sportsball 0.5.4__tar.gz → 0.5.6__tar.gz

{sportsball-0.5.4/sportsball.egg-info → sportsball-0.5.6}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: sportsball
-Version: 0.5.4
+Version: 0.5.6
 Summary: A library for pulling in and normalising sports stats.
 Home-page: https://github.com/8W9aG/sportsball
 Author: Will Sackfield

{sportsball-0.5.4 → sportsball-0.5.6}/setup.py

@@ -26,7 +26,7 @@ def install_requires() -> typing.List[str]:
 
 setup(
     name='sportsball',
-    version='0.5.4',
+    version='0.5.6',
     description='A library for pulling in and normalising sports stats.',
     long_description=long_description,
     long_description_content_type='text/markdown',

{sportsball-0.5.4 → sportsball-0.5.6}/sportsball/__init__.py

@@ -4,4 +4,4 @@
 import cchardet  # type: ignore
 import lxml  # noqa: ignore
 
-__VERSION__ = "0.5.4"
+__VERSION__ = "0.5.6"

{sportsball-0.5.4 → sportsball-0.5.6}/sportsball/data/coach_model.py

@@ -5,7 +5,7 @@ from typing import Literal
 
 from pydantic import BaseModel, Field
 
-from .field_type import TYPE_KEY, FieldType
+from .field_type import FFILL_KEY, TYPE_KEY, FieldType
 
 COACH_IDENTIFIER_COLUMN: Literal["identifier"] = "identifier"
 COACH_NAME_COLUMN: Literal["name"] = "name"
@@ -22,5 +22,7 @@ class CoachModel(BaseModel):
         alias=COACH_IDENTIFIER_COLUMN,
     )
     name: str = Field(..., alias=COACH_NAME_COLUMN)
-    birth_date: datetime.date | None = Field(..., alias=COACH_BIRTH_DATE_COLUMN)
+    birth_date: datetime.date | None = Field(
+        ..., json_schema_extra={FFILL_KEY: True}, alias=COACH_BIRTH_DATE_COLUMN
+    )
     age: int | None = Field(..., alias=COACH_AGE_COLUMN)

{sportsball-0.5.4 → sportsball-0.5.6}/sportsball/data/combined/combined_coach_model.py

@@ -1,11 +1,16 @@
 """Combined coach model."""
 
+from typing import Any
+
 from ..coach_model import CoachModel
+from ..field_type import FFILL_KEY
 from .null_check import is_null
 
 
 def create_combined_coach_model(
-    coach_models: list[CoachModel], identifier: str
+    coach_models: list[CoachModel],
+    identifier: str,
+    coach_ffill: dict[str, dict[str, Any]],
 ) -> CoachModel:
     """Create a coach model by combining many coach models."""
     name = None
@@ -23,9 +28,22 @@ def create_combined_coach_model(
             age = coach_model_age
     if name is None:
         raise ValueError("name is null.")
-    return CoachModel(
+    coach_model = CoachModel(
         identifier=identifier,
         name=name,
         birth_date=birth_date,
         age=age,
     )
+
+    coach_instance_ffill = coach_ffill.get(identifier, {})
+    for field_name, field in coach_model.model_fields.items():
+        extra = field.json_schema_extra or {}
+        if extra.get(FFILL_KEY, False):  # type: ignore
+            current_value = getattr(coach_model, field_name)
+            if current_value is None:
+                setattr(coach_model, field_name, coach_instance_ffill.get(field_name))
+            else:
+                coach_instance_ffill[field_name] = current_value
+    coach_ffill[identifier] = coach_instance_ffill
+
+    return coach_model

{sportsball-0.5.4 → sportsball-0.5.6}/sportsball/data/combined/combined_game_model.py

