Develop discriminating pytest tests with Copilot

Pytest can collect the fixture’s existing unittest.TestCase tests and new pytest functions. A green collection UI is not proof that a regression assertion executed.

Lab briefing

[!NOTE] A lightweight data object isolates repository search, not JSON loading.

Concepts · First task · Evidence checklist · Reset

Learning objectives

• Use the existing test conventions and a selected interpreter.
• Test JsonLoanRepository.get_loan, not a mock of that method.
• Verify found/missing IDs and non-mutation.
• Demonstrate a test failing for an intentional wrong assertion.

Before you start

Prepare 04-pytest following the setup guide. Use the supplied Python testing fixture. Select an environment on the work drive. Install requirements.txt in that copy only if pytest is missing, then run commands from library.

Concepts and use cases

get_loan iterates the injected object’s loans collection and returns the matching loan or None. A lightweight object with a loans attribute is sufficient for that unit test; actual JsonData loading and saving are separate component tests.

Exercise scenario

The service tests do not directly validate repository search. Add tests that distinguish an existing loan from a missing one without modifying fixture JSON.

Task 1 - Inspect and collect

python -m pytest --collect-only -q tests
python -m pytest -q tests

Record actual test names/counts. Read infrastructure/json_loan_repository.py, application_core/entities/loan.py, and the service tests before adding anything.

Task 2 - Define cases and a narrow seam

Example seam:

from types import SimpleNamespace
from infrastructure.json_loan_repository import JsonLoanRepository

# Supply actual Loan objects constructed with the fields required by the fixture.
data = SimpleNamespace(loans=[])
repository = JsonLoanRepository(data)
assert repository.get_loan(999) is None

This isolates the repository method, not the JSON parser. Do not report it as a filesystem integration test.

Task 3 - Plan and write the tests

Ask:

Inspect get_loan and the Loan entity. Propose found/missing/empty/non-mutation cases.
Use the real repository and a minimal data object; do not mock get_loan itself.
Distinguish current wrong-type behavior from a new validation requirement.

1. Review the plan and add tests/test_json_loan_repository.py.
2. Use pytest fixtures to avoid duplicated setup when they improve clarity.
3. Assert fields or object identity as required by the current contract.
4. Keep dates controlled and unrelated service behavior out of scope.
5. Do not add broad exception handlers to make tests pass.

Task 4 - Prove detection

python -m pytest -q tests/test_json_loan_repository.py
python -m pytest -q tests

1. Confirm the focused file actually contains collected tests.
2. Temporarily make a found-ID assertion expect None; verify failure.
3. Restore the assertion and rerun.
4. Use the Test Explorer only after selecting the same interpreter/import root.
5. Record limitations: these tests do not prove JSON loading, persistence, security, or all loan-service behavior.

Verify your work

• Found/missing/empty cases exercise the actual repository.
• Search does not mutate the source collection.
• The wrong assertion fails with nonzero exit.
• Interpreter, working directory and commands are reproducible.
• No real JSON data or other project was modified.

Troubleshooting

ModuleNotFoundError usually means the wrong root or interpreter. 0 tests collected is not success. If a mock returns exactly what you configured, ask whether the method under test ran at all.

Independent practice

Add a component test for corrupt JSON using a temporary fixture and characterize the actual loader behavior. Do not silently assume a printed error is an exception.

Reset

Restore only your new test file and local test configuration. Close the disposable workspace; keep shared interpreters intact.

Official references

• Python testing in VS Code
• Python environments
• pytest fixtures