Warning

Figuring out the correct field type from an untagged dump is a matter of guesswork and can be error-prone. A more explicit approach like https://github.com/Bachmann1234/marshmallow-polyfield/ is better. This project remains here to avoid breaking existing software.

Usage

To use marshmallow-union in a project import marshmallow_union.Union and pass it a list of marshmallow.fields.Field instances.

When deserializing, each field will be tried in sequence until one succeeds, not raising marshmallow.exceptions.ValidationError. If none of them is successful, marshmallow.exceptions.ValidationError is raised.

When serializing, each field will be tried in sequence until one succeeds, not raising any exception. If none of them is successful, marshmallow_union.ExceptionGroup is raised, containing the values of the raised exceptions.

See the API reference for more details.

import marshmallow
import marshmallow_union


class PersonSchema(marshmallow.Schema):
    name = marshmallow.fields.String()
    number_or_numbers = marshmallow_union.Union(
        [
            marshmallow.fields.List(marshmallow.fields.Integer()),
            marshmallow.fields.Integer(),
        ],
            reverse_serialize_candidates=True,
    )



input_data = {"name": "Alice", "number_or_numbers": 25}
schema = PersonSchema()

loaded = schema.load(input_data)
dumped = schema.dump(loaded)

assert dumped == input_data