import streamlit as st
import seaborn as sns
import matplotlib.pyplot as plt
import pandas as pd
import numpy as np

# Load data
@st.cache_data # Cache the data loading to speed up app performance
def load_data():
    # Load pre-processed data from csv file
    df = pd.read_csv("processed_data.csv")  # replace with your dataset

   # Explicitly cast metric columns to numeric types after reading from CSV
    try:
        df['total_price'] = df['total_price'].astype(float)
        df['quantity'] = df['quantity'].astype(int)
    except Exception as e:
        # Fallback for logging if a type conversion fails unexpectedly
        print(f"Error during type casting in app.py: {e}")


    # Ensure order_time is treated as datetime for plots (important since it loses dtype during CSV save/load)
    if 'order_time' in df.columns:
        df['order_time'] = pd.to_datetime(df['order_time'])
    
    return df

# Code to create Streamlit app
def app():
    # Title for the app
    st.title("Pizza Sales Data Analysis Dashboard")
    
    # Load and cache the data
    df = load_data()


    # Calculate key metrics
    # Write a code snippet to calculate key metrics from the pizza orders dataframe, including the 
    # total number of unique orders, total revenue generated, the most popular pizza size, the most 
    # frequent pizza category, total pizzas sold

    # Calculate key metrics from dataset
    total_orders = df['order_id'].nunique()
    total_revenue = df['total_price'].sum()
    most_popular_pizza_size = df['pizza_size'].value_counts().idxmax()
    most_frequent_pizza_category = df['pizza_category'].value_counts().idxmax()
    most_popular_pizza_name = df['pizza_name'].value_counts().idxmax()
    total_pizzas_sold = df['quantity'].sum()
             

     # Sidebar with key metrics
    # Write a code snippet to display key metrics in the sidebar of a Streamlit application. 
    # Show the total number of orders, total revenue (formatted as currency), the most popular
    # pizza size, the most popular pizza category, and the total number of pizzas sold 
    # using the st.sidebar.metric function.

    # Generate Sidebar on dashboard with key metrics 
    st.sidebar.header("Key Metrics")
    st.sidebar.metric("Total Orders", total_orders)
    st.sidebar.metric("Total Revenue", f"${total_revenue:,.2f}")
    st.sidebar.metric("Most Popular Pizza Size", most_popular_pizza_size )
    st.sidebar.metric("Most Popular Pizza Category", most_frequent_pizza_category)
    st.sidebar.metric("Most Popular Pizza Name", most_popular_pizza_name)
    st.sidebar.metric("Total Pizzas Sold", total_pizzas_sold)

  
    # Plot Configurations for the Reports Dashboard

    plots = [
        {"title": "Top Selling Pizzas (by Quantity)", "type": "bar_sorted", "x": "pizza_name", "y": "quantity", "top": 5},
        {"title": "Quantity of Pizzas Sold by Hour of the Day", "type": "bar", "x": "order_hour", "y": "quantity"},    
        {"title": "Quantity of Pizzas Sold by Category and Time of the Day", "type": "count", "x": "pizza_category", "hue": "time_of_day"},   
        {"title": "Quantity of Pizzas Sold by Size and Time of the Day", "type": "count", "x": "pizza_size", "hue": "time_of_day"},  
        {"title": "Monthly Revenue Trends by Pizza Category", "type": "line", "x": "order_month", "y": "total_price", "hue": "pizza_category", "estimator": "sum", "marker": "o"}, 
    ]

    # Iterate through the plots dictionary above to display them
    for plot in plots:
        st.header(plot["title"])
        fig, ax = plt.subplots(figsize=(10, 6))
        
        if plot["type"] == "bar_sorted":
            # Bar Plot: Top Selling Pizzas (Custom sorting)
            plot_var = df.groupby(plot["x"])[plot["y"]].sum().reset_index().sort_values(by=plot["y"], ascending=False).head(plot["top"])
            sns.barplot(data=plot_var, x=plot["x"], y=plot["y"], ax=ax, palette="plasma")
            ax.set_xticklabels(ax.get_xticklabels(), rotation=45, ha='right')
            plt.tight_layout()
            
        elif plot["type"] == "bar":
            # Bar Plot: Pizzas sold per hour of day (No sorting)
            hourly_var = df.groupby(plot["x"])[plot["y"]].sum().reset_index()           
            sns.barplot(data=hourly_var, x=plot["x"], y=plot["y"], ax=ax, palette="rocket", order=np.arange(24))
          
            # Set explicit ticks from 0 to 23 for clarity
            ax.set_xticks(np.arange(0, 24, 2)) # Show every other tick for less clutter
            ax.tick_params(axis='x', rotation=0)

        elif plot["type"] == "count":
            # Count Plot: Category or Size breakdown
            sns.countplot(data=df, x=plot["x"], hue=plot["hue"], ax=ax)
            ax.legend(bbox_to_anchor=(1.05, 1), loc=2, borderaxespad=0.) # Move legend outside
            
        elif plot["type"] == "line":
            # Line Plot: Monthly Revenue Trends
            # Ensure order_month is treated as categorical for the plot if plotting trends is needed.
            sns.lineplot(data=df, x=plot["x"], y=plot["y"], hue=plot["hue"], estimator=plot["estimator"], errorbar=None, marker=plot["marker"], ax=ax)
            ax.set_xticks(df[plot["x"]].unique()) # Ensure ticks are on actual month values
            
        
        # Axis and Final Display Cleanup
        ax.set_xlabel(" ".join(plot["x"].split("_")).title())
        
        if "y" in plot.keys():
            # Customize y-label for revenue plots
            if plot["title"] == "Monthly Revenue Trends by Pizza Category":
                 ax.set_ylabel("Total Revenue (USD)")
            elif plot["y"] != "quantity":
                ax.set_ylabel(" ".join(plot["y"].split("_")).title())
            else:
                 ax.set_ylabel("Quantity Sold")
        elif plot["type"] == "count":
             ax.set_ylabel("Order Count")
        
        st.pyplot(fig)
        plt.close(fig) # Close the figure to free up memory


if __name__ == "__main__":
    app()