Update app.py

app.py

@@ -6,6 +6,8 @@ from langchain_community.vectorstores import Chroma
 from langchain_community.embeddings import HuggingFaceEmbeddings
 from langchain_groq import ChatGroq
 from langchain_community.document_loaders import PyPDFLoader
+from langchain.memory import ConversationBufferMemory
+from langchain.chains import ConversationalRetrievalChain
 
 # Configuration
 HF_REPO_ID = "Shami96/7solar-documentation"
@@ -36,35 +38,46 @@ def initialize_components():
     )
     chunks = text_splitter.split_documents(documents)
     embeddings = HuggingFaceEmbeddings(model_name="all-MiniLM-L6-v2")
-    return Chroma.from_documents(chunks, embeddings)
+    vectorstore = Chroma.from_documents(chunks, embeddings)
+    
+    # Initialize LLM
+    llm = ChatGroq(
+        model_name="llama3-70b-8192",
+        temperature=0.3
+    )
+    
+    # Create conversation memory
+    memory = ConversationBufferMemory(
+        memory_key="chat_history",
+        return_messages=True
+    )
+    
+    # Create retrieval chain with memory
+    qa_chain = ConversationalRetrievalChain.from_llm(
+        llm=llm,
+        retriever=vectorstore.as_retriever(),
+        memory=memory,
+        chain_type="stuff"
+    )
+    
+    return qa_chain
 
 # Chat function
 def respond(message, history):
     try:
         # Initialize if not already done
-        if 'vector_db' not in globals():
-            global vector_db
-            vector_db = initialize_components()
+        if 'qa_chain' not in globals():
+            global qa_chain
+            qa_chain = initialize_components()
         
         # Handle greetings
         if message.lower() in ["hi", "hello", "hey"]:
             return "Hello! I'm your 7Solar assistant. How can I help you today?"
         
-        # Search documents
-        docs = vector_db.similarity_search(message, k=3)
-        if not docs:
-            return "I couldn't find relevant information. Please try another question about 7Solar."
-        
-        # Generate response
-        llm = ChatGroq(
-            model_name="llama3-70b-8192",
-            temperature=0.3
-        )
-        context = "\n\n".join([doc.page_content for doc in docs])
-        response = llm.invoke(
-            f"Using only this context:\n{context}\n\nQuestion: {message}\nAnswer:"
-        )
-        return response.content
+        # Get response with conversation context
+        result = qa_chain({"question": message})
+        return result["answer"]
+    
     except Exception as e:
         return f"An error occurred: {str(e)}"