Normalizes the HTML in old documentation from help.adobe.com that is Creative Commons licensed (to make it easier to clean up and convert into other formats, like Markdown, with tools like pandoc)

normalize-help-adobe-com-html.js

document.body.querySelector("#ahpod")?.remove();
document.body.querySelector("#mboxScriptContainer")?.remove();
let h1 = document.querySelector("#content_wrapper h1:first-of-type");
let article = document.querySelector(
  "table#inner_content_table td:first-of-type"
);
article.querySelector("div:first-of-type")?.remove();
article.querySelector("#chcPromo")?.remove();
article.querySelectorAll("script")?.forEach((element) => element.remove());
article.querySelector("#userprefs")?.remove();
// article.querySelector("#related")?.remove();
article.querySelector("#footer")?.remove();
article.querySelector("#minitoc")?.remove();
article
  .querySelectorAll('a[name^="WS"')
  ?.forEach((element) => element.remove());
article
  .querySelectorAll('a[href^="#top"]')
  ?.forEach((element) => element.remove());
article
  .querySelectorAll('*[width^="NaN%"]')
  ?.forEach((element) => element.removeAttribute("width"));
article
  .querySelectorAll("a")
  ?.forEach((element) => element.removeAttribute("target"));
document.body.querySelectorAll("*")?.forEach((element) => {
  element.removeAttribute("class");
  element.removeAttribute("id");
  element.removeAttribute("style");
  element.removeAttribute("valign");
  element.removeAttribute("headers");
  element.removeAttribute("border");
  element.removeAttribute("cellpadding");
  element.removeAttribute("cellspacing");
  element.removeAttribute("align");
  element.removeAttribute("valign");
});
article
  .querySelectorAll("a")
  ?.forEach((element) => element.removeAttribute("onclick"));
document.body.innerHTML = "";
document.body.appendChild(h1);
document.body.appendChild(article);
document.body.innerHTML = document.body.innerHTML
  .replaceAll(/( |\n)<samp>\s+/g, "$1<samp>") //no whitespace after <samp>
  .replaceAll(/\s+<\/samp>/g, "</samp>") // no whitespace before </samp>
  .replaceAll(/(<\/\w+>)\s+(\.|,|:|\))/g, "$1$2") // no whitespace between end tag and punctuation
  .replaceAll(/\s+(<\/a>)/g, "$1"); // no whitespace before </a>