auth.py

def validate_token(secret, auth_token):
  # do the token check logic... (using settings.PARTNER_ID)

def validate_token_from_yola(auth_token):
    yola = Yola()
    partner = yola.get_partner(settings.PARTNER_ID)
    validate_token(partner['shared_secret_key'], auth_token)

I guess I psyched myself out with how much set up the auth test was having to do for what is basically a ==

That's a good instinct. If you have places where collaboration and business logic mix, your tests are probably ugly, and a better design is probably hiding. In this gist I split that logic up:

validate_token is now all about the logic. So the tests can just pass in dicts and assert agains the == check.

validate_token_for_yola_user is now all about collaboration. This is where mocks should be used. The tests should mock Yola and validate_token and assert get_user and get_partner and validate_token are called correctly.