Fix: readArticle supports ALL URLs via /api/article scrape, no external links"

app_run.py

@@ -1,4 +1,4 @@
-"""Wrapper: faster + relevant hashtag sources via Google News search."""
+"""Wrapper: relevant hashtag + ALL articles readable in-app via scrape."""
 from app_final import *
 from app_final import app, f6, f5, rt, PATCH_INJECT, UNIFIED_INJECT_FIXED, HIGHLIGHT_FULL_OVERRIDE, EXTRA_WALL_FIX
 from fastapi.responses import HTMLResponse, JSONResponse
@@ -11,7 +11,6 @@ import re, html as html_lib
 def _clean(s):return re.sub(r"\s+"," ",html_lib.unescape(str(s or ""))).strip()
 
 def _google_news_search(topic, limit=8):
-    """Search Google News RSS for topic-specific articles — much more relevant than generic RSS pool."""
     items=[]
     try:
         url='https://news.google.com/rss/search?q='+quote(topic+' tin tức')+'&hl=vi&gl=VN&ceid=VN:vi'
@@ -22,9 +21,7 @@ def _google_news_search(topic, limit=8):
             link=_clean(it.find('link').get_text(strip=True) if it.find('link') else '')
             src=_clean(it.find('source').get_text(' ',strip=True) if it.find('source') else '')
             if not title or not link:continue
-            # Relevance check: topic keywords must appear in title
-            topic_lower=topic.lower()
-            title_lower=title.lower()
+            topic_lower=topic.lower();title_lower=title.lower()
             topic_words=[w for w in re.findall(r'[A-Za-zÀ-ỹ0-9]+',topic_lower) if len(w)>1]
             match_count=sum(1 for w in topic_words if w in title_lower)
             if match_count==0 and topic_lower not in title_lower:continue
@@ -34,7 +31,6 @@ def _google_news_search(topic, limit=8):
     return items
 
 def _duckduckgo_search(topic, limit=6):
-    """Fallback search via DuckDuckGo HTML."""
     items=[]
     try:
         url='https://html.duckduckgo.com/html/?q='+quote(topic+' Việt Nam tin tức')
@@ -45,7 +41,6 @@ def _duckduckgo_search(topic, limit=6):
             if not a:continue
             title=_clean(a.get_text(' ',strip=True))
             href=a.get('href','')
-            # Unwrap DDG redirect
             if 'duckduckgo.com/l/' in href:
                 from urllib.parse import parse_qs,urlparse,unquote
                 qs=parse_qs(urlparse(href if href.startswith('http') else 'https:'+href).query)
@@ -59,7 +54,6 @@ def _duckduckgo_search(topic, limit=6):
     except:pass
     return items
 
-# Override endpoints
 app.router.routes=[r for r in app.router.routes if not (
     (getattr(r,'path',None)=='/api/hashtag/sources' and 'GET' in getattr(r,'methods',set())) or
     (getattr(r,'path',None)=='/' and 'GET' in getattr(r,'methods',set()))
@@ -67,17 +61,13 @@ app.router.routes=[r for r in app.router.routes if not (
 
 @app.get('/api/hashtag/sources')
 def _hashtag_relevant(topic:str=Query(...)):
-    """Return RELEVANT sources by searching Google News + DuckDuckGo for the specific topic."""
     sources=_google_news_search(topic,8)
-    if len(sources)<3:
-        sources+=_duckduckgo_search(topic,6-len(sources))
-    # Dedupe by URL
+    if len(sources)<3:sources+=_duckduckgo_search(topic,6-len(sources))
     seen=set();unique=[]
     for s in sources:
         if s['url'] not in seen:seen.add(s['url']);unique.append(s)
     return JSONResponse({'sources':unique[:8],'topic':topic})
 
-# Extra JS: override showHashtagSources with spinner + lazy img + topic input integration
 FAST_HASHTAG_JS = r'''
 <style>
 .hashtag-loading{display:flex;align-items:center;gap:8px;padding:12px;color:#888;font-size:12px}
@@ -88,6 +78,67 @@ FAST_HASHTAG_JS = r'''
 (function(){
 function esc(s){return String(s||'').replace(/[&<>"']/g,function(m){return{'&':'&amp;','<':'&lt;','>':'&gt;','"':'&quot;',"'":'&#39;'}[m]});}
 
