RecentChangesUpdateJob.php

Go to the documentation of this file.

<?php
use MediaWiki\MediaWikiServices;
use Wikimedia\Rdbms\DBReplicationWaitError;
 
class RecentChangesUpdateJob extends Job {
    function __construct( Title $title, array $params ) {
        parent::__construct( 'recentChangesUpdate', $title, $params );
 
        if ( !isset( $params['type'] ) ) {
            throw new Exception( "Missing 'type' parameter." );
        }
 
        $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() {
        global $wgRCMaxAge, $wgUpdateRowsPerQuery;
 
        $lockKey = wfWikiID() . ':recentchanges-prune';
 
        $dbw = wfGetDB( DB_MASTER );
        if ( !$dbw->lockIsFree( $lockKey, __METHOD__ )
            || !$dbw->lock( $lockKey, __METHOD__, 1 )
        ) {
            return; // already in progress
        }
 
        $factory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
        $ticket = $factory->getEmptyTransactionTicket( __METHOD__ );
        $cutoff = $dbw->timestamp( time() - $wgRCMaxAge );
        do {
            $rcIds = [];
            $rows = [];
            $res = $dbw->select( 'recentchanges',
                RecentChange::selectFields(),
                [ 'rc_timestamp < ' . $dbw->addQuotes( $cutoff ) ],
                __METHOD__,
                [ 'LIMIT' => $wgUpdateRowsPerQuery ]
            );
            foreach ( $res as $row ) {
                $rcIds[] = $row->rc_id;
                $rows[] = $row;
            }
            if ( $rcIds ) {
                $dbw->delete( 'recentchanges', [ 'rc_id' => $rcIds ], __METHOD__ );
                Hooks::run( 'RecentChangesPurgeRows', [ $rows ] );
                // There might be more, so try waiting for replica DBs
                try {
                    $factory->commitAndWaitForReplication(
                        __METHOD__, $ticket, [ 'timeout' => 3 ]
                    );
                } catch ( DBReplicationWaitError $e ) {
                    // Another job will continue anyway
                    break;
                }
            }
        } while ( $rcIds );
 
        $dbw->unlock( $lockKey, __METHOD__ );
    }
 
    protected function updateActiveUsers() {
        global $wgActiveUserDays;
 
        // Users that made edits at least this many days ago are "active"
        $days = $wgActiveUserDays;
        // Pull in the full window of active users in this update
        $window = $wgActiveUserDays * 86400;
 
        $dbw = wfGetDB( DB_MASTER );
        // JobRunner uses DBO_TRX, but doesn't call begin/commit itself;
        // onTransactionIdle() will run immediately since there is no trx.
        $dbw->onTransactionIdle(
            function () use ( $dbw, $days, $window ) {
                $factory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
                $ticket = $factory->getEmptyTransactionTicket( __METHOD__ );
                // Avoid disconnect/ping() cycle that makes locks fall off
                $dbw->setSessionOptions( [ 'connTimeout' => 900 ] );
 
                $lockKey = wfWikiID() . '-activeusers';
                if ( !$dbw->lockIsFree( $lockKey, __METHOD__ ) || !$dbw->lock( $lockKey, __METHOD__, 1 ) ) {
                    // Exclusive update (avoids duplicate entries)… it's usually fine to just drop out here,
                    // if the Job is already running.
                    return;
                }
 
                $nowUnix = time();
                // Get the last-updated timestamp for the cache
                $cTime = $dbw->selectField( 'querycache_info',
                    'qci_timestamp',
                    [ 'qci_type' => 'activeusers' ]
                );
                $cTimeUnix = $cTime ? 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 for sanity.
                // 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' ],
                    [ 'rc_user_text', 'lastedittime' => 'MAX(rc_timestamp)' ],
                    [
                        'rc_user > 0', // 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' => [ 'rc_user_text' ],
                        'ORDER BY' => 'NULL' // avoid filesort
                    ]
                );
                $names = [];
                foreach ( $res as $row ) {
                    $names[$row->rc_user_text] = $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_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' => 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__
                );
 
                $dbw->unlock( $lockKey, __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__
                );
            },
            __METHOD__
        );
    }
}