import numpy
import marshmallow
[docs]class FeatureValuesValidator(object):
"""Class implementing validator for the feature values"""
[docs] @classmethod
def validate(cls, values):
"""
Validates the feature values.
:param values: values to be validated
:type values: Any
:return: validated values
:rtype: Any
"""
# Validate the feature values
if values is None:
raise marshmallow.ValidationError(f"Missing data for required field.", "features.values")
if not isinstance(values, numpy.ndarray):
raise marshmallow.ValidationError(f"Not a valid numpy.array.", "features.values")
# Ensure the subjects-dimension for a rank-one array
values = numpy.atleast_2d(values)
# Return the validated feature values
return values
[docs]class FeatureLabelsValidator(object):
"""Class implementing validator for the feature labels"""
[docs] @classmethod
def validate(cls, labels, values):
"""
Validates the feature labels.
:param labels: labels to be validated
:type labels: Any
:param values: values to be referenced
:type values: Any
:return: validated values
:rtype: Any
"""
# Validate the feature labels
if not isinstance(labels, (tuple, list)):
raise marshmallow.ValidationError(f"Not a valid (tuple, list).", "features.labels")
if not (len(labels) == values.shape[-1]):
raise marshmallow.ValidationError(
f"Not a valid shape (must match the values). The API expects the same number of labels "
f"as the shape of the last dimension of the features (for more information, check the "
f"documentation or the docstring for the featurizer resource (resources.featurizer).",
"features.labels")
# Return the validated feature labels
return labels