fix noise formula check (#49)

* fixups

* fixup scientific notation

* even better

* Update lint.py

Author: FFroehlich

petab/lint.py

@@ -526,6 +526,24 @@ def assert_parameter_estimate_is_boolean(parameter_df: pd.DataFrame) -> None:
                 f"Expected 0 or 1 but got {estimate} in {ESTIMATE} column.")
 
 
+def is_scalar_float(x: Any):
+    """
+    Checks whether input is a number or can be transformed into a number
+    via float
+    :param x:
+        input
+    :return:
+        True if is or can be converted to number, False otherwise.
+    """
+    if isinstance(x, numbers.Number):
+        return True
+    try:
+        float(x)
+        return True
+    except (ValueError, TypeError):
+        return False
+
+
 def measurement_table_has_timepoint_specific_mappings(
         measurement_df: pd.DataFrame,
         allow_scalar_numeric_noise_parameters: bool = False,
@@ -553,23 +571,6 @@ def measurement_table_has_timepoint_specific_mappings(
     # since we edit it, copy it first
     measurement_df = copy.deepcopy(measurement_df)
 
-    def is_scalar_float(x: Any):
-        """
-        Checks whether input is a number or can be transformed into a number
-        via float
-        :param x:
-            input
-        :return:
-            True if is or can be converted to number, False otherwise.
-        """
-        if isinstance(x, numbers.Number):
-            return True
-        try:
-            float(x)
-            return True
-        except (ValueError, TypeError):
-            return False
-
     # mask numeric values
     for col, allow_scalar_numeric in [
         (OBSERVABLE_PARAMETERS, allow_scalar_numeric_observable_parameters),
@@ -622,7 +623,8 @@ def observable_table_has_nontrivial_noise_formula(
     """
 
     return not observable_df[NOISE_FORMULA].apply(
-        lambda x: re.match(r'^[\w_\.]+$', x) is not None
+        lambda x: is_scalar_float(x) or
+        re.match(r'^[\w]+$', str(x)) is not None
     ).all()
 
 

tests/test_lint.py

@@ -86,11 +86,14 @@ def test_observable_table_has_nontrivial_noise_formula():
     # Ensure we fail if we have nontrivial noise formulas
 
     observable_df = pd.DataFrame(data={
-        OBSERVABLE_ID: ['0obsPar1noisePar', '2obsPar0noisePar'],
+        OBSERVABLE_ID: ['0obsPar1noisePar', '2obsPar0noisePar',
+                        '3obsPar0noisePar'],
         OBSERVABLE_FORMULA: ['1.0',
+                             '1.0',
                              '1.0'],
         NOISE_FORMULA: ['noiseParameter1_0obsPar1noisePar + 3.0',
-                        '1.0']
+                        1e18,
+                        '1e18']
     })
 
     assert lint.observable_table_has_nontrivial_noise_formula(observable_df)\