e

M3U8/scrape/livetvsx.py

@@ -4,12 +4,21 @@ import json
 import ssl
 import xml.etree.ElementTree as ET
 from datetime import datetime, timedelta
+from functools import partial
 from pathlib import Path
 
 import httpx
-from playwright.async_api import Request, async_playwright
+from playwright.async_api import async_playwright
 
-from .utils import LOGOS, TZ, get_logger, now, safe_process_event
+from .utils import (
+    LOGOS,
+    TZ,
+    capture_req,
+    get_logger,
+    load_ts_cache,
+    now,
+    safe_process_event,
+)
 
 log = get_logger(__name__)
 
@@ -62,21 +71,6 @@ async def get_cert(client: httpx.AsyncClient) -> ssl.SSLContext:
     return ssl.create_default_context(cafile=CERT_FILE)
 
 
-def load_cache() -> dict[str, dict[str, str | float]]:
-    try:
-        data: dict[str, dict[str, str | float]] = json.loads(
-            CACHE_FILE.read_text(encoding="utf-8")
-        )
-
-        return {
-            k: v
-            for k, v in data.items()
-            if now.timestamp() - data[k].get("timestamp", 0) < 14400  # 4 hours
-        }
-    except (FileNotFoundError, json.JSONDecodeError):
-        return {}
-
-
 async def fetch_xml_stream(url: str, ssl_ctx: ssl.SSLContext) -> io.BytesIO:
     buffer = io.BytesIO()
 
@@ -104,44 +98,38 @@ async def process_event(url: str, url_num: int) -> str | None:
         context = await browser.new_context(
             ignore_https_errors=True  # website doesn't send valid certs
         )
-        ev_page = await context.new_page()
+
+        page = await context.new_page()
 
         captured: list[str] = []
 
         got_one = asyncio.Event()
 
-        def capture_req(req: Request) -> None:
-            if (
-                ".m3u8" in req.url
-                and "amazonaws" not in req.url
-                and "knitcdn" not in req.url
-            ):
-                captured.append(req.url)
-                got_one.set()
+        handler = partial(capture_req, captured=captured, got_one=got_one)
 
         popup = None
 
         try:
-            await ev_page.goto(
+            await page.goto(
                 url,
                 wait_until="domcontentloaded",
                 timeout=10_000,
             )
 
-            btn = await ev_page.query_selector(".lnkhdr > tbody > tr > td:nth-child(2)")
+            btn = await page.query_selector(".lnkhdr > tbody > tr > td:nth-child(2)")
 
             if btn:
                 try:
                     await btn.click()
 
-                    await ev_page.wait_for_timeout(500)
+                    await page.wait_for_timeout(500)
                 except Exception as e:
                     log.debug(f"URL {url_num}) Failed to click Browser Links tab: {e}")
                     return
             else:
                 log.warning(f"URL {url_num}) Browser Links tab not found")
 
-            link_img = await ev_page.query_selector(
+            link_img = await page.query_selector(
                 "tr:nth-child(2) > td:nth-child(1) td:nth-child(6) img"
             )
 
@@ -149,10 +137,10 @@ async def process_event(url: str, url_num: int) -> str | None:
                 log.warning(f"URL {url_num}) No browser link to click.")
                 return
 
-            ev_page.on("request", capture_req)
+            page.on("request", handler)
 
             try:
-                async with ev_page.expect_popup(timeout=5_000) as popup_info:
+                async with page.expect_popup(timeout=5_000) as popup_info:
                     try:
                         await link_img.click()
                     except Exception as e:
@@ -162,22 +150,20 @@ async def process_event(url: str, url_num: int) -> str | None:
 
                 popup = await popup_info.value
 
-                popup.on("request", capture_req)
+                popup.on("request", handler)
             except Exception:
 
                 try:
                     await link_img.click()
                 except Exception as e:
                     log.debug(f"URL {url_num}) Fallback click failed: {e}")
-                    return
 
             wait_task = asyncio.create_task(got_one.wait())
 
             try:
                 await asyncio.wait_for(wait_task, timeout=1.5e1)
-
             except asyncio.TimeoutError:
-                log.warning(f"URL {url_num}) Timed out waiting for m3u8.")
+                log.warning(f"URL {url_num}) Timed out waiting for M3U8.")
                 return
 
             finally:
@@ -189,32 +175,32 @@ async def process_event(url: str, url_num: int) -> str | None:
                     except asyncio.CancelledError:
                         pass
 
-            ev_page.remove_listener("request", capture_req)
+            page.remove_listener("request", handler)
 
             if popup:
-                popup.remove_listener("request", capture_req)
+                popup.remove_listener("request", handler)
 
                 await popup.close()
 
-            await ev_page.close()
+            await page.close()
 
             if captured:
                 log.info(f"URL {url_num}) Captured M3U8")
 
                 return captured[-1]
 
-            log.warning(f"URL {url_num}) No m3u8 captured in popup or inline playback.")
+            log.warning(f"URL {url_num}) No M3U8 captured in popup or inline playback.")
             return
         except Exception:
             try:
-                ev_page.remove_listener("request", capture_req)
+                page.remove_listener("request", handler)
 
                 if popup:
-                    popup.remove_listener("request", capture_req)
+                    popup.remove_listener("request", handler)
 
                     await popup.close()
 
-                await ev_page.close()
+                await page.close()
             except Exception:
                 pass
 
@@ -283,7 +269,7 @@ async def main(client: httpx.AsyncClient) -> None:
 
     cert = await get_cert(client)
 
-    cached_urls = load_cache()
+    cached_urls = load_ts_cache(CACHE_FILE, 14400)
     cached_count = len(cached_urls)
 
     log.info(f"Collected {cached_count} event(s) from cache")