7e92b4db12323f6d8b53c46dc454bf20268513f9
[lhc/web/wiklou.git] / includes / filerepo / FileRepo.php
1 <?php
2 /**
3 * Base code for file repositories.
4 *
5 * @file
6 * @ingroup FileRepo
7 */
8
9 /**
10 * Base class for file repositories
11 *
12 * @ingroup FileRepo
13 */
14 class FileRepo {
15 const FILES_ONLY = 1;
16
17 const DELETE_SOURCE = 1;
18 const OVERWRITE = 2;
19 const OVERWRITE_SAME = 4;
20 const SKIP_LOCKING = 8;
21
22 /** @var FileBackendBase */
23 protected $backend;
24 /** @var Array Map of zones to config */
25 protected $zones = array();
26
27 var $thumbScriptUrl, $transformVia404;
28 var $descBaseUrl, $scriptDirUrl, $scriptExtension, $articleUrl;
29 var $fetchDescription, $initialCapital;
30 var $pathDisclosureProtection = 'simple'; // 'paranoid'
31 var $descriptionCacheExpiry, $url, $thumbUrl;
32 var $hashLevels, $deletedHashLevels;
33
34 /**
35 * Factory functions for creating new files
36 * Override these in the base class
37 */
38 var $fileFactory = array( 'UnregisteredLocalFile', 'newFromTitle' );
39 var $oldFileFactory = false;
40 var $fileFactoryKey = false, $oldFileFactoryKey = false;
41
42 function __construct( $info ) {
43 // Required settings
44 $this->name = $info['name'];
45 if ( $info['backend'] instanceof FileBackendBase ) {
46 $this->backend = $info['backend']; // useful for testing
47 } else {
48 $this->backend = FileBackendGroup::singleton()->get( $info['backend'] );
49 }
50
51 // Optional settings that can have no value
52 $optionalSettings = array(
53 'descBaseUrl', 'scriptDirUrl', 'articleUrl', 'fetchDescription',
54 'thumbScriptUrl', 'pathDisclosureProtection', 'descriptionCacheExpiry',
55 'scriptExtension'
56 );
57 foreach ( $optionalSettings as $var ) {
58 if ( isset( $info[$var] ) ) {
59 $this->$var = $info[$var];
60 }
61 }
62
63 // Optional settings that have a default
64 $this->initialCapital = isset( $info['initialCapital'] )
65 ? $info['initialCapital']
66 : MWNamespace::isCapitalized( NS_FILE );
67 $this->url = isset( $info['url'] )
68 ? $info['url']
69 : false; // a subclass may set the URL (e.g. ForeignAPIRepo)
70 if ( isset( $info['thumbUrl'] ) ) {
71 $this->thumbUrl = $info['thumbUrl'];
72 } else {
73 $this->thumbUrl = $this->url ? "{$this->url}/thumb" : false;
74 }
75 $this->hashLevels = isset( $info['hashLevels'] )
76 ? $info['hashLevels']
77 : 2;
78 $this->deletedHashLevels = isset( $info['deletedHashLevels'] )
79 ? $info['deletedHashLevels']
80 : $this->hashLevels;
81 $this->transformVia404 = !empty( $info['transformVia404'] );
82 $this->zones = isset( $info['zones'] )
83 ? $info['zones']
84 : array();
85 // Give defaults for the basic zones...
86 foreach ( array( 'public', 'thumb', 'temp', 'deleted' ) as $zone ) {
87 if ( !isset( $this->zones[$zone] ) ) {
88 $this->zones[$zone] = array(
89 'container' => "{$this->name}-{$zone}",
90 'directory' => '' // container root
91 );
92 }
93 }
94 }
95
96 /**
97 * Get the file backend instance
98 *
99 * @return FileBackendBase
100 */
101 public function getBackend() {
102 return $this->backend;
103 }
104
105 /**
106 * Prepare a single zone or list of zones for usage.
107 * See initDeletedDir() for additional setup needed for the 'deleted' zone.
108 *
109 * @param $doZones Array Only do a particular zones
110 * @return Status
111 */
112 protected function initZones( $doZones = array() ) {
113 $status = $this->newGood();
114 foreach ( (array)$doZones as $zone ) {
115 $root = $this->getZonePath( $zone );
116 if ( $root === null ) {
117 throw new MWException( "No '$zone' zone defined in the {$this->name} repo." );
118 }
119 }
120 return $status;
121 }
122
123 /**
124 * Take all available measures to prevent web accessibility of new deleted
125 * directories, in case the user has not configured offline storage
126 *
127 * @param $dir string
128 * @return void
129 */
130 protected function initDeletedDir( $dir ) {
131 $this->backend->secure( // prevent web access & dir listings
132 array( 'dir' => $dir, 'noAccess' => true, 'noListing' => true ) );
133 }
134
135 /**
136 * Determine if a string is an mwrepo:// URL
137 *
138 * @param $url string
139 * @return bool
140 */
141 public static function isVirtualUrl( $url ) {
142 return substr( $url, 0, 9 ) == 'mwrepo://';
143 }
144
145 /**
146 * Get a URL referring to this repository, with the private mwrepo protocol.
147 * The suffix, if supplied, is considered to be unencoded, and will be
148 * URL-encoded before being returned.
