import streamlit as st
import joblib
import pickle
import pandas as pd
import numpy as np
from model_interface.hf_model_store import get_artifact_path


def expense_forecasting_farmerp():
    st.title("Expense Forecasting")
    model = joblib.load(
        get_artifact_path("13_Expense_forecasting(FarmERP)/Expense_model2.pkl"
    ))

    encoders = joblib.load(
        get_artifact_path("13_Expense_forecasting(FarmERP)/labels_expense.pkl"
    ))

    reference_df = joblib.load(
        get_artifact_path("13_Expense_forecasting(FarmERP)/expense_reference.pkl"
    ))


    # PAGE CONFIG


    # st.set_page_config(
    #     page_title="Expense Forecasting App",
    #     layout="wide"
    # )

    # st.title("🌾 Expense & Harvest Forecasting - VegPro")


    # SIDEBAR INPUTS (WITH 'All' OPTION)


    st.sidebar.header("Enter Crop & Field Details")

    # 1. Site Name
    site_options = ["All"] + sorted(reference_df["Site_Name"].unique())
    site_name = st.sidebar.selectbox("Site Name", site_options)

    if site_name != "All":
        site_df = reference_df[reference_df["Site_Name"] == site_name]
    else:
        site_df = reference_df.copy()

    # 2. Plot Name
    plot_options = ["All"] + sorted(site_df["Plot_Name"].unique())
    plot_name = st.sidebar.selectbox("Plot Name", plot_options)

    if plot_name != "All":
        plot_df = site_df[site_df["Plot_Name"] == plot_name]
    else:
        plot_df = site_df.copy()

    # 3. Crop Name
    crop_options = ["All"] + sorted(plot_df["Crop_Name"].unique())
    crop_name = st.sidebar.selectbox("Crop Name", crop_options)

    if crop_name != "All":
        filtered_df = plot_df[plot_df["Crop_Name"] == crop_name]
    else:
        filtered_df = plot_df.copy()


    # PREDICTION


    if st.sidebar.button("🔮 Predict"):

        if filtered_df.empty:
            st.warning("No records found for selected inputs.")
            st.stop()

        results = []

        # Detect area column automatically
        area_col = [c for c in filtered_df.columns if "area" in c.lower()][0]

        for _, row in filtered_df.iterrows():

            area_acres = float(row[area_col])

            input_df = pd.DataFrame({
                "Site_Name": [row["Site_Name"]],
                "Plot_Name": [row["Plot_Name"]],
                "SubPlot_Name": [row["SubPlot_Name"]],
                "Crop_Name": [row["Crop_Name"]],
                "Crop_Type": [row["Crop_Type"]],
                "Variety_Name": [row["Variety_Name"]],
                "Area_acres": [area_acres]
            })

            # Encode categorical columns
            for col, encoder in encoders.items():
                if input_df[col].iloc[0] not in encoder.classes_:
                    st.error(f"❌ Unknown value in '{col}': {input_df[col].iloc[0]}")
                    st.stop()
                input_df[col] = encoder.transform(input_df[col])

            # Model prediction
            predictions = model.predict(input_df)

            total_expense = float(predictions[0][0])
            total_harvested_qty = float(predictions[0][1])

            results.append({
                "Site Name": row["Site_Name"],
                "Plot Name": row["Plot_Name"],
                "Sub Plot Name": row["SubPlot_Name"],
                "Crop Name": row["Crop_Name"],
                "Crop Type": row["Crop_Type"],
                "Variety Name": row["Variety_Name"],
                "Area (Acres)": area_acres,
                "Total Estimated Production Qty(Kg)": 0,
                "Total Harvested Qty (Kg)": round(total_harvested_qty, 2),
                "Total Expenses (KES)": round(total_expense, 2),
                #"Cost Per Unit": 0
            })

        result_df = pd.DataFrame(results)

    
        # DISPLAY RESULTS
        

        st.subheader("📊 Forecast Results")
        st.dataframe(result_df, use_container_width=True)

        
        # SUMMARY METRICS
        

        st.subheader("📈 Overall Summary")

        col1, col2, col3 = st.columns(3)

        total_expenses = result_df["Total Expenses (KES)"].sum()
        total_qty = result_df["Total Harvested Qty (Kg)"].sum()

        col1.metric("Total Expenses (KES)", round(total_expenses, 2))
        col2.metric("Total Harvested Qty (Kg)", round(total_qty, 2))
        #col3.metric("Avg Cost Per Unit", 0)