Upload vtv_scraper.py

vtv_scraper.py

@@ -1,6 +1,7 @@
 """
 VTV Channels Scraper
-Fetches stream URLs from hd.xemtv.net for VTV1-VTV10 + VTV Can Tho
+Fetches stream URLs from hd.xemtv.net for VTV1-VTV9 + VTV Cần Thơ
+Uses fptplay CDN as primary source (accessed via backend proxy with Referer header)
 """
 import requests, re, time, threading
 from bs4 import BeautifulSoup
@@ -11,34 +12,31 @@ UA = {
     "Referer": "https://hd.xemtv.net/",
 }
 
-# Channel page URLs on xemtv.net
+# Channel page URLs on xemtv.net (for scraping)
 XEMTV_CHANNELS = {
     "VTV1": "https://hd.xemtv.net/kenh-vtv1-truc-tuyen-hot.html",
     "VTV2": "https://hd.xemtv.net/kenh-vtv2-truc-tuyen-hot.html",
     "VTV3": "https://hd.xemtv.net/kenh-vtv3-truc-tuyen-hot.html",
     "VTV4": "https://hd.xemtv.net/kenh-vtv4-truc-tuyen-hot.html",
     "VTV5": "https://hd.xemtv.net/kenh-vtv5-truc-tuyen-hot.html",
-    "VTV CẦN THƠ": "https://hd.xemtv.net/kenh-vtv-can-tho-truc-tuyen-hot.html",
-    "VTV6": "https://hd.xemtv.net/kenh-vtv6-truc-tuyen-hot.html",
+    "VTV CẦN THƠ": "https://hd.xemtv.net/kenh-vtv10-truc-tuyen-hot.html",
     "VTV7": "https://hd.xemtv.net/kenh-vtv7-truc-tuyen-hot.html",
     "VTV8": "https://hd.xemtv.net/kenh-vtv8-truc-tuyen-hot.html",
     "VTV9": "https://hd.xemtv.net/kenh-vtv9-truc-tuyen-hot.html",
-    "VTV10": "https://hd.xemtv.net/kenh-vtv10-truc-tuyen-hot.html",
 }
 
-# CDN fallback streams (HoichainTV)
+# Primary CDN streams (fptplay) — these require Referer: https://fptplay.vn/
+# The backend proxy (/api/proxy/hls) adds this header
 CDN_STREAMS = {
-    "VTV1": "https://vips-livecdn.hoichanhtv.com/vtv1/index.m3u8",
-    "VTV2": "https://vips-livecdn.hoichanhtv.com/vtv2/index.m3u8",
-    "VTV3": "https://vips-livecdn.hoichanhtv.com/vtv3/index.m3u8",
-    "VTV4": "https://vips-livecdn.hoichanhtv.com/vtv4/index.m3u8",
-    "VTV5": "https://vips-livecdn.hoichanhtv.com/vtv5/index.m3u8",
-    "VTV CẦN THƠ": "https://vips-livecdn.hoichanhtv.com/vtv6/index.m3u8",
-    "VTV6": "https://vips-livecdn.hoichanhtv.com/vtv6/index.m3u8",
-    "VTV7": "https://vips-livecdn.hoichanhtv.com/vtv7/index.m3u8",
-    "VTV8": "https://vips-livecdn.hoichanhtv.com/vtv8/index.m3u8",
-    "VTV9": "https://vips-livecdn.hoichanhtv.com/vtv9/index.m3u8",
-    "VTV10": "https://vips-livecdn.hoichanhtv.com/vtv10/index.m3u8",
+    "VTV1": "https://live.fptplay53.net/fnxch2/vtv1hd_abr.smil/chunklist.m3u8",
+    "VTV2": "https://live.fptplay53.net/fnxch2/vtv2hd_abr.smil/chunklist.m3u8",
+    "VTV3": "https://live.fptplay53.net/fnxch2/vtv3hd_abr.smil/chunklist.m3u8",
+    "VTV4": "https://live.fptplay53.net/fnxch2/vtv4hd_abr.smil/chunklist.m3u8",
+    "VTV5": "https://live-a.fptplay53.net/live/media/VTV5HD/live_hls_avc/index.m3u8",
+    "VTV CẦN THƠ": "https://live.fptplay53.net/fnxch2/vtvcantho_abr.smil/chunklist.m3u8",
+    "VTV7": "https://live.fptplay53.net/fnxhd1/vtv7hd_vhls.smil/chunklist_b5000000.m3u8",
+    "VTV8": "https://live.fptplay53.net/epzhd1/vtv8hd_vhls.smil/chunklist.m3u8",
+    "VTV9": "https://live.fptplay53.net/fnxhd1/vtv9hd_vhls.smil/chunklist.m3u8",
 }
 
 _vtv_cache = {}
@@ -78,7 +76,6 @@ def extract_m3u8_from_html(html):
         m = re.search(pat, html, re.IGNORECASE)
         if m:
             url = m.group(1).strip()
-            # Validate it looks like a real m3u8
             if '.m3u8' in url and len(url) > 20:
                 return url
     return None
@@ -87,28 +84,28 @@ def extract_m3u8_from_html(html):
 def fetch_vtv_stream(channel_name):
     """Fetch m3u8 stream URL for a VTV channel from xemtv.net."""
     channel_name = channel_name.upper().strip()
-    
+
     # Normalize name
     name_map = {
         'VTVCT': 'VTV CẦN THƠ', 'VTV-CAN-THO': 'VTV CẦN THƠ',
-        'VTV CAN THO': 'VTV CẦN THƠ', 'VTVCT': 'VTV CẦN THƠ',
+        'VTV CAN THO': 'VTV CẦN THƠ', 'VTV6': 'VTV CẦN THƠ',
+        'VTCT': 'VTV CẦN THƠ',
     }
     channel_name = name_map.get(channel_name, channel_name)
-    
+
     # Check cache
     cached = _cached(channel_name)
     if cached:
         return cached
-    
+
     page_url = XEMTV_CHANNELS.get(channel_name)
     if not page_url:
-        # Try CDN fallback
         cdn = CDN_STREAMS.get(channel_name)
         if cdn:
             _set_cache(channel_name, cdn)
             return cdn
         return None
-    
+
     try:
         r = requests.get(page_url, headers=UA, timeout=15, allow_redirects=True)
         if r.status_code == 200:
@@ -116,15 +113,15 @@ def fetch_vtv_stream(channel_name):
             if m3u8:
                 _set_cache(channel_name, m3u8)
                 return m3u8
-    except Exception as e:
+    except:
         pass
-    
+
     # Fallback to CDN
     cdn = CDN_STREAMS.get(channel_name)
     if cdn:
         _set_cache(channel_name, cdn)
         return cdn
-    
+
     return None