Merge "maintenance: Script to rename titles for Unicode uppercasing changes"
[lhc/web/wiklou.git] / includes / filerepo / FileRepo.php
1 <?php
2 /**
3 * @defgroup FileRepo File Repository
4 *
5 * @brief This module handles how MediaWiki interacts with filesystems.
6 *
7 * @details
8 */
9
10 use MediaWiki\MediaWikiServices;
11
12 /**
13 * Base code for file repositories.
14 *
15 * This program is free software; you can redistribute it and/or modify
16 * it under the terms of the GNU General Public License as published by
17 * the Free Software Foundation; either version 2 of the License, or
18 * (at your option) any later version.
19 *
20 * This program is distributed in the hope that it will be useful,
21 * but WITHOUT ANY WARRANTY; without even the implied warranty of
22 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
23 * GNU General Public License for more details.
24 *
25 * You should have received a copy of the GNU General Public License along
26 * with this program; if not, write to the Free Software Foundation, Inc.,
27 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
28 * http://www.gnu.org/copyleft/gpl.html
29 *
30 * @file
31 * @ingroup FileRepo
32 */
33
34 /**
35 * Base class for file repositories
36 *
37 * @ingroup FileRepo
38 */
39 class FileRepo {
40 const DELETE_SOURCE = 1;
41 const OVERWRITE = 2;
42 const OVERWRITE_SAME = 4;
43 const SKIP_LOCKING = 8;
44
45 const NAME_AND_TIME_ONLY = 1;
46
47 /** @var bool Whether to fetch commons image description pages and display
48 * them on the local wiki */
49 public $fetchDescription;
50
51 /** @var int */
52 public $descriptionCacheExpiry;
53
54 /** @var bool */
55 protected $hasSha1Storage = false;
56
57 /** @var bool */
58 protected $supportsSha1URLs = false;
59
60 /** @var FileBackend */
61 protected $backend;
62
63 /** @var array Map of zones to config */
64 protected $zones = [];
65
66 /** @var string URL of thumb.php */
67 protected $thumbScriptUrl;
68
69 /** @var bool Whether to skip media file transformation on parse and rely
70 * on a 404 handler instead. */
71 protected $transformVia404;
72
73 /** @var string URL of image description pages, e.g.
74 * https://en.wikipedia.org/wiki/File:
75 */
76 protected $descBaseUrl;
77
78 /** @var string URL of the MediaWiki installation, equivalent to
79 * $wgScriptPath, e.g. https://en.wikipedia.org/w
80 */
81 protected $scriptDirUrl;
82
83 /** @var string Equivalent to $wgArticlePath, e.g. https://en.wikipedia.org/wiki/$1 */
84 protected $articleUrl;
85
86 /** @var bool Equivalent to $wgCapitalLinks (or $wgCapitalLinkOverrides[NS_FILE],
87 * determines whether filenames implicitly start with a capital letter.
88 * The current implementation may give incorrect description page links
89 * when the local $wgCapitalLinks and initialCapital are mismatched.
90 */
91 protected $initialCapital;
92
93 /** @var string May be 'paranoid' to remove all parameters from error
94 * messages, 'none' to leave the paths in unchanged, or 'simple' to
95 * replace paths with placeholders. Default for LocalRepo is
96 * 'simple'.
97 */
98 protected $pathDisclosureProtection = 'simple';
99
100 /** @var string|false Public zone URL. */
101 protected $url;
102
103 /** @var string The base thumbnail URL. Defaults to "<url>/thumb". */
104 protected $thumbUrl;
105
106 /** @var int The number of directory levels for hash-based division of files */
107 protected $hashLevels;
108
109 /** @var int The number of directory levels for hash-based division of deleted files */
110 protected $deletedHashLevels;
111
112 /** @var int File names over this size will use the short form of thumbnail
113 * names. Short thumbnail names only have the width, parameters, and the
114 * extension.
115 */
116 protected $abbrvThreshold;
117
118 /** @var string The URL of the repo's favicon, if any */
119 protected $favicon;
120
121 /** @var bool Whether all zones should be private (e.g. private wiki repo) */
122 protected $isPrivate;
123
124 /** @var callable Override these in the base class */
125 protected $fileFactory = [ UnregisteredLocalFile::class, 'newFromTitle' ];
126 /** @var callable|false Override these in the base class */
127 protected $oldFileFactory = false;
128 /** @var callable|false Override these in the base class */
129 protected $fileFactoryKey = false;
130 /** @var callable|false Override these in the base class */
131 protected $oldFileFactoryKey = false;
132
133 /** @var string URL of where to proxy thumb.php requests to.
134 * Example: http://127.0.0.1:8888/wiki/dev/thumb/
135 */
136 protected $thumbProxyUrl;
137 /** @var string Secret key to pass as an X-Swift-Secret header to the proxied thumb service */
138 protected $thumbProxySecret;
139
140 /** @var WANObjectCache */
141 protected $wanCache;
142
143 /**
144 * @param array|null $info
145 * @throws MWException
146 */
147 public function __construct( array $info = null ) {
148 // Verify required settings presence
149 if (
150 $info === null
151 || !array_key_exists( 'name', $info )
152 || !array_key_exists( 'backend', $info )
153 ) {
154 throw new MWException( __CLASS__ .
155 " requires an array of options having both 'name' and 'backend' keys.\n" );
156 }
157
158 // Required settings
159 $this->name = $info['name'];
160 if ( $info['backend'] instanceof FileBackend ) {
161 $this->backend = $info['backend']; // useful for testing
162 } else {
163 $this->backend = FileBackendGroup::singleton()->get( $info['backend'] );
164 }
165
166 // Optional settings that can have no value
167 $optionalSettings = [
168 'descBaseUrl', 'scriptDirUrl', 'articleUrl', 'fetchDescription',
169 'thumbScriptUrl', 'pathDisclosureProtection', 'descriptionCacheExpiry',
170 'favicon', 'thumbProxyUrl', 'thumbProxySecret',
171 ];
172 foreach ( $optionalSettings as $var ) {
173 if ( isset( $info[$var] ) ) {
174 $this->$var = $info[$var];
175 }
176 }
177
178 // Optional settings that have a default
179 $this->initialCapital = $info['initialCapital'] ??
180 MediaWikiServices::getInstance()->getNamespaceInfo()->isCapitalized( NS_FILE );
181 $this->url = $info['url'] ?? false; // a subclass may set the URL (e.g. ForeignAPIRepo)
182 if ( isset( $info['thumbUrl'] ) ) {
183 $this->thumbUrl = $info['thumbUrl'];
184 } else {
185 $this->thumbUrl = $this->url ? "{$this->url}/thumb" : false;
186 }
187 $this->hashLevels = $info['hashLevels'] ?? 2;
188 $this->deletedHashLevels = $info['deletedHashLevels'] ?? $this->hashLevels;
189 $this->transformVia404 = !empty( $info['transformVia404'] );
190 $this->abbrvThreshold = $info['abbrvThreshold'] ?? 255;
191 $this->isPrivate = !empty( $info['isPrivate'] );
192 // Give defaults for the basic zones...
193 $this->zones = $info['zones'] ?? [];
194 foreach ( [ 'public', 'thumb', 'transcoded', 'temp', 'deleted' ] as $zone ) {
195 if ( !isset( $this->zones[$zone]['container'] ) ) {
196 $this->zones[$zone]['container'] = "{$this->name}-{$zone}";
197 }
198 if ( !isset( $this->zones[$zone]['directory'] ) ) {
199 $this->zones[$zone]['directory'] = '';
200 }
201 if ( !isset( $this->zones[$zone]['urlsByExt'] ) ) {
202 $this->zones[$zone]['urlsByExt'] = [];
203 }
204 }
205
206 $this->supportsSha1URLs = !empty( $info['supportsSha1URLs'] );
207
208 $this->wanCache = $info['wanCache'] ?? WANObjectCache::newEmpty();
209 }
210
211 /**
212 * Get the file backend instance. Use this function wisely.
213 *
214 * @return FileBackend
215 */
216 public function getBackend() {
217 return $this->backend;
218 }
219
220 /**
221 * Get an explanatory message if this repo is read-only.
222 * This checks if an administrator disabled writes to the backend.
