3 * @defgroup FileRepo File Repository
5 * @brief This module handles how MediaWiki interacts with filesystems.
11 * Base code for file repositories.
13 * This program is free software; you can redistribute it and/or modify
14 * it under the terms of the GNU General Public License as published by
15 * the Free Software Foundation; either version 2 of the License, or
16 * (at your option) any later version.
18 * This program is distributed in the hope that it will be useful,
19 * but WITHOUT ANY WARRANTY; without even the implied warranty of
20 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
21 * GNU General Public License for more details.
23 * You should have received a copy of the GNU General Public License along
24 * with this program; if not, write to the Free Software Foundation, Inc.,
25 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
26 * http://www.gnu.org/copyleft/gpl.html
33 * Base class for file repositories
38 const DELETE_SOURCE
= 1;
40 const OVERWRITE_SAME
= 4;
41 const SKIP_LOCKING
= 8;
43 /** @var FileBackend */
45 /** @var Array Map of zones to config */
46 protected $zones = array();
48 var $thumbScriptUrl, $transformVia404;
49 var $descBaseUrl, $scriptDirUrl, $scriptExtension, $articleUrl;
50 var $fetchDescription, $initialCapital;
51 var $pathDisclosureProtection = 'simple'; // 'paranoid'
52 var $descriptionCacheExpiry, $url, $thumbUrl;
53 var $hashLevels, $deletedHashLevels;
54 protected $abbrvThreshold;
57 * Factory functions for creating new files
58 * Override these in the base class
60 var $fileFactory = array( 'UnregisteredLocalFile', 'newFromTitle' );
61 var $oldFileFactory = false;
62 var $fileFactoryKey = false, $oldFileFactoryKey = false;
65 * @param $info array|null
68 public function __construct( array $info = null ) {
69 // Verify required settings presence
72 ||
!array_key_exists( 'name', $info )
73 ||
!array_key_exists( 'backend', $info )
75 throw new MWException( __CLASS__
. " requires an array of options having both 'name' and 'backend' keys.\n" );
79 $this->name
= $info['name'];
80 if ( $info['backend'] instanceof FileBackend
) {
81 $this->backend
= $info['backend']; // useful for testing
83 $this->backend
= FileBackendGroup
::singleton()->get( $info['backend'] );
86 // Optional settings that can have no value
87 $optionalSettings = array(
88 'descBaseUrl', 'scriptDirUrl', 'articleUrl', 'fetchDescription',
89 'thumbScriptUrl', 'pathDisclosureProtection', 'descriptionCacheExpiry',
92 foreach ( $optionalSettings as $var ) {
93 if ( isset( $info[$var] ) ) {
94 $this->$var = $info[$var];
98 // Optional settings that have a default
99 $this->initialCapital
= isset( $info['initialCapital'] )
100 ?
$info['initialCapital']
101 : MWNamespace
::isCapitalized( NS_FILE
);
102 $this->url
= isset( $info['url'] )
104 : false; // a subclass may set the URL (e.g. ForeignAPIRepo)
105 if ( isset( $info['thumbUrl'] ) ) {
106 $this->thumbUrl
= $info['thumbUrl'];
108 $this->thumbUrl
= $this->url ?
"{$this->url}/thumb" : false;
110 $this->hashLevels
= isset( $info['hashLevels'] )
111 ?
$info['hashLevels']
113 $this->deletedHashLevels
= isset( $info['deletedHashLevels'] )
114 ?
$info['deletedHashLevels']
116 $this->transformVia404
= !empty( $info['transformVia404'] );
117 $this->abbrvThreshold
= isset( $info['abbrvThreshold'] )
118 ?
$info['abbrvThreshold']
120 $this->isPrivate
= !empty( $info['isPrivate'] );
121 // Give defaults for the basic zones...
122 $this->zones
= isset( $info['zones'] ) ?
$info['zones'] : array();
123 foreach ( array( 'public', 'thumb', 'temp', 'deleted' ) as $zone ) {
124 if ( !isset( $this->zones
[$zone]['container'] ) ) {
125 $this->zones
[$zone]['container'] = "{$this->name}-{$zone}";
127 if ( !isset( $this->zones
[$zone]['directory'] ) ) {
128 $this->zones
[$zone]['directory'] = '';
130 if ( !isset( $this->zones
[$zone]['urlsByExt'] ) ) {
131 $this->zones
[$zone]['urlsByExt'] = array();
137 * Get the file backend instance. Use this function wisely.
139 * @return FileBackend
141 public function getBackend() {
142 return $this->backend
;
146 * Get an explanatory message if this repo is read-only.
147 * This checks if an administrator disabled writes to the backend.
149 * @return string|bool Returns false if the repo is not read-only
151 public function getReadOnlyReason() {
152 return $this->backend
->getReadOnlyReason();
156 * Check if a single zone or list of zones is defined for usage
158 * @param $doZones Array Only do a particular zones
159 * @throws MWException
162 protected function initZones( $doZones = array() ) {
163 $status = $this->newGood();
164 foreach ( (array)$doZones as $zone ) {
165 $root = $this->getZonePath( $zone );
166 if ( $root === null ) {
167 throw new MWException( "No '$zone' zone defined in the {$this->name} repo." );
174 * Determine if a string is an mwrepo:// URL
179 public static function isVirtualUrl( $url ) {
180 return substr( $url, 0, 9 ) == 'mwrepo://';
184 * Get a URL referring to this repository, with the private mwrepo protocol.
185 * The suffix, if supplied, is considered to be unencoded, and will be
186 * URL-encoded before being returned.
188 * @param $suffix string|bool
191 public function getVirtualUrl( $suffix = false ) {
192 $path = 'mwrepo://' . $this->name
;
193 if ( $suffix !== false ) {
194 $path .= '/' . rawurlencode( $suffix );
200 * Get the URL corresponding to one of the four basic zones
202 * @param $zone String One of: public, deleted, temp, thumb
203 * @param $ext String|null Optional file extension
204 * @return String or false
206 public function getZoneUrl( $zone, $ext = null ) {
207 if ( in_array( $zone, array( 'public', 'temp', 'thumb' ) ) ) { // standard public zones
208 if ( $ext !== null && isset( $this->zones
[$zone]['urlsByExt'][$ext] ) ) {
209 return $this->zones
[$zone]['urlsByExt'][$ext]; // custom URL for extension/zone
210 } elseif ( isset( $this->zones
[$zone]['url'] ) ) {
211 return $this->zones
[$zone]['url']; // custom URL for zone
218 return "{$this->url}/temp";
220 return false; // no public URL
222 return $this->thumbUrl
;
229 * Get the thumb zone URL configured to be handled by scripts like thumb_handler.php.
