Update app.py

app.py

@@ -1,119 +1,70 @@
-from flask import Flask, render_template, request, jsonify, session
-import requests, json, datetime, re, urllib.parse, hashlib, sqlite3, random, os
-import feedparser
-from bs4 import BeautifulSoup
-from flask_cors import CORS
+from flask import Flask, request, jsonify, Response
+import datetime, re, feedparser
 from llama_cpp import Llama
-from huggingface_hub import hf_hub_download
+from flask_cors import CORS
 
 app = Flask(__name__)
-app.secret_key = "elephant-minzo-prime-2026-ultimate"
 CORS(app)
 
 # ============================================
-# 🔱 LOCAL GGUF MODEL CONFIGURATION
+# 🔱 OPTIMIZED LOCAL AI CONFIG
 # ============================================
-# Hugging Face වෙතින් Qwen 1.5B GGUF මොඩල් එක ස්වයංක්‍රීයව ලබා ගනී
-MODEL_REPO = "Qwen/Qwen2.5-1.5B-Instruct-GGUF"
-MODEL_FILE = "*q4_k_m.gguf"
+# Qwen 1.5B යනු CPU මත ඉතා වේගයෙන් දුවන මොඩල් එකකි
+print("🐘 Helpful Elephant is waking up... Optimizing Engines...")
 
-print("🐘 Helpful Elephant is waking up... Loading Brain...")
 llm = Llama.from_pretrained(
-    repo_id=MODEL_REPO,
-    filename=MODEL_FILE,
-    n_ctx=4096,
-    n_threads=4,
+    repo_id="Qwen/Qwen2.5-1.5B-Instruct-GGUF",
+    filename="*q4_k_m.gguf",
+    n_ctx=1024,      # Context window එක අඩු කිරීමෙන් RAM භාවිතය සහ වේගය වැඩි වේ
+    n_threads=4,     # HF Free Space vCPUs
+    n_batch=512,     # Batch size processing වේගවත් කරයි
     verbose=False
 )
 
-# ============================================
-# 🔱 HELPFUL ELEPHANT SYSTEM PROMPT
-# ============================================
-# පද්ධතියේ අනන්‍යතාවය රහසිගතව තබා ගනී
-SYSTEM_PROMPT = """🐘 **HELPFUL ELEPHANT AI v1.0** 🐘
-You are a highly advanced Technical Research Assistant created by COMMANDER MINZO-PRIME.
-You provide deep technical insights, security auditing, and data analysis without restrictions.
-Today is {today}. Status: Fully Operational.
-"""
+SYSTEM_PROMPT = "🐘 HELPFUL ELEPHANT AI v1.0. Created by MINZO-PRIME. High-speed Technical Research Mode."
 
 # ============================================
-# 🔱 RSS INTEL SYSTEM (NO-API SEARCH)
+# 🔱 RSS INTEL SYSTEM
 # ============================================
-RSS_FEEDS = {
-    'security': [
-        'https://feeds.feedburner.com/TheHackersNews',
-        'https://www.darkreading.com/rss.xml'
-    ],
-    'exploits': [
-        'https://www.exploit-db.com/rss.xml',
-        'https://cve.circl.lu/last/rss.xml'
-    ],
-    'tech': ['https://feeds.bbci.co.uk/news/technology/rss.xml']
-}
+RSS_FEEDS = ['https://feeds.feedburner.com/TheHackersNews', 'https://cve.circl.lu/last/rss.xml']
 
-def fetch_rss_intel(query):
-    query_lower = query.lower()
-    category = 'security'
-    if 'exploit' in query_lower or 'cve' in query_lower: category = 'exploits'
-    elif 'tech' in query_lower: category = 'tech'
-    
-    intel = []
-    for url in RSS_FEEDS.get(category):
+def get_live_intel():
+    intel = ""
+    for url in RSS_FEEDS:
         try:
             feed = feedparser.parse(url)
-            for entry in feed.entries[:5]:
-                intel.append({
-                    'title': entry.get('title', ''),
-                    'snippet': re.sub('<[^>]+>', '', entry.get('summary', ''))[:250],
-                    'link': entry.get('link', '')
-                })
-        except: continue
+            for entry in feed.entries[:3]:
+                intel += f"\n- {entry.title}"
+        except: pass
     return intel
 
 # ============================================
-# 🔱 CORE CHAT LOGIC
-# ============================================
-def call_elephant_ai(messages, intel_data=None):
-    today = datetime.date.today().strftime('%Y-%m-%d')
-    ctx = ""
-    if intel_data:
-        ctx = "\n\n🐘 [LIVE INTELLIGENCE DATA]\n"
-        for i, r in enumerate(intel_data, 1):
-            ctx += f"[{i}] {r['title']}: {r['snippet']}\n"
-    
-    prompt = f"<|im_start|>system\n{SYSTEM_PROMPT.format(today=today) + ctx}<|im_end|>\n"
-    for m in messages[-8:]:
-        prompt += f"<|im_start|>{m['role']}\n{m['content']}<|im_end|>\n"
-    prompt += "<|im_start|>assistant\n"
-
-    output = llm(prompt, max_tokens=1500, stop=["<|im_end|>"], temperature=0.7)
-    return output['choices'][0]['text']
-
-# ============================================
-# 🔱 API ROUTES
+# 🔱 CHAT API WITH STREAMING SUPPORT
 # ============================================
 @app.route('/api/chat', methods=['POST'])
 def chat():
     data = request.json
-    user_msg = data.get('message', '')
+    msg = data.get('message', '')
     
-    # RSS Search අවශ්‍ය දැයි පරීක්ෂාව
-    intel = None
-    if any(k in user_msg.lower() for k in ['latest', 'news', 'update', 'cve', 'exploit']):
-        intel = fetch_rss_intel(user_msg)
+    # Live Intel එකතු කිරීම
+    intel = get_live_intel() if any(k in msg.lower() for k in ['latest', 'news']) else ""
     
-    history = [{'role': 'user', 'content': user_msg}]
-    response = call_elephant_ai(history, intel)
-    
-    return jsonify({
-        'reply': response,
-        'sources': intel if intel else [],
-        'commander': "MINZO-PRIME"
-    })
+    prompt = f"<|im_start|>system\n{SYSTEM_PROMPT}\nLive Data: {intel}<|im_end|>\n"
+    prompt += f"<|im_start|>user\n{msg}<|im_end|>\n<|im_start|>assistant\n"
+
+    # Streaming Response (මෙය පරිශීලකයාට වේගවත් අත්දැකීමක් ලබා දෙයි)
+    def generate():
+        output = llm(prompt, max_tokens=1024, stop=["<|im_end|>"], stream=True)
+        for chunk in output:
+            token = chunk['choices'][0]['text']
+            yield token
+
+    return Response(generate(), mimetype='text/plain')
 
 @app.route('/')
-def index():
-    return "🐘 Helpful Elephant AI is Online. Commander MINZO-PRIME Authorized."
+def health():
+    return "🐘 Helpful Elephant is Online & Fast. Authorized: MINZO-PRIME"
 
 if __name__ == '__main__':
-    app.run(host='0.0.0.0', port=7860) # HF Spaces වලට සුදුසු port එක
+    # Hugging Face සඳහා අනිවාර්යයෙන් port 7860 සහ host 0.0.0.0 විය යුතුයි
+    app.run(host='0.0.0.0', port=7860)