app.py

import pandas as pd
import plotly.graph_objects as go
from plotly.subplots import make_subplots
import gradio as gr
import ast
from functools import lru_cache
from collections import Counter
import requests
import os
import time

# --- Constants and Mappings (Unchanged) ---
BODY_ORDER = ['Very light-bodied', 'Light-bodied', 'Medium-bodied', 'Full-bodied', 'Very full-bodied']
ACIDITY_ORDER = ['Low', 'Medium', 'High']
BODY_MAPPING = {'Very light-bodied': 1, 'Light-bodied': 2, 'Medium-bodied': 3, 'Full-bodied': 4, 'Very full-bodied': 5}
WINE_TYPE_ORDER = {'Red': 2, 'Rosé': 1, 'White': 0}
SAMPLE_THRESHOLDS = {
    'Very Common (250+)': 250,
    'Common (100+)': 100, 
    'Uncommon (50+)': 50, 
    'Rare (20+)': 20
}
COUNTRY_FLAGS = {
    'United States': '🇺🇸', 'France': '🇫🇷', 'Italy': '🇮🇹', 'Spain': '🇪🇸', 'Germany': '🇩🇪', 'Australia': '🇦🇺',
    'Chile': '🇨🇱', 'Argentina': '🇦🇷', 'Portugal': '🇵🇹', 'South Africa': '🇿🇦', 'New Zealand': '🇳🇿',
    'Austria': '🇦🇹', 'Greece': '🇬🇷', 'Hungary': '🇭🇺', 'Croatia': '🇭🇷', 'Slovenia': '🇸🇮', 'Canada': '🇨🇦',
    'Brazil': '🇧🇷', 'Uruguay': '🇺🇾', 'Israel': '🇮🇱', 'Lebanon': '🇱🇧', 'Turkey': '🇹🇷', 'Bulgaria': '🇧🇬',
    'Romania': '🇷🇴', 'Georgia': '🇬🇪', 'Moldova': '🇲🇩', 'Switzerland': '🇨🇭', 'England': '🏴'
}
FOOD_EMOJIS = {
    # Meat - Specific Animals
    'Beef': '🐄', 'Pork': '🐷', 'Lamb': '🐑', 'Veal': '🐄', 'Ham': '🐷',
    'Poultry': '🐔', 'Chicken': '🐔', 'Duck': '🦆',
    'Game Meat': '🦌', 'Meat': '🥩',
    
    # Cured/Processed Meats
    'Cured Meat': '🥓', 'Cold Cuts': '🥪', 'Barbecue': '🔥', 'Grilled': '🔥', 'Roast': '🍖',
    
    # Fish & Seafood - Specific Types
    'Rich Fish': '🐟', 'Lean Fish': '🐟', 'Fish': '🐟', 'Codfish': '🐟',
    'Shellfish': '🦐', 'Seafood': '🦞', 'Sushi': '🍣', 'Sashimi': '🍣',
    
    # Cheese - Different Types
    'Cheese': '🧀', 'Soft Cheese': '🧀', 'Hard Cheese': '🧀', 'Blue Cheese': '🧀',
    'Goat Cheese': '🐐', 'Maturated Cheese': '🧀', 'Mild Cheese': '🧀', 'Medium-cured Cheese': '🧀',
    
    # Pasta & Italian
    'Pasta': '🍝', 'Tagliatelle': '🍝', 'Lasagna': '🍝', 'Risotto': '🍚',
    'Pizza': '🍕', 'Eggplant Parmigiana': '🍆',
    
    # Asian Food
    'Asian Food': '🥢', 'Curry Chicken': '🍛', 'Yakissoba': '🍜', 'Paella': '🥘',
    
    # Vegetables & Vegetarian
    'Vegetarian': '🥬', 'Salad': '🥗', 'Mushrooms': '🍄', 'Beans': '🫘',
    'Baked Potato': '🥔', 'French Fries': '🍟',
    
