Skip to content

Drop-in @pytest.mark.parametrize replacement working with unittest.TestCase

License

Notifications You must be signed in to change notification settings

Bobronium/parametrize

Repository files navigation

CI PyPi Python Versions Code style: black

Drop-in @pytest.mark.parametrize replacement working with unittest.TestCase

Why?

You want to start using @pytest.mark.parametrize, but can't simply drop unittest.TestCase because you have tons of self.assert's, setUp's tearDown's to rewrite?

With @parametrize you can start parameterizing your tests now, and get rid of unittest.TestCase later if needed.

Usage

Simple example from pytest docs adapted to unittest

import unittest
from parametrize import parametrize

class TestSomething(unittest.TestCase):

    @parametrize('test_input,expected', [("3+5", 8), ("2+4", 6)])
    def test_eval(self, test_input, expected):
        self.assertEqual(expected, eval(test_input))
$ python -m unittest test.py -v
test_eval[2+4-6] (test.TestSomething) ... ok
test_eval[3+5-8] (test.TestSomething) ... ok
test_eval[6*9-42] (test.TestSomething) ... FAIL

======================================================================
FAIL: test_eval[6*9-42] (test.TestSomething)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "parametrize/parametrize.py", line 261, in parametrized_method
    return parametrized_func(*args, **kwargs)
  File "test.py", line 8, in test_eval
    self.assertEqual(expected, eval(test_input))
AssertionError: 42 != 54

----------------------------------------------------------------------
Ran 3 tests in 0.001s

FAILED (failures=1)
You don't need to use additional decorators, custom base classes or metaclasses.

Stacking parametrize decorators:

import unittest
from parametrize import parametrize

class TestSomething(unittest.TestCase):
    
    @parametrize("x", [0, 1])
    @parametrize("y", [2, 3])
    def test_foo(self, x, y):
        pass

test_foo will be called with: (x=0, y=2), (x=1, y=2), (x=0, y=3), and (x=1, y=3):

$ python -m unittest test.py -v
test_foo[2-0] (test.TestSomething) ... ok
test_foo[2-1] (test.TestSomething) ... ok
test_foo[3-0] (test.TestSomething) ... ok
test_foo[3-1] (test.TestSomething) ... ok

----------------------------------------------------------------------
Ran 4 tests in 0.000s

OK
Note: even though the tests are always generated in the same order, the execution order is not guaranteed

Compatibility

Any @parametrize decorator can be converted to @pytest.mark.parametrize just by changing its name. @pytest.mark.parametrize decorator can be converted to @parametrize as long as pytest.param, indirect, ids and scope are not used.

@parametrize works with both unittest and pytest. However, pytest is recommended due to limitations when using unittest in cli.

Parametrized tests are properly detected and handled by PyCharm. They are displayed as if they were parametrized with @pytest.mark.parametrize.

Limitations

Since @parametrize does some kind of magic under the hood, there are some limitations you need to consider. It's likely you will never face most of them, but if you will, @parametrize will let you know with an error:

  • All parametrization must be done via decorators

    ✅ OK

    @parametrize('a', (1, 2))
    def f(a):
        ...

    ❌ Won't work:

    def f(a):
        ...
    
    parametrize('a', (1, 2))(f)
    RuntimeError: Unable to find any parametrizes in decorators, please rewrite decorator name to match any of detected names @{'parametrize'}
  • All other decorators must be defined before parametrize decorators

    ✅ OK:

    @parametrize("a", (1, 2))
    @parametrize("b", (2, 3))
    @mock.patch(f"{__name__}.bar", "foo")
    def f(a, b):
        return a, b

    ❌ Won't work:

    @mock.patch(f"{__name__}.bar", "foo")
    @parametrize("a", (1, 2))
    @parametrize("b", (2, 3))
    def f(a, b):
        return a, b
    TypeError: @mock.patch(f"{__name__}.bar", "foo") must be defined before any of @{'parametrize'} decorators
  • If you assign parametrized decorator to variable, it must be accessible from locals() or globals() namespaces:

    ✅ OK:

    a_parameters = parametrize("a", (4, 5))  # defined in module
    def func():  
        class TestSomething:
            b_parameters = parametrize("b", (1, 2, 3))
    
            @b_parameters  # b_parameters found in locals()
            @a_parameters  # a_parameters found in globals()
            def test_method(self, a, b):
                ...

    ❌ Won't work:

    def func():
        # defined in function scope
        a_parameters = parametrize("a", (4, 5))
    
        class TestSomething:
            print('a_parameters' in {**globals(), **locals()})  # False
    
            @a_parameters  # accessed in class body scope
            def test_method(self, a, b):
                ...
    RuntimeError: Unable to find any parametrizes in decorators, please rewrite decorator name to match any of detected names @{'parametrize'}  
  • Parametrized method can be ran from command line only via pytest:

    $ cat test.py

    import unittest
    from parametrize import parametrize
    
    class TestSomething(unittest.TestCase):
        @parametrize('a', (1, 2))
        def test_something(self, a):
            self.assertIsInstance(a, int)

    ✅ OK:

    $ pytest test.py::TestSomething::test_something -v

     ...    
     test.py::TestSomething.test_something[1] ✓                                       50% █████     
     test.py::TestSomething.test_something[2] ✓                                      100% ██████████
    
    Results (0.07s):
           2 passed

    ❌ Won't work:

    $ python -m unittest test.TestSomething.test_something

    Traceback (most recent call last):
      ...
    TypeError: don't know how to make test from: test_something[...]
  • @parametrize cannot be used in interactive environments like REPL (It works in IPython though)

  • @parametrize cannot be used in cythonized code

About

Drop-in @pytest.mark.parametrize replacement working with unittest.TestCase

Resources

License

Stars

Watchers

Forks

Packages

No packages published