How do you write maintainable tests with pytest?

pytest discovers functions named test_ and runs them with fixtures injected by name. Fixtures can be scoped to a function, module or session, which keeps expensive setup shared without leaking state between tests. Plain assert statements are rewritten to show the compared values on failure.

import pytest

@pytest.fixture
def account():
    return Account(owner="Asha", balance=100)

@pytest.mark.parametrize(
    "amount,expected",
    [(50, 50), (100, 0), (150, -50)],
)
def test_withdraw(account, amount, expected):
    account.withdraw(amount)
    assert account.balance == expected

def test_rejects_negative(account):
    with pytest.raises(ValueError):
        account.withdraw(-1)

Practices that scale

• Test observable behaviour, not private implementation details
• Use tmp_path for filesystem tests and monkeypatch for environment or attributes
• Mark slow or integration tests so the fast suite stays quick in CI
• Use coverage to find untested branches, not as a target to game

Fixtures should be explicit about their scope. A session-scoped fixture that mutates shared state will create order-dependent failures. Keep tests independent so they can run in any order and in parallel.