from docsifer.core.html_cleaner import clean_html def test_removes_style_script_noscript() -> None: html = ( "" "

hi

" ) out = clean_html(html) assert " None: html = ( "" '
also-hidden
' '
spaced
' '' "

visible

" ) out = clean_html(html) assert "secret" not in out assert "also-hidden" not in out assert "spaced" not in out assert "aria" not in out assert "visible" in out def test_empty_input_returns_empty() -> None: assert clean_html("") == "" def test_malformed_html_does_not_raise() -> None: out = clean_html("<<<<><><>") assert isinstance(out, str)