populateChangeTagDef.php

Go to the documentation of this file.

<?php
require_once __DIR__ . '/Maintenance.php';
 
class PopulateChangeTagDef extends LoggedUpdateMaintenance {
    protected $lbFactory;
 
    public function __construct() {
        parent::__construct();
        $this->addDescription( 'Populate and improve accuracy of change_tag_def statistics' );
        $this->addOption( 'dry-run', 'Print debug info instead of actually deleting' );
        $this->setBatchSize( 1000 );
        $this->addOption(
            'sleep',
            'Sleep time (in seconds) between every batch, defaults to zero',
            false,
            true
        );
        $this->addOption( 'populate-only', 'Do not update change_tag_def table' );
        $this->addOption( 'set-user-tags-only', 'Only update ctd_user_defined from valid_tag table' );
    }
 
    protected function doDBUpdates() {
        $this->lbFactory = $this->getServiceContainer()->getDBLoadBalancerFactory();
        $this->setBatchSize( $this->getOption( 'batch-size', $this->getBatchSize() ) );
 
        $dbr = $this->lbFactory->getMainLB()->getConnectionRef( DB_REPLICA );
        if ( $dbr->fieldExists(
                'change_tag',
                'ct_tag',
                __METHOD__
            )
        ) {
            if ( $this->hasOption( 'set-user-tags-only' ) ) {
                $this->setUserDefinedTags();
                return true;
            }
            if ( !$this->hasOption( 'populate-only' ) ) {
                $this->updateCountTag();
            }
            $this->backpopulateChangeTagId();
            $this->setUserDefinedTags();
        } else {
            $this->updateCountTagId();
        }
 
        // TODO: Implement
        // $this->cleanZeroCountRows();
 
        return true;
    }
 
    private function setUserDefinedTags() {
        $dbr = $this->lbFactory->getMainLB()->getConnectionRef( DB_REPLICA );
 
        $userTags = null;
        if ( $dbr->tableExists( 'valid_tag', __METHOD__ ) ) {
            $userTags = $dbr->newSelectQueryBuilder()
                ->select( 'vt_tag' )
                ->from( 'valid_tag' )
                ->caller( __METHOD__ )->fetchFieldValues();
        }
 
        if ( !$userTags ) {
            $this->output( "No user defined tags to set, moving on...\n" );
            return;
        }
 
        if ( $this->hasOption( 'dry-run' ) ) {
            $this->output(
                'These tags will have ctd_user_defined=1 : ' . implode( ', ', $userTags ) . "\n"
            );
            return;
        }
 
        $dbw = $this->lbFactory->getMainLB()->getConnectionRef( DB_PRIMARY );
 
        $dbw->update(
            'change_tag_def',
            [ 'ctd_user_defined' => 1 ],
            [ 'ctd_name' => $userTags ],
            __METHOD__
        );
        $this->lbFactory->waitForReplication();
        $this->output( "Finished setting user defined tags in change_tag_def table\n" );
    }
 
    private function updateCountTagId() {
        $dbr = $this->lbFactory->getMainLB()->getConnectionRef( DB_REPLICA );
 
        // This query can be pretty expensive, don't run it on master
        $res = $dbr->newSelectQueryBuilder()
            ->select( [ 'ct_tag_id', 'hitcount' => 'count(*)' ] )
            ->from( 'change_tag' )
            ->groupBy( 'ct_tag_id' )
            ->caller( __METHOD__ )->fetchResultSet();
 
        $dbw = $this->lbFactory->getMainLB()->getConnectionRef( DB_PRIMARY );
 
        foreach ( $res as $row ) {
            if ( !$row->ct_tag_id ) {
                continue;
            }
 
            if ( $this->hasOption( 'dry-run' ) ) {
                $this->output( 'This row will be updated: ' . implode( ', ', $row ) . "\n" );
                continue;
            }
 
            $dbw->update(
                'change_tag_def',
                [ 'ctd_count' => $row->hitcount ],
                [ 'ctd_id' => $row->ct_tag_id ],
                __METHOD__
            );
        }
        $this->lbFactory->waitForReplication();
    }
 
    private function updateCountTag() {
        $dbr = $this->lbFactory->getMainLB()->getConnectionRef( DB_REPLICA );
 
