def get_data(html):
soup = BeautifulSoup(html, 'html.parser')
data = []
# data in all table rows inside div#allPlays>table
for tr in soup.find(id='allPlays')('tr'):
(period, team, eventName, eventDesc, vid) = [util.clean_str(td.text) for td in tr('td')]
r = IceTrackerRow()
if period == 'Shootout':
r.period = 5
r.time = None
eventName = 'Shootout ' + eventName
else:
period, time = period.split(' ')
r.period = util.get_numeric_period(period)
r.time = time
if len(team) == 0: team = None
r.team = team
# period end should really have an event type
if eventDesc in ('shootout complete', 'period end'): eventName = 'Period End'
r.event_type = eventName
r.event_description = eventDesc
r.details = event_handlers.handler[eventName](eventDesc)
data.append(r)
return data
def get_team_periods_from_table(table):
periods = []
# skip heading row, total row
for tr in table('tr')[1:-1]:
(period, goals, shots, penalties, penalty_minutes) = \
[util.clean_nbsp(td.text).strip() for td in tr('td')]
p = Period()
# period can also be inferred by order
p.period = util.get_numeric_period(period)
p.goals = int(goals)
p.shots = int(shots)
p.penalties = int(penalties)
p.penalty_minutes = int(penalty_minutes)
periods.append(p)
return periods
def get_team_penalties_from_table(table):
penalties = []
for tr in table('tr', recursive=False)[1:]:
(number, period, time, player, minutes, penalty_type) = \
[util.clean_nbsp(td.text).strip() for td in tr('td', recursive=False)]
p = Penalty()
p.period = util.get_numeric_period(period)
p.time = time
try: p.player = util.get_integer(player)
# Can have whole-team penalties with no player
except ValueError: p.player = None
p.minutes = int(minutes)
p.penalty_type = penalty_type
penalties.append(p)
return penalties
