from mtdata.forecast import forecast_registry as fr
def test_check_requirements_marks_mlf_rf_unavailable_and_normalizes_sklearn(monkeypatch):
checked_names = []
def fake_find_spec(name):
checked_names.append(name)
return object() if name == "sklearn" else None
monkeypatch.setattr(fr, "_MLF_AVAILABLE", False)
monkeypatch.setattr(fr._importlib_util, "find_spec", fake_find_spec)
available, reqs = fr._check_requirements("mlf_rf", ["scikit-learn>=1.6"])
assert available is False
assert "mlforecast, scikit-learn" in reqs
assert "sklearn" in checked_names
def test_check_requirements_strips_versions_and_maps_python_dotenv(monkeypatch):
checked_names = []
def fake_find_spec(name):
checked_names.append(name)
return object()
monkeypatch.setattr(fr._importlib_util, "find_spec", fake_find_spec)
available, reqs = fr._check_requirements(
"custom_method",
["python-dotenv>=1.0.0", "numpy==2.0.0"],
)
assert available is True
assert reqs == ["python-dotenv>=1.0.0", "numpy==2.0.0"]
assert "dotenv" in checked_names
assert "numpy" in checked_names
def test_get_forecast_methods_data_assembles_categories_and_skips_broken(monkeypatch):
class GoodMethod:
"""Good method summary."""
supports_features = {"price": True, "return": True, "volatility": False, "ci": True}
required_packages = ["numpy>=1.0"]
PARAMS = [{"name": "window", "type": "int"}]
category = "Classic"
class MissingParamsMethod:
supports_features = None
required_packages = []
PARAMS = {"invalid": "shape"}
category = None
class BrokenMethod:
def __init__(self):
raise RuntimeError("construction failed")
class FakeRegistry:
@staticmethod
def get_all_method_names():
return ["good", "broken", "ensemble", "missing_params"]
@staticmethod
def get_class(name):
mapping = {
"good": GoodMethod,
"broken": BrokenMethod,
"missing_params": MissingParamsMethod,
}
return mapping[name]
def fake_check_requirements(method, requires):
if method == "missing_params":
return False, ["some_pkg"]
return True, list(requires)
monkeypatch.setattr(fr, "ForecastRegistry", FakeRegistry)
monkeypatch.setattr(fr, "_ensure_registry_loaded", lambda: None)
monkeypatch.setattr(fr, "_check_requirements", fake_check_requirements)
data = fr.get_forecast_methods_data()
methods = {m["method"]: m for m in data["methods"]}
assert data["total"] == 3
assert "broken" not in methods
assert methods["good"]["description"] == "Good method summary."
assert methods["good"]["params"] == [{"name": "window", "type": "int"}]
assert methods["good"]["available"] is True
assert methods["missing_params"]["available"] is False
assert methods["missing_params"]["params"] == []
assert methods["missing_params"]["supports"] == {
"price": True,
"return": True,
"volatility": False,
"ci": False,
}
assert "good" in data["categories"]["classic"]
assert "missing_params" in data["categories"]["unknown"]
assert "ensemble" in data["categories"]["ensemble"]
