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_Type'] = str(change_event_name(event['result']['eventTypeId']))
play['Event_Desc'] = event['result']['description']
# play['Event_Num'] = event['about']['eventIdx']
# commented out as eventIdx in json does not match event num from html
play['Period'] = event['about']['period']
play['Time_Elapsed'] = convert_to_seconds(event['about']['periodTime'])
play['Away_Score'] = event['about']['goals']['away']
play['Home_Score'] = event['about']['goals']['home']
if event['result']['eventTypeId'] == "SHOT":
if "Snap" in event['result']['secondaryType']:
play["Secondary_Type"] = "Snap"
elif "Slap" in event['result']['secondaryType']:
play["Secondary_Type"] = "Slap"
elif "Backhand" in event['result']['secondaryType']:
play["Secondary_Type"] = "Backhand"
elif "Tip-In" in event['result']['secondaryType']:
play["Secondary_Type"] = "Tip-In"
elif "Wrap-around" in event['result']['secondaryType']:
play["Secondary_Type"] = "Wrap-around"
elif "Deflected" in event['result']['secondaryType']:
play["Secondary_Type"] = "Deflected"
if event['result']['eventTypeId'] == "MISSED_SHOT":
play["Secondary_Type"] = event['result']['description'].split('-')[1]
if event['result']['eventTypeId'] == "PENALTY":
play["Secondary_Type"] = event['result']['secondaryType']
play["Penl_Length"] = event['result']['penaltyMinutes']
if event['result']['eventTypeId'] == "GOAL":
play["Secondary_Type"] = event['result']['secondaryType']
# If there's a players key that means an event occurred on the play.
if 'players' in event.keys():
for i in range(len(event['players'])):
play['P{}_Name'.format(i + 1)] = fix_name(
event['players'][i]['player']['fullName'].upper())
play['P{}_Id'.format(i + 1)] = str(
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 analyze_shifts(shift, name, team):
"""
Analyze shifts for each player when using.
Prior to this each player (in a dictionary) has a list with each entry being a shift.
This function is only used for the html
:param shift: info on shift
:param name: player name
:param team: given team
:return: dict with info for shift
"""
shifts = dict()
shifts['Player'] = name.upper()
shifts['Period'] = '4' if shift[1] == 'OT' else shift[1]
shifts['Team'] = fix_team(team.strip(' '))
shifts['Shift'] = shift[0]
shifts['Start'] = convert_to_seconds(shift[2].split('/')[0])
shifts['End'] = convert_to_seconds(shift[3].split('/')[0])
shifts['Duration'] = convert_to_seconds(shift[4].split('/')[0])
return shifts
def parse_shift(shift):
"""
Parse shift for json
:param shift: json for shift
:return: dict with shift info
"""
shift_dict = dict()
name = fix_name(' '.join([shift['firstName'].strip(' ').upper(), shift['lastName'].strip(' ').upper()]))
shift_dict['Period'] = shift['period']
shift_dict['Player'] = name
shift_dict['Team'] = shift['teamAbbrev']
shift_dict['Shift'] = shift['shiftNumber']
if shift['eventDescription'] is None:
shift_dict['Start'] = convert_to_seconds(shift['startTime'])
shift_dict['End'] = convert_to_seconds(shift['endTime'])
shift_dict['Duration'] = convert_to_seconds(shift['duration'])
else:
shift_dict = dict()
return shift_dict
def parse_event(event):
"""
Parse each event. In the string each field is separated by a '~'.
Relevant for here: The first two are the x and y coordinates. And the 4th and 5th are the time elapsed and period.
