MediaWiki fundraising/REL1_35
FileBackendStore.php
Go to the documentation of this file.
1<?php
24use Wikimedia\AtEase\AtEase;
25use Wikimedia\Timestamp\ConvertibleTimestamp;
26
41abstract class FileBackendStore extends FileBackend {
43 protected $memCache;
45 protected $srvCache;
47 protected $cheapCache;
49 protected $expensiveCache;
50
52 protected $shardViaHashLevels = [];
53
55 protected $mimeCallback;
56
57 protected $maxFileSize = 4294967296; // integer bytes (4GiB)
58
59 protected const CACHE_TTL = 10; // integer; TTL in seconds for process cache entries
60 protected const CACHE_CHEAP_SIZE = 500; // integer; max entries in "cheap cache"
61 protected const CACHE_EXPENSIVE_SIZE = 5; // integer; max entries in "expensive cache"
62
64 protected static $RES_ABSENT = false;
66 protected static $RES_ERROR = null;
67
69 protected static $ABSENT_NORMAL = 'FNE-N';
71 protected static $ABSENT_LATEST = 'FNE-L';
72
86 public function __construct( array $config ) {
87 parent::__construct( $config );
88 $this->mimeCallback = $config['mimeCallback'] ?? null;
89 $this->srvCache = new EmptyBagOStuff(); // disabled by default
90 $this->memCache = WANObjectCache::newEmpty(); // disabled by default
91 $this->cheapCache = new MapCacheLRU( self::CACHE_CHEAP_SIZE );
92 $this->expensiveCache = new MapCacheLRU( self::CACHE_EXPENSIVE_SIZE );
93 }
94
102 final public function maxFileSizeInternal() {
103 return $this->maxFileSize;
104 }
105
116 abstract public function isPathUsableInternal( $storagePath );
117
136 final public function createInternal( array $params ) {
138 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
139
140 if ( strlen( $params['content'] ) > $this->maxFileSizeInternal() ) {
141 $status = $this->newStatus( 'backend-fail-maxsize',
142 $params['dst'], $this->maxFileSizeInternal() );
143 } else {
144 $status = $this->doCreateInternal( $params );
145 $this->clearCache( [ $params['dst'] ] );
146 if ( $params['dstExists'] ?? true ) {
147 $this->deleteFileCache( $params['dst'] ); // persistent cache
148 }
149 }
150
151 return $status;
152 }
153
159 abstract protected function doCreateInternal( array $params );
160
179 final public function storeInternal( array $params ) {
181 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
182
183 if ( filesize( $params['src'] ) > $this->maxFileSizeInternal() ) {
184 $status = $this->newStatus( 'backend-fail-maxsize',
185 $params['dst'], $this->maxFileSizeInternal() );
186 } else {
187 $status = $this->doStoreInternal( $params );
188 $this->clearCache( [ $params['dst'] ] );
189 if ( $params['dstExists'] ?? true ) {
190 $this->deleteFileCache( $params['dst'] ); // persistent cache
191 }
192 }
193
194 return $status;
195 }
196
202 abstract protected function doStoreInternal( array $params );
203
223 final public function copyInternal( array $params ) {
225 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
226
227 $status = $this->doCopyInternal( $params );
228 $this->clearCache( [ $params['dst'] ] );
229 if ( $params['dstExists'] ?? true ) {
230 $this->deleteFileCache( $params['dst'] ); // persistent cache
231 }
232
233 return $status;
234 }
235
241 abstract protected function doCopyInternal( array $params );
242
257 final public function deleteInternal( array $params ) {
259 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
260
261 $status = $this->doDeleteInternal( $params );
262 $this->clearCache( [ $params['src'] ] );
263 $this->deleteFileCache( $params['src'] ); // persistent cache
264 return $status;
265 }
266
272 abstract protected function doDeleteInternal( array $params );
273
293 final public function moveInternal( array $params ) {
295 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
296
297 $status = $this->doMoveInternal( $params );
298 $this->clearCache( [ $params['src'], $params['dst'] ] );
299 $this->deleteFileCache( $params['src'] ); // persistent cache
300 if ( $params['dstExists'] ?? true ) {
301 $this->deleteFileCache( $params['dst'] ); // persistent cache
302 }
303
304 return $status;
305 }
306
313 protected function doMoveInternal( array $params ) {
314 unset( $params['async'] ); // two steps, won't work here :)
315 $nsrc = FileBackend::normalizeStoragePath( $params['src'] );
316 $ndst = FileBackend::normalizeStoragePath( $params['dst'] );
317 // Copy source to dest
318 $status = $this->copyInternal( $params );
319 if ( $nsrc !== $ndst && $status->isOK() ) {
320 // Delete source (only fails due to races or network problems)
321 $status->merge( $this->deleteInternal( [ 'src' => $params['src'] ] ) );
322 $status->setResult( true, $status->value ); // ignore delete() errors
323 }
324
325 return $status;
326 }
327
342 final public function describeInternal( array $params ) {
344 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
345
346 if ( count( $params['headers'] ) ) {
347 $status = $this->doDescribeInternal( $params );
348 $this->clearCache( [ $params['src'] ] );
349 $this->deleteFileCache( $params['src'] ); // persistent cache
350 } else {
351 $status = $this->newStatus(); // nothing to do
352 }
353
354 return $status;
355 }
356
363 protected function doDescribeInternal( array $params ) {
364 return $this->newStatus();
365 }
366
374 final public function nullInternal( array $params ) {
375 return $this->newStatus();
376 }
377
378 final public function concatenate( array $params ) {
380 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
381 $status = $this->newStatus();
382
383 // Try to lock the source files for the scope of this function
385 $scopeLockS = $this->getScopedFileLocks( $params['srcs'], LockManager::LOCK_UW, $status );
386 if ( $status->isOK() ) {
387 // Actually do the file concatenation...
388 $start_time = microtime( true );
389 $status->merge( $this->doConcatenate( $params ) );
390 $sec = microtime( true ) - $start_time;
391 if ( !$status->isOK() ) {
392 $this->logger->error( static::class . "-{$this->name}" .
393 " failed to concatenate " . count( $params['srcs'] ) . " file(s) [$sec sec]" );
394 }
395 }
396
397 return $status;
398 }
399
406 protected function doConcatenate( array $params ) {
407 $status = $this->newStatus();
408 $tmpPath = $params['dst']; // convenience
409 unset( $params['latest'] ); // sanity
410
411 // Check that the specified temp file is valid...
412 AtEase::suppressWarnings();
413 $ok = ( is_file( $tmpPath ) && filesize( $tmpPath ) == 0 );
414 AtEase::restoreWarnings();
415 if ( !$ok ) { // not present or not empty
416 $status->fatal( 'backend-fail-opentemp', $tmpPath );
417
418 return $status;
419 }
420
421 // Get local FS versions of the chunks needed for the concatenation...
422 $fsFiles = $this->getLocalReferenceMulti( $params );
423 foreach ( $fsFiles as $path => &$fsFile ) {
424 if ( !$fsFile ) { // chunk failed to download?
425 $fsFile = $this->getLocalReference( [ 'src' => $path ] );
426 if ( !$fsFile ) { // retry failed?
427 $status->fatal( 'backend-fail-read', $path );
428
429 return $status;
430 }
431 }
432 }
433 unset( $fsFile ); // unset reference so we can reuse $fsFile
434
435 // Get a handle for the destination temp file
436 $tmpHandle = fopen( $tmpPath, 'ab' );
437 if ( $tmpHandle === false ) {
438 $status->fatal( 'backend-fail-opentemp', $tmpPath );
439
440 return $status;
441 }
442
443 // Build up the temp file using the source chunks (in order)...
