Testing with Pytest and TDD

Testing is one of the most important practices in software development. It gives you confidence that your code works, and it prevents bugs from reaching production.

Why Testing Matters

• Catches bugs early — cheaper to fix in development than in production
• Documents behavior — tests show what the code should do
• Enables refactoring — change code without fear of breaking things
• Speeds up development — automated tests are faster than manual checking

Testing for AI Validation

With AI-generated code (Copilot, Claude, ChatGPT), testing is critical:

• AI can generate code that looks correct but has subtle bugs
• Tests validate that AI-generated code actually works
• Give AI your tests as acceptance criteria — it produces better code
• Tests are the “single most impactful practice” for agentic coding (Anthropic)

The Testing Pyramid

        /  E2E  \          ← Few, slow, expensive
       /----------\
      / Integration \      ← Some, medium speed
     /----------------\
    /    Unit Tests     \  ← Many, fast, cheap
   /____________________\

Test-Driven Development (TDD)

TDD means writing tests before writing code. The cycle:

1. RED    → Write a failing test
2. GREEN  → Write minimum code to pass
3. REFACTOR → Clean up the code
↻ Repeat

Benefits of TDD:

• Forces you to think about design first
• You get 100% test coverage by default
• Small, focused iterations
• Acts as living documentation

Pytest Basics

Pytest is the most popular Python testing framework. Simple and powerful.

Install

pip install pytest

Write a Test

# calculator.py
def add(a, b):
    return a + b

# test_calculator.py
from calculator import add

def test_add_positive():
    assert add(2, 3) == 5

def test_add_negative():
    assert add(-1, -1) == -2

def test_add_zero():
    assert add(0, 0) == 0

Run Tests

pytest                    # run all tests
pytest test_calculator.py # run specific file
pytest -v                 # verbose output
pytest -k "test_add"      # run tests matching pattern

Key Pytest Features

Fixtures — Reusable Setup

import pytest

@pytest.fixture
def database():
    db = create_test_db()
    yield db
    db.cleanup()

def test_insert(database):
    database.insert({"name": "Alice"})
    assert database.count() == 1

Parametrize — Multiple Inputs

@pytest.mark.parametrize("a, b, expected", [
    (1, 2, 3),
    (0, 0, 0),
    (-1, 1, 0),
])
def test_add(a, b, expected):
    assert add(a, b) == expected

Markers — Categorize Tests

@pytest.mark.slow
def test_heavy_computation():
    ...

# Run: pytest -m "not slow"

Mocking

from unittest.mock import patch

def test_api_call():
    with patch("module.requests.get") as mock_get:
        mock_get.return_value.status_code = 200
        result = fetch_data()
        assert result is not None

Project Structure

project/
├── src/
│   └── calculator.py
├── tests/
│   ├── conftest.py        ← shared fixtures
│   ├── test_calculator.py
│   └── test_integration.py
├── pytest.ini             ← pytest config
└── requirements.txt

pytest.ini Example

[pytest]
testpaths = tests
markers =
    slow: marks tests as slow
    integration: integration tests
addopts = -v --tb=short

Summary