Literal type usage

[martin-thoma]

The Literal type was introduced with PEP 586 in Python 3.8 and can be used with Pydantic. With typing-extensions the Literal type can also be used in Python 3.6+.

Would it be ok to introduce Literal?

[overcat]

Hi, I just checked out PEP-586, can you tell me where you plan to introduce it?

Thank you.

[martin-thoma]

Here: https://github.com/StellarCN/stellar-model/blob/main/stellar_model/model/horizon/operations.py#L133-L134

asset_type: Literal["native", "credit_alphanum4", "credit_alphanum12"]
type: Literal["payment"]
type_i: Literal[1]

I'm uncertain why type and type_i are not part of the model, but at least for asset_type it makes sense.

[overcat]

I think using Enum is a better choice here.

from enum import Enum

from pydantic import BaseModel

class AssetType(Enum):
    native = "native"
    credit_alphanum4 = "credit_alphanum4"
    credit_alphanum12 = "credit_alphanum12"


class DemoOp(BaseModel):
    asset_type: AssetType

demo_op = DemoOp.parse_obj({'asset_type': 'credit_alphanum4'})
assert demo_op.asset_type == AssetType.credit_alphanum4
# If we use Literal here, the user still has to compare strings here
# assert demo_op.asset_type == 'credit_alphanum4'