chatbot history added

llm_setup/llm_setup.py

@@ -44,17 +44,23 @@ class LLMService:
         Initializes the conversational RAG chain.
         """
         # Initialize RAG (Retrieval-Augmented Generation) chain
-        prompt = ChatPromptTemplate(input_variables=['context', 'question'], messages=[HumanMessagePromptTemplate(
-            prompt=PromptTemplate(input_variables=['context', 'question'], template=self.system_prompt))])
+        prompt = ChatPromptTemplate(input_variables=['history','context', 'question'], messages=[HumanMessagePromptTemplate(
+            prompt=PromptTemplate(input_variables=['history','context', 'question'], template="{history}\nContext: {context}\nQuestion: {question}"))])
 
         # Initialize conversational RAG chain
         self._conversational_rag_chain = (
-                {"context": self._web_retriever | format_documents, "question": RunnablePassthrough()}
+                {"context": self._web_retriever | format_documents, "question": RunnablePassthrough(), "history":self.get_chat_history}
                 | prompt
                 | self.llm
                 | StrOutputParser()
         )
 
+    def get_chat_history(self):
+        """
+        Retrieves the last 3 chat messages formatted as a string.
+        """
+        return "\n".join(self.chat_history) if self.chat_history else "No prior conversation."    
+
     def conversational_rag_chain(self):
         """
         Returns the initialized conversational RAG chain.
@@ -64,6 +70,20 @@ class LLMService:
         """
         return self._conversational_rag_chain
 
+    def update_chat_history(self, user_input: str, llm_response: str):
+        """
+        Updates the chat history with the latest question and response.
+        """
+        self.chat_history.append(f"User: {user_input}\nAI: {llm_response}")
+
+    def ask_question(self, question: str):
+        """
+        Processes a user question using the conversational RAG chain and updates history.
+        """
+        response = self._conversational_rag_chain.invoke(question)
+        self.update_chat_history(question, response)
+        return response
+    
     def get_llm(self) -> ChatGoogleGenerativeAI:
         """
         Returns the LLM instance.