e

M3U8/fetch.py

@@ -54,7 +54,7 @@ async def main() -> None:
         asyncio.create_task(streamfree.scrape(network.client)),
         asyncio.create_task(strmd.scrape(network.client)),
         asyncio.create_task(tvpass.scrape(network.client)),
-        #asyncio.create_task(volo.scrape(network.client)),
+        # asyncio.create_task(volo.scrape(network.client)),
         asyncio.create_task(watchfooty.scrape(network.client)),
     ]
 

M3U8/scrapers/roxie.py

@@ -18,6 +18,15 @@ HTML_CACHE = Cache("roxie-html.json", exp=19_800)
 
 BASE_URL = "https://roxiestreams.live"
 
+valid_sports = {
+    "fighting": "Fighting",
+    "mlb": "MLB",
+    "motorsports": "Racing",
+    "nba": "NBA",
+    "nfl": "American Football",
+    "soccer": "Soccer",
+}
+
 
 async def process_event(
     client: httpx.AsyncClient,
@@ -47,6 +56,7 @@ async def process_event(
 async def refresh_html_cache(
     client: httpx.AsyncClient,
     url: str,
+    sport: str,
     now_ts: float,
 ) -> dict[str, str | float]:
 
@@ -60,11 +70,6 @@ async def refresh_html_cache(
 
     soup = HTMLParser(r.text)
 
-    pattern = re.compile(r'Upcoming ([^"]*) Events', re.IGNORECASE)
-
-    h2_title = soup.css_first("h2").text(strip=True)
-
-    sport = sport_name[1] if (sport_name := pattern.search(h2_title)) else "Event"
     events = {}
 
     for row in soup.css("table#eventsTable tbody tr"):
@@ -85,10 +90,12 @@ async def refresh_html_cache(
 
         event_dt = Time.from_str(data_start, timezone="PST")
 
-        key = f"[{sport}] {event} (ROXIE)"
+        event_sport = valid_sports[sport]
+
+        key = f"[{event_sport}] {event} (ROXIE)"
 
         events[key] = {
-            "sport": sport,
+            "sport": event_sport,
             "event": event,
             "link": href,
             "event_ts": event_dt.timestamp(),
@@ -100,7 +107,7 @@ async def refresh_html_cache(
 
 async def get_events(
     client: httpx.AsyncClient,
-    sport_urls: list[str],
+    sport_urls: dict[str, str],
     cached_keys: set[str],
 ) -> list[dict[str, str]]:
 
@@ -111,9 +118,10 @@ async def get_events(
             refresh_html_cache(
                 client,
                 url,
+                sport,
                 now.timestamp(),
             )
-            for url in sport_urls
+            for sport, url in sport_urls.items()
         ]
 
         results = await asyncio.gather(*tasks)
@@ -148,10 +156,7 @@ async def scrape(client: httpx.AsyncClient) -> None:
 
     log.info(f'Scraping from "{BASE_URL}"')
 
-    sport_urls = [
-        urljoin(BASE_URL, sport)
-        for sport in ["fighting", "mlb", "motorsports", "nba", "nfl", "soccer"]
-    ]
+    sport_urls = {sport: urljoin(BASE_URL, sport) for sport in valid_sports}
 
     events = await get_events(
         client,