Merge "Remove m prefixes from private variables"
[lhc/web/wiklou.git] / includes / filebackend / FileBackendStore.php
1 <?php
2 /**
3 * Base class for all backends using particular storage medium.
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
19 *
20 * @file
21 * @ingroup FileBackend
22 * @author Aaron Schulz
23 */
24
25 /**
26 * @brief Base class for all backends using particular storage medium.
27 *
28 * This class defines the methods as abstract that subclasses must implement.
29 * Outside callers should *not* use functions with "Internal" in the name.
30 *
31 * The FileBackend operations are implemented using basic functions
32 * such as storeInternal(), copyInternal(), deleteInternal() and the like.
33 * This class is also responsible for path resolution and sanitization.
34 *
35 * @ingroup FileBackend
36 * @since 1.19
37 */
38 abstract class FileBackendStore extends FileBackend {
39 /** @var BagOStuff */
40 protected $memCache;
41 /** @var ProcessCacheLRU Map of paths to small (RAM/disk) cache items */
42 protected $cheapCache;
43 /** @var ProcessCacheLRU Map of paths to large (RAM/disk) cache items */
44 protected $expensiveCache;
45
46 /** @var array Map of container names to sharding config */
47 protected $shardViaHashLevels = array();
48
49 /** @var callback Method to get the MIME type of files */
50 protected $mimeCallback;
51
52 protected $maxFileSize = 4294967296; // integer bytes (4GiB)
53
54 const CACHE_TTL = 10; // integer; TTL in seconds for process cache entries
55 const CACHE_CHEAP_SIZE = 500; // integer; max entries in "cheap cache"
56 const CACHE_EXPENSIVE_SIZE = 5; // integer; max entries in "expensive cache"
57
58 /**
59 * @see FileBackend::__construct()
60 * Additional $config params include:
61 * - mimeCallback : Callback that takes (storage path, content, file system path) and
62 * returns the MIME type of the file or 'unknown/unknown'. The file
63 * system path parameter should be used if the content one is null.
64 *
65 * @param array $config
66 */
67 public function __construct( array $config ) {
68 parent::__construct( $config );
69 $this->mimeCallback = isset( $config['mimeCallback'] )
70 ? $config['mimeCallback']
71 : function ( $storagePath, $content, $fsPath ) {
72 // @TODO: handle the case of extension-less files using the contents
73 return StreamFile::contentTypeFromPath( $storagePath ) ? : 'unknown/unknown';
74 };
75 $this->memCache = new EmptyBagOStuff(); // disabled by default
76 $this->cheapCache = new ProcessCacheLRU( self::CACHE_CHEAP_SIZE );
77 $this->expensiveCache = new ProcessCacheLRU( self::CACHE_EXPENSIVE_SIZE );
78 }
79
80 /**
81 * Get the maximum allowable file size given backend
82 * medium restrictions and basic performance constraints.
83 * Do not call this function from places outside FileBackend and FileOp.
84 *
85 * @return int Bytes
86 */
87 final public function maxFileSizeInternal() {
88 return $this->maxFileSize;
89 }
90
91 /**
92 * Check if a file can be created or changed at a given storage path.
93 * FS backends should check if the parent directory exists, files can be
94 * written under it, and that any file already there is writable.
95 * Backends using key/value stores should check if the container exists.
96 *
97 * @param string $storagePath
98 * @return bool
99 */
100 abstract public function isPathUsableInternal( $storagePath );
101
102 /**
103 * Create a file in the backend with the given contents.
104 * This will overwrite any file that exists at the destination.
105 * Do not call this function from places outside FileBackend and FileOp.
106 *
107 * $params include:
108 * - content : the raw file contents
109 * - dst : destination storage path
110 * - headers : HTTP header name/value map
111 * - async : Status will be returned immediately if supported.
112 * If the status is OK, then its value field will be
113 * set to a FileBackendStoreOpHandle object.
114 * - dstExists : Whether a file exists at the destination (optimization).
115 * Callers can use "false" if no existing file is being changed.
116 *
117 * @param array $params
118 * @return Status
119 */
120 final public function createInternal( array $params ) {
121 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
122 if ( strlen( $params['content'] ) > $this->maxFileSizeInternal() ) {
123 $status = Status::newFatal( 'backend-fail-maxsize',
124 $params['dst'], $this->maxFileSizeInternal() );
125 } else {
126 $status = $this->doCreateInternal( $params );
127 $this->clearCache( array( $params['dst'] ) );
128 if ( !isset( $params['dstExists'] ) || $params['dstExists'] ) {
129 $this->deleteFileCache( $params['dst'] ); // persistent cache
130 }
131 }
132
133 return $status;
134 }
135
136 /**
137 * @see FileBackendStore::createInternal()
138 * @param array $params
139 * @return Status
140 */
141 abstract protected function doCreateInternal( array $params );
142
143 /**
144 * Store a file into the backend from a file on disk.
145 * This will overwrite any file that exists at the destination.
146 * Do not call this function from places outside FileBackend and FileOp.
147 *
148 * $params include:
149 * - src : source path on disk
150 * - dst : destination storage path
151 * - headers : HTTP header name/value map
152 * - async : Status will be returned immediately if supported.
153 * If the status is OK, then its value field will be
154 * set to a FileBackendStoreOpHandle object.
155 * - dstExists : Whether a file exists at the destination (optimization).
156 * Callers can use "false" if no existing file is being changed.
157 *
158 * @param array $params
159 * @return Status
160 */
161 final public function storeInternal( array $params ) {
162 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
163 if ( filesize( $params['src'] ) > $this->maxFileSizeInternal() ) {
164 $status = Status::newFatal( 'backend-fail-maxsize',
165 $params['dst'], $this->maxFileSizeInternal() );
166 } else {
167 $status = $this->doStoreInternal( $params );
168 $this->clearCache( array( $params['dst'] ) );
169 if ( !isset( $params['dstExists'] ) || $params['dstExists'] ) {
170 $this->deleteFileCache( $params['dst'] ); // persistent cache
171 }
172 }
173
174 return $status;
175 }
176
177 /**
178 * @see FileBackendStore::storeInternal()
179 * @param array $params
180 * @return Status
181 */
182 abstract protected function doStoreInternal( array $params );
183
184 /**
185 * Copy a file from one storage path to another in the backend.
186 * This will overwrite any file that exists at the destination.
187 * Do not call this function from places outside FileBackend and FileOp.
188 *
189 * $params include:
190 * - src : source storage path
191 * - dst : destination storage path
192 * - ignoreMissingSource : do nothing if the source file does not exist
193 * - headers : HTTP header name/value map
194 * - async : Status will be returned immediately if supported.
195 * If the status is OK, then its value field will be
196 * set to a FileBackendStoreOpHandle object.
197 * - dstExists : Whether a file exists at the destination (optimization).
198 * Callers can use "false" if no existing file is being changed.
199 *
200 * @param array $params
201 * @return Status
202 */
203 final public function copyInternal( array $params ) {
204 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
205 $status = $this->doCopyInternal( $params );
206 $this->clearCache( array( $params['dst'] ) );
207 if ( !isset( $params['dstExists'] ) || $params['dstExists'] ) {
208 $this->deleteFileCache( $params['dst'] ); // persistent cache
209 }
210
211 return $status;
212 }
213
214 /**
215 * @see FileBackendStore::copyInternal()
216 * @param array $params
217 * @return Status
218 */
219 abstract protected function doCopyInternal( array $params );
220
221 /**
222 * Delete a file at the storage path.
223 * Do not call this function from places outside FileBackend and FileOp.
224 *
225 * $params include:
226 * - src : source storage path
227 * - ignoreMissingSource : do nothing if the source file does not exist
228 * - async : Status will be returned immediately if supported.
229 * If the status is OK, then its value field will be
230 * set to a FileBackendStoreOpHandle object.
231 *
232 * @param array $params
233 * @return Status
234 */
235 final public function deleteInternal( array $params ) {
236 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
237 $status = $this->doDeleteInternal( $params );
238 $this->clearCache( array( $params['src'] ) );
239 $this->deleteFileCache( $params['src'] ); // persistent cache
240 return $status;
241 }
242
243 /**
244 * @see FileBackendStore::deleteInternal()
245 * @param array $params
246 * @return Status
247 */
248 abstract protected function doDeleteInternal( array $params );
249
250 /**
251 * Move a file from one storage path to another in the backend.
