app.py

import gradio as gr
import requests
import re
import json
from urllib.parse import urlparse, urljoin
import time
import random

# Mock contacts database (same as your API)
CONTACTS_DB = [
    {
        "id": 1,
        "first_name": "Sarah",
        "last_name": "Chen",
        "email": "sarah.chen@techflowsolutions.com",
        "phone": "+1-555-0101",
        "job_title": "CTO",
        "company": "TechFlow Solutions",
        "website": "https://techflowsolutions.com"
    },
    {
        "id": 2,
        "first_name": "Marcus",
        "last_name": "Rodriguez",
        "email": "m.rodriguez@techflowsolutions.com",
        "phone": "+1-555-0102",
        "job_title": "Senior Developer",
        "company": "TechFlow Solutions",
        "website": "https://techflowsolutions.com"
    },
    {
        "id": 3,
        "first_name": "Emma",
        "last_name": "Thompson",
        "email": "emma@greenleafconsult.com",
        "phone": "+1-555-0201",
        "job_title": "Managing Partner",
        "company": "GreenLeaf Consulting",
        "website": "https://greenleafconsult.com"
    },
    {
        "id": 4,
        "first_name": "David",
        "last_name": "Park",
        "email": "david.park@greenleafconsult.com",
        "phone": "+1-555-0202",
        "job_title": "Environmental Analyst",
        "company": "GreenLeaf Consulting",
        "website": "https://greenleafconsult.com"
    },
    {
        "id": 5,
        "first_name": "Jessica",
        "last_name": "Williams",
        "email": "jessica@blueskymarketing.net",
        "phone": "+1-555-0301",
        "job_title": "Creative Director",
        "company": "BlueSky Marketing",
        "website": "https://blueskymarketing.net"
    },
    {
        "id": 6,
        "first_name": "Ryan",
        "last_name": "Mitchell",
        "email": "ryan.mitchell@blueskymarketing.net",
        "phone": "+1-555-0302",
        "job_title": "Account Manager",
        "company": "BlueSky Marketing",
        "website": "https://blueskymarketing.net"
    },
    {
        "id": 7,
        "first_name": "Lisa",
        "last_name": "Zhang",
        "email": "l.zhang@quantumdynamics.org",
        "phone": "+1-555-0401",
        "job_title": "Research Director",
        "company": "Quantum Dynamics Corp",
        "website": "https://quantumdynamics.org"
    },
    {
        "id": 8,
        "first_name": "Ahmed",
        "last_name": "Hassan",
        "email": "ahmed.hassan@quantumdynamics.org",
        "phone": "+1-555-0402",
        "job_title": "Quantum Engineer",
        "company": "Quantum Dynamics Corp",
        "website": "https://quantumdynamics.org"
    },
    {
        "id": 9,
        "first_name": "Maria",
        "last_name": "Gonzalez",
        "email": "maria@stellarlogistics.biz",
        "phone": "+1-555-0501",
        "job_title": "Operations Manager",
        "company": "Stellar Logistics",
        "website": "https://stellarlogistics.biz"
    },
    {
        "id": 10,
        "first_name": "James",
        "last_name": "O'Connor",
        "email": "james.oconnor@stellarlogistics.biz",
        "phone": "+1-555-0502",
        "job_title": "Fleet Coordinator",
        "company": "Stellar Logistics",
        "website": "https://stellarlogistics.biz"
    },
    {
        "id": 11,
        "first_name": "Robert",
        "last_name": "Kim",
        "email": "robert.kim@nexusfinancial.pro",
        "phone": "+1-555-0601",
        "job_title": "Senior Advisor",
        "company": "Nexus Financial",
        "website": "https://nexusfinancial.pro"
    },
    {
        "id": 12,
        "first_name": "Catherine",
        "last_name": "Lee",
        "email": "catherine@nexusfinancial.pro",
        "phone": "+1-555-0602",
        "job_title": "Investment Analyst",
        "company": "Nexus Financial",
        "website": "https://nexusfinancial.pro"
    },
    {
        "id": 13,
        "first_name": "Michael",
        "last_name": "Johnson",
        "email": "m.johnson@horizonhealth.care",
        "phone": "+1-555-0701",
        "job_title": "Chief Medical Officer",
        "company": "Horizon Health Systems",
        "website": "https://horizonhealth.care"
    },
    {
        "id": 14,
        "first_name": "Jennifer",
        "last_name": "Adams",
        "email": "jennifer.adams@horizonhealth.care",
        "phone": "+1-555-0702",
        "job_title": "Head Nurse",
        "company": "Horizon Health Systems",
        "website": "https://horizonhealth.care"
    },
    {
        "id": 15,
        "first_name": "Tony",
        "last_name": "Ricci",
        "email": "tony.ricci@phoenixmfg.com",
        "phone": "+1-555-0801",
        "job_title": "Plant Manager",
        "company": "Phoenix Manufacturing",
        "website": "https://phoenixmfg.com"
    },
    {
        "id": 16,
        "first_name": "Linda",
        "last_name": "Martinez",
        "email": "linda.martinez@phoenixmfg.com",
        "phone": "+1-555-0802",
        "job_title": "Quality Control Supervisor",
        "company": "Phoenix Manufacturing",
        "website": "https://phoenixmfg.com"
    },
    {
        "id": 17,
        "first_name": "Patricia",
        "last_name": "White",
        "email": "patricia.white@alpineeducation.edu",
        "phone": "+1-555-0901",
        "job_title": "Director of Programs",
        "company": "Alpine Education Group",
        "website": "https://alpineeducation.edu"
    },
    {
        "id": 18,
        "first_name": "Kevin",
        "last_name": "Brown",
        "email": "kevin.brown@alpineeducation.edu",
        "phone": "+1-555-0902",
        "job_title": "Curriculum Specialist",
        "company": "Alpine Education Group",
        "website": "https://alpineeducation.edu"
    },
    {
        "id": 19,
        "first_name": "Sophia",
        "last_name": "Taylor",
        "email": "sophia@crimsoncreative.studio",
        "phone": "+1-555-1001",
        "job_title": "Art Director",
        "company": "Crimson Creative Studio",
        "website": "https://crimsoncreative.studio"
    },
    {
        "id": 20,
        "first_name": "Alex",
        "last_name": "Cooper",
        "email": "alex.cooper@crimsoncreative.studio",
        "phone": "+1-555-1002",
        "job_title": "Graphic Designer",
        "company": "Crimson Creative Studio",
        "website": "https://crimsoncreative.studio"
    }
]

