<?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
}
public function execute() {
+ $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 );
*/
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;
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" );
+ $i = 0;
+
foreach ( $res as $row ) {
if ( !( ++$i % $reportingInterval ) ) {
$this->output( "$i\n" );
- wfWaitForSlaves( $maxLag );
+ wfWaitForSlaves();
}
$this->fixRedirect( $row->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 );
+ wfWaitForSlaves();
}
- if ( $redirectsOnly )
+ if ( $redirectsOnly ) {
$this->fixRedirect( $row->page_id );
- else
- $this->fixLinksFromArticle( $row->page_id );
+ } else {
+ self::fixLinksFromArticle( $row->page_id );
+ }
}
} else {
if ( !$end ) {
}
$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( $maxLag );
+ wfWaitForSlaves();
}
$this->fixRedirect( $id );
}
$this->output( "Starting from page_id $start of $end.\n" );
for ( $id = $start; $id <= $end; $id++ ) {
-
+
if ( !( $id % $reportingInterval ) ) {
$this->output( "$id\n" );
- wfWaitForSlaves( $maxLag );
+ wfWaitForSlaves();
}
- $this->fixLinksFromArticle( $id );
+ self::fixLinksFromArticle( $id );
}
}
}
* @param $id int The page_id of the redirect
*/
private function fixRedirect( $id ) {
- global $wgTitle, $wgArticle;
-
- $wgTitle = Title::newFromID( $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 ) ) {
- // $wgTitle is not a redirect
+ $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 {
- $wgArticle->updateRedirectOn( $dbw, $rt );
+ $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;
}
+ $dbw->begin();
+
$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();
}
- /*
+ /**
* Removes non-existing links from pages from pagelinks, imagelinks,
- * categorylinks, templatelinks and externallinks tables.
+ * categorylinks, templatelinks, externallinks, interwikilinks, langlinks and redirect tables.
*
* @param $maxLag
* @param $batchSize The size of deletion batches
* @author Merlijn van Deen <valhallasw@arctus.nl>
*/
private function deleteLinksFromNonexistent( $maxLag = 0, $batchSize = 100 ) {
- wfWaitForSlaves( $maxLag );
+ wfWaitForSlaves();
$dbw = wfGetDB( DB_MASTER );
'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 ) {
$counter = 0;
$list = array();
$this->output( "0.." );
-
foreach ( $results as $row ) {
$counter++;
$list[] = $row->$field;
if ( ( $counter % $batchSize ) == 0 ) {
- wfWaitForSlaves( 5 );
+ wfWaitForSlaves();
$dbw->delete( $table, array( $field => $list ), __METHOD__ );
$this->output( $counter . ".." );
}
$maintClass = 'RefreshLinks';
-require_once( DO_MAINTENANCE );
+require_once( RUN_MAINTENANCE_IF_MAIN );