Make room for preloadFileStat() call in FileBackend::doOperationsInternal
[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 // Strongly consistent backends can automatically set "latest"
652 $stat['latest'] = isset( $stat['latest'] ) ? $stat['latest'] : $latest;
653 $this->cheapCache->set( $path, 'stat', $stat );
654 $this->setFileCache( $path, $stat ); // update persistent cache
655 if ( isset( $stat['sha1'] ) ) { // some backends store SHA-1 as metadata
656 $this->cheapCache->set( $path, 'sha1',
657 array( 'hash' => $stat['sha1'], 'latest' => $latest ) );
658 }
659 if ( isset( $stat['xattr'] ) ) { // some backends store headers/metadata
660 $stat['xattr'] = self::normalizeXAttributes( $stat['xattr'] );
661 $this->cheapCache->set( $path, 'xattr',
662 array( 'map' => $stat['xattr'], 'latest' => $latest ) );
663 }
664 } elseif ( $stat === false ) { // file does not exist
665 $this->cheapCache->set( $path, 'stat', $latest ? 'NOT_EXIST_LATEST' : 'NOT_EXIST' );
666 $this->cheapCache->set( $path, 'xattr', array( 'map' => false, 'latest' => $latest ) );
667 $this->cheapCache->set( $path, 'sha1', array( 'hash' => false, 'latest' => $latest ) );
668 wfDebug( __METHOD__ . ": File $path does not exist.\n" );
669 } else { // an error occurred
670 wfDebug( __METHOD__ . ": Could not stat file $path.\n" );
671 }
672
673 return $stat;
674 }
675
676 /**
677 * @see FileBackendStore::getFileStat()
678 */
679 abstract protected function doGetFileStat( array $params );
680
681 public function getFileContentsMulti( array $params ) {
682 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
683
684 $params = $this->setConcurrencyFlags( $params );
685 $contents = $this->doGetFileContentsMulti( $params );
686
687 return $contents;
688 }
689
690 /**
691 * @see FileBackendStore::getFileContentsMulti()
692 * @param array $params
693 * @return array
694 */
695 protected function doGetFileContentsMulti( array $params ) {
696 $contents = array();
697 foreach ( $this->doGetLocalReferenceMulti( $params ) as $path => $fsFile ) {
698 wfSuppressWarnings();
699 $contents[$path] = $fsFile ? file_get_contents( $fsFile->getPath() ) : false;
700 wfRestoreWarnings();
701 }
702
703 return $contents;
704 }
705
706 final public function getFileXAttributes( array $params ) {
707 $path = self::normalizeStoragePath( $params['src'] );
708 if ( $path === null ) {
709 return false; // invalid storage path
710 }
711 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
712 $latest = !empty( $params['latest'] ); // use latest data?
713 if ( $this->cheapCache->has( $path, 'xattr', self::CACHE_TTL ) ) {
714 $stat = $this->cheapCache->get( $path, 'xattr' );
715 // If we want the latest data, check that this cached
716 // value was in fact fetched with the latest available data.
717 if ( !$latest || $stat['latest'] ) {
718 return $stat['map'];
719 }
720 }
721 wfProfileIn( __METHOD__ . '-miss' );
722 wfProfileIn( __METHOD__ . '-miss-' . $this->name );
723 $fields = $this->doGetFileXAttributes( $params );
724 $fields = is_array( $fields ) ? self::normalizeXAttributes( $fields ) : false;
725 wfProfileOut( __METHOD__ . '-miss-' . $this->name );
726 wfProfileOut( __METHOD__ . '-miss' );
727 $this->cheapCache->set( $path, 'xattr', array( 'map' => $fields, 'latest' => $latest ) );
728
729 return $fields;
730 }
731
732 /**
733 * @see FileBackendStore::getFileXAttributes()
734 * @return bool|string
735 */
736 protected function doGetFileXAttributes( array $params ) {
737 return array( 'headers' => array(), 'metadata' => array() ); // not supported
738 }
739
740 final public function getFileSha1Base36( array $params ) {
741 $path = self::normalizeStoragePath( $params['src'] );
742 if ( $path === null ) {
743 return false; // invalid storage path
744 }
745 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
746 $latest = !empty( $params['latest'] ); // use latest data?
747 if ( $this->cheapCache->has( $path, 'sha1', self::CACHE_TTL ) ) {
748 $stat = $this->cheapCache->get( $path, 'sha1' );
749 // If we want the latest data, check that this cached
750 // value was in fact fetched with the latest available data.
751 if ( !$latest || $stat['latest'] ) {
752 return $stat['hash'];
753 }
754 }
755 wfProfileIn( __METHOD__ . '-miss-' . $this->name );
756 $hash = $this->doGetFileSha1Base36( $params );
757 wfProfileOut( __METHOD__ . '-miss-' . $this->name );
758 $this->cheapCache->set( $path, 'sha1', array( 'hash' => $hash, 'latest' => $latest ) );
759
760 return $hash;
761 }
762
763 /**
764 * @see FileBackendStore::getFileSha1Base36()
765 * @param array $params
766 * @return bool|string
767 */
768 protected function doGetFileSha1Base36( array $params ) {
769 $fsFile = $this->getLocalReference( $params );
770 if ( !$fsFile ) {
771 return false;
772 } else {
773 return $fsFile->getSha1Base36();
774 }
775 }
776
777 final public function getFileProps( array $params ) {
778 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
779 $fsFile = $this->getLocalReference( $params );
780 $props = $fsFile ? $fsFile->getProps() : FSFile::placeholderProps();
781
782 return $props;
783 }
784
785 final public function getLocalReferenceMulti( array $params ) {
786 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
787
788 $params = $this->setConcurrencyFlags( $params );
789
790 $fsFiles = array(); // (path => FSFile)
791 $latest = !empty( $params['latest'] ); // use latest data?
792 // Reuse any files already in process cache...
793 foreach ( $params['srcs'] as $src ) {
794 $path = self::normalizeStoragePath( $src );
795 if ( $path === null ) {
796 $fsFiles[$src] = null; // invalid storage path
797 } elseif ( $this->expensiveCache->has( $path, 'localRef' ) ) {
798 $val = $this->expensiveCache->get( $path, 'localRef' );
799 // If we want the latest data, check that this cached
800 // value was in fact fetched with the latest available data.
801 if ( !$latest || $val['latest'] ) {
802 $fsFiles[$src] = $val['object'];
803 }
804 }
805 }
806 // Fetch local references of any remaning files...
807 $params['srcs'] = array_diff( $params['srcs'], array_keys( $fsFiles ) );
808 foreach ( $this->doGetLocalReferenceMulti( $params ) as $path => $fsFile ) {
809 $fsFiles[$path] = $fsFile;
810 if ( $fsFile ) { // update the process cache...
811 $this->expensiveCache->set( $path, 'localRef',
812 array( 'object' => $fsFile, 'latest' => $latest ) );
813 }
814 }
815
816 return $fsFiles;
817 }
818
819 /**
820 * @see FileBackendStore::getLocalReferenceMulti()
821 * @param array $params
822 * @return array
823 */
824 protected function doGetLocalReferenceMulti( array $params ) {
825 return $this->doGetLocalCopyMulti( $params );
826 }
827
828 final public function getLocalCopyMulti( array $params ) {
829 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
830
831 $params = $this->setConcurrencyFlags( $params );
832 $tmpFiles = $this->doGetLocalCopyMulti( $params );
833
834 return $tmpFiles;
835 }
836
837 /**
838 * @see FileBackendStore::getLocalCopyMulti()
839 * @param array $params
840 * @return array
841 */
842 abstract protected function doGetLocalCopyMulti( array $params );
843
844 /**
845 * @see FileBackend::getFileHttpUrl()
846 * @param array $params
847 * @return string|null
848 */
849 public function getFileHttpUrl( array $params ) {
850 return null; // not supported
851 }
852
853 final public function streamFile( array $params ) {
854 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
855 $status = Status::newGood();
856
857 $info = $this->getFileStat( $params );
858 if ( !$info ) { // let StreamFile handle the 404
859 $status->fatal( 'backend-fail-notexists', $params['src'] );
860 }
861
862 // Set output buffer and HTTP headers for stream
863 $extraHeaders = isset( $params['headers'] ) ? $params['headers'] : array();
864 $res = StreamFile::prepareForStream( $params['src'], $info, $extraHeaders );
865 if ( $res == StreamFile::NOT_MODIFIED ) {
866 // do nothing; client cache is up to date
867 } elseif ( $res == StreamFile::READY_STREAM ) {
868 wfProfileIn( __METHOD__ . '-send-' . $this->name );
869 $status = $this->doStreamFile( $params );
870 wfProfileOut( __METHOD__ . '-send-' . $this->name );
871 if ( !$status->isOK() ) {
872 // Per bug 41113, nasty things can happen if bad cache entries get
873 // stuck in cache. It's also possible that this error can come up
874 // with simple race conditions. Clear out the stat cache to be safe.
875 $this->clearCache( array( $params['src'] ) );
876 $this->deleteFileCache( $params['src'] );
877 trigger_error( "Bad stat cache or race condition for file {$params['src']}." );
878 }
879 } else {
880 $status->fatal( 'backend-fail-stream', $params['src'] );
881 }
882
883 return $status;
884 }
885
886 /**
887 * @see FileBackendStore::streamFile()
888 * @param array $params
889 * @return Status
890 */
891 protected function doStreamFile( array $params ) {
892 $status = Status::newGood();
893
894 $fsFile = $this->getLocalReference( $params );
895 if ( !$fsFile ) {
896 $status->fatal( 'backend-fail-stream', $params['src'] );
897 } elseif ( !readfile( $fsFile->getPath() ) ) {
898 $status->fatal( 'backend-fail-stream', $params['src'] );
899 }
900
901 return $status;
902 }
903
904 final public function directoryExists( array $params ) {
905 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
906 if ( $dir === null ) {
907 return false; // invalid storage path
908 }
909 if ( $shard !== null ) { // confined to a single container/shard
910 return $this->doDirectoryExists( $fullCont, $dir, $params );
911 } else { // directory is on several shards
912 wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
913 list( , $shortCont, ) = self::splitStoragePath( $params['dir'] );
914 $res = false; // response
915 foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
916 $exists = $this->doDirectoryExists( "{$fullCont}{$suffix}", $dir, $params );
917 if ( $exists ) {
918 $res = true;
919 break; // found one!
920 } elseif ( $exists === null ) { // error?
921 $res = null; // if we don't find anything, it is indeterminate
922 }
923 }
924
925 return $res;
926 }
927 }
928
929 /**
930 * @see FileBackendStore::directoryExists()
931 *
932 * @param string $container Resolved container name
933 * @param string $dir Resolved path relative to container
934 * @param array $params
935 * @return bool|null
936 */
937 abstract protected function doDirectoryExists( $container, $dir, array $params );
938
939 final public function getDirectoryList( array $params ) {
940 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
941 if ( $dir === null ) { // invalid storage path
942 return null;
943 }
944 if ( $shard !== null ) {
945 // File listing is confined to a single container/shard
946 return $this->getDirectoryListInternal( $fullCont, $dir, $params );
947 } else {
948 wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
949 // File listing spans multiple containers/shards
950 list( , $shortCont, ) = self::splitStoragePath( $params['dir'] );
951
952 return new FileBackendStoreShardDirIterator( $this,
953 $fullCont, $dir, $this->getContainerSuffixes( $shortCont ), $params );
954 }
955 }
956
957 /**
958 * Do not call this function from places outside FileBackend
959 *
960 * @see FileBackendStore::getDirectoryList()
961 *
962 * @param string $container Resolved container name
963 * @param string $dir Resolved path relative to container
964 * @param array $params
965 * @return Traversable|Array|null Returns null on failure
966 */
967 abstract public function getDirectoryListInternal( $container, $dir, array $params );
968
969 final public function getFileList( array $params ) {
970 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
971 if ( $dir === null ) { // invalid storage path
972 return null;
973 }
974 if ( $shard !== null ) {
975 // File listing is confined to a single container/shard
976 return $this->getFileListInternal( $fullCont, $dir, $params );
977 } else {
978 wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
979 // File listing spans multiple containers/shards
980 list( , $shortCont, ) = self::splitStoragePath( $params['dir'] );
981
982 return new FileBackendStoreShardFileIterator( $this,
983 $fullCont, $dir, $this->getContainerSuffixes( $shortCont ), $params );
984 }
985 }
986
987 /**
988 * Do not call this function from places outside FileBackend
989 *
990 * @see FileBackendStore::getFileList()
991 *
992 * @param string $container Resolved container name
993 * @param string $dir Resolved path relative to container
994 * @param array $params
995 * @return Traversable|Array|null Returns null on failure
996 */
997 abstract public function getFileListInternal( $container, $dir, array $params );
998
999 /**
1000 * Return a list of FileOp objects from a list of operations.
