Merge "Use $wgUser in ResourceLoaderUserGroupsModule when possible."
[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 /**
11 * Base code for file repositories.
12 *
13 * @file
14 * @ingroup FileRepo
15 */
16
17 /**
18 * Base class for file repositories
19 *
20 * @ingroup FileRepo
21 */
22 class FileRepo {
23 const DELETE_SOURCE = 1;
24 const OVERWRITE = 2;
25 const OVERWRITE_SAME = 4;
26 const SKIP_LOCKING = 8;
27
28 /** @var FileBackend */
29 protected $backend;
30 /** @var Array Map of zones to config */
31 protected $zones = array();
32
33 var $thumbScriptUrl, $transformVia404;
34 var $descBaseUrl, $scriptDirUrl, $scriptExtension, $articleUrl;
35 var $fetchDescription, $initialCapital;
36 var $pathDisclosureProtection = 'simple'; // 'paranoid'
37 var $descriptionCacheExpiry, $url, $thumbUrl;
38 var $hashLevels, $deletedHashLevels;
39
40 /**
41 * Factory functions for creating new files
42 * Override these in the base class
43 */
44 var $fileFactory = array( 'UnregisteredLocalFile', 'newFromTitle' );
45 var $oldFileFactory = false;
46 var $fileFactoryKey = false, $oldFileFactoryKey = false;
47
48 function __construct( array $info = null ) {
49 // Verify required settings presence
50 if(
51 $info === null
52 || !array_key_exists( 'name', $info )
53 || !array_key_exists( 'backend', $info )
54 ) {
55 throw new MWException( __CLASS__ . " requires an array of options having both 'name' and 'backend' keys.\n" );
56 }
57
58 // Required settings
59 $this->name = $info['name'];
60 if ( $info['backend'] instanceof FileBackend ) {
61 $this->backend = $info['backend']; // useful for testing
62 } else {
63 $this->backend = FileBackendGroup::singleton()->get( $info['backend'] );
64 }
65
66 // Optional settings that can have no value
67 $optionalSettings = array(
68 'descBaseUrl', 'scriptDirUrl', 'articleUrl', 'fetchDescription',
69 'thumbScriptUrl', 'pathDisclosureProtection', 'descriptionCacheExpiry',
70 'scriptExtension'
71 );
72 foreach ( $optionalSettings as $var ) {
73 if ( isset( $info[$var] ) ) {
74 $this->$var = $info[$var];
75 }
76 }
77
78 // Optional settings that have a default
79 $this->initialCapital = isset( $info['initialCapital'] )
80 ? $info['initialCapital']
81 : MWNamespace::isCapitalized( NS_FILE );
82 $this->url = isset( $info['url'] )
83 ? $info['url']
84 : false; // a subclass may set the URL (e.g. ForeignAPIRepo)
85 if ( isset( $info['thumbUrl'] ) ) {
86 $this->thumbUrl = $info['thumbUrl'];
87 } else {
88 $this->thumbUrl = $this->url ? "{$this->url}/thumb" : false;
89 }
90 $this->hashLevels = isset( $info['hashLevels'] )
91 ? $info['hashLevels']
92 : 2;
93 $this->deletedHashLevels = isset( $info['deletedHashLevels'] )
94 ? $info['deletedHashLevels']
95 : $this->hashLevels;
96 $this->transformVia404 = !empty( $info['transformVia404'] );
97 $this->zones = isset( $info['zones'] )
98 ? $info['zones']
99 : array();
100 // Give defaults for the basic zones...
101 foreach ( array( 'public', 'thumb', 'temp', 'deleted' ) as $zone ) {
102 if ( !isset( $this->zones[$zone] ) ) {
103 $this->zones[$zone] = array(
104 'container' => "{$this->name}-{$zone}",
105 'directory' => '' // container root
106 );
107 }
108 }
109 }
110
111 /**
112 * Get the file backend instance. Use this function wisely.
113 *
114 * @return FileBackend
115 */
116 public function getBackend() {
117 return $this->backend;
118 }
119
120 /**
121 * Get an explanatory message if this repo is read-only.
122 * This checks if an administrator disabled writes to the backend.
123 *
124 * @return string|bool Returns false if the repo is not read-only
125 */
126 public function getReadOnlyReason() {
127 return $this->backend->getReadOnlyReason();
128 }
129
130 /**
131 * Check if a single zone or list of zones is defined for usage
132 *
133 * @param $doZones Array Only do a particular zones
134 * @return Status
135 */
136 protected function initZones( $doZones = array() ) {
137 $status = $this->newGood();
138 foreach ( (array)$doZones as $zone ) {
139 $root = $this->getZonePath( $zone );
140 if ( $root === null ) {
141 throw new MWException( "No '$zone' zone defined in the {$this->name} repo." );
142 }
143 }
144 return $status;
145 }
146
147 /**
148 * Take all available measures to prevent web accessibility of new deleted
149 * directories, in case the user has not configured offline storage
150 *
151 * @param $dir string
152 * @return void
153 */
154 protected function initDeletedDir( $dir ) {
155 $this->backend->secure( // prevent web access & dir listings
156 array( 'dir' => $dir, 'noAccess' => true, 'noListing' => true ) );
157 }
158
159 /**
160 * Determine if a string is an mwrepo:// URL
161 *
162 * @param $url string
163 * @return bool
164 */
165 public static function isVirtualUrl( $url ) {
166 return substr( $url, 0, 9 ) == 'mwrepo://';
167 }
168
169 /**
170 * Get a URL referring to this repository, with the private mwrepo protocol.
171 * The suffix, if supplied, is considered to be unencoded, and will be
172 * URL-encoded before being returned.
173 *
174 * @param $suffix string
175 * @return string
176 */
177 public function getVirtualUrl( $suffix = false ) {
178 $path = 'mwrepo://' . $this->name;
179 if ( $suffix !== false ) {
180 $path .= '/' . rawurlencode( $suffix );
181 }
182 return $path;
183 }
184
185 /**
186 * Get the URL corresponding to one of the four basic zones
187 *
188 * @param $zone String: one of: public, deleted, temp, thumb
189 * @return String or false
190 */
191 public function getZoneUrl( $zone ) {
192 switch ( $zone ) {
193 case 'public':
194 return $this->url;
195 case 'temp':
196 return "{$this->url}/temp";
197 case 'deleted':
198 return false; // no public URL
199 case 'thumb':
200 return $this->thumbUrl;
201 default:
202 return false;
203 }
204 }
205
206 /**
207 * Get the backend storage path corresponding to a virtual URL.
208 * Use this function wisely.
209 *
210 * @param $url string
211 * @return string
212 */
213 public function resolveVirtualUrl( $url ) {
214 if ( substr( $url, 0, 9 ) != 'mwrepo://' ) {
215 throw new MWException( __METHOD__.': unknown protocol' );
216 }
217 $bits = explode( '/', substr( $url, 9 ), 3 );
218 if ( count( $bits ) != 3 ) {
219 throw new MWException( __METHOD__.": invalid mwrepo URL: $url" );
220 }
221 list( $repo, $zone, $rel ) = $bits;
222 if ( $repo !== $this->name ) {
223 throw new MWException( __METHOD__.": fetching from a foreign repo is not supported" );
224 }
225 $base = $this->getZonePath( $zone );
226 if ( !$base ) {
227 throw new MWException( __METHOD__.": invalid zone: $zone" );
228 }
229 return $base . '/' . rawurldecode( $rel );
230 }
231
232 /**
233 * The the storage container and base path of a zone
234 *
235 * @param $zone string
236 * @return Array (container, base path) or (null, null)
237 */
238 protected function getZoneLocation( $zone ) {
239 if ( !isset( $this->zones[$zone] ) ) {
240 return array( null, null ); // bogus
241 }
242 return array( $this->zones[$zone]['container'], $this->zones[$zone]['directory'] );
243 }
244
245 /**
246 * Get the storage path corresponding to one of the zones
247 *
248 * @param $zone string
249 * @return string|null Returns null if the zone is not defined
250 */
251 public function getZonePath( $zone ) {
252 list( $container, $base ) = $this->getZoneLocation( $zone );
253 if ( $container === null || $base === null ) {
254 return null;
255 }
256 $backendName = $this->backend->getName();
257 if ( $base != '' ) { // may not be set
258 $base = "/{$base}";
259 }
260 return "mwstore://$backendName/{$container}{$base}";
261 }
262
263 /**
264 * Create a new File object from the local repository
265 *
266 * @param $title Mixed: Title object or string
267 * @param $time Mixed: Time at which the image was uploaded.
268 * If this is specified, the returned object will be an
269 * instance of the repository's old file class instead of a
270 * current file. Repositories not supporting version control
271 * should return false if this parameter is set.
272 * @return File|null A File, or null if passed an invalid Title
273 */
274 public function newFile( $title, $time = false ) {
275 $title = File::normalizeTitle( $title );
276 if ( !$title ) {
277 return null;
278 }
279 if ( $time ) {
280 if ( $this->oldFileFactory ) {
281 return call_user_func( $this->oldFileFactory, $title, $this, $time );
282 } else {
283 return false;
284 }
285 } else {
286 return call_user_func( $this->fileFactory, $title, $this );
287 }
288 }
289
290 /**
291 * Find an instance of the named file created at the specified time
292 * Returns false if the file does not exist. Repositories not supporting
293 * version control should return false if the time is specified.
294 *
295 * @param $title Mixed: Title object or string
296 * @param $options array Associative array of options:
297 * time: requested time for a specific file version, or false for the
298 * current version. An image object will be returned which was
299 * created at the specified time (which may be archived or current).
300 *
301 * ignoreRedirect: If true, do not follow file redirects
302 *
303 * private: If true, return restricted (deleted) files if the current
304 * user is allowed to view them. Otherwise, such files will not
305 * be found.
306 * @return File|bool False on failure
307 */
308 public function findFile( $title, $options = array() ) {
309 $title = File::normalizeTitle( $title );
310 if ( !$title ) {
311 return false;
312 }
313 $time = isset( $options['time'] ) ? $options['time'] : false;
314 # First try the current version of the file to see if it precedes the timestamp
315 $img = $this->newFile( $title );
316 if ( !$img ) {
317 return false;
318 }
319 if ( $img->exists() && ( !$time || $img->getTimestamp() == $time ) ) {
320 return $img;
321 }
322 # Now try an old version of the file
323 if ( $time !== false ) {
324 $img = $this->newFile( $title, $time );
325 if ( $img && $img->exists() ) {
326 if ( !$img->isDeleted( File::DELETED_FILE ) ) {
327 return $img; // always OK
328 } elseif ( !empty( $options['private'] ) && $img->userCan( File::DELETED_FILE ) ) {
329 return $img;
330 }
331 }
332 }
333
334 # Now try redirects
335 if ( !empty( $options['ignoreRedirect'] ) ) {
336 return false;
337 }
338 $redir = $this->checkRedirect( $title );
339 if ( $redir && $title->getNamespace() == NS_FILE) {
340 $img = $this->newFile( $redir );
341 if ( !$img ) {
342 return false;
343 }
344 if ( $img->exists() ) {
345 $img->redirectedFrom( $title->getDBkey() );
346 return $img;
347 }
348 }
349 return false;
350 }
351
352 /**
353 * Find many files at once.
354 *
355 * @param $items array An array of titles, or an array of findFile() options with
356 * the "title" option giving the title. Example:
357 *
358 * $findItem = array( 'title' => $title, 'private' => true );
359 * $findBatch = array( $findItem );
360 * $repo->findFiles( $findBatch );
361 * @return array
362 */
363 public function findFiles( array $items ) {
364 $result = array();
365 foreach ( $items as $item ) {
366 if ( is_array( $item ) ) {
367 $title = $item['title'];
368 $options = $item;
369 unset( $options['title'] );
370 } else {
371 $title = $item;
372 $options = array();
373 }
374 $file = $this->findFile( $title, $options );
375 if ( $file ) {
376 $result[$file->getTitle()->getDBkey()] = $file;
377 }
378 }
379 return $result;
380 }
381
382 /**
383 * Find an instance of the file with this key, created at the specified time
384 * Returns false if the file does not exist. Repositories not supporting
385 * version control should return false if the time is specified.
386 *
387 * @param $sha1 String base 36 SHA-1 hash
388 * @param $options array Option array, same as findFile().
389 * @return File|bool False on failure
390 */
391 public function findFileFromKey( $sha1, $options = array() ) {
392 $time = isset( $options['time'] ) ? $options['time'] : false;
393 # First try to find a matching current version of a file...