230 * This is probably only useful for internal requests, such as from a fast frontend server
231 * to a slower backend server.
233 * Large sites may use a different host name for uploads than for wikis. In any case, the
234 * wiki configuration is needed in order to use thumb.php. To avoid extracting the wiki ID
235 * from the URL path, one can configure thumb_handler.php to recognize a special path on the
236 * same host name as the wiki that is used for viewing thumbnails.
238 * @param $zone String: one of: public, deleted, temp, thumb
239 * @return String or false
241 public function getZoneHandlerUrl( $zone ) {
242 if ( isset( $this->zones
[$zone]['handlerUrl'] )
243 && in_array( $zone, array( 'public', 'temp', 'thumb' ) ) )
245 return $this->zones
[$zone]['handlerUrl'];
251 * Get the backend storage path corresponding to a virtual URL.
252 * Use this function wisely.
255 * @throws MWException
258 public function resolveVirtualUrl( $url ) {
259 if ( substr( $url, 0, 9 ) != 'mwrepo://' ) {
260 throw new MWException( __METHOD__
.': unknown protocol' );
262 $bits = explode( '/', substr( $url, 9 ), 3 );
263 if ( count( $bits ) != 3 ) {
264 throw new MWException( __METHOD__
.": invalid mwrepo URL: $url" );
266 list( $repo, $zone, $rel ) = $bits;
267 if ( $repo !== $this->name
) {
268 throw new MWException( __METHOD__
.": fetching from a foreign repo is not supported" );
270 $base = $this->getZonePath( $zone );
272 throw new MWException( __METHOD__
.": invalid zone: $zone" );
274 return $base . '/' . rawurldecode( $rel );
278 * The the storage container and base path of a zone
280 * @param $zone string
281 * @return Array (container, base path) or (null, null)
283 protected function getZoneLocation( $zone ) {
284 if ( !isset( $this->zones
[$zone] ) ) {
285 return array( null, null ); // bogus
287 return array( $this->zones
[$zone]['container'], $this->zones
[$zone]['directory'] );
291 * Get the storage path corresponding to one of the zones
293 * @param $zone string
294 * @return string|null Returns null if the zone is not defined
296 public function getZonePath( $zone ) {
297 list( $container, $base ) = $this->getZoneLocation( $zone );
298 if ( $container === null ||
$base === null ) {
301 $backendName = $this->backend
->getName();
302 if ( $base != '' ) { // may not be set
305 return "mwstore://$backendName/{$container}{$base}";
309 * Create a new File object from the local repository
311 * @param $title Mixed: Title object or string
312 * @param $time Mixed: Time at which the image was uploaded.
313 * If this is specified, the returned object will be an
314 * instance of the repository's old file class instead of a
315 * current file. Repositories not supporting version control
316 * should return false if this parameter is set.
317 * @return File|null A File, or null if passed an invalid Title
319 public function newFile( $title, $time = false ) {
320 $title = File
::normalizeTitle( $title );
325 if ( $this->oldFileFactory
) {
326 return call_user_func( $this->oldFileFactory
, $title, $this, $time );
331 return call_user_func( $this->fileFactory
, $title, $this );
336 * Find an instance of the named file created at the specified time
337 * Returns false if the file does not exist. Repositories not supporting
338 * version control should return false if the time is specified.
340 * @param $title Mixed: Title object or string
341 * @param $options array Associative array of options:
342 * time: requested time for a specific file version, or false for the
343 * current version. An image object will be returned which was
344 * created at the specified time (which may be archived or current).
346 * ignoreRedirect: If true, do not follow file redirects
348 * private: If true, return restricted (deleted) files if the current
349 * user is allowed to view them. Otherwise, such files will not
351 * @return File|bool False on failure
353 public function findFile( $title, $options = array() ) {
354 $title = File
::normalizeTitle( $title );
358 $time = isset( $options['time'] ) ?
$options['time'] : false;
359 # First try the current version of the file to see if it precedes the timestamp
360 $img = $this->newFile( $title );
364 if ( $img->exists() && ( !$time ||
$img->getTimestamp() == $time ) ) {
367 # Now try an old version of the file
368 if ( $time !== false ) {
369 $img = $this->newFile( $title, $time );
370 if ( $img && $img->exists() ) {
371 if ( !$img->isDeleted( File
::DELETED_FILE
) ) {
372 return $img; // always OK
373 } elseif ( !empty( $options['private'] ) && $img->userCan( File
::DELETED_FILE
) ) {
380 if ( !empty( $options['ignoreRedirect'] ) ) {
383 $redir = $this->checkRedirect( $title );
384 if ( $redir && $title->getNamespace() == NS_FILE
) {
385 $img = $this->newFile( $redir );
389 if ( $img->exists() ) {
390 $img->redirectedFrom( $title->getDBkey() );
398 * Find many files at once.
400 * @param $items array An array of titles, or an array of findFile() options with
401 * the "title" option giving the title. Example:
403 * $findItem = array( 'title' => $title, 'private' => true );
404 * $findBatch = array( $findItem );
405 * $repo->findFiles( $findBatch );
408 public function findFiles( array $items ) {
410 foreach ( $items as $item ) {
411 if ( is_array( $item ) ) {
412 $title = $item['title'];
414 unset( $options['title'] );
419 $file = $this->findFile( $title, $options );
421 $result[$file->getTitle()->getDBkey()] = $file;
428 * Find an instance of the file with this key, created at the specified time
429 * Returns false if the file does not exist. Repositories not supporting
430 * version control should return false if the time is specified.
432 * @param $sha1 String base 36 SHA-1 hash
433 * @param $options array Option array, same as findFile().
434 * @return File|bool False on failure
436 public function findFileFromKey( $sha1, $options = array() ) {
437 $time = isset( $options['time'] ) ?
$options['time'] : false;
438 # First try to find a matching current version of a file...