    # Desserts & Sweets
    'Sweet Dessert': '🍰', 'Dessert': '🍰', 'Fruit Dessert': '🍓',
    'Cake': '🍰', 'Cookies': '🍪', 'Chocolate': '🍫', 'Cream': '🍨',
    'Citric Dessert': '🍋', 'Spiced Fruit Cake': '🍰',
    
    # Fruits & Nuts
    'Fruit': '🍇', 'Dried Fruits': '🍇', 'Chestnut': '🌰',
    
    # Appetizers & Snacks
    'Appetizer': '🍤', 'Snack': '🥨', 'Aperitif': '🥂',
    
    # Soups & Stews
    'Light Stews': '🍲', 'Soufflé': '🥄',
    
    # Dishes & Preparations
    'Spicy Food': '🌶️', 'Tomato Dishes': '🍅'
}


# --- Data Download Function ---
def download_data():
    """Downloads the dataset from Google Drive if not already present."""
    csv_filename = 'XWines_Full_100K_wines.csv'
    
    if os.path.exists(csv_filename):
        print(f"Using existing dataset: {csv_filename}")
        return csv_filename
    
    # Convert Google Drive share link to direct download link
    file_id = '1uEEipmKNxdiKUAhjH-K14JOSQ2BLRFss'
    download_url = f'https://drive.google.com/uc?export=download&id={file_id}'
    
    print(f"Downloading dataset from Google Drive...")
    try:
        response = requests.get(download_url, stream=True)
        response.raise_for_status()
        
        with open(csv_filename, 'wb') as f:
            for chunk in response.iter_content(chunk_size=8192):
                f.write(chunk)
        
        print(f"Dataset downloaded successfully: {csv_filename}")
        return csv_filename
    except Exception as e:
        raise Exception(f"Failed to download dataset: {str(e)}")

# --- OPTIMIZATION 1: Data Loading & Pre-processing ---
@lru_cache(maxsize=1)
def load_and_preprocess_data():
    """Loads and performs expensive one-time preprocessing on the dataset."""
    start_time = time.time()
    print("[TIMING] Starting data loading and preprocessing...")
    
    csv_filename = download_data()
    print(f"[TIMING] File check/download completed in {time.time() - start_time:.2f}s")
    
    try:
        csv_start = time.time()
        print("[TIMING] Loading CSV data...")
        # Use efficient data types and only load needed columns if possible
        df = pd.read_csv(csv_filename, low_memory=False)
        print(f"[TIMING] CSV loaded in {time.time() - csv_start:.2f}s - {len(df):,} wine records")
    except FileNotFoundError:
        raise FileNotFoundError(f"CSV file '{csv_filename}' not found.")

    def parse_list_string(s):
        if not isinstance(s, str) or not s.strip():
            return []
        try:
            # Fast path for common patterns
            s = s.strip()
            if s.startswith('[') and s.endswith(']'):
                return ast.literal_eval(s)
            return []
        except (ValueError, SyntaxError):
            return []

    # Vectorized string processing for better performance
    parse_start = time.time()
    print("[TIMING] Starting string parsing...")
    
    df['grapes_list'] = df['Grapes'].fillna('[]').apply(parse_list_string)
    print(f"[TIMING] Grapes parsing completed in {time.time() - parse_start:.2f}s")
    
    harmonize_start = time.time()
    df['harmonize_list'] = df['Harmonize'].fillna('[]').apply(parse_list_string)
    print(f"[TIMING] Harmonize parsing completed in {time.time() - harmonize_start:.2f}s")
    
    derived_start = time.time()
    df['main_grape'] = df['grapes_list'].apply(lambda x: x[0] if x else 'Unknown')
    df['num_grapes'] = df['grapes_list'].apply(len)
    df['body_numeric'] = df['Body'].map(BODY_MAPPING)
    print(f"[TIMING] Derived columns completed in {time.time() - derived_start:.2f}s")
    
    total_time = time.time() - start_time
    print(f"[TIMING] Total preprocessing completed in {total_time:.2f}s")
    return df


