Escaping \( delimiter when used outside of math expressions?

[oddhack]

We're using asciidoctor + auto-render extension + offline pre-processing with nodejs to pre-convert math to HTML, and running into an issue where some '\( ... \)' markup in text that isn't math (a regexp example) is getting treated as a math delimiter and messing up the rendering. The \( math delimiters are getting generated from somewhere inside the asciidoctor pipeline and aren't directly under our control.

Is there a way to tag regions of the document so the auto-render extension won't run on them? That may be doable via a CSS style on the regexp, for example. Other alternatives get uglier (trying to use funky Unicode to get the offending markup through to the HTML renderer without the auto-render extension seeing it? Actually changing the regexp? Hacking asciidoctor internals?)

[edemaine]

Perhaps you can use the ignoredTags or ignoredClasses options in auto-render? See this documentation.

If you want to change the regexes, you could use [(] and [)] in place of \( and \). But maybe that's not an option given that they "aren't directly under [y]our control".

Alternatively, you could change your KaTeX delimiters to ignore \( and \), and use something else in their place ($...$ for example, unless you have regexes with $s in them...)

The "proper" way to fix this would be to embed KaTeX into asciidoctor so that they're processed together and proper escaping can happen. Maybe look at some plugins like https://github.com/jirutka/asciidoctor-katex

[oddhack]

Perhaps you can use the ignoredTags or ignoredClasses options in auto-render? See this documentation.

If you want to change the regexes, you could use [(] and [)] in place of \( and \). But maybe that's not an option given that they "aren't directly under [y]our control".

Alternatively, you could change your KaTeX delimiters to ignore \( and \), and use something else in their place ($...$ for example, unless you have regexes with $s in them...)

The "proper" way to fix this would be to embed KaTeX into asciidoctor so that they're processed together and proper escaping can happen. Maybe look at some plugins like https://github.com/jirutka/asciidoctor-katex

Thanks. Eventually I realized moving to POSIX EREs allowed using non-escaped () and given that it's one line in a 3,000 page document, that was sufficient for the time being.

I'd love to see katex more tightly integrated with asciidoctor eventually. Today we have to use a Ruby extension to asciidoctor to replace the MathJax <script> tags with equivalent KaTeX font/script links, then do the nodejs postprocess to avoid runtime rendering and the major performance hit that imposes on our specs.