Bug in RE_COMMENT results in removal of uncommented content

[mvastola]

I just ran into an issue with this library where (what I think is) valid, uncommented HTML is being removed at

inline-source/src/parse.js

Line 25 in bc91b4f

const html = context.html.replace(RE_COMMENT, '');

due to the RE_COMMENT regex here:

inline-source/src/parse.js

Line 16 in bc91b4f

const RE_COMMENT = /(<!--[^[i][\S\s]+?--\s?>)/gm;

A minimal example is this HTML fragment.

<!--
--><img src="images/foo.svg"><!-- comment -->

This regex will match the entire fragment (including the uncommented <img src="..."> tag) and causes inline-source to ignore the image.

I'm having trouble groking the regex personally (as I'm not sure what the i is for, and I can't make sense of the mismatched brackets), but the problem seems to be due to the use of a newline to separate the open/close tags for the comment, without any other characters.

If I change the + into a *, the regex works for me, but -- as I'm not clear on exactly how this regex works -- I can't speak to if that creates any false positives or what the best solution is.

Unfortunately my use case ingests HTML generated by a third party library, meaning I can't easily generate the HTML without the useless comment.

Thanks for your help!

[mvastola]

Oh wait. I just realized the second [ appears to be recognized in JS as a literal [. If that's the case, while I'm still not sure why the i is in the pattern, the following regex seems likely to do the trick:

/(<!--(?!=[\[i])[\S\s]+?--\s?>)/

Happy to create a PR if you'd like.

Note: I escaped the literal [ for ease of understanding above, but I imagine that can be removed if desired.