update_ahmia_blacklist.py 944 B

1234567891011121314151617181920212223242526272829303132333435
  1. #!/usr/bin/env python
  2. # lint: pylint
  3. # SPDX-License-Identifier: AGPL-3.0-or-later
  4. """This script saves `Ahmia's blacklist`_ for onion sites.
  5. Output file: :origin:`searx/data/ahmia_blacklist.txt` (:origin:`CI Update data
  6. ... <.github/workflows/data-update.yml>`).
  7. .. _Ahmia's blacklist: https://ahmia.fi/blacklist/
  8. """
  9. from os.path import join
  10. import requests
  11. from searx import searx_dir
  12. URL = 'https://ahmia.fi/blacklist/banned/'
  13. def fetch_ahmia_blacklist():
  14. resp = requests.get(URL, timeout=3.0)
  15. if resp.status_code != 200:
  16. raise Exception("Error fetching Ahmia blacklist, HTTP code " + resp.status_code)
  17. return resp.text.split()
  18. def get_ahmia_blacklist_filename():
  19. return join(join(searx_dir, "data"), "ahmia_blacklist.txt")
  20. if __name__ == '__main__':
  21. blacklist = fetch_ahmia_blacklist()
  22. with open(get_ahmia_blacklist_filename(), "w", encoding='utf-8') as f:
  23. f.write('\n'.join(blacklist))