ReadPDFx - OCR PDF MCP Server

Official MCP SDK STDIO Server - MCP Protocol 2025-06-18 Compliant

ReadPDFx is a comprehensive MCP (Model Context Protocol) server that provides intelligent OCR and PDF processing capabilities using the official MCP SDK with STDIO transport. It automatically detects whether a PDF contains digital text or scanned images and applies the appropriate processing method.

⚡ Quick Start (STDIO Server)

1. Install Dependencies

pip install -r requirements.txt

2. Validate Installation

# Test imports and tools
python validate_tools.py

3. Client Integration

The server runs via STDIO protocol - configure your MCP client:

Claude Desktop:

{
  "mcpServers": {
    "ocr-pdf": {
      "command": "python",
      "args": ["d:/AI/MCP/python/ocr_pdf_mcp/mcp_server_stdio.py"],
      "env": {}
    }
  }
}

🚀 Features

• 🎯 Official MCP SDK: Built with official FastMCP framework
• 📡 STDIO Transport: Standard MCP protocol over STDIO
• 🧠 Smart PDF Processing: Automatically detects digital vs scanned content
• 🔧 5 OCR Tools: Text extraction, OCR processing, combined operations
• 🌐 Universal Client Support: Claude Desktop, LM Studio, Continue.dev, Cursor
• ⚡ Lightweight: ~200 lines vs 800+ in HTTP implementation
• 🛡️ Production Ready: Comprehensive error handling and logging
• 📋 Auto Tool Registration: Decorators handle tool discovery

🔧 Installation

Prerequisites

• Python 3.8+
• Tesseract OCR

Windows

# Install Python dependencies
pip install -r requirements.txt

# Install Tesseract
choco install tesseract

macOS

pip install -r requirements.txt
brew install tesseract

Linux

pip install -r requirements.txt
sudo apt-get install tesseract-ocr

📋 Available Tools

1. Smart PDF Processing

Intelligent processing with automatic OCR detection:

{
  "name": "process_pdf_smart",
  "arguments": {
    "pdf_path": "/path/to/document.pdf",
    "language": "eng"
  }
}

2. PDF Text Extraction

Direct text extraction from digital PDFs:

{
  "name": "extract_pdf_text", 
  "arguments": {
    "pdf_path": "/path/to/document.pdf",
    "page_range": "1-5"
  }
}

3. OCR Processing

OCR on image files:

{
  "name": "perform_ocr",
  "arguments": {
    "image_path": "/path/to/image.png",
    "language": "eng"
  }
}

4. PDF Structure Analysis

Analyze document structure and metadata:

{
  "name": "analyze_pdf_structure",
  "arguments": {
    "pdf_path": "/path/to/document.pdf"
  }
}

5. Batch Processing

Process multiple files:

{
  "name": "batch_process_pdfs",
  "arguments": {
    "input_directory": "/path/to/pdfs/",
    "output_directory": "/path/to/output/",
    "file_pattern": "*.pdf"
  }
}

🔌 Client Integration

Claude Desktop

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "readpdfx": {
      "command": "python",
      "args": ["path/to/readpdfx/run.py"],
      "env": {
        "PYTHONPATH": "path/to/readpdfx"
      }
    }
  }
}

LM Studio

Configure MCP server with:

• Command: python
• Args: path/to/readpdfx/run.py
• URL: http://localhost:8000 (HTTP mode)

Continue.dev

Add to config.json:

{
  "contextProviders": [
    {
      "name": "mcp",
      "params": {
        "command": "python",
        "args": ["path/to/readpdfx/run.py"]
      }
    }
  ]
}

Cursor

Configure in settings.json:

{
  "mcp.servers": {
    "readpdfx": {
      "command": "python",
      "args": ["path/to/readpdfx/run.py"]
    }
  }
}

📁 See client-configs/ for detailed integration guides.

🌐 API Endpoints

MCP Protocol Endpoints

• POST /mcp/initialize - Initialize MCP session
• POST /mcp/tools/list - List available tools
• POST /mcp/tools/call - Call MCP tools
• GET /mcp/manifest - Get MCP manifest

HTTP Endpoints

• GET /health - Health check
• POST /jsonrpc - JSON-RPC 2.0 endpoint
• GET /docs - API documentation
• GET /tools - Tools discovery

🔧 Configuration

Environment Variables

MCP_SERVER_HOST=localhost      # Server host
MCP_SERVER_PORT=8000           # Server port  
TESSERACT_CMD=/usr/bin/tesseract  # Tesseract path
PYTHONPATH=.                   # Python path

Config Files

• mcp.json - MCP Protocol configuration
• mcp-config.yaml - YAML configuration
• pyproject.toml - Python project config
• package.json - Node.js compatibility

🐳 Docker & Kubernetes

Docker Deployment

Quick Start with Docker

# Build and run with Docker
docker build -t ocr-pdf-mcp .
docker run -p 8000:8000 -v ./pdf-test:/app/pdf-test:ro ocr-pdf-mcp

# Or use Docker Compose
docker-compose up -d

Automated Docker Deployment

# Linux/macOS
./scripts/docker-deploy.sh run

# Windows
scripts\docker-deploy.bat run

Available Docker commands:

• build - Build Docker image only
• run - Build and run container (default)
• start - Start container (assumes image exists)
• stop - Stop running container
• logs - Show container logs
• clean - Stop container and remove image
• status - Show container status

Kubernetes Deployment

Deploy to Kubernetes

# Quick deployment
./scripts/k8s-deploy.sh deploy

# Manual deployment
kubectl apply -f k8s/ -n ocr-pdf-mcp

Kubernetes Resources

• Deployment: k8s/deployment.yaml - Main application deployment
• Service: k8s/deployment.yaml - Service exposure
• Ingress: k8s/ingress.yaml - External access
• ConfigMap: k8s/configmap.yaml - Configuration management
• HPA: k8s/hpa.yaml - Horizontal Pod Autoscaler

Kubernetes Commands

# Scale deployment
kubectl scale deployment ocr-pdf-mcp --replicas=5 -n ocr-pdf-mcp

# Port forward for local access
kubectl port-forward svc/ocr-pdf-mcp-service 8000:80 -n ocr-pdf-mcp

# View logs
kubectl logs -f deployment/ocr-pdf-mcp -n ocr-pdf-mcp

# Check status
kubectl get pods,svc,ingress -n ocr-pdf-mcp

Production Considerations

Multi-stage Build

Use Dockerfile.prod for optimized production builds:

docker build -f Dockerfile.prod -t ocr-pdf-mcp:prod .

Environment Variables

# Docker
docker run -e LOG_LEVEL=INFO -e CORS_ORIGINS="*" ocr-pdf-mcp

# Kubernetes - update ConfigMap
kubectl edit configmap ocr-pdf-mcp-config -n ocr-pdf-mcp

Persistent Storage

# Add to deployment.yaml
volumeMounts:
- name: pdf-storage
  mountPath: /app/pdf-test
volumes:
- name: pdf-storage
  persistentVolumeClaim:
    claimName: pdf-storage-pvc

🧪 Testing

Run Tests

python test_mcp_server.py

Manual Testing

# Health check
curl http://localhost:8000/health

# List tools  
curl -X POST http://localhost:8000/mcp/tools/list \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc": "2.0", "method": "tools/list", "id": 1}'

# Call tool
curl -X POST http://localhost:8000/mcp/tools/call \
  -H "Content-Type: application/json" \
  -d '{
    "jsonrpc": "2.0", 
    "method": "tools/call",
    "params": {
      "name": "process_pdf_smart",
      "arguments": {"pdf_path": "/path/to/test.pdf"}
    },
    "id": 1
  }'

📊 Performance

• Startup Time: < 2 seconds
• Memory Usage: ~50MB base
• Throughput: 10+ PDFs/minute
• Concurrent Requests: Up to 100
• File Size Limit: 100MB per file

🛠️ Development

Development Mode

python run_server.py --dev --port 8000

Project Structure

readpdfx/
├── run.py                 # Simple production runner
├── run_server.py          # Advanced runner with options  
├── mcp_server.py          # Core MCP server
├── mcp_tools.py           # MCP tools implementation
├── mcp_types.py           # MCP Protocol types
├── mcp_server_runner.py   # HTTP server runner
├── client-configs/        # Client integration guides
├── backup/                # Legacy files
└── tests/                 # Test files

Adding New Tools

1. Define tool schema in mcp_tools.py
2. Implement tool handler method
3. Register tool in MCPToolsRegistry
4. Update tests and documentation

🐛 Troubleshooting

Common Issues

Server won't start

# Check port availability
netstat -an | grep 8000

# Try different port
python run_server.py --port 8001

OCR not working

# Check Tesseract installation
tesseract --version

# Install language data
tesseract --list-langs

Permission errors

• Ensure read access to PDF files
• Check write permissions for output directory
• Run with appropriate user privileges

Connection timeout

• Verify server is running: curl http://localhost:8000/health
• Check firewall settings
• Try HTTP instead of direct MCP connection

Debug Mode

python run_server.py --dev

📈 Monitoring

Health Check

curl http://localhost:8000/health

Metrics (Future)

• Request count and latency
• Tool usage statistics
• Error rates and types
• Resource utilization

🤝 Contributing

1. Fork the repository
2. Create feature branch: git checkout -b feature/new-tool
3. Make changes and add tests
4. Submit pull request

Development Setup

git clone https://github.com/irev/mcp-readpdfx.git
cd readpdfx
pip install -r requirements-dev.txt
python test_mcp_server.py

📄 License

MIT License - see LICENSE file.

🔗 Links

• Repository: https://github.com/irev/mcp-readpdfx
• Issues: https://github.com/irev/mcp-readpdfx/issues
• Documentation: https://github.com/irev/mcp-readpdfx#readme
• MCP Protocol: Model Context Protocol Specification

🏆 Acknowledgments

• MCP Protocol Team for the specification
• FastAPI for the web framework
• Tesseract OCR for text recognition
• PyPDF2 and pdfplumber for PDF processing

---

Made with ❤️ for the MCP community