World Cup RSS all articles + persistent AI rewrite wall APIs

main.py

@@ -72,6 +72,20 @@ def _save_short_db(db):
     except:pass
 
 def _short_default():return {"views":0,"likes":0,"shares":0,"comments":[]}
+WALL_FILE = "/data/wall_posts.json" if os.path.isdir("/data") else "/app/wall_posts.json"
+def _load_wall():
+    try:
+        if os.path.exists(WALL_FILE):
+            with open(WALL_FILE,"r",encoding="utf-8") as f:return json.load(f)
+    except:pass
+    return []
+def _save_wall(posts):
+    try:
+        os.makedirs(os.path.dirname(WALL_FILE),exist_ok=True)
+        tmp=WALL_FILE+".tmp"
+        with open(tmp,"w",encoding="utf-8") as f:json.dump(posts[:100],f,ensure_ascii=False)
+        os.replace(tmp,WALL_FILE)
+    except:pass
 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}
 HL_LEAGUES = {"premier-league":{"path":"anh/premier-league","name":"Premier League","emoji":"🏴󠁧󠁢󠁥󠁮󠁧󠁿"},"fa-cup":{"path":"anh/fa-cup","name":"FA Cup","emoji":"🏆"},"bundesliga":{"path":"duc/bundesliga","name":"Bundesliga","emoji":"🇩🇪"},"serie-a":{"path":"italy/serie-a","name":"Serie A","emoji":"🇮🇹"},"la-liga":{"path":"tay-ban-nha/la-liga","name":"La Liga","emoji":"🇪🇸"},"champions-league":{"path":"cup-chau-au/uefa-champions-league","name":"Champions League","emoji":"⭐"},"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":"🌍"}}
@@ -541,7 +555,20 @@ def scrape_bbc_article(url):
     except:return None
 
 def scrape_ttvh_worldcup():
-    """Scrape World Cup 2026 articles from The Thao Van Hoa."""
+    """Scrape all World Cup 2026 articles from The Thao Van Hoa RSS."""
+    try:
+        r=requests.get("https://thethaovanhoa.vn/rss/world-cup-2026.rss",headers=HEADERS,timeout=15);r.encoding="utf-8"
+        soup=BeautifulSoup(r.text,"xml");arts=[];seen=set()
+        for it in soup.find_all("item"):
+            title=(it.find("title").get_text(strip=True) if it.find("title") else "")
+            link=(it.find("link").get_text(strip=True) if it.find("link") else "")
+            desc=(it.find("description").get_text(" ",strip=True) if it.find("description") else "")
+            img="";ds=BeautifulSoup(desc,"lxml");im=ds.find("img")
+            if im:img=im.get("src","") or im.get("data-src","")
+            if title and link and link not in seen:
+                seen.add(link);arts.append({"title":title,"link":link,"img":img,"source":"ttvh"})
+        if arts:return arts
+    except:pass
     try:
         soup=_get("https://thethaovanhoa.vn/world-cup-2026.htm");arts=[];seen=set()
         for a in soup.find_all("a",href=True):
@@ -646,6 +673,43 @@ def api_article(url:str=Query(...)):
     elif"thethaovanhoa.vn" in url:data=scrape_ttvh_article(url)
     else:data=None
     return JSONResponse(data if data else{"error":"not supported"})
+def _article_by_url(url):
+    if "vnexpress.net" in url:return scrape_vne_article(url)
+    if "bbc.com" in url:return scrape_bbc_article(url)
+    if "dantri.com.vn" in url:return scrape_dantri_article(url)
+    if "genk.vn" in url:return scrape_genk_article(url)
+    if "thethaovanhoa.vn" in url:return scrape_ttvh_article(url)
+    return None
+
+def _ai_rewrite_article(data):
+    title=(data or {}).get("title","");summary=(data or {}).get("summary","")
+    ps=[b.get("text","") for b in (data or {}).get("body",[]) if b.get("type")=="p" and len(b.get("text",""))>30]
+    lead=summary or (ps[0] if ps else "")
+    points=[]
+    for p in ps[:5]:
+        t=p.strip()
+        if len(t)>180:t=t[:177]+"..."
+        points.append(t)
+    rewritten="Bản tin AI viết lại: "+title+"\n\n"+lead
+    if points:rewritten += "\n\nĐiểm chính:\n" + "\n".join(["• "+x for x in points])
+    return rewritten
+
+@app.get("/api/wall")
+def api_wall():return JSONResponse({"posts":_load_wall()[:50]})
+
+@app.post("/api/rewrite_share")
+async def api_rewrite_share(request:Request):
+    try:body=await request.json()
+    except:body={}
+    url=str(body.get("url","")).strip()
+    if not url:return JSONResponse({"error":"missing url"},status_code=400)
+    data=_article_by_url(url)
+    if not data or not data.get("title") or not data.get("body"):
+        return JSONResponse({"error":"Không đọc được bài viết"},status_code=422)
+    post={"id":hashlib.md5((url+str(time.time())).encode()).hexdigest()[:12],"url":url,"title":data.get("title",""),"img":data.get("og_image","") or "","text":_ai_rewrite_article(data),"ts":int(time.time()),"source":data.get("source","")}
+    posts=_load_wall();posts.insert(0,post);_save_wall(posts)
+    return JSONResponse({"post":post})
+
 @app.get("/v")
 async def video_share(url:str=Query(default=""),title:str=Query(default="VNEWS Video"),img:str=Query(default=""),type:str=Query(default="highlights")):
     decoded_url=unquote(url);decoded_title=unquote(title)