394 if ( $this->fileFactoryKey ) {
395 $img = call_user_func( $this->fileFactoryKey, $sha1, $this, $time );
396 } else {
397 return false; // find-by-sha1 not supported
398 }
399 if ( $img && $img->exists() ) {
400 return $img;
401 }
402 # Now try to find a matching old version of a file...
403 if ( $time !== false && $this->oldFileFactoryKey ) { // find-by-sha1 supported?
404 $img = call_user_func( $this->oldFileFactoryKey, $sha1, $this, $time );
405 if ( $img && $img->exists() ) {
406 if ( !$img->isDeleted( File::DELETED_FILE ) ) {
407 return $img; // always OK
408 } elseif ( !empty( $options['private'] ) && $img->userCan( File::DELETED_FILE ) ) {
409 return $img;
410 }
411 }
412 }
413 return false;
414 }
415
416 /**
417 * Get an array or iterator of file objects for files that have a given
418 * SHA-1 content hash.
419 *
420 * STUB
421 * @return array
422 */
423 public function findBySha1( $hash ) {
424 return array();
425 }
426
427 /**
428 * Get the public root URL of the repository
429 *
430 * @return string
431 */
432 public function getRootUrl() {
433 return $this->url;
434 }
435
436 /**
437 * Get the URL of thumb.php
438 *
439 * @return string
440 */
441 public function getThumbScriptUrl() {
442 return $this->thumbScriptUrl;
443 }
444
445 /**
446 * Returns true if the repository can transform files via a 404 handler
447 *
448 * @return bool
449 */
450 public function canTransformVia404() {
451 return $this->transformVia404;
452 }
453
454 /**
455 * Get the name of an image from its title object
456 *
457 * @param $title Title
458 * @return String
459 */
460 public function getNameFromTitle( Title $title ) {
461 global $wgContLang;
462 if ( $this->initialCapital != MWNamespace::isCapitalized( NS_FILE ) ) {
463 $name = $title->getUserCaseDBKey();
464 if ( $this->initialCapital ) {
465 $name = $wgContLang->ucfirst( $name );
466 }
467 } else {
468 $name = $title->getDBkey();
469 }
470 return $name;
471 }
472
473 /**
474 * Get the public zone root storage directory of the repository
475 *
476 * @return string
477 */
478 public function getRootDirectory() {
479 return $this->getZonePath( 'public' );
480 }
481
482 /**
483 * Get a relative path including trailing slash, e.g. f/fa/
484 * If the repo is not hashed, returns an empty string
485 *
486 * @param $name string Name of file
487 * @return string
488 */
489 public function getHashPath( $name ) {
490 return self::getHashPathForLevel( $name, $this->hashLevels );
491 }
492
493 /**
494 * Get a relative path including trailing slash, e.g. f/fa/
495 * If the repo is not hashed, returns an empty string
496 *
497 * @param $suffix string Basename of file from FileRepo::storeTemp()
498 * @return string
499 */
500 public function getTempHashPath( $suffix ) {
501 $parts = explode( '!', $suffix, 2 ); // format is <timestamp>!<name> or just <name>
502 $name = isset( $parts[1] ) ? $parts[1] : $suffix; // hash path is not based on timestamp
503 return self::getHashPathForLevel( $name, $this->hashLevels );
504 }
505
506 /**
507 * @param $name
508 * @param $levels
509 * @return string
510 */
511 protected static function getHashPathForLevel( $name, $levels ) {
512 if ( $levels == 0 ) {
513 return '';
514 } else {
515 $hash = md5( $name );
516 $path = '';
517 for ( $i = 1; $i <= $levels; $i++ ) {
518 $path .= substr( $hash, 0, $i ) . '/';
519 }
520 return $path;
521 }
522 }
523
524 /**
525 * Get the number of hash directory levels
526 *
527 * @return integer
528 */
529 public function getHashLevels() {
530 return $this->hashLevels;
531 }
532
533 /**
534 * Get the name of this repository, as specified by $info['name]' to the constructor
535 *
536 * @return string
537 */
538 public function getName() {
539 return $this->name;
540 }
541
542 /**
543 * Make an url to this repo
544 *
545 * @param $query mixed Query string to append
546 * @param $entry string Entry point; defaults to index
547 * @return string|bool False on failure
548 */
549 public function makeUrl( $query = '', $entry = 'index' ) {
550 if ( isset( $this->scriptDirUrl ) ) {
551 $ext = isset( $this->scriptExtension ) ? $this->scriptExtension : '.php';
552 return wfAppendQuery( "{$this->scriptDirUrl}/{$entry}{$ext}", $query );
553 }
554 return false;
555 }
556
557 /**
558 * Get the URL of an image description page. May return false if it is
559 * unknown or not applicable. In general this should only be called by the
560 * File class, since it may return invalid results for certain kinds of
561 * repositories. Use File::getDescriptionUrl() in user code.
562 *
563 * In particular, it uses the article paths as specified to the repository
564 * constructor, whereas local repositories use the local Title functions.
565 *
566 * @param $name string
567 * @return string
568 */
569 public function getDescriptionUrl( $name ) {
570 $encName = wfUrlencode( $name );
571 if ( !is_null( $this->descBaseUrl ) ) {
572 # "http://example.com/wiki/Image:"
573 return $this->descBaseUrl . $encName;
574 }
575 if ( !is_null( $this->articleUrl ) ) {
576 # "http://example.com/wiki/$1"
577 #
578 # We use "Image:" as the canonical namespace for
579 # compatibility across all MediaWiki versions.
580 return str_replace( '$1',
581 "Image:$encName", $this->articleUrl );
582 }
583 if ( !is_null( $this->scriptDirUrl ) ) {
584 # "http://example.com/w"
585 #
586 # We use "Image:" as the canonical namespace for
587 # compatibility across all MediaWiki versions,
588 # and just sort of hope index.php is right. ;)
589 return $this->makeUrl( "title=Image:$encName" );
590 }
591 return false;
592 }
593
594 /**
595 * Get the URL of the content-only fragment of the description page. For
596 * MediaWiki this means action=render. This should only be called by the
597 * repository's file class, since it may return invalid results. User code
598 * should use File::getDescriptionText().
599 *
600 * @param $name String: name of image to fetch
601 * @param $lang String: language to fetch it in, if any.
