BUGFIX for regular expression

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@1261 6c8d7289-2bf4-0310-a012-ef5d649a1542
yacy · Dec 28, 2005 · 8eb4181 · 8eb4181
1 parent 106cd29
commit 8eb4181
Showing 1 changed file with 5 additions and 1 deletion.
diff --git a/source/de/anomic/htmlFilter/htmlFilterContentScraper.java b/source/de/anomic/htmlFilter/htmlFilterContentScraper.java
@@ -164,14 +164,18 @@ public static String urlNormalform(URL url) {
 
         // (this is different from previous normal forms where a '/' must not appear in root paths; here it must appear. Makes everything easier.)
         if (path.length() == 0 || path.charAt(0) != '/') path = "/" + path;
-
+        
         Pattern pathPattern = Pattern.compile("(/[^/\\.]+/)[.]{2}(?=/)|/\\.(?=/)|/(?=/)");
         Matcher matcher = pathPattern.matcher(path);
         while (matcher.find()) {
             path = matcher.replaceAll("");
             matcher.reset(path);
         }
 
+        while (path.startsWith("/../")) {
+            path = path.substring(3);
+        }
+
         if (defaultPort) return url.getProtocol() + "://" + url.getHost() + path;
         return url.getProtocol() + "://" + url.getHost() + ":" + url.getPort() + path;
     }