app.py

import spaces
import gradio as gr
from pdf2image import convert_from_path
import pytesseract
from PIL import Image
import os

@spaces.GPU
def convert(pdf_file):
    pages = convert_from_path(pdf_file)
    markdown_output = ""
    metadata = {}  # Opcional: puedes extraer metadata con PyMuPDF si lo deseas

    for idx, page_image in enumerate(pages):
        # Realizar OCR
        text = pytesseract.image_to_string(page_image)

        if text.strip() == "":
            # Si no hay texto, insertar un enlace vacío
            markdown_output += f"[imagen]()\n\n"
        else:
            markdown_output += text.strip() + "\n\n"

    return markdown_output.strip(), metadata

gr.Interface(
    convert,
    inputs=[
        gr.File(label="Upload PDF", type="filepath"),
    ],
    outputs=[
        gr.Text(label="Markdown"),
        gr.JSON(label="Metadata"),
    ],
).launch()