[fdiff.thirdparty.fdifflib] refactor line number tests for autojunk heuristic

moved into the autojunk block so that execution only occurs if it is required

Author: chrissimpkins

lib/fdiff/thirdparty/fdifflib.py

@@ -352,18 +352,17 @@ def __chain_b(self):
         self.bpopular = popular = set()
         n = len(b)
 
-        # Define the number of identical lines that must be included
-        # in order to consider the string "popular" for the autojunk
-        # strategy.  Edited from upstream Python difflib to decrease
-        # the number of lines as file sizes become larger
-        if n >= 20000:
-            ntest = n // 10000 + 1
-        elif n >= 2000:
-            ntest = n // 1000 + 1
-        elif n >= 200:
-            ntest = n // 100 + 1
-
         if self.autojunk and n >= 200:
+            # Define the number of identical lines that must be included
+            # in order to consider the string "popular" for the autojunk
+            # strategy.  Edited from upstream Python difflib to decrease
+            # the number of lines as file sizes become larger
+            if n >= 20000:
+                ntest = n // 10000 + 1
+            elif n >= 2000:
+                ntest = n // 1000 + 1
+            else:
+                ntest = n // 100 + 1
             for elt, idxs in b2j.items():
                 if len(idxs) > ntest:
                     popular.add(elt)