I have a yaml file full of Kedro parameters like:

data_extraction:
    param1: 5
    param2: "hello"

My nodes then take in inputs like

def mynode(data_extraction_params: dict): ...

and my pipeline does

node(func=mynode, inputs=["params:data_extraction"], ...

I want to use Pydantic models to specify for each node what the inputs are, kind of like structs in Typescript. So essentially I'd like to do:

class dataExtractionParams(BaseModel):
    param1: int
    param2: str
def mynode(data_extraction_params: dataExtractionParams): ...

How can I most easily do this? That way in the function I can use the model rather than a dict.

The closest thing to this that I've seen is

typing.Unpack

- see here

Please see this thread above: https://kedro-org.slack.com/archives/C03RKP2LW64/p1724738166844939

Apart from that, if you want to pass a native dataclass as an argument to

node

, it's not supported. there is a on-going discussion here: https://github.com/kedro-org/kedro/discussions/3808