app.py

# app.py
import fitz                     # PyMuPDF
from markdownify import markdownify as md
import json
import gradio as gr

def convert_pdf_to_markdown(path):
    """Extract each page as HTML, convert to Markdown."""
    doc = fitz.open(path)
    pages_md = []
    for i, page in enumerate(doc, start=1):
        html = page.get_text("html") or ""
        # Clean conversion: collapse multiple newlines
        page_md = md(html).strip()
        pages_md.append({"page": i, "markdown": page_md})
    return pages_md

def process_upload(pdf_file, output_format):
    """
    pdf_file: tempfile-like object from Gradio
    output_format: "markdown" or "json"
    """
    # Convert and collect
    pages = convert_pdf_to_markdown(pdf_file.name)
    
    if output_format == "markdown":
        # Join all pages
        full_md = "\n\n---\n\n".join(p["markdown"] for p in pages)
        return full_md
    else:
        # Return pretty JSON
        return json.dumps({"pages": pages}, indent=2, ensure_ascii=False)

# Gradio interface
demo = gr.Interface(
    fn=process_upload,
    inputs=[
        gr.File(label="Upload your PDF", file_types=[".pdf"]),
        gr.Radio(choices=["markdown", "json"],
                 value="markdown",
                 label="Output format")
    ],
    outputs=gr.Code(label="Converted Output"),
    title="PDF → Markdown/JSON Converter",
    description=(
        "Upload a PDF and get back a professionally converted Markdown "
        "or a structured JSON with each page’s Markdown. "
        "PDFs with images or complex tables may still need manual review."
    ),
    examples=[
        # you can add example PDFs here if desired
    ]
)

if __name__ == "__main__":
    demo.launch(server_name="0.0.0.0", server_port=7860)