def schedule_for_month(url):
response = requests.get(url=url)
response.raise_for_status()
page = SchedulePage(html=html.fromstring(html=response.content))
parser = ScheduledGamesParser(
start_time_parser=ScheduledStartTimeParser(),
team_name_parser=TeamNameParser(team_names_to_teams=TEAM_NAME_TO_TEAM),
)
return parser.parse_games(games=page.rows)
def setUp(self):
self.october_2001_html = requests.get(
'https://www.basketball-reference.com/leagues/NBA_2001_games.html'
).text
self.october_2018_html = requests.get(
'https://www.basketball-reference.com/leagues/NBA_2018_games-october.html'
).text
self.schedule_with_future_games_html_file = open(
os.path.join(os.path.dirname(__file__),
'./NBA_2019_games-april.html'))
self.schedule_with_future_games_html = self.schedule_with_future_games_html_file.read(
)
self.parser = ScheduledGamesParser(
start_time_parser=ScheduledStartTimeParser(),
team_name_parser=TeamNameParser(
team_names_to_teams=TEAM_NAME_TO_TEAM),
)
def season_schedule(season_end_year):
url = '{BASE_URL}/leagues/NBA_{season_end_year}_games.html'.format(
BASE_URL=BASE_URL, season_end_year=season_end_year)
response = requests.get(url=url)
response.raise_for_status()
page = SchedulePage(html=html.fromstring(html=response.content))
parser = ScheduledGamesParser(
start_time_parser=ScheduledStartTimeParser(),
team_name_parser=TeamNameParser(team_names_to_teams=TEAM_NAME_TO_TEAM),
)
season_schedule_values = parser.parse_games(games=page.rows)
for month_url_path in page.other_months_schedule_urls:
url = '{BASE_URL}{month_url_path}'.format(
BASE_URL=BASE_URL, month_url_path=month_url_path)
monthly_schedule = schedule_for_month(url=url)
season_schedule_values.extend(monthly_schedule)
return season_schedule_values
class TestScheduleParser(TestCase):
def setUp(self):
self.october_2001_html = requests.get(
'https://www.basketball-reference.com/leagues/NBA_2001_games.html'
).text
self.october_2018_html = requests.get(
'https://www.basketball-reference.com/leagues/NBA_2018_games-october.html'
).text
self.schedule_with_future_games_html_file = open(
os.path.join(os.path.dirname(__file__),
'./NBA_2019_games-april.html'))
self.schedule_with_future_games_html = self.schedule_with_future_games_html_file.read(
)
self.parser = ScheduledGamesParser(
start_time_parser=ScheduledStartTimeParser(),
team_name_parser=TeamNameParser(
team_names_to_teams=TEAM_NAME_TO_TEAM),
)
def tearDown(self):
self.schedule_with_future_games_html_file.close()
def test_parse_october_2001_schedule_for_month_url_paths_(self):
page = SchedulePage(html=html.fromstring(self.october_2001_html))
expected_urls = [
"/leagues/NBA_2001_games-november.html",
"/leagues/NBA_2001_games-december.html",
"/leagues/NBA_2001_games-january.html",
"/leagues/NBA_2001_games-february.html",
"/leagues/NBA_2001_games-march.html",
"/leagues/NBA_2001_games-april.html",
"/leagues/NBA_2001_games-may.html",
"/leagues/NBA_2001_games-june.html",
]
