Add __all__ (suggested by Raymond Hettinger).

Rename 'whitespace' global to '_whitespace' -- it's not part of the
public interface.

Lib/textwrap.py

@@ -12,6 +12,8 @@ __revision__ = "$Id$"
 
 import string, re
 
+__all__ = ['TextWrapper', 'wrap', 'fill']
+
 # Hardcode the recognized whitespace characters to the US-ASCII
 # whitespace characters.  The main reason for doing this is that in
 # ISO-8859-1, 0xa0 is non-breaking whitespace, so in certain locales
@@ -20,7 +22,7 @@ import string, re
 # same as any other whitespace char, which is clearly wrong (it's a
 # *non-breaking* space), 2) possibly cause problems with Unicode,
 # since 0xa0 is not in range(128).
-whitespace = '\t\n\x0b\x0c\r '
+_whitespace = '\t\n\x0b\x0c\r '
 
 class TextWrapper:
     """
@@ -58,11 +60,11 @@ class TextWrapper:
         be broken, and some lines might be longer than 'width'.
     """
 
-    whitespace_trans = string.maketrans(whitespace, ' ' * len(whitespace))
+    whitespace_trans = string.maketrans(_whitespace, ' ' * len(_whitespace))
 
     unicode_whitespace_trans = {}
     uspace = ord(u' ')
-    for x in map(ord, whitespace):
+    for x in map(ord, _whitespace):
         unicode_whitespace_trans[x] = uspace
 
     # This funky little regex is just the trick for splitting