def scrape_games(games, if_scrape_shifts, data_format='csv'):
"""
Scrape a list of games
:param games: list of game_ids
:param if_scrape_shifts: Boolean indicating whether to also scrape shifts
:param data_format: format you want data in - csv or json (csv is default)
:param preseason: Boolean indicating whether include preseason games (default if False)
:return: Json string or None
"""
check_data_format(data_format)
json_dfs = dict() # Holds json of data if choose to return that
# Create List of game_id's and dates
games_list = json_schedule.get_dates(games)
# Scrape pbp and shifts
pbp_df, shifts_df = scrape_list_of_games(games_list, if_scrape_shifts)
if data_format.lower() == 'csv':
to_csv(str(random.randint(1, 101)), pbp_df, shifts_df)
else:
if pbp_df is not None:
json_dfs['pbp'] = pbp_df.to_dict('records')
if shifts_df is not None:
json_dfs['shifts'] = shifts_df.to_dict('records')
# Print all errors associated with scrape call
game_scraper.print_errors()
if len(json_dfs.keys()) > 0:
return json.dumps(json_dfs)
def scrape_seasons(seasons,
if_scrape_shifts,
data_format='csv',
preseason=False):
"""
Given list of seasons it scrapes all the seasons
:param seasons: list of seasons
:param if_scrape_shifts: Boolean indicating whether to also scrape shifts
:param data_format: format you want data in - csv or json (csv is default)
:param preseason: Boolean indicating whether include preseason games (default if False)
:return: Json string or None
"""
check_data_format(data_format)
json_dfs = {
'pbp': dict(),
'shifts': dict()
} # Holds json of data if choose to return that
for season in seasons:
from_date = '-'.join([str(season), '9', '1'])
to_date = '-'.join([str(season + 1), '7', '1'])
games = json_schedule.scrape_schedule(from_date, to_date, preseason)
pbp_df, shifts_df = scrape_list_of_games(games, if_scrape_shifts)
if data_format.lower() == 'csv':
to_csv(str(season) + str(season + 1), pbp_df, shifts_df)
else:
if pbp_df is not None:
json_dfs['pbp'][str(season)] = pbp_df.to_dict('records')
if shifts_df is not None:
json_dfs['shifts'][str(season)] = shifts_df.to_dict('records')
# Print all errors associated with scrape call
game_scraper.print_errors()
# If we have something in there that means json was chosen
if len(json_dfs.keys()) > 0:
return json.dumps(json_dfs)
def scrape_date_range(from_date,
to_date,
if_scrape_shifts,
data_format='csv',
preseason=False):
"""
Scrape games in given date range
:param from_date: date you want to scrape from
:param to_date: date you want to scrape to
:param if_scrape_shifts: Boolean indicating whether to also scrape shifts
:param data_format: format you want data in - csv or json (csv is default)
:param preseason: Boolean indicating whether include preseason games (default if False)
:return: Json string or None
"""
check_data_format(data_format)
check_valid_dates(from_date, to_date)
json_dfs = dict() # Holds json of data if choose to return that
games = json_schedule.scrape_schedule(from_date, to_date, preseason)
pbp_df, shifts_df = scrape_list_of_games(games, if_scrape_shifts)
if data_format.lower() == 'csv':
to_csv(from_date + '--' + to_date, pbp_df, shifts_df)
else:
if pbp_df is not None:
json_dfs['pbp'] = pbp_df.to_dict('records')
if shifts_df is not None:
json_dfs['shifts'] = shifts_df.to_dict('records')
# Print all errors associated with scrape call
game_scraper.print_errors()
# If we have something in there that means json was chosen
if len(json_dfs.keys()) > 0:
return json.dumps(json_dfs)