@@ -2,6 +2,7 @@
 
 # pylint: disable=too-many-locals,line-too-long,too-many-arguments,too-many-branches,too-many-statements
 import logging
+from typing import Any
 
 import requests
 
@@ -38,6 +39,9 @@ def _team_models(
     player_identity_map: dict[str, str],
     names: dict[str, str],
     coach_names: dict[str, str],
+    player_ffill: dict[str, dict[str, Any]],
+    team_ffill: dict[str, dict[str, Any]],
+    coach_ffill: dict[str, dict[str, Any]],
 ) -> list[TeamModel]:
     team_models: dict[str, list[TeamModel]] = {}
     for game_model in game_models:
@@ -54,7 +58,16 @@ def _team_models(
                     team_model
                 ]
     return [
-        create_combined_team_model(v, k, player_identity_map, names, coach_names)  # pyright: ignore
+        create_combined_team_model(
+            v,
+            k,
+            player_identity_map,
+            names,
+            coach_names,
+            player_ffill,
+            team_ffill,
+            coach_ffill,
+        )  # pyright: ignore
         for k, v in team_models.items()
     ]
 
@@ -68,11 +81,21 @@ def create_combined_game_model(
     names: dict[str, str],
     coach_names: dict[str, str],
     last_game_number: int | None,
+    player_ffill: dict[str, dict[str, Any]],
+    team_ffill: dict[str, dict[str, Any]],
+    coach_ffill: dict[str, dict[str, Any]],
 ) -> GameModel:
     """Create a game model by combining many game models."""
     venue_models, full_venue_identity = _venue_models(game_models, venue_identity_map)
     full_team_models = _team_models(
-        game_models, team_identity_map, player_identity_map, names, coach_names
+        game_models,
+        team_identity_map,
+        player_identity_map,
+        names,
+        coach_names,
+        player_ffill,
+        team_ffill,
+        coach_ffill,
     )
     attendance = None
     end_dt = None

{sportsball-0.5.4 → sportsball-0.5.6}/sportsball/data/combined/combined_league_model.py

@@ -91,17 +91,23 @@ class CombinedLeagueModel(LeagueModel):
                 games[key] = games.get(key, []) + [game_model]
         names: dict[str, str] = {}
         coach_names: dict[str, str] = {}
+        player_ffill: dict[str, dict[str, Any]] = {}
+        team_ffill: dict[str, dict[str, Any]] = {}
+        coach_ffill: dict[str, dict[str, Any]] = {}
         last_game_number = None
         for game_models in games.values():
             game_model = create_combined_game_model(  # type: ignore
-                game_models,
-                self.venue_identity_map(),
-                team_identity_map,
-                self.player_identity_map(),
-                self.session,
-                names,
-                coach_names,
-                last_game_number,
+                game_models=game_models,
+                venue_identity_map=self.venue_identity_map(),
+                team_identity_map=team_identity_map,
+                player_identity_map=self.player_identity_map(),
+                session=self.session,
+                names=names,
+                coach_names=coach_names,
+                last_game_number=last_game_number,
+                player_ffill=player_ffill,
+                team_ffill=team_ffill,
+                coach_ffill=coach_ffill,
             )
             last_game_number = game_model.game_number
             yield game_model

{sportsball-0.5.4 → sportsball-0.5.6}/sportsball/data/combined/combined_player_model.py

@@ -1,12 +1,17 @@
 """Combined player model."""
 
 # pylint: disable=too-many-locals,too-many-branches,too-many-statements,duplicate-code
+from typing import Any
+
+from ..field_type import FFILL_KEY
 from ..player_model import PlayerModel
 from .null_check import is_null
 
 
 def create_combined_player_model(
-    player_models: list[PlayerModel], identifier: str
+    player_models: list[PlayerModel],
+    identifier: str,
+    player_ffill: dict[str, dict[str, Any]],
 ) -> PlayerModel:
     """Create a player model by combining many player models."""
     jersey = None
@@ -234,7 +239,8 @@ def create_combined_player_model(
         raise ValueError("name is null")
     if species is None:
         raise ValueError("species is null")
-    return PlayerModel(
+
+    player_model = PlayerModel(
         identifier=identifier,
         jersey=jersey,
         kicks=kicks,
@@ -291,3 +297,16 @@ def create_combined_player_model(
         game_score=game_score,
         point_differential=point_differential,
     )
+
+    player_instance_ffill = player_ffill.get(identifier, {})
+    for field_name, field in player_model.model_fields.items():
+        extra = field.json_schema_extra or {}
+        if extra.get(FFILL_KEY, False):  # type: ignore
+            current_value = getattr(player_model, field_name)
+            if current_value is None:
+                setattr(player_model, field_name, player_instance_ffill.get(field_name))
+            else:
+                player_instance_ffill[field_name] = current_value
+    player_ffill[identifier] = player_instance_ffill
+
+    return player_model

