app.py

import gradio as gr
import pandas as pd
import numpy as np
from typing import List, Dict
import os

# Create static directory if it doesn't exist
if not os.path.exists('static'):
    os.makedirs('static')

# Move thumbs directory into static if it exists
if os.path.exists('thumbs'):
    import shutil
    if not os.path.exists('static/thumbs'):
        shutil.move('thumbs', 'static/')
    else:
        # If static/thumbs already exists, merge the contents
        for file in os.listdir('thumbs'):
            shutil.move(os.path.join('thumbs', file), 'static/thumbs/')
        os.rmdir('thumbs')

class CSVViewer:
    def __init__(self, csv_path: str):
        self.df = pd.read_csv(csv_path)
        self.current_page = 0
        self.rows_per_page = 1
        self.total_pages = len(self.df)
        
    def get_page_data(self, page_num: int) -> Dict:
        if page_num < 0 or page_num >= self.total_pages:
            return None
            
        row = self.df.iloc[page_num]
        
        # Get product information
        product_info = {
            'Web link': str(row['Web link']),
            'Product Code': str(row['Product Code']),
            'Product Name': str(row['Product Name'])
        }
        
        # Separate original and AI-predicted columns
        original_data = {}
        ai_data = {}
        
        for col in self.df.columns:
            if 'Score' in col or 'Distance' in col or col  == '':
                continue  # Skip score and distance columns
                
            if col in ['Web link', 'Product Code', 'Product Name']:
                continue  # Skip these columns as they're shown at the top

            if col in ["Length", "Style", "Sleeve Length", "Neckline", "Pattern", "Fabric", "Features", "Closure", "Body Fit", "Occasion", "Season"]:
                original_data[col] = str(row[col])
            if col.endswith('_AI'):
                ai_data[col] = str(row[col])
                
        
        return {
            'product_info': product_info,
            'original_data': original_data,
            'ai_data': ai_data,
            'page_num': page_num + 1,
            'total_pages': self.total_pages
        }

def create_page_view(page_num: int):
    viewer = CSVViewer('attr.csv')
    page_data = viewer.get_page_data(page_num)
    
    if not page_data:
        return "Invalid page number"
    
    # Get product image
    product_code = page_data['product_info']['Product Code']
    
    # Find matching image file
    image_files = [f for f in os.listdir('static/thumbs') if f.startswith(product_code)]
    if image_files:
        # Use Gradio's static file serving
        image_path = f"/static/thumbs/{image_files[0]}"
        print(f"Found image: {image_path}")
    else:
        image_path = None
    
    # Create HTML content
    html = f"<h3>Page {page_data['page_num']} of {page_data['total_pages']}</h3>"
    html += "<div style='margin-bottom: 20px; display: flex; gap: 20px;'>"
    
    # Product info section
    html += "<div>"
    if image_path:
        html += f"<img src='{image_path}' alt='Product Image' style='max-width: 100%; border-radius: 5px;'>"
    for key, value in page_data['product_info'].items():
        html += f"<p><strong>{key}:</strong> {value}</p>"
    html += "</div></div>"
    
    html += "<div style='display: flex; gap: 20px;'>"
    
    # Comparison section
    html += "<div style='flex: 1; border: 1px solid #ddd; padding: 15px; border-radius: 5px;'>"
    html += "<h4>Comparison</h4>"
    html += "<table style='width: 100%;'>"
    html += "<tr style='background-color: #f5f5f5;'><th>Attribute</th><th>Original</th><th>AI Prediction</th><th>Status</th></tr>"
    
    # Add comparison rows
    for key, orig_value in page_data['original_data'].items():
        # Find matching AI prediction
        ai_key = key + '_AI'
        ai_value = page_data['ai_data'].get(ai_key, '')
        
        # Skip if AI value is empty
        if not ai_value.strip():
            continue
            
        # Clean up values by removing brackets and quotes
        orig_value = str(orig_value).strip('[]').strip('"').strip()
        ai_value = str(ai_value).strip('[]').strip('"').strip()
        
        # Handle array values
        try:
            # Try to parse as arrays
            orig_items = [item.strip() for item in orig_value.split(',') if item.strip()]
            ai_items = [item.strip() for item in ai_value.split(',') if item.strip()]
            
            # Check if all original items are in AI items
            if all(item.lower() in [x.lower() for x in ai_items] for item in orig_items):
                if len(ai_items) > len(orig_items):
                    status = "✓ All and more"
                    status_color = "#4CAF50"
                    row_color = "#e8f5e9"
                else:
                    status = "✓ Match"
                    status_color = "#4CAF50"
                    row_color = "#e8f5e9"
            else:
                status = "✗ Mismatch"
                status_color = "#f44336"
                row_color = "#ffebee"
                
            # Format the values nicely
            orig_display = ', '.join(orig_items)
            ai_display = ', '.join(ai_items)
            
        except:
            # Handle non-array values
            if orig_value.lower() == ai_value.lower():
                status = "✓ Match"
                status_color = "#4CAF50"
                row_color = "#e8f5e9"
                orig_display = orig_value
                ai_display = ai_value
            else:
                status = "✗ Mismatch"
                status_color = "#f44336"
                row_color = "#ffebee"
                orig_display = orig_value
                ai_display = ai_value
            
        # Add row with comparison
        html += f"<tr style='background-color: {row_color};'>"
        html += f"<td style='width: 150px;'>{key}</td>"
        html += f"<td>{orig_display}</td>"
        html += f"<td>{ai_display}</td>"
        html += f"<td style='color: {status_color}; font-weight: bold;'>{status}</td>"
        html += "</tr>"
    
    html += "</table></div>"
    
    # Original data section
    html += "<div style='flex: 1; border: 1px solid #ddd; padding: 15px; border-radius: 5px;'>"
    html += "<h4>Original Data</h4>"
    html += "<table style='width: 100%;'>"
    for key, value in page_data['original_data'].items():
        html += f"<tr><td style='width: 150px;'>{key}</td><td>{value}</td></tr>"
    html += "</table></div>"
    
    # AI data section
    html += "<div style='flex: 1; border: 1px solid #ddd; padding: 15px; border-radius: 5px;'>"
    html += "<h4>AI Predicted Data</h4>"
    html += "<table style='width: 100%;'>"
    for key, value in page_data['ai_data'].items():
        html += f"<tr><td style='width: 150px;'>{key}</td><td>{value}</td></tr>"
    html += "</table></div>"
    
    html += "</div>"
    return html

# Create Gradio interface
demo = gr.Blocks()
with demo:
    gr.Markdown("# CSV Data Viewer")
    gr.Markdown("Navigate through CSV data pages to compare original and AI-predicted values")
    
    page_num = gr.Slider(0, 100, step=1, label="Page Number")
    html_output = gr.HTML()
    
    page_num.change(create_page_view, [page_num], [html_output])

# Launch Gradio interface
demo.launch(server_name="0.0.0.0", server_port=7860, share=True)