:param event: string with info
:return: return dict with relevant info
"""
info = dict()
fields = event.split('~')
# Shootouts screw everything up so don't bother...coordinates don't matter there either way
if fields[4] == '5':
return None
info['xC'] = int(fields[0])
info['yC'] = int(fields[1])
info['Time_Elapsed'] = convert_to_seconds(fields[3])
info['Period'] = fields[4]
info['Event_Type'] = event_type(fields[8].upper())
return info
def parse_pbp(gameid):
"""
parse cleaned html and create pd Dataframe
:param gameid: game id
:return: df where each row is an event in pb
"""
html = get_pbp(gameid)
events = clean_html_pbp(html)
teams = dict()
teams['away'] = events[0][6][:3]
teams['home'] = events[0][7][:3]
score = {'home': 0, 'away': 0}
pbp_events = []
event_types = {
'PERIOD_START': 'PSTR',
'FACEOFF': 'FAC',
'BLOCKED_SHOT': 'BLOCK',
'GAME_END': 'GEND',
'GIVEAWAY': 'GIVE',
'GOAL': 'GOAL',
'HIT': 'HIT',
'MISSED_SHOT': 'MISS',
'PERIOD_END': 'PEND',
'SHOT': 'SHOT',
'STOP': 'STOP',
'TAKEAWAY': 'TAKE',
'PENALTY': 'PENL',
'Early Intermission Start': 'EISTR',
'Early Intermission End': 'EIEND',
'Shootout Completed': 'SOC',
}
for i in events[:len(events)]:
if i[4] in event_types.values() and i[0] != '#':
play = dict()
play['Home_Score'] = score['home']
play['Away_Score'] = score['away']
# play['Team_away'] = teams['away']
# play['Team_home'] = teams['home']
play['Event_Num'] = i[0]
play['Period'] = i[1]
# play['strength'] = i[2]
play['Time_Elapsed'] = convert_to_seconds(i[3])
play['Event_Type'] = i[4]
play['Event_Desc'] = i[5]
if 'Off. Zone' in i[5] and i[5][:3] == teams['home']:
play['Home_Zone'] = 'OZ'
play['Away_Zone'] = 'DZ'
elif 'Off. Zone' in i[5] and i[5][:3] == teams['away']:
play['Home_Zone'] = 'DZ'
play['Away_Zone'] = 'OZ'
elif 'Def. Zone' in i[5] and i[5][:3] == teams['home'] and i[
4] == 'BLOCK':
play['Home_Zone'] = 'OZ'
play['Away_Zone'] = 'DZ'
elif 'Def. Zone' in i[5] and i[5][:3] == teams[
'home'] and i[4] != 'BLOCK':
play['Home_Zone'] = 'DZ'
play['Away_Zone'] = 'OZ'
elif 'Def. Zone' in i[5] and i[5][:3] == teams['away'] and i[
4] == 'BLOCK':
play['Home_Zone'] = 'DZ'
play['Away_Zone'] = 'OZ'
elif 'Def. Zone' in i[5] and i[5][:3] == teams[
'away'] and i[4] != 'BLOCK':
play['Home_Zone'] = 'OZ'
play['Away_Zone'] = 'DZ'
elif 'Neu. Zone' in i[5]:
play['Home_Zone'] = 'NZ'
play['Away_Zone'] = 'NZ'
if len(i[6]) > 0 and i[6][len(i[6]) - 1][2] == 'G':
play['Away_Strength'] = len(i[6]) - 1
play['G_Pulled_Away'] = False
elif len(i[6]) > 0 and i[6][len(i[6]) - 1][2] != 'G':
play['Away_Strength'] = len(i[6])
play['G_Pulled_Away'] = True
if len(i[7]) > 0 and i[7][len(i[7]) - 1][2] == 'G':
play['Home_Strength'] = len(i[7]) - 1
play['G_Pulled_Home'] = False
elif len(i[7]) > 0 and i[7][len(i[7]) - 1][2] != 'G':
play['Home_Strength'] = len(i[7])
play['G_Pulled_Home'] = True
for player in i[6]:
n = i[6].index(player) + 1
play['A{}Name'.format(n)] = fix_name(player[0])
play['A{}Num'.format(n)] = player[1]
play['A{}Pos'.format(n)] = player[2]
for player in i[7]:
n = i[7].index(player) + 1
play['H{}Name'.format(n)] = fix_name(player[0])
play['H{}Num'.format(n)] = player[1]
play['H{}Pos'.format(n)] = player[2]
if 'Penalty Shot' in i[5]:
play['Penalty_Shot'] = True
else:
play['Penalty_Shot'] = False
if i[4] == 'FAC':
# MTL won Neu. Zone - MTL #11 GOMEZ vs TOR #37 BRENT
play['P1_Team'] = i[5][:3]
for value in teams.values():
if value != i[5][:3]:
play['P2_Team'] = value
regex = re.compile(r'(.{3})\s+#(\d+)')
desc = regex.findall(i[5]) # [[Team, num], [Team, num]]
for player in desc:
if player[0] == i[5][:3]:
play['P1_Num'] = player[1]
else:
play['P2_Num'] = player[1]
penl = {
'Hooking': 'Hooking',
'Holding': 'Holding',
'Holding the stick': 'Holding the stick',
'Interference': 'Interference',
'Roughing': 'Roughing',
'Tripping': 'Tripping',
'Unsportsmanlike conduct': 'Unsportsmanlike conduct',
'Illegal equipment': 'Illegal equipment',
'Diving': 'Embellishment',
'Embellishment': 'Embellishment',
'Broken stick': 'Broken stick',
'Delaying Game-Puck over glass':
'Delaying Game-Puck over glass',
'Delay Gm - Face-off Violation':
'Delay Gm - Face-off Violation',
'Delay of game': 'Delay of game',
'Delaying the game': 'Delay of game',
'Delay of game - bench': 'Delay of game - bench',
'Delaying Game-Ill. play goalie':
'Delaying Game-Ill. play goalie',
'Delaying Game-Smothering puck':
'Delaying Game-Smothering puck',
'Goalie leave crease': 'Goalie leave crease',
'Face-off violation-bench': 'Face-off violation-bench',
'Bench': 'Bench',
'Illegal stick': 'Illegal stick',
'Closing hand on puck': 'Closing hand on puck',
'Throwing stick': 'Throwing stick',
'Too many men/ice - bench': 'Too many men/ice - bench',
'Abusive language - bench': 'Abuse of officials - bench',
'Abuse of officials - bench': 'Abuse of officials - bench',
'Interference on goalkeeper': 'Interference on goalkeeper',
'Hi-sticking': 'High-sticking',
'Hi stick - double minor': 'High-sticking',
'Cross checking': 'Cross checking',
'Cross check - double minor': 'Cross checking',
'Slashing': 'Slashing',
'Charging': 'Charging',
'Boarding': 'Boarding',
'Kneeing': 'Kneeing',
'Clipping': 'Clipping',
'Elbowing': 'Elbowing',
'Spearing': 'Spearing',
'Butt ending': 'Butt ending',
'Head butting': 'Head butting',
'Illegal check to head': 'Illegal check to head',
'Fighting': 'Fighting',
'Instigator': 'Instigator',
'Abuse of officials': 'Abuse of officials',
'Aggressor': 'Aggressor',
'PS-': "Penalty Shot",
'PS -': "Penalty Shot",
'Misconduct': 'Misconduct',
'Game misconduct': 'Game misconduct',
'Game Misconduct': 'Game misconduct',
'Match penalty': 'Game Misconduct',
'Checking from behind': 'Checking from behind'
}
if i[4] == 'PENL':
for pen in penl:
if pen in i[5]:
play['Secondary_Type'] = penl[pen]
elif pen not in penl:
play['Secondary_Type'] = 'misc'
if '2 min' in i[5]:
play['Penl_Length'] = 2
elif '4 min' in i[5]:
play['Penl_Length'] = 4
elif '5 min' in i[5]:
play['Penl_Length'] = 5
elif '10 min' in i[5]:
play['Penl_Length'] = 10
play['P1_Team'] = i[5][:3]
for value in teams.values():
if value != i[5][:3]:
play['P2_Team'] = value
if 'TEAM' in i[5]:
play['P1_Num'] = i[5][:3]
else:
regex = re.compile(r'(.{3})\s+#(\d+)')
desc = regex.findall(i[5]) # [[Team, num], [Team, num]]
for player in desc:
if player[0] == i[5][:3]:
play['P1_Num'] = player[1]
elif player[0] != i[5][:3]:
play['P2_Num'] = player[1]
if i[4] == 'HIT':
play['P1_Team'] = i[5][:3]
for value in teams.values():
if value != i[5][:3]:
play['P2_Team'] = value
regex = re.compile(r'(.{3})\s+#(\d+)')
desc = regex.findall(i[5]) # [[Team, num], [Team, num]]
for player in desc:
if player[0] == i[5][:3]:
play['P1_Num'] = player[1]
elif player[0] != i[5][:3]:
play['P2_Num'] = player[1]
shot_types = [
'Wrist', 'Snap', 'Slap', 'Deflected', 'Tip-In', 'Backhand',
'Wrap-around'
]
if i[4] == 'BLOCK':
play['P2_Team'] = i[5][:3]
for value in teams.values():
if value != i[5][:3]:
play['P1_Team'] = value
for shot in shot_types:
if shot in i[5]:
play['Secondary_Type'] = shot
regex = re.compile(r'(.{3})\s+#(\d+)')
desc = regex.findall(i[5]) # [[Team, num], [Team, num]]
for player in desc:
if player[0] == i[5][:3]:
play['P2_Num'] = player[1]
elif player[0] != i[5][:3]:
play['P1_Num'] = player[1]
if i[4] == 'SHOT':
play['P1_Team'] = i[5][:3]
for value in teams.values():
if value != i[5][:3]:
play['P4_Team'] = value
for shot in shot_types:
if shot in i[5]:
play['Secondary_Type'] = shot
regex = re.compile(r'#(\d+)')
desc = regex.search(i[5]).groups() # num
play['P1_Num'] = desc[0]
for key, value in teams.items():
if value != i[5][:3] and key == 'away':
play['P4_Num'] = i[6][len(i[6]) - 1][1]
elif value != i[5][:3] and key == 'home':
play['P4_Num'] = i[7][len(i[7]) - 1][1]
dist = i[5].split(',')
feet = re.findall('\d+', dist[len(dist) - 1][1:])
play['Dist'] = feet[0]
miss_types = ['Over Net', 'Wide of Net', 'Crossbar', 'Goalpost']
if i[4] == 'MISS':
play['P1_Team'] = i[5][:3]
for value in teams.values():
if value != i[5][:3]:
play['P4_Team'] = value
for shot in shot_types:
if shot in i[5]:
play['Secondary_Type'] = shot
for miss in miss_types:
if miss in i[5]:
play['Tertiary_Type'] = miss
regex = re.compile(r'#(\d+)')
desc = regex.search(i[5]).groups() # num
play['P1_Num'] = desc[0]
for key, value in teams.items():
if value != i[5][:3] and key == 'away':
play['P4_Num'] = i[6][len(i[6]) - 1][1]
elif value != i[5][:3] and key == 'home':
play['P4_Num'] = i[7][len(i[7]) - 1][1]
dist = i[5].split(',')
feet = re.findall('\d+', dist[len(dist) - 1][1:])
play['Dist'] = feet[0]
if i[4] == 'GIVE' or i[4] == 'TAKE':
play['P1_Team'] = i[5][:3]