223 *
224 * @return string|bool Returns false if the repo is not read-only
225 */
226 public function getReadOnlyReason() {
227 return $this->backend->getReadOnlyReason();
228 }
229
230 /**
231 * Check if a single zone or list of zones is defined for usage
232 *
233 * @param string[]|string $doZones Only do a particular zones
234 * @throws MWException
235 * @return Status
236 */
237 protected function initZones( $doZones = [] ) {
238 $status = $this->newGood();
239 foreach ( (array)$doZones as $zone ) {
240 $root = $this->getZonePath( $zone );
241 if ( $root === null ) {
242 throw new MWException( "No '$zone' zone defined in the {$this->name} repo." );
243 }
244 }
245
246 return $status;
247 }
248
249 /**
250 * Determine if a string is an mwrepo:// URL
251 *
252 * @param string $url
253 * @return bool
254 */
255 public static function isVirtualUrl( $url ) {
256 return substr( $url, 0, 9 ) == 'mwrepo://';
257 }
258
259 /**
260 * Get a URL referring to this repository, with the private mwrepo protocol.
261 * The suffix, if supplied, is considered to be unencoded, and will be
262 * URL-encoded before being returned.
263 *
264 * @param string|bool $suffix
265 * @return string
266 */
267 public function getVirtualUrl( $suffix = false ) {
268 $path = 'mwrepo://' . $this->name;
269 if ( $suffix !== false ) {
270 $path .= '/' . rawurlencode( $suffix );
271 }
272
273 return $path;
274 }
275
276 /**
277 * Get the URL corresponding to one of the four basic zones
278 *
279 * @param string $zone One of: public, deleted, temp, thumb
280 * @param string|null $ext Optional file extension
281 * @return string|bool
282 */
283 public function getZoneUrl( $zone, $ext = null ) {
284 if ( in_array( $zone, [ 'public', 'thumb', 'transcoded' ] ) ) {
285 // standard public zones
286 if ( $ext !== null && isset( $this->zones[$zone]['urlsByExt'][$ext] ) ) {
287 // custom URL for extension/zone
288 return $this->zones[$zone]['urlsByExt'][$ext];
289 } elseif ( isset( $this->zones[$zone]['url'] ) ) {
290 // custom URL for zone
291 return $this->zones[$zone]['url'];
292 }
293 }
294 switch ( $zone ) {
295 case 'public':
296 return $this->url;
297 case 'temp':
298 case 'deleted':
299 return false; // no public URL
300 case 'thumb':
301 return $this->thumbUrl;
302 case 'transcoded':
303 return "{$this->url}/transcoded";
304 default:
305 return false;
306 }
307 }
308
309 /**
310 * @return bool Whether non-ASCII path characters are allowed
311 */
312 public function backendSupportsUnicodePaths() {
313 return (bool)( $this->getBackend()->getFeatures() & FileBackend::ATTR_UNICODE_PATHS );
314 }
315
316 /**
317 * Get the backend storage path corresponding to a virtual URL.
318 * Use this function wisely.
319 *
320 * @param string $url
321 * @throws MWException
322 * @return string
323 */
324 public function resolveVirtualUrl( $url ) {
325 if ( substr( $url, 0, 9 ) != 'mwrepo://' ) {
326 throw new MWException( __METHOD__ . ': unknown protocol' );
327 }
328 $bits = explode( '/', substr( $url, 9 ), 3 );
329 if ( count( $bits ) != 3 ) {
330 throw new MWException( __METHOD__ . ": invalid mwrepo URL: $url" );
331 }
332 list( $repo, $zone, $rel ) = $bits;
333 if ( $repo !== $this->name ) {
334 throw new MWException( __METHOD__ . ": fetching from a foreign repo is not supported" );
335 }
336 $base = $this->getZonePath( $zone );
337 if ( !$base ) {
338 throw new MWException( __METHOD__ . ": invalid zone: $zone" );
339 }
340
341 return $base . '/' . rawurldecode( $rel );
342 }
343
344 /**
345 * The the storage container and base path of a zone
346 *
347 * @param string $zone
348 * @return array (container, base path) or (null, null)
349 */
350 protected function getZoneLocation( $zone ) {
351 if ( !isset( $this->zones[$zone] ) ) {
352 return [ null, null ]; // bogus
353 }
354
355 return [ $this->zones[$zone]['container'], $this->zones[$zone]['directory'] ];
356 }
357
358 /**
359 * Get the storage path corresponding to one of the zones
360 *
361 * @param string $zone
362 * @return string|null Returns null if the zone is not defined
363 */
364 public function getZonePath( $zone ) {
365 list( $container, $base ) = $this->getZoneLocation( $zone );
366 if ( $container === null || $base === null ) {
367 return null;
368 }
369 $backendName = $this->backend->getName();
370 if ( $base != '' ) { // may not be set
371 $base = "/{$base}";
372 }
373
374 return "mwstore://$backendName/{$container}{$base}";
375 }
376
377 /**
378 * Create a new File object from the local repository
379 *
380 * @param Title|string $title Title object or string
381 * @param bool|string $time Time at which the image was uploaded. If this
382 * is specified, the returned object will be an instance of the
383 * repository's old file class instead of a current file. Repositories
384 * not supporting version control should return false if this parameter
385 * is set.
386 * @return File|null A File, or null if passed an invalid Title
387 */
388 public function newFile( $title, $time = false ) {
389 $title = File::normalizeTitle( $title );
390 if ( !$title ) {
391 return null;
392 }
393 if ( $time ) {
394 if ( $this->oldFileFactory ) {
395 return call_user_func( $this->oldFileFactory, $title, $this, $time );
396 } else {
397 return null;
398 }
399 } else {
400 return call_user_func( $this->fileFactory, $title, $this );
401 }
402 }
403
404 /**
405 * Find an instance of the named file created at the specified time
406 * Returns false if the file does not exist. Repositories not supporting
407 * version control should return false if the time is specified.
408 *
409 * @param Title|string $title Title object or string
410 * @param array $options Associative array of options:
411 * time: requested time for a specific file version, or false for the
412 * current version. An image object will be returned which was
413 * created at the specified time (which may be archived or current).
414 * ignoreRedirect: If true, do not follow file redirects
415 * private: If true, return restricted (deleted) files if the current
416 * user is allowed to view them. Otherwise, such files will not
417 * be found. If a User object, use that user instead of the current.
418 * latest: If true, load from the latest available data into File objects
419 * @return File|bool False on failure
420 */
421 public function findFile( $title, $options = [] ) {
422 $title = File::normalizeTitle( $title );
423 if ( !$title ) {
424 return false;
425 }
426 if ( isset( $options['bypassCache'] ) ) {
427 $options['latest'] = $options['bypassCache']; // b/c
428 }
429 $time = $options['time'] ?? false;
430 $flags = !empty( $options['latest'] ) ? File::READ_LATEST : 0;
431 # First try the current version of the file to see if it precedes the timestamp
432 $img = $this->newFile( $title );
433 if ( !$img ) {
434 return false;
435 }
436 $img->load( $flags );
437 if ( $img->exists() && ( !$time || $img->getTimestamp() == $time ) ) {
438 return $img;
439 }
440 # Now try an old version of the file
441 if ( $time !== false ) {
442 $img = $this->newFile( $title, $time );
443 if ( $img ) {
444 $img->load( $flags );
445 if ( $img->exists() ) {
446 if ( !$img->isDeleted( File::DELETED_FILE ) ) {
447 return $img; // always OK
448 } elseif ( !empty( $options['private'] ) &&
449 $img->userCan( File::DELETED_FILE,
450 $options['private'] instanceof User ? $options['private'] : null
451 )
452 ) {
453 return $img;
454 }
455 }
456 }
457 }
458
459 # Now try redirects
460 if ( !empty( $options['ignoreRedirect'] ) ) {
461 return false;
462 }
463 $redir = $this->checkRedirect( $title );
464 if ( $redir && $title->getNamespace() == NS_FILE ) {
465 $img = $this->newFile( $redir );
466 if ( !$img ) {
467 return false;
468 }
469 $img->load( $flags );
470 if ( $img->exists() ) {
471 $img->redirectedFrom( $title->getDBkey() );
472
473 return $img;
474 }
475 }
476
477 return false;
478 }
479
480 /**
481 * Find many files at once.
