app.py

import os
import requests
import streamlit as st
from langchain_huggingface import HuggingFaceEndpoint
from langchain_core.prompts import PromptTemplate
from langchain_core.output_parsers import StrOutputParser
from transformers import pipeline  # for Sentiment Analysis
NASA_API_KEY = os.getenv("NASA_API_KEY")

model_id = "mistralai/Mistral-7B-Instruct-v0.3"

# Initialize sentiment analysis pipeline
sentiment_analyzer = pipeline("sentiment-analysis")

def get_llm_hf_inference(model_id=model_id, max_new_tokens=128, temperature=0.1):
    llm = HuggingFaceEndpoint(
        repo_id=model_id,
        max_new_tokens=max_new_tokens,
        temperature=temperature,
        token=os.getenv("HF_TOKEN")  # Hugging Face token from environment variable
    )
    return llm

def get_nasa_apod():
    """
    Fetch the Astronomy Picture of the Day (APOD) from the NASA API.
    """
    url = f"https://api.nasa.gov/planetary/apod?api_key={NASA_API_KEY}"
    response = requests.get(url)
    if response.status_code == 200:
        data = response.json()
        return f"Title: {data['title']}\nExplanation: {data['explanation']}\nURL: {data['url']}"
    else:
        return "I couldn't fetch data from NASA right now. Please try again later."

def analyze_sentiment(user_text):
    """
    Analyzes the sentiment of the user's input to adjust responses.
    """
    result = sentiment_analyzer(user_text)[0]
    sentiment = result['label']
    return sentiment

def predict_action(user_text):
    """
    Predicts actions based on user input (e.g., fetch space info or general knowledge).
    """
    if "NASA" in user_text or "space" in user_text:
        return "nasa_info"
    if "weather" in user_text:
        return "weather_info"
    return "general_query"

def generate_follow_up(user_text):
    """
    Generates a relevant follow-up question based on the user's input.
    """
    prompt_text = (
        f"Given the user's message: '{user_text}', ask one natural follow-up question "
        "that suggests a related topic or offers user the opportunity to go in a new direction."
    )

    hf = get_llm_hf_inference(max_new_tokens=64, temperature=0.7)
    chat = hf.invoke(input=prompt_text)
    
    return chat.strip()

def get_response(system_message, chat_history, user_text, 
                 eos_token_id=['User'], max_new_tokens=256, get_llm_hf_kws={}):
    sentiment = analyze_sentiment(user_text)
    action = predict_action(user_text)

    if action == "nasa_info":
        nasa_response = get_nasa_apod()
        chat_history.append({'role': 'user', 'content': user_text})
        chat_history.append({'role': 'assistant', 'content': nasa_response})
        
        follow_up = generate_follow_up(user_text)
        chat_history.append({'role': 'assistant', 'content': follow_up})
        return f"{nasa_response}\n\n{follow_up}", chat_history

    hf = get_llm_hf_inference(max_new_tokens=max_new_tokens, temperature=0.1)

    prompt = PromptTemplate.from_template(
        (
            "[INST] {system_message}"
            "\nCurrent Conversation:\n{chat_history}\n\n"
            "\nUser: {user_text}.\n [/INST]"
            "\nAI:"
        )
    )
    chat = prompt | hf.bind(skip_prompt=True) | StrOutputParser(output_key='content')
    response = chat.invoke(input=dict(system_message=system_message, user_text=user_text, chat_history=chat_history))
    response = response.split("AI:")[-1]

    chat_history.append({'role': 'user', 'content': user_text})
    chat_history.append({'role': 'assistant', 'content': response})

    # Modify response based on sentiment analysis (e.g., offer help for negative sentiments)
    if sentiment == "NEGATIVE":
        response += "\nI'm sorry to hear that. How can I assist you further?"

    follow_up = generate_follow_up(user_text)
    chat_history.append({'role': 'assistant', 'content': follow_up})

    return f"{response}\n\n{follow_up}", chat_history

# Streamlit setup
st.set_page_config(page_title="HuggingFace ChatBot", page_icon="🤗")
st.title("NASA Personal Assistant")
st.markdown(f"*This chatbot uses {model_id} and NASA's APIs to provide information and responses.*")

# Initialize session state
if "chat_history" not in st.session_state:
    st.session_state.chat_history = [{"role": "assistant", "content": "Hello! How can I assist you today?"}]
    
# Sidebar for settings
if st.sidebar.button("Reset Chat"):
    st.session_state.chat_history = [{"role": "assistant", "content": "Hello! How can I assist you today?"}]

# Main chat interface
user_input = st.chat_input(placeholder="Type your message here...")
if user_input:
    response, st.session_state.chat_history = get_response(
        system_message="You are a helpful AI assistant.",
        user_text=user_input,
        chat_history=st.session_state.chat_history,
        max_new_tokens=128
    )
    # Display messages
    for message in st.session_state.chat_history:
        st.chat_message(message["role"]).write(message["content"])




if st.button("Send"):
    if user_input:
        response, follow_up, st.session_state.chat_history, image_url = get_response(
            system_message="You are a helpful AI assistant.",
            user_text=user_input,
            chat_history=st.session_state.chat_history
        )

        # Display response
        st.markdown(f"<div class='assistant-msg'><strong>HAL:</strong> {response}</div>", unsafe_allow_html=True)

        # Display NASA image if available
        if image_url:
            st.image(image_url, caption="NASA Image of the Day")

        # Follow-up question suggestions
        follow_up_options = [follow_up, "Explain differently", "Give me an example"]
        selected_option = st.radio("What would you like to do next?", follow_up_options)

        if st.button("Continue"):
            if selected_option:
                response, _, st.session_state.chat_history, _ = get_response(
                    system_message="You are a helpful AI assistant.",
                    user_text=selected_option,
                    chat_history=st.session_state.chat_history
                )
                st.markdown(f"<div class='assistant-msg'><strong>HAL:</strong> {response}</div>", unsafe_allow_html=True)