252 * This will overwrite any file that exists at the destination.
253 * Do not call this function from places outside FileBackend and FileOp.
254 *
255 * $params include:
256 * - src : source storage path
257 * - dst : destination storage path
258 * - ignoreMissingSource : do nothing if the source file does not exist
259 * - headers : HTTP header name/value map
260 * - async : Status will be returned immediately if supported.
261 * If the status is OK, then its value field will be
262 * set to a FileBackendStoreOpHandle object.
263 * - dstExists : Whether a file exists at the destination (optimization).
264 * Callers can use "false" if no existing file is being changed.
265 *
266 * @param array $params
267 * @return Status
268 */
269 final public function moveInternal( array $params ) {
270 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
271 $status = $this->doMoveInternal( $params );
272 $this->clearCache( array( $params['src'], $params['dst'] ) );
273 $this->deleteFileCache( $params['src'] ); // persistent cache
274 if ( !isset( $params['dstExists'] ) || $params['dstExists'] ) {
275 $this->deleteFileCache( $params['dst'] ); // persistent cache
276 }
277
278 return $status;
279 }
280
281 /**
282 * @see FileBackendStore::moveInternal()
283 * @param array $params
284 * @return Status
285 */
286 protected function doMoveInternal( array $params ) {
287 unset( $params['async'] ); // two steps, won't work here :)
288 $nsrc = FileBackend::normalizeStoragePath( $params['src'] );
289 $ndst = FileBackend::normalizeStoragePath( $params['dst'] );
290 // Copy source to dest
291 $status = $this->copyInternal( $params );
292 if ( $nsrc !== $ndst && $status->isOK() ) {
293 // Delete source (only fails due to races or network problems)
294 $status->merge( $this->deleteInternal( array( 'src' => $params['src'] ) ) );
295 $status->setResult( true, $status->value ); // ignore delete() errors
296 }
297
298 return $status;
299 }
300
301 /**
302 * Alter metadata for a file at the storage path.
303 * Do not call this function from places outside FileBackend and FileOp.
304 *
305 * $params include:
306 * - src : source storage path
307 * - headers : HTTP header name/value map
308 * - async : Status will be returned immediately if supported.
309 * If the status is OK, then its value field will be
310 * set to a FileBackendStoreOpHandle object.
311 *
312 * @param array $params
313 * @return Status
314 */
315 final public function describeInternal( array $params ) {
316 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
317 if ( count( $params['headers'] ) ) {
318 $status = $this->doDescribeInternal( $params );
319 $this->clearCache( array( $params['src'] ) );
320 $this->deleteFileCache( $params['src'] ); // persistent cache
321 } else {
322 $status = Status::newGood(); // nothing to do
323 }
324
325 return $status;
326 }
327
328 /**
329 * @see FileBackendStore::describeInternal()
330 * @param array $params
331 * @return Status
332 */
333 protected function doDescribeInternal( array $params ) {
334 return Status::newGood();
335 }
336
337 /**
338 * No-op file operation that does nothing.
339 * Do not call this function from places outside FileBackend and FileOp.
340 *
341 * @param array $params
342 * @return Status
343 */
344 final public function nullInternal( array $params ) {
345 return Status::newGood();
346 }
347
348 final public function concatenate( array $params ) {
349 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
350 $status = Status::newGood();
351
352 // Try to lock the source files for the scope of this function
353 $scopeLockS = $this->getScopedFileLocks( $params['srcs'], LockManager::LOCK_UW, $status );
354 if ( $status->isOK() ) {
355 // Actually do the file concatenation...
356 $start_time = microtime( true );
357 $status->merge( $this->doConcatenate( $params ) );
358 $sec = microtime( true ) - $start_time;
359 if ( !$status->isOK() ) {
360 wfDebugLog( 'FileOperation', get_class( $this ) . " failed to concatenate " .
361 count( $params['srcs'] ) . " file(s) [$sec sec]" );
362 }
363 }
364
365 return $status;
366 }
367
368 /**
369 * @see FileBackendStore::concatenate()
370 * @param array $params
371 * @return Status
372 */
373 protected function doConcatenate( array $params ) {
374 $status = Status::newGood();
375 $tmpPath = $params['dst']; // convenience
376 unset( $params['latest'] ); // sanity
377
378 // Check that the specified temp file is valid...
379 wfSuppressWarnings();
380 $ok = ( is_file( $tmpPath ) && filesize( $tmpPath ) == 0 );
381 wfRestoreWarnings();
382 if ( !$ok ) { // not present or not empty
383 $status->fatal( 'backend-fail-opentemp', $tmpPath );
384
385 return $status;
386 }
387
388 // Get local FS versions of the chunks needed for the concatenation...
389 $fsFiles = $this->getLocalReferenceMulti( $params );
390 foreach ( $fsFiles as $path => &$fsFile ) {
391 if ( !$fsFile ) { // chunk failed to download?
392 $fsFile = $this->getLocalReference( array( 'src' => $path ) );
393 if ( !$fsFile ) { // retry failed?
394 $status->fatal( 'backend-fail-read', $path );
395
396 return $status;
397 }
398 }
399 }
400 unset( $fsFile ); // unset reference so we can reuse $fsFile
401
402 // Get a handle for the destination temp file
403 $tmpHandle = fopen( $tmpPath, 'ab' );
404 if ( $tmpHandle === false ) {
405 $status->fatal( 'backend-fail-opentemp', $tmpPath );
406
407 return $status;
408 }
409
410 // Build up the temp file using the source chunks (in order)...
