hostname_replace.py 1.6 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546
  1. # SPDX-License-Identifier: AGPL-3.0-or-later
  2. import re
  3. from urllib.parse import urlunparse, urlparse
  4. from searx import settings
  5. from searx.plugins import logger
  6. from flask_babel import gettext
  7. name = gettext('Hostname replace')
  8. description = gettext('Rewrite result hostnames or remove results based on the hostname')
  9. default_on = False
  10. preference_section = 'general'
  11. plugin_id = 'hostname_replace'
  12. replacements = {re.compile(p): r for (p, r) in settings[plugin_id].items()} if plugin_id in settings else {}
  13. logger = logger.getChild(plugin_id)
  14. parsed = 'parsed_url'
  15. _url_fields = ['iframe_src', 'audio_src']
  16. def on_result(request, search, result):
  17. for (pattern, replacement) in replacements.items():
  18. if parsed in result:
  19. if pattern.search(result[parsed].netloc):
  20. # to keep or remove this result from the result list depends
  21. # (only) on the 'parsed_url'
  22. if not replacement:
  23. return False
  24. result[parsed] = result[parsed]._replace(netloc=pattern.sub(replacement, result[parsed].netloc))
  25. result['url'] = urlunparse(result[parsed])
  26. for url_field in _url_fields:
  27. if result.get(url_field):
  28. url_src = urlparse(result[url_field])
  29. if pattern.search(url_src.netloc):
  30. if not replacement:
  31. del result[url_field]
  32. else:
  33. url_src = url_src._replace(netloc=pattern.sub(replacement, url_src.netloc))
  34. result[url_field] = urlunparse(url_src)
  35. return True