Convert tabs to spaces and remove trailing whitespace in code files

tabs-to-spaces.py

#! /usr/bin/python3

import argparse
import functools
import multiprocessing
import pathlib
import os
import sys


def _gather_file_list(path: str, patterns: list):
    files = []
    for pattern in patterns:
        files.extend(pathlib.Path(path).rglob(pattern))

    return files


def _parse_arguments():
    parser = argparse.ArgumentParser(
        prog='tabs-to-spaces',
        description='Convert tabs to spaces.')

    parser.add_argument('--tab-width',
                        dest='tab_width',
                        default=4,
                        help='Number of spaces used to replace a tab.')
    parser.add_argument('--trim-eol',
                        dest='trim_eol',
                        default=True,
                        help='Determines if trailing whitespace should be removed.')
    parser.add_argument('--patterns',
                        dest='pattern_list',
                        nargs='+',
                        default=[],
                        help='List of file pattern to process..')
    parser.add_argument('--directory',
                        dest='directory',
                        type=str,
                        help='Directory to process.')

    args = parser.parse_args()

    if not args.directory or not args.pattern_list:
        parser.print_help(sys.stderr)
        return None
    else:
        args.directory = os.path.abspath(args.directory)

        print('Settings:')
        print('\tTab width : ', args.tab_width)
        print('\tTrim lines: ', args.trim_eol)
        print('\tPatterns  : ', args.pattern_list)
        print('\tDirectory : ', args.directory)

        return args


def _process_file(path: str, tab_width: int, trim: bool):
    contents = None

    with open(path, 'r') as in_file:
        contents = in_file.readlines()

    if contents:
        for line in contents:
            line = line.replace('\t', ' ' * tab_width)

            if trim:
                line = line.rstrip()

        with open(path, 'w') as out_file:
            out_file.writelines(contents)


def main():
    args = _parse_arguments()
    if not args:
        return -1

    file_list = _gather_file_list(args.directory, args.pattern_list)
    if not file_list:
        return -1

    with multiprocessing.Pool() as pool:
        pool.map(functools.partial(_process_file,
                 tab_width=args.tab_width,
                 trim=args.trim_eol),
                 file_list)

    return 0


if __name__ == "__main__":
    main()