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
24use MediaWiki\HookContainer\ProtectedHookAccessorTrait;
25use MediaWiki\MediaWikiServices;
26use MediaWiki\Permissions\Authority;
27use MediaWiki\Permissions\PermissionStatus;
28use MediaWiki\Shell\Shell;
29use MediaWiki\User\UserIdentity;
30
31/**
32 * @defgroup Upload Upload related
33 */
34
35/**
36 * @ingroup Upload
37 *
38 * UploadBase and subclasses are the backend of MediaWiki's file uploads.
39 * The frontends are formed by ApiUpload and SpecialUpload.
40 *
41 * @stable to extend
42 *
43 * @author Brion Vibber
44 * @author Bryan Tong Minh
45 * @author Michael Dale
46 */
47abstract class UploadBase {
48	use ProtectedHookAccessorTrait;
49
50	/** @var string|null Local file system path to the file to upload (or a local copy) */
51	protected $mTempPath;
52	/** @var TempFSFile|null Wrapper to handle deleting the temp file */
53	protected $tempFileObj;
54	/** @var string|null */
55	protected $mDesiredDestName;
56	/** @var string|null */
57	protected $mDestName;
58	/** @var bool|null */
59	protected $mRemoveTempFile;
60	/** @var string|null */
61	protected $mSourceType;
62	/** @var Title|bool */
63	protected $mTitle = false;
64	/** @var int */
65	protected $mTitleError = 0;
66	/** @var string|null */
67	protected $mFilteredName;
68	/** @var string|null */
69	protected $mFinalExtension;
70	/** @var LocalFile|null */
71	protected $mLocalFile;
72	/** @var UploadStashFile|null */
73	protected $mStashFile;
74	/** @var int|null */
75	protected $mFileSize;
76	/** @var array|null */
77	protected $mFileProps;
78	/** @var string[] */
79	protected $mBlackListedExtensions;
80	/** @var bool|null */
81	protected $mJavaDetected;
82	/** @var string|null */
83	protected $mSVGNSError;
84
85	protected static $safeXmlEncodings = [
86		'UTF-8',
87		'ISO-8859-1',
88		'ISO-8859-2',
89		'UTF-16',
90		'UTF-32',
91		'WINDOWS-1250',
92		'WINDOWS-1251',
93		'WINDOWS-1252',
94		'WINDOWS-1253',
95		'WINDOWS-1254',
96		'WINDOWS-1255',
97		'WINDOWS-1256',
98		'WINDOWS-1257',
99		'WINDOWS-1258',
100	];
101
102	public const SUCCESS = 0;
103	public const OK = 0;
104	public const EMPTY_FILE = 3;
105	public const MIN_LENGTH_PARTNAME = 4;
106	public const ILLEGAL_FILENAME = 5;
107	public const OVERWRITE_EXISTING_FILE = 7; # Not used anymore; handled by verifyTitlePermissions()
108	public const FILETYPE_MISSING = 8;
109	public const FILETYPE_BADTYPE = 9;
110	public const VERIFICATION_ERROR = 10;
111	public const HOOK_ABORTED = 11;
112	public const FILE_TOO_LARGE = 12;
113	public const WINDOWS_NONASCII_FILENAME = 13;
114	public const FILENAME_TOO_LONG = 14;
115
116	/**
117	 * @param int $error
118	 * @return string
119	 */
120	public function getVerificationErrorCode( $error ) {
121		$code_to_status = [
122			self::EMPTY_FILE => 'empty-file',
123			self::FILE_TOO_LARGE => 'file-too-large',
124			self::FILETYPE_MISSING => 'filetype-missing',
125			self::FILETYPE_BADTYPE => 'filetype-banned',
126			self::MIN_LENGTH_PARTNAME => 'filename-tooshort',
127			self::ILLEGAL_FILENAME => 'illegal-filename',
128			self::OVERWRITE_EXISTING_FILE => 'overwrite',
129			self::VERIFICATION_ERROR => 'verification-error',
130			self::HOOK_ABORTED => 'hookaborted',
131			self::WINDOWS_NONASCII_FILENAME => 'windows-nonascii-filename',
132			self::FILENAME_TOO_LONG => 'filename-toolong',
133		];
134		return $code_to_status[$error] ?? 'unknown-error';
135	}
136
137	/**
138	 * Returns true if uploads are enabled.
139	 * Can be override by subclasses.
140	 * @stable to override
141	 * @return bool
142	 */
143	public static function isEnabled() {
144		global $wgEnableUploads;
145
146		return $wgEnableUploads && wfIniGetBool( 'file_uploads' );
147	}
148
149	/**
150	 * Returns true if the user can use this upload module or else a string
151	 * identifying the missing permission.
152	 * Can be overridden by subclasses.
153	 *
154	 * @param Authority $performer
155	 * @return bool|string
156	 */
157	public static function isAllowed( Authority $performer ) {
158		foreach ( [ 'upload', 'edit' ] as $permission ) {
159			if ( !$performer->isAllowed( $permission ) ) {
160				return $permission;
161			}
162		}
163
164		return true;
165	}
166
167	/**
168	 * Returns true if the user has surpassed the upload rate limit, false otherwise.
169	 *
170	 * @param User $user
171	 * @return bool
172	 */
173	public static function isThrottled( $user ) {
174		return $user->pingLimiter( 'upload' );
175	}
176
177	/** @var string[] Upload handlers. Should probably just be a global. */
178	private static $uploadHandlers = [ 'Stash', 'File', 'Url' ];
179
180	/**
181	 * Create a form of UploadBase depending on wpSourceType and initializes it
182	 *
183	 * @param WebRequest &$request
184	 * @param string|null $type
185	 * @return null|self
186	 */
187	public static function createFromRequest( &$request, $type = null ) {
188		$type = $type ?: $request->getVal( 'wpSourceType', 'File' );
189
190		if ( !$type ) {
191			return null;
192		}
193
194		// Get the upload class
195		$type = ucfirst( $type );
196
197		// Give hooks the chance to handle this request
198		/** @var self|null $className */
199		$className = null;
200		Hooks::runner()->onUploadCreateFromRequest( $type, $className );
201		if ( $className === null ) {
202			$className = 'UploadFrom' . $type;
203			wfDebug( __METHOD__ . ": class name: $className" );
204			if ( !in_array( $type, self::$uploadHandlers ) ) {
205				return null;
206			}
207		}
208
209		// Check whether this upload class is enabled
210		if ( !$className::isEnabled() ) {
211			return null;
212		}
213
214		// Check whether the request is valid
215		if ( !$className::isValidRequest( $request ) ) {
216			return null;
217		}
218
219		/** @var self $handler */
220		$handler = new $className;
221
222		$handler->initializeFromRequest( $request );
223
224		return $handler;
225	}
226
227	/**
228	 * Check whether a request if valid for this handler
229	 * @param WebRequest $request
230	 * @return bool
231	 */
232	public static function isValidRequest( $request ) {
233		return false;
234	}
235
236	/**
237	 * @stable to call
238	 */
239	public function __construct() {
240	}
241
242	/**
243	 * Returns the upload type. Should be overridden by child classes
244	 *
245	 * @since 1.18
246	 * @stable to override
247	 * @return string
248	 */
249	public function getSourceType() {
250		return null;
251	}
252
253	/**
254	 * @param string $name The desired destination name
255	 * @param string $tempPath
256	 * @param int|null $fileSize
257	 * @param bool $removeTempFile (false) remove the temporary file?
258	 * @throws MWException
259	 */
260	public function initializePathInfo( $name, $tempPath, $fileSize, $removeTempFile = false ) {
261		$this->mDesiredDestName = $name;
262		if ( FileBackend::isStoragePath( $tempPath ) ) {
263			throw new MWException( __METHOD__ . " given storage path `$tempPath`." );
264		}
265
266		$this->setTempFile( $tempPath, $fileSize );
267		$this->mRemoveTempFile = $removeTempFile;
268	}
269
270	/**
271	 * Initialize from a WebRequest. Override this in a subclass.
272	 *
273	 * @param WebRequest &$request
274	 */
275	abstract public function initializeFromRequest( &$request );
276
277	/**
278	 * @param string $tempPath File system path to temporary file containing the upload
279	 * @param int|null $fileSize
280	 */
281	protected function setTempFile( $tempPath, $fileSize = null ) {
282		$this->mTempPath = $tempPath;
283		$this->mFileSize = $fileSize ?: null;
284		if ( strlen( $this->mTempPath ) && file_exists( $this->mTempPath ) ) {
285			$this->tempFileObj = new TempFSFile( $this->mTempPath );
286			if ( !$fileSize ) {
287				$this->mFileSize = filesize( $this->mTempPath );
288			}
289		} else {
290			$this->tempFileObj = null;
291		}
292	}
293
294	/**
295	 * Fetch the file. Usually a no-op
296	 * @stable to override
297	 * @return Status
298	 */
299	public function fetchFile() {
300		return Status::newGood();
301	}
302
303	/**
304	 * Return true if the file is empty
305	 * @return bool
306	 */
307	public function isEmptyFile() {
308		return empty( $this->mFileSize );
309	}
310
311	/**
312	 * Return the file size
313	 * @return int
314	 */
315	public function getFileSize() {
316		return $this->mFileSize;
317	}
318
319	/**
320	 * Get the base 36 SHA1 of the file
321	 * @stable to override
322	 * @return string|false
323	 */
324	public function getTempFileSha1Base36() {
325		return FSFile::getSha1Base36FromPath( $this->mTempPath );
326	}
327
328	/**
329	 * @param string $srcPath The source path
330	 * @return string|bool The real path if it was a virtual URL Returns false on failure
331	 */
332	public function getRealPath( $srcPath ) {
333		$repo = MediaWikiServices::getInstance()->getRepoGroup()->getLocalRepo();
334		if ( FileRepo::isVirtualUrl( $srcPath ) ) {
335			/** @todo Just make uploads work with storage paths UploadFromStash
336			 *  loads files via virtual URLs.
337			 */
338			$tmpFile = $repo->getLocalCopy( $srcPath );
339			if ( $tmpFile ) {
340				$tmpFile->bind( $this ); // keep alive with $this
341			}
342			$path = $tmpFile ? $tmpFile->getPath() : false;
343		} else {
344			$path = $srcPath;
345		}
346
347		return $path;
348	}
349
350	/**
351	 * Verify whether the upload is sane.