411 foreach ( $fsFiles as $virtualSource => $fsFile ) {
412 // Get a handle to the local FS version
413 $sourceHandle = fopen( $fsFile->getPath(), 'rb' );
414 if ( $sourceHandle === false ) {
415 fclose( $tmpHandle );
416 $status->fatal( 'backend-fail-read', $virtualSource );
417
418 return $status;
419 }
420 // Append chunk to file (pass chunk size to avoid magic quotes)
421 if ( !stream_copy_to_stream( $sourceHandle, $tmpHandle ) ) {
422 fclose( $sourceHandle );
423 fclose( $tmpHandle );
424 $status->fatal( 'backend-fail-writetemp', $tmpPath );
425
426 return $status;
427 }
428 fclose( $sourceHandle );
429 }
430 if ( !fclose( $tmpHandle ) ) {
431 $status->fatal( 'backend-fail-closetemp', $tmpPath );
432
433 return $status;
434 }
435
436 clearstatcache(); // temp file changed
437
438 return $status;
439 }
440
441 final protected function doPrepare( array $params ) {
442 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
443 $status = Status::newGood();
444
445 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
446 if ( $dir === null ) {
447 $status->fatal( 'backend-fail-invalidpath', $params['dir'] );
448
449 return $status; // invalid storage path
450 }
451
452 if ( $shard !== null ) { // confined to a single container/shard
453 $status->merge( $this->doPrepareInternal( $fullCont, $dir, $params ) );
454 } else { // directory is on several shards
455 wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
456 list( , $shortCont, ) = self::splitStoragePath( $params['dir'] );
457 foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
458 $status->merge( $this->doPrepareInternal( "{$fullCont}{$suffix}", $dir, $params ) );
459 }
460 }
461
462 return $status;
463 }
464
465 /**
466 * @see FileBackendStore::doPrepare()
467 * @param $container
468 * @param string $dir
469 * @param array $params
470 * @return Status
471 */
472 protected function doPrepareInternal( $container, $dir, array $params ) {
473 return Status::newGood();
474 }
475
476 final protected function doSecure( array $params ) {
477 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
478 $status = Status::newGood();
479
480 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
481 if ( $dir === null ) {
482 $status->fatal( 'backend-fail-invalidpath', $params['dir'] );
483
484 return $status; // invalid storage path
485 }
486
487 if ( $shard !== null ) { // confined to a single container/shard
488 $status->merge( $this->doSecureInternal( $fullCont, $dir, $params ) );
489 } else { // directory is on several shards
490 wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
491 list( , $shortCont, ) = self::splitStoragePath( $params['dir'] );
492 foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
493 $status->merge( $this->doSecureInternal( "{$fullCont}{$suffix}", $dir, $params ) );
494 }
495 }
496
497 return $status;
498 }
499
500 /**
501 * @see FileBackendStore::doSecure()
502 * @param $container
503 * @param string $dir
504 * @param array $params
505 * @return Status
506 */
507 protected function doSecureInternal( $container, $dir, array $params ) {
508 return Status::newGood();
509 }
510
511 final protected function doPublish( array $params ) {
512 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
513 $status = Status::newGood();
514
515 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
516 if ( $dir === null ) {
517 $status->fatal( 'backend-fail-invalidpath', $params['dir'] );
518
519 return $status; // invalid storage path
520 }
521
522 if ( $shard !== null ) { // confined to a single container/shard
523 $status->merge( $this->doPublishInternal( $fullCont, $dir, $params ) );
524 } else { // directory is on several shards
525 wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
526 list( , $shortCont, ) = self::splitStoragePath( $params['dir'] );
527 foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
528 $status->merge( $this->doPublishInternal( "{$fullCont}{$suffix}", $dir, $params ) );
529 }
530 }
531
532 return $status;
533 }
534
535 /**
536 * @see FileBackendStore::doPublish()
537 * @param $container
538 * @param string $dir
539 * @param array $params
540 * @return Status
541 */
542 protected function doPublishInternal( $container, $dir, array $params ) {
543 return Status::newGood();
544 }
545
546 final protected function doClean( array $params ) {
547 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
548 $status = Status::newGood();
549
550 // Recursive: first delete all empty subdirs recursively
551 if ( !empty( $params['recursive'] ) && !$this->directoriesAreVirtual() ) {
552 $subDirsRel = $this->getTopDirectoryList( array( 'dir' => $params['dir'] ) );
553 if ( $subDirsRel !== null ) { // no errors
554 foreach ( $subDirsRel as $subDirRel ) {
555 $subDir = $params['dir'] . "/{$subDirRel}"; // full path
556 $status->merge( $this->doClean( array( 'dir' => $subDir ) + $params ) );
557 }
558 unset( $subDirsRel ); // free directory for rmdir() on Windows (for FS backends)
559 }
560 }
561
562 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
563 if ( $dir === null ) {
564 $status->fatal( 'backend-fail-invalidpath', $params['dir'] );
565
566 return $status; // invalid storage path
567 }
568
569 // Attempt to lock this directory...
570 $filesLockEx = array( $params['dir'] );
571 $scopedLockE = $this->getScopedFileLocks( $filesLockEx, LockManager::LOCK_EX, $status );
572 if ( !$status->isOK() ) {
573 return $status; // abort
574 }
575
576 if ( $shard !== null ) { // confined to a single container/shard
577 $status->merge( $this->doCleanInternal( $fullCont, $dir, $params ) );
578 $this->deleteContainerCache( $fullCont ); // purge cache
579 } else { // directory is on several shards
580 wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
581 list( , $shortCont, ) = self::splitStoragePath( $params['dir'] );
582 foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
583 $status->merge( $this->doCleanInternal( "{$fullCont}{$suffix}", $dir, $params ) );
584 $this->deleteContainerCache( "{$fullCont}{$suffix}" ); // purge cache
585 }
586 }
587
588 return $status;
589 }
590
591 /**
592 * @see FileBackendStore::doClean()
593 * @param $container
594 * @param string $dir
595 * @param array $params
596 * @return Status
597 */
598 protected function doCleanInternal( $container, $dir, array $params ) {
599 return Status::newGood();
600 }
601
602 final public function fileExists( array $params ) {
603 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
604 $stat = $this->getFileStat( $params );
605
606 return ( $stat === null ) ? null : (bool)$stat; // null => failure
607 }
608
609 final public function getFileTimestamp( array $params ) {
610 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
611 $stat = $this->getFileStat( $params );
612
613 return $stat ? $stat['mtime'] : false;
614 }
615
616 final public function getFileSize( array $params ) {
617 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
618 $stat = $this->getFileStat( $params );
619
620 return $stat ? $stat['size'] : false;
621 }
622
623 final public function getFileStat( array $params ) {
624 $path = self::normalizeStoragePath( $params['src'] );
625 if ( $path === null ) {
626 return false; // invalid storage path
627 }
628 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
629 $latest = !empty( $params['latest'] ); // use latest data?
630 if ( !$this->cheapCache->has( $path, 'stat', self::CACHE_TTL ) ) {
631 $this->primeFileCache( array( $path ) ); // check persistent cache
632 }
633 if ( $this->cheapCache->has( $path, 'stat', self::CACHE_TTL ) ) {
634 $stat = $this->cheapCache->get( $path, 'stat' );
635 // If we want the latest data, check that this cached
636 // value was in fact fetched with the latest available data.
637 if ( is_array( $stat ) ) {
638 if ( !$latest || $stat['latest'] ) {
639 return $stat;
640 }
641 } elseif ( in_array( $stat, array( 'NOT_EXIST', 'NOT_EXIST_LATEST' ) ) ) {
642 if ( !$latest || $stat === 'NOT_EXIST_LATEST' ) {
643 return false;
644 }
645 }
646 }
647 wfProfileIn( __METHOD__ . '-miss-' . $this->name );
648 $stat = $this->doGetFileStat( $params );
649 wfProfileOut( __METHOD__ . '-miss-' . $this->name );
650 if ( is_array( $stat ) ) { // file exists
651 $stat['latest'] = $latest;
652 $this->cheapCache->set( $path, 'stat', $stat );
653 $this->setFileCache( $path, $stat ); // update persistent cache
654 if ( isset( $stat['sha1'] ) ) { // some backends store SHA-1 as metadata
655 $this->cheapCache->set( $path, 'sha1',
656 array( 'hash' => $stat['sha1'], 'latest' => $latest ) );
657 }
658 } elseif ( $stat === false ) { // file does not exist
659 $this->cheapCache->set( $path, 'stat', $latest ? 'NOT_EXIST_LATEST' : 'NOT_EXIST' );
660 $this->cheapCache->set( $path, 'sha1', // the SHA-1 must be false too
661 array( 'hash' => false, 'latest' => $latest ) );
662 wfDebug( __METHOD__ . ": File $path does not exist.\n" );
663 } else { // an error occurred
664 wfDebug( __METHOD__ . ": Could not stat file $path.\n" );
665 }
666
667 return $stat;
668 }
669
670 /**
671 * @see FileBackendStore::getFileStat()
672 */
673 abstract protected function doGetFileStat( array $params );
674
675 public function getFileContentsMulti( array $params ) {
676 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
677
678 $params = $this->setConcurrencyFlags( $params );
679 $contents = $this->doGetFileContentsMulti( $params );
680
681 return $contents;
682 }
683
684 /**
685 * @see FileBackendStore::getFileContentsMulti()
686 * @param array $params
687 * @return array
688 */
689 protected function doGetFileContentsMulti( array $params ) {
690 $contents = array();
691 foreach ( $this->doGetLocalReferenceMulti( $params ) as $path => $fsFile ) {
692 wfSuppressWarnings();
693 $contents[$path] = $fsFile ? file_get_contents( $fsFile->getPath() ) : false;
694 wfRestoreWarnings();
695 }
696
697 return $contents;
698 }
699
700 final public function getFileSha1Base36( array $params ) {
701 $path = self::normalizeStoragePath( $params['src'] );
702 if ( $path === null ) {
703 return false; // invalid storage path
704 }
705 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
706 $latest = !empty( $params['latest'] ); // use latest data?
707 if ( $this->cheapCache->has( $path, 'sha1', self::CACHE_TTL ) ) {
708 $stat = $this->cheapCache->get( $path, 'sha1' );
709 // If we want the latest data, check that this cached
710 // value was in fact fetched with the latest available data.
