# -*- encoding: utf-8 -*- # @Author: SWHL # @Contact: liekkaskono@163.com from typing import List import gradio as gr from gradio_pdf import PDF from rapidocr_pdf import RapidOCRPDF custom_css = """ body {font-family: body {font-family: 'Helvetica Neue', Helvetica;} .gr-button {background-color: #4CAF50; color: white; border: none; padding: 10px 20px; border-radius: 5px;} .gr-button:hover {background-color: #45a049;} .gr-textbox {margin-bottom: 15px;} .example-button {background-color: #1E90FF; color: white; border: none; padding: 8px 15px; border-radius: 5px; margin: 5px;} .example-button:hover {background-color: #FF4500;} .tall-radio .gr-radio-item {padding: 15px 0; min-height: 50px; display: flex; align-items: center;} .tall-radio label {font-size: 16px;} .output-image, .input-image, .image-preview {height: 300px !important} """ pdf_extracter = RapidOCRPDF() def get_pdf_result(pdf_path: str, page_num: int = 0): return pdf_extracter(pdf_path, page_num_list=[page_num]) with gr.Blocks( title="RapidOCR 📄 PDF", css="custom_css", theme=gr.themes.Soft() ) as demo: gr.Markdown( "

RapidOCR 📄 PDF

" ) gr.HTML( """
PyPI SemVer2.0 GitHub
""" ) pdf_path = PDF(label="Upload a PDF", interactive=True) page_num = gr.Number(value=0, label="Select Extract Page", minimum=0, maximum=5) btn = gr.Button("Run") results = gr.Dataframe( label="PDF Extract Results", headers=["Index", "Txt", "Score"], datatype=["number", "str", "number"], show_copy_button=True, show_fullscreen_button=True, ) btn.click(get_pdf_result, inputs=[pdf_path, page_num], outputs=[results]) if __name__ == "__main__": demo.launch()