From: Aryeh Gregor <simetrical@users.mediawiki.org>
Date: Wed, 5 Mar 2008 03:50:12 +0000 (+0000)
Subject: Add --old-redirects-only option to maintenance/refreshLinks.php, to add old redirects... 
X-Git-Tag: 1.31.0-rc.0~49249
X-Git-Url: https://git.cyclocoop.org/%7B%24www_url%7Dadmin/compta/comptes/ajouter.php?a=commitdiff_plain;h=6b79f5995a0f6d19460537310621882d89f0902c;p=lhc%2Fweb%2Fwiklou.git

Add --old-redirects-only option to maintenance/refreshLinks.php, to add old redirects to the redirect table.  This may be worth trying out on at least some of the smaller wikis.  It seems to work correctly, and pretty quickly for a couple thousand redirects to fix, although truncating the redirect table and running it inevitably gives a different number (probably due to incorrect redirect table entries to start with).  With the limit options, it might be tried on bigger wikis too.
---

diff --git a/RELEASE-NOTES b/RELEASE-NOTES
index 7aa01ce36e..ed8418d241 100644
--- a/RELEASE-NOTES
+++ b/RELEASE-NOTES
@@ -36,7 +36,8 @@ it from source control: http://www.mediawiki.org/wiki/Download_from_SVN
 * (bug 13112) Added Special:RelatedChanges alias to Special:Recentchangeslinked
 * (bug 13130) Moved edit token and autosummary fields above edit tools to
   reduce broken form submissions
-
+* Add --old-redirects-only option to maintenance/refreshLinks.php, to add old
+  redirects to the redirect table
 
 === Bug fixes in 1.13 ===
 
diff --git a/maintenance/refreshLinks.inc b/maintenance/refreshLinks.inc
index e89db8aa03..c471d0eece 100644
--- a/maintenance/refreshLinks.inc
+++ b/maintenance/refreshLinks.inc
@@ -8,7 +8,7 @@
 define( "REPORTING_INTERVAL", 100 );
 #define( "REPORTING_INTERVAL", 1 );
 
-function refreshLinks( $start, $newOnly = false, $maxLag = false, $end = 0, $redirectsOnly = false ) {
+function refreshLinks( $start, $newOnly = false, $maxLag = false, $end = 0, $redirectsOnly = false, $oldRedirectsOnly = false ) {
 	global $wgUser, $wgParser, $wgUseImageResize, $wgUseTidy;
 
 	$fname = 'refreshLinks';
@@ -25,15 +25,36 @@ function refreshLinks( $start, $newOnly = false, $maxLag = false, $end = 0, $red
 	$wgUseImageResize = false;
 	$wgUseTidy = false;
 
-	$what = ($redirectsOnly)? "redirects" : "links";
+	$what = $redirectsOnly ? "redirects" : "links";
+
+	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"),
+			$fname
+		);
+		$num = $dbr->numRows( $res );
+		print "Refreshing $num old redirects from $start...\n";
 
-	if ( $newOnly ) {
+		while( $row = $dbr->fetchObject( $res ) ) {
+			if ( !( ++$i % REPORTING_INTERVAL ) ) {
+				print "$i\n";
+				wfWaitForSlaves( $maxLag );
+			}
+			fixRedirect( $row->page_id );
+		}
+	} elseif( $newOnly ) {
 		print "Refreshing $what from ";
 		$res = $dbr->select( 'page',
 			array( 'page_id' ),
 			array(
 				'page_is_new' => 1,
-				"page_id > $start" ),
+				"page_id >= $start" ),
 			$fname
 		);
 		$num = $dbr->numRows( $res );
diff --git a/maintenance/refreshLinks.php b/maintenance/refreshLinks.php
index 67fb77aebf..3b4780ee02 100644
--- a/maintenance/refreshLinks.php
+++ b/maintenance/refreshLinks.php
@@ -11,13 +11,20 @@ require_once( "refreshLinks.inc" );
 
 if( isset( $options['help'] ) ) {
 	echo <<<TEXT
-Usage: php refreshLinks.php [<start>] [-e <end>] [-m <maxlag>] [--help] 
-
-    --help      : This help message
-    --dfn-only  : Delete links from nonexistent articles only
-    -m <number> : Maximum replication lag
-    <start>     : First page id to refresh
-    -e <number> : Last page id to refresh
+Usage:
+    php refreshLinks.php --help
+    php refreshLinks.php [<start>] [-e <end>] [-m <maxlag>] [--dfn-only]
+                         [--new-only] [--redirects-only]
+    php refreshLinks.php [<start>] [-e <end>] [-m <maxlag>] --old-redirects-only
+
+    --help               : This help message
+    --dfn-only           : Delete links from nonexistent articles only
+    --new-only           : Only affect articles with just a single edit
+    --redirects-only     : Only fix redirects, not all links
+    --old-redirects-only : Only fix redirects with no redirect table entry
+    -m <number>          : Maximum replication lag
+    <start>              : First page id to refresh
+    -e <number>          : Last page id to refresh
 
 TEXT;
 	exit(0);
@@ -32,7 +39,7 @@ if ( !$options['dfn-only'] ) {
 		$start = 1;
 	}
 
-	refreshLinks( $start, $options['new-only'], $options['m'], $options['e'], $options['redirects-only'] );
+	refreshLinks( $start, $options['new-only'], $options['m'], $options['e'], $options['redirects-only'], $options['old-redirects-only'] );
 }
 // this bit's bad for replication: disabling temporarily
 // --brion 2005-07-16