import streamlit as st
import os
from langchain_huggingface import HuggingFaceEndpoint,HuggingFacePipeline,ChatHuggingFace

# Secure token loading
hf_token = os.getenv("KEYS")
if not hf_token:
    raise ValueError("Environment variable 'KEYS' is not set.")
os.environ["HF_TOKEN"] = hf_token
os.environ["HUGGINGFACEHUB_API_KEY"] = hf_token

# Custom CSS
custom_css = """
<style>
    html, body, [data-testid="stAppViewContainer"] {
        background: url('https://wallpapercave.com/wp/wp2581376.jpg') no-repeat center center fixed;
        background-size: cover;
        font-family: Arial, sans-serif;
        color: #f5f5f5;
    }

    /* Optional overlay for readability */
    [data-testid="stAppViewContainer"]::before {
        content: "";
        position: fixed;
        top: 0;
        left: 0;
        width: 100%;
        height: 100%;
        background: rgba(0, 0, 0, 0.4); /* dark overlay */
        z-index: -1;
    }

    .button {
        display: inline-block;
        padding: 10px 20px;
        margin: 10px;
        font-size: 16px;
        text-decoration: none;
        color: white;
        border-radius: 5px;
        background-color:#E1D5FC;
        transition: background-color 0.3s ease;
    }

    .button:hover {
        background-color: #e68b7f;
    }

    .chat-reply {
            color: #00ffcc;
            background-color: #1e1e1e;
            padding: 10px;
            border-radius: 8px;
            margin-bottom: 10px;
        }
</style>
"""


st.markdown(custom_css, unsafe_allow_html=True)

mentoring = """
<div>
    <a href="?page=python" class="button">Python</a>
    <a href="?page=eda" class="button">Data Analysis</a>
    <a href="?page=stats" class="button">Statistics</a>
    <a href="?page=ml" class="button">Machine Learning</a>
    <a href="?page=dl" class="button">Deep Learning</a>
    <a href="?page=excel" class="button">MS Excel</a>
    <a href="?page=power_bi" class="button">Power BI</a>
</div>
"""
llama_model = HuggingFaceEndpoint(repo_id = "meta-llama/Llama-3.1-8B-Instruct",provider = 'novita',temperature = 0.7,max_new_tokens = 10,task="conversational")
llama = ChatHuggingFace(llm = llama_model,repo_id = "meta-llama/Llama-3.1-8B-Instruct",provider = 'novita',temperature = 0.7,max_new_tokens = 10,task="conversational")

deepseek_model = HuggingFaceEndpoint(repo_id = "deepseek-ai/DeepSeek-R1",provider = 'novita',temperature = 0.7,max_new_tokens = 10,task="conversational")
deepseek = ChatHuggingFace(llm = deepseek_model,repo_id = "deepseek-ai/DeepSeek-R1",provider = 'novita',temperature = 0.7,max_new_tokens = 10,task="conversational")


from langchain.prompts import ChatPromptTemplate,SystemMessagePromptTemplate,HumanMessagePromptTemplate
# Page definitions
def main_page():
    st.title("Welcome To Innomatics AI Mentor Support")
    st.write("Here is the list of modules we offer to support you in your learning journey...")
    st.markdown(mentoring, unsafe_allow_html=True)

def python_page(): 
    st.title("Hi, Have doubts in Python? 🤔 Let's solve them together 😎🤩")

    exps = st.number_input("Choose a mentor based on their experience: 0-20 years")
    doubts = st.text_input("Explain your doubt")

    if doubts:
        # Set up the prompt
        messages = [
            SystemMessagePromptTemplate.from_template("You have an experience of {exp} years in Python."),
            HumanMessagePromptTemplate.from_template("Provide a clear solution for the following doubt: {doubt}")
        ]

        prompt = ChatPromptTemplate.from_messages(messages)
        formatted_prompt = prompt.format(exp=exps, doubt=doubts)


        # Get response
        response = llama.invoke(formatted_prompt)

        # Display in Streamlit
        st.markdown("### 🧠 Solution:")
        st.markdown(f'<div class="chat-reply">{response.content}</div>', unsafe_allow_html=True)

        #st.write(response.content)


    
    
def eda_page(): 
    st.title("Data Analysis Mentoring")
    exps=st.number_input("Choose a mentor based on their experience: 0-20")
    doubts = st.text_input("Explain your doubt")
    if doubts:
        # Set up the prompt
        messages = [
            SystemMessagePromptTemplate.from_template("You have an experience of {exp} years in Data Analysis."),
            HumanMessagePromptTemplate.from_template("Provide a clear solution for the following doubt: {doubt}")
        ]

        prompt = ChatPromptTemplate.from_messages(messages)
        formatted_prompt = prompt.format(exp=exps, doubt=doubts)