1001 * Do not call this function from places outside FileBackend.
1002 *
1003 * The result must have the same number of items as the input.
1004 * An exception is thrown if an unsupported operation is requested.
1005 *
1006 * @param array $ops Same format as doOperations()
1007 * @return array List of FileOp objects
1008 * @throws FileBackendError
1009 */
1010 final public function getOperationsInternal( array $ops ) {
1011 $supportedOps = array(
1012 'store' => 'StoreFileOp',
1013 'copy' => 'CopyFileOp',
1014 'move' => 'MoveFileOp',
1015 'delete' => 'DeleteFileOp',
1016 'create' => 'CreateFileOp',
1017 'describe' => 'DescribeFileOp',
1018 'null' => 'NullFileOp'
1019 );
1020
1021 $performOps = array(); // array of FileOp objects
1022 // Build up ordered array of FileOps...
1023 foreach ( $ops as $operation ) {
1024 $opName = $operation['op'];
1025 if ( isset( $supportedOps[$opName] ) ) {
1026 $class = $supportedOps[$opName];
1027 // Get params for this operation
1028 $params = $operation;
1029 // Append the FileOp class
1030 $performOps[] = new $class( $this, $params );
1031 } else {
1032 throw new FileBackendError( "Operation '$opName' is not supported." );
1033 }
1034 }
1035
1036 return $performOps;
1037 }
1038
1039 /**
1040 * Get a list of storage paths to lock for a list of operations
1041 * Returns an array with LockManager::LOCK_UW (shared locks) and
1042 * LockManager::LOCK_EX (exclusive locks) keys, each corresponding
1043 * to a list of storage paths to be locked. All returned paths are
1044 * normalized.
1045 *
1046 * @param array $performOps List of FileOp objects
1047 * @return array (LockManager::LOCK_UW => path list, LockManager::LOCK_EX => path list)
1048 */
1049 final public function getPathsToLockForOpsInternal( array $performOps ) {
1050 // Build up a list of files to lock...
1051 $paths = array( 'sh' => array(), 'ex' => array() );
1052 foreach ( $performOps as $fileOp ) {
1053 $paths['sh'] = array_merge( $paths['sh'], $fileOp->storagePathsRead() );
1054 $paths['ex'] = array_merge( $paths['ex'], $fileOp->storagePathsChanged() );
1055 }
1056 // Optimization: if doing an EX lock anyway, don't also set an SH one
1057 $paths['sh'] = array_diff( $paths['sh'], $paths['ex'] );
1058 // Get a shared lock on the parent directory of each path changed
1059 $paths['sh'] = array_merge( $paths['sh'], array_map( 'dirname', $paths['ex'] ) );
1060
1061 return array(
1062 LockManager::LOCK_UW => $paths['sh'],
1063 LockManager::LOCK_EX => $paths['ex']
1064 );
1065 }
1066
1067 public function getScopedLocksForOps( array $ops, Status $status ) {
1068 $paths = $this->getPathsToLockForOpsInternal( $this->getOperationsInternal( $ops ) );
1069
1070 return array( $this->getScopedFileLocks( $paths, 'mixed', $status ) );
1071 }
1072
1073 final protected function doOperationsInternal( array $ops, array $opts ) {
1074 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
1075 $status = Status::newGood();
1076
1077 // Fix up custom header name/value pairs...
1078 $ops = array_map( array( $this, 'stripInvalidHeadersFromOp' ), $ops );
1079
1080 // Build up a list of FileOps...
1081 $performOps = $this->getOperationsInternal( $ops );
1082
1083 // Acquire any locks as needed...
1084 if ( empty( $opts['nonLocking'] ) ) {
1085 // Build up a list of files to lock...
1086 $paths = $this->getPathsToLockForOpsInternal( $performOps );
1087 // Try to lock those files for the scope of this function...
1088 $scopeLock = $this->getScopedFileLocks( $paths, 'mixed', $status );
1089 if ( !$status->isOK() ) {
1090 return $status; // abort
1091 }
1092 }
1093
1094 // Clear any file cache entries (after locks acquired)
1095 if ( empty( $opts['preserveCache'] ) ) {
1096 $this->clearCache();
1097 }
1098
1099 // Build the list of paths involved
1100 $paths = array();
1101 foreach ( $performOps as $op ) {
1102 $paths = array_merge( $paths, $op->storagePathsRead() );
1103 $paths = array_merge( $paths, $op->storagePathsChanged() );
1104 }
1105
1106 // Enlarge the cache to fit the stat entries of these files
1107 $this->cheapCache->resize( max( 2 * count( $paths ), self::CACHE_CHEAP_SIZE ) );
1108
1109 // Load from the persistent container caches
1110 $this->primeContainerCache( $paths );
1111 // Get the latest stat info for all the files (having locked them)
1112 $this->preloadFileStat( array( 'srcs' => $paths, 'latest' => true ) );
1113
1114 // Actually attempt the operation batch...
1115 $opts = $this->setConcurrencyFlags( $opts );
1116 $subStatus = FileOpBatch::attempt( $performOps, $opts, $this->fileJournal );
1117
1118 // Merge errors into status fields
1119 $status->merge( $subStatus );
1120 $status->success = $subStatus->success; // not done in merge()
1121
1122 // Shrink the stat cache back to normal size
1123 $this->cheapCache->resize( self::CACHE_CHEAP_SIZE );
1124
1125 return $status;
1126 }
1127
1128 final protected function doQuickOperationsInternal( array $ops ) {
1129 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
1130 $status = Status::newGood();
1131
1132 // Fix up custom header name/value pairs...
