Fix issue on prefer_longuer processor when matches are not separated

rebulk/processors.py

@@ -22,8 +22,10 @@ def conflict_prefer_longer(matches):
 
         for i in range(*match.span):
             conflicting_matches.update(matches.starting(i))
-            conflicting_matches.update(matches.ending(i))
+            if i != match.span[0]:
+                conflicting_matches.update(matches.ending(i))
 
+        conflicting_matches.remove(match)
         if conflicting_matches:
             # keep the match only if it's the longest
             for conflicting_match in filter(lambda match: not match.private, conflicting_matches):

rebulk/test/test_processors.py

@@ -36,3 +36,22 @@ def test_conflict_prefer_longer():
     values = [x.value for x in processed_matches]
 
     assert values == ["ijklmnopqrst"]
+
+    input_string = "123456789"
+
+    pattern = StringPattern("123", "456789")
+    matches = Matches(pattern.matches(input_string))
+
+    processed_matches = conflict_prefer_longer(matches)
+
+    values = [x.value for x in processed_matches]
+    assert values == ["123", "456789"]
+
+    pattern = StringPattern("123456", "789")
+    matches = Matches(pattern.matches(input_string))
+
+    processed_matches = conflict_prefer_longer(matches)
+
+    values = [x.value for x in processed_matches]
+    assert values == ["123456", "789"]
+