439 if ( $this->fileFactoryKey
) {
440 $img = call_user_func( $this->fileFactoryKey
, $sha1, $this, $time );
442 return false; // find-by-sha1 not supported
444 if ( $img && $img->exists() ) {
447 # Now try to find a matching old version of a file...
448 if ( $time !== false && $this->oldFileFactoryKey
) { // find-by-sha1 supported?
449 $img = call_user_func( $this->oldFileFactoryKey
, $sha1, $this, $time );
450 if ( $img && $img->exists() ) {
451 if ( !$img->isDeleted( File
::DELETED_FILE
) ) {
452 return $img; // always OK
453 } elseif ( !empty( $options['private'] ) && $img->userCan( File
::DELETED_FILE
) ) {
462 * Get an array or iterator of file objects for files that have a given
463 * SHA-1 content hash.
469 public function findBySha1( $hash ) {
474 * Get an array of arrays or iterators of file objects for files that
475 * have the given SHA-1 content hashes.
477 * @param $hashes array An array of hashes
478 * @return array An Array of arrays or iterators of file objects and the hash as key
480 public function findBySha1s( array $hashes ) {
482 foreach ( $hashes as $hash ) {
483 $files = $this->findBySha1( $hash );
484 if ( count( $files ) ) {
485 $result[$hash] = $files;
492 * Get the public root URL of the repository
494 * @deprecated since 1.20
497 public function getRootUrl() {
498 return $this->getZoneUrl( 'public' );
502 * Get the URL of thumb.php
506 public function getThumbScriptUrl() {
507 return $this->thumbScriptUrl
;
511 * Returns true if the repository can transform files via a 404 handler
515 public function canTransformVia404() {
516 return $this->transformVia404
;
520 * Get the name of an image from its title object
522 * @param $title Title
525 public function getNameFromTitle( Title
$title ) {
527 if ( $this->initialCapital
!= MWNamespace
::isCapitalized( NS_FILE
) ) {
528 $name = $title->getUserCaseDBKey();
529 if ( $this->initialCapital
) {
530 $name = $wgContLang->ucfirst( $name );
533 $name = $title->getDBkey();
539 * Get the public zone root storage directory of the repository
543 public function getRootDirectory() {
544 return $this->getZonePath( 'public' );
548 * Get a relative path including trailing slash, e.g. f/fa/
549 * If the repo is not hashed, returns an empty string
551 * @param $name string Name of file
554 public function getHashPath( $name ) {
555 return self
::getHashPathForLevel( $name, $this->hashLevels
);
559 * Get a relative path including trailing slash, e.g. f/fa/
560 * If the repo is not hashed, returns an empty string
562 * @param $suffix string Basename of file from FileRepo::storeTemp()
565 public function getTempHashPath( $suffix ) {
566 $parts = explode( '!', $suffix, 2 ); // format is <timestamp>!<name> or just <name>
567 $name = isset( $parts[1] ) ?
$parts[1] : $suffix; // hash path is not based on timestamp
568 return self
::getHashPathForLevel( $name, $this->hashLevels
);
576 protected static function getHashPathForLevel( $name, $levels ) {
577 if ( $levels == 0 ) {
580 $hash = md5( $name );
582 for ( $i = 1; $i <= $levels; $i++
) {
583 $path .= substr( $hash, 0, $i ) . '/';
590 * Get the number of hash directory levels
594 public function getHashLevels() {
595 return $this->hashLevels
;
599 * Get the name of this repository, as specified by $info['name]' to the constructor
603 public function getName() {
608 * Make an url to this repo
610 * @param $query mixed Query string to append
611 * @param $entry string Entry point; defaults to index
612 * @return string|bool False on failure
614 public function makeUrl( $query = '', $entry = 'index' ) {
615 if ( isset( $this->scriptDirUrl
) ) {
616 $ext = isset( $this->scriptExtension
) ?
$this->scriptExtension
: '.php';
617 return wfAppendQuery( "{$this->scriptDirUrl}/{$entry}{$ext}", $query );
623 * Get the URL of an image description page. May return false if it is
624 * unknown or not applicable. In general this should only be called by the
625 * File class, since it may return invalid results for certain kinds of
626 * repositories. Use File::getDescriptionUrl() in user code.
628 * In particular, it uses the article paths as specified to the repository
629 * constructor, whereas local repositories use the local Title functions.
631 * @param $name string
634 public function getDescriptionUrl( $name ) {
635 $encName = wfUrlencode( $name );
636 if ( !is_null( $this->descBaseUrl
) ) {
637 # "http://example.com/wiki/Image:"
638 return $this->descBaseUrl
. $encName;
640 if ( !is_null( $this->articleUrl
) ) {
641 # "http://example.com/wiki/$1"
643 # We use "Image:" as the canonical namespace for
644 # compatibility across all MediaWiki versions.
645 return str_replace( '$1',
646 "Image:$encName", $this->articleUrl
);
648 if ( !is_null( $this->scriptDirUrl
) ) {
649 # "http://example.com/w"
651 # We use "Image:" as the canonical namespace for
652 # compatibility across all MediaWiki versions,
653 # and just sort of hope index.php is right. ;)
654 return $this->makeUrl( "title=Image:$encName" );
660 * Get the URL of the content-only fragment of the description page. For
661 * MediaWiki this means action=render. This should only be called by the
662 * repository's file class, since it may return invalid results. User code
663 * should use File::getDescriptionText().
665 * @param $name String: name of image to fetch
666 * @param $lang String: language to fetch it in, if any.
669 public function getDescriptionRenderUrl( $name, $lang = null ) {
670 $query = 'action=render';
671 if ( !is_null( $lang ) ) {
672 $query .= '&uselang=' . $lang;
674 if ( isset( $this->scriptDirUrl
) ) {
675 return $this->makeUrl(
677 wfUrlencode( 'Image:' . $name ) .
680 $descUrl = $this->getDescriptionUrl( $name );
682 return wfAppendQuery( $descUrl, $query );
690 * Get the URL of the stylesheet to apply to description pages
692 * @return string|bool False on failure
694 public function getDescriptionStylesheetUrl() {
695 if ( isset( $this->scriptDirUrl
) ) {
696 return $this->makeUrl( 'title=MediaWiki:Filepage.css&' .
697 wfArrayToCGI( Skin
::getDynamicStylesheetQuery() ) );
703 * Store a file to a given destination.