149 *
150 * @param $suffix string
151 * @return string
152 */
153 public function getVirtualUrl( $suffix = false ) {
154 $path = 'mwrepo://' . $this->name;
155 if ( $suffix !== false ) {
156 $path .= '/' . rawurlencode( $suffix );
157 }
158 return $path;
159 }
160
161 /**
162 * Get the URL corresponding to one of the four basic zones
163 *
164 * @param $zone String: one of: public, deleted, temp, thumb
165 * @return String or false
166 */
167 public function getZoneUrl( $zone ) {
168 switch ( $zone ) {
169 case 'public':
170 return $this->url;
171 case 'temp':
172 return "{$this->url}/temp";
173 case 'deleted':
174 return false; // no public URL
175 case 'thumb':
176 return $this->thumbUrl;
177 default:
178 return false;
179 }
180 }
181
182 /**
183 * Get the backend storage path corresponding to a virtual URL
184 *
185 * @param $url string
186 * @return string
187 */
188 function resolveVirtualUrl( $url ) {
189 if ( substr( $url, 0, 9 ) != 'mwrepo://' ) {
190 throw new MWException( __METHOD__.': unknown protocol' );
191 }
192 $bits = explode( '/', substr( $url, 9 ), 3 );
193 if ( count( $bits ) != 3 ) {
194 throw new MWException( __METHOD__.": invalid mwrepo URL: $url" );
195 }
196 list( $repo, $zone, $rel ) = $bits;
197 if ( $repo !== $this->name ) {
198 throw new MWException( __METHOD__.": fetching from a foreign repo is not supported" );
199 }
200 $base = $this->getZonePath( $zone );
201 if ( !$base ) {
202 throw new MWException( __METHOD__.": invalid zone: $zone" );
203 }
204 return $base . '/' . rawurldecode( $rel );
205 }
206
207 /**
208 * The the storage container and base path of a zone
209 *
210 * @param $zone string
211 * @return Array (container, base path) or (null, null)
212 */
213 protected function getZoneLocation( $zone ) {
214 if ( !isset( $this->zones[$zone] ) ) {
215 return array( null, null ); // bogus
216 }
217 return array( $this->zones[$zone]['container'], $this->zones[$zone]['directory'] );
218 }
219
220 /**
221 * Get the storage path corresponding to one of the zones
222 *
223 * @param $zone string
224 * @return string|null
225 */
226 public function getZonePath( $zone ) {
227 list( $container, $base ) = $this->getZoneLocation( $zone );
228 if ( $container === null || $base === null ) {
229 return null;
230 }
231 $backendName = $this->backend->getName();
232 if ( $base != '' ) { // may not be set
233 $base = "/{$base}";
234 }
235 return "mwstore://$backendName/{$container}{$base}";
236 }
237
238 /**
239 * Create a new File object from the local repository
240 *
241 * @param $title Mixed: Title object or string
242 * @param $time Mixed: Time at which the image was uploaded.
243 * If this is specified, the returned object will be an
244 * instance of the repository's old file class instead of a
245 * current file. Repositories not supporting version control
246 * should return false if this parameter is set.
247 * @return File|null A File, or null if passed an invalid Title
248 */
249 public function newFile( $title, $time = false ) {
250 $title = File::normalizeTitle( $title );
251 if ( !$title ) {
252 return null;
253 }
254 if ( $time ) {
255 if ( $this->oldFileFactory ) {
256 return call_user_func( $this->oldFileFactory, $title, $this, $time );
257 } else {
258 return false;
259 }
260 } else {
261 return call_user_func( $this->fileFactory, $title, $this );
262 }
263 }
264
265 /**
266 * Find an instance of the named file created at the specified time
267 * Returns false if the file does not exist. Repositories not supporting
268 * version control should return false if the time is specified.
269 *
270 * @param $title Mixed: Title object or string
271 * @param $options array Associative array of options:
272 * time: requested time for an archived image, or false for the
273 * current version. An image object will be returned which was
274 * created at the specified time.
275 *
276 * ignoreRedirect: If true, do not follow file redirects
277 *
278 * private: If true, return restricted (deleted) files if the current
279 * user is allowed to view them. Otherwise, such files will not
280 * be found.
281 * @return File|false
282 */
283 public function findFile( $title, $options = array() ) {
284 $title = File::normalizeTitle( $title );
285 if ( !$title ) {
286 return false;
287 }
288 $time = isset( $options['time'] ) ? $options['time'] : false;
289 # First try the current version of the file to see if it precedes the timestamp
290 $img = $this->newFile( $title );
291 if ( !$img ) {
292 return false;
293 }
294 if ( $img->exists() && ( !$time || $img->getTimestamp() == $time ) ) {
295 return $img;
296 }
297 # Now try an old version of the file
298 if ( $time !== false ) {
299 $img = $this->newFile( $title, $time );
300 if ( $img && $img->exists() ) {
301 if ( !$img->isDeleted( File::DELETED_FILE ) ) {
302 return $img; // always OK
303 } elseif ( !empty( $options['private'] ) && $img->userCan( File::DELETED_FILE ) ) {
304 return $img;
305 }
306 }
307 }
308
309 # Now try redirects
310 if ( !empty( $options['ignoreRedirect'] ) ) {
311 return false;
312 }
313 $redir = $this->checkRedirect( $title );
314 if ( $redir && $title->getNamespace() == NS_FILE) {
315 $img = $this->newFile( $redir );
316 if ( !$img ) {
317 return false;
318 }
319 if ( $img->exists() ) {
320 $img->redirectedFrom( $title->getDBkey() );
321 return $img;
322 }
323 }
324 return false;
325 }
326
327 /**
328 * Find many files at once.
329 *
330 * @param $items An array of titles, or an array of findFile() options with
331 * the "title" option giving the title. Example:
332 *
333 * $findItem = array( 'title' => $title, 'private' => true );
334 * $findBatch = array( $findItem );
335 * $repo->findFiles( $findBatch );
336 * @return array
337 */
338 public function findFiles( $items ) {
339 $result = array();
340 foreach ( $items as $item ) {
341 if ( is_array( $item ) ) {
342 $title = $item['title'];
343 $options = $item;
344 unset( $options['title'] );
345 } else {
346 $title = $item;
347 $options = array();
348 }
349 $file = $this->findFile( $title, $options );
350 if ( $file ) {
351 $result[$file->getTitle()->getDBkey()] = $file;
352 }
353 }
354 return $result;
355 }
356
357 /**
358 * Find an instance of the file with this key, created at the specified time
359 * Returns false if the file does not exist. Repositories not supporting
360 * version control should return false if the time is specified.
361 *
362 * @param $sha1 String base 36 SHA-1 hash
363 * @param $options Option array, same as findFile().
364 * @return File|false
365 */
366 public function findFileFromKey( $sha1, $options = array() ) {
367 $time = isset( $options['time'] ) ? $options['time'] : false;
368
369 # First try to find a matching current version of a file...
