oamg/leapp-repository/leapp-unit-tests

Writing Unit Tests

File placement

Tests for actors

Tests live in the tests directory under an actor:

repos/system_upgrade/<leapp_repo_dir>/actors/<actor_name>/
├── actor.py
├── libraries/<actor_name>.py
└── tests/test_<actor_name>.py      # <-- tests go here

Tests for shared libraries

Tests for shared libraries are in the tests sub-directory next to the shared library, like:

repos/system_upgrade/<leapp_repo_dir>/
├── <libA>.py
├── tests/test_<libA>.py      # <-- tests for libA go here
├── <pathB>/<libB>.py
└── <pathB>/tests/test_<libB>.py      # <-- tests for libB go here

Key testing utilities

As mocked objects, use functions and classes present in leapp.libraries.common.testutils - especially:

Standard test pattern

Test the library, not the actor class directly:

from leapp.libraries.actor import <actor_lib>
from leapp.libraries.common.testutils import CurrentActorMocked, produce_mocked
from leapp.libraries.stdlib import api
from leapp.models import <ConsumedModel>, <ProducedModel>
def test_<scenario>(monkeypatch):
    # 1. Mock current_actor context
    monkeypatch.setattr(api, 'current_actor', CurrentActorMocked())
    # 2. Mock consumed messages
    monkeypatch.setattr(api, 'consume', lambda x: iter([<ConsumedModel>(...)]))
    # 3. Mock produce
    monkeypatch.setattr(api, 'produce', produce_mocked())
    # 4. Call library function
    <actor_lib>.process()
    # 5. Assert on produced messages
    assert api.produce.called
    assert isinstance(api.produce.model_instances[0], <ProducedModel>)

Rules

• Test the library, not actor.py — the process() method should be a thin wrapper.
• Pass consumed messages via CurrentActorMocked(msgs=[...]).
• Mock file I/O instead of creating temp files when possible. Use the leapp_tmpdir fixture when tempfiles are needed.
• Use monkeypatch.setattr over unittest.mock.patch — monkeypatch is the project convention.
• Cover at least: happy path, edge/error cases, and inhibitor conditions if applicable.
• If you intentionally skip a test scenario, add a comment explaining why.
• Use consistent string quoting within a file.
• Use pytest.mark.parametrize where possible.
• Never use type() when creating mock classes.
• Use module-level constants with "_" prefix.

• Name test functions clearly after the behavior: test_inhibits_when_fips_enabled not test_fips_1.
• Look at existing tests in repos/system_upgrade/common/actors/ for project conventions before writing new ones.

Running tests

See skills/leapp-test-runner/ for full commands. Quick single-actor run:

ACTOR=<actor_name> make dev_test_no_lint