Pylance inference on pandera's Dataframe

[Cameronwood611]

I have a simple block of code where I use Pylance for static type checking.

Side note: I'm attempting to be "pydantic" here and not have to use the Series key but I don't think there's anyway around it, please let me know if so.

from pathlib import Path
from typing import cast
import pandas as pd
import pandera as pa
from pandera.typing import DataFrame, Series


class CarSchema(pa.SchemaModel):
    """ Schema for car. """

    make: Series[str]
    model: str

Here, I expect to have Series[str] for cars.make, but I get Series[Unknown] instead:

@pa.check_types(lazy=True)
def get_cars(path: Path) -> DataFrame[CarSchema]:
    df = pd.read_csv(path, dtype=object)
    return cast(DataFrame[CarSchema], df)
cars = get_cars(Path("./"))
cars.make # ==> Series[Unknown]

I could try the following to get Series[str] but now I'm not treating it like a DataFrame (so indexing by cars["make"] will show an error) :

@pa.check_types(lazy=True)
def get_cars(path: Path) -> CarSchema:
    df = pd.read_csv(path, dtype=object)
    return cast(CarSchema, df)
cars = get_cars(Path("./cars.csv"))
cars.make # ==> Series[str]

I understand that pandera isn't completely compatible with static type checkers yet but I did want to ask if there was anything I could do about this?