sportsball 0.3.59__tar.gz → 0.3.61__tar.gz

{sportsball-0.3.59/sportsball.egg-info → sportsball-0.3.61}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: sportsball
-Version: 0.3.59
+Version: 0.3.61
 Summary: A library for pulling in and normalising sports stats.
 Home-page: https://github.com/8W9aG/sportsball
 Author: Will Sackfield

{sportsball-0.3.59 → sportsball-0.3.61}/setup.py

@@ -26,7 +26,7 @@ def install_requires() -> typing.List[str]:
 
 setup(
     name='sportsball',
-    version='0.3.59',
+    version='0.3.61',
     description='A library for pulling in and normalising sports stats.',
     long_description=long_description,
     long_description_content_type='text/markdown',

{sportsball-0.3.59 → sportsball-0.3.61}/sportsball/__init__.py

@@ -4,4 +4,4 @@
 import cchardet  # type: ignore
 import lxml  # noqa: ignore
 
-__VERSION__ = "0.3.59"
+__VERSION__ = "0.3.61"

sportsball-0.3.61/sportsball/data/afl/afl/afl_afl_league_model.py

@@ -0,0 +1,137 @@
+"""AFL AFL league model."""
+
+# pylint: disable=too-many-statements,protected-access,too-many-locals,bare-except,too-many-branches
+import datetime
+import logging
+import re
+from typing import Iterator
+
+import requests_cache
+from bs4 import BeautifulSoup
+from playwright.sync_api import sync_playwright
+
+from ....playwright import ensure_install
+from ...game_model import GameModel
+from ...league import League
+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]
+) -> 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(
+            "span", {"class": re.compile(".*team-lineups__team-name.*")}
+        ):
+            team_names.append(span.get_text().strip())
+        teams_players: list[list[tuple[str, str, str, str]]] = [[], []]
+        for div_positions_row in div.find_all(
+            "div", {"class": re.compile(".*team-lineups__positions-row.*")}
+        ):
+            row_players: list[list[tuple[str, str, str, str]]] = []
+            for div_team_players in div_positions_row.find_all(
+                "div",
+                {"class": re.compile(".*team-lineups__positions-players.*")},
+            ):
+                team_players: list[tuple[str, str, str, str]] = []
+                for a in div_team_players.find_all(
+                    "a", {"class": re.compile(".*js-player-profile-link.*")}
+                ):
+                    player_id = "afl:" + a.get("data-player-id")
+                    first_name = a.get("data-first-name")
+                    second_name = a.get("data-surname")
+                    player_number = None
+                    for span_player_number in a.find_all(
+                        "span",
+                        {"class": re.compile(".*team-lineups__player-number.*")},
+                    ):
+                        player_number = (
+                            span_player_number.get_text()
+                            .strip()
+                            .replace("[", "")
+                            .replace("]", "")
+                        )
+                    if player_number is None:
+                        raise ValueError("player_number is null")
+                    team_players.append(
+                        (player_id, player_number, first_name, second_name)
+                    )
+                row_players.append(team_players)
+                if len(row_players) >= 2:
+                    break
+            for count, row_players_list in enumerate(row_players):
+                try:
+                    teams_players[count].extend(row_players_list)
+                except IndexError as exc:
+                    # logging.error(html)
+                    logging.error(len(team_players))
+                    logging.error(count)
+                    raise exc
+        dt = None
+        for time in div.find_all(
+            "time", {"class": re.compile(".*match-list-alt__header-time.*")}
+        ):
+            dt = datetime.datetime.fromtimestamp(float(time.get("data-date")) / 1000.0)
+        venue_name: str | None = None
+        for span in div.find_all(
+            "span", {"class": re.compile(".*match-list-alt__header-venue.*")}
+        ):
+            venue_name = span.get_text().strip().replace(",", "")
+        if venue_name is None:
+            raise ValueError("venue_name is null")
+        if dt is None:
+            raise ValueError("dt is null")
+        yield create_afl_afl_game_model(
+            team_names, teams_players, dt, venue_name, session, ladder
+        )
+
+
+class AFLAFLLeagueModel(LeagueModel):
+    """AFL AFL implementation of the league model."""
+
+    def __init__(
+        self, session: requests_cache.CachedSession, position: int | None = None
+    ) -> None:
+        super().__init__(League.AFL, session)
+
+    @property
+    def _ladder(self) -> list[str]:
+        ladder = []
+        ensure_install()
+        with sync_playwright() as p:
+            browser = p.chromium.launch()
+            context = browser.new_context()
+            page = context.new_page()
+            try:
+                page.goto(
+                    "https://www.afl.com.au/ladder",
+                    wait_until="networkidle",
+                    timeout=60000.0,
+                )
+            except:  # noqa: E722
+                logging.warning("Ladder URL timed out.")
+            soup = BeautifulSoup(page.content(), "lxml")
+            for span in soup.find_all(
+                "span", {"class": re.compile(".*stats-table__club-name.*")}
+            ):
+                team_name = span.get_text().strip()
+                if team_name in ladder:
+                    continue
+                ladder.append(team_name)
+        return ladder
+
+    @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()
+            page.goto(
+                "https://www.afl.com.au/matches/team-lineups", wait_until="networkidle"
+            )
+            yield from _parse_game_info(page.content(), self.session, ladder)