1133 $ops = array_map( array( $this, 'stripInvalidHeadersFromOp' ), $ops );
1134
1135 // Clear any file cache entries
1136 $this->clearCache();
1137
1138 $supportedOps = array( 'create', 'store', 'copy', 'move', 'delete', 'describe', 'null' );
1139 $async = ( $this->parallelize === 'implicit' && count( $ops ) > 1 );
1140 $maxConcurrency = $this->concurrency; // throttle
1141
1142 $statuses = array(); // array of (index => Status)
1143 $fileOpHandles = array(); // list of (index => handle) arrays
1144 $curFileOpHandles = array(); // current handle batch
1145 // Perform the sync-only ops and build up op handles for the async ops...
1146 foreach ( $ops as $index => $params ) {
1147 if ( !in_array( $params['op'], $supportedOps ) ) {
1148 throw new FileBackendError( "Operation '{$params['op']}' is not supported." );
1149 }
1150 $method = $params['op'] . 'Internal'; // e.g. "storeInternal"
1151 $subStatus = $this->$method( array( 'async' => $async ) + $params );
1152 if ( $subStatus->value instanceof FileBackendStoreOpHandle ) { // async
1153 if ( count( $curFileOpHandles ) >= $maxConcurrency ) {
1154 $fileOpHandles[] = $curFileOpHandles; // push this batch
1155 $curFileOpHandles = array();
1156 }
1157 $curFileOpHandles[$index] = $subStatus->value; // keep index
1158 } else { // error or completed
1159 $statuses[$index] = $subStatus; // keep index
1160 }
1161 }
1162 if ( count( $curFileOpHandles ) ) {
1163 $fileOpHandles[] = $curFileOpHandles; // last batch
1164 }
1165 // Do all the async ops that can be done concurrently...
1166 foreach ( $fileOpHandles as $fileHandleBatch ) {
1167 $statuses = $statuses + $this->executeOpHandlesInternal( $fileHandleBatch );
1168 }
1169 // Marshall and merge all the responses...
1170 foreach ( $statuses as $index => $subStatus ) {
1171 $status->merge( $subStatus );
1172 if ( $subStatus->isOK() ) {
1173 $status->success[$index] = true;
1174 ++$status->successCount;
1175 } else {
1176 $status->success[$index] = false;
1177 ++$status->failCount;
1178 }
1179 }
1180
1181 return $status;
1182 }
1183
1184 /**
1185 * Execute a list of FileBackendStoreOpHandle handles in parallel.
1186 * The resulting Status object fields will correspond
1187 * to the order in which the handles where given.
1188 *
1189 * @param array $fileOpHandles
1190 * @throws FileBackendError
1191 * @internal param array $handles List of FileBackendStoreOpHandle objects
1192 * @return array Map of Status objects
1193 */
1194 final public function executeOpHandlesInternal( array $fileOpHandles ) {
1195 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
1196
1197 foreach ( $fileOpHandles as $fileOpHandle ) {
1198 if ( !( $fileOpHandle instanceof FileBackendStoreOpHandle ) ) {
1199 throw new FileBackendError( "Given a non-FileBackendStoreOpHandle object." );
1200 } elseif ( $fileOpHandle->backend->getName() !== $this->getName() ) {
1201 throw new FileBackendError( "Given a FileBackendStoreOpHandle for the wrong backend." );
1202 }
1203 }
1204 $res = $this->doExecuteOpHandlesInternal( $fileOpHandles );
1205 foreach ( $fileOpHandles as $fileOpHandle ) {
1206 $fileOpHandle->closeResources();
1207 }
1208
1209 return $res;
1210 }
1211
1212 /**
1213 * @see FileBackendStore::executeOpHandlesInternal()
1214 * @param array $fileOpHandles
1215 * @throws FileBackendError
1216 * @return array List of corresponding Status objects
1217 */
1218 protected function doExecuteOpHandlesInternal( array $fileOpHandles ) {
1219 if ( count( $fileOpHandles ) ) {
1220 throw new FileBackendError( "This backend supports no asynchronous operations." );
1221 }
1222
1223 return array();
1224 }
1225
1226 /**
1227 * Strip long HTTP headers from a file operation.
1228 * Most headers are just numbers, but some are allowed to be long.
1229 * This function is useful for cleaning up headers and avoiding backend
1230 * specific errors, especially in the middle of batch file operations.
1231 *
1232 * @param array $op Same format as doOperation()
1233 * @return array
1234 */
1235 protected function stripInvalidHeadersFromOp( array $op ) {
1236 static $longs = array( 'Content-Disposition' );
1237 if ( isset( $op['headers'] ) ) { // op sets HTTP headers
1238 foreach ( $op['headers'] as $name => $value ) {
1239 $maxHVLen = in_array( $name, $longs ) ? INF : 255;
1240 if ( strlen( $name ) > 255 || strlen( $value ) > $maxHVLen ) {
1241 trigger_error( "Header '$name: $value' is too long." );
1242 unset( $op['headers'][$name] );
1243 } elseif ( !strlen( $value ) ) {
1244 $op['headers'][$name] = ''; // null/false => ""
1245 }
1246 }
1247 }
1248
1249 return $op;
1250 }
1251
1252 final public function preloadCache( array $paths ) {
1253 $fullConts = array(); // full container names
1254 foreach ( $paths as $path ) {
1255 list( $fullCont, , ) = $this->resolveStoragePath( $path );
1256 $fullConts[] = $fullCont;
1257 }
1258 // Load from the persistent file and container caches
1259 $this->primeContainerCache( $fullConts );
1260 $this->primeFileCache( $paths );
1261 }
1262
1263 final public function clearCache( array $paths = null ) {
1264 if ( is_array( $paths ) ) {
1265 $paths = array_map( 'FileBackend::normalizeStoragePath', $paths );
1266 $paths = array_filter( $paths, 'strlen' ); // remove nulls
1267 }
1268 if ( $paths === null ) {
1269 $this->cheapCache->clear();
1270 $this->expensiveCache->clear();
1271 } else {
1272 foreach ( $paths as $path ) {
1273 $this->cheapCache->clear( $path );
1274 $this->expensiveCache->clear( $path );
1275 }
1276 }
1277 $this->doClearCache( $paths );
1278 }
1279
1280 /**
1281 * Clears any additional stat caches for storage paths
1282 *
1283 * @see FileBackend::clearCache()
1284 *
1285 * @param array $paths Storage paths (optional)
1286 */
1287 protected function doClearCache( array $paths = null ) {
1288 }
1289
1290 final public function preloadFileStat( array $params ) {
1291 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
1292
1293 $params['concurrency'] = ( $this->parallelize !== 'off' ) ? $this->concurrency : 1;
1294 $stats = $this->doGetFileStatMulti( $params );
1295 if ( $stats === null ) {
1296 return; // not supported
1297 }
1298
1299 $latest = !empty( $params['latest'] ); // use latest data?
