RecentChangesUpdateJob.php

Go to the documentation of this file.

<?php
use MediaWiki\MainConfigNames;
use MediaWiki\MediaWikiServices;
 
class RecentChangesUpdateJob extends Job {
    public function __construct( Title $title, array $params ) {
        parent::__construct( 'recentChangesUpdate', $title, $params );
 
        if ( !isset( $params['type'] ) ) {
            throw new Exception( "Missing 'type' parameter." );
        }
 
        $this->executionFlags |= self::JOB_NO_EXPLICIT_TRX_ROUND;
        $this->removeDuplicates = true;
    }
 
    final public static function newPurgeJob() {
        return new self(
            SpecialPage::getTitleFor( 'Recentchanges' ), [ 'type' => 'purge' ]
        );
    }
 
    final public static function newCacheUpdateJob() {
        return new self(
            SpecialPage::getTitleFor( 'Recentchanges' ), [ 'type' => 'cacheUpdate' ]
        );
    }
 
    public function run() {
        if ( $this->params['type'] === 'purge' ) {
            $this->purgeExpiredRows();
        } elseif ( $this->params['type'] === 'cacheUpdate' ) {
            $this->updateActiveUsers();
        } else {
            throw new InvalidArgumentException(
                "Invalid 'type' parameter '{$this->params['type']}'." );
        }
 
        return true;
    }
 
    protected function purgeExpiredRows() {
        $rcMaxAge = MediaWikiServices::getInstance()->getMainConfig()->get(
            MainConfigNames::RCMaxAge );
        $updateRowsPerQuery = MediaWikiServices::getInstance()->getMainConfig()->get(
            MainConfigNames::UpdateRowsPerQuery );
        $dbw = wfGetDB( DB_PRIMARY );
        $lockKey = $dbw->getDomainID() . ':recentchanges-prune';
        if ( !$dbw->lock( $lockKey, __METHOD__, 0 ) ) {
            // already in progress
            return;
        }
 
        $factory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
        $ticket = $factory->getEmptyTransactionTicket( __METHOD__ );
        $cutoff = $dbw->timestamp( time() - $rcMaxAge );
        $rcQuery = RecentChange::getQueryInfo();
        do {
            $rcIds = [];
            $rows = [];
            $res = $dbw->select(
                $rcQuery['tables'],
                $rcQuery['fields'],
                [ 'rc_timestamp < ' . $dbw->addQuotes( $cutoff ) ],
                __METHOD__,
                [ 'LIMIT' => $updateRowsPerQuery ],
                $rcQuery['joins']
            );
            foreach ( $res as $row ) {
                $rcIds[] = $row->rc_id;
                $rows[] = $row;
            }
            if ( $rcIds ) {
                $dbw->delete( 'recentchanges', [ 'rc_id' => $rcIds ], __METHOD__ );
                Hooks::runner()->onRecentChangesPurgeRows( $rows );
                // There might be more, so try waiting for replica DBs
                if ( !$factory->commitAndWaitForReplication(
                    __METHOD__, $ticket, [ 'timeout' => 3 ]
                ) ) {
                    // Another job will continue anyway
                    break;
                }
            }
        } while ( $rcIds );
 
        $dbw->unlock( $lockKey, __METHOD__ );
    }
 
    protected function updateActiveUsers() {
        $activeUserDays = MediaWikiServices::getInstance()->getMainConfig()->get(
            MainConfigNames::ActiveUserDays );
 
        // Users that made edits at least this many days ago are "active"
        $days = $activeUserDays;
        // Pull in the full window of active users in this update
        $window = $activeUserDays * 86400;
 
        $dbw = wfGetDB( DB_PRIMARY );
        $factory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
        $ticket = $factory->getEmptyTransactionTicket( __METHOD__ );
 
        $lockKey = $dbw->getDomainID() . '-activeusers';
        if ( !$dbw->lock( $lockKey, __METHOD__, 0 ) ) {
            // Exclusive update (avoids duplicate entries)… it's usually fine to just
            // drop out here, if the Job is already running.
            return;
        }
 
        // Long-running queries expected
        $dbw->setSessionOptions( [ 'connTimeout' => 900 ] );
 