{sportsball-0.3.59 → sportsball-0.3.61}/sportsball/data/google/google_address_model.py

@@ -2287,6 +2287,15 @@ NEW_ORLEANS_CITY = SportsballGeocodeTuple(
     housenumber="",
     country="USA",
 )
+PETCO_PARK = SportsballGeocodeTuple(
+    city="San Diego",
+    state="CA",
+    postal="92101",
+    lat=32.7073,
+    lng=-117.1566,
+    housenumber="100",
+    country="USA",
+)
 _CACHED_GEOCODES: dict[str, Any] = {
     "S.C.G. - Australia": SCG,
     "Victoria Park - Australia": SportsballGeocodeTuple(
@@ -4035,15 +4044,7 @@ _CACHED_GEOCODES: dict[str, Any] = {
     ),
     "Riccardo Silva Stadium - Miami - FL - 33199": PITBULL_STADIUM,
     "Ryan Field - Evanston - IL - 60208": LANNY_AND_SHARON_MARTIN_STADIUM,
-    "Petco Park - San Diego - CA - 92101": SportsballGeocodeTuple(
-        city="San Diego",
-        state="CA",
-        postal="92101",
-        lat=32.7073,
-        lng=-117.1566,
-        housenumber="100",
-        country="USA",
-    ),
+    "Petco Park - San Diego - CA - 92101": PETCO_PARK,
     "Providence Park - Portland - OR - 97205": SportsballGeocodeTuple(
         city="Portland",
         state="OR",
@@ -12709,6 +12710,45 @@ _CACHED_GEOCODES: dict[str, Any] = {
         country="USA",
     ),
     "New Orleans , Louisiana": NEW_ORLEANS_CITY,
+    "First Bank & Trust Arena, Brookings, South Dakota": SportsballGeocodeTuple(
+        city="Brookings",
+        state="SD",
+        postal="57007",
+        lat=44.317984,
+        lng=-96.779058,
+        housenumber="1165",
+        country="USA",
+    ),
+    "Cambria County War Memorial Arena, Johnstown, Pennsylvania": SportsballGeocodeTuple(
+        city="Johnstown",
+        state="PA",
+        postal="15901",
+        lat=40.32296,
+        lng=-78.922654,
+        housenumber="326",
+        country="USA",
+    ),
+    "Petco Park, San Diego, California": PETCO_PARK,
+    "Esby Gym, Glassboro, New Jersey": SportsballGeocodeTuple(
+        city="Glassboro",
+        state="NJ",
+        postal="08028",
+        lat=39.7106001,
+        lng=-75.1186627,
+        housenumber="300",
+        country="USA",
+    ),
+    "Boone, North Carolina": SportsballGeocodeTuple(
+        city="Boone",
+        state="NC",
+        postal="",
+        lat=36.211111,
+        lng=-81.666944,
+        housenumber="",
+        country="USA",
+    ),
+    "FedExForum - Memphis, Tennessee - United States": FEDEX_FORUM,
+    "Kia Center, Orlando, FL, EUA": KIA_CENTER,
 }
 
 

{sportsball-0.3.59 → sportsball-0.3.61}/sportsball/data/sportsreference/sportsreference_game_model.py

@@ -179,27 +179,33 @@ def _find_old_dt(
             test_row = df.iat[0, 0]
             test_row_2 = df.iat[1, 0]
 