370 if ( $this->fileFactoryKey ) {
371 $img = call_user_func( $this->fileFactoryKey, $sha1, $this, $time );
372 } else {
373 return false; // find-by-sha1 not supported
374 }
375 if ( $img && $img->exists() ) {
376 return $img;
377 }
378 # Now try to find a matching old version of a file...
379 if ( $time !== false && $this->oldFileFactoryKey ) { // find-by-sha1 supported?
380 $img = call_user_func( $this->oldFileFactoryKey, $sha1, $this, $time );
381 if ( $img && $img->exists() ) {
382 if ( !$img->isDeleted( File::DELETED_FILE ) ) {
383 return $img; // always OK
384 } elseif ( !empty( $options['private'] ) && $img->userCan( File::DELETED_FILE ) ) {
385 return $img;
386 }
387 }
388 }
389 return false;
390 }
391
392 /**
393 * Get an array or iterator of file objects for files that have a given
394 * SHA-1 content hash.
395 *
396 * STUB
397 */
398 public function findBySha1( $hash ) {
399 return array();
400 }
401
402 /**
403 * Get the public root URL of the repository
404 *
405 * @return string|false
406 */
407 public function getRootUrl() {
408 return $this->url;
409 }
410
411 /**
412 * Returns true if the repository uses a multi-level directory structure
413 *
414 * @return string
415 */
416 public function isHashed() {
417 return (bool)$this->hashLevels;
418 }
419
420 /**
421 * Get the URL of thumb.php
422 *
423 * @return string
424 */
425 public function getThumbScriptUrl() {
426 return $this->thumbScriptUrl;
427 }
428
429 /**
430 * Returns true if the repository can transform files via a 404 handler
431 *
432 * @return bool
433 */
434 public function canTransformVia404() {
435 return $this->transformVia404;
436 }
437
438 /**
439 * Get the name of an image from its title object
440 *
441 * @param $title Title
442 */
443 public function getNameFromTitle( Title $title ) {
444 global $wgContLang;
445 if ( $this->initialCapital != MWNamespace::isCapitalized( NS_FILE ) ) {
446 $name = $title->getUserCaseDBKey();
447 if ( $this->initialCapital ) {
448 $name = $wgContLang->ucfirst( $name );
449 }
450 } else {
451 $name = $title->getDBkey();
452 }
453 return $name;
454 }
455
456 /**
457 * Get the public zone root storage directory of the repository
458 *
459 * @return string
460 */
461 public function getRootDirectory() {
462 return $this->getZonePath( 'public' );
463 }
464
465 /**
466 * Get a relative path including trailing slash, e.g. f/fa/
467 * If the repo is not hashed, returns an empty string
468 *
469 * @param $name string
470 * @return string
471 */
472 public function getHashPath( $name ) {
473 return self::getHashPathForLevel( $name, $this->hashLevels );
474 }
475
476 /**
477 * @param $name
478 * @param $levels
479 * @return string
480 */
481 static function getHashPathForLevel( $name, $levels ) {
482 if ( $levels == 0 ) {
483 return '';
484 } else {
485 $hash = md5( $name );
486 $path = '';
487 for ( $i = 1; $i <= $levels; $i++ ) {
488 $path .= substr( $hash, 0, $i ) . '/';
489 }
490 return $path;
491 }
492 }
493
494 /**
495 * Get the number of hash directory levels
496 *
497 * @return integer
498 */
499 public function getHashLevels() {
500 return $this->hashLevels;
501 }
502
503 /**
504 * Get the name of this repository, as specified by $info['name]' to the constructor
505 *
506 * @return string
507 */
508 public function getName() {
509 return $this->name;
510 }
511
512 /**
513 * Make an url to this repo
514 *
515 * @param $query mixed Query string to append
516 * @param $entry string Entry point; defaults to index
517 * @return string|false
518 */
519 public function makeUrl( $query = '', $entry = 'index' ) {
520 if ( isset( $this->scriptDirUrl ) ) {
521 $ext = isset( $this->scriptExtension ) ? $this->scriptExtension : '.php';
522 return wfAppendQuery( "{$this->scriptDirUrl}/{$entry}{$ext}", $query );
523 }
524 return false;
525 }
526
527 /**
528 * Get the URL of an image description page. May return false if it is
529 * unknown or not applicable. In general this should only be called by the
530 * File class, since it may return invalid results for certain kinds of
531 * repositories. Use File::getDescriptionUrl() in user code.
532 *
533 * In particular, it uses the article paths as specified to the repository
534 * constructor, whereas local repositories use the local Title functions.
535 *
536 * @param $name string
537 * @return string
538 */
539 public function getDescriptionUrl( $name ) {
540 $encName = wfUrlencode( $name );
541 if ( !is_null( $this->descBaseUrl ) ) {
542 # "http://example.com/wiki/Image:"
543 return $this->descBaseUrl . $encName;
544 }
545 if ( !is_null( $this->articleUrl ) ) {
546 # "http://example.com/wiki/$1"
547 #
548 # We use "Image:" as the canonical namespace for
549 # compatibility across all MediaWiki versions.
550 return str_replace( '$1',
551 "Image:$encName", $this->articleUrl );
552 }
553 if ( !is_null( $this->scriptDirUrl ) ) {
554 # "http://example.com/w"
555 #
556 # We use "Image:" as the canonical namespace for
557 # compatibility across all MediaWiki versions,
558 # and just sort of hope index.php is right. ;)
559 return $this->makeUrl( "title=Image:$encName" );
560 }
561 return false;
562 }
563
564 /**
565 * Get the URL of the content-only fragment of the description page. For
566 * MediaWiki this means action=render. This should only be called by the
567 * repository's file class, since it may return invalid results. User code
568 * should use File::getDescriptionText().
569 *
570 * @param $name String: name of image to fetch
571 * @param $lang String: language to fetch it in, if any.
572 * @return string
573 */
574 public function getDescriptionRenderUrl( $name, $lang = null ) {
575 $query = 'action=render';
576 if ( !is_null( $lang ) ) {
577 $query .= '&uselang=' . $lang;
578 }
579 if ( isset( $this->scriptDirUrl ) ) {
580 return $this->makeUrl(
581 'title=' .
