Local test lab

The cannectors repository ships a local test lab under test-lab/. It spins up a Postgres + WireMock stack via Docker Compose, loads fixtures, and gives you a deterministic sandbox to run pipelines against — no calls to real external services.

Why it exists

--dry-run still hits the real input (its source API, your database). That's great for a one-off smoke test, less great for:

• CI that needs to be deterministic
• Pipelines that aren't built yet (the source doesn't exist)
• Reproducing a bug with a specific record shape

The test lab solves all three by mocking the source and the destination behind URLs your pipelines can point at.

Layout

cannectors/test-lab/
├── docker-compose.yml         postgres + wiremock
├── pipelines/                 example pipelines pointing at the stack
├── scenarios/                 named scenarios with expected outputs
├── wiremock/__files/          response bodies served by WireMock
├── wiremock/mappings/         WireMock request → response rules
├── postgres/                  init SQL for the source/dest databases
└── run.py                     scenario runner

Bringing the stack up

cd cannectors
make test-lab-up      # docker compose up -d, waits for healthchecks

After ~10 seconds, WireMock is at http://localhost:8080 and Postgres is at postgres://lab:lab@localhost:5432/lab.

Running a scenario

make test-lab-run     # runs every scenario under test-lab/scenarios/

Each scenario:

1. Resets WireMock mappings and Postgres tables to a known state.
2. Runs a specific pipeline from test-lab/pipelines/.
3. Checks the result against expected output (records in Postgres, requests in WireMock's journal).

The runner is test-lab/run.py — a Python script, ~500 lines. Reading the scenarios is the fastest way to learn the patterns.

Tearing it down

make test-lab-down    # docker compose down -v

The -v flag drops volumes too, so the next make test-lab-up starts clean.

CI usage

In CI, run the lab as part of the integration step:

- run: make test-lab-up
- run: make test-lab-run
- run: make test-lab-down
  if: always()

The compose stack starts under 15s on a warm runner. Total scenario time is a few seconds.

See also