
273 }
274 $path = $tmpFile ? $tmpFile->getPath() : false;
275 } else {
276 $path = $srcPath;
277 }
278 wfProfileOut( __METHOD__ );
279
280 return $path;
281 }
282
283 /**
284 * Verify whether the upload is sane.
285 * @return mixed Const self::OK or else an array with error information
286 */
287 public function verifyUpload() {
288 wfProfileIn( __METHOD__ );
289
290 /**
291 * If there was no filename or a zero size given, give up quick.
292 */
293 if ( $this->isEmptyFile() ) {
294 wfProfileOut( __METHOD__ );
295
296 return array( 'status' => self::EMPTY_FILE );
297 }
298
299 /**
300 * Honor $wgMaxUploadSize
301 */
302 $maxSize = self::getMaxUploadSize( $this->getSourceType() );
303 if ( $this->mFileSize > $maxSize ) {
304 wfProfileOut( __METHOD__ );
305
306 return array(
307 'status' => self::FILE_TOO_LARGE,
308 'max' => $maxSize,
309 );
310 }
311
312 /**
313 * Look at the contents of the file; if we can recognize the
314 * type but it's corrupt or data of the wrong type, we should
315 * probably not accept it.
316 */
317 $verification = $this->verifyFile();
318 if ( $verification !== true ) {
319 wfProfileOut( __METHOD__ );
320
321 return array(
322 'status' => self::VERIFICATION_ERROR,
323 'details' => $verification
324 );
325 }
326
327 /**
328 * Make sure this file can be created
329 */
330 $result = $this->validateName();
331 if ( $result !== true ) {
332 wfProfileOut( __METHOD__ );
333
334 return $result;
335 }
336
337 $error = '';
338 if ( !wfRunHooks( 'UploadVerification',
339 array( $this->mDestName, $this->mTempPath, &$error ) )
340 ) {
341 wfProfileOut( __METHOD__ );
342
343 return array( 'status' => self::HOOK_ABORTED, 'error' => $error );
344 }
345
346 wfProfileOut( __METHOD__ );
347
348 return array( 'status' => self::OK );
349 }
350
351 /**
352 * Verify that the name is valid and, if necessary, that we can overwrite
353 *
354 * @return mixed True if valid, otherwise and array with 'status'
355 * and other keys
356 */
357 public function validateName() {
358 $nt = $this->getTitle();
359 if ( is_null( $nt ) ) {
360 $result = array( 'status' => $this->mTitleError );
361 if ( $this->mTitleError == self::ILLEGAL_FILENAME ) {
362 $result['filtered'] = $this->mFilteredName;
363 }
364 if ( $this->mTitleError == self::FILETYPE_BADTYPE ) {
365 $result['finalExt'] = $this->mFinalExtension;
366 if ( count( $this->mBlackListedExtensions ) ) {
367 $result['blacklistedExt'] = $this->mBlackListedExtensions;
368 }
369 }
370
371 return $result;
372 }
373 $this->mDestName = $this->getLocalFile()->getName();
374
375 return true;
376 }
377
378 /**
379 * Verify the MIME type.
380 *
381 * @note Only checks that it is not an evil MIME. The "does it have
382 * correct extension given its MIME type?" check is in verifyFile.
383 * in `verifyFile()` that MIME type and file extension correlate.
384 * @param string $mime Representing the MIME
385 * @return mixed True if the file is verified, an array otherwise
386 */
387 protected function verifyMimeType( $mime ) {
388 global $wgVerifyMimeType;
389 wfProfileIn( __METHOD__ );
390 if ( $wgVerifyMimeType ) {
391 wfDebug( "mime: <$mime> extension: <{$this->mFinalExtension}>\n" );
392 global $wgMimeTypeBlacklist;
393 if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
394 wfProfileOut( __METHOD__ );
395
396 return array( 'filetype-badmime', $mime );
397 }
398
399 # Check what Internet Explorer would detect
400 $fp = fopen( $this->mTempPath, 'rb' );
401 $chunk = fread( $fp, 256 );
402 fclose( $fp );
403
404 $magic = MimeMagic::singleton();
405 $extMime = $magic->guessTypesForExtension( $this->mFinalExtension );
406 $ieTypes = $magic->getIEMimeTypes( $this->mTempPath, $chunk, $extMime );
407 foreach ( $ieTypes as $ieType ) {
408 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
409 wfProfileOut( __METHOD__ );
410
411 return array( 'filetype-bad-ie-mime', $ieType );
412 }
413 }
414 }
415
416 wfProfileOut( __METHOD__ );
417
418 return true;
419 }
420
421 /**
422 * Verifies that it's ok to include the uploaded file
423 *
424 * @return mixed True of the file is verified, array otherwise.
425 */
426 protected function verifyFile() {
427 global $wgVerifyMimeType;
428 wfProfileIn( __METHOD__ );
429
430 $status = $this->verifyPartialFile();
431 if ( $status !== true ) {
432 wfProfileOut( __METHOD__ );
433
434 return $status;
435 }
436
437 $this->mFileProps = FSFile::getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
438 $mime = $this->mFileProps['mime'];
439
440 if ( $wgVerifyMimeType ) {
441 # XXX: Missing extension will be caught by validateName() via getTitle()
442 if ( $this->mFinalExtension != '' && !$this->verifyExtension( $mime, $this->mFinalExtension ) ) {
443 wfProfileOut( __METHOD__ );
444
445 return array( 'filetype-mime-mismatch', $this->mFinalExtension, $mime );
446 }
447 }
448
449 $handler = MediaHandler::getHandler( $mime );
450 if ( $handler ) {
451 $handlerStatus = $handler->verifyUpload( $this->mTempPath );
452 if ( !$handlerStatus->isOK() ) {
453 $errors = $handlerStatus->getErrorsArray();
454 wfProfileOut( __METHOD__ );
455
456 return reset( $errors );
457 }
458 }
459
460 wfRunHooks( 'UploadVerifyFile', array( $this, $mime, &$status ) );
461 if ( $status !== true ) {
462 wfProfileOut( __METHOD__ );
463
464 return $status;
465 }
466
467 wfDebug( __METHOD__ . ": all clear; passing.\n" );
468 wfProfileOut( __METHOD__ );
469
470 return true;
471 }
472
473 /**
474 * A verification routine suitable for partial files
475 *
476 * Runs the blacklist checks, but not any checks that may
477 * assume the entire file is present.
478 *
479 * @return mixed True for valid or array with error message key.
480 */
481 protected function verifyPartialFile() {
482 global $wgAllowJavaUploads, $wgDisableUploadScriptChecks;
483 wfProfileIn( __METHOD__ );
484
485 # getTitle() sets some internal parameters like $this->mFinalExtension
486 $this->getTitle();
487
488 $this->mFileProps = FSFile::getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
489
490 # check MIME type, if desired
491 $mime = $this->mFileProps['file-mime'];
492 $status = $this->verifyMimeType( $mime );
493 if ( $status !== true ) {
494 wfProfileOut( __METHOD__ );
495
496 return $status;
497 }
498
499 # check for htmlish code and javascript
500 if ( !$wgDisableUploadScriptChecks ) {
501 if ( self::detectScript( $this->mTempPath, $mime, $this->mFinalExtension ) ) {
502 wfProfileOut( __METHOD__ );
503
504 return array( 'uploadscripted' );
505 }
506 if ( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
507 $svgStatus = $this->detectScriptInSvg( $this->mTempPath );
508 if ( $svgStatus !== false ) {
509 wfProfileOut( __METHOD__ );
510
511 return $svgStatus;
512 }
513 }
514 }
515
516 # Check for Java applets, which if uploaded can bypass cross-site
517 # restrictions.
518 if ( !$wgAllowJavaUploads ) {
519 $this->mJavaDetected = false;
520 $zipStatus = ZipDirectoryReader::read( $this->mTempPath,
521 array( $this, 'zipEntryCallback' ) );
522 if ( !$zipStatus->isOK() ) {
523 $errors = $zipStatus->getErrorsArray();
524 $error = reset( $errors );
525 if ( $error[0] !== 'zip-wrong-format' ) {
526 wfProfileOut( __METHOD__ );
527
528 return $error;
529 }
530 }
531 if ( $this->mJavaDetected ) {
532 wfProfileOut( __METHOD__ );
533
534 return array( 'uploadjava' );
535 }
536 }
537
538 # Scan the uploaded file for viruses
539 $virus = $this->detectVirus( $this->mTempPath );
540 if ( $virus ) {
541 wfProfileOut( __METHOD__ );
542
543 return array( 'uploadvirus', $virus );
544 }
545
546 wfProfileOut( __METHOD__ );
547
548 return true;
549 }
550
551 /**
552 * Callback for ZipDirectoryReader to detect Java class files.
553 *
554 * @param array $entry
555 */
556 function zipEntryCallback( $entry ) {
557 $names = array( $entry['name'] );
558
559 // If there is a null character, cut off the name at it, because JDK's
560 // ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name
561 // were constructed which had ".class\0" followed by a string chosen to
562 // make the hash collide with the truncated name, that file could be
563 // returned in response to a request for the .class file.
564 $nullPos = strpos( $entry['name'], "\000" );
565 if ( $nullPos !== false ) {
566 $names[] = substr( $entry['name'], 0, $nullPos );
567 }
568
569 // If there is a trailing slash in the file name, we have to strip it,
570 // because that's what ZIP_GetEntry() does.
571 if ( preg_grep( '!\.class/?$!', $names ) ) {
572 $this->mJavaDetected = true;
573 }
574 }
575
576 /**
577 * Alias for verifyTitlePermissions. The function was originally
578 * 'verifyPermissions', but that suggests it's checking the user, when it's
579 * really checking the title + user combination.
580 *
581 * @param User $user User object to verify the permissions against
582 * @return mixed An array as returned by getUserPermissionsErrors or true
583 * in case the user has proper permissions.
584 */
585 public function verifyPermissions( $user ) {
586 return $this->verifyTitlePermissions( $user );
587 }
588
589 /**
590 * Check whether the user can edit, upload and create the image. This
591 * checks only against the current title; if it returns errors, it may
592 * very well be that another title will not give errors. Therefore
593 * isAllowed() should be called as well for generic is-user-blocked or
594 * can-user-upload checking.
595 *
596 * @param User $user User object to verify the permissions against
597 * @return mixed An array as returned by getUserPermissionsErrors or true
598 * in case the user has proper permissions.
599 */
600 public function verifyTitlePermissions( $user ) {
601 /**
602 * If the image is protected, non-sysop users won't be able
603 * to modify it by uploading a new revision.
604 */
605 $nt = $this->getTitle();
606 if ( is_null( $nt ) ) {
607 return true;
608 }
609 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
610 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
611 if ( !$nt->exists() ) {
612 $permErrorsCreate = $nt->getUserPermissionsErrors( 'create', $user );
613 } else {
614 $permErrorsCreate = array();
615 }
616 if ( $permErrors || $permErrorsUpload || $permErrorsCreate ) {
617 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
618 $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
619
620 return $permErrors;
621 }
622
623 $overwriteError = $this->checkOverwrite( $user );
624 if ( $overwriteError !== true ) {
625 return array( $overwriteError );
626 }
627
628 return true;
629 }
630
631 /**
632 * Check for non fatal problems with the file.
633 *
634 * This should not assume that mTempPath is set.
635 *
636 * @return array Array of warnings
637 */
638 public function checkWarnings() {
639 global $wgLang;
640 wfProfileIn( __METHOD__ );
641
642 $warnings = array();
643
644 $localFile = $this->getLocalFile();
645 $filename = $localFile->getName();
646
647 /**
648 * Check whether the resulting filename is different from the desired one,
649 * but ignore things like ucfirst() and spaces/underscore things
650 */
651 $comparableName = str_replace( ' ', '_', $this->mDesiredDestName );
652 $comparableName = Title::capitalize( $comparableName, NS_FILE );
653
654 if ( $this->mDesiredDestName != $filename && $comparableName != $filename ) {
655 $warnings['badfilename'] = $filename;
656 // Debugging for bug 62241
657 wfDebugLog( 'upload', "Filename: '$filename', mDesiredDestName: "
658 . "'$this->mDesiredDestName', comparableName: '$comparableName'" );
659 }
660
661 // Check whether the file extension is on the unwanted list
662 global $wgCheckFileExtensions, $wgFileExtensions;
663 if ( $wgCheckFileExtensions ) {
664 $extensions = array_unique( $wgFileExtensions );
665 if ( !$this->checkFileExtension( $this->mFinalExtension, $extensions ) ) {
666 $warnings['filetype-unwanted-type'] = array( $this->mFinalExtension,
667 $wgLang->commaList( $extensions ), count( $extensions ) );
668 }
669 }
670
671 global $wgUploadSizeWarning;
672 if ( $wgUploadSizeWarning && ( $this->mFileSize > $wgUploadSizeWarning ) ) {
673 $warnings['large-file'] = array( $wgUploadSizeWarning, $this->mFileSize );
674 }
675
676 if ( $this->mFileSize == 0 ) {
677 $warnings['emptyfile'] = true;
678 }
679
680 $exists = self::getExistsWarning( $localFile );
681 if ( $exists !== false ) {
682 $warnings['exists'] = $exists;
683 }
684
685 // Check dupes against existing files
686 $hash = $this->getTempFileSha1Base36();
687 $dupes = RepoGroup::singleton()->findBySha1( $hash );
688 $title = $this->getTitle();
689 // Remove all matches against self
690 foreach ( $dupes as $key => $dupe ) {
691 if ( $title->equals( $dupe->getTitle() ) ) {
692 unset( $dupes[$key] );
693 }
694 }
695 if ( $dupes ) {
696 $warnings['duplicate'] = $dupes;
697 }
698
699 // Check dupes against archives
700 $archivedImage = new ArchivedFile( null, 0, "{$hash}.{$this->mFinalExtension}" );
701 if ( $archivedImage->getID() > 0 ) {
702 if ( $archivedImage->userCan( File::DELETED_FILE ) ) {
703 $warnings['duplicate-archive'] = $archivedImage->getName();
704 } else {
705 $warnings['duplicate-archive'] = '';
706 }
707 }
708
709 wfProfileOut( __METHOD__ );
710
711 return $warnings;
712 }
713
714 /**
715 * Really perform the upload. Stores the file in the local repo, watches
716 * if necessary and runs the UploadComplete hook.
717 *
718 * @param string $comment
719 * @param string $pageText
720 * @param bool $watch
721 * @param User $user
722 *
723 * @return Status Indicating the whether the upload succeeded.
724 */
725 public function performUpload( $comment, $pageText, $watch, $user ) {
726 wfProfileIn( __METHOD__ );
727
728 $status = $this->getLocalFile()->upload(
729 $this->mTempPath,
730 $comment,
731 $pageText,
732 File::DELETE_SOURCE,
733 $this->mFileProps,
734 false,
735 $user
736 );
737
738 if ( $status->isGood() ) {
739 if ( $watch ) {
740 WatchAction::doWatch(
741 $this->getLocalFile()->getTitle(),
742 $user,
743 WatchedItem::IGNORE_USER_RIGHTS
744 );
745 }
746 wfRunHooks( 'UploadComplete', array( &$this ) );
747
748 $this->postProcessUpload();
749 }
750
751 wfProfileOut( __METHOD__ );
752
753 return $status;
754 }
755
756 /**
757 * Perform extra steps after a successful upload.
