app.py

import subprocess
import sys

# Install the 'marker' package from GitHub if not already installed
try:
    import marker
except ImportError:
    subprocess.check_call([sys.executable, "-m", "pip", "install", "git+https://github.com/VikParuchuri/marker.git"])

# Verify the marker package is installed and check its contents
import marker
print("Available modules in marker:", dir(marker))

from marker.pdf import PDF  # Updated import path
import os
import gradio as gr

def convert_pdf(input_file, output_format):
    """
    Convert a PDF file to the specified format.

    Args:
        input_file: Uploaded PDF file.
        output_format: Desired output format (Markdown, HTML, JSON).

    Returns:
        Path to the converted file.
    """
    pdf = PDF(input_file.name)  # Initialize the PDF object

    output_file_path = f"output.{output_format.split(' ')[0].lower()}"

    if output_format == "Markdown (.md)":
        with open(output_file_path, "w") as f:
            f.write(pdf.to_markdown())
    elif output_format == "HTML (.html)":
        with open(output_file_path, "w") as f:
            f.write(pdf.to_html())
    elif output_format == "JSON (.json)":
        with open(output_file_path, "w") as f:
            f.write(pdf.to_json())
    else:
        return "Unsupported output format!"

    return output_file_path

output_format_dropdown = gr.inputs.Dropdown(
    ["Markdown (.md)", "HTML (.html)", "JSON (.json)"],
    label="Select Output File Format",
)
file_input = gr.inputs.File(label="Upload PDF File", type="file")

output_file = gr.outputs.File(label="Download Converted File")

gr_interface = gr.Interface(
    fn=convert_pdf,
    inputs=[file_input, output_format_dropdown],
    outputs=output_file,
    title="PDF Converter",
    description="Upload a PDF file and select the desired output format (Markdown, HTML, or JSON).",
)

gr_interface.launch()