1300 foreach ( $stats as $path => $stat ) {
1301 $path = FileBackend::normalizeStoragePath( $path );
1302 if ( $path === null ) {
1303 continue; // this shouldn't happen
1304 }
1305 if ( is_array( $stat ) ) { // file exists
1306 // Strongly consistent backends can automatically set "latest"
1307 $stat['latest'] = isset( $stat['latest'] ) ? $stat['latest'] : $latest;
1308 $this->cheapCache->set( $path, 'stat', $stat );
1309 $this->setFileCache( $path, $stat ); // update persistent cache
1310 if ( isset( $stat['sha1'] ) ) { // some backends store SHA-1 as metadata
1311 $this->cheapCache->set( $path, 'sha1',
1312 array( 'hash' => $stat['sha1'], 'latest' => $latest ) );
1313 }
1314 if ( isset( $stat['xattr'] ) ) { // some backends store headers/metadata
1315 $stat['xattr'] = self::normalizeXAttributes( $stat['xattr'] );
1316 $this->cheapCache->set( $path, 'xattr',
1317 array( 'map' => $stat['xattr'], 'latest' => $latest ) );
1318 }
1319 } elseif ( $stat === false ) { // file does not exist
1320 $this->cheapCache->set( $path, 'stat',
1321 $latest ? 'NOT_EXIST_LATEST' : 'NOT_EXIST' );
1322 $this->cheapCache->set( $path, 'xattr',
1323 array( 'map' => false, 'latest' => $latest ) );
1324 $this->cheapCache->set( $path, 'sha1',
1325 array( 'hash' => false, 'latest' => $latest ) );
1326 wfDebug( __METHOD__ . ": File $path does not exist.\n" );
1327 } else { // an error occurred
1328 wfDebug( __METHOD__ . ": Could not stat file $path.\n" );
1329 }
1330 }
1331 }
1332
1333 /**
1334 * Get file stat information (concurrently if possible) for several files
1335 *
1336 * @see FileBackend::getFileStat()
1337 *
1338 * @param array $params Parameters include:
1339 * - srcs : list of source storage paths
1340 * - latest : use the latest available data
1341 * @return array|null Map of storage paths to array|bool|null (returns null if not supported)
1342 * @since 1.23
1343 */
1344 protected function doGetFileStatMulti( array $params ) {
1345 return null; // not supported
1346 }
1347
1348 /**
1349 * Is this a key/value store where directories are just virtual?
1350 * Virtual directories exists in so much as files exists that are
1351 * prefixed with the directory path followed by a forward slash.
1352 *
1353 * @return bool
1354 */
1355 abstract protected function directoriesAreVirtual();
1356
1357 /**
1358 * Check if a container name is valid.
1359 * This checks for for length and illegal characters.
1360 *
1361 * @param string $container
1362 * @return bool
1363 */
1364 final protected static function isValidContainerName( $container ) {
1365 // This accounts for Swift and S3 restrictions while leaving room
1366 // for things like '.xxx' (hex shard chars) or '.seg' (segments).
1367 // This disallows directory separators or traversal characters.
1368 // Note that matching strings URL encode to the same string;
1369 // in Swift, the length restriction is *after* URL encoding.
1370 return preg_match( '/^[a-z0-9][a-z0-9-_]{0,199}$/i', $container );
1371 }
1372
1373 /**
1374 * Splits a storage path into an internal container name,
1375 * an internal relative file name, and a container shard suffix.
1376 * Any shard suffix is already appended to the internal container name.
1377 * This also checks that the storage path is valid and within this backend.
1378 *
1379 * If the container is sharded but a suffix could not be determined,
1380 * this means that the path can only refer to a directory and can only
1381 * be scanned by looking in all the container shards.
1382 *
1383 * @param string $storagePath
1384 * @return array (container, path, container suffix) or (null, null, null) if invalid
1385 */
1386 final protected function resolveStoragePath( $storagePath ) {
1387 list( $backend, $container, $relPath ) = self::splitStoragePath( $storagePath );
1388 if ( $backend === $this->name ) { // must be for this backend
1389 $relPath = self::normalizeContainerPath( $relPath );
1390 if ( $relPath !== null ) {
1391 // Get shard for the normalized path if this container is sharded
1392 $cShard = $this->getContainerShard( $container, $relPath );
1393 // Validate and sanitize the relative path (backend-specific)
1394 $relPath = $this->resolveContainerPath( $container, $relPath );
1395 if ( $relPath !== null ) {
1396 // Prepend any wiki ID prefix to the container name
1397 $container = $this->fullContainerName( $container );
1398 if ( self::isValidContainerName( $container ) ) {
1399 // Validate and sanitize the container name (backend-specific)
1400 $container = $this->resolveContainerName( "{$container}{$cShard}" );
1401 if ( $container !== null ) {
1402 return array( $container, $relPath, $cShard );
1403 }
1404 }
1405 }
1406 }
1407 }
1408
1409 return array( null, null, null );
1410 }
1411
1412 /**
1413 * Like resolveStoragePath() except null values are returned if
1414 * the container is sharded and the shard could not be determined
1415 * or if the path ends with '/'. The later case is illegal for FS
1416 * backends and can confuse listings for object store backends.
1417 *
1418 * This function is used when resolving paths that must be valid
1419 * locations for files. Directory and listing functions should
1420 * generally just use resolveStoragePath() instead.
1421 *
1422 * @see FileBackendStore::resolveStoragePath()
1423 *
1424 * @param string $storagePath
1425 * @return array (container, path) or (null, null) if invalid
1426 */
1427 final protected function resolveStoragePathReal( $storagePath ) {
1428 list( $container, $relPath, $cShard ) = $this->resolveStoragePath( $storagePath );
1429 if ( $cShard !== null && substr( $relPath, -1 ) !== '/' ) {
1430 return array( $container, $relPath );
1431 }
1432
1433 return array( null, null );
1434 }
1435
1436 /**
1437 * Get the container name shard suffix for a given path.
