pyweek game grabber

gistfile1.py

#!/usr/bin/env python

from bs4 import BeautifulSoup


INDEX_FILE = 'index.html'
SCRIPT_FILE = 'get_games.sh'


soup = BeautifulSoup(open(INDEX_FILE).read())
script = open(SCRIPT_FILE, 'w')


def process_game(th, td):
    team_name = th.contents[-1].strip()[1:-1]
    urls = [a.get('href') for a in td.find_all('a')]
    urls = [url for url in urls if "media.pyweek.org" in url]
    script.write('mkdir %s\npushd %s\n' % (team_name, team_name))
    for url in urls:
        script.write('wget -c "%s"\n' % url)
    script.write('popd\n\n')


rows = soup.find_all('tr')

while rows:
    th = rows[0].th
    td = rows[1].td
    rows = rows [2:]
    process_game(th, td)


script.close()