Generic Types

Python generic types are supported, but require to_python_value and to_preserialization_data implementations similar to custom types. The class's CONVERTERS attribute will have the appropriate datafile converters placed in it for the specified generic types. You should use these in your implementation to convert the marshalled data.

from typing import Generic, List, TypeVar

from datafiles import Missing, converters, datafile
from datafiles.utils import dedent


S = TypeVar("S")
T = TypeVar("T")


class Pair(Generic[S, T], converters.Converter):
    first: S
    second: T

    def __init__(self, first: S, second: T) -> None:
        self.first = first
        self.second = second

    @classmethod
    def to_python_value(cls, deserialized_data, *, target_object=None):
        paired = zip(cls.CONVERTERS, deserialized_data)
        values = [convert.to_python_value(val) for convert, val in paired]
        return cls(*values)

    @classmethod
    def to_preserialization_data(cls, python_value, *, default_to_skip=None):
        values = [python_value.first, python_value.second]
        paired = zip(cls.CONVERTERS, values)
        return [
            convert.to_preserialization_data(val)
            for convert, val in paired
        ]

@datafile("sample.yml")
class Dictish:
    contents: List[Pair[str, converters.Number]]

which can be constructed like so:

dictish = Dictish([Pair("a", 1), Pair("pi", 3.14)])

to save this sample.yml file:

contents:
  -   - a
      - 1
  -   - pi
      - 3.14

An example of using generic types can be found in this notebook.