def test_contract(self):
james = lebron_james()
assert james.contract
assert len(james.contract.years) == 3
assert all([year.season for year in james.contract.years])
assert all([year.amount for year in james.contract.years])
first_year = james.contract.years[0]
assert first_year.season == "2019-20"
assert first_year.amount == 37436858
assert first_year.option is None
third_year = james.contract.years[2]
assert third_year.season == "2021-22"
assert third_year.amount == 41002273
# assert third_year.option == "Player"
chamberlain = wilt_chamberlain()
assert not chamberlain.contract
shayok = PlayerPageScraper(get_resource("shayok.html")).get_content()
assert not shayok.contract
simons = PlayerPageScraper(
get_resource("anfernee_simons.html")).get_content()
assert simons.contract.years[0].option == "player"
assert simons.contract.years[1].option == "early termination"
assert simons.contract.years[2].option == "team"
def test_stats_where_partially_available(self):
mullin = PlayerPageScraper(
get_resource("chris_mullin.html")).get_content()
seasons = list(mullin.seasons)
rookie_season = seasons[0]
assert rookie_season.stats.shooting_data is None
final_season = seasons[-1]
assert final_season.stats.shooting_data
assert final_season.stats.shooting_data.avg_distance == 18.8
def test_game_log(self):
# can't store this page locally because we depend on a hyperlink inside
seasons = list(
PlayerPageScraper(
"https://www.basketball-reference.com/players/m/mbengdj01.html"
).get_content().seasons)
gl = seasons[0].game_logs()
assert len([game for game in gl
if game.played]) == seasons[0].stats.games_played
def test_career_stats_encapsulated(self):
mullin = PlayerPageScraper(
get_resource("chris_mullin.html")).get_content()
career_stats = mullin.career_stats
first_season = next(mullin.seasons)
first_playoffs = next(mullin.playoffs)
for attr in ["age", "all_star", "team", "season"]:
assert not hasattr(career_stats, attr)
assert hasattr(first_season, attr)
assert hasattr(first_playoffs, attr)
def test_playoff_games(self):
lbj = PlayerPageScraper(
"https://www.basketball-reference.com/players/j/jamesle01.html"
).get_content()
po = next(lbj.playoffs)
assert po.stats.points == 400
games = list(po.game_logs())
assert len(games) == 13
game: GameLog = games[0]
assert game.date == date(2006, 4, 22)
assert game.points == 32
assert game.assists == 11
def test_shooting_data(self):
lbj = PlayerPageScraper(
"https://www.basketball-reference.com/players/j/jamesle01.html"
).get_content()
first_season = next(lbj.seasons)
shooting_data = first_season.stats.shooting_data
assert shooting_data.heaves_made == 0
assert shooting_data.heaves_attempted == 3
assert shooting_data.corner_three_point_fgp == 0.323
assert shooting_data.corner_three_point_fga == 0.286
assert shooting_data.three_point_fga_assisted == 0.746
assert shooting_data.dunks_made == 91
assert shooting_data.dunks_fga == 0.064
assert shooting_data.two_point_fga_assisted == 0.442
assert shooting_data.fgp_by_distance
assert shooting_data.fga_by_distance
assert shooting_data.fga_by_distance.two_point == 0.855
assert shooting_data.fga_by_distance.zero_three == 0.315
assert shooting_data.fga_by_distance.three_ten == 0.168
assert shooting_data.fga_by_distance.ten_sixteen == 0.161
assert shooting_data.fga_by_distance.sixteen_three_pt == 0.211
assert shooting_data.fga_by_distance.three_point == 0.145
assert shooting_data.fgp_by_distance.two_point == 0.438
assert shooting_data.fgp_by_distance.zero_three == 0.604
assert shooting_data.fgp_by_distance.three_ten == 0.356
assert shooting_data.fgp_by_distance.ten_sixteen == 0.313
assert shooting_data.fgp_by_distance.sixteen_three_pt == 0.352
assert shooting_data.fgp_by_distance.three_point == 0.290
career_sd = lbj.career_stats.shooting_data
assert career_sd.dunks_made == 1898
assert career_sd.dunks_fga == 0.08
assert career_sd.heaves_made == 2
assert career_sd.heaves_attempted == 34
assert career_sd.two_point_fga_assisted == 0.347
assert career_sd.three_point_fga_assisted == 0.469
assert career_sd.corner_three_point_fga == 0.129
assert career_sd.corner_three_point_fgp == 0.387
