Incorrect schema for non-integer Primary keys using PrimaryKeyRelatedField

[BeryJu]

With the default PrimaryKeyRelatedFieldConverter (https://github.com/maykinmedia/drf-jsonschema-serializer/blob/main/drf_jsonschema_serializer/converters.py#L267), using PrimaryKeyRelatedField with a foreign key field which does not use Integer primary keys creates an incorrect schema with {"type": "integer"}.

A potential fix could look something like this, however I haven't found a good method to test for all string-based fields:

@converter
class PrimaryKeyRelatedFieldConverter:
    """Custom primary key field converter which is aware of non-integer based PKs

    This is not an exhaustive fix for other non-int PKs, however in authentik we either
    use UUIDs or ints"""

    field_class = PrimaryKeyRelatedField

    def convert(self, field: PrimaryKeyRelatedField):
        model: Model = field.queryset.model
        pk_field = model._meta.pk
        if isinstance(pk_field, fields.UUIDField):
            return {"type": "string", "format": "uuid"}
        return {"type": "integer"}

[sergei-maertens]

I think you're on the right path - but rather than doing isinstance checks, you can probably grab the PK (model) field type, put that through the DRF map of model -> serializer field type and then use that result to look up the "new" field type in the converter registry again. That way you have something that is dynamic and automatic.

There may be some edge cases still, but those can be solved with subclassing in the project where you create the edge cases. What do you think?

[BeryJu]

Oh yeah thats a good idea, I forgot that such a map existed. Let me change the example fix to that and open a PR with that