You're using Python 3, and strings (str-type) in Python 3 are Unicode objects. create_string_buffer creates C char arrays, so you are required to pass a bytes object. If you have a str object encode with an appropriate encoding to create a bytes object. Note there is a create_unicode_buffer that creates C wchar arrays as well and takes Python 3 str objects.
Python 3.2.1 (default, Jul 10 2011, 21:51:15) [MSC v.1500 32 bit (Intel)] on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import ctypes
>>> ctypes.create_unicode_buffer('abcd')
<ctypes.c_wchar_Array_5 object at 0x00C2D300>
>>> ctypes.create_string_buffer(b'abcd')
<ctypes.c_char_Array_5 object at 0x00C2D1C0>
>>> ctypes.create_string_buffer('abcd'.encode('utf-8'))
<ctypes.c_char_Array_5 object at 0x00C2D300>
In regards to the 2nd part of your question, do you want to edit the existing string or return an entirely new string allocated on the heap? If the former, use create_string_buffer to pass a mutable string to the function, then modifying it in place will work. Here's a simple example calling _strupr() from the Windows MSVCRT library:
Python 3.2.1 (default, Jul 10 2011, 21:51:15) [MSC v.1500 32 bit (Intel)] on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import ctypes
>>> c = ctypes.CDLL('msvcr90.dll')
>>> strupr = c._strupr
>>> a=ctypes.create_string_buffer(b'abc')
>>> strupr.restype=ctypes.c_char_p
>>> strupr(a)
b'ABC'
Note setting the restype (result type) of the function to a c_char_p tells ctypes to convert the return value to a bytes object. Also note that _strupr converts a string in place, so a mutable string buffer must be passed. Only pass Python byte strings to C functions that take const char*, since modifying Python strings directly via ctypes is BadTM.
str(not that that's the problem here). Shadowing builtins is evil.