From f88b6fd3ec20cd246cd3c984c0038183d1219e70 Mon Sep 17 00:00:00 2001
From: Tim Starling <tstarling@users.mediawiki.org>
Date: Thu, 26 Jan 2006 11:25:26 +0000
Subject: [PATCH] Don't use tidy. Command line option to specify the end ID.
 Clear the link cache after each article to keep memory usage down.

---
 maintenance/refreshLinks.inc | 19 +++++++++++++------
 maintenance/refreshLinks.php | 12 +++++++++---
 2 files changed, 22 insertions(+), 9 deletions(-)

diff --git a/maintenance/refreshLinks.inc b/maintenance/refreshLinks.inc
index 093c27192d..04125c12df 100644
--- a/maintenance/refreshLinks.inc
+++ b/maintenance/refreshLinks.inc
@@ -7,9 +7,10 @@
 
 /** */
 define( "REPORTING_INTERVAL", 100 );
+#define( "REPORTING_INTERVAL", 1 );
 
-function refreshLinks( $start, $newOnly = false, $maxLag = false ) {
-	global $wgUser, $wgParser, $wgUseImageResize;
+function refreshLinks( $start, $newOnly = false, $maxLag = false, $end = 0 ) {
+	global $wgUser, $wgParser, $wgUseImageResize, $wgUseTidy;
 
 	$fname = 'refreshLinks';
 	$dbr =& wfGetDB( DB_SLAVE );
@@ -24,6 +25,7 @@ function refreshLinks( $start, $newOnly = false, $maxLag = false ) {
 
 	# Don't generate thumbnail images
 	$wgUseImageResize = false;
+	$wgUseTidy = false;
 
 	if ( $newOnly ) {
 		print "Refreshing links from ";
@@ -48,7 +50,9 @@ function refreshLinks( $start, $newOnly = false, $maxLag = false ) {
 		}
 	} else {
 		print "Refreshing link table.\n";
-		$end = $dbr->selectField( 'page', 'max(page_id)', false );
+		if ( !$end ) {
+			$end = $dbr->selectField( 'page', 'max(page_id)', false );
+		}
 		print("Starting from page_id $start of $end.\n");
 
 		for ($id = $start; $id <= $end; $id++) {
@@ -65,11 +69,14 @@ function refreshLinks( $start, $newOnly = false, $maxLag = false ) {
 }
 
 function fixLinksFromArticle( $id ) {
-	global $wgTitle, $wgArticle, $wgOut, $wgParser;
-
+	global $wgTitle, $wgArticle, $wgOut, $wgParser, $wgLinkCache;
+	
 	$wgTitle = Title::newFromID( $id );
 	$dbw =& wfGetDB( DB_MASTER );
 
+	$linkCache =& LinkCache::singleton();
+	$linkCache->clear();
+	
 	if ( is_null( $wgTitle ) ) {
 		return;
 	}
@@ -83,7 +90,7 @@ function fixLinksFromArticle( $id ) {
 	$options = new ParserOptions;
 	$parserOutput = $wgParser->parse( $revision->getText(), $wgTitle, $options, true, true, $revision->getId() );
 	$update = new LinksUpdate( $wgTitle, $parserOutput );
-	$update->doDumbUpdate();
+	$update->doUpdate();
 	$dbw->immediateCommit();
 }
 
diff --git a/maintenance/refreshLinks.php b/maintenance/refreshLinks.php
index f37c8e874f..e59124aaf5 100644
--- a/maintenance/refreshLinks.php
+++ b/maintenance/refreshLinks.php
@@ -6,7 +6,7 @@
  */
 
 /** */
-$optionsWithArgs = array( 'm' );
+$optionsWithArgs = array( 'm', 'e' );
 require_once( "commandLine.inc" );
 require_once( "refreshLinks.inc" );
 
@@ -19,8 +19,14 @@ if ( !$options['dfn-only'] ) {
 		$start = 1;
 	}
 
-	refreshLinks( $start, $options['new-only'], $options['m'] );
+	refreshLinks( $start, $options['new-only'], $options['m'], $options['e'] );
+}
+// this bit's bad for replication: disabling temporarily
+// --brion 2005-07-16
+//deleteLinksFromNonexistent();
+
+if ( $options['globals'] ) {
+	print_r( $GLOBALS );
 }
-deleteLinksFromNonexistent();
 
 ?>
-- 
2.20.1