update M3U8

- edit scraping for streamsgate.py

M3U8/fetch.py

@@ -69,7 +69,7 @@ async def main() -> None:
                 asyncio.create_task(sportzone.scrape(xtrnl_brwsr)),
                 asyncio.create_task(streamcenter.scrape(hdl_brwsr)),
                 # asyncio.create_task(streamhub.scrape(xtrnl_brwsr)),
-                # asyncio.create_task(streamsgate.scrape(xtrnl_brwsr)),
+                asyncio.create_task(streamsgate.scrape(xtrnl_brwsr)),
                 asyncio.create_task(timstreams.scrape(xtrnl_brwsr)),
             ]
 

M3U8/scrapers/streamsgate.py

@@ -1,10 +1,10 @@
-import json
+import asyncio
-import re
 from functools import partial
+from itertools import chain
 from typing import Any
+from urllib.parse import urljoin
 
 from playwright.async_api import Browser
-from selectolax.parser import HTMLParser
 
 from .utils import Cache, Time, get_logger, leagues, network
 
@@ -16,95 +16,101 @@ TAG = "STRMSGATE"
 
 CACHE_FILE = Cache(TAG, exp=10_800)
 
-BASE_URL = "https://streamingon.org/index.php"
+API_FILE = Cache(f"{TAG}-api", exp=19_800)
+
+BASE_URL = "https://streamingon.org"
+
+SPORT_URLS = [
+    urljoin(BASE_URL, f"data/{sport}.json")
+    for sport in [
+        # "cfb",
+        "mlb",
+        "nba",
+        # "nfl",
+        "nhl",
+        "soccer",
+        "ufc",
+    ]
+]
+
+
+def get_event(t1: str, t2: str) -> str:
+    match t1:
+        case "RED ZONE":
+            return "NFL RedZone"
+
+        case "TBD":
+            return "TBD"
+
+        case _:
+            return f"{t1.strip()} vs {t2.strip()}"
+
+
+async def refresh_api_cache(now_ts: float) -> list[dict[str, Any]]:
+    tasks = [network.request(url, log=log) for url in SPORT_URLS]
+
+    results = await asyncio.gather(*tasks)
+
+    if not (data := [*chain.from_iterable(r.json() for r in results if r)]):
+        return [{"timestamp": now_ts}]
+
+    for ev in data:
+        ev["ts"] = ev.pop("timestamp")
+
+    data[-1]["timestamp"] = now_ts
+
+    return data
 
 
 async def get_events(cached_keys: list[str]) -> list[dict[str, str]]:
     now = Time.clean(Time.now())
 
+    if not (api_data := API_FILE.load(per_entry=False, index=-1)):
+        log.info("Refreshing API cache")
+
+        api_data = await refresh_api_cache(now.timestamp())
+
+        API_FILE.write(api_data)
+
     events = []
 
-    if not (
+    start_dt = now.delta(hours=-1)
-        html_data := await network.request(
+    end_dt = now.delta(minutes=5)
-            BASE_URL,
-            params={
-                "sport": "all",
-                "league": "all",
-                "sort": "time",
-                "stream": "available",
-                "day": "all",
-            },
-            log=log,
-        )
-    ):
-        return events
 
-    link_data_ptrn = re.compile(r"var\s+linkData\s+=\s+({.*?});", re.I | re.S)
+    for stream_group in api_data:
+        date = stream_group.get("time")
 
-    if not (match := link_data_ptrn.search(html_data.text)):
+        sport = stream_group.get("league")
-        log.warning("No `linkData` variable found.")
-        return events
 
-    link_data: dict[str, dict[str, Any]] = json.loads(match[1])
+        t1, t2 = stream_group.get("away"), stream_group.get("home")
 
-    start_dt = now.delta(minutes=-30)
+        event = get_event(t1, t2)
-    end_dt = now.delta(minutes=30)
 
-    soup = HTMLParser(html_data.content)
+        if not (date and sport):
-
-    for body in soup.css(".sport-body"):
-        if not (date_elem := body.css_first(".date-label")):
             continue
 
-        event_date = date_elem.text(strip=True)
+        if f"[{sport}] {event} ({TAG})" in cached_keys:
+            continue
 
-        for card in soup.css(".game-card"):
+        event_dt = Time.from_str(date, timezone="UTC")
-            if not (event_id := card.attributes.get("data-id")):
-                continue
 
-            if not (league_elem := card.css_first(".card-league")):
+        if not start_dt <= event_dt <= end_dt:
-                continue
+            continue
 
-            if not (teams := card.css(".card-teams .card-team-name")):
+        if not (streams := stream_group.get("streams")):
-                continue
+            continue
 
-            if not (time_elem := card.css_first(".card-time")):
+        if not (url := streams[0].get("url")):
-                continue
+            continue
 
-            event_dt = Time.from_str(
+        events.append(
-                f"{event_date} {time_elem.text(strip=True)}",
+            {
-                timezone="CET",
+                "sport": sport,
-            )
+                "event": event,
-
+                "link": url,
-            if not start_dt <= event_dt <= end_dt:
+                "timestamp": event_dt.timestamp(),
-                continue
+            }
-
+        )
-            sport = league_elem.text(strip=True)
-
-            team_1, team_2 = (team.text(strip=True) for team in teams)
-
-            event_name = f"{team_2} vs {team_1}"
-
-            if f"[{sport}] {event_name} ({TAG})" in cached_keys:
-                continue
-
-            if not (event_info := link_data.get(event_id)):
-                continue
-
-            if not (stream_links := event_info.get("streamLinks")):
-                continue
-
-            if not (url := stream_links[0].get("url")):
-                continue
-
-            events.append(
-                {
-                    "sport": sport,
-                    "event": event_name,
-                    "link": url,
-                    "timestamp": now.timestamp(),
-                }
-            )
 
     return events
 
@@ -167,6 +173,8 @@ async def scrape(browser: Browser) -> None:
                     if url:
                         valid_count += 1
 
+                        entry["url"] = url.split("&e")[0]
+
                         urls[key] = entry
 
         log.info(f"Collected and cached {valid_count - cached_count} new event(s)")

readme.md

@@ -1,13 +1,14 @@
-## Base Log @ 2026-03-20 09:04 UTC
+## Base Log @ 2026-03-21 08:55 UTC
 
-### ✅ Working Streams: 158<br>❌ Dead Streams: 4
+### ✅ Working Streams: 157<br>❌ Dead Streams: 5
 
 | Channel | Error (Code) | Link |
 | ------- | ------------ | ---- |
 | Altitude Sports | HTTP Error (403) | `http://mytvstream.net:8080/live/30550113/30550113/79545.m3u8` |
-| Hallmark Family | HTTP Error (000) | `http://mytvstream.net:8080/live/30550113/30550113/9307.m3u8` |
+| Hallmark Family | HTTP Error (403) | `http://mytvstream.net:8080/live/30550113/30550113/9307.m3u8` |
 | NBC Sports California | HTTP Error (403) | `http://mytvstream.net:8080/live/30550113/30550113/20940.m3u8` |
 | NHL Network | HTTP Error (000) | `http://mytvstream.net:8080/live/30550113/30550113/20179.m3u8` |
+| Pop TV | HTTP Error (403) | `http://mytvstream.net:8080/live/30550113/30550113/20976.m3u8` |
 ---
 #### Base Channels URL
 ```