602 * @return string
603 */
604 public function getDescriptionRenderUrl( $name, $lang = null ) {
605 $query = 'action=render';
606 if ( !is_null( $lang ) ) {
607 $query .= '&uselang=' . $lang;
608 }
609 if ( isset( $this->scriptDirUrl ) ) {
610 return $this->makeUrl(
611 'title=' .
612 wfUrlencode( 'Image:' . $name ) .
613 "&$query" );
614 } else {
615 $descUrl = $this->getDescriptionUrl( $name );
616 if ( $descUrl ) {
617 return wfAppendQuery( $descUrl, $query );
618 } else {
619 return false;
620 }
621 }
622 }
623
624 /**
625 * Get the URL of the stylesheet to apply to description pages
626 *
627 * @return string|bool False on failure
628 */
629 public function getDescriptionStylesheetUrl() {
630 if ( isset( $this->scriptDirUrl ) ) {
631 return $this->makeUrl( 'title=MediaWiki:Filepage.css&' .
632 wfArrayToCGI( Skin::getDynamicStylesheetQuery() ) );
633 }
634 return false;
635 }
636
637 /**
638 * Store a file to a given destination.
639 *
640 * @param $srcPath String: source FS path, storage path, or virtual URL
641 * @param $dstZone String: destination zone
642 * @param $dstRel String: destination relative path
643 * @param $flags Integer: bitwise combination of the following flags:
644 * self::DELETE_SOURCE Delete the source file after upload
645 * self::OVERWRITE Overwrite an existing destination file instead of failing
646 * self::OVERWRITE_SAME Overwrite the file if the destination exists and has the
647 * same contents as the source
648 * self::SKIP_LOCKING Skip any file locking when doing the store
649 * @return FileRepoStatus
650 */
651 public function store( $srcPath, $dstZone, $dstRel, $flags = 0 ) {
652 $this->assertWritableRepo(); // fail out if read-only
653
654 $status = $this->storeBatch( array( array( $srcPath, $dstZone, $dstRel ) ), $flags );
655 if ( $status->successCount == 0 ) {
656 $status->ok = false;
657 }
658
659 return $status;
660 }
661
662 /**
663 * Store a batch of files
664 *
665 * @param $triplets Array: (src, dest zone, dest rel) triplets as per store()
666 * @param $flags Integer: bitwise combination of the following flags:
667 * self::DELETE_SOURCE Delete the source file after upload
668 * self::OVERWRITE Overwrite an existing destination file instead of failing
669 * self::OVERWRITE_SAME Overwrite the file if the destination exists and has the
670 * same contents as the source
671 * self::SKIP_LOCKING Skip any file locking when doing the store
672 * @return FileRepoStatus
673 */
674 public function storeBatch( array $triplets, $flags = 0 ) {
675 $this->assertWritableRepo(); // fail out if read-only
676
677 $status = $this->newGood();
678 $backend = $this->backend; // convenience
679
680 $operations = array();
681 $sourceFSFilesToDelete = array(); // cleanup for disk source files
682 // Validate each triplet and get the store operation...
683 foreach ( $triplets as $triplet ) {
684 list( $srcPath, $dstZone, $dstRel ) = $triplet;
685 wfDebug( __METHOD__
686 . "( \$src='$srcPath', \$dstZone='$dstZone', \$dstRel='$dstRel' )\n"
687 );
688
689 // Resolve destination path
690 $root = $this->getZonePath( $dstZone );
691 if ( !$root ) {
692 throw new MWException( "Invalid zone: $dstZone" );
693 }
694 if ( !$this->validateFilename( $dstRel ) ) {
695 throw new MWException( 'Validation error in $dstRel' );
696 }
697 $dstPath = "$root/$dstRel";
698 $dstDir = dirname( $dstPath );
699 // Create destination directories for this triplet
700 if ( !$backend->prepare( array( 'dir' => $dstDir ) )->isOK() ) {
701 return $this->newFatal( 'directorycreateerror', $dstDir );
702 }
703
704 if ( $dstZone == 'deleted' ) {
705 $this->initDeletedDir( $dstDir );
706 }
707
708 // Resolve source to a storage path if virtual
709 if ( self::isVirtualUrl( $srcPath ) ) {
710 $srcPath = $this->resolveVirtualUrl( $srcPath );
711 }
712
713 // Get the appropriate file operation
714 if ( FileBackend::isStoragePath( $srcPath ) ) {
715 $opName = ( $flags & self::DELETE_SOURCE ) ? 'move' : 'copy';
716 } else {
717 $opName = 'store';
718 if ( $flags & self::DELETE_SOURCE ) {
719 $sourceFSFilesToDelete[] = $srcPath;
720 }
721 }
722 $operations[] = array(
723 'op' => $opName,
724 'src' => $srcPath,
725 'dst' => $dstPath,
726 'overwrite' => $flags & self::OVERWRITE,
727 'overwriteSame' => $flags & self::OVERWRITE_SAME,
728 );
729 }
730
731 // Execute the store operation for each triplet
732 $opts = array( 'force' => true );
733 if ( $flags & self::SKIP_LOCKING ) {
734 $opts['nonLocking'] = true;
735 }
736 $status->merge( $backend->doOperations( $operations, $opts ) );
737 // Cleanup for disk source files...
738 foreach ( $sourceFSFilesToDelete as $file ) {
739 wfSuppressWarnings();
740 unlink( $file ); // FS cleanup
741 wfRestoreWarnings();
742 }
743
744 return $status;
745 }
746
747 /**
748 * Deletes a batch of files.
749 * Each file can be a (zone, rel) pair, virtual url, storage path.
750 * It will try to delete each file, but ignores any errors that may occur.
751 *
752 * @param $pairs array List of files to delete
753 * @param $flags Integer: bitwise combination of the following flags:
754 * self::SKIP_LOCKING Skip any file locking when doing the deletions
755 * @return FileRepoStatus
756 */
757 public function cleanupBatch( array $files, $flags = 0 ) {
758 $this->assertWritableRepo(); // fail out if read-only
759
760 $status = $this->newGood();
761
762 $operations = array();
763 foreach ( $files as $path ) {
764 if ( is_array( $path ) ) {
765 // This is a pair, extract it
766 list( $zone, $rel ) = $path;
767 $path = $this->getZonePath( $zone ) . "/$rel";
768 } else {
769 // Resolve source to a storage path if virtual
770 if ( self::isVirtualUrl( $path ) ) {
771 $path = $this->resolveVirtualUrl( $path );
772 }
773 }
774 $operations[] = array( 'op' => 'delete', 'src' => $path );
775 }
776 // Actually delete files from storage...
