3 * This program is free software; you can redistribute it and/or modify
4 * it under the terms of the GNU General Public License as published by
5 * the Free Software Foundation; either version 2 of the License, or
6 * (at your option) any later version.
8 * This program is distributed in the hope that it will be useful,
9 * but WITHOUT ANY WARRANTY; without even the implied warranty of
10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11 * GNU General Public License for more details.
13 * You should have received a copy of the GNU General Public License along
14 * with this program; if not, write to the Free Software Foundation, Inc.,
15 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
16 * http://www.gnu.org/copyleft/gpl.html
21 use MediaWiki\MediaWikiServices
;
22 use MediaWiki\Revision\RevisionRecord
;
23 use MediaWiki\Revision\RevisionStore
;
24 use MediaWiki\Storage\SqlBlobStore
;
25 use Wikimedia\Assert\Assert
;
26 use Wikimedia\Rdbms\IResultWrapper
;
27 use Wikimedia\Rdbms\IDatabase
;
30 * Used to show archived pages and eventually restore them.
37 protected $fileStatus;
40 protected $revisionStatus;
45 public function __construct( $title, Config
$config = null ) {
46 if ( is_null( $title ) ) {
47 throw new MWException( __METHOD__
. ' given a null title.' );
49 $this->title
= $title;
50 if ( $config === null ) {
51 wfDebug( __METHOD__
. ' did not have a Config object passed to it' );
52 $config = MediaWikiServices
::getInstance()->getMainConfig();
54 $this->config
= $config;
58 * @return RevisionStore
60 private function getRevisionStore() {
61 // TODO: Refactor: delete()/undelete() should live in a PageStore service;
62 // Methods in PageArchive and RevisionStore that deal with archive revisions
63 // should move into an ArchiveStore service (but could still be implemented
64 // together with RevisionStore).
65 return MediaWikiServices
::getInstance()->getRevisionStore();
68 public function doesWrites() {
73 * List all deleted pages recorded in the archive table. Returns result
74 * wrapper with (ar_namespace, ar_title, count) fields, ordered by page
77 * @deprecated since 1.32.
79 * @return IResultWrapper
81 public static function listAllPages() {
82 wfDeprecated( __METHOD__
, '1.32' );
84 $dbr = wfGetDB( DB_REPLICA
);
86 return self
::listPages( $dbr, '' );
90 * List deleted pages recorded in the archive matching the
91 * given term, using search engine archive.
92 * Returns result wrapper with (ar_namespace, ar_title, count) fields.
94 * @param string $term Search term
95 * @return IResultWrapper
97 public static function listPagesBySearch( $term ) {
98 $title = Title
::newFromText( $term );
100 $ns = $title->getNamespace();
101 $termMain = $title->getText();
102 $termDb = $title->getDBkey();
104 // Prolly won't work too good
105 // @todo handle bare namespace names cleanly?
107 $termMain = $termDb = $term;
110 // Try search engine first
111 $engine = MediaWikiServices
::getInstance()->newSearchEngine();
112 $engine->setLimitOffset( 100 );
113 $engine->setNamespaces( [ $ns ] );
114 $results = $engine->searchArchiveTitle( $termMain );
115 if ( !$results->isOK() ) {
118 $results = $results->getValue();
122 // Fall back to regular prefix search
123 return self
::listPagesByPrefix( $term );
126 $dbr = wfGetDB( DB_REPLICA
);
127 $condTitles = array_unique( array_map( function ( Title
$t ) {
128 return $t->getDBkey();
131 'ar_namespace' => $ns,
132 $dbr->makeList( [ 'ar_title' => $condTitles ], LIST_OR
) . " OR ar_title " .
133 $dbr->buildLike( $termDb, $dbr->anyString() )
136 return self
::listPages( $dbr, $conds );
140 * List deleted pages recorded in the archive table matching the
141 * given title prefix.
142 * Returns result wrapper with (ar_namespace, ar_title, count) fields.
144 * @param string $prefix Title prefix
145 * @return IResultWrapper
147 public static function listPagesByPrefix( $prefix ) {
148 $dbr = wfGetDB( DB_REPLICA
);
150 $title = Title
::newFromText( $prefix );
152 $ns = $title->getNamespace();
153 $prefix = $title->getDBkey();
155 // Prolly won't work too good
156 // @todo handle bare namespace names cleanly?
