Update app.py

app.py

@@ -1,66 +1,44 @@
 import os
 from flask import Flask, render_template, request, jsonify
-from flask_cors import CORS
+import chromadb
 from huggingface_hub import InferenceClient
-from duckduckgo_search import DDGS # Direct import for stability
+from sentence_transformers import SentenceTransformer
 
 app = Flask(__name__)
-CORS(app)
 
-client = InferenceClient(
-    model="meta-llama/Llama-3.2-3B-Instruct",
-    token=os.getenv("HF_TOKEN")
-)
+client = InferenceClient(model="meta-llama/Llama-3.2-3B-Instruct", token=os.getenv("HF_TOKEN"))
 
-def get_live_data(query):
-    """Fetches real-time snippets from the web."""
-    try:
-        with DDGS() as ddgs:
-            results = [r['body'] for r in ddgs.text(query, max_results=3)]
-            return "\n".join(results)
-    except Exception:
-        return "No real-time data available at the moment."
+embed_model = SentenceTransformer("all-MiniLM-L6-v2")
+db_client = chromadb.PersistentClient(path="./my_vector_db")
+collection = db_client.get_or_create_collection(name="research_docs")
 
-@app.route('/')
-def index():
-    return render_template('index.html')
+def add_to_knowledge_base(text_list):
+    embeddings = embed_model.encode(text_list).tolist()
+    ids = [f"id_{i}" for i in range(len(text_list))]
+    collection.add(documents=text_list, embeddings=embeddings, ids=ids)
+
+initial_data = [
+    "The 2026 World AI Summit took place in Tokyo, focusing on decentralized models.",
+    "Nexus AI Research OS was launched in January 2026 as a browser-based tool.",
+]
+add_to_knowledge_base(initial_data)
 
 @app.route('/ask', methods=['POST'])
 def ask():
-    try:
-        data = request.get_json()
-        user_query = data.get("query")
-
-        # 1. Get real-time info
-        web_context = get_live_data(user_query)
-
-        # 2. Build the 'Augmented' Prompt
-        system_prompt = f"""You are a helpful research assistant. 
-        Use the following real-time web context to answer the user's request. 
-        If the context is empty, rely on your internal knowledge.
-        
-        WEB CONTEXT:
-        {web_context}
-        """
-
-        # 3. Call Llama
-        response = ""
-        for message in client.chat_completion(
-            messages=[
-                {"role": "system", "content": system_prompt},
-                {"role": "user", "content": user_query}
-            ],
-            max_tokens=800,
-            stream=True,
-        ):
-            token = message.choices[0].delta.content
-            if token:
-                response += token
-
-        return jsonify({"answer": response})
-
-    except Exception as e:
-        return jsonify({"answer": f"System Error: {str(e)}"}), 500
+    user_query = request.get_json().get("query")
+    
+    query_vec = embed_model.encode([user_query]).tolist()
+    results = collection.query(query_embeddings=query_vec, n_results=2)
+    retrieved_text = "\n".join(results['documents'][0])
+
+    prompt = f"Use this data to answer: {retrieved_text}\n\nQuestion: {user_query}"
+    
+    response = client.chat_completion(
+        messages=[{"role": "user", "content": prompt}],
+        max_tokens=500
+    )
+    
+    return jsonify({"answer": response.choices[0].message.content})
 
 if __name__ == "__main__":
     app.run(host="0.0.0.0", port=7860)