352	 *
353	 * Return a status array representing the outcome of the verification.
354	 * Possible keys are:
355	 * - 'status': set to self::OK in case of success, or to one of the error constants defined in
356	 *   this class in case of failure
357	 * - 'max': set to the maximum allowed file size ($wgMaxUploadSize) if the upload is too large
358	 * - 'details': set to error details if the file type is valid but contents are corrupt
359	 * - 'filtered': set to the sanitized file name if the requested file name is invalid
360	 * - 'finalExt': set to the file's file extension if it is not an allowed file extension
361	 * - 'blacklistedExt': set to the list of blacklisted file extensions if the current file extension
362	 *    is not allowed for uploads and the blacklist is not empty
363	 *
364	 * @stable to override
365	 * @return mixed[] array representing the result of the verification
366	 */
367	public function verifyUpload() {
368		/**
369		 * If there was no filename or a zero size given, give up quick.
370		 */
371		if ( $this->isEmptyFile() ) {
372			return [ 'status' => self::EMPTY_FILE ];
373		}
374
375		/**
376		 * Honor $wgMaxUploadSize
377		 */
378		$maxSize = self::getMaxUploadSize( $this->getSourceType() );
379		if ( $this->mFileSize > $maxSize ) {
380			return [
381				'status' => self::FILE_TOO_LARGE,
382				'max' => $maxSize,
383			];
384		}
385
386		/**
387		 * Look at the contents of the file; if we can recognize the
388		 * type but it's corrupt or data of the wrong type, we should
389		 * probably not accept it.
390		 */
391		$verification = $this->verifyFile();
392		if ( $verification !== true ) {
393			return [
394				'status' => self::VERIFICATION_ERROR,
395				'details' => $verification
396			];
397		}
398
399		/**
400		 * Make sure this file can be created
401		 */
402		$result = $this->validateName();
403		if ( $result !== true ) {
404			return $result;
405		}
406
407		return [ 'status' => self::OK ];
408	}
409
410	/**
411	 * Verify that the name is valid and, if necessary, that we can overwrite
412	 *
413	 * @return array|bool True if valid, otherwise an array with 'status'
414	 * and other keys
415	 */
416	public function validateName() {
417		$nt = $this->getTitle();
418		if ( $nt === null ) {
419			$result = [ 'status' => $this->mTitleError ];
420			if ( $this->mTitleError == self::ILLEGAL_FILENAME ) {
421				$result['filtered'] = $this->mFilteredName;
422			}
423			if ( $this->mTitleError == self::FILETYPE_BADTYPE ) {
424				$result['finalExt'] = $this->mFinalExtension;
425				if ( count( $this->mBlackListedExtensions ) ) {
426					$result['blacklistedExt'] = $this->mBlackListedExtensions;
427				}
428			}
429
430			return $result;
431		}
432		$this->mDestName = $this->getLocalFile()->getName();
433
434		return true;
435	}
436
437	/**
438	 * Verify the MIME type.
439	 *
440	 * @note Only checks that it is not an evil MIME. The "does it have
441	 *  correct extension given its MIME type?" check is in verifyFile.
442	 *  in `verifyFile()` that MIME type and file extension correlate.
443	 * @param string $mime Representing the MIME
444	 * @return array|bool True if the file is verified, an array otherwise
445	 */
446	protected function verifyMimeType( $mime ) {
447		global $wgVerifyMimeType, $wgVerifyMimeTypeIE;
448		if ( $wgVerifyMimeType ) {
449			wfDebug( "mime: <$mime> extension: <{$this->mFinalExtension}>" );
450			global $wgMimeTypeExclusions;
451			if ( self::checkFileExtension( $mime, $wgMimeTypeExclusions ) ) {
452				return [ 'filetype-badmime', $mime ];
453			}
454
455			if ( $wgVerifyMimeTypeIE ) {
456				# Check what Internet Explorer would detect
457				$fp = fopen( $this->mTempPath, 'rb' );
458				if ( $fp ) {
459					$chunk = fread( $fp, 256 );
460					fclose( $fp );
461
462					$magic = MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer();
463					$extMime = $magic->getMimeTypeFromExtensionOrNull( (string)$this->mFinalExtension ) ?? '';
464					$ieTypes = $magic->getIEMimeTypes( $this->mTempPath, $chunk, $extMime );
465					foreach ( $ieTypes as $ieType ) {
466						if ( self::checkFileExtension( $ieType, $wgMimeTypeExclusions ) ) {
467							return [ 'filetype-bad-ie-mime', $ieType ];
468						}
469					}
470				}
471			}
472		}
473
474		return true;
475	}
476
477	/**
478	 * Verifies that it's ok to include the uploaded file
479	 *
480	 * @return array|bool True of the file is verified, array otherwise.
481	 */
482	protected function verifyFile() {
483		global $wgVerifyMimeType, $wgDisableUploadScriptChecks;
484
485		$status = $this->verifyPartialFile();
486		if ( $status !== true ) {
487			return $status;
488		}
489
490		$mwProps = new MWFileProps( MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer() );
491		$this->mFileProps = $mwProps->getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
492		$mime = $this->mFileProps['mime'];
493
494		if ( $wgVerifyMimeType ) {
495			# XXX: Missing extension will be caught by validateName() via getTitle()
496			if ( (string)$this->mFinalExtension !== '' &&
497				!$this->verifyExtension( $mime, $this->mFinalExtension )
498			) {
499				return [ 'filetype-mime-mismatch', $this->mFinalExtension, $mime ];
500			}
501		}
502
503		# check for htmlish code and javascript
504		if ( !$wgDisableUploadScriptChecks ) {
505			if ( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
506				$svgStatus = $this->detectScriptInSvg( $this->mTempPath, false );
507				if ( $svgStatus !== false ) {
508					return $svgStatus;
509				}
510			}
511		}
512
513		$handler = MediaHandler::getHandler( $mime );
514		if ( $handler ) {
515			$handlerStatus = $handler->verifyUpload( $this->mTempPath );
516			if ( !$handlerStatus->isOK() ) {
517				$errors = $handlerStatus->getErrorsArray();
518
519				return reset( $errors );
520			}
521		}
522
523		$error = true;
524		$this->getHookRunner()->onUploadVerifyFile( $this, $mime, $error );
525		if ( $error !== true ) {
526			if ( !is_array( $error ) ) {
527				$error = [ $error ];
528			}
529			return $error;
530		}
531
532		wfDebug( __METHOD__ . ": all clear; passing." );
533
534		return true;
535	}
536
537	/**
538	 * A verification routine suitable for partial files
539	 *
540	 * Runs the blacklist checks, but not any checks that may
541	 * assume the entire file is present.
542	 *
543	 * @return array|bool True if the file is valid, else an array with error message key.
544	 */
545	protected function verifyPartialFile() {
546		global $wgAllowJavaUploads, $wgDisableUploadScriptChecks;
547
548		# getTitle() sets some internal parameters like $this->mFinalExtension
549		$this->getTitle();
550
551		$mwProps = new MWFileProps( MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer() );
552		$this->mFileProps = $mwProps->getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
553
554		# check MIME type, if desired
555		$mime = $this->mFileProps['file-mime'];
556		$status = $this->verifyMimeType( $mime );
557		if ( $status !== true ) {
558			return $status;
559		}
560
561		# check for htmlish code and javascript
562		if ( !$wgDisableUploadScriptChecks ) {
563			if ( self::detectScript( $this->mTempPath, $mime, $this->mFinalExtension ) ) {
564				return [ 'uploadscripted' ];
565			}
566			if ( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
567				$svgStatus = $this->detectScriptInSvg( $this->mTempPath, true );
568				if ( $svgStatus !== false ) {
569					return $svgStatus;
570				}
571			}
572		}
573
574		# Check for Java applets, which if uploaded can bypass cross-site
575		# restrictions.
576		if ( !$wgAllowJavaUploads ) {
577			$this->mJavaDetected = false;
578			$zipStatus = ZipDirectoryReader::read( $this->mTempPath,
579				[ $this, 'zipEntryCallback' ] );
580			if ( !$zipStatus->isOK() ) {
581				$errors = $zipStatus->getErrorsArray();
582				$error = reset( $errors );
583				if ( $error[0] !== 'zip-wrong-format' ) {
584					return $error;
585				}
586			}
587			if ( $this->mJavaDetected ) {
588				return [ 'uploadjava' ];
589			}
590		}
591
592		# Scan the uploaded file for viruses
593		$virus = $this->detectVirus( $this->mTempPath );
594		if ( $virus ) {
595			return [ 'uploadvirus', $virus ];
596		}
597
598		return true;
599	}
600
601	/**
602	 * Callback for ZipDirectoryReader to detect Java class files.
603	 *
604	 * @param array $entry
605	 */
606	public function zipEntryCallback( $entry ) {
607		$names = [ $entry['name'] ];
608
609		// If there is a null character, cut off the name at it, because JDK's
610		// ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name
611		// were constructed which had ".class\0" followed by a string chosen to
612		// make the hash collide with the truncated name, that file could be
613		// returned in response to a request for the .class file.
614		$nullPos = strpos( $entry['name'], "\000" );
615		if ( $nullPos !== false ) {
616			$names[] = substr( $entry['name'], 0, $nullPos );
617		}
618
619		// If there is a trailing slash in the file name, we have to strip it,
620		// because that's what ZIP_GetEntry() does.
621		if ( preg_grep( '!\.class/?$!', $names ) ) {
622			$this->mJavaDetected = true;
623		}
624	}
625
626	/**
627	 * Alias for verifyTitlePermissions. The function was originally
628	 * 'verifyPermissions', but that suggests it's checking the user, when it's
629	 * really checking the title + user combination.
630	 *
631	 * @param Authority $performer to verify the permissions against
632	 * @return array|bool An array as returned by getPermissionErrors or true
633	 *   in case the user has proper permissions.
634	 */
635	public function verifyPermissions( Authority $performer ) {
636		return $this->verifyTitlePermissions( $performer );
637	}
638
639	/**
640	 * Check whether the user can edit, upload and create the image. This
641	 * checks only against the current title; if it returns errors, it may
642	 * very well be that another title will not give errors. Therefore
643	 * isAllowed() should be called as well for generic is-user-blocked or
644	 * can-user-upload checking.