705 * @param $srcPath String: source file system path, storage path, or virtual URL
706 * @param $dstZone String: destination zone
707 * @param $dstRel String: destination relative path
708 * @param $flags Integer: bitwise combination of the following flags:
709 * self::DELETE_SOURCE Delete the source file after upload
710 * self::OVERWRITE Overwrite an existing destination file instead of failing
711 * self::OVERWRITE_SAME Overwrite the file if the destination exists and has the
712 * same contents as the source
713 * self::SKIP_LOCKING Skip any file locking when doing the store
714 * @return FileRepoStatus
716 public function store( $srcPath, $dstZone, $dstRel, $flags = 0 ) {
717 $this->assertWritableRepo(); // fail out if read-only
719 $status = $this->storeBatch( array( array( $srcPath, $dstZone, $dstRel ) ), $flags );
720 if ( $status->successCount
== 0 ) {
728 * Store a batch of files
730 * @param $triplets Array: (src, dest zone, dest rel) triplets as per store()
731 * @param $flags Integer: bitwise combination of the following flags:
732 * self::DELETE_SOURCE Delete the source file after upload
733 * self::OVERWRITE Overwrite an existing destination file instead of failing
734 * self::OVERWRITE_SAME Overwrite the file if the destination exists and has the
735 * same contents as the source
736 * self::SKIP_LOCKING Skip any file locking when doing the store
737 * @throws MWException
738 * @return FileRepoStatus
740 public function storeBatch( array $triplets, $flags = 0 ) {
741 $this->assertWritableRepo(); // fail out if read-only
743 $status = $this->newGood();
744 $backend = $this->backend
; // convenience
746 $operations = array();
747 $sourceFSFilesToDelete = array(); // cleanup for disk source files
748 // Validate each triplet and get the store operation...
749 foreach ( $triplets as $triplet ) {
750 list( $srcPath, $dstZone, $dstRel ) = $triplet;
752 . "( \$src='$srcPath', \$dstZone='$dstZone', \$dstRel='$dstRel' )\n"
755 // Resolve destination path
756 $root = $this->getZonePath( $dstZone );
758 throw new MWException( "Invalid zone: $dstZone" );
760 if ( !$this->validateFilename( $dstRel ) ) {
761 throw new MWException( 'Validation error in $dstRel' );
763 $dstPath = "$root/$dstRel";
764 $dstDir = dirname( $dstPath );
765 // Create destination directories for this triplet
766 if ( !$this->initDirectory( $dstDir )->isOK() ) {
767 return $this->newFatal( 'directorycreateerror', $dstDir );
770 // Resolve source to a storage path if virtual
771 $srcPath = $this->resolveToStoragePath( $srcPath );
773 // Get the appropriate file operation
774 if ( FileBackend
::isStoragePath( $srcPath ) ) {
775 $opName = ( $flags & self
::DELETE_SOURCE
) ?
'move' : 'copy';
778 if ( $flags & self
::DELETE_SOURCE
) {
779 $sourceFSFilesToDelete[] = $srcPath;
782 $operations[] = array(
786 'overwrite' => $flags & self
::OVERWRITE
,
787 'overwriteSame' => $flags & self
::OVERWRITE_SAME
,
791 // Execute the store operation for each triplet
792 $opts = array( 'force' => true );
793 if ( $flags & self
::SKIP_LOCKING
) {
794 $opts['nonLocking'] = true;
796 $status->merge( $backend->doOperations( $operations, $opts ) );
797 // Cleanup for disk source files...
798 foreach ( $sourceFSFilesToDelete as $file ) {
799 wfSuppressWarnings();
800 unlink( $file ); // FS cleanup
808 * Deletes a batch of files.
809 * Each file can be a (zone, rel) pair, virtual url, storage path.
810 * It will try to delete each file, but ignores any errors that may occur.
812 * @param $files array List of files to delete
813 * @param $flags Integer: bitwise combination of the following flags:
814 * self::SKIP_LOCKING Skip any file locking when doing the deletions
815 * @return FileRepoStatus
817 public function cleanupBatch( array $files, $flags = 0 ) {
818 $this->assertWritableRepo(); // fail out if read-only
820 $status = $this->newGood();
822 $operations = array();
823 foreach ( $files as $path ) {
824 if ( is_array( $path ) ) {
825 // This is a pair, extract it
826 list( $zone, $rel ) = $path;
827 $path = $this->getZonePath( $zone ) . "/$rel";
829 // Resolve source to a storage path if virtual
830 $path = $this->resolveToStoragePath( $path );
832 $operations[] = array( 'op' => 'delete', 'src' => $path );
834 // Actually delete files from storage...
835 $opts = array( 'force' => true );
836 if ( $flags & self
::SKIP_LOCKING
) {
837 $opts['nonLocking'] = true;
839 $status->merge( $this->backend
->doOperations( $operations, $opts ) );
845 * Import a file from the local file system into the repo.
846 * This does no locking nor journaling and overrides existing files.
847 * This function can be used to write to otherwise read-only foreign repos.
848 * This is intended for copying generated thumbnails into the repo.
850 * @param $src string Source file system path, storage path, or virtual URL
851 * @param $dst string Virtual URL or storage path
852 * @param $disposition string|null Content-Disposition if given and supported
853 * @return FileRepoStatus
855 final public function quickImport( $src, $dst, $disposition = null ) {
856 return $this->quickImportBatch( array( array( $src, $dst, $disposition ) ) );
860 * Purge a file from the repo. This does no locking nor journaling.
861 * This function can be used to write to otherwise read-only foreign repos.
862 * This is intended for purging thumbnails.
864 * @param $path string Virtual URL or storage path
865 * @return FileRepoStatus
867 final public function quickPurge( $path ) {
868 return $this->quickPurgeBatch( array( $path ) );
872 * Deletes a directory if empty.
873 * This function can be used to write to otherwise read-only foreign repos.
875 * @param $dir string Virtual URL (or storage path) of directory to clean
878 public function quickCleanDir( $dir ) {
879 $status = $this->newGood();
880 $status->merge( $this->backend
->clean(
881 array( 'dir' => $this->resolveToStoragePath( $dir ) ) ) );
887 * Import a batch of files from the local file system into the repo.