        # Get response
        response = deepseek.invoke(formatted_prompt)

        # Display in Streamlit
        st.markdown("### 🧠 Solution:")
        st.markdown(f'<div class="chat-reply">{response.content}</div>', unsafe_allow_html=True)

def stats_page(): 
    st.title("Statistics Mentoring")
    exps=st.number_input("Choose a mentor based on their experience: 0-20")
    doubts = st.text_input("Explain your doubt")
    if doubts:
        # Set up the prompt
        messages = [
            SystemMessagePromptTemplate.from_template("You have an experience of {exp} years in Statistical Analysis."),
            HumanMessagePromptTemplate.from_template("Provide a clear solution for the following doubt: {doubt}")
        ]

        prompt = ChatPromptTemplate.from_messages(messages)
        formatted_prompt = prompt.format(exp=exps, doubt=doubts)


        # Get response
        response = llama.invoke(formatted_prompt)

        # Display in Streamlit
        st.markdown("### 🧠 Solution:")
        st.markdown(f'<div class="chat-reply">{response.content}</div>', unsafe_allow_html=True)

def ml_page(): 
    st.title("Machine Learning Mentoring")
    exps=st.number_input("Choose a mentor based on their experience: 0-20")
    doubts = st.text_input("Explain your doubt")
    if doubts:
        # Set up the prompt
        messages = [
            SystemMessagePromptTemplate.from_template("You have an experience of {exp} years in ML."),
            HumanMessagePromptTemplate.from_template("Provide a clear solution for the following doubt: {doubt}")
        ]

        prompt = ChatPromptTemplate.from_messages(messages)
        formatted_prompt = prompt.format(exp=exps, doubt=doubts)


        # Get response
        response = deepseek.invoke(formatted_prompt)

        # Display in Streamlit
        st.markdown("### 🧠 Solution:")
        st.markdown(f'<div class="chat-reply">{response.content}</div>', unsafe_allow_html=True)

def dl_page(): 
    st.title("Deep Learning Mentoring")
    exps=st.number_input("Choose a mentor based on their experience: 0-20")
    doubts = st.text_input("Explain your doubt")
    if doubts:
        # Set up the prompt
        messages = [
            SystemMessagePromptTemplate.from_template("You have an experience of {exp} years in Deep learning."),
            HumanMessagePromptTemplate.from_template("Provide a clear solution for the following doubt: {doubt}")
        ]

        prompt = ChatPromptTemplate.from_messages(messages)
        formatted_prompt = prompt.format(exp=exps, doubt=doubts)


        # Get response
        response = llama.invoke(formatted_prompt)

        # Display in Streamlit
        st.markdown("### 🧠 Solution:")
        st.markdown(f'<div class="chat-reply">{response.content}</div>', unsafe_allow_html=True)

def excel_page(): 
    st.title("MS Excel Mentoring")
    exps=st.number_input("Choose a mentor based on their experience: 0-20")
    doubts = st.text_input("Explain your doubt")
    if doubts:
        # Set up the prompt
        messages = [
            SystemMessagePromptTemplate.from_template("You have an experience of {exp} years in Excel."),
            HumanMessagePromptTemplate.from_template("Provide a clear solution for the following doubt: {doubt}")
        ]

        prompt = ChatPromptTemplate.from_messages(messages)
        formatted_prompt = prompt.format(exp=exps, doubt=doubts)


        # Get response
        response = deepseek.invoke(formatted_prompt)

        # Display in Streamlit
        st.markdown("### 🧠 Solution:")
        st.markdown(f'<div class="chat-reply">{response.content}</div>', unsafe_allow_html=True)

def power_bi_page(): 
    st.title("Power BI Mentoring")
    exps=st.number_input("Choose a mentor based on their experience: 0-20")
    doubts = st.text_input("Explain your doubt")
    if doubts:
        # Set up the prompt
        messages = [
            SystemMessagePromptTemplate.from_template("You have an experience of {exp} years in Power BI."),
            HumanMessagePromptTemplate.from_template("Provide a clear solution for the following doubt: {doubt}")
        ]

        prompt = ChatPromptTemplate.from_messages(messages)
        formatted_prompt = prompt.format(exp=exps, doubt=doubts)


        # Get response
        response = llama.invoke(formatted_prompt)

        # Display in Streamlit
        st.markdown("### 🧠 Solution:")
        st.markdown(f'<div class="chat-reply">{response.content}</div>', unsafe_allow_html=True)


# Routing
page = st.query_params.get("page", "main")

if page == "python":
    python_page()
elif page == "eda":
    eda_page()
elif page == "stats":
    stats_page()
elif page == "ml":
    ml_page()
elif page == "dl":
    dl_page()
elif page == "excel":
    excel_page()
elif page == "power_bi":
    power_bi_page()
else:
    main_page()