pandera

pandera performs data validation on dataframe-like objects (pandas, polars, dask modin, spark etc).

import pandera as pa
from pandera.typing import Series
 
# pydantic-style syntax
class Schema(pa.DataFrameModel):
    column1: int = pa.Field(le=10)
    column2: float = pa.Field(lt=-1.2)
    column3: str = pa.Field(str_startswith="value_")
 
	# more specific custom rule
	@pa.check("column3")
    def column_3_check(cls, series: Series[str]) -> Series[bool]:
        """Check that column3 values have two elements after being split with '_'"""
        return series.str.split("_", expand=True).shape[1] == 2
 
Schema.validate(df)

pandera integrates with beartype.