AI-Driven Universal Web Data Extraction Platform

A production-grade, MCP-enabled universal web scraping platform with MongoDB storage and advanced anti-bot (antigravity) mechanisms.

🎯 Features

• Dual Scraping Engines: Static (Requests + BeautifulSoup) and Dynamic (Playwright)

• Auto-Detection: Automatically selects the appropriate scraper based on page content

• Anti-Bot Protection: User-Agent rotation, rate limiting, robots.txt compliance, stealth mode

• MongoDB Storage: Persists all scraped data with full metadata

• MCP Integration: Exposes scraping as tools for LLM invocation

• Export Options: JSON and CSV export capabilities

Related MCP server: mcp-web-search

📁 Project Structure

d:\mcp\
├── requirements.txt          # Python dependencies
├── config.py                 # Configuration settings
├── main.py                   # FastAPI MCP server entry point
├── scraper/
│   ├── static_scraper.py     # Requests + BeautifulSoup scraper
│   ├── dynamic_scraper.py    # Playwright scraper
│   └── strategy_selector.py  # Auto-detection logic
├── antigravity/
│   ├── user_agents.py        # User-Agent rotation
│   ├── throttle.py           # Request delays & rate limiting
│   ├── robots_validator.py   # robots.txt compliance
│   └── stealth.py            # Playwright stealth configuration
├── database/
│   ├── mongodb.py            # MongoDB connection & operations
│   └── models.py             # Pydantic data models
├── mcp/
│   └── tools.py              # MCP tool definitions
├── utils/
│   ├── normalizer.py         # Data normalization
│   └── exporter.py           # CSV/JSON export
├── tests/                    # Test suite
└── docs/
    └── README.md             # This file

🚀 Quick Start

1. Install Dependencies

cd d:\mcp
pip install -r requirements.txt
playwright install chromium

2. Start MongoDB

Ensure MongoDB is running on localhost:27017 (or update MONGODB_URI in config.py).

3. Run the Server

python main.py

The server will start at http://localhost:8000.

4. Test the API

Open http://localhost:8000/docs for interactive Swagger documentation.

🔌 API Endpoints

Example Scrape Request

curl -X POST "http://localhost:8000/scrape" \
  -H "Content-Type: application/json" \
  -d '{"url": "https://example.com", "auto_detect": true}'

🧠 MCP Tool Usage

The platform exposes a scrape_website tool via MCP:

# Tool Schema
{
    "name": "scrape_website",
    "parameters": {
        "url": "string (required)",
        "dynamic": "boolean (default: false)",
        "auto_detect": "boolean (default: true)",
        "store_in_mongodb": "boolean (default: true)"
    }
}

🛡️ Anti-Bot (Antigravity) Features

1. User-Agent Rotation: 20+ realistic browser User-Agents

2. Request Throttling: 1-5 second random delays between requests

3. Rate Limiting: Max 10 requests per domain per minute

4. robots.txt Compliance: Respects crawling restrictions

5. Playwright Stealth Mode: Disables automation detection flags

📊 MongoDB Schema

scraped_data Collection

{
  "_id": "ObjectId",
  "url": "string",
  "scraped_at": "ISO timestamp",
  "scraper_type": "static | dynamic",
  "content": {
    "title": "string",
    "text": "string",
    "links": ["string"]
  },
  "metadata": {
    "status_code": "number",
    "response_time": "number",
    "user_agent": "string"
  }
}

scrape_logs Collection

{
  "url": "string",
  "timestamp": "ISO timestamp",
  "success": "boolean",
  "error": "string | null"
}

🧪 Running Tests

cd d:\mcp
pytest tests/ -v

⚖️ Ethical Considerations

• Always respects robots.txt directives

• Implements polite crawling with delays

• Only scrapes publicly accessible content

• Rate limiting prevents server overload

• Designed for responsible use

📋 Limitations

• Cannot bypass authentication or CAPTCHAs

• JavaScript-heavy SPAs may require dynamic scraping

• Some sites may detect and block scraping despite stealth measures

• Rate limiting may slow down bulk operations

🔮 Future Scope

• Proxy rotation support

• CAPTCHA solving integration

• Distributed scraping with task queues

• Advanced content extraction (structured data, tables)

• Scheduled/recurring scrapes

• WebSocket real-time updates

📄 License

This project is for educational purposes.