645	 *
646	 * @param Authority $performer to verify the permissions against
647	 * @return array|bool An array as returned by getPermissionErrors or true
648	 *   in case the user has proper permissions.
649	 */
650	public function verifyTitlePermissions( Authority $performer ) {
651		/**
652		 * If the image is protected, non-sysop users won't be able
653		 * to modify it by uploading a new revision.
654		 */
655		$nt = $this->getTitle();
656		if ( $nt === null ) {
657			return true;
658		}
659
660		$status = PermissionStatus::newEmpty();
661		$performer->authorizeWrite( 'edit', $nt, $status );
662		$performer->authorizeWrite( 'upload', $nt, $status );
663		if ( !$status->isGood() ) {
664			return $status->toLegacyErrorArray();
665		}
666
667		$overwriteError = $this->checkOverwrite( $performer );
668		if ( $overwriteError !== true ) {
669			return [ $overwriteError ];
670		}
671
672		return true;
673	}
674
675	/**
676	 * Check for non fatal problems with the file.
677	 *
678	 * This should not assume that mTempPath is set.
679	 *
680	 * @param User|null $user Accepted since 1.35
681	 *
682	 * @return mixed[] Array of warnings
683	 */
684	public function checkWarnings( $user = null ) {
685		if ( $user === null ) {
686			// TODO check uses and hard deprecate
687			$user = RequestContext::getMain()->getUser();
688		}
689
690		$warnings = [];
691
692		$localFile = $this->getLocalFile();
693		$localFile->load( File::READ_LATEST );
694		$filename = $localFile->getName();
695		$hash = $this->getTempFileSha1Base36();
696
697		$badFileName = $this->checkBadFileName( $filename, $this->mDesiredDestName );
698		if ( $badFileName !== null ) {
699			$warnings['badfilename'] = $badFileName;
700		}
701
702		$unwantedFileExtensionDetails = $this->checkUnwantedFileExtensions( (string)$this->mFinalExtension );
703		if ( $unwantedFileExtensionDetails !== null ) {
704			$warnings['filetype-unwanted-type'] = $unwantedFileExtensionDetails;
705		}
706
707		$fileSizeWarnings = $this->checkFileSize( $this->mFileSize );
708		if ( $fileSizeWarnings ) {
709			$warnings = array_merge( $warnings, $fileSizeWarnings );
710		}
711
712		$localFileExistsWarnings = $this->checkLocalFileExists( $localFile, $hash );
713		if ( $localFileExistsWarnings ) {
714			$warnings = array_merge( $warnings, $localFileExistsWarnings );
715		}
716
717		if ( $this->checkLocalFileWasDeleted( $localFile ) ) {
718			$warnings['was-deleted'] = $filename;
719		}
720
721		// If a file with the same name exists locally then the local file has already been tested
722		// for duplication of content
723		$ignoreLocalDupes = isset( $warnings['exists'] );
724		$dupes = $this->checkAgainstExistingDupes( $hash, $ignoreLocalDupes );
725		if ( $dupes ) {
726			$warnings['duplicate'] = $dupes;
727		}
728
729		$archivedDupes = $this->checkAgainstArchiveDupes( $hash, $user );
730		if ( $archivedDupes !== null ) {
731			$warnings['duplicate-archive'] = $archivedDupes;
732		}
733
734		return $warnings;
735	}
736
737	/**
738	 * Convert the warnings array returned by checkWarnings() to something that
739	 * can be serialized. File objects will be converted to an associative array
740	 * with the following keys:
741	 *
742	 *   - fileName: The name of the file
743	 *   - timestamp: The upload timestamp
744	 *
745	 * @param mixed[] $warnings
746	 * @return mixed[]
747	 */
748	public static function makeWarningsSerializable( $warnings ) {
749		array_walk_recursive( $warnings, static function ( &$param, $key ) {
750			if ( $param instanceof File ) {
751				$param = [
752					'fileName' => $param->getName(),
753					'timestamp' => $param->getTimestamp()
754				];
755			} elseif ( is_object( $param ) ) {
756				throw new InvalidArgumentException(
757					'UploadBase::makeWarningsSerializable: ' .
758					'Unexpected object of class ' . get_class( $param ) );
759			}
760		} );
761		return $warnings;
762	}
763
764	/**
765	 * Check whether the resulting filename is different from the desired one,
766	 * but ignore things like ucfirst() and spaces/underscore things
767	 *
768	 * @param string $filename
769	 * @param string $desiredFileName
770	 *
771	 * @return string|null String that was determined to be bad or null if the filename is okay
772	 */
773	private function checkBadFileName( $filename, $desiredFileName ) {
774		$comparableName = str_replace( ' ', '_', $desiredFileName );
775		$comparableName = Title::capitalize( $comparableName, NS_FILE );
776
777		if ( $desiredFileName != $filename && $comparableName != $filename ) {
778			return $filename;
779		}
780
781		return null;
782	}
783
784	/**
785	 * @param string $fileExtension The file extension to check
786	 *
787	 * @return array|null array with the following keys:
788	 *                    0 => string The final extension being used
789	 *                    1 => string[] The extensions that are allowed
790	 *                    2 => int The number of extensions that are allowed.
791	 */
792	private function checkUnwantedFileExtensions( $fileExtension ) {
793		global $wgCheckFileExtensions, $wgFileExtensions, $wgLang;
794
795		if ( $wgCheckFileExtensions ) {
796			$extensions = array_unique( $wgFileExtensions );
797			if ( !$this->checkFileExtension( $fileExtension, $extensions ) ) {
798				return [
799					$fileExtension,
800					$wgLang->commaList( $extensions ),
801					count( $extensions )
802				];
803			}
804		}
805
806		return null;
807	}
808
809	/**
810	 * @param int $fileSize
811	 *
812	 * @return array warnings
813	 */
814	private function checkFileSize( $fileSize ) {
815		global $wgUploadSizeWarning;
816
817		$warnings = [];
818
819		if ( $wgUploadSizeWarning && ( $fileSize > $wgUploadSizeWarning ) ) {
820			$warnings['large-file'] = [
821				Message::sizeParam( $wgUploadSizeWarning ),
822				Message::sizeParam( $fileSize ),
823			];
824		}
825
826		if ( $fileSize == 0 ) {
827			$warnings['empty-file'] = true;
828		}
829
830		return $warnings;
831	}
832
833	/**
834	 * @param LocalFile $localFile
835	 * @param string|false $hash sha1 hash of the file to check
836	 *
837	 * @return array warnings
838	 */
839	private function checkLocalFileExists( LocalFile $localFile, $hash ) {
840		$warnings = [];
841
842		$exists = self::getExistsWarning( $localFile );
843		if ( $exists !== false ) {
844			$warnings['exists'] = $exists;
845
846			// check if file is an exact duplicate of current file version
847			if ( $hash !== false && $hash === $localFile->getSha1() ) {
848				$warnings['no-change'] = $localFile;
849			}
850
851			// check if file is an exact duplicate of older versions of this file
852			$history = $localFile->getHistory();
853			foreach ( $history as $oldFile ) {
854				if ( $hash === $oldFile->getSha1() ) {
855					$warnings['duplicate-version'][] = $oldFile;
856				}
857			}
858		}
859
860		return $warnings;
861	}
862
863	private function checkLocalFileWasDeleted( LocalFile $localFile ) {
864		return $localFile->wasDeleted() && !$localFile->exists();
865	}
866
867	/**
868	 * @param string|false $hash sha1 hash of the file to check
869	 * @param bool $ignoreLocalDupes True to ignore local duplicates
870	 *
871	 * @return File[] Duplicate files, if found.
872	 */
873	private function checkAgainstExistingDupes( $hash, $ignoreLocalDupes ) {
874		if ( $hash === false ) {
875			return [];
876		}
877		$dupes = MediaWikiServices::getInstance()->getRepoGroup()->findBySha1( $hash );
878		$title = $this->getTitle();
879		foreach ( $dupes as $key => $dupe ) {
880			if (
881				( $dupe instanceof LocalFile ) &&
882				$ignoreLocalDupes &&
883				$title->equals( $dupe->getTitle() )
884			) {
885				unset( $dupes[$key] );
886			}
887		}
888
889		return $dupes;
890	}
891
892	/**
893	 * @param string|false $hash sha1 hash of the file to check
894	 * @param Authority $performer
895	 *
896	 * @return string|null Name of the dupe or empty string if discovered (depending on visibility)
897	 *                     null if the check discovered no dupes.
898	 */
899	private function checkAgainstArchiveDupes( $hash, Authority $performer ) {
900		if ( $hash === false ) {
901			return null;
902		}
903		$archivedFile = new ArchivedFile( null, 0, '', $hash );
904		if ( $archivedFile->getID() > 0 ) {
905			if ( $archivedFile->userCan( File::DELETED_FILE, $performer ) ) {
906				return $archivedFile->getName();
907			} else {
908				return '';
909			}
910		}
911
912		return null;
913	}
914
915	/**
916	 * Really perform the upload. Stores the file in the local repo, watches
917	 * if necessary and runs the UploadComplete hook.
918	 *
919	 * @param string $comment
920	 * @param string $pageText
921	 * @param bool $watch Whether the file page should be added to user's watchlist.
922	 *   (This doesn't check $user's permissions.)
923	 * @param User $user
924	 * @param string[] $tags Change tags to add to the log entry and page revision.
925	 *   (This doesn't check $user's permissions.)
926	 * @param string|null $watchlistExpiry Optional watchlist expiry timestamp in any format
927	 *   acceptable to wfTimestamp().
928	 * @return Status Indicating the whether the upload succeeded.
929	 *
930	 * @since 1.35 Accepts $watchlistExpiry parameter.
931	 */
932	public function performUpload(
933		$comment, $pageText, $watch, $user, $tags = [], ?string $watchlistExpiry = null
934	) {
935		$this->getLocalFile()->load( File::READ_LATEST );
936		$props = $this->mFileProps;
937
938		$error = null;
939		$this->getHookRunner()->onUploadVerifyUpload( $this, $user, $props, $comment, $pageText, $error );
940		if ( $error ) {
941			if ( !is_array( $error ) ) {
942				$error = [ $error ];
943			}
944			return Status::newFatal( ...$error );
945		}
946
947		$status = $this->getLocalFile()->upload(
948			$this->mTempPath,
949			$comment,
950			$pageText,
951			File::DELETE_SOURCE,
952			$props,
953			false,
954			$user,
955			$tags
956		);
957
958		if ( $status->isGood() ) {
959			if ( $watch ) {
960				MediaWikiServices::getInstance()->getWatchlistManager()->addWatchIgnoringRights(
961					$user,
962					$this->getLocalFile()->getTitle(),
963					$watchlistExpiry
964				);
965			}
966			$this->getHookRunner()->onUploadComplete( $this );
967
968			$this->postProcessUpload();
969		}
970
971		return $status;
972	}
973
974	/**
975	 * Perform extra steps after a successful upload.
