def parse_sheet(url):
sheet_page_html = urllib2.urlopen(url).read()
soup = BeautifulSoup(sheet_page_html)
gameid_re = re.compile("gameid=(.*?)&")
gameid_str = soup.find("form", action=gameid_re)['action']
gameid = gameid_re.search(gameid_str).group(1)
matchup = Matchup()
matchup.game_id = gameid
matchup.save()
all_matchup_trends = matchup.matchup_trends.all().delete()
#gameid_form = soup.find("form", action=re.compile("gameid=(.*?)&"))
team_trends = soup.findAll(text=re.compile('.*Recent ATS Trends.*'))
for team_trend in team_trends:
team_name = team_trend.string.split("-")[0].strip()
logger.info("processing team: " + team_name)
matchup_table = team_trend.findParent("table")
trend_rows = matchup_table.findAll("tr")
new_trends = list()
for idx in range(4, len(trend_rows)):
trend_row = trend_rows[idx]
data_columns = trend_row.findAll("td")
#instantiate the matchup and set the rigth matchup
matchupTrend = MatchupTrend()
matchupTrend.game_matchup = matchup
matchupTrend.team = team_name
matchupTrend.description = data_columns[0].string.strip()
logger.debug("trend description: " + matchupTrend.description)
matchupTrend.current_win, matchupTrend.current_loss = return_win_loss(data_columns[1])
matchupTrend.last3_win, matchupTrend.last3_loss = return_win_loss(data_columns[2])
matchupTrend.since1992_win, matchupTrend.since1992_loss = return_win_loss(data_columns[3])
matchupTrend.save()
new_trends.append( matchupTrend )
matchup.matchup_trends = new_trends
def parse_matchup(matchup_data):
matchup_table = matchup_data.find("table")
team_rx = re.compile(".*college-football/teams/team-page.cfm/team.*")
teams = matchup_table.findAll("a", href=team_rx)
away_team, home_team = get_teams(teams)
if away_team == None or home_team == None:
#if we dont find a team go onto the next team matchup
return
matchup_date = None
try:
date_re = re.compile("date/(.*)")
anchor_tag = matchup_table.find(href=date_re)
matchup_date_str = date_re.search(anchor_tag['href']).group(1)
matchup_date = datetime.datetime.strptime(matchup_date_str,"%m-%d-%y")
except:
logger.exception( "matchup date could not be found" )
logger.error( "anchor tag: " + str(anchor_tag ) )
logger.error( "matchup: " + matchup_table.prettify() )
return
matchup = Matchup()
insider_game_id = "%s-@-%s" % (away_team.name.replace(' ','-'), home_team.name.replace(' ','-'))
found_matchups = Matchup.objects.filter(insider_game_id=insider_game_id, gametime=matchup_date, season=YEAR)
if len(found_matchups) > 0:
matchup = found_matchups[0]
else:
logger.info("!! could not find matchup for: " +insider_game_id + " and date: " + str(matchup_date_str) )
matchup.insider_game_id = insider_game_id
matchup.gametime = matchup_date
matchup.home_team = home_team
matchup.away_team = away_team
matchup.season = int(YEAR)
get_lines_and_score( matchup, matchup_table )
logger.info("final score was: %s to %s for game: %s " % (str(matchup.away_score), str(matchup.home_score), matchup.insider_game_id) )
try:
if not DEBUG:
matchup.save()
except:
logger.exception("problem saving matchup for game id: " + matchup.insider_game_id)
logger.exception("path: " + path)
exit(1)
def parse_matchup(matchup_data):
matchup_table = matchup_data.findParent("table")
team_rx = re.compile(".*college-football/teams/team-page.cfm/team.*")
teams = matchup_table.findAll("a", href=team_rx)
away_team, home_team = get_teams(teams)
if away_team == None or home_team == None:
logger.error("couldn't find teams")
logger.error(matchup_table)
return
date_re = re.compile("date/(.*?)/time")
anchor_tag = matchup_table.find(href=date_re)
matchup_date_str = date_re.search(anchor_tag['href']).group(1)
matchup_date = time.strptime(matchup_date_str,"%m-%d-%y")
matchup = Matchup()
insider_game_id = "%s-@-%s" % (away_team.name.replace(' ','-'), home_team.name.replace(' ','-'))
gametime = strftime("%Y-%m-%d", matchup_date)
found_matchups = Matchup.objects.filter(insider_game_id=insider_game_id, gametime=gametime, season=year)
if len(found_matchups) > 1:
logger.error("!! there can't be more than one matchup for:%s and time %s",(insider_game_id, gametime))
if len(found_matchups) == 1:
matchup = found_matchups[0]
matchup.insider_game_id = insider_game_id
matchup.gametime = gametime
matchup.home_team = home_team
matchup.away_team = away_team
matchup.season = year
get_lines( matchup, matchup_table )
try:
matchup.save()
except:
logger.exception("problem saving matchup for game id: " + matchup.insider_game_id)
exit(1)