711 if ( !$latest || $stat['latest'] ) {
712 return $stat['hash'];
713 }
714 }
715 wfProfileIn( __METHOD__ . '-miss-' . $this->name );
716 $hash = $this->doGetFileSha1Base36( $params );
717 wfProfileOut( __METHOD__ . '-miss-' . $this->name );
718 $this->cheapCache->set( $path, 'sha1', array( 'hash' => $hash, 'latest' => $latest ) );
719
720 return $hash;
721 }
722
723 /**
724 * @see FileBackendStore::getFileSha1Base36()
725 * @param array $params
726 * @return bool|string
727 */
728 protected function doGetFileSha1Base36( array $params ) {
729 $fsFile = $this->getLocalReference( $params );
730 if ( !$fsFile ) {
731 return false;
732 } else {
733 return $fsFile->getSha1Base36();
734 }
735 }
736
737 final public function getFileProps( array $params ) {
738 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
739 $fsFile = $this->getLocalReference( $params );
740 $props = $fsFile ? $fsFile->getProps() : FSFile::placeholderProps();
741
742 return $props;
743 }
744
745 final public function getLocalReferenceMulti( array $params ) {
746 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
747
748 $params = $this->setConcurrencyFlags( $params );
749
750 $fsFiles = array(); // (path => FSFile)
751 $latest = !empty( $params['latest'] ); // use latest data?
752 // Reuse any files already in process cache...
753 foreach ( $params['srcs'] as $src ) {
754 $path = self::normalizeStoragePath( $src );
755 if ( $path === null ) {
756 $fsFiles[$src] = null; // invalid storage path
757 } elseif ( $this->expensiveCache->has( $path, 'localRef' ) ) {
758 $val = $this->expensiveCache->get( $path, 'localRef' );
759 // If we want the latest data, check that this cached
760 // value was in fact fetched with the latest available data.
761 if ( !$latest || $val['latest'] ) {
762 $fsFiles[$src] = $val['object'];
763 }
764 }
765 }
766 // Fetch local references of any remaning files...
767 $params['srcs'] = array_diff( $params['srcs'], array_keys( $fsFiles ) );
768 foreach ( $this->doGetLocalReferenceMulti( $params ) as $path => $fsFile ) {
769 $fsFiles[$path] = $fsFile;
770 if ( $fsFile ) { // update the process cache...
771 $this->expensiveCache->set( $path, 'localRef',
772 array( 'object' => $fsFile, 'latest' => $latest ) );
773 }
774 }
775
776 return $fsFiles;
777 }
778
779 /**
780 * @see FileBackendStore::getLocalReferenceMulti()
781 * @param array $params
782 * @return array
783 */
784 protected function doGetLocalReferenceMulti( array $params ) {
785 return $this->doGetLocalCopyMulti( $params );
786 }
787
788 final public function getLocalCopyMulti( array $params ) {
789 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
790
791 $params = $this->setConcurrencyFlags( $params );
792 $tmpFiles = $this->doGetLocalCopyMulti( $params );
793
794 return $tmpFiles;
795 }
796
797 /**
798 * @see FileBackendStore::getLocalCopyMulti()
799 * @param array $params
800 * @return array
801 */
802 abstract protected function doGetLocalCopyMulti( array $params );
803
804 /**
805 * @see FileBackend::getFileHttpUrl()
806 * @param array $params
807 * @return string|null
808 */
809 public function getFileHttpUrl( array $params ) {
810 return null; // not supported
811 }
812
813 final public function streamFile( array $params ) {
814 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
815 $status = Status::newGood();
816
817 $info = $this->getFileStat( $params );
818 if ( !$info ) { // let StreamFile handle the 404
819 $status->fatal( 'backend-fail-notexists', $params['src'] );
820 }
821
822 // Set output buffer and HTTP headers for stream
823 $extraHeaders = isset( $params['headers'] ) ? $params['headers'] : array();
824 $res = StreamFile::prepareForStream( $params['src'], $info, $extraHeaders );
825 if ( $res == StreamFile::NOT_MODIFIED ) {
826 // do nothing; client cache is up to date
827 } elseif ( $res == StreamFile::READY_STREAM ) {
828 wfProfileIn( __METHOD__ . '-send-' . $this->name );
829 $status = $this->doStreamFile( $params );
830 wfProfileOut( __METHOD__ . '-send-' . $this->name );
831 if ( !$status->isOK() ) {
832 // Per bug 41113, nasty things can happen if bad cache entries get
833 // stuck in cache. It's also possible that this error can come up
834 // with simple race conditions. Clear out the stat cache to be safe.
835 $this->clearCache( array( $params['src'] ) );
836 $this->deleteFileCache( $params['src'] );
837 trigger_error( "Bad stat cache or race condition for file {$params['src']}." );
838 }
839 } else {
840 $status->fatal( 'backend-fail-stream', $params['src'] );
841 }
842
843 return $status;
844 }
845
846 /**
847 * @see FileBackendStore::streamFile()
848 * @param array $params
849 * @return Status
850 */
851 protected function doStreamFile( array $params ) {
852 $status = Status::newGood();
853
854 $fsFile = $this->getLocalReference( $params );
855 if ( !$fsFile ) {
856 $status->fatal( 'backend-fail-stream', $params['src'] );
857 } elseif ( !readfile( $fsFile->getPath() ) ) {
858 $status->fatal( 'backend-fail-stream', $params['src'] );
859 }
860
861 return $status;
862 }
863
864 final public function directoryExists( array $params ) {
865 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
866 if ( $dir === null ) {
867 return false; // invalid storage path
868 }
869 if ( $shard !== null ) { // confined to a single container/shard
870 return $this->doDirectoryExists( $fullCont, $dir, $params );
871 } else { // directory is on several shards
872 wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
873 list( , $shortCont, ) = self::splitStoragePath( $params['dir'] );
874 $res = false; // response
875 foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
876 $exists = $this->doDirectoryExists( "{$fullCont}{$suffix}", $dir, $params );
877 if ( $exists ) {
878 $res = true;
879 break; // found one!
880 } elseif ( $exists === null ) { // error?
881 $res = null; // if we don't find anything, it is indeterminate
882 }
883 }
884
885 return $res;
886 }
887 }
888
889 /**
890 * @see FileBackendStore::directoryExists()
891 *
892 * @param string $container Resolved container name
893 * @param string $dir Resolved path relative to container
894 * @param array $params
895 * @return bool|null
896 */
897 abstract protected function doDirectoryExists( $container, $dir, array $params );
898
899 final public function getDirectoryList( array $params ) {
900 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
901 if ( $dir === null ) { // invalid storage path
902 return null;
903 }
904 if ( $shard !== null ) {
905 // File listing is confined to a single container/shard
906 return $this->getDirectoryListInternal( $fullCont, $dir, $params );
907 } else {
908 wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
909 // File listing spans multiple containers/shards
910 list( , $shortCont, ) = self::splitStoragePath( $params['dir'] );
911
912 return new FileBackendStoreShardDirIterator( $this,
913 $fullCont, $dir, $this->getContainerSuffixes( $shortCont ), $params );
914 }
915 }
916
917 /**
918 * Do not call this function from places outside FileBackend
919 *
920 * @see FileBackendStore::getDirectoryList()
921 *
922 * @param string $container Resolved container name
923 * @param string $dir Resolved path relative to container
924 * @param array $params
925 * @return Traversable|Array|null Returns null on failure
926 */
927 abstract public function getDirectoryListInternal( $container, $dir, array $params );
928
929 final public function getFileList( array $params ) {
930 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
931 if ( $dir === null ) { // invalid storage path
932 return null;
933 }
934 if ( $shard !== null ) {
935 // File listing is confined to a single container/shard
936 return $this->getFileListInternal( $fullCont, $dir, $params );
937 } else {
938 wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
939 // File listing spans multiple containers/shards
940 list( , $shortCont, ) = self::splitStoragePath( $params['dir'] );
941
942 return new FileBackendStoreShardFileIterator( $this,
943 $fullCont, $dir, $this->getContainerSuffixes( $shortCont ), $params );
944 }
945 }
946
947 /**
948 * Do not call this function from places outside FileBackend
949 *
950 * @see FileBackendStore::getFileList()
951 *
952 * @param string $container Resolved container name
953 * @param string $dir Resolved path relative to container
954 * @param array $params
955 * @return Traversable|Array|null Returns null on failure
956 */
957 abstract public function getFileListInternal( $container, $dir, array $params );
958
959 /**
960 * Return a list of FileOp objects from a list of operations.