976	 *
977	 * @stable to override
978	 * @since  1.25
979	 */
980	public function postProcessUpload() {
981	}
982
983	/**
984	 * Returns the title of the file to be uploaded. Sets mTitleError in case
985	 * the name was illegal.
986	 *
987	 * @return Title|null The title of the file or null in case the name was illegal
988	 */
989	public function getTitle() {
990		if ( $this->mTitle !== false ) {
991			return $this->mTitle;
992		}
993		if ( !is_string( $this->mDesiredDestName ) ) {
994			$this->mTitleError = self::ILLEGAL_FILENAME;
995			$this->mTitle = null;
996
997			return $this->mTitle;
998		}
999		/* Assume that if a user specified File:Something.jpg, this is an error
1000		 * and that the namespace prefix needs to be stripped of.
1001		 */
1002		$title = Title::newFromText( $this->mDesiredDestName );
1003		if ( $title && $title->getNamespace() === NS_FILE ) {
1004			$this->mFilteredName = $title->getDBkey();
1005		} else {
1006			$this->mFilteredName = $this->mDesiredDestName;
1007		}
1008
1009		# oi_archive_name is max 255 bytes, which include a timestamp and an
1010		# exclamation mark, so restrict file name to 240 bytes.
1011		if ( strlen( $this->mFilteredName ) > 240 ) {
1012			$this->mTitleError = self::FILENAME_TOO_LONG;
1013			$this->mTitle = null;
1014
1015			return $this->mTitle;
1016		}
1017
1018		/**
1019		 * Chop off any directories in the given filename. Then
1020		 * filter out illegal characters, and try to make a legible name
1021		 * out of it. We'll strip some silently that Title would die on.
1022		 */
1023		$this->mFilteredName = wfStripIllegalFilenameChars( $this->mFilteredName );
1024		/* Normalize to title form before we do any further processing */
1025		$nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
1026		if ( $nt === null ) {
1027			$this->mTitleError = self::ILLEGAL_FILENAME;
1028			$this->mTitle = null;
1029
1030			return $this->mTitle;
1031		}
1032		$this->mFilteredName = $nt->getDBkey();
1033
1034		/**
1035		 * We'll want to blacklist against *any* 'extension', and use
1036		 * only the final one for the whitelist.
1037		 */
1038		list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName );
1039
1040		if ( $ext !== [] ) {
1041			$this->mFinalExtension = trim( end( $ext ) );
1042		} else {
1043			$this->mFinalExtension = '';
1044
1045			// No extension, try guessing one from the temporary file
1046			// FIXME: Sometimes we mTempPath isn't set yet here, possibly due to an unrealistic
1047			// or incomplete test case in UploadBaseTest (T272328)
1048			if ( $this->mTempPath !== null ) {
1049				$magic = MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer();
1050				$mime = $magic->guessMimeType( $this->mTempPath );
1051				if ( $mime !== 'unknown/unknown' ) {
1052					# Get a space separated list of extensions
1053					$mimeExt = $magic->getExtensionFromMimeTypeOrNull( $mime );
1054					if ( $mimeExt !== null ) {
1055						# Set the extension to the canonical extension
1056						$this->mFinalExtension = $mimeExt;
1057
1058						# Fix up the other variables
1059						$this->mFilteredName .= ".{$this->mFinalExtension}";
1060						$nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
1061						$ext = [ $this->mFinalExtension ];
1062					}
1063				}
1064			}
1065		}
1066
1067		// Don't allow users to override the list of prohibited file extensions (check file extension)
1068		global $wgCheckFileExtensions, $wgStrictFileExtensions;
1069		global $wgFileExtensions, $wgProhibitedFileExtensions;
1070
1071		$blackListedExtensions = self::checkFileExtensionList( $ext, $wgProhibitedFileExtensions );
1072
1073		if ( $this->mFinalExtension == '' ) {
1074			$this->mTitleError = self::FILETYPE_MISSING;
1075			$this->mTitle = null;
1076
1077			return $this->mTitle;
1078		} elseif ( $blackListedExtensions ||
1079			( $wgCheckFileExtensions && $wgStrictFileExtensions &&
1080				!$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) )
1081		) {
1082			$this->mBlackListedExtensions = $blackListedExtensions;
1083			$this->mTitleError = self::FILETYPE_BADTYPE;
1084			$this->mTitle = null;
1085
1086			return $this->mTitle;
1087		}
1088
1089		// Windows may be broken with special characters, see T3780
1090		if ( !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() )
1091			&& !MediaWikiServices::getInstance()->getRepoGroup()
1092				->getLocalRepo()->backendSupportsUnicodePaths()
1093		) {
1094			$this->mTitleError = self::WINDOWS_NONASCII_FILENAME;
1095			$this->mTitle = null;
1096
1097			return $this->mTitle;
1098		}
1099
1100		# If there was more than one "extension", reassemble the base
1101		# filename to prevent bogus complaints about length
1102		if ( count( $ext ) > 1 ) {
1103			$iterations = count( $ext ) - 1;
1104			for ( $i = 0; $i < $iterations; $i++ ) {
1105				$partname .= '.' . $ext[$i];
1106			}
1107		}
1108
1109		if ( strlen( $partname ) < 1 ) {
1110			$this->mTitleError = self::MIN_LENGTH_PARTNAME;
1111			$this->mTitle = null;
1112
1113			return $this->mTitle;
1114		}
1115
1116		$this->mTitle = $nt;
1117
1118		return $this->mTitle;
1119	}
1120
1121	/**
1122	 * Return the local file and initializes if necessary.
1123	 *
1124	 * @stable to override
1125	 * @return LocalFile|null
1126	 */
1127	public function getLocalFile() {
1128		if ( $this->mLocalFile === null ) {
1129			$nt = $this->getTitle();
1130			$this->mLocalFile = $nt === null
1131				? null
1132				: MediaWikiServices::getInstance()->getRepoGroup()->getLocalRepo()->newFile( $nt );
1133		}
1134
1135		return $this->mLocalFile;
1136	}
1137
1138	/**
1139	 * @return UploadStashFile|null
1140	 */
1141	public function getStashFile() {
1142		return $this->mStashFile;
1143	}
1144
1145	/**
1146	 * Like stashFile(), but respects extensions' wishes to prevent the stashing. verifyUpload() must
1147	 * be called before calling this method (unless $isPartial is true).
1148	 *
1149	 * Upload stash exceptions are also caught and converted to an error status.
1150	 *
1151	 * @since 1.28
1152	 * @stable to override
1153	 * @param User $user
1154	 * @param bool $isPartial Pass `true` if this is a part of a chunked upload (not a complete file).
1155	 * @return Status If successful, value is an UploadStashFile instance
1156	 */
1157	public function tryStashFile( User $user, $isPartial = false ) {
1158		if ( !$isPartial ) {
1159			$error = $this->runUploadStashFileHook( $user );
1160			if ( $error ) {
1161				return Status::newFatal( ...$error );
1162			}
1163		}
1164		try {
1165			$file = $this->doStashFile( $user );
1166			return Status::newGood( $file );
1167		} catch ( UploadStashException $e ) {
1168			return Status::newFatal( 'uploadstash-exception', get_class( $e ), $e->getMessage() );
1169		}
1170	}
1171
1172	/**
1173	 * @param User $user
1174	 * @return array|null Error message and parameters, null if there's no error
1175	 */
1176	protected function runUploadStashFileHook( User $user ) {
1177		$props = $this->mFileProps;
1178		$error = null;
1179		$this->getHookRunner()->onUploadStashFile( $this, $user, $props, $error );
1180		if ( $error && !is_array( $error ) ) {
1181			$error = [ $error ];
1182		}
1183		return $error;
1184	}
1185
1186	/**
1187	 * Implementation for stashFile() and tryStashFile().
1188	 *
1189	 * @stable to override
1190	 * @param User|null $user
1191	 * @return UploadStashFile Stashed file
1192	 */
1193	protected function doStashFile( User $user = null ) {
1194		$stash = MediaWikiServices::getInstance()->getRepoGroup()
1195			->getLocalRepo()->getUploadStash( $user );
1196		$file = $stash->stashFile( $this->mTempPath, $this->getSourceType() );
1197		$this->mStashFile = $file;
1198
1199		return $file;
1200	}
1201
1202	/**
1203	 * If we've modified the upload file we need to manually remove it
1204	 * on exit to clean up.
1205	 */
1206	public function cleanupTempFile() {
1207		if ( $this->mRemoveTempFile && $this->tempFileObj ) {
1208			// Delete when all relevant TempFSFile handles go out of scope
1209			wfDebug( __METHOD__ . ": Marked temporary file '{$this->mTempPath}' for removal" );
1210			$this->tempFileObj->autocollect();
1211		}
1212	}
1213
1214	public function getTempPath() {
1215		return $this->mTempPath;
1216	}
1217
1218	/**
1219	 * Split a file into a base name and all dot-delimited 'extensions'
1220	 * on the end. Some web server configurations will fall back to
1221	 * earlier pseudo-'extensions' to determine type and execute
1222	 * scripts, so the blacklist needs to check them all.
1223	 *
1224	 * @param string $filename
1225	 * @return array [ string, string[] ]
1226	 */
1227	public static function splitExtensions( $filename ) {
1228		$bits = explode( '.', $filename );
1229		$basename = array_shift( $bits );
1230
1231		return [ $basename, $bits ];
1232	}
1233
1234	/**
1235	 * Perform case-insensitive match against a list of file extensions.
1236	 * Returns true if the extension is in the list.
1237	 *
1238	 * @param string $ext
1239	 * @param array $list
1240	 * @return bool
1241	 */
1242	public static function checkFileExtension( $ext, $list ) {
1243		return in_array( strtolower( $ext ), $list );
1244	}
1245
1246	/**
1247	 * Perform case-insensitive match against a list of file extensions.
