refreshLinks.php

Go to the documentation of this file.

<?php
require_once( dirname( __FILE__ ) . '/Maintenance.php' );

class RefreshLinks extends Maintenance {
        public function __construct() {
                parent::__construct();
                $this->mDescription = "Refresh link tables";
                $this->addOption( 'dfn-only', 'Delete links from nonexistent articles only' );
                $this->addOption( 'new-only', 'Only affect articles with just a single edit' );
                $this->addOption( 'redirects-only', 'Only fix redirects, not all links' );
                $this->addOption( 'old-redirects-only', 'Only fix redirects with no redirect table entry' );
                $this->addOption( 'm', 'Maximum replication lag', false, true );
                $this->addOption( 'e', 'Last page id to refresh', false, true );
                $this->addArg( 'start', 'Page_id to start from, default 1', false );
                $this->setBatchSize( 100 );
        }

        public function execute() {
                $max = $this->getOption( 'm', 0 );
                if ( !$this->hasOption( 'dfn-only' ) ) {
                        $start = $this->getArg( 0, 1 );
                        $new = $this->getOption( 'new-only', false );
                        $end = $this->getOption( 'e', 0 );
                        $redir = $this->getOption( 'redirects-only', false );
                        $oldRedir = $this->getOption( 'old-redirects-only', false );
                        $this->doRefreshLinks( $start, $new, $max, $end, $redir, $oldRedir );
                }
                $this->deleteLinksFromNonexistent( $max, $this->mBatchSize );
        }

        private function doRefreshLinks( $start, $newOnly = false, $maxLag = false,
                                                $end = 0, $redirectsOnly = false, $oldRedirectsOnly = false ) {
                global $wgParser, $wgUseTidy;

                $reportingInterval = 100;
                $dbr = wfGetDB( DB_SLAVE );
                $start = intval( $start );

                // Give extensions a chance to optimize settings
                wfRunHooks( 'MaintenanceRefreshLinksInit', array( $this ) );

                # Don't generate extension images (e.g. Timeline)
                $wgParser->clearTagHooks();

                # Don't use HTML tidy
                $wgUseTidy = false;

                $what = $redirectsOnly ? "redirects" : "links";

                if ( $oldRedirectsOnly ) {
                        # This entire code path is cut-and-pasted from below.  Hurrah.

                        $conds = array(
                                "page_is_redirect=1",
                                "rd_from IS NULL"
                        );

                        if ( $end == 0 ) {
                                $conds[] = "page_id >= $start";
                        } else {
                                $conds[] = "page_id BETWEEN $start AND $end";
                        }

                        $res = $dbr->select(
                                array( 'page', 'redirect' ),
                                'page_id',
                                $conds,
                                __METHOD__,
                                array(),
                                array( 'redirect' => array( "LEFT JOIN", "page_id=rd_from" ) )
                        );
                        $num = $dbr->numRows( $res );
                        $this->output( "Refreshing $num old redirects from $start...\n" );

                        $i = 0;

                        foreach ( $res as $row ) {
                                if ( !( ++$i % $reportingInterval ) ) {
                                        $this->output( "$i\n" );
                                        wfWaitForSlaves();
                                }
                                $this->fixRedirect( $row->page_id );
                        }
                } elseif ( $newOnly ) {
                        $this->output( "Refreshing $what from " );
                        $res = $dbr->select( 'page',
                                array( 'page_id' ),
                                array(
                                        'page_is_new' => 1,
                                        "page_id >= $start" ),
                                __METHOD__
                        );
                        $num = $dbr->numRows( $res );
                        $this->output( "$num new articles...\n" );

                        $i = 0;
                        foreach ( $res as $row ) {
                                if ( !( ++$i % $reportingInterval ) ) {
                                        $this->output( "$i\n" );
                                        wfWaitForSlaves();
                                }
                                if ( $redirectsOnly ) {
                                        $this->fixRedirect( $row->page_id );
                                } else {
                                        self::fixLinksFromArticle( $row->page_id );
                                }
                        }
                } else {
                        if ( !$end ) {
                                $maxPage = $dbr->selectField( 'page', 'max(page_id)', false );
                                $maxRD = $dbr->selectField( 'redirect', 'max(rd_from)', false );
                                $end = max( $maxPage, $maxRD );
                        }
                        $this->output( "Refreshing redirects table.\n" );
                        $this->output( "Starting from page_id $start of $end.\n" );

