Convert lognormal dist validation to exception-based approach

src/semeio/fmudesign/design_distributions.py

@@ -47,24 +47,29 @@ def parse_and_validate_normal_params(
     return tuple(params)
 
 
-def _check_dist_params_lognormal(dist_params: Sequence[str]) -> tuple[bool, str]:
+def parse_and_validate_lognormal_params(
+    dist_params: Sequence[int | str | float],
+) -> tuple[float, float]:
     if len(dist_params) != 2:
-        status = False
-        msg = (
+        raise ValueError(
             "Lognormal distribution must have 2 parameters, "
-            "but had " + str(len(dist_params)) + " parameters. "
+            f"but had {len(dist_params)} parameters."
         )
-    elif not (is_number(dist_params[0]) and is_number(dist_params[1])):
-        status = False
-        msg = "Parameters for lognormal distribution must be numbers. "
-    elif float(dist_params[1]) < 0:
-        status = False
-        msg = "Lognormal distribution must have stddev >= 0. "
-    else:
-        status = True
-        msg = ""
+    try:
+        mean, stddev = [float(p) for p in dist_params]
+    except (ValueError, TypeError) as e:
+        raise ValueError(
+            f"All parameters must be convertible to numbers. Got: {dist_params}"
+        ) from e
 
-    return status, msg
+    if np.any(np.isnan([mean, stddev])):
+        raise ValueError(f"Parameters cannot be NaN. Got: {[mean, stddev]}")
+
+    if stddev < 0:
+        raise ValueError(
+            f"Stddev for lognormal distribution must be >= 0. Got: {stddev}"
+        )
+    return mean, stddev
 
 
 def parse_and_validate_uniform_params(
@@ -227,7 +232,7 @@ def draw_values_normal(
 
 
 def draw_values_lognormal(
-    dist_parameters: Sequence[str],
+    dist_parameters: Sequence[int | str | float],
     numreals: int,
     rng: np.random.Generator,
     normalscoresamples: npt.NDArray[Any] | None = None,
@@ -241,12 +246,7 @@ def draw_values_lognormal(
     Returns:
         list of values
     """
-    status, msg = _check_dist_params_lognormal(dist_parameters)
-    if not status:
-        raise ValueError(msg)
-
-    mean = float(dist_parameters[0])
-    sigma = float(dist_parameters[1])
+    mean, sigma = parse_and_validate_lognormal_params(dist_parameters)
     if normalscoresamples is not None:
         values = scipy.stats.lognorm.ppf(
             scipy.stats.norm.cdf(normalscoresamples),

tests/fmudesign/test_design_distributions.py

@@ -81,20 +81,58 @@ def test_that_invalid_truncation_bounds_raise_ordering_error(self):
             dists.parse_and_validate_normal_params([0, 1, 1, 0])
 
 
-def test_check_dist_params_lognormal():
-    """Test lognormal dist param checker"""
-    assert not dists._check_dist_params_lognormal([])[0]
-    assert not dists._check_dist_params_lognormal(())[0]
+class TestLognormalDistribution:
+    def test_that_empty_list_raises_parameter_count_error(self):
+        with pytest.raises(
+            ValueError,
+            match="Lognormal distribution must have 2 parameters, but had 0 parameters.",
+        ):
+            dists.parse_and_validate_lognormal_params([])
+
+    def test_that_empty_tuple_raises_parameter_count_error(self):
+        with pytest.raises(
+            ValueError,
+            match="Lognormal distribution must have 2 parameters, but had 0 parameters.",
+        ):
+            dists.parse_and_validate_lognormal_params(())
+
+    def test_that_single_parameter_raises_count_error(self):
+        with pytest.raises(
+            ValueError,
+            match="Lognormal distribution must have 2 parameters, but had 1 parameters.",
+        ):
+            dists.parse_and_validate_lognormal_params([0])
+
+    def test_that_three_parameters_raises_count_error(self):
+        with pytest.raises(
+            ValueError,
+            match="Lognormal distribution must have 2 parameters, but had 3 parameters.",
+        ):
+            dists.parse_and_validate_lognormal_params([0, 0, 0])
 
-    assert not dists._check_dist_params_lognormal([0])[0]
-    assert not dists._check_dist_params_lognormal([0, 0, 0])[0]
+    def test_that_non_numeric_parameters_raise_conversion_error(self):
+        with pytest.raises(
+            ValueError,
+            match=r"All parameters must be convertible to numbers. Got: \['mean', 'mu'\]",
+        ):
+            dists.parse_and_validate_lognormal_params(["mean", "mu"])
 
-    assert not dists._check_dist_params_lognormal(["mean", "mu"])[0]
+    def test_that_valid_lognormal_parameters_return_float_tuple(self):
+        assert dists.parse_and_validate_lognormal_params([0, 1]) == (0.0, 1.0)
 
-    assert dists._check_dist_params_lognormal([0, 1])[0]
-    assert not dists._check_dist_params_lognormal([0, -1])[0]
+    def test_that_negative_stddev_raises_validation_error(self):
+        with pytest.raises(
+            ValueError,
+            match="Stddev for lognormal distribution must be >= 0. Got: -1.0",
+        ):
+            dists.parse_and_validate_lognormal_params([0, -1])
+
+    def test_that_zero_stddev_is_accepted(self):
+        assert dists.parse_and_validate_lognormal_params([0, 0]) == (0.0, 0.0)
 
-    assert dists._check_dist_params_lognormal([0, 0])[0]  # edge case
+    def test_that_nan_parameter_raises_error(self):
+        with pytest.raises(ValueError, match="Parameters cannot be NaN"):
+            dists.parse_and_validate_lognormal_params([0, np.nan])
 
 
 class TestUniformDistribution: