purgeList.php

Go to the documentation of this file.

<?php
use MediaWiki\MainConfigNames;
use MediaWiki\MediaWikiServices;
 
require_once __DIR__ . '/Maintenance.php';
 
class PurgeList extends Maintenance {
    private $namespaceId;
    private $allNamespaces;
    private $doDbTouch;
    private $delay;
 
    public function __construct() {
        parent::__construct();
        $this->addDescription( "Send purge requests for listed pages to CDN.\n"
            . "By default this expects a list of URLs or page names from STDIN. "
            . "To query the database for input, use --namespace or --all-namespaces instead."
        );
        $this->addOption( 'namespace', 'Purge pages with this namespace number', false, true );
        $this->addOption( 'all-namespaces', 'Purge pages in all namespaces', false, false );
        $this->addOption( 'db-touch', 'Update the page.page_touched database field', false, false );
        $this->addOption( 'delay', 'Number of seconds to delay between each purge', false, true );
        $this->addOption( 'verbose', 'Show more output', false, false, 'v' );
        $this->setBatchSize( 100 );
    }
 
    public function execute() {
        $this->namespaceId = $this->getOption( 'namespace' );
        $this->allNamespaces = $this->hasOption( 'all-namespaces' );
        $this->doDbTouch = $this->hasOption( 'db-touch' );
        $this->delay = intval( $this->getOption( 'delay', '0' ) );
 
        $conf = $this->getConfig();
        if ( ( $this->namespaceId !== null || $this->allNamespaces )
            && $this->doDbTouch
            && $conf->get( MainConfigNames::MiserMode )
        ) {
            $this->fatalError( 'Prevented mass db-invalidation (MiserMode is enabled).' );
        }
 
        if ( $this->allNamespaces ) {
            $this->purgeNamespace( false );
        } elseif ( $this->namespaceId !== null ) {
            $this->purgeNamespace( intval( $this->namespaceId ) );
        } else {
            $this->doPurge();
        }
        $this->output( "Done!\n" );
    }
 
    private function doPurge() {
        $stdin = $this->getStdin();
        $urls = [];
        $htmlCacheUpdater = MediaWikiServices::getInstance()->getHtmlCacheUpdater();
 
        while ( !feof( $stdin ) ) {
            $page = trim( fgets( $stdin ) );
            if ( preg_match( '%^https?://%', $page ) ) {
                $urls[] = $page;
            } elseif ( $page !== '' ) {
                $title = Title::newFromText( $page );
                if ( $title ) {
                    $newUrls = $htmlCacheUpdater->getUrls( $title );
 
                    foreach ( $newUrls as $url ) {
                        $this->output( "$url\n" );
                    }
 
                    $urls = array_merge( $urls, $newUrls );
 
                    if ( $this->doDbTouch ) {
                        $title->invalidateCache();
                    }
                } else {
                    $this->output( "(Invalid title '$page')\n" );
                }
            }
        }
        $this->output( "Purging " . count( $urls ) . " urls\n" );
        $this->sendPurgeRequest( $urls );
    }
 
    private function purgeNamespace( $namespace = false ) {
        $dbr = $this->getDB( DB_REPLICA );
        $htmlCacheUpdater = MediaWikiServices::getInstance()->getHtmlCacheUpdater();
        $startId = 0;
        if ( $namespace === false ) {
            $conds = [];
        } else {
            $conds = [ 'page_namespace' => $namespace ];
        }
        while ( true ) {
            $res = $dbr->select( 'page',
                [ 'page_id', 'page_namespace', 'page_title' ],
                $conds + [ 'page_id > ' . $dbr->addQuotes( $startId ) ],
                __METHOD__,
                [
                    'LIMIT' => $this->getBatchSize(),
                    'ORDER BY' => 'page_id'
 
                ]
            );
            if ( !$res->numRows() ) {
                break;
            }
            $urls = [];
            foreach ( $res as $row ) {
                $title = Title::makeTitle( $row->page_namespace, $row->page_title );
                $urls = array_merge( $urls, $htmlCacheUpdater->getUrls( $title ) );
                $startId = $row->page_id;
            }
            $this->sendPurgeRequest( $urls );
        }
    }
 
    private function sendPurgeRequest( $urls ) {
        $hcu = MediaWikiServices::getInstance()->getHtmlCacheUpdater();
        if ( $this->delay > 0 ) {
            foreach ( $urls as $url ) {
                if ( $this->hasOption( 'verbose' ) ) {
                    $this->output( $url . "\n" );
                }
                $hcu->purgeUrls( $url, $hcu::PURGE_NAIVE );
                sleep( $this->delay );
            }
        } else {
            if ( $this->hasOption( 'verbose' ) ) {
                $this->output( implode( "\n", $urls ) . "\n" );
            }
            $hcu->purgeUrls( $urls, $hcu::PURGE_NAIVE );
        }
    }
}
 
$maintClass = PurgeList::class;
require_once RUN_MAINTENANCE_IF_MAIN;