[itecz26]

canonical() of a combined selector, :is() and :where() strips a leading * from its argument with lstrip("*"), which also eats the * of a *| namespace wildcard: parse('foo *|bar')[0].canonical() returns foo |bar, so "any namespace" turns into "no namespace" and re-parsing it translates to descendant::bar instead of descendant::*[local-name() = 'bar'], matching different elements. Selector.canonical() and Negation.canonical() already only strip a * followed by #, ., [ or :; this uses the same check at the three lstrip sites, with round-trip cases added to test_css_export.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 99.79%. Comparing base (ec61e4b) to head (ba2a704).

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #204   +/-   ##
=======================================
  Coverage   99.79%   99.79%           
=======================================
  Files           3        3           
  Lines         987      987           
  Branches      162      162           
=======================================
  Hits          985      985           
  Misses          1        1           
  Partials        1        1           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[wRAR]

Can you please look into missing coverage?

[itecz26]

the missed line was the strip in CombinedSelector.canonical(): before, lstrip("*") ran on any subselector longer than one char, so it was always covered as a no-op, but nothing tested an actual *.foo after a combinator. added foo *.bar and foo > *[bar] cases, which cover it, and patch coverage is back to master's level (the remaining miss at parser.py:1065 is pre-existing).