161 'ar_namespace' => $ns,
162 'ar_title' . $dbr->buildLike( $prefix, $dbr->anyString() ),
165 return self
::listPages( $dbr, $conds );
169 * @param IDatabase $dbr
170 * @param string|array $condition
171 * @return bool|IResultWrapper
173 protected static function listPages( $dbr, $condition ) {
179 'count' => 'COUNT(*)'
184 'GROUP BY' => [ 'ar_namespace', 'ar_title' ],
185 'ORDER BY' => [ 'ar_namespace', 'ar_title' ],
192 * List the revisions of the given page. Returns result wrapper with
193 * various archive table fields.
195 * @return IResultWrapper
197 public function listRevisions() {
198 $revisionStore = $this->getRevisionStore();
199 $queryInfo = $revisionStore->getArchiveQueryInfo();
202 'ar_namespace' => $this->title
->getNamespace(),
203 'ar_title' => $this->title
->getDBkey(),
206 // NOTE: ordering by ar_timestamp and ar_id, to remove ambiguity.
207 // XXX: Ideally, we would be ordering by ar_timestamp and ar_rev_id, but since we
208 // don't have an index on ar_rev_id, that causes a file sort.
209 $options = [ 'ORDER BY' => 'ar_timestamp DESC, ar_id DESC' ];
211 ChangeTags
::modifyDisplayQuery(
212 $queryInfo['tables'],
213 $queryInfo['fields'],
220 $dbr = wfGetDB( DB_REPLICA
);
222 $queryInfo['tables'],
223 $queryInfo['fields'],
232 * List the deleted file revisions for this page, if it's a file page.
233 * Returns a result wrapper with various filearchive fields, or null
234 * if not a file page.
236 * @return IResultWrapper
237 * @todo Does this belong in Image for fuller encapsulation?
239 public function listFiles() {
240 if ( $this->title
->getNamespace() != NS_FILE
) {
244 $dbr = wfGetDB( DB_REPLICA
);
245 $fileQuery = ArchivedFile
::getQueryInfo();
247 $fileQuery['tables'],
248 $fileQuery['fields'],
249 [ 'fa_name' => $this->title
->getDBkey() ],
251 [ 'ORDER BY' => 'fa_timestamp DESC' ],
257 * Return a Revision object containing data for the deleted revision.
259 * @deprecated since 1.32, use getArchivedRevision() instead.
261 * @param string $timestamp
262 * @return Revision|null
264 public function getRevision( $timestamp ) {
265 $dbr = wfGetDB( DB_REPLICA
);
266 $rec = $this->getRevisionByConditions(
267 [ 'ar_timestamp' => $dbr->timestamp( $timestamp ) ]
269 return $rec ?
new Revision( $rec ) : null;
273 * Return the archived revision with the given ID.
276 * @return Revision|null
278 public function getArchivedRevision( $revId ) {
279 // Protect against code switching from getRevision() passing in a timestamp.
280 Assert
::parameterType( 'integer', $revId, '$revId' );
282 $rec = $this->getRevisionByConditions( [ 'ar_rev_id' => $revId ] );
283 return $rec ?
new Revision( $rec ) : null;
287 * @param array $conditions
288 * @param array $options
290 * @return RevisionRecord|null
292 private function getRevisionByConditions( array $conditions, array $options = [] ) {
293 $dbr = wfGetDB( DB_REPLICA
);
294 $arQuery = $this->getRevisionStore()->getArchiveQueryInfo();
296 $conditions = $conditions +
[
297 'ar_namespace' => $this->title
->getNamespace(),
298 'ar_title' => $this->title
->getDBkey(),
301 $row = $dbr->selectRow(
311 return $this->getRevisionStore()->newRevisionFromArchiveRow( $row, 0, $this->title
);
318 * Return the most-previous revision, either live or deleted, against
319 * the deleted revision given by timestamp.
321 * May produce unexpected results in case of history merges or other
322 * unusual time issues.
