Create a struct called MarkdownFlavor with static properties with an array of the renderers to use.

public struct MarkdownFlavor {
    public static commonMark(renderers: [...])
    public static gfm(renderers: [...])


    public let renderers: [SyntaxRenderer]
}

We use a struct instead of an enum so it can be extended and have other cases added to it.

Support dollars ($InlineMath$, $$DisplayMath$$) syntax as Markdown delimiters for LaTeX math expressions. Possibly also add single backslash \(InlineMath\), \[DisplayMath\] and double backslash \\(InlineMath\\), \\[DisplayMath\\] syntax. Provide an enable|disable for each implemented syntax.

Such LaTeX math sections would be passed through to html for rendering by a library such as MathJax (more comprehensive) or KaTeX (faster).

Various Markdown editing environments support LaTeX math. (e.g. Atom, MacDown, MWeb). It would be nice if markdown containing LaTeX math could be processed by SwiftMark. :-)

Aside, but related FYI, Apple has added LaTeX mathematical equation support to Pages, Numbers, Keynote, and iBooks Author. Apple: HT207569, HT202501

In my use case, dollars are most frequently used in Markdown with the double backslash as a distant second. I don't use single backslash notation except for some testing & evaluation. One such test set is here: 004.55_Markdown-CLI-Evaluation.

There is also a community syntax survey of Math in MarkDown on GitHub. The appendix summary table shows the dollars to be the a widely used Markdown math delimiter syntax.

• Modularity
• Increased speed
• Replace MarkdownRenderer with Markdown

As the algorithm is currently written, the rendering is probably fairly slow. We need to run some performance tests and try some other methods of rendering:

• Using an array of bits (with the vapor/bits package)
• Using an array of Strings and reducing it

The way that the renderers are registered is incredibly inefficient. This should be changed.

The GFM spec states that many blocks should be automatically ended if you reach the end of a container block. This could be done if you parse the container blocks first and then their contents second, along with having the proper regex pattern at the end of your pattern (ending with $).