directory-api-client

directory-api-client

Great API client.

Installation

$ pip install directory-api-client

The api client expects the following settings:

Setting	Notes
DIRECTORY_API_CLIENT_BASE_URL
DIRECTORY_API_CLIENT_API_KEY	Unique to client. Retrieved during the on-boarding process.
DIRECTORY_API_CLIENT_SENDER_ID	Unique to client. Retrieved during the on-boarding process.
DIRECTORY_API_CLIENT_DEFAULT_TIMEOUT

The following directory client core settings also apply to directory cms client:

Setting	Notes
DIRECTORY_CLIENT_CORE_CACHE_EXPIRE_SECONDS	Duration to store the retrieved content in the cache.
DIRECTORY_CLIENT_CORE_CACHE_LOG_THROTTLING_SECONDS	Duration to throttle log events for a given url for.

And the caching expects the following key in CACHES setting: api_fallback

Once that is done the API client can be used:

from directory_api_client.client import api_client

Development

$ git clone https://github.com/uktrade/directory-api-client
$ cd directory-api-client
$ [create virtual environment and activate]
$ make test_requirements

Use make checks to validate the codebase with black and isort, in dry-run mode Use make autoformat to run black and isort in file-updating mode

Fallback cache

Where feasible the response is cached to the client's fallback cache. This allows retrieval later if API returns non successful response or times out.

When enabling the fallback cache on a handler make sure that the request's querystring or url path are unique per user, otherwise the User B's details could be leaked User A.

For example, api_client.company.profile_retrieve looks up the company for the authenticated user. The authentication header is not used when generating the cache key for the response. This means for that endpoint the querystring and url are the same for all users, so the cache key would therefore also be the same for all users. This means if API was down then all users would see the company details for the last user to successfully retrieve their company.

Publish to PyPI

The package should be published to PyPI on merge to master. If you need to do it locally then get the credentials from rattic and add the environment variables to your host machine:

Setting
DIRECTORY_PYPI_USERNAME
DIRECTORY_PYPI_PASSWORD

Then run the following command:

$ make publish