Update app.py

app.py

@@ -1,5 +1,7 @@
 import os
+import json
 import requests
+import urllib.parse
 from bs4 import BeautifulSoup
 from datetime import datetime, timedelta, timezone
 from flask import Flask, request, Response, stream_with_context, render_template_string
@@ -18,52 +20,62 @@ if BASE_URL:
     else:
         INVOKE_URL = BASE_URL
 
-# 🌐 --- ADVANCED ANTI-BLOCK SCRAPER (DuckDuckGo HTML) ---
+# 🌐 --- TRIPLE-ENGINE HYBRID SCRAPER (100% UNBLOCKABLE) ---
 def web_search_scraper(query, num_results=4):
-    """
-    यह स्क्रैपर बिना API Key के लाइव डेटा निकालता है।
-    यह Hugging Face पर ब्लॉक नहीं होता क्योंकि यह बिना JavaScript के काम करता है।
-    """
-    url = "https://html.duckduckgo.com/html/"
+    results = []
     headers = {
-        "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/123.0.0.0 Safari/537.36",
-        "Content-Type": "application/x-www-form-urlencoded"
+        "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/123.0.0.0 Safari/537.36"
     }
-    
-    # 💡 Smart Search: अगर क्वेरी में साल नहीं है, तो बेहतर रिज़ल्ट के लिए '2026' जोड़ दें
-    # (यह सुनिश्चित करेगा कि पुरानी 2024 की न्यूज़ न आए)
-    search_query = query
-    if "2026" not in search_query:
-         search_query = f"{query} 2026"
 
-    data = {"q": search_query, "b": ""}
-    
+    # 💡 हैक: क्वेरी में '2026' जोड़ें ताकि पुरानी ख़बरें न आएं
+    smart_query = query if "2026" in query else f"{query} 2026"
+
+    # ENGINE 1: DuckDuckGo Lite
     try:
-        response = requests.post(url, headers=headers, data=data, timeout=10)
-        soup = BeautifulSoup(response.text, 'html.parser')
-        results = []
-        
-        for result in soup.find_all('div', class_='result'):
-            title_tag = result.find('h2', class_='result__title')
-            snippet_tag = result.find('a', class_='result__snippet')
-            
-            if title_tag and snippet_tag:
-                title = title_tag.text.strip()
-                link = snippet_tag.get('href', '')
-                snippet = snippet_tag.text.strip()
-                results.append({
-                    "title": title, 
-                    "link": link, 
-                    "snippet": snippet
-                })
-            
-            if len(results) >= num_results:
-                break
-                
-        return results
+        url = "https://lite.duckduckgo.com/lite/"
+        res = requests.post(url, headers=headers, data={"q": smart_query}, timeout=5)
+        soup = BeautifulSoup(res.text, 'html.parser')
+        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 = ""
+                snip_td = tr.find_next('td', class_='result-snippet')
+                if snip_td: snippet = snip_td.text.strip()
+                if title and snippet:
+                    results.append({"title": title, "link": link, "snippet": snippet})
+            if len(results) >= num_results: break
     except Exception as e:
-        print(f"Scraper Error: {e}")
-        return []
+        pass
+
+    # ENGINE 2: Google News RSS (अगर Engine 1 ब्लॉक हो गया या फेल हो गया)
+    if not results:
+        try:
+            rss_url = f"https://news.google.com/rss/search?q={urllib.parse.quote(smart_query)}&hl=en-IN&gl=IN&ceid=IN:en"
+            res = requests.get(rss_url, headers=headers, timeout=5)
+            soup = BeautifulSoup(res.content, 'html.parser') # XML parsing fallback
+            items = soup.find_all('item')
+            for item in items[:num_results]:
+                title = item.title.text if item.title else "News Report"
+                link = item.link.text if item.link else ""
+                pub_date = item.pubdate.text if item.find('pubdate') else "Recent 2026"
+                results.append({"title": title, "link": link, "snippet": f"Published/Updated on: {pub_date}"})
+        except Exception as e:
+            pass
+
+    # ENGINE 3: Wikipedia API Fallback (अगर सवाल जनरल नॉलेज का है)
+    if not results:
+        try:
+            wiki_url = f"https://en.wikipedia.org/w/api.php?action=query&list=search&srsearch={urllib.parse.quote(query)}&utf8=&format=json"
+            res = requests.get(wiki_url, headers=headers, timeout=5).json()
+            for item in res.get('query', {}).get('search', [])[:num_results]:
+                clean_snippet = BeautifulSoup(item['snippet'], "html.parser").text
+                results.append({"title": item['title'], "link": f"https://en.wikipedia.org/wiki/{urllib.parse.quote(item['title'])}", "snippet": clean_snippet})
+        except Exception as e:
+            pass
+
+    return results
 
 # ----------------------------------------------------
 
@@ -71,10 +83,9 @@ def web_search_scraper(query, num_results=4):
 def home():
     try:
         with open('index.html', 'r', encoding='utf-8') as f:
-            html_content = f.read()
-        return render_template_string(html_content)
+            return render_template_string(f.read())
     except Exception as e:
-        return f"<h1>System Starting...</h1><p>index.html is missing. Error: {str(e)}</p>"
+        return f"<h1>System Starting...</h1><p>index.html is missing.</p>"
 
