[FIX] VnExpress Short format scraper fallback for JS-lazy-loaded articles

rewrite_slide.py

@@ -31,11 +31,14 @@ def _clean(s): return re.sub(r'\s+', ' ', str(s or '')).strip()
 
 
 def _scrape_article_full(url):
-    """Scrape article: extract paragraphs + ALL images."""
+    """Scrape article: extract paragraphs + ALL images.
+    Handles both normal HTML articles AND VnExpress JS-lazy-loaded Short format.
+    """
     try:
         r = requests.get(url, headers=UA, timeout=15, allow_redirects=True)
         r.encoding = 'utf-8'
-        soup = BeautifulSoup(r.text, 'lxml')
+        html = r.text
+        soup = BeautifulSoup(html, 'lxml')
         for tag in soup.find_all(['script', 'style', 'nav', 'footer', 'aside', 'form']): tag.decompose()
         
         # Title
@@ -48,11 +51,11 @@ def _scrape_article_full(url):
         og_img = ogi.get('content', '') if ogi else ''
         if og_img and og_img.startswith('//'): og_img = 'https:' + og_img
         
-        # Find content block
+        # Find content block (normal article layout)
         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
+            es = soup.select_one(sel)
+            if es and len(es.find_all('p')) >= 2: block = es; break
         if not block: block = soup.body or soup
         
         # Extract paragraphs and images IN ORDER
@@ -79,6 +82,42 @@ def _scrape_article_full(url):
                             images.append(src)
                             seen_imgs.add(src)
         
+        # If no meaningful paragraphs found, try VnExpress Short format
+        # (JS-lazy-loaded articles where body class is "page-short")
+        if len(paragraphs) < 2:
+            body = soup.find('body')
+            body_class = ' '.join(body.get('class', [])) if body else ''
+            
+            if 'page-short' in body_class or not paragraphs:
+                paragraphs = []
+                images = []
+                seen_imgs.clear()
+                
+                # Extract dataCurrentDetail for title + lead
+                m = re.search(r'var dataCurrentDetail\s*=\s*({.*?});', html, re.DOTALL)
+                if m:
+                    try:
+                        d = json.loads(m.group(1))
+                        lead = _clean(d.get('lead', ''))
+                        if lead and len(lead) > 50:
+                            paragraphs.append(lead)
+                        thumb = d.get('thumbnail_url', '') or ''
+                        if thumb and thumb not in seen_imgs:
+                            images.append(thumb)
+                            seen_imgs.add(thumb)
+                    except: pass
+                
+                # Also get OG description
+                ogd = soup.find('meta', property='og:description')
+                if ogd:
+                    desc = _clean(ogd.get('content', ''))
+                    if desc and len(desc) > 50 and desc not in paragraphs:
+                        paragraphs.append(desc)
+                
+                # Use title as paragraph if nothing else
+                if not paragraphs and title:
+                    paragraphs.append(title[:200] + '.')
+        
         return {'title': _clean(title), 'paragraphs': paragraphs, 'images': images, 'og_img': og_img}
     except Exception as e:
         return None