444 foreach ( $fsFiles as $virtualSource => $fsFile ) {
445 // Get a handle to the local FS version
446 $sourceHandle = fopen( $fsFile->getPath(), 'rb' );
447 if ( $sourceHandle === false ) {
448 fclose( $tmpHandle );
449 $status->fatal( 'backend-fail-read', $virtualSource );
450
451 return $status;
452 }
453 // Append chunk to file (pass chunk size to avoid magic quotes)
454 if ( !stream_copy_to_stream( $sourceHandle, $tmpHandle ) ) {
455 fclose( $sourceHandle );
456 fclose( $tmpHandle );
457 $status->fatal( 'backend-fail-writetemp', $tmpPath );
458
459 return $status;
460 }
461 fclose( $sourceHandle );
462 }
463 if ( !fclose( $tmpHandle ) ) {
464 $status->fatal( 'backend-fail-closetemp', $tmpPath );
465
466 return $status;
467 }
468
469 clearstatcache(); // temp file changed
470
471 return $status;
472 }
473
474 final protected function doPrepare( array $params ) {
476 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
477 $status = $this->newStatus();
478
479 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
480 if ( $dir === null ) {
481 $status->fatal( 'backend-fail-invalidpath', $params['dir'] );
482
483 return $status; // invalid storage path
484 }
485
486 if ( $shard !== null ) { // confined to a single container/shard
487 $status->merge( $this->doPrepareInternal( $fullCont, $dir, $params ) );
488 } else { // directory is on several shards
489 $this->logger->debug( __METHOD__ . ": iterating over all container shards." );
490 list( , $shortCont, ) = self::splitStoragePath( $params['dir'] );
491 foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
492 $status->merge( $this->doPrepareInternal( "{$fullCont}{$suffix}", $dir, $params ) );
493 }
494 }
495
496 return $status;
497 }
498
507 protected function doPrepareInternal( $container, $dir, array $params ) {
508 return $this->newStatus();
509 }
510
511 final protected function doSecure( array $params ) {
513 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
514 $status = $this->newStatus();
515
516 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
517 if ( $dir === null ) {
518 $status->fatal( 'backend-fail-invalidpath', $params['dir'] );
519
520 return $status; // invalid storage path
521 }
522
523 if ( $shard !== null ) { // confined to a single container/shard
524 $status->merge( $this->doSecureInternal( $fullCont, $dir, $params ) );
525 } else { // directory is on several shards
526 $this->logger->debug( __METHOD__ . ": iterating over all container shards." );
527 list( , $shortCont, ) = self::splitStoragePath( $params['dir'] );
528 foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
529 $status->merge( $this->doSecureInternal( "{$fullCont}{$suffix}", $dir, $params ) );
530 }
531 }
532
533 return $status;
534 }
535
544 protected function doSecureInternal( $container, $dir, array $params ) {
545 return $this->newStatus();
546 }
547
548 final protected function doPublish( array $params ) {
550 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
551 $status = $this->newStatus();
552
553 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
554 if ( $dir === null ) {
555 $status->fatal( 'backend-fail-invalidpath', $params['dir'] );
556
557 return $status; // invalid storage path
558 }
559
560 if ( $shard !== null ) { // confined to a single container/shard
561 $status->merge( $this->doPublishInternal( $fullCont, $dir, $params ) );
562 } else { // directory is on several shards
563 $this->logger->debug( __METHOD__ . ": iterating over all container shards." );
564 list( , $shortCont, ) = self::splitStoragePath( $params['dir'] );
565 foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
566 $status->merge( $this->doPublishInternal( "{$fullCont}{$suffix}", $dir, $params ) );
567 }
568 }
569
570 return $status;
571 }
572
581 protected function doPublishInternal( $container, $dir, array $params ) {
582 return $this->newStatus();
583 }
584
585 final protected function doClean( array $params ) {
587 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
588 $status = $this->newStatus();
589
590 // Recursive: first delete all empty subdirs recursively
591 if ( !empty( $params['recursive'] ) && !$this->directoriesAreVirtual() ) {
592 $subDirsRel = $this->getTopDirectoryList( [ 'dir' => $params['dir'] ] );
593 if ( $subDirsRel !== null ) { // no errors
594 foreach ( $subDirsRel as $subDirRel ) {
595 $subDir = $params['dir'] . "/{$subDirRel}"; // full path
596 $status->merge( $this->doClean( [ 'dir' => $subDir ] + $params ) );
597 }
598 unset( $subDirsRel ); // free directory for rmdir() on Windows (for FS backends)
599 }
600 }
601
602 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
603 if ( $dir === null ) {
604 $status->fatal( 'backend-fail-invalidpath', $params['dir'] );
605
606 return $status; // invalid storage path
607 }
608
609 // Attempt to lock this directory...
610 $filesLockEx = [ $params['dir'] ];
612 $scopedLockE = $this->getScopedFileLocks( $filesLockEx, LockManager::LOCK_EX, $status );
613 if ( !$status->isOK() ) {
614 return $status; // abort
615 }
616
617 if ( $shard !== null ) { // confined to a single container/shard
618 $status->merge( $this->doCleanInternal( $fullCont, $dir, $params ) );
619 $this->deleteContainerCache( $fullCont ); // purge cache
620 } else { // directory is on several shards
621 $this->logger->debug( __METHOD__ . ": iterating over all container shards." );
622 list( , $shortCont, ) = self::splitStoragePath( $params['dir'] );
623 foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
624 $status->merge( $this->doCleanInternal( "{$fullCont}{$suffix}", $dir, $params ) );
625 $this->deleteContainerCache( "{$fullCont}{$suffix}" ); // purge cache
626 }
627 }
628
629 return $status;
630 }
631
640 protected function doCleanInternal( $container, $dir, array $params ) {
641 return $this->newStatus();
642 }
643
644 final public function fileExists( array $params ) {
646 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
647
648 $stat = $this->getFileStat( $params );
649 if ( is_array( $stat ) ) {
650 return true;
651 }
652
653 return ( $stat === self::$RES_ABSENT ) ? false : self::EXISTENCE_ERROR;
654 }
655
656 final public function getFileTimestamp( array $params ) {
658 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
659
660 $stat = $this->getFileStat( $params );
661 if ( is_array( $stat ) ) {
662 return $stat['mtime'];
663 }
664
665 return self::TIMESTAMP_FAIL; // all failure cases
666 }
667
668 final public function getFileSize( array $params ) {
670 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
671
672 $stat = $this->getFileStat( $params );
673 if ( is_array( $stat ) ) {
674 return $stat['size'];
675 }
676
677 return self::SIZE_FAIL; // all failure cases
678 }
679
680 final public function getFileStat( array $params ) {
682 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
683
684 $path = self::normalizeStoragePath( $params['src'] );
685 if ( $path === null ) {
686 return self::STAT_ERROR; // invalid storage path
687 }
688
689 // Whether to bypass cache except for process cache entries loaded directly from
690 // high consistency backend queries (caller handles any cache flushing and locking)
691 $latest = !empty( $params['latest'] );
692 // Whether to ignore cache entries missing the SHA-1 field for existing files
693 $requireSHA1 = !empty( $params['requireSHA1'] );
694
695 $stat = $this->cheapCache->getField( $path, 'stat', self::CACHE_TTL );
696 // Load the persistent stat cache into process cache if needed
697 if ( !$latest ) {
698 if (
699 // File stat is not in process cache
700 $stat === null ||
701 // Key/value store backends might opportunistically set file stat process
702 // cache entries from object listings that do not include the SHA-1. In that
703 // case, loading the persistent stat cache will likely yield the SHA-1.