888 * This does no locking nor journaling and overrides existing files.
889 * This function can be used to write to otherwise read-only foreign repos.
890 * This is intended for copying generated thumbnails into the repo.
892 * All path parameters may be a file system path, storage path, or virtual URL.
893 * When "dispositions" are given they are used as Content-Disposition if supported.
895 * @param $triples Array List of (source path, destination path, disposition)
896 * @return FileRepoStatus
898 public function quickImportBatch( array $triples ) {
899 $status = $this->newGood();
900 $operations = array();
901 foreach ( $triples as $triple ) {
902 list( $src, $dst ) = $triple;
903 $src = $this->resolveToStoragePath( $src );
904 $dst = $this->resolveToStoragePath( $dst );
905 $operations[] = array(
906 'op' => FileBackend
::isStoragePath( $src ) ?
'copy' : 'store',
909 'disposition' => isset( $triple[2] ) ?
$triple[2] : null
911 $status->merge( $this->initDirectory( dirname( $dst ) ) );
913 $status->merge( $this->backend
->doQuickOperations( $operations ) );
919 * Purge a batch of files from the repo.
920 * This function can be used to write to otherwise read-only foreign repos.
921 * This does no locking nor journaling and is intended for purging thumbnails.
923 * @param $paths Array List of virtual URLs or storage paths
924 * @return FileRepoStatus
926 public function quickPurgeBatch( array $paths ) {
927 $status = $this->newGood();
928 $operations = array();
929 foreach ( $paths as $path ) {
930 $operations[] = array(
932 'src' => $this->resolveToStoragePath( $path ),
933 'ignoreMissingSource' => true
936 $status->merge( $this->backend
->doQuickOperations( $operations ) );
942 * Pick a random name in the temp zone and store a file to it.
943 * Returns a FileRepoStatus object with the file Virtual URL in the value,
944 * file can later be disposed using FileRepo::freeTemp().
946 * @param $originalName String: the base name of the file as specified
947 * by the user. The file extension will be maintained.
948 * @param $srcPath String: the current location of the file.
949 * @return FileRepoStatus object with the URL in the value.
951 public function storeTemp( $originalName, $srcPath ) {
952 $this->assertWritableRepo(); // fail out if read-only
954 $date = gmdate( "YmdHis" );
955 $hashPath = $this->getHashPath( $originalName );
956 $dstRel = "{$hashPath}{$date}!{$originalName}";
957 $dstUrlRel = $hashPath . $date . '!' . rawurlencode( $originalName );
958 $virtualUrl = $this->getVirtualUrl( 'temp' ) . '/' . $dstUrlRel;
960 $result = $this->quickImport( $srcPath, $virtualUrl );
961 $result->value
= $virtualUrl;
967 * Remove a temporary file or mark it for garbage collection
969 * @param $virtualUrl String: the virtual URL returned by FileRepo::storeTemp()
970 * @return Boolean: true on success, false on failure
972 public function freeTemp( $virtualUrl ) {
973 $this->assertWritableRepo(); // fail out if read-only
975 $temp = $this->getVirtualUrl( 'temp' );
976 if ( substr( $virtualUrl, 0, strlen( $temp ) ) != $temp ) {
977 wfDebug( __METHOD__
.": Invalid temp virtual URL\n" );
981 return $this->quickPurge( $virtualUrl )->isOK();
985 * Concatenate a list of temporary files into a target file location.
987 * @param $srcPaths Array Ordered list of source virtual URLs/storage paths
988 * @param $dstPath String Target file system path
989 * @param $flags Integer: bitwise combination of the following flags:
990 * self::DELETE_SOURCE Delete the source files
991 * @param $callback Closure Optional callback function (see FileBackend::concatenate())
992 * @return FileRepoStatus
994 public function concatenate( array $srcPaths, $dstPath, $flags = 0, Closure
$callback = null ) {
995 $this->assertWritableRepo(); // fail out if read-only
997 $status = $this->newGood();
1000 foreach ( $srcPaths as $srcPath ) {
1001 // Resolve source to a storage path if virtual
1002 $source = $this->resolveToStoragePath( $srcPath );
1003 $sources[] = $source; // chunk to merge
1006 // Concatenate the chunks into one FS file
1007 $params = array( 'srcs' => $sources, 'dst' => $dstPath, 'callback' => $callback );
1008 $status->merge( $this->backend
->concatenate( $params ) );
1009 if ( !$status->isOK() ) {
1013 // Delete the sources if required
1014 if ( $flags & self
::DELETE_SOURCE
) {
1015 $status->merge( $this->quickPurgeBatch( $srcPaths ) );
1018 // Make sure status is OK, despite any quickPurgeBatch() fatals
1019 $status->setResult( true );
1025 * Copy or move a file either from a storage path, virtual URL,
1026 * or file system path, into this repository at the specified destination location.
1028 * Returns a FileRepoStatus object. On success, the value contains "new" or
1029 * "archived", to indicate whether the file was new with that name.
1031 * Options to $options include:
1032 * - headers : name/value map of HTTP headers to use in response to GET/HEAD requests
1034 * @param $srcPath String: the source file system path, storage path, or URL
1035 * @param $dstRel String: the destination relative path
1036 * @param $archiveRel String: the relative path where the existing file is to
1037 * be archived, if there is one. Relative to the public zone root.
1038 * @param $flags Integer: bitfield, may be FileRepo::DELETE_SOURCE to indicate
1039 * that the source file should be deleted if possible
1040 * @param $options Array Optional additional parameters
1041 * @return FileRepoStatus
1043 public function publish(
1044 $srcPath, $dstRel, $archiveRel, $flags = 0, array $options = array()
1046 $this->assertWritableRepo(); // fail out if read-only
1048 $status = $this->publishBatch(
1049 array( array( $srcPath, $dstRel, $archiveRel, $options ) ), $flags );
1050 if ( $status->successCount
== 0 ) {
1051 $status->ok
= false;
1053 if ( isset( $status->value
[0] ) ) {
1054 $status->value
= $status->value
[0];
1056 $status->value
= false;
1063 * Publish a batch of files
1065 * @param $ntuples Array: (source, dest, archive) triplets or
1066 * (source, dest, archive, options) 4-tuples as per publish().