                        for ( $id = $start; $id <= $end; $id++ ) {

                                if ( !( $id % $reportingInterval ) ) {
                                        $this->output( "$id\n" );
                                        wfWaitForSlaves();
                                }
                                $this->fixRedirect( $id );
                        }

                        if ( !$redirectsOnly ) {
                                $this->output( "Refreshing links table.\n" );
                                $this->output( "Starting from page_id $start of $end.\n" );

                                for ( $id = $start; $id <= $end; $id++ ) {

                                        if ( !( $id % $reportingInterval ) ) {
                                                $this->output( "$id\n" );
                                                wfWaitForSlaves();
                                        }
                                        self::fixLinksFromArticle( $id );
                                }
                        }
                }
        }

        private function fixRedirect( $id ) {
                $title = Title::newFromID( $id );
                $dbw = wfGetDB( DB_MASTER );

                if ( is_null( $title ) ) {
                        // This page doesn't exist (any more)
                        // Delete any redirect table entry for it
                        $dbw->delete( 'redirect', array( 'rd_from' => $id ),
                                __METHOD__ );
                        return;
                }

                $page = WikiPage::factory( $title );
                $rt = $page->getRedirectTarget();

                if ( $rt === null ) {
                        // $title is not a redirect
                        // Delete any redirect table entry for it
                        $dbw->delete( 'redirect', array( 'rd_from' => $id ),
                                __METHOD__ );
                }
        }

        public static function fixLinksFromArticle( $id ) {
                global $wgParser;

                $title = Title::newFromID( $id );
                $dbw = wfGetDB( DB_MASTER );

                LinkCache::singleton()->clear();

                if ( is_null( $title ) ) {
                        return;
                }

                $revision = Revision::newFromTitle( $title );
                if ( !$revision ) {
                        return;
                }

                $dbw->begin();

                $options = new ParserOptions;
                $parserOutput = $wgParser->parse( $revision->getText(), $title, $options, true, true, $revision->getId() );
                $update = new LinksUpdate( $title, $parserOutput, false );
                $update->doUpdate();
                $dbw->commit();
        }

        private function deleteLinksFromNonexistent( $maxLag = 0, $batchSize = 100 ) {
                wfWaitForSlaves();

                $dbw = wfGetDB( DB_MASTER );

                $lb = wfGetLBFactory()->newMainLB();
                $dbr = $lb->getConnection( DB_SLAVE );
                $dbr->bufferResults( false );

                $linksTables = array( // table name => page_id field
                        'pagelinks' => 'pl_from',
                        'imagelinks' => 'il_from',
                        'categorylinks' => 'cl_from',
                        'templatelinks' => 'tl_from',
                        'externallinks' => 'el_from',
                        'iwlinks' => 'iwl_from',
                        'langlinks' => 'll_from',
                        'redirect' => 'rd_from',
                        'page_props' => 'pp_page',
                );

                foreach ( $linksTables as $table => $field ) {
                        $this->output( "Retrieving illegal entries from $table... " );

                        // SELECT DISTINCT( $field ) FROM $table LEFT JOIN page ON $field=page_id WHERE page_id IS NULL;
                        $results = $dbr->select( array( $table, 'page' ),
                                                  $field,
                                                  array( 'page_id' => null ),
                                                  __METHOD__,
                                                  'DISTINCT',
                                                  array( 'page' => array( 'LEFT JOIN', "$field=page_id" ) )
                        );

                        $counter = 0;
                        $list = array();
                        $this->output( "0.." );
                        foreach ( $results as $row ) {
                                $counter++;
                                $list[] = $row->$field;
                                if ( ( $counter % $batchSize ) == 0 ) {
                                        wfWaitForSlaves();
                                        $dbw->delete( $table, array( $field => $list ), __METHOD__ );

                                        $this->output( $counter . ".." );
                                        $list = array();
                                }
                        }
                        $this->output( $counter );
                        if ( count( $list ) > 0 ) {
                                $dbw->delete( $table, array( $field => $list ), __METHOD__ );
                        }
                        $this->output( "\n" );
                }
                $lb->closeAll();
        }
}

$maintClass = 'RefreshLinks';
require_once( RUN_MAINTENANCE_IF_MAIN );