-            if (
-                "Prev Game" in test_row
-                or "Next Game" in test_row
-                or "Lost" in test_row
-                or "PM," in test_row_2
-            ):
-                date_venue_split = df.iat[1, 0].split()
-                current_idx = 5
-                try:
-                    dt = parse(" ".join(date_venue_split[:current_idx]))
-                except dateutil.parser._parser.ParserError:  # type: ignore
+            try:
+                if (
+                    "Prev Game" in test_row
+                    or "Next Game" in test_row
+                    or "Lost" in test_row
+                    or "PM," in test_row_2
+                ):
+                    date_venue_split = df.iat[1, 0].split()
+                    current_idx = 5
                     try:
-                        current_idx = 3
                         dt = parse(" ".join(date_venue_split[:current_idx]))
                     except dateutil.parser._parser.ParserError:  # type: ignore
-                        dt = parse(
-                            " ".join(",".join(test_row.split(",")[1:]).split()[:3])
-                        )
-                venue_name = " ".join(date_venue_split[current_idx:])
-                _process_team_row(df)
-                break
+                        try:
+                            current_idx = 3
+                            dt = parse(" ".join(date_venue_split[:current_idx]))
+                        except dateutil.parser._parser.ParserError:  # type: ignore
+                            dt = parse(
+                                " ".join(",".join(test_row.split(",")[1:]).split()[:3])
+                            )
+                    venue_name = " ".join(date_venue_split[current_idx:])
+                    _process_team_row(df)
+                    break
+            except TypeError as exc:
+                logging.error(test_row)
+                logging.error(response.url)
+                logging.error(response.text)
+                raise exc
 
     if dt is None:
         title_tag = soup.find("title")
@@ -305,7 +311,7 @@ def _create_sportsreference_game_model(
 
     # If the page_title is bad, try fetching from a non wayback source
     if page_title is not None:
-        if page_title.get_text().strip() == "File Not Found":
+        if "File Not Found" in page_title.get_text().strip():
             session.cache.delete(urls=[url])
             response = session.get(url, headers={X_NO_WAYBACK: "1"})
             response.raise_for_status()

{sportsball-0.3.59 → sportsball-0.3.61}/sportsball/data/sportsreference/sportsreference_team_model.py

@@ -90,6 +90,20 @@ def _create_sportsreference_team_model(
     assists: dict[str, int],
     turnovers: dict[str, int],
 ) -> TeamModel:
+    if url == "https://www.sports-reference.com/cbb/schools/alaska-anchorage/2016.html":
+        name = "Alaska-Anchorage"
+        return TeamModel(
+            identifier=name,
+            name=name,
+            location="Alaska",
+            players=[],
+            odds=[],
+            points=points,
+            ladder_rank=None,
+            news=create_google_news_models(name, session, dt, league),
+            social=create_x_social_model(name, session, dt),
+        )
+
     headers = {}
     if url in _NON_WAYBACK_URLS:
         headers = {X_NO_WAYBACK: "1"}

{sportsball-0.3.59 → sportsball-0.3.61/sportsball.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: sportsball
-Version: 0.3.59
+Version: 0.3.61
 Summary: A library for pulling in and normalising sports stats.
 Home-page: https://github.com/8W9aG/sportsball
 Author: Will Sackfield

{sportsball-0.3.59 → sportsball-0.3.61}/sportsball.egg-info/SOURCES.txt

@@ -165,6 +165,8 @@ tests/data/player_model_test.py
 tests/data/social_model_test.py
 tests/data/team_model_test.py
 tests/data/afl/__init__.py
+tests/data/afl/afl/__init__.py
+tests/data/afl/afl/afl_afl_league_model_test.py
 tests/data/afl/afltables/__init__.py
 tests/data/afl/afltables/afl_afltables_game_model_test.py
 tests/data/afl/afltables/afl_afltables_league_model_test.py

sportsball-0.3.61/tests/data/afl/afl/afl_afl_league_model_test.py

@@ -0,0 +1,37 @@
+"""Tests for the afl AFL league model class."""
+import unittest
+import os
+import datetime
+
+import requests_cache
+from sportsball.data.afl.afl.afl_afl_league_model import _parse_game_info
+
+
+class TestAFLAFLLeagueModel(unittest.TestCase):
+
+    def setUp(self):
+        self.dir = os.path.dirname(__file__)
+        self._session = requests_cache.CachedSession(backend="memory")
+
+    def test_parse_game_info(self):
+        with open(os.path.join(self.dir, "team-lineups.html")) as handle:
+            list(_parse_game_info(handle.read(), self._session, [
+                "Melbourne",
+                "Richmond",
+                "Collingwood",
+                "Essendon",
+                "Fremantle",
+                "Adelaide Crows",
+                "St Kilda",
+                "Brisbane Lions",
+                "Port Adelaide",
+                "North Melbourne",
+                "GWS GIANTS",
+                "Western Bulldogs",
+                "Gold Coast SUNS",
+                "Sydney Swans",
+                "Carlton",
+                "Geelong Cats",
+                "Hawthorn",
+                "West Coast Eagles",
+            ]))

