def parse_event(event):
"""
Parses a single event when the info is in a json format
:param event: json of event
:return: dictionary with the info
"""
play = dict()
play['event_id'] = event['about']['eventIdx']
play['period'] = event['about']['period']
play['event'] = str(change_event_name(event['result']['eventTypeId']))
play['seconds_elapsed'] = shared.convert_to_seconds(event['about']['periodTime'])
# If there's a players key that means an event occurred on the play.
if 'players' in event.keys():
play['p1_name'] = shared.fix_name(event['players'][0]['player']['fullName'])
play['p1_ID'] = event['players'][0]['player']['id']
for i in range(len(event['players'])):
if event['players'][i]['playerType'] != 'Goalie':
play['p{}_name'.format(i + 1)] = shared.fix_name(event['players'][i]['player']['fullName'].upper())
play['p{}_ID'.format(i + 1)] = event['players'][i]['player']['id']
# Coordinates aren't always there
try:
play['xC'] = event['coordinates']['x']
play['yC'] = event['coordinates']['y']
except KeyError:
play['xC'] = ''
play['yC'] = ''
return play
def populate_players(event_dict, players, away_players, home_players):
"""
Populate away and home player info (and num skaters on each side)
:param event_dict: dict with event info
:param players: all players in game and info
:param away_players: players for away team
:param home_players: players for home team
:return: None
"""
for venue in ['home', 'away']:
for j in range(6):
try:
if venue == "home":
name = shared.fix_name(home_players[j][0].upper())
foo = 'Home'
else:
name = shared.fix_name(away_players[j][0].upper())
foo = 'Away'
event_dict['{}Player{}'.format(venue, j + 1)] = name
event_dict['{}Player{}_id'.format(
venue, j + 1)] = players[foo][name]['id']
except KeyError:
event_dict['{}Player{}_id'.format(venue, j + 1)] = 'NA'
except IndexError:
event_dict['{}Player{}'.format(venue, j + 1)] = ''
event_dict['{}Player{}_id'.format(venue, j + 1)] = ''
# Did this because above method assumes the goalie is at end of player list
for x in away_players:
if x[2] == 'G':
event_dict['Away_Goalie'] = shared.fix_name(x[0].upper())
try:
event_dict['Away_Goalie_Id'] = players['Away'][
event_dict['Away_Goalie']]['id']
except KeyError:
event_dict['Away_Goalie_Id'] = 'NA'
else:
event_dict['Away_Goalie'] = ''
event_dict['Away_Goalie_Id'] = ''
for x in home_players:
if x[2] == 'G':
event_dict['Home_Goalie'] = shared.fix_name(x[0].upper())
try:
event_dict['Home_Goalie_Id'] = players['Home'][
event_dict['Home_Goalie']]['id']
except KeyError:
event_dict['Home_Goalie_Id'] = 'NA'
else:
event_dict['Home_Goalie'] = ''
event_dict['Home_Goalie_Id'] = ''
event_dict['Away_Players'] = len(away_players)
event_dict['Home_Players'] = len(home_players)
def combine_players_lists(json_players, roster_players, game_id):
"""
Combine the json list of players (which contains id's) with the list in the roster html
:param json_players: dict of all players with id's
:param roster_players: dict with home and and away keys for players
:param game_id: id of game
:return: dict containing home and away keys -> which contains list of info on each player
"""
players = {'Home': dict(), 'Away': dict()}
for venue in players.keys():
for player in roster_players[venue]:
try:
name = shared.fix_name(player[2])
player_id = json_players[name]['id']
players[venue][name] = {'id': player_id, 'number': player[0]}
except KeyError:
# If he was listed as a scratch and not a goalie (check_goalie deals with goalies)
# As a whole the scratch list shouldn't be trusted but if a player is missing an id # and is on the
# scratch list I'm willing to assume that he didn't play
if not player[3] and player[1] != 'G':
player.extend([game_id])
players_missing_ids.extend([[player[2], player[4]]])
players[venue][name] = {'id': 'NA', 'number': player[0]}
return players
def get_players_json(players_json):
"""
Return dict of players for that game
:param players_json: players section of json
:return: dict of players->keys are the name (in uppercase)
"""
players = dict()
for key in players_json.keys():
name = shared.fix_name(players_json[key]['fullName'].upper())
players[name] = {
'id': ' ',
'last_name': players_json[key]['lastName'].upper()
}
try:
players[name]['id'] = players_json[key]['id']
except KeyError:
shared.print_warning(
'{name} is missing an ID number in the pbp json'.format(
name=name))
players[name]['id'] = 'NA'
return players
def parse_shift(shift):
"""
Parse shift for json
:param shift: json for shift
:return: dict with shift info
"""
shift_dict = dict()
name = shared.fix_name(' '.join([
shift['firstName'].strip(' ').upper(),
shift['lastName'].strip(' ').upper()
]))
shift_dict['Player'] = name
shift_dict['Player_Id'] = shift['playerId']
shift_dict['Period'] = shift['period']
shift_dict['Team'] = fix_team_tricode(shift['teamAbbrev'])
# At the end of the json they list when all the goal events happened. They are the only one's which have their
# eventDescription be not null
if shift['eventDescription'] is None:
shift_dict['Start'] = shared.convert_to_seconds(shift['startTime'])
shift_dict['End'] = shared.convert_to_seconds(shift['endTime'])
shift_dict['Duration'] = shared.convert_to_seconds(shift['duration'])
else:
shift_dict = dict()
return shift_dict
def parse_html(html, player_ids, game_id):
"""
Parse the html
Note: Don't f**k with this!!! I'm not exactly sure how or why but it works.
:param html: cleaned up html
:param player_ids: dict of home and away players
:param game_id: id for game
:return: DataFrame with info
"""
columns = [
'Game_Id', 'Player', 'Player_Id', 'Period', 'Team', 'Start', 'End',
'Duration'
]
df = pd.DataFrame(columns=columns)
td, teams = get_soup(html)
team = teams[0]
home_team = teams[1]
players = dict()
# The list 'td' is laid out with player name followed by every component of each shift. Each shift contains:
# shift #, Period, begin, end, and duration. The shift event isn't included.
for t in td:
t = t.get_text()
if ',' in t: # If it has a comma in it we know it's a player's name...so add player to dict
name = t
# Just format the name normally...it's coded as: 'num last_name, first_name'
name = name.split(',')
name = ' '.join([name[1].strip(' '), name[0][2:].strip(' ')])
name = shared.fix_name(name)
players[name] = dict()
players[name]['number'] = name[0][:2].strip()
players[name]['Shifts'] = []
else:
# Here we add all the shifts to whatever player we are up to
players[name]['Shifts'].extend([t])
for key in players.keys():
# Create a list of lists (each length 5)...corresponds to 5 columns in html shifts
players[key]['Shifts'] = [
players[key]['Shifts'][i:i + 5]
for i in range(0, len(players[key]['Shifts']), 5)
]
# Parse each shift
shifts = [
analyze_shifts(shift, key, team, home_team, player_ids)
for shift in players[key]['Shifts']
]
df = df.append(shifts, ignore_index=True)
df['Game_Id'] = str(game_id)[5:]
return df
def get_players_json(json):
"""
Return dict of players for that game
:param json: gameData section of json
:return: dict of players->keys are the name (in uppercase)
"""
players = dict()
players_json = json['players']
for key in players_json.keys():
name = shared.fix_name(players_json[key]['fullName'].upper())
players[name] = {'id': ' '}
try:
players[name]['id'] = players_json[key]['id']
except KeyError:
print(name, ' is missing an ID number in the pbp json')
players[name]['id'] = 'NA'
return players