Maintenance script fixes

[lhc/web/wiklou.git] / maintenance / rebuildlinks.inc

diff --git a/maintenance/rebuildlinks.inc b/maintenance/rebuildlinks.inc
index 047476e..6a6bc26 100644 (file)
--- a/maintenance/rebuildlinks.inc
+++ b/maintenance/rebuildlinks.inc
@@ -5,9 +5,12 @@
 # See rebuildlinks.php, for example.
 #
 
+# Turn this on if you've got memory to burn
+$wgUseMemoryTables = false;
+
 function rebuildLinkTablesPass1()
 {
-       global $wgLang;
+       global $wgLang, $wgUseMemoryTables;
        $count = 0;
        print "Rebuilding link tables (pass 1).\n";
 
@@ -18,10 +21,11 @@ function rebuildLinkTablesPass1()
   rl_f_id int(8) unsigned NOT NULL default 0,
   rl_f_title varchar(255) binary NOT NULL default '',
   rl_to varchar(255) binary NOT NULL default '',
-  INDEX rl_to (rl_to) ) TYPE=MyISAM";
+  INDEX rl_to (rl_to) )";
+       if( $wgUseMemoryTables ) $sql .= " TYPE=heap";
        wfQuery( $sql, DB_WRITE );
 
-       $sql = "LOCK TABLES cur READ, rebuildlinks WRITE, interwiki READ";
+       $sql = "LOCK TABLES cur READ, rebuildlinks WRITE, interwiki READ, user_newtalk READ";
        wfQuery( $sql, DB_WRITE );
 
        $sql = "DELETE FROM rebuildlinks";
@@ -48,6 +52,11 @@ function rebuildLinkTablesPass1()
                        $first = true;
                        $sql = "INSERT INTO rebuildlinks (rl_f_id,rl_f_title,rl_to) VALUES ";
                        for ( $i = 0; $i < $numlinks; ++$i ) {
+                               if( preg_match( '/^(http|https|ftp|mailto|news):/', $m[1][$i] ) ) {
+                                       # an URL link; not for us!
+                                       continue;
+                               }
+                               # FIXME: Handle subpage links
                                $nt = Title::newFromText( $m[1][$i] );
                                if (! $nt)
                                {
@@ -55,6 +64,18 @@ function rebuildLinkTablesPass1()
                                        print "error in '$ns:{$row->cur_title}' :\t'$txt'\n";
                                        continue;
                                }
+                               if( $nt->getInterwiki() != "" ) {
+                                       # Interwiki links are not stored in the link tables
+                                       continue;
+                               }
+                               if( $nt->getNamespace() == Namespace::getSpecial() ) {
+                                       # Special links not stored in link tables
+                                       continue;
+                               }
+                               if( $nt->getNamespace() == Namespace::getMedia() ) {
+                                       # treat media: links as image: links
+                                       $nt = Title::makeTitle( Namespace::getImage(), $nt->getDBkey() );
+                               }
 
                                if (!$first)
                                        $sql .= ",";
@@ -138,7 +159,7 @@ function rebuildLinkTablesPass2()
                $to = addslashes( $row->rl_to );
 
                if ( 0 == $id ) {
-                       $sql = "SELECT rl_f_id FROM rebuildlinks WHERE rl_to='{$to}'";
+                       $sql = "SELECT DISTINCT rl_f_id FROM rebuildlinks WHERE rl_to='{$to}'";
                        $res2 = wfQuery( $sql, DB_WRITE );
 
                        $sql = "INSERT INTO brokenlinks (bl_from,bl_to) VALUES ";
@@ -156,7 +177,7 @@ function rebuildLinkTablesPass2()
                        wfFreeResult( $res2 );
                        if ( ! $first ) { wfQuery( $sql, DB_WRITE ); }
                } else {
-                       $sql = "SELECT rl_f_title FROM rebuildlinks WHERE rl_to='{$to}'";
+                       $sql = "SELECT DISTINCT rl_f_title FROM rebuildlinks WHERE rl_to='{$to}'";
                        $res2 = wfQuery( $sql, DB_WRITE );
 
                        $sql = "INSERT INTO links (l_from,l_to) VALUES ";