purgeChangedPages.php

Go to the documentation of this file.

<?php
require_once __DIR__ . '/Maintenance.php';
 
use MediaWiki\MediaWikiServices;
use MediaWiki\Title\Title;
use Wikimedia\Rdbms\IResultWrapper;
 
class PurgeChangedPages extends Maintenance {
 
    public function __construct() {
        parent::__construct();
        $this->addDescription( 'Send purge requests for edits in date range to squid/varnish' );
        $this->addOption( 'starttime', 'Starting timestamp', true, true );
        $this->addOption( 'endtime', 'Ending timestamp', true, true );
        $this->addOption( 'htcp-dest', 'HTCP announcement destination (IP:port)', false, true );
        $this->addOption( 'sleep-per-batch', 'Milliseconds to sleep between batches', false, true );
        $this->addOption( 'dry-run', 'Do not send purge requests' );
        $this->addOption( 'verbose', 'Show more output', false, false, 'v' );
        $this->setBatchSize( 100 );
    }
 
    public function execute() {
        global $wgHTCPRouting;
 
        if ( $this->hasOption( 'htcp-dest' ) ) {
            $parts = explode( ':', $this->getOption( 'htcp-dest' ), 2 );
            if ( count( $parts ) < 2 ) {
                // Add default htcp port
                $parts[] = '4827';
            }
 
            // Route all HTCP messages to provided host:port
            $wgHTCPRouting = [
                '' => [ 'host' => $parts[0], 'port' => $parts[1] ],
            ];
            if ( $this->hasOption( 'verbose' ) ) {
                $this->output( "HTCP broadcasts to {$parts[0]}:{$parts[1]}\n" );
            }
        }
 
        $dbr = $this->getDB( DB_REPLICA );
        $minTime = $dbr->timestamp( $this->getOption( 'starttime' ) );
        $maxTime = $dbr->timestamp( $this->getOption( 'endtime' ) );
 
        if ( $maxTime < $minTime ) {
            $this->error( "\nERROR: starttime after endtime\n" );
            $this->maybeHelp( true );
        }
 
        $stuckCount = 0;
        while ( true ) {
            // Adjust bach size if we are stuck in a second that had many changes
            $bSize = ( $stuckCount + 1 ) * $this->getBatchSize();
 
            $res = $dbr->select(
                [ 'page', 'revision' ],
                [
                    'rev_timestamp',
                    'page_namespace',
                    'page_title',
                ],
                [
                    "rev_timestamp > " . $dbr->addQuotes( $minTime ),
                    "rev_timestamp <= " . $dbr->addQuotes( $maxTime ),
                    // Only get rows where the revision is the latest for the page.
                    // Other revisions would be duplicate and we don't need to purge if
                    // there has been an edit after the interesting time window.
                    "page_latest = rev_id",
                ],
                __METHOD__,
                [ 'ORDER BY' => 'rev_timestamp', 'LIMIT' => $bSize ],
                [
                    'page' => [ 'JOIN', 'rev_page=page_id' ],
                ]
            );
 
            if ( !$res->numRows() ) {
                // nothing more found so we are done
                break;
            }
 
            // Kludge to not get stuck in loops for batches with the same timestamp
            [ $rows, $lastTime ] = $this->pageableSortedRows( $res, 'rev_timestamp', $bSize );
            if ( !count( $rows ) ) {
                ++$stuckCount;
                continue;
            }
            // Reset suck counter
            $stuckCount = 0;
 
            $this->output( "Processing changes from {$minTime} to {$lastTime}.\n" );
 
            // Advance past the last row next time
            $minTime = $lastTime;
 
            // Create list of URLs from page_namespace + page_title
            $urls = [];
            foreach ( $rows as $row ) {
                $title = Title::makeTitle( $row->page_namespace, $row->page_title );
                $urls[] = $title->getInternalURL();
            }
 
            if ( $this->hasOption( 'dry-run' ) || $this->hasOption( 'verbose' ) ) {
                $this->output( implode( "\n", $urls ) . "\n" );
                if ( $this->hasOption( 'dry-run' ) ) {
                    continue;
                }
            }
 
            // Send batch of purge requests out to CDN servers
            $hcu = MediaWikiServices::getInstance()->getHtmlCacheUpdater();
            $hcu->purgeUrls( $urls, $hcu::PURGE_NAIVE );
 
            if ( $this->hasOption( 'sleep-per-batch' ) ) {
                // sleep-per-batch is milliseconds, usleep wants micro seconds.
                usleep( 1000 * (int)$this->getOption( 'sleep-per-batch' ) );
            }
        }
 
        $this->output( "Done!\n" );
    }
 
    protected function pageableSortedRows( IResultWrapper $res, $column, $limit ) {
        $rows = iterator_to_array( $res, false );
 
        // Nothing to do
        if ( !$rows ) {
            return [ [], null ];
        }
 
        $lastValue = end( $rows )->$column;
        if ( count( $rows ) < $limit ) {
            return [ $rows, $lastValue ];
        }
 
        for ( $i = count( $rows ) - 1; $i >= 0; --$i ) {
            if ( $rows[$i]->$column !== $lastValue ) {
                break;
            }
 
            unset( $rows[$i] );
        }
 
        // No more rows left
        if ( !$rows ) {
            return [ [], null ];
        }
 
        return [ $rows, end( $rows )->$column ];
    }
}
 
$maintClass = PurgeChangedPages::class;
require_once RUN_MAINTENANCE_IF_MAIN;