Ponysearch/searx/plugins/tracker_url_remover.py

43 lines
1.1 KiB
Python
Raw Normal View History

# SPDX-License-Identifier: AGPL-3.0-or-later
# pylint: disable=missing-module-docstring
import re
from urllib.parse import urlunparse, parse_qsl, urlencode
from flask_babel import gettext
regexes = {
re.compile(r'utm_[^&]+'),
re.compile(r'(wkey|wemail)[^&]*'),
re.compile(r'(_hsenc|_hsmi|hsCtaTracking|__hssc|__hstc|__hsfp)[^&]*'),
re.compile(r'&$'),
}
name = gettext('Tracker URL remover')
description = gettext('Remove trackers arguments from the returned URL')
default_on = True
2017-02-12 15:06:01 +01:00
preference_section = 'privacy'
def on_result(_request, _search, result):
if 'parsed_url' not in result:
return True
query = result['parsed_url'].query
2015-06-18 14:27:15 +02:00
if query == "":
return True
parsed_query = parse_qsl(query)
changes = 0
2019-10-14 15:09:39 +02:00
for i, (param_name, _) in enumerate(list(parsed_query)):
for reg in regexes:
if reg.match(param_name):
parsed_query.pop(i - changes)
changes += 1
result['parsed_url'] = result['parsed_url']._replace(query=urlencode(parsed_query))
result['url'] = urlunparse(result['parsed_url'])
break
return True