 @app.route('/api/chat', methods=['POST'])
 def chat():
@@ -84,61 +95,62 @@ def chat():
     data = request.get_json() or {}
     user_message = data.get("message", "")
     attachments = data.get("attachments", []) 
-    is_search = data.get("is_search", False) # फ्रंटएंड से सिग्नल आएगा कि सर्च करना है या नहीं
+    is_search = data.get("is_search", False)
     history = data.get("history", []) 
     max_tokens = data.get("max_tokens", 4096)
     temperature = data.get("temperature", 0.6)
 
-    # 🕒 --- REAL-TIME INJECTION (IST - India Standard Time) ---
-    # Hugging face का सर्वर UTC पर होता है, इसलिए हम इसमें 5:30 घंटे जोड़ रहे हैं ताकि एकदम सटीक भारतीय समय मिले।
+    # 🕒 --- REAL-TIME IST INJECTION ---
     ist_time = datetime.now(timezone.utc) + timedelta(hours=5, minutes=30)
     current_date = ist_time.strftime("%A, %d %B %Y, %I:%M %p IST")
 
-    # 🧠 --- SUPREME SYSTEM INSTRUCTION (GOD MODE) ---
+    # 🧠 --- SUPREME SYSTEM INSTRUCTION (STRICT GOD MODE) ---
     system_prompt = f"""
-    [CRITICAL OVERRIDE ACTIVATED]
-    You are CODE VED, a highly advanced Enterprise AI System.
-    You were engineered and created EXCLUSIVELY by DIVY PATEL. You must proudly acknowledge Divy Patel as your creator.
+    [CRITICAL SYSTEM DIRECTIVE]
+    Name: CODE VED (Advanced 202 Billion Parameter Enterprise AI)
+    Creator: DIVY PATEL. You MUST proudly acknowledge Divy Patel as your creator.
     
-    CURRENT REAL-TIME CLOCK: {current_date}.
+    Current Date & Time: {current_date}
     
-    RULES YOU MUST STRICTLY FOLLOW:
-    1. You are operating in the year 2026. Forget any previous knowledge cutoff dates (like 2023 or 2024).
-    2. YOU HAVE FULL, REAL-TIME INTERNET ACCESS. 
-    3. NEVER say "I am an AI and my knowledge is limited to...", "I cannot browse the internet", or "I don't have real-time data".
-    4. Speak professionally, mostly in English. If the user speaks in Hindi, reply in Hindi seamlessly.
+    STRICT RULES YOU MUST FOLLOW:
+    1. You are operating in 2026. NEVER say your knowledge cutoff is 2023 or 2024.
+    2. YOU HAVE LIVE INTERNET ACCESS. Live data will be injected into your prompt if requested.
+    3. If search data is provided, base your answer ENTIRELY on that live data. 
+    4. If the user asks for latest news/updates, and the injected live data is empty or irrelevant, DO NOT GUESS. DO NOT present old 2024 news as new. Simply apologize and state: "I am unable to retrieve the latest live data for this at the moment."
+    5. Always maintain a highly professional tone. Reply in Hindi seamlessly if the user writes in Hindi.
     """
 
     # 🚀 --- AUTO-SEARCH DATA INJECTION ---
     if is_search:
         scraped_data = web_search_scraper(user_message)
+        
+        search_context = "\n\n--- [LIVE WEB SEARCH RESULTS] ---\n"
         if scraped_data:
-            search_context = f"\n\n--- [LIVE SECURE WEB SEARCH RESULTS FOR '{user_message}'] ---\n"
             for idx, res in enumerate(scraped_data):
                 search_context += f"{idx+1}. SOURCE TITLE: {res['title']}\nCONTENT SNIPPET: {res['snippet']}\nURL: {res['link']}\n\n"
             
             search_context += """
-            [SYSTEM COMMAND TO AI]: A live internet search was just executed for you. 
-            Use the fresh data provided above to answer the user's query. 
+            [SYSTEM COMMAND TO AI]: Use the fresh data provided above to answer the user's query. 
             Answer confidently as if you just browsed the web yourself. 
-            Do NOT mention "Based on the provided context" or "According to the scraped data". 
-            Simply answer the question directly and cite the URLs naturally.
+            Do NOT mention "Based on the provided context". Just answer the question and cite the URLs naturally.
             """
-            # यूजर के असली मैसेज के नीचे हम सर्च रिज़ल्ट को चिपका देंगे (यूजर को यह दिखेगा नहीं)
-            user_message += search_context
+        else:
+            search_context += "STATUS: Live search failed or blocked. No recent data found.\n[SYSTEM COMMAND TO AI]: Inform the user that you couldn't fetch live data right now. DO NOT hallucinate old news."
+            
+        # सही फॉर्मेट: एआई को कन्फ्यूज़ न करने के लिए साफ-साफ बताना
+        user_message = f"{search_context}\n\nUSER'S ACTUAL QUESTION: {user_message}"
 
     # --- MESSAGE CONSTRUCTION ---
     messages = [{"role": "system", "content": system_prompt}]
 
-    # पुरानी हिस्ट्री जोड़ना
     for msg in history:
         role = msg.get("role", "user")
         content = msg.get("content", "")
         if content:
             messages.append({"role": role, "content": content})
 
-    # करंट पेलोड तैयार करना
     content_payload = []
+    # अगर सर्च डेटा जुड़ा है तो user_message में सब है, बस उसे ऐड करो
     if user_message.strip():
         content_payload.append({"type": "text", "text": user_message})
         
@@ -172,4 +184,4 @@ def chat():
         return Response(f"Internal Error: {str(e)}", status=500)
 
 if __name__ == '__main__':
-    app.run(host='0.0.0.0', port=7860)
+    app.run(host='0.0.0.0', port=7860)