pandas-contract 0.11.3

Define input and output columns for functions working on pandas dataframes.

pandas-contract

Provide decorators to check functions arguments and return values using pandas DataFrame.

The decorators utilize the pandera.io library to validate data types and constraints of the input arguments and output values of functions.

Documentation

Documentation on pandas-contract.readthedocs.io

Installation

pip install pandas-contract

Setup

See Setup for first-time setup information.

Usage

ℹ️ Info: Generally, the standard abbreviations for the package imports are

import pandas as pd
import pandas_contract as pc
import pandera.pandas as pa

Check Dataframe structure

The following defines a function that takes a DataFrame with a column 'x' of type integer as input and returns a DataFrame with the column 'x' of type string as output.

The Pandera.io documentation provides a full overview of the DataFrame/DataSeries checks.

@pc.argument("df", pa.DataFrameSchema({"x": pa.Int}))
@pc.result(pa.DataFrameSchema({"x": pa.String}))
def col_x_to_string(df: pd.DataFrame) -> pd.DataFrame:
    """Convert column x to string"""
    return df.assign(x=df["x"].astype(str))

Dynamic Arguments and return values

Required columns and arguments can also be specified dynamically using a function that returns a schema.

@pc.argument("df", pa.DataFrameSchema(
    {pc.from_arg("col"): pa.Column()})
)
@pc.result(pa.DataFrameSchema({pc.from_arg("col"): pa.String}))
def col_to_string(df: pd.DataFrame, col: str) -> pd.DataFrame:
    return df.assign(**{col: df[col].astype(str)})

Multiple columns in function argument

The decorator also supports multiple columns from the function argument.

@pc.argument("df", pa.DataFrameSchema(
        {pc.from_arg("cols"): pa.Column()}
    )
)
@pc.result(pa.DataFrameSchema({pc.from_arg("cols"): pa.String}))
def cols_to_string(df: pd.DataFrame, cols: list[str]) -> pd.DataFrame:
    return df.assign(**{col: df[col].astype(str) for col in cols})

Retrieve dataframes from a more complex argument

Sometimes the dataframe is not a direct argument of the function, but is part of a more complex argument. In this case, the decorator argument key can be used to specify the key of the dataframe in the argument.

If key is a callable, it will be called with the argument and the result will be used as the dataframe. Otherwise, it will be used as a key to retrieve the dataframe from the argument, i.e. arg[key].

Dataframe result is wrapped within another object

@pc.result(key="data")
def into_dict():
    """Dataframe wrapped in a dict"""
    return dict(data=pd.DataFrame())

See Key Type for more information and examples.