scripts/merge-csv-lists.py

#!/usr/bin/python
import argparse


def unique_combined_list(input1, input2):
    # Split each input by comma and trim whitespace
    list1 = [item.strip().title() for item in input1.split(",")]
    list2 = [item.strip().title() for item in input2.split(",")]

    # Combine lists
    combined_list = list1 + list2

    # Create an empty list to store the final unique names
    final_list = []

    # Check for reversed names
    for name in combined_list:
        parts = name.split()

        # If the name has two words, check for its reversed variant
        if len(parts) == 2:
            first, last = parts
            reversed_name = f"{last} {first}"

            # If neither the name nor its reversed variant is in the final list, add the name
            if name not in final_list and reversed_name not in final_list:
                final_list.append(name)
        # If it's a single-word name, simply add it if it's not in the final list
        else:
            if name not in final_list:
                final_list.append(name)

    # Flatten the list into individual words
    flattened_words = [word for name in combined_list for word in name.split()]

    # Sort the list based on the criteria discussed above
    sorted_list = sorted(
        final_list,
        key=lambda x: (flattened_words.index(x.split()[0]), combined_list.index(x)),
    )

    # Convert the list back to a comma-separated string
    output = ",".join(sorted_list)

    return output


def main():
    # Create an argument parser
    parser = argparse.ArgumentParser(
        description="Combine two comma-separated lists into one unique sorted list."
    )

    # Add arguments for the two input lists
    parser.add_argument("list1", type=str, help="The first comma-separated list.")
    parser.add_argument("list2", type=str, help="The second comma-separated list.")

    # Parse the arguments
    args = parser.parse_args()

    # Get the unique combined list
    result = unique_combined_list(args.list1, args.list2)

    print(result)


if __name__ == "__main__":
    main()