sportsball 0.10.12__tar.gz → 0.10.13__tar.gz

{sportsball-0.10.12/sportsball.egg-info → sportsball-0.10.13}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: sportsball
-Version: 0.10.12
+Version: 0.10.13
 Summary: A library for pulling in and normalising sports stats.
 Home-page: https://github.com/8W9aG/sportsball
 Author: Will Sackfield
@@ -47,7 +47,7 @@ Requires-Dist: playwright>=1.51.0
 Requires-Dist: cchardet>=2.2.0a2
 Requires-Dist: lxml>=5.3.0
 Requires-Dist: gender-guesser>=0.4.0
-Requires-Dist: scrapesession>=0.0.10
+Requires-Dist: scrapesession>=0.0.11
 Requires-Dist: pyhigh>=0.0.6
 Dynamic: author
 Dynamic: author-email

{sportsball-0.10.12 → sportsball-0.10.13}/requirements.txt

@@ -34,5 +34,5 @@ playwright>=1.51.0
 cchardet>=2.2.0a2
 lxml>=5.3.0
 gender-guesser>=0.4.0
-scrapesession>=0.0.10
+scrapesession>=0.0.11
 pyhigh>=0.0.6

{sportsball-0.10.12 → sportsball-0.10.13}/setup.py

@@ -26,7 +26,7 @@ def install_requires() -> typing.List[str]:
 
 setup(
     name='sportsball',
-    version='0.10.12',
+    version='0.10.13',
     description='A library for pulling in and normalising sports stats.',
     long_description=long_description,
     long_description_content_type='text/markdown',

{sportsball-0.10.12 → sportsball-0.10.13}/sportsball/__init__.py

@@ -4,4 +4,4 @@
 import cchardet  # type: ignore
 import lxml  # noqa: ignore
 
-__VERSION__ = "0.10.12"
+__VERSION__ = "0.10.13"

{sportsball-0.10.12 → sportsball-0.10.13}/sportsball/data/afl/afl/afl_afl_league_model.py

@@ -15,7 +15,7 @@ from scrapesession.scrapesession import ScrapeSession  # type: ignore
 from ....playwright import ensure_install
 from ...game_model import VERSION, GameModel
 from ...league import League
-from ...league_model import LeagueModel
+from ...league_model import SHUTDOWN_FLAG, LeagueModel
 from ..position import Position, position_from_str
 from .afl_afl_game_model import create_afl_afl_game_model, parse_players_v1
 
@@ -28,6 +28,8 @@ def _parse_v1(
     playwright: Playwright,
 ) -> Iterator[GameModel]:
     for div in soup.find_all("div", {"class": re.compile(".*js-match-list-item.*")}):
+        if SHUTDOWN_FLAG.is_set():
+            return
         team_names = []
         for span in div.find_all(
             "span", {"class": re.compile(".*team-lineups__team-name.*")}
@@ -77,6 +79,8 @@ def _parse_v2_soup(
     tuple[list[str], list[list[tuple[str, str, str, str, Position]]], str, str]
 ]:
     for div in soup.find_all("div", {"class": "team-lineups__item"}):
+        if SHUTDOWN_FLAG.is_set():
+            return
         team_names = []
         for span in div.find_all(
             "span", {"class": re.compile(".*team-lineups-header__name.*")}
@@ -218,12 +222,18 @@ class AFLAFLLeagueModel(LeagueModel):
 
     @property
     def games(self) -> Iterator[GameModel]:
-        ladder = self._ladder
-        ensure_install()
-        with sync_playwright() as p:
-            browser = p.chromium.launch()
-            context = browser.new_context()
-            page = context.new_page()
-            url = "https://www.afl.com.au/matches/team-lineups"
-            page.goto(url, wait_until="networkidle")
-            yield from _parse_game_info(page.content(), self.session, ladder, url, p)
+        try:
+            ladder = self._ladder
+            ensure_install()
+            with sync_playwright() as p:
+                browser = p.chromium.launch()
+                context = browser.new_context()
+                page = context.new_page()
+                url = "https://www.afl.com.au/matches/team-lineups"
+                page.goto(url, wait_until="networkidle")
+                yield from _parse_game_info(
+                    page.content(), self.session, ladder, url, p
+                )
+        except Exception as exc:
+            SHUTDOWN_FLAG.set()
+            raise exc