1248	 * Returns an array of matching extensions.
1249	 *
1250	 * @param string[] $ext
1251	 * @param string[] $list
1252	 * @return string[]
1253	 */
1254	public static function checkFileExtensionList( $ext, $list ) {
1255		return array_intersect( array_map( 'strtolower', $ext ), $list );
1256	}
1257
1258	/**
1259	 * Checks if the MIME type of the uploaded file matches the file extension.
1260	 *
1261	 * @param string $mime The MIME type of the uploaded file
1262	 * @param string $extension The filename extension that the file is to be served with
1263	 * @return bool
1264	 */
1265	public static function verifyExtension( $mime, $extension ) {
1266		$magic = MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer();
1267
1268		if ( !$mime || $mime == 'unknown' || $mime == 'unknown/unknown' ) {
1269			if ( !$magic->isRecognizableExtension( $extension ) ) {
1270				wfDebug( __METHOD__ . ": passing file with unknown detected mime type; " .
1271					"unrecognized extension '$extension', can't verify" );
1272
1273				return true;
1274			} else {
1275				wfDebug( __METHOD__ . ": rejecting file with unknown detected mime type; " .
1276					"recognized extension '$extension', so probably invalid file" );
1277
1278				return false;
1279			}
1280		}
1281
1282		$match = $magic->isMatchingExtension( $extension, $mime );
1283
1284		if ( $match === null ) {
1285			if ( $magic->getMimeTypesFromExtension( $extension ) !== [] ) {
1286				wfDebug( __METHOD__ . ": No extension known for $mime, but we know a mime for $extension" );
1287
1288				return false;
1289			} else {
1290				wfDebug( __METHOD__ . ": no file extension known for mime type $mime, passing file" );
1291
1292				return true;
1293			}
1294		} elseif ( $match ) {
1295			wfDebug( __METHOD__ . ": mime type $mime matches extension $extension, passing file" );
1296
1297			/** @todo If it's a bitmap, make sure PHP or ImageMagick resp. can handle it! */
1298			return true;
1299		} else {
1300			wfDebug( __METHOD__
1301				. ": mime type $mime mismatches file extension $extension, rejecting file" );
1302
1303			return false;
1304		}
1305	}
1306
1307	/**
1308	 * Heuristic for detecting files that *could* contain JavaScript instructions or
1309	 * things that may look like HTML to a browser and are thus
1310	 * potentially harmful. The present implementation will produce false
1311	 * positives in some situations.
1312	 *
1313	 * @param string $file Pathname to the temporary upload file
1314	 * @param string $mime The MIME type of the file
1315	 * @param string|null $extension The extension of the file
1316	 * @return bool True if the file contains something looking like embedded scripts
1317	 */
1318	public static function detectScript( $file, $mime, $extension ) {
1319		# ugly hack: for text files, always look at the entire file.
1320		# For binary field, just check the first K.
1321
1322		$isText = strpos( $mime, 'text/' ) === 0;
1323		if ( $isText ) {
1324			$chunk = file_get_contents( $file );
1325		} else {
1326			$fp = fopen( $file, 'rb' );
1327			if ( !$fp ) {
1328				return false;
1329			}
1330			$chunk = fread( $fp, 1024 );
1331			fclose( $fp );
1332		}
1333
1334		$chunk = strtolower( $chunk );
1335
1336		if ( !$chunk ) {
1337			return false;
1338		}
1339
1340		# decode from UTF-16 if needed (could be used for obfuscation).
1341		if ( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
1342			$enc = 'UTF-16BE';
1343		} elseif ( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
1344			$enc = 'UTF-16LE';
1345		} else {
1346			$enc = null;
1347		}
1348
1349		if ( $enc !== null ) {
1350			$chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
1351		}
1352
1353		$chunk = trim( $chunk );
1354
1355		/** @todo FIXME: Convert from UTF-16 if necessary! */
1356		wfDebug( __METHOD__ . ": checking for embedded scripts and HTML stuff" );
1357
1358		# check for HTML doctype
1359		if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
1360			return true;
1361		}
1362
1363		// Some browsers will interpret obscure xml encodings as UTF-8, while
1364		// PHP/expat will interpret the given encoding in the xml declaration (T49304)
1365		if ( $extension == 'svg' || strpos( $mime, 'image/svg' ) === 0 ) {
1366			if ( self::checkXMLEncodingMissmatch( $file ) ) {
1367				return true;
1368			}
1369		}
1370
1371		// Quick check for HTML heuristics in old IE and Safari.
1372		//
1373		// The exact heuristics IE uses are checked separately via verifyMimeType(), so we
1374		// don't need them all here as it can cause many false positives.
1375		//
1376		// Check for `<script` and such still to forbid script tags and embedded HTML in SVG:
1377		$tags = [
1378			'<body',
1379			'<head',
1380			'<html', # also in safari
1381			'<script', # also in safari
1382		];
1383
1384		foreach ( $tags as $tag ) {
1385			if ( strpos( $chunk, $tag ) !== false ) {
1386				wfDebug( __METHOD__ . ": found something that may make it be mistaken for html: $tag" );
1387
1388				return true;
1389			}
1390		}
1391
1392		/*
1393		 * look for JavaScript
1394		 */
1395
1396		# resolve entity-refs to look at attributes. may be harsh on big files... cache result?
1397		$chunk = Sanitizer::decodeCharReferences( $chunk );
1398
1399		# look for script-types
1400		if ( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
1401			wfDebug( __METHOD__ . ": found script types" );
1402
1403			return true;
1404		}
1405
1406		# look for html-style script-urls
1407		if ( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1408			wfDebug( __METHOD__ . ": found html-style script urls" );
1409
1410			return true;
1411		}
1412
1413		# look for css-style script-urls
1414		if ( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1415			wfDebug( __METHOD__ . ": found css-style script urls" );
1416
1417			return true;
1418		}
1419
1420		wfDebug( __METHOD__ . ": no scripts found" );
1421
1422		return false;
1423	}
1424
1425	/**
1426	 * Check a whitelist of xml encodings that are known not to be interpreted differently
1427	 * by the server's xml parser (expat) and some common browsers.
1428	 *
1429	 * @param string $file Pathname to the temporary upload file
1430	 * @return bool True if the file contains an encoding that could be misinterpreted
1431	 */
1432	public static function checkXMLEncodingMissmatch( $file ) {
1433		global $wgSVGMetadataCutoff;
1434		$contents = file_get_contents( $file, false, null, 0, $wgSVGMetadataCutoff );
1435		$encodingRegex = '!encoding[ \t\n\r]*=[ \t\n\r]*[\'"](.*?)[\'"]!si';
1436
1437		if ( preg_match( "!<\?xml\b(.*?)\?>!si", $contents, $matches ) ) {
1438			if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1439				&& !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1440			) {
1441				wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'" );
1442
1443				return true;
1444			}
1445		} elseif ( preg_match( "!<\?xml\b!si", $contents ) ) {
1446			// Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1447			// bytes. There shouldn't be a legitimate reason for this to happen.
1448			wfDebug( __METHOD__ . ": Unmatched XML declaration start" );
1449
1450			return true;
1451		} elseif ( substr( $contents, 0, 4 ) == "\x4C\x6F\xA7\x94" ) {
1452			// EBCDIC encoded XML
1453			wfDebug( __METHOD__ . ": EBCDIC Encoded XML" );
1454
1455			return true;
1456		}
1457
1458		// It's possible the file is encoded with multi-byte encoding, so re-encode attempt to
1459		// detect the encoding in case is specifies an encoding not whitelisted in self::$safeXmlEncodings
1460		$attemptEncodings = [ 'UTF-16', 'UTF-16BE', 'UTF-32', 'UTF-32BE' ];
1461		foreach ( $attemptEncodings as $encoding ) {
1462			Wikimedia\suppressWarnings();
1463			$str = iconv( $encoding, 'UTF-8', $contents );
1464			Wikimedia\restoreWarnings();
1465			if ( $str != '' && preg_match( "!<\?xml\b(.*?)\?>!si", $str, $matches ) ) {
1466				if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1467					&& !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1468				) {
1469					wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'" );
1470
1471					return true;
1472				}
1473			} elseif ( $str != '' && preg_match( "!<\?xml\b!si", $str ) ) {
1474				// Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1475				// bytes. There shouldn't be a legitimate reason for this to happen.
1476				wfDebug( __METHOD__ . ": Unmatched XML declaration start" );
1477
1478				return true;
1479			}
1480		}
1481
1482		return false;
1483	}
1484
1485	/**
1486	 * @param string $filename
1487	 * @param bool $partial
1488	 * @return bool|array
1489	 */
1490	protected function detectScriptInSvg( $filename, $partial ) {
1491		$this->mSVGNSError = false;
1492		$check = new XmlTypeCheck(
1493			$filename,
1494			[ $this, 'checkSvgScriptCallback' ],
1495			true,
1496			[
1497				'processing_instruction_handler' => [ __CLASS__, 'checkSvgPICallback' ],
1498				'external_dtd_handler' => [ __CLASS__, 'checkSvgExternalDTD' ],
1499			]
1500		);
1501		if ( $check->wellFormed !== true ) {
1502			// Invalid xml (T60553)
1503			// But only when non-partial (T67724)
1504			return $partial ? false : [ 'uploadinvalidxml' ];
1505		} elseif ( $check->filterMatch ) {
1506			if ( $this->mSVGNSError ) {
1507				return [ 'uploadscriptednamespace', $this->mSVGNSError ];
1508			}
1509
1510			return $check->filterMatchType;
1511		}
1512
1513		return false;
1514	}
1515
1516	/**
1517	 * Callback to filter SVG Processing Instructions.
1518	 * @param string $target Processing instruction name
1519	 * @param string $data Processing instruction attribute and value
1520	 * @return bool|array
1521	 */
1522	public static function checkSvgPICallback( $target, $data ) {
1523		// Don't allow external stylesheets (T59550)
1524		if ( preg_match( '/xml-stylesheet/i', $target ) ) {
1525			return [ 'upload-scripted-pi-callback' ];
1526		}
1527
1528		return false;
1529	}
1530
1531	/**
1532	 * Verify that DTD urls referenced are only the standard dtds
1533	 *
1534	 * Browsers seem to ignore external dtds. However just to be on the
1535	 * safe side, only allow dtds from the svg standard.