{sportsball-0.5.4 → sportsball-0.5.6}/sportsball/data/combined/combined_team_model.py

@@ -1,10 +1,12 @@
 """Combined team model."""
 
-# pylint: disable=too-many-locals,too-many-branches,too-many-statements
+# pylint: disable=too-many-locals,too-many-branches,too-many-statements,too-many-arguments
 import re
 import unicodedata
+from typing import Any
 
 from ..coach_model import CoachModel
+from ..field_type import FFILL_KEY
 from ..news_model import NewsModel
 from ..odds_model import OddsModel
 from ..player_model import PlayerModel
@@ -30,6 +32,9 @@ def create_combined_team_model(
     player_identity_map: dict[str, str],
     names: dict[str, str],
     coach_names: dict[str, str],
+    player_ffill: dict[str, dict[str, Any]],
+    team_ffill: dict[str, dict[str, Any]],
+    coach_ffill: dict[str, dict[str, Any]],
 ) -> TeamModel:
     """Create a team model by combining many team models."""
     location = None
@@ -99,12 +104,12 @@ def create_combined_team_model(
         if not is_null(team_model_end_dt):
             end_dt = team_model_end_dt
 
-    return TeamModel(
+    team_model = TeamModel(
         identifier=identifier,
         name=team_models[0].name,
         location=location,
         players=[  # pyright: ignore
-            create_combined_player_model(v, k) for k, v in players.items()
+            create_combined_player_model(v, k, player_ffill) for k, v in players.items()
         ],
         odds=[x[0] for x in odds.values()],
         points=points,
@@ -112,7 +117,22 @@ def create_combined_team_model(
         news=sorted(news.values(), key=lambda x: x.published),
         social=sorted(social.values(), key=lambda x: x.published),
         field_goals=field_goals,
-        coaches=[create_combined_coach_model(v, k) for k, v in coaches.items()],
+        coaches=[
+            create_combined_coach_model(v, k, coach_ffill) for k, v in coaches.items()
+        ],
         lbw=lbw,
         end_dt=end_dt,
     )
+
+    team_instance_ffill = team_ffill.get(identifier, {})
+    for field_name, field in team_model.model_fields.items():
+        extra = field.json_schema_extra or {}
+        if extra.get(FFILL_KEY, False):  # type: ignore
+            current_value = getattr(team_model, field_name)
+            if current_value is None:
+                setattr(team_model, field_name, team_instance_ffill.get(field_name))
+            else:
+                team_instance_ffill[field_name] = current_value
+    team_ffill[identifier] = team_instance_ffill
+
+    return team_model

{sportsball-0.5.4 → sportsball-0.5.6}/sportsball/data/field_type.py

@@ -3,6 +3,7 @@
 from enum import StrEnum
 
 TYPE_KEY = "type"
+FFILL_KEY = "ffill"
 
 
 class FieldType(StrEnum):

{sportsball-0.5.4 → sportsball-0.5.6}/sportsball/data/google/google_address_model.py