{sportsball-0.10.12 → sportsball-0.10.13}/sportsball/data/afl/afltables/afl_afltables_league_model.py

@@ -15,7 +15,7 @@ from scrapesession.scrapesession import ScrapeSession  # type: ignore
 
 from ...game_model import GameModel
 from ...league import League
-from ...league_model import LeagueModel
+from ...league_model import SHUTDOWN_FLAG, LeagueModel
 from ...season_type import SeasonType
 from .afl_afltables_game_model import create_afl_afltables_game_model
 
@@ -82,6 +82,8 @@ class AFLAFLTablesLeagueModel(LeagueModel):
                 if "Venue:" in td_text:
                     current_dt = _find_dt(td_text, season_url)
             for a in table.find_all("a", href=True):
+                if SHUTDOWN_FLAG.is_set():
+                    return
                 if a.get_text().strip().lower() == "match stats":
                     url = urllib.parse.urljoin(season_url, a.get("href"))
                     if url in urls_duplicates:
@@ -149,20 +151,24 @@ class AFLAFLTablesLeagueModel(LeagueModel):
 
     @property
     def games(self) -> Iterator[GameModel]:
-        with self.session.cache_disabled():
-            with self.session.wayback_disabled():
-                response = self.session.get(_SEASON_URL)
-        response.raise_for_status()
-        soup = BeautifulSoup(response.text, "lxml")
-        with tqdm.tqdm(position=self.position) as pbar:
-            for table in soup.find_all("table"):
-                for tr in table.find_all("tr"):
-                    for td in tr.find_all("td"):
-                        for a in td.find_all("a"):
-                            url = urllib.parse.urljoin(_SEASON_URL, a.get("href"))
-                            yield from self._produce_games(
-                                url, SeasonType.REGULAR, pbar
-                            )
-                            yield from self._produce_games(
-                                url, SeasonType.POSTSEASON, pbar
-                            )
+        try:
+            with self.session.cache_disabled():
+                with self.session.wayback_disabled():
+                    response = self.session.get(_SEASON_URL)
+            response.raise_for_status()
+            soup = BeautifulSoup(response.text, "lxml")
+            with tqdm.tqdm(position=self.position) as pbar:
+                for table in soup.find_all("table"):
+                    for tr in table.find_all("tr"):
+                        for td in tr.find_all("td"):
+                            for a in td.find_all("a"):
+                                url = urllib.parse.urljoin(_SEASON_URL, a.get("href"))
+                                yield from self._produce_games(
+                                    url, SeasonType.REGULAR, pbar
+                                )
+                                yield from self._produce_games(
+                                    url, SeasonType.POSTSEASON, pbar
+                                )
+        except Exception as exc:
+            SHUTDOWN_FLAG.set()
+            raise exc

{sportsball-0.10.12 → sportsball-0.10.13}/sportsball/data/aussportsbetting/aussportsbetting_league_model.py

@@ -11,7 +11,7 @@ from scrapesession.scrapesession import ScrapeSession  # type: ignore
 
 from ..game_model import GameModel
 from ..league import League
-from ..league_model import LeagueModel
+from ..league_model import SHUTDOWN_FLAG, LeagueModel
 from .aussportsbetting_game_model import create_aussportsbetting_game_model
 
 
@@ -98,17 +98,23 @@ class AusSportsBettingLeagueModel(LeagueModel):
 
     @property
     def games(self) -> Iterator[GameModel]:
-        with self.session.cache_disabled():
-            response = self.session.get(self._spreadsheet_url)
-        response.raise_for_status()
-        workbook = load_workbook(filename=BytesIO(response.content))
-        ws = workbook.active
-        if ws is None:
-            raise ValueError("ws is null.")
-        with tqdm.tqdm(position=self.position) as pbar:
-            for row in ws.iter_rows():
-                game_model = self._row_to_game(row)
-                if game_model is not None:
-                    pbar.update(1)
-                    pbar.set_description(f"AusSportsBetting - {game_model.dt}")
-                    yield game_model
+        try:
+            with self.session.cache_disabled():
+                response = self.session.get(self._spreadsheet_url)
+            response.raise_for_status()
+            workbook = load_workbook(filename=BytesIO(response.content))
+            ws = workbook.active
+            if ws is None:
+                raise ValueError("ws is null.")
+            with tqdm.tqdm(position=self.position) as pbar:
+                for row in ws.iter_rows():
+                    if SHUTDOWN_FLAG.is_set():
+                        return
+                    game_model = self._row_to_game(row)
+                    if game_model is not None:
+                        pbar.update(1)
+                        pbar.set_description(f"AusSportsBetting - {game_model.dt}")
+                        yield game_model
+        except Exception as exc:
+            SHUTDOWN_FLAG.set()
+            raise exc

{sportsball-0.10.12 → sportsball-0.10.13}/sportsball/data/espn/espn_league_model.py

@@ -8,7 +8,7 @@ from scrapesession.scrapesession import ScrapeSession  # type: ignore
 
 from ..game_model import GameModel
 from ..league import League
-from ..league_model import LeagueModel
+from ..league_model import SHUTDOWN_FLAG, LeagueModel
 from ..season_type import SeasonType
 from .espn_game_model import create_espn_game_model
 
@@ -136,6 +136,8 @@ class ESPNLeagueModel(LeagueModel):
             weeks_response.raise_for_status()
             weeks = weeks_response.json()
             for item in weeks["items"]:
+                if SHUTDOWN_FLAG.is_set():
+                    return
                 if cache_disabled:
                     with self.session.cache_disabled():
                         week_response = self.session.get(item["$ref"])
@@ -153,35 +155,43 @@ class ESPNLeagueModel(LeagueModel):
 
     @property
     def games(self) -> Iterator[GameModel]:
-        with self.session.wayback_disabled():
-            page = 1
-            first = True
-            with tqdm.tqdm(position=self.position) as pbar:
-                while True:
-                    if page == 1:
-                        with self.session.cache_disabled():
+        try:
+            with self.session.wayback_disabled():
+                page = 1
+                first = True
+                with tqdm.tqdm(position=self.position) as pbar:
+                    while True:
+                        if page == 1:
+                            with self.session.cache_disabled():
+                                response = self.session.get(
+                                    self._start_url + f"&page={page}"
+                                )
+                        else:
                             response = self.session.get(
                                 self._start_url + f"&page={page}"
                             )
-                    else:
-                        response = self.session.get(self._start_url + f"&page={page}")
-                    response.raise_for_status()
-                    seasons = response.json()
-                    for item in seasons.get("items", []):
-                        season_response = self.session.get(item["$ref"])
-                        season_response.raise_for_status()
-                        season_json = season_response.json()
-
-                        for season_item in season_json["types"]["items"]:
-                            season_type_response = self.session.get(season_item["$ref"])
-                            season_type_response.raise_for_status()
-                            season_type_json = season_type_response.json()
-
-                            yield from self._produce_week_games(
-                                season_type_json, page, pbar, first
-                            )
-                        first = False
-
-                    if page >= seasons.get("pageCount", 0):
-                        break
-                    page += 1
+                        response.raise_for_status()
+                        seasons = response.json()
+                        for item in seasons.get("items", []):
+                            season_response = self.session.get(item["$ref"])
+                            season_response.raise_for_status()
+                            season_json = season_response.json()
+
+                            for season_item in season_json["types"]["items"]:
+                                season_type_response = self.session.get(
+                                    season_item["$ref"]
+                                )
+                                season_type_response.raise_for_status()
+                                season_type_json = season_type_response.json()
+
+                                yield from self._produce_week_games(
+                                    season_type_json, page, pbar, first
+                                )
+                            first = False
+
+                        if page >= seasons.get("pageCount", 0):
+                            break
+                        page += 1
+        except Exception as exc:
+            SHUTDOWN_FLAG.set()
+            raise exc

{sportsball-0.10.12 → sportsball-0.10.13}/sportsball/data/google/google_address_model.py

@@ -18544,6 +18544,16 @@ _CACHED_GEOCODES: dict[str, Any] = {
         housenumber="",
         country="USA",
     ),
+    "Canadian Tire Centre, Ottawa, ON, Canada": SportsballGeocodeTuple(
+        city="Ottawa",
+        state="ON",
+        postal="",
+        lat=45.296944,
+        lng=-75.927222,
+        housenumber="",
+        country="Canada",
+    ),
+    "American Airlines Center": AMERICAN_AIRLINES_ARENA,
 }
 
 

sportsball-0.10.13/sportsball/data/hkjc/hkjc/hkjc_hkjc_league_model.py

@@ -0,0 +1,111 @@
+"""The HKJC league model."""
+
+# pylint: disable=line-too-long,too-many-statements
+import datetime
+import http
+import urllib.parse
+from typing import Iterator
+from urllib.parse import urlparse
+
+import tqdm
+from bs4 import BeautifulSoup
+from scrapesession.scrapesession import ScrapeSession  # type: ignore
+
+from ...game_model import GameModel
+from ...league import League
+from ...league_model import SHUTDOWN_FLAG, LeagueModel
+from .hkjc_hkjc_game_model import (RACE_COURSE_QUERY_KEY, RACE_DATE_QUERY_KEY,
+                                   RACE_NUMBER_QUERY_KEY,
+                                   create_hkjc_hkjc_game_model)
+from .hkjc_hkjc_venue_model import HAPPY_VALLEY_VENUE_CODE, SHA_TIN_VENUE_CODE
+
+_BAD_URLS = {
+    "https://common.hkjc.com/racing/information/English/Racing/LocalResults.aspx?RaceDate=2025/03/05&Racecourse=HV&RaceNo=9",
+    "https://common.hkjc.com/racing/information/English/Racing/LocalResults.aspx?RaceDate=1992/04/22&Racecourse=HV&RaceNo=4",
+    "https://common.hkjc.com/racing/information/English/Racing/LocalResults.aspx?RaceDate=1992/04/22&Racecourse=HV&RaceNo=5",
+    "https://common.hkjc.com/racing/information/English/Racing/LocalResults.aspx?RaceDate=1992/04/22&Racecourse=HV&RaceNo=6",
+}
+
+
+class HKJCHKJCLeagueModel(LeagueModel):
+    """HKJC HKJC implementation of the league model."""
+
+    def __init__(self, session: ScrapeSession, position: int | None = None) -> None:
+        super().__init__(League.HKJC, session, position=position)
+
+    @classmethod
+    def name(cls) -> str:
+        return "hkjc-league-model"
+
+    @property
+    def games(self) -> Iterator[GameModel]:  # type: ignore
+        try:
+
+            def games_for_date(dt: datetime.date) -> Iterator[GameModel]:
+                dt_str = dt.strftime("%Y/%m/%d")
+                for course in [SHA_TIN_VENUE_CODE, HAPPY_VALLEY_VENUE_CODE]:
+                    with self.session.wayback_disabled():
+                        response = self.session.get(
+                            f"https://racing.hkjc.com/racing/information/English/Racing/LocalResults.aspx?{RACE_DATE_QUERY_KEY}={dt_str}&{RACE_COURSE_QUERY_KEY}={course}&{RACE_NUMBER_QUERY_KEY}=1",
+                        )
+                    response.raise_for_status()
+                    game_model = create_hkjc_hkjc_game_model(
+                        self.session, response.text, response.url
+                    )
+                    if game_model is not None:
+                        yield game_model
+                    soup = BeautifulSoup(response.text, "lxml")
+                    seen_urls = {response.url}
+                    for a in soup.find_all("a", href=True):
+                        if SHUTDOWN_FLAG.is_set():
+                            return
+                        url = urllib.parse.urljoin(response.url, a.get("href"))
+                        if url in seen_urls:
+                            continue
+                        o = urlparse(url)
+                        if not o.path.endswith("/LocalResults.aspx"):
+                            continue
+                        if url in _BAD_URLS:
+                            continue
+                        query = urllib.parse.parse_qs(o.query)
+                        if RACE_NUMBER_QUERY_KEY not in query:
+                            continue
+                        if query[RACE_COURSE_QUERY_KEY][0] != course:
+                            continue
+                        with self.session.wayback_disabled():
+                            response = self.session.get(url)
+                        if response.status_code == http.HTTPStatus.NOT_FOUND:
+                            continue
+                        response.raise_for_status()
+                        game_model = create_hkjc_hkjc_game_model(
+                            self.session, response.text, response.url
+                        )
+                        if game_model is not None:
+                            yield game_model
+                        seen_urls.add(url)
+
+            today_date = datetime.datetime.today().date()
+            last_scraped_date: datetime.date | None = None
+            current_date = today_date
+
+            with tqdm.tqdm(position=self.position) as pbar:
+                while (
+                    last_scraped_date is None
+                    or (last_scraped_date - current_date).days < 360
+                ):
+                    if current_date >= today_date:
+                        with self.session.cache_disabled():
+                            for game_model in games_for_date(current_date):
+                                last_scraped_date = current_date
+                                pbar.update(1)
+                                yield game_model
+                    else:
+                        for game_model in games_for_date(current_date):
+                            last_scraped_date = current_date
+                            pbar.update(1)
+                            yield game_model
+                    pbar.set_description(f"HKJC {current_date}")
+                    current_date -= datetime.timedelta(days=1)
+        except Exception as exc:
+            SHUTDOWN_FLAG.set()
+            raise exc