# --- OPTIMIZATION 2: Vectorized Data Aggregation ---
def get_top_food_pairings(harmonize_list, top_n=3):
    """Get top N food pairings with emojis and names - optimized version."""
    # Flatten list more efficiently
    all_pairings = []
    for sublist in harmonize_list:
        if isinstance(sublist, list):
            all_pairings.extend(sublist)
    
    if not all_pairings:
        return {'emojis': '🍽️', 'names': 'General'}
    
    top_items = Counter(all_pairings).most_common(top_n)
    emojis = ''.join([FOOD_EMOJIS.get(item[0], '🍽️') for item in top_items])
    names = ', '.join([item[0] for item in top_items])
    return {'emojis': emojis, 'names': names}


def aggregate_wine_data(df, wine_types, max_grape_count, min_samples_choice, regional_grouping):
    """Filters and aggregates wine data using efficient, vectorized pandas operations."""
    agg_start = time.time()
    print(f"[TIMING] Starting aggregation with {len(df):,} records...")
    
    filtered_df = df.copy()

    if wine_types and 'All' not in wine_types:
        filtered_df = filtered_df[filtered_df['Type'].isin(wine_types)]
    if max_grape_count < 5:
        filtered_df = filtered_df[filtered_df['num_grapes'] <= max_grape_count]

    group_by_cols = ['main_grape', 'Type']
    if regional_grouping:
        group_by_cols.append('Country')

    groupby_start = time.time()
    agg_df = filtered_df.groupby(group_by_cols).agg(
        count=('ABV', 'size'),
        avg_fullness=('body_numeric', 'mean'),
        abv_list=('ABV', list),
        body_list=('Body', list),
        acidity_list=('Acidity', list),
        harmonize_list=('harmonize_list', list),
        region_count=('RegionName', 'nunique'),
        winery_count=('WineryName', 'nunique')
    ).reset_index()
    print(f"[TIMING] GroupBy aggregation completed in {time.time() - groupby_start:.2f}s")

    min_samples = SAMPLE_THRESHOLDS[min_samples_choice]
    agg_df = agg_df[agg_df['count'] >= min_samples].copy()

    if agg_df.empty:
        return agg_df

    # Optimized distribution calculation
    def calc_distribution(values_list, categories):
        if not values_list:
            return {cat: 0.0 for cat in categories}
        
        counts = pd.Series(values_list).value_counts(normalize=True) * 100
        return {cat: counts.get(cat, 0.0) for cat in categories}
    
    dist_start = time.time()
    agg_df['body_dist'] = agg_df['body_list'].apply(
        lambda x: calc_distribution(x, BODY_ORDER))
    agg_df['acid_dist'] = agg_df['acidity_list'].apply(
        lambda x: calc_distribution(x, ACIDITY_ORDER))
    print(f"[TIMING] Distribution calculations completed in {time.time() - dist_start:.2f}s")

    # Pre-compute food pairings more efficiently
    pairing_start = time.time()
    pairing_data = []
    for harmonize_list in agg_df['harmonize_list']:
        pairing_data.append(get_top_food_pairings(harmonize_list))
    agg_df['pairing_data'] = pairing_data
    agg_df['pairing_emoji'] = agg_df['pairing_data'].apply(lambda x: x['emojis'])
    agg_df['pairing_names'] = agg_df['pairing_data'].apply(lambda x: x['names'])
    print(f"[TIMING] Food pairing calculations completed in {time.time() - pairing_start:.2f}s")
    
    final_start = time.time()
    agg_df['wine_type_order'] = agg_df['Type'].map(WINE_TYPE_ORDER)
    agg_df = agg_df.sort_values(by=['wine_type_order', 'avg_fullness'], ascending=[False, True])
    print(f"[TIMING] Final sorting completed in {time.time() - final_start:.2f}s")
    
    total_agg_time = time.time() - agg_start
    print(f"[TIMING] Total aggregation completed in {total_agg_time:.2f}s - {len(agg_df)} combinations")
    return agg_df