582 wfUrlencode( 'Image:' . $name ) .
583 "&$query" );
584 } else {
585 $descUrl = $this->getDescriptionUrl( $name );
586 if ( $descUrl ) {
587 return wfAppendQuery( $descUrl, $query );
588 } else {
589 return false;
590 }
591 }
592 }
593
594 /**
595 * Get the URL of the stylesheet to apply to description pages
596 *
597 * @return string|false
598 */
599 public function getDescriptionStylesheetUrl() {
600 if ( isset( $this->scriptDirUrl ) ) {
601 return $this->makeUrl( 'title=MediaWiki:Filepage.css&' .
602 wfArrayToCGI( Skin::getDynamicStylesheetQuery() ) );
603 }
604 return false;
605 }
606
607 /**
608 * Store a file to a given destination.
609 *
610 * @param $srcPath String: source FS path, storage path, or virtual URL
611 * @param $dstZone String: destination zone
612 * @param $dstRel String: destination relative path
613 * @param $flags Integer: bitwise combination of the following flags:
614 * self::DELETE_SOURCE Delete the source file after upload
615 * self::OVERWRITE Overwrite an existing destination file instead of failing
616 * self::OVERWRITE_SAME Overwrite the file if the destination exists and has the
617 * same contents as the source
618 * self::SKIP_LOCKING Skip any file locking when doing the store
619 * @return FileRepoStatus
620 */
621 public function store( $srcPath, $dstZone, $dstRel, $flags = 0 ) {
622 $status = $this->storeBatch( array( array( $srcPath, $dstZone, $dstRel ) ), $flags );
623 if ( $status->successCount == 0 ) {
624 $status->ok = false;
625 }
626 return $status;
627 }
628
629 /**
630 * Store a batch of files
631 *
632 * @param $triplets Array: (src, dest zone, dest rel) triplets as per store()
633 * @param $flags Integer: bitwise combination of the following flags:
634 * self::DELETE_SOURCE Delete the source file after upload
635 * self::OVERWRITE Overwrite an existing destination file instead of failing
636 * self::OVERWRITE_SAME Overwrite the file if the destination exists and has the
637 * same contents as the source
638 * self::SKIP_LOCKING Skip any file locking when doing the store
639 * @return FileRepoStatus
640 */
641 public function storeBatch( $triplets, $flags = 0 ) {
642 $backend = $this->backend; // convenience
643
644 $status = $this->newGood();
645
646 $operations = array();
647 $sourceFSFilesToDelete = array(); // cleanup for disk source files
648 // Validate each triplet and get the store operation...
649 foreach ( $triplets as $triplet ) {
650 list( $srcPath, $dstZone, $dstRel ) = $triplet;
651
652 // Resolve destination path
653 $root = $this->getZonePath( $dstZone );
654 if ( !$root ) {
655 throw new MWException( "Invalid zone: $dstZone" );
656 }
657 if ( !$this->validateFilename( $dstRel ) ) {
658 throw new MWException( 'Validation error in $dstRel' );
659 }
660 $dstPath = "$root/$dstRel";
661 $dstDir = dirname( $dstPath );
662
663 // Create destination directories for this triplet
664 if ( !$backend->prepare( array( 'dir' => $dstDir ) )->isOK() ) {
665 return $this->newFatal( 'directorycreateerror', $dstDir );
666 }
667
668 if ( $dstZone == 'deleted' ) {
669 $this->initDeletedDir( $dstDir );
670 }
671
672 // Resolve source to a storage path if virtual
673 if ( self::isVirtualUrl( $srcPath ) ) {
674 $srcPath = $this->resolveVirtualUrl( $srcPath );
675 }
676
677 // Get the appropriate file operation
678 if ( FileBackend::isStoragePath( $srcPath ) ) {
679 $opName = ( $flags & self::DELETE_SOURCE ) ? 'move' : 'copy';
680 } else {
681 $opName = 'store';
682 if ( $flags & self::DELETE_SOURCE ) {
683 $sourceFSFilesToDelete[] = $srcPath;
684 }
685 }
686 $operations[] = array(
687 'op' => $opName,
688 'src' => $srcPath,
689 'dst' => $dstPath,
690 'overwrite' => $flags & self::OVERWRITE,
691 'overwriteSame' => $flags & self::OVERWRITE_SAME,
692 );
693 }
694
695 // Execute the store operation for each triplet
696 $opts = array( 'force' => true );
697 if ( $flags & self::SKIP_LOCKING ) {
698 $opts['nonLocking'] = true;
699 }
700 $status->merge( $backend->doOperations( $operations, $opts ) );
701 // Cleanup for disk source files...
702 foreach ( $sourceFSFilesToDelete as $file ) {
703 wfSuppressWarnings();
704 unlink( $file ); // FS cleanup
705 wfRestoreWarnings();
706 }
707
708 return $status;
709 }
710
711 /**
712 * Deletes a batch of files.
713 * Each file can be a (zone, rel) pair, virtual url, storage path, or FS path.
714 * It will try to delete each file, but ignores any errors that may occur.
715 *
716 * @param $pairs array List of files to delete
717 * @return void
718 */
719 public function cleanupBatch( $files ) {
720 $operations = array();
721 $sourceFSFilesToDelete = array(); // cleanup for disk source files
722 foreach ( $files as $file ) {
723 if ( is_array( $file ) ) {
724 // This is a pair, extract it
725 list( $zone, $rel ) = $file;
726 $root = $this->getZonePath( $zone );
727 $path = "$root/$rel";
728 } else {
729 if ( self::isVirtualUrl( $file ) ) {
730 // This is a virtual url, resolve it
731 $path = $this->resolveVirtualUrl( $file );
732 } else {
733 // This is a full file name
734 $path = $file;
735 }
736 }
737 // Get a file operation if needed
738 if ( FileBackend::isStoragePath( $path ) ) {
739 $operations[] = array(
740 'op' => 'delete',
741 'src' => $path,
742 );
743 } else {
744 $sourceFSFilesToDelete[] = $path;
745 }
746 }
747 // Actually delete files from storage...
