How can I skip a test if another test fails with py.test?

Question

Let's say I have these test functions:

def test_function_one():
    assert # etc...

def test_function_two():
    # should only run if test_function_one passes
    assert # etc.

How can I make sure that test_function_two only runs if test_function_one passes (I'm hoping that it's possible)?

Edit: I need this because test two is using the property that test one verifies.

Can you explain why you need this? Is the first test setting up something the second one uses? That's generally bad. — loganfsmyth, May 05 '12 at 17:59
This is typically a sign of a brittle test, a test that depends on more things than the unit it tests, are you *sure* you need to do this? It would (probably) be better to refactor the test and/or code under test to avoid this dependency. — Lasse V. Karlsen, May 05 '12 at 18:01
@loganfsmyth No the first test is not setting up anything, but the second test *does* use the property that the first test verifies. — Te-jé Rodgers, May 05 '12 at 18:07
@tjd.rodgers But then doesn't it make sense that the second test would also fail if it's inputs are incorrect. Why would you not want to run it? — loganfsmyth, May 05 '12 at 18:11
@loganfsmyth I'd thought of that, but I didn't want a situation where the second test fails for no apparent reason and pollutes the test results while the problem really is in the first test. — Te-jé Rodgers, May 05 '12 at 18:19
I think you want a single test, which makes two assertions. If the first assertion fails, the second will not execute. — Martin Del Vecchio, May 30 '12 at 14:47
Don't worry about the push-back on the question. I reckon it's a good question worth asking, even if the answer isn't what you may have originally expected. There's no such thing as a bad question. — Craig McQueen, Aug 31 '12 at 03:20
I think this kind of dependency between tests is fine: I would rather see (after waiting a short time) in my test output a failed `test_dashboard_x_present` and 10 skipped test cases than (after waiting a long time) 11 failed `test_dashboard_x_....` and having to look over them to notice what's going on. — DaveFar, Nov 05 '20 at 07:38

score 10 · Answer 1 · edited Jul 09 '21 at 08:24

10

You can use plugin for pytest called pytest-dependency.

The code can look like this:

import pytest

@pytest.mark.dependency()   #First test have to have mark too
def test_function_one():
    assert 0, "Deliberate fail"

@pytest.mark.dependency(depends=["test_function_one"])
def test_function_two():
    pass   #but will be skipped because first function failed

edited Jul 09 '21 at 08:24

maxt026

63
10

answered Aug 23 '17 at 08:24

Jakub Wagner

428
1
5
15

1

this indeed skips `test_function_two` for me, but it *also skips when `test_function_one` passes* – Gulzar Jan 02 '20 at 09:18
1

@Gulzar I had the same issue and realised that you need to also decorate the function that is being depended on, even if it depends on nothing. In this example, make sure you have `@pytest.mark.dependency()` on `test_function_one`. – William Laroche Feb 13 '20 at 11:47

score 1 · Answer 2 · answered Aug 31 '12 at 03:16

1

I think that a solution for yout would be to mock the value that the test1 is setting.

Ideally tests should be independent, so try to mock that value so you can always run test2 whenever you want, in fact, you should also simulate (mock) dirty values so you can see how test2 behaves if it receives unexpected data.

answered Aug 31 '12 at 03:16

Juan Antonio Gomez Moriano

13,103
10
47
65

In theory what you say is true. But in my experience by mocking the result of the first test you end up reimplementing the functionality of the first function / class instead of reusing that code. – Akzidenzgrotesk Oct 16 '18 at 08:09

score 1 · Answer 3 · answered Jul 19 '20 at 10:15

1

I'm using the plugin for pytest called pytest-dependency.

Adding to the stated above - if you are using tests inside test classes - you got to add the test class name to the function test name.

For example:

import pytest


class TestFoo:

    @pytest.mark.dependency()
    def test_A(self):
        assert False

    @pytest.mark.dependency(depends=['TestFoo::test_A'])
    def test_B(self):
        assert True

So if test_A fails - test_B won't run. and if test_A passes - test_B will run.

answered Jul 19 '20 at 10:15

sheldonzy

5,505
9
48
86

pytest is not honoring dependency, its running the dependent steps even if it fails, no impact at all. I am using pytest 7.1.1 https://stackoverflow.com/questions/71832895/pytest-dependency-not-working-tests-run-even-if-the-dependent-tests-fail – Anmol Parida Apr 22 '22 at 07:51

score 0 · Answer 4 · edited Mar 08 '23 at 14:59

0

When you have class defined make sure you use classname along with the testcase name else it will not work

class Test_class():
    @pytest.mark.dependency()
    def test1(self):
        assert False

    @pytest.mark.dependency(depends=["classname::test1"])
    def test2(self):
        print("lastname")

Output: failed=1, skipped=1

edited Mar 08 '23 at 14:59

Minh-Long Luu

2,393
1
17
39

answered Mar 03 '23 at 10:41

Chetan Savant

1

Please format code as code, not plain text – Gameplay Mar 08 '23 at 08:11

score -1 · Answer 5 · answered May 30 '12 at 14:49

-1

I think this is what you want:

def test_function():
    assert # etc...
    assert # etc...

This meets your requirement that the second "test" (assertion) runs only if the first "test" (assertion) passes.

answered May 30 '12 at 14:49

Martin Del Vecchio

3,558
2
27
36

How can I skip a test if another test fails with py.test?

5 Answers5