self.assertEqual(page.other_months_schedule_urls, expected_urls)
def test_parse_october_2001_schedule(self):
page = SchedulePage(html=html.fromstring(self.october_2001_html))
parsed_schedule = self.parser.parse_games(games=page.rows)
first_game = parsed_schedule[0]
expected_datetime = pytz.timezone("US/Eastern") \
.localize(datetime(year=2000, month=10, day=31, hour=19, minute=30)) \
.astimezone(pytz.utc)
self.assertEqual(len(parsed_schedule), 13)
self.assertTrue(
abs(first_game["start_time"] -
expected_datetime) < timedelta(seconds=1))
self.assertEqual(first_game["away_team"], Team.CHARLOTTE_HORNETS)
self.assertEqual(first_game["home_team"], Team.ATLANTA_HAWKS)
self.assertEqual(first_game["away_team_score"], 106)
self.assertEqual(first_game["home_team_score"], 82)
def test_parse_october_2018_schedule(self):
page = SchedulePage(html=html.fromstring(self.october_2018_html))
parsed_schedule = self.parser.parse_games(games=page.rows)
self.assertEqual(len(parsed_schedule), 104)
def test_parse_future_game(self):
page = SchedulePage(
html=html.fromstring(self.schedule_with_future_games_html))
parsed_schedule = self.parser.parse_games(games=page.rows)
first_game = parsed_schedule[0]
expected_first_game_start_time = pytz.timezone("US/Eastern") \
.localize(datetime(year=2019, month=4, day=1, hour=19, minute=30)) \
.astimezone(pytz.utc)
self.assertIsNotNone(parsed_schedule)
self.assertEqual(len(parsed_schedule), 79)
self.assertEqual(first_game["start_time"],
expected_first_game_start_time)
self.assertEqual(first_game["away_team"], Team.MIAMI_HEAT)
self.assertEqual(first_game["home_team"], Team.BOSTON_CELTICS)
self.assertIsNone(first_game["away_team_score"])
self.assertIsNone(first_game["home_team_score"])
def __init__(self):
self.team_abbreviation_parser = TeamAbbreviationParser(
abbreviations_to_teams=TEAM_ABBREVIATIONS_TO_TEAM)
self.league_abbreviation_parser = LeagueAbbreviationParser(
abbreviations_to_league=LEAGUE_ABBREVIATIONS_TO_LEAGUE)
self.location_abbreviation_parser = LocationAbbreviationParser(
abbreviations_to_locations=LOCATION_ABBREVIATIONS_TO_POSITION, )
self.outcome_abbreviation_parser = OutcomeAbbreviationParser(
abbreviations_to_outcomes=OUTCOME_ABBREVIATIONS_TO_OUTCOME, )
self.outcome_parser = PlayerBoxScoreOutcomeParser(
outcome_abbreviation_parser=self.outcome_abbreviation_parser)
self.period_details_parser = PeriodDetailsParser(
regulation_periods_count=4)
self.period_timestamp_parser = PeriodTimestampParser(
timestamp_format=ParserService.PLAY_BY_PLAY_TIMESTAMP_FORMAT)
self.position_abbreviation_parser = PositionAbbreviationParser(
abbreviations_to_positions=POSITION_ABBREVIATIONS_TO_POSITION, )
self.seconds_played_parser = SecondsPlayedParser()
self.scores_parser = ScoresParser(
scores_regex=ParserService.PLAY_BY_PLAY_SCORES_REGEX)
self.search_result_name_parser = SearchResultNameParser()
self.search_result_location_parser = ResourceLocationParser(
resource_location_regex=ParserService.