@@ -3232,6 +3232,51 @@ MADISON = SportsballGeocodeTuple(
     housenumber="",
     country="USA",
 )
+SAN_JOSE = SportsballGeocodeTuple(
+    city="San Jose",
+    state="CA",
+    postal="",
+    lat=37.336111,
+    lng=-121.890556,
+    housenumber="",
+    country="USA",
+)
+MIAMI_ARENA = SportsballGeocodeTuple(
+    city="Miami",
+    state="FL",
+    postal="",
+    lat=25.781111,
+    lng=-80.194722,
+    housenumber="721",
+    country="USA",
+)
+SILVIO_O_CONTE_FORUM = SportsballGeocodeTuple(
+    city="Chestnut Hill",
+    state="MA",
+    postal="",
+    lat=42.335,
+    lng=-71.1675,
+    housenumber="",
+    country="USA",
+)
+ROGERS_CENTER = SportsballGeocodeTuple(
+    city="Toronto",
+    state="ON",
+    postal="M5V 1J1",
+    lat=43.6415757,
+    lng=-79.3942991,
+    housenumber="580",
+    country="Canada",
+)
+BIRMINGHAM_JEFFERSON_CIVIC_CENTER = SportsballGeocodeTuple(
+    city="Birmingham",
+    state="AL",
+    postal="35203",
+    lat=33.524,
+    lng=-86.812,
+    housenumber="2100",
+    country="USA",
+)
 _CACHED_GEOCODES: dict[str, Any] = {
     "S.C.G. - Australia": SCG,
     "Victoria Park - Australia": SportsballGeocodeTuple(
@@ -3502,15 +3547,7 @@ _CACHED_GEOCODES: dict[str, Any] = {
         housenumber="580",
         country="Australia",
     ),
-    "Rogers Centre - Toronto - ON": SportsballGeocodeTuple(
-        city="Toronto",
-        state="ON",
-        postal="M5V 1J1",
-        lat=43.6415757,
-        lng=-79.3942991,
-        housenumber="580",
-        country="Canada",
-    ),
+    "Rogers Centre - Toronto - ON": ROGERS_CENTER,
     "Euroa - Australia": SportsballGeocodeTuple(
         city="Euroa",
         state="VIC",
@@ -6788,15 +6825,7 @@ _CACHED_GEOCODES: dict[str, Any] = {
         housenumber="2168",
         country="USA",
     ),
-    "Silvio O. Conte Forum, Chestnut Hill, Massachusetts": SportsballGeocodeTuple(
-        city="Chestnut Hill",
-        state="MA",
-        postal="",
-        lat=42.335,
-        lng=-71.1675,
-        housenumber="",
-        country="USA",
-    ),
+    "Silvio O. Conte Forum, Chestnut Hill, Massachusetts": SILVIO_O_CONTE_FORUM,
     "Millett Hall, Oxford, Ohio": SportsballGeocodeTuple(
         city="Oxford",
         state="OH",
@@ -7987,15 +8016,7 @@ _CACHED_GEOCODES: dict[str, Any] = {
     "Charlotte Coliseum": CHARLOTTE_COLISEUM,
     "Toyota Center, Houston, Texas": TOYOTA_CENTRE,
     "American Airlines Center, Dallas, Texas": AMERICAN_AIRLINES_ARENA,
-    "Miami Arena": SportsballGeocodeTuple(
-        city="Miami",
-        state="FL",
-        postal="",
-        lat=25.781111,
-        lng=-80.194722,
-        housenumber="721",
-        country="USA",
-    ),
+    "Miami Arena": MIAMI_ARENA,
     "Al McGuire Center, Milwaukee, Wisconsin": SportsballGeocodeTuple(
         city="Milwaukee",
         state="WI",
