Extract text chunks from a given Crowdin page

extract_crowdin_text.py

"""
Extract source and translation texts from a given Crowdin page. The output file
generated by this script is suitable for doing translation work on a mobile 
device that is occasionally offline.

This script assumes that you've created a custom Firefox profile that is
already logged into your Crowdin account, and that you've provided the path for
it below.

"""

url = 'https://crowdin.com/translate/django-girls-tutorial/53/en-zhcn'


import codecs
import os
import os.path as op
import sys
from selenium import webdriver
from selenium.webdriver.firefox.firefox_profile import FirefoxProfile


browser = None


def main():
    global browser
    text_pairs = []

    browser = webdriver.Firefox(firefox_profile=get_profile())
    browser.get(url)

    while True:
        pair = get_text_pair()
        if text_pairs and pair == text_pairs[-1]:
            break

        text_pairs.append(pair)

        next_button = browser.find_element_by_id('next_translation')
        next_button.click()
        print pair 

    browser.quit()

    with codecs.open('text.txt', 'w', 'utf-8') as fp:
        fp.write(url + '\n\n')
        for source, translated in text_pairs:
            if not translated:
                translated = '-'
            fp.write(source + '\n')
            fp.write(translated + '\n\n')

    print '\nWrote results to text.txt'

def get_text_pair():
    el = browser.find_element_by_css_selector('#source_phrase_container')
    source_text = el.text

    el = browser.find_element_by_css_selector('#translation')
    translated_text = el.get_attribute('value')

    return source_text, translated_text


def get_profile():
    """
    Returns the FirefoxProfile object for a profile that has the name 'selenium'
    in it.

    """
    home = os.environ['HOME']
    if sys.platform == 'darwin':
        profile_dir = '%s/Library/Application Support/Firefox/Profiles' % home
    else:
        profile_dir = '%/.mozilla/firefox'

    for profile in os.listdir(profile_dir):
        if 'selenium' in profile:
            profile_path = op.join(profile_dir, profile)
            return FirefoxProfile(profile_path)

    return None


if __name__ == '__main__':
    main()