758 *
759 * @since 1.25
760 */
761 public function postProcessUpload() {
762 global $wgUploadThumbnailRenderMap;
763
764 $jobs = array();
765
766 $sizes = $wgUploadThumbnailRenderMap;
767 rsort( $sizes );
768
769 $file = $this->getLocalFile();
770
771 foreach ( $sizes as $size ) {
772 if ( $file->isVectorized()
773 || $file->getWidth() > $size ) {
774 $jobs[] = new ThumbnailRenderJob( $file->getTitle(), array(
775 'transformParams' => array( 'width' => $size ),
776 ) );
777 }
778 }
779
780 if ( $jobs ) {
781 JobQueueGroup::singleton()->push( $jobs );
782 }
783 }
784
785 /**
786 * Returns the title of the file to be uploaded. Sets mTitleError in case
787 * the name was illegal.
788 *
789 * @return Title The title of the file or null in case the name was illegal
790 */
791 public function getTitle() {
792 if ( $this->mTitle !== false ) {
793 return $this->mTitle;
794 }
795 /* Assume that if a user specified File:Something.jpg, this is an error
796 * and that the namespace prefix needs to be stripped of.
797 */
798 $title = Title::newFromText( $this->mDesiredDestName );
799 if ( $title && $title->getNamespace() == NS_FILE ) {
800 $this->mFilteredName = $title->getDBkey();
801 } else {
802 $this->mFilteredName = $this->mDesiredDestName;
803 }
804
805 # oi_archive_name is max 255 bytes, which include a timestamp and an
806 # exclamation mark, so restrict file name to 240 bytes.
807 if ( strlen( $this->mFilteredName ) > 240 ) {
808 $this->mTitleError = self::FILENAME_TOO_LONG;
809 $this->mTitle = null;
810
811 return $this->mTitle;
812 }
813
814 /**
815 * Chop off any directories in the given filename. Then
816 * filter out illegal characters, and try to make a legible name
817 * out of it. We'll strip some silently that Title would die on.
818 */
819 $this->mFilteredName = wfStripIllegalFilenameChars( $this->mFilteredName );
820 /* Normalize to title form before we do any further processing */
821 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
822 if ( is_null( $nt ) ) {
823 $this->mTitleError = self::ILLEGAL_FILENAME;
824 $this->mTitle = null;
825
826 return $this->mTitle;
827 }
828 $this->mFilteredName = $nt->getDBkey();
829
830 /**
831 * We'll want to blacklist against *any* 'extension', and use
832 * only the final one for the whitelist.
833 */
834 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName );
835
836 if ( count( $ext ) ) {
837 $this->mFinalExtension = trim( $ext[count( $ext ) - 1] );
838 } else {
839 $this->mFinalExtension = '';
840
841 # No extension, try guessing one
842 $magic = MimeMagic::singleton();
843 $mime = $magic->guessMimeType( $this->mTempPath );
844 if ( $mime !== 'unknown/unknown' ) {
845 # Get a space separated list of extensions
846 $extList = $magic->getExtensionsForType( $mime );
847 if ( $extList ) {
848 # Set the extension to the canonical extension
849 $this->mFinalExtension = strtok( $extList, ' ' );
850
851 # Fix up the other variables
852 $this->mFilteredName .= ".{$this->mFinalExtension}";
853 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
854 $ext = array( $this->mFinalExtension );
855 }
856 }
857 }
858
859 /* Don't allow users to override the blacklist (check file extension) */
860 global $wgCheckFileExtensions, $wgStrictFileExtensions;
861 global $wgFileExtensions, $wgFileBlacklist;
862
863 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
864
865 if ( $this->mFinalExtension == '' ) {
866 $this->mTitleError = self::FILETYPE_MISSING;
867 $this->mTitle = null;
868
869 return $this->mTitle;
870 } elseif ( $blackListedExtensions ||
871 ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
872 !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) )
873 ) {
874 $this->mBlackListedExtensions = $blackListedExtensions;
875 $this->mTitleError = self::FILETYPE_BADTYPE;
876 $this->mTitle = null;
877
878 return $this->mTitle;
879 }
880
881 // Windows may be broken with special characters, see bug 1780
882 if ( !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() )
883 && !RepoGroup::singleton()->getLocalRepo()->backendSupportsUnicodePaths()
884 ) {
885 $this->mTitleError = self::WINDOWS_NONASCII_FILENAME;
886 $this->mTitle = null;
887
888 return $this->mTitle;
889 }
890
891 # If there was more than one "extension", reassemble the base
892 # filename to prevent bogus complaints about length
893 if ( count( $ext ) > 1 ) {
894 $iterations = count( $ext ) - 1;
895 for ( $i = 0; $i < $iterations; $i++ ) {
896 $partname .= '.' . $ext[$i];
897 }
898 }
899
900 if ( strlen( $partname ) < 1 ) {
901 $this->mTitleError = self::MIN_LENGTH_PARTNAME;
902 $this->mTitle = null;
903
904 return $this->mTitle;
905 }
906
907 $this->mTitle = $nt;
908
909 return $this->mTitle;
910 }
911
912 /**
913 * Return the local file and initializes if necessary.
914 *
915 * @return LocalFile|UploadStashFile|null
916 */
917 public function getLocalFile() {
918 if ( is_null( $this->mLocalFile ) ) {
919 $nt = $this->getTitle();
920 $this->mLocalFile = is_null( $nt ) ? null : wfLocalFile( $nt );
921 }
922
923 return $this->mLocalFile;
924 }
925
926 /**
927 * If the user does not supply all necessary information in the first upload
928 * form submission (either by accident or by design) then we may want to
929 * stash the file temporarily, get more information, and publish the file
930 * later.
931 *
932 * This method will stash a file in a temporary directory for later
933 * processing, and save the necessary descriptive info into the database.
934 * This method returns the file object, which also has a 'fileKey' property
935 * which can be passed through a form or API request to find this stashed
936 * file again.
937 *
938 * @param User $user
939 * @return UploadStashFile Stashed file
940 */
941 public function stashFile( User $user = null ) {
942 // was stashSessionFile
943 wfProfileIn( __METHOD__ );
944
945 $stash = RepoGroup::singleton()->getLocalRepo()->getUploadStash( $user );
946 $file = $stash->stashFile( $this->mTempPath, $this->getSourceType() );
947 $this->mLocalFile = $file;
948
949 wfProfileOut( __METHOD__ );
950
951 return $file;
952 }
953
954 /**
955 * Stash a file in a temporary directory, returning a key which can be used
956 * to find the file again. See stashFile().
957 *
958 * @return string File key
959 */
960 public function stashFileGetKey() {
961 return $this->stashFile()->getFileKey();
962 }
963
964 /**
965 * alias for stashFileGetKey, for backwards compatibility
966 *
967 * @return string File key
968 */
969 public function stashSession() {
970 return $this->stashFileGetKey();
971 }
972
973 /**
974 * If we've modified the upload file we need to manually remove it
975 * on exit to clean up.
