app.py

import os
import tempfile

import pandas as pd
import streamlit as st

from executor import execute_code
from llm import generate_code
from prompt import build_prompt

st.set_page_config(page_title="LLM Data Analyst", layout="wide")

st.title("📊 LLM-Powered Data Analyst")

st.markdown("Upload a CSV file and ask questions about it.")

# File uploader
uploaded_file = st.file_uploader("Upload CSV file", type=["csv"])

csv_path = st.session_state.get("csv_path")

if uploaded_file is not None:
    # Persist the uploaded file to disk so generated code can read it
    with tempfile.NamedTemporaryFile(delete=False, suffix=".csv") as tmp:
        tmp.write(uploaded_file.getbuffer())
        csv_path = os.path.abspath(tmp.name)

    st.session_state["csv_path"] = csv_path

    df = pd.read_csv(csv_path, encoding="latin1")
    st.session_state["csv_columns"] = list(df.columns)

    st.subheader("Dataset Preview")
    st.dataframe(df.head())
elif csv_path:
    # Reuse previously uploaded file during the same session
    df = pd.read_csv(csv_path, encoding="latin1")
    st.subheader("Dataset Preview")
    st.dataframe(df.head())
else:
    st.info("Please upload a CSV file to begin.")

user_query = st.text_input(
    "Enter your analysis request:",
    placeholder="e.g. Show total sales per region as a bar chart"
)

if st.button("Run Analysis"):
    if not user_query:
        st.warning("Please enter a query.")
    elif not csv_path:
        st.warning("Please upload a CSV file first.")
    else:
        with st.spinner("Generating code with LLM..."):
            prompt = build_prompt(user_query, csv_path)
            code = generate_code(prompt)

        st.subheader("🧠 Generated Code")
        st.code(code, language="python")

        with st.spinner("Executing code..."):
            try:
                result = execute_code(code)

                if os.path.exists("output.png"):
                    st.subheader("📈 Visualization")
                    st.image("output.png")

                if result["result_df"] is not None:
                    st.subheader("📋 Result Table")
                    st.dataframe(result["result_df"])

                st.success("Analysis completed successfully.")

            except Exception as e:
                st.error(f"Execution failed: {e}")