        $nowUnix = time();
        // Get the last-updated timestamp for the cache
        $cTime = $dbw->selectField( 'querycache_info',
            'qci_timestamp',
            [ 'qci_type' => 'activeusers' ],
            __METHOD__
        );
        $cTimeUnix = $cTime ? (int)wfTimestamp( TS_UNIX, $cTime ) : 1;
 
        // Pick the date range to fetch from. This is normally from the last
        // update to till the present time, but has a limited window.
        // If the window is limited, multiple runs are need to fully populate it.
        $sTimestamp = max( $cTimeUnix, $nowUnix - $days * 86400 );
        $eTimestamp = min( $sTimestamp + $window, $nowUnix );
 
        // Get all the users active since the last update
        $res = $dbw->select(
            [ 'recentchanges', 'actor' ],
            [
                'actor_name',
                'lastedittime' => 'MAX(rc_timestamp)'
            ],
            [
                'actor_user IS NOT NULL', // actual accounts
                'rc_type != ' . $dbw->addQuotes( RC_EXTERNAL ), // no wikidata
                'rc_log_type IS NULL OR rc_log_type != ' . $dbw->addQuotes( 'newusers' ),
                'rc_timestamp >= ' . $dbw->addQuotes( $dbw->timestamp( $sTimestamp ) ),
                'rc_timestamp <= ' . $dbw->addQuotes( $dbw->timestamp( $eTimestamp ) )
            ],
            __METHOD__,
            [
                'GROUP BY' => 'actor_name',
                'ORDER BY' => 'NULL' // avoid filesort
            ],
            [
                'actor' => [ 'JOIN', 'actor_id=rc_actor' ]
            ]
        );
        $names = [];
        foreach ( $res as $row ) {
            $names[$row->actor_name] = $row->lastedittime;
        }
 
        // Find which of the recently active users are already accounted for
        if ( count( $names ) ) {
            $res = $dbw->select( 'querycachetwo',
                [ 'user_name' => 'qcc_title' ],
                [
                    'qcc_type' => 'activeusers',
                    'qcc_namespace' => NS_USER,
                    'qcc_title' => array_map( 'strval', array_keys( $names ) ),
                    'qcc_value >= ' . $dbw->addQuotes( $nowUnix - $days * 86400 ), // TS_UNIX
                ],
                __METHOD__
            );
            // Note: In order for this to be actually consistent, we would need
            // to update these rows with the new lastedittime.
            foreach ( $res as $row ) {
                unset( $names[$row->user_name] );
            }
        }
 
        // Insert the users that need to be added to the list
        if ( count( $names ) ) {
            $newRows = [];
            foreach ( $names as $name => $lastEditTime ) {
                $newRows[] = [
                    'qcc_type' => 'activeusers',
                    'qcc_namespace' => NS_USER,
                    'qcc_title' => $name,
                    'qcc_value' => (int)wfTimestamp( TS_UNIX, $lastEditTime ),
                    'qcc_namespacetwo' => 0, // unused
                    'qcc_titletwo' => '' // unused
                ];
            }
            foreach ( array_chunk( $newRows, 500 ) as $rowBatch ) {
                $dbw->insert( 'querycachetwo', $rowBatch, __METHOD__ );
                $factory->commitAndWaitForReplication( __METHOD__, $ticket );
            }
        }
 
        // If a transaction was already started, it might have an old
        // snapshot, so kludge the timestamp range back as needed.
        $asOfTimestamp = min( $eTimestamp, (int)$dbw->trxTimestamp() );
 
        // Touch the data freshness timestamp
        $dbw->replace(
            'querycache_info',
            'qci_type',
            [ 'qci_type' => 'activeusers',
                'qci_timestamp' => $dbw->timestamp( $asOfTimestamp ) ], // not always $now
            __METHOD__
        );
 
        // Rotate out users that have not edited in too long (according to old data set)
        $dbw->delete( 'querycachetwo',
            [
                'qcc_type' => 'activeusers',
                'qcc_value < ' . $dbw->addQuotes( $nowUnix - $days * 86400 ) // TS_UNIX
            ],
            __METHOD__
        );
 
        if ( !MediaWikiServices::getInstance()->getMainConfig()->get( MainConfigNames::MiserMode ) ) {
            SiteStatsUpdate::cacheUpdate( $dbw );
        }
 
        $dbw->unlock( $lockKey, __METHOD__ );
    }
}