Made a more sophisticated encoding heuristic that runs at module load.

1 files changed, 28 insertions, 1 deletions

diff --git a/setuptools/svn_utils.py b/setuptools/svn_utils.py
index 224d11ea..98a1ff2e 100644
--- a/setuptools/svn_utils.py
+++ b/setuptools/svn_utils.py
@@ -5,6 +5,7 @@ from distutils import log
 import xml.dom.pulldom

 import shlex

 import locale

+import codecs

 import unicodedata

 import warnings

 from setuptools.compat import unicode, bytes

@@ -72,6 +73,32 @@ def joinpath(prefix, *suffix):
         return os.path.join(*suffix)

     return os.path.join(prefix, *suffix)

 

+def determine_console_encoding():

+    try:

+        #try for the preferred encoding

+        encoding = locale.getpreferredencoding()

+    

+        #see if the locale.getdefaultlocale returns null

+        #some versions of python\platforms return US-ASCII

+        #when it cannot determine an encoding

+        if not encoding or encoding == "US-ASCII":

+            encoding = locale.getdefaultlocale()[1]

+        

+        if encoding:

+            codecs.lookup(encoding)  # make sure a lookup error is not made

+            

+    except locale.Error, LookupError:

+        encoding = None

+        

+    #olders pythons defaulted to this

+    if not encoding:

+        return "US-ASCII"

+    elif encoding.startswith("mac-") and os.platform == "darwin":

+        return "utf-8"

+    else:

+        return encoding

+    

+_console_encoding = determine_console_encoding()

 

 def decode_as_string(text, encoding=None):

     """

@@ -87,7 +114,7 @@ def decode_as_string(text, encoding=None):
     #text should be a byte string    

 

     if encoding is None:

-        encoding = locale.getpreferredencoding()

+        encoding = _console_encoding

 

     if not isinstance(text, unicode):

         text = text.decode(encoding)