smoldocling/README.md

Smoldocling CLI

A command-line interface for processing document images and PDFs using Smoldocling-256-preview model.

Installation

1. Clone this repository
2. Install the required dependencies:

   pip install -r requirements.txt
   

Usage

The CLI supports processing one or multiple document images and PDFs at once. The processed output will be saved as HTML files.

Basic usage:

python smoldocling_cli.py input1.png input2.jpg input3.pdf

Specify output directory:

python smoldocling_cli.py -o custom_output input1.png document.pdf

Arguments

• input_files: One or more input files (images or PDFs) to process
• -o, --output-dir: Output directory for HTML files (default: 'output')

Example

python smoldocling_cli.py document1.png document2.pdf -o processed_docs

This will:

1. Process document1.png and generate document1.html
2. Process document2.pdf and generate document2.html (containing all pages in a single file)
3. Create a directory called 'processed_docs' if it doesn't exist
4. Save all HTML files in the processed_docs directory

Notes

• The script will automatically create the output directory if it doesn't exist
• Each input image file will generate a corresponding HTML file with the same name (but .html extension)
• PDF files will generate a single HTML file containing all pages
• Currently, PDF processing is limited to the first 3 pages due to model limitations
• Failed processing of one file won't stop the processing of other files
• Error messages will be printed to stderr
• The model is loaded only once for processing multiple files