482 *
483 * @param array $items An array of titles, or an array of findFile() options with
484 * the "title" option giving the title. Example:
485 *
486 * $findItem = [ 'title' => $title, 'private' => true ];
487 * $findBatch = [ $findItem ];
488 * $repo->findFiles( $findBatch );
489 *
490 * No title should appear in $items twice, as the result use titles as keys
491 * @param int $flags Supports:
492 * - FileRepo::NAME_AND_TIME_ONLY : return a (search title => (title,timestamp)) map.
493 * The search title uses the input titles; the other is the final post-redirect title.
494 * All titles are returned as string DB keys and the inner array is associative.
495 * @return array Map of (file name => File objects) for matches
496 */
497 public function findFiles( array $items, $flags = 0 ) {
498 $result = [];
499 foreach ( $items as $item ) {
500 if ( is_array( $item ) ) {
501 $title = $item['title'];
502 $options = $item;
503 unset( $options['title'] );
504 } else {
505 $title = $item;
506 $options = [];
507 }
508 $file = $this->findFile( $title, $options );
509 if ( $file ) {
510 $searchName = File::normalizeTitle( $title )->getDBkey(); // must be valid
511 if ( $flags & self::NAME_AND_TIME_ONLY ) {
512 $result[$searchName] = [
513 'title' => $file->getTitle()->getDBkey(),
514 'timestamp' => $file->getTimestamp()
515 ];
516 } else {
517 $result[$searchName] = $file;
518 }
519 }
520 }
521
522 return $result;
523 }
524
525 /**
526 * Find an instance of the file with this key, created at the specified time
527 * Returns false if the file does not exist. Repositories not supporting
528 * version control should return false if the time is specified.
529 *
530 * @param string $sha1 Base 36 SHA-1 hash
531 * @param array $options Option array, same as findFile().
532 * @return File|bool False on failure
533 */
534 public function findFileFromKey( $sha1, $options = [] ) {
535 $time = $options['time'] ?? false;
536 # First try to find a matching current version of a file...
537 if ( !$this->fileFactoryKey ) {
538 return false; // find-by-sha1 not supported
539 }
540 $img = call_user_func( $this->fileFactoryKey, $sha1, $this, $time );
541 if ( $img && $img->exists() ) {
542 return $img;
543 }
544 # Now try to find a matching old version of a file...
545 if ( $time !== false && $this->oldFileFactoryKey ) { // find-by-sha1 supported?
546 $img = call_user_func( $this->oldFileFactoryKey, $sha1, $this, $time );
547 if ( $img && $img->exists() ) {
548 if ( !$img->isDeleted( File::DELETED_FILE ) ) {
549 return $img; // always OK
550 } elseif ( !empty( $options['private'] ) &&
551 $img->userCan( File::DELETED_FILE,
552 $options['private'] instanceof User ? $options['private'] : null
553 )
554 ) {
555 return $img;
556 }
557 }
558 }
559
560 return false;
561 }
562
563 /**
564 * Get an array or iterator of file objects for files that have a given
565 * SHA-1 content hash.
566 *
567 * STUB
568 * @param string $hash SHA-1 hash
569 * @return File[]
570 */
571 public function findBySha1( $hash ) {
572 return [];
573 }
574
575 /**
576 * Get an array of arrays or iterators of file objects for files that
577 * have the given SHA-1 content hashes.
578 *
579 * @param string[] $hashes An array of hashes
580 * @return array[] An Array of arrays or iterators of file objects and the hash as key
581 */
582 public function findBySha1s( array $hashes ) {
583 $result = [];
584 foreach ( $hashes as $hash ) {
585 $files = $this->findBySha1( $hash );
586 if ( count( $files ) ) {
587 $result[$hash] = $files;
588 }
589 }
590
591 return $result;
592 }
593
594 /**
595 * Return an array of files where the name starts with $prefix.
596 *
597 * STUB
598 * @param string $prefix The prefix to search for
599 * @param int $limit The maximum amount of files to return
600 * @return LocalFile[]
601 */
602 public function findFilesByPrefix( $prefix, $limit ) {
603 return [];
604 }
605
606 /**
607 * Get the URL of thumb.php
608 *
609 * @return string
610 */
611 public function getThumbScriptUrl() {
612 return $this->thumbScriptUrl;
613 }
614
615 /**
616 * Get the URL thumb.php requests are being proxied to
617 *
618 * @return string
619 */
620 public function getThumbProxyUrl() {
621 return $this->thumbProxyUrl;
622 }
623
624 /**
625 * Get the secret key for the proxied thumb service
626 *
627 * @return string
628 */
629 public function getThumbProxySecret() {
630 return $this->thumbProxySecret;
631 }
632
633 /**
634 * Returns true if the repository can transform files via a 404 handler
635 *
636 * @return bool
637 */
638 public function canTransformVia404() {
639 return $this->transformVia404;
640 }
641
642 /**
643 * Get the name of a file from its title object
644 *
645 * @param Title $title
646 * @return string
647 */
648 public function getNameFromTitle( Title $title ) {
649 if (
650 $this->initialCapital !=
651 MediaWikiServices::getInstance()->getNamespaceInfo()->isCapitalized( NS_FILE )
652 ) {
653 $name = $title->getUserCaseDBKey();
654 if ( $this->initialCapital ) {
655 $name = MediaWikiServices::getInstance()->getContentLanguage()->ucfirst( $name );
656 }
657 } else {
658 $name = $title->getDBkey();
659 }
660
661 return $name;
662 }
663
664 /**
665 * Get the public zone root storage directory of the repository
666 *
667 * @return string
668 */
669 public function getRootDirectory() {
670 return $this->getZonePath( 'public' );
671 }
672
673 /**
674 * Get a relative path including trailing slash, e.g. f/fa/
675 * If the repo is not hashed, returns an empty string
676 *
677 * @param string $name Name of file
678 * @return string
679 */
680 public function getHashPath( $name ) {
681 return self::getHashPathForLevel( $name, $this->hashLevels );
682 }
683
684 /**
685 * Get a relative path including trailing slash, e.g. f/fa/
686 * If the repo is not hashed, returns an empty string
687 *
688 * @param string $suffix Basename of file from FileRepo::storeTemp()
689 * @return string
690 */
691 public function getTempHashPath( $suffix ) {
692 $parts = explode( '!', $suffix, 2 ); // format is <timestamp>!<name> or just <name>
693 $name = $parts[1] ?? $suffix; // hash path is not based on timestamp
694 return self::getHashPathForLevel( $name, $this->hashLevels );
695 }
696
697 /**
698 * @param string $name
699 * @param int $levels
700 * @return string
701 */
702 protected static function getHashPathForLevel( $name, $levels ) {
703 if ( $levels == 0 ) {
704 return '';
705 } else {
706 $hash = md5( $name );
707 $path = '';
708 for ( $i = 1; $i <= $levels; $i++ ) {
709 $path .= substr( $hash, 0, $i ) . '/';
710 }
711
712 return $path;
713 }
714 }
715
716 /**
717 * Get the number of hash directory levels
718 *
719 * @return int
720 */
721 public function getHashLevels() {
722 return $this->hashLevels;
723 }
724
725 /**
726 * Get the name of this repository, as specified by $info['name]' to the constructor
727 *
728 * @return string
729 */
730 public function getName() {
731 return $this->name;
732 }
733
734 /**
735 * Make an url to this repo
736 *
737 * @param string|string[] $query Query string to append
738 * @param string $entry Entry point; defaults to index
739 * @return string|bool False on failure
740 */
741 public function makeUrl( $query = '', $entry = 'index' ) {
742 if ( isset( $this->scriptDirUrl ) ) {
743 return wfAppendQuery( "{$this->scriptDirUrl}/{$entry}.php", $query );
744 }
745
746 return false;
747 }
748
749 /**
750 * Get the URL of an image description page. May return false if it is
751 * unknown or not applicable. In general this should only be called by the
752 * File class, since it may return invalid results for certain kinds of
753 * repositories. Use File::getDescriptionUrl() in user code.
754 *
755 * In particular, it uses the article paths as specified to the repository
756 * constructor, whereas local repositories use the local Title functions.
757 *
758 * @param string $name
759 * @return string|false
760 */
761 public function getDescriptionUrl( $name ) {
762 $encName = wfUrlencode( $name );
763 if ( !is_null( $this->descBaseUrl ) ) {
764 # "http://example.com/wiki/File:"
765 return $this->descBaseUrl . $encName;
766 }
767 if ( !is_null( $this->articleUrl ) ) {
768 # "http://example.com/wiki/$1"
769 # We use "Image:" as the canonical namespace for
770 # compatibility across all MediaWiki versions.
