e

M3U8/scrape/livetvsx.py

@@ -5,20 +5,16 @@ import ssl
 import xml.etree.ElementTree as ET
 from datetime import datetime, timedelta
 from pathlib import Path
-from typing import Any
 
 import httpx
 from playwright.async_api import Request, async_playwright
 
-from .tvpass import TZ, logos
-from .utils.logger import get_logger
+from .utils import LOGOS, TZ, get_logger, safe_process_event
 
 log = get_logger(__name__)
 
 urls: dict[str, dict[str, str]] = {}
 
-tvp_sports = set(logos.keys())
-
 BASE_URL = "https://cdn.livetv861.me/rss/upcoming_en.xml"
 
 CERT_BUNDL_URLS = [
@@ -32,23 +28,7 @@ CERT_FILE = Path(__file__).parent / "utils" / "cached-ca.pem"
 
 CACHE_FILE = Path(__file__).parent / "caches" / "livetvsx.json"
 
-
-async def safe_process_event(fn, url_num: int, timeout=20) -> Any | None:
-    task = asyncio.create_task(fn())
-
-    try:
-        return await asyncio.wait_for(task, timeout=timeout)
-    except asyncio.TimeoutError:
-        log.warning(f"URL {url_num}) Timed out after {timeout}s, skipping event")
-
-        task.cancel()
-
-        try:
-            await task
-        except asyncio.CancelledError:
-            pass
-        except Exception as e:
-            log.debug(f"URL {url_num}) Ignore exception after timeout: {e}")
+exist_sprts = set(LOGOS.keys())
 
 
 async def write_to_cert(client: httpx.AsyncClient, url: str, cert: Path) -> None:
@@ -86,15 +66,13 @@ async def get_cert(client: httpx.AsyncClient) -> ssl.SSLContext:
 
 def load_cache() -> dict[str, dict[str, str | str]]:
     try:
-        data = json.loads(CACHE_FILE.read_text(encoding="utf-8"))
+        data: dict = json.loads(CACHE_FILE.read_text(encoding="utf-8"))
 
         now = datetime.now(TZ).timestamp()
 
-        return {
-            k: v
-            for k, v in data.items()
-            if now - v.get("timestamp", 0) < timedelta(hours=4).total_seconds()
-        }
+        age: float = now - data.get("timestamp", 0)
+
+        return {k: v for k, v in data.items() if age < 14400}  # 4 hours
     except (FileNotFoundError, json.JSONDecodeError):
         return {}
 
@@ -163,15 +141,17 @@ async def parse_feed(
                     elem.clear()
                     continue
 
-                elif not tvp_sports & {sport, event}:
-                    events.append(
-                        {
-                            "sport": sport,
-                            "event": event,
-                            "title": title,
-                            "link": link,
-                        }
-                    )
+                if not exist_sprts & {sport, event}:
+                    continue
+
+                events.append(
+                    {
+                        "sport": sport,
+                        "event": event,
+                        "title": title,
+                        "link": link,
+                    }
+                )
 
             elem.clear()
 
@@ -288,7 +268,7 @@ async def process_event(url: str, url_num: int) -> str | None:
 
             log.warning(f"URL {url_num}) No m3u8 captured in popup or inline playback.")
             return
-        except Exception as e:
+        except Exception:
             try:
                 ev_page.remove_listener("request", capture_req)
 
@@ -310,10 +290,9 @@ async def main(client: httpx.AsyncClient) -> None:
     cert = await get_cert(client)
 
     cached_urls = load_cache()
-    cached_keys = set(cached_urls.keys())
     cached_count = len(cached_urls)
 
-    events = await parse_feed(BASE_URL, cert, cached_keys)
+    events = await parse_feed(BASE_URL, cert, set(cached_urls.keys()))
 
     log.info(f"Processing {len(events)} URLs")
 
@@ -328,13 +307,15 @@ async def main(client: httpx.AsyncClient) -> None:
         key = f"[{sport}: {event}] {title}"
 
         url = await safe_process_event(
-            lambda: process_event(link, url_num=num), url_num=num
+            lambda: process_event(link, url_num=num),
+            url_num=num,
+            log=log,
         )
 
         if url:
             entry = {
                 "url": url,
-                "logo": logos.get(
+                "logo": LOGOS.get(
                     sport,
                     "https://i.gyazo.com/ec27417a9644ae517196494afa72d2b9.png",
                 ),
@@ -349,4 +330,4 @@ async def main(client: httpx.AsyncClient) -> None:
 
     log.info(f"Cached {cached_count} event(s)")
 
-    log.info(f"Collected {new_count} new event(s)")
+    log.info(f"Collected {new_count} event(s)")