sportsball 0.3.117__tar.gz → 0.3.119__tar.gz

{sportsball-0.3.117/sportsball.egg-info → sportsball-0.3.119}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: sportsball
-Version: 0.3.117
+Version: 0.3.119
 Summary: A library for pulling in and normalising sports stats.
 Home-page: https://github.com/8W9aG/sportsball
 Author: Will Sackfield

{sportsball-0.3.117 → sportsball-0.3.119}/setup.py

@@ -26,7 +26,7 @@ def install_requires() -> typing.List[str]:
 
 setup(
     name='sportsball',
-    version='0.3.117',
+    version='0.3.119',
     description='A library for pulling in and normalising sports stats.',
     long_description=long_description,
     long_description_content_type='text/markdown',

{sportsball-0.3.117 → sportsball-0.3.119}/sportsball/__init__.py

@@ -4,4 +4,4 @@
 import cchardet  # type: ignore
 import lxml  # noqa: ignore
 
-__VERSION__ = "0.3.117"
+__VERSION__ = "0.3.119"

{sportsball-0.3.117 → sportsball-0.3.119}/sportsball/data/afl/afl/afl_afl_game_model.py

@@ -2,12 +2,12 @@
 
 # pylint: disable=too-many-statements,protected-access,too-many-arguments,bare-except
 import datetime
-import logging
+import re
 
 import pytest_is_running
 import requests_cache
 from bs4 import BeautifulSoup
-from playwright.sync_api import sync_playwright
+from playwright.sync_api import Playwright
 
 from ....playwright import ensure_install
 from ...game_model import GameModel
@@ -16,36 +16,49 @@ from .afl_afl_team_model import create_afl_afl_team_model
 from .afl_afl_venue_model import create_afl_afl_venue_model
 
 
-def _extract_odds(html: str) -> list[float]:
+def _extract_odds(soup: BeautifulSoup) -> list[float]:
     odds = []
-    soup = BeautifulSoup(html, "lxml")
     for span in soup.find_all("span", {"class": "betting-button__money"}):
         odds.append(float(span.get_text().strip().replace("$", "")))
     return odds
 
 
+def _extract_dt(soup: BeautifulSoup) -> datetime.datetime:
+    for div in soup.find_all("div", {"class": re.compile(".*js-match-start-time.*")}):
+        start_time = div.get("data-start-time")
+        return datetime.datetime.fromisoformat(start_time)
+    raise ValueError("Unable to find datetime")
+
+
+def _parse(html: str) -> tuple[list[float], datetime.datetime]:
+    soup = BeautifulSoup(html, "lxml")
+    return _extract_odds(soup), _extract_dt(soup)
+
+
 def create_afl_afl_game_model(
     team_names: list[str],
     players: list[list[tuple[str, str, str, str]]],
-    dt: datetime.datetime,
+    dt: datetime.datetime | None,
     venue_name: str,
     session: requests_cache.CachedSession,
     ladder: list[str],
     url: str | None,
+    playwright: Playwright,
 ) -> GameModel:
     """Create a game model from AFL Tables."""
-    odds = []
+    odds: list[float] = []
     if url is not None and not pytest_is_running.is_running():
         ensure_install()
-        with sync_playwright() as p:
-            browser = p.chromium.launch()
-            context = browser.new_context()
-            page = context.new_page()
-            try:
-                page.goto(url, wait_until="networkidle")
-            except:  # noqa: E722
-                logging.warning("Ladder URL timed out.")
-            odds = _extract_odds(page.content())
+        browser = playwright.chromium.launch()
+        context = browser.new_context()
+        page = context.new_page()
+        try:
+            page.goto(url, wait_until="networkidle")
+        except:  # noqa: E722
+            pass
+        odds, dt = _parse(page.content())
+    if dt is None:
+        raise ValueError("dt is null")
 
     venue_model = create_afl_afl_venue_model(venue_name, session, dt)
     teams = [

{sportsball-0.3.117 → sportsball-0.3.119}/sportsball/data/afl/afl/afl_afl_league_model.py