961 * Do not call this function from places outside FileBackend.
962 *
963 * The result must have the same number of items as the input.
964 * An exception is thrown if an unsupported operation is requested.
965 *
966 * @param array $ops Same format as doOperations()
967 * @return array List of FileOp objects
968 * @throws MWException
969 */
970 final public function getOperationsInternal( array $ops ) {
971 $supportedOps = array(
972 'store' => 'StoreFileOp',
973 'copy' => 'CopyFileOp',
974 'move' => 'MoveFileOp',
975 'delete' => 'DeleteFileOp',
976 'create' => 'CreateFileOp',
977 'describe' => 'DescribeFileOp',
978 'null' => 'NullFileOp'
979 );
980
981 $performOps = array(); // array of FileOp objects
982 // Build up ordered array of FileOps...
983 foreach ( $ops as $operation ) {
984 $opName = $operation['op'];
985 if ( isset( $supportedOps[$opName] ) ) {
986 $class = $supportedOps[$opName];
987 // Get params for this operation
988 $params = $operation;
989 // Append the FileOp class
990 $performOps[] = new $class( $this, $params );
991 } else {
992 throw new MWException( "Operation '$opName' is not supported." );
993 }
994 }
995
996 return $performOps;
997 }
998
999 /**
1000 * Get a list of storage paths to lock for a list of operations
1001 * Returns an array with LockManager::LOCK_UW (shared locks) and
1002 * LockManager::LOCK_EX (exclusive locks) keys, each corresponding
1003 * to a list of storage paths to be locked. All returned paths are
1004 * normalized.
1005 *
1006 * @param array $performOps List of FileOp objects
1007 * @return array (LockManager::LOCK_UW => path list, LockManager::LOCK_EX => path list)
1008 */
1009 final public function getPathsToLockForOpsInternal( array $performOps ) {
1010 // Build up a list of files to lock...
1011 $paths = array( 'sh' => array(), 'ex' => array() );
1012 foreach ( $performOps as $fileOp ) {
1013 $paths['sh'] = array_merge( $paths['sh'], $fileOp->storagePathsRead() );
1014 $paths['ex'] = array_merge( $paths['ex'], $fileOp->storagePathsChanged() );
1015 }
1016 // Optimization: if doing an EX lock anyway, don't also set an SH one
1017 $paths['sh'] = array_diff( $paths['sh'], $paths['ex'] );
1018 // Get a shared lock on the parent directory of each path changed
1019 $paths['sh'] = array_merge( $paths['sh'], array_map( 'dirname', $paths['ex'] ) );
1020
1021 return array(
1022 LockManager::LOCK_UW => $paths['sh'],
1023 LockManager::LOCK_EX => $paths['ex']
1024 );
1025 }
1026
1027 public function getScopedLocksForOps( array $ops, Status $status ) {
1028 $paths = $this->getPathsToLockForOpsInternal( $this->getOperationsInternal( $ops ) );
1029
1030 return array( $this->getScopedFileLocks( $paths, 'mixed', $status ) );
1031 }
1032
1033 final protected function doOperationsInternal( array $ops, array $opts ) {
1034 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
1035 $status = Status::newGood();
1036
1037 // Fix up custom header name/value pairs...
1038 $ops = array_map( array( $this, 'stripInvalidHeadersFromOp' ), $ops );
1039
1040 // Build up a list of FileOps...
1041 $performOps = $this->getOperationsInternal( $ops );
1042
1043 // Acquire any locks as needed...
1044 if ( empty( $opts['nonLocking'] ) ) {
1045 // Build up a list of files to lock...
1046 $paths = $this->getPathsToLockForOpsInternal( $performOps );
1047 // Try to lock those files for the scope of this function...
1048 $scopeLock = $this->getScopedFileLocks( $paths, 'mixed', $status );
1049 if ( !$status->isOK() ) {
1050 return $status; // abort
1051 }
1052 }
1053
1054 // Clear any file cache entries (after locks acquired)
1055 if ( empty( $opts['preserveCache'] ) ) {
1056 $this->clearCache();
1057 }
1058
1059 // Load from the persistent file and container caches
1060 $this->primeFileCache( $performOps );
1061 $this->primeContainerCache( $performOps );
1062
1063 // Actually attempt the operation batch...
1064 $opts = $this->setConcurrencyFlags( $opts );
1065 $subStatus = FileOpBatch::attempt( $performOps, $opts, $this->fileJournal );
1066
1067 // Merge errors into status fields
1068 $status->merge( $subStatus );
1069 $status->success = $subStatus->success; // not done in merge()
1070
1071 return $status;
1072 }
1073
1074 final protected function doQuickOperationsInternal( array $ops ) {
1075 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
1076 $status = Status::newGood();
1077
1078 // Fix up custom header name/value pairs...
1079 $ops = array_map( array( $this, 'stripInvalidHeadersFromOp' ), $ops );
1080
1081 // Clear any file cache entries
1082 $this->clearCache();
1083
1084 $supportedOps = array( 'create', 'store', 'copy', 'move', 'delete', 'describe', 'null' );
1085 $async = ( $this->parallelize === 'implicit' && count( $ops ) > 1 );
1086 $maxConcurrency = $this->concurrency; // throttle
1087
1088 $statuses = array(); // array of (index => Status)
1089 $fileOpHandles = array(); // list of (index => handle) arrays
1090 $curFileOpHandles = array(); // current handle batch
1091 // Perform the sync-only ops and build up op handles for the async ops...
1092 foreach ( $ops as $index => $params ) {
1093 if ( !in_array( $params['op'], $supportedOps ) ) {
1094 throw new MWException( "Operation '{$params['op']}' is not supported." );
1095 }
1096 $method = $params['op'] . 'Internal'; // e.g. "storeInternal"
1097 $subStatus = $this->$method( array( 'async' => $async ) + $params );
1098 if ( $subStatus->value instanceof FileBackendStoreOpHandle ) { // async
1099 if ( count( $curFileOpHandles ) >= $maxConcurrency ) {
1100 $fileOpHandles[] = $curFileOpHandles; // push this batch
1101 $curFileOpHandles = array();
1102 }
1103 $curFileOpHandles[$index] = $subStatus->value; // keep index
1104 } else { // error or completed
1105 $statuses[$index] = $subStatus; // keep index
1106 }
1107 }
1108 if ( count( $curFileOpHandles ) ) {
1109 $fileOpHandles[] = $curFileOpHandles; // last batch
1110 }
1111 // Do all the async ops that can be done concurrently...
1112 foreach ( $fileOpHandles as $fileHandleBatch ) {
1113 $statuses = $statuses + $this->executeOpHandlesInternal( $fileHandleBatch );
1114 }
1115 // Marshall and merge all the responses...
1116 foreach ( $statuses as $index => $subStatus ) {
1117 $status->merge( $subStatus );
1118 if ( $subStatus->isOK() ) {
1119 $status->success[$index] = true;
1120 ++$status->successCount;
1121 } else {
1122 $status->success[$index] = false;
1123 ++$status->failCount;
1124 }
1125 }
1126
1127 return $status;
1128 }
1129
1130 /**
1131 * Execute a list of FileBackendStoreOpHandle handles in parallel.
1132 * The resulting Status object fields will correspond
1133 * to the order in which the handles where given.
