Upload app_v2_entry_patch.py

app_v2_entry_patch.py

@@ -0,0 +1,194 @@
+"""Patch to add rewrite endpoints to app_v2_entry.py - append to bottom of file."""
+# This file contains the code that should be appended to app_v2_entry.py
+# to enable the Rewrite functionality on Tường AI
+
+# ===== REWRITE / ARTICLE-TO-WALL ENDPOINTS =====
+import random as _random2
+from urllib.parse import quote as _quote2
+
+_UA_RW = {'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36', 'Accept-Language': 'vi-VN,vi;q=0.9'}
+
+def _scrape_article_for_rewrite(url):
+    """Scrape article: extract title, paragraphs, images, OG image."""
+    try:
+        r = req.get(url, headers=_UA_RW, timeout=15, allow_redirects=True)
+        r.encoding = 'utf-8'
+        soup = BeautifulSoup(r.text, 'lxml')
+        for tag in soup.find_all(['script', 'style', 'nav', 'footer', 'aside', 'form']):
+            tag.decompose()
+        h1 = soup.find('h1')
+        ogt = soup.find('meta', property='og:title')
+        title = (h1.get_text(strip=True) if h1 else '') or (ogt.get('content', '') if ogt else '')
+        ogi = soup.find('meta', property='og:image')
+        og_img = ogi.get('content', '') if ogi else ''
+        if og_img and og_img.startswith('//'):
+            og_img = 'https:' + og_img
+        block = None
+        for sel in ['article', '.singular-content', '.detail-content', '.fck_detail', '.content-detail', '.knc-content', 'main', '.cms-body', '.article__body']:
+            el = soup.select_one(sel)
+            if el and len(el.find_all('p')) >= 2:
+                block = el
+                break
+        if not block:
+            block = soup.body or soup
+        paragraphs = []
+        images = []
+        seen_imgs = set()
+        if og_img and og_img not in seen_imgs:
+            images.append(og_img)
+            seen_imgs.add(og_img)
+        for el in block.find_all(['p', 'h2', 'h3', 'figure', 'img'], recursive=True):
+            if el.name == 'p':
+                t = _clean(el.get_text(strip=True))
+                if t and len(t) > 40:
+                    paragraphs.append(t)
+            elif el.name in ('figure', 'img'):
+                im = el if el.name == 'img' else el.find('img')
+                if im:
+                    src = im.get('data-src') or im.get('src') or im.get('data-original') or ''
+                    if src and 'base64' not in src:
+                        if src.startswith('//'):
+                            src = 'https:' + src
+                        if src not in seen_imgs:
+                            images.append(src)
+                            seen_imgs.add(src)
+        return {'title': _clean(title), 'paragraphs': paragraphs, 'images': images, 'og_img': og_img}
+    except Exception:
+        return None
+
+
+def _extract_key_points_rw(paragraphs, max_points=5):
+    """Extract key points from paragraphs."""
+    points = []
+    for p in paragraphs:
+        if len(points) >= max_points:
+            break
+        m = re.match(r'^(.+?[.!?])\s', p)
+        if m:
+            sentence = m.group(1)
+        else:
+            sentence = p[:150] + ('.' if not p.endswith('.') else '')
+        if len(sentence) < 30:
+            continue
+        if any(sentence[:50] in existing for existing in points):
+            continue
+        points.append(sentence)
+    return points
+
+
+@app.post("/api/rewrite_slide")
+async def api_rewrite_slide(request: Request):
+    """Fast rewrite as SLIDES - no AI needed, instant response."""
