def test_get_game_logs(self):
expected_columns = [
'DATE', 'AGE', 'TEAM', 'HOME/AWAY', 'OPPONENT', 'RESULT', 'GS',
'MP', 'FG', 'FGA', 'FG%', '3P', '3PA', '3P%', 'FT', 'FTA', 'FT%',
'ORB', 'DRB', 'TRB', 'AST', 'STL', 'BLK', 'TOV', 'PF', 'PTS',
'GAME_SCORE', '+/-'
]
df = get_game_logs('Stephen Curry', '2015-10-11', '2016-10-11')
self.assertCountEqual(list(df.columns), expected_columns)
df = get_game_logs('Pau Gasol',
'2010-01-12',
'2010-01-20',
playoffs=False)
self.assertEqual(len(df), 2)
def view_mvp_candidates():
mvp_candidates = [
'LeBron James', 'Giannis Antetokounmpo', 'James Harden',
'Anthony Davis', 'Kawhi Leonard', 'Jimmy Butler', 'Chris Paul'
]
display_rank = 1
for mvp_candidate in mvp_candidates:
print(f"Our Ranking #{display_rank}: {mvp_candidate}")
display_rank += 1
print()
mvp_candidate = input("Whose stats would like to see? ")
# ADD Check Condition Here.
# Loop if Player not one of the above.
# Error when entry is 1.
game_logs = get_game_logs(mvp_candidate,
'2019-10-22',
'2020-03-01',
playoffs=False)
game_stats = get_stats(mvp_candidate,
stat_type='PER_GAME',
playoffs=False,
career=False)
print(f"-----------{mvp_candidate}'s 2019-2020 GAME LOGS-----------")
print(game_logs)
print(f"-----------{mvp_candidate}'s CAREER GAME STATS-----------")
print(game_stats)
def player_game_logs(player, start_date, end_date, playoffs):
print("Saving into excel sheet... ")
writer = ExcelWriter("./Excel-Sheets/%s_GAME_LOG.xlsx" % player)
player_game_log_df = get_game_logs(player, start_date, end_date, playoffs)
player_game_log_df.to_excel(writer, sheet_name=player, startrow=0, startcol=0, index=False, header=True)
writer.save()
def get_players_game_log(players):
"""
scrapes the game log for the list of players and creates a CSV for each
:param players: players to scrape data
:return: void
"""
for player in players:
# TODO: create cache to download only missing parts of the data.
# TODO: can make easy cache to get last dates or cpmlex for all missing data
try:
temp = get_game_logs(player,
"2020-10-10",
"2021-10-10",
playoffs=False)
path = r"C:\Users\gilad\PycharmProjects\nba\log-" + str(
player) + "20-21.csv"
temp.to_csv(path, index=False)
except:
print("no data for " + player)
def user_input():
team_or_player = input("What stats would you like to see: (1)Team or (2)Player ")
if team_or_player == '1':
user_team = input("What team would you like to view? ")
user_season = int(input("Enter the desired season: "))
user_stats = input("Enter the number for what stat you would like to see: \n1.) Roster List\n"
"2.) Team Stats\n"
"3.) Team Opponent Stats\n"
"4.) Roster Stats\n"
"5.) Misc Stats\n")
print_or_excel = input("Would you like to save information in an excel sheet (Y/N)? ").upper()
if user_stats == '1':
if print_or_excel == 'Y':
roster_list(user_team, user_season)
print("Saved")
elif print_or_excel == 'N':
print(get_roster(user_team, user_season))
elif user_stats == '2':
data_format = input("Enter data format: (TOTAL | PER_GAME): ").upper()
if print_or_excel == 'Y':
team_stats(user_team, user_season, data_format)
print(get_team_stats(user_team, user_season, data_format))
elif print_or_excel == 'N':
print(get_team_stats(user_team, user_season, data_format))
elif user_stats == '3':
data_format = input("Enter data format: (TOTAL | PER_GAME: ").upper()
if print_or_excel == 'Y':
team_opp_stats(user_team, user_season, data_format)
print(get_opp_stats(user_team, user_season, data_format))
elif print_or_excel == 'N' or 'n':
print(get_opp_stats(user_team, user_season, data_format))
elif user_stats == '4':
user_playoffs = input("Would you like playoff roster stats? (Y/N) ")
data_format = input("Enter data format: (TOTAL | PER_GAME): ").upper()
playoffs = False
if user_playoffs == 'Y':
playoffs = True
if print_or_excel == 'Y':
roster_stats(user_team, user_season, data_format, playoffs)
print(get_roster_stats(user_team, user_season, data_format, playoffs))
elif print_or_excel == 'N':
print(get_roster_stats(user_team, user_season, data_format, playoffs))
elif user_stats == '5':
if print_or_excel == 'Y':
team_misc(user_team, user_season)
print(get_team_misc(user_team, user_season))
elif print_or_excel == 'N':
print(get_team_misc(user_team, user_season))
elif team_or_player == '2':
user_player = input('What player do you like to view? ')
print_or_excel = input("Save into excel sheet? (Y/N): ").upper()
user_stats = input("Enter the number for what stat you would like to see: \n"
"1.) Player Stats\n"
"2.) Game Logs\n")
user_playoffs = input("Playoff Stats? (Y/N): ").upper()
if user_stats == '1':
stat_type = input("Enter stat type: (PER_GAME | PER_MINUTE | PER_POSS | ADVANCED): ").upper()
user_career = input("Career Stats? (Y/N): ").upper()
if user_playoffs == 'Y':
playoffs = True
else:
playoffs = False
if user_career == 'Y':