1134 *
1135 * @param array $fileOpHandles
1136 * @throws MWException
1137 * @internal param array $handles List of FileBackendStoreOpHandle objects
1138 * @return array Map of Status objects
1139 */
1140 final public function executeOpHandlesInternal( array $fileOpHandles ) {
1141 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
1142 foreach ( $fileOpHandles as $fileOpHandle ) {
1143 if ( !( $fileOpHandle instanceof FileBackendStoreOpHandle ) ) {
1144 throw new MWException( "Given a non-FileBackendStoreOpHandle object." );
1145 } elseif ( $fileOpHandle->backend->getName() !== $this->getName() ) {
1146 throw new MWException( "Given a FileBackendStoreOpHandle for the wrong backend." );
1147 }
1148 }
1149 $res = $this->doExecuteOpHandlesInternal( $fileOpHandles );
1150 foreach ( $fileOpHandles as $fileOpHandle ) {
1151 $fileOpHandle->closeResources();
1152 }
1153
1154 return $res;
1155 }
1156
1157 /**
1158 * @see FileBackendStore::executeOpHandlesInternal()
1159 * @param array $fileOpHandles
1160 * @throws MWException
1161 * @return array List of corresponding Status objects
1162 */
1163 protected function doExecuteOpHandlesInternal( array $fileOpHandles ) {
1164 if ( count( $fileOpHandles ) ) {
1165 throw new MWException( "This backend supports no asynchronous operations." );
1166 }
1167
1168 return array();
1169 }
1170
1171 /**
1172 * Strip long HTTP headers from a file operation.
1173 * Most headers are just numbers, but some are allowed to be long.
1174 * This function is useful for cleaning up headers and avoiding backend
1175 * specific errors, especially in the middle of batch file operations.
1176 *
1177 * @param array $op Same format as doOperation()
1178 * @return array
1179 */
1180 protected function stripInvalidHeadersFromOp( array $op ) {
1181 static $longs = array( 'Content-Disposition' );
1182 if ( isset( $op['headers'] ) ) { // op sets HTTP headers
1183 foreach ( $op['headers'] as $name => $value ) {
1184 $maxHVLen = in_array( $name, $longs ) ? INF : 255;
1185 if ( strlen( $name ) > 255 || strlen( $value ) > $maxHVLen ) {
1186 trigger_error( "Header '$name: $value' is too long." );
1187 unset( $op['headers'][$name] );
1188 } elseif ( !strlen( $value ) ) {
1189 $op['headers'][$name] = ''; // null/false => ""
1190 }
1191 }
1192 }
1193
1194 return $op;
1195 }
1196
1197 final public function preloadCache( array $paths ) {
1198 $fullConts = array(); // full container names
1199 foreach ( $paths as $path ) {
1200 list( $fullCont, , ) = $this->resolveStoragePath( $path );
1201 $fullConts[] = $fullCont;
1202 }
1203 // Load from the persistent file and container caches
1204 $this->primeContainerCache( $fullConts );
1205 $this->primeFileCache( $paths );
1206 }
1207
1208 final public function clearCache( array $paths = null ) {
1209 if ( is_array( $paths ) ) {
1210 $paths = array_map( 'FileBackend::normalizeStoragePath', $paths );
1211 $paths = array_filter( $paths, 'strlen' ); // remove nulls
1212 }
1213 if ( $paths === null ) {
1214 $this->cheapCache->clear();
1215 $this->expensiveCache->clear();
1216 } else {
1217 foreach ( $paths as $path ) {
1218 $this->cheapCache->clear( $path );
1219 $this->expensiveCache->clear( $path );
1220 }
1221 }
1222 $this->doClearCache( $paths );
1223 }
1224
1225 /**
1226 * Clears any additional stat caches for storage paths
1227 *
1228 * @see FileBackend::clearCache()
1229 *
1230 * @param array $paths Storage paths (optional)
1231 */
1232 protected function doClearCache( array $paths = null ) {
1233 }
1234
1235 /**
1236 * Is this a key/value store where directories are just virtual?
1237 * Virtual directories exists in so much as files exists that are
1238 * prefixed with the directory path followed by a forward slash.
1239 *
1240 * @return bool
1241 */
1242 abstract protected function directoriesAreVirtual();
1243
1244 /**
1245 * Check if a container name is valid.
1246 * This checks for for length and illegal characters.
1247 *
1248 * @param string $container
1249 * @return bool
1250 */
1251 final protected static function isValidContainerName( $container ) {
1252 // This accounts for Swift and S3 restrictions while leaving room
1253 // for things like '.xxx' (hex shard chars) or '.seg' (segments).
1254 // This disallows directory separators or traversal characters.
1255 // Note that matching strings URL encode to the same string;
1256 // in Swift, the length restriction is *after* URL encoding.
1257 return preg_match( '/^[a-z0-9][a-z0-9-_]{0,199}$/i', $container );
1258 }
1259
1260 /**
1261 * Splits a storage path into an internal container name,
1262 * an internal relative file name, and a container shard suffix.
1263 * Any shard suffix is already appended to the internal container name.
1264 * This also checks that the storage path is valid and within this backend.
1265 *
1266 * If the container is sharded but a suffix could not be determined,
1267 * this means that the path can only refer to a directory and can only
1268 * be scanned by looking in all the container shards.
1269 *
1270 * @param string $storagePath
1271 * @return array (container, path, container suffix) or (null, null, null) if invalid
1272 */
1273 final protected function resolveStoragePath( $storagePath ) {
1274 list( $backend, $container, $relPath ) = self::splitStoragePath( $storagePath );
1275 if ( $backend === $this->name ) { // must be for this backend
1276 $relPath = self::normalizeContainerPath( $relPath );
1277 if ( $relPath !== null ) {
1278 // Get shard for the normalized path if this container is sharded
1279 $cShard = $this->getContainerShard( $container, $relPath );
1280 // Validate and sanitize the relative path (backend-specific)
1281 $relPath = $this->resolveContainerPath( $container, $relPath );
1282 if ( $relPath !== null ) {
1283 // Prepend any wiki ID prefix to the container name
1284 $container = $this->fullContainerName( $container );
1285 if ( self::isValidContainerName( $container ) ) {
1286 // Validate and sanitize the container name (backend-specific)
1287 $container = $this->resolveContainerName( "{$container}{$cShard}" );
1288 if ( $container !== null ) {
1289 return array( $container, $relPath, $cShard );
1290 }
1291 }
1292 }
1293 }
1294 }
1295
1296 return array( null, null, null );
1297 }
1298
1299 /**
1300 * Like resolveStoragePath() except null values are returned if
1301 * the container is sharded and the shard could not be determined
1302 * or if the path ends with '/'. The later case is illegal for FS
1303 * backends and can confuse listings for object store backends.
1304 *
1305 * This function is used when resolving paths that must be valid
1306 * locations for files. Directory and listing functions should
1307 * generally just use resolveStoragePath() instead.
1308 *
1309 * @see FileBackendStore::resolveStoragePath()
1310 *
1311 * @param string $storagePath
1312 * @return array (container, path) or (null, null) if invalid
1313 */
1314 final protected function resolveStoragePathReal( $storagePath ) {
1315 list( $container, $relPath, $cShard ) = $this->resolveStoragePath( $storagePath );
1316 if ( $cShard !== null && substr( $relPath, -1 ) !== '/' ) {
1317 return array( $container, $relPath );
1318 }
1319
1320 return array( null, null );
1321 }
1322
1323 /**
1324 * Get the container name shard suffix for a given path.
1325 * Any empty suffix means the container is not sharded.
1326 *
1327 * @param string $container Container name
1328 * @param string $relPath Storage path relative to the container
1329 * @return string|null Returns null if shard could not be determined
1330 */
1331 final protected function getContainerShard( $container, $relPath ) {
1332 list( $levels, $base, $repeat ) = $this->getContainerHashLevels( $container );
1333 if ( $levels == 1 || $levels == 2 ) {
1334 // Hash characters are either base 16 or 36
1335 $char = ( $base == 36 ) ? '[0-9a-z]' : '[0-9a-f]';
1336 // Get a regex that represents the shard portion of paths.
1337 // The concatenation of the captures gives us the shard.
1338 if ( $levels === 1 ) { // 16 or 36 shards per container
1339 $hashDirRegex = '(' . $char . ')';
1340 } else { // 256 or 1296 shards per container
1341 if ( $repeat ) { // verbose hash dir format (e.g. "a/ab/abc")
1342 $hashDirRegex = $char . '/(' . $char . '{2})';
1343 } else { // short hash dir format (e.g. "a/b/c")
1344 $hashDirRegex = '(' . $char . ')/(' . $char . ')';
1345 }
1346 }
1347 // Allow certain directories to be above the hash dirs so as
1348 // to work with FileRepo (e.g. "archive/a/ab" or "temp/a/ab").
