Use fixtures to isolate test data from production data

A reliable test owns its inputs. Reading a live spreadsheet or production database makes results unstable and can expose or alter real records.

Prove a fresh copy

The engineering foundations lab mutates a copied fixture.

from engineering_cases import fixture_isolation_case

result = fixture_isolation_case()
assert result["mutated_copy"][0]["amount"] == 99
assert result["fixture"][0]["amount"] == 10
assert result["fixture_unchanged"] is True
assert result["production_records_used"] is False

Use minimal synthetic rows, temporary directories and disposable local databases. Scope mutable fixtures so one test cannot contaminate another. Factories help vary only the fields a test cares about; snapshots need careful review when schemas change.

Configuration should fail closed if the test process points at a production hostname or account. Seed randomness and clock values. Separate contract-test credentials and clean up records with unique run IDs when a sandbox is necessary.

The FDE for Freshers course treats safe test data as part of professional engineering practice.

Exercise

Build three pytest fixtures for records, temporary files and SQLite. Deliberately mutate one and show the next test receives clean state.

Continue learning

This article is part of the FDE engineering foundations sequence. Use the neighbouring tasks when you need the prerequisite or the next application.

• Review the prerequisite or neighbouring task in Write unit tests around a business rule.
• Continue with Mock an external API without testing your own mock.

Reference: pytest fixtures.