Changed find_lexer_class_for_filename() to use the lexer name as secondary criteria when sorting for rating. This ensures that even when multiple lexers return the same rating the same lexer wins. Before the lexer with the lowest hashcode in the LEXERS dictionary won, which could return different results between repeated invocations of Python. This also fixes issue #1284.

1 files changed, 2 insertions, 2 deletions

diff --git a/pygments/lexers/__init__.py b/pygments/lexers/__init__.py
index d64f163f..0bd799e1 100644
--- a/pygments/lexers/__init__.py
+++ b/pygments/lexers/__init__.py
@@ -149,8 +149,8 @@ def find_lexer_class_for_filename(_fn, code=None):
         # gets turned into 0.0.  Run scripts/detect_missing_analyse_text.py
         # to find lexers which need it overridden.
         if code:
-            return cls.analyse_text(code) + bonus
-        return cls.priority + bonus
+            return cls.analyse_text(code) + bonus, cls.__name__
+        return cls.priority + bonus, cls.__name__
 
     if matches:
         matches.sort(key=get_rating)