748 $opts = array( 'force' => true );
749 $this->backend->doOperations( $operations, $opts );
750 // Cleanup for disk source files...
751 foreach ( $sourceFSFilesToDelete as $file ) {
752 wfSuppressWarnings();
753 unlink( $file ); // FS cleanup
754 wfRestoreWarnings();
755 }
756 }
757
758 /**
759 * Pick a random name in the temp zone and store a file to it.
760 * Returns a FileRepoStatus object with the URL in the value.
761 *
762 * @param $originalName String: the base name of the file as specified
763 * by the user. The file extension will be maintained.
764 * @param $srcPath String: the current location of the file.
765 * @return FileRepoStatus object with the URL in the value.
766 */
767 public function storeTemp( $originalName, $srcPath ) {
768 $date = gmdate( "YmdHis" );
769 $hashPath = $this->getHashPath( $originalName );
770 $dstRel = "{$hashPath}{$date}!{$originalName}";
771 $dstUrlRel = $hashPath . $date . '!' . rawurlencode( $originalName );
772
773 $result = $this->store( $srcPath, 'temp', $dstRel, self::SKIP_LOCKING );
774 $result->value = $this->getVirtualUrl( 'temp' ) . '/' . $dstUrlRel;
775 return $result;
776 }
777
778 /**
779 * Concatenate a list of files into a target file location.
780 *
781 * @param $srcPaths Array Ordered list of source virtual URLs/storage paths
782 * @param $dstPath String Target file system path
783 * @param $flags Integer: bitwise combination of the following flags:
784 * self::DELETE_SOURCE Delete the source files
785 * @return FileRepoStatus
786 */
787 function concatenate( $srcPaths, $dstPath, $flags = 0 ) {
788 $status = $this->newGood();
789
790 $sources = array();
791 $deleteOperations = array(); // post-concatenate ops
792 foreach ( $srcPaths as $srcPath ) {
793 // Resolve source to a storage path if virtual
794 $source = $this->resolveToStoragePath( $srcPath );
795 $sources[] = $source; // chunk to merge
796 if ( $flags & self::DELETE_SOURCE ) {
797 $deleteOperations[] = array( 'op' => 'delete', 'src' => $source );
798 }
799 }
800
801 // Concatenate the chunks into one FS file
802 $params = array( 'srcs' => $sources, 'dst' => $dstPath );
803 $status->merge( $this->backend->concatenate( $params ) );
804 if ( !$status->isOK() ) {
805 return $status;
806 }
807
808 // Delete the sources if required
809 if ( $deleteOperations ) {
810 $opts = array( 'force' => true );
811 $status->merge( $this->backend->doOperations( $deleteOperations, $opts ) );
812 }
813
814 // Make sure status is OK, despite any $deleteOperations fatals
815 $status->setResult( true );
816
817 return $status;
818 }
819
820 /**
821 * Remove a temporary file or mark it for garbage collection
822 *
823 * @param $virtualUrl String: the virtual URL returned by storeTemp
824 * @return Boolean: true on success, false on failure
825 */
826 public function freeTemp( $virtualUrl ) {
827 $temp = "mwrepo://{$this->name}/temp";
828 if ( substr( $virtualUrl, 0, strlen( $temp ) ) != $temp ) {
829 wfDebug( __METHOD__.": Invalid temp virtual URL\n" );
830 return false;
831 }
832 $path = $this->resolveVirtualUrl( $virtualUrl );
833 $op = array( 'op' => 'delete', 'src' => $path );
834 $status = $this->backend->doOperation( $op );
835 return $status->isOK();
836 }
837
838 /**
839 * Copy or move a file either from a storage path, virtual URL,
840 * or FS path, into this repository at the specified destination location.
841 *
842 * Returns a FileRepoStatus object. On success, the value contains "new" or
843 * "archived", to indicate whether the file was new with that name.
844 *
845 * @param $srcPath String: the source FS path, storage path, or URL
846 * @param $dstRel String: the destination relative path
847 * @param $archiveRel String: the relative path where the existing file is to
848 * be archived, if there is one. Relative to the public zone root.
849 * @param $flags Integer: bitfield, may be FileRepo::DELETE_SOURCE to indicate
850 * that the source file should be deleted if possible
851 */
852 public function publish( $srcPath, $dstRel, $archiveRel, $flags = 0 ) {
853 $status = $this->publishBatch( array( array( $srcPath, $dstRel, $archiveRel ) ), $flags );
854 if ( $status->successCount == 0 ) {
855 $status->ok = false;
856 }
857 if ( isset( $status->value[0] ) ) {
858 $status->value = $status->value[0];
859 } else {
860 $status->value = false;
861 }
862 return $status;
863 }
864
865 /**
866 * Publish a batch of files
867 *
868 * @param $triplets Array: (source, dest, archive) triplets as per publish()
869 * @param $flags Integer: bitfield, may be FileRepo::DELETE_SOURCE to indicate
870 * that the source files should be deleted if possible
871 * @return FileRepoStatus
872 */
873 public function publishBatch( $triplets, $flags = 0 ) {
874 $backend = $this->backend; // convenience
875
876 // Try creating directories
877 $status = $this->initZones( 'public' );
878 if ( !$status->isOK() ) {
879 return $status;
880 }
881
882 $status = $this->newGood( array() );
883
884 $operations = array();
885 $sourceFSFilesToDelete = array(); // cleanup for disk source files
886 // Validate each triplet and get the store operation...
