Update app.py

app.py

@@ -45,22 +45,44 @@ llm_name = "gpt-3.5-turbo-0301"
 vectordb = initialize.initialize()
 
 
-history = "Hello !"   
+
+chat_history = []
+
+# For getting source documents
+def get_file(source_documents):
+  files = set()
+  for doc in source_documents:
+    file = os.path.basename(doc.metadata['source'])
+    files.add(file)
+    # Print unique filenames
+  return list(set(files))
+
 
 def chat_query(question, history):
 
-    query = f"As an experienced Electrical Engineer, answer pointwise to the Question : {question} and consider your previous reply as : {history[:-1]}, if any."
+    
+    query = f"""As an experienced Electrical Engineer, please provide an elaborate, precise, and pointwise answer politely to the question: {question}. 
+              Also, Please consider the provided chat history: {history[:-1]}
+              Ensure that your response is detailed, accurate, and addresses each aspect of the question thoroughly. 
+              If the context of the question doesn't align with the chat history, please provide your response in a fresh manner. """
+
     #llm = ChatOpenAI(model=llm_name, temperature=0.1, api_key = OPENAI_API_KEY)
     llm = GoogleGenerativeAI(model="gemini-1.0-pro", google_api_key=GEMINI_API_KEY)
 
     # Conversation Retrival Chain with Memory
-    memory = ConversationBufferMemory(memory_key="chat_history", return_messages=True)
+    #memory = ConversationBufferMemory(memory_key="chat_history", return_messages=True)
     retriever=vectordb.as_retriever()
-    qa = ConversationalRetrievalChain.from_llm(llm, retriever=retriever, memory=memory)
+    qa = ConversationalRetrievalChain.from_llm(llm, retriever=retriever, return_source_documents=True)
 
     # Replace input() with question variable for Gradio
-    result = qa({"question": query})
-    return result['answer']
+    result = qa({"question": query, "chat_history" : history[:-1]})
+
+    
+    source_docs = result["source_documents"]
+    file_name = get_file(source_docs)[0]
+
+
+    return result["answer"] + "\n\nSource : " + file_name
 
     # Chatbot only answers based on Documents
     # qa = VectorDBQA.from_chain_type(llm=OpenAI(openai_api_key = OPENAI_API_KEY, ), chain_type="stuff", vectorstore=vectordb)