LinksDeletionUpdate.php

Go to the documentation of this file.

<?php
use MediaWiki\MediaWikiServices;
use Wikimedia\ScopedCallback;
use Wikimedia\Rdbms\IDatabase;
 
class LinksDeletionUpdate extends DataUpdate implements EnqueueableDataUpdate {
    protected $page;
    protected $pageId;
    protected $timestamp;
 
    private $db;
 
    function __construct( WikiPage $page, $pageId = null, $timestamp = null ) {
        parent::__construct();
 
        $this->page = $page;
        if ( $pageId ) {
            $this->pageId = $pageId; // page ID at time of deletion
        } elseif ( $page->exists() ) {
            $this->pageId = $page->getId();
        } else {
            throw new InvalidArgumentException( "Page ID not known. Page doesn't exist?" );
        }
 
        $this->timestamp = $timestamp ?: wfTimestampNow();
    }
 
    public function doUpdate() {
        $services = MediaWikiServices::getInstance();
        $config = $services->getMainConfig();
        $lbFactory = $services->getDBLoadBalancerFactory();
        $batchSize = $config->get( 'UpdateRowsPerQuery' );
 
        // Page may already be deleted, so don't just getId()
        $id = $this->pageId;
 
        if ( $this->ticket ) {
            // Make sure all links update threads see the changes of each other.
            // This handles the case when updates have to batched into several COMMITs.
            $scopedLock = LinksUpdate::acquirePageLock( $this->getDB(), $id );
            if ( !$scopedLock ) {
                throw new RuntimeException( "Could not acquire lock for page ID '{$id}'." );
            }
        }
 
        $title = $this->page->getTitle();
        $dbw = $this->getDB(); // convenience
 
        // Delete restrictions for it
        $dbw->delete( 'page_restrictions', [ 'pr_page' => $id ], __METHOD__ );
 
        // Fix category table counts
        $cats = $dbw->selectFieldValues(
            'categorylinks',
            'cl_to',
            [ 'cl_from' => $id ],
            __METHOD__
        );
        $catBatches = array_chunk( $cats, $batchSize );
        foreach ( $catBatches as $catBatch ) {
            $this->page->updateCategoryCounts( [], $catBatch, $id );
            if ( count( $catBatches ) > 1 ) {
                // Only sacrifice atomicity if necessary due to size
                $lbFactory->commitAndWaitForReplication(
                    __METHOD__, $this->ticket, [ 'domain' => $dbw->getDomainID() ]
                );
            }
        }
 
        // Refresh counts on categories that should be empty now
        if ( $title->getNamespace() === NS_CATEGORY ) {
            // T166757: do the update after the main job DB commit
            DeferredUpdates::addCallableUpdate( function () use ( $title ) {
                $cat = Category::newFromName( $title->getDBkey() );
                $cat->refreshCountsIfEmpty();
            } );
        }
 
        $this->batchDeleteByPK(
            'pagelinks',
            [ 'pl_from' => $id ],
            [ 'pl_from', 'pl_namespace', 'pl_title' ],
            $batchSize
        );
        $this->batchDeleteByPK(
            'imagelinks',
            [ 'il_from' => $id ],
            [ 'il_from', 'il_to' ],
            $batchSize
        );
        $this->batchDeleteByPK(
            'categorylinks',
            [ 'cl_from' => $id ],
            [ 'cl_from', 'cl_to' ],
            $batchSize
        );
        $this->batchDeleteByPK(
            'templatelinks',
            [ 'tl_from' => $id ],
            [ 'tl_from', 'tl_namespace', 'tl_title' ],
            $batchSize
        );
        $this->batchDeleteByPK(
            'externallinks',
            [ 'el_from' => $id ],
            [ 'el_id' ],
            $batchSize
        );
        $this->batchDeleteByPK(
            'langlinks',
            [ 'll_from' => $id ],
            [ 'll_from', 'll_lang' ],
            $batchSize
        );
        $this->batchDeleteByPK(
            'iwlinks',
            [ 'iwl_from' => $id ],
            [ 'iwl_from', 'iwl_prefix', 'iwl_title' ],
            $batchSize
        );
 
        // Delete any redirect entry or page props entries
        $dbw->delete( 'redirect', [ 'rd_from' => $id ], __METHOD__ );
        $dbw->delete( 'page_props', [ 'pp_page' => $id ], __METHOD__ );
 
        // Find recentchanges entries to clean up...
        $rcIdsForTitle = $dbw->selectFieldValues(
            'recentchanges',
            'rc_id',
            [
                'rc_type != ' . RC_LOG,
                'rc_namespace' => $title->getNamespace(),
                'rc_title' => $title->getDBkey(),
                'rc_timestamp < ' .
                    $dbw->addQuotes( $dbw->timestamp( $this->timestamp ) )
            ],
            __METHOD__
        );
        $rcIdsForPage = $dbw->selectFieldValues(
            'recentchanges',
            'rc_id',
            [ 'rc_type != ' . RC_LOG, 'rc_cur_id' => $id ],
            __METHOD__
        );
 
        // T98706: delete by PK to avoid lock contention with RC delete log insertions
        $rcIdBatches = array_chunk( array_merge( $rcIdsForTitle, $rcIdsForPage ), $batchSize );
        foreach ( $rcIdBatches as $rcIdBatch ) {
            $dbw->delete( 'recentchanges', [ 'rc_id' => $rcIdBatch ], __METHOD__ );
            if ( count( $rcIdBatches ) > 1 ) {
                $lbFactory->commitAndWaitForReplication(
                    __METHOD__, $this->ticket, [ 'domain' => $dbw->getDomainID() ]
                );
            }
        }
 
        // Commit and release the lock (if set)
        ScopedCallback::consume( $scopedLock );
    }
 
    private function batchDeleteByPK( $table, array $conds, array $pk, $bSize ) {
        $services = MediaWikiServices::getInstance();
        $lbFactory = $services->getDBLoadBalancerFactory();
        $dbw = $this->getDB(); // convenience
 
        $res = $dbw->select( $table, $pk, $conds, __METHOD__ );
 
        $pkDeleteConds = [];
        foreach ( $res as $row ) {
            $pkDeleteConds[] = $dbw->makeList( (array)$row, LIST_AND );
            if ( count( $pkDeleteConds ) >= $bSize ) {
                $dbw->delete( $table, $dbw->makeList( $pkDeleteConds, LIST_OR ), __METHOD__ );
                $lbFactory->commitAndWaitForReplication(
                    __METHOD__, $this->ticket, [ 'domain' => $dbw->getDomainID() ]
                );
                $pkDeleteConds = [];
            }
        }
 
        if ( $pkDeleteConds ) {
            $dbw->delete( $table, $dbw->makeList( $pkDeleteConds, LIST_OR ), __METHOD__ );
        }
    }
 
    protected function getDB() {
        if ( !$this->db ) {
            $this->db = wfGetDB( DB_MASTER );
        }
 
        return $this->db;
    }
 
    public function getAsJobSpecification() {
        return [
            'wiki' => WikiMap::getWikiIdFromDomain( $this->getDB()->getDomainID() ),
            'job'  => new JobSpecification(
                'deleteLinks',
                [ 'pageId' => $this->pageId, 'timestamp' => $this->timestamp ],
                [ 'removeDuplicates' => true ],
                $this->page->getTitle()
            )
        ];
    }
}