From 69b8d61c4725c686bb5fb618f0399b047bfb811b Mon Sep 17 00:00:00 2001
From: Michael Peter Christen <mc@yacy.net>
Date: Tue, 12 Nov 2013 15:54:54 +0100
Subject: [PATCH] fix for search requests in GSA interface which contain
 'funny' characters (like ':' etc.)

---
 htroot/gsa/searchresult.java                |  2 +-
 source/net/yacy/search/query/QueryGoal.java | 13 ++++++++++++-
 2 files changed, 13 insertions(+), 2 deletions(-)

diff --git a/htroot/gsa/searchresult.java b/htroot/gsa/searchresult.java
index 1a145cf94..e88714076 100644
--- a/htroot/gsa/searchresult.java
+++ b/htroot/gsa/searchresult.java
@@ -163,7 +163,7 @@ public class searchresult {
                 CollectionSchema.last_modified.getSolrFieldName() + ',' +
                 CollectionSchema.size_i.getSolrFieldName());
         post.put("hl", "true");
-        post.put("hl.q", originalQuery);
+        post.put("hl.q", qg.getIncludeString());
         post.put("hl.fl", CollectionSchema.h1_txt.getSolrFieldName() + "," + CollectionSchema.h2_txt.getSolrFieldName() + "," + CollectionSchema.text_t.getSolrFieldName());
         post.put("hl.alternateField", CollectionSchema.description_txt.getSolrFieldName());
         post.put("hl.simple.pre", "<b>");
diff --git a/source/net/yacy/search/query/QueryGoal.java b/source/net/yacy/search/query/QueryGoal.java
index 50861de59..51687a581 100644
--- a/source/net/yacy/search/query/QueryGoal.java
+++ b/source/net/yacy/search/query/QueryGoal.java
@@ -45,7 +45,7 @@ public class QueryGoal {
     private static char space = ' ';
     private static char sq = '\'';
     private static char dq = '"';
-    private static String seps = ".,/&_";
+    private static String seps = ".:;#'*`,!$%()=?^<>/&_";
     
     private String query_original;
     private HandleSet include_hashes, exclude_hashes;
@@ -165,6 +165,17 @@ public class QueryGoal {
         return exclude_hashes;
     }
    
+    /**
+     * the include string may be useful (and better) for highlight/snippet computation 
+     * @return the query string containing only the positive literals (includes) and without whitespace characters
+     */
+    public String getIncludeString() {
+        if (this.include_strings.size() == 0) return "";
+        StringBuilder sb = new StringBuilder(10 * include_strings.size());
+        for (String s: this.include_strings) sb.append(s).append(' ');
+        return sb.toString().substring(0, sb.length() - 1);
+    }
+   
     public ArrayList<String> getIncludeStrings() {
         return include_strings;
     }