import os
import json
import requests
import pandas as pd
import streamlit as st

st.set_page_config(page_title="SuperKart Forecast UI", layout="wide")

DEFAULT_BACKEND = os.environ.get("BACKEND_URL", "http://127.0.0.1:8000")
BACKEND_URL = st.sidebar.text_input("Backend base URL", DEFAULT_BACKEND)
st.sidebar.caption("Example: https://<your-backend-space>.hf.space")

if st.sidebar.button("Test Backend Health"):
    try:
        r = requests.get(f"{BACKEND_URL}/health", timeout=15)
        if r.ok:
            st.sidebar.success(r.json())
        else:
            st.sidebar.error(f"{r.status_code}: {r.text}")
    except Exception as e:
        st.sidebar.error(f"Failed: {e}")

st.title("SuperKart Sales Forecast")
st.write("Fill in the features below. This UI uses **Store_Establishment_Year** (not Store_Age).")


with st.form("single_pred"):
    col1, col2, col3 = st.columns(3)

    with col1:
        store_est_year = st.number_input("Store_Establishment_Year", min_value=1900, max_value=2100, value=2014, step=1)
        product_weight = st.number_input("Product_Weight (kg)", min_value=0.0, value=0.45, step=0.01, format="%.2f")
        product_alloc_area = st.number_input("Product_Allocated_Area (0-1)", min_value=0.0, max_value=1.0, value=0.08, step=0.01, format="%.2f")
        product_mrp = st.number_input("Product_MRP", min_value=0.0, value=1200.0, step=1.0, format="%.2f")

    with col2:
        store_size = st.selectbox("Store_Size", ["High", "Medium", "Low"])
        store_loc_city_type = st.selectbox("Store_Location_City_Type", ["Tier 1", "Tier 2", "Tier 3"])
        store_type = st.selectbox("Store_Type", ["Supermarket Type 1", "Supermarket Type 2", "Departmental Store", "Food Mart"])

    with col3:
        product_sugar_content = st.selectbox("Product_Sugar_Content", ["Regular", "Low Sugar", "No Sugar"])
        product_type = st.selectbox(
            "Product_Type",
            [
                "Meat", "Snack foods", "Hard drinks", "Dairy", "Canned", "Soft drinks",
                "Health and hygiene", "Baking goods", "Bread", "Breakfast", "Frozen foods",
                "Fruits and vegetables", "Household", "Seafood", "Starchy foods", "Others"
            ],
            index=1
        )

    submitted = st.form_submit_button("Predict")

    payload = {
        "Store_Establishment_Year": int(store_est_year),
        "Product_Weight": float(product_weight),
        "Product_Allocated_Area": float(product_alloc_area),
        "Product_MRP": float(product_mrp),
        "Store_Size": store_size,
        "Store_Location_City_Type": store_loc_city_type,
        "Store_Type": store_type,
        "Product_Sugar_Content": product_sugar_content,
        "Product_Type": product_type,
    }

    with st.expander("Request JSON preview"):
        st.code(json.dumps(payload, indent=2), language="json")

    if submitted:
        try:
            r = requests.post(f"{BACKEND_URL}/predict", json=payload, timeout=25)
            if r.ok:
                st.success(r.json())
            else:
                st.error(f"{r.status_code}: {r.text}")
        except Exception as e:
            st.error(f"Prediction failed: {e}")


st.markdown("---")
st.subheader("Batch prediction via CSV")
st.caption("Include training columns (e.g., Store_Establishment_Year, Product_Weight, Product_Allocated_Area, Product_MRP, Store_Size, Store_Location_City_Type, Store_Type, Product_Sugar_Content, Product_Type).")

uploaded = st.file_uploader("Upload CSV", type=["csv"])
if uploaded is not None:
    try:
        df_preview = pd.read_csv(uploaded)
        st.write("Preview:", df_preview.head())
        uploaded.seek(0)
        files = {"file": ("batch.csv", uploaded, "text/csv")}
        r = requests.post(f"{BACKEND_URL}/predict-csv", files=files, timeout=60)
        if r.ok:
            resp = r.json()
            st.success(f"Rows: {resp.get('n')}")
            st.json(resp if len(str(resp)) < 1200 else {"message": "Predictions received.", "keys": list(resp.keys())})
        else:
            st.error(f"{r.status_code}: {r.text}")
    except Exception as e:
        st.error(f"CSV prediction failed: {e}")