benchmarkParse.php

Go to the documentation of this file.

<?php
// @codeCoverageIgnoreStart
require_once __DIR__ . '/../Maintenance.php';
// @codeCoverageIgnoreEnd
 
use MediaWiki\Linker\LinkTarget;
use MediaWiki\Maintenance\Maintenance;
use MediaWiki\Page\LinkCache;
use MediaWiki\Revision\RevisionRecord;
use MediaWiki\Revision\SlotRecord;
use MediaWiki\Title\Title;
use Wikimedia\Rdbms\SelectQueryBuilder;
use Wikimedia\Timestamp\TimestampFormat as TS;
 
class BenchmarkParse extends Maintenance {
    private $templateTimestamp = null;
 
    private $clearLinkCache = false;
 
    private $linkCache;
 
    private $idCache = [];
 
    public function __construct() {
        parent::__construct();
        $this->addDescription( 'Benchmark parse operation' );
        $this->addArg( 'title', 'The name of the page to parse' );
        $this->addOption( 'warmup', 'Repeat the parse operation this number of times to warm the cache',
            false, true );
        $this->addOption( 'loops', 'Number of times to repeat parse operation post-warmup',
            false, true );
        $this->addOption( 'page-time',
            'Use the version of the page which was current at the given time',
            false, true );
        $this->addOption( 'tpl-time',
            'Use templates which were current at the given time (except that moves and ' .
            'deletes are not handled properly)',
            false, true );
        $this->addOption( 'reset-linkcache', 'Reset the LinkCache after every parse.',
            false, false );
    }
 
    public function execute() {
        if ( $this->hasOption( 'tpl-time' ) ) {
            $this->templateTimestamp = wfTimestamp( TS::MW, strtotime( $this->getOption( 'tpl-time' ) ) );
            $hookContainer = $this->getHookContainer();
            $hookContainer->register( 'BeforeParserFetchTemplateRevisionRecord', [ $this, 'onFetchTemplate' ] );
        }
 
        $this->clearLinkCache = $this->hasOption( 'reset-linkcache' );
        // Set as a member variable to avoid function calls when we're timing the parse
        $this->linkCache = $this->getServiceContainer()->getLinkCache();
 
        $title = Title::newFromText( $this->getArg( 0 ) );
        if ( !$title ) {
            $this->fatalError( "Invalid title" );
        }
 
        $revLookup = $this->getServiceContainer()->getRevisionLookup();
        if ( $this->hasOption( 'page-time' ) ) {
            $pageTimestamp = wfTimestamp( TS::MW, strtotime( $this->getOption( 'page-time' ) ) );
            $id = $this->getRevIdForTime( $title, $pageTimestamp );
            if ( !$id ) {
                $this->fatalError( "The page did not exist at that time" );
            }
 
            $revision = $revLookup->getRevisionById( (int)$id );
        } else {
            $revision = $revLookup->getRevisionByTitle( $title );
        }
 
        if ( !$revision ) {
            $this->fatalError( "Unable to load revision, incorrect title?" );
        }
 
        $warmup = $this->getOption( 'warmup', 1 );
        for ( $i = 0; $i < $warmup; $i++ ) {
            $this->runParser( $revision );
        }
 
        $loops = $this->getOption( 'loops', 1 );
        if ( $loops < 1 ) {
            $this->fatalError( 'Invalid number of loops specified' );
        }
        $startUsage = getrusage();
        $startTime = microtime( true );
        for ( $i = 0; $i < $loops; $i++ ) {
            $this->runParser( $revision );
        }
        $endUsage = getrusage();
        $endTime = microtime( true );
 
        printf( "CPU time = %.3f s, wall clock time = %.3f s\n",
            // CPU time
            ( $endUsage['ru_utime.tv_sec'] + $endUsage['ru_utime.tv_usec'] * 1e-6
            - $startUsage['ru_utime.tv_sec'] - $startUsage['ru_utime.tv_usec'] * 1e-6 ) / $loops,
            // Wall clock time
            ( $endTime - $startTime ) / $loops
        );
    }
 
    private function getRevIdForTime( Title $title, $timestamp ) {
        $dbr = $this->getReplicaDB();
 
        $id = $dbr->newSelectQueryBuilder()
            ->select( 'rev_id' )
            ->from( 'revision' )
            ->join( 'page', null, 'rev_page=page_id' )
            ->where( [ 'page_namespace' => $title->getNamespace(), 'page_title' => $title->getDBkey() ] )
            ->andWhere( $dbr->expr( 'rev_timestamp', '<=', $timestamp ) )
            ->orderBy( 'rev_timestamp', SelectQueryBuilder::SORT_DESC )
            ->caller( __METHOD__ )->fetchField();
 
        return $id;
    }
 
    private function runParser( RevisionRecord $revision ) {
        $content = $revision->getContent( SlotRecord::MAIN );
        $contentRenderer = $this->getServiceContainer()->getContentRenderer();
        // @phan-suppress-next-line PhanTypeMismatchArgumentNullable - $content is not null here
        $contentRenderer->getParserOutput( $content, $revision->getPage(), $revision );
        if ( $this->clearLinkCache ) {
            $this->linkCache->clear();
        }
    }
 
    private function onFetchTemplate(
        ?LinkTarget $contextTitle,
        LinkTarget $titleTarget,
        bool &$skip,
        ?RevisionRecord &$revRecord
    ): bool {
        $title = Title::newFromLinkTarget( $titleTarget );
 
        $pdbk = $title->getPrefixedDBkey();
        if ( !isset( $this->idCache[$pdbk] ) ) {
            $proposedId = $this->getRevIdForTime( $title, $this->templateTimestamp );
            $this->idCache[$pdbk] = $proposedId;
        }
        if ( $this->idCache[$pdbk] !== false ) {
            $revLookup = $this->getServiceContainer()->getRevisionLookup();
            $revRecord = $revLookup->getRevisionById( $this->idCache[$pdbk] );
        }
 
        return true;
    }
}
 
// @codeCoverageIgnoreStart
$maintClass = BenchmarkParse::class;
require_once RUN_MAINTENANCE_IF_MAIN;
// @codeCoverageIgnoreEnd