def _retrieve_all_teams(year, season_file=None):
"""
Find and create Team instances for all teams in the given season.
For a given season, parses the specified NBA stats table and finds all
requested stats. Each team then has a Team instance created which includes
all requested stats and a few identifiers, such as the team's name and
abbreviation. All of the individual Team instances are added to a list.
Parameters
----------
year : string
The requested year to pull stats from.
season_file : string (optional)
Link with filename to the local season page.
Returns
-------
tuple
Returns a ``tuple`` of the team_data_dict and year which represent all
stats for all teams, and the given year that should be used to pull
stats from, respectively.
"""
team_data_dict = {}
if not year:
year = utils._find_year_for_season('nba')
# Given the delays to the NBA season in 2020, the default season
# selection logic is no longer valid after the original season should
# have concluded. In this case, the previous season should be pulled
# instead.
if year == 2021:
try:
doc = pq(SEASON_PAGE_URL % year)
except HTTPError:
year = str(int(year) - 1)
# If stats for the requested season do not exist yet (as is the case
# right before a new season begins), attempt to pull the previous
# year's stats. If it exists, use the previous year instead.
if not utils._url_exists(SEASON_PAGE_URL % year) and \
utils._url_exists(SEASON_PAGE_URL % str(int(year) - 1)):
year = str(int(year) - 1)
doc = utils._pull_page(SEASON_PAGE_URL % year, season_file)
teams_list = utils._get_stats_table(doc, 'div#all_team-stats-base')
opp_teams_list = utils._get_stats_table(doc, 'div#all_opponent-stats-base')
if not teams_list and not opp_teams_list:
utils._no_data_found()
return None, None
for stats_list in [teams_list, opp_teams_list]:
team_data_dict = _add_stats_data(stats_list, team_data_dict)
return team_data_dict, year
def _retrieve_all_teams(year, season_page, offensive_stats, defensive_stats):
"""
Find and create Team instances for all teams in the given season.
For a given season, parses the specified NCAAF stats table and finds
all requested stats. Each team then has a Team instance created which
includes all requested stats and a few identifiers, such as the team's
name and abbreviation. All of the individual Team instances are added
to a list.
Note that this method is called directly once Teams is invoked and does
not need to be called manually.
Parameters
----------
year : string
The requested year to pull stats from.
season_page : string (optional)
Link with filename to the local season stats page.
offensive_stats : string (optional)
Link with filename to the local offensive stats page.
defensive_stats : string (optional)
Link with filename to the local defensive stats page.
Returns
-------
tuple
Returns a ``tuple`` of the team_data_dict and year which represent all
stats for all teams, and the given year that should be used to pull
stats from, respectively.
"""
team_data_dict = {}
if not year:
year = utils._find_year_for_season('ncaaf')
# If stats for the requested season do not exist yet (as is the case
# right before a new season begins), attempt to pull the previous
# year's stats. If it exists, use the previous year instead.
if not utils._url_exists(SEASON_PAGE_URL % year) and \
utils._url_exists(SEASON_PAGE_URL % str(int(year) - 1)):
year = str(int(year) - 1)
doc = utils._pull_page(SEASON_PAGE_URL % year, season_page)
teams_list = utils._get_stats_table(doc, 'div#div_standings')
offense_doc = utils._pull_page(OFFENSIVE_STATS_URL % year, offensive_stats)
offense_list = utils._get_stats_table(offense_doc, 'table#offense')
defense_doc = utils._pull_page(DEFENSIVE_STATS_URL % year, defensive_stats)
defense_list = utils._get_stats_table(defense_doc, 'table#defense')
if not teams_list and not offense_list and not defense_list:
utils._no_data_found()
for stats_list in [teams_list, offense_list, defense_list]:
team_data_dict = _add_stats_data(stats_list, team_data_dict)
return team_data_dict, year
def _retrieve_all_teams(year, standings_file=None, teams_file=None):
"""
Find and create Team instances for all teams in the given season.
For a given season, parses the specified MLB stats table and finds all
requested stats. Each team then has a Team instance created which includes
all requested stats and a few identifiers, such as the team's name and
abbreviation. All of the individual Team instances are added to a list.
