How to handle returned value if an exception happens in a library code

def parse_ET(self, ETObj):

Python style guide recommends lowercase_with_underscores for both function and parameter names

    if ETObj == None: return None

Use is None to check for None. However, consider whether you really want to support None as a parameter. Why would someone pass None to this function? Do they really want a None in return?

    dict_of_dict = collections.defaultdict(dict)

Avoid names that describe the datastructure. Choose names that describe what you are putting in it.

    try:
        for doc in ETObj.iter("result"):
            id = doc.attrib.get("id")
            for elem in doc.iter("attrib"):
                dict_of_dict[id].setdefault(elem.attrib.get("name"), elem.text)

Can there be duplicate "id" and "name"? If not you don't need to use defaultdict or setdefault

    except Exception, ex:

You should almost never catch generic exceptions. Only catch the actual exceptions you want.

        logging.exception("%s:%s" % (self.__class__, str(ex)))

Don't log errors and try to continue on as if nothing has happened. If your function fails, it should raise an exception. In this case, you probably shouldn't even catch the exception.

    finally:
        return dict_of_docs

finally is for cleanup tasks. Under no circumstances should you putting a return in it.