Fixed bug causing 'what links here' on image pages to break. Fixed bug causing proble...

diff --git a/maintenance/rebuildlinks.inc b/maintenance/rebuildlinks.inc
index 6259182..7db35a2 100644 (file)
--- a/maintenance/rebuildlinks.inc
+++ b/maintenance/rebuildlinks.inc
@@ -97,7 +97,7 @@ function rebuildLinkTables()
                $numlinks = preg_match_all( "/\\[\\[([{$tc}]+)(]|\\|)/", $text,
                  $m, PREG_PATTERN_ORDER );
 
-               $seen_links = array(); // seen links in this article
+               $seen_dbtitles = array(); // seen links (normalized and with ns, see below) 
                $titles_ready_for_insertion = array();
                $titles_needing_curdata = array();
                $titles_needing_curdata_pos = array();
@@ -105,12 +105,6 @@ function rebuildLinkTables()
 
                for ( $i = 0 ; $i < $numlinks; ++$i ) {
                        $link = $m[1][$i];
-
-                       // We're only interested in the link once per article
-                       if( isset( $seen_links[$link] ) )
-                               continue;
-                       $seen_links[$link] = 1;
-
                        if( preg_match( '/^(http|https|ftp|mailto|news):/', $m[1][$i] ) ) {
                                # an URL link; not for us!
                                continue;
@@ -119,13 +113,26 @@ function rebuildLinkTables()
                        # FIXME: Handle subpage links
                        $nt = $titleCache->get( $link );
                        if( $nt != false ){
-                               $titles_ready_for_insertion[] = $nt;                            
+                               // Only process each unique link once per page
+                               $nt_key = $nt->getDBkey() . $nt->getNamespace();
+                               if( isset( $seen_dbtitles[$nt_key] ) )
+                                       continue;
+                               $seen_dbtitles[$nt_key] = 1;
+
+                               $titles_ready_for_insertion[] = $nt;
                        } else {
                                $nt = Title::newFromText( $link );
                                if (! $nt) {
-                                       print "\nerror in '$ns:{$from_full_title}': '$link'\n";
+                                       print "\nInvalid link in page '$ns:{$from_full_title}': '$link'\n";
                                        continue;
                                }
+                               
+                               // Only process each unique link once per page
+                               $nt_key = $nt->getDBkey() . $nt->getNamespace();
+                               if( isset( $seen_dbtitles[$nt_key] ) )
+                                       continue;
+                               $seen_dbtitles[$nt_key] = 1;
+
                                if( $nt->getInterwiki() != "" ) {
                                        # Interwiki links are not stored in the link tables
                                        continue;
@@ -152,8 +159,7 @@ function rebuildLinkTables()
                        $parts = array();
                        foreach ($titles_needing_curdata as $nt ) {
                                $parts[] = " (cur_namespace = " . $nt->getNamespace() . " AND " .
-                                       "cur_title='" . wfStrencode( $nt->getDBkey() ) . "' AND ".
-                                       "cur_namespace=" . intval( $nt->getNamespace() ) . ")";
+                                       "cur_title='" . wfStrencode( $nt->getDBkey() ) . "')";
                        }
                        $sql = "SELECT cur_title, cur_id FROM cur WHERE " . implode(" OR ", $parts);
                        $res = wfQuery( $sql, DB_WRITE );
@@ -169,13 +175,15 @@ function rebuildLinkTables()
                }
 
                foreach ( $titles_ready_for_insertion as $nt ) {
-                       $dest = addslashes( $nt->getPrefixedDBkey() );
+                       $dest_noslashes = $nt->getPrefixedDBkey();
+                       $dest = addslashes( $dest_noslashes ); 
                        $dest_id = $nt->getArticleID();
                        $from = $from_full_title_with_slashes;
 
                        # print "\nLINK '$from_full_title' ($from_id) -> '$dest' ($dest_id)\n";
-                       if ( 0 == strncmp( "$ins:", $from_full_title, $inslen ) ) { 
-                               $iname = addslashes( substr( $from_full_title, $inslen ) );
+
+                       if ( 0 == strncmp( "$ins:", $dest_noslashes, $inslen ) ) { 
+                               $iname = addslashes( substr( $dest_noslashes, $inslen ) );
                                $imagelinks_inserter->insert( "('{$from}','{$iname}')" );
                        } else if ( 0 == $dest_id ) {
                                $brokenlinks_inserter->insert( "({$from_id},'{$dest}')" );