777 $opts = array( 'force' => true );
778 if ( $flags & self::SKIP_LOCKING ) {
779 $opts['nonLocking'] = true;
780 }
781 $status->merge( $this->backend->doOperations( $operations, $opts ) );
782
783 return $status;
784 }
785
786 /**
787 * Pick a random name in the temp zone and store a file to it.
788 * Returns a FileRepoStatus object with the file Virtual URL in the value,
789 * file can later be disposed using FileRepo::freeTemp().
790 *
791 * @param $originalName String: the base name of the file as specified
792 * by the user. The file extension will be maintained.
793 * @param $srcPath String: the current location of the file.
794 * @return FileRepoStatus object with the URL in the value.
795 */
796 public function storeTemp( $originalName, $srcPath ) {
797 $this->assertWritableRepo(); // fail out if read-only
798
799 $date = gmdate( "YmdHis" );
800 $hashPath = $this->getHashPath( $originalName );
801 $dstRel = "{$hashPath}{$date}!{$originalName}";
802 $dstUrlRel = $hashPath . $date . '!' . rawurlencode( $originalName );
803
804 $result = $this->store( $srcPath, 'temp', $dstRel, self::SKIP_LOCKING );
805 $result->value = $this->getVirtualUrl( 'temp' ) . '/' . $dstUrlRel;
806
807 return $result;
808 }
809
810 /**
811 * Concatenate a list of files into a target file location.
812 *
813 * @param $srcPaths Array Ordered list of source virtual URLs/storage paths
814 * @param $dstPath String Target file system path
815 * @param $flags Integer: bitwise combination of the following flags:
816 * self::DELETE_SOURCE Delete the source files
817 * @return FileRepoStatus
818 */
819 public function concatenate( array $srcPaths, $dstPath, $flags = 0 ) {
820 $this->assertWritableRepo(); // fail out if read-only
821
822 $status = $this->newGood();
823
824 $sources = array();
825 $deleteOperations = array(); // post-concatenate ops
826 foreach ( $srcPaths as $srcPath ) {
827 // Resolve source to a storage path if virtual
828 $source = $this->resolveToStoragePath( $srcPath );
829 $sources[] = $source; // chunk to merge
830 if ( $flags & self::DELETE_SOURCE ) {
831 $deleteOperations[] = array( 'op' => 'delete', 'src' => $source );
832 }
833 }
834
835 // Concatenate the chunks into one FS file
836 $params = array( 'srcs' => $sources, 'dst' => $dstPath );
837 $status->merge( $this->backend->concatenate( $params ) );
838 if ( !$status->isOK() ) {
839 return $status;
840 }
841
842 // Delete the sources if required
843 if ( $deleteOperations ) {
844 $opts = array( 'force' => true );
845 $status->merge( $this->backend->doOperations( $deleteOperations, $opts ) );
846 }
847
848 // Make sure status is OK, despite any $deleteOperations fatals
849 $status->setResult( true );
850
851 return $status;
852 }
853
854 /**
855 * Remove a temporary file or mark it for garbage collection
856 *
857 * @param $virtualUrl String: the virtual URL returned by FileRepo::storeTemp()
858 * @return Boolean: true on success, false on failure
859 */
860 public function freeTemp( $virtualUrl ) {
861 $this->assertWritableRepo(); // fail out if read-only
862
863 $temp = "mwrepo://{$this->name}/temp";
864 if ( substr( $virtualUrl, 0, strlen( $temp ) ) != $temp ) {
865 wfDebug( __METHOD__.": Invalid temp virtual URL\n" );
866 return false;
867 }
868 $path = $this->resolveVirtualUrl( $virtualUrl );
869
870 return $this->cleanupBatch( array( $path ), self::SKIP_LOCKING )->isOK();
871 }
872
873 /**
874 * Copy or move a file either from a storage path, virtual URL,
875 * or FS path, into this repository at the specified destination location.
876 *
877 * Returns a FileRepoStatus object. On success, the value contains "new" or
878 * "archived", to indicate whether the file was new with that name.
879 *
880 * @param $srcPath String: the source FS path, storage path, or URL
881 * @param $dstRel String: the destination relative path
882 * @param $archiveRel String: the relative path where the existing file is to
883 * be archived, if there is one. Relative to the public zone root.
884 * @param $flags Integer: bitfield, may be FileRepo::DELETE_SOURCE to indicate
885 * that the source file should be deleted if possible
886 * @return FileRepoStatus
887 */
888 public function publish( $srcPath, $dstRel, $archiveRel, $flags = 0 ) {
889 $this->assertWritableRepo(); // fail out if read-only
890
891 $status = $this->publishBatch( array( array( $srcPath, $dstRel, $archiveRel ) ), $flags );
892 if ( $status->successCount == 0 ) {
893 $status->ok = false;
894 }
895 if ( isset( $status->value[0] ) ) {
896 $status->value = $status->value[0];
897 } else {
898 $status->value = false;
899 }
900
901 return $status;
902 }
903
904 /**
905 * Publish a batch of files
906 *
907 * @param $triplets Array: (source, dest, archive) triplets as per publish()
908 * @param $flags Integer: bitfield, may be FileRepo::DELETE_SOURCE to indicate
909 * that the source files should be deleted if possible
910 * @return FileRepoStatus
911 */
912 public function publishBatch( array $triplets, $flags = 0 ) {
913 $this->assertWritableRepo(); // fail out if read-only
914
915 $backend = $this->backend; // convenience
916 // Try creating directories
917 $status = $this->initZones( 'public' );
918 if ( !$status->isOK() ) {
919 return $status;
920 }
921
922 $status = $this->newGood( array() );
923
924 $operations = array();
925 $sourceFSFilesToDelete = array(); // cleanup for disk source files
926 // Validate each triplet and get the store operation...