@@ -10350,15 +10371,7 @@ _CACHED_GEOCODES: dict[str, Any] = {
         housenumber="",
         country="USA",
     ),
-    "San Jose, California": SportsballGeocodeTuple(
-        city="San Jose",
-        state="CA",
-        postal="",
-        lat=37.336111,
-        lng=-121.890556,
-        housenumber="",
-        country="USA",
-    ),
+    "San Jose, California": SAN_JOSE,
     "Columbia, South Carolina": SportsballGeocodeTuple(
         city="Columbia",
         state="SC",
@@ -13497,15 +13510,7 @@ _CACHED_GEOCODES: dict[str, Any] = {
         housenumber="6400",
         country="USA",
     ),
-    "Birmingham-Jefferson Civic Center, Birmingham, Alabama": SportsballGeocodeTuple(
-        city="Birmingham",
-        state="AL",
-        postal="35203",
-        lat=33.524,
-        lng=-86.812,
-        housenumber="2100",
-        country="USA",
-    ),
+    "Birmingham-Jefferson Civic Center, Birmingham, Alabama": BIRMINGHAM_JEFFERSON_CIVIC_CENTER,
     "War Memorial Gymnasium, Maui, Hawaii": SportsballGeocodeTuple(
         city="Maui",
         state="HI",
@@ -15693,12 +15698,73 @@ _CACHED_GEOCODES: dict[str, Any] = {
         housenumber="",
         country="USA",
     ),
+    "Danville, Virginia, United States": SportsballGeocodeTuple(
+        city="Danville",
+        state="VA",
+        postal="",
+        lat=36.587222,
+        lng=-79.404444,
+        housenumber="",
+        country="USA",
+    ),
+    "Antioch, CA, USA": SportsballGeocodeTuple(
+        city="Antioch",
+        state="CA",
+        postal="",
+        lat=38.005,
+        lng=-121.805833,
+        housenumber="",
+        country="USA",
+    ),
+    "San Jose, CA, USA": SAN_JOSE,
+    "Miami Arena, Miami, Florida": MIAMI_ARENA,
+    "Silvio O. Conte Forum, Boston, Massachusetts": SILVIO_O_CONTE_FORUM,
+    "IND": SportsballGeocodeTuple(
+        city="",
+        state="",
+        postal="",
+        lat=21.0,
+        lng=78.0,
+        housenumber="",
+        country="India",
+    ),
+    "Tokyo Dome, Tokyo, Japan": SportsballGeocodeTuple(
+        city="Tokyo",
+        state="",
+        postal="",
+        lat=35.705556,
+        lng=139.751944,
+        housenumber="3",
+        country="Japan",
+    ),
+    "NH": SportsballGeocodeTuple(
+        city="",
+        state="",
+        postal="",
+        lat=52.0,
+        lng=6.0,
+        housenumber="",
+        country="Netherlands",
+    ),
+    "Federal Way, WA, USA": SportsballGeocodeTuple(
+        city="Federal Way",
+        state="WA",
+        postal="",
+        lat=47.313333,
+        lng=-122.372222,
+        housenumber="",
+        country="USA",
+    ),
+    "SkyDome, Toronto, Canada": ROGERS_CENTER,
+    "CoreStates Center, Philadelphia, Pennsylvania": SPECTRUM,
+    "BJCC Arena, Birmingham, Alabama": BIRMINGHAM_JEFFERSON_CIVIC_CENTER,
 }
 