1536	 *
1537	 * @param string $type PUBLIC or SYSTEM
1538	 * @param string $publicId The well-known public identifier for the dtd
1539	 * @param string $systemId The url for the external dtd
1540	 * @return bool|array
1541	 */
1542	public static function checkSvgExternalDTD( $type, $publicId, $systemId ) {
1543		// This doesn't include the XHTML+MathML+SVG doctype since we don't
1544		// allow XHTML anyways.
1545		$allowedDTDs = [
1546			'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd',
1547			'http://www.w3.org/TR/2001/REC-SVG-20010904/DTD/svg10.dtd',
1548			'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-basic.dtd',
1549			'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-tiny.dtd',
1550			// https://phabricator.wikimedia.org/T168856
1551			'http://www.w3.org/TR/2001/PR-SVG-20010719/DTD/svg10.dtd',
1552		];
1553		if ( $type !== 'PUBLIC'
1554			|| !in_array( $systemId, $allowedDTDs )
1555			|| strpos( $publicId, "-//W3C//" ) !== 0
1556		) {
1557			return [ 'upload-scripted-dtd' ];
1558		}
1559		return false;
1560	}
1561
1562	/**
1563	 * @todo Replace this with a whitelist filter!
1564	 * @param string $element
1565	 * @param array $attribs
1566	 * @param string|null $data
1567	 * @return bool|array
1568	 */
1569	public function checkSvgScriptCallback( $element, $attribs, $data = null ) {
1570		list( $namespace, $strippedElement ) = $this->splitXmlNamespace( $element );
1571
1572		// We specifically don't include:
1573		// http://www.w3.org/1999/xhtml (T62771)
1574		static $validNamespaces = [
1575			'',
1576			'adobe:ns:meta/',
1577			'http://creativecommons.org/ns#',
1578			'http://inkscape.sourceforge.net/dtd/sodipodi-0.dtd',
1579			'http://ns.adobe.com/adobeillustrator/10.0/',
1580			'http://ns.adobe.com/adobesvgviewerextensions/3.0/',
1581			'http://ns.adobe.com/extensibility/1.0/',
1582			'http://ns.adobe.com/flows/1.0/',
1583			'http://ns.adobe.com/illustrator/1.0/',
1584			'http://ns.adobe.com/imagereplacement/1.0/',
1585			'http://ns.adobe.com/pdf/1.3/',
1586			'http://ns.adobe.com/photoshop/1.0/',
1587			'http://ns.adobe.com/saveforweb/1.0/',
1588			'http://ns.adobe.com/variables/1.0/',
1589			'http://ns.adobe.com/xap/1.0/',
1590			'http://ns.adobe.com/xap/1.0/g/',
1591			'http://ns.adobe.com/xap/1.0/g/img/',
1592			'http://ns.adobe.com/xap/1.0/mm/',
1593			'http://ns.adobe.com/xap/1.0/rights/',
1594			'http://ns.adobe.com/xap/1.0/stype/dimensions#',
1595			'http://ns.adobe.com/xap/1.0/stype/font#',
1596			'http://ns.adobe.com/xap/1.0/stype/manifestitem#',
1597			'http://ns.adobe.com/xap/1.0/stype/resourceevent#',
1598			'http://ns.adobe.com/xap/1.0/stype/resourceref#',
1599			'http://ns.adobe.com/xap/1.0/t/pg/',
1600			'http://purl.org/dc/elements/1.1/',
1601			'http://purl.org/dc/elements/1.1',
1602			'http://schemas.microsoft.com/visio/2003/svgextensions/',
1603			'http://sodipodi.sourceforge.net/dtd/sodipodi-0.dtd',
1604			'http://taptrix.com/inkpad/svg_extensions',
1605			'http://web.resource.org/cc/',
1606			'http://www.freesoftware.fsf.org/bkchem/cdml',
1607			'http://www.inkscape.org/namespaces/inkscape',
1608			'http://www.opengis.net/gml',
1609			'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
1610			'http://www.w3.org/2000/svg',
1611			'http://www.w3.org/tr/rec-rdf-syntax/',
1612			'http://www.w3.org/2000/01/rdf-schema#',
1613		];
1614
1615		// Inkscape mangles namespace definitions created by Adobe Illustrator.
1616		// This is nasty but harmless. (T144827)
1617		$isBuggyInkscape = preg_match( '/^&(#38;)*ns_[a-z_]+;$/', $namespace );
1618
1619		if ( !( $isBuggyInkscape || in_array( $namespace, $validNamespaces ) ) ) {
1620			wfDebug( __METHOD__ . ": Non-svg namespace '$namespace' in uploaded file." );
1621			/** @todo Return a status object to a closure in XmlTypeCheck, for MW1.21+ */
1622			$this->mSVGNSError = $namespace;
1623
1624			return true;
1625		}
1626
1627		/*
1628		 * check for elements that can contain javascript
1629		 */
1630		if ( $strippedElement == 'script' ) {
1631			wfDebug( __METHOD__ . ": Found script element '$element' in uploaded file." );
1632
1633			return [ 'uploaded-script-svg', $strippedElement ];
1634		}
1635
1636		# e.g., <svg xmlns="http://www.w3.org/2000/svg">
1637		#  <handler xmlns:ev="http://www.w3.org/2001/xml-events" ev:event="load">alert(1)</handler> </svg>
1638		if ( $strippedElement == 'handler' ) {
1639			wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file." );
1640
1641			return [ 'uploaded-script-svg', $strippedElement ];
1642		}
1643
1644		# SVG reported in Feb '12 that used xml:stylesheet to generate javascript block
1645		if ( $strippedElement == 'stylesheet' ) {
1646			wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file." );
1647
1648			return [ 'uploaded-script-svg', $strippedElement ];
1649		}
1650
1651		# Block iframes, in case they pass the namespace check
1652		if ( $strippedElement == 'iframe' ) {
1653			wfDebug( __METHOD__ . ": iframe in uploaded file." );
1654
1655			return [ 'uploaded-script-svg', $strippedElement ];
1656		}
1657
1658		# Check <style> css
1659		if ( $strippedElement == 'style'
1660			&& self::checkCssFragment( Sanitizer::normalizeCss( $data ) )
1661		) {
1662			wfDebug( __METHOD__ . ": hostile css in style element." );
1663			return [ 'uploaded-hostile-svg' ];
1664		}
1665
1666		foreach ( $attribs as $attrib => $value ) {
1667			$stripped = $this->stripXmlNamespace( $attrib );
1668			$value = strtolower( $value );
1669
1670			if ( substr( $stripped, 0, 2 ) == 'on' ) {
1671				wfDebug( __METHOD__
1672					. ": Found event-handler attribute '$attrib'='$value' in uploaded file." );
1673
1674				return [ 'uploaded-event-handler-on-svg', $attrib, $value ];
1675			}
1676
1677			# Do not allow relative links, or unsafe url schemas.
1678			# For <a> tags, only data:, http: and https: and same-document
1679			# fragment links are allowed. For all other tags, only data:
1680			# and fragment are allowed.
1681			if ( $stripped == 'href'
1682				&& $value !== ''
1683				&& strpos( $value, 'data:' ) !== 0
1684				&& strpos( $value, '#' ) !== 0
1685			) {
1686				if ( !( $strippedElement === 'a'
1687					&& preg_match( '!^https?://!i', $value ) )
1688				) {
1689					wfDebug( __METHOD__ . ": Found href attribute <$strippedElement "
1690						. "'$attrib'='$value' in uploaded file." );
1691
1692					return [ 'uploaded-href-attribute-svg', $strippedElement, $attrib, $value ];
1693				}
1694			}
1695
1696			# only allow data: targets that should be safe. This prevents vectors like,
1697			# image/svg, text/xml, application/xml, and text/html, which can contain scripts
1698			if ( $stripped == 'href' && strncasecmp( 'data:', $value, 5 ) === 0 ) {
1699				// rfc2397 parameters. This is only slightly slower than (;[\w;]+)*.
1700				// phpcs:ignore Generic.Files.LineLength
1701				$parameters = '(?>;[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+=(?>[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+|"(?>[\0-\x0c\x0e-\x21\x23-\x5b\x5d-\x7f]+|\\\\[\0-\x7f])*"))*(?:;base64)?';
1702
1703				if ( !preg_match( "!^data:\s*image/(gif|jpeg|jpg|png)$parameters,!i", $value ) ) {
1704					wfDebug( __METHOD__ . ": Found href to unwhitelisted data: uri "
1705						. "\"<$strippedElement '$attrib'='$value'...\" in uploaded file." );
1706					return [ 'uploaded-href-unsafe-target-svg', $strippedElement, $attrib, $value ];
1707				}
1708			}
1709
1710			# Change href with animate from (http://html5sec.org/#137).