927 foreach ( $triplets as $i => $triplet ) {
928 list( $srcPath, $dstRel, $archiveRel ) = $triplet;
929 // Resolve source to a storage path if virtual
930 if ( substr( $srcPath, 0, 9 ) == 'mwrepo://' ) {
931 $srcPath = $this->resolveVirtualUrl( $srcPath );
932 }
933 if ( !$this->validateFilename( $dstRel ) ) {
934 throw new MWException( 'Validation error in $dstRel' );
935 }
936 if ( !$this->validateFilename( $archiveRel ) ) {
937 throw new MWException( 'Validation error in $archiveRel' );
938 }
939
940 $publicRoot = $this->getZonePath( 'public' );
941 $dstPath = "$publicRoot/$dstRel";
942 $archivePath = "$publicRoot/$archiveRel";
943
944 $dstDir = dirname( $dstPath );
945 $archiveDir = dirname( $archivePath );
946 // Abort immediately on directory creation errors since they're likely to be repetitive
947 if ( !$backend->prepare( array( 'dir' => $dstDir ) )->isOK() ) {
948 return $this->newFatal( 'directorycreateerror', $dstDir );
949 }
950 if ( !$backend->prepare( array( 'dir' => $archiveDir ) )->isOK() ) {
951 return $this->newFatal( 'directorycreateerror', $archiveDir );
952 }
953
954 // Archive destination file if it exists
955 if ( $backend->fileExists( array( 'src' => $dstPath ) ) ) {
956 // Check if the archive file exists
957 // This is a sanity check to avoid data loss. In UNIX, the rename primitive
958 // unlinks the destination file if it exists. DB-based synchronisation in
959 // publishBatch's caller should prevent races. In Windows there's no
960 // problem because the rename primitive fails if the destination exists.
961 if ( $backend->fileExists( array( 'src' => $archivePath ) ) ) {
962 $operations[] = array( 'op' => 'null' );
963 continue;
964 } else {
965 $operations[] = array(
966 'op' => 'move',
967 'src' => $dstPath,
968 'dst' => $archivePath
969 );
970 }
971 $status->value[$i] = 'archived';
972 } else {
973 $status->value[$i] = 'new';
974 }
975 // Copy (or move) the source file to the destination
976 if ( FileBackend::isStoragePath( $srcPath ) ) {
977 if ( $flags & self::DELETE_SOURCE ) {
978 $operations[] = array(
979 'op' => 'move',
980 'src' => $srcPath,
981 'dst' => $dstPath
982 );
983 } else {
984 $operations[] = array(
985 'op' => 'copy',
986 'src' => $srcPath,
987 'dst' => $dstPath
988 );
989 }
990 } else { // FS source path
991 $operations[] = array(
992 'op' => 'store',
993 'src' => $srcPath,
994 'dst' => $dstPath
995 );
996 if ( $flags & self::DELETE_SOURCE ) {
997 $sourceFSFilesToDelete[] = $srcPath;
998 }
999 }
1000 }
1001
1002 // Execute the operations for each triplet
1003 $opts = array( 'force' => true );
1004 $status->merge( $backend->doOperations( $operations, $opts ) );
1005 // Cleanup for disk source files...
1006 foreach ( $sourceFSFilesToDelete as $file ) {
1007 wfSuppressWarnings();
1008 unlink( $file ); // FS cleanup
1009 wfRestoreWarnings();
1010 }
1011
1012 return $status;
1013 }
1014
1015 /**
1016 * Checks existence of a a file
1017 *
1018 * @param $file string Virtual URL (or storage path) of file to check
1019 * @return bool
1020 */
1021 public function fileExists( $file ) {
1022 $result = $this->fileExistsBatch( array( $file ) );
1023 return $result[0];
1024 }
1025
1026 /**
1027 * Checks existence of an array of files.
1028 *
1029 * @param $files Array: Virtual URLs (or storage paths) of files to check
1030 * @return array|bool Either array of files and existence flags, or false
1031 */
1032 public function fileExistsBatch( array $files ) {
1033 $result = array();
1034 foreach ( $files as $key => $file ) {
1035 if ( self::isVirtualUrl( $file ) ) {
1036 $file = $this->resolveVirtualUrl( $file );
1037 }
1038 $result[$key] = $this->backend->fileExists( array( 'src' => $file ) );
1039 }
1040 return $result;
1041 }
1042
1043 /**
1044 * Move a file to the deletion archive.
1045 * If no valid deletion archive exists, this may either delete the file
1046 * or throw an exception, depending on the preference of the repository
1047 *
1048 * @param $srcRel Mixed: relative path for the file to be deleted
1049 * @param $archiveRel Mixed: relative path for the archive location.
1050 * Relative to a private archive directory.
1051 * @return FileRepoStatus object
1052 */
1053 public function delete( $srcRel, $archiveRel ) {
1054 $this->assertWritableRepo(); // fail out if read-only
1055
1056 return $this->deleteBatch( array( array( $srcRel, $archiveRel ) ) );
1057 }
1058
1059 /**
1060 * Move a group of files to the deletion archive.
1061 *
1062 * If no valid deletion archive is configured, this may either delete the
1063 * file or throw an exception, depending on the preference of the repository.
1064 *
1065 * The overwrite policy is determined by the repository -- currently LocalRepo
1066 * assumes a naming scheme in the deleted zone based on content hash, as
1067 * opposed to the public zone which is assumed to be unique.
1068 *
1069 * @param $sourceDestPairs Array of source/destination pairs. Each element
1070 * is a two-element array containing the source file path relative to the
1071 * public root in the first element, and the archive file path relative
1072 * to the deleted zone root in the second element.
1073 * @return FileRepoStatus
1074 */
1075 public function deleteBatch( array $sourceDestPairs ) {
1076 $this->assertWritableRepo(); // fail out if read-only
1077
1078 // Try creating directories
1079 $status = $this->initZones( array( 'public', 'deleted' ) );
1080 if ( !$status->isOK() ) {
1081 return $status;
1082 }
1083
1084 $status = $this->newGood();
1085
1086 $backend = $this->backend; // convenience
1087 $operations = array();
1088 // Validate filenames and create archive directories
1089 foreach ( $sourceDestPairs as $pair ) {
1090 list( $srcRel, $archiveRel ) = $pair;
1091 if ( !$this->validateFilename( $srcRel ) ) {
1092 throw new MWException( __METHOD__.':Validation error in $srcRel' );
1093 } elseif ( !$this->validateFilename( $archiveRel ) ) {
1094 throw new MWException( __METHOD__.':Validation error in $archiveRel' );
1095 }
1096
1097 $publicRoot = $this->getZonePath( 'public' );
1098 $srcPath = "{$publicRoot}/$srcRel";
1099
1100 $deletedRoot = $this->getZonePath( 'deleted' );
1101 $archivePath = "{$deletedRoot}/{$archiveRel}";
1102 $archiveDir = dirname( $archivePath ); // does not touch FS
1103
1104 // Create destination directories
1105 if ( !$backend->prepare( array( 'dir' => $archiveDir ) )->isOK() ) {
1106 return $this->newFatal( 'directorycreateerror', $archiveDir );
1107 }
1108 $this->initDeletedDir( $archiveDir );
1109
1110 $operations[] = array(
1111 'op' => 'move',
1112 'src' => $srcPath,
1113 'dst' => $archivePath,
1114 // We may have 2+ identical files being deleted,
1115 // all of which will map to the same destination file
1116 'overwriteSame' => true // also see bug 31792
1117 );
1118 }
1119
1120 // Move the files by execute the operations for each pair.
