Get Top Words!

topwords.py

import argparse
import re
from typing import Dict


def args():
    parser = argparse.ArgumentParser()
    parser.add_argument(
        '-f', '--file', help='Name of file to count words from')
    parser.add_argument('-n', '--N', help='Number of top items', type=int)
    return parser.parse_args()


def _get_words(lines) -> Dict[str, int]:
    words = {}
    for l in lines:
        ws = l.lower().split()
        ws = [re.sub('[()“”‘#.,?!]', '', w) for w in ws]
        for w in ws:
            if not w in words.keys():
                words[w] = 1
            elif w in words.keys():
                words[w] = words.get(w)+1
    return words


def _get_top(words: dict, howmany: int = 10):
    sorted_words = sorted(words, key=words.get, reverse=True)
    for i in range(howmany):
        key = sorted_words[i]
        print(key, words.get(key))


def main():
    argu = args()
    f = open(argu.file, 'r')
    lines = f.readlines()
    words = _get_words(lines=lines)
    _get_top(words=words, howmany=argu.N)


if __name__ == "__main__":
    main()