Unit-test fixtures and reproducibility — DedSec Project

What is happening

A good test specifies setup, action and expected observation. Use small deterministic inputs instead of relying on random output or live websites.

Practical investigation

For code that talks to external services, mock the HTTP boundary. Unit tests should fail because your logic changed, not because a network provider is offline.

Common traps

Test the error path as carefully as success. Preserve a regression test whenever you fix a bug found in production.

Adaptable demonstration

python -m unittest discover -v

Make the negative case observable

from tempfile import TemporaryDirectory
from pathlib import Path
import json

with TemporaryDirectory() as folder:
    path = Path(folder) / 'settings.json'
    path.write_text('{broken', encoding='utf8')
    before = path.read_bytes()
    try:
        json.loads(path.read_text(encoding='utf8'))
    except json.JSONDecodeError:
        pass
    else:
        raise AssertionError('malformed JSON was accepted')
    assert path.read_bytes() == before

This fixture checks both rejection and the preservation of the input. A test that only confirms an exception could miss code that corrupts the file before failing. Extend it with valid empty configuration, a Unicode value and a wrong-type top-level value. Parsing valid JSON does not imply the resulting configuration has the intended schema.

Keep mutable fixtures local to the test, not in a shared module-level list. Inject clocks and transport functions for expiry or network behavior. Assert the application's response and side effects rather than every internal helper call, so a harmless refactor does not require rewriting the whole suite.