benchmark.py 2.0 KB

1234567891011121314151617181920212223242526272829303132333435363738394041
  1. import zipfile
  2. import os
  3. import shutil
  4. code_path = os.environ.get('GITHUB_WORKSPACE')
  5. pdf_dev_path = "/home/quyuan/data"
  6. pdf_res_path = "/home/quyuan/code/Magic-PDF/Magic-PDF/Magic-PDF/ci/magic-pdf"
  7. def test_cli():
  8. cmd = 'cd %s && export PYTHONPATH=. && find %s -type f -name "*.pdf" | xargs -I{} python magic_pdf/cli/magicpdf.py pdf-command --pdf {}' % (code_path, pdf_dev_path)
  9. os.system(cmd)
  10. for annotaion_name in os.listdir(os.join(pdf_dev_path, "output")):
  11. if annotaion_name.endswith('.pdf'):
  12. for pdf_res_path in os.listdir(pdf_res_path):
  13. if ".md" in os.join(pdf_res_path, annotaion_name, "auto"):
  14. prefix = annotaion_name.split('_')[-2]
  15. if not os.path.exists(os.join(pdf_dev_path, prefix)):
  16. os.makedirs(os.join(pdf_dev_path, prefix))
  17. shutil.copy(os.join(pdf_res_path, annotaion_name, "auto", annotaion_name + ".md"), os.join(pdf_dev_path, prefix, annotaion_name + ".md"))
  18. def calculate_score():
  19. cmd = "cd %s && export PYTHONPATH=. && python tools/clean_photo.py --tool_name annotations --download_dir %s" % pdf_dev_path
  20. os.system(cmd)
  21. cmd = "cd %s && export PYTHONPATH=. && python tools/clean_photo.py --tool_name magicpdf --download_dir %s" % (pdf_dev_path)
  22. os.system(cmd)
  23. cmd = "cd %s && export PYTHONPATH=. && python tools/markdown_calculate.py --tool_name pdf-command --download_dir %s --results %s" % (pdf_dev_path, os.join(pdf_dev_path, "result.json"))
  24. os.system(cmd)
  25. def extrat_zip(zip_file_path, extract_to_path):
  26. if zipfile.is_zipfile(zip_file_path):
  27. with zipfile.ZipFile(zip_file_path, 'r') as zip_ref:
  28. zip_ref.extractall(extract_to_path)
  29. print(f'Files extracted to {extract_to_path}')
  30. else:
  31. print(f'{zip_file_path} is not a zip file')
  32. if __name__ == "__main__":
  33. extrat_zip(os.join(pdf_dev_path, 'output.zip'), os.join(pdf_dev_path,'datasets'))
  34. test_cli()
  35. calculate_score()