887 foreach ( $triplets as $i => $triplet ) {
888 list( $srcPath, $dstRel, $archiveRel ) = $triplet;
889 // Resolve source to a storage path if virtual
890 if ( substr( $srcPath, 0, 9 ) == 'mwrepo://' ) {
891 $srcPath = $this->resolveVirtualUrl( $srcPath );
892 }
893 if ( !$this->validateFilename( $dstRel ) ) {
894 throw new MWException( 'Validation error in $dstRel' );
895 }
896 if ( !$this->validateFilename( $archiveRel ) ) {
897 throw new MWException( 'Validation error in $archiveRel' );
898 }
899
900 $publicRoot = $this->getZonePath( 'public' );
901 $dstPath = "$publicRoot/$dstRel";
902 $archivePath = "$publicRoot/$archiveRel";
903
904 $dstDir = dirname( $dstPath );
905 $archiveDir = dirname( $archivePath );
906 // Abort immediately on directory creation errors since they're likely to be repetitive
907 if ( !$backend->prepare( array( 'dir' => $dstDir ) )->isOK() ) {
908 return $this->newFatal( 'directorycreateerror', $dstDir );
909 }
910 if ( !$backend->prepare( array( 'dir' => $archiveDir ) )->isOK() ) {
911 return $this->newFatal( 'directorycreateerror', $archiveDir );
912 }
913
914 // Archive destination file if it exists
915 if ( $backend->fileExists( array( 'src' => $dstPath ) ) ) {
916 // Check if the archive file exists
917 // This is a sanity check to avoid data loss. In UNIX, the rename primitive
918 // unlinks the destination file if it exists. DB-based synchronisation in
919 // publishBatch's caller should prevent races. In Windows there's no
920 // problem because the rename primitive fails if the destination exists.
921 if ( $backend->fileExists( array( 'src' => $archivePath ) ) ) {
922 $operations[] = array( 'op' => 'null' );
923 continue;
924 } else {
925 $operations[] = array(
926 'op' => 'move',
927 'src' => $dstPath,
928 'dst' => $archivePath
929 );
930 }
931 $status->value[$i] = 'archived';
932 } else {
933 $status->value[$i] = 'new';
934 }
935 // Copy (or move) the source file to the destination
936 if ( FileBackend::isStoragePath( $srcPath ) ) {
937 if ( $flags & self::DELETE_SOURCE ) {
938 $operations[] = array(
939 'op' => 'move',
940 'src' => $srcPath,
941 'dst' => $dstPath
942 );
943 } else {
944 $operations[] = array(
945 'op' => 'copy',
946 'src' => $srcPath,
947 'dst' => $dstPath
948 );
949 }
950 } else { // FS source path
951 $operations[] = array(
952 'op' => 'store',
953 'src' => $srcPath,
954 'dst' => $dstPath
955 );
956 if ( $flags & self::DELETE_SOURCE ) {
957 $sourceFSFilesToDelete[] = $srcPath;
958 }
959 }
960 }
961
962 // Execute the operations for each triplet
963 $opts = array( 'force' => true );
964 $status->merge( $backend->doOperations( $operations, $opts ) );
965 // Cleanup for disk source files...
966 foreach ( $sourceFSFilesToDelete as $file ) {
967 wfSuppressWarnings();
968 unlink( $file ); // FS cleanup
969 wfRestoreWarnings();
970 }
971
972 return $status;
973 }
974
975 /**
976 * Checks existence of a a file
977 *
978 * @param $file Virtual URL (or storage path) of file to check
979 * @param $flags Integer: bitwise combination of the following flags:
980 * self::FILES_ONLY Mark file as existing only if it is a file (not directory)
981 * @return bool
982 */
983 public function fileExists( $file, $flags = 0 ) {
984 $result = $this->fileExistsBatch( array( $file ), $flags );
985 return $result[0];
986 }
987
988 /**
989 * Checks existence of an array of files.
990 *
991 * @param $files Array: Virtual URLs (or storage paths) of files to check
992 * @param $flags Integer: bitwise combination of the following flags:
993 * self::FILES_ONLY Mark file as existing only if it is a file (not directory)
994 * @return Either array of files and existence flags, or false
995 */
996 public function fileExistsBatch( $files, $flags = 0 ) {
997 $result = array();
998 foreach ( $files as $key => $file ) {
999 if ( self::isVirtualUrl( $file ) ) {
1000 $file = $this->resolveVirtualUrl( $file );
1001 }
1002 if ( FileBackend::isStoragePath( $file ) ) {
1003 $result[$key] = $this->backend->fileExists( array( 'src' => $file ) );
1004 } else {
1005 if ( $flags & self::FILES_ONLY ) {
1006 $result[$key] = is_file( $file ); // FS only
1007 } else {
1008 $result[$key] = file_exists( $file ); // FS only
1009 }
1010 }
1011 }
1012
1013 return $result;
1014 }
1015
1016 /**
1017 * Move a file to the deletion archive.
1018 * If no valid deletion archive exists, this may either delete the file
1019 * or throw an exception, depending on the preference of the repository
1020 *
1021 * @param $srcRel Mixed: relative path for the file to be deleted
1022 * @param $archiveRel Mixed: relative path for the archive location.
1023 * Relative to a private archive directory.
1024 * @return FileRepoStatus object
1025 */
1026 public function delete( $srcRel, $archiveRel ) {
1027 return $this->deleteBatch( array( array( $srcRel, $archiveRel ) ) );
1028 }
1029
1030 /**
1031 * Move a group of files to the deletion archive.
1032 *
1033 * If no valid deletion archive is configured, this may either delete the
1034 * file or throw an exception, depending on the preference of the repository.
1035 *
1036 * The overwrite policy is determined by the repository -- currently LocalRepo
1037 * assumes a naming scheme in the deleted zone based on content hash, as
1038 * opposed to the public zone which is assumed to be unique.
1039 *
1040 * @param $sourceDestPairs Array of source/destination pairs. Each element
1041 * is a two-element array containing the source file path relative to the
1042 * public root in the first element, and the archive file path relative
1043 * to the deleted zone root in the second element.
