feat: add transformations for latex in oxhugofm (#510)

Hrishikesh Barman

2023-09-29 2f99339dcf93ef50b766263297785a32d9c35250

feat: add transformations for latex in oxhugofm (#510)

ox-hugo currently supports the following syntax for latex equations:
- https://orgmode.org/manual/LaTeX-fragments.html
- https://ox-hugo.scripter.co/doc/equations

This syntax is supported by mathjax as is mentioned in the ox-hugo documentation.

But quartz uses remark-math which has some issues with the \( \) syntax.
See https://github.com/remarkjs/remark-math/issues/39

This change adds few more transformations to the OxHugoFlavouredMarkdown
plugin, which makes a best effort conversion of this syntax into what
the Quartz Latex transformer plugin supports.

With these changes, the generated files show latex formatting with
default quartz configuration.

Sidenote on `\_` escape by ox-hugo:

ox-hugo escapes, _ using \_, we match against it after we transform
equations into what quartz supports($$ and $).

This could be achieved using lookaround like regex as follows
```js
(?<=(\$|\$\$)[\s\S]*) -> Positive lookbehind for $ or $$
\\_ -> Matches \_
(?=[\s\S]*(?:\1)) Positive lookahead for $ or $$ if matched
const escapedUnderscoreRegex = new RegExp(/(?<=(\$|\$\$)[\s\S]*)\\_(?=[\s\S]*(?:\1))/, "g")
````

But since lookahead/behind can slow things down on large files, we just
look up all equations with $ and $$ delimiters and then try replacing \_

 docs/features/OxHugo compatibility.md

@@ -32,6 +32,7 @@
  - `replaceFigureWithMdImg`: Whether to replace `<figure/>` with `![]()`
- Formatting
  - `removeHugoShortcode`: Whether to remove hugo shortcode syntax (`{{}}`)
  - `replaceOrgLatex`: Whether to replace org-mode formatting for latex fragments with what `Plugin.Latex` supports.

> [!warning]
>

 quartz/plugins/transformers/oxhugofm.ts

@@ -9,6 +9,9 @@
  removeHugoShortcode: boolean
  /** Replace <figure/> with ![]() */
  replaceFigureWithMdImg: boolean

  /** Replace org latex fragments with $ and $$ */
  replaceOrgLatex: boolean
}

const defaultOptions: Options = {
@@ -16,12 +19,27 @@
  removePredefinedAnchor: true,
  removeHugoShortcode: true,
  replaceFigureWithMdImg: true,
  replaceOrgLatex: true,
}

const relrefRegex = new RegExp(/\[([^\]]+)\]\(\{\{< relref "([^"]+)" >\}\}\)/, "g")
const predefinedHeadingIdRegex = new RegExp(/(.*) {#(?:.*)}/, "g")
const hugoShortcodeRegex = new RegExp(/{{(.*)}}/, "g")
const figureTagRegex = new RegExp(/< ?figure src="(.*)" ?>/, "g")
// \\\\\( -> matches \\(
// (.+?) -> Lazy match for capturing the equation
// \\\\\) -> matches \\)
const inlineLatexRegex = new RegExp(/\\\\\((.+?)\\\\\)/, "g")
// (?:\\begin{equation}|\\\\\(|\\\\\[) -> start of equation
// ([\s\S]*?) -> Matches the block equation
// (?:\\\\\]|\\\\\)|\\end{equation}) -> end of equation
const blockLatexRegex = new RegExp(
  /(?:\\begin{equation}|\\\\\(|\\\\\[)([\s\S]*?)(?:\\\\\]|\\\\\)|\\end{equation})/,
  "g",
)
// \$\$[\s\S]*?\$\$ -> Matches block equations
// \$.*?\$ -> Matches inline equations
const quartzLatexRegex = new RegExp(/\$\$[\s\S]*?\$\$|\$.*?\$/, "g")

/**
 * ox-hugo is an org exporter backend that exports org files to hugo-compatible
@@ -67,6 +85,23 @@
          return `![](${src})`
        })
      }

      if (opts.replaceOrgLatex) {
        src = src.toString()
        src = src.replaceAll(inlineLatexRegex, (value, ...capture) => {
          const [eqn] = capture
          return `$${eqn}$`
        })
        src = src.replaceAll(blockLatexRegex, (value, ...capture) => {
          const [eqn] = capture
          return `$$${eqn}$$`
        })

        // ox-hugo escapes _ as \_
        src = src.replaceAll(quartzLatexRegex, (value) => {
          return value.replaceAll("\\_", "_")
        })
      }
      return src
    },
  }