SiteStatsUpdate.php

Go to the documentation of this file.

<?php
use MediaWiki\MainConfigNames;
use MediaWiki\MediaWikiServices;
use MediaWiki\SiteStats\SiteStats;
use Wikimedia\Assert\Assert;
use Wikimedia\Rdbms\IDatabase;
 
class SiteStatsUpdate implements DeferrableUpdate, MergeableUpdate {
    protected $edits = 0;
    protected $pages = 0;
    protected $articles = 0;
    protected $users = 0;
    protected $images = 0;
 
    private const SHARDS_OFF = 1;
    public const SHARDS_ON = 10;
 
    private const COUNTERS = [
        'ss_total_edits'   => 'edits',
        'ss_total_pages'   => 'pages',
        'ss_good_articles' => 'articles',
        'ss_users'         => 'users',
        'ss_images'        => 'images'
    ];
 
    public function __construct( $views, $edits, $good, $pages = 0, $users = 0 ) {
        $this->edits = $edits;
        $this->articles = $good;
        $this->pages = $pages;
        $this->users = $users;
    }
 
    public function merge( MergeableUpdate $update ) {
        Assert::parameterType( __CLASS__, $update, '$update' );
        '@phan-var SiteStatsUpdate $update';
 
        foreach ( self::COUNTERS as $field ) {
            $this->$field += $update->$field;
        }
    }
 
    public static function factory( array $deltas ) {
        $update = new self( 0, 0, 0 );
 
        foreach ( $deltas as $name => $unused ) {
            if ( !in_array( $name, self::COUNTERS ) ) { // T187585
                throw new UnexpectedValueException( __METHOD__ . ": no field called '$name'" );
            }
        }
 
        foreach ( self::COUNTERS as $field ) {
            $update->$field = $deltas[$field] ?? 0;
        }
 
        return $update;
    }
 
    public function doUpdate() {
        $services = MediaWikiServices::getInstance();
        $stats = $services->getStatsdDataFactory();
        $shards = $services->getMainConfig()->get( MainConfigNames::MultiShardSiteStats ) ?
            self::SHARDS_ON : self::SHARDS_OFF;
 
        $deltaByType = [];
        foreach ( self::COUNTERS as $type ) {
            $delta = $this->$type;
            if ( $delta !== 0 ) {
                $stats->updateCount( "site.$type", $delta );
            }
            $deltaByType[$type] = $delta;
        }
 
        ( new AutoCommitUpdate(
            $services->getDBLoadBalancerFactory()->getPrimaryDatabase(),
            __METHOD__,
            static function ( IDatabase $dbw, $fname ) use ( $deltaByType, $shards ) {
                $set = [];
                $initValues = [];
                if ( $shards > 1 ) {
                    $shard = mt_rand( 1, $shards );
                } else {
                    $shard = 1;
                }
 
                $hasNegativeDelta = false;
                foreach ( self::COUNTERS as $field => $type ) {
                    $delta = (int)$deltaByType[$type];
                    $initValues[$field] = $delta;
                    if ( $delta > 0 ) {
                        $set[] = "$field=" . $dbw->buildGreatest(
                            [ $field => $dbw->addIdentifierQuotes( $field ) . '+' . abs( $delta ) ],
                            0
                        );
                    } elseif ( $delta < 0 ) {
                        $hasNegativeDelta = true;
                        $set[] = "$field=" . $dbw->buildGreatest(
                            [ 'new' => $dbw->addIdentifierQuotes( $field ) . '-' . abs( $delta ) ],
                            0
                        );
                    }
                }
 
                if ( $set ) {
                    if ( $hasNegativeDelta ) {
                        $dbw->newUpdateQueryBuilder()
                            ->update( 'site_stats' )
                            ->set( $set )
                            ->where( [ 'ss_row_id' => $shard ] )
                            ->caller( $fname )->execute();
                    } else {
                        $dbw->newInsertQueryBuilder()
                            ->insertInto( 'site_stats' )
                            ->row( array_merge( [ 'ss_row_id' => $shard ], $initValues ) )
                            ->onDuplicateKeyUpdate()
                            ->uniqueIndexFields( [ 'ss_row_id' ] )
                            ->set( $set )
                            ->caller( $fname )->execute();
                    }
                }
            }
        ) )->doUpdate();
 
        // Invalidate cache used by parser functions
        SiteStats::unload();
    }
 
    public static function cacheUpdate( IDatabase $dbw ) {
        $services = MediaWikiServices::getInstance();
        $config = $services->getMainConfig();
 
        $dbr = $services->getDBLoadBalancerFactory()->getReplicaDatabase( false, 'vslow' );
        # Get non-bot users than did some recent action other than making accounts.
        # If account creation is included, the number gets inflated ~20+ fold on enwiki.
        $activeUsers = $dbr->newSelectQueryBuilder()
            ->select( 'COUNT(DISTINCT rc_actor)' )
            ->from( 'recentchanges' )
            ->join( 'actor', 'actor', 'actor_id=rc_actor' )
            ->where( [
                'rc_type != ' . $dbr->addQuotes( RC_EXTERNAL ), // Exclude external (Wikidata)
                'actor_user IS NOT NULL',
                'rc_bot' => 0,
                'rc_log_type != ' . $dbr->addQuotes( 'newusers' ) . ' OR rc_log_type IS NULL',
                'rc_timestamp >= ' . $dbr->addQuotes(
                    $dbr->timestamp( time() - $config->get( MainConfigNames::ActiveUserDays ) * 24 * 3600 ) ),
            ] )
            ->caller( __METHOD__ )
            ->fetchField();
        $dbw->newUpdateQueryBuilder()
            ->update( 'site_stats' )
            ->set( [ 'ss_active_users' => intval( $activeUsers ) ] )
            ->where( [ 'ss_row_id' => 1 ] )
            ->caller( __METHOD__ )->execute();
 
        // Invalid cache used by parser functions
        SiteStats::unload();
 
        return $activeUsers;
    }
}