translate a srt file with google translate

translate-srt.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-

import sys,json,urllib
from pysrt import SubRipFile, SubRipItem

url_string = 'http://ajax.googleapis.com/ajax/services/language/'

def detect(text):
    query = dict(v='1.0', q=text)
    file = urllib.urlopen(url_string+'detect?%s' % urllib.urlencode(query))
    result = json.loads(file.read())
    return result['responseData']['language'] if result['responseData'] else None

def translate(text, input_language, output_language):
    source = detect(text) 
    target = [input_language, output_language][source == input_language]
    query = dict(v='1.0', q=text,langpair = '%s|%s' % (source, target))
    file = urllib.urlopen(url_string+'translate?%s' % urllib.urlencode(query))
    result = json.loads(file.read())
    return result['responseData']['translatedText'] if result['responseData'] else None

def translate_srt_file(input_file, output_file, input_language, output_language):
    """translate a srt file from a language to another"""
    print 'processing file', input_file
    subs = SubRipFile.open(input_file)
    for sentence in subs:
        sentence.text = translate(sentence.text, input_language, output_language) + '\n'
    subs.save(output_file, 'utf-8')

def main():
    if len(sys.argv) < 4:
        info = '''
            translate a subtitle file from a language to another.
            Most of the code to use google translate API was taken there :
              https://github.com/zhendi/google-translate-cli

            usage:
              translate-srt.py input_file output_file input_language output_language

            example:
              translate-srt.py ./titanic-en.srt ./titanic-fr.srt en fr
            '''
        print info
    else:
        translate_srt_file(*sys.argv[-4:])

if __name__ == '__main__':
    main()