app.py

import gradio as gr
from docling.document_converter import DocumentConverter
import spaces


def convert_document(file, method):
    if method == "Docling":
        # Load document and convert it using Docling
        converter = DocumentConverter()
        result = converter.convert(file.name)
    
        # Check available attributes in DoclingDocument
        available_attributes = dir(result.document)
        document = result.document
    
    
        # Output
        converted_text = result.document.export_to_markdown()
        metadata = {
            "Available Attributes": available_attributes
        }
    
        return converted_text, metadata
    elif method == "Marker":
        return 'unsupported method', ''
    else:
        return 'unknown method', ''

with gr.Blocks() as app:
    gr.Markdown("# Document Converter")
    gr.Markdown("Upload a document, choose the backend, and get the converted text with metadata.")

    file_input = gr.File(label="Upload Document")
    method_input = gr.Radio(["Docling", "Marker"], label="Choose Conversion Backend")
    output_text = gr.Textbox(label="Converted Document")
    output_metadata = gr.JSON(label="Metadata")

    convert_button = gr.Button("Convert")
    convert_button.click(
        convert_document,
        inputs=[file_input, method_input],
        outputs=[output_text, output_metadata]
    )

app.launch(debug=True, show_error=True)