1349 // They must be 2+ chars to avoid any hash directory ambiguity.
1350 $m = array();
1351 if ( preg_match( "!^(?:[^/]{2,}/)*$hashDirRegex(?:/|$)!", $relPath, $m ) ) {
1352 return '.' . implode( '', array_slice( $m, 1 ) );
1353 }
1354
1355 return null; // failed to match
1356 }
1357
1358 return ''; // no sharding
1359 }
1360
1361 /**
1362 * Check if a storage path maps to a single shard.
1363 * Container dirs like "a", where the container shards on "x/xy",
1364 * can reside on several shards. Such paths are tricky to handle.
1365 *
1366 * @param string $storagePath Storage path
1367 * @return bool
1368 */
1369 final public function isSingleShardPathInternal( $storagePath ) {
1370 list( , , $shard ) = $this->resolveStoragePath( $storagePath );
1371
1372 return ( $shard !== null );
1373 }
1374
1375 /**
1376 * Get the sharding config for a container.
1377 * If greater than 0, then all file storage paths within
1378 * the container are required to be hashed accordingly.
1379 *
1380 * @param string $container
1381 * @return array (integer levels, integer base, repeat flag) or (0, 0, false)
1382 */
1383 final protected function getContainerHashLevels( $container ) {
1384 if ( isset( $this->shardViaHashLevels[$container] ) ) {
1385 $config = $this->shardViaHashLevels[$container];
1386 $hashLevels = (int)$config['levels'];
1387 if ( $hashLevels == 1 || $hashLevels == 2 ) {
1388 $hashBase = (int)$config['base'];
1389 if ( $hashBase == 16 || $hashBase == 36 ) {
1390 return array( $hashLevels, $hashBase, $config['repeat'] );
1391 }
1392 }
1393 }
1394
1395 return array( 0, 0, false ); // no sharding
1396 }
1397
1398 /**
1399 * Get a list of full container shard suffixes for a container
1400 *
1401 * @param string $container
1402 * @return array
1403 */
1404 final protected function getContainerSuffixes( $container ) {
1405 $shards = array();
1406 list( $digits, $base ) = $this->getContainerHashLevels( $container );
1407 if ( $digits > 0 ) {
1408 $numShards = pow( $base, $digits );
1409 for ( $index = 0; $index < $numShards; $index++ ) {
1410 $shards[] = '.' . wfBaseConvert( $index, 10, $base, $digits );
1411 }
1412 }
1413
1414 return $shards;
1415 }
1416
1417 /**
1418 * Get the full container name, including the wiki ID prefix
1419 *
1420 * @param string $container
1421 * @return string
1422 */
1423 final protected function fullContainerName( $container ) {
1424 if ( $this->wikiId != '' ) {
1425 return "{$this->wikiId}-$container";
1426 } else {
1427 return $container;
1428 }
1429 }
1430
1431 /**
1432 * Resolve a container name, checking if it's allowed by the backend.
1433 * This is intended for internal use, such as encoding illegal chars.
1434 * Subclasses can override this to be more restrictive.
1435 *
1436 * @param string $container
1437 * @return string|null
1438 */
1439 protected function resolveContainerName( $container ) {
1440 return $container;
1441 }
1442
1443 /**
1444 * Resolve a relative storage path, checking if it's allowed by the backend.
1445 * This is intended for internal use, such as encoding illegal chars or perhaps
1446 * getting absolute paths (e.g. FS based backends). Note that the relative path
1447 * may be the empty string (e.g. the path is simply to the container).
1448 *
1449 * @param string $container Container name
1450 * @param string $relStoragePath Storage path relative to the container
1451 * @return string|null Path or null if not valid
1452 */
1453 protected function resolveContainerPath( $container, $relStoragePath ) {
1454 return $relStoragePath;
1455 }
1456
1457 /**
1458 * Get the cache key for a container
1459 *
1460 * @param string $container Resolved container name
1461 * @return string
1462 */
1463 private function containerCacheKey( $container ) {
1464 return wfMemcKey( 'backend', $this->getName(), 'container', $container );
1465 }
1466
1467 /**
1468 * Set the cached info for a container
1469 *
1470 * @param string $container Resolved container name
1471 * @param array $val Information to cache
1472 */
1473 final protected function setContainerCache( $container, array $val ) {
1474 $this->memCache->add( $this->containerCacheKey( $container ), $val, 14 * 86400 );
1475 }
1476
1477 /**
1478 * Delete the cached info for a container.
1479 * The cache key is salted for a while to prevent race conditions.
1480 *
1481 * @param string $container Resolved container name
1482 */
1483 final protected function deleteContainerCache( $container ) {
1484 if ( !$this->memCache->set( $this->containerCacheKey( $container ), 'PURGED', 300 ) ) {
1485 trigger_error( "Unable to delete stat cache for container $container." );
1486 }
1487 }
1488
1489 /**
1490 * Do a batch lookup from cache for container stats for all containers
1491 * used in a list of container names, storage paths, or FileOp objects.
1492 * This loads the persistent cache values into the process cache.
1493 *
1494 * @param array $items
1495 */
1496 final protected function primeContainerCache( array $items ) {
1497 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
1498
1499 $paths = array(); // list of storage paths
1500 $contNames = array(); // (cache key => resolved container name)
1501 // Get all the paths/containers from the items...
1502 foreach ( $items as $item ) {
1503 if ( $item instanceof FileOp ) {
1504 $paths = array_merge( $paths, $item->storagePathsRead() );
1505 $paths = array_merge( $paths, $item->storagePathsChanged() );
1506 } elseif ( self::isStoragePath( $item ) ) {
1507 $paths[] = $item;
1508 } elseif ( is_string( $item ) ) { // full container name
1509 $contNames[$this->containerCacheKey( $item )] = $item;
1510 }
1511 }
1512 // Get all the corresponding cache keys for paths...
1513 foreach ( $paths as $path ) {
1514 list( $fullCont, , ) = $this->resolveStoragePath( $path );
1515 if ( $fullCont !== null ) { // valid path for this backend
1516 $contNames[$this->containerCacheKey( $fullCont )] = $fullCont;
1517 }
1518 }
1519
1520 $contInfo = array(); // (resolved container name => cache value)
1521 // Get all cache entries for these container cache keys...
1522 $values = $this->memCache->getMulti( array_keys( $contNames ) );
1523 foreach ( $values as $cacheKey => $val ) {
1524 $contInfo[$contNames[$cacheKey]] = $val;
1525 }
1526
1527 // Populate the container process cache for the backend...
1528 $this->doPrimeContainerCache( array_filter( $contInfo, 'is_array' ) );
1529 }
1530
1531 /**
1532 * Fill the backend-specific process cache given an array of
1533 * resolved container names and their corresponding cached info.
1534 * Only containers that actually exist should appear in the map.
1535 *
1536 * @param array $containerInfo Map of resolved container names to cached info
1537 */
1538 protected function doPrimeContainerCache( array $containerInfo ) {
1539 }
1540
1541 /**
1542 * Get the cache key for a file path
1543 *
1544 * @param string $path Normalized storage path
1545 * @return string
1546 */
1547 private function fileCacheKey( $path ) {
1548 return wfMemcKey( 'backend', $this->getName(), 'file', sha1( $path ) );
1549 }
1550
1551 /**
1552 * Set the cached stat info for a file path.
1553 * Negatives (404s) are not cached. By not caching negatives, we can skip cache
1554 * salting for the case when a file is created at a path were there was none before.
1555 *
1556 * @param string $path Storage path
1557 * @param array $val Stat information to cache
1558 */
1559 final protected function setFileCache( $path, array $val ) {
1560 $path = FileBackend::normalizeStoragePath( $path );
1561 if ( $path === null ) {
1562 return; // invalid storage path
1563 }
1564 $age = time() - wfTimestamp( TS_UNIX, $val['mtime'] );
1565 $ttl = min( 7 * 86400, max( 300, floor( .1 * $age ) ) );
1566 $this->memCache->add( $this->fileCacheKey( $path ), $val, $ttl );
1567 }
1568
1569 /**
1570 * Delete the cached stat info for a file path.
