<?php
/**
+ * Refresh link tables.
+ *
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* @ingroup Maintenance
*/
-require_once( dirname(__FILE__) . '/Maintenance.php' );
+require_once( dirname( __FILE__ ) . '/Maintenance.php' );
class RefreshLinks extends Maintenance {
public function __construct() {
$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' );
+ $this->addArg( 'start', 'Page_id to start from, default 1', false );
$this->setBatchSize( 100 );
}
public function execute() {
- if( !$this->hasOption( 'dfn-only' ) ) {
+ $max = $this->getOption( 'm', 0 );
+ if ( !$this->hasOption( 'dfn-only' ) ) {
$start = $this->getArg( 0, 1 );
$new = $this->getOption( 'new-only', false );
- $max = $this->getOption( 'm', false );
$end = $this->getOption( 'e', 0 );
$redir = $this->getOption( 'redirects-only', false );
$oldRedir = $this->getOption( 'old-redirects-only', false );
* @param $redirectsOnly bool Only fix redirects
* @param $oldRedirectsOnly bool Only fix redirects without redirect entries
*/
- private function doRefreshLinks( $start, $newOnly = false, $maxLag = false,
+ private function doRefreshLinks( $start, $newOnly = false, $maxLag = false,
$end = 0, $redirectsOnly = false, $oldRedirectsOnly = false ) {
- global $wgUser, $wgParser, $wgUseTidy;
+ global $wgParser, $wgUseTidy;
$reportingInterval = 100;
$dbr = wfGetDB( DB_SLAVE );
$start = intval( $start );
- # Don't generate TeX PNGs (lack of a sensible current directory causes errors anyway)
- $wgUser->setOption('math', MW_MATH_SOURCE);
+ // Give extensions a chance to optimize settings
+ wfRunHooks( 'MaintenanceRefreshLinksInit', array( $this ) );
# Don't generate extension images (e.g. Timeline)
- if( method_exists( $wgParser, "clearTagHooks" ) ) {
- $wgParser->clearTagHooks();
- }
+ $wgParser->clearTagHooks();
# Don't use HTML tidy
$wgUseTidy = false;
$what = $redirectsOnly ? "redirects" : "links";
- if( $oldRedirectsOnly ) {
+ if ( $oldRedirectsOnly ) {
# This entire code path is cut-and-pasted from below. Hurrah.
- $res = $dbr->query(
- "SELECT page_id ".
- "FROM page ".
- "LEFT JOIN redirect ON page_id=rd_from ".
- "WHERE page_is_redirect=1 AND rd_from IS NULL AND ".
- ($end == 0 ? "page_id >= $start"
- : "page_id BETWEEN $start AND $end"),
- __METHOD__
+
+ $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" );
- foreach( $res as $row ) {
+ $i = 0;
+
+ foreach ( $res as $row ) {
if ( !( ++$i % $reportingInterval ) ) {
$this->output( "$i\n" );
wfWaitForSlaves( $maxLag );
}
$this->fixRedirect( $row->page_id );
}
- } elseif( $newOnly ) {
+ } elseif ( $newOnly ) {
$this->output( "Refreshing $what from " );
$res = $dbr->select( 'page',
array( 'page_id' ),
);
$num = $dbr->numRows( $res );
$this->output( "$num new articles...\n" );
-
+
$i = 0;
foreach ( $res as $row ) {
if ( !( ++$i % $reportingInterval ) ) {
$this->output( "$i\n" );
wfWaitForSlaves( $maxLag );
}
- if($redirectsOnly)
+ if ( $redirectsOnly ) {
$this->fixRedirect( $row->page_id );
- else
- $this->fixLinksFromArticle( $row->page_id );
+ } else {
+ self::fixLinksFromArticle( $row->page_id );
+ }
}
} else {
if ( !$end ) {
- $end = $dbr->selectField( 'page', 'max(page_id)', false );
+ $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) ) {
+
+ for ( $id = $start; $id <= $end; $id++ ) {
+
+ if ( !( $id % $reportingInterval ) ) {