704 ( $requireSHA1 && is_array( $stat ) && !isset( $stat['sha1'] ) )
705 ) {
706 $this->primeFileCache( [ $path ] );
707 // Get any newly process-cached entry
708 $stat = $this->cheapCache->getField( $path, 'stat', self::CACHE_TTL );
709 }
710 }
711
712 if ( is_array( $stat ) ) {
713 if (
714 ( !$latest || $stat['latest'] ) &&
715 ( !$requireSHA1 || isset( $stat['sha1'] ) )
716 ) {
717 return $stat;
718 }
719 } elseif ( $stat === self::$ABSENT_LATEST ) {
720 return self::STAT_ABSENT;
721 } elseif ( $stat === self::$ABSENT_NORMAL ) {
722 if ( !$latest ) {
723 return self::STAT_ABSENT;
724 }
725 }
726
727 // Load the file stat from the backend and update caches
728 $stat = $this->doGetFileStat( $params );
729 $this->ingestFreshFileStats( [ $path => $stat ], $latest );
730
731 if ( is_array( $stat ) ) {
732 return $stat;
733 }
734
735 return ( $stat === self::$RES_ERROR ) ? self::STAT_ERROR : self::STAT_ABSENT;
736 }
737
745 final protected function ingestFreshFileStats( array $stats, $latest ) {
746 $success = true;
747
748 foreach ( $stats as $path => $stat ) {
749 if ( is_array( $stat ) ) {
750 // Strongly consistent backends might automatically set this flag
751 $stat['latest'] = $stat['latest'] ?? $latest;
752
753 $this->cheapCache->setField( $path, 'stat', $stat );
754 if ( isset( $stat['sha1'] ) ) {
755 // Some backends store the SHA-1 hash as metadata
756 $this->cheapCache->setField(
757 $path,
758 'sha1',
759 [ 'hash' => $stat['sha1'], 'latest' => $latest ]
760 );
761 }
762 if ( isset( $stat['xattr'] ) ) {
763 // Some backends store custom headers/metadata
764 $stat['xattr'] = self::normalizeXAttributes( $stat['xattr'] );
765 $this->cheapCache->setField(
766 $path,
767 'xattr',
768 [ 'map' => $stat['xattr'], 'latest' => $latest ]
769 );
770 }
771 // Update persistent cache (@TODO: set all entries in one batch)
772 $this->setFileCache( $path, $stat );
773 } elseif ( $stat === self::$RES_ABSENT ) {
774 $this->cheapCache->setField(
775 $path,
776 'stat',
777 $latest ? self::$ABSENT_LATEST : self::$ABSENT_NORMAL
778 );
779 $this->cheapCache->setField(
780 $path,
781 'xattr',
782 [ 'map' => self::XATTRS_FAIL, 'latest' => $latest ]
783 );
784 $this->cheapCache->setField(
785 $path,
786 'sha1',
787 [ 'hash' => self::SHA1_FAIL, 'latest' => $latest ]
788 );
789 $this->logger->debug(
790 __METHOD__ . ': File {path} does not exist',
791 [ 'path' => $path ]
792 );
793 } else {
794 $success = false;
795 $this->logger->error(
796 __METHOD__ . ': Could not stat file {path}',
797 [ 'path' => $path ]
798 );
799 }
800 }
801
802 return $success;
803 }
804
809 abstract protected function doGetFileStat( array $params );
810
811 public function getFileContentsMulti( array $params ) {
813 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
814
815 $params = $this->setConcurrencyFlags( $params );
816 $contents = $this->doGetFileContentsMulti( $params );
817 foreach ( $contents as $path => $content ) {
818 if ( !is_string( $content ) ) {
819 $contents[$path] = self::CONTENT_FAIL; // used for all failure cases
820 }
821 }
822
823 return $contents;
824 }
825
832 protected function doGetFileContentsMulti( array $params ) {
833 $contents = [];
834 foreach ( $this->doGetLocalReferenceMulti( $params ) as $path => $fsFile ) {
835 if ( $fsFile instanceof FSFile ) {
836 AtEase::suppressWarnings();
837 $content = file_get_contents( $fsFile->getPath() );
838 AtEase::restoreWarnings();
839 $contents[$path] = is_string( $content ) ? $content : self::$RES_ERROR;
840 } elseif ( $fsFile === self::$RES_ABSENT ) {
841 $contents[$path] = self::$RES_ABSENT;
842 } else {
843 $contents[$path] = self::$RES_ERROR;
844 }
845 }
846
847 return $contents;
848 }
849
850 final public function getFileXAttributes( array $params ) {
852 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
853
854 $path = self::normalizeStoragePath( $params['src'] );
855 if ( $path === null ) {
856 return self::XATTRS_FAIL; // invalid storage path
857 }
858 $latest = !empty( $params['latest'] ); // use latest data?
859 if ( $this->cheapCache->hasField( $path, 'xattr', self::CACHE_TTL ) ) {
860 $stat = $this->cheapCache->getField( $path, 'xattr' );
861 // If we want the latest data, check that this cached
862 // value was in fact fetched with the latest available data.
863 if ( !$latest || $stat['latest'] ) {
864 return $stat['map'];
865 }
866 }
867 $fields = $this->doGetFileXAttributes( $params );
868 if ( is_array( $fields ) ) {
869 $fields = self::normalizeXAttributes( $fields );
870 $this->cheapCache->setField(
871 $path,
872 'xattr',
873 [ 'map' => $fields, 'latest' => $latest ]
874 );
875 } elseif ( $fields === self::$RES_ABSENT ) {
876 $this->cheapCache->setField(
877 $path,
878 'xattr',
879 [ 'map' => self::XATTRS_FAIL, 'latest' => $latest ]
880 );
881 } else {
882 $fields = self::XATTRS_FAIL; // used for all failure cases
883 }
884
885 return $fields;
886 }
887
894 protected function doGetFileXAttributes( array $params ) {
895 return [ 'headers' => [], 'metadata' => [] ]; // not supported
896 }
897
898 final public function getFileSha1Base36( array $params ) {
900 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
901
902 $path = self::normalizeStoragePath( $params['src'] );
903 if ( $path === null ) {
904 return self::SHA1_FAIL; // invalid storage path
905 }
906 $latest = !empty( $params['latest'] ); // use latest data?
907 if ( $this->cheapCache->hasField( $path, 'sha1', self::CACHE_TTL ) ) {
908 $stat = $this->cheapCache->getField( $path, 'sha1' );
909 // If we want the latest data, check that this cached
910 // value was in fact fetched with the latest available data.
911 if ( !$latest || $stat['latest'] ) {
912 return $stat['hash'];
913 }
914 }
915 $sha1 = $this->doGetFileSha1Base36( $params );
916 if ( is_string( $sha1 ) ) {
917 $this->cheapCache->setField(
918 $path,
919 'sha1',
920 [ 'hash' => $sha1, 'latest' => $latest ]
921 );
922 } elseif ( $sha1 === self::$RES_ABSENT ) {
923 $this->cheapCache->setField(
924 $path,
925 'sha1',
926 [ 'hash' => self::SHA1_FAIL, 'latest' => $latest ]
927 );
928 } else {
929 $sha1 = self::SHA1_FAIL; // used for all failure cases
930 }
931
932 return $sha1;
933 }
934
941 protected function doGetFileSha1Base36( array $params ) {
942 $fsFile = $this->getLocalReference( $params );
943 if ( $fsFile instanceof FSFile ) {
944 $sha1 = $fsFile->getSha1Base36();
945
946 return is_string( $sha1 ) ? $sha1 : self::$RES_ERROR;
947 }
948
949 return ( $fsFile === self::$RES_ERROR ) ? self::$RES_ERROR : self::$RES_ABSENT;
950 }
951
952 final public function getFileProps( array $params ) {
954 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
955
956 $fsFile = $this->getLocalReference( $params );
957
958 return $fsFile ? $fsFile->getProps() : FSFile::placeholderProps();
959 }
960
961 final public function getLocalReferenceMulti( array $params ) {
963 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
964
965 $params = $this->setConcurrencyFlags( $params );
966
967 $fsFiles = []; // (path => FSFile)
968 $latest = !empty( $params['latest'] ); // use latest data?
969 // Reuse any files already in process cache...
970 foreach ( $params['srcs'] as $src ) {
972 if ( $path === null ) {
973 $fsFiles[$src] = null; // invalid storage path
974 } elseif ( $this->expensiveCache->hasField( $path, 'localRef' ) ) {
975 $val = $this->expensiveCache->getField( $path, 'localRef' );
976 // If we want the latest data, check that this cached
977 // value was in fact fetched with the latest available data.