771 return str_replace( '$1',
772 "Image:$encName", $this->articleUrl );
773 }
774 if ( !is_null( $this->scriptDirUrl ) ) {
775 # "http://example.com/w"
776 # We use "Image:" as the canonical namespace for
777 # compatibility across all MediaWiki versions,
778 # and just sort of hope index.php is right. ;)
779 return $this->makeUrl( "title=Image:$encName" );
780 }
781
782 return false;
783 }
784
785 /**
786 * Get the URL of the content-only fragment of the description page. For
787 * MediaWiki this means action=render. This should only be called by the
788 * repository's file class, since it may return invalid results. User code
789 * should use File::getDescriptionText().
790 *
791 * @param string $name Name of image to fetch
792 * @param string|null $lang Language to fetch it in, if any.
793 * @return string|false
794 */
795 public function getDescriptionRenderUrl( $name, $lang = null ) {
796 $query = 'action=render';
797 if ( !is_null( $lang ) ) {
798 $query .= '&uselang=' . urlencode( $lang );
799 }
800 if ( isset( $this->scriptDirUrl ) ) {
801 return $this->makeUrl(
802 'title=' .
803 wfUrlencode( 'Image:' . $name ) .
804 "&$query" );
805 } else {
806 $descUrl = $this->getDescriptionUrl( $name );
807 if ( $descUrl ) {
808 return wfAppendQuery( $descUrl, $query );
809 } else {
810 return false;
811 }
812 }
813 }
814
815 /**
816 * Get the URL of the stylesheet to apply to description pages
817 *
818 * @return string|bool False on failure
819 */
820 public function getDescriptionStylesheetUrl() {
821 if ( isset( $this->scriptDirUrl ) ) {
822 // Must match canonical query parameter order for optimum caching
823 // See Title::getCdnUrls
824 return $this->makeUrl( 'title=MediaWiki:Filepage.css&action=raw&ctype=text/css' );
825 }
826
827 return false;
828 }
829
830 /**
831 * Store a file to a given destination.
832 *
833 * @param string $srcPath Source file system path, storage path, or virtual URL
834 * @param string $dstZone Destination zone
835 * @param string $dstRel Destination relative path
836 * @param int $flags Bitwise combination of the following flags:
837 * self::OVERWRITE Overwrite an existing destination file instead of failing
838 * self::OVERWRITE_SAME Overwrite the file if the destination exists and has the
839 * same contents as the source
840 * self::SKIP_LOCKING Skip any file locking when doing the store
841 * @return Status
842 */
843 public function store( $srcPath, $dstZone, $dstRel, $flags = 0 ) {
844 $this->assertWritableRepo(); // fail out if read-only
845
846 $status = $this->storeBatch( [ [ $srcPath, $dstZone, $dstRel ] ], $flags );
847 if ( $status->successCount == 0 ) {
848 $status->setOK( false );
849 }
850
851 return $status;
852 }
853
854 /**
855 * Store a batch of files
856 *
857 * @param array $triplets (src, dest zone, dest rel) triplets as per store()
858 * @param int $flags Bitwise combination of the following flags:
859 * self::OVERWRITE Overwrite an existing destination file instead of failing
860 * self::OVERWRITE_SAME Overwrite the file if the destination exists and has the
861 * same contents as the source
862 * self::SKIP_LOCKING Skip any file locking when doing the store
863 * @throws MWException
864 * @return Status
865 */
866 public function storeBatch( array $triplets, $flags = 0 ) {
867 $this->assertWritableRepo(); // fail out if read-only
868
869 if ( $flags & self::DELETE_SOURCE ) {
870 throw new InvalidArgumentException( "DELETE_SOURCE not supported in " . __METHOD__ );
871 }
872
873 $status = $this->newGood();
874 $backend = $this->backend; // convenience
875
876 $operations = [];
877 // Validate each triplet and get the store operation...
878 foreach ( $triplets as $triplet ) {
879 list( $srcPath, $dstZone, $dstRel ) = $triplet;
880 wfDebug( __METHOD__
881 . "( \$src='$srcPath', \$dstZone='$dstZone', \$dstRel='$dstRel' )\n"
882 );
883
884 // Resolve destination path
885 $root = $this->getZonePath( $dstZone );
886 if ( !$root ) {
887 throw new MWException( "Invalid zone: $dstZone" );
888 }
889 if ( !$this->validateFilename( $dstRel ) ) {
890 throw new MWException( 'Validation error in $dstRel' );
891 }
892 $dstPath = "$root/$dstRel";
893 $dstDir = dirname( $dstPath );
894 // Create destination directories for this triplet
895 if ( !$this->initDirectory( $dstDir )->isOK() ) {
896 return $this->newFatal( 'directorycreateerror', $dstDir );
897 }
898
899 // Resolve source to a storage path if virtual
900 $srcPath = $this->resolveToStoragePath( $srcPath );
901
902 // Get the appropriate file operation
903 if ( FileBackend::isStoragePath( $srcPath ) ) {
904 $opName = 'copy';
905 } else {
906 $opName = 'store';
907 }
908 $operations[] = [
909 'op' => $opName,
910 'src' => $srcPath,
911 'dst' => $dstPath,
912 'overwrite' => $flags & self::OVERWRITE,
913 'overwriteSame' => $flags & self::OVERWRITE_SAME,
914 ];
915 }
916
917 // Execute the store operation for each triplet
918 $opts = [ 'force' => true ];
919 if ( $flags & self::SKIP_LOCKING ) {
920 $opts['nonLocking'] = true;
921 }
922 $status->merge( $backend->doOperations( $operations, $opts ) );
923
924 return $status;
925 }
926
927 /**
928 * Deletes a batch of files.
929 * Each file can be a (zone, rel) pair, virtual url, storage path.
930 * It will try to delete each file, but ignores any errors that may occur.
931 *
932 * @param string[] $files List of files to delete
933 * @param int $flags Bitwise combination of the following flags:
934 * self::SKIP_LOCKING Skip any file locking when doing the deletions
935 * @return Status
936 */
937 public function cleanupBatch( array $files, $flags = 0 ) {
938 $this->assertWritableRepo(); // fail out if read-only
939
940 $status = $this->newGood();
941
942 $operations = [];
943 foreach ( $files as $path ) {
944 if ( is_array( $path ) ) {
945 // This is a pair, extract it
946 list( $zone, $rel ) = $path;
947 $path = $this->getZonePath( $zone ) . "/$rel";
948 } else {
949 // Resolve source to a storage path if virtual
950 $path = $this->resolveToStoragePath( $path );
951 }
952 $operations[] = [ 'op' => 'delete', 'src' => $path ];
953 }
954 // Actually delete files from storage...
955 $opts = [ 'force' => true ];
956 if ( $flags & self::SKIP_LOCKING ) {
957 $opts['nonLocking'] = true;
958 }
959 $status->merge( $this->backend->doOperations( $operations, $opts ) );
960
961 return $status;
962 }
963
964 /**
965 * Import a file from the local file system into the repo.
966 * This does no locking nor journaling and overrides existing files.
967 * This function can be used to write to otherwise read-only foreign repos.
968 * This is intended for copying generated thumbnails into the repo.
969 *
970 * @param string|FSFile $src Source file system path, storage path, or virtual URL
971 * @param string $dst Virtual URL or storage path
972 * @param array|string|null $options An array consisting of a key named headers
973 * listing extra headers. If a string, taken as content-disposition header.
974 * (Support for array of options new in 1.23)
975 * @return Status
976 */
977 final public function quickImport( $src, $dst, $options = null ) {
978 return $this->quickImportBatch( [ [ $src, $dst, $options ] ] );
979 }
980
981 /**
982 * Purge a file from the repo. This does no locking nor journaling.
983 * This function can be used to write to otherwise read-only foreign repos.
984 * This is intended for purging thumbnails.
985 *
986 * @param string $path Virtual URL or storage path
987 * @return Status
988 */
989 final public function quickPurge( $path ) {
990 return $this->quickPurgeBatch( [ $path ] );
991 }
992
993 /**
994 * Deletes a directory if empty.
995 * This function can be used to write to otherwise read-only foreign repos.
