Overhaul encoding handling in cmdline even more.

Now the encoding guessed for the input file will be used for an output file.

We now always read and write to the terminal .buffer on Python 3, which allows
us to override the terminal encoding and use our guessing algorithm.

1 files changed, 1 insertions, 1 deletions

diff --git a/pygments/lexer.py b/pygments/lexer.py
index d93cb284..5b3ad358 100644
--- a/pygments/lexer.py
+++ b/pygments/lexer.py
@@ -149,7 +149,7 @@ class Lexer(object):
         """
         if not isinstance(text, text_type):
             if self.encoding == 'guess':
-                text = guess_decode(text)
+                text, _ = guess_decode(text)
             elif self.encoding == 'chardet':
                 try:
                     import chardet