change markdown parsing dependency

Use the one used by https://github.com/cerbos/antora-llm-generator

Author: osfameron

lib/markdown-for-llm.js

@@ -1,67 +1,16 @@
 'use strict'
 
-const {convertHtmlToMarkdown} = require('dom-to-semantic-markdown')
-const {JSDOM} = require('jsdom')
-const File = require('vinyl')
-
-function overrideElementProcessing (element) {
+const { NodeHtmlMarkdown } = require('node-html-markdown')
+const nhm = new NodeHtmlMarkdown()
 
-  if (element.tagName?.toLowerCase() === 'a') {
-    if (element.className === 'anchor') {
-      return [{type: 'custom', blank: true}]
-    }
-    let href = element.getAttribute('href')
-    const hasProtocol = /^[a-z]+:\/\//i
-    if (href && !href.match(hasProtocol)) {
-      href = href.replace(/\.html/, '.md')
-      const content = toMarkdown(element.innerHTML || href)
-      return [{type: 'link', href, content}]
-    }
-  }
-
-  if (element.classList?.contains("admonitionblock")) {
-    element.classList.remove('admonitionblock')
-    const admonition = element.className.toUpperCase()
-    const content = toMarkdown(
-      element.querySelector("td.content").innerHTML)
-
-    return [{
-      type: 'custom', 
-      admonition,
-      content
-    }]
-  }
-}
-
-function renderCustomNode (node) {
-  if (node.blank) {
-    return ''
-  }
-  if (node.admonition) {
-    const body = node.content.split('\n').map(line => `> ${line}`).join('\n')
-    return `\n> [!${node.admonition}]\n${body}\n\n`
-  }
-}
-
-function toMarkdown (html) {
-  const dom = new JSDOM(html)
-  const markdown = convertHtmlToMarkdown(
-    html,
-    {
-      overrideDOMParser: new dom.window.DOMParser(),
-      overrideElementProcessing,
-      renderCustomNode
-    }
-  )
-  dom.window.close()
-  return markdown
-}
+const File = require('vinyl')
 
 function markdownify(page) {
   const html = page.contents.toString()
 
   const link = `[View original HTML](${page.pub.url})\n\n`
-  const markdown = toMarkdown(html)
+
+  const markdown = link + nhm.translate(html)
 
   page.out.path = page.out.path.replace(/\.html$/, '.md')
   page.pub.url = page.pub.url.replace(/\.html$/, '.md')
@@ -106,8 +55,18 @@ module.exports.register = function ({ playbook, config }) {
         && page.pub
         && page.out)
 
-    for (const page of pages) {
-      markdownify(page)
+    const CHUNK_SIZE = 100
+
+    for (let i = 0; i < pages.length; i += CHUNK_SIZE) {
+      const chunk = pages.slice(i, i + CHUNK_SIZE)
+
+      for (const page of chunk) {
+        markdownify(page)
+      }
+
+      if (i + CHUNK_SIZE < pages.length) {
+        await new Promise(resolve => setImmediate(resolve))
+      }
     }
   })
  this.once('beforePublish', async ({ siteCatalog }) => {