976 */
977 public function cleanupTempFile() {
978 if ( $this->mRemoveTempFile && $this->mTempPath && file_exists( $this->mTempPath ) ) {
979 wfDebug( __METHOD__ . ": Removing temporary file {$this->mTempPath}\n" );
980 unlink( $this->mTempPath );
981 }
982 }
983
984 public function getTempPath() {
985 return $this->mTempPath;
986 }
987
988 /**
989 * Split a file into a base name and all dot-delimited 'extensions'
990 * on the end. Some web server configurations will fall back to
991 * earlier pseudo-'extensions' to determine type and execute
992 * scripts, so the blacklist needs to check them all.
993 *
994 * @param string $filename
995 * @return array
996 */
997 public static function splitExtensions( $filename ) {
998 $bits = explode( '.', $filename );
999 $basename = array_shift( $bits );
1000
1001 return array( $basename, $bits );
1002 }
1003
1004 /**
1005 * Perform case-insensitive match against a list of file extensions.
1006 * Returns true if the extension is in the list.
1007 *
1008 * @param string $ext
1009 * @param array $list
1010 * @return bool
1011 */
1012 public static function checkFileExtension( $ext, $list ) {
1013 return in_array( strtolower( $ext ), $list );
1014 }
1015
1016 /**
1017 * Perform case-insensitive match against a list of file extensions.
1018 * Returns an array of matching extensions.
1019 *
1020 * @param array $ext
1021 * @param array $list
1022 * @return bool
1023 */
1024 public static function checkFileExtensionList( $ext, $list ) {
1025 return array_intersect( array_map( 'strtolower', $ext ), $list );
1026 }
1027
1028 /**
1029 * Checks if the MIME type of the uploaded file matches the file extension.
1030 *
1031 * @param string $mime The MIME type of the uploaded file
1032 * @param string $extension The filename extension that the file is to be served with
1033 * @return bool
1034 */
1035 public static function verifyExtension( $mime, $extension ) {
1036 $magic = MimeMagic::singleton();
1037
1038 if ( !$mime || $mime == 'unknown' || $mime == 'unknown/unknown' ) {
1039 if ( !$magic->isRecognizableExtension( $extension ) ) {
1040 wfDebug( __METHOD__ . ": passing file with unknown detected mime type; " .
1041 "unrecognized extension '$extension', can't verify\n" );
1042
1043 return true;
1044 } else {
1045 wfDebug( __METHOD__ . ": rejecting file with unknown detected mime type; " .
1046 "recognized extension '$extension', so probably invalid file\n" );
1047
1048 return false;
1049 }
1050 }
1051
1052 $match = $magic->isMatchingExtension( $extension, $mime );
1053
1054 if ( $match === null ) {
1055 if ( $magic->getTypesForExtension( $extension ) !== null ) {
1056 wfDebug( __METHOD__ . ": No extension known for $mime, but we know a mime for $extension\n" );
1057
1058 return false;
1059 } else {
1060 wfDebug( __METHOD__ . ": no file extension known for mime type $mime, passing file\n" );
1061
1062 return true;
1063 }
1064 } elseif ( $match === true ) {
1065 wfDebug( __METHOD__ . ": mime type $mime matches extension $extension, passing file\n" );
1066
1067 /** @todo If it's a bitmap, make sure PHP or ImageMagick resp. can handle it! */
1068 return true;
1069 } else {
1070 wfDebug( __METHOD__
1071 . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
1072
1073 return false;
1074 }
1075 }
1076
1077 /**
1078 * Heuristic for detecting files that *could* contain JavaScript instructions or
1079 * things that may look like HTML to a browser and are thus
1080 * potentially harmful. The present implementation will produce false
1081 * positives in some situations.
1082 *
1083 * @param string $file Pathname to the temporary upload file
1084 * @param string $mime The MIME type of the file
1085 * @param string $extension The extension of the file
1086 * @return bool True if the file contains something looking like embedded scripts
1087 */
1088 public static function detectScript( $file, $mime, $extension ) {
1089 global $wgAllowTitlesInSVG;
1090 wfProfileIn( __METHOD__ );
1091
1092 # ugly hack: for text files, always look at the entire file.
1093 # For binary field, just check the first K.
1094
1095 if ( strpos( $mime, 'text/' ) === 0 ) {
1096 $chunk = file_get_contents( $file );
1097 } else {
1098 $fp = fopen( $file, 'rb' );
1099 $chunk = fread( $fp, 1024 );
1100 fclose( $fp );
1101 }
1102
1103 $chunk = strtolower( $chunk );
1104
1105 if ( !$chunk ) {
1106 wfProfileOut( __METHOD__ );
1107
1108 return false;
1109 }
1110
1111 # decode from UTF-16 if needed (could be used for obfuscation).
1112 if ( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
1113 $enc = 'UTF-16BE';
1114 } elseif ( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
1115 $enc = 'UTF-16LE';
1116 } else {
1117 $enc = null;
1118 }
1119
1120 if ( $enc ) {
1121 $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
1122 }
1123
1124 $chunk = trim( $chunk );
1125
1126 /** @todo FIXME: Convert from UTF-16 if necessary! */
1127 wfDebug( __METHOD__ . ": checking for embedded scripts and HTML stuff\n" );
1128
1129 # check for HTML doctype
1130 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
1131 wfProfileOut( __METHOD__ );
1132
1133 return true;
1134 }
1135
1136 // Some browsers will interpret obscure xml encodings as UTF-8, while
1137 // PHP/expat will interpret the given encoding in the xml declaration (bug 47304)
1138 if ( $extension == 'svg' || strpos( $mime, 'image/svg' ) === 0 ) {
1139 if ( self::checkXMLEncodingMissmatch( $file ) ) {
1140 wfProfileOut( __METHOD__ );
1141
1142 return true;
1143 }
1144 }
1145
1146 /**
1147 * Internet Explorer for Windows performs some really stupid file type
1148 * autodetection which can cause it to interpret valid image files as HTML
1149 * and potentially execute JavaScript, creating a cross-site scripting
1150 * attack vectors.
1151 *
1152 * Apple's Safari browser also performs some unsafe file type autodetection
1153 * which can cause legitimate files to be interpreted as HTML if the
1154 * web server is not correctly configured to send the right content-type
1155 * (or if you're really uploading plain text and octet streams!)
1156 *
1157 * Returns true if IE is likely to mistake the given file for HTML.
1158 * Also returns true if Safari would mistake the given file for HTML
1159 * when served with a generic content-type.
1160 */
1161 $tags = array(
1162 '<a href',
1163 '<body',
1164 '<head',
1165 '<html', #also in safari
1166 '<img',
1167 '<pre',
1168 '<script', #also in safari
1169 '<table'
1170 );
1171
1172 if ( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
1173 $tags[] = '<title';
1174 }
1175
1176 foreach ( $tags as $tag ) {
1177 if ( false !== strpos( $chunk, $tag ) ) {
1178 wfDebug( __METHOD__ . ": found something that may make it be mistaken for html: $tag\n" );
1179 wfProfileOut( __METHOD__ );
1180
1181 return true;
1182 }
1183 }
1184
1185 /*
1186 * look for JavaScript
1187 */
1188
1189 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
1190 $chunk = Sanitizer::decodeCharReferences( $chunk );
1191
1192 # look for script-types
1193 if ( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
1194 wfDebug( __METHOD__ . ": found script types\n" );
1195 wfProfileOut( __METHOD__ );
1196
1197 return true;
1198 }
1199
1200 # look for html-style script-urls
1201 if ( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1202 wfDebug( __METHOD__ . ": found html-style script urls\n" );
1203 wfProfileOut( __METHOD__ );
1204
1205 return true;
1206 }
1207
1208 # look for css-style script-urls
1209 if ( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1210 wfDebug( __METHOD__ . ": found css-style script urls\n" );
1211 wfProfileOut( __METHOD__ );
1212
1213 return true;
1214 }
1215
1216 wfDebug( __METHOD__ . ": no scripts found\n" );
1217 wfProfileOut( __METHOD__ );
1218
1219 return false;
1220 }
1221
1222 /**
1223 * Check a whitelist of xml encodings that are known not to be interpreted differently
1224 * by the server's xml parser (expat) and some common browsers.
1225 *
1226 * @param string $file Pathname to the temporary upload file
1227 * @return bool True if the file contains an encoding that could be misinterpreted
1228 */
1229 public static function checkXMLEncodingMissmatch( $file ) {
1230 global $wgSVGMetadataCutoff;
1231 $contents = file_get_contents( $file, false, null, -1, $wgSVGMetadataCutoff );
1232 $encodingRegex = '!encoding[ \t\n\r]*=[ \t\n\r]*[\'"](.*?)[\'"]!si';
1233
1234 if ( preg_match( "!<\?xml\b(.*?)\?>!si", $contents, $matches ) ) {
1235 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1236 && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1237 ) {
1238 wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1239
1240 return true;
1241 }
1242 } elseif ( preg_match( "!<\?xml\b!si", $contents ) ) {
1243 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1244 // bytes. There shouldn't be a legitimate reason for this to happen.
1245 wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1246
1247 return true;
1248 } elseif ( substr( $contents, 0, 4 ) == "\x4C\x6F\xA7\x94" ) {
1249 // EBCDIC encoded XML
1250 wfDebug( __METHOD__ . ": EBCDIC Encoded XML\n" );
1251
1252 return true;
1253 }
1254
1255 // It's possible the file is encoded with multi-byte encoding, so re-encode attempt to
1256 // detect the encoding in case is specifies an encoding not whitelisted in self::$safeXmlEncodings
1257 $attemptEncodings = array( 'UTF-16', 'UTF-16BE', 'UTF-32', 'UTF-32BE' );
1258 foreach ( $attemptEncodings as $encoding ) {
1259 wfSuppressWarnings();
1260 $str = iconv( $encoding, 'UTF-8', $contents );
1261 wfRestoreWarnings();
1262 if ( $str != '' && preg_match( "!<\?xml\b(.*?)\?>!si", $str, $matches ) ) {
1263 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1264 && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1265 ) {
1266 wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1267
1268 return true;
1269 }
1270 } elseif ( $str != '' && preg_match( "!<\?xml\b!si", $str ) ) {
1271 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1272 // bytes. There shouldn't be a legitimate reason for this to happen.
1273 wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1274
1275 return true;
1276 }
1277 }
1278
1279 return false;
1280 }
1281
1282 /**
1283 * @param string $filename
1284 * @return mixed False of the file is verified (does not contain scripts), array otherwise.
1285 */
1286 protected function detectScriptInSvg( $filename ) {
1287 $this->mSVGNSError = false;
1288 $check = new XmlTypeCheck(
1289 $filename,
1290 array( $this, 'checkSvgScriptCallback' ),
1291 true,
1292 array( 'processing_instruction_handler' => 'UploadBase::checkSvgPICallback' )
1293 );
1294 if ( $check->wellFormed !== true ) {
1295 // Invalid xml (bug 58553)
1296 return array( 'uploadinvalidxml' );
1297 } elseif ( $check->filterMatch ) {
1298 if ( $this->mSVGNSError ) {
1299 return array( 'uploadscriptednamespace', $this->mSVGNSError );
1300 }
1301
1302 return array( 'uploadscripted' );
1303 }
1304
1305 return false;
1306 }
1307
1308 /**
1309 * Callback to filter SVG Processing Instructions.
1310 * @param string $target Processing instruction name
1311 * @param string $data Processing instruction attribute and value
1312 * @return bool (true if the filter identified something bad)
1313 */
1314 public static function checkSvgPICallback( $target, $data ) {
1315 // Don't allow external stylesheets (bug 57550)
1316 if ( preg_match( '/xml-stylesheet/i', $target ) ) {
1317 return true;
1318 }
1319
1320 return false;
1321 }
1322
1323 /**
1324 * @todo Replace this with a whitelist filter!
1325 * @param string $element
1326 * @param array $attribs
1327 * @return bool
1328 */
1329 public function checkSvgScriptCallback( $element, $attribs, $data = null ) {
1330
1331 list( $namespace, $strippedElement ) = $this->splitXmlNamespace( $element );
1332
1333 // We specifically don't include:
1334 // http://www.w3.org/1999/xhtml (bug 60771)
1335 static $validNamespaces = array(
1336 '',
1337 'adobe:ns:meta/',
1338 'http://creativecommons.org/ns#',
1339 'http://inkscape.sourceforge.net/dtd/sodipodi-0.dtd',
1340 'http://ns.adobe.com/adobeillustrator/10.0/',
1341 'http://ns.adobe.com/adobesvgviewerextensions/3.0/',
1342 'http://ns.adobe.com/extensibility/1.0/',
1343 'http://ns.adobe.com/flows/1.0/',
1344 'http://ns.adobe.com/illustrator/1.0/',
1345 'http://ns.adobe.com/imagereplacement/1.0/',
1346 'http://ns.adobe.com/pdf/1.3/',
1347 'http://ns.adobe.com/photoshop/1.0/',
1348 'http://ns.adobe.com/saveforweb/1.0/',
1349 'http://ns.adobe.com/variables/1.0/',
1350 'http://ns.adobe.com/xap/1.0/',
1351 'http://ns.adobe.com/xap/1.0/g/',
1352 'http://ns.adobe.com/xap/1.0/g/img/',
1353 'http://ns.adobe.com/xap/1.0/mm/',
1354 'http://ns.adobe.com/xap/1.0/rights/',
1355 'http://ns.adobe.com/xap/1.0/stype/dimensions#',
1356 'http://ns.adobe.com/xap/1.0/stype/font#',
1357 'http://ns.adobe.com/xap/1.0/stype/manifestitem#',
1358 'http://ns.adobe.com/xap/1.0/stype/resourceevent#',
1359 'http://ns.adobe.com/xap/1.0/stype/resourceref#',
1360 'http://ns.adobe.com/xap/1.0/t/pg/',
1361 'http://purl.org/dc/elements/1.1/',
1362 'http://purl.org/dc/elements/1.1',
1363 'http://schemas.microsoft.com/visio/2003/svgextensions/',
1364 'http://sodipodi.sourceforge.net/dtd/sodipodi-0.dtd',
1365 'http://taptrix.com/inkpad/svg_extensions',
1366 'http://web.resource.org/cc/',
1367 'http://www.freesoftware.fsf.org/bkchem/cdml',
1368 'http://www.inkscape.org/namespaces/inkscape',
1369 'http://www.opengis.net/gml',
1370 'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
1371 'http://www.w3.org/2000/svg',
1372 'http://www.w3.org/tr/rec-rdf-syntax/',
1373 );
1374
1375 if ( !in_array( $namespace, $validNamespaces ) ) {
1376 wfDebug( __METHOD__ . ": Non-svg namespace '$namespace' in uploaded file.\n" );
1377 /** @todo Return a status object to a closure in XmlTypeCheck, for MW1.21+ */
1378 $this->mSVGNSError = $namespace;
1379
1380 return true;
1381 }
1382
1383 /*
1384 * check for elements that can contain javascript
1385 */
1386 if ( $strippedElement == 'script' ) {
1387 wfDebug( __METHOD__ . ": Found script element '$element' in uploaded file.\n" );
1388
1389 return true;
1390 }
1391
1392 # e.g., <svg xmlns="http://www.w3.org/2000/svg">
1393 # <handler xmlns:ev="http://www.w3.org/2001/xml-events" ev:event="load">alert(1)</handler> </svg>
1394 if ( $strippedElement == 'handler' ) {
1395 wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1396
1397 return true;
1398 }
1399
1400 # SVG reported in Feb '12 that used xml:stylesheet to generate javascript block
1401 if ( $strippedElement == 'stylesheet' ) {
1402 wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1403
1404 return true;
1405 }
1406
1407 # Block iframes, in case they pass the namespace check