@@ -8,9 +8,9 @@ import urllib.parse
 from typing import Iterator
 
 import requests_cache
-from bs4 import BeautifulSoup
+from bs4 import BeautifulSoup, Tag
 from dateutil.parser import parse
-from playwright.sync_api import sync_playwright
+from playwright.sync_api import Playwright, sync_playwright
 
 from ....playwright import ensure_install
 from ...game_model import GameModel
@@ -19,10 +19,13 @@ from ...league_model import LeagueModel
 from .afl_afl_game_model import create_afl_afl_game_model
 
 
-def _parse_game_info(
-    html: str, session: requests_cache.CachedSession, ladder: list[str], html_url: str
+def _parse_v1(
+    soup: BeautifulSoup,
+    session: requests_cache.CachedSession,
+    ladder: list[str],
+    html_url: str,
+    playwright: Playwright,
 ) -> Iterator[GameModel]:
-    soup = BeautifulSoup(html, "lxml")
     for div in soup.find_all("div", {"class": re.compile(".*js-match-list-item.*")}):
         team_names = []
         for span in div.find_all(
@@ -95,10 +98,101 @@ def _parse_game_info(
         ):
             url = urllib.parse.urljoin(html_url, a.get("href"))
         yield create_afl_afl_game_model(
-            team_names, teams_players, dt, venue_name, session, ladder, url
+            team_names,
+            teams_players,
+            dt,
+            venue_name,
+            session,
+            ladder,
+            url,
+            playwright,
         )
 
 
+def _parse_v2(
+    soup: BeautifulSoup,
+    session: requests_cache.CachedSession,
+    ladder: list[str],
+    html_url: str,
+    playwright: Playwright,
+) -> Iterator[GameModel]:
+    for div in soup.find_all("div", {"class": "team-lineups__item"}):
+        team_names = []
+        for span in div.find_all(
+            "span", {"class": re.compile(".*team-lineups-header__name.*")}
+        ):
+            team_names = [x.strip() for x in span.get_text().strip().split(" v ")]
+        team_players: list[list[tuple[str, str, str, str]]] = [[], []]
+
+        def process_a_player(
+            a: Tag, idx: int, team_players: list[list[tuple[str, str, str, str]]]
+        ) -> list[list[tuple[str, str, str, str]]]:
+            href = a.get("href")
+            if not isinstance(href, str):
+                raise ValueError("href is not a str")
+            player_id = "afl:" + href.split("/")[2]
+            name = a.get("title")
+            if not isinstance(name, str):
+                raise ValueError("name is not a str")
+            first_name, sur_name = name.split()
+            jersey = None
+            for div_shirt in a.find_all(
+                "div", {"class": re.compile(".*team-lineups__player-entry--shirt.*")}
+            ):
+                jersey = div_shirt.get_text().strip()
+            if jersey is None:
+                raise ValueError("jersey is null")
+            team_players[idx].append((player_id, first_name, sur_name, jersey))
+            return team_players
+
+        for a in div.find_all(
+            "a", {"class": re.compile(".*team-lineups__player-entry--home-team.*")}
+        ):
+            team_players = process_a_player(a, 0, team_players)
+        for a in div.find_all(
+            "a", {"class": re.compile(".*team-lineups__player-entry--away-team.*")}
+        ):
+            team_players = process_a_player(a, 1, team_players)
+
+        venue_name = None
+        for div_info in div.find_all("div", {"class": "team-lineups-header__info"}):
+            header_info = div_info.get_text().strip()
+            venue_name = header_info.split("･")[1].strip()
+        if venue_name is None:
+            raise ValueError("venue_name is null")
+
+        url = None
+        for a in div.find_all("a", {"class": "team-lineups-header"}):
+            url = urllib.parse.urljoin(html_url, a.get("href"))
+
+        yield create_afl_afl_game_model(
+            team_names,
+            team_players,
+            None,
+            venue_name,
+            session,
+            ladder,
+            url,
+            playwright,
+        )
+
+
+def _parse_game_info(
+    html: str,
+    session: requests_cache.CachedSession,
+    ladder: list[str],
+    html_url: str,
+    playwright: Playwright,
+) -> Iterator[GameModel]:
+    soup = BeautifulSoup(html, "lxml")
+    found = False
+    for game_model in _parse_v1(soup, session, ladder, html_url, playwright):
+        found = True
+        yield game_model
+    if not found:
+        yield from _parse_v2(soup, session, ladder, html_url, playwright)
+
+
 class AFLAFLLeagueModel(LeagueModel):
     """AFL AFL implementation of the league model."""
 
