src/streamlit_app.py

import streamlit as st
import pandas as pd
import plotly.express as px
from pathlib import Path

st.set_page_config(
    page_title="Food Delivery Analytics",
    page_icon="🍔",
    layout="wide"
)

def find_file(filename: str) -> Path:
    candidates = [
        Path("/app") / filename,
        Path("/app/src") / filename,
        Path(__file__).resolve().parent / filename,
        Path(__file__).resolve().parent.parent / filename,
    ]
    for path in candidates:
        if path.exists():
            return path
    raise FileNotFoundError(
        f"Could not find {filename}. Checked: " + ", ".join(str(p) for p in candidates)
    )

@st.cache_data
def load_data():
    real_df = pd.read_csv(find_file("real_world_driver_data.csv"))
    metrics_df = pd.read_csv(find_file("synthetic_delivery_metrics.csv"))
    reviews_df = pd.read_csv(find_file("synthetic_customer_reviews.csv"))
    features_df = pd.read_csv(find_file("synthetic_driver_features.csv"))
    monthly_df = pd.read_csv(find_file("synthetic_monthly_delivery_series.csv"))
    return real_df, metrics_df, reviews_df, features_df, monthly_df

real_df, metrics_df, reviews_df, features_df, monthly_df = load_data()

monthly_df["month"] = pd.to_datetime(monthly_df["month"])

# KPIs
total_drivers = features_df["driver_id"].nunique()
total_deliveries = int(metrics_df["deliveries_completed"].sum())
avg_rating = round(features_df["Delivery_person_Ratings"].mean(), 2)
total_reviews = int(reviews_df.shape[0])

# Sidebar
st.sidebar.title("Filters")
city_options = ["All"] + sorted(features_df["City"].dropna().unique().tolist())
vehicle_options = ["All"] + sorted(features_df["Type_of_vehicle"].dropna().unique().tolist())

selected_city = st.sidebar.selectbox("City", city_options)
selected_vehicle = st.sidebar.selectbox("Vehicle Type", vehicle_options)

filtered_features = features_df.copy()

if selected_city != "All":
    filtered_features = filtered_features[filtered_features["City"] == selected_city]

if selected_vehicle != "All":
    filtered_features = filtered_features[filtered_features["Type_of_vehicle"] == selected_vehicle]

filtered_driver_ids = filtered_features["driver_id"].unique()

filtered_metrics = metrics_df[metrics_df["driver_id"].isin(filtered_driver_ids)].copy()
filtered_reviews = reviews_df[reviews_df["driver_id"].isin(filtered_driver_ids)].copy()

filtered_total_drivers = filtered_features["driver_id"].nunique()
filtered_total_deliveries = int(filtered_metrics["deliveries_completed"].sum()) if not filtered_metrics.empty else 0
filtered_avg_rating = round(filtered_features["Delivery_person_Ratings"].mean(), 2) if not filtered_features.empty else 0
filtered_total_reviews = int(filtered_reviews.shape[0])

# Header
st.markdown("""
<h1 style='text-align: center;'>🍔 Food Delivery Analytics Dashboard</h1>
<p style='text-align: center; font-size:18px;'>
Analyze driver performance, customer sentiment, and delivery trends
</p>
""", unsafe_allow_html=True)

st.markdown("---")

# KPI cards
col1, col2, col3, col4 = st.columns(4)
col1.metric("🚚 Drivers", filtered_total_drivers)
col2.metric("📦 Deliveries", f"{filtered_total_deliveries:,}")
col3.metric("⭐ Avg Rating", filtered_avg_rating)
col4.metric("💬 Reviews", f"{filtered_total_reviews:,}")

st.markdown("---")

tab1, tab2, tab3, tab4 = st.tabs([
    "Overview",
    "Driver Performance",
    "Customer Reviews",
    "Monthly Trends"
])

with tab1:
    st.subheader("Dataset Overview")
    st.write("This dashboard combines driver performance, delivery activity, and customer sentiment data.")

