translate

translate_str.py

#!/usr/bin/env python3

"""
Quick'n'dirty script to translate subs using Yandex.

It all started because I didn't have enough time to finish a Japanese movie
that I was watching heading back home from DEF CON 27.
Anyway, I managed to get a copy of the movie (**cough cough**) but my 日本語
is really bad and the only subs I could find were in Chinese.
So I decided to bear the automatic translation for the remaining 30 minutes
that I had left.

Usage: ./translate_str.py /path/to/subs.srt zh-en

(c) 2019 Paolo Montesel
"""

import re
import sys

import yandex_translate


API_KEY = YOUR_YANDEX_API_KEY


TMPL = r'''{}
{} --> {}
{}
'''


def main(fp, lan):
    translate = yandex_translate.YandexTranslate(API_KEY)
    with open(fp, 'r') as f:
        ss = re.split('(\\d+)\n(\\S+) --> (\\S+)\s+', f.read())
        start = 0
        for i in range(100):
            try:
                start = int(ss[i])
                break
            except:
                pass
        ss = ss[start:]
        for i in range(0, len(ss), 4):
            seq, start, end, text = ss[i:i+4]
            seq = int(seq)
            text = text.strip()
            text = translate.translate(text, lan)['text']
            print(TMPL.format(seq, start, end, text[0]))
            assert len(text) == 1


if __name__ == '__main__':
    main(sys.argv[1], sys.argv[2])