1067 * @param $flags Integer: bitfield, may be FileRepo::DELETE_SOURCE to indicate
1068 * that the source files should be deleted if possible
1069 * @throws MWException
1070 * @return FileRepoStatus
1072 public function publishBatch( array $ntuples, $flags = 0 ) {
1073 $this->assertWritableRepo(); // fail out if read-only
1075 $backend = $this->backend
; // convenience
1076 // Try creating directories
1077 $status = $this->initZones( 'public' );
1078 if ( !$status->isOK() ) {
1082 $status = $this->newGood( array() );
1084 $operations = array();
1085 $sourceFSFilesToDelete = array(); // cleanup for disk source files
1086 // Validate each triplet and get the store operation...
1087 foreach ( $ntuples as $i => $ntuple ) {
1088 list( $srcPath, $dstRel, $archiveRel ) = $ntuple;
1089 $options = isset( $ntuple[3] ) ?
$ntuple[3] : array();
1090 // Resolve source to a storage path if virtual
1091 $srcPath = $this->resolveToStoragePath( $srcPath );
1092 if ( !$this->validateFilename( $dstRel ) ) {
1093 throw new MWException( 'Validation error in $dstRel' );
1095 if ( !$this->validateFilename( $archiveRel ) ) {
1096 throw new MWException( 'Validation error in $archiveRel' );
1099 $publicRoot = $this->getZonePath( 'public' );
1100 $dstPath = "$publicRoot/$dstRel";
1101 $archivePath = "$publicRoot/$archiveRel";
1103 $dstDir = dirname( $dstPath );
1104 $archiveDir = dirname( $archivePath );
1105 // Abort immediately on directory creation errors since they're likely to be repetitive
1106 if ( !$this->initDirectory( $dstDir )->isOK() ) {
1107 return $this->newFatal( 'directorycreateerror', $dstDir );
1109 if ( !$this->initDirectory($archiveDir )->isOK() ) {
1110 return $this->newFatal( 'directorycreateerror', $archiveDir );
1113 // Set any desired headers to be use in GET/HEAD responses
1114 $headers = isset( $options['headers'] ) ?
$options['headers'] : array();
1116 // Archive destination file if it exists.
1117 // This will check if the archive file also exists and fail if does.
1118 // This is a sanity check to avoid data loss. On Windows and Linux,
1119 // copy() will overwrite, so the existence check is vulnerable to
1120 // race conditions unless an functioning LockManager is used.
1121 // LocalFile also uses SELECT FOR UPDATE for synchronization.
1122 $operations[] = array(
1125 'dst' => $archivePath,
1126 'ignoreMissingSource' => true
1129 // Copy (or move) the source file to the destination
1130 if ( FileBackend
::isStoragePath( $srcPath ) ) {
1131 if ( $flags & self
::DELETE_SOURCE
) {
1132 $operations[] = array(
1136 'overwrite' => true, // replace current
1137 'headers' => $headers
1140 $operations[] = array(
1144 'overwrite' => true, // replace current
1145 'headers' => $headers
1148 } else { // FS source path
1149 $operations[] = array(
1153 'overwrite' => true, // replace current
1154 'headers' => $headers
1156 if ( $flags & self
::DELETE_SOURCE
) {
1157 $sourceFSFilesToDelete[] = $srcPath;
1162 // Execute the operations for each triplet
1163 $status->merge( $backend->doOperations( $operations ) );
1164 // Find out which files were archived...
1165 foreach ( $ntuples as $i => $ntuple ) {
1166 list( $srcPath, $dstRel, $archiveRel ) = $ntuple;
1167 $archivePath = $this->getZonePath( 'public' ) . "/$archiveRel";
1168 if ( $this->fileExists( $archivePath ) ) {
1169 $status->value
[$i] = 'archived';
1171 $status->value
[$i] = 'new';
1174 // Cleanup for disk source files...
1175 foreach ( $sourceFSFilesToDelete as $file ) {
1176 wfSuppressWarnings();
1177 unlink( $file ); // FS cleanup
1178 wfRestoreWarnings();
1185 * Creates a directory with the appropriate zone permissions.
1186 * Callers are responsible for doing read-only and "writable repo" checks.
1188 * @param $dir string Virtual URL (or storage path) of directory to clean
1191 protected function initDirectory( $dir ) {
1192 $path = $this->resolveToStoragePath( $dir );
1193 list( $b, $container, $r ) = FileBackend
::splitStoragePath( $path );
1195 $params = array( 'dir' => $path );
1196 if ( $this->isPrivate ||
$container === $this->zones
['deleted']['container'] ) {
1197 # Take all available measures to prevent web accessibility of new deleted
1198 # directories, in case the user has not configured offline storage
1199 $params = array( 'noAccess' => true, 'noListing' => true ) +
$params;
1202 return $this->backend
->prepare( $params );
1206 * Deletes a directory if empty.
1208 * @param $dir string Virtual URL (or storage path) of directory to clean
1211 public function cleanDir( $dir ) {
1212 $this->assertWritableRepo(); // fail out if read-only
1214 $status = $this->newGood();
1215 $status->merge( $this->backend
->clean(
1216 array( 'dir' => $this->resolveToStoragePath( $dir ) ) ) );
1222 * Checks existence of a a file
1224 * @param $file string Virtual URL (or storage path) of file to check
1227 public function fileExists( $file ) {
1228 $result = $this->fileExistsBatch( array( $file ) );
1233 * Checks existence of an array of files.
1235 * @param $files Array: Virtual URLs (or storage paths) of files to check
1236 * @return array|bool Either array of files and existence flags, or false
1238 public function fileExistsBatch( array $files ) {
1240 foreach ( $files as $key => $file ) {
1241 $file = $this->resolveToStoragePath( $file );
1242 $result[$key] = $this->backend
->fileExists( array( 'src' => $file ) );
1248 * Move a file to the deletion archive.
1249 * If no valid deletion archive exists, this may either delete the file
1250 * or throw an exception, depending on the preference of the repository
1252 * @param $srcRel Mixed: relative path for the file to be deleted
1253 * @param $archiveRel Mixed: relative path for the archive location.
1254 * Relative to a private archive directory.
1255 * @return FileRepoStatus object
1257 public function delete( $srcRel, $archiveRel ) {
1258 $this->assertWritableRepo(); // fail out if read-only
1260 return $this->deleteBatch( array( array( $srcRel, $archiveRel ) ) );
1264 * Move a group of files to the deletion archive.
1266 * If no valid deletion archive is configured, this may either delete the
1267 * file or throw an exception, depending on the preference of the repository.
1269 * The overwrite policy is determined by the repository -- currently LocalRepo
1270 * assumes a naming scheme in the deleted zone based on content hash, as
1271 * opposed to the public zone which is assumed to be unique.
1273 * @param $sourceDestPairs Array of source/destination pairs. Each element
1274 * is a two-element array containing the source file path relative to the
1275 * public root in the first element, and the archive file path relative
1276 * to the deleted zone root in the second element.
1277 * @throws MWException
1278 * @return FileRepoStatus
1280 public function deleteBatch( array $sourceDestPairs ) {
1281 $this->assertWritableRepo(); // fail out if read-only
1283 // Try creating directories
1284 $status = $this->initZones( array( 'public', 'deleted' ) );
1285 if ( !$status->isOK() ) {
1289 $status = $this->newGood();
1291 $backend = $this->backend
; // convenience
1292 $operations = array();
1293 // Validate filenames and create archive directories
1294 foreach ( $sourceDestPairs as $pair ) {
1295 list( $srcRel, $archiveRel ) = $pair;
1296 if ( !$this->validateFilename( $srcRel ) ) {
1297 throw new MWException( __METHOD__
.':Validation error in $srcRel' );
1298 } elseif ( !$this->validateFilename( $archiveRel ) ) {
1299 throw new MWException( __METHOD__
.':Validation error in $archiveRel' );
1302 $publicRoot = $this->getZonePath( 'public' );
1303 $srcPath = "{$publicRoot}/$srcRel";
1305 $deletedRoot = $this->getZonePath( 'deleted' );
1306 $archivePath = "{$deletedRoot}/{$archiveRel}";
1307 $archiveDir = dirname( $archivePath ); // does not touch FS
1309 // Create destination directories
1310 if ( !$this->initDirectory( $archiveDir )->isOK() ) {
1311 return $this->newFatal( 'directorycreateerror', $archiveDir );
1314 $operations[] = array(
1317 'dst' => $archivePath,
1318 // We may have 2+ identical files being deleted,
1319 // all of which will map to the same destination file
1320 'overwriteSame' => true // also see bug 31792
1324 // Move the files by execute the operations for each pair.
1325 // We're now committed to returning an OK result, which will
1326 // lead to the files being moved in the DB also.
1327 $opts = array( 'force' => true );
1328 $status->merge( $backend->doOperations( $operations, $opts ) );
1334 * Delete files in the deleted directory if they are not referenced in the filearchive table
1338 public function cleanupDeletedBatch( array $storageKeys ) {
1339 $this->assertWritableRepo();
1343 * Get a relative path for a deletion archive key,
1344 * e.g. s/z/a/ for sza251lrxrc1jad41h5mgilp8nysje52.jpg
1346 * @param $key string
1347 * @throws MWException
1350 public function getDeletedHashPath( $key ) {
1351 if ( strlen( $key ) < 31 ) {
1352 throw new MWException( "Invalid storage key '$key'." );
1355 for ( $i = 0; $i < $this->deletedHashLevels
; $i++
) {
1356 $path .= $key[$i] . '/';
1362 * If a path is a virtual URL, resolve it to a storage path.
1363 * Otherwise, just return the path as it is.
1365 * @param $path string
1367 * @throws MWException
1369 protected function resolveToStoragePath( $path ) {
1370 if ( $this->isVirtualUrl( $path ) ) {
1371 return $this->resolveVirtualUrl( $path );
1377 * Get a local FS copy of a file with a given virtual URL/storage path.
1378 * Temporary files may be purged when the file object falls out of scope.
1380 * @param $virtualUrl string
1381 * @return TempFSFile|null Returns null on failure
1383 public function getLocalCopy( $virtualUrl ) {
1384 $path = $this->resolveToStoragePath( $virtualUrl );
1385 return $this->backend
->getLocalCopy( array( 'src' => $path ) );
1389 * Get a local FS file with a given virtual URL/storage path.
1390 * The file is either an original or a copy. It should not be changed.
1391 * Temporary files may be purged when the file object falls out of scope.
1393 * @param $virtualUrl string
1394 * @return FSFile|null Returns null on failure.
1396 public function getLocalReference( $virtualUrl ) {
1397 $path = $this->resolveToStoragePath( $virtualUrl );
1398 return $this->backend
->getLocalReference( array( 'src' => $path ) );
1402 * Get properties of a file with a given virtual URL/storage path.
1403 * Properties should ultimately be obtained via FSFile::getProps().
1405 * @param $virtualUrl string
1408 public function getFileProps( $virtualUrl ) {
1409 $path = $this->resolveToStoragePath( $virtualUrl );
1410 return $this->backend
->getFileProps( array( 'src' => $path ) );
1414 * Get the timestamp of a file with a given virtual URL/storage path
1416 * @param $virtualUrl string
1417 * @return string|bool False on failure
1419 public function getFileTimestamp( $virtualUrl ) {
1420 $path = $this->resolveToStoragePath( $virtualUrl );
1421 return $this->backend
->getFileTimestamp( array( 'src' => $path ) );
1425 * Get the size of a file with a given virtual URL/storage path
1427 * @param $virtualUrl string
1428 * @return integer|bool False on failure
1430 public function getFileSize( $virtualUrl ) {
1431 $path = $this->resolveToStoragePath( $virtualUrl );
1432 return $this->backend
->getFileSize( array( 'src' => $path ) );
1436 * Get the sha1 (base 36) of a file with a given virtual URL/storage path
1438 * @param $virtualUrl string
1439 * @return string|bool
1441 public function getFileSha1( $virtualUrl ) {
1442 $path = $this->resolveToStoragePath( $virtualUrl );
1443 return $this->backend
->getFileSha1Base36( array( 'src' => $path ) );
1447 * Attempt to stream a file with the given virtual URL/storage path
1449 * @param $virtualUrl string
1450 * @param $headers Array Additional HTTP headers to send on success
1451 * @return bool Success
1453 public function streamFile( $virtualUrl, $headers = array() ) {
1454 $path = $this->resolveToStoragePath( $virtualUrl );
1455 $params = array( 'src' => $path, 'headers' => $headers );
1456 return $this->backend
->streamFile( $params )->isOK();
1460 * Call a callback function for every public regular file in the repository.
