Rewrite fuzzy rule to not contain the trailing ? when querystring is …

…empty
openzim · Mar 18, 2024 · 60d174b · 60d174b
1 parent fcaf1ad
commit 60d174b
Show file tree

Hide file tree

Showing 2 changed files with 2 additions and 2 deletions.
diff --git a/src/warc2zim/url_rewriting.py b/src/warc2zim/url_rewriting.py
@@ -71,7 +71,7 @@
  r"=[^&]+).*",
  "replace": r"youtube.fuzzy.replayweb.page/\1\2",
  },
- {"pattern": r"([^?]+\?)[\d]+$", "replace": r"\1"},
+ {"pattern": r"([^?]+)\?[\d]+$", "replace": r"\1"},
  {
  "pattern": r"(?:www\.)?youtube(?:-nocookie)?\.com\/(youtubei\/[^?]+).*(videoId["
  r"^&]+).*",

diff --git a/tests/test_fuzzy_rules.py b/tests/test_fuzzy_rules.py
@@ -87,7 +87,7 @@ def test_fuzzyrules_google_video_infos(google_video_info_case):
  params=[
  ContentForTests(
  "www.example.com/page?1234",
- "www.example.com/page?",
+ "www.example.com/page",
  ),
  ContentForTests(
  "www.example.com/page?foo=1234",