978 if ( !$latest || $val['latest'] ) {
979 $fsFiles[$src] = $val['object'];
980 }
981 }
982 }
983 // Fetch local references of any remaning files...
984 $params['srcs'] = array_diff( $params['srcs'], array_keys( $fsFiles ) );
985 foreach ( $this->doGetLocalReferenceMulti( $params ) as $path => $fsFile ) {
986 if ( $fsFile instanceof FSFile ) {
987 $fsFiles[$path] = $fsFile;
988 $this->expensiveCache->setField(
989 $path,
990 'localRef',
991 [ 'object' => $fsFile, 'latest' => $latest ]
992 );
993 } else {
994 $fsFiles[$path] = null; // used for all failure cases
995 }
996 }
997
998 return $fsFiles;
999 }
1000
1007 protected function doGetLocalReferenceMulti( array $params ) {
1008 return $this->doGetLocalCopyMulti( $params );
1009 }
1010
1011 final public function getLocalCopyMulti( array $params ) {
1013 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
1014
1015 $params = $this->setConcurrencyFlags( $params );
1016 $tmpFiles = $this->doGetLocalCopyMulti( $params );
1017 foreach ( $tmpFiles as $path => $tmpFile ) {
1018 if ( !$tmpFile ) {
1019 $tmpFiles[$path] = null; // used for all failure cases
1020 }
1021 }
1022
1023 return $tmpFiles;
1024 }
1025
1031 abstract protected function doGetLocalCopyMulti( array $params );
1032
1039 public function getFileHttpUrl( array $params ) {
1040 return self::TEMPURL_ERROR; // not supported
1041 }
1042
1043 final public function streamFile( array $params ) {
1045 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
1046 $status = $this->newStatus();
1047
1048 // Always set some fields for subclass convenience
1049 $params['options'] = $params['options'] ?? [];
1050 $params['headers'] = $params['headers'] ?? [];
1051
1052 // Don't stream it out as text/html if there was a PHP error
1053 if ( ( empty( $params['headless'] ) || $params['headers'] ) && headers_sent() ) {
1054 print "Headers already sent, terminating.\n";
1055 $status->fatal( 'backend-fail-stream', $params['src'] );
1056 return $status;
1057 }
1058
1059 $status->merge( $this->doStreamFile( $params ) );
1060
1061 return $status;
1062 }
1063
1070 protected function doStreamFile( array $params ) {
1071 $status = $this->newStatus();
1072
1073 $flags = 0;
1074 $flags |= !empty( $params['headless'] ) ? HTTPFileStreamer::STREAM_HEADLESS : 0;
1075 $flags |= !empty( $params['allowOB'] ) ? HTTPFileStreamer::STREAM_ALLOW_OB : 0;
1076
1077 $fsFile = $this->getLocalReference( $params );
1078 if ( $fsFile ) {
1079 $streamer = new HTTPFileStreamer(
1080 $fsFile->getPath(),
1081 [
1082 'obResetFunc' => $this->obResetFunc,
1083 'streamMimeFunc' => $this->streamMimeFunc
1084 ]
1085 );
1086 $res = $streamer->stream( $params['headers'], true, $params['options'], $flags );
1087 } else {
1088 $res = false;
1089 HTTPFileStreamer::send404Message( $params['src'], $flags );
1090 }
1091
1092 if ( !$res ) {
1093 $status->fatal( 'backend-fail-stream', $params['src'] );
1094 }
1095
1096 return $status;
1097 }
1098
1099 final public function directoryExists( array $params ) {
1100 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
1101 if ( $dir === null ) {
1102 return self::EXISTENCE_ERROR; // invalid storage path
1103 }
1104 if ( $shard !== null ) { // confined to a single container/shard
1105 return $this->doDirectoryExists( $fullCont, $dir, $params );
1106 } else { // directory is on several shards
1107 $this->logger->debug( __METHOD__ . ": iterating over all container shards." );
1108 list( , $shortCont, ) = self::splitStoragePath( $params['dir'] );
1109 $res = false; // response
1110 foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
1111 $exists = $this->doDirectoryExists( "{$fullCont}{$suffix}", $dir, $params );
1112 if ( $exists === true ) {
1113 $res = true;
1114 break; // found one!
1115 } elseif ( $exists === self::$RES_ERROR ) {
1116 $res = self::EXISTENCE_ERROR;
1117 }
1118 }
1119
1120 return $res;
1121 }
1122 }
1123
1132 abstract protected function doDirectoryExists( $container, $dir, array $params );
1133
1134 final public function getDirectoryList( array $params ) {
1135 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
1136 if ( $dir === null ) {
1137 return self::EXISTENCE_ERROR; // invalid storage path
1138 }
1139 if ( $shard !== null ) {
1140 // File listing is confined to a single container/shard
1141 return $this->getDirectoryListInternal( $fullCont, $dir, $params );
1142 } else {
1143 $this->logger->debug( __METHOD__ . ": iterating over all container shards." );
1144 // File listing spans multiple containers/shards
1145 list( , $shortCont, ) = self::splitStoragePath( $params['dir'] );
1146
1147 return new FileBackendStoreShardDirIterator( $this,
1148 $fullCont, $dir, $this->getContainerSuffixes( $shortCont ), $params );
1149 }
1150 }
1151
1162 abstract public function getDirectoryListInternal( $container, $dir, array $params );
1163
1164 final public function getFileList( array $params ) {
1165 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
1166 if ( $dir === null ) {
1167 return self::LIST_ERROR; // invalid storage path
1168 }
1169 if ( $shard !== null ) {
1170 // File listing is confined to a single container/shard
1171 return $this->getFileListInternal( $fullCont, $dir, $params );
1172 } else {
1173 $this->logger->debug( __METHOD__ . ": iterating over all container shards." );
1174 // File listing spans multiple containers/shards
1175 list( , $shortCont, ) = self::splitStoragePath( $params['dir'] );
1176
1177 return new FileBackendStoreShardFileIterator( $this,
1178 $fullCont, $dir, $this->getContainerSuffixes( $shortCont ), $params );
1179 }
1180 }
1181
1192 abstract public function getFileListInternal( $container, $dir, array $params );
1193
1205 final public function getOperationsInternal( array $ops ) {
1206 $supportedOps = [
1207 'store' => StoreFileOp::class,
1208 'copy' => CopyFileOp::class,
1209 'move' => MoveFileOp::class,
1210 'delete' => DeleteFileOp::class,
1211 'create' => CreateFileOp::class,
1212 'describe' => DescribeFileOp::class,
1213 'null' => NullFileOp::class
1214 ];
1215
1216 $performOps = []; // array of FileOp objects
1217 // Build up ordered array of FileOps...
1218 foreach ( $ops as $operation ) {
1219 $opName = $operation['op'];
1220 if ( isset( $supportedOps[$opName] ) ) {
1221 $class = $supportedOps[$opName];
1222 // Get params for this operation
1223 $params = $operation;
1224 // Append the FileOp class
1225 $performOps[] = new $class( $this, $params, $this->logger );
1226 } else {
1227 throw new FileBackendError( "Operation '$opName' is not supported." );
1228 }
1229 }
1230
1231 return $performOps;
1232 }
1233
1244 final public function getPathsToLockForOpsInternal( array $performOps ) {
1245 // Build up a list of files to lock...
