app.py

import os
import gradio as gr
import main

#os.environ["CUDA_VISIBLE_DEVICES"]='0'
#os.environ["USE_GPU"]="True"


def predict_from_pdf(pdf_file):
    upload_dir = "./catalogue/"
    os.makedirs(upload_dir, exist_ok=True)

    try:
        # Save the uploaded file to a temporary location
        dest_path = os.path.join(upload_dir, pdf_file.name)
        with open(dest_path, "wb") as f:
            with open(pdf_file.name, "rb") as uploaded_file:
                f.write(uploaded_file.read())

        # Process the PDF
        df, response = main.process_pdf_catalog(dest_path)
        return df, response
    except Exception as e:
        return None, f"Error: {str(e)}"


pdf_examples = [
    ["examples/flexpocket.pdf"],
    ["examples/ASICS_Catalog.pdf"],
]

demo = gr.Interface(
    fn=predict_from_pdf,
    inputs=gr.File(label="Upload PDF Catalog"),
    outputs=["json", "text"],
    examples=pdf_examples,
    title="Open Source PDF Catalog Parser",
    description="Efficient PDF catalog processing using PyMuPDF and OpenLLM",
    article="Uses MinerU for layout analysis and DeepSeek-7B for structured extraction"
)

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