1711			if ( $stripped === 'attributename'
1712				&& $strippedElement === 'animate'
1713				&& $this->stripXmlNamespace( $value ) == 'href'
1714			) {
1715				wfDebug( __METHOD__ . ": Found animate that might be changing href using from "
1716					. "\"<$strippedElement '$attrib'='$value'...\" in uploaded file." );
1717
1718				return [ 'uploaded-animate-svg', $strippedElement, $attrib, $value ];
1719			}
1720
1721			# use set/animate to add event-handler attribute to parent
1722			if ( ( $strippedElement == 'set' || $strippedElement == 'animate' )
1723				&& $stripped == 'attributename'
1724				&& substr( $value, 0, 2 ) == 'on'
1725			) {
1726				wfDebug( __METHOD__ . ": Found svg setting event-handler attribute with "
1727					. "\"<$strippedElement $stripped='$value'...\" in uploaded file." );
1728
1729				return [ 'uploaded-setting-event-handler-svg', $strippedElement, $stripped, $value ];
1730			}
1731
1732			# use set to add href attribute to parent element
1733			if ( $strippedElement == 'set'
1734				&& $stripped == 'attributename'
1735				&& strpos( $value, 'href' ) !== false
1736			) {
1737				wfDebug( __METHOD__ . ": Found svg setting href attribute '$value' in uploaded file." );
1738
1739				return [ 'uploaded-setting-href-svg' ];
1740			}
1741
1742			# use set to add a remote / data / script target to an element
1743			if ( $strippedElement == 'set'
1744				&& $stripped == 'to'
1745				&& preg_match( '!(http|https|data|script):!sim', $value )
1746			) {
1747				wfDebug( __METHOD__ . ": Found svg setting attribute to '$value' in uploaded file." );
1748
1749				return [ 'uploaded-wrong-setting-svg', $value ];
1750			}
1751
1752			# use handler attribute with remote / data / script
1753			if ( $stripped == 'handler' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1754				wfDebug( __METHOD__ . ": Found svg setting handler with remote/data/script "
1755					. "'$attrib'='$value' in uploaded file." );
1756
1757				return [ 'uploaded-setting-handler-svg', $attrib, $value ];
1758			}
1759
1760			# use CSS styles to bring in remote code
1761			if ( $stripped == 'style'
1762				&& self::checkCssFragment( Sanitizer::normalizeCss( $value ) )
1763			) {
1764				wfDebug( __METHOD__ . ": Found svg setting a style with "
1765					. "remote url '$attrib'='$value' in uploaded file." );
1766				return [ 'uploaded-remote-url-svg', $attrib, $value ];
1767			}
1768
1769			# Several attributes can include css, css character escaping isn't allowed
1770			$cssAttrs = [ 'font', 'clip-path', 'fill', 'filter', 'marker',
1771				'marker-end', 'marker-mid', 'marker-start', 'mask', 'stroke' ];
1772			if ( in_array( $stripped, $cssAttrs )
1773				&& self::checkCssFragment( $value )
1774			) {
1775				wfDebug( __METHOD__ . ": Found svg setting a style with "
1776					. "remote url '$attrib'='$value' in uploaded file." );
1777				return [ 'uploaded-remote-url-svg', $attrib, $value ];
1778			}
1779
1780			# image filters can pull in url, which could be svg that executes scripts
1781			# Only allow url( "#foo" ). Do not allow url( http://example.com )
1782			if ( $strippedElement == 'image'
1783				&& $stripped == 'filter'
1784				&& preg_match( '!url\s*\(\s*["\']?[^#]!sim', $value )
1785			) {
1786				wfDebug( __METHOD__ . ": Found image filter with url: "
1787					. "\"<$strippedElement $stripped='$value'...\" in uploaded file." );
1788
1789				return [ 'uploaded-image-filter-svg', $strippedElement, $stripped, $value ];
1790			}
1791		}
1792
1793		return false; // No scripts detected
1794	}
1795
1796	/**
1797	 * Check a block of CSS or CSS fragment for anything that looks like
1798	 * it is bringing in remote code.
1799	 * @param string $value a string of CSS
1800	 * @return bool true if the CSS contains an illegal string, false if otherwise
1801	 */
1802	private static function checkCssFragment( $value ) {
1803		# Forbid external stylesheets, for both reliability and to protect viewer's privacy
1804		if ( stripos( $value, '@import' ) !== false ) {
1805			return true;
1806		}
1807
1808		# We allow @font-face to embed fonts with data: urls, so we snip the string
1809		# 'url' out so this case won't match when we check for urls below
1810		$pattern = '!(@font-face\s*{[^}]*src:)url(\("data:;base64,)!im';
1811		$value = preg_replace( $pattern, '$1$2', $value );
1812
1813		# Check for remote and executable CSS. Unlike in Sanitizer::checkCss, the CSS
1814		# properties filter and accelerator don't seem to be useful for xss in SVG files.
1815		# Expression and -o-link don't seem to work either, but filtering them here in case.
1816		# Additionally, we catch remote urls like url("http:..., url('http:..., url(http:...,
1817		# but not local ones such as url("#..., url('#..., url(#....
1818		if ( preg_match( '!expression
1819				| -o-link\s*:
1820				| -o-link-source\s*:
1821				| -o-replace\s*:!imx', $value ) ) {
1822			return true;
1823		}
1824
1825		if ( preg_match_all(
1826				"!(\s*(url|image|image-set)\s*\(\s*[\"']?\s*[^#]+.*?\))!sim",
1827				$value,
1828				$matches
1829			) !== 0
1830		) {
1831			# TODO: redo this in one regex. Until then, url("#whatever") matches the first
1832			foreach ( $matches[1] as $match ) {
1833				if ( !preg_match( "!\s*(url|image|image-set)\s*\(\s*(#|'#|\"#)!im", $match ) ) {
1834					return true;
1835				}
1836			}
1837		}
1838
1839		if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ) {
1840			return true;
1841		}
1842
1843		return false;
1844	}
1845
1846	/**
1847	 * Divide the element name passed by the xml parser to the callback into URI and prifix.
1848	 * @param string $element
1849	 * @return array Containing the namespace URI and prefix
1850	 */
1851	private static function splitXmlNamespace( $element ) {
1852		// 'http://www.w3.org/2000/svg:script' -> [ 'http://www.w3.org/2000/svg', 'script' ]
1853		$parts = explode( ':', strtolower( $element ) );
1854		$name = array_pop( $parts );
1855		$ns = implode( ':', $parts );
1856
1857		return [ $ns, $name ];
1858	}
1859
1860	/**
1861	 * @param string $name
1862	 * @return string
1863	 */
1864	private function stripXmlNamespace( $name ) {
1865		// 'http://www.w3.org/2000/svg:script' -> 'script'
1866		$parts = explode( ':', strtolower( $name ) );
1867
1868		return array_pop( $parts );
1869	}
1870
1871	/**
1872	 * Generic wrapper function for a virus scanner program.
1873	 * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
1874	 * $wgAntivirusRequired may be used to deny upload if the scan fails.
1875	 *
1876	 * @param string $file Pathname to the temporary upload file
1877	 * @return bool|null|string False if not virus is found, null if the scan fails or is disabled,
1878	 *   or a string containing feedback from the virus scanner if a virus was found.
1879	 *   If textual feedback is missing but a virus was found, this function returns true.
1880	 */
1881	public static function detectVirus( $file ) {
1882		global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
1883
1884		if ( !$wgAntivirus ) {
1885			wfDebug( __METHOD__ . ": virus scanner disabled" );
1886
1887			return null;
1888		}
1889
1890		if ( !$wgAntivirusSetup[$wgAntivirus] ) {
1891			wfDebug( __METHOD__ . ": unknown virus scanner: $wgAntivirus" );
1892			$wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
1893				[ 'virus-badscanner', $wgAntivirus ] );
1894
1895			return wfMessage( 'virus-unknownscanner' )->text() . " $wgAntivirus";
1896		}
1897
1898		# look up scanner configuration
1899		$command = $wgAntivirusSetup[$wgAntivirus]['command'];
1900		$exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
1901		$msgPattern = $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ?? null;
1902
1903		if ( strpos( $command, "%f" ) === false ) {
1904			# simple pattern: append file to scan
1905			$command .= " " . Shell::escape( $file );
1906		} else {
1907			# complex pattern: replace "%f" with file to scan
1908			$command = str_replace( "%f", Shell::escape( $file ), $command );
1909		}
1910
1911		wfDebug( __METHOD__ . ": running virus scan: $command " );
1912
1913		# execute virus scanner
1914		$exitCode = false;
1915
1916		# NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1917		#      that does not seem to be worth the pain.
1918		#      Ask me (Duesentrieb) about it if it's ever needed.
1919		$output = wfShellExecWithStderr( $command, $exitCode );
1920
1921		# map exit code to AV_xxx constants.
1922		$mappedCode = $exitCode;
1923		if ( $exitCodeMap ) {
1924			if ( isset( $exitCodeMap[$exitCode] ) ) {
1925				$mappedCode = $exitCodeMap[$exitCode];
1926			} elseif ( isset( $exitCodeMap["*"] ) ) {
1927				$mappedCode = $exitCodeMap["*"];
1928			}
1929		}
1930
1931		/* NB: AV_NO_VIRUS is 0 but AV_SCAN_FAILED is false,
1932		 * so we need the strict equalities === and thus can't use a switch here
1933		 */
1934		if ( $mappedCode === AV_SCAN_FAILED ) {
1935			# scan failed (code was mapped to false by $exitCodeMap)
1936			wfDebug( __METHOD__ . ": failed to scan $file (code $exitCode)." );
1937
1938			$output = $wgAntivirusRequired
1939				? wfMessage( 'virus-scanfailed', [ $exitCode ] )->text()
1940				: null;
1941		} elseif ( $mappedCode === AV_SCAN_ABORTED ) {
1942			# scan failed because filetype is unknown (probably imune)
1943			wfDebug( __METHOD__ . ": unsupported file type $file (code $exitCode)." );
1944			$output = null;
1945		} elseif ( $mappedCode === AV_NO_VIRUS ) {
1946			# no virus found
1947			wfDebug( __METHOD__ . ": file passed virus scan." );
1948			$output = false;
1949		} else {
1950			$output = trim( $output );
1951
1952			if ( !$output ) {
1953				$output = true; # if there's no output, return true
1954			} elseif ( $msgPattern ) {
1955				$groups = [];
1956				if ( preg_match( $msgPattern, $output, $groups ) && $groups[1] ) {
1957					$output = $groups[1];
1958				}
1959			}
1960
1961			wfDebug( __METHOD__ . ": FOUND VIRUS! scanner feedback: $output" );
1962		}
1963
1964		return $output;
1965	}
1966
1967	/**
1968	 * Check if there's an overwrite conflict and, if so, if restrictions
1969	 * forbid this user from performing the upload.
1970	 *
1971	 * @param Authority $performer
1972	 *
1973	 * @return bool|array
1974	 */
1975	private function checkOverwrite( Authority $performer ) {
1976		// First check whether the local file can be overwritten
1977		$file = $this->getLocalFile();
1978		$file->load( File::READ_LATEST );
1979		if ( $file->exists() ) {
1980			if ( !self::userCanReUpload( $performer, $file ) ) {
1981				return [ 'fileexists-forbidden', $file->getName() ];
1982			} else {
1983				return true;
1984			}
1985		}
1986
1987		$services = MediaWikiServices::getInstance();
1988
1989		/* Check shared conflicts: if the local file does not exist, but
1990		 * RepoGroup::findFile finds a file, it exists in a shared repository.