1408 if ( $strippedElement == 'iframe' ) {
1409 wfDebug( __METHOD__ . ": iframe in uploaded file.\n" );
1410
1411 return true;
1412 }
1413
1414 # Check <style> css
1415 if ( $strippedElement == 'style'
1416 && self::checkCssFragment( Sanitizer::normalizeCss( $data ) )
1417 ) {
1418 wfDebug( __METHOD__ . ": hostile css in style element.\n" );
1419 return true;
1420 }
1421
1422 foreach ( $attribs as $attrib => $value ) {
1423 $stripped = $this->stripXmlNamespace( $attrib );
1424 $value = strtolower( $value );
1425
1426 if ( substr( $stripped, 0, 2 ) == 'on' ) {
1427 wfDebug( __METHOD__
1428 . ": Found event-handler attribute '$attrib'='$value' in uploaded file.\n" );
1429
1430 return true;
1431 }
1432
1433 # href with non-local target (don't allow http://, javascript:, etc)
1434 if ( $stripped == 'href'
1435 && strpos( $value, 'data:' ) !== 0
1436 && strpos( $value, '#' ) !== 0
1437 ) {
1438 if ( !( $strippedElement === 'a'
1439 && preg_match( '!^https?://!im', $value ) )
1440 ) {
1441 wfDebug( __METHOD__ . ": Found href attribute <$strippedElement "
1442 . "'$attrib'='$value' in uploaded file.\n" );
1443
1444 return true;
1445 }
1446 }
1447
1448 # href with embedded svg as target
1449 if ( $stripped == 'href' && preg_match( '!data:[^,]*image/svg[^,]*,!sim', $value ) ) {
1450 wfDebug( __METHOD__ . ": Found href to embedded svg "
1451 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1452
1453 return true;
1454 }
1455
1456 # href with embedded (text/xml) svg as target
1457 if ( $stripped == 'href' && preg_match( '!data:[^,]*text/xml[^,]*,!sim', $value ) ) {
1458 wfDebug( __METHOD__ . ": Found href to embedded svg "
1459 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1460
1461 return true;
1462 }
1463
1464 # Change href with animate from (http://html5sec.org/#137). This doesn't seem
1465 # possible without embedding the svg, but filter here in case.
1466 if ( $stripped == 'from'
1467 && $strippedElement === 'animate'
1468 && !preg_match( '!^https?://!im', $value )
1469 ) {
1470 wfDebug( __METHOD__ . ": Found animate that might be changing href using from "
1471 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1472
1473 return true;
1474 }
1475
1476 # use set/animate to add event-handler attribute to parent
1477 if ( ( $strippedElement == 'set' || $strippedElement == 'animate' )
1478 && $stripped == 'attributename'
1479 && substr( $value, 0, 2 ) == 'on'
1480 ) {
1481 wfDebug( __METHOD__ . ": Found svg setting event-handler attribute with "
1482 . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1483
1484 return true;
1485 }
1486
1487 # use set to add href attribute to parent element
1488 if ( $strippedElement == 'set'
1489 && $stripped == 'attributename'
1490 && strpos( $value, 'href' ) !== false
1491 ) {
1492 wfDebug( __METHOD__ . ": Found svg setting href attribute '$value' in uploaded file.\n" );
1493
1494 return true;
1495 }
1496
1497 # use set to add a remote / data / script target to an element
1498 if ( $strippedElement == 'set'
1499 && $stripped == 'to'
1500 && preg_match( '!(http|https|data|script):!sim', $value )
1501 ) {
1502 wfDebug( __METHOD__ . ": Found svg setting attribute to '$value' in uploaded file.\n" );
1503
1504 return true;
1505 }
1506
1507 # use handler attribute with remote / data / script
1508 if ( $stripped == 'handler' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1509 wfDebug( __METHOD__ . ": Found svg setting handler with remote/data/script "
1510 . "'$attrib'='$value' in uploaded file.\n" );
1511
1512 return true;
1513 }
1514
1515 # use CSS styles to bring in remote code
1516 if ( $stripped == 'style'
1517 && self::checkCssFragment( Sanitizer::normalizeCss( $value ) )
1518 ) {
1519 wfDebug( __METHOD__ . ": Found svg setting a style with "
1520 . "remote url '$attrib'='$value' in uploaded file.\n" );
1521 return true;
1522 }
1523
1524 # Several attributes can include css, css character escaping isn't allowed
1525 $cssAttrs = array( 'font', 'clip-path', 'fill', 'filter', 'marker',
1526 'marker-end', 'marker-mid', 'marker-start', 'mask', 'stroke' );
1527 if ( in_array( $stripped, $cssAttrs )
1528 && self::checkCssFragment( $value )
1529 ) {
1530 wfDebug( __METHOD__ . ": Found svg setting a style with "
1531 . "remote url '$attrib'='$value' in uploaded file.\n" );
1532 return true;
1533 }
1534
1535 # image filters can pull in url, which could be svg that executes scripts
1536 if ( $strippedElement == 'image'
1537 && $stripped == 'filter'
1538 && preg_match( '!url\s*\(!sim', $value )
1539 ) {
1540 wfDebug( __METHOD__ . ": Found image filter with url: "
1541 . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1542
1543 return true;
1544 }
1545 }
1546
1547 return false; //No scripts detected
1548 }
1549
1550 /**
1551 * Check a block of CSS or CSS fragment for anything that looks like
1552 * it is bringing in remote code.
1553 * @param string $value a string of CSS
1554 * @param bool $propOnly only check css properties (start regex with :)
1555 * @return bool true if the CSS contains an illegal string, false if otherwise
1556 */
1557 private static function checkCssFragment( $value ) {
1558
1559 # Forbid external stylesheets, for both reliability and to protect viewer's privacy
1560 if ( strpos( $value, '@import' ) !== false ) {
1561 return true;
1562 }
1563
1564 # We allow @font-face to embed fonts with data: urls, so we snip the string
1565 # 'url' out so this case won't match when we check for urls below
1566 $pattern = '!(@font-face\s*{[^}]*src:)url(\("data:;base64,)!im';
1567 $value = preg_replace( $pattern, '$1$2', $value );
1568
1569 # Check for remote and executable CSS. Unlike in Sanitizer::checkCss, the CSS
1570 # properties filter and accelerator don't seem to be useful for xss in SVG files.
1571 # Expression and -o-link don't seem to work either, but filtering them here in case.
1572 # Additionally, we catch remote urls like url("http:..., url('http:..., url(http:...,
1573 # but not local ones such as url("#..., url('#..., url(#....
1574 if ( preg_match( '!expression
1575 | -o-link\s*:
1576 | -o-link-source\s*:
1577 | -o-replace\s*:!imx', $value ) ) {
1578 return true;
1579 }
1580
1581 if ( preg_match_all(
1582 "!(\s*(url|image|image-set)\s*\(\s*[\"']?\s*[^#]+.*?\))!sim",
1583 $value,
1584 $matches
1585 ) !== 0
1586 ) {
1587 # TODO: redo this in one regex. Until then, url("#whatever") matches the first
1588 foreach ( $matches[1] as $match ) {
1589 if ( !preg_match( "!\s*(url|image|image-set)\s*\(\s*(#|'#|\"#)!im", $match ) ) {
1590 return true;
1591 }
1592 }
1593 }
1594
1595 if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ) {
1596 return true;
1597 }
1598
1599 return false;
1600 }
1601
1602 /**
1603 * Divide the element name passed by the xml parser to the callback into URI and prifix.
