Engineering Craft › Testing

Unit Test

A fast test of one small piece of code in isolation.

Also known as: unit testing, unit tests

A unit test checks one small piece of code (usually a function or a class) in isolation, runs in milliseconds and doesn’t touch the network, disk or a real database.

def test_apply_discount():
    assert apply_discount(price=200, percent=25) == 150

Anatomy: arrange, act, assert

def test_cannot_withdraw_more_than_balance():
    account = Account(balance=50)                    # arrange: set up

    with pytest.raises(InsufficientFunds):
        account.withdraw(80)                         # act and assert on the outcome

    assert account.balance == 50                     # nothing changed

(See arrange-act-assert.)

What makes a good one

• Fast and independent. Tests don’t depend on order or on each other.
• Deterministic. Same input, same result every time. No real clock, randomness or network (flaky tests destroy trust).
• Tests behavior, not implementation, so refactoring doesn’t break it.
• One idea per test, with a name that says what failed: test_expired_coupon_is_rejected.
• Covers edge cases: empty input, zero, negative numbers, None, boundaries (see off-by-one errors).

What they can’t do

They won’t tell you your pieces work together. For that, see integration tests. Mix many unit tests with fewer larger ones, following the testing pyramid.

When a dependency is slow or unpredictable, replace it with a mock or fake, but sparingly. Run unit tests constantly; they’re your fastest safety net.