1991		 */
1992		$file = $services->getRepoGroup()->findFile( $this->getTitle(), [ 'latest' => true ] );
1993		if ( $file && !$performer->isAllowed( 'reupload-shared' )
1994		) {
1995			return [ 'fileexists-shared-forbidden', $file->getName() ];
1996		}
1997
1998		return true;
1999	}
2000
2001	/**
2002	 * Check if a user is the last uploader
2003	 *
2004	 * @param Authority $performer
2005	 * @param File $img
2006	 * @return bool
2007	 */
2008	public static function userCanReUpload( Authority $performer, File $img ) {
2009		if ( $performer->isAllowed( 'reupload' ) ) {
2010			return true; // non-conditional
2011		} elseif ( !$performer->isAllowed( 'reupload-own' ) ) {
2012			return false;
2013		}
2014
2015		if ( !( $img instanceof LocalFile ) ) {
2016			return false;
2017		}
2018
2019		return $performer->getUser()->equals( $img->getUploader( File::RAW ) );
2020	}
2021
2022	/**
2023	 * Helper function that does various existence checks for a file.
2024	 * The following checks are performed:
2025	 * - The file exists
2026	 * - Article with the same name as the file exists
2027	 * - File exists with normalized extension
2028	 * - The file looks like a thumbnail and the original exists
2029	 *
2030	 * @param File $file The File object to check
2031	 * @return array|bool False if the file does not exist, else an array
2032	 */
2033	public static function getExistsWarning( $file ) {
2034		if ( $file->exists() ) {
2035			return [ 'warning' => 'exists', 'file' => $file ];
2036		}
2037
2038		if ( $file->getTitle()->getArticleID() ) {
2039			return [ 'warning' => 'page-exists', 'file' => $file ];
2040		}
2041
2042		if ( !strpos( $file->getName(), '.' ) ) {
2043			$partname = $file->getName();
2044			$extension = '';
2045		} else {
2046			$n = strrpos( $file->getName(), '.' );
2047			$extension = substr( $file->getName(), $n + 1 );
2048			$partname = substr( $file->getName(), 0, $n );
2049		}
2050		$normalizedExtension = File::normalizeExtension( $extension );
2051		$localRepo = MediaWikiServices::getInstance()->getRepoGroup()->getLocalRepo();
2052
2053		if ( $normalizedExtension != $extension ) {
2054			// We're not using the normalized form of the extension.
2055			// Normal form is lowercase, using most common of alternate
2056			// extensions (eg 'jpg' rather than 'JPEG').
2057
2058			// Check for another file using the normalized form...
2059			$nt_lc = Title::makeTitle( NS_FILE, "{$partname}.{$normalizedExtension}" );
2060			$file_lc = $localRepo->newFile( $nt_lc );
2061
2062			if ( $file_lc->exists() ) {
2063				return [
2064					'warning' => 'exists-normalized',
2065					'file' => $file,
2066					'normalizedFile' => $file_lc
2067				];
2068			}
2069		}
2070
2071		// Check for files with the same name but a different extension
2072		$similarFiles = $localRepo->findFilesByPrefix( "{$partname}.", 1 );
2073		if ( count( $similarFiles ) ) {
2074			return [
2075				'warning' => 'exists-normalized',
2076				'file' => $file,
2077				'normalizedFile' => $similarFiles[0],
2078			];
2079		}
2080
2081		if ( self::isThumbName( $file->getName() ) ) {
2082			# Check for filenames like 50px- or 180px-, these are mostly thumbnails
2083			$nt_thb = Title::newFromText(
2084				substr( $partname, strpos( $partname, '-' ) + 1 ) . '.' . $extension,
2085				NS_FILE
2086			);
2087			$file_thb = $localRepo->newFile( $nt_thb );
2088			if ( $file_thb->exists() ) {
2089				return [
2090					'warning' => 'thumb',
2091					'file' => $file,
2092					'thumbFile' => $file_thb
2093				];
2094			} else {
2095				// File does not exist, but we just don't like the name
2096				return [
2097					'warning' => 'thumb-name',
2098					'file' => $file,
2099					'thumbFile' => $file_thb
2100				];
2101			}
2102		}
2103
2104		foreach ( self::getFilenamePrefixBlacklist() as $prefix ) {
2105			if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
2106				return [
2107					'warning' => 'bad-prefix',
2108					'file' => $file,
2109					'prefix' => $prefix
2110				];
2111			}
2112		}
2113
2114		return false;
2115	}
2116
2117	/**
2118	 * Helper function that checks whether the filename looks like a thumbnail
2119	 * @param string $filename
2120	 * @return bool
2121	 */
2122	public static function isThumbName( $filename ) {
2123		$n = strrpos( $filename, '.' );
2124		$partname = $n ? substr( $filename, 0, $n ) : $filename;
2125
2126		return (
2127				substr( $partname, 3, 3 ) == 'px-' ||
2128				substr( $partname, 2, 3 ) == 'px-'
2129			) &&
2130			preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) );
2131	}
2132
2133	/**
2134	 * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
2135	 *
2136	 * @return string[] List of prefixes
2137	 */
2138	public static function getFilenamePrefixBlacklist() {
2139		$list = [];
2140		$message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
2141		if ( !$message->isDisabled() ) {
2142			$lines = explode( "\n", $message->plain() );
2143			foreach ( $lines as $line ) {
2144				// Remove comment lines
2145				$comment = substr( trim( $line ), 0, 1 );
2146				if ( $comment == '#' || $comment == '' ) {
2147					continue;
2148				}
2149				// Remove additional comments after a prefix
2150				$comment = strpos( $line, '#' );
2151				if ( $comment > 0 ) {
2152					$line = substr( $line, 0, $comment - 1 );
2153				}
2154				$list[] = trim( $line );
2155			}
2156		}
2157
2158		return $list;
2159	}
2160
2161	/**
2162	 * Gets image info about the file just uploaded.
2163	 *
2164	 * Also has the effect of setting metadata to be an 'indexed tag name' in
2165	 * returned API result if 'metadata' was requested. Oddly, we have to pass
2166	 * the "result" object down just so it can do that with the appropriate
2167	 * format, presumably.
2168	 *
2169	 * @param ApiResult $result
2170	 * @return array Image info
2171	 */
2172	public function getImageInfo( $result ) {
2173		$localFile = $this->getLocalFile();
2174		$stashFile = $this->getStashFile();
2175		// Calling a different API module depending on whether the file was stashed is less than optimal.
2176		// In fact, calling API modules here at all is less than optimal. Maybe it should be refactored.
2177		if ( $stashFile ) {
2178			$imParam = ApiQueryStashImageInfo::getPropertyNames();
2179			$info = ApiQueryStashImageInfo::getInfo( $stashFile, array_fill_keys( $imParam, true ), $result );
2180		} else {
2181			$imParam = ApiQueryImageInfo::getPropertyNames();
2182			$info = ApiQueryImageInfo::getInfo( $localFile, array_fill_keys( $imParam, true ), $result );
2183		}
2184
2185		return $info;
2186	}
2187
2188	/**
2189	 * @param array $error
2190	 * @return Status
2191	 */
2192	public function convertVerifyErrorToStatus( $error ) {
2193		$code = $error['status'];
2194		unset( $code['status'] );
2195
2196		return Status::newFatal( $this->getVerificationErrorCode( $code ), $error );
2197	}
2198
2199	/**
2200	 * Get MediaWiki's maximum uploaded file size for given type of upload, based on
2201	 * $wgMaxUploadSize.
2202	 *
2203	 * @param null|string $forType
2204	 * @return int
2205	 */
2206	public static function getMaxUploadSize( $forType = null ) {
2207		global $wgMaxUploadSize;
2208
2209		if ( is_array( $wgMaxUploadSize ) ) {
2210			if ( $forType !== null && isset( $wgMaxUploadSize[$forType] ) ) {
2211				return $wgMaxUploadSize[$forType];
2212			} else {
2213				return $wgMaxUploadSize['*'];
2214			}
2215		} else {
2216			return intval( $wgMaxUploadSize );
2217		}
2218	}
2219
2220	/**
2221	 * Get the PHP maximum uploaded file size, based on ini settings. If there is no limit or the
2222	 * limit can't be guessed, returns a very large number (PHP_INT_MAX).
2223	 *
2224	 * @since 1.27
2225	 * @return int
2226	 */
2227	public static function getMaxPhpUploadSize() {
2228		$phpMaxFileSize = wfShorthandToInteger(
2229			ini_get( 'upload_max_filesize' ),
2230			PHP_INT_MAX
2231		);
2232		$phpMaxPostSize = wfShorthandToInteger(
2233			ini_get( 'post_max_size' ),
2234			PHP_INT_MAX
2235		) ?: PHP_INT_MAX;
2236		return min( $phpMaxFileSize, $phpMaxPostSize );
2237	}
2238
2239	/**
2240	 * Get the current status of a chunked upload (used for polling)
2241	 *
2242	 * The value will be read from cache.
2243	 *
2244	 * @param UserIdentity $user
2245	 * @param string $statusKey
2246	 * @return Status[]|bool
2247	 */
2248	public static function getSessionStatus( UserIdentity $user, $statusKey ) {
2249		$store = self::getUploadSessionStore();
2250		$key = self::getUploadSessionKey( $store, $user, $statusKey );
2251
2252		return $store->get( $key );
2253	}
2254
2255	/**
2256	 * Set the current status of a chunked upload (used for polling)
2257	 *
2258	 * The value will be set in cache for 1 day
2259	 *
2260	 * Avoid triggering this method on HTTP GET/HEAD requests
2261	 *
2262	 * @param UserIdentity $user
2263	 * @param string $statusKey
2264	 * @param array|bool $value
2265	 * @return void
2266	 */
2267	public static function setSessionStatus( UserIdentity $user, $statusKey, $value ) {
2268		$store = self::getUploadSessionStore();
2269		$key = self::getUploadSessionKey( $store, $user, $statusKey );
2270
2271		if ( $value === false ) {
2272			$store->delete( $key );
2273		} else {
2274			$store->set( $key, $value, $store::TTL_DAY );
2275		}
2276	}
2277
2278	/**
2279	 * @param BagOStuff $store
2280	 * @param UserIdentity $user
2281	 * @param string $statusKey
2282	 * @return string
2283	 */
2284	private static function getUploadSessionKey( BagOStuff $store, UserIdentity $user, $statusKey ) {
2285		return $store->makeKey(
2286			'uploadstatus',
2287			$user->getId() ?: md5( $user->getName() ),
2288			$statusKey
2289		);
2290	}
2291
2292	/**
2293	 * @return BagOStuff
2294	 */
2295	private static function getUploadSessionStore() {
2296		return ObjectCache::getInstance( 'db-replicated' );
2297	}
2298}
2299