from schemas import DatasetSchema, ColumnSchema from validator import ColumnValidator def ensure_matching_inputs( df_a: pd.DataFrame, df_b: pd.DataFrame, *, schema: Optional[DatasetSchema] = None, name_a: str = "input_a", name_b: str = "input_b", ) -> Tuple[pd.DataFrame, pd.DataFrame, DatasetSchema]: if schema is None: inferred_columns = [ ColumnSchema(name=col, dtype=str(df_a[col].dtype)) for col in df_a.columns ] schema = DatasetSchema( name="inferred_input_schema", columns=inferred_columns, strict_order=True, strict_types=False, ) validator = ColumnValidator(schema) validator.validate(df_a, dataset_name=name_a) validator.validate(df_b, dataset_name=name_b) aligned_a = validator.align_columns(df_a) aligned_b = validator.align_columns(df_b) return aligned_a, aligned_b, schema def ensure_output_matches( output_df: pd.DataFrame, *, schema: DatasetSchema, output_name: str = "output", ) -> pd.DataFrame: validator = ColumnValidator(schema) validator.validate(output_df, dataset_name=output_name) return validator.align_columns(output_df)