1461 * This only acts on the current version of files, not any old versions.
1462 * May use either the database or the filesystem.
1464 * @param $callback Array|string
1467 public function enumFiles( $callback ) {
1468 $this->enumFilesInStorage( $callback );
1472 * Call a callback function for every public file in the repository.
1473 * May use either the database or the filesystem.
1475 * @param $callback Array|string
1478 protected function enumFilesInStorage( $callback ) {
1479 $publicRoot = $this->getZonePath( 'public' );
1480 $numDirs = 1 << ( $this->hashLevels
* 4 );
1481 // Use a priori assumptions about directory structure
1482 // to reduce the tree height of the scanning process.
1483 for ( $flatIndex = 0; $flatIndex < $numDirs; $flatIndex++
) {
1484 $hexString = sprintf( "%0{$this->hashLevels}x", $flatIndex );
1485 $path = $publicRoot;
1486 for ( $hexPos = 0; $hexPos < $this->hashLevels
; $hexPos++
) {
1487 $path .= '/' . substr( $hexString, 0, $hexPos +
1 );
1489 $iterator = $this->backend
->getFileList( array( 'dir' => $path ) );
1490 foreach ( $iterator as $name ) {
1491 // Each item returned is a public file
1492 call_user_func( $callback, "{$path}/{$name}" );
1498 * Determine if a relative path is valid, i.e. not blank or involving directory traveral
1500 * @param $filename string
1503 public function validateFilename( $filename ) {
1504 if ( strval( $filename ) == '' ) {
1507 return FileBackend
::isPathTraversalFree( $filename );
1511 * Get a callback function to use for cleaning error message parameters
1515 function getErrorCleanupFunction() {
1516 switch ( $this->pathDisclosureProtection
) {
1518 case 'simple': // b/c
1519 $callback = array( $this, 'passThrough' );
1521 default: // 'paranoid'
1522 $callback = array( $this, 'paranoidClean' );
1528 * Path disclosure protection function
1530 * @param $param string
1533 function paranoidClean( $param ) {
1538 * Path disclosure protection function
1540 * @param $param string
1543 function passThrough( $param ) {
1548 * Create a new fatal error
1550 * @return FileRepoStatus
1552 public function newFatal( $message /*, parameters...*/ ) {
1553 $params = func_get_args();
1554 array_unshift( $params, $this );
1555 return MWInit
::callStaticMethod( 'FileRepoStatus', 'newFatal', $params );
1559 * Create a new good result
1561 * @param $value null|string
1562 * @return FileRepoStatus
1564 public function newGood( $value = null ) {
1565 return FileRepoStatus
::newGood( $this, $value );
1569 * Checks if there is a redirect named as $title. If there is, return the
1570 * title object. If not, return false.
1573 * @param $title Title of image
1576 public function checkRedirect( Title
$title ) {
1581 * Invalidates image redirect cache related to that image
1582 * Doesn't do anything for repositories that don't support image redirects.
1585 * @param $title Title of image
1587 public function invalidateImageRedirect( Title
$title ) {}
1590 * Get the human-readable name of the repo
1594 public function getDisplayName() {
1595 // We don't name our own repo, return nothing
1596 if ( $this->isLocal() ) {
1599 // 'shared-repo-name-wikimediacommons' is used when $wgUseInstantCommons = true
1600 return wfMessageFallback( 'shared-repo-name-' . $this->name
, 'shared-repo' )->text();
1604 * Get the portion of the file that contains the origin file name.
1605 * If that name is too long, then the name "thumbnail.<ext>" will be given.
1607 * @param $name string
1610 public function nameForThumb( $name ) {
1611 if ( strlen( $name ) > $this->abbrvThreshold
) {
1612 $ext = FileBackend
::extensionFromPath( $name );
1613 $name = ( $ext == '' ) ?
'thumbnail' : "thumbnail.$ext";
1619 * Returns true if this the local file repository.
1623 public function isLocal() {
1624 return $this->getName() == 'local';
1628 * Get a key on the primary cache for this repository.
1629 * Returns false if the repository's cache is not accessible at this site.
1630 * The parameters are the parts of the key, as for wfMemcKey().
1635 public function getSharedCacheKey( /*...*/ ) {
1640 * Get a key for this repo in the local cache domain. These cache keys are
1641 * not shared with remote instances of the repo.
1642 * The parameters are the parts of the key, as for wfMemcKey().
1646 public function getLocalCacheKey( /*...*/ ) {
1647 $args = func_get_args();
1648 array_unshift( $args, 'filerepo', $this->getName() );
1649 return call_user_func_array( 'wfMemcKey', $args );
1653 * Get an temporary FileRepo associated with this repo.
1654 * Files will be created in the temp zone of this repo and
1655 * thumbnails in a /temp subdirectory in thumb zone of this repo.
1656 * It will have the same backend as this repo.
1658 * @return TempFileRepo
1660 public function getTempRepo() {
1661 return new TempFileRepo( array(
1662 'name' => "{$this->name}-temp",
1663 'backend' => $this->backend
,
1666 'container' => $this->zones
['temp']['container'],
1667 'directory' => $this->zones
['temp']['directory']
1670 'container' => $this->zones
['thumb']['container'],
1671 'directory' => ( $this->zones
['thumb']['directory'] == '' )
1673 : $this->zones
['thumb']['directory'] . '/temp'
1676 'url' => $this->getZoneUrl( 'temp' ),
1677 'thumbUrl' => $this->getZoneUrl( 'thumb' ) . '/temp',
1678 'hashLevels' => $this->hashLevels
// performance
1683 * Get an UploadStash associated with this repo.
1685 * @return UploadStash
1687 public function getUploadStash() {
1688 return new UploadStash( $this );
1692 * Throw an exception if this repo is read-only by design.
1693 * This does not and should not check getReadOnlyReason().
1696 * @throws MWException
1698 protected function assertWritableRepo() {}
1702 * FileRepo for temporary files created via FileRepo::getTempRepo()
1704 class TempFileRepo
extends FileRepo
{
1705 public function getTempRepo() {
1706 throw new MWException( "Cannot get a temp repo from a temp repo." );