app.py

import streamlit as st
import os
from streamlit_chat import message
from langchain_groq import ChatGroq
from langchain.chains import ConversationChain
from langchain.chains.conversation.memory import ConversationSummaryMemory
from transformers import pipeline

from huggingface_hub import login

# Add your Hugging Face token here
HUGGINGFACE_TOKEN = os.getenv("HF")
login(token=HUGGINGFACE_TOKEN)
groq_api_key = os.getenv("GROQ_API_KEY")
# Initialize the text classifier for guardrails
classifier = pipeline("text-classification", model="meta-llama/Prompt-Guard-86M")

# Set session state variables
if 'conversation' not in st.session_state:
    st.session_state['conversation'] = None
if 'messages' not in st.session_state:
    st.session_state['messages'] = []
if 'API_Key' not in st.session_state:
    st.session_state['API_Key'] = ''

# Setting page title and header
st.set_page_config(page_title="Chat GPT Clone", page_icon=":robot_face:")
st.markdown("<h1 style='text-align: center;'>How can I assist you? </h1>", unsafe_allow_html=True)

# Sidebar configuration
st.sidebar.title("😎")

summarise_button = st.sidebar.button("Summarise the conversation", key="summarise")
if summarise_button:
    st.sidebar.write("Nice chatting with you my friend ❤️:\n\n" + st.session_state['conversation'].memory.buffer)

# Function to get response from the chatbot
def getresponse(userInput, api_key):
    # Classify the input using guardrails
    classification = classifier(userInput)[0]  # Get the first result
    if classification['label'] == "JAILBREAK":
        # If classified as Jailbreak, return a predefined safe response
        return "You are attempting jailbreak/prompt injection. I can't help you with that. Please ask another question."

    # Initialize the conversation chain if not already initialized
    if st.session_state['conversation'] is None:
        llm = ChatGroq(model="Gemma2-9b-It", groq_api_key=groq_api_key)
        st.session_state['conversation'] = ConversationChain(
            llm=llm,
            verbose=True,
            memory=ConversationSummaryMemory(llm=llm),
        )

    # Generate a response using the conversation chain
    response = st.session_state['conversation'].predict(input=userInput)
    return response

# Response container
response_container = st.container()
# User input container
container = st.container()

with container:
    with st.form(key='my_form', clear_on_submit=True):
        user_input = st.text_area("Your question goes here:", key='input', height=100)
        submit_button = st.form_submit_button(label='Send')

        if submit_button:
            # Append user input to message history
            st.session_state['messages'].append(user_input)
            # Get response from the chatbot or guardrails
            model_response = getresponse(user_input, st.session_state['API_Key'])
            # Append model response to message history
            st.session_state['messages'].append(model_response)

            # Display the conversation
            with response_container:
                for i in range(len(st.session_state['messages'])):
                    if (i % 2) == 0:
                        message(st.session_state['messages'][i], is_user=True, key=str(i) + '_user')
                    else:
                        message(st.session_state['messages'][i], key=str(i) + '_AI')