def ytd(self):
print("\nFINDING YEAR TO DATE'S DATA:")
date_obj = DateManipulation()
prev_date = '2018-03-29'
last_date = date_obj.get_date_for_num_days(2)
#print(prev_date)
file = open(self.output_file, 'w')
data = batting_stats_range(prev_date, last_date)
for i, row in data.iterrows():
file.write(str(row))
file.write("\n")
file.close()
player = self.create_player()
player.print_player()
return data
def write_batting_for_single_date(self, output_file):
data = batting_stats_range(self.find_prev_date(),)
file = open(output_file, 'w')
for i, row in data.iterrows():
file.write(str(row))
file.write("\n")
file.close()
return data
def write_batting_for_date(self, outputfile):
print(self.find_prev_date())
data = batting_stats_range("2018-03-29", self.find_prev_date())
file = open(outputfile, 'w')
for i, row in data.iterrows():
file.write(str(row))
file.write("\n")
file.close()
return data
def write_data_to_text(self, years_before):
file = open(self.output_file, 'w')
calc_year = self.year - years_before
data = batting_stats_range(
str(calc_year) + "-03-25",
str(calc_year) + "-10-01")
for i, row in data.iterrows():
file.write(str(row))
file.write("\n")
file.close()
return data
def get_raw_stats(season_days_so_far, len_days, ct):
df_lst = []
for i in range(len(season_days_so_far)):
ct += 1
d = season_days_so_far[i]
try:
tmp_df = _pybaseball.batting_stats_range(str(d), )
tmp_df['DATE'] = d
pct_complete = int(round(ct/len_days * 100, 0))
print('Compiling... ', d)
print("{}%".format(pct_complete))
clear_output(wait=True)
df_lst.append(tmp_df)
except IndexError:
continue
return df_lst, ct
def write_file_for_dates(self, prev_date, late_date):
date_obj = DateManipulation()
#print(prev_date)
file = open(self.output_file, 'w')
data = batting_stats_range(prev_date, late_date)
for i, row in data.iterrows():
file.write(str(row))
file.write("\n")
file.close()
return data
def get_raw_stats(season_days_so_far, len_days, ct):
df_lst = []
for i in range(len(season_days_so_far)):
ct += 1
#print(ct, len_days)
d = season_days_so_far[i]
try:
tmp_df = _pybaseball.batting_stats_range(str(d), )
tmp_df['DATE'] = d
pct_complete = int(round(ct / len_days * 100, 0))
print('FIRST-TIME-USE INITIALIZE')
print('processing... ', d)
print("{}%".format(pct_complete))
clear_output(wait=True)
df_lst.append(tmp_df)
except IndexError:
continue
return _pd.concat(df_lst, ignore_index=True), ct
def write_file_for_range(self, num_days):
date_obj = DateManipulation()
prev_date = date_obj.get_date_for_num_days(num_days)
last_date = date_obj.get_date_for_num_days(1)
#print(prev_date)
file = open(self.output_file, 'w')
data = batting_stats_range(prev_date, last_date)
for i, row in data.iterrows():
file.write(str(row))
file.write("\n")
file.close()
return data
currentYear = currentDate.year
currentDay = date.today()
recentDay = currentDay - timedelta(days=7)
currentDay = str(currentDay)
recentDay = str(recentDay)
else:
currentYear = currentDate.year -1
currentDay = None
recentDay = None
pitchingData = pitching_stats(currentYear)
battingData = batting_stats(currentYear)
if currentDay is not None:
recentPitchingData = pitching_stats_range(recentDay, currentDay)
recentBattingData = batting_stats_range(recentDay, currentDay)
recentPitchingDataFile = open("../public/json/pitcherRankingsRecent.json", "w")
recentPitchingDataFile.write(json.dumps(json.loads(recentPitchingData.reset_index().to_json(orient='index')), indent=2))
recentPitchingDataFile.close()
recentBattingDataFile = open("../public/json/batterRankingsRecent.json", "w")
recentBattingDataFile.write(json.dumps(json.loads(recentBattingData.reset_index().to_json(orient='index')), indent=2))
recentBattingDataFile.close()
else:
recentPitchingDataFile = open("../public/json/pitcherRankingsRecent.json", "w")
recentPitchingDataFile.write('{}')
recentPitchingDataFile.close()
recentBattingDataFile = open("../public/json/batterRankingsRecent.json", "w")
for month in months:
print(
f'Now Downloading : start_dt = {year}-{month}-01, end_dt={year}-{month}-{months_days[month]}'
)
start_dt = f'{year}-{month}-01'
end_dt = f'{year}-{month}-{months_days[month]}'
filename = f'{year}-{month}_batting.csv'
file_path = folder_path.joinpath(filename)
if file_path.exists():
print('-' * 40)
print(f'{filename} already dowaloaded !')
print('-' * 40)
continue
try:
print('-' * 40)
print(f'Start with {filename}')
print('-' * 40)
data = batting_stats_range(start_dt=start_dt, end_dt=end_dt)
data.to_csv(file_path, encoding='utf_8_sig')
print('-' * 40)
print(f'Done with {filename}')
print('-' * 40)
except Exception as e:
print(f'\nError\n{e}')
with open(folder_path.joinpath('errorlog.txt'), 'a') as logfile:
logfile.write(f'\noccur at {start_dt} ~ {end_dt}')
continue
time.sleep(10)
end_date = today
# get a datestring to use to cycle through to collect data
date_string = pd.date_range(start_date, end_date, freq='D')
'''SRART COMPILING'''
#loop through all the days
for active_date in date_string:
#change the format of 'date' to a 'string' since that what the website needs
if (active_date.month > 2) & (active_date.month < 12):
loop_date = active_date.strftime("%Y-%m-%d")
#try to scrape data, on error go to next day and try again
try:
#get stats
batting = batting_stats_range(loop_date, )
pitching = pitching_stats_range(loop_date, )
#now clean up column names by adding 'p_' and 'b_' to pitching and batting stats
batting_specific_columns = [
(i, 'b_' + i) for i in batting.iloc[:, 6:].columns.values
]
batting.rename(columns=dict(batting_specific_columns),
inplace=True)
pitching_specific_columns = [
(i, 'p_' + i) for i in pitching.iloc[:, 6:].columns.values
]
pitching.rename(columns=dict(pitching_specific_columns),
inplace=True)
import pybaseball
import pandas as pd
# Gather all hitter data from 2015 thru 2020
data = pybaseball.batting_stats_range('2008-01-01', '2020-12-31')
# Cull list to only include hitters w/ @ least 1610 PAs
# Desired sample size based on XBH stabilization (https://library.fangraphs.com/principles/sample-size/)
data = data[data.PA >= 1610]
# Saving this data to a csv file for easier access
data.to_csv('./data/gather_hitters.csv')
def getAllPlayers(year):
data = batting_stats_range(year + '-04-01', year + '-09-30')
df = pd.DataFrame(data)
allPlayers = df['Name'].tolist()
return allPlayers
def test_batting_stats_range_single_date():
stats = batting_stats_range('2019-05-01', )
assert not stats.empty