+    body = await request.json()
+    url = _clean(body.get("url", ""))
+    context = body.get("context", "")
+    if not url and not context:
+        return JSONResponse({"error": "Cần URL hoặc nội dung"}, status_code=400)
+    data = None
+    if url and url.startswith("http"):
+        data = _scrape_article_for_rewrite(url)
+    if not data and context:
+        paragraphs = [_clean(p) for p in context.split('\n') if len(_clean(p)) > 40]
+        data = {'title': paragraphs[0][:80] if paragraphs else 'Bài viết', 'paragraphs': paragraphs, 'images': [], 'og_img': ''}
+    if not data or not data.get('paragraphs'):
+        return JSONResponse({"error": "Không đọc được bài viết"}, status_code=422)
+    points = _extract_key_points_rw(data['paragraphs'], max_points=6)
+    if not points:
+        return JSONResponse({"error": "Không tìm được ý chính"}, status_code=422)
+    images = data.get('images', [])
+    slides = []
+    for i, point in enumerate(points):
+        img = images[i] if i < len(images) else (images[-1] if images else '')
+        if img and 'cdnphoto.dantri' in img:
+            img = '/api/proxy/img?url=' + _quote2(img, safe='')
+        slides.append({'text': point, 'image': img, 'index': i + 1})
+    summary_text = '\n\n'.join([f"• {s['text']}" for s in slides])
+    post = {
+        "id": str(int(time.time() * 1000)) + str(_random2.randint(100, 999)),
+        "title": data['title'],
+        "text": summary_text,
+        "img": images[0] if images else '',
+        "url": url,
+        "kind": "slide_summary",
+        "slides": slides,
+        "images": images[:10],
+        "video": "",
+        "voice": "hoaimy",
+        "emotion": "trung_tinh",
+        "ts": int(time.time())
+    }
+    posts = _load_wall_posts()
+    posts.insert(0, post)
+    _save_wall_posts(posts)
+    return JSONResponse({"post": post, "slides": slides})
+
+
+@app.post("/api/rewrite_share")
+async def api_rewrite_share(request: Request):
+    """Rewrite article and post to Tường AI (with AI fallback to extractive)."""
+    body = await request.json()
+    url = _clean(body.get("url", ""))
+    ctx = _clean(body.get("context", ""))
+    if not url and not ctx:
+        return JSONResponse({"error": "Cần URL hoặc nội dung"}, status_code=400)
+    data = None
+    if url and url.startswith("http"):
+        data = _scrape_article_for_rewrite(url)
+    if not data and ctx:
+        paragraphs = [_clean(p) for p in ctx.split('\n') if len(_clean(p)) > 40]
+        data = {'title': paragraphs[0][:80] if paragraphs else 'Bài viết', 'paragraphs': paragraphs, 'images': [], 'og_img': ''}
+    if not data or not data.get('paragraphs'):
+        return JSONResponse({"error": "Không đọc được bài viết"}, status_code=422)
+    raw_text = '\n'.join(data['paragraphs'])
+    if len(raw_text) < 50:
+        raw_text = ctx[:14000]
+    if len(raw_text) < 50:
+        return JSONResponse({"error": "Bài viết quá ngắn"}, status_code=422)
+    domain = ''
+    try:
+        from urllib.parse import urlparse
+        domain = urlparse(url).netloc.replace('www.', '')
+    except:
+        pass
+    ai_text = None
+    try:
+        import ai_ext
+        if hasattr(ai_ext, 'qwen_generate'):
+            prompt = f'Tóm tắt đăng Tường AI:\nTiêu đề: {data["title"]}\n{raw_text[:14000]}\n\n4-6 ý chính. Cuối ghi nguồn.'
+            ai_text = await ai_ext.qwen_generate(prompt, max_tokens=1000)
+    except Exception:
+        pass
+    if not ai_text or len(ai_text) < 80:
+        key_pts = _extract_key_points_rw(data['paragraphs'], max_points=6)
+        if key_pts:
+            ai_text = '\n\n'.join([f"• {p}" for p in key_pts])
+        else:
+            ai_text = f"Tóm tắt: {data['title']}\n\n{raw_text[:1200]}\n\nNguồn: {domain}"
+    images = data.get('images', [])
+    post = {
+        "id": str(int(time.time() * 1000)) + str(_random2.randint(100, 999)),
+        "title": data['title'],
+        "text": ai_text,
+        "img": images[0] if images else '',
+        "url": url,
+        "kind": "rewrite",
+        "images": images[:10],
+        "video": "",
+        "voice": "hoaimy",
+        "emotion": "trung_tinh",
+        "ts": int(time.time())
+    }
+    posts = _load_wall_posts()
+    posts.insert(0, post)
+    _save_wall_posts(posts)
+    return JSONResponse({"post": post})
+
+
+@app.post("/api/url_wall")
+async def api_url_wall(request: Request):
+    """Submit URL to add to Tường AI."""
+    body = await request.json()
+    url = _clean(body.get("url", ""))
+    if not url or not url.startswith('http'):
+        return JSONResponse({"error": "URL không hợp lệ"}, status_code=400)
+    return await api_rewrite_share(request)