1246 $paths = [ 'sh' => [], 'ex' => [] ];
1247 foreach ( $performOps as $fileOp ) {
1248 $paths['sh'] = array_merge( $paths['sh'], $fileOp->storagePathsRead() );
1249 $paths['ex'] = array_merge( $paths['ex'], $fileOp->storagePathsChanged() );
1250 }
1251 // Optimization: if doing an EX lock anyway, don't also set an SH one
1252 $paths['sh'] = array_diff( $paths['sh'], $paths['ex'] );
1253 // Get a shared lock on the parent directory of each path changed
1254 $paths['sh'] = array_merge( $paths['sh'], array_map( 'dirname', $paths['ex'] ) );
1255
1256 return [
1257 LockManager::LOCK_UW => $paths['sh'],
1258 LockManager::LOCK_EX => $paths['ex']
1259 ];
1260 }
1261
1262 public function getScopedLocksForOps( array $ops, StatusValue $status ) {
1263 $paths = $this->getPathsToLockForOpsInternal( $this->getOperationsInternal( $ops ) );
1264
1265 return $this->getScopedFileLocks( $paths, 'mixed', $status );
1266 }
1267
1268 final protected function doOperationsInternal( array $ops, array $opts ) {
1270 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
1271 $status = $this->newStatus();
1272
1273 // Fix up custom header name/value pairs
1274 $ops = array_map( [ $this, 'sanitizeOpHeaders' ], $ops );
1275 // Build up a list of FileOps and involved paths
1276 $fileOps = $this->getOperationsInternal( $ops );
1277 $pathsUsed = [];
1278 foreach ( $fileOps as $fileOp ) {
1279 $pathsUsed = array_merge( $pathsUsed, $fileOp->storagePathsReadOrChanged() );
1280 }
1281
1282 // Acquire any locks as needed for the scope of this function
1283 if ( empty( $opts['nonLocking'] ) ) {
1284 $pathsByLockType = $this->getPathsToLockForOpsInternal( $fileOps );
1286 $scopeLock = $this->getScopedFileLocks( $pathsByLockType, 'mixed', $status );
1287 if ( !$status->isOK() ) {
1288 return $status; // abort
1289 }
1290 }
1291
1292 // Clear any file cache entries (after locks acquired)
1293 if ( empty( $opts['preserveCache'] ) ) {
1294 $this->clearCache( $pathsUsed );
1295 }
1296
1297 // Enlarge the cache to fit the stat entries of these files
1298 $this->cheapCache->setMaxSize( max( 2 * count( $pathsUsed ), self::CACHE_CHEAP_SIZE ) );
1299
1300 // Load from the persistent container caches
1301 $this->primeContainerCache( $pathsUsed );
1302 // Get the latest stat info for all the files (having locked them)
1303 $ok = $this->preloadFileStat( [ 'srcs' => $pathsUsed, 'latest' => true ] );
1304
1305 if ( $ok ) {
1306 // Actually attempt the operation batch...
1307 $opts = $this->setConcurrencyFlags( $opts );
1308 $subStatus = FileOpBatch::attempt( $fileOps, $opts, $this->fileJournal );
1309 } else {
1310 // If we could not even stat some files, then bail out
1311 $subStatus = $this->newStatus( 'backend-fail-internal', $this->name );
1312 foreach ( $ops as $i => $op ) { // mark each op as failed
1313 $subStatus->success[$i] = false;
1314 ++$subStatus->failCount;
1315 }
1316 $this->logger->error( static::class . "-{$this->name} " .
1317 " stat failure; aborted operations: " . FormatJson::encode( $ops ) );
1318 }
1319
1320 // Merge errors into StatusValue fields
1321 $status->merge( $subStatus );
1322 $status->success = $subStatus->success; // not done in merge()
1323
1324 // Shrink the stat cache back to normal size
1325 $this->cheapCache->setMaxSize( self::CACHE_CHEAP_SIZE );
1326
1327 return $status;
1328 }
1329
1330 final protected function doQuickOperationsInternal( array $ops, array $opts ) {
1332 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
1333 $status = $this->newStatus();
1334
1335 // Fix up custom header name/value pairs
1336 $ops = array_map( [ $this, 'sanitizeOpHeaders' ], $ops );
1337 // Build up a list of FileOps and involved paths
1338 $fileOps = $this->getOperationsInternal( $ops );
1339 $pathsUsed = [];
1340 foreach ( $fileOps as $fileOp ) {
1341 $pathsUsed = array_merge( $pathsUsed, $fileOp->storagePathsReadOrChanged() );
1342 }
1343
1344 // Clear any file cache entries for involved paths
1345 $this->clearCache( $pathsUsed );
1346
1347 // Parallel ops may be disabled in config due to dependencies (e.g. needing popen())
1348 $async = ( $this->parallelize === 'implicit' && count( $ops ) > 1 );
1349 $maxConcurrency = $this->concurrency; // throttle
1351 $statuses = []; // array of (index => StatusValue)
1352 $fileOpHandles = []; // list of (index => handle) arrays
1353 $curFileOpHandles = []; // current handle batch
1354 // Perform the sync-only ops and build up op handles for the async ops...
1355 foreach ( $fileOps as $index => $fileOp ) {
1356 $subStatus = $async
1357 ? $fileOp->attemptAsyncQuick()
1358 : $fileOp->attemptQuick();
1359 if ( $subStatus->value instanceof FileBackendStoreOpHandle ) { // async
1360 if ( count( $curFileOpHandles ) >= $maxConcurrency ) {
1361 $fileOpHandles[] = $curFileOpHandles; // push this batch
1362 $curFileOpHandles = [];
1363 }
1364 $curFileOpHandles[$index] = $subStatus->value; // keep index
1365 } else { // error or completed
1366 $statuses[$index] = $subStatus; // keep index
1367 }
1368 }
1369 if ( count( $curFileOpHandles ) ) {
1370 $fileOpHandles[] = $curFileOpHandles; // last batch
1371 }
1372 // Do all the async ops that can be done concurrently...
1373 foreach ( $fileOpHandles as $fileHandleBatch ) {
1374 $statuses += $this->executeOpHandlesInternal( $fileHandleBatch );
1375 }
1376 // Marshall and merge all the responses...
