Default anchorPattern selects too many links when URI contains (fn|footnote|note)

[kevin1]

The default anchorPattern is /(fn|footnote|note)[:\-_\d]/gi. This could select too many links when the URI of the page contains the strings fn, footnote, or note.

For example, a page containing the following element is served at https://footnote.example.com/:

<a href="#">go to top</a>

Then in the following code, pattern will match because link.href is https://footnote.example.com/#

littlefoot/src/dom/document.ts

Lines 69 to 71 in b2f470f

	return queryAll<HTMLAnchorElement>(document, scope + ' a[href*="#"]').filter(
	(link) => (link.href + link.rel).match(pattern),
	)

I believe this can be fixed by setting the default anchorPattern to /#(fn|footnote|note)[:\-_\d]/gi

[goblindegook]

The setup still tries to match the URL fragment with an existing footnote at the bottom at the document, though. So while it is less efficient because it picks too many link candidates, I suspect it will not lead to incorrect footnotes being created. But I could be wrong... do you have a counter example?

[kevin1]

Hey @goblindegook - sorry I should be more clear, it causes the footnote matching to fail entirely so none of the footnotes are converted by littlefoot. The document remains unmodified. The footnote matching code emits the following error:

[Error] SyntaxError: '#' is not a valid selector.
	querySelectorAll (littlefoot.min.js:2:4794)
	C (littlefoot.min.js:2:4794)
	(anonymous function) (littlefoot.min.js:2:6602)
	map
	B (littlefoot.min.js:2:6525)
	tt (littlefoot.min.js:2:9786)
	(anonymous function) (index:51)

because findReference would parse the href https://footnote.example.com/# into a query of simply #

littlefoot/src/dom/document.ts

Lines 82 to 85 in b2f470f

	const fragment = link.href.split('#')[1]!
	const related = queryAll<E>(
	document,
	'#' + window.CSS.escape(fragment),

[goblindegook]

Thank you, that's clear now. I'll work on a fix.