Some fiddling with the grouping algorithm

Author: Bytekeeper

src/main/java/org/stt/text/CommonPrefixGrouper.java

@@ -42,18 +42,29 @@ public List<Group> getGroupsOf(String text) {
         int i = 0;
         int start = 0;
         while (i < n && node != null) {
-            char lastChar = (char) -1;
-            while (i < n && node != null && (node.numChildren() <= 1 || i < start + 3)) {
-                lastChar = chars[i];
-                node = node.child(lastChar);
+            int lastGood = i;
+            do {
+                if (!Character.isWhitespace(chars[i])) {
+                    lastGood = i;
+                }
+                node = node.child(chars[i]);
                 i++;
+            } while (i < n && node != null && node.numChildren() <= 1);
+            do {
+                if (lastGood >= i && node != null) {
+                    node = node.child(chars[i]);
+                    i++;
+                }
+                lastGood++;
             }
-            while (i < n && node != null && lastChar != ' ') {
-                lastChar = chars[i];
-                node = node.child(lastChar);
+            while (lastGood < n && (!Character.isWhitespace(chars[lastGood]) || lastGood - start < 3));
+            groups.add(new Group(Type.MATCH, text.substring(start, lastGood), new IntRange(start, lastGood)));
+            while (i < n && Character.isWhitespace(chars[i])) {
+                if (node != null) {
+                    node = node.child(chars[i]);
+                }
                 i++;
             }
-            groups.add(new Group(Type.MATCH, text.substring(start, i), new IntRange(start, i)));
             start = i;
         }
         if (i < n) {

src/test/java/org/stt/text/CommonPrefixGrouperTest.java

@@ -99,7 +99,7 @@ public void shouldFindGroupsWithSpaces() {
         List<String> result = groupsAsString(firstComment);
 
 		// THEN
-        assertThat(result, is(Arrays.asList("group subgroup ", "one")));
+        assertThat(result, is(Arrays.asList("group subgroup", "one")));
 
 	}
 
@@ -117,8 +117,8 @@ public void shouldFindSubGroups() {
 
 		// THEN
 		assertThat(withThreeGroups,
-                is(Arrays.asList("group subgroup ", "one")));
-        assertThat(withTwoGroups, is(Arrays.asList("group subgroup2 ", "one")));
+                is(Arrays.asList("group subgroup", "one")));
+        assertThat(withTwoGroups, is(Arrays.asList("group subgroup2", "one")));
     }
 
 	@Test
@@ -131,7 +131,7 @@ public void shouldFindLongestCommonPrefix() {
         List<String> groups = groupsAsString(firstComment);
 
 		// THEN
-        assertThat(groups, is(Arrays.asList("group ", "one")));
+        assertThat(groups, is(Arrays.asList("group", "one")));
 
 	}
 
@@ -161,7 +161,7 @@ public void shouldCutGroupAtShorterItem()
         List<String> result = groupsAsString("aaaa bbbb cccc dddd");
 
 		// THEN
-        assertThat(result, is(Arrays.asList("aaaa ", "bbbb ", "cccc ", "dddd")));
+        assertThat(result, is(Arrays.asList("aaaa", "bbbb", "cccc", "dddd")));
     }
 
     private void givenReaderReturnsItemsWithComment(