From 33e798b01b3d8b45aeae31c085a072260ad351df Mon Sep 17 00:00:00 2001 From: Markus Heiser Date: Mon, 20 Oct 2025 11:20:33 +0200 Subject: [fix] TrackerPatternsDB.clean_url: don't delete query argument from new_url (#5339) The query argument for URLs like: - 'http://example.org?q=' --> query_str is 'q=' - 'http://example.org?/foo/bar' --> query_str is 'foo/bar' is a *simple string* and not a key/value dict. This string may only be removed from the URL if one of the patterns matches. BTW get_pretty_url(): keep such a *simple string* in the path element. Closes: https://github.com/searxng/searxng/issues/5299 Signed-off-by: Markus Heiser --- searx/webapp.py | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'searx/webapp.py') diff --git a/searx/webapp.py b/searx/webapp.py index 218959a9c..d7bb37717 100755 --- a/searx/webapp.py +++ b/searx/webapp.py @@ -356,6 +356,12 @@ def get_pretty_url(parsed_url: urllib.parse.ParseResult): path = parsed_url.path path = path[:-1] if len(path) > 0 and path[-1] == '/' else path path = unquote(path.replace("/", " › ")) + + # Keep the query argument for URLs like: + # - 'http://example.org?/foo/bar' --> parsed_url.query is 'foo/bar' + query_args: list[tuple[str, str]] = list(urllib.parse.parse_qsl(parsed_url.query)) + if not query_args and parsed_url.query: + path += (" › .." if len(parsed_url.query) > 24 else " › ") + parsed_url.query[-24:] return [parsed_url.scheme + "://" + parsed_url.netloc, path] -- cgit v1.2.3