Parameters
----------
year : string
The requested year to pull stats from.
standings_file : string (optional)
Link with filename to the local standings page.
teams_file : string (optional)
Link with filename to the local teams page.
Returns
-------
tuple
Returns a ``tuple`` of the team_data_dict and year which represent all
stats for all teams, and the given year that should be used to pull
stats from, respectively.
"""
team_data_dict = {}
if not year:
year = utils._find_year_for_season('mlb')
# If stats for the requested season do not exist yet (as is the case
# right before a new season begins), attempt to pull the previous
# year's stats. If it exists, use the previous year instead.
if not utils._url_exists(STANDINGS_URL % year) and \
utils._url_exists(STANDINGS_URL % str(int(year) - 1)):
year = str(int(year) - 1)
doc = utils._pull_page(STANDINGS_URL % year, standings_file)
div_prefix = 'div#all_expanded_standings_overall'
standings = utils._get_stats_table(doc, div_prefix)
doc = utils._pull_page(TEAM_STATS_URL % year, teams_file)
div_prefix = 'div#all_teams_standard_%s'
batting_stats = utils._get_stats_table(doc, div_prefix % 'batting')
pitching_stats = utils._get_stats_table(doc, div_prefix % 'pitching')
if not standings and not batting_stats and not pitching_stats:
utils._no_data_found()
return None, None
for stats_list in [standings, batting_stats, pitching_stats]:
team_data_dict = _add_stats_data(stats_list, team_data_dict)
return team_data_dict, year
def _pull_schedule(self, abbreviation, year):
"""
Download and create objects for the team's schedule.
Given a team abbreviation and season, first download the team's
schedule page and convert to a PyQuery object, then create a Game
instance for every game in the team's schedule and append it to the
'_games' property.
Parameters
----------
abbreviation : string
A team's short name, such as 'DET' for the Detroit Pistons.
year : string
The requested year to pull stats from.
"""
if not year:
year = utils._find_year_for_season('nba')
# Given the delays to the NBA season in 2020, the default season
# selection logic is no longer valid after the original season
# should have concluded. In this case, the previous season should
# be pulled instead.
if year == 2021:
try:
doc = pq(SCHEDULE_URL % (abbreviation.lower(), year))
except HTTPError:
year = str(int(year) - 1)
# If stats for the requested season do not exist yet (as is the
# case right before a new season begins), attempt to pull the
# previous year's stats. If it exists, use the previous year
# instead.
if not utils._url_exists(SCHEDULE_URL % (abbreviation.lower(),
year)) and \
utils._url_exists(SCHEDULE_URL % (abbreviation.lower(),
str(int(year) - 1))):
year = str(int(year) - 1)
doc = pq(SCHEDULE_URL % (abbreviation, year))
schedule = utils._get_stats_table(doc, 'table#games')
if not schedule:
utils._no_data_found()
return
self._add_games_to_schedule(schedule)
if 'id="games_playoffs"' in str(doc):
playoffs = utils._get_stats_table(doc, 'table#games_playoffs')
self._add_games_to_schedule(playoffs, True)
def _pull_schedule(self, abbreviation, year):
"""
Download and create objects for the team's schedule.
Given a team abbreviation and season, first download the team's
schedule page and convert to a PyQuery object, then create a Game
instance for every game in the team's schedule and append it to the
'_games' property.
Parameters
----------
abbreviation : string
A team's short name, such as 'PURDUE' for the Purdue Boilermakers.
year : string
The requested year to pull stats from.
"""
if not year:
year = utils._find_year_for_season('ncaab')
# If stats for the requested season do not exist yet (as is the
# case right before a new season begins), attempt to pull the
# previous year's stats. If it exists, use the previous year
# instead.
if not utils._url_exists(SCHEDULE_URL % (abbreviation.lower(),
year)) and \
utils._url_exists(SCHEDULE_URL % (abbreviation.lower(),
str(int(year) - 1))):
year = str(int(year) - 1)
doc = pq(SCHEDULE_URL % (abbreviation.lower(), year))
schedule = utils._get_stats_table(doc, 'table#schedule')
if not schedule:
utils._no_data_found()
return
for item in schedule:
if 'class="thead"' in str(item):
continue
game = Game(item)
self._games.append(game)
def _pull_schedule(self, abbreviation, year):
"""
Download and create objects for the team's schedule.
