From ec24a0c85a1ec7be117ca869caa7bce7c33e3fed Mon Sep 17 00:00:00 2001
From: reger <reger18@arcor.de>
Date: Thu, 24 Mar 2016 19:26:38 +0100
Subject: [PATCH] add test case for optimized toTokens()

---
 .../cora/document/id/MultiProtocolURL.java    | 12 +++--------
 .../document/id/MultiProtocolURLTest.java     | 21 +++++++++++++++++++
 2 files changed, 24 insertions(+), 9 deletions(-)

diff --git a/source/net/yacy/cora/document/id/MultiProtocolURL.java b/source/net/yacy/cora/document/id/MultiProtocolURL.java
index 081296228..64570dd4c 100644
--- a/source/net/yacy/cora/document/id/MultiProtocolURL.java
+++ b/source/net/yacy/cora/document/id/MultiProtocolURL.java
@@ -980,8 +980,8 @@ public class MultiProtocolURL implements Serializable, Comparable<MultiProtocolU
 
     /**
      * create word tokens for parser. Find CamelCases and separate these words
-     * resulting words are not ordered by appearance, but all
-     * @return
+     * resulting words are not ordered by appearance, but all in sequence
+     * @return string with unique tokens
      */
     public static String toTokens(final String s) {
         // remove all non-character & non-number
@@ -992,14 +992,8 @@ public class MultiProtocolURL implements Serializable, Comparable<MultiProtocolU
             if ((c >= '0' && c <='9') || (c >= 'a' && c <='z') || (c >= 'A' && c <='Z')) sb.append(c); else sb.append(' ');
         }
 
-        String t = sb.toString();
-
-        // remove all double-spaces
-        int p;
-        while ((p = t.indexOf("  ",0)) >= 0) t = t.substring(0, p) + t.substring(p + 1);
-
         // split the string into tokens and add all camel-case splitting
-        final String[] u = CommonPattern.SPACES.split(t);
+        final String[] u = CommonPattern.SPACES.split(sb);
         final Set<String> token = new LinkedHashSet<String>();
         for (final String r: u) token.add(r);
         for (final String r: u) token.addAll(parseCamelCase(r));
diff --git a/test/java/net/yacy/cora/document/id/MultiProtocolURLTest.java b/test/java/net/yacy/cora/document/id/MultiProtocolURLTest.java
index b3c3abf14..9ada0285c 100644
--- a/test/java/net/yacy/cora/document/id/MultiProtocolURLTest.java
+++ b/test/java/net/yacy/cora/document/id/MultiProtocolURLTest.java
@@ -235,6 +235,27 @@ public class MultiProtocolURLTest {
             assertEquals(testurls.get(s),result);
         }
     }
+
+     /**
+     * Test of toTokens method, of class MultiProtocolURL.
+     */
+    @Test
+    public void testToTokens() {
+        // test string pairs which should generate equal results
+        String[][] testString = new String[][]{
+            {"abc", "abc "},
+            {" cde", "cde"},
+            {"  efg", "efg "},
+            {"hij hij", " hij "},
+            {"klm          mno", "klm@mno"},
+            {"abc/cde?fff", "abc\\cde-fff "} };
+        String result1, result2;
+        for (String[] s : testString) {
+            result1 = MultiProtocolURL.toTokens(s[0]);
+            result2 = MultiProtocolURL.toTokens(s[1]);
+            assertEquals("input: "+s[0]+"="+s[1],result1, result2);
+        }
+    }
 }