# --- OPTIMIZATION 3: Efficient & Clean Chart Creation ---
def create_wine_chart(chart_data, regional_grouping):
    """Creates the Plotly figure with optimized traces and layout."""
    chart_start = time.time()
    print(f"[TIMING] Starting chart creation with {len(chart_data)} rows...")
    
    if chart_data.empty:
        fig = go.Figure()
        fig.add_annotation(text="No data available with current filters.", xref="paper", yref="paper", x=0.5, y=0.5,
                           showarrow=False)
        return fig

    num_rows = len(chart_data)

    # Pre-compute labels more efficiently
    prep_start = time.time()
    wine_type_emojis = {'Red': '🍷', 'White': '🥂', 'Rosé': '🌸', 'Sparkling': '🍾'}
    chart_data = chart_data.copy()  # Avoid SettingWithCopyWarning
    chart_data['wine_emoji'] = chart_data['Type'].map(wine_type_emojis).fillna('🍷')
    
    if regional_grouping:
        chart_data['flag'] = chart_data['Country'].map(COUNTRY_FLAGS).fillna('🌍')
        # Vectorized string concatenation instead of apply
        chart_data['grape_label'] = chart_data['wine_emoji'] + ' ' + chart_data['main_grape'] + ' ' + chart_data['flag']
    else:
        chart_data['grape_label'] = chart_data['wine_emoji'] + ' ' + chart_data['main_grape']
    print(f"[TIMING] Label preparation completed in {time.time() - prep_start:.2f}s")

    y_labels = chart_data['grape_label'].tolist()

    fig = make_subplots(
        rows=1, cols=5,
        specs=[[{}, {"type": "bar"}, {"type": "bar"}, {"type": "box"}, {}]],
        column_widths=[0.25, 0.22, 0.22, 0.16, 0.05],
        horizontal_spacing=0.025,
        shared_yaxes=True
    )

    # Pre-compute hover texts more efficiently
    hover_start = time.time()
    country_part = chart_data['Country'] if regional_grouping else 'Global'
    hover_texts = (
        '<b>' + chart_data['main_grape'] + ' (' + country_part.astype(str) + ')</b><br>' +
        'Wineries: ' + chart_data['winery_count'].astype(str) + '<br>' +
        'Regions: ' + chart_data['region_count'].astype(str) + '<br>' +
        'Total Wines: ' + chart_data['count'].apply(lambda x: f"{x:,}")
    ).tolist()
    print(f"[TIMING] Hover text preparation completed in {time.time() - hover_start:.2f}s")
    fig.add_trace(go.Bar(
        y=y_labels, x=[1] * num_rows, orientation='h',
        marker_color='rgba(0,0,0,0)', showlegend=False,
        hoverinfo='text', hovertext=hover_texts
    ), row=1, col=1)
    fig.add_trace(go.Scatter(
        y=y_labels, x=[0.03] * num_rows, mode='text',
        text=y_labels, textposition='middle right',
        textfont={'size': 22, 'color': '#1A1A1A'},
        hoverinfo='none', showlegend=False
    ), row=1, col=1)

    # Optimize body distribution traces
    body_start = time.time()
    body_colors = {'Very light-bodied': '#FFB6C1', 'Light-bodied': '#CD5C5C', 'Medium-bodied': '#C13636',
                   'Full-bodied': '#8B0000', 'Very full-bodied': '#4B0000'}
    
    # Pre-extract all body values at once
    body_data = {}
    for body_type in BODY_ORDER:
        body_data[body_type] = [d.get(body_type, 0) for d in chart_data['body_dist']]
    
    for body_type in BODY_ORDER:
        fig.add_trace(go.Bar(
            y=y_labels, x=body_data[body_type], name=body_type, orientation='h',
            marker_color=body_colors.get(body_type), showlegend=False,
            hovertemplate=f"{body_type}: %{{x:.1f}}%<extra></extra>"
        ), row=1, col=2)
    print(f"[TIMING] Body traces completed in {time.time() - body_start:.2f}s")