+// === OVERRIDE readArticle to support ALL URLs via /api/article scrape ===
+var _origReadArticle=window.readArticle;
+window.readArticle=async function(url,source){
+  // Always try to read in-app first via /api/article
+  showView('view-article');
+  var el=document.getElementById('view-article');
+  el.innerHTML='<div class="loading">Đang tải bài viết...</div>';
+  try{
+    var r=await fetch('/api/article?url='+encodeURIComponent(url));
+    var data=await r.json();
+    if(data&&!data.error&&data.body&&data.body.length){
+      window._currentArticle={url:url,data:data};
+      var h='<button class="back-btn" onclick="switchCat(\'home\')">← Quay lại</button><div class="article-view"><h1 class="article-title">'+esc(data.title)+'</h1>';
+      if(data.summary)h+='<div class="article-summary">'+esc(data.summary)+'</div>';
+      var seenImgs={};
+      data.body.forEach(function(b){
+        if(b.type==='p')h+='<p class="article-p">'+b.text+'</p>';
+        else if(b.type==='img'&&b.src&&!seenImgs[b.src]){seenImgs[b.src]=1;h+='<img class="article-img" src="'+esc(b.src)+'">';}
+        else if(b.type==='heading')h+='<h2 class="article-h2">'+esc(b.text)+'</h2>';
+      });
+      h+='<div class="article-actions"><button class="primary" onclick="doRewriteArticle(this)">🤖 Rewrite AI đăng tường</button><button onclick="doShare(\''+esc(data.title)+'\',\''+esc(url)+'\',\''+esc(data.og_image||'')+'\')">📤 Chia sẻ</button><button onclick="window.open(\''+esc(url)+'\',\'_blank\')">🔗 Gốc</button></div>';
+      h+='<div class="article-ai-ask"><h3 style="font-size:14px;color:#5cb87a">🤖 Hỏi AI</h3><textarea id="article-ai-question" placeholder="Hỏi về bài viết..."></textarea><button onclick="askArticleAI()">Hỏi</button><div id="article-ai-answer" class="article-ai-answer"></div></div>';
+      h+='</div>';
+      el.innerHTML=h;
+      window.scrollTo(0,0);
+      return;
+    }
+  }catch(e){}
+  // Fallback: try original readArticle for supported domains
+  if(_origReadArticle){
+    try{await _origReadArticle(url,source);return;}catch(e){}
+  }
+  // Last fallback: show link to open externally
+  el.innerHTML='<button class="back-btn" onclick="switchCat(\'home\')">← Quay lại</button><div class="loading"><p>Không đọc được bài này trong app.</p><a href="'+esc(url)+'" target="_blank" style="color:#5cb87a">Mở link gốc →</a></div>';
+};
+
+// doRewriteArticle for articles opened from hashtag sources
+window.doRewriteArticle=async function(btn){
+  var url=(window._currentArticle&&window._currentArticle.url)||'';
+  if(!url){alert('Không có URL');return;}
+  var ctx=document.querySelector('.article-view')?.innerText?.slice(0,14000)||'';
+  btn.disabled=true;btn.textContent='Đang rewrite...';
+  try{
+    var r=await fetch('/api/rewrite_share',{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({url:url,context:ctx})});
+    var j=await r.json();
+    if(!r.ok||j.error)throw new Error(j.error||'Lỗi');
+    alert('Đã rewrite và đăng lên Tường AI!');
+  }catch(e){alert(e.message);}
+  finally{btn.disabled=false;btn.textContent='🤖 Rewrite AI đăng tường';}
+};
+
+window.askArticleAI=async function(){
+  var q=document.getElementById('article-ai-question')?.value.trim();
+  if(!q)return alert('Nhập câu hỏi');
+  var a=document.getElementById('article-ai-answer');a.textContent='Đang hỏi...';
+  var url=(window._currentArticle&&window._currentArticle.url)||'';
+  var ctx=document.querySelector('.article-view')?.innerText?.slice(0,12000)||'';
+  try{var r=await fetch('/api/article/ask',{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({url:url,question:q,context:ctx})});var j=await r.json();a.textContent=j.answer||'Không trả lời được';}catch(e){a.textContent='Lỗi: '+e.message}
+};
+
+// === HASHTAG SOURCES ===
 window.showHashtagSources=async function(topic){
   var home=document.getElementById('view-home');if(!home)return;
   document.getElementById('hashtag-sources-box')?.remove();
@@ -102,13 +153,14 @@ window.showHashtagSources=async function(topic){
     if(!sources.length){box.innerHTML='<h3>🔍 '+esc(topic)+'</h3><div style="color:#888;font-size:12px;padding:8px">Không tìm được nguồn liên quan</div>';return;}
     var h='<h3>🔍 '+esc(topic)+' <span style="font-size:10px;color:#888">('+sources.length+' nguồn)</span></h3>';
     sources.forEach(function(s,i){
-      h+='<div class="hashtag-src-item" onclick="if(typeof readArticle===\'function\')readArticle(\''+esc(s.url||'')+'\')">';
+      h+='<div class="hashtag-src-item" onclick="readArticle(\''+esc(s.url||'')+'\')">';
       h+='<div class="hashtag-src-img" id="ht-img-'+i+'"></div>';
       h+='<div class="hashtag-src-text"><div class="hashtag-src-title">'+esc(s.title)+'</div><div class="hashtag-src-via">'+esc(s.via||'')+'</div>'+(s.snippet?'<div style="font-size:10px;color:#999;margin-top:2px;display:-webkit-box;-webkit-line-clamp:2;-webkit-box-orient:vertical;overflow:hidden">'+esc(s.snippet)+'</div>':'')+'</div>';
       h+='</div>';
     });
     h+='<button class="hashtag-rewrite-btn" onclick="rewriteHashtagTopic(\''+esc(topic)+'\')">🤖 Rewrite AI tổng hợp nguồn & đăng tường</button>';
     box.innerHTML=h;
+    // Lazy load images
     sources.forEach(function(s,i){
       if(!s.url)return;
       fetch('/api/article?url='+encodeURIComponent(s.url)).then(function(r){return r.json()}).then(function(d){
@@ -118,19 +170,27 @@ window.showHashtagSources=async function(topic){
   }catch(e){box.innerHTML='<h3>🔍 '+esc(topic)+'</h3><div style="color:#e74c3c;font-size:12px;padding:8px">Lỗi: '+esc(e.message)+'</div>';}
 };
 