1377 foreach ( $statuses as $index => $subStatus ) {
1378 $status->merge( $subStatus );
1379 if ( $subStatus->isOK() ) {
1380 $status->success[$index] = true;
1381 ++$status->successCount;
1382 } else {
1383 $status->success[$index] = false;
1384 ++$status->failCount;
1385 }
1386 }
1387
1388 $this->clearCache( $pathsUsed );
1389
1390 return $status;
1391 }
1392
1402 final public function executeOpHandlesInternal( array $fileOpHandles ) {
1404 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
1405
1406 foreach ( $fileOpHandles as $fileOpHandle ) {
1407 if ( !( $fileOpHandle instanceof FileBackendStoreOpHandle ) ) {
1408 throw new InvalidArgumentException( "Expected FileBackendStoreOpHandle object." );
1409 } elseif ( $fileOpHandle->backend->getName() !== $this->getName() ) {
1410 throw new InvalidArgumentException( "Expected handle for this file backend." );
1411 }
1412 }
1413
1414 $statuses = $this->doExecuteOpHandlesInternal( $fileOpHandles );
1415 foreach ( $fileOpHandles as $fileOpHandle ) {
1416 $fileOpHandle->closeResources();
1417 }
1418
1419 return $statuses;
1420 }
1421
1431 protected function doExecuteOpHandlesInternal( array $fileOpHandles ) {
1432 if ( count( $fileOpHandles ) ) {
1433 throw new FileBackendError( "Backend does not support asynchronous operations." );
1434 }
1435
1436 return [];
1437 }
1438
1450 protected function sanitizeOpHeaders( array $op ) {
1451 static $longs = [ 'content-disposition' ];
1452
1453 if ( isset( $op['headers'] ) ) { // op sets HTTP headers
1454 $newHeaders = [];
1455 foreach ( $op['headers'] as $name => $value ) {
1456 $name = strtolower( $name );
1457 $maxHVLen = in_array( $name, $longs ) ? INF : 255;
1458 if ( strlen( $name ) > 255 || strlen( $value ) > $maxHVLen ) {
1459 $this->logger->error( "Header '{header}' is too long.", [
1460 'filebackend' => $this->name,
1461 'header' => "$name: $value",
1462 ] );
1463 } else {
1464 $newHeaders[$name] = strlen( $value ) ? $value : ''; // null/false => ""
1465 }
1466 }
1467 $op['headers'] = $newHeaders;
1468 }
1469
1470 return $op;
1471 }
1472
1473 final public function preloadCache( array $paths ) {
1474 $fullConts = []; // full container names
1475 foreach ( $paths as $path ) {
1476 list( $fullCont, , ) = $this->resolveStoragePath( $path );
1477 $fullConts[] = $fullCont;
1478 }
1479 // Load from the persistent file and container caches
1480 $this->primeContainerCache( $fullConts );
1481 $this->primeFileCache( $paths );
1482 }
1483
1484 final public function clearCache( array $paths = null ) {
1485 if ( is_array( $paths ) ) {
1486 $paths = array_map( 'FileBackend::normalizeStoragePath', $paths );
1487 $paths = array_filter( $paths, 'strlen' ); // remove nulls
1488 }
1489 if ( $paths === null ) {
1490 $this->cheapCache->clear();
1491 $this->expensiveCache->clear();
1492 } else {
1493 foreach ( $paths as $path ) {
1494 $this->cheapCache->clear( $path );
1495 $this->expensiveCache->clear( $path );
1496 }
1497 }
1498 $this->doClearCache( $paths );
1499 }
1500
1509 protected function doClearCache( array $paths = null ) {
1510 }
1511
1512 final public function preloadFileStat( array $params ) {
1514 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
1515
1516 $params['concurrency'] = ( $this->parallelize !== 'off' ) ? $this->concurrency : 1;
1517 $stats = $this->doGetFileStatMulti( $params );
1518 if ( $stats === null ) {
1519 return true; // not supported
1520 }
1521
1522 // Whether this queried the backend in high consistency mode
1523 $latest = !empty( $params['latest'] );
1524
1525 return $this->ingestFreshFileStats( $stats, $latest );
1526 }
1527
1540 protected function doGetFileStatMulti( array $params ) {
1541 return null; // not supported
1542 }
1543
1551 abstract protected function directoriesAreVirtual();
1552
1563 final protected static function isValidShortContainerName( $container ) {
1564 // Suffixes like '.xxx' (hex shard chars) or '.seg' (file segments)
1565 // might be used by subclasses. Reserve the dot character for sanity.
1566 // The only way dots end up in containers (e.g. resolveStoragePath)
1567 // is due to the wikiId container prefix or the above suffixes.
1568 return self::isValidContainerName( $container ) && !preg_match( '/[.]/', $container );
1569 }
1570
1580 final protected static function isValidContainerName( $container ) {
1581 // This accounts for NTFS, Swift, and Ceph restrictions
1582 // and disallows directory separators or traversal characters.
1583 // Note that matching strings URL encode to the same string;
1584 // in Swift/Ceph, the length restriction is *after* URL encoding.
1585 return (bool)preg_match( '/^[a-z0-9][a-z0-9-_.]{0,199}$/i', $container );
1586 }
1587
1601 final protected function resolveStoragePath( $storagePath ) {
1602 list( $backend, $shortCont, $relPath ) = self::splitStoragePath( $storagePath );
1603 if ( $backend === $this->name ) { // must be for this backend
1604 $relPath = self::normalizeContainerPath( $relPath );
1605 if ( $relPath !== null && self::isValidShortContainerName( $shortCont ) ) {
1606 // Get shard for the normalized path if this container is sharded
1607 $cShard = $this->getContainerShard( $shortCont, $relPath );
1608 // Validate and sanitize the relative path (backend-specific)
1609 $relPath = $this->resolveContainerPath( $shortCont, $relPath );
1610 if ( $relPath !== null ) {
1611 // Prepend any domain ID prefix to the container name
1612 $container = $this->fullContainerName( $shortCont );
1613 if ( self::isValidContainerName( $container ) ) {
1614 // Validate and sanitize the container name (backend-specific)
1615 $container = $this->resolveContainerName( "{$container}{$cShard}" );
1616 if ( $container !== null ) {
1617 return [ $container, $relPath, $cShard ];
1618 }
1619 }
1620 }
1621 }
1622 }
1623
1624 return [ null, null, null ];
1625 }
1626
1642 final protected function resolveStoragePathReal( $storagePath ) {
1643 list( $container, $relPath, $cShard ) = $this->resolveStoragePath( $storagePath );
1644 if ( $cShard !== null && substr( $relPath, -1 ) !== '/' ) {
1645 return [ $container, $relPath ];
1646 }
1647
1648 return [ null, null ];
1649 }
1650
1659 final protected function getContainerShard( $container, $relPath ) {
1660 list( $levels, $base, $repeat ) = $this->getContainerHashLevels( $container );
1661 if ( $levels == 1 || $levels == 2 ) {
1662 // Hash characters are either base 16 or 36
1663 $char = ( $base == 36 ) ? '[0-9a-z]' : '[0-9a-f]';
1664 // Get a regex that represents the shard portion of paths.
1665 // The concatenation of the captures gives us the shard.
1666 if ( $levels === 1 ) { // 16 or 36 shards per container
1667 $hashDirRegex = '(' . $char . ')';
1668 } else { // 256 or 1296 shards per container
1669 if ( $repeat ) { // verbose hash dir format (e.g. "a/ab/abc")
1670 $hashDirRegex = $char . '/(' . $char . '{2})';
1671 } else { // short hash dir format (e.g. "a/b/c")
1672 $hashDirRegex = '(' . $char . ')/(' . $char . ')';
1673 }
1674 }
1675 // Allow certain directories to be above the hash dirs so as
1676 // to work with FileRepo (e.g. "archive/a/ab" or "temp/a/ab").
1677 // They must be 2+ chars to avoid any hash directory ambiguity.
1678 $m = [];
1679 if ( preg_match( "!^(?:[^/]{2,}/)*$hashDirRegex(?:/|$)!", $relPath, $m ) ) {
1680 return '.' . implode( '', array_slice( $m, 1 ) );
1681 }
1682
1683 return null; // failed to match
1684 }
1685
1686 return ''; // no sharding
1687 }
1688
1697 final public function isSingleShardPathInternal( $storagePath ) {
1698 list( , , $shard ) = $this->resolveStoragePath( $storagePath );
1699
1700 return ( $shard !== null );
1701 }
1702
1711 final protected function getContainerHashLevels( $container ) {
1712 if ( isset( $this->shardViaHashLevels[$container] ) ) {
1713 $config = $this->shardViaHashLevels[$container];
1714 $hashLevels = (int)$config['levels'];
1715 if ( $hashLevels == 1 || $hashLevels == 2 ) {
1716 $hashBase = (int)$config['base'];
1717 if ( $hashBase == 16 || $hashBase == 36 ) {
1718 return [ $hashLevels, $hashBase, $config['repeat'] ];
1719 }
1720 }
1721 }
1722
1723 return [ 0, 0, false ]; // no sharding
1724 }
1725
1732 final protected function getContainerSuffixes( $container ) {
1733 $shards = [];
1734 list( $digits, $base ) = $this->getContainerHashLevels( $container );
1735 if ( $digits > 0 ) {
1736 $numShards = $base ** $digits;
1737 for ( $index = 0; $index < $numShards; $index++ ) {
1738 $shards[] = '.' . Wikimedia\base_convert( $index, 10, $base, $digits );
1739 }
1740 }
1741
1742 return $shards;
1743 }
1744
1751 final protected function fullContainerName( $container ) {
1752 if ( $this->domainId != '' ) {
1753 return "{$this->domainId}-$container";
1754 } else {
1755 return $container;
1756 }
1757 }
1758
1768 protected function resolveContainerName( $container ) {
1769 return $container;
1770 }
1771
1783 protected function resolveContainerPath( $container, $relStoragePath ) {
1784 return $relStoragePath;
1785 }
1786
1793 private function containerCacheKey( $container ) {
1794 return "filebackend:{$this->name}:{$this->domainId}:container:{$container}";
1795 }
1796
1803 final protected function setContainerCache( $container, array $val ) {
1804 $this->memCache->set( $this->containerCacheKey( $container ), $val, 14 * 86400 );
1805 }
1806
1813 final protected function deleteContainerCache( $container ) {
1814 if ( !$this->memCache->delete( $this->containerCacheKey( $container ), 300 ) ) {
1815 $this->logger->warning( "Unable to delete stat cache for container {container}.",
1816 [ 'filebackend' => $this->name, 'container' => $container ]
1817 );
1818 }
1819 }
1820
1828 final protected function primeContainerCache( array $items ) {
1830 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
1831
1832 $paths = []; // list of storage paths
1833 $contNames = []; // (cache key => resolved container name)
1834 // Get all the paths/containers from the items...