@@ -126,7 +220,7 @@ class AFLAFLLeagueModel(LeagueModel):
                     timeout=60000.0,
                 )
             except:  # noqa: E722
-                logging.warning("Ladder URL timed out.")
+                pass
             soup = BeautifulSoup(page.content(), "lxml")
             for span in soup.find_all(
                 "span", {"class": re.compile(".*stats-table__club-name.*")}
@@ -135,6 +229,7 @@ class AFLAFLLeagueModel(LeagueModel):
                 if team_name in ladder:
                     continue
                 ladder.append(team_name)
+            logging.info("Found ladder: %s", ",".join(ladder))
         return ladder
 
     @property
@@ -147,4 +242,4 @@ class AFLAFLLeagueModel(LeagueModel):
             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)
+            yield from _parse_game_info(page.content(), self.session, ladder, url, p)

{sportsball-0.3.117 → sportsball-0.3.119}/sportsball/data/afl/combined/afl_combined_league_model.py

@@ -251,6 +251,15 @@ AFL_VENUE_IDENTITY_MAP = {
     "Ondrej Nepela Arena": TIPOS_ARENA,
     "Barossa Park": BAROSSA_PARK,
     "Ninja Stadium": BELLERIVE_OVAL,
+    # AFL
+    "TIO Stadium, Darwin": MARRARA_OVAL,
+    "SCG, Sydney": SCG,
+    "MCG, Melbourne": MCG,
+    "Adelaide Oval, Adelaide": ADELAIDE_OVAL,
+    "ENGIE Stadium, Sydney": STADIUM_AUSTRALIA,
+    "Marvel Stadium, Melbourne": DOCKLANDS,
+    "Gabba, Brisbane": GABBA,
+    "Optus Stadium, Perth": PERTH,
 }
 
 

{sportsball-0.3.117 → sportsball-0.3.119}/sportsball/data/google/google_address_model.py

@@ -13864,6 +13864,24 @@ _CACHED_GEOCODES: dict[str, Any] = {
     ),
     "Cox Arena, San Diego, California": VIEJAS_ARENA,
     "The Omni, Atlanta, Georgia": OMNI_COLISEUM,
+    "TD Banknorth Sports Center, Hamden, Connecticut": TD_BANK_SPORTS_CENTER,
+    "The Event Center, San Jose, California": SportsballGeocodeTuple(
+        city="San Jose",
+        state="CA",
+        postal="95192",
+        lat=37.33536,
+        lng=-121.87985,
+        housenumber="290",
+        country="USA",
+    ),
+    "TIO Stadium, Darwin": TIO_STADIUM,
+    "SCG, Sydney": SCG,
+    "MCG, Melbourne": MCG,
+    "Adelaide Oval, Adelaide": ADELAIDE_OVAL,
+    "ENGIE Stadium, Sydney": ENGIE_STADIUM,
+    "Marvel Stadium, Melbourne": MARVEL_STADIUM,
+    "Gabba, Brisbane": GABBA,
+    "Optus Stadium, Perth": OPTUS_STADIUM,
 }
 
 

{sportsball-0.3.117 → sportsball-0.3.119/sportsball.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: sportsball
-Version: 0.3.117
+Version: 0.3.119
 Summary: A library for pulling in and normalising sports stats.
 Home-page: https://github.com/8W9aG/sportsball
 Author: Will Sackfield

{sportsball-0.3.117 → sportsball-0.3.119}/tests/data/afl/afl/afl_afl_league_model_test.py

@@ -34,7 +34,7 @@ class TestAFLAFLLeagueModel(unittest.TestCase):
                 "Geelong Cats",
                 "Hawthorn",
                 "West Coast Eagles",
-            ], "https://www.afl.com.au/matches/team-lineups"))
+            ], "https://www.afl.com.au/matches/team-lineups", None))
             first_game = game_models[0]
             self.assertEqual(len(first_game.teams), 2)
             for team in first_game.teams: