hyp_python_client
Python API client for Hyp.
Installation
pip install hyp-python-clientUsage
The most basic usage of the client is to call the Hyp API endpoints for assignment of a participant into an experiment variant and for conversion of a participant in a particular experiment:
from hyp_client.v1 import HypClient
# This is a real access token for a dummy data account on Hyp. Feel free to use
# it to test out the API.
client = HypClient(access_token="PRODUCTION/HYP/5ab8d3d8-6eca-4e11-9203-1b64faea1f33")
client.assignment(participant_id="fuzzybear", experiment_id=8)
# {'payload': {'variant_id': 18, 'variant_name': 'v2'}, 'message': 'success', 'status_code': 200}
client.conversion(participant_id="fuzzybear", experiment_id=8)
# {'payload': {'converted': True}, 'message': 'success', 'status_code': 200}
# No experiment found
client.assignment(participant_id="fuzzybear", experiment_id=13)
# {'payload': '', 'message': 'No experiment with ID 13 was found.', 'status_code': 404}
client.conversion(participant_id="fuzzybear", experiment_id=13)
# {'payload': '', 'message': 'No variant assignment for participant fuzzybear in experiment 8 was found. Participants must be assigned to a variant before conversion can be recorded.', 'status_code': 404}The client also provides convenience methods for the common task of calling the API, parsing the response, and handling errors. For example in your application you could write code like the following:
from hyp_client.v1 import HypClient
client = HypClient("PRODUCTION/HYP/5ab8d3d8-6eca-4e11-9203-1b64faea1f33")
response = client.assignment(participant_id="fuzzybear", experiment_id=8)
variant = "Red button color"
if response["message"] == "success":
variant = response["payload"]["variant_name"]To avoid this boilerplate you can use the try_assignment and try_conversion
methods:
variant = self.client.try_assignment(participant_id="fuzzybear", experiment_id=8, fallback="some variant name")
# API call was successful, `variant` is "v2"
variant = self.client.try_assignment(participant_id="fuzzybear", experiment_id=13, fallback="some variant name")
# API call was unsuccessful because there is no experiment with ID 13.
# `variant` is "some variant name"
converted = self.client.try_conversion(participant_id="fuzzybear", experiment_id=8)
# API call was successful, `converted` is True
converted = self.client.try_conversion(participant_id="fuzzybear", experiment_id=13)
# API call was unsuccessful both because there is no experiment with ID 13 and
# there is no assignment for "fuzzybear" in that experiment.
# `converted` is False.These methods will call the API and return the relevant result if the API call
was successful. In the case of try_assignment the result is the name of the
variant. In the case of try_conversion the result is the boolean True, confirming
that the participant was marked as converted for that experiment.
If the API call returns a non-200 response code for some reason try_assignment
will return the provided fallback and try_conversion will return False.
Regardless of whether or not the API call is successful, these methods also log
out info about what has happened. They use a logger named "hyp_python_client"
and log out at the info level if the API call is successful and at the warning
level if the call was unsuccessful. The log messages look like this:
variant = self.client.try_assignment(participant_id="fuzzybear", experiment_id=8, fallback="some variant name")
# => Successfully got assignment for participant fuzzybear in experiment 8.
variant = self.client.try_assignment(participant_id="fuzzybear", experiment_id=13, fallback="some variant name")
# => Failed to get assignment for participant fuzzybear in experiment 13 Returning fallback some variant name.
converted = self.client.try_conversion(participant_id="fuzzybear", experiment_id=8)
# => Successfully converted participant fuzzybear in experiment 8.
converted = self.client.try_conversion(participant_id="fuzzybear", experiment_id=13)
# => Failed to convert participant fuzzybear in experiment 13.Developing and running the tests
We recommend using a virtual environment to isolate your installed packages.
python -m pip install -r requirements.txt
python -m unittest discover tests
![dependabot[bot]](https://avatars.githubusercontent.com/u/49699333?size=120)