QA Wiki

Appearance

Python for QA ​

Python is the QA's Swiss army knife: test data scripts, internal tools, results analysis and, with pytest and requests, a complete and very readable testing stack. Even if your main suite is Java or TypeScript, Python shows up sooner or later.

pytest in five minutes ​

A test is a function that starts with test_ and uses plain assert:

python
def test_active_order():
    order = create_service_order(status="active")
    assert order.status == "active"

The three pieces worth knowing:

  • Fixtures — reusable setup and teardown, injected by parameter name:
python
import pytest

@pytest.fixture
def api_client():
    client = ApiClient(base_url=BASE_URL)
    yield client        # everything after the yield is the teardown
    client.close()

def test_health(api_client):
    assert api_client.get("/health").status_code == 200
python
@pytest.mark.parametrize("msisdn", ["12345", "abcdefghi", "+34-600", ""])
def test_invalid_msisdn_rejected(api_client, msisdn):
    r = api_client.post("/service-orders", json={"productId": "mobile-20gb", "msisdn": msisdn})
    assert r.status_code == 400
  • Markers (@pytest.mark.smoke) — labeling and filtering execution (pytest -m smoke), like tags in other frameworks.

requests for APIs ​

python
import requests

order = {"customerId": "C-100", "productId": "fiber-1gbps"}
r = requests.post(f"{BASE_URL}/service-orders", json=order, timeout=10)
assert r.status_code == 201
assert r.json()["status"] == "created"

With pytest + requests + jsonschema you get the lightweight equivalent of REST Assured: the anatomy of an API test is the same, only the syntax changes.

What else I use it for ​

  • Generating test data (files, bulk payloads, random data with faker).
  • Support scripts: cleaning environments, comparing responses between environments, processing logs or results CSVs.
  • Internal tools: in my case, the sharding optimization model that later became CI Shard Advisor started as a Python script.

When to pick Python as the suite's language ​

Pick Python if…Pick Java/TS if…
The team already speaks it (data, scripts, Python backend)The suite lives next to Java/TS product code
You prioritize readability and writing speedYou want the product stack's typing and tooling
Testing is mostly APIs and dataThere's a lot of browser E2E (Playwright TS is first-class)

Key idea

Frameworks change, concepts don't: fixtures are setup/teardown, parametrize is data-driven and markers are tags. If you master those ideas in one stack, in Python you're only missing the syntax — and it's the friendliest one to learn.

References ​