 
 def _create_google_address_model(
     query: str, session: requests_cache.CachedSession, dt: datetime.datetime | None
 ) -> AddressModel:
+    query = query.replace("\n", "")
     g = _CACHED_GEOCODES.get(query)
     if g is None:
         logging.warning("Failed to find query: %s", query)

{sportsball-0.5.4 → sportsball-0.5.6}/sportsball/data/hkjc/hkjc/hkjc_hkjc_game_model.py

@@ -52,7 +52,11 @@ def _create_hkjc_hkjc_game_model(
 
     o = urlparse(url)
     query = urllib.parse.parse_qs(o.query)
-    dt = parse(query[RACE_DATE_QUERY_KEY][0])
+    try:
+        dt = parse(query[RACE_DATE_QUERY_KEY][0])
+    except KeyError:
+        logging.error(url)
+        return None
     game_number = int(query[RACE_NUMBER_QUERY_KEY][0])
     venue_code = query[RACE_COURSE_QUERY_KEY][0]
 
@@ -69,7 +73,7 @@ def _create_hkjc_hkjc_game_model(
         if count == 1:
             race_track = ""
             try:
-                race_track = df.iat[2, 2].split("-")[0].strip()
+                race_track = str(df.iat[2, 2]).split("-", maxsplit=1)[0].strip()
             except IndexError:
                 logging.error(url)
                 raise
@@ -86,7 +90,7 @@ def _create_hkjc_hkjc_game_model(
             for _, row in df.iterrows():
                 horse_name = None
                 try:
-                    horse_name = row["Horse"].split("(")[0].strip()
+                    horse_name = "(".join(row["Horse"].split("(")[:-1]).strip()
                 except KeyError:
                     logging.error(url)
                     continue
@@ -141,6 +145,8 @@ def _create_hkjc_hkjc_game_model(
                     lbw = 2.4 * 0.07
                 elif lbw_str == "+SH":
                     lbw = 2.4 * 0.2 * 0.5
+                elif lbw_str in {"HD+", "+HD"}:
+                    lbw = 2.4 * 0.2 * 1.5
                 elif lbw_str and lbw_str != "-" and lbw_str != "---":
                     if lbw is not None:
                         if "-" in lbw_str:
@@ -190,6 +196,8 @@ def _create_hkjc_hkjc_game_model(
                 if place is None:
                     continue
                 if horse_url is None:
+                    logging.error(url)
+                    logging.error(row)
                     raise ValueError("horse_url is null")
 
                 team_model = create_hkjc_hkjc_team_model(
@@ -210,18 +218,30 @@ def _create_hkjc_hkjc_game_model(
         elif count == 3:
             for _, row in df.iterrows():
                 row_df = row.to_frame()
-                pool = row_df.iat[0, 0].strip()
+                pool = str(row_df.iat[0, 0]).strip()
 
-                dividend_str = str(row_df.iat[2, 0]).strip().lower()
-                if dividend_str == "details":
-                    continue
-                if dividend_str == "refund":
-                    continue
-                if dividend_str == "not win":
-                    continue
-                if dividend_str == "detail":
+                try:
+                    dividend_str = str(row_df.iat[2, 0]).strip().lower()
+                    if dividend_str == "details":
+                        continue
+                    if dividend_str == "refund":
+                        continue
+                    if dividend_str == "not win":
+                        continue
+                    if dividend_str == "detail":
+                        continue
+                    dividend_str = (
+                        dividend_str.split("/", maxsplit=1)[0]
+                        .replace(",", "")
+                        .replace("$", "")
+                        .replace("(", "")
+                        .replace(")", "")
+                    )
+                    if dividend_str == "win":
+                        continue
+                except IndexError:
+                    logging.error(url)
                     continue
-                dividend_str = dividend_str.split("/", maxsplit=1)[0].replace(",", "")
                 dividend = float(dividend_str)
 
                 combination = None

{sportsball-0.5.4 → sportsball-0.5.6}/sportsball/data/league_model.py

@@ -41,6 +41,8 @@ def _normalize_tz(df: pd.DataFrame) -> pd.DataFrame:
     tqdm.tqdm.pandas(desc="Timezone Conversions")
 
     # Check each row to see if they have the correct timezone
+    dt_cols = set()
+
     def apply_tz(row: pd.Series) -> pd.Series:
         if tz_column not in row:
             return row
@@ -53,6 +55,7 @@ def _normalize_tz(df: pd.DataFrame) -> pd.DataFrame:
         }
         datetime_cols.add(GAME_DT_COLUMN)
         for col in datetime_cols:
+            dt_cols.add(col)
             dt = row[col]  # type: ignore
             if isinstance(dt, (datetime.date, datetime.datetime)):
                 dt = pd.to_datetime(dt)
@@ -65,6 +68,9 @@ def _normalize_tz(df: pd.DataFrame) -> pd.DataFrame:
 
         return row
 
+    for dt_col in dt_cols:
+        df[dt_col] = pd.to_datetime(df[dt_col])
+
     return df.progress_apply(apply_tz, axis=1)  # type: ignore
 
 

{sportsball-0.5.4 → sportsball-0.5.6}/sportsball/data/nfl/position.py

@@ -13,3 +13,4 @@ class Position(StrEnum):
     OFFENSIVE_LINESMAN = "OL"
     SAFETY = "S"
     WIDE_RECEIVER = "WR"
+    GUARD = "G"