Auto scheduler: rewrite AI + short at 7/13/19 VN time from hot topics + latest articles

app_v2_entry.py

@@ -1282,4 +1282,205 @@ def _bg():
         time.sleep(90)
 threading.Thread(target=_bg,daemon=True).start()
 
-app.mount('/static',StaticFiles(directory=STATIC_DIR),name='vnews_static')
+# ===== AUTO SCHEDULER: rewrite AI + short at 7/13/19 VN time =====
+_AUTO_SCHEDULE_TIMES = [(7, '07:00'), (13, '13:00'), (19, '19:00')]
+_AUTO_LOG = os.path.join(DATA_DIR, 'auto_rewrite_log.json')
+
+def _load_auto_log():
+    try:
+        if os.path.exists(_AUTO_LOG):
+            with open(_AUTO_LOG, 'r') as f:
+                return json.load(f)
+    except: pass
+    return {}
+
+def _save_auto_log(log):
+    try:
+        tmp = _AUTO_LOG + '.tmp'
+        with open(tmp, 'w') as f:
+            json.dump(log, f)
+        os.replace(tmp, _AUTO_LOG)
+    except: pass
+
+async def _auto_fetch_short(post_id):
+    """Try to auto-generate a short for a post."""
+    try:
+        import httpx
+        async with httpx.AsyncClient(timeout=180) as cl:
+            r = await cl.post(
+                f"http://localhost:7860/api/ai/short/{post_id}",
+                json={"voice":"vi-VN-HoaiMyNeural","emotion":"neutral","speed":1.2},
+                headers={"Content-Type":"application/json"}
+            )
+            if r.status_code < 300:
+                sj = r.json()
+                if sj.get('video'):
+                    posts = _load_wall_posts()
+                    for p in posts:
+                        if p.get('id') == post_id:
+                            p['video'] = sj['video']
+                            break
+                    _save_wall_posts(posts)
+                    return True
+    except: pass
+    return False
+
+async def _auto_rewrite_one(topic, slot_label):
+    """Rewrite one topic: find articles, summarize, post to wall, trigger short."""
+    from urllib.parse import quote as _q
+    items = _search_all(topic, limit=6)
+    if not items:
+        # fallback: Google RSS
+        try:
+            r = req.get(f"https://news.google.com/rss/search?q={_q(topic)}&hl=vi&gl=VN&ceid=VN:vi",
+                       headers={'User-Agent':'Mozilla/5.0'}, timeout=8)
+            r.encoding = 'utf-8'
+            soup = BeautifulSoup(r.text, 'xml')
+            for it in soup.find_all('item')[:5]:
+                t = _clean(it.find('title').get_text(' ',strip=True) if it.find('title') else '')
+                lk = it.find('link').get_text(strip=True) if it.find('link') else ''
+                if t and lk: items.append({'title':t,'url':lk,'via':'Google News'})
+        except: pass
+    if not items:
+        return False
+    
+    item = items[0]  # best match
+    url = item.get('url', '')
+    title = item.get('title', topic)
+    if not url.startswith('http'):
+        return False
+    
+    data = _scrape_article_for_rewrite(url)
+    if not data or not data.get('paragraphs'):
+        return False
+    
+    raw_text = '\n'.join(data['paragraphs'])
+    ai_text = None
+    
+    # Try AI generation
+    try:
+        import ai_ext
+        prompt = f"Tóm tắt tin tức (tự động {slot_label}):\nTiêu đề: {data['title']}\n{raw_text[:10000]}\n\n4-6 ý chính dạng bullet. Cuối ghi nguồn."
+        ai_text = await ai_ext.qwen_generate(prompt, max_tokens=1000)
+    except: pass
+    
+    if not ai_text or len(ai_text) < 80:
+        pts = data['paragraphs'][:6]
+        ai_text = '\n\n'.join([f"• {p[:300]}" for p in pts])
+        via = item.get('via', '') or urlparse(url).netloc.replace('www.', '')
+        ai_text += f"\n\nNguồn tham khảo: {via}"
+    
+    # Build slides
+    images = data.get('images', [])
+    pts = data['paragraphs'][:10]
+    slides = []
+    for i, p in enumerate(pts[:8]):
+        img = images[i] if i < len(images) else (images[-1] if images else data.get('og_img', ''))
+        slides.append({'text': p[:300], 'image': img, 'index': i + 1})
+    
+    post_id = str(int(time.time() * 1000)) + str(random.randint(100, 999))
+    post = {
+        "id": post_id, "title": data.get('title', title)[:200],
+        "text": ai_text, "img": images[0] if images else data.get('og_img', ''),
+        "url": url, "kind": "auto_rewrite", "slides": slides,
+        "images": images[:10], "video": "",
+        "voice": "vi-VN-HoaiMyNeural", "emotion": "neutral",
+        "language": "vietnamese", "ts": int(time.time()),
+        "auto_scheduled": True, "slot": slot_label,
+    }
+    
+    posts = _load_wall_posts()
+    posts.insert(0, post)
+    _save_wall_posts(posts)
+    
+    # Trigger short generation async
+    threading.Thread(target=lambda: asyncio.run(_auto_fetch_short(post_id)), daemon=True).start()
+    return True
+
+async def _do_scheduled_run(slot_label):
+    """Main scheduled run: top 3 hot topics + latest articles."""
+    print(f"[auto] Starting scheduled rewrite for {slot_label}")
+    
+    # 1) Get top 3 hot topics
+    topics = _get_hot_topics()[:3]
+    job_topics = [t['topic'] for t in topics if t.get('topic')]
+    
+    # 2) Add latest articles from RSS
+    for feed_url in ['https://vnexpress.net/rss/tin-moi-nhat.rss',
+                     'https://dantri.com.vn/rss/home.rss',
+                     'https://vietnamnet.vn/rss/tin-moi-nhat.rss']:
+        try:
+            r = req.get(feed_url, headers={'User-Agent':'Mozilla/5.0'}, timeout=5)
+            r.encoding = 'utf-8'
+            soup = BeautifulSoup(r.text, 'xml')
+            for it in soup.find_all('item')[:3]:
+                t = _clean(it.find('title').get_text(' ',strip=True) if it.find('title') else '')
+                if t and len(t) > 10 and t[:50] not in ' '.join(job_topics):
+                    job_topics.append(t[:60])
+        except: pass
+    
+    # 3) Execute jobs
+    results = []
+    for jt in job_topics[:6]:
+        try:
+            ok = await asyncio.wait_for(_auto_rewrite_one(jt, slot_label), timeout=90)
+            results.append((jt, ok))
+        except: results.append((jt, False))
+        await asyncio.sleep(2)
+    
+    # Log
+    from datetime import datetime, timezone, timedelta
+    VN_TZ_SCHED = timezone(timedelta(hours=7))
+    today_str = datetime.now(VN_TZ_SCHED).strftime('%Y-%m-%d')
+    log = _load_auto_log()
+    if today_str not in log: log[today_str] = {}
+    log[today_str][slot_label] = {
+        'time': datetime.now(VN_TZ_SCHED).strftime('%H:%M:%S'),
+        'count': sum(1 for _, ok in results if ok),
+        'total': len(results),
+    }
+    _save_auto_log(log)
+    print(f"[auto] Done {slot_label}: {sum(1 for _, ok in results if ok)}/{len(results)} posts")
+
+def _scheduler_loop():
+    """Check every 60s; trigger at 7:00, 13:00, 19:00 VN time."""
+    time.sleep(35)
+    from datetime import datetime, timezone, timedelta
+    VN_TZ_SCHED = timezone(timedelta(hours=7))
+    
+    _last_run_date = ""
+    _last_run_slots = set()
+    
+    while True:
+        try:
+            now = datetime.now(VN_TZ_SCHED)
+            today = now.strftime('%Y-%m-%d')
+            hour = now.hour
+            minute = now.minute
+            
+            if today != _last_run_date:
+                _last_run_date = today
+                _last_run_slots = set()
+            
+            slot = None
+            for h, label in _AUTO_SCHEDULE_TIMES:
+                if hour == h and 0 <= minute < 5:
+                    slot = label
+                    break
+            
+            if slot and slot not in _last_run_slots:
+                _last_run_slots.add(slot)
+                loop = asyncio.new_event_loop()
+                asyncio.set_event_loop(loop)
+                try:
+                    loop.run_until_complete(_do_scheduled_run(slot))
+                finally:
+                    loop.close()
+        except Exception as e:
+            print(f"[auto] Loop error: {e}")
+        
+        time.sleep(60)
+
+threading.Thread(target=_scheduler_loop, daemon=True, name='auto-rewrite-scheduler').start()
+
+app.mount('/static',StaticFiles(directory=STATIC_DIR),name='vnews_static')