career = True
else:
career = False
if print_or_excel == "Y":
player_stats(user_player, stat_type, playoffs, career)
print(get_stats(user_player, stat_type, playoffs, career))
elif print_or_excel == "N":
print(get_stats(user_player, stat_type, playoffs, career))
elif user_stats == '2':
user_start_date = input("Enter start date: (YYYY-MM-DD): ").upper()
user_end_date = input("Enter end date: (YYYY-MM-DD): ")
if user_playoffs == 'Y':
playoffs = True
else:
playoffs = False
if print_or_excel == "Y":
player_game_logs(user_player, user_start_date, user_end_date, playoffs)
print(get_game_logs(user_player, user_start_date, user_end_date, playoffs))
elif print_or_excel == "N":
print(get_game_logs(user_player, user_start_date, user_end_date, playoffs))
html = urlopen(url)
except:
match = True
soup = BeautifulSoup(html, features="html.parser")
player_page = soup.find('h1', itemprop="name").getText().strip()
if player_page == player_name:
return url
id_num += 1
return
def get_player_url_endpoint(player_name, num):
first_name, last_name = player_name.split(' ')
url_endpoint = last_name[0] + '/'
digit_id = '0' + str(num)
if len(last_name) < 5:
url_endpoint += last_name + first_name[:2] + digit_id + '.html'
else:
url_endpoint += last_name[:5] + first_name[:2] + digit_id + '.html'
# print("url_endpoint:", url_endpoint.lower())
return url_endpoint.lower()
# scrape_player_season_data('Ray Allen')
stats = get_game_logs('Ray Allen', '2010-01-01', '2014-04-01', False)
print(stats)
# players_file.close()
print('\n\nGetting Game Logs\n\n')
#Get_game_logs for each player and Save into .csv file
players = iter(allPlayers)
for player in players:
#Maybe dont need
if player is None:
next(players)
print(f'\n\n{player} logs \n')
spacesRemoved = player.replace(" ", "")
if spacesRemoved.isalpha() == False:
next(players)
else:
gameLogs = get_game_logs(player, oneYear, dateString)
#Save gamelogs to .csv
if gameLogs is None:
next(players)
else:
gameLogs.to_csv(f'PlayerGameLogs/{player}.csv', index=False)
print(gameLogs)
###########################################################################
###################WORK IN PROGRESS ########################################
#######Trying to figure out how to save a dataframe to SQL
# https://www.dataquest.io/blog/sql-insert-tutorial/
# #One Player in Game logs
# gameLog = get_game_logs('Stephen Curry', oneYear, dateString)
s = get_roster_stats('GSW', 2019, data_format='PER_GAME', playoffs=False)
print(s)
s = get_team_misc('GSW', 2019)
print(s)
from basketball_reference_scraper.players import get_stats, get_game_logs
s = get_stats('Stephen Curry',
stat_type='PER_GAME',
playoffs=False,
career=False)
print(s)
df = get_game_logs('LeBron James', '2010-01-19', '2014-01-20', playoffs=False)
print(df)
from basketball_reference_scraper.seasons import get_schedule, get_standings
s = get_schedule(2018, playoffs=False)
print(s)
s = get_standings(date='2020-01-06')
print(s)
from basketball_reference_scraper.box_scores import get_box_scores
s = get_box_scores('2020-01-13',
'CHI',
'BOS',
# Loop for pulling roster data
'''
output = []
for team in teams:
df1 = get_roster(team, 2021)
df1 = df1[['PLAYER','POS']]
df1['TEAM'] = team
output.append(df1)
results = pd.concat(output,ignore_index=True)
results.to_csv('2021 Roster.csv')
'''
df = pd.read_csv('2020 Playoff Roster.csv', index_col = 0)
df.drop_duplicates(subset=['PLAYER'],keep='first',inplace=True)
df = list(df['PLAYER'])
#df = df[165:]
#print(get_game_logs('Joel Embiid','2020-08-01', '2020-10-11', playoffs=True))
for i,player in enumerate(df):
if i == 0:
game_log = get_game_logs(player, '2020-08-01', '2020-10-11', playoffs=True)
df_to_add = get_game_logs(player, '2020-08-01', '2020-10-11', playoffs=True)
game_log = pd.concat([game_log, df_to_add])
game_log.to_csv('2019-2020 Game Log Playoffs.csv')
from basketball_reference_scraper.teams import get_roster, get_team_stats, get_opp_stats, get_roster_stats, get_team_misc
from basketball_reference_scraper.players import get_stats, get_game_logs
import pandas as pd
#Read Roster CSV
df = pd.read_csv('2019 Playoff Roster.csv', index_col=0)
# for Keep, we want last, because a player could have been traded to another team. They would only play for Playoffs for that last team
df.drop_duplicates(subset=['PLAYER'], keep='last', inplace=True)
#pd.set_option('display.max_rows', df.shape[0]+1)
#print(df.duplicated())
df = list(df['PLAYER'])
#df = df[50:]
#print(get_game_logs('Jayson Tatum','2019-04-13', '2019-06-13', playoffs=True))
for i, player in enumerate(df):
if i == 0:
game_log = get_game_logs(player,
'2018-10-16',
'2019-04-10',
playoffs=False)
df_to_add = get_game_logs(player,
'2018-10-16',
'2019-04-10',
playoffs=False)
game_log = pd.concat([game_log, df_to_add])
game_log.to_csv('2018-2019 Game Log.csv')
