Astral ty Integration

[iot-resister]

Not sure what need's to be done, if anything, but it would great if pandera supported astral's new typechecker.

Thanks for considering it!

[cosmicBboy]

hi @iot-resister yes I've been keeping track of ty! What kind of integration would you imagine between pandera and ty? Are you thinking something similar to the mypy integration? https://pandera.readthedocs.io/en/stable/mypy_integration.html

[iot-resister]

Thanks for the quick response! Yes, I'm thinking of something similar to the mypy integration, but adapted for Ty's specific characteristics and capabilities.

The ideal would be making Pandera's core typing features would work

# This should work in both mypy and ty
from pandera.typing import DataFrame, Series

class UserSchema(pa.DataFrameModel):
    id: Series[int] = pa.Field(ge=1)
    name: Series[str] = pa.Field(str_length=(1, 50))

def process_users(df: DataFrame[UserSchema]) -> DataFrame[UserSchema]:
    return df.head(10)

• Fix import resolution issues with pandera.typing
  • getting Ty Error: Member 'DataFrame' of module 'pandera.typing' is possibly unbound)
• Ensure basic pa.Field() parameters type-check correctly
  • getting **Ty Error**: Expected 'dict[str, Any] | None', found 'tuple[Literal[1] , Literal[50]]'
• Maintain pandas method access on typed DataFrames
  • getting # Error: Type 'DataFrameBase[Unknown]' has no attribute 'head'