    # Optimize acidity distribution traces
    acid_start = time.time()
    acid_colors = {'Low': '#F5F5DC', 'Medium': '#DAA520', 'High': '#B8860B'}
    
    # Pre-extract all acidity values at once
    acid_data = {}
    for acid_type in ACIDITY_ORDER:
        acid_data[acid_type] = [d.get(acid_type, 0) for d in chart_data['acid_dist']]
    
    for acid_type in ACIDITY_ORDER:
        fig.add_trace(go.Bar(
            y=y_labels, x=acid_data[acid_type], name=acid_type, orientation='h',
            marker_color=acid_colors.get(acid_type), showlegend=False,
            hovertemplate=f"{acid_type} acidity: %{{x:.1f}}%<extra></extra>"
        ), row=1, col=3)
    print(f"[TIMING] Acidity traces completed in {time.time() - acid_start:.2f}s")

    # Optimize box plot creation
    box_start = time.time()
    box_colors = {'Red': '#8B0000', 'White': '#DAA520', 'Rosé': '#CD5C5C', 'Sparkling': '#9370DB'}
    
    # Create box plots more efficiently
    for idx, (_, row) in enumerate(chart_data.iterrows()):
        abv_values = row['abv_list']
        color = box_colors.get(row['Type'], '#6A5ACD')
        fig.add_trace(go.Box(
            y=[y_labels[idx]] * len(abv_values), x=abv_values, name=row['Type'], orientation='h',
            showlegend=False, marker_color=color, line_color=color,
            hovertemplate=f"ABV: %{{x:.1f}}%<extra></extra>"
        ), row=1, col=4)
    print(f"[TIMING] Box plot traces completed in {time.time() - box_start:.2f}s")

    fig.add_trace(go.Scatter(
        y=y_labels, x=[0.5] * num_rows, mode='text',
        text=chart_data['pairing_emoji'], textposition='middle center',
        textfont={'size': 32}, showlegend=False,
        hoverinfo='text', hovertext=chart_data['pairing_names']
    ), row=1, col=5)

    fig.update_layout(
        title={
            'text': "Wine Characteristics by Grape Variety",
            'x': 0.5,
            'font': {'size': 26, 'color': '#1A1A1A'}
        },
        height=max(600, num_rows * 55),
        barmode='stack', showlegend=False, plot_bgcolor='#FAFAFA', paper_bgcolor='#E8E9EA',
        margin=dict(l=30, r=30, t=100, b=60), boxgap=0.3, bargap=0.2
    )

    column_titles = ["Grape Variety", "Body Profile (%)", "Acidity Profile (%)", "Alcohol (ABV %)", "Food Pairing"]
    # Add column titles
    title_start = time.time()
    for i, title in enumerate(column_titles, 1):
        domain = fig.layout[f'xaxis{i if i > 1 else ""}'].domain
        fig.add_annotation(
            x=(domain[0] + domain[1]) / 2, y=1.02,
            xref="paper", yref="paper", text=f"<b>{title}</b>",
            xanchor='center', showarrow=False, font={'size': 20, 'color': '#1A1A1A'}
        )
    print(f"[TIMING] Column titles completed in {time.time() - title_start:.2f}s")

    # Update axes formatting
    axes_start = time.time()
    for i in range(1, 6):
        fig.update_yaxes(showticklabels=False, showgrid=False, zeroline=False, row=1, col=i)
        # Show x-axis labels only for body, acidity, and ABV columns
        if i in [2, 3, 4]:
            fig.update_xaxes(showticklabels=True, showgrid=True, gridcolor='rgba(0,0,0,0.1)', 
                           zeroline=False, title_text="", row=1, col=i)
        else:
            fig.update_xaxes(showticklabels=False, showgrid=False, zeroline=False, title_text="", row=1, col=i)
    print(f"[TIMING] Axes formatting completed in {time.time() - axes_start:.2f}s")

