Find the year an album was published

gistfile1.py

import requests
import pyquery
import re
import sys
import urllib

SEARCH_URL = 'http://ajax.googleapis.com/ajax/services/search/web'


def unquote_all(url):
    new_url = urllib.unquote(url)
    if new_url == url:
        return new_url
    else:
        return unquote_all(new_url)


def get_year(album):
    query = '{0} album wiki'.format(album)
    res = requests.get(SEARCH_URL, params={'q': query, 'v': '1.0'})
    url = res.json['responseData']['results'][0]['url']

    # Hack to deal with the David Bowie Album '"Hereos"'
    # special characters in the url were breaking the script
    url = unquote_all(url)

    print "Looking up on wiki at {}".format(url)

    text = pyquery.PyQuery(requests.get(url).text).find('.published').text()
    year = re.search('(\d{4})', text).groups()[0]
    return year


if __name__ == '__main__':
    print get_year(' '.join(sys.argv[1:]))