+window.rewriteHashtagTopic=async function(topic){
+  var btn=event?.target;if(btn){btn.disabled=true;btn.textContent='Đang tổng hợp...';}
+  try{
+    var r=await fetch('/api/topic_post',{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({topic:topic})});
+    var j=await r.json();if(!r.ok||j.error)throw new Error(j.error||'Lỗi');
+    if(btn)btn.textContent='✅ Đã đăng lên Tường AI!';
+    setTimeout(function(){document.getElementById('hashtag-sources-box')?.remove();},2000);
+  }catch(e){if(btn){btn.disabled=false;btn.textContent='❌ '+e.message;}}
+};
+
 window.createTopicPost=function(){
   var inp=document.getElementById('ai-topic-input');
   var topic=(inp&&inp.value||'').trim();
   if(!topic){alert('Nhập chủ đề trước');return;}
-  showHashtagSources(topic);
-  if(inp)inp.value='';
+  showHashtagSources(topic);if(inp)inp.value='';
 };
 window.createTopicPostFinal5=function(){
   var inp=document.getElementById('ai-topic-input-final5')||document.getElementById('ai-topic-input');
   var topic=(inp&&inp.value||'').trim();
   if(!topic){alert('Nhập chủ đề trước');return;}
-  showHashtagSources(topic);
-  if(inp)inp.value='';
+  showHashtagSources(topic);if(inp)inp.value='';
 };
 })();
 </script>