X-Git-Url: https://git.heureux-cyclage.org/?a=blobdiff_plain;ds=sidebyside;f=maintenance%2FrefreshLinks.inc;h=48d99712afd4f6da44c0a970118d7687c2078225;hb=63a6b8e0a3740660912c6052549933ac6e1cc277;hp=feb91ef214b231d3ee0371393ec133607ae4bd61;hpb=87cdeb11e83cbface41e1b0a380243cbfb36e5b3;p=lhc%2Fweb%2Fwiklou.git

diff --git a/maintenance/refreshLinks.inc b/maintenance/refreshLinks.inc
index feb91ef214..48d99712af 100644
--- a/maintenance/refreshLinks.inc
+++ b/maintenance/refreshLinks.inc
@@ -1,117 +1,154 @@
 <?php
 /**
  * @todo document
- * @package MediaWiki
- * @subpackage Maintenance
+ * @addtogroup Maintenance
  */
 
 /** */
 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, $redirectsOnly = false, $oldRedirectsOnly = false ) {
+	global $wgUser, $wgParser, $wgUseTidy;
 
 	$fname = 'refreshLinks';
-	$dbr =& wfGetDB( DB_SLAVE );
-	$dbw =& wfGetDB( DB_MASTER );
+	$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);
 
 	# Don't generate extension images (e.g. Timeline)
-	$wgParser->mTagHooks = array();
-	
-	# Don't generate thumbnail images
-	$wgUseImageResize = false;
+	$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.
+		$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 ) {
-		print "Refreshing links from ";
+		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_id' ),
 			array(
 				'page_is_new' => 1,
-				"page_id > $start" ),
+				"page_id >= $start" ),
 			$fname
 		);
 		$num = $dbr->numRows( $res );
 		print "$num new articles...\n";
-		
+
 		$i = 0;
 		while ( $row = $dbr->fetchObject( $res ) ) {
 			if ( !( ++$i % REPORTING_INTERVAL ) ) {
 				print "$i\n";
 				wfWaitForSlaves( $maxLag );
 			}
-
-			fixLinksFromArticle( $row->page_id );
+			if($redirectsOnly)
+				fixRedirect( $row->page_id );
+			else
+				fixLinksFromArticle( $row->page_id );
 		}
 	} else {
-		print "Refreshing link table.\n";
-		$end = $dbr->selectField( 'page', 'max(page_id)', false );
+		print "Refreshing $what table.\n";
+		if ( !$end ) {
+			$end = $dbr->selectField( 'page', 'max(page_id)', false );
+		}
 		print("Starting from page_id $start of $end.\n");
 
 		for ($id = $start; $id <= $end; $id++) {
-			
+
 			if ( !($id % REPORTING_INTERVAL) ) {
 				print "$id\n";
 				wfWaitForSlaves( $maxLag );
 			}
-			fixLinksFromArticle( $id );
+			if($redirectsOnly)
+				fixRedirect( $id );
+			else
+				fixLinksFromArticle( $id );
 		}
-		
+	}
+}
+
+function fixRedirect( $id ){
+	global $wgTitle, $wgArticle;
+
+	$wgTitle = Title::newFromID( $id );
+	$dbw = wfGetDB( DB_MASTER );
 
+	if ( is_null( $wgTitle ) ) {
+		return;
 	}
+	$wgArticle = new Article($wgTitle);
+
+	$rt = $wgArticle->followRedirect();
+
+	if($rt == false || !is_object($rt))
+		return;
+
+	$wgArticle->updateRedirectOn($dbw,$rt);
 }
 
 function fixLinksFromArticle( $id ) {
-	global $wgTitle, $wgArticle, $wgLinkCache, $wgOut;
+	global $wgTitle, $wgParser;
 	
 	$wgTitle = Title::newFromID( $id );
-	$dbw =& wfGetDB( DB_MASTER );
+	$dbw = wfGetDB( DB_MASTER );
+
+	$linkCache =& LinkCache::singleton();
+	$linkCache->clear();
 	
 	if ( is_null( $wgTitle ) ) {
 		return;
 	}
 	$dbw->begin();
 
-	$wgArticle = new Article( $wgTitle );
-	$text = $wgArticle->getContent( true );
-	$wgLinkCache = new LinkCache;
-	$wgLinkCache->forUpdate( true );
-	
-	global $wgLinkHolders;
-	$wgLinkHolders = array(
-		'namespaces' => array(),
-		'dbkeys' => array(),
-		'queries' => array(),
-		'texts' => array(),
-		'titles' => array()
-	);
-
-
-	# Parse the text and replace links with placeholders
-	$wgOut->addWikiText( $text );
-	
-	# Look up the links in the DB and add them to the link cache
-	$wgOut->clearHTML();
+	$revision = Revision::newFromTitle( $wgTitle );
+	if ( !$revision ) {
+		return;
+	}
 
-	$linksUpdate = new LinksUpdate( $id, $wgTitle->getPrefixedDBkey() );
-	$linksUpdate->doDumbUpdate();
+	$options = new ParserOptions;
+	$parserOutput = $wgParser->parse( $revision->getText(), $wgTitle, $options, true, true, $revision->getId() );
+	$update = new LinksUpdate( $wgTitle, $parserOutput, false );
+	$update->doUpdate();
 	$dbw->immediateCommit();
 }
 
 function deleteLinksFromNonexistent( $maxLag = 0 ) {
 	$fname = 'deleteLinksFromNonexistent';
-	
+
 	wfWaitForSlaves( $maxLag );
 
-	$dbw =& wfGetDB( DB_WRITE );
-	
-	$linksTables = array( 
+	$dbw = wfGetDB( DB_MASTER );
+
+	$linksTables = array(
 		'pagelinks' => 'pl_from',
 		'imagelinks' => 'il_from',
 		'categorylinks' => 'cl_from',
+		'templatelinks' => 'tl_from',
+		'externallinks' => 'el_from',
 	);
 
 	$page = $dbw->tableName( 'page' );
@@ -129,7 +166,7 @@ function deleteLinksFromNonexistent( $maxLag = 0 ) {
 
 		$pTable = $dbw->tableName( $table );
 		$sql = "DELETE $pTable FROM $pTable LEFT JOIN $page ON page_id=$field WHERE page_id IS NULL";
-		
+
 		print "Deleting $table from non-existent articles...";
 		$dbw->query( $sql, $fname );
 		print " fixed " .$dbw->affectedRows() . " row(s)\n";