From 325ba7bfb87792b7849f39701c37f197fdc1e28e Mon Sep 17 00:00:00 2001
From: lotus <lotus@6c8d7289-2bf4-0310-a012-ef5d649a1542>
Date: Wed, 26 Nov 2008 16:41:38 +0000
Subject: [PATCH] only query words with length > 2 this is not complete, yet

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@5368 6c8d7289-2bf4-0310-a012-ef5d649a1542
---
 htroot/yacysearchitem.java                     |  2 +-
 source/de/anomic/plasma/plasmaSearchQuery.java | 17 ++++++++++++-----
 2 files changed, 13 insertions(+), 6 deletions(-)

diff --git a/htroot/yacysearchitem.java b/htroot/yacysearchitem.java
index 00d516489..6b49446b0 100644
--- a/htroot/yacysearchitem.java
+++ b/htroot/yacysearchitem.java
@@ -227,7 +227,7 @@ public class yacysearchitem {
                     ((yacyURL.probablyRootURL(result.hash())) ? ", probablyRootURL" : "") + 
                     (((wordURL = yacyURL.probablyWordURL(result.hash(), query[0])) != null) ? ", probablyWordURL=" + wordURL.toNormalform(false, true) : ""));
             final plasmaSnippetCache.TextSnippet snippet = result.textSnippet();
-            prop.put("content_snippet", (snippet == null) ? "" : snippet.getLineMarked(theQuery.queryHashes));
+            prop.put("content_snippet", (snippet == null) ? "" : snippet.getLineMarked(theQuery.fullqueryHashes));
             serverProfiling.update("SEARCH", new plasmaProfiling.searchEvent(theQuery.id(true), plasmaSearchEvent.FINALIZATION + "-" + item, 0, 0));
             
             return prop;
diff --git a/source/de/anomic/plasma/plasmaSearchQuery.java b/source/de/anomic/plasma/plasmaSearchQuery.java
index 50c40f0b8..854eccb42 100644
--- a/source/de/anomic/plasma/plasmaSearchQuery.java
+++ b/source/de/anomic/plasma/plasmaSearchQuery.java
@@ -56,7 +56,7 @@ public final class plasmaSearchQuery {
     public static final kelondroBitfield catchall_constraint = new kelondroBitfield(4, "______");
     
     public String queryString;
-    public TreeSet<String> queryHashes, excludeHashes;
+    public TreeSet<String> fullqueryHashes, queryHashes, excludeHashes;
     public int linesPerPage, offset;
     public String prefer;
     public int contentdom;
@@ -92,6 +92,7 @@ public final class plasmaSearchQuery {
     		final TreeSet<String>[] cq = cleanQuery(queryString);
     		this.queryHashes = indexWord.words2hashes(cq[0]);
     		this.excludeHashes = indexWord.words2hashes(cq[1]);
+    		this.fullqueryHashes = indexWord.words2hashes(cq[2]);
     	}
     	this.ranking = ranking;
         this.maxDistance = Integer.MAX_VALUE;
@@ -129,6 +130,7 @@ public final class plasmaSearchQuery {
 		this.queryString = queryString;
 		this.queryHashes = queryHashes;
 		this.excludeHashes = excludeHashes;
+		this.fullqueryHashes = queryHashes; //FIXME: refactor this method to get the proper hashes
 		this.ranking = ranking;
 		this.maxDistance = maxDistance;
 		this.prefer = prefer;
@@ -235,7 +237,7 @@ public final class plasmaSearchQuery {
     
     @SuppressWarnings("unchecked")
     public static TreeSet<String>[] cleanQuery(String querystring) {
-    	// returns two sets: a query set and a exclude set
+    	// returns three sets: a query set, a exclude set and a full query set
     	if ((querystring == null) || (querystring.length() == 0)) return new TreeSet[]{new TreeSet<String>(kelondroNaturalOrder.naturalComparator), new TreeSet<String>(kelondroNaturalOrder.naturalComparator)};
         
         // convert Umlaute
@@ -245,22 +247,27 @@ public final class plasmaSearchQuery {
             while ((c = querystring.indexOf(seps.charAt(i))) >= 0) { querystring = querystring.substring(0, c) + (((c + 1) < querystring.length()) ? (" " + querystring.substring(c + 1)) : ""); }
         }
         
+        String s;
         // the string is clean now, but we must generate a set out of it
         final TreeSet<String> query = new TreeSet<String>(kelondroNaturalOrder.naturalComparator);
         final TreeSet<String> exclude = new TreeSet<String>(kelondroNaturalOrder.naturalComparator);
+        final TreeSet<String> fullquery = new TreeSet<String>(kelondroNaturalOrder.naturalComparator);
         final String[] a = querystring.split(" ");
         for (int i = 0; i < a.length; i++) {
         	if (a[i].startsWith("-")) {
         		exclude.add(a[i].substring(1));
         	} else {
         		while ((c = a[i].indexOf('-')) >= 0) {
-        			query.add(a[i].substring(0, c));
+        			s = a[i].substring(0, c);
+        			if(s.length() > 2) query.add(s);
+        			fullquery.add(s);
         			a[i] = a[i].substring(c + 1);
         		}
-        		if (a[i].length() > 0) query.add(a[i]);
+        		if (a[i].length() > 2) query.add(a[i]);
+        		fullquery.add(a[i]);
         	}
         }
-        return new TreeSet[]{query, exclude};
+        return new TreeSet[]{query, exclude, fullquery};
     }
     
     public String queryString(final boolean encodeHTML) {