996 *
997 * @param string $dir Virtual URL (or storage path) of directory to clean
998 * @return Status
999 */
1000 public function quickCleanDir( $dir ) {
1001 $status = $this->newGood();
1002 $status->merge( $this->backend->clean(
1003 [ 'dir' => $this->resolveToStoragePath( $dir ) ] ) );
1004
1005 return $status;
1006 }
1007
1008 /**
1009 * Import a batch of files from the local file system into the repo.
1010 * This does no locking nor journaling and overrides existing files.
1011 * This function can be used to write to otherwise read-only foreign repos.
1012 * This is intended for copying generated thumbnails into the repo.
1013 *
1014 * All path parameters may be a file system path, storage path, or virtual URL.
1015 * When "headers" are given they are used as HTTP headers if supported.
1016 *
1017 * @param array $triples List of (source path or FSFile, destination path, disposition)
1018 * @return Status
1019 */
1020 public function quickImportBatch( array $triples ) {
1021 $status = $this->newGood();
1022 $operations = [];
1023 foreach ( $triples as $triple ) {
1024 list( $src, $dst ) = $triple;
1025 if ( $src instanceof FSFile ) {
1026 $op = 'store';
1027 } else {
1028 $src = $this->resolveToStoragePath( $src );
1029 $op = FileBackend::isStoragePath( $src ) ? 'copy' : 'store';
1030 }
1031 $dst = $this->resolveToStoragePath( $dst );
1032
1033 if ( !isset( $triple[2] ) ) {
1034 $headers = [];
1035 } elseif ( is_string( $triple[2] ) ) {
1036 // back-compat
1037 $headers = [ 'Content-Disposition' => $triple[2] ];
1038 } elseif ( is_array( $triple[2] ) && isset( $triple[2]['headers'] ) ) {
1039 $headers = $triple[2]['headers'];
1040 } else {
1041 $headers = [];
1042 }
1043
1044 $operations[] = [
1045 'op' => $op,
1046 'src' => $src,
1047 'dst' => $dst,
1048 'headers' => $headers
1049 ];
1050 $status->merge( $this->initDirectory( dirname( $dst ) ) );
1051 }
1052 $status->merge( $this->backend->doQuickOperations( $operations ) );
1053
1054 return $status;
1055 }
1056
1057 /**
1058 * Purge a batch of files from the repo.
1059 * This function can be used to write to otherwise read-only foreign repos.
1060 * This does no locking nor journaling and is intended for purging thumbnails.
1061 *
1062 * @param array $paths List of virtual URLs or storage paths
1063 * @return Status
1064 */
1065 public function quickPurgeBatch( array $paths ) {
1066 $status = $this->newGood();
1067 $operations = [];
1068 foreach ( $paths as $path ) {
1069 $operations[] = [
1070 'op' => 'delete',
1071 'src' => $this->resolveToStoragePath( $path ),
1072 'ignoreMissingSource' => true
1073 ];
1074 }
1075 $status->merge( $this->backend->doQuickOperations( $operations ) );
1076
1077 return $status;
1078 }
1079
1080 /**
1081 * Pick a random name in the temp zone and store a file to it.
1082 * Returns a Status object with the file Virtual URL in the value,
1083 * file can later be disposed using FileRepo::freeTemp().
1084 *
1085 * @param string $originalName The base name of the file as specified
1086 * by the user. The file extension will be maintained.
1087 * @param string $srcPath The current location of the file.
1088 * @return Status Object with the URL in the value.
1089 */
1090 public function storeTemp( $originalName, $srcPath ) {
1091 $this->assertWritableRepo(); // fail out if read-only
1092
1093 $date = MWTimestamp::getInstance()->format( 'YmdHis' );
1094 $hashPath = $this->getHashPath( $originalName );
1095 $dstUrlRel = $hashPath . $date . '!' . rawurlencode( $originalName );
1096 $virtualUrl = $this->getVirtualUrl( 'temp' ) . '/' . $dstUrlRel;
1097
1098 $result = $this->quickImport( $srcPath, $virtualUrl );
1099 $result->value = $virtualUrl;
1100
1101 return $result;
1102 }
1103
1104 /**
1105 * Remove a temporary file or mark it for garbage collection
1106 *
1107 * @param string $virtualUrl The virtual URL returned by FileRepo::storeTemp()
1108 * @return bool True on success, false on failure
1109 */
1110 public function freeTemp( $virtualUrl ) {
1111 $this->assertWritableRepo(); // fail out if read-only
1112
1113 $temp = $this->getVirtualUrl( 'temp' );
1114 if ( substr( $virtualUrl, 0, strlen( $temp ) ) != $temp ) {
1115 wfDebug( __METHOD__ . ": Invalid temp virtual URL\n" );
1116
1117 return false;
1118 }
1119
1120 return $this->quickPurge( $virtualUrl )->isOK();
1121 }
1122
1123 /**
1124 * Concatenate a list of temporary files into a target file location.
1125 *
1126 * @param array $srcPaths Ordered list of source virtual URLs/storage paths
1127 * @param string $dstPath Target file system path
1128 * @param int $flags Bitwise combination of the following flags:
1129 * self::DELETE_SOURCE Delete the source files on success
1130 * @return Status
1131 */
1132 public function concatenate( array $srcPaths, $dstPath, $flags = 0 ) {
1133 $this->assertWritableRepo(); // fail out if read-only
1134
1135 $status = $this->newGood();
1136
1137 $sources = [];
1138 foreach ( $srcPaths as $srcPath ) {
1139 // Resolve source to a storage path if virtual
1140 $source = $this->resolveToStoragePath( $srcPath );
1141 $sources[] = $source; // chunk to merge
1142 }
1143
1144 // Concatenate the chunks into one FS file
1145 $params = [ 'srcs' => $sources, 'dst' => $dstPath ];
1146 $status->merge( $this->backend->concatenate( $params ) );
1147 if ( !$status->isOK() ) {
1148 return $status;
1149 }
1150
1151 // Delete the sources if required
1152 if ( $flags & self::DELETE_SOURCE ) {
1153 $status->merge( $this->quickPurgeBatch( $srcPaths ) );
1154 }
1155
1156 // Make sure status is OK, despite any quickPurgeBatch() fatals
1157 $status->setResult( true );
1158
1159 return $status;
1160 }
1161
1162 /**
1163 * Copy or move a file either from a storage path, virtual URL,
1164 * or file system path, into this repository at the specified destination location.
1165 *
1166 * Returns a Status object. On success, the value contains "new" or
1167 * "archived", to indicate whether the file was new with that name.
1168 *
1169 * Options to $options include:
1170 * - headers : name/value map of HTTP headers to use in response to GET/HEAD requests
1171 *
1172 * @param string|FSFile $src The source file system path, storage path, or URL
1173 * @param string $dstRel The destination relative path
1174 * @param string $archiveRel The relative path where the existing file is to
1175 * be archived, if there is one. Relative to the public zone root.
1176 * @param int $flags Bitfield, may be FileRepo::DELETE_SOURCE to indicate
1177 * that the source file should be deleted if possible
1178 * @param array $options Optional additional parameters
1179 * @return Status
1180 */
1181 public function publish(
1182 $src, $dstRel, $archiveRel, $flags = 0, array $options = []
1183 ) {
1184 $this->assertWritableRepo(); // fail out if read-only
1185
1186 $status = $this->publishBatch(
1187 [ [ $src, $dstRel, $archiveRel, $options ] ], $flags );
1188 if ( $status->successCount == 0 ) {
1189 $status->setOK( false );
1190 }
1191 $status->value = $status->value[0] ?? false;
1192
1193 return $status;
1194 }
1195
1196 /**
1197 * Publish a batch of files
1198 *
1199 * @param array $ntuples (source, dest, archive) triplets or
1200 * (source, dest, archive, options) 4-tuples as per publish().
1201 * @param int $flags Bitfield, may be FileRepo::DELETE_SOURCE to indicate
1202 * that the source files should be deleted if possible
1203 * @throws MWException
1204 * @return Status
1205 */
1206 public function publishBatch( array $ntuples, $flags = 0 ) {
1207 $this->assertWritableRepo(); // fail out if read-only
1208
1209 $backend = $this->backend; // convenience
1210 // Try creating directories
1211 $status = $this->initZones( 'public' );
1212 if ( !$status->isOK() ) {
1213 return $status;
1214 }
1215
1216 $status = $this->newGood( [] );
1217
1218 $operations = [];
1219 $sourceFSFilesToDelete = []; // cleanup for disk source files
1220 // Validate each triplet and get the store operation...
