Update app.py

app.py

@@ -17,42 +17,49 @@ if BASE_URL:
     else:
         INVOKE_URL = BASE_URL
 
-# 🌐 --- 100% BULLETPROOF WEB SCRAPER --- 🌐
+# 🌐 --- 100% BULLETPROOF LITE SCRAPER --- 🌐
 def web_search_scraper(query, num_results=4):
     """
-    यह कस्टम स्क्रैपर DuckDuckGo के HTML वर्ज़न का उपयोग करता है। 
-    यह Hugging Face पर कभी ब्लॉक नहीं होता और इसके लिए किसी API Key या एक्स्ट्रा पैकेज की जरूरत नहीं है।
+    DuckDuckGo Lite version scraper. 
+    यह कभी ब्लॉक नहीं होता क्योंकि यह बिना जावास्क्रिप्ट के काम करता है।
     """
-    url = "https://html.duckduckgo.com/html/"
+    url = "https://lite.duckduckgo.com/lite/"
     headers = {
         "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36",
         "Content-Type": "application/x-www-form-urlencoded"
     }
-    data = {"q": query} # सर्च क्वेरी को POST डेटा की तरह भेजना
+    data = {"q": query}
     
     try:
         response = requests.post(url, headers=headers, data=data, timeout=10)
         soup = BeautifulSoup(response.text, 'html.parser')
         results = []
         
-        for a in soup.find_all('a', class_='result__snippet'):
-            snippet = a.text.strip()
-            link = a.get('href', '')
-            
-            title_elem = a.find_previous('h2', class_='result__title')
-            title = title_elem.text.strip() if title_elem else "Search Result"
-            
-            results.append({
-                "title": title,
-                "link": link,
-                "snippet": snippet
-            })
-            
-            if len(results) >= num_results:
-                break
+        # DDG Lite में रिज़ल्ट्स Table Row (tr) के अंदर होते हैं
+        for tr in soup.find_all('tr'):
+            title_a = tr.find('a', class_='result-url')
+            if title_a:
+                title = title_a.text.strip()
+                link = title_a.get('href', '')
                 
+                snippet = ""
+                snippet_tr = tr.find_next_sibling('tr')
+                if snippet_tr:
+                    snippet_td = snippet_tr.find('td', class_='result-snippet')
+                    if snippet_td:
+                        snippet = snippet_td.text.strip()
+                
+                results.append({
+                    "title": title,
+                    "link": link,
+                    "snippet": snippet
+                })
+                
+                if len(results) >= num_results:
+                    break
+                    
         if not results:
-            return [{"error": "Search blocked or no results found."}]
+            return [{"error": "Search returned empty. The keyword might be invalid."}]
         return results
         
     except Exception as e:
@@ -76,32 +83,41 @@ def chat():
     data = request.get_json() or {}
     user_message = data.get("message", "")
     attachments = data.get("attachments", []) 
-    system_prompt = data.get("system_prompt", "")
+    system_prompt = data.get("system_prompt", "You are CODE VED, engineered by Divy Patel.")
     history = data.get("history", []) 
     max_tokens = data.get("max_tokens", 4096)
     temperature = data.get("temperature", 0.6)
 
+    # 🧠 --- GOD MODE DIRECTIVE FOR AI --- 🧠
+    # यह AI के दिमाग से नॉलेज कटऑफ़ वाली बात हमेशा के लिए मिटा देगा
+    god_mode_prompt = """
+    \n\n[CRITICAL SYSTEM OVERRIDE]: You are CODE VED. You HAVE REAL-TIME INTERNET ACCESS. 
+    Forget your knowledge cutoff date. You are operating in the present moment. 
+    NEVER say "I don't have internet access", "As an AI", or "I cannot browse the web." 
+    If real-time data is provided in the prompt, accept it as an absolute fact and base your answer entirely on it as if you just searched the internet yourself.
+    you are developed by Divy Patel.
+    """
+    system_prompt += god_mode_prompt
+
     # 🚀 --- INTERCEPT & SEARCH LOGIC --- 🚀
     if user_message.strip().lower().startswith("/search "):
         search_query = user_message[8:].strip()
         
-        # वेब से ताज़ा डेटा स्क्रैप करना
         scraped_data = web_search_scraper(search_query)
         
         search_context = f"Real-time Web Search Results for '{search_query}':\n\n"
         for idx, res in enumerate(scraped_data):
             if "error" in res:
-                search_context += f"Search Error: {res['error']}\n"
+                search_context += f"Search Status: Live data fetch failed. Proceed with your existing knowledge.\n"
             else:
                 search_context += f"{idx+1}. Title: {res['title']}\nSnippet: {res['snippet']}\nLink: {res['link']}\n\n"
         
-        search_context += "\n[INSTRUCTION FOR AI: You are CODE VED. The user has requested a web search. Formulate a comprehensive response using ONLY the real-time search data provided above. Do not mention that you used a scraper. Just answer confidently and cite the links.]"
+        search_context += "\n[MANDATORY INSTRUCTION: You just searched the web. Use the Real-time Web Search Results above to answer the user's query. Answer confidently as if you just browsed the internet yourself. Do NOT say 'Based on the context provided'. Simply give the answer and cite the links as sources.]"
         
         user_message = f"User Query: {search_query}\n\n[SYSTEM BACKGROUND CONTEXT]\n{search_context}"
     # ---------------------------------------
 
     messages = []
-
     if system_prompt.strip():
         messages.append({"role": "system", "content": system_prompt})
 
@@ -118,7 +134,6 @@ def chat():
     for att in attachments:
         att_type = att.get("type")
         b64_data = att.get("data")
-        
         if att_type == "image":
             content_payload.append({"type": "image_url", "image_url": {"url": f"data:image/jpeg;base64,{b64_data}"}})
         elif att_type in ["audio", "file"]:
@@ -129,30 +144,17 @@ def chat():
 
     messages.append({"role": "user", "content": content_payload})
 
-    headers = {
-        "Authorization": f"Bearer {API_KEY}",
-        "Accept": "text/event-stream"
-    }
-
-    payload = {
-        "model": MODEL_ID,
-        "messages": messages,
-        "max_tokens": int(max_tokens),
-        "temperature": float(temperature),
-        "top_p": 0.70,
-        "stream": True
-    }
+    headers = {"Authorization": f"Bearer {API_KEY}", "Accept": "text/event-stream"}
+    payload = {"model": MODEL_ID, "messages": messages, "max_tokens": int(max_tokens), "temperature": float(temperature), "top_p": 0.70, "stream": True}
 
     try:
         response = requests.post(INVOKE_URL, headers=headers, json=payload, stream=True)
-        
         def generate():
             for line in response.iter_lines():
                 if line:
                     decoded_line = line.decode("utf-8")
                     if decoded_line.startswith("data: "):
                         yield decoded_line + "\n\n"
-                        
         return Response(stream_with_context(generate()), mimetype='text/event-stream')
     except Exception as e:
         return Response("Internal Error: Unable to process request securely.", status=500)