Skip to content

testing.md

Latest commit

 

History

History
70 lines (44 loc) · 3.3 KB

testing.md

File metadata and controls

70 lines (44 loc) · 3.3 KB
Raw

Testing

This document describes the test strategy and how to run tests for IRIDIUM.

Python tests

Tests live under tests/. The test runner is pytest. Run from the repo root:

pytest tests/ -v

Unit tests

  • test_schemas.py: Pydantic schema validation (sensor events, route request, forecast segment, equity, anomaly). Ensures contracts accept valid data and reject invalid data.
  • test_forecasting.py: Forecasting pipeline; horizon, segment list, and segment field ranges.
  • test_routing.py: Routing service; response shape and alternative segments.
  • test_equity.py: Equity score computation; default districts and composite score range.
  • test_anomaly.py: Anomaly detector; list shape, since filter, and required fields.
  • test_ingestion.py: Ingestion batch validation and run_ingestion with synthetic dirs.

Integration tests

  • tests/integration/test_api_health.py: FastAPI TestClient against the full app. Covers GET /health, /version, /api/v1/network/graph, /api/v1/forecast/congestion, POST /api/v1/routing/plan, GET /api/v1/equity/score, GET /api/v1/anomalies, POST /api/v1/ingestion/events. No live server required; the client mounts the app.

To run only integration tests:

pytest tests/integration/ -v

Test data

Unit and integration tests use the in-memory digital twin and synthetic data under data/synthetic/. No database or external services are required for the baseline suite.

Frontend tests

Frontend tests use Vitest and React Testing Library. Run from the repo root or from apps/web:

cd apps/web && npm run test:run

Current coverage includes a smoke test for the App (layout and navigation). Additional component or API-boundary tests can be added under apps/web/src.

Coverage

Coverage is measured over apps, packages, and services. Install pytest-cov (e.g. pip install pytest-cov or use the dev extra), then run:

pytest tests/ --cov=apps --cov=packages --cov=services --cov-report=term-missing --cov-report=html
  • term-missing: prints missing line numbers per file.
  • html: writes htmlcov/ for a browser report.

Configuration lives in pyproject.toml under [tool.coverage.run] and [tool.coverage.report]. The build is configured with fail_under = 100; use # pragma: no cover on defensive or hard-to-test branches (e.g. HTTP error returns, optional imports) so that the main paths stay covered and the report reaches 100%.

  • Warnings: Deprecation warnings for datetime.utcnow() are suppressed via filterwarnings in [tool.pytest.ini_options]; production code uses datetime.now(timezone.utc).
  • Exclude lines: exclude_lines in coverage skip pragma: no cover, def __repr__, raise NotImplementedError, if TYPE_CHECKING:, if __name__ == "__main__":, and except ImportError: so optional or CLI-only code does not lower the percentage.

CI

GitHub Actions run lint and tests as defined in .github/workflows/ci.yml. The workflow runs pytest and, if present, frontend tests. Ensure all tests pass before merging.

What is not tested

  • End-to-end browser tests (e.g. Playwright) are not included in the baseline.
  • Performance and load tests are not included.
  • Federated learning and ST-GNN code paths are not yet implemented; tests will be added when those components exist.