sportsball-0.3.59/sportsball/data/afl/afl/afl_afl_league_model.py

@@ -1,132 +0,0 @@
-"""AFL AFL league model."""
-
-# pylint: disable=too-many-statements,protected-access,too-many-locals,bare-except
-import datetime
-import logging
-import re
-from typing import Iterator
-
-import requests_cache
-from bs4 import BeautifulSoup
-from playwright.sync_api import sync_playwright
-
-from ....playwright import ensure_install
-from ...game_model import GameModel
-from ...league import League
-from ...league_model import LeagueModel
-from .afl_afl_game_model import create_afl_afl_game_model
-
-
-class AFLAFLLeagueModel(LeagueModel):
-    """AFL AFL implementation of the league model."""
-
-    def __init__(
-        self, session: requests_cache.CachedSession, position: int | None = None
-    ) -> None:
-        super().__init__(League.AFL, session)
-
-    @property
-    def _ladder(self) -> list[str]:
-        ladder = []
-        ensure_install()
-        with sync_playwright() as p:
-            browser = p.chromium.launch()
-            context = browser.new_context()
-            page = context.new_page()
-            try:
-                page.goto(
-                    "https://www.afl.com.au/ladder",
-                    wait_until="networkidle",
-                    timeout=60000.0,
-                )
-            except:  # noqa: E722
-                logging.warning("Ladder URL timed out.")
-            soup = BeautifulSoup(page.content(), "lxml")
-            for span in soup.find_all(
-                "span", {"class": re.compile(".*stats-table__club-name.*")}
-            ):
-                team_name = span.get_text().strip()
-                if team_name in ladder:
-                    continue
-                ladder.append(team_name)
-        return ladder
-
-    @property
-    def games(self) -> Iterator[GameModel]:
-        # https://www.afl.com.au/matches/team-lineups
-        ladder = self._ladder
-        ensure_install()
-        with sync_playwright() as p:
-            browser = p.chromium.launch()
-            context = browser.new_context()
-            page = context.new_page()
-            page.goto(
-                "https://www.afl.com.au/matches/team-lineups", wait_until="networkidle"
-            )
-            soup = BeautifulSoup(page.content(), "lxml")
-            for div in soup.find_all(
-                "div", {"class": re.compile(".*js-match-list-item.*")}
-            ):
-                team_names = []
-                for span in div.find_all(
-                    "span", {"class": re.compile(".*team-lineups__team-name.*")}
-                ):
-                    team_names.append(span.get_text().strip())
-                teams_players: list[list[tuple[str, str, str, str]]] = [[], []]
-                for div_positions_row in div.find_all(
-                    "div", {"class": re.compile(".*team-lineups__positions-row.*")}
-                ):
-                    row_players: list[list[tuple[str, str, str, str]]] = []
-                    for div_team_players in div_positions_row.find_all(
-                        "div",
-                        {"class": re.compile(".*team-lineups__positions-players.*")},
-                    ):
-                        team_players: list[tuple[str, str, str, str]] = []
-                        for a in div_team_players.find_all(
-                            "a", {"class": re.compile(".*js-player-profile-link.*")}
-                        ):
-                            player_id = "afl:" + a.get("data-player-id")
-                            first_name = a.get("data-first-name")
-                            second_name = a.get("data-surname")
-                            player_number = None
-                            for span_player_number in a.find_all(
-                                "span",
-                                {
-                                    "class": re.compile(
-                                        ".*team-lineups__player-number.*"
-                                    )
-                                },
-                            ):
-                                player_number = (
-                                    span_player_number.get_text()
-                                    .strip()
-                                    .replace("[", "")
-                                    .replace("]", "")
-                                )
-                            if player_number is None:
-                                raise ValueError("player_number is null")
-                            team_players.append(
-                                (player_id, player_number, first_name, second_name)
-                            )
-                        row_players.append(team_players)
-                    for count, row_players_list in enumerate(row_players):
-                        teams_players[count].extend(row_players_list)
-                dt = None
-                for time in div.find_all(
-                    "time", {"class": re.compile(".*match-list-alt__header-time.*")}
-                ):
-                    dt = datetime.datetime.fromtimestamp(
-                        float(time.get("data-date")) / 1000.0
-                    )
-                venue_name: str | None = None
-                for span in div.find_all(
-                    "span", {"class": re.compile(".*match-list-alt__header-venue.*")}
-                ):
-                    venue_name = span.get_text().strip().replace(",", "")
-                if venue_name is None:
-                    raise ValueError("venue_name is null")
-                if dt is None:
-                    raise ValueError("dt is null")
-                yield create_afl_afl_game_model(
-                    team_names, teams_players, dt, venue_name, self.session, ladder
-                )