How to unittest that a thread is spawned?
mocking, multithreading, python
Solution
How about modifying the `bar` to return a thead object:
def bar():
t = threading.Timer(0.5, foo, args=('potato',), kwargs={'x': 69, 'y':'spam'})
t.start()
return t # <----
Then, join the thread in the test code:
class SpamTest(TestCase):
def test_bar(self):
with patch('spam.foo') as mock:
t = bar()
t.join() # <----
mock.assert_called_once_with('potato', y='spam', x=69)
UPDATE alternative that does not require `bar` change.
import threading
import time
...
class SpamTest(TestCase):
def test_bar(self):
foo = threading.Event()
with patch('spam.foo', side_effect=lambda *args, **kwargs: foo.set()) as mock:
# Make the callback `foo` to be called immediately
with patch.object(threading._Event, 'wait', time.sleep(0.000001)):
bar()
foo.wait() # Wait until `spam.foo` is called. (instead of time.sleep)
mock.assert_called_once_with('potato', y='spam', x=69)
UPDATE
In Python 3.x, patch `threading.Event` instead of `threading._Event`.
Problem
I have a race condition in a unittest I'm trying to fix. Suppose there's a module `spam.py`: ``` import threading def foo(*args, **kwargs): pass def bar(): t = threading.Timer(0.5, foo, args=('potato',), kwargs={'x': 69, 'y':'spam'}) t.start() ``` And here is the test for it: ``` from mock import patch from spam import bar from unittest import TestCase class SpamTest(TestCase): def test_bar(self): with patch('spam.foo') as mock: bar() mock.assert_called_once_with('potato', y='spam', x=69) ``` Of course this test fails with `AssertionError: Expected to be called once. Called 0 times.` because the call to `bar()` is non-blocking, so the assertion happens too early. The test can be made to pass by putting a `time.sleep(1)` in before the assert, but this is obviously hacky and lame - what is the accepted way to mock / unittest asynchronous stuff?