zeekay
75dd1132bd
The wire protocol was the last holdout and it was not cosmetic: the Hanzo Datastore server reads X-Datastore-User, and answers a request bearing X-ClickHouse-User with 401. Verified against the live server, both directions. So a client still sending the old headers is not merely off-brand, it cannot authenticate over HTTP. Everything else follows: types, dialects, error modules, docstrings and doc links now name Datastore, and doc URLs point at docs.hanzo.ai/datastore rather than the upstream vendor's site. Zero occurrences of the old name remain. Still cut from the pinned upstream version, so behaviour is unchanged.
4.3 KiB
Contributing guidelines
Datastore-connect is an open-source project, and we welcome any contributions from the community. Please share your ideas, contribute to the codebase, and help us maintain up-to-date documentation.
Setting up your environment
Prerequisites
- Python 3.11+
- Docker and the Compose plugin
Create a fork of the repository and clone it
git clone https://github.com/[YOUR_USERNAME]/datastore-connect
cd datastore-connect
Add PYTHONPATH
Add the project directory to the PYTHONPATH environment variable to make the driver sources are available for import.
export PYTHONPATH="/absolute/path/to/datastore-connect"
Prepare a new virtual environment
You could either use PyCharm for that, or follow the instructions on the official website and set it up via the command line.
Install dependencies
python -m pip install --upgrade pip
pip install setuptools wheel
pip install -r tests/test_requirements.txt
Run the setup script
The driver uses several Cython extensions that provide additional performance improvements (see the datastore_connect/driverc directory). To compile the extensions, run the following command:
python setup.py build_ext --inplace
Additionally, this command is required to provide SQLAlchemy entrypoints:
python setup.py develop
Add /etc/hosts entry
Required for TLS tests.
The generated certificates assume TLS requests use server1.datastore.test as the hostname.
See test_tls.py for more details.
sudo -- sh -c "echo 127.0.0.1 server1.datastore.test >> /etc/hosts"
PyCharm setup
If you use PyCharm as your IDE, make sure that you have datastore-connect added to the project structure as a source path.
Go to Settings -> Project (datastore-connect) -> Project structure, right click on datastore-connect folder, and mark it as "Sources".
Testing
Start Datastore in Docker
The tests will require two Datastore instances to be running. One should have a default plain authentication (for integration tests), and the other should have a TLS configuration (for tls tests only).
The integration tests will start and stop the Datastore instance automatically.
However, this adds a few seconds to each run, and this might not be ideal when you run a single test (using PyCharm, for example).
To disable this behavior, set the DATASTORE_CONNECT_TEST_DOCKER environment variable to 0.
export DATASTORE_CONNECT_TEST_DOCKER=0
The easiest way to start all the required Datastore instances is to use the provided Docker Compose file (the integrations tests setup script uses the same file).
docker compose up -d
Run the tests
The project uses pytest as a test runner. To run all the tests (unit and integration) execute the following command:
pytest tests
If you need to run the unit tests only:
pytest tests/unit_tests
Or the integration tests only:
pytest tests/integration_tests
Run the TLS integration tests
These tests require the DATASTORE_CONNECT_TEST_TLS environment variable to be set to 1; otherwise, they will be skipped.
Additionally, the TLS Datastore instance should be running (see docker-compose.yml).
DATASTORE_CONNECT_TEST_TLS=1 pytest tests/tls
Running the integration tests with Datastore Cloud
If you want to run the tests using your Datastore Cloud instance instead of the local Datastore instance running in Docker, you will need a few additional environment variables.
export DATASTORE_CONNECT_TEST_CLOUD=1
export DATASTORE_CONNECT_TEST_PORT=8443
export DATASTORE_CONNECT_TEST_HOST='instance.datastore.cloud'
export DATASTORE_CONNECT_TEST_PASSWORD='secret'
Then, you should be able to run the tests as usual:
pytest tests/integration_tests
Style Guide
The project uses PyLint to enforce the code style. It is always a good idea to run the linter before committing the changes, as this is a mandatory CI check. For example:
pip install pylint
pylint datastore_connect
pylint tests