324 * @param string $timestamp
325 * @return Revision|null Null when there is no previous revision
327 public function getPreviousRevision( $timestamp ) {
328 $dbr = wfGetDB( DB_REPLICA
);
330 // Check the previous deleted revision...
331 $row = $dbr->selectRow( 'archive',
332 [ 'ar_rev_id', 'ar_timestamp' ],
333 [ 'ar_namespace' => $this->title
->getNamespace(),
334 'ar_title' => $this->title
->getDBkey(),
336 $dbr->addQuotes( $dbr->timestamp( $timestamp ) ) ],
339 'ORDER BY' => 'ar_timestamp DESC',
341 $prevDeleted = $row ?
wfTimestamp( TS_MW
, $row->ar_timestamp
) : false;
342 $prevDeletedId = $row ?
intval( $row->ar_rev_id
) : null;
344 $row = $dbr->selectRow( [ 'page', 'revision' ],
345 [ 'rev_id', 'rev_timestamp' ],
347 'page_namespace' => $this->title
->getNamespace(),
348 'page_title' => $this->title
->getDBkey(),
349 'page_id = rev_page',
351 $dbr->addQuotes( $dbr->timestamp( $timestamp ) ) ],
354 'ORDER BY' => 'rev_timestamp DESC',
356 $prevLive = $row ?
wfTimestamp( TS_MW
, $row->rev_timestamp
) : false;
357 $prevLiveId = $row ?
intval( $row->rev_id
) : null;
359 if ( $prevLive && $prevLive > $prevDeleted ) {
360 // Most prior revision was live
361 $rec = $this->getRevisionStore()->getRevisionById( $prevLiveId );
362 $rec = $rec ?
new Revision( $rec ) : null;
363 } elseif ( $prevDeleted ) {
364 // Most prior revision was deleted
365 $rec = $this->getArchivedRevision( $prevDeletedId );
374 * Get the text from an archive row containing ar_text_id.
376 * @deprecated since 1.32. In the MCR schema, ar_text_id no longer exists.
377 * Calling code should switch to getArchiveRevision().
379 * @todo remove in 1.33
381 * @param object $row Database row
384 public function getTextFromRow( $row ) {
385 wfDeprecated( __METHOD__
, '1.32' );
387 if ( empty( $row->ar_text_id
) ) {
388 throw new InvalidArgumentException( '$row->ar_text_id must be set and not empty!' );
391 $address = SqlBlobStore
::makeAddressFromTextId( $row->ar_text_id
);
392 $blobStore = MediaWikiServices
::getInstance()->getBlobStore();
394 return $blobStore->getBlob( $address );
398 * Fetch (and decompress if necessary) the stored text of the most
399 * recently edited deleted revision of the page.
401 * If there are no archived revisions for the page, returns NULL.
403 * @note this bypasses any audience checks.
405 * @deprecated since 1.32. For compatibility with the MCR schema,
406 * calling code should switch to getLastRevisionId() and getArchiveRevision().
408 * @todo remove in 1.33
410 * @return string|null
412 public function getLastRevisionText() {
413 wfDeprecated( __METHOD__
, '1.32' );
415 $revId = $this->getLastRevisionId();
418 $rev = $this->getArchivedRevision( $revId );
419 $content = $rev->getContent( RevisionRecord
::RAW
);
420 return $content->serialize();
427 * Returns the ID of the latest deleted revision.
429 * @return int|false The revision's ID, or false if there is no deleted revision.
431 public function getLastRevisionId() {
432 $dbr = wfGetDB( DB_REPLICA
);
433 $revId = $dbr->selectField(
436 [ 'ar_namespace' => $this->title
->getNamespace(),
437 'ar_title' => $this->title
->getDBkey() ],
439 [ 'ORDER BY' => 'ar_timestamp DESC, ar_id DESC' ]
442 return $revId ?
intval( $revId ) : false;
446 * Quick check if any archived revisions are present for the page.
447 * This says nothing about whether the page currently exists in the page table or not.
451 public function isDeleted() {
452 $dbr = wfGetDB( DB_REPLICA
);
453 $row = $dbr->selectRow(
455 '1', // We don't care about the value. Allow the database to optimize.