1221 foreach ( $ntuples as $ntuple ) {
1222 list( $src, $dstRel, $archiveRel ) = $ntuple;
1223 $srcPath = ( $src instanceof FSFile ) ? $src->getPath() : $src;
1224
1225 $options = $ntuple[3] ?? [];
1226 // Resolve source to a storage path if virtual
1227 $srcPath = $this->resolveToStoragePath( $srcPath );
1228 if ( !$this->validateFilename( $dstRel ) ) {
1229 throw new MWException( 'Validation error in $dstRel' );
1230 }
1231 if ( !$this->validateFilename( $archiveRel ) ) {
1232 throw new MWException( 'Validation error in $archiveRel' );
1233 }
1234
1235 $publicRoot = $this->getZonePath( 'public' );
1236 $dstPath = "$publicRoot/$dstRel";
1237 $archivePath = "$publicRoot/$archiveRel";
1238
1239 $dstDir = dirname( $dstPath );
1240 $archiveDir = dirname( $archivePath );
1241 // Abort immediately on directory creation errors since they're likely to be repetitive
1242 if ( !$this->initDirectory( $dstDir )->isOK() ) {
1243 return $this->newFatal( 'directorycreateerror', $dstDir );
1244 }
1245 if ( !$this->initDirectory( $archiveDir )->isOK() ) {
1246 return $this->newFatal( 'directorycreateerror', $archiveDir );
1247 }
1248
1249 // Set any desired headers to be use in GET/HEAD responses
1250 $headers = $options['headers'] ?? [];
1251
1252 // Archive destination file if it exists.
1253 // This will check if the archive file also exists and fail if does.
1254 // This is a sanity check to avoid data loss. On Windows and Linux,
1255 // copy() will overwrite, so the existence check is vulnerable to
1256 // race conditions unless a functioning LockManager is used.
1257 // LocalFile also uses SELECT FOR UPDATE for synchronization.
1258 $operations[] = [
1259 'op' => 'copy',
1260 'src' => $dstPath,
1261 'dst' => $archivePath,
1262 'ignoreMissingSource' => true
1263 ];
1264
1265 // Copy (or move) the source file to the destination
1266 if ( FileBackend::isStoragePath( $srcPath ) ) {
1267 if ( $flags & self::DELETE_SOURCE ) {
1268 $operations[] = [
1269 'op' => 'move',
1270 'src' => $srcPath,
1271 'dst' => $dstPath,
1272 'overwrite' => true, // replace current
1273 'headers' => $headers
1274 ];
1275 } else {
1276 $operations[] = [
1277 'op' => 'copy',
1278 'src' => $srcPath,
1279 'dst' => $dstPath,
1280 'overwrite' => true, // replace current
1281 'headers' => $headers
1282 ];
1283 }
1284 } else { // FS source path
1285 $operations[] = [
1286 'op' => 'store',
1287 'src' => $src, // prefer FSFile objects
1288 'dst' => $dstPath,
1289 'overwrite' => true, // replace current
1290 'headers' => $headers
1291 ];
1292 if ( $flags & self::DELETE_SOURCE ) {
1293 $sourceFSFilesToDelete[] = $srcPath;
1294 }
1295 }
1296 }
1297
1298 // Execute the operations for each triplet
1299 $status->merge( $backend->doOperations( $operations ) );
1300 // Find out which files were archived...
1301 foreach ( $ntuples as $i => $ntuple ) {
1302 list( , , $archiveRel ) = $ntuple;
1303 $archivePath = $this->getZonePath( 'public' ) . "/$archiveRel";
1304 if ( $this->fileExists( $archivePath ) ) {
1305 $status->value[$i] = 'archived';
1306 } else {
1307 $status->value[$i] = 'new';
1308 }
1309 }
1310 // Cleanup for disk source files...
1311 foreach ( $sourceFSFilesToDelete as $file ) {
1312 Wikimedia\suppressWarnings();
1313 unlink( $file ); // FS cleanup
1314 Wikimedia\restoreWarnings();
1315 }
1316
1317 return $status;
1318 }
1319
1320 /**
1321 * Creates a directory with the appropriate zone permissions.
1322 * Callers are responsible for doing read-only and "writable repo" checks.
1323 *
1324 * @param string $dir Virtual URL (or storage path) of directory to clean
1325 * @return Status
1326 */
1327 protected function initDirectory( $dir ) {
1328 $path = $this->resolveToStoragePath( $dir );
1329 list( , $container, ) = FileBackend::splitStoragePath( $path );
1330
1331 $params = [ 'dir' => $path ];
1332 if ( $this->isPrivate
1333 || $container === $this->zones['deleted']['container']
1334 || $container === $this->zones['temp']['container']
1335 ) {
1336 # Take all available measures to prevent web accessibility of new deleted
1337 # directories, in case the user has not configured offline storage
1338 $params = [ 'noAccess' => true, 'noListing' => true ] + $params;
1339 }
1340
1341 $status = $this->newGood();
1342 $status->merge( $this->backend->prepare( $params ) );
1343
1344 return $status;
1345 }
1346
1347 /**
1348 * Deletes a directory if empty.
1349 *
1350 * @param string $dir Virtual URL (or storage path) of directory to clean
1351 * @return Status
1352 */
1353 public function cleanDir( $dir ) {
1354 $this->assertWritableRepo(); // fail out if read-only
1355
1356 $status = $this->newGood();
1357 $status->merge( $this->backend->clean(
1358 [ 'dir' => $this->resolveToStoragePath( $dir ) ] ) );
1359
1360 return $status;
1361 }
1362
1363 /**
1364 * Checks existence of a file
1365 *
1366 * @param string $file Virtual URL (or storage path) of file to check
1367 * @return bool
1368 */
1369 public function fileExists( $file ) {
1370 $result = $this->fileExistsBatch( [ $file ] );
1371
1372 return $result[0];
1373 }
1374
1375 /**
1376 * Checks existence of an array of files.
1377 *
1378 * @param string[] $files Virtual URLs (or storage paths) of files to check
1379 * @return array Map of files and existence flags, or false
1380 */
1381 public function fileExistsBatch( array $files ) {
1382 $paths = array_map( [ $this, 'resolveToStoragePath' ], $files );
1383 $this->backend->preloadFileStat( [ 'srcs' => $paths ] );
1384
1385 $result = [];
1386 foreach ( $files as $key => $file ) {
1387 $path = $this->resolveToStoragePath( $file );
1388 $result[$key] = $this->backend->fileExists( [ 'src' => $path ] );
1389 }
1390
1391 return $result;
1392 }
1393
1394 /**
1395 * Move a file to the deletion archive.
1396 * If no valid deletion archive exists, this may either delete the file
1397 * or throw an exception, depending on the preference of the repository
1398 *
1399 * @param mixed $srcRel Relative path for the file to be deleted
1400 * @param mixed $archiveRel Relative path for the archive location.
1401 * Relative to a private archive directory.
1402 * @return Status
1403 */
1404 public function delete( $srcRel, $archiveRel ) {
1405 $this->assertWritableRepo(); // fail out if read-only
1406
1407 return $this->deleteBatch( [ [ $srcRel, $archiveRel ] ] );
1408 }
1409
1410 /**
1411 * Move a group of files to the deletion archive.
1412 *
1413 * If no valid deletion archive is configured, this may either delete the
1414 * file or throw an exception, depending on the preference of the repository.
1415 *
1416 * The overwrite policy is determined by the repository -- currently LocalRepo
1417 * assumes a naming scheme in the deleted zone based on content hash, as
1418 * opposed to the public zone which is assumed to be unique.
1419 *
1420 * @param array $sourceDestPairs Array of source/destination pairs. Each element
1421 * is a two-element array containing the source file path relative to the
1422 * public root in the first element, and the archive file path relative
1423 * to the deleted zone root in the second element.
