Question

I have a process that pickles a dictionary using Python 3.2. I then need to unpickle this dictionary using Python 2.7 or 2.6. The problem is that when transferring between python versions I get a dictionary full of unicode data which upsets the Python API I am trying to feed it into.

Pickling in Python 3.2:

myDict = {'a': 'first', 'b': 'second', 'c': 'third'}
with open(file, 'wb') as f:
    pickle.dump(myDict, f, 2)

Unpickling in Python 2.6:

with open(file, f) as f:
    myDict = pickle.load(f)

returns: {u’a’: u’first’, u’c’: u’third’, u’b’: u’second’}

How can I get back exactly what I put in (i.e. not unicode)?

Answer 1

You’re actually getting back exactly what you put in, since strings in Python 3 are unicode

To get str‘s, you can convert the keys and values in the dictionary:

strDict = dict((k.encode(), v.encode()) for k, v in myDict.iteritems())