Add MathJax Support - #477

Hello and thank you for your pull request. Support for math sequences is one of the longest standing and most highly requested features. There have been long discussions on what the delimiters for inline and block math sequences should be, but we have unfortunately not yet come to a decision. The latex-style you've implemented is certainly a contender, but there are some drawbacks relating to unintended sequences and graceful fallback.

Before we can move forward with any implementation, we'd have to agree on a format (read: delimiters) and a spec. That said, I did have a quick look at the code and the implementation generally looks good. If we do choose to go with dollars, this PR should make an excellent basis.

You have prompted me to try and form a proposal so we can hopefully finally get this sorted. I will include you in future discussion. Thanks again!

Hi @marcusklaas, actually I've noticed the discussions. I desperately need the math support, so I rushed to implement a basic working version for my personal needs. It's true that I haven't consider too much about the edges cases and this one shouldn't be a complete and sound solution.

I am a heavy markdown user and this is certainly one of the best implementation. It would be perfect if it has any kind of math support.

Thanks

Any content on a line that starts with $$ is currently ignored, and something like this, which I think is relatively common:

$$ \LaTeX $$

breaks the rest of the document.

Just mentioning to keep track. Hopefully we can get this merged soon!

Edit: Another bug is that math content is not escaped in the HTML renderer. This is left untouched and the <b> is parsed as a tag:

$a<b>c$

Hey @marcusklaas did you ever put that proposal together? Or what is the status of planning for this?

#6 is a mess of bikeshedding over the delimiter. $...$ and $$...$$ is basically ubiquitous at this point (used by StackExchange, Pandoc, Jupyter, and now GitHub $E=mc^2$) so should definitely be included. Choosing other delimiters shouldn't block implementation work since they could be added later.

Assuming that logic is satisfactory, what is blocking this from being merged? There hasn't been too much of a discussion on KaTeX vs MathJax and pre- vs. post- rendering, but the work here is already mostly done, and could be adapted later if desired.

@litao91 are you interested in doing work to make this mergable, if needed? I see you are at least keeping this branch up to date with upstream

Closed in favor of #734.