From 25f9c350335cfd9c221e3f72e3e8e3a0a31c1a54 Mon Sep 17 00:00:00 2001
From: Michael Peter Christen <mc@yacy.net>
Date: Fri, 27 Dec 2013 00:34:55 +0100
Subject: [PATCH] add patch which shall prevent that naive search mistakes like
 usage of regular expressions cause no results. Usage of '*' followed by a dot
 or any expression will now cause that this expression is used as a filetype
 search.

---
 htroot/yacysearch.java                        |  7 +++-
 .../net/yacy/search/query/QueryModifier.java  | 36 +++++++++++--------
 2 files changed, 28 insertions(+), 15 deletions(-)

diff --git a/htroot/yacysearch.java b/htroot/yacysearch.java
index adb16d9e5..f2a577ef3 100644
--- a/htroot/yacysearch.java
+++ b/htroot/yacysearch.java
@@ -370,7 +370,12 @@ public class yacysearch {
             
             int stp = querystring.indexOf('*');
             if (stp >= 0) {
-                querystring = querystring.substring(0, stp) + Segment.catchallString + querystring.substring(stp + 1);
+                // if the star appears as a single entry, use the catchallstring
+                if (querystring.length() == 1) {
+                    querystring = Segment.catchallString;
+                } else {
+                    querystring = querystring.replace('*', ' ').replaceAll("  ", " ");
+                }
             }
             if ( querystring.indexOf("/near", 0) >= 0 ) {
                 querystring = querystring.replace("/near", "");
diff --git a/source/net/yacy/search/query/QueryModifier.java b/source/net/yacy/search/query/QueryModifier.java
index 362ece108..d17871348 100644
--- a/source/net/yacy/search/query/QueryModifier.java
+++ b/source/net/yacy/search/query/QueryModifier.java
@@ -27,6 +27,7 @@ import org.apache.solr.common.params.MultiMapSolrParams;
 
 import net.yacy.cora.document.id.DigestURL;
 import net.yacy.cora.util.CommonPattern;
+import net.yacy.search.index.Segment;
 import net.yacy.search.schema.CollectionSchema;
 import net.yacy.server.serverObjects;
 
@@ -74,21 +75,11 @@ public class QueryModifier {
             add("/file");
         }
         
+        // parse 'common search mistakes' like guessed regular expressions
+        querystring = filetypeParser(querystring, "*");
+        
         // parse filetype
-        final int ftp = querystring.indexOf("filetype:", 0);
-        if ( ftp >= 0 ) {
-            int ftb = querystring.indexOf(' ', ftp);
-            if ( ftb == -1 ) {
-                ftb = querystring.length();
-            }
-            filetype = querystring.substring(ftp + 9, ftb);
-            querystring = querystring.replace("filetype:" + filetype, "");
-            while ( !filetype.isEmpty() && filetype.charAt(0) == '.' ) {
-                filetype = filetype.substring(1);
-            }
-            add("filetype:" + filetype);
-            if (filetype.isEmpty()) filetype = null;
-        }
+        querystring = filetypeParser(querystring, "filetype:");
         
         // parse site
         final int sp = querystring.indexOf("site:", 0);
@@ -141,6 +132,23 @@ public class QueryModifier {
         return querystring.trim();
     }
     
+    private String filetypeParser(String querystring, final String filetypePrefix) {
+        final int ftp = querystring.indexOf(filetypePrefix, 0);
+        if ( ftp >= 0 ) {
+            int ftb = querystring.indexOf(' ', ftp);
+            if ( ftb < 0 ) ftb = querystring.length();
+            filetype = querystring.substring(ftp + filetypePrefix.length(), ftb);
+            querystring = querystring.replace(filetypePrefix + filetype, "");
+            while ( !filetype.isEmpty() && filetype.charAt(0) == '.' ) {
+                filetype = filetype.substring(1);
+            }
+            add("filetype:" + filetype);
+            if (filetype.isEmpty()) filetype = null;
+            if (querystring.length() == 0) querystring = "*";
+        }
+        return querystring;
+    }
+    
     public void add(String m) {
         if (modifier.length() > 0 && modifier.charAt(modifier.length() - 1) != ' ' && m != null && m.length() > 0) modifier.append(' ');
         if (m != null) modifier.append(m);