1835 foreach ( $items as $item ) {
1836 if ( self::isStoragePath( $item ) ) {
1837 $paths[] = $item;
1838 } elseif ( is_string( $item ) ) { // full container name
1839 $contNames[$this->containerCacheKey( $item )] = $item;
1840 }
1841 }
1842 // Get all the corresponding cache keys for paths...
1843 foreach ( $paths as $path ) {
1844 list( $fullCont, , ) = $this->resolveStoragePath( $path );
1845 if ( $fullCont !== null ) { // valid path for this backend
1846 $contNames[$this->containerCacheKey( $fullCont )] = $fullCont;
1847 }
1848 }
1849
1850 $contInfo = []; // (resolved container name => cache value)
1851 // Get all cache entries for these container cache keys...
1852 $values = $this->memCache->getMulti( array_keys( $contNames ) );
1853 foreach ( $values as $cacheKey => $val ) {
1854 $contInfo[$contNames[$cacheKey]] = $val;
1855 }
1856
1857 // Populate the container process cache for the backend...
1858 $this->doPrimeContainerCache( array_filter( $contInfo, 'is_array' ) );
1859 }
1860
1869 protected function doPrimeContainerCache( array $containerInfo ) {
1870 }
1871
1878 private function fileCacheKey( $path ) {
1879 return "filebackend:{$this->name}:{$this->domainId}:file:" . sha1( $path );
1880 }
1881
1890 final protected function setFileCache( $path, array $val ) {
1892 if ( $path === null ) {
1893 return; // invalid storage path
1894 }
1895 $mtime = ConvertibleTimestamp::convert( TS_UNIX, $val['mtime'] );
1896 $ttl = $this->memCache->adaptiveTTL( $mtime, 7 * 86400, 300, 0.1 );
1897 $key = $this->fileCacheKey( $path );
1898 // Set the cache unless it is currently salted.
1899 $this->memCache->set( $key, $val, $ttl );
1900 }
1901
1910 final protected function deleteFileCache( $path ) {
1912 if ( $path === null ) {
1913 return; // invalid storage path
1914 }
1915 if ( !$this->memCache->delete( $this->fileCacheKey( $path ), 300 ) ) {
1916 $this->logger->warning( "Unable to delete stat cache for file {path}.",
1917 [ 'filebackend' => $this->name, 'path' => $path ]
1918 );
1919 }
1920 }
1921
1929 final protected function primeFileCache( array $items ) {
1931 $ps = $this->scopedProfileSection( __METHOD__ . "-{$this->name}" );
1932
1933 $paths = []; // list of storage paths
1934 $pathNames = []; // (cache key => storage path)
1935 // Get all the paths/containers from the items...
1936 foreach ( $items as $item ) {
1937 if ( self::isStoragePath( $item ) ) {
1938 $paths[] = FileBackend::normalizeStoragePath( $item );
1939 }
1940 }
1941 // Get rid of any paths that failed normalization
1942 $paths = array_filter( $paths, 'strlen' ); // remove nulls
1943 // Get all the corresponding cache keys for paths...
1944 foreach ( $paths as $path ) {
1945 list( , $rel, ) = $this->resolveStoragePath( $path );
1946 if ( $rel !== null ) { // valid path for this backend
1947 $pathNames[$this->fileCacheKey( $path )] = $path;
1948 }
1949 }
1950 // Get all cache entries for these file cache keys.
1951 // Note that negatives are not cached by getFileStat()/preloadFileStat().
1952 $values = $this->memCache->getMulti( array_keys( $pathNames ) );
1953 // Load all of the results into process cache...
1954 foreach ( array_filter( $values, 'is_array' ) as $cacheKey => $stat ) {
1955 $path = $pathNames[$cacheKey];
1956 // Sanity; this flag only applies to stat info loaded directly
1957 // from a high consistency backend query to the process cache
1958 unset( $stat['latest'] );
1959
1960 $this->cheapCache->setField( $path, 'stat', $stat );
1961 if ( isset( $stat['sha1'] ) && strlen( $stat['sha1'] ) == 31 ) {
1962 // Some backends store SHA-1 as metadata
1963 $this->cheapCache->setField(
1964 $path,
1965 'sha1',
1966 [ 'hash' => $stat['sha1'], 'latest' => false ]
1967 );
1968 }
1969 if ( isset( $stat['xattr'] ) && is_array( $stat['xattr'] ) ) {
1970 // Some backends store custom headers/metadata
1971 $stat['xattr'] = self::normalizeXAttributes( $stat['xattr'] );
1972 $this->cheapCache->setField(
1973 $path,
1974 'xattr',
1975 [ 'map' => $stat['xattr'], 'latest' => false ]
1976 );
1977 }
1978 }
1979 }
1980
1988 final protected static function normalizeXAttributes( array $xattr ) {
1989 $newXAttr = [ 'headers' => [], 'metadata' => [] ];
1990
1991 foreach ( $xattr['headers'] as $name => $value ) {
1992 $newXAttr['headers'][strtolower( $name )] = $value;
1993 }
1994
1995 foreach ( $xattr['metadata'] as $name => $value ) {
1996 $newXAttr['metadata'][strtolower( $name )] = $value;
1997 }
1998
1999 return $newXAttr;
2000 }
2001
2008 final protected function setConcurrencyFlags( array $opts ) {
2009 $opts['concurrency'] = 1; // off
2010 if ( $this->parallelize === 'implicit' ) {
2011 if ( $opts['parallelize'] ?? true ) {
2012 $opts['concurrency'] = $this->concurrency;
2013 }
2014 } elseif ( $this->parallelize === 'explicit' ) {
2015 if ( !empty( $opts['parallelize'] ) ) {
2016 $opts['concurrency'] = $this->concurrency;
2017 }
2018 }
2019
2020 return $opts;
2021 }
2022
2032 protected function getContentType( $storagePath, $content, $fsPath ) {
2033 if ( $this->mimeCallback ) {
2034 return call_user_func_array( $this->mimeCallback, func_get_args() );
2035 }
2036
2037 $mime = ( $fsPath !== null ) ? mime_content_type( $fsPath ) : false;
2038 return $mime ?: 'unknown/unknown';
2039 }
2040}
Class representing a cache/ephemeral data store.
Definition BagOStuff.php:71
A BagOStuff object with no objects in it.
Class representing a non-directory file on the file system.
Definition FSFile.php:32
static placeholderProps()
Placeholder file properties to use for files that don't exist.
Definition FSFile.php:150
File backend exception for checked exceptions (e.g.
FileBackendStore helper class for performing asynchronous file operations.