    col_a, col_b = st.columns(2)

    with col_a:
        if not filtered_features.empty:
            fig_overview_1 = px.histogram(
                filtered_features,
                x="performance_tier",
                title="Performance Tier Distribution"
            )
            st.plotly_chart(fig_overview_1, use_container_width=True)

    with col_b:
        if not filtered_features.empty:
            city_counts = filtered_features["City"].value_counts().reset_index()
            city_counts.columns = ["City", "count"]

            fig_overview_2 = px.bar(
                city_counts,
                x="City",
                y="count",
                title="Drivers by City"
            )
            st.plotly_chart(fig_overview_2, use_container_width=True)

    st.markdown("### Data Preview")
    st.dataframe(filtered_features.head(10), use_container_width=True)

with tab2:
    st.subheader("Driver Performance")

    if not filtered_features.empty:
        top_drivers = filtered_features.sort_values("total_deliveries", ascending=False).head(10)

        fig1 = px.bar(
            top_drivers,
            x="driver_id",
            y="total_deliveries",
            color="performance_tier",
            title="Top 10 Drivers by Total Deliveries"
        )
        st.plotly_chart(fig1, use_container_width=True)

        fig2 = px.scatter(
            filtered_features,
            x="avg_delivery_time",
            y="Delivery_person_Ratings",
            color="sentiment_label",
            hover_name="driver_id",
            title="Average Delivery Time vs Driver Rating"
        )
        st.plotly_chart(fig2, use_container_width=True)
    else:
        st.warning("No driver data available for the selected filters.")

with tab3:
    st.subheader("Customer Reviews")

    if not filtered_reviews.empty:
        col_a, col_b = st.columns(2)

        with col_a:
            sentiment_counts = filtered_reviews["sentiment_label"].value_counts().reset_index()
            sentiment_counts.columns = ["sentiment_label", "count"]

            fig3 = px.pie(
                sentiment_counts,
                names="sentiment_label",
                values="count",
                title="Customer Review Sentiment Distribution"
            )
            st.plotly_chart(fig3, use_container_width=True)

        with col_b:
            avg_rating_by_sentiment = (
                filtered_reviews.groupby("sentiment_label", as_index=False)["driver_rating"]
                .mean()
            )

            fig4 = px.bar(
                avg_rating_by_sentiment,
                x="sentiment_label",
                y="driver_rating",
                title="Average Rating by Sentiment"
            )
            st.plotly_chart(fig4, use_container_width=True)

        st.write("Sample reviews:")
        st.dataframe(
            filtered_reviews[["driver_id", "sentiment_label", "driver_rating", "review_text"]].head(15),
            use_container_width=True
        )
    else:
        st.warning("No review data available for the selected filters.")

with tab4:
    st.subheader("Monthly Delivery Trends")

    if not filtered_metrics.empty:
        monthly_trend = (
            filtered_metrics.groupby("month", as_index=False)["deliveries_completed"]
            .sum()
            .rename(columns={"deliveries_completed": "total_deliveries"})
        )
        monthly_trend["month"] = pd.to_datetime(monthly_trend["month"])

        fig4 = px.line(
            monthly_trend,
            x="month",
            y="total_deliveries",
            markers=True,
            title="Total Deliveries Over Time"
        )
        st.plotly_chart(fig4, use_container_width=True)
    else:
        st.warning("No monthly delivery data available for the selected filters.")

import requests

st.markdown("---")
st.subheader("🤖 AI Insight")

webhook_url = "https://jq7hhh.app.n8n.cloud/webhook/3dc59db1-0e3d-4b73-bdc1-bbe5168feb6f"

if st.button("Generate AI Insight", key="n8n_ai_insight_button"):

    payload = {
        "city": selected_city,
        "vehicle_type": selected_vehicle,
        "drivers": int(filtered_total_drivers),
        "deliveries": int(filtered_total_deliveries),
        "avg_rating": float(filtered_avg_rating),
        "reviews": int(filtered_total_reviews)
    }

    try:
        response = requests.post(webhook_url, json=payload, timeout=20)

        if response.status_code == 200:
            result = response.json()
            st.success(result.get("Insight", "Insight generated successfully."))
        else:
            st.error(f"Webhook error: {response.status_code}")
            st.write(response.text)

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