$this->output( "$id\n" );
wfWaitForSlaves( $maxLag );
}
$this->fixRedirect( $id );
}
- if(!$redirectsOnly) {
+ 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) ) {
+ for ( $id = $start; $id <= $end; $id++ ) {
+
+ if ( !( $id % $reportingInterval ) ) {
$this->output( "$id\n" );
wfWaitForSlaves( $maxLag );
}
- $this->fixLinksFromArticle( $id );
+ self::fixLinksFromArticle( $id );
}
}
}
* Update the redirect entry for a given page
* @param $id int The page_id of the redirect
*/
- private function fixRedirect( $id ){
- global $wgTitle, $wgArticle;
-
- $wgTitle = Title::newFromID( $id );
+ private function fixRedirect( $id ) {
+ $title = Title::newFromID( $id );
$dbw = wfGetDB( DB_MASTER );
-
- if ( is_null( $wgTitle ) ) {
+
+ 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;
}
- $wgArticle = new Article($wgTitle);
-
- $rt = $wgArticle->followRedirect();
-
- if($rt == false || !is_object($rt))
- return;
-
- $wgArticle->updateRedirectOn($dbw,$rt);
+ $article = new Article( $title );
+
+ $rt = $article->followRedirect();
+
+ if ( !$rt || !is_object( $rt ) ) {
+ // $title is not a redirect
+ // Delete any redirect table entry for it
+ $dbw->delete( 'redirect', array( 'rd_from' => $id ),
+ __METHOD__ );
+ } else {
+ $article->updateRedirectOn( $dbw, $rt );
+ }
}
/**
* Run LinksUpdate for all links on a given page_id
* @param $id int The page_id
*/
- private function fixLinksFromArticle( $id ) {
- global $wgTitle, $wgParser;
+ public static function fixLinksFromArticle( $id ) {
+ global $wgParser;
- $wgTitle = Title::newFromID( $id );
+ $title = Title::newFromID( $id );
$dbw = wfGetDB( DB_MASTER );
- $linkCache =& LinkCache::singleton();
- $linkCache->clear();
+ LinkCache::singleton()->clear();
- if ( is_null( $wgTitle ) ) {
+ if ( is_null( $title ) ) {
return;
}
$dbw->begin();
- $revision = Revision::newFromTitle( $wgTitle );
+ $revision = Revision::newFromTitle( $title );
if ( !$revision ) {
return;
}
$options = new ParserOptions;
- $parserOutput = $wgParser->parse( $revision->getText(), $wgTitle, $options, true, true, $revision->getId() );
- $update = new LinksUpdate( $wgTitle, $parserOutput, false );
+ $parserOutput = $wgParser->parse( $revision->getText(), $title, $options, true, true, $revision->getId() );
+ $update = new LinksUpdate( $title, $parserOutput, false );
$update->doUpdate();
$dbw->commit();
}
// SELECT DISTINCT( $field ) FROM $table LEFT JOIN page ON $field=page_id WHERE page_id IS NULL;
$results = $dbr->select( array( $table, 'page' ),
$field,
- array('page_id' => null ),
+ array( 'page_id' => null ),
__METHOD__,
'DISTINCT',
- array( 'page' => array( 'LEFT JOIN', "$field=page_id"))
+ array( 'page' => array( 'LEFT JOIN', "$field=page_id" ) )
);
$counter = 0;
$list = array();
$this->output( "0.." );
- foreach( $results as $row ) {
+ foreach ( $results as $row ) {
$counter++;
$list[] = $row->$field;
if ( ( $counter % $batchSize ) == 0 ) {
- wfWaitForSlaves(5);
+ wfWaitForSlaves( 5 );
$dbw->delete( $table, array( $field => $list ), __METHOD__ );
$this->output( $counter . ".." );
}
}
$this->output( $counter );
- if (count($list) > 0) {
+ if ( count( $list ) > 0 ) {
$dbw->delete( $table, array( $field => $list ), __METHOD__ );
}
$this->output( "\n" );
}
$maintClass = 'RefreshLinks';
-require_once( DO_MAINTENANCE );
+require_once( RUN_MAINTENANCE_IF_MAIN );