1044 * @return FileRepoStatus
1045 */
1046 public function deleteBatch( $sourceDestPairs ) {
1047 $backend = $this->backend; // convenience
1048
1049 // Try creating directories
1050 $status = $this->initZones( array( 'public', 'deleted' ) );
1051 if ( !$status->isOK() ) {
1052 return $status;
1053 }
1054
1055 $status = $this->newGood();
1056
1057 $operations = array();
1058 // Validate filenames and create archive directories
1059 foreach ( $sourceDestPairs as $pair ) {
1060 list( $srcRel, $archiveRel ) = $pair;
1061 if ( !$this->validateFilename( $srcRel ) ) {
1062 throw new MWException( __METHOD__.':Validation error in $srcRel' );
1063 }
1064 if ( !$this->validateFilename( $archiveRel ) ) {
1065 throw new MWException( __METHOD__.':Validation error in $archiveRel' );
1066 }
1067
1068 $publicRoot = $this->getZonePath( 'public' );
1069 $srcPath = "{$publicRoot}/$srcRel";
1070
1071 $deletedRoot = $this->getZonePath( 'deleted' );
1072 $archivePath = "{$deletedRoot}/{$archiveRel}";
1073 $archiveDir = dirname( $archivePath ); // does not touch FS
1074
1075 // Create destination directories
1076 if ( !$backend->prepare( array( 'dir' => $archiveDir ) )->isOK() ) {
1077 return $this->newFatal( 'directorycreateerror', $archiveDir );
1078 }
1079 $this->initDeletedDir( $archiveDir );
1080
1081 $operations[] = array(
1082 'op' => 'move',
1083 'src' => $srcPath,
1084 'dst' => $archivePath,
1085 // We may have 2+ identical files being deleted,
1086 // all of which will map to the same destination file
1087 'overwriteSame' => true
1088 );
1089 }
1090
1091 // Move the files by execute the operations for each pair.
1092 // We're now committed to returning an OK result, which will
1093 // lead to the files being moved in the DB also.
1094 $opts = array( 'force' => true );
1095 $status->merge( $backend->doOperations( $operations, $opts ) );
1096
1097 return $status;
1098 }
1099
1100 /**
1101 * Get a relative path for a deletion archive key,
1102 * e.g. s/z/a/ for sza251lrxrc1jad41h5mgilp8nysje52.jpg
1103 *
1104 * @return string
1105 */
1106 public function getDeletedHashPath( $key ) {
1107 $path = '';
1108 for ( $i = 0; $i < $this->deletedHashLevels; $i++ ) {
1109 $path .= $key[$i] . '/';
1110 }
1111 return $path;
1112 }
1113
1114 /**
1115 * If a path is a virtual URL, resolve it to a storage path.
1116 * Otherwise, just return the path as it is.
1117 *
1118 * @param $path string
1119 * @return string
1120 * @throws MWException
1121 */
1122 protected function resolveToStoragePath( $path ) {
1123 if ( $this->isVirtualUrl( $path ) ) {
1124 return $this->resolveVirtualUrl( $path );
1125 }
1126 return $path;
1127 }
1128
1129 /**
1130 * Get a local FS copy of a file with a given virtual URL/storage path.
1131 * Temporary files may be purged when the file object falls out of scope.
1132 *
1133 * @param $virtualUrl string
1134 * @return TempFSFile|null Returns null on failure
1135 */
1136 public function getLocalCopy( $virtualUrl ) {
1137 $path = $this->resolveToStoragePath( $virtualUrl );
1138 return $this->backend->getLocalCopy( array( 'src' => $path ) );
1139 }
1140
1141 /**
1142 * Get a local FS file with a given virtual URL/storage path.
1143 * The file is either an original or a copy. It should not be changed.
1144 * Temporary files may be purged when the file object falls out of scope.
1145 *
1146 * @param $virtualUrl string
1147 * @return FSFile|null Returns null on failure.
1148 */
1149 public function getLocalReference( $virtualUrl ) {
1150 $path = $this->resolveToStoragePath( $virtualUrl );
1151 return $this->backend->getLocalReference( array( 'src' => $path ) );
1152 }
1153
1154 /**
1155 * Get properties of a file with a given virtual URL/storage path.
1156 * Properties should ultimately be obtained via FSFile::getProps().
1157 *
1158 * @param $virtualUrl string
1159 * @return Array
1160 */
1161 public function getFileProps( $virtualUrl ) {
1162 $path = $this->resolveToStoragePath( $virtualUrl );
1163 return $this->backend->getFileProps( array( 'src' => $path ) );
1164 }
1165
1166 /**
1167 * Get the timestamp of a file with a given virtual URL/storage path
1168 *
1169 * @param $virtualUrl string
1170 * @return string|false
1171 */
1172 public function getFileTimestamp( $virtualUrl ) {
1173 $path = $this->resolveToStoragePath( $virtualUrl );
1174 return $this->backend->getFileTimestamp( array( 'src' => $path ) );
1175 }
1176
1177 /**
1178 * Get the sha1 of a file with a given virtual URL/storage path
1179 *
1180 * @param $virtualUrl string
1181 * @return string|false
1182 */
1183 public function getFileSha1( $virtualUrl ) {
1184 $path = $this->resolveToStoragePath( $virtualUrl );
1185 $tmpFile = $this->backend->getLocalReference( array( 'src' => $path ) );
1186 if ( !$tmpFile ) {
1187 return false;
1188 }
1189 return $tmpFile->getSha1Base36();
1190 }
1191
1192 /**
1193 * Attempt to stream a file with the given virtual URL/storage path
1194 *
1195 * @param $virtualUrl string
1196 * @param $headers Array Additional HTTP headers to send on success
1197 * @return bool Success
1198 */
1199 public function streamFile( $virtualUrl, $headers = array() ) {
1200 $path = $this->resolveToStoragePath( $virtualUrl );
1201 $params = array( 'src' => $path, 'headers' => $headers );
1202 return $this->backend->streamFile( $params )->isOK();
1203 }
1204
1205 /**
1206 * Call a callback function for every public regular file in the repository.
1207 * This only acts on the current version of files, not any old versions.
1208 * May use either the database or the filesystem.
1209 *
1210 * @param $callback Array|string
1211 * @return void
1212 */
1213 public function enumFiles( $callback ) {
1214 $this->enumFilesInStorage( $callback );
1215 }
1216
1217 /**
1218 * Call a callback function for every public file in the repository.
1219 * May use either the database or the filesystem.
1220 *
1221 * @param $callback Array|string
1222 * @return void
1223 */
1224 protected function enumFilesInStorage( $callback ) {
1225 $publicRoot = $this->getZonePath( 'public' );
1226 $numDirs = 1 << ( $this->hashLevels * 4 );
1227 // Use a priori assumptions about directory structure
1228 // to reduce the tree height of the scanning process.
