Docling Document Extractor

A FastAPI-based application that extracts layout, tables, and content from documents (PDFs) using native Docling features.

Features

• Advanced Table Extraction: Uses TableFormer in ACCURATE mode for high-fidelity table structure recognition.
• OCR Support: Built-in OCR for scanned documents and images.
• Image Analysis: Automatic classification and description of figures and images.
• Markdown Export: Converts documents to Markdown with embedded table structures and image captions.
• Structured Data: Provides access to the underlying structured data model of the document.

Installation

1. Clone the repository:

   git clone <repository_url>
   cd Doc_extracter

2. Create a virtual environment (recommended):

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies:

   pip install -r requirements.txt

Configuration

1. Copy the example environment file:

   cp .env.example .env

2. Open .env and configure your credentials if using cloud-based features (e.g., specific enrichment services). Note: logic for loading .env is not currently in main.py, so you may need to export these variables in your shell or use python-dotenv if required by underlying libraries.

Usage

Starting the Server

Run the FastAPI server using uvicorn:

uvicorn app.main:app --reload

The API will be available at http://127.0.0.1:8000.

Using the Client

A simple client script is provided to test the extraction:

python client_example.py <path_to_document.pdf>

Example:

python client_example.py sample.pdf

API Documentation

POST /extract

Extracts content from an uploaded document.

• Request: multipart/form-data with a file field.
• Response: JSON object containing:
  • markdown: The extracted text in Markdown format.
  • structured_data: The raw structured data from Docling.

Visit http://127.0.0.1:8000/docs for the interactive Swagger UI.