SEARCH_RESULT_RESOURCE_LOCATION_REGEX)
self.team_name_parser = TeamNameParser(
team_names_to_teams=TEAM_NAME_TO_TEAM)
self.play_by_plays_parser = PlayByPlaysParser(
period_details_parser=self.period_details_parser,
period_timestamp_parser=self.period_timestamp_parser,
scores_parser=self.scores_parser,
)
self.player_box_scores_parser = PlayerBoxScoresParser(
team_abbreviation_parser=self.team_abbreviation_parser,
location_abbreviation_parser=self.location_abbreviation_parser,
outcome_abbreviation_parser=self.outcome_abbreviation_parser,
seconds_played_parser=self.seconds_played_parser)
self.player_data_parser = PlayerDataParser(
search_result_location_parser=self.search_result_location_parser,
league_abbreviation_parser=self.league_abbreviation_parser,
)
self.player_season_box_scores_parser = PlayerSeasonBoxScoresParser(
team_abbreviation_parser=self.team_abbreviation_parser,
location_abbreviation_parser=self.location_abbreviation_parser,
outcome_parser=self.outcome_parser,
seconds_played_parser=self.seconds_played_parser)
self.player_season_totals_parser = PlayerSeasonTotalsParser(
position_abbreviation_parser=self.position_abbreviation_parser,
team_abbreviation_parser=self.team_abbreviation_parser,
)
self.player_advanced_season_totals_parser = PlayerAdvancedSeasonTotalsParser(
team_abbreviation_parser=self.team_abbreviation_parser,
position_abbreviation_parser=self.position_abbreviation_parser,
)
self.scheduled_start_time_parser = ScheduledStartTimeParser()
self.scheduled_games_parser = ScheduledGamesParser(
start_time_parser=self.scheduled_start_time_parser,
team_name_parser=self.team_name_parser,
)
self.search_results_parser = SearchResultsParser(
search_result_name_parser=self.search_result_name_parser,
search_result_location_parser=self.search_result_location_parser,
league_abbreviation_parser=self.league_abbreviation_parser,
)
self.team_totals_parser = TeamTotalsParser(
team_abbreviation_parser=self.team_abbreviation_parser)
self.division_name_parser = DivisionNameParser(divisions=Division)
self.team_standings_parser = TeamStandingsParser(teams=Team)
self.conference_division_standings_parser = ConferenceDivisionStandingsParser(
division_name_parser=self.division_name_parser,
team_standings_parser=self.team_standings_parser,
divisions_to_conferences=DIVISIONS_TO_CONFERENCES,
)
class ParserService:
PLAY_BY_PLAY_TIMESTAMP_FORMAT = "%M:%S.%f"
PLAY_BY_PLAY_SCORES_REGEX = "(?P<away_team_score>[0-9]+)-(?P<home_team_score>[0-9]+)"
SEARCH_RESULT_RESOURCE_LOCATION_REGEX = '(https?:\/\/www\.basketball-reference\.com\/)?(?P<resource_type>.+?(?=\/)).*\/(?P<resource_identifier>.+).html'
def __init__(self):
self.team_abbreviation_parser = TeamAbbreviationParser(
abbreviations_to_teams=TEAM_ABBREVIATIONS_TO_TEAM)
self.league_abbreviation_parser = LeagueAbbreviationParser(
abbreviations_to_league=LEAGUE_ABBREVIATIONS_TO_LEAGUE)
self.location_abbreviation_parser = LocationAbbreviationParser(
abbreviations_to_locations=LOCATION_ABBREVIATIONS_TO_POSITION, )
self.outcome_abbreviation_parser = OutcomeAbbreviationParser(
abbreviations_to_outcomes=OUTCOME_ABBREVIATIONS_TO_OUTCOME, )
self.outcome_parser = PlayerBoxScoreOutcomeParser(
outcome_abbreviation_parser=self.outcome_abbreviation_parser)
self.period_details_parser = PeriodDetailsParser(
regulation_periods_count=4)
self.period_timestamp_parser = PeriodTimestampParser(
timestamp_format=ParserService.PLAY_BY_PLAY_TIMESTAMP_FORMAT)
self.position_abbreviation_parser = PositionAbbreviationParser(
abbreviations_to_positions=POSITION_ABBREVIATIONS_TO_POSITION, )
self.seconds_played_parser = SecondsPlayedParser()
self.scores_parser = ScoresParser(
scores_regex=ParserService.PLAY_BY_PLAY_SCORES_REGEX)
self.search_result_name_parser = SearchResultNameParser()
self.search_result_location_parser = ResourceLocationParser(
resource_location_regex=ParserService.
