Working with Tests

These tests use Pytest.

Running and Using Tests

These tests assume that openfhe-python is installed in the current python environment, which you can check by importing openfhe.

python -c "__import__('openfhe')"

and that the pytest package is installed, either through pip or by installing python3-pytest in the operating system package manager.

Specific to the OpenFHE unit tests

Some tests are marked with @pytest.mark.long if they are not meant to run on Github Actions. Run these locally with:

pytest --run-long
pytest --run-all

General Pytest usage

This is a quick reminder of pytest's features. To test a particular file:

pytest test_particular_file.py

Test all functions matching a name. For instance, this would pick up test_add_two_numbers:

pytest -k add

As a reminder, pytest can be helpful for debugging. This command-line option shows debug output from logging statements.

pytest --log-cli-level=debug

If a test is failing, pytest can drop into the debugger when an exception happens.

pytest --pdb

Guidelines for Writing Tests

Mark long-running tests with long -- These tests run with default settings on Github Actions, which can be underpowered, so there is a way to mark tests that can be run by hand or on other automation servers.

@pytest.mark.long
def test_ckks_large_context():
    assert true

The goal is for the Github Actions tests to reassure a committer that they have not broken the Python wrapper.

Import OpenFHE as fhe -- Unit tests tend to use more imports than most code, for instance JSON, which conflicts with an OpenFHE name, so qualify imports in the tests.

import openfhe as fhe

def test_something():
    parameters = fhe.CCParamsCKKSRNS()

Use logging instead of print statements -- Pytest has nice support for making logging statements visible, in the case that you are using tests for debugging.

import logging

LOGGER = logging.getLogger("test_file_name")

def test_something():
    arg = 3
    LOGGER.debug("My message has an argument %s", arg)