456 [ 'ar_namespace' => $this->title
->getNamespace(),
457 'ar_title' => $this->title
->getDBkey() ],
465 * Restore the given (or all) text and file revisions for the page.
466 * Once restored, the items will be removed from the archive tables.
467 * The deletion log will be updated with an undeletion notice.
469 * This also sets Status objects, $this->fileStatus and $this->revisionStatus
470 * (depending what operations are attempted).
472 * @param array $timestamps Pass an empty array to restore all revisions,
473 * otherwise list the ones to undelete.
474 * @param string $comment
475 * @param array $fileVersions
476 * @param bool $unsuppress
477 * @param User|null $user User performing the action, or null to use $wgUser
478 * @param string|string[]|null $tags Change tags to add to log entry
479 * ($user should be able to add the specified tags before this is called)
480 * @return array|bool array(number of file revisions restored, number of image revisions
481 * restored, log message) on success, false on failure.
483 public function undelete( $timestamps, $comment = '', $fileVersions = [],
484 $unsuppress = false, User
$user = null, $tags = null
486 // If both the set of text revisions and file revisions are empty,
487 // restore everything. Otherwise, just restore the requested items.
488 $restoreAll = empty( $timestamps ) && empty( $fileVersions );
490 $restoreText = $restoreAll ||
!empty( $timestamps );
491 $restoreFiles = $restoreAll ||
!empty( $fileVersions );
493 if ( $restoreFiles && $this->title
->getNamespace() == NS_FILE
) {
494 $img = wfLocalFile( $this->title
);
495 $img->load( File
::READ_LATEST
);
496 $this->fileStatus
= $img->restore( $fileVersions, $unsuppress );
497 if ( !$this->fileStatus
->isOK() ) {
500 $filesRestored = $this->fileStatus
->successCount
;
505 if ( $restoreText ) {
506 $this->revisionStatus
= $this->undeleteRevisions( $timestamps, $unsuppress, $comment );
507 if ( !$this->revisionStatus
->isOK() ) {
511 $textRestored = $this->revisionStatus
->getValue();
518 if ( !$textRestored && !$filesRestored ) {
519 wfDebug( "Undelete: nothing undeleted...\n" );
524 if ( $user === null ) {
529 $logEntry = new ManualLogEntry( 'delete', 'restore' );
530 $logEntry->setPerformer( $user );
531 $logEntry->setTarget( $this->title
);
532 $logEntry->setComment( $comment );
533 $logEntry->setTags( $tags );
534 $logEntry->setParameters( [
536 'revisions' => $textRestored,
537 'files' => $filesRestored,
541 Hooks
::run( 'ArticleUndeleteLogEntry', [ $this, &$logEntry, $user ] );
543 $logid = $logEntry->insert();
544 $logEntry->publish( $logid );
546 return [ $textRestored, $filesRestored, $comment ];
550 * This is the meaty bit -- It restores archived revisions of the given page
551 * to the revision table.
553 * @param array $timestamps Pass an empty array to restore all revisions,
554 * otherwise list the ones to undelete.
555 * @param bool $unsuppress Remove all ar_deleted/fa_deleted restrictions of seletected revs
556 * @param string $comment
557 * @throws ReadOnlyError
558 * @return Status Status object containing the number of revisions restored on success
560 private function undeleteRevisions( $timestamps, $unsuppress = false, $comment = '' ) {
561 if ( wfReadOnly() ) {
562 throw new ReadOnlyError();
565 $dbw = wfGetDB( DB_MASTER
);
566 $dbw->startAtomic( __METHOD__
);
568 $restoreAll = empty( $timestamps );
570 # Does this page already exist? We'll have to update it...
571 $article = WikiPage
::factory( $this->title
);
572 # Load latest data for the current page (T33179)
573 $article->loadPageData( 'fromdbmaster' );
574 $oldcountable = $article->isCountable();
576 $page = $dbw->selectRow( 'page',
577 [ 'page_id', 'page_latest' ],
578 [ 'page_namespace' => $this->title
->getNamespace(),
579 'page_title' => $this->title
->getDBkey() ],
581 [ 'FOR UPDATE' ] // lock page
586 # Page already exists. Import the history, and if necessary
587 # we'll update the latest revision field in the record.
