Python deserialization fails for union types

[Thoralf-M]

Description

Deserialization fails for union types if the data is not always matching the first type.
For example address: Ed25519Address | AliasAddress | NFTAddress only works if the type matches the Ed25519Address variant, for AliasAddress or NFTAddress an error is returned.

Didn't found a solution yet, but maybe can be done with something similar to this

    # https://github.com/lidatong/dataclasses-json/issues/303
    address: Ed25519Address | AliasAddress | NFTAddress = field(
        metadata=config(
            decoder=lambda d: 
            match d["type"]:
                    case AddressType.Ed25519:
                        Ed25519Address.from_dict(d)
                    case ...

If not probably a custom _from_dict_custom(), similar to

iota-sdk/bindings/python/iota_sdk/types/transaction_options.py

Line 31 in 1add7b7

def _to_dict_custom(config: Dict[str, any]) -> Dict[str, any]:

can be used

https://github.com/lidatong/dataclasses-json/pull/478 might help

Motivation

Make deserialization work correct

Are you planning to do it yourself in a pull request?

No.

[thibault-martinez]

Only happening in 2.0?

[Thoralf-M]

Yes, most probably introduced in 80a5667

[thibault-martinez]

Closed by #1335