Skip to content

README.md

Latest commit

 

History

History
93 lines (66 loc) · 3.37 KB

README.md

File metadata and controls

93 lines (66 loc) · 3.37 KB
Raw

Integration / end-to-end tests

These tests run against a real test tenant using credentials stored securely in your OS keychain. They are opt-in and self-skip when no test environment is configured, so the default pytest run stays green without them.

One-time setup

1. Install keyring support

pip install -e .[secrets]

2. Store your test-tenant credentials in the OS keychain

Use a dedicated tabcloud-test service name to keep test creds separate from prod:

tabcloud-secrets set tabcloud-test extractor-pat     # TCM PAT for the test tenant
tabcloud-secrets set tabcloud-test publish-pat       # only if testing the publish path
tabcloud-secrets doctor                              # verify the backend works

(You'll be prompted for each value; nothing is echoed to the terminal.)

3. Create your test config

cp config/config.test.ini.example config/config.test.ini

Then edit config/config.test.ini and fill in your test tenant's server_url, token_name, and site_uris. The token_secret fields already reference the keyring entries you created above — no secrets go in the file.

config/config.test.ini is gitignored (it names your real tenant); only the .example template is committed.

Running

# Default: unit tests only — integration tests are excluded automatically.
pytest

# Run ONLY integration tests (requires the setup above):
pytest -m integration -v

# Just the fast connectivity smoke tests (verify auth works):
pytest -m integration -k connectivity -v

# Full end-to-end pipeline (extractor → transformer → loader):
pytest -m integration -k e2e -v -s

# Run EVERYTHING (unit + integration), ignoring the default filter:
pytest -m ""

# Point at a different config (e.g. a second tenant):
TABCLOUD_TEST_CONFIG=~/other_tenant.ini pytest -m integration

What's here

File Purpose
conftest.py Fixtures: extractor_credentials (resolves the PAT via common.secrets), temp_storage, runner_env. All derive from test_config, so they skip when no config is present.
test_connectivity.py Fast smoke tests — authenticate and list sites against the test tenant.
test_e2e_pipeline.py Drives python -m runner --config config.test.ini as a subprocess; asserts a .hyper is produced and that the secret never leaks into output.
mappings/ Custom transformer + loader YAML scoped to the test tenant (small entity set, test-specific output names). Edit these for new use cases.

Writing new E2E use cases

  1. Add or edit a transformer mapping in mappings/ (mirror the production ones in mappings/transformer/).
  2. Reference it from mappings/test_loader_bundle.yml (a new tables: entry) or create a new loader bundle.
  3. Point config/config.test.ini [mappings] at your new files.
  4. Add a test in test_e2e_pipeline.py (or a new test_*.py) that runs the runner and asserts on the output. Mark it with pytestmark = pytest.mark.integration.

CI / headless servers

Keyring is usually unavailable on CI runners. Switch the references in your CI config to environment variables and set them as CI secrets:

# config.test.ini (CI variant)
[extractor]
token_secret = env:TABCLOUD_TEST_EXTRACTOR_PAT

The same common.secrets.resolve_secret() call resolves env: and keyring: identically, so no test code changes are needed.