589 # Get the time span of this page
590 $previousTimestamp = $dbw->selectField( 'revision', 'rev_timestamp',
591 [ 'rev_id' => $page->page_latest
],
594 if ( $previousTimestamp === false ) {
595 wfDebug( __METHOD__
. ": existing page refers to a page_latest that does not exist\n" );
597 $status = Status
::newGood( 0 );
598 $status->warning( 'undeleterevision-missing' );
599 $dbw->endAtomic( __METHOD__
);
604 # Have to create a new article...
606 $previousTimestamp = 0;
610 'ar_namespace' => $this->title
->getNamespace(),
611 'ar_title' => $this->title
->getDBkey(),
613 if ( !$restoreAll ) {
614 $oldWhere['ar_timestamp'] = array_map( [ &$dbw, 'timestamp' ], $timestamps );
617 $revisionStore = $this->getRevisionStore();
618 $queryInfo = $revisionStore->getArchiveQueryInfo();
619 $queryInfo['tables'][] = 'revision';
620 $queryInfo['fields'][] = 'rev_id';
621 $queryInfo['joins']['revision'] = [ 'LEFT JOIN', 'ar_rev_id=rev_id' ];
624 * Select each archived revision...
626 $result = $dbw->select(
627 $queryInfo['tables'],
628 $queryInfo['fields'],
632 [ 'ORDER BY' => 'ar_timestamp' ],
636 $rev_count = $result->numRows();
638 wfDebug( __METHOD__
. ": no revisions to restore\n" );
640 $status = Status
::newGood( 0 );
641 $status->warning( "undelete-no-results" );
642 $dbw->endAtomic( __METHOD__
);
647 // We use ar_id because there can be duplicate ar_rev_id even for the same
648 // page. In this case, we may be able to restore the first one.
649 $restoreFailedArIds = [];
651 // Map rev_id to the ar_id that is allowed to use it. When checking later,
652 // if it doesn't match, the current ar_id can not be restored.
654 // Value can be an ar_id or -1 (-1 means no ar_id can use it, since the
655 // rev_id is taken before we even start the restore).
656 $allowedRevIdToArIdMap = [];
658 $latestRestorableRow = null;
660 foreach ( $result as $row ) {
661 if ( $row->ar_rev_id
) {
662 // rev_id is taken even before we start restoring.
663 if ( $row->ar_rev_id
=== $row->rev_id
) {
664 $restoreFailedArIds[] = $row->ar_id
;
665 $allowedRevIdToArIdMap[$row->ar_rev_id
] = -1;
667 // rev_id is not taken yet in the DB, but it might be taken
668 // by a prior revision in the same restore operation. If
669 // not, we need to reserve it.
670 if ( isset( $allowedRevIdToArIdMap[$row->ar_rev_id
] ) ) {
671 $restoreFailedArIds[] = $row->ar_id
;
673 $allowedRevIdToArIdMap[$row->ar_rev_id
] = $row->ar_id
;
674 $latestRestorableRow = $row;
678 // If ar_rev_id is null, there can't be a collision, and a
679 // rev_id will be chosen automatically.
680 $latestRestorableRow = $row;
684 $result->seek( 0 ); // move back
687 if ( $latestRestorableRow !== null ) {
688 $oldPageId = (int)$latestRestorableRow->ar_page_id
; // pass this to ArticleUndelete hook
690 // Grab the content to check consistency with global state before restoring the page.
691 // XXX: The only current use case is Wikibase, which tries to enforce uniqueness of
692 // certain things across all pages. There may be a better way to do that.
693 $revision = $revisionStore->newRevisionFromArchiveRow(
694 $latestRestorableRow,
699 // TODO: use User::newFromUserIdentity from If610c68f4912e
700 // TODO: The User isn't used for anything in prepareSave()! We should drop it.
701 $user = User
::newFromName( $revision->getUser( RevisionRecord
::RAW
)->getName(), false );
703 foreach ( $revision->getSlotRoles() as $role ) {
704 $content = $revision->getContent( $role, RevisionRecord
::RAW
);
706 // NOTE: article ID may not be known yet. prepareSave() should not modify the database.
