Docling Document Processing System

A full-stack web application for processing documents using IBM's Docling library with AI-powered analysis and Q&A capabilities.

Features

• Document Processing: Upload and process various document formats (PDF, DOCX, PPTX, XLSX, HTML, Images)
• Advanced OCR: Text extraction from images and scanned documents
• AI Analysis: Document scoring for AI readiness, content quality, and search optimization
• Q&A System: Interactive questions and answers about processed documents
• Multiple Export Formats: Download results as Markdown, JSON, or HTML
• Modern UI: Responsive web interface with dark mode support

Architecture

• Frontend: HTML/CSS/JavaScript with Tailwind CSS
• Backend: FastAPI server with Docling integration
• Processing: IBM Docling for document conversion and analysis

Quick Start

Prerequisites

• Python 3.8 or higher
• pip (Python package installer)

Installation & Running

1. Clone or download this repository

2. Navigate to the Backend directory:

   cd Backend

3. Run the startup script:

   python start_server.py

   This script will:

   • Install all required dependencies
   • Set up Docling and its dependencies
   • Start the FastAPI server

4. Access the application:

   • Open your browser and go to: http://localhost:8000
   • API documentation: http://localhost:8000/docs

Usage

Processing Documents

1. Upload a Document:

   • Drag and drop a file or click to select
   • Supported formats: PDF, DOCX, PPTX, XLSX, HTML, Images (PNG, TIFF, JPEG)

2. Configure Processing Options:

   • Choose export format (Markdown, JSON, Plain Text)
   • Set processing threads
   • Enable/disable OCR and table structure detection
   • Configure enrichment features (code understanding, formula extraction, etc.)

3. Process the Document:

   • Click "Extract Information"
   • Wait for processing to complete (progress indicator will show)

4. View Results:

   • Download processed results as a ZIP file
   • Get AI readiness scores
   • Ask questions about the document content

API Endpoints

The backend provides a REST API with the following main endpoints:

• POST /api/upload - Upload and process documents
• GET /api/status/{doc_id} - Check processing status
• GET /api/download/{doc_id} - Download results
• POST /api/score/{doc_id} - Get document scoring
• POST /api/qa/{doc_id} - Document Q&A
• DELETE /api/cleanup/{doc_id} - Clean up temporary files

Configuration Options

Processing Options

• Export Format: Choose output format (Markdown, JSON, Text)
• Threading: Number of CPU threads for processing
• OCR Settings: Engine selection (EasyOCR, Tesseract) and language
• Enrichment Features:
  • Code Understanding: Analyze code blocks
  • Formula Understanding: Extract mathematical formulas
  • Picture Classification: Classify images and diagrams
  • Picture Description: Generate image descriptions

OCR Enhancement Options

• OCR Enhancement: Improve text extraction quality
• Auto Deskew: Correct rotated or skewed text
• Noise Reduction: Remove image noise
• Contrast Enhancement: Improve text visibility

Document Scoring

The system provides three scoring metrics:

1. AI Readiness (0-100): How well-structured the document is for machine processing
2. Content Quality (0-100): Text clarity, grammar, and coherence
3. Search Readiness (0-100): Keyword usage and searchability

Troubleshooting

Common Issues

1. Server Won't Start:

   • Check if Python 3.8+ is installed
   • Ensure port 8000 is not in use
   • Run python start_server.py from the Backend directory

2. Document Processing Fails:

   • Check file format is supported
   • Ensure file is not corrupted
   • Try reducing processing threads if memory issues occur

3. Dependencies Issues:

   • Try installing requirements manually: pip install -r requirements.txt
   • Check Python version compatibility

Manual Installation

If the startup script doesn't work, install dependencies manually:

# Install core dependencies
pip install fastapi uvicorn[standard] python-multipart pydantic

# Install Docling
pip install docling

# Start server
python api_server.py

Development

Project Structure

├── Backend/
│   ├── api_server.py          # Main FastAPI application
│   ├── start_server.py        # Startup script
│   ├── requirements.txt       # Python dependencies
│   └── docling/              # Docling library (cloned)
├── Frontend/
│   └── Frontend.html         # Web interface
└── README.md                 # This file

Adding Features

1. Backend: Modify api_server.py to add new endpoints
2. Frontend: Update Frontend.html for UI changes
3. Processing: Extend Docling integration in the API server

License

This project integrates with IBM's Docling library. Please check Docling's license terms for commercial usage.

Support

For issues related to:

• Document processing: Check Docling documentation
• Web interface: Review browser console for errors
• API errors: Check server logs and /docs endpoint