Introduction
When we write unittest, it comes the moment when we cross the road of some services that we can't test. So the moment comes when we have to use Mock
In the following text, you should spot some evidence related to django ;)
Function to mock
here is the piece of code I will "mock" :
def save_data(self, trigger_id, **data):
"""
let's save the data
:param trigger_id: trigger ID from which to save data
:param data: the data to check to be used and save
:type trigger_id: int
:type data: dict
:return: the status of the save statement
:rtype: boolean
"""
status = False
# set the title and content of the data
title, content = super(ServiceTwitter, self).save_data(
trigger_id, **data)
if data.get('link') and len(data.get('link')) > 0:
content = str("{title} {link}").format(
title=title, link=data.get('link'))
content += self.get_tags(trigger_id)
try:
self.twitter_api.update_status(status=content)
status = True
except Exception as inst:
logger.critical("Twitter ERR {}".format(inst))
update_result(trigger_id, msg=inst)
status = False
return status
Mock a complet function
the piece of unittest with the mock applied to save_data :
from unittest.mock import patch
def test_save_data(self):
token = self.token
trigger_id = self.trigger_id
content = 'foobar #tag'
self.data['title'] = 'a title'
self.data['link'] = 'http://domain.ltd'
self.assertTrue(token)
self.assertTrue(isinstance(trigger_id, int))
self.assertIn('text', self.data)
self.assertNotEqual(self.data['text'], '')
with patch.object(ServiceTwitter, 'save_data') as mock_save_data:
se = ServiceTwitter(self.token)
se.save_data(trigger_id, **self.data)
mock_save_data.assert_called_once_with(trigger_id, **self.data)
then the testing show us
coverage run --source='.' manage.py test -v2
...
test_save_data (th_twitter.tests.ServiceTwitterTest) ... ok
...
Fine !
But the coverage report (in html) shows us
and the %
[foxmask:~/DjangoVirtualEnv/django-trigger-happy/django-th] [django-trigger-happy] coverage report -m |grep twitter
th_twitter/__init__.py 2 0 100%
th_twitter/forms.py 12 0 100%
th_twitter/models.py 21 0 100%
th_twitter/my_twitter.py 117 56 52% 119-121, 138-173, 190-209, 219-231, 241-250, 256, 273-278
th_twitter/tests.py 80 0 100%
Mock one FunctionB in a FunctionA
It's fine, but in our quest of the perfect tests and to be sharper, we would like to test the content of save_data and only mock the function that makes the call to the Twitter API (named Twython).
To do so we can use the manager like previously, or a decorator.
Just have a look :
from unittest.mock import patch
@patch.object(Twython, 'update_status')
def test_save_data(self, mock1):
self.create_twitter()
token = self.token
trigger_id = self.trigger_id
content = 'foobar #tag'
self.data['title'] = 'a title'
self.data['link'] = 'http://domain.ltd'
self.assertTrue(token)
self.assertTrue(isinstance(trigger_id, int))
self.assertIn('text', self.data)
self.assertNotEqual(self.data['text'], '')
se = ServiceTwitter(self.token)
se.save_data(trigger_id, **self.data)
mock1.assert_called_once_with(status=content)
And this time the coverage report show us
and the %
[foxmask:~/DjangoVirtualEnv/django-trigger-happy/django-th] [django-trigger-happy] coverage report -m |grep twitter
th_twitter/__init__.py 2 0 100%
th_twitter/forms.py 12 0 100%
th_twitter/models.py 21 0 100%
th_twitter/my_twitter.py 117 40 66% 119-121, 138-173, 205-208, 241-250, 256, 273-278
th_twitter/tests.py 80 0 100%
Mock (2 or more functions) FunctionB and FunctionC in a FunctionA
Let's suppose you want to mock several functions in save_data, we will do something like this :
from unittest.mock import patch
# be careful with the order of the decorator
@patch.object(Twython, 'update_status') # will go to mock2
@patch.object(AnotherService, 'other_method') # will go to mock1
def test_save_data(self, mock1, mock2):
self.create_twitter()
token = self.token
trigger_id = self.trigger_id
content = 'foobar #tag'
self.data['title'] = 'a title'
self.data['link'] = 'http://domain.ltd'
self.assertTrue(token)
self.assertTrue(isinstance(trigger_id, int))
self.assertIn('text', self.data)
self.assertNotEqual(self.data['text'], '')
se = ServiceTwitter(self.token)
se.save_data(trigger_id, **self.data)
mock1.assert_called_once_with()
mock2.assert_called_once_with(status=content)
/!\ Here, be really very carefull with the order of the decorator:
if the parms for Twython.update_status and AnotherService.other_method are the same, we can write twice the same like mock1.assert_called_once(status=content) and mock2.assert_called_once(status=content) (for example)
but if they don't, be sure to set the right parm to the right 'mock'
End
Hope this will be helpful like it was for me as I spent a lot of time to find and test that ;)
If you want to dig that topic have a look at the doc