Unit testing with unittest

The concept

A unit test checks one small piece of behavior automatically and reports pass or fail. A suite of fast, repeatable tests tells you quickly when a change breaks something. Tests also serve as executable documentation: they show how a function is expected to behave, including in unusual cases.

Python's built-in unittest module groups tests in classes that inherit from TestCase. Each method whose name starts with test is one test. A good test has three parts, often called arrange, act, and assert: set up the input, call the code, and check the result. Choose cases deliberately: a typical value, boundary values such as 0 and 100, and invalid input that should raise an error, checked with assertRaises.

The example prices are integer cents so that arithmetic is exact: floating-point numbers are a poor fit for money. The program loads the tests, runs them with a runner that writes its report to an in-memory stream, and prints how many ran and whether they all passed. In a real project you would run them with python -m unittest or another test runner such as pytest, on every change.

Code example

import io
import unittest


def apply_discount(price_cents, percent):
    if not 0 <= percent <= 100:
        raise ValueError("percent must be between 0 and 100")
    return price_cents * (100 - percent) // 100


class DiscountTests(unittest.TestCase):
    def test_typical_discount(self):
        self.assertEqual(apply_discount(8000, 25), 6000)

    def test_zero_discount_keeps_price(self):
        self.assertEqual(apply_discount(1999, 0), 1999)

    def test_invalid_percent_is_rejected(self):
        with self.assertRaises(ValueError):
            apply_discount(1000, 120)


suite = unittest.defaultTestLoader.loadTestsFromTestCase(DiscountTests)
result = unittest.TextTestRunner(stream=io.StringIO()).run(suite)
print(result.testsRun, result.wasSuccessful())

3 True

Follow the code

Open the walkthrough to inspect the example one step at a time.

—

—

—

3 True

Behind the scenes

Cost and common mistakes

A test run takes time proportional to the number of tests and the cost of the code they call. Unit tests should run in milliseconds.

• Tests that depend on the clock, the network, random values, or each other are flaky. Make each test independent and deterministic.
• Checking only the happy path misses most bugs. Test boundaries and invalid input too.
• A passing test suite shows that the tested cases work, not that the program is correct.

Try it yourself

Add a test for apply_discount(1000, 100). What should it return?

Further reading

• Python unittest module (external)
• SWEBOK Guide (IEEE Computer Society) (external)