1424 * @throws MWException
1425 * @return Status
1426 */
1427 public function deleteBatch( array $sourceDestPairs ) {
1428 $this->assertWritableRepo(); // fail out if read-only
1429
1430 // Try creating directories
1431 $status = $this->initZones( [ 'public', 'deleted' ] );
1432 if ( !$status->isOK() ) {
1433 return $status;
1434 }
1435
1436 $status = $this->newGood();
1437
1438 $backend = $this->backend; // convenience
1439 $operations = [];
1440 // Validate filenames and create archive directories
1441 foreach ( $sourceDestPairs as $pair ) {
1442 list( $srcRel, $archiveRel ) = $pair;
1443 if ( !$this->validateFilename( $srcRel ) ) {
1444 throw new MWException( __METHOD__ . ':Validation error in $srcRel' );
1445 } elseif ( !$this->validateFilename( $archiveRel ) ) {
1446 throw new MWException( __METHOD__ . ':Validation error in $archiveRel' );
1447 }
1448
1449 $publicRoot = $this->getZonePath( 'public' );
1450 $srcPath = "{$publicRoot}/$srcRel";
1451
1452 $deletedRoot = $this->getZonePath( 'deleted' );
1453 $archivePath = "{$deletedRoot}/{$archiveRel}";
1454 $archiveDir = dirname( $archivePath ); // does not touch FS
1455
1456 // Create destination directories
1457 if ( !$this->initDirectory( $archiveDir )->isOK() ) {
1458 return $this->newFatal( 'directorycreateerror', $archiveDir );
1459 }
1460
1461 $operations[] = [
1462 'op' => 'move',
1463 'src' => $srcPath,
1464 'dst' => $archivePath,
1465 // We may have 2+ identical files being deleted,
1466 // all of which will map to the same destination file
1467 'overwriteSame' => true // also see T33792
1468 ];
1469 }
1470
1471 // Move the files by execute the operations for each pair.
1472 // We're now committed to returning an OK result, which will
1473 // lead to the files being moved in the DB also.
1474 $opts = [ 'force' => true ];
1475 $status->merge( $backend->doOperations( $operations, $opts ) );
1476
1477 return $status;
1478 }
1479
1480 /**
1481 * Delete files in the deleted directory if they are not referenced in the filearchive table
1482 *
1483 * STUB
1484 * @param string[] $storageKeys
1485 */
1486 public function cleanupDeletedBatch( array $storageKeys ) {
1487 $this->assertWritableRepo();
1488 }
1489
1490 /**
1491 * Get a relative path for a deletion archive key,
1492 * e.g. s/z/a/ for sza251lrxrc1jad41h5mgilp8nysje52.jpg
1493 *
1494 * @param string $key
1495 * @throws MWException
1496 * @return string
1497 */
1498 public function getDeletedHashPath( $key ) {
1499 if ( strlen( $key ) < 31 ) {
1500 throw new MWException( "Invalid storage key '$key'." );
1501 }
1502 $path = '';
1503 for ( $i = 0; $i < $this->deletedHashLevels; $i++ ) {
1504 $path .= $key[$i] . '/';
1505 }
1506
1507 return $path;
1508 }
1509
1510 /**
1511 * If a path is a virtual URL, resolve it to a storage path.
1512 * Otherwise, just return the path as it is.
1513 *
1514 * @param string $path
1515 * @return string
1516 * @throws MWException
1517 */
1518 protected function resolveToStoragePath( $path ) {
1519 if ( self::isVirtualUrl( $path ) ) {
1520 return $this->resolveVirtualUrl( $path );
1521 }
1522
1523 return $path;
1524 }
1525
1526 /**
1527 * Get a local FS copy of a file with a given virtual URL/storage path.
1528 * Temporary files may be purged when the file object falls out of scope.
1529 *
1530 * @param string $virtualUrl
1531 * @return TempFSFile|null Returns null on failure
1532 */
1533 public function getLocalCopy( $virtualUrl ) {
1534 $path = $this->resolveToStoragePath( $virtualUrl );
1535
1536 return $this->backend->getLocalCopy( [ 'src' => $path ] );
1537 }
1538
1539 /**
1540 * Get a local FS file with a given virtual URL/storage path.
1541 * The file is either an original or a copy. It should not be changed.
1542 * Temporary files may be purged when the file object falls out of scope.
1543 *
1544 * @param string $virtualUrl
1545 * @return FSFile|null Returns null on failure.
1546 */
1547 public function getLocalReference( $virtualUrl ) {
1548 $path = $this->resolveToStoragePath( $virtualUrl );
1549
1550 return $this->backend->getLocalReference( [ 'src' => $path ] );
1551 }
1552
1553 /**
1554 * Get properties of a file with a given virtual URL/storage path.
1555 * Properties should ultimately be obtained via FSFile::getProps().
1556 *
1557 * @param string $virtualUrl
1558 * @return array
1559 */
1560 public function getFileProps( $virtualUrl ) {
1561 $fsFile = $this->getLocalReference( $virtualUrl );
1562 $mwProps = new MWFileProps( MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer() );
1563 if ( $fsFile ) {
1564 $props = $mwProps->getPropsFromPath( $fsFile->getPath(), true );
1565 } else {
1566 $props = $mwProps->newPlaceholderProps();
1567 }
1568
1569 return $props;
1570 }
1571
1572 /**
1573 * Get the timestamp of a file with a given virtual URL/storage path
1574 *
1575 * @param string $virtualUrl
1576 * @return string|bool False on failure
1577 */
1578 public function getFileTimestamp( $virtualUrl ) {
1579 $path = $this->resolveToStoragePath( $virtualUrl );
1580
1581 return $this->backend->getFileTimestamp( [ 'src' => $path ] );
1582 }
1583
1584 /**
1585 * Get the size of a file with a given virtual URL/storage path
1586 *
1587 * @param string $virtualUrl
1588 * @return int|bool False on failure
1589 */
1590 public function getFileSize( $virtualUrl ) {
1591 $path = $this->resolveToStoragePath( $virtualUrl );
1592
1593 return $this->backend->getFileSize( [ 'src' => $path ] );
1594 }
1595
1596 /**
1597 * Get the sha1 (base 36) of a file with a given virtual URL/storage path
1598 *
1599 * @param string $virtualUrl
1600 * @return string|bool
1601 */
1602 public function getFileSha1( $virtualUrl ) {
1603 $path = $this->resolveToStoragePath( $virtualUrl );
1604
1605 return $this->backend->getFileSha1Base36( [ 'src' => $path ] );
1606 }
1607
1608 /**
1609 * Attempt to stream a file with the given virtual URL/storage path
1610 *
1611 * @param string $virtualUrl
1612 * @param array $headers Additional HTTP headers to send on success
1613 * @param array $optHeaders HTTP request headers (if-modified-since, range, ...)
1614 * @return Status
1615 * @since 1.27
1616 */
1617 public function streamFileWithStatus( $virtualUrl, $headers = [], $optHeaders = [] ) {
1618 $path = $this->resolveToStoragePath( $virtualUrl );
1619 $params = [ 'src' => $path, 'headers' => $headers, 'options' => $optHeaders ];
1620
1621 // T172851: HHVM does not flush the output properly, causing OOM
1622 ob_start( null, 1048576 );
1623 ob_implicit_flush( true );
1624
1625 $status = $this->newGood();
1626 $status->merge( $this->backend->streamFile( $params ) );
1627
1628 // T186565: Close the buffer, unless it has already been closed
1629 // in HTTPFileStreamer::resetOutputBuffers().
1630 if ( ob_get_status() ) {
1631 ob_end_flush();
1632 }
1633
1634 return $status;
1635 }
1636
1637 /**
1638 * Call a callback function for every public regular file in the repository.
1639 * This only acts on the current version of files, not any old versions.
1640 * May use either the database or the filesystem.
1641 *
1642 * @param callable $callback
1643 * @return void
1644 */
1645 public function enumFiles( $callback ) {
1646 $this->enumFilesInStorage( $callback );
1647 }
1648
1649 /**
1650 * Call a callback function for every public file in the repository.
1651 * May use either the database or the filesystem.
1652 *
1653 * @param callable $callback
1654 * @return void
1655 */
1656 protected function enumFilesInStorage( $callback ) {
1657 $publicRoot = $this->getZonePath( 'public' );
1658 $numDirs = 1 << ( $this->hashLevels * 4 );
1659 // Use a priori assumptions about directory structure
1660 // to reduce the tree height of the scanning process.