707 $status = $content->prepareSave( $article, 0, -1, $user );
708 if ( !$status->isOK() ) {
709 $dbw->endAtomic( __METHOD__
);
716 $newid = false; // newly created page ID
717 $restored = 0; // number of revisions restored
718 /** @var RevisionRecord|null $revision */
721 // If there are no restorable revisions, we can skip most of the steps.
722 if ( $latestRestorableRow === null ) {
723 $failedRevisionCount = $rev_count;
726 // Check the state of the newest to-be version...
728 && ( $latestRestorableRow->ar_deleted
& RevisionRecord
::DELETED_TEXT
)
730 $dbw->endAtomic( __METHOD__
);
732 return Status
::newFatal( "undeleterevdel" );
734 // Safe to insert now...
735 $newid = $article->insertOn( $dbw, $latestRestorableRow->ar_page_id
);
736 if ( $newid === false ) {
737 // The old ID is reserved; let's pick another
738 $newid = $article->insertOn( $dbw );
742 // Check if a deleted revision will become the current revision...
743 if ( $latestRestorableRow->ar_timestamp
> $previousTimestamp ) {
744 // Check the state of the newest to-be version...
746 && ( $latestRestorableRow->ar_deleted
& RevisionRecord
::DELETED_TEXT
)
748 $dbw->endAtomic( __METHOD__
);
750 return Status
::newFatal( "undeleterevdel" );
755 $pageId = $article->getId();
758 foreach ( $result as $row ) {
759 // Check for key dupes due to needed archive integrity.
760 if ( $row->ar_rev_id
&& $allowedRevIdToArIdMap[$row->ar_rev_id
] !== $row->ar_id
) {
763 // Insert one revision at a time...maintaining deletion status
764 // unless we are specifically removing all restrictions...
765 $revision = $revisionStore->newRevisionFromArchiveRow(
770 'page_id' => $pageId,
771 'deleted' => $unsuppress ?
0 : $row->ar_deleted
775 // This will also copy the revision to ip_changes if it was an IP edit.
776 $revisionStore->insertRevisionOn( $revision, $dbw );
780 $legacyRevision = new Revision( $revision );
781 Hooks
::run( 'ArticleRevisionUndeleted',
782 [ &$this->title
, $legacyRevision, $row->ar_page_id
] );
783 $restoredPages[$row->ar_page_id
] = true;
786 // Now that it's safely stored, take it out of the archive
787 // Don't delete rows that we failed to restore
788 $toDeleteConds = $oldWhere;
789 $failedRevisionCount = count( $restoreFailedArIds );
790 if ( $failedRevisionCount > 0 ) {
791 $toDeleteConds[] = 'ar_id NOT IN ( ' . $dbw->makeList( $restoreFailedArIds ) . ' )';
794 $dbw->delete( 'archive',
799 $status = Status
::newGood( $restored );
801 if ( $failedRevisionCount > 0 ) {
803 wfMessage( 'undeleterevision-duplicate-revid', $failedRevisionCount ) );
806 // Was anything restored at all?
808 $created = (bool)$newid;
809 // Attach the latest revision to the page...
810 // XXX: updateRevisionOn should probably move into a PageStore service.
811 $wasnew = $article->updateIfNewerOn( $dbw, $legacyRevision );
812 if ( $created ||
$wasnew ) {
813 // Update site stats, link tables, etc
814 // TODO: use DerivedPageDataUpdater from If610c68f4912e!
815 $article->doEditUpdates(
817 User
::newFromName( $revision->getUser( RevisionRecord
::RAW
)->getName(), false ),
819 'created' => $created,
820 'oldcountable' => $oldcountable,
826 Hooks
::run( 'ArticleUndelete',
827 [ &$this->title
, $created, $comment, $oldPageId, $restoredPages ] );
828 if ( $this->title
->getNamespace() == NS_FILE
) {
829 DeferredUpdates
::addUpdate(
830 new HTMLCacheUpdate( $this->title
, 'imagelinks', 'file-restore' )
835 $dbw->endAtomic( __METHOD__
);
843 public function getFileStatus() {
844 return $this->fileStatus
;
850 public function getRevisionStatus() {
851 return $this->revisionStatus
;