# Copyright (c) Opendatalab. All rights reserved. import base64 import json import os import time import zipfile from pathlib import Path import re import uuid import pymupdf # os.system('pip install -U magic-pdf==0.10.5') os.system('pip uninstall -y magic-pdf') os.system('pip install git+https://github.com/opendatalab/MinerU.git@dev') os.system('wget https://github.com/opendatalab/MinerU/raw/dev/scripts/download_models_hf.py -O download_models_hf.py') os.system('python download_models_hf.py') with open('/home/user/magic-pdf.json', 'r') as file: data = json.load(file) # 修改处1：允许cpu和gpu自动选择，cuda不可用自动回落到cpu data['device-mode'] = "cuda" if os.system("command -v nvidia-smi > /dev/null") == 0 else "cpu" if os.getenv('apikey'): data['llm-aided-config']['title_aided']['api_key'] = os.getenv('apikey') data['llm-aided-config']['title_aided']['enable'] = True with open('/home/user/magic-pdf.json', 'w') as file: json.dump(data, file, indent=4) from gradio_pdf import PDF import gradio as gr from loguru import logger from magic_pdf.data.data_reader_writer import FileBasedDataReader from magic_pdf.libs.hash_utils import compute_sha256 from magic_pdf.tools.common import do_parse, prepare_env def read_fn(path): disk_rw = FileBasedDataReader(os.path.dirname(path)) return disk_rw.read(os.path.basename(path)) def parse_pdf(doc_path, output_dir, end_page_id, is_ocr, layout_mode, formula_enable, table_enable, language): os.makedirs(output_dir, exist_ok=True) try: file_name = f"{str(Path(doc_path).stem)}_{time.time()}" pdf_data = read_fn(doc_path) parse_method = "ocr" if is_ocr else "auto" local_image_dir, local_md_dir = prepare_env(output_dir, file_name, parse_method) do_parse( output_dir, file_name, pdf_data, [], parse_method, False, end_page_id=end_page_id, layout_model=layout_mode, formula_enable=formula_enable, table_enable=table_enable, lang=language, f_dump_orig_pdf=False, ) return local_md_dir, file_name except Exception as e: logger.exception(e) def compress_directory_to_zip(directory_path, output_zip_path): try: with zipfile.ZipFile(output_zip_path, 'w', zipfile.ZIP_DEFLATED) as zipf: for root, dirs, files in os.walk(directory_path): for file in files: file_path = os.path.join(root, file) arcname = os.path.relpath(file_path, directory_path) zipf.write(file_path, arcname) return 0 except Exception as e: logger.exception(e) return -1 def image_to_base64(image_path): with open(image_path, "rb") as image_file: return base64.b64encode(image_file.read()).decode('utf-8') def replace_image_with_base64(markdown_text, image_dir_path): pattern = r'\!\[(?:[^\]]*)\]\(([^)]+)\)' # 使用lambda简化函数 return re.sub(pattern, lambda match: f"![{match.group(1)}](data:image/jpeg;base64,{image_to_base64(os.path.join(image_dir_path, match.group(1)))})", markdown_text) def to_markdown(file_path, end_pages, is_ocr, layout_mode, formula_enable, table_enable, language): file_path = to_pdf(file_path) # 修改处2：不再限制最大20页 local_md_dir, file_name = parse_pdf(file_path, './output', end_pages - 1, is_ocr, layout_mode, formula_enable, table_enable, language) archive_zip_path = os.path.join("./output", compute_sha256(local_md_dir) + ".zip") zip_archive_success = compress_directory_to_zip(local_md_dir, archive_zip_path) if zip_archive_success == 0: logger.info("压缩成功") else: logger.error("压缩失败") md_path = os.path.join(local_md_dir, file_name + ".md") with open(md_path, 'r', encoding='utf-8') as f: txt_content = f.read() md_content = replace_image_with_base64(txt_content, local_md_dir) new_pdf_path = os.path.join(local_md_dir, file_name + "_layout.pdf") return md_content, txt_content, archive_zip_path, new_pdf_path latex_delimiters = [{"left": "$$", "right": "$$", "display": True}, {"left": '$', "right": '$', "display": False}] def init_model(): from magic_pdf.model.doc_analyze_by_custom_model import ModelSingleton try: model_manager = ModelSingleton() model_manager.get_model(False, False) model_manager.get_model(True, False) return 0 except Exception as e: logger.exception(e) return -1 model_init = init_model() logger.info(f"model_init: {model_init}") with open("header.html", "r") as file: header = file.read() latin_lang = ['af', 'az', 'bs', 'cs', 'cy', 'da', 'de', 'es', 'et', 'fr', 'ga', 'hr', 'hu', 'id', 'is', 'it', 'ku', 'la', 'lt', 'lv', 'mi', 'ms', 'mt', 'nl', 'no', 'oc', 'pi', 'pl', 'pt', 'ro', 'rs_latin', 'sk', 'sl', 'sq', 'sv', 'sw', 'tl', 'tr', 'uz', 'vi', 'french', 'german'] arabic_lang = ['ar', 'fa', 'ug', 'ur'] cyrillic_lang = ['ru', 'rs_cyrillic', 'be', 'bg', 'uk', 'mn', 'abq', 'ady', 'kbd', 'ava', 'dar', 'inh', 'che', 'lbe', 'lez', 'tab'] devanagari_lang = ['hi', 'mr', 'ne', 'bh', 'mai', 'ang', 'bho', 'mah', 'sck', 'new', 'gom', 'sa', 'bgc'] other_lang = ['ch', 'en', 'korean', 'japan', 'chinese_cht', 'ta', 'te', 'ka'] all_lang = ['', 'auto'] + other_lang + latin_lang + arabic_lang + cyrillic_lang + devanagari_lang def to_pdf(file_path): with pymupdf.open(file_path) as f: if f.is_pdf: return file_path pdf_bytes = f.convert_to_pdf() tmp_file_path = os.path.join(os.path.dirname(file_path), f"{uuid.uuid4()}.pdf") with open(tmp_file_path, 'wb') as tmp_pdf_file: tmp_pdf_file.write(pdf_bytes) return tmp_file_path if __name__ == "__main__": with gr.Blocks() as demo: gr.HTML(header) with gr.Row(): with gr.Column(scale=5): file = gr.File(label="Please upload a PDF or image", file_types=[".pdf", ".png", ".jpeg", ".jpg"]) # 修改处3：gradio界面页数最大改为99999 max_pages = gr.Slider(1, 99999, 10, step=1, label='Max convert pages') with gr.Row(): layout_mode = gr.Dropdown(["doclayout_yolo"], label="Layout model", value="doclayout_yolo") language = gr.Dropdown(all_lang, label="Language", value='auto') with gr.Row(): formula_enable = gr.Checkbox(label="Enable formula recognition", value=True) is_ocr = gr.Checkbox(label="Force enable OCR", value=False) table_enable = gr.Checkbox(label="Enable table recognition(test)", value=True) with gr.Row(): change_bu = gr.Button("Convert") clear_bu = gr.ClearButton(value="Clear") pdf_show = PDF(label='PDF preview', interactive=False, visible=True, height=800) with gr.Column(scale=5): output_file = gr.File(label="convert result", interactive=False) md = gr.Markdown() md_text = gr.TextArea(lines=45) file.change(fn=to_pdf, inputs=file, outputs=pdf_show) change_bu.click(fn=to_markdown, inputs=[file, max_pages, is_ocr, layout_mode, formula_enable, table_enable, language], outputs=[md, md_text, output_file, pdf_show]) clear_bu.add([file, md, pdf_show, md_text, output_file, is_ocr]) demo.launch(ssr_mode=True)