assert career_sd.avg_distance == 12.0
first_playoffs = next(lbj.playoffs)
assert first_playoffs.stats.shooting_data.avg_distance == 12.1
assert first_playoffs.stats.shooting_data.dunks_made == 13
def test_stats_where_none_available(self):
chamberlain = PlayerPageScraper(
get_resource("chamberlain.html")).get_content()
rookie_season = next(chamberlain.seasons)
assert rookie_season.stats.rebounds is None
assert rookie_season.stats.offensive_rebounds is None
assert rookie_season.stats.defensive_rebounds is None
assert rookie_season.stats.turnovers is None
assert rookie_season.stats.three_fg_percentage is None
assert rookie_season.stats.three_fg_made is None
assert rookie_season.stats.three_fg_attempted is None
assert rookie_season.stats.shooting_data is None
advanced_stats = rookie_season.stats.advanced
assert advanced_stats.value_over_replacement_player is None
assert advanced_stats.ft_attempt_rate == 0.429
assert advanced_stats.assist_percentage is None
assert advanced_stats.offensive_rebound_percentage is None
assert advanced_stats.defensive_rebound_percentage is None
assert advanced_stats.usage_percentage is None
def scrape_all(self, _max: int = None) -> Iterator[Player]:
urls = list(set(self._urls[0:_max] if _max else self._urls))
with concurrent.futures.ThreadPoolExecutor() as executor:
for player in executor.map(lambda u: PlayerPageScraper(u).get_content(), urls):
yield player
def get_by_name(name: str) -> Optional[Player]:
result = Search.search_players(name)
if not result:
return None
url = result[0][-1]
return PlayerPageScraper(url).get_content()
def get_by_id(_id: str) -> Player:
url = f"https://www.basketball-reference.com/players/{_id[0]}/{_id}.html"
return PlayerPageScraper(url).get_content()
def get_by_url(url: str) -> Player:
return PlayerPageScraper(url).get_content()
def as_player(self) -> Player:
from bballer.scrapers.PlayerPageScraper import PlayerPageScraper
scr = PlayerPageScraper(self.url)
return scr.get_content()
def test_all_star(self):
erving = PlayerPageScraper(
get_resource("julius_erving.html")).get_content()
assert len([
season for season in list(erving.seasons) if season.all_star
]) == 11
def wilt_chamberlain():
return PlayerPageScraper(get_resource("chamberlain.html")).get_content()
def test_seasons(self):
erving = PlayerPageScraper(
get_resource("julius_erving.html")).get_content()
assert len(list(erving.seasons)) == 11 # ABA seasons must be discarded
class TestStats:
player = PlayerPageScraper(
get_resource("carmelo_anthony.html")).get_content()
seasons = list(player.seasons)
rookie_season = seasons[0]
def test_season_stats(self):
rookie_season = self.rookie_season
assert repr(rookie_season).endswith("(2004, DEN)")
assert rookie_season.stats.position == "SF"
assert rookie_season.season == 2004
assert rookie_season.stats.games_started == 82
assert rookie_season.stats.games_played == 82
assert rookie_season.stats.minutes_played == 2995
assert rookie_season.stats.fg_attempted == 1465
assert rookie_season.stats.fg_made == 624
assert rookie_season.stats.three_fg_made == 69
assert rookie_season.stats.three_fg_attempted == 214
assert rookie_season.stats.two_fg_attempted == 1251
assert rookie_season.stats.two_fg_made == 555
assert rookie_season.stats.effective_fg_percentage == 0.449
assert rookie_season.stats.ft_made == 408
assert rookie_season.stats.ft_attempted == 525
assert rookie_season.stats.offensive_rebounds == 183
assert rookie_season.stats.defensive_rebounds == 315
assert rookie_season.stats.rebounds == 498
assert round(rookie_season.stats.two_fg_percentage, 3) == 0.444
assert rookie_season.stats.fg_percentage == 0.426
assert rookie_season.stats.three_fg_percentage == 0.322
assert rookie_season.stats.free_throw_percentage == 0.777
assert not rookie_season.all_star
assert rookie_season.stats.assists == 227
assert rookie_season.stats.blocks == 41
assert rookie_season.stats.turnovers == 247
assert rookie_season.stats.steals == 97
assert rookie_season.stats.fouls == 225
assert rookie_season.stats.points == 1725
def test_advanced_stats(self):
rookie_season = self.rookie_season
