Remove folder matching name recursively and file with specific extension

dataset_cleanup.py

import os
import shutil
from concurrent.futures import ThreadPoolExecutor

def remove_file(file_path):
    try:
        os.remove(file_path)
        print(f"File '{file_path}' removed successfully.")
    except Exception as e:
        print(f"Failed to remove file '{file_path}': {e}")

def remove_folder(folder_path):
    try:
        shutil.rmtree(folder_path)
        print(f"Folder '{folder_path}' removed successfully.")
    except Exception as e:
        print(f"Failed to remove folder '{folder_path}': {e}")

def remove_folders_and_files(root_dir, folder_names_to_remove, file_extension_to_remove, num_threads=1):
    with ThreadPoolExecutor(max_workers=num_threads) as executor:
        for dirpath, dirnames, filenames in os.walk(root_dir, topdown=True):
            for filename in filenames:
                if filename.endswith(file_extension_to_remove):
                    file_path = os.path.join(dirpath, filename)
                    executor.submit(remove_file, file_path)

            for dirname in dirnames[:]:
                if dirname in folder_names_to_remove:
                    folder_path = os.path.join(dirpath, dirname)
                    executor.submit(remove_folder, folder_path)

if __name__ == "__main__":
    root_directory = "/path/to/your/root_directory"  # Replace with the actual root directory path
    folders_to_remove = ["folder_name_1", "folder_name_2", "folder_name_3"]  # Replace with the folder names you want to remove
    file_extension_to_remove = ".txt"  # Replace with the file extension you want to remove
    num_threads = 4  # Replace with the desired number of threads

    remove_folders_and_files(root_directory, folders_to_remove, file_extension_to_remove, num_threads)