Base class for all backends using particular storage medium.
preloadFileStat(array $params)
Preload file stat information (concurrently if possible) into in-process cache.
containerCacheKey( $container)
Get the cache key for a container.
doGetLocalReferenceMulti(array $params)
resolveContainerName( $container)
Resolve a container name, checking if it's allowed by the backend.
static normalizeXAttributes(array $xattr)
Normalize file headers/metadata to the FileBackend::getFileXAttributes() format.
getLocalReferenceMulti(array $params)
Like getLocalReference() except it takes an array of storage paths and yields an order-preserved map ...
setFileCache( $path, array $val)
Set the cached stat info for a file path.
doPublish(array $params)
moveInternal(array $params)
Move a file from one storage path to another in the backend.
setContainerCache( $container, array $val)
Set the cached info for a container.
isPathUsableInternal( $storagePath)
Check if a file can be created or changed at a given storage path in the backend.
doGetFileContentsMulti(array $params)
static string $ABSENT_NORMAL
File does not exist according to a normal stat query.
doStreamFile(array $params)
doSecure(array $params)
MapCacheLRU $expensiveCache
Map of paths to large (RAM/disk) cache items.
doClean(array $params)
getFileTimestamp(array $params)
Get the last-modified timestamp of the file at a storage path.
doGetLocalCopyMulti(array $params)
getScopedLocksForOps(array $ops, StatusValue $status)
Get an array of scoped locks needed for a batch of file operations.
doCleanInternal( $container, $dir, array $params)
callable $mimeCallback
Method to get the MIME type of files.
getPathsToLockForOpsInternal(array $performOps)
Get a list of storage paths to lock for a list of operations Returns an array with LockManager::LOCK_...
executeOpHandlesInternal(array $fileOpHandles)
Execute a list of FileBackendStoreOpHandle handles in parallel.
doPrepareInternal( $container, $dir, array $params)
concatenate(array $params)
Concatenate a list of storage files into a single file system file.
storeInternal(array $params)
Store a file into the backend from a file on disk.
getFileProps(array $params)
Get the properties of the content of the file at a storage path in the backend.
getDirectoryList(array $params)
Get an iterator to list all directories under a storage directory.
doQuickOperationsInternal(array $ops, array $opts)
resolveStoragePath( $storagePath)
Splits a storage path into an internal container name, an internal relative file name,...
copyInternal(array $params)
Copy a file from one storage path to another in the backend.
doStoreInternal(array $params)
directoriesAreVirtual()
Is this a key/value store where directories are just virtual? Virtual directories exists in so much a...
doGetFileStat(array $params)
getFileStat(array $params)
Get quick information about a file at a storage path in the backend.
MapCacheLRU $cheapCache
Map of paths to small (RAM/disk) cache items.
resolveStoragePathReal( $storagePath)
Like resolveStoragePath() except null values are returned if the container is sharded and the shard c...
clearCache(array $paths=null)
Invalidate any in-process file stat and property cache.
doPublishInternal( $container, $dir, array $params)
doGetFileStatMulti(array $params)
Get file stat information (concurrently if possible) for several files Stable to override.
getFileSha1Base36(array $params)
Get a SHA-1 hash of the content of the file at a storage path in the backend.
static isValidContainerName( $container)
Check if a full container name is valid.
ingestFreshFileStats(array $stats, $latest)
Ingest file stat entries that just came from querying the backend (not cache)
getFileListInternal( $container, $dir, array $params)
Do not call this function from places outside FileBackend.
doCopyInternal(array $params)
getLocalCopyMulti(array $params)
Like getLocalCopy() except it takes an array of storage paths and yields an order preserved-map of st...
getFileContentsMulti(array $params)
Like getFileContents() except it takes an array of storage paths and returns an order preserved map o...
doClearCache(array $paths=null)
Clears any additional stat caches for storage paths Stable to override.
fullContainerName( $container)
Get the full container name, including the domain ID prefix.
primeFileCache(array $items)
Do a batch lookup from cache for file stats for all paths used in a list of storage paths or FileOp o...
doOperationsInternal(array $ops, array $opts)
isSingleShardPathInternal( $storagePath)
Check if a storage path maps to a single shard.
directoryExists(array $params)
Check if a directory exists at a given storage path.
doDeleteInternal(array $params)
static string $ABSENT_LATEST
File does not exist according to a "latest"-mode stat query.
getContainerSuffixes( $container)
Get a list of full container shard suffixes for a container.
primeContainerCache(array $items)
Do a batch lookup from cache for container stats for all containers used in a list of container names...
sanitizeOpHeaders(array $op)
Normalize and filter HTTP headers from a file operation.
doDescribeInternal(array $params)
doGetFileSha1Base36(array $params)
deleteInternal(array $params)
Delete a file at the storage path.
doPrimeContainerCache(array $containerInfo)
Fill the backend-specific process cache given an array of resolved container names and their correspo...
doMoveInternal(array $params)
maxFileSizeInternal()
Get the maximum allowable file size given backend medium restrictions and basic performance constrain...
describeInternal(array $params)
Alter metadata for a file at the storage path.
streamFile(array $params)
Stream the content of the file at a storage path in the backend.
getOperationsInternal(array $ops)
Return a list of FileOp objects from a list of operations.
deleteFileCache( $path)
Delete the cached stat info for a file path.
setConcurrencyFlags(array $opts)
Set the 'concurrency' option from a list of operation options.
getFileXAttributes(array $params)
Get metadata about a file at a storage path in the backend.
getContentType( $storagePath, $content, $fsPath)
Get the content type to use in HEAD/GET requests for a file Stable to override.
preloadCache(array $paths)
Preload persistent file stat cache and property cache into in-process cache.
resolveContainerPath( $container, $relStoragePath)
Resolve a relative storage path, checking if it's allowed by the backend.
doDirectoryExists( $container, $dir, array $params)
WANObjectCache $memCache
static false $RES_ABSENT
Idiom for "no result due to missing file" (since 1.34)
doPrepare(array $params)
getFileHttpUrl(array $params)
__construct(array $config)
doGetFileXAttributes(array $params)
static null $RES_ERROR
Idiom for "no result due to I/O errors" (since 1.34)
doExecuteOpHandlesInternal(array $fileOpHandles)
nullInternal(array $params)
No-op file operation that does nothing.
getFileSize(array $params)
Get the size (bytes) of a file at a storage path in the backend.
array $shardViaHashLevels
Map of container names to sharding config.
fileCacheKey( $path)
Get the cache key for a file path.
doConcatenate(array $params)
doSecureInternal( $container, $dir, array $params)
doCreateInternal(array $params)
fileExists(array $params)
Check if a file exists at a storage path in the backend.
getDirectoryListInternal( $container, $dir, array $params)
Do not call this function from places outside FileBackend.
static isValidShortContainerName( $container)
Check if a short container name is valid.
getContainerHashLevels( $container)
Get the sharding config for a container.
getContainerShard( $container, $relPath)
Get the container name shard suffix for a given path.
deleteContainerCache( $container)
Delete the cached info for a container.
createInternal(array $params)
Create a file in the backend with the given contents.
getFileList(array $params)
Get an iterator to list all stored files under a storage directory.
Base class for all file backend classes (including multi-write backends).
string $name
Unique backend name.
static splitStoragePath( $storagePath)
Split a storage path into a backend name, a container name, and a relative file path.
static normalizeContainerPath( $path)
Validate and normalize a relative storage path.
getScopedFileLocks(array $paths, $type, StatusValue $status, $timeout=0)
Lock the files at the given storage paths in the backend.
getTopDirectoryList(array $params)
Same as FileBackend::getDirectoryList() except only lists directories that are immediately under the ...
scopedProfileSection( $section)
newStatus(... $args)
Yields the result of the status wrapper callback on either:
static normalizeStoragePath( $storagePath)
Normalize a storage path by cleaning up directory separators.
int $concurrency
How many operations can be done in parallel.
getLocalReference(array $params)
Returns a file system file, identical in content to the file at a storage path.
static attempt(array $performOps, array $opts, FileJournal $journal)
Attempt to perform a series of file operations.
Functions related to the output of file content.
static send404Message( $fname, $flags=0)
Send out a standard 404 message for a file.
Handles a simple LRU key/value map with a maximum number of entries.
Generic operation result class Has warning/error list, boolean status and arbitrary value.
Multi-datacenter aware caching interface.
while(( $__line=Maintenance::readconsole()) !==false) print
Definition eval.php:64
$content
Definition router.php:76
$mime
Definition router.php:60