assert repr(rookie_season.stats.advanced) == "AdvancedStatLine(2004)"
assert rookie_season.stats.advanced.assist_percentage == 13.8
assert rookie_season.stats.advanced.player_efficiency_rating == 17.6
assert rookie_season.stats.advanced.true_shooting_percentage == 0.509
assert rookie_season.stats.advanced.three_fg_attempt_rate == 0.146
assert rookie_season.stats.advanced.ft_attempt_rate == 0.358
assert rookie_season.stats.advanced.offensive_rebound_percentage == 6.8
assert rookie_season.stats.advanced.defensive_rebound_percentage == 12.1
assert rookie_season.stats.advanced.total_rebound_percentage == 9.4
assert rookie_season.stats.advanced.steal_percentage == 1.7
assert rookie_season.stats.advanced.block_percentage == 1.0
assert rookie_season.stats.advanced.turnover_percentage == 12.7
assert rookie_season.stats.advanced.usage_percentage == 28.5
assert rookie_season.stats.advanced.offensive_win_shares == 3.7
assert rookie_season.stats.advanced.defensive_win_shares == 2.4
assert rookie_season.stats.advanced.win_shares == 6.1
assert rookie_season.stats.advanced.win_shares_per_48 == 0.098
assert rookie_season.stats.advanced.offensive_box_plus_minus == 1.2
assert rookie_season.stats.advanced.defensive_box_plus_minus == -1.2
assert rookie_season.stats.advanced.box_plus_minus == 0
assert rookie_season.stats.advanced.value_over_replacement_player == 1.6
def test_stats_where_none_available(self):
chamberlain = PlayerPageScraper(
get_resource("chamberlain.html")).get_content()
rookie_season = next(chamberlain.seasons)
assert rookie_season.stats.rebounds is None
assert rookie_season.stats.offensive_rebounds is None
assert rookie_season.stats.defensive_rebounds is None
assert rookie_season.stats.turnovers is None
assert rookie_season.stats.three_fg_percentage is None
assert rookie_season.stats.three_fg_made is None
assert rookie_season.stats.three_fg_attempted is None
assert rookie_season.stats.shooting_data is None
advanced_stats = rookie_season.stats.advanced
assert advanced_stats.value_over_replacement_player is None
assert advanced_stats.ft_attempt_rate == 0.429
assert advanced_stats.assist_percentage is None
assert advanced_stats.offensive_rebound_percentage is None
assert advanced_stats.defensive_rebound_percentage is None
assert advanced_stats.usage_percentage is None
def test_stats_where_partially_available(self):
mullin = PlayerPageScraper(
get_resource("chris_mullin.html")).get_content()
seasons = list(mullin.seasons)
rookie_season = seasons[0]
assert rookie_season.stats.shooting_data is None
final_season = seasons[-1]
assert final_season.stats.shooting_data
assert final_season.stats.shooting_data.avg_distance == 18.8
def test_career_stats_advanced(self):
mullin = PlayerPageScraper(
get_resource("chris_mullin.html")).get_content()
assert mullin.career_stats.advanced.usage_percentage
def test_career_stats_encapsulated(self):
mullin = PlayerPageScraper(
get_resource("chris_mullin.html")).get_content()
career_stats = mullin.career_stats
first_season = next(mullin.seasons)
first_playoffs = next(mullin.playoffs)
for attr in ["age", "all_star", "team", "season"]:
assert not hasattr(career_stats, attr)
assert hasattr(first_season, attr)
assert hasattr(first_playoffs, attr)
def julius_erving():
return PlayerPageScraper(get_resource("julius_erving.html")).get_content()
def test_career_stats_advanced(self):
mullin = PlayerPageScraper(
get_resource("chris_mullin.html")).get_content()
assert mullin.career_stats.advanced.usage_percentage
def lebron_james():
return PlayerPageScraper(get_resource("lebron_james.html")).get_content()
def carmelo_anthony():
return PlayerPageScraper(
get_resource("carmelo_anthony.html")).get_content()
def test_historical_player(self):
player = PlayerPageScraper(
get_resource("tom_hawkins.html")).get_content()
assert player.name == "Tom Hawkins"
assert player.date_of_birth == date(1936, 12, 22)
def test_get_playoff_totals(self):
erving = PlayerPageScraper(
get_resource("julius_erving.html")).get_content()
playoffs = list(erving.playoffs)
assert len(playoffs) == 11
assert sum([season.stats.points for season in playoffs]) == 3088