Fix ROOT CAUSE: segments are PNG-wrapped TS - proxy strips 188-byte PNG header, rewrites m3u8 segment URLs through server

main.py

@@ -20,8 +20,6 @@ _cache_ttl_live = 60
 _cache_ttl_yt = 1800
 PRIORITY_LEAGUES = ["Ngoại Hạng Anh","FA Cup","Champions League","LaLiga","Copa del Rey","Serie A","Bundesliga","Ligue 1","V-League"]
 LEAGUE_IDS = {"nha":27110,"laliga":27233,"seriea":27044,"bundesliga":26891,"ligue1":27212}
-
-# ===== HIGHLIGHT LEAGUE CONFIG =====
 HL_LEAGUES = {
     "premier-league": {"path": "anh/premier-league", "name": "Premier League", "emoji": "🏴󠁧󠁢󠁥󠁮󠁧󠁿"},
     "fa-cup": {"path": "anh/fa-cup", "name": "FA Cup", "emoji": "🏆"},
@@ -32,7 +30,6 @@ HL_LEAGUES = {
     "europa-league": {"path": "cup-chau-au/uefa-europa-league", "name": "Europa League", "emoji": "🟠"},
     "world-cup": {"path": "the-gioi/world-cup-qualifiers", "name": "World Cup 2026", "emoji": "🌍"},
 }
-
 def _cached(key, fn, ttl=None):
     now=time.time();t=ttl or _cache_ttl
     if key in _cache and now-_cache[key]["t"]<t:return _cache[key]["d"]
@@ -74,22 +71,52 @@ def _parse_match_from_li(li, status_type="live"):
     league=league_el.get_text(strip=True) if league_el else ""
     return{"home":home_el.get_text(strip=True),"away":away_el.get_text(strip=True),"score":score or"VS","minute":minute,"league":league,"time":time_el.get_text(strip=True) if time_el else "","event_id":event_id,"home_logo":home_logo.get("src","") if home_logo else "","away_logo":away_logo.get("src","") if away_logo else "","status":status_type}
 
-# ===== VIDEO PROXY (fix CORS) =====
+# ===== VIDEO PROXY (fix CORS + PNG-wrapped TS segments) =====
 @app.get("/api/proxy/m3u8")
 def proxy_m3u8(url: str = Query(...)):
-    """Proxy m3u8 manifest to add CORS headers. Segments have CORS already."""
+    """Proxy m3u8 manifest: add CORS + rewrite segment URLs to go through /api/proxy/seg"""
     try:
         r = requests.get(url, headers=HEADERS, timeout=15)
         if r.status_code != 200:
             return Response(status_code=502, content="upstream error")
+        # Rewrite segment URLs to proxy through our server (strips PNG header)
+        lines = r.text.strip().split('\n')
+        rewritten = []
+        for line in lines:
+            if line.startswith('#') or not line.strip():
+                rewritten.append(line)
+            else:
+                # This is a segment URL - rewrite to our proxy
+                rewritten.append("/api/proxy/seg?url=" + quote(line.strip(), safe=""))
+        content = '\n'.join(rewritten)
         return Response(
-            content=r.content,
+            content=content.encode('utf-8'),
             media_type="application/vnd.apple.mpegurl",
             headers={"Access-Control-Allow-Origin": "*", "Cache-Control": "public, max-age=300"}
         )
     except:
         return Response(status_code=502, content="proxy error")
 
+@app.get("/api/proxy/seg")
+def proxy_segment(url: str = Query(...)):
+    """Proxy HLS segment: strips 188-byte PNG header wrapper, returns clean MPEG-TS."""
+    try:
+        r = requests.get(url, headers=HEADERS, timeout=30)
+        if r.status_code != 200:
+            return Response(status_code=502, content="upstream error")
+        data = r.content
+        # TikTok CDN wraps TS in PNG: first 188 bytes = 1x1 PNG, rest = MPEG-TS
+        # Detect and strip: check if starts with PNG magic + byte 188 is 0x47 (TS sync)
+        if len(data) > 188 and data[0:4] == b'\x89PNG' and data[188] == 0x47:
+            data = data[188:]
+        return Response(
+            content=data,
+            media_type="video/mp2t",
+            headers={"Access-Control-Allow-Origin": "*", "Cache-Control": "public, max-age=3600"}
+        )
+    except:
+        return Response(status_code=502, content="proxy error")
+
 @app.get("/api/proxy/video")
 def proxy_video(url: str = Query(...), request: Request = None):
     """Proxy MP4/video for sources that block cross-origin (BDP). Supports range requests."""
@@ -117,7 +144,6 @@ def proxy_video(url: str = Query(...), request: Request = None):
 
 # ===== XEMLAIBONGDA HIGHLIGHTS =====
 def _scrape_xemlaibongda_page(page_path):
-    """Scrape highlights from a specific xemlaibongda page (league or homepage)."""
     try:
         url = f"https://xemlaibongda.top/{page_path}" if page_path else "https://xemlaibongda.top/"
         r=requests.get(url,headers=HEADERS,timeout=15)
@@ -147,16 +173,13 @@ def _scrape_xemlaibongda_page(page_path):
     except:return[]
 
 def scrape_xemlaibongda():
-    """Get all highlight videos from homepage."""
     return _scrape_xemlaibongda_page("")
 
 def scrape_highlights_by_league(league_key):
-    """Get highlights for a specific league."""
     if league_key not in HL_LEAGUES:return[]
     return _scrape_xemlaibongda_page(HL_LEAGUES[league_key]["path"])
 
 def scrape_all_league_highlights():
-    """Fetch highlights for all leagues in parallel."""
     results = {}
     def _fetch(key):
         return key, scrape_highlights_by_league(key)
@@ -165,13 +188,11 @@ def scrape_all_league_highlights():
         for f in as_completed(futs):
             try:
                 key, vids = f.result()
-                if vids:
-                    results[key] = vids[:6]
+                if vids:results[key] = vids[:6]
             except:pass
     return results
 
 def extract_xemlaibongda_video(url):
-    """Extract HLS m3u8 URL from xemlaibongda article page."""
     try:
         r=requests.get(url,headers=HEADERS,timeout=15)
         if r.status_code!=200:return None
@@ -282,17 +303,14 @@ def api_highlights():
     return JSONResponse(_cached("xemlaibongda_hl",scrape_xemlaibongda,ttl=_cache_ttl))
 @app.get("/api/highlights/leagues")
 def api_highlights_leagues():
-    """Get highlights grouped by league for homepage display."""
     return JSONResponse(_cached("hl_leagues",scrape_all_league_highlights,ttl=_cache_ttl))
 @app.get("/api/highlights/{league}")
 def api_highlights_league(league:str):
-    """Get highlights for a specific league."""
     if league not in HL_LEAGUES:
         return JSONResponse({"error":"league not found"})
     return JSONResponse(_cached(f"hl_{league}",lambda:scrape_highlights_by_league(league),ttl=_cache_ttl))
 @app.get("/api/highlights_config")
 def api_highlights_config():
-    """Return league config for frontend."""
     return JSONResponse(HL_LEAGUES)
 @app.get("/api/video_url")
 def api_video_url(url:str=Query(...)):
@@ -340,7 +358,7 @@ def api_bdp_videos():
             return arts[:20]
         except:return[]
     return JSONResponse(_cached("bdp_videos",_f))
-# ===== NEWS =====
+# ===== NEWS (unchanged) =====
 def scrape_vne(cat_url):
     try:
         soup=_get(cat_url);arts=[]