{sportsball-0.10.12 → sportsball-0.10.13}/sportsball/data/league_model.py

@@ -2,6 +2,7 @@
 
 # pylint: disable=line-too-long
 import datetime
+import threading
 from typing import Iterator, get_args, get_origin
 
 import pandas as pd
@@ -19,6 +20,7 @@ from .model import Model
 from .venue_model import VENUE_ADDRESS_COLUMN
 
 LEAGUE_COLUMN = "league"
+SHUTDOWN_FLAG = threading.Event()
 
 
 def _clear_column_list(df: pd.DataFrame) -> pd.DataFrame:

{sportsball-0.10.12 → sportsball-0.10.13}/sportsball/data/nba/nba/nba_nba_league_model.py

@@ -13,7 +13,7 @@ from scrapesession.scrapesession import ScrapeSession  # type: ignore
 
 from ...game_model import GameModel
 from ...league import League
-from ...league_model import LeagueModel
+from ...league_model import SHUTDOWN_FLAG, LeagueModel
 from .nba_nba_game_model import create_nba_nba_game_model
 
 
@@ -83,6 +83,8 @@ class NBANBALeagueModel(LeagueModel):
         pbar: tqdm.tqdm,
     ) -> Iterator[GameModel]:
         for _, row in all_games.iterrows():
+            if SHUTDOWN_FLAG.is_set():
+                return
             season_id = row["SEASON_ID"]
             dt = parse(row["GAME_DATE"])
             season_info = seasons.get(
@@ -112,30 +114,34 @@ class NBANBALeagueModel(LeagueModel):
 
     @property
     def games(self) -> Iterator[GameModel]:
-        to_date = datetime.datetime.today().date()
-        seasons: dict[str, NBANBALeagueModel._SeasonInfo] = {}
-        first_call = False
-        with tqdm.tqdm(position=self.position) as pbar:
-            while True:
-                next_date = to_date - relativedelta(years=1)
-                if not first_call:
-                    with self.session.cache_disabled():
+        try:
+            to_date = datetime.datetime.today().date()
+            seasons: dict[str, NBANBALeagueModel._SeasonInfo] = {}
+            first_call = False
+            with tqdm.tqdm(position=self.position) as pbar:
+                while True:
+                    next_date = to_date - relativedelta(years=1)
+                    if not first_call:
+                        with self.session.cache_disabled():
+                            result = leaguegamefinder.LeagueGameFinder(
+                                league_id_nullable=self._league_id,
+                                date_from_nullable=next_date.strftime("%m/%d/%Y"),
+                                date_to_nullable=to_date.strftime("%m/%d/%Y"),
+                            )
+                        first_call = True
+                    else:
                         result = leaguegamefinder.LeagueGameFinder(
                             league_id_nullable=self._league_id,
                             date_from_nullable=next_date.strftime("%m/%d/%Y"),
                             date_to_nullable=to_date.strftime("%m/%d/%Y"),
                         )
-                    first_call = True
-                else:
-                    result = leaguegamefinder.LeagueGameFinder(
-                        league_id_nullable=self._league_id,
-                        date_from_nullable=next_date.strftime("%m/%d/%Y"),
-                        date_to_nullable=to_date.strftime("%m/%d/%Y"),
-                    )
 
