Modularize tests code

[nghnam]

Fix issue #26

Output

.2019-07-19 10:54:50,904 - INFO - test logprint
2019-07-19 10:54:50,904 - INFO - test log print
.....2019-07-19 10:54:50,906 - INFO - Model:
2019-07-19 10:54:50,906 - INFO - r.r: sub, obj, act
2019-07-19 10:54:50,906 - INFO - p.p: sub, obj, act
2019-07-19 10:54:50,906 - INFO - e.e: some(where (p_eft == allow))
2019-07-19 10:54:50,906 - INFO - m.m: r_sub == r_obj.Owner
2019-07-19 10:54:50,906 - INFO - Request: alice, {'Owner': 'alice', 'id': 'data1'}, write ---> True
............................
----------------------------------------------------------------------
Ran 34 tests in 0.022s

OK

[leeqvip]

There is no add_role_policy function, it is included in the add_policy function.
Modularize test cases are acceptable.

[hsluoyz]

Hi @techoner , I think the current test_add_policy() only tests p policy. It doesn't test g policy.

def test_add_policy(self):
        m = Model()
        m.load_model(get_examples("basic_model.conf"))

        rule = ['admin', 'domain1', 'data1', 'read']

        self.assertFalse(m.has_policy('p', 'p', rule))

        m.add_policy('p', 'p', rule)
        self.assertTrue(m.has_policy('p', 'p', rule))

So add_role_policy() can test g. It's not bad.

[leeqvip]

@hsluoyz , It can be placed together in test_add_policy.

[hsluoyz]

@techoner , OK, can you merge this PR then add a commit to merge this add_role_policy() function into add_policy() based on your point of view? I think it may be faster than letting the PR author modify it.