Testing

Test Categories

Dry-Run Tests

Test tools without the rendering engine by passing cmd=None. Every tool should handle this:

def test_my_tool_dry_run():
    from tools.mygroup.mytool import run
    result = run(None, "arg1", "arg2")
    assert result is not None

Mock Tests

Use the mock_cmd fixture for testing rendering interactions:

def test_my_tool_mock(mock_cmd):
    from tools.mygroup.mytool import run
    result = run(mock_cmd, "protein")
    mock_cmd.show.assert_called_once()

Integration Tests

Require the full rendering engine — auto-skipped in CI:

@pytest.mark.integration
def test_my_tool_real():
    import codemol
    # ... test with full rendering engine

Key Fixtures

Defined in tests/conftest.py:

Fixture	Scope	Description
all_tools	Session	Full tool registry via discover_tools()
mock_cmd	Function	MagicMock with standard rendering return values
reset_measurement_registry	Function (autouse)	Clears measurement registry
reset_selection_overrides	Function (autouse)	Clears alias overrides

mock_cmd Default Returns

cmd.count_atoms.return_value = 10
cmd.distance.return_value = 3.5
cmd.angle.return_value = 109.5
cmd.get_dihedral.return_value = -60.0
cmd.get_names.return_value = []
cmd.get_object_list.return_value = ["test_obj"]
cmd.get_model.return_value = MagicMock(atom=[])

Running Tests

# All non-integration tests
pytest tests/ -v

# Specific test file
pytest tests/test_tools.py -v

# Specific test
pytest tests/test_tools.py::test_distance -v

# With coverage
pytest tests/ --cov=codemol --cov=tools

# Integration tests only (needs rendering engine)
pytest tests/ -v -m integration

Test Data

Test data lives in tests/:

tests/
├── conftest.py           # Fixtures
├── docking/
│   ├── docking_amazon/   # GOLD docking results
│   └── docking_brazil/   # GOLD docking results
└── projects/             # Sample project files

Writing Good Tests

Test the contract, not the implementation

Test that run() returns the right string and calls the right rendering methods — don't test internal helper functions.

Use dry-run for smoke tests

A dry-run test catches import errors, missing dependencies, and argument mismatches without needing the rendering engine.

Reset shared state

The autouse fixtures handle measurement registry and selection overrides. If your tool uses other shared state, add a reset fixture.