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['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 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
"""
home_players = dict()
for player in roster_players['Home']:
try:
name = shared.fix_name(player[2])
id = json_players[name]['id']
home_players[name] = {'id': id, 'number': player[0]}
except KeyError:
# This usually means it's the backup goalie (who didn't play) so it's no big deal with them
if player[1] != 'G':
players_missing_ids.extend([player, game_id])
home_players[name] = {'id': 'NA', 'number': player[0]}
away_players = dict()
for player in roster_players['Away']:
try:
name = shared.fix_name(player[2])
id = json_players[name]['id']
away_players[name] = {'id': id, 'number': player[0]}
except KeyError:
if player[1] != 'G':
players_missing_ids.extend([player, game_id])
away_players[name] = {'id': 'NA', 'number': player[0]}
return {'Home': home_players, 'Away': away_players}
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['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'] = ''
"""
# Sometimes they record events for shots in the wrong zone (or maybe not)...so change it
if play['xC'] != 'Na' and play['yC'] != 'Na':
if play['Ev_Team'] == home_team:
# X should be negative in 1st and 3rd for home_team
if (play['Period'] == 1 or play['Period'] == 3) and play['xC'] > 0:
play['xC'] = -int(play['xC'])
play['yC'] = -int(play['yC'])
elif play['Period'] == 2 and play['xC'] < 0:
play['xC'] = -int(play['xC'])
play['yC'] = -int(play['yC'])
else:
# X should be positive in 1st and 3rd for away_team
if (play['Period'] == 1 or play['Period'] == 3) and play['xC'] < 0:
play['xC'] = -int(play['xC'])
play['yC'] = -int(play['yC'])
elif play['Period'] == 2 and play['xC'] > 0:
play['xC'] = -int(play['xC'])
play['yC'] = -int(play['yC'])
"""
return play
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
:param html: cleaned up html
:param player_ids: dict of home and away players
:param game_id:
:return: DataFrame with info
"""
columns = [
'Game_Id', 'Player', 'Player_Id', 'Period', 'Team', 'Start', 'End',
'Duration'
]
df = pd.DataFrame(columns=columns)
soup = BeautifulSoup(html.content, "lxml")
teams = get_teams(soup)
team = teams[0]
home_team = teams[1]
td = soup.findAll(
True, {'class': ['playerHeading + border', 'lborder + bborder']})
"""
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.
"""
players = dict()
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)
]
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)
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')
players[name]['id'] = 'NA'
return players
def parse_event(event, players, home_team, if_plays_in_json, current_score):
"""
Receievs an event and parses it
:param event:
:param players: players in game
:param home_team:
:param if_plays_in_json: If the pbp json contains the plays
:param current_score: current score for both teams
:return: dict with info
"""
event_dict = dict()
away_players = event[6]
home_players = event[7]
try:
event_dict['Period'] = int(event[1])
except ValueError:
event_dict['Period'] = 0
event_dict['Description'] = event[5]
event_dict['Event'] = str(event[4])
if event_dict['Event'] in [
'GOAL', 'SHOT', 'MISS', 'BLOCK', 'PENL', 'FAC', 'HIT', 'TAKE',
'GIVE'
]:
event_dict['Ev_Team'] = event[5].split(
)[0] # Split the description and take the first thing (which is the team)
# If it's a goal change the score
if event[4] == 'GOAL':
if event_dict['Ev_Team'] == home_team:
current_score['Home'] += 1
else:
current_score['Away'] += 1
event_dict['Home_Score'] = current_score['Home']
event_dict['Away_Score'] = current_score['Away']
# Populate away and home player info
for j in range(6):
try:
name = shared.fix_name(away_players[j][0].upper())
event_dict['awayPlayer{}'.format(j + 1)] = name
event_dict['awayPlayer{}_id'.format(
j + 1)] = players['Away'][name]['id']
except KeyError:
event_dict['awayPlayer{}_id'.format(j + 1)] = 'NA'
except IndexError:
event_dict['awayPlayer{}'.format(j + 1)] = ''
event_dict['awayPlayer{}_id'.format(j + 1)] = ''
try:
name = shared.fix_name(home_players[j][0].upper())
event_dict['homePlayer{}'.format(j + 1)] = name
event_dict['homePlayer{}_id'.format(
j + 1)] = players['Home'][name]['id']
except KeyError:
event_dict['homePlayer{}_id'.format(j + 1)] = 'NA'
except IndexError:
event_dict['homePlayer{}'.format(j + 1)] = ''
event_dict['homePlayer{}_id'.format(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)
try:
home_skaters = event_dict['Home_Players'] - 1 if event_dict[
'Home_Goalie'] != '' else len(home_players)
away_skaters = event_dict['Away_Players'] - 1 if event_dict[
'Away_Goalie'] != '' else len(away_players)
except KeyError:
# Getting a key error here means that home/away goalie isn't there..which means home/away players are empty
home_skaters = 0
away_skaters = 0
event_dict['Strength'] = 'x'.join([str(home_skaters), str(away_skaters)])
event_dict['Ev_Zone'] = which_zone(event[5])
if 'PENL' in event[4]:
event_dict['Type'] = get_penalty(event[5])
else:
event_dict['Type'] = shot_type(event[5]).upper()
event_dict['Time_Elapsed'] = str(event[3])
if event[3] != '':
event_dict['Seconds_Elapsed'] = shared.convert_to_seconds(event[3])
else:
event_dict['Seconds_Elapsed'] = ''
# I like getting the event players from the json
if not if_plays_in_json:
if event_dict['Event'] in [
'GOAL', 'SHOT', 'MISS', 'BLOCK', 'PENL', 'FAC', 'HIT', 'TAKE',
'GIVE'
]:
event_dict.update(get_event_players(
event, players, home_team)) # Add players involves in event
return [event_dict, current_score]
def populate_players(event_dict, players, away_players, home_players):
"""
Populate away and home player info (and num skaters on each side)
NOTE: Could probably do this in a much neater way...
: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 j in range(6):
try:
name = shared.fix_name(away_players[j][0].upper())
event_dict['awayPlayer{}'.format(j + 1)] = name
event_dict['awayPlayer{}_id'.format(
j + 1)] = players['Away'][name]['id']
except KeyError:
event_dict['awayPlayer{}_id'.format(j + 1)] = 'NA'
except IndexError:
event_dict['awayPlayer{}'.format(j + 1)] = ''
event_dict['awayPlayer{}_id'.format(j + 1)] = ''
try:
name = shared.fix_name(home_players[j][0].upper())
event_dict['homePlayer{}'.format(j + 1)] = name
event_dict['homePlayer{}_id'.format(
j + 1)] = players['Home'][name]['id']
except KeyError:
event_dict['homePlayer{}_id'.format(j + 1)] = 'NA'
except IndexError:
event_dict['homePlayer{}'.format(j + 1)] = ''
event_dict['homePlayer{}_id'.format(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)