def setUp(self):
self.parser = PlayerDataParser(
search_result_location_parser=ResourceLocationParser(
resource_location_regex=ParserService.
SEARCH_RESULT_RESOURCE_LOCATION_REGEX, ),
league_abbreviation_parser=LeagueAbbreviationParser(
abbreviations_to_league=LEAGUE_ABBREVIATIONS_TO_LEAGUE),
)
def search(term):
response = requests.get(
url="{BASE_URL}/search/search.fcgi".format(BASE_URL=BASE_URL),
params={"search": term}
)
response.raise_for_status()
player_results = []
if response.url.startswith("{BASE_URL}/search/search.fcgi".format(BASE_URL=BASE_URL)):
page = SearchPage(html=html.fromstring(response.content))
parser = SearchResultsParser(
search_result_name_parser=SearchResultNameParser(),
search_result_location_parser=ResourceLocationParser(
resource_location_regex=SEARCH_RESULT_RESOURCE_LOCATION_REGEX,
),
league_abbreviation_parser=LeagueAbbreviationParser(
abbreviations_to_league=LEAGUE_ABBREVIATIONS_TO_LEAGUE,
),
)
parsed_results = parser.parse(nba_aba_baa_players=page.nba_aba_baa_players)
player_results += parsed_results["players"]
while page.nba_aba_baa_players_pagination_url is not None:
response = requests.get(
url="{BASE_URL}/search/{pagination_url}".format(
BASE_URL=BASE_URL,
pagination_url=page.nba_aba_baa_players_pagination_url
)
)
response.raise_for_status()
page = SearchPage(html=html.fromstring(response.content))
parsed_results = parser.parse(nba_aba_baa_players=page.nba_aba_baa_players)
player_results += parsed_results["players"]
elif response.url.startswith("{BASE_URL}/players".format(BASE_URL=BASE_URL)):
page = PlayerPage(html=html.fromstring(response.content))
data = PlayerData(
name=page.name,
resource_location=response.url,
league_abbreviations=set([row.league_abbreviation for row in page.totals_table.rows])
)
parser = PlayerDataParser(
search_result_location_parser=ResourceLocationParser(
resource_location_regex=SEARCH_RESULT_RESOURCE_LOCATION_REGEX,
),
league_abbreviation_parser=LeagueAbbreviationParser(abbreviations_to_league=LEAGUE_ABBREVIATIONS_TO_LEAGUE),
)
player_results += [parser.parse(player=data)]
return {
"players": player_results
}
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)
class TestPlayerDataParser(TestCase):
def setUp(self):
self.parser = PlayerDataParser(
search_result_location_parser=ResourceLocationParser(
resource_location_regex=ParserService.
SEARCH_RESULT_RESOURCE_LOCATION_REGEX, ),
league_abbreviation_parser=LeagueAbbreviationParser(
abbreviations_to_league=LEAGUE_ABBREVIATIONS_TO_LEAGUE),
)
def test_parse_name(self):
parsed_player = self.parser.parse(
player=PlayerData(name="jaebaebae",
resource_location='/players/b/bryanko01.html',
league_abbreviations=[]))
self.assertEqual(parsed_player["name"], "jaebaebae")
def test_parse_resource_location(self):
parsed_player = self.parser.parse(
player=PlayerData(name="jaebaebae",
resource_location='/players/b/bryanko01.html',
league_abbreviations=[]))
self.assertEqual(parsed_player["identifier"], "bryanko01")
def test_parse_league_abbreviations_for_single_nba_abbreviation(self):
parsed_player = self.parser.parse(
player=PlayerData(name="jaebaebae",
resource_location='/players/b/bryanko01.html',
league_abbreviations=["NBA"]))
self.assertEqual(parsed_player["leagues"],
{League.NATIONAL_BASKETBALL_ASSOCIATION})
def test_parse_league_abbreviations_for_single_aba_abbreviation(self):
parsed_player = self.parser.parse(
player=PlayerData(name="jaebaebae",
resource_location='/players/b/bryanko01.html',
league_abbreviations=["ABA"]))
self.assertEqual(parsed_player["leagues"],
{League.AMERICAN_BASKETBALL_ASSOCIATION})
def test_parse_league_abbreviations_for_single_baa_abbreviation(self):
parsed_player = self.parser.parse(
player=PlayerData(name="jaebaebae",
resource_location='/players/b/bryanko01.html',
league_abbreviations=["BAA"]))
self.assertEqual(parsed_player["leagues"],
{League.BASKETBALL_ASSOCIATION_OF_AMERICA})
def test_parse_league_abbreviations_for_multiple_nba_abbreviations(self):
parsed_player = self.parser.parse(
player=PlayerData(name="jaebaebae",
resource_location='/players/b/bryanko01.html',
league_abbreviations=["NBA", "NBA"]))
self.assertEqual(parsed_player["leagues"],
{League.NATIONAL_BASKETBALL_ASSOCIATION})
def test_parse_league_abbreviations_for_multiple_aba_abbreviations(self):
parsed_player = self.parser.parse(
player=PlayerData(name="jaebaebae",
resource_location='/players/b/bryanko01.html',
league_abbreviations=["ABA", "ABA"]))
self.assertEqual(parsed_player["leagues"],
{League.AMERICAN_BASKETBALL_ASSOCIATION})
def test_parse_league_abbreviations_for_multiple_baa_abbreviations(self):
parsed_player = self.parser.parse(
player=PlayerData(name="jaebaebae",
resource_location='/players/b/bryanko01.html',
league_abbreviations=["BAA", "BAA"]))
self.assertEqual(parsed_player["leagues"],
{League.BASKETBALL_ASSOCIATION_OF_AMERICA})
def test_parse_league_abbreviations_for_multiple_nba_and_aba_and_baa_abbreviations(
self):
parsed_player = self.parser.parse(player=PlayerData(
name="jaebaebae",
resource_location='/players/b/bryanko01.html',
league_abbreviations=["NBA", "ABA", "BAA", "NBA", "ABA", "BAA"]))
self.assertEqual(
parsed_player["leagues"], {
League.NATIONAL_BASKETBALL_ASSOCIATION,
League.AMERICAN_BASKETBALL_ASSOCIATION,
League.BASKETBALL_ASSOCIATION_OF_AMERICA
})