    # Final axis configuration
    final_axes_start = time.time()
    fig.update_yaxes(categoryorder="array", categoryarray=y_labels, autorange=False, range=[-0.5, num_rows - 0.5],
                     row=1, col=1)
    print(f"[TIMING] Final axis configuration completed in {time.time() - final_axes_start:.2f}s")

    # Add alternating row backgrounds for better readability
    bg_start = time.time()
    for i in range(num_rows):
        if i % 2 == 1:
            fig.add_hrect(y0=i - 0.5, y1=i + 0.5, fillcolor="#F0F2F3", layer="below", line_width=0, row=1, col="all")
    print(f"[TIMING] Background rectangles completed in {time.time() - bg_start:.2f}s")

    chart_time = time.time() - chart_start
    print(f"[TIMING] Chart creation completed in {chart_time:.2f}s")
    return fig


# --- Gradio Interface Logic ---
def update_dashboard(wine_types, max_grape_count, min_samples_choice, regional_grouping,
                     progress=gr.Progress(track_tqdm=True)):
    """Main function to update dashboard."""
    dashboard_start = time.time()
    print(f"[TIMING] Starting dashboard update...")
    
    progress(0, desc="Loading and processing data...")
    df = load_and_preprocess_data()

    progress(0.5, desc="Filtering and aggregating...")
    chart_data = aggregate_wine_data(df, wine_types, max_grape_count, min_samples_choice, regional_grouping)

    progress(0.8, desc="Creating chart...")
    fig = create_wine_chart(chart_data, regional_grouping)

    total_combinations = len(chart_data)
    total_wines = chart_data['count'].sum() if not chart_data.empty else 0
    min_samples = SAMPLE_THRESHOLDS[min_samples_choice]
    grouping_type = "grape+region" if regional_grouping else "grape+type"
    summary = f"📊 Showing **{total_combinations}** {grouping_type} combinations from **{total_wines:,}** wines (min {min_samples} samples each)"

    total_dashboard_time = time.time() - dashboard_start
    print(f"[TIMING] Total dashboard update completed in {total_dashboard_time:.2f}s")
    return fig, summary


# Create Gradio interface
def create_interface():
    with gr.Blocks(title="Wine Analysis Dashboard", theme=gr.themes.Soft()) as demo:
        gr.Markdown("# 🍷 Wine Characteristics Dashboard")

        with gr.Row():
            wine_type_filter = gr.CheckboxGroup(
                choices=['Red', 'White', 'Rosé', 'Sparkling', 'Dessert', 'Dessert/Port'],
                value=['Red'], label="🍷 Wine Types"
            )
            max_grape_slider = gr.Slider(
                minimum=1, maximum=5, step=1, value=1, label="🍇 Max Grapes per Wine",
                info="1: Varietals, 5: All Blends"
            )
            min_samples_choice = gr.Radio(
                choices=list(SAMPLE_THRESHOLDS.keys()), value='Very Common (250+)',
                label="Minimum Sample Size (Wines per Variety)",
            )
            regional_grouping = gr.Checkbox(
                value=True, label="Split By Country"
            )

        summary_text = gr.Markdown()
        wine_plot = gr.Plot()

        inputs = [wine_type_filter, max_grape_slider, min_samples_choice, regional_grouping]
        outputs = [wine_plot, summary_text]

        # Auto-update when any input changes
        for input_component in inputs:
            input_component.change(update_dashboard, inputs=inputs, outputs=outputs)
        
        demo.load(fn=update_dashboard, inputs=inputs, outputs=outputs)

    return demo


if __name__ == "__main__":
    app_interface = create_interface()
    app_interface.launch()