docs/lib/get-english-headings.js

import { fromMarkdown } from 'mdast-util-from-markdown'
import { toString } from 'mdast-util-to-string'
import { visit } from 'unist-util-visit'
import findPage from './find-page.js'

// for any translated page, first get corresponding English markdown
// then get the headings on both the translated and English pageMap
// finally, create a map of translation:English for all headings on the page
export default function getEnglishHeadings(page, context) {
  // Special handling for glossaries, because their headings are
  // generated programatically.
  if (page.relativePath.endsWith('/github-glossary.md')) {
    // Return an object of `{ localized-term: english-slug }`
    const languageGlossary = context.site.data.glossaries.external
    return languageGlossary.reduce((prev, curr) => {
      prev[curr.term] = curr.slug
      return prev
    }, {})
  }

  const translatedHeadings = getHeadings(page.markdown)
  if (!translatedHeadings.length) return

  const englishPage = findPage(
    `/en/${page.relativePath.replace(/.md$/, '')}`,
    context.pages,
    context.redirects
  )
  if (!englishPage) return

  // FIX  there may be bugs if English headings are updated before Crowdin syncs up :/
  const englishHeadings = getHeadings(englishPage.markdown)
  if (!englishHeadings.length) return

  // return a map from translation:English
  return Object.assign(
    ...translatedHeadings.map((k, i) => ({
      [k]: englishHeadings[i],
    }))
  )
}

function getHeadings(markdown) {
  const ast = fromMarkdown(markdown)
  const headings = []

  visit(ast, (node) => {
    if (node.type !== 'heading') return
    if (![2, 3, 4].includes(node.depth)) return
    headings.push(toString(node))
  })

  return headings
}