def run(self):
DB = store.Store()
soup = BeautifulSoup(Fetcher.fetch(self.url))
for link in soup.findAll('a'):
if link['href'].find('gid_') >= 0:
gid = link['href'].rstrip('/')
g = game.Game(gid)
if (g.game_type != self.gametype):
continue;
g.save()
game_id = g.game_id
ab = atbats.AtBats(gid, game_id)
ab.save()
chart = hitchart.HitChart(gid, game_id)
chart.save()
batters = players.Batters(gid, game_id)
batters.save()
pitchers = players.Pitchers(gid, game_id)
pitchers.save()
def run(self):
DB = store.Store()
soup = BeautifulSoup(Fetcher.fetch(self.url))
for link in soup.findAll('a'):
if link['href'].find('gid_') >= 0:
gid = link['href'].rstrip('/')
g = game.Game(gid)
if (g.game_type != self.gametype):
continue
g.save()
game_id = g.game_id
ab = atbats.AtBats(gid, game_id)
ab.save()
chart = hitchart.HitChart(gid, game_id)
chart.save()
batters = players.Batters(gid, game_id)
batters.save()
pitchers = players.Pitchers(gid, game_id)
pitchers.save()
def getMonths(year, start=1):
months = []
url = '%syear_%4d/' % (CONSTANTS.BASE, year)
soup = BeautifulSoup(Fetcher.fetch(url))
for link in soup.findAll('a'):
if link['href'].find('month') >= 0:
month = int(link['href'].replace('month_', '').rstrip('/'))
if month >= start:
months.append(month)
return months
def getMonths(year, start = 1):
months = []
url = '%syear_%4d/' % (CONSTANTS.BASE, year)
soup = BeautifulSoup(Fetcher.fetch(url))
for link in soup.findAll('a'):
if link['href'].find('month') >= 0:
month = int(link['href'].replace('month_', '').rstrip('/'))
if month >= start:
months.append(month)
return months
def getDays(year, month, start=1):
days = []
url = '%syear_%4d/month_%02d/' % (CONSTANTS.BASE, year, month)
soup = BeautifulSoup(Fetcher.fetch(url))
for link in soup.findAll('a'):
if link['href'].find('day') >= 0:
try:
day = int(link['href'].replace('day_', '').rstrip('/'))
except:
# sometimes gameday will have like a '26_bak' directory
continue
if day >= start:
days.append(day)
return days
def getDays(year, month, start = 1):
days = []
url = '%syear_%4d/month_%02d/' % (CONSTANTS.BASE, year, month)
soup = BeautifulSoup(Fetcher.fetch(url))
for link in soup.findAll('a'):
if link['href'].find('day') >= 0:
try:
day = int(link['href'].replace('day_', '').rstrip('/'))
except:
# sometimes gameday will have like a '26_bak' directory
continue
if day >= start:
days.append(day)
return days
log.addHandler(logging.StreamHandler())
else:
log.setLevel(logging.ERROR)
log.addHandler(logging.StreamHandler())
logfilename = './log.txt'
filelog = logging.FileHandler(logfilename, 'a')
filelog.setLevel(logging.ERROR)
filelog.setFormatter(formatter)
log.addHandler(filelog)
CONSTANTS.BASE = CONSTANTS.BASE.replace('%TYPE%', TYPE)
url = '%syear_%4d/' % (CONSTANTS.BASE, YEAR)
try:
soup = BeautifulSoup(Fetcher.fetch(url))
except TypeError, e:
print 'Could not fetch %s' % url
raise SystemExit
if MONTH is None:
if startmonth:
months = getMonths(YEAR, startmonth)
else:
months = getMonths(YEAR)
else:
months = MONTH
for month in months:
if DAY is None:
if startday:
log.addHandler(logging.StreamHandler())
else:
log.setLevel(logging.ERROR)
log.addHandler(logging.StreamHandler())
logfilename = './' + args.errors
filelog = logging.FileHandler(logfilename, 'a')
filelog.setLevel(logging.ERROR)
filelog.setFormatter(formatter)
log.addHandler(filelog)
CONSTANTS.BASE = CONSTANTS.BASE.replace('%LEAGUE%', args.league)
url = '%syear_%4d/' % (CONSTANTS.BASE, args.year)
try:
soup = BeautifulSoup(Fetcher.fetch(url))
except TypeError, e:
print 'Could not fetch %s' % url
raise SystemExit
if args.month is None:
if startmonth:
months = getMonths(args.year, startmonth)
else:
months = getMonths(args.year)
else:
months = args.month
for month in months:
if args.day is None:
if startday:
