bugfix for http://www.yacy-forum.de/viewtopic.php?t=1706

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@1300 6c8d7289-2bf4-0310-a012-ef5d649a1542
yacy · Jan 9, 2006 · 3abd843 · 3abd843
1 parent c3284c2
commit 3abd843
Showing 1 changed file with 5 additions and 2 deletions.
diff --git a/source/de/anomic/htmlFilter/htmlFilterContentScraper.java b/source/de/anomic/htmlFilter/htmlFilterContentScraper.java
@@ -160,18 +160,21 @@ public static String urlNormalform(URL url) {
         } else if (url.getProtocol().equals("https")) {
             if (url.getPort() < 0 || url.getPort() == 443) { defaultPort = true; }
         }
-        String path = url.getFile();
+        String path = url.getPath();
 
         // (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);
         }
 
+        String query = url.getQuery().replaceAll("[\"\\/:*?<>|]", "_");
+        if (query != null) { path = path.concat("_").concat(query); }
+
         if (defaultPort) return url.getProtocol() + "://" + url.getHost() + path;
         return url.getProtocol() + "://" + url.getHost() + ":" + url.getPort() + path;
     }