not supported instances of 'str' and 'int' in multiprocessing

I'm trying to run a multiprocess task in a function that receives the dataframe and the condition and return a new dataframe.

My problem is that I receiving the "not supported instances of 'str' and 'int'" when I call the function.

import multiprocessing
import pandas as pd


def operation(data, condition):  
    .
    . No problem in the function since I tested it isolated
    .
    .
 return pd.DataFrame(result_of_operation)

if __name__ == '__main__':  

    data = pd.read_csv(r"C:\Users\max\test.csv", index_col=False, encoding='UTF-8')
    
    column_names = ["A", "B", "C"]
    new_df = pd.DataFrame(columns = column_names)
    
    condition = ['orange', 'banana', 'apple']
    
    with multiprocessing.Pool(8) as p:
        for x in range(0, len(condition)):
            new_df.append(p.map(operation, data, condition[x]),ignore_index=True)

I believe it is a problem with my map operation since it works if I call the function by itself, like:

for x in range(0, len(condition)):
     new_df.append(operation(data, condition[x]),ignore_index=True)

Solution

It looks like the way you're passing parameters to map() is wrong.

Try this:

from functools import partial

with multiprocessing.Pool() as p:
  new_df = pd.concat(p.map(partial(operation, data), condition), ignore_index=True)

Perhaps to make things even clearer, here's a complete working example. Note that append() is being deprecated so I use the preferred concat() here:

from pandas import DataFrame, concat
from concurrent.futures import ProcessPoolExecutor
from functools import partial


def process(columns, data):
    return DataFrame(data, columns=columns)


def main():
    data = [[0, 1, 2], [3, 4, 5], [6, 7, 8, 9]]
    with ProcessPoolExecutor(max_workers=3) as p:
        newdf = concat(p.map(partial(process, ['A']), data), ignore_index=True)
        print(newdf)


if __name__ == '__main__':
    main()

Output: