sportsball 0.3.30__tar.gz → 0.3.32__tar.gz

{sportsball-0.3.30/sportsball.egg-info → sportsball-0.3.32}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: sportsball
-Version: 0.3.30
+Version: 0.3.32
 Summary: A library for pulling in and normalising sports stats.
 Home-page: https://github.com/8W9aG/sportsball
 Author: Will Sackfield

{sportsball-0.3.30 → sportsball-0.3.32}/setup.py

@@ -26,7 +26,7 @@ def install_requires() -> typing.List[str]:
 
 setup(
     name='sportsball',
-    version='0.3.30',
+    version='0.3.32',
     description='A library for pulling in and normalising sports stats.',
     long_description=long_description,
     long_description_content_type='text/markdown',

{sportsball-0.3.30 → sportsball-0.3.32}/sportsball/__init__.py

@@ -1,3 +1,3 @@
 """The main module for sportsball."""
 
-__VERSION__ = "0.3.30"
+__VERSION__ = "0.3.32"

{sportsball-0.3.30 → sportsball-0.3.32}/sportsball/data/google/google_address_model.py

@@ -1441,6 +1441,15 @@ TED_CONSTANT_CONVOCATION_CENTER = SportsballGeocodeTuple(
     housenumber="4320",
     country="USA",
 )
+STONY_BROOK_ARENA = SportsballGeocodeTuple(
+    city="Stony Brook",
+    state="NY",
+    postal="11790",
+    lat=40.91741,
+    lng=-73.125955,
+    housenumber="100",
+    country="USA",
+)
 _CACHED_GEOCODES: dict[str, Any] = {
     "S.C.G. - Australia": SCG,
     "Victoria Park - Australia": SportsballGeocodeTuple(
@@ -5554,15 +5563,7 @@ _CACHED_GEOCODES: dict[str, Any] = {
         housenumber="4000",
         country="USA",
     ),
-    "Island Federal Credit Union, Stony Brook, New York": SportsballGeocodeTuple(
-        city="Stony Brook",
-        state="NY",
-        postal="11790",
-        lat=40.91741,
-        lng=-73.125955,
-        housenumber="100",
-        country="USA",
-    ),
+    "Island Federal Credit Union, Stony Brook, New York": STONY_BROOK_ARENA,
     "Cajundome, Lafayette, Louisiana": SportsballGeocodeTuple(
         city="Lafayette",
         state="LA",
@@ -7588,15 +7589,7 @@ _CACHED_GEOCODES: dict[str, Any] = {
         housenumber="11200",
         country="USA",
     ),
-    "Island Federal Credit Union Arena, Stony Brook, New York": SportsballGeocodeTuple(
-        city="Stony Brook",
-        state="NY",
-        postal="11790",
-        lat=40.91741,
-        lng=-73.125955,
-        housenumber="100",
-        country="USA",
-    ),
+    "Island Federal Credit Union Arena, Stony Brook, New York": STONY_BROOK_ARENA,
     "Phoenix, Arizona": SportsballGeocodeTuple(
         city="Phoenix",
         state="AZ",
@@ -7834,6 +7827,7 @@ _CACHED_GEOCODES: dict[str, Any] = {
         housenumber="",
         country="USA",
     ),
+    "Island Federal Credit Union Arena, New York, NY, USA": STONY_BROOK_ARENA,
 }
 
 

{sportsball-0.3.30 → sportsball-0.3.32}/sportsball/data/oddsportal/oddsportal_league_model.py

@@ -13,6 +13,7 @@ import requests_cache
 import tqdm
 from bs4 import BeautifulSoup
 
+from ...proxy_session import X_NO_WAYBACK
 from ..game_model import GameModel
 from ..league import League
 from ..league_model import LeagueModel
@@ -30,6 +31,17 @@ USA = "usa"
 NCAA = "ncaa"
 