1438 * Any empty suffix means the container is not sharded.
1439 *
1440 * @param string $container Container name
1441 * @param string $relPath Storage path relative to the container
1442 * @return string|null Returns null if shard could not be determined
1443 */
1444 final protected function getContainerShard( $container, $relPath ) {
1445 list( $levels, $base, $repeat ) = $this->getContainerHashLevels( $container );
1446 if ( $levels == 1 || $levels == 2 ) {
1447 // Hash characters are either base 16 or 36
1448 $char = ( $base == 36 ) ? '[0-9a-z]' : '[0-9a-f]';
1449 // Get a regex that represents the shard portion of paths.
1450 // The concatenation of the captures gives us the shard.
1451 if ( $levels === 1 ) { // 16 or 36 shards per container
1452 $hashDirRegex = '(' . $char . ')';
1453 } else { // 256 or 1296 shards per container
1454 if ( $repeat ) { // verbose hash dir format (e.g. "a/ab/abc")
1455 $hashDirRegex = $char . '/(' . $char . '{2})';
1456 } else { // short hash dir format (e.g. "a/b/c")
1457 $hashDirRegex = '(' . $char . ')/(' . $char . ')';
1458 }
1459 }
1460 // Allow certain directories to be above the hash dirs so as
1461 // to work with FileRepo (e.g. "archive/a/ab" or "temp/a/ab").
1462 // They must be 2+ chars to avoid any hash directory ambiguity.
1463 $m = array();
1464 if ( preg_match( "!^(?:[^/]{2,}/)*$hashDirRegex(?:/|$)!", $relPath, $m ) ) {
1465 return '.' . implode( '', array_slice( $m, 1 ) );
1466 }
1467
1468 return null; // failed to match
1469 }
1470
1471 return ''; // no sharding
1472 }
1473
1474 /**
1475 * Check if a storage path maps to a single shard.
1476 * Container dirs like "a", where the container shards on "x/xy",
1477 * can reside on several shards. Such paths are tricky to handle.
1478 *
1479 * @param string $storagePath Storage path
1480 * @return bool
1481 */
1482 final public function isSingleShardPathInternal( $storagePath ) {
1483 list( , , $shard ) = $this->resolveStoragePath( $storagePath );
1484
1485 return ( $shard !== null );
1486 }
1487
1488 /**
1489 * Get the sharding config for a container.
1490 * If greater than 0, then all file storage paths within
1491 * the container are required to be hashed accordingly.
1492 *
1493 * @param string $container
1494 * @return array (integer levels, integer base, repeat flag) or (0, 0, false)
1495 */
1496 final protected function getContainerHashLevels( $container ) {
1497 if ( isset( $this->shardViaHashLevels[$container] ) ) {
1498 $config = $this->shardViaHashLevels[$container];
1499 $hashLevels = (int)$config['levels'];
1500 if ( $hashLevels == 1 || $hashLevels == 2 ) {
1501 $hashBase = (int)$config['base'];
1502 if ( $hashBase == 16 || $hashBase == 36 ) {
1503 return array( $hashLevels, $hashBase, $config['repeat'] );
1504 }
1505 }
1506 }
1507
1508 return array( 0, 0, false ); // no sharding
1509 }
1510
1511 /**
1512 * Get a list of full container shard suffixes for a container
1513 *
1514 * @param string $container
1515 * @return array
1516 */
1517 final protected function getContainerSuffixes( $container ) {
1518 $shards = array();
1519 list( $digits, $base ) = $this->getContainerHashLevels( $container );
1520 if ( $digits > 0 ) {
1521 $numShards = pow( $base, $digits );
1522 for ( $index = 0; $index < $numShards; $index++ ) {
1523 $shards[] = '.' . wfBaseConvert( $index, 10, $base, $digits );
1524 }
1525 }
1526
1527 return $shards;
1528 }
1529
1530 /**
1531 * Get the full container name, including the wiki ID prefix
1532 *
1533 * @param string $container
1534 * @return string
1535 */
1536 final protected function fullContainerName( $container ) {
1537 if ( $this->wikiId != '' ) {
1538 return "{$this->wikiId}-$container";
1539 } else {
1540 return $container;
1541 }
1542 }
1543
1544 /**
1545 * Resolve a container name, checking if it's allowed by the backend.
1546 * This is intended for internal use, such as encoding illegal chars.
1547 * Subclasses can override this to be more restrictive.
1548 *
1549 * @param string $container
1550 * @return string|null
1551 */
1552 protected function resolveContainerName( $container ) {
1553 return $container;
1554 }
1555
1556 /**
1557 * Resolve a relative storage path, checking if it's allowed by the backend.
1558 * This is intended for internal use, such as encoding illegal chars or perhaps
1559 * getting absolute paths (e.g. FS based backends). Note that the relative path
1560 * may be the empty string (e.g. the path is simply to the container).
1561 *
1562 * @param string $container Container name
1563 * @param string $relStoragePath Storage path relative to the container
1564 * @return string|null Path or null if not valid
1565 */
1566 protected function resolveContainerPath( $container, $relStoragePath ) {
1567 return $relStoragePath;
1568 }
1569
1570 /**
1571 * Get the cache key for a container
1572 *
1573 * @param string $container Resolved container name
1574 * @return string
1575 */
1576 private function containerCacheKey( $container ) {
1577 return "filebackend:{$this->name}:{$this->wikiId}:container:{$container}";
1578 }
1579
1580 /**
1581 * Set the cached info for a container
1582 *
1583 * @param string $container Resolved container name
1584 * @param array $val Information to cache
1585 */
1586 final protected function setContainerCache( $container, array $val ) {
1587 $this->memCache->add( $this->containerCacheKey( $container ), $val, 14 * 86400 );
1588 }
1589
1590 /**
1591 * Delete the cached info for a container.
1592 * The cache key is salted for a while to prevent race conditions.
1593 *
1594 * @param string $container Resolved container name
1595 */
1596 final protected function deleteContainerCache( $container ) {
1597 if ( !$this->memCache->set( $this->containerCacheKey( $container ), 'PURGED', 300 ) ) {
1598 trigger_error( "Unable to delete stat cache for container $container." );
1599 }
1600 }
1601
1602 /**
1603 * Do a batch lookup from cache for container stats for all containers
1604 * used in a list of container names or storage paths objects.