Given a team abbreviation and season, first download the team's
schedule page and convert to a PyQuery object, then create a Game
instance for every game in the team's schedule and append it to the
'_games' property.
Parameters
----------
abbreviation : string
A team's short name, such as 'NWE' for the New England Patriots.
year : string
The requested year to pull stats from.
"""
if not year:
year = utils._find_year_for_season('nfl')
# If stats for the requested season do not exist yet (as is the
# case right before a new season begins), attempt to pull the
# previous year's stats. If it exists, use the previous year
# instead.
if not utils._url_exists(SCHEDULE_URL % (abbreviation.lower(),
year)) and \
utils._url_exists(SCHEDULE_URL % (abbreviation.lower(),
str(int(year) - 1))):
year = str(int(year) - 1)
doc = pq(SCHEDULE_URL % (abbreviation.lower(), year))
schedule = utils._get_stats_table(doc, 'table#gamelog%s' % year)
if not schedule:
utils._no_data_found()
return
self._add_games_to_schedule(schedule, REGULAR_SEASON, year)
if 'playoff_gamelog%s' % year in str(doc):
playoffs = utils._get_stats_table(doc,
'table#playoff_gamelog%s' % year)
self._add_games_to_schedule(playoffs, POST_SEASON, year)
def _retrieve_all_teams(year,
basic_stats=None,
basic_opp_stats=None,
adv_stats=None,
adv_opp_stats=None):
"""
Find and create Team instances for all teams in the given season.
For a given season, parses the specified NCAAB stats table and finds all
requested stats. Each team then has a Team instance created which includes
all requested stats and a few identifiers, such as the team's name and
abbreviation. All of the individual Team instances are added to a list.
Note that this method is called directly once Teams is invoked and does not
need to be called manually.
Parameters
----------
year : string
The requested year to pull stats from.
basic_stats : string (optional)
Link with filename to the local basic stats page.
basic_opp_stats : string (optional)
Link with filename to the local basic opponent stats page.
adv_stats : string (optional)
Link with filename to the local advanved stats page.
adv_opp_stats : string (optional)
Link with filename to the local advanced opponents stats page.
Returns
-------
tuple
Returns a ``tuple`` of the team_data_dict and year which represent all
stats for all teams, and the given year that should be used to pull
stats from, respectively.
"""
team_data_dict = {}
if not year:
year = utils._find_year_for_season('ncaab')
# If stats for the requested season do not exist yet (as is the case
# right before a new season begins), attempt to pull the previous
# year's stats. If it exists, use the previous year instead.
if not utils._url_exists(BASIC_STATS_URL % year) and \
utils._url_exists(BASIC_STATS_URL % str(int(year) - 1)):
year = str(int(year) - 1)
doc = utils._pull_page(BASIC_STATS_URL % year, basic_stats)
teams_list = utils._get_stats_table(doc, 'table#basic_school_stats')
doc = utils._pull_page(BASIC_OPPONENT_STATS_URL % year, basic_opp_stats)
opp_list = utils._get_stats_table(doc, 'table#basic_opp_stats')
doc = utils._pull_page(ADVANCED_STATS_URL % year, adv_stats)
adv_teams_list = utils._get_stats_table(doc, 'table#adv_school_stats')
doc = utils._pull_page(ADVANCED_OPPONENT_STATS_URL % year, adv_opp_stats)
adv_opp_list = utils._get_stats_table(doc, 'table#adv_opp_stats')
if not teams_list and not opp_list and not adv_teams_list \
and not adv_opp_list:
utils._no_data_found()
return None, None
for stats_list in [teams_list, opp_list, adv_teams_list, adv_opp_list]:
team_data_dict = _add_stats_data(stats_list, team_data_dict)
return team_data_dict, year
def test_invalid_url_exception_returns_false(self, *args, **kwargs):
response = utils._url_exists('http://www.exception.com')
assert not response
def test_404_url_returns_false(self, *args, **kwargs):
response = utils._url_exists('http://www.404.com/doesnt/exist')
assert not response
def test_valid_url_returns_true(self, *args, **kwargs):
response = utils._url_exists('http://www.good_url.com/this/is/valid')
assert response