1661 for ( $flatIndex = 0; $flatIndex < $numDirs; $flatIndex++ ) {
1662 $hexString = sprintf( "%0{$this->hashLevels}x", $flatIndex );
1663 $path = $publicRoot;
1664 for ( $hexPos = 0; $hexPos < $this->hashLevels; $hexPos++ ) {
1665 $path .= '/' . substr( $hexString, 0, $hexPos + 1 );
1666 }
1667 $iterator = $this->backend->getFileList( [ 'dir' => $path ] );
1668 foreach ( $iterator as $name ) {
1669 // Each item returned is a public file
1670 call_user_func( $callback, "{$path}/{$name}" );
1671 }
1672 }
1673 }
1674
1675 /**
1676 * Determine if a relative path is valid, i.e. not blank or involving directory traveral
1677 *
1678 * @param string $filename
1679 * @return bool
1680 */
1681 public function validateFilename( $filename ) {
1682 if ( strval( $filename ) == '' ) {
1683 return false;
1684 }
1685
1686 return FileBackend::isPathTraversalFree( $filename );
1687 }
1688
1689 /**
1690 * Get a callback function to use for cleaning error message parameters
1691 *
1692 * @return callable
1693 */
1694 function getErrorCleanupFunction() {
1695 switch ( $this->pathDisclosureProtection ) {
1696 case 'none':
1697 case 'simple': // b/c
1698 $callback = [ $this, 'passThrough' ];
1699 break;
1700 default: // 'paranoid'
1701 $callback = [ $this, 'paranoidClean' ];
1702 }
1703 return $callback;
1704 }
1705
1706 /**
1707 * Path disclosure protection function
1708 *
1709 * @param string $param
1710 * @return string
1711 */
1712 function paranoidClean( $param ) {
1713 return '[hidden]';
1714 }
1715
1716 /**
1717 * Path disclosure protection function
1718 *
1719 * @param string $param
1720 * @return string
1721 */
1722 function passThrough( $param ) {
1723 return $param;
1724 }
1725
1726 /**
1727 * Create a new fatal error
1728 *
1729 * @param string $message
1730 * @return Status
1731 */
1732 public function newFatal( $message /*, parameters...*/ ) {
1733 $status = Status::newFatal( ...func_get_args() );
1734 $status->cleanCallback = $this->getErrorCleanupFunction();
1735
1736 return $status;
1737 }
1738
1739 /**
1740 * Create a new good result
1741 *
1742 * @param null|mixed $value
1743 * @return Status
1744 */
1745 public function newGood( $value = null ) {
1746 $status = Status::newGood( $value );
1747 $status->cleanCallback = $this->getErrorCleanupFunction();
1748
1749 return $status;
1750 }
1751
1752 /**
1753 * Checks if there is a redirect named as $title. If there is, return the
1754 * title object. If not, return false.
1755 * STUB
1756 *
1757 * @param Title $title Title of image
1758 * @return bool
1759 */
1760 public function checkRedirect( Title $title ) {
1761 return false;
1762 }
1763
1764 /**
1765 * Invalidates image redirect cache related to that image
1766 * Doesn't do anything for repositories that don't support image redirects.
1767 *
1768 * STUB
1769 * @param Title $title Title of image
1770 */
1771 public function invalidateImageRedirect( Title $title ) {
1772 }
1773
1774 /**
1775 * Get the human-readable name of the repo
1776 *
1777 * @return string
1778 */
1779 public function getDisplayName() {
1780 global $wgSitename;
1781
1782 if ( $this->isLocal() ) {
1783 return $wgSitename;
1784 }
1785
1786 // 'shared-repo-name-wikimediacommons' is used when $wgUseInstantCommons = true
1787 return wfMessageFallback( 'shared-repo-name-' . $this->name, 'shared-repo' )->text();
1788 }
1789
1790 /**
1791 * Get the portion of the file that contains the origin file name.
1792 * If that name is too long, then the name "thumbnail.<ext>" will be given.
1793 *
1794 * @param string $name
1795 * @return string
1796 */
1797 public function nameForThumb( $name ) {
1798 if ( strlen( $name ) > $this->abbrvThreshold ) {
1799 $ext = FileBackend::extensionFromPath( $name );
1800 $name = ( $ext == '' ) ? 'thumbnail' : "thumbnail.$ext";
1801 }
1802
1803 return $name;
1804 }
1805
1806 /**
1807 * Returns true if this the local file repository.
1808 *
1809 * @return bool
1810 */
1811 public function isLocal() {
1812 return $this->getName() == 'local';
1813 }
1814
1815 /**
1816 * Get a key on the primary cache for this repository.
1817 * Returns false if the repository's cache is not accessible at this site.
1818 * The parameters are the parts of the key.
1819 *
1820 * STUB
1821 * @return bool
1822 */
1823 public function getSharedCacheKey( /*...*/ ) {
1824 return false;
1825 }
1826
1827 /**
1828 * Get a key for this repo in the local cache domain. These cache keys are
1829 * not shared with remote instances of the repo.
1830 * The parameters are the parts of the key.
1831 *
1832 * @return string
1833 */
1834 public function getLocalCacheKey( /*...*/ ) {
1835 $args = func_get_args();
1836 array_unshift( $args, 'filerepo', $this->getName() );
1837
1838 return $this->wanCache->makeKey( ...$args );
1839 }
1840
1841 /**
1842 * Get a temporary private FileRepo associated with this repo.
1843 *
1844 * Files will be created in the temp zone of this repo.
1845 * It will have the same backend as this repo.
1846 *
1847 * @return TempFileRepo
1848 */
1849 public function getTempRepo() {
1850 return new TempFileRepo( [
1851 'name' => "{$this->name}-temp",
1852 'backend' => $this->backend,
1853 'zones' => [
1854 'public' => [
1855 // Same place storeTemp() uses in the base repo, though
1856 // the path hashing is mismatched, which is annoying.
1857 'container' => $this->zones['temp']['container'],
1858 'directory' => $this->zones['temp']['directory']
1859 ],
1860 'thumb' => [
1861 'container' => $this->zones['temp']['container'],
1862 'directory' => $this->zones['temp']['directory'] == ''
1863 ? 'thumb'
1864 : $this->zones['temp']['directory'] . '/thumb'
1865 ],
1866 'transcoded' => [
1867 'container' => $this->zones['temp']['container'],
1868 'directory' => $this->zones['temp']['directory'] == ''
1869 ? 'transcoded'
1870 : $this->zones['temp']['directory'] . '/transcoded'
1871 ]
1872 ],
1873 'hashLevels' => $this->hashLevels, // performance
1874 'isPrivate' => true // all in temp zone
1875 ] );
1876 }
1877
1878 /**
1879 * Get an UploadStash associated with this repo.
1880 *
1881 * @param User|null $user
1882 * @return UploadStash
1883 */
1884 public function getUploadStash( User $user = null ) {
1885 return new UploadStash( $this, $user );
1886 }
1887
1888 /**
1889 * Throw an exception if this repo is read-only by design.
1890 * This does not and should not check getReadOnlyReason().
1891 *
1892 * @return void
1893 * @throws MWException
1894 */
1895 protected function assertWritableRepo() {
1896 }
1897
1898 /**
1899 * Return information about the repository.
1900 *
1901 * @return array
1902 * @since 1.22
1903 */
1904 public function getInfo() {
1905 $ret = [
1906 'name' => $this->getName(),
1907 'displayname' => $this->getDisplayName(),
1908 'rootUrl' => $this->getZoneUrl( 'public' ),
1909 'local' => $this->isLocal(),
1910 ];
1911
1912 $optionalSettings = [
1913 'url', 'thumbUrl', 'initialCapital', 'descBaseUrl', 'scriptDirUrl', 'articleUrl',
1914 'fetchDescription', 'descriptionCacheExpiry', 'favicon'
1915 ];
1916 foreach ( $optionalSettings as $k ) {
1917 if ( isset( $this->$k ) ) {
1918 $ret[$k] = $this->$k;
1919 }
1920 }
1921
1922 return $ret;
1923 }
1924
1925 /**
1926 * Returns whether or not storage is SHA-1 based
1927 * @return bool
1928 */
1929 public function hasSha1Storage() {
1930 return $this->hasSha1Storage;
1931 }
1932
1933 /**
1934 * Returns whether or not repo supports having originals SHA-1s in the thumb URLs
1935 * @return bool
1936 */
1937 public function supportsSha1URLs() {
1938 return $this->supportsSha1URLs;
1939 }
1940 }