Simple script to see if Andrew Bynum played last night for the Philadelphia 76ers

did_bynum_play.py

import urllib2
from lxml.html import parse, make_links_absolute
from StringIO import StringIO
import re
from datetime import date, timedelta
# Helper Function to Scrape Pages #


def get_html_tree(url):
    html = urllib2.urlopen(url).read()
    abs_links = make_links_absolute(html, url)
    return parse(StringIO(abs_links))


def find_date(url):
    """Finds date in URLs with 8 digit format of YYYYMMDD"""
    m = re.search(r'(\d{4})(\d{2})(\d{2})', url)
    return date(int(m.group(1)), int(m.group(2)), int(m.group(3)))


def game_yesterday(date):
    if date < date.today() - timedelta(1):
        return None
    else:
        return 1


def did_bynum_play(boxscore_tree):
    """Goes through inactive players looking for Bynum"""
    bynum_check = None
    statuses = [player.text_content() for player in boxscore_tree.xpath('///ul[@id="nbaGIPlyrStatus"]')]
    for status in statuses:
        if re.search("Bynum", status):
            bynum_check = 1
            continue
    if bynum_check:
        print "Bynum did not play!"
    else:
        print "Bynum might have played (or this program is broken)."

if __name__ == '__main__':
    # Sixers Homepage (Find most Recent Box Score) #
    home = 'http://www.nba.com/sixers/'
    home_tree = get_html_tree(home)

    # Find most recent box-score #
    link_tag = home_tree.xpath('///a[@class="boxscore"]')[0].iterlinks()
    boxscore_link = [elem[2] for elem in link_tag][0]

    # Check if game was played yesterday #
    gamedate = find_date(boxscore_link)
    game_check = game_yesterday(gamedate)
    if game_check == None:
        print "No game for Bynum to play in yesterday!"
        raise SystemExit

    # Go to last Box Score #
    boxscore = get_html_tree(boxscore_link)

    # See if Bynum Played #
    did_bynum_play(boxscore)