--- src/java/org/apache/nutch/parse/ParseOutputFormat.java.orig	2006-05-20 02:20:51.000000000 +0200
+++ src/java/org/apache/nutch/parse/ParseOutputFormat.java	2006-05-20 19:40:05.000000000 +0200
@@ -27,6 +27,7 @@
 import org.apache.nutch.net.*;
 
 import java.io.*;
+import java.net.*;
 import java.util.ArrayList;
 import java.util.logging.Logger;
 
@@ -54,6 +55,7 @@
     final float interval = job.getFloat("db.default.fetch.interval", 30f);
     final float extscore = job.getFloat("db.score.link.external", 1.0f);
     final boolean countFiltered = job.getBoolean("db.score.count.filtered", false);
+    final boolean ignoreExternalLinks = job.getBoolean("crawl.ignore.external.links", false);
     
     Path text =
       new Path(new Path(job.getOutputPath(), ParseText.DIR_NAME), name);
@@ -78,7 +80,10 @@
           throws IOException {
           
           Parse parse = (Parse)value;
-          
+          String fromUrl = key.toString();
+          String fromHost = null; 
+          String toHost = null;
+                             
           textOut.append(key, new ParseText(parse.getText()));
           
           ParseData parseData = parse.getData();
@@ -97,6 +102,16 @@
           // collect outlinks for subsequent db update
           Outlink[] links = parseData.getOutlinks();
 
+          if (ignoreExternalLinks) {
+            try {
+              fromHost = new URL(fromUrl).getHost().toLowerCase();
+            } catch (MalformedURLException e) {
+              fromHost = null;
+            }
+          } else {
+            fromHost = null;
+          }
+
           String[] toUrls = new String[links.length];
           int validCount = 0;
           for (int i = 0; i < links.length; i++) {
@@ -114,6 +129,18 @@
           // compute score contributions and adjustment to the original score
           for (int i = 0; i < toUrls.length; i++) {
             if (toUrls[i] == null) continue;
+
+            if (ignoreExternalLinks) {
+              try {
+                toHost = new URL(toUrls[i]).getHost().toLowerCase();
+              } catch (MalformedURLException e) {
+                toHost = null;
+              }
+              if (toHost == null || !toHost.equals(fromHost)) { // external links
+                continue; // skip it
+              }
+            }
+
             CrawlDatum target = new CrawlDatum(CrawlDatum.STATUS_LINKED, interval);
             UTF8 targetUrl = new UTF8(toUrls[i]);
             adjust = null;
--- conf/nutch-default.xml.orig	2006-05-20 02:20:50.000000000 +0200
+++ conf/nutch-default.xml	2006-05-20 19:46:22.000000000 +0200
@@ -330,6 +330,13 @@
   required that a local caching DNS be used.</description>
 </property>
 
+<property>
+  <name>crawl.ignore.external.links</name>
+  <value>false</value>
+  <description>If true all links where hostname changes between page and
+  link-target are ignored.</description>
+</property>
+
 <!-- fetcher properties -->
 
 <property>