1605 * This loads the persistent cache values into the process cache.
1606 *
1607 * @param array $items
1608 */
1609 final protected function primeContainerCache( array $items ) {
1610 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
1611
1612 $paths = array(); // list of storage paths
1613 $contNames = array(); // (cache key => resolved container name)
1614 // Get all the paths/containers from the items...
1615 foreach ( $items as $item ) {
1616 if ( self::isStoragePath( $item ) ) {
1617 $paths[] = $item;
1618 } elseif ( is_string( $item ) ) { // full container name
1619 $contNames[$this->containerCacheKey( $item )] = $item;
1620 }
1621 }
1622 // Get all the corresponding cache keys for paths...
1623 foreach ( $paths as $path ) {
1624 list( $fullCont, , ) = $this->resolveStoragePath( $path );
1625 if ( $fullCont !== null ) { // valid path for this backend
1626 $contNames[$this->containerCacheKey( $fullCont )] = $fullCont;
1627 }
1628 }
1629
1630 $contInfo = array(); // (resolved container name => cache value)
1631 // Get all cache entries for these container cache keys...
1632 $values = $this->memCache->getMulti( array_keys( $contNames ) );
1633 foreach ( $values as $cacheKey => $val ) {
1634 $contInfo[$contNames[$cacheKey]] = $val;
1635 }
1636
1637 // Populate the container process cache for the backend...
1638 $this->doPrimeContainerCache( array_filter( $contInfo, 'is_array' ) );
1639 }
1640
1641 /**
1642 * Fill the backend-specific process cache given an array of
1643 * resolved container names and their corresponding cached info.
1644 * Only containers that actually exist should appear in the map.
1645 *
1646 * @param array $containerInfo Map of resolved container names to cached info
1647 */
1648 protected function doPrimeContainerCache( array $containerInfo ) {
1649 }
1650
1651 /**
1652 * Get the cache key for a file path
1653 *
1654 * @param string $path Normalized storage path
1655 * @return string
1656 */
1657 private function fileCacheKey( $path ) {
1658 return "filebackend:{$this->name}:{$this->wikiId}:file:" . sha1( $path );
1659 }
1660
1661 /**
1662 * Set the cached stat info for a file path.
1663 * Negatives (404s) are not cached. By not caching negatives, we can skip cache
1664 * salting for the case when a file is created at a path were there was none before.
1665 *
1666 * @param string $path Storage path
1667 * @param array $val Stat information to cache
1668 */
1669 final protected function setFileCache( $path, array $val ) {
1670 $path = FileBackend::normalizeStoragePath( $path );
1671 if ( $path === null ) {
1672 return; // invalid storage path
1673 }
1674 $age = time() - wfTimestamp( TS_UNIX, $val['mtime'] );
1675 $ttl = min( 7 * 86400, max( 300, floor( .1 * $age ) ) );
1676 $key = $this->fileCacheKey( $path );
1677 // Set the cache unless it is currently salted with the value "PURGED".
1678 // Using add() handles this except it also is a no-op in that case where
1679 // the current value is not "latest" but $val is, so use CAS in that case.
1680 if ( !$this->memCache->add( $key, $val, $ttl ) && !empty( $val['latest'] ) ) {
1681 $this->memCache->merge(
1682 $key,
1683 function( BagOStuff $cache, $key, $cValue ) use ( $val ) {
1684 return ( is_array( $cValue ) && empty( $cValue['latest'] ) )
1685 ? $val // update the stat cache with the lastest info
1686 : false; // do nothing (cache is salted or some error happened)
1687 },
1688 $ttl,
1689 1
1690 );
1691 }
1692 }
1693
1694 /**
1695 * Delete the cached stat info for a file path.
1696 * The cache key is salted for a while to prevent race conditions.
1697 * Since negatives (404s) are not cached, this does not need to be called when
1698 * a file is created at a path were there was none before.
1699 *
1700 * @param string $path Storage path
1701 */
1702 final protected function deleteFileCache( $path ) {
1703 $path = FileBackend::normalizeStoragePath( $path );
1704 if ( $path === null ) {
1705 return; // invalid storage path
1706 }
1707 if ( !$this->memCache->set( $this->fileCacheKey( $path ), 'PURGED', 300 ) ) {
1708 trigger_error( "Unable to delete stat cache for file $path." );
1709 }
1710 }
1711
1712 /**
1713 * Do a batch lookup from cache for file stats for all paths
1714 * used in a list of storage paths or FileOp objects.
1715 * This loads the persistent cache values into the process cache.
1716 *
1717 * @param array $items List of storage paths
1718 */
1719 final protected function primeFileCache( array $items ) {
1720 $section = new ProfileSection( __METHOD__ . "-{$this->name}" );
1721
1722 $paths = array(); // list of storage paths
1723 $pathNames = array(); // (cache key => storage path)
1724 // Get all the paths/containers from the items...
1725 foreach ( $items as $item ) {
1726 if ( self::isStoragePath( $item ) ) {
1727 $paths[] = FileBackend::normalizeStoragePath( $item );
1728 }
1729 }
1730 // Get rid of any paths that failed normalization...
1731 $paths = array_filter( $paths, 'strlen' ); // remove nulls
1732 // Get all the corresponding cache keys for paths...
1733 foreach ( $paths as $path ) {
1734 list( , $rel, ) = $this->resolveStoragePath( $path );
1735 if ( $rel !== null ) { // valid path for this backend
1736 $pathNames[$this->fileCacheKey( $path )] = $path;
1737 }
1738 }
1739 // Get all cache entries for these container cache keys...
1740 $values = $this->memCache->getMulti( array_keys( $pathNames ) );
1741 foreach ( $values as $cacheKey => $val ) {
1742 if ( is_array( $val ) ) {
1743 $path = $pathNames[$cacheKey];
1744 $this->cheapCache->set( $path, 'stat', $val );
1745 if ( isset( $val['sha1'] ) ) { // some backends store SHA-1 as metadata
1746 $this->cheapCache->set( $path, 'sha1',
1747 array( 'hash' => $val['sha1'], 'latest' => $val['latest'] ) );
1748 }
1749 if ( isset( $val['xattr'] ) ) { // some backends store headers/metadata
1750 $val['xattr'] = self::normalizeXAttributes( $val['xattr'] );
1751 $this->cheapCache->set( $path, 'xattr',
1752 array( 'map' => $val['xattr'], 'latest' => $val['latest'] ) );
1753 }
1754 }
1755 }
1756 }
1757
1758 /**
1759 * Normalize file headers/metadata to the FileBackend::getFileXAttributes() format
1760 *
1761 * @param array $xattr
1762 * @return array
1763 * @since 1.22
1764 */
1765 final protected static function normalizeXAttributes( array $xattr ) {
1766 $newXAttr = array( 'headers' => array(), 'metadata' => array() );
1767
1768 foreach ( $xattr['headers'] as $name => $value ) {
1769 $newXAttr['headers'][strtolower( $name )] = $value;
1770 }
1771
1772 foreach ( $xattr['metadata'] as $name => $value ) {
1773 $newXAttr['metadata'][strtolower( $name )] = $value;
1774 }
1775
1776 return $newXAttr;
1777 }
1778
1779 /**
1780 * Set the 'concurrency' option from a list of operation options
1781 *
1782 * @param array $opts Map of operation options
1783 * @return array
1784 */
1785 final protected function setConcurrencyFlags( array $opts ) {
1786 $opts['concurrency'] = 1; // off
1787 if ( $this->parallelize === 'implicit' ) {
1788 if ( !isset( $opts['parallelize'] ) || $opts['parallelize'] ) {
1789 $opts['concurrency'] = $this->concurrency;
1790 }
1791 } elseif ( $this->parallelize === 'explicit' ) {
1792 if ( !empty( $opts['parallelize'] ) ) {
1793 $opts['concurrency'] = $this->concurrency;
1794 }
1795 }
1796
1797 return $opts;
1798 }
1799
1800 /**
1801 * Get the content type to use in HEAD/GET requests for a file
1802 *
1803 * @param string $storagePath
1804 * @param string|null $content File data
1805 * @param string|null $fsPath File system path
1806 * @return MIME type
1807 */
1808 protected function getContentType( $storagePath, $content, $fsPath ) {
1809 return call_user_func_array( $this->mimeCallback, func_get_args() );
1810 }
1811 }
1812
1813 /**
1814 * FileBackendStore helper class for performing asynchronous file operations.
1815 *
1816 * For example, calling FileBackendStore::createInternal() with the "async"
1817 * param flag may result in a Status that contains this object as a value.
1818 * This class is largely backend-specific and is mostly just "magic" to be
1819 * passed to FileBackendStore::executeOpHandlesInternal().
1820 */
1821 abstract class FileBackendStoreOpHandle {
1822 /** @var array */
1823 public $params = array(); // params to caller functions
1824 /** @var FileBackendStore */
1825 public $backend;
1826 /** @var array */
1827 public $resourcesToClose = array();
1828
1829 public $call; // string; name that identifies the function called
1830
1831 /**
1832 * Close all open file handles
1833 */
1834 public function closeResources() {
1835 array_map( 'fclose', $this->resourcesToClose );
1836 }
1837 }
1838
1839 /**
1840 * FileBackendStore helper function to handle listings that span container shards.
1841 * Do not use this class from places outside of FileBackendStore.
1842 *
1843 * @ingroup FileBackend
1844 */
1845 abstract class FileBackendStoreShardListIterator extends FilterIterator {
1846 /** @var FileBackendStore */
1847 protected $backend;
1848
1849 /** @var array */
1850 protected $params;
1851
1852 /** @var string Full container name */
1853 protected $container;
1854
1855 /** @var string Resolved relative path */
1856 protected $directory;
1857
1858 /** @var array */
1859 protected $multiShardPaths = array(); // (rel path => 1)
1860
1861 /**
1862 * @param FileBackendStore $backend
1863 * @param string $container Full storage container name
1864 * @param string $dir Storage directory relative to container
1865 * @param array $suffixes List of container shard suffixes
1866 * @param array $params
1867 */
1868 public function __construct(
1869 FileBackendStore $backend, $container, $dir, array $suffixes, array $params
1870 ) {
1871 $this->backend = $backend;
1872 $this->container = $container;
1873 $this->directory = $dir;
1874 $this->params = $params;
1875
1876 $iter = new AppendIterator();
1877 foreach ( $suffixes as $suffix ) {
1878 $iter->append( $this->listFromShard( $this->container . $suffix ) );
1879 }
1880
1881 parent::__construct( $iter );
1882 }
1883
1884 public function accept() {
1885 $rel = $this->getInnerIterator()->current(); // path relative to given directory
1886 $path = $this->params['dir'] . "/{$rel}"; // full storage path
1887 if ( $this->backend->isSingleShardPathInternal( $path ) ) {
1888 return true; // path is only on one shard; no issue with duplicates
1889 } elseif ( isset( $this->multiShardPaths[$rel] ) ) {
1890 // Don't keep listing paths that are on multiple shards
1891 return false;
1892 } else {
1893 $this->multiShardPaths[$rel] = 1;
1894
1895 return true;
1896 }
1897 }
1898
1899 public function rewind() {
1900 parent::rewind();
1901 $this->multiShardPaths = array();
1902 }
1903
1904 /**
1905 * Get the list for a given container shard
1906 *
1907 * @param string $container Resolved container name
1908 * @return Iterator
1909 */
1910 abstract protected function listFromShard( $container );
1911 }
1912
1913 /**
1914 * Iterator for listing directories
1915 */
1916 class FileBackendStoreShardDirIterator extends FileBackendStoreShardListIterator {
1917 protected function listFromShard( $container ) {
1918 $list = $this->backend->getDirectoryListInternal(
1919 $container, $this->directory, $this->params );
1920 if ( $list === null ) {
1921 return new ArrayIterator( array() );
1922 } else {
1923 return is_array( $list ) ? new ArrayIterator( $list ) : $list;
1924 }
1925 }
1926 }
1927
1928 /**
1929 * Iterator for listing regular files
1930 */
1931 class FileBackendStoreShardFileIterator extends FileBackendStoreShardListIterator {
1932 protected function listFromShard( $container ) {
1933 $list = $this->backend->getFileListInternal(
1934 $container, $this->directory, $this->params );
1935 if ( $list === null ) {
1936 return new ArrayIterator( array() );
1937 } else {
1938 return is_array( $list ) ? new ArrayIterator( $list ) : $list;
1939 }
1940 }
1941 }