def test_all(): year = '20142015' report_type = 'PL' game_type = '02' for x in range (750,1150): game_num = Operations.pad_game_num(x) game_info = GameHeader.harvest(year, game_num, report_type, game_type) game_personnel = Roster.harvest(year, game_num) game_summary = GameSummary.harvest(year, game_num, game_info, game_personnel) yield check_refs, game_personnel, game_summary
def harvest(year, game_num):
"""
Extract roster information from a html file on the
local machine and create database entries
"""
game_info = GameHeader.harvest(year, game_num, "RO", "02")
tree = Operations.germinate_report_seed(year, game_num, "RO", "02")
tables = tree.xpath("//table//table//table//table")
away_roster = chop_ind_roster_branch(tables, "away", game_info, year)
home_roster = chop_ind_roster_branch(tables, "home", game_info, year)
away_coach, home_coach = chop_coach_branch(tables)
away_coach.team = game_info.away_team
home_coach.team = game_info.home_team
referees, linesmen = chop_officials_branch(tables)
return GamePersonnel(away_roster, home_roster, away_coach, home_coach, referees, linesmen)
def harvest(year, game_num):
game_info = GameHeader.harvest (year, game_num, 'RO', '02')
away_full_name = Operations.team_acronym_to_uppercase(
game_info.away_team)
home_full_name = Operations.team_acronym_to_uppercase(
game_info.home_team)
tree = Operations.germinate_report_seed (year, game_num, "ES", '02')
tables = tree.xpath('//table[@class="tablewidth" and @align="center"]/tr/td/table[@width="100%"]')
rows = tables[3].xpath('./tr')
roster = []
team_acronym = game_info.away_team
for item in rows:
if item.xpath('./td/text()')[0] == home_full_name:
away_roster = roster
roster = []
team_acronym = game_info.home_team
elif item.get('class') == 'evenColor' or\
item.get('class') == 'oddColor':
fields = item.xpath('./td/text()')
for index, field in enumerate(fields):
if field == u'\xa0':
field = '0'
if index == 0:
number = field
elif index == 1:
position = field
elif index == 2:
name_raw = field.split(', ')
first_name = name_raw[1]
last_name = name_raw[0]
elif index == 3:
goals = field
elif index == 4:
assists = field
elif index == 5:
points = field
elif index == 6:
plus_minus = field
elif index == 7:
num_penalties = field
elif index == 8:
pim = field
elif index == 9:
total_minutes = field
elif index == 10:
num_shifts = field
elif index == 11:
avg_shift_length = field
elif index == 12:
powerplay_minutes = field
elif index == 13:
shorthanded_minutes = field
elif index == 14:
evenstrength_minutes = field
elif index == 15:
shots = field
elif index == 16:
attempts_blocked = field
elif index == 17:
missed_shots = field
elif index == 18:
hits = field
elif index == 19:
give_aways = field
elif index == 20:
take_aways = field
elif index == 21:
blocked_shots = field
elif index == 22:
faceoff_wins = field
elif index == 23:
faceoff_losses = field
elif index == 24:
faceoff_percentage = field
playerid = Operations.get_playerid(first_name, last_name,
team_acronym, year, position)
roster.append (ES_Player(team_acronym, number, position,
first_name, last_name, goals, assists, points, plus_minus,
num_penalties, pim, total_minutes, num_shifts,
avg_shift_length, powerplay_minutes, shorthanded_minutes,
evenstrength_minutes, shots, attempts_blocked, missed_shots,
hits, give_aways, take_aways, blocked_shots, faceoff_wins,
faceoff_losses, faceoff_percentage, playerid))
home_roster = roster
return away_roster, home_roster
game_personnel.home_roster) pruned_events = [] for index, event in enumerate(raw_events): pruned_event = prune_event(index, raw_events, game_personnel) pruned_events.append(pruned_event) #if pruned_event.event_type == 'HIT': # print pruned_event return PlayByPlay(raw_events, pruned_events) if __name__ == '__main__': year = '20142015' report_type = 'PL' game_type = '02' for game_num_raw in range (7,8): game_num = Operations.pad_game_num (game_num_raw) game_info = GameHeader.harvest(year, game_num, report_type, game_type) game_personnel = Roster.harvest (year, game_num) temp_pbp = harvest(year, game_num, report_type, game_type, game_info, game_personnel) print game_info for item in temp_pbp.pruned_events: if item.event_type == 'PENL': print item