-                all_games = _combine_team_games(result.get_data_frames()[0])
-                all_games = all_games.sort_values(by="GAME_DATE", ascending=True)
-                if all_games.empty:
-                    break
-                yield from self._produce_games(all_games, seasons, pbar)
-                to_date = next_date
+                    all_games = _combine_team_games(result.get_data_frames()[0])
+                    all_games = all_games.sort_values(by="GAME_DATE", ascending=True)
+                    if all_games.empty:
+                        break
+                    yield from self._produce_games(all_games, seasons, pbar)
+                    to_date = next_date
+        except Exception as exc:
+            SHUTDOWN_FLAG.set()
+            raise exc

sportsball-0.10.13/sportsball/data/nba/nbacom/nba_nbacom_league_model.py

@@ -0,0 +1,46 @@
+"""NBA NBA.com league model."""
+
+# pylint: disable=too-many-statements,protected-access,too-many-locals,bare-except,too-many-branches,duplicate-code
+import datetime
+import logging
+from typing import Iterator
+
+from scrapesession.scrapesession import ScrapeSession  # type: ignore
+
+from ...game_model import VERSION, GameModel
+from ...league import League
+from ...league_model import SHUTDOWN_FLAG, LeagueModel
+from .nba_nbacom_game_model import create_nba_nbacom_game_model
+
+
+class NBANBAComLeagueModel(LeagueModel):
+    """NBA NBA.com implementation of the league model."""
+
+    def __init__(self, session: ScrapeSession, position: int | None = None) -> None:
+        super().__init__(League.AFL, session, position=position)
+
+    @classmethod
+    def name(cls) -> str:
+        return "nba-nbacom-league-model"
+
+    @property
+    def games(self) -> Iterator[GameModel]:
+        try:
+            with self.session.cache_disabled():
+                date_slug = datetime.datetime.today().date().strftime("%Y%m%d")
+                response = self.session.get(
+                    f"https://stats.nba.com/js/data/leaders/00_daily_lineups_{date_slug}.json"
+                )
+                if response.ok:
+                    lineup = response.json()
+                    for game in lineup["games"]:
+                        if SHUTDOWN_FLAG.is_set():
+                            return
+                        yield create_nba_nbacom_game_model(
+                            game=game, session=self.session, version=VERSION
+                        )
+                else:
+                    logging.warning("Failed to fetch NBA daily lineups.")
+        except Exception as exc:
+            SHUTDOWN_FLAG.set()
+            raise exc

{sportsball-0.10.12 → sportsball-0.10.13}/sportsball/data/oddsportal/oddsportal_league_model.py

@@ -15,7 +15,7 @@ from scrapesession.scrapesession import ScrapeSession  # type: ignore
 
 from ..game_model import GameModel
 from ..league import League
-from ..league_model import LeagueModel
+from ..league_model import SHUTDOWN_FLAG, LeagueModel
 from .decrypt import fetch_data
 from .oddsportal_game_model import create_oddsportal_game_model
 
@@ -63,6 +63,8 @@ def _process_results_pages(
         if d.get("total") == 0:
             return
         for row in d.get("rows", []):
+            if SHUTDOWN_FLAG.is_set():
+                return
             game_model = create_oddsportal_game_model(
                 session, urllib.parse.urljoin(url, row["url"]), league, False
             )
@@ -165,6 +167,10 @@ class OddsPortalLeagueModel(LeagueModel):
 
     @property
     def games(self) -> Iterator[GameModel]:
-        with tqdm.tqdm(position=self.position) as pbar:
-            yield from self._find_next(pbar)
-            yield from self._find_previous(pbar)
+        try:
+            with tqdm.tqdm(position=self.position) as pbar:
+                yield from self._find_next(pbar)
+                yield from self._find_previous(pbar)
+        except Exception as exc:
+            SHUTDOWN_FLAG.set()
+            raise exc