Add all VNE categories + VnExpress video + DanTri tin nong slide

main.py

@@ -24,6 +24,66 @@ def _cached(key, fn):
 def _get(url,headers=None):
     h=headers or HEADERS;r=requests.get(url,headers=h,timeout=15);r.encoding="utf-8"
     return BeautifulSoup(r.text,"lxml")
+
+# ===== DANTRI TIN NONG =====
+def scrape_dantri_hot():
+    """Scrape tin nóng from dantri.com.vn"""
+    try:
+        soup=_get("https://dantri.com.vn/tin-nong.htm")
+        arts=[];seen=set()
+        for a in soup.find_all("a",href=True):
+            href=a.get("href","");title=a.get("title","") or a.get_text(strip=True)
+            if not title or len(title)<15 or"javascript:" in href:continue
+            if not href.startswith("http"):href="https://dantri.com.vn"+href
+            if href in seen or not href.endswith(".htm"):continue
+            if href=="https://dantri.com.vn/tin-nong.htm":continue
+            seen.add(href)
+            img=a.find("img")
+            if not img and a.parent:img=a.parent.find("img")
+            img_src=""
+            if img:img_src=img.get("data-src") or img.get("src","")
+            if img_src and("base64" in img_src or len(img_src)<10):img_src=""
+            arts.append({"title":title,"link":href,"img":img_src,"source":"dantri"})
+            if len(arts)>=15:break
+        # OG images for missing
+        def _og(art):
+            if art["img"]:return
+            try:
+                r2=requests.get(art["link"],headers=HEADERS,timeout=8);r2.encoding="utf-8"
+                og=BeautifulSoup(r2.text,"lxml").find("meta",property="og:image")
+                art["img"]=og.get("content","") if og else ""
+            except:pass
+        need=[a for a in arts if not a["img"]][:5]
+        if need:
+            with ThreadPoolExecutor(3) as ex:list(ex.map(_og,need))
+        return arts[:12]
+    except:return[]
+
+# ===== VNEXPRESS VIDEO (VnEgo) =====
+def scrape_vne_video():
+    """Scrape VnExpress video section."""
+    try:
+        soup=_get("https://vnexpress.net/video")
+        arts=[];seen=set()
+        for a in soup.find_all("a",href=True):
+            href=a.get("href","");title=a.get("title","") or a.get_text(strip=True)
+            if not title or len(title)<10 or"javascript:" in href:continue
+            if"/video/" not in href:continue
+            if href in seen:continue
+            seen.add(href)
+            img=a.find("img")
+            if not img and a.parent:img=a.parent.find("img")
+            img_src=""
+            if img:img_src=img.get("data-src") or img.get("src","")
+            if img_src and("blank" in img_src or"base64" in img_src):
+                source=a.find("source") or (a.parent.find("source") if a.parent else None)
+                if source:img_src=source.get("srcset","").split(",")[0].strip().split(" ")[0]
+            if not img_src or len(img_src)<10:img_src=""
+            arts.append({"title":title,"link":href,"img":img_src,"source":"vne-video"})
+            if len(arts)>=15:break
+        return arts[:12]
+    except:return[]
+
 def scrape_vne(cat_url):
     try:
         soup=_get(cat_url);arts=[]
@@ -195,7 +255,10 @@ def extract_bdp_video(url):
         source=video.find("source")
         return{"src":source.get("src","") if source else "","poster":video.get("poster","")}
     except:return None
-VNE_CATS={"thoi-su":("https://vnexpress.net/thoi-su","Thời Sự"),"the-gioi":("https://vnexpress.net/the-gioi","Thế Giới"),"kinh-doanh":("https://vnexpress.net/kinh-doanh","Kinh Doanh"),"cong-nghe":("https://vnexpress.net/so-hoa","Công Nghệ"),"the-thao":("https://vnexpress.net/the-thao","Thể Thao"),"giai-tri":("https://vnexpress.net/giai-tri","Giải Trí"),"suc-khoe":("https://vnexpress.net/suc-khoe","Sức Khỏe")}
+
+# All VnExpress categories
+VNE_CATS={"thoi-su":("https://vnexpress.net/thoi-su","Thời Sự"),"the-gioi":("https://vnexpress.net/the-gioi","Thế Giới"),"kinh-doanh":("https://vnexpress.net/kinh-doanh","Kinh Doanh"),"cong-nghe":("https://vnexpress.net/so-hoa","Công Nghệ"),"the-thao":("https://vnexpress.net/the-thao","Thể Thao"),"giai-tri":("https://vnexpress.net/giai-tri","Giải Trí"),"suc-khoe":("https://vnexpress.net/suc-khoe","Sức Khỏe"),"giao-duc":("https://vnexpress.net/giao-duc","Giáo Dục"),"phap-luat":("https://vnexpress.net/phap-luat","Pháp Luật"),"du-lich":("https://vnexpress.net/du-lich","Du Lịch"),"xe":("https://vnexpress.net/oto-xe-may","Xe"),"doi-song":("https://vnexpress.net/doi-song","Đời Sống")}
+
 @app.get("/api/homepage")
 def api_homepage():
     def _f():
@@ -225,6 +288,10 @@ def api_categories():
 def api_highlights():return JSONResponse(_cached("highlights",scrape_24h_highlights))
 @app.get("/api/shorts")
 def api_shorts():return JSONResponse(_cached("shorts",scrape_24h_shorts))
+@app.get("/api/vne_video")
+def api_vne_video():return JSONResponse(_cached("vne_video",scrape_vne_video))
+@app.get("/api/dantri_hot")
+def api_dantri_hot():return JSONResponse(_cached("dantri_hot",scrape_dantri_hot))
 @app.get("/api/bdp_videos")
 def api_bdp_videos():return JSONResponse(_cached("bdp_videos",scrape_bdp_videos))
 @app.get("/api/video_url")
@@ -244,13 +311,13 @@ async def video_share(url:str=Query(default=""),title:str=Query(default="VNEWS V
     og_image=unquote(img) if img else "https://s1.vnecdn.net/vnexpress/restruct/i/v9505/logo_default.jpg"
     decoded_url=unquote(url);decoded_title=unquote(title)
     redirect_script=f'<script>localStorage.setItem("pending_video",JSON.stringify({{"url":"{decoded_url}","type":"{type}"}}));location.href="{SPACE_URL}";</script>' if decoded_url else f'<script>location.href="{SPACE_URL}";</script>'
-    return HTMLResponse(f'''<!DOCTYPE html><html><head><meta charset="utf-8"><title>{decoded_title} - VNEWS</title><meta property="og:title" content="{decoded_title}"><meta property="og:image" content="{og_image}"><meta property="og:type" content="video.other"><meta property="og:site_name" content="VNEWS"><meta property="og:description" content="🎬 {decoded_title} - Xem trên VNEWS"><meta name="twitter:card" content="summary_large_image"><meta name="twitter:title" content="{decoded_title}"><meta name="twitter:image" content="{og_image}"></head><body style="background:#111;color:#fff;font-family:sans-serif;text-align:center;padding:40px"><p>⏳ Đang mở video...</p>{redirect_script}</body></html>''')
+    return HTMLResponse(f'''<!DOCTYPE html><html><head><meta charset="utf-8"><title>{decoded_title} - VNEWS</title><meta property="og:title" content="{decoded_title}"><meta property="og:image" content="{og_image}"><meta property="og:type" content="video.other"><meta property="og:site_name" content="VNEWS"></head><body style="background:#111;color:#fff;text-align:center;padding:40px"><p>⏳ Đang mở video...</p>{redirect_script}</body></html>''')
 @app.get("/s")
 async def share_redirect(url:str=Query(default=""),title:str=Query(default="VNEWS"),img:str=Query(default="")):
     og_image=unquote(img) if img else "https://s1.vnecdn.net/vnexpress/restruct/i/v9505/logo_default.jpg"
     decoded_url=unquote(url)
     redirect_script=f'<script>localStorage.setItem("pending_article","{decoded_url}");location.href="{SPACE_URL}";</script>' if decoded_url else f'<script>location.href="{SPACE_URL}";</script>'
-    return HTMLResponse(f'''<!DOCTYPE html><html><head><meta charset="utf-8"><title>{unquote(title)}</title><meta property="og:title" content="{unquote(title)}"><meta property="og:image" content="{og_image}"><meta property="og:url" content="{SPACE_URL}/s?url={url}&title={title}&img={img}"><meta property="og:type" content="article"><meta property="og:site_name" content="VNEWS"><meta property="og:description" content="{unquote(title)} - Đọc trên VNEWS"><meta name="twitter:card" content="summary_large_image"><meta name="twitter:title" content="{unquote(title)}"><meta name="twitter:image" content="{og_image}"></head><body>{redirect_script}</body></html>''')
+    return HTMLResponse(f'''<!DOCTYPE html><html><head><meta charset="utf-8"><title>{unquote(title)}</title><meta property="og:title" content="{unquote(title)}"><meta property="og:image" content="{og_image}"><meta property="og:type" content="article"><meta property="og:site_name" content="VNEWS"></head><body>{redirect_script}</body></html>''')
 @app.get("/")
 async def index():
     with open("/app/static/index.html","r",encoding="utf-8") as f:return HTMLResponse(content=f.read())