Sometimes you want to overhaul a chunk of code and don't want to stare at a broken test. You could comment it out. But pytest provides an easier (and more feature-ful) alternative for skipping tests.

We'll show this in action while implementing:

• The ability to add multiple guardians at once

• A concept of the "primary guardian" of a player

Bulk Guardians

Players usually have more than one Guardian. When you have a list of Guardians, you might prefer a different method that lets you add them all at once.

Let's implement this in TDD fashion, by first writing a test_add_guardians test in test_player.py that fails:

def test_add_guardians():
    p = Player('Tatiana', 'Jones')
    # Add one guardian
    g1 = Guardian('Mary', 'Jones')
    p.add_guardian(g1)
    # Later, add some more
    g2 = Guardian('Joanie', 'Johnson')
    g3 = Guardian('Jerry', 'Johnson')
    p.add_guardians([g2, g3])
    assert [g1, g2, g3] == p.guardians

We don't have a method add_guardians and so the test fails. Perhaps we are busy on something else and we'd like pytest to not yell at us, but we don't want to comment out or (worse) delete the test.

Instead, let's import pytest and put a decorator for the skip marker on that test:

import pytest

from laxleague.guardian import Guardian
from laxleague.player import Player


def test_construction():
    p = Player('Tatiana', 'Jones')
    assert 'Tatiana' == p.first_name
    assert 'Jones' == p.last_name
    assert [] == p.guardians


def test_add_guardian():
    g = Guardian('Mary', 'Jones')
    p = Player('Tatiana', 'Jones')
    p.add_guardian(g)
    assert [g] == p.guardians


@pytest.mark.skip(reason='Have not yet implemented method')
def test_add_guardians():
    p = Player('Tatiana', 'Jones')

    # Add one guardian
    g1 = Guardian('Mary', 'Jones')
    p.add_guardian(g1)

    # Later, add some more
    g2 = Guardian('Joanie', 'Johnson')
    g3 = Guardian('Jerry', 'Johnson')
    p.add_guardians([g2, g3])

    assert [g1, g2, g3] == p.guardians

Remember, we don't have to manually type the import...just start typing @pyt and let PyCharm autocomplete using Ctrl-Space Ctrl-Space.

Now when the tests run (automatically, thanks to Toggle auto-test), they don't fail, but they do indicate a test was ignored:

With our failing test in place, let's implement the missing method. In player.py, clone the existing add_guardian method, then change its arguments and implementation:

from dataclasses import dataclass, field
from typing import List

from laxleague.guardian import Guardian


@dataclass
class Player:
    """ A lacrosse player in the league """

    first_name: str
    last_name: str
    guardians: List[Guardian] = field(default_factory=list)

    def add_guardian(self, guardian: Guardian):
        self.guardians.append(guardian)

    def add_guardians(self, guardians: List[Guardian]):
        self.guardians.extend(guardians)

We can now remove the skip marker and the test passes. Remember to remove the now-unused pytest import in test_player.py using Optimize Imports.

Some Typing Cleanup

Eager readers might have spotted a type hinting flaw: our code breaks the Be liberal in what you accept, and conservative in what you return rule.

That is, our new method wants a List. When really, it will take a Python "iterable".

For example, our test passes in a list of guardians. It's immutable. Might as well change it to be a tuple:

    p.add_guardians((g2, g3))

Doing so breaks Python type checking:

Let's change our add_guardians to accept any kind of Iterable:

from dataclasses import dataclass, field
from typing import List, Iterable

from laxleague.guardian import Guardian


@dataclass
class Player:
    """ A lacrosse player in the league """

    first_name: str
    last_name: str
    guardians: List[Guardian] = field(default_factory=list)

    def add_guardian(self, guardian: Guardian):
        self.guardians.append(guardian)

    def add_guardians(self, guardians: Iterable[Guardian]):
        self.guardians.extend(guardians)

Tests still pass and type checking now passes.

Primary Guardian

For the second feature, let's use the same process: write a failing test, temporarily mark it with skip, then implement it and remove the mark.

Our feature will work like this: whichever guardian is added first is the primary guardian. In test_player.py we add test_primary_guardian, with the mark already in place:

@pytest.mark.skip
def test_primary_guardian():
    p = Player('Tatiana', 'Jones')
    # Add one guardian
    g1 = Guardian('Mary', 'Jones')
    p.add_guardian(g1)
    # Later, add some more
    g2 = Guardian('Joanie', 'Johnson')
    g3 = Guardian('Jerry', 'Johnson')
    p.add_guardians((g2, g3))
    assert g1 == p.primary_guardian

Now time for the implementation. We're doing this as a Python "property", so add the following in player.py:

from dataclasses import dataclass, field
from typing import List, Iterable

from laxleague.guardian import Guardian


@dataclass
class Player:
    """ A lacrosse player in the league """

    first_name: str
    last_name: str
    guardians: List[Guardian] = field(default_factory=list)

    def add_guardian(self, guardian: Guardian):
        self.guardians.append(guardian)

    def add_guardians(self, guardians: Iterable[Guardian]):
        self.guardians.extend(guardians)

    @property
    def primary_guardian(self):
        return self.guardians[0]

Tip: Use the property LiveTemplate in PyCharm to speed up the generation of a property.

Remove the @pytest.mark.skip mark from test_primary_guardian and the test now passes.

Conclusion

The pytest.mark.skip facility, with related skipIf and xFail, have a broad set of uses. As you mature in test writing, start to include other people, and have tests that execute in different environments, you'll put them to good use.