Using TypeDict with djangorestframework_camel_case sets required fields to optional

[saschametz]

Hi, when using TypeDict in combination with the drf_spectacular.contrib.djangorestframework_camel_case.camelize_serializer_fields hook all fields that are converted from snake_case to camelCase are set to optional.

Serializer:

class RankData(TypedDict):
    value: int
    created_at: str
    name: str
    rank_by_value: bool


class RankSerializer(serializers.Serializer):
    rank_data = serializers.SerializerMethodField()

    def get_rank_data(self) -> RankData:
        return RankData(
            value=290,
            created_at="2022-01-01",
            name="Checkpoint A",
            rank_by_value=False,
        )

Settings:

SPECTACULAR_SETTINGS = {
    "POSTPROCESSING_HOOKS": [
        "drf_spectacular.hooks.postprocess_schema_enums",
        "drf_spectacular.contrib.djangorestframework_camel_case.camelize_serializer_fields",
    ],
}

This produces the following schema:

-> All names that are converted from snake_case to camelCase are set to optional.

Removing "drf_spectacular.contrib.djangorestframework_camel_case.camelize_serializer_fields" from the settings produces the correct schema (all fields are required):

[tfranzel]

Looks like the string replacement does not account for nested objects properly. It goes through the nested props properly, but it fails on nested required list.

[saschametz]

Thanks for the quick fix @tfranzel
I can confirm it's working now.

[saschametz]

I just noticed one small issue. After adding a nested child_rank it doesn't convert to camelCase anymore. That's only the case when using List[ChildRankData], but not ChildRankData (without a list).

Serializer:

class ChildRankData(TypedDict):
    value: int
    created_at: str


class RankData(TypedDict):
    value: int
    created_at: str
    name: str
    rank_by_value: bool
    child_rank: List[ChildRankData]


class RankSerializer(serializers.Serializer):
    rank_data = serializers.SerializerMethodField()

    def get_rank_data(self) -> RankData:
        return RankData(
            value=290,
            created_at="2022-01-01",
            name="Checkpoint A",
            rank_by_value=False,
            child_rank=[ChildRankData(value=123, created_at="022-02-02")],
        )

Schema:

[tfranzel]

ahh yes, I forgot about that case, since we now do translation manually.

[tfranzel]

did a drive-by bugfix that came to my attention. this should work now as expected.

[saschametz]

Everything works perfectly now 👍