def extract_domain(url):
    """Extract domain from URL"""
    try:
        parsed = urlparse(url)
        domain = parsed.netloc.lower()
        # Remove www. if present
        if domain.startswith('www.'):
            domain = domain[4:]
        return domain
    except:
        return ""

def find_contacts_by_website(website_url):
    """Find contacts that match the website domain"""
    target_domain = extract_domain(website_url)
    if not target_domain:
        return []
    
    print(f"Searching for domain: {target_domain}")  # Debug info
    
    matching_contacts = []
    for contact in CONTACTS_DB:
        contact_domain = extract_domain(contact['website'])
        print(f"Comparing with: {contact_domain}")  # Debug info
        
        # Exact domain match or subdomain match
        if target_domain == contact_domain or target_domain in contact_domain or contact_domain in target_domain:
            matching_contacts.append(contact)
    
    print(f"Found {len(matching_contacts)} matching contacts")  # Debug info
    return matching_contacts

def simulate_website_scraping(url):
    """Simulate scraping a website and finding contact information"""
    # Add some delay to simulate real scraping
    time.sleep(random.uniform(1, 2))
    
    # Find matching contacts from our database
    contacts = find_contacts_by_website(url)
    
    # Only return contacts if we found exact matches
    # Don't return random contacts if no match found
    return contacts

def search_website_contacts(website_url, max_results=10):
    """Main function to search for contacts on a website"""
    if not website_url:
        return "Please enter a website URL", ""
    
    # Clean up URL
    if not website_url.startswith(('http://', 'https://')):
        website_url = 'https://' + website_url
    
    try:
        # Simulate finding contacts
        contacts = simulate_website_scraping(website_url)
        
        if not contacts:
            return f"No contacts found on {website_url}. \n\nThis website is not in our contact database. Try one of the sample websites listed below, or the website might not have publicly available contact information.", ""
        
        # Limit results
        contacts = contacts[:max_results]
        
        # Format results
        results_text = f"Found {len(contacts)} contacts on {website_url}:\n\n"
        
        contact_data = []
        for i, contact in enumerate(contacts, 1):
            results_text += f"**Contact {i}:**\n"
            results_text += f"• First Name: {contact['first_name']}\n"
            results_text += f"• Last Name: {contact['last_name']}\n"
            results_text += f"• Job Title: {contact['job_title']}\n"
            results_text += f"• Email: {contact['email']}\n"
            results_text += f"• Phone: {contact['phone']}\n"
            results_text += f"• Company: {contact['company']}\n\n"
            