1229 for ( $flatIndex = 0; $flatIndex < $numDirs; $flatIndex++ ) {
1230 $hexString = sprintf( "%0{$this->hashLevels}x", $flatIndex );
1231 $path = $publicRoot;
1232 for ( $hexPos = 0; $hexPos < $this->hashLevels; $hexPos++ ) {
1233 $path .= '/' . substr( $hexString, 0, $hexPos + 1 );
1234 }
1235 $iterator = $this->backend->getFileList( array( 'dir' => $path ) );
1236 foreach ( $iterator as $name ) {
1237 // Each item returned is a public file
1238 call_user_func( $callback, "{$path}/{$name}" );
1239 }
1240 }
1241 }
1242
1243 /**
1244 * Determine if a relative path is valid, i.e. not blank or involving directory traveral
1245 *
1246 * @param $filename string
1247 * @return bool
1248 */
1249 public function validateFilename( $filename ) {
1250 if ( strval( $filename ) == '' ) {
1251 return false;
1252 }
1253 if ( wfIsWindows() ) {
1254 $filename = strtr( $filename, '\\', '/' );
1255 }
1256 /**
1257 * Use the same traversal protection as Title::secureAndSplit()
1258 */
1259 if ( strpos( $filename, '.' ) !== false &&
1260 ( $filename === '.' || $filename === '..' ||
1261 strpos( $filename, './' ) === 0 ||
1262 strpos( $filename, '../' ) === 0 ||
1263 strpos( $filename, '/./' ) !== false ||
1264 strpos( $filename, '/../' ) !== false ) )
1265 {
1266 return false;
1267 } else {
1268 return true;
1269 }
1270 }
1271
1272 /**
1273 * Get a callback function to use for cleaning error message parameters
1274 *
1275 * @return Array
1276 */
1277 function getErrorCleanupFunction() {
1278 switch ( $this->pathDisclosureProtection ) {
1279 case 'none':
1280 $callback = array( $this, 'passThrough' );
1281 break;
1282 case 'simple':
1283 $callback = array( $this, 'simpleClean' );
1284 break;
1285 default: // 'paranoid'
1286 $callback = array( $this, 'paranoidClean' );
1287 }
1288 return $callback;
1289 }
1290
1291 /**
1292 * Path disclosure protection function
1293 *
1294 * @param $param string
1295 * @return string
1296 */
1297 function paranoidClean( $param ) {
1298 return '[hidden]';
1299 }
1300
1301 /**
1302 * Path disclosure protection function
1303 *
1304 * @param $param string
1305 * @return string
1306 */
1307 function simpleClean( $param ) {
1308 global $IP;
1309 if ( !isset( $this->simpleCleanPairs ) ) {
1310 $this->simpleCleanPairs = array(
1311 $IP => '$IP', // sanity
1312 );
1313 }
1314 return strtr( $param, $this->simpleCleanPairs );
1315 }
1316
1317 /**
1318 * Path disclosure protection function
1319 *
1320 * @param $param string
1321 * @return string
1322 */
1323 function passThrough( $param ) {
1324 return $param;
1325 }
1326
1327 /**
1328 * Create a new fatal error
1329 *
1330 * @return FileRepoStatus
1331 */
1332 function newFatal( $message /*, parameters...*/ ) {
1333 $params = func_get_args();
1334 array_unshift( $params, $this );
1335 return MWInit::callStaticMethod( 'FileRepoStatus', 'newFatal', $params );
1336 }
1337
1338 /**
1339 * Create a new good result
1340 *
1341 * @return FileRepoStatus
1342 */
1343 function newGood( $value = null ) {
1344 return FileRepoStatus::newGood( $this, $value );
1345 }
1346
1347 /**
1348 * Delete files in the deleted directory if they are not referenced in the filearchive table
1349 *
1350 * STUB
1351 */
1352 public function cleanupDeletedBatch( $storageKeys ) {}
1353
1354 /**
1355 * Checks if there is a redirect named as $title. If there is, return the
1356 * title object. If not, return false.
1357 * STUB
1358 *
1359 * @param $title Title of image
1360 * @return Bool
1361 */
1362 public function checkRedirect( Title $title ) {
1363 return false;
1364 }
1365
1366 /**
1367 * Invalidates image redirect cache related to that image
1368 * Doesn't do anything for repositories that don't support image redirects.
1369 *
1370 * STUB
1371 * @param $title Title of image
1372 */
1373 public function invalidateImageRedirect( Title $title ) {}
1374
1375 /**
1376 * Get the human-readable name of the repo
1377 *
1378 * @return string
1379 */
1380 public function getDisplayName() {
1381 // We don't name our own repo, return nothing
1382 if ( $this->isLocal() ) {
1383 return null;
1384 }
1385 // 'shared-repo-name-wikimediacommons' is used when $wgUseInstantCommons = true
1386 return wfMessageFallback( 'shared-repo-name-' . $this->name, 'shared-repo' )->text();
1387 }
1388
1389 /**
1390 * Returns true if this the local file repository.
1391 *
1392 * @return bool
1393 */
1394 public function isLocal() {
1395 return $this->getName() == 'local';
1396 }
1397
1398 /**
1399 * Get a key on the primary cache for this repository.
1400 * Returns false if the repository's cache is not accessible at this site.
1401 * The parameters are the parts of the key, as for wfMemcKey().
1402 *
1403 * STUB
1404 */
1405 function getSharedCacheKey( /*...*/ ) {
1406 return false;
1407 }
1408
1409 /**
1410 * Get a key for this repo in the local cache domain. These cache keys are
1411 * not shared with remote instances of the repo.
1412 * The parameters are the parts of the key, as for wfMemcKey().
1413 *
1414 * @return string
1415 */
1416 function getLocalCacheKey( /*...*/ ) {
1417 $args = func_get_args();
1418 array_unshift( $args, 'filerepo', $this->getName() );
1419 return call_user_func_array( 'wfMemcKey', $args );
1420 }
1421
1422 /**
1423 * Get an UploadStash associated with this repo.
1424 *
1425 * @return UploadStash
1426 */
1427 public function getUploadStash() {
1428 return new UploadStash( $this );
1429 }
1430 }