pages/model.py

import streamlit as st 
import os 
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
from io import StringIO
from langchain_huggingface import HuggingFaceEndpoint, ChatHuggingFace


os.environ["HF_TOKEN"]=os.getenv('HF_Token')
os.environ["HUGGINGFACEHUB_API_KEY"]=os.getenv('HF_Token')

st.set_page_config(page_title="InsightGenie – AI-Powered CSV Explorer", layout="wide")
st.title("🧙 InsightGenie")
st.markdown("**Explore your CSV like magic – Ask, analyze, and visualize with AI.**")

if "qa_conversations" not in st.session_state:
    st.session_state.qa_conversations = []

uploaded_csv = st.file_uploader("📂 Upload your CSV file to begin", type=["csv"])

if uploaded_csv:
    try:
        data = pd.read_csv(uploaded_csv)
        st.success("✅ Data loaded successfully!")
        
        st.header("🔎 Dataset Overview")
        st.markdown(f"- **Rows and Columns:** {data.shape[0]} rows × {data.shape[1]} columns")
        st.markdown("**📌 Column Names:**")
        st.write(data.columns.tolist())

        col1, col2 = st.columns(2)
        with col1:
            st.markdown("**🧩 Missing Values**")
            st.dataframe(data.isnull().sum(), height=200)
        with col2:
            st.markdown("**🔢 Data Types**")
            st.dataframe(data.dtypes, height=200)

    except Exception as e:
        st.error(f"❌ Failed to read the file: {e}")
        st.stop()

    st.header("💬 Ask InsightGenie")
    user_question = st.text_input("Type your question about the dataset here:")

    genie_endpoint = HuggingFaceEndpoint(
        repo_id="deepseek-ai/DeepSeek-R1",
        provider="nebius",
        temperature=0.5,
        max_new_tokens=150,
        task="conversational"
    )

    genie_chatbot = ChatHuggingFace(
        llm=genie_endpoint,
        repo_id=genie_endpoint.repo_id,
        provider=genie_endpoint.provider,
        temperature=0.5,
        max_new_tokens=150,
        task="conversational"
    )

    if user_question:
        sample_data = data.head(50).to_csv(index=False)

        prompt = f"""
You are a skilled data assistant named InsightGenie. A user has uploaded a dataset and asked a question.
Answer clearly. If the question involves charts or graphs, provide appropriate Python code using matplotlib or seaborn.

Here’s a preview of the dataset:
{sample_data}

User question:
{user_question}
"""

        with st.spinner("🔍 Generating response..."):
            try:
                model_response = genie_chatbot.invoke([{"role": "user", "content": prompt}])
                bot_reply = model_response.content if hasattr(model_response, "content") else model_response

                st.session_state.qa_conversations.append((user_question, bot_reply))

                st.markdown("### 🧠 Genie Says")
                st.write(bot_reply)

                # Auto-plot for simple queries
                if "plot" in user_question.lower():
                    with st.expander("📊 Auto-generated Plot"):
                        try:
                            numeric_cols = data.select_dtypes(include='number').columns.tolist()
                            if len(numeric_cols) >= 2:
                                fig, ax = plt.subplots()
                                sns.lineplot(data=data, x=numeric_cols[0], y=numeric_cols[1], ax=ax)
                                ax.set_title(f"{numeric_cols[1]} vs {numeric_cols[0]}")
                                st.pyplot(fig)
                            else:
                                st.info("⚠️ Not enough numeric columns to generate a plot.")
                        except Exception as e:
                            st.error(f"❌ Plotting error: {e}")
            except Exception as e:
                st.error(f"❌ Error generating AI response: {e}")

    if st.session_state.qa_conversations:
        st.header("📚 Chat History")
        for user_q, ai_a in reversed(st.session_state.qa_conversations):
            st.markdown(f"**🧑‍💻 You:** {user_q}")
            st.markdown(f"**🤖 InsightGenie:** {ai_a}")
            st.markdown("---")