[Python] When writing unit tests in python, why are mocks in subsequent tests not overwriting...

I am trying to write unit tests that involve mocking several libraries in different ways for each test. When I run each test individually, they all pass, but when I run them all together, many of them fail. The reason they fail is that when a method is mocked in two tests, the mock from the second test is ignored, and the mocked function always behaves as defined by the first test.

I have made a demo project to demonstrate the issue:

File Structure


/.
├── src
│ └── main.py
└── test
├── __init__.py
├── test_base.py
└── test.py

Files


main.py

import boto3

s3 = boto3.client('s3')

def main():
return s3.list_buckets()


__init__.py

from .test import *


test_base.py

import unittest
from unittest.mock import MagicMock, patch

class TestBase(unittest.TestCase):
def setUp(self):
self.patch_boto3_client = patch('boto3.client', autospec=True)

self.mock_boto3_client = self.patch_boto3_client.start()

# BOTO3

# Set up mock boto3 clients
self.mock_s3 = MagicMock()

# Configure mock_boto3_client to return our mock clients
self.mock_boto3_client.side_effect = lambda service: {
's3': self.mock_s3,
}[service]

def tearDown(self) -> None:
self.patch_boto3_client.stop()


test.py

from .test_base import TestBase

class Test(TestBase):
def test_1(self):
from src.main import main

self.mock_s3.list_buckets.return_value = 'test_1'

response = main()

self.assertEqual(response, 'test_1')


def test_2(self):
from src.main import main

self.mock_s3.list_buckets.return_value = 'test_2'

response = main()

self.assertEqual(response, 'test_2')

Results


$ python3 -m coverage run -m unittest test
.F
======================================================================
FAIL: test_2 (test.test.Test.test_2)
----------------------------------------------------------------------
Traceback (most recent call last):
File ".../test/test.py", line 21, in test_2
self.assertEqual(response, 'test_2')
AssertionError: 'test_1' != 'test_2'
- test_1
? ^
+ test_2
? ^


----------------------------------------------------------------------
Ran 2 tests in 0.292s

FAILED (failures=1)


I have tried patching by using decorators, and manually starting and stopping patch objects as shown in my code.

How can I make my unit tests respect updates I make to mocked functions?

Continue reading...