regex = re.compile(r'#(\d+)')
desc = regex.search(i[5]).groups() # num
play['P1_Num'] = desc[0]
if i[4] == 'GOAL':
play['P1_Team'] = i[5][:3]
for value in teams.values():
if value != i[5][:3]:
play['P4_Team'] = value
for shot in shot_types:
if shot in i[5]:
play['Secondary_Type'] = shot
for key, value in teams.items():
if value != i[5][:3] and key == 'away':
play['P4_Num'] = i[6][len(i[6]) - 1][1]
elif value != i[5][:3] and key == 'home':
play['P4_Num'] = i[7][len(i[7]) - 1][1]
regex = re.compile(r'#(\d+)\s+')
desc = regex.findall(
i[5]) # [num] -> ranging from 1 to 3 indices
play['P1_Num'] = desc[0]
if len(desc) >= 2:
play['P2_Num'] = desc[1]
play['P2_Team'] = i[5][:3]
if len(desc) == 3:
play['P3_Num'] = desc[2]
play['P3_Team'] = i[5][:3]
dist = i[5].split(',')
for d in dist:
if 'ft' in d:
feet = re.findall('\d+', d[1:])
play['Dist'] = feet[0]
for key, value in teams.items():
if value == i[5][:3] and key == 'away':
score['away'] += 1
elif value == i[5][:3] and key == 'home':
score['home'] += 1
if i[4] == 'STOP':
if 'ICING' in i[5]:
play['Secondary_Type'] = 'Icing'
elif 'OFFSIDE' in i[5]:
play['Secondary_Type'] = 'Offside'
pbp_events.append(play)
pbp = pd.DataFrame(pbp_events)
pbp['Game_Id'] = gameid
'''
try:
pbp['xC'] = get_coords(gameid)['xC']
except:
pass
try:
pbp['yC'] = get_coords(gameid)['yC']
except:
pass
'''
columns = [
'Game_Id', 'Period', 'Event_Num', 'Event_Type', 'Secondary_Type',
'Tertiary_Type', 'Penl_Length', 'Event_Desc', 'Time_Elapsed',
'P1_Team', 'P1_Num', 'P2_Team', 'P2_Num', 'P3_Team', 'P3_Num',
'P4_Team', 'P4_Num', 'Home_Score', 'Away_Score', 'Home_Zone',
'Away_Zone', 'Home_Strength', 'Away_Strength', 'G_Pulled_Home',
'G_Pulled_Away', 'Dist', 'H1Name', 'H1Num', 'H1Pos', 'H2Name', 'H2Num',
'H2Pos', 'H3Name', 'H3Num', 'H3Pos', 'H4Name', 'H4Num', 'H4Pos',
'H5Name', 'H5Num', 'H5Pos', 'H6Name', 'H6Num', 'H6Pos', 'A1Name',
'A1Num', 'A1Pos', 'A2Name', 'A2Num', 'A2Pos', 'A3Name', 'A3Num',
'A3Pos', 'A4Name', 'A4Num', 'A4Pos', 'A5Name', 'A5Num', 'A5Pos',
'A6Name', 'A6Num', 'A6Pos', 'Penalty_Shot'
]
pbp = pbp.reindex_axis(columns, axis=1)
# add S/V adjustment coefficients for each pbp event
def adjustment(row):
if row['Home_Score'] - row['Away_Score'] >= 3:
return 1.132, 0.895
elif row['Home_Score'] - row['Away_Score'] == 2:
return 1.074, 0.936
elif row['Home_Score'] - row['Away_Score'] == 1:
return 1.026, 0.975
elif row['Home_Score'] - row['Away_Score'] == 0:
return 0.970, 1.032
elif row['Home_Score'] - row['Away_Score'] == -1:
return 0.915, 1.103
elif row['Home_Score'] - row['Away_Score'] == -2:
return 0.882, 1.154
elif row['Home_Score'] - row['Away_Score'] <= -3:
return 0.850, 1.214
else:
return np.NaN, np.NaN
pbp['Home_Adjustment'] = pbp.apply(lambda row: adjustment(row)[0], axis=1)
pbp['Away_Adjustment'] = pbp.apply(lambda row: adjustment(row)[1], axis=1)
return pbp