1604 * @param string $element
1605 * @return array Containing the namespace URI and prefix
1606 */
1607 private static function splitXmlNamespace( $element ) {
1608 // 'http://www.w3.org/2000/svg:script' -> array( 'http://www.w3.org/2000/svg', 'script' )
1609 $parts = explode( ':', strtolower( $element ) );
1610 $name = array_pop( $parts );
1611 $ns = implode( ':', $parts );
1612
1613 return array( $ns, $name );
1614 }
1615
1616 /**
1617 * @param string $name
1618 * @return string
1619 */
1620 private function stripXmlNamespace( $name ) {
1621 // 'http://www.w3.org/2000/svg:script' -> 'script'
1622 $parts = explode( ':', strtolower( $name ) );
1623
1624 return array_pop( $parts );
1625 }
1626
1627 /**
1628 * Generic wrapper function for a virus scanner program.
1629 * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
1630 * $wgAntivirusRequired may be used to deny upload if the scan fails.
1631 *
1632 * @param string $file Pathname to the temporary upload file
1633 * @return mixed False if not virus is found, null if the scan fails or is disabled,
1634 * or a string containing feedback from the virus scanner if a virus was found.
1635 * If textual feedback is missing but a virus was found, this function returns true.
1636 */
1637 public static function detectVirus( $file ) {
1638 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
1639 wfProfileIn( __METHOD__ );
1640
1641 if ( !$wgAntivirus ) {
1642 wfDebug( __METHOD__ . ": virus scanner disabled\n" );
1643 wfProfileOut( __METHOD__ );
1644
1645 return null;
1646 }
1647
1648 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
1649 wfDebug( __METHOD__ . ": unknown virus scanner: $wgAntivirus\n" );
1650 $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
1651 array( 'virus-badscanner', $wgAntivirus ) );
1652 wfProfileOut( __METHOD__ );
1653
1654 return wfMessage( 'virus-unknownscanner' )->text() . " $wgAntivirus";
1655 }
1656
1657 # look up scanner configuration
1658 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
1659 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
1660 $msgPattern = isset( $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ) ?
1661 $wgAntivirusSetup[$wgAntivirus]['messagepattern'] : null;
1662
1663 if ( strpos( $command, "%f" ) === false ) {
1664 # simple pattern: append file to scan
1665 $command .= " " . wfEscapeShellArg( $file );
1666 } else {
1667 # complex pattern: replace "%f" with file to scan
1668 $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
1669 }
1670
1671 wfDebug( __METHOD__ . ": running virus scan: $command \n" );
1672
1673 # execute virus scanner
1674 $exitCode = false;
1675
1676 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1677 # that does not seem to be worth the pain.
1678 # Ask me (Duesentrieb) about it if it's ever needed.
1679 $output = wfShellExecWithStderr( $command, $exitCode );
1680
1681 # map exit code to AV_xxx constants.
1682 $mappedCode = $exitCode;
1683 if ( $exitCodeMap ) {
1684 if ( isset( $exitCodeMap[$exitCode] ) ) {
1685 $mappedCode = $exitCodeMap[$exitCode];
1686 } elseif ( isset( $exitCodeMap["*"] ) ) {
1687 $mappedCode = $exitCodeMap["*"];
1688 }
1689 }
1690
1691 /* NB: AV_NO_VIRUS is 0 but AV_SCAN_FAILED is false,
1692 * so we need the strict equalities === and thus can't use a switch here
1693 */
1694 if ( $mappedCode === AV_SCAN_FAILED ) {
1695 # scan failed (code was mapped to false by $exitCodeMap)
1696 wfDebug( __METHOD__ . ": failed to scan $file (code $exitCode).\n" );
1697
1698 $output = $wgAntivirusRequired
1699 ? wfMessage( 'virus-scanfailed', array( $exitCode ) )->text()
1700 : null;
1701 } elseif ( $mappedCode === AV_SCAN_ABORTED ) {
1702 # scan failed because filetype is unknown (probably imune)
1703 wfDebug( __METHOD__ . ": unsupported file type $file (code $exitCode).\n" );
1704 $output = null;
1705 } elseif ( $mappedCode === AV_NO_VIRUS ) {
1706 # no virus found
1707 wfDebug( __METHOD__ . ": file passed virus scan.\n" );
1708 $output = false;
1709 } else {
1710 $output = trim( $output );
1711
1712 if ( !$output ) {
1713 $output = true; #if there's no output, return true
1714 } elseif ( $msgPattern ) {
1715 $groups = array();
1716 if ( preg_match( $msgPattern, $output, $groups ) ) {
1717 if ( $groups[1] ) {
1718 $output = $groups[1];
1719 }
1720 }
1721 }
1722
1723 wfDebug( __METHOD__ . ": FOUND VIRUS! scanner feedback: $output \n" );
1724 }
1725
1726 wfProfileOut( __METHOD__ );
1727
1728 return $output;
1729 }
1730
1731 /**
1732 * Check if there's an overwrite conflict and, if so, if restrictions
1733 * forbid this user from performing the upload.
1734 *
1735 * @param User $user
1736 *
1737 * @return mixed True on success, array on failure
1738 */
1739 private function checkOverwrite( $user ) {
1740 // First check whether the local file can be overwritten
1741 $file = $this->getLocalFile();
1742 if ( $file->exists() ) {
1743 if ( !self::userCanReUpload( $user, $file ) ) {
1744 return array( 'fileexists-forbidden', $file->getName() );
1745 } else {
1746 return true;
1747 }
1748 }
1749
1750 /* Check shared conflicts: if the local file does not exist, but
1751 * wfFindFile finds a file, it exists in a shared repository.
1752 */
1753 $file = wfFindFile( $this->getTitle() );
1754 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1755 return array( 'fileexists-shared-forbidden', $file->getName() );
1756 }
1757
1758 return true;
1759 }
1760
1761 /**
1762 * Check if a user is the last uploader
1763 *
1764 * @param User $user
1765 * @param string $img Image name
1766 * @return bool
1767 */
1768 public static function userCanReUpload( User $user, $img ) {
1769 if ( $user->isAllowed( 'reupload' ) ) {
1770 return true; // non-conditional
1771 }
1772 if ( !$user->isAllowed( 'reupload-own' ) ) {
1773 return false;
1774 }
1775 if ( is_string( $img ) ) {
1776 $img = wfLocalFile( $img );
1777 }
1778 if ( !( $img instanceof LocalFile ) ) {
1779 return false;
1780 }
1781
1782 return $user->getId() == $img->getUser( 'id' );
1783 }
1784
1785 /**
1786 * Helper function that does various existence checks for a file.