SEARCH_RESULT_RESOURCE_LOCATION_REGEX)
self.team_name_parser = TeamNameParser(
team_names_to_teams=TEAM_NAME_TO_TEAM)
self.play_by_plays_parser = PlayByPlaysParser(
period_details_parser=self.period_details_parser,
period_timestamp_parser=self.period_timestamp_parser,
scores_parser=self.scores_parser,
)
self.player_box_scores_parser = PlayerBoxScoresParser(
team_abbreviation_parser=self.team_abbreviation_parser,
location_abbreviation_parser=self.location_abbreviation_parser,
outcome_abbreviation_parser=self.outcome_abbreviation_parser,
seconds_played_parser=self.seconds_played_parser)
self.player_data_parser = PlayerDataParser(
search_result_location_parser=self.search_result_location_parser,
league_abbreviation_parser=self.league_abbreviation_parser,
)
self.player_season_box_scores_parser = PlayerSeasonBoxScoresParser(
team_abbreviation_parser=self.team_abbreviation_parser,
location_abbreviation_parser=self.location_abbreviation_parser,
outcome_parser=self.outcome_parser,
seconds_played_parser=self.seconds_played_parser)
self.player_season_totals_parser = PlayerSeasonTotalsParser(
position_abbreviation_parser=self.position_abbreviation_parser,
team_abbreviation_parser=self.team_abbreviation_parser,
)
self.player_advanced_season_totals_parser = PlayerAdvancedSeasonTotalsParser(
team_abbreviation_parser=self.team_abbreviation_parser,
position_abbreviation_parser=self.position_abbreviation_parser,
)
self.scheduled_start_time_parser = ScheduledStartTimeParser()
self.scheduled_games_parser = ScheduledGamesParser(
start_time_parser=self.scheduled_start_time_parser,
team_name_parser=self.team_name_parser,
)
self.search_results_parser = SearchResultsParser(
search_result_name_parser=self.search_result_name_parser,
search_result_location_parser=self.search_result_location_parser,
league_abbreviation_parser=self.league_abbreviation_parser,
)
self.team_totals_parser = TeamTotalsParser(
team_abbreviation_parser=self.team_abbreviation_parser)
self.division_name_parser = DivisionNameParser(divisions=Division)
self.team_standings_parser = TeamStandingsParser(teams=Team)
self.conference_division_standings_parser = ConferenceDivisionStandingsParser(
division_name_parser=self.division_name_parser,
team_standings_parser=self.team_standings_parser,
divisions_to_conferences=DIVISIONS_TO_CONFERENCES,
)
def parse_division_standings(self, standings):
return self.conference_division_standings_parser.parse(
division_standings=standings)
def parse_play_by_plays(self, play_by_plays, away_team_name,
home_team_name):
return self.play_by_plays_parser.parse(
play_by_plays=play_by_plays,
away_team=self.team_name_parser.parse_team_name(
team_name=away_team_name),
home_team=self.team_name_parser.parse_team_name(
team_name=home_team_name),
)
def parse_player_box_scores(self, box_scores):
return self.player_box_scores_parser.parse(box_scores=box_scores)
def parse_player_season_box_scores(self, box_scores):
return self.player_season_box_scores_parser.parse(
box_scores=box_scores)
def parse_player_advanced_season_totals_parser(self, totals):
return self.player_advanced_season_totals_parser.parse(totals=totals)
def parse_player_season_totals(self, totals):
return self.player_season_totals_parser.parse(totals=totals)
def parse_scheduled_games(self, games):
return self.scheduled_games_parser.parse_games(games)
def parse_team_totals(self, first_team_totals, second_team_totals):
return self.team_totals_parser.parse(
first_team_totals=first_team_totals,
second_team_totals=second_team_totals)
def parse_player_search_results(self, nba_aba_baa_players):
return self.search_results_parser.parse(
nba_aba_baa_players=nba_aba_baa_players)
def parse_player_data(self, player):
return self.player_data_parser.parse(player=player)