1121 // We're now committed to returning an OK result, which will
1122 // lead to the files being moved in the DB also.
1123 $opts = array( 'force' => true );
1124 $status->merge( $backend->doOperations( $operations, $opts ) );
1125
1126 return $status;
1127 }
1128
1129 /**
1130 * Delete files in the deleted directory if they are not referenced in the filearchive table
1131 *
1132 * STUB
1133 */
1134 public function cleanupDeletedBatch( array $storageKeys ) {
1135 $this->assertWritableRepo();
1136 }
1137
1138 /**
1139 * Get a relative path for a deletion archive key,
1140 * e.g. s/z/a/ for sza251lrxrc1jad41h5mgilp8nysje52.jpg
1141 *
1142 * @return string
1143 */
1144 public function getDeletedHashPath( $key ) {
1145 $path = '';
1146 for ( $i = 0; $i < $this->deletedHashLevels; $i++ ) {
1147 $path .= $key[$i] . '/';
1148 }
1149 return $path;
1150 }
1151
1152 /**
1153 * If a path is a virtual URL, resolve it to a storage path.
1154 * Otherwise, just return the path as it is.
1155 *
1156 * @param $path string
1157 * @return string
1158 * @throws MWException
1159 */
1160 protected function resolveToStoragePath( $path ) {
1161 if ( $this->isVirtualUrl( $path ) ) {
1162 return $this->resolveVirtualUrl( $path );
1163 }
1164 return $path;
1165 }
1166
1167 /**
1168 * Get a local FS copy of a file with a given virtual URL/storage path.
1169 * Temporary files may be purged when the file object falls out of scope.
1170 *
1171 * @param $virtualUrl string
1172 * @return TempFSFile|null Returns null on failure
1173 */
1174 public function getLocalCopy( $virtualUrl ) {
1175 $path = $this->resolveToStoragePath( $virtualUrl );
1176 return $this->backend->getLocalCopy( array( 'src' => $path ) );
1177 }
1178
1179 /**
1180 * Get a local FS file with a given virtual URL/storage path.
1181 * The file is either an original or a copy. It should not be changed.
1182 * Temporary files may be purged when the file object falls out of scope.
1183 *
1184 * @param $virtualUrl string
1185 * @return FSFile|null Returns null on failure.
1186 */
1187 public function getLocalReference( $virtualUrl ) {
1188 $path = $this->resolveToStoragePath( $virtualUrl );
1189 return $this->backend->getLocalReference( array( 'src' => $path ) );
1190 }
1191
1192 /**
1193 * Get properties of a file with a given virtual URL/storage path.
1194 * Properties should ultimately be obtained via FSFile::getProps().
1195 *
1196 * @param $virtualUrl string
1197 * @return Array
1198 */
1199 public function getFileProps( $virtualUrl ) {
1200 $path = $this->resolveToStoragePath( $virtualUrl );
1201 return $this->backend->getFileProps( array( 'src' => $path ) );
1202 }
1203
1204 /**
1205 * Get the timestamp of a file with a given virtual URL/storage path
1206 *
1207 * @param $virtualUrl string
1208 * @return string|bool False on failure
1209 */
1210 public function getFileTimestamp( $virtualUrl ) {
1211 $path = $this->resolveToStoragePath( $virtualUrl );
1212 return $this->backend->getFileTimestamp( array( 'src' => $path ) );
1213 }
1214
1215 /**
1216 * Get the sha1 of a file with a given virtual URL/storage path
1217 *
1218 * @param $virtualUrl string
1219 * @return string|bool
1220 */
1221 public function getFileSha1( $virtualUrl ) {
1222 $path = $this->resolveToStoragePath( $virtualUrl );
1223 $tmpFile = $this->backend->getLocalReference( array( 'src' => $path ) );
1224 if ( !$tmpFile ) {
1225 return false;
1226 }
1227 return $tmpFile->getSha1Base36();
1228 }
1229
1230 /**
1231 * Attempt to stream a file with the given virtual URL/storage path
1232 *
1233 * @param $virtualUrl string
1234 * @param $headers Array Additional HTTP headers to send on success
1235 * @return bool Success
1236 */
1237 public function streamFile( $virtualUrl, $headers = array() ) {
1238 $path = $this->resolveToStoragePath( $virtualUrl );
1239 $params = array( 'src' => $path, 'headers' => $headers );
1240 return $this->backend->streamFile( $params )->isOK();
1241 }
1242
1243 /**
1244 * Call a callback function for every public regular file in the repository.
1245 * This only acts on the current version of files, not any old versions.
1246 * May use either the database or the filesystem.
1247 *
1248 * @param $callback Array|string
1249 * @return void
1250 */
1251 public function enumFiles( $callback ) {
1252 $this->enumFilesInStorage( $callback );
1253 }
1254
1255 /**
1256 * Call a callback function for every public file in the repository.
1257 * May use either the database or the filesystem.
1258 *
1259 * @param $callback Array|string
1260 * @return void
1261 */
1262 protected function enumFilesInStorage( $callback ) {
1263 $publicRoot = $this->getZonePath( 'public' );
1264 $numDirs = 1 << ( $this->hashLevels * 4 );
1265 // Use a priori assumptions about directory structure
1266 // to reduce the tree height of the scanning process.
