compareParserCache.php

Go to the documentation of this file.

<?php
require_once __DIR__ . '/Maintenance.php';
 
use MediaWiki\MediaWikiServices;
 
class CompareParserCache extends Maintenance {
    public function __construct() {
        parent::__construct();
        $this->addDescription( 'Parse random pages and compare output to cache.' );
        $this->addOption( 'namespace', 'Page namespace number', true, true );
        $this->addOption( 'maxpages', 'Number of pages to try', true, true );
    }
 
    public function execute() {
        $pages = $this->getOption( 'maxpages' );
 
        $dbr = $this->getDB( DB_REPLICA );
 
        $totalsec = 0.0;
        $scanned = 0;
        $withcache = 0;
        $withdiff = 0;
        $parserCache = MediaWikiServices::getInstance()->getParserCache();
        $renderer = MediaWikiServices::getInstance()->getRevisionRenderer();
        while ( $pages-- > 0 ) {
            $row = $dbr->selectRow( 'page',
                // @todo Title::selectFields() or Title::getQueryInfo() or something
                [
                    'page_namespace', 'page_title', 'page_id',
                    'page_len', 'page_is_redirect', 'page_latest',
                ],
                [
                    'page_namespace' => $this->getOption( 'namespace' ),
                    'page_is_redirect' => 0,
                    'page_random >= ' . wfRandom()
                ],
                __METHOD__,
                [
                    'ORDER BY' => 'page_random',
                ]
            );
 
            if ( !$row ) {
                continue;
            }
            ++$scanned;
 
            $title = Title::newFromRow( $row );
            $page = WikiPage::factory( $title );
            $revision = $page->getRevision()->getRevisionRecord();
            $parserOptions = $page->makeParserOptions( 'canonical' );
 
            $parserOutputOld = $parserCache->get( $page, $parserOptions );
 
            if ( $parserOutputOld ) {
                $t1 = microtime( true );
                $parserOutputNew = $renderer->getRenderedRevision( $revision, $parserOptions )
                    ->getRevisionParserOutput();
 
                $sec = microtime( true ) - $t1;
                $totalsec += $sec;
 
                $this->output( "Parsed '{$title->getPrefixedText()}' in $sec seconds.\n" );
 
                $this->output( "Found cache entry found for '{$title->getPrefixedText()}'..." );
                $oldHtml = trim( preg_replace( '#<!-- .+-->#Us', '', $parserOutputOld->getText() ) );
                $newHtml = trim( preg_replace( '#<!-- .+-->#Us', '', $parserOutputNew->getText() ) );
                $diff = wfDiff( $oldHtml, $newHtml );
                if ( strlen( $diff ) ) {
                    $this->output( "differences found:\n\n$diff\n\n" );
                    ++$withdiff;
                } else {
                    $this->output( "No differences found.\n" );
                }
                ++$withcache;
            } else {
                $this->output( "No parser cache entry found for '{$title->getPrefixedText()}'.\n" );
            }
        }
 
        $ave = $totalsec ? $totalsec / $scanned : 0;
        $this->output( "Checked $scanned pages; $withcache had prior cache entries.\n" );
        $this->output( "Pages with differences found: $withdiff\n" );
        $this->output( "Average parse time: $ave sec\n" );
    }
}
 
$maintClass = CompareParserCache::class;
require_once RUN_MAINTENANCE_IF_MAIN;