1571 * The cache key is salted for a while to prevent race conditions.
1572 * Since negatives (404s) are not cached, this does not need to be called when
1573 * a file is created at a path were there was none before.
1574 *
1575 * @param string $path Storage path
1576 */
1577 final protected function deleteFileCache( $path ) {
1578 $path = FileBackend::normalizeStoragePath( $path );
1579 if ( $path === null ) {
1580 return; // invalid storage path
1581 }
1582 if ( !$this->memCache->set( $this->fileCacheKey( $path ), 'PURGED', 300 ) ) {
1583 trigger_error( "Unable to delete stat cache for file $path." );
1584 }
1585 }
1586
1587 /**
1588 * Do a batch lookup from cache for file stats for all paths
1589 * used in a list of storage paths or FileOp objects.
1590 * This loads the persistent cache values into the process cache.
1591 *
1592 * @param array $items List of storage paths or FileOps
1593 */
1594 final protected function primeFileCache( array $items ) {
1595 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
1596
1597 $paths = array(); // list of storage paths
1598 $pathNames = array(); // (cache key => storage path)
1599 // Get all the paths/containers from the items...
1600 foreach ( $items as $item ) {
1601 if ( $item instanceof FileOp ) {
1602 $paths = array_merge( $paths, $item->storagePathsRead() );
1603 $paths = array_merge( $paths, $item->storagePathsChanged() );
1604 } elseif ( self::isStoragePath( $item ) ) {
1605 $paths[] = FileBackend::normalizeStoragePath( $item );
1606 }
1607 }
1608 // Get rid of any paths that failed normalization...
1609 $paths = array_filter( $paths, 'strlen' ); // remove nulls
1610 // Get all the corresponding cache keys for paths...
1611 foreach ( $paths as $path ) {
1612 list( , $rel, ) = $this->resolveStoragePath( $path );
1613 if ( $rel !== null ) { // valid path for this backend
1614 $pathNames[$this->fileCacheKey( $path )] = $path;
1615 }
1616 }
1617 // Get all cache entries for these container cache keys...
1618 $values = $this->memCache->getMulti( array_keys( $pathNames ) );
1619 foreach ( $values as $cacheKey => $val ) {
1620 if ( is_array( $val ) ) {
1621 $path = $pathNames[$cacheKey];
1622 $this->cheapCache->set( $path, 'stat', $val );
1623 if ( isset( $val['sha1'] ) ) { // some backends store SHA-1 as metadata
1624 $this->cheapCache->set( $path, 'sha1',
1625 array( 'hash' => $val['sha1'], 'latest' => $val['latest'] ) );
1626 }
1627 }
1628 }
1629 }
1630
1631 /**
1632 * Set the 'concurrency' option from a list of operation options
1633 *
1634 * @param array $opts Map of operation options
1635 * @return array
1636 */
1637 final protected function setConcurrencyFlags( array $opts ) {
1638 $opts['concurrency'] = 1; // off
1639 if ( $this->parallelize === 'implicit' ) {
1640 if ( !isset( $opts['parallelize'] ) || $opts['parallelize'] ) {
1641 $opts['concurrency'] = $this->concurrency;
1642 }
1643 } elseif ( $this->parallelize === 'explicit' ) {
1644 if ( !empty( $opts['parallelize'] ) ) {
1645 $opts['concurrency'] = $this->concurrency;
1646 }
1647 }
1648
1649 return $opts;
1650 }
1651
1652 /**
1653 * Get the content type to use in HEAD/GET requests for a file
1654 *
1655 * @param string $storagePath
1656 * @param string|null $content File data
1657 * @param string|null $fsPath File system path
1658 * @return MIME type
1659 */
1660 protected function getContentType( $storagePath, $content, $fsPath ) {
1661 return call_user_func_array( $this->mimeCallback, func_get_args() );
1662 }
1663 }
1664
1665 /**
1666 * FileBackendStore helper class for performing asynchronous file operations.
1667 *
1668 * For example, calling FileBackendStore::createInternal() with the "async"
1669 * param flag may result in a Status that contains this object as a value.
1670 * This class is largely backend-specific and is mostly just "magic" to be
1671 * passed to FileBackendStore::executeOpHandlesInternal().
1672 */
1673 abstract class FileBackendStoreOpHandle {
1674 /** @var array */
1675 public $params = array(); // params to caller functions
1676 /** @var FileBackendStore */
1677 public $backend;
1678 /** @var array */
1679 public $resourcesToClose = array();
1680
1681 public $call; // string; name that identifies the function called
1682
1683 /**
1684 * Close all open file handles
1685 */
1686 public function closeResources() {
1687 array_map( 'fclose', $this->resourcesToClose );
1688 }
1689 }
1690
1691 /**
1692 * FileBackendStore helper function to handle listings that span container shards.
1693 * Do not use this class from places outside of FileBackendStore.
1694 *
1695 * @ingroup FileBackend
1696 */
1697 abstract class FileBackendStoreShardListIterator extends FilterIterator {
1698 /** @var FileBackendStore */
1699 protected $backend;
1700
1701 /** @var array */
1702 protected $params;
1703
1704 /** @var string Full container name */
1705 protected $container;
1706
1707 /** @var string Resolved relative path */
1708 protected $directory;
1709
1710 /** @var array */
1711 protected $multiShardPaths = array(); // (rel path => 1)
1712
1713 /**
1714 * @param FileBackendStore $backend
1715 * @param string $container Full storage container name
1716 * @param string $dir Storage directory relative to container
1717 * @param array $suffixes List of container shard suffixes
1718 * @param array $params
1719 */
1720 public function __construct(
1721 FileBackendStore $backend, $container, $dir, array $suffixes, array $params
1722 ) {
1723 $this->backend = $backend;
1724 $this->container = $container;
1725 $this->directory = $dir;
1726 $this->params = $params;
1727
1728 $iter = new AppendIterator();
1729 foreach ( $suffixes as $suffix ) {
1730 $iter->append( $this->listFromShard( $this->container . $suffix ) );
1731 }
1732
1733 parent::__construct( $iter );
1734 }
1735
1736 public function accept() {
1737 $rel = $this->getInnerIterator()->current(); // path relative to given directory
1738 $path = $this->params['dir'] . "/{$rel}"; // full storage path
1739 if ( $this->backend->isSingleShardPathInternal( $path ) ) {
1740 return true; // path is only on one shard; no issue with duplicates
1741 } elseif ( isset( $this->multiShardPaths[$rel] ) ) {
1742 // Don't keep listing paths that are on multiple shards
1743 return false;
1744 } else {
1745 $this->multiShardPaths[$rel] = 1;
1746
1747 return true;
1748 }
1749 }
1750
1751 public function rewind() {
1752 parent::rewind();
1753 $this->multiShardPaths = array();
1754 }
1755
1756 /**
1757 * Get the list for a given container shard
1758 *
1759 * @param string $container Resolved container name
1760 * @return Iterator
1761 */
1762 abstract protected function listFromShard( $container );
1763 }
1764
1765 /**
1766 * Iterator for listing directories
1767 */
1768 class FileBackendStoreShardDirIterator extends FileBackendStoreShardListIterator {
1769 protected function listFromShard( $container ) {
1770 $list = $this->backend->getDirectoryListInternal(
1771 $container, $this->directory, $this->params );
1772 if ( $list === null ) {
1773 return new ArrayIterator( array() );
1774 } else {
1775 return is_array( $list ) ? new ArrayIterator( $list ) : $list;
1776 }
1777 }
1778 }
1779
1780 /**
1781 * Iterator for listing regular files
1782 */
1783 class FileBackendStoreShardFileIterator extends FileBackendStoreShardListIterator {
1784 protected function listFromShard( $container ) {
1785 $list = $this->backend->getFileListInternal(
1786 $container, $this->directory, $this->params );
1787 if ( $list === null ) {
1788 return new ArrayIterator( array() );
1789 } else {
1790 return is_array( $list ) ? new ArrayIterator( $list ) : $list;
1791 }
1792 }
1793 }