            # Prepare data for CSV/table format
            contact_data.append([
                contact['first_name'],
                contact['last_name'],
                contact['job_title'],
                contact['email'],
                contact['phone'],
                contact['company']
            ])
        
        # Create a simple table format for the second output
        table_text = "First Name,Last Name,Job Title,Email,Phone,Company\n"
        for contact in contacts:
            table_text += f"{contact['first_name']},{contact['last_name']},{contact['job_title']},{contact['email']},{contact['phone']},{contact['company']}\n"
        
        return results_text, table_text
        
    except Exception as e:
        return f"Error searching website: {str(e)}", ""

def get_all_available_websites():
    """Get list of all available websites from the database"""
    websites = list(set([contact['website'] for contact in CONTACTS_DB]))
    return "\n".join(sorted(websites))

# Create Gradio interface
with gr.Blocks(title="Website Contact Finder", theme=gr.themes.Soft()) as app:
    gr.HTML("""
    <div style="text-align: center; padding: 20px;">
        <h1>🔍 Website Contact Finder</h1>
        <p>Search any website to find contact information including names, job titles, phone numbers, and emails</p>
    </div>
    """)
    
    with gr.Row():
        with gr.Column(scale=2):
            website_input = gr.Textbox(
                label="Website URL",
                placeholder="Enter website URL (e.g., techflowsolutions.com or https://greenleafconsult.com)",
                value=""
            )
            
            max_results = gr.Slider(
                minimum=1,
                maximum=20,
                value=5,
                step=1,
                label="Maximum Results"
            )
            
            search_btn = gr.Button("🔍 Search Contacts", variant="primary", size="lg")
            
        with gr.Column(scale=1):
            gr.HTML("""
            <div style="background: #f0f0f0; padding: 15px; border-radius: 10px;">
                <h3>💡 Tips:</h3>
                <ul>
                    <li>Enter any website URL</li>
                    <li>Works with or without https://</li>
                    <li>Try our sample websites below</li>
                    <li>Results include full contact details</li>
                </ul>
            </div>
            """)
    
    with gr.Row():
        results_display = gr.Textbox(
            label="Contact Results",
            lines=15,
            max_lines=30,
            show_copy_button=True
        )
        
        csv_output = gr.Textbox(
            label="CSV Format (Copy & Paste into Excel)",
            lines=15,
            max_lines=30,
            show_copy_button=True
        )
    
    # Sample websites section
    with gr.Accordion("📋 Available Sample Websites", open=False):
        sample_websites = gr.Textbox(
            label="Sample Websites from Database",
            value=get_all_available_websites(),
            lines=10,
            interactive=False
        )
    
    # Quick search buttons for sample websites
    with gr.Row():
        gr.HTML("<h3>🚀 Quick Search Sample Websites:</h3>")
    
    with gr.Row():
        quick_btn1 = gr.Button("TechFlow Solutions", size="sm")
        quick_btn2 = gr.Button("GreenLeaf Consulting", size="sm")
        quick_btn3 = gr.Button("BlueSky Marketing", size="sm")
        quick_btn4 = gr.Button("Quantum Dynamics", size="sm")
    
    with gr.Row():
        quick_btn5 = gr.Button("Stellar Logistics", size="sm")
        quick_btn6 = gr.Button("Nexus Financial", size="sm")
        quick_btn7 = gr.Button("Horizon Health", size="sm")
        quick_btn8 = gr.Button("Phoenix Manufacturing", size="sm")
    
    # Event handlers
    search_btn.click(
        fn=search_website_contacts,
        inputs=[website_input, max_results],
        outputs=[results_display, csv_output]
    )
    
    # Quick search button handlers
    quick_btn1.click(lambda: "techflowsolutions.com", outputs=website_input)
    quick_btn2.click(lambda: "greenleafconsult.com", outputs=website_input)
    quick_btn3.click(lambda: "blueskymarketing.net", outputs=website_input)
    quick_btn4.click(lambda: "quantumdynamics.org", outputs=website_input)
    quick_btn5.click(lambda: "stellarlogistics.biz", outputs=website_input)
    quick_btn6.click(lambda: "nexusfinancial.pro", outputs=website_input)
    quick_btn7.click(lambda: "horizonhealth.care", outputs=website_input)
    quick_btn8.click(lambda: "phoenixmfg.com", outputs=website_input)
    
    # Examples
    gr.Examples(
        examples=[
            ["techflowsolutions.com", 5],
            ["https://greenleafconsult.com", 3],
            ["blueskymarketing.net", 2],
            ["quantumdynamics.org", 4]
        ],
        inputs=[website_input, max_results]
    )

if __name__ == "__main__":
    app.launch()