1787 * The following checks are performed:
1788 * - The file exists
1789 * - Article with the same name as the file exists
1790 * - File exists with normalized extension
1791 * - The file looks like a thumbnail and the original exists
1792 *
1793 * @param File $file The File object to check
1794 * @return mixed False if the file does not exists, else an array
1795 */
1796 public static function getExistsWarning( $file ) {
1797 if ( $file->exists() ) {
1798 return array( 'warning' => 'exists', 'file' => $file );
1799 }
1800
1801 if ( $file->getTitle()->getArticleID() ) {
1802 return array( 'warning' => 'page-exists', 'file' => $file );
1803 }
1804
1805 if ( $file->wasDeleted() && !$file->exists() ) {
1806 return array( 'warning' => 'was-deleted', 'file' => $file );
1807 }
1808
1809 if ( strpos( $file->getName(), '.' ) == false ) {
1810 $partname = $file->getName();
1811 $extension = '';
1812 } else {
1813 $n = strrpos( $file->getName(), '.' );
1814 $extension = substr( $file->getName(), $n + 1 );
1815 $partname = substr( $file->getName(), 0, $n );
1816 }
1817 $normalizedExtension = File::normalizeExtension( $extension );
1818
1819 if ( $normalizedExtension != $extension ) {
1820 // We're not using the normalized form of the extension.
1821 // Normal form is lowercase, using most common of alternate
1822 // extensions (eg 'jpg' rather than 'JPEG').
1823 //
1824 // Check for another file using the normalized form...
1825 $nt_lc = Title::makeTitle( NS_FILE, "{$partname}.{$normalizedExtension}" );
1826 $file_lc = wfLocalFile( $nt_lc );
1827
1828 if ( $file_lc->exists() ) {
1829 return array(
1830 'warning' => 'exists-normalized',
1831 'file' => $file,
1832 'normalizedFile' => $file_lc
1833 );
1834 }
1835 }
1836
1837 // Check for files with the same name but a different extension
1838 $similarFiles = RepoGroup::singleton()->getLocalRepo()->findFilesByPrefix(
1839 "{$partname}.", 1 );
1840 if ( count( $similarFiles ) ) {
1841 return array(
1842 'warning' => 'exists-normalized',
1843 'file' => $file,
1844 'normalizedFile' => $similarFiles[0],
1845 );
1846 }
1847
1848 if ( self::isThumbName( $file->getName() ) ) {
1849 # Check for filenames like 50px- or 180px-, these are mostly thumbnails
1850 $nt_thb = Title::newFromText(
1851 substr( $partname, strpos( $partname, '-' ) + 1 ) . '.' . $extension,
1852 NS_FILE
1853 );
1854 $file_thb = wfLocalFile( $nt_thb );
1855 if ( $file_thb->exists() ) {
1856 return array(
1857 'warning' => 'thumb',
1858 'file' => $file,
1859 'thumbFile' => $file_thb
1860 );
1861 } else {
1862 // File does not exist, but we just don't like the name
1863 return array(
1864 'warning' => 'thumb-name',
1865 'file' => $file,
1866 'thumbFile' => $file_thb
1867 );
1868 }
1869 }
1870
1871 foreach ( self::getFilenamePrefixBlacklist() as $prefix ) {
1872 if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
1873 return array(
1874 'warning' => 'bad-prefix',
1875 'file' => $file,
1876 'prefix' => $prefix
1877 );
1878 }
1879 }
1880
1881 return false;
1882 }
1883
1884 /**
1885 * Helper function that checks whether the filename looks like a thumbnail
1886 * @param string $filename
1887 * @return bool
1888 */
1889 public static function isThumbName( $filename ) {
1890 $n = strrpos( $filename, '.' );
1891 $partname = $n ? substr( $filename, 0, $n ) : $filename;
1892
1893 return (
1894 substr( $partname, 3, 3 ) == 'px-' ||
1895 substr( $partname, 2, 3 ) == 'px-'
1896 ) &&
1897 preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) );
1898 }
1899
1900 /**
1901 * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
1902 *
1903 * @return array List of prefixes
1904 */
1905 public static function getFilenamePrefixBlacklist() {
1906 $blacklist = array();
1907 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
1908 if ( !$message->isDisabled() ) {
1909 $lines = explode( "\n", $message->plain() );
1910 foreach ( $lines as $line ) {
1911 // Remove comment lines
1912 $comment = substr( trim( $line ), 0, 1 );
1913 if ( $comment == '#' || $comment == '' ) {
1914 continue;
1915 }
1916 // Remove additional comments after a prefix
1917 $comment = strpos( $line, '#' );
1918 if ( $comment > 0 ) {
1919 $line = substr( $line, 0, $comment - 1 );
1920 }
1921 $blacklist[] = trim( $line );
1922 }
1923 }
1924
1925 return $blacklist;
1926 }
1927
1928 /**
1929 * Gets image info about the file just uploaded.
1930 *
1931 * Also has the effect of setting metadata to be an 'indexed tag name' in
1932 * returned API result if 'metadata' was requested. Oddly, we have to pass
1933 * the "result" object down just so it can do that with the appropriate
1934 * format, presumably.
1935 *
1936 * @param ApiResult $result
1937 * @return array Image info
1938 */
1939 public function getImageInfo( $result ) {
1940 $file = $this->getLocalFile();
1941 /** @todo This cries out for refactoring.
1942 * We really want to say $file->getAllInfo(); here.
1943 * Perhaps "info" methods should be moved into files, and the API should
1944 * just wrap them in queries.
1945 */
1946 if ( $file instanceof UploadStashFile ) {
1947 $imParam = ApiQueryStashImageInfo::getPropertyNames();
1948 $info = ApiQueryStashImageInfo::getInfo( $file, array_flip( $imParam ), $result );
1949 } else {
1950 $imParam = ApiQueryImageInfo::getPropertyNames();
1951 $info = ApiQueryImageInfo::getInfo( $file, array_flip( $imParam ), $result );
1952 }
1953
1954 return $info;
1955 }
1956
1957 /**
1958 * @param array $error
1959 * @return Status
1960 */
1961 public function convertVerifyErrorToStatus( $error ) {
1962 $code = $error['status'];
1963 unset( $code['status'] );
1964
1965 return Status::newFatal( $this->getVerificationErrorCode( $code ), $error );
1966 }
1967
1968 /**
1969 * @param null|string $forType
1970 * @return int
1971 */
1972 public static function getMaxUploadSize( $forType = null ) {
1973 global $wgMaxUploadSize;
1974
1975 if ( is_array( $wgMaxUploadSize ) ) {
1976 if ( !is_null( $forType ) && isset( $wgMaxUploadSize[$forType] ) ) {
1977 return $wgMaxUploadSize[$forType];
1978 } else {
1979 return $wgMaxUploadSize['*'];
1980 }
1981 } else {
1982 return intval( $wgMaxUploadSize );
1983 }
1984 }
1985
1986 /**
1987 * Get the current status of a chunked upload (used for polling)
1988 *
1989 * The value will be read from cache.
1990 *
1991 * @param User $user
1992 * @param string $statusKey
1993 * @return Status[]|bool
1994 */
1995 public static function getSessionStatus( User $user, $statusKey ) {
1996 $key = wfMemcKey( 'uploadstatus', $user->getId() ?: md5( $user->getName() ), $statusKey );
1997
1998 return wfGetCache( CACHE_ANYTHING )->get( $key );
1999 }
2000
2001 /**
2002 * Set the current status of a chunked upload (used for polling)
2003 *
2004 * The value will be set in cache for 1 day
2005 *
2006 * @param User $user
2007 * @param string $statusKey
2008 * @param array|bool $value
2009 * @return void
2010 */
2011 public static function setSessionStatus( User $user, $statusKey, $value ) {
2012 $key = wfMemcKey( 'uploadstatus', $user->getId() ?: md5( $user->getName() ), $statusKey );
2013
2014 $cache = wfGetCache( CACHE_ANYTHING );
2015 if ( $value === false ) {
2016 $cache->delete( $key );
2017 } else {
2018 $cache->set( $key, $value, 86400 );
2019 }
2020 }
2021 }