Avoid stopwords as topwords, configurable

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@3454 6c8d7289-2bf4-0310-a012-ef5d649a1542

htroot/yacysearch.java

@@ -367,6 +367,13 @@ public class yacysearch {
                             kelondroMSetTools.excludeDestructive(topwords, plasmaSwitchboard.badwords);
                         }
 
+						//avoid stopwords being topwords
+                        if (env.getConfig("filterOutStopwordsFromTopwords", "true").equals("true")) {
+                        if ((plasmaSwitchboard.stopwords != null) && (plasmaSwitchboard.stopwords.size() > 0)) {
+                            kelondroMSetTools.excludeDestructive(topwords, plasmaSwitchboard.stopwords);
+                        	}
+                        }
+						
                         String word;
                         hintcount = 0;
                         final Iterator iter = topwords.iterator();

yacy.init

@@ -462,6 +462,11 @@ xsstopw=true
 xdstopw=true
 xpstopw=true
 
+# Topwords filtering
+# If set to true, all stopwords (stopwords.yacy) are filtered from the topwords
+# Change to false if requesting hits from peers with modified stopwords-file and using the unchanged client-version
+filterOutStopwordsFromTopwords=true
+
 # performance-settings
 # delay-times for permanent loops (milliseconds)
 # the idlesleep is the pause that an proces sleeps if the last call to the