 
+def _find_ids(text: str) -> tuple[str, str]:
+    sentinel = "var pageOutrightsVar = '"
+    sanitised_text = text[text.find(sentinel) + len(sentinel) :]
+    sanitised_text = sanitised_text[: sanitised_text.find("'")]
+    try:
+        page_outrights = json.loads(sanitised_text)
+    except json.decoder.JSONDecodeError as exc:
+        raise exc
+    return str(page_outrights["sid"]), page_outrights["id"]
+
+
 def _process_results_pages(
     url: str,
     session: requests_cache.CachedSession,
@@ -39,11 +51,7 @@ def _process_results_pages(
     response: requests.Response,
 ) -> Iterator[GameModel]:
     # Fetch first page
-    sanitised_text = response.text[response.text.find("var pageOutrightsVar = '") :]
-    sanitised_text = sanitised_text[: sanitised_text.find("'")]
-    page_outrights = json.loads(sanitised_text)
-    sports_id = page_outrights["sid"]
-    oddsportal_id = page_outrights["id"]
+    sports_id, oddsportal_id = _find_ids(response.text)
 
     current_page = 1
     total_pages = None
@@ -53,14 +61,17 @@ def _process_results_pages(
         dat_url = f"https://www.oddsportal.com/ajax-sport-country-tournament-archive_/{sports_id}/{oddsportal_id}/X134529032X0X0X0X0X0X0X0X0X0X0X0X0X0X0X0X0X0X512X32X0X0X0X0X0X0X131072X0X2048/1/-5/page/{current_page}//"
         parsed_data = fetch_data(dat_url, session, url, soup)
         d = parsed_data["d"]
-        for row in d["rows"]:
+        if d.get("total") == 0:
+            return
+        for row in d.get("rows", []):
             game_model = create_oddsportal_game_model(
                 session, urllib.parse.urljoin(url, row["url"]), league, False
             )
             pbar.update(1)
             pbar.set_description(f"OddsPortal {game_model.dt}")
             yield game_model
-        total_pages = d["pagination"]["pages"]
+        pagination = d["pagination"]
+        total_pages = pagination["pages"]
         current_page += 1
 
 
@@ -118,17 +129,13 @@ class OddsPortalLeagueModel(LeagueModel):
             url = queued_urls.pop()
             if url in seen_urls:
                 continue
-            logging.debug("processing url: %s", url)
             seen_urls.add(url)
 
-            if url.endswith(standard_suffix):
-                with self.session.cache_disabled():
-                    response = self.session.get(url)
-            else:
-                response = self.session.get(url)
+            with self.session.cache_disabled():
+                response = self.session.get(url, headers={X_NO_WAYBACK: "1"})
             response.raise_for_status()
 
-            soup = BeautifulSoup(response.text, "xml")
+            soup = BeautifulSoup(response.text, "html.parser")
 
             # Find next URLs
             for option in soup.find_all("option"):

{sportsball-0.3.30 → sportsball-0.3.32/sportsball.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: sportsball
-Version: 0.3.30
+Version: 0.3.32
 Summary: A library for pulling in and normalising sports stats.
 Home-page: https://github.com/8W9aG/sportsball
 Author: Will Sackfield

{sportsball-0.3.30 → sportsball-0.3.32}/sportsball.egg-info/SOURCES.txt

@@ -191,6 +191,7 @@ tests/data/nfl/sportsdb/__init__.py
 tests/data/nfl/sportsdb/nfl_sportsdb_league_model_test.py
 tests/data/oddsportal/__init__.py
 tests/data/oddsportal/oddsportal_game_model_test.py
+tests/data/oddsportal/oddsportal_league_model_test.py
 tests/data/sportsdb/__init__.py
 tests/data/sportsdb/sportsdb_game_model_test.py
 tests/data/sportsdb/sportsdb_league_model_test.py

sportsball-0.3.32/tests/data/oddsportal/oddsportal_league_model_test.py

@@ -0,0 +1,19 @@
+"""Tests for the oddsportal league model class."""
+import os
+import unittest
+
+import requests_cache
+from sportsball.data.oddsportal.oddsportal_league_model import _find_ids
+
+
+class TestOddsPortalLeagueModel(unittest.TestCase):
+
+    def setUp(self):
+        self.session = requests_cache.CachedSession(backend="memory")
+        self.dir = os.path.dirname(__file__)
+
+    def test_current_html_ids(self):
+        with open(os.path.join(self.dir, "results.html"), "r") as handle:
+            sports_id, oddsportal_id = _find_ids(handle.read())
+            self.assertEqual(sports_id, "18")
+            self.assertEqual(oddsportal_id, "6y3HOkJ7")