1267 for ( $flatIndex = 0; $flatIndex < $numDirs; $flatIndex++ ) {
1268 $hexString = sprintf( "%0{$this->hashLevels}x", $flatIndex );
1269 $path = $publicRoot;
1270 for ( $hexPos = 0; $hexPos < $this->hashLevels; $hexPos++ ) {
1271 $path .= '/' . substr( $hexString, 0, $hexPos + 1 );
1272 }
1273 $iterator = $this->backend->getFileList( array( 'dir' => $path ) );
1274 foreach ( $iterator as $name ) {
1275 // Each item returned is a public file
1276 call_user_func( $callback, "{$path}/{$name}" );
1277 }
1278 }
1279 }
1280
1281 /**
1282 * Determine if a relative path is valid, i.e. not blank or involving directory traveral
1283 *
1284 * @param $filename string
1285 * @return bool
1286 */
1287 public function validateFilename( $filename ) {
1288 if ( strval( $filename ) == '' ) {
1289 return false;
1290 }
1291 return FileBackend::isPathTraversalFree( $filename );
1292 }
1293
1294 /**
1295 * Get a callback function to use for cleaning error message parameters
1296 *
1297 * @return Array
1298 */
1299 function getErrorCleanupFunction() {
1300 switch ( $this->pathDisclosureProtection ) {
1301 case 'none':
1302 case 'simple': // b/c
1303 $callback = array( $this, 'passThrough' );
1304 break;
1305 default: // 'paranoid'
1306 $callback = array( $this, 'paranoidClean' );
1307 }
1308 return $callback;
1309 }
1310
1311 /**
1312 * Path disclosure protection function
1313 *
1314 * @param $param string
1315 * @return string
1316 */
1317 function paranoidClean( $param ) {
1318 return '[hidden]';
1319 }
1320
1321 /**
1322 * Path disclosure protection function
1323 *
1324 * @param $param string
1325 * @return string
1326 */
1327 function passThrough( $param ) {
1328 return $param;
1329 }
1330
1331 /**
1332 * Create a new fatal error
1333 *
1334 * @return FileRepoStatus
1335 */
1336 public function newFatal( $message /*, parameters...*/ ) {
1337 $params = func_get_args();
1338 array_unshift( $params, $this );
1339 return MWInit::callStaticMethod( 'FileRepoStatus', 'newFatal', $params );
1340 }
1341
1342 /**
1343 * Create a new good result
1344 *
1345 * @return FileRepoStatus
1346 */
1347 public function newGood( $value = null ) {
1348 return FileRepoStatus::newGood( $this, $value );
1349 }
1350
1351 /**
1352 * Checks if there is a redirect named as $title. If there is, return the
1353 * title object. If not, return false.
1354 * STUB
1355 *
1356 * @param $title Title of image
1357 * @return Bool
1358 */
1359 public function checkRedirect( Title $title ) {
1360 return false;
1361 }
1362
1363 /**
1364 * Invalidates image redirect cache related to that image
1365 * Doesn't do anything for repositories that don't support image redirects.
1366 *
1367 * STUB
1368 * @param $title Title of image
1369 */
1370 public function invalidateImageRedirect( Title $title ) {}
1371
1372 /**
1373 * Get the human-readable name of the repo
1374 *
1375 * @return string
1376 */
1377 public function getDisplayName() {
1378 // We don't name our own repo, return nothing
1379 if ( $this->isLocal() ) {
1380 return null;
1381 }
1382 // 'shared-repo-name-wikimediacommons' is used when $wgUseInstantCommons = true
1383 return wfMessageFallback( 'shared-repo-name-' . $this->name, 'shared-repo' )->text();
1384 }
1385
1386 /**
1387 * Returns true if this the local file repository.
1388 *
1389 * @return bool
1390 */
1391 public function isLocal() {
1392 return $this->getName() == 'local';
1393 }
1394
1395 /**
1396 * Get a key on the primary cache for this repository.
1397 * Returns false if the repository's cache is not accessible at this site.
1398 * The parameters are the parts of the key, as for wfMemcKey().
1399 *
1400 * STUB
1401 * @return bool
1402 */
1403 public function getSharedCacheKey( /*...*/ ) {
1404 return false;
1405 }
1406
1407 /**
1408 * Get a key for this repo in the local cache domain. These cache keys are
1409 * not shared with remote instances of the repo.
1410 * The parameters are the parts of the key, as for wfMemcKey().
1411 *
1412 * @return string
1413 */
1414 public function getLocalCacheKey( /*...*/ ) {
1415 $args = func_get_args();
1416 array_unshift( $args, 'filerepo', $this->getName() );
1417 return call_user_func_array( 'wfMemcKey', $args );
1418 }
1419
1420 /**
1421 * Get an temporary FileRepo associated with this repo.
1422 * Files will be created in the temp zone of this repo and
1423 * thumbnails in a /temp subdirectory in thumb zone of this repo.
1424 * It will have the same backend as this repo.
1425 *
1426 * @return TempFileRepo
1427 */
1428 public function getTempRepo() {
1429 return new TempFileRepo( array(
1430 'name' => "{$this->name}-temp",
1431 'backend' => $this->backend,
1432 'zones' => array(
1433 'public' => array(
1434 'container' => $this->zones['temp']['container'],
1435 'directory' => $this->zones['temp']['directory']
1436 ),
1437 'thumb' => array(
1438 'container' => $this->zones['thumb']['container'],
1439 'directory' => ( $this->zones['thumb']['directory'] == '' )
1440 ? 'temp'
1441 : $this->zones['thumb']['directory'] . '/temp'
1442 )
1443 ),
1444 'url' => $this->getZoneUrl( 'temp' ),
1445 'thumbUrl' => $this->getZoneUrl( 'thumb' ) . '/temp',
1446 'hashLevels' => $this->hashLevels // performance
1447 ) );
1448 }
1449
1450 /**
1451 * Get an UploadStash associated with this repo.
1452 *
1453 * @return UploadStash
1454 */
1455 public function getUploadStash() {
1456 return new UploadStash( $this );
1457 }
1458
1459 /**
1460 * Throw an exception if this repo is read-only by design.
1461 * This does not and should not check getReadOnlyReason().
1462 *
1463 * @return void
1464 * @throws MWException
1465 */
1466 protected function assertWritableRepo() {}
1467 }
1468
1469 /**
1470 * FileRepo for temporary files created via FileRepo::getTempRepo()
1471 */
1472 class TempFileRepo extends FileRepo {
1473 public function getTempRepo() {
1474 throw new MWException( "Cannot get a temp repo from a temp repo." );
1475 }
1476 }