        // This query can be pretty expensive, don't run it on master
        $res = $dbr->newSelectQueryBuilder()
            ->select( [ 'ct_tag', 'hitcount' => 'count(*)' ] )
            ->from( 'change_tag' )
            ->groupBy( 'ct_tag' )
            ->caller( __METHOD__ )->fetchResultSet();
 
        $dbw = $this->lbFactory->getMainLB()->getConnectionRef( DB_PRIMARY );
 
        foreach ( $res as $row ) {
            // Hygiene check
            if ( !$row->ct_tag ) {
                continue;
            }
 
            if ( $this->hasOption( 'dry-run' ) ) {
                $this->output( 'This row will be updated: ' . $row->ct_tag . $row->hitcount . "\n" );
                continue;
            }
            $dbw->newInsertQueryBuilder()
                ->insertInto( 'change_tag_def' )
                ->row( [
                    'ctd_name' => $row->ct_tag,
                    'ctd_user_defined' => 0,
                    'ctd_count' => $row->hitcount
                ] )
                ->onDuplicateKeyUpdate()
                ->uniqueIndexFields( [ 'ctd_name' ] )
                ->set( [ 'ctd_count' => $row->hitcount ] )
                ->caller( __METHOD__ )->execute();
        }
        $this->lbFactory->waitForReplication();
    }
 
    private function backpopulateChangeTagId() {
        $dbr = $this->lbFactory->getMainLB()->getConnectionRef( DB_REPLICA );
        $changeTagDefs = $dbr->newSelectQueryBuilder()
            ->select( [ 'ctd_name', 'ctd_id' ] )
            ->from( 'change_tag_def' )
            ->orderBy( 'ctd_id' )
            ->caller( __METHOD__ )->fetchResultSet();
 
        foreach ( $changeTagDefs as $row ) {
            $this->backpopulateChangeTagPerTag( $row->ctd_name, $row->ctd_id );
        }
    }
 
    private function backpopulateChangeTagPerTag( $tagName, $tagId ) {
        $dbr = $this->lbFactory->getMainLB()->getConnectionRef( DB_REPLICA );
        $dbw = $this->lbFactory->getMainLB()->getConnectionRef( DB_PRIMARY );
        $sleep = (int)$this->getOption( 'sleep', 0 );
        $lastId = 0;
        $this->output( "Starting to add ct_tag_id = {$tagId} for ct_tag = {$tagName}\n" );
        while ( true ) {
            // Given that indexes might not be there, it's better to use replica
            $ids = $dbr->newSelectQueryBuilder()
                ->select( 'ct_id' )
                ->from( 'change_tag' )
                ->where( [ 'ct_tag' => $tagName, 'ct_tag_id' => null, 'ct_id > ' . $lastId ] )
                ->orderBy( 'ct_id' )
                ->limit( $this->getBatchSize() )
                ->caller( __METHOD__ )->fetchFieldValues();
 
            if ( !$ids ) {
                break;
            }
            $lastId = end( $ids );
 
            if ( $this->hasOption( 'dry-run' ) ) {
                $this->output(
                    "These ids will be changed to have \"{$tagId}\" as tag id: " . implode( ', ', $ids ) . "\n"
                );
                continue;
            } else {
                $this->output( "Updating ct_tag_id = {$tagId} up to row ct_id = {$lastId}\n" );
            }
 
            $dbw->update(
                'change_tag',
                [ 'ct_tag_id' => $tagId ],
                [ 'ct_id' => $ids ],
                __METHOD__
            );
 
            $this->lbFactory->waitForReplication();
            if ( $sleep > 0 ) {
                sleep( $sleep );
            }
        }
 
        $this->output( "Finished adding ct_tag_id = {$tagId} for ct_tag = {$tagName}\n" );
    }
 
    protected function getUpdateKey() {
        return __CLASS__;
    }
}
 
$maintClass = PopulateChangeTagDef::class;
require_once RUN_MAINTENANCE_IF_MAIN;