semi-throwaway code to parse a wordpress readme.txt-style formated document

parse-wp-changelog.js

// ==UserScript==
// @name     WP Changelog Parser
// @version  0.1
// @match    *://*/*/changelog.txt
// @match    file:///*/*/changelog.txt
// @grant    none
// ==/UserScript==

var doc = document
var btnText, btnHTML
btnText = doc.createElement('button')
btnHTML = doc.createElement('button')
btnText.innerText = 'Plaintext'
btnHTML.innerText = 'Parsed HTML'
doc.body.prepend(btnText)
doc.body.prepend(btnHTML)

btnHTML.onclick = function(){
  convert()
  document.querySelector('#html').style.display = 'block'
}
btnText.onclick = function(){
  var doc = document
  var html = doc.querySelector('#html')
  if (html) html.style.display = 'none'
  doc.querySelector('pre').style.display = 'block'
}

function convert() {
	var doc = document,
		elm = doc.createElement.bind(doc)
	var pre = doc.querySelector('pre') // TODO: check if this mandated by HTML spec
	var html = doc.querySelector('#html') || doc.body.appendChild(elm('div'))
	html.id = 'html', html.innerHTML = ''

	pre.style.display = 'none'
	var lines = pre.textContent.split('\n')
	for (var line of lines) {
		var inList = false
		var lastList = html.querySelector('ul:last-child')
		var [,type,text] = line.match(/^([*=]+)?(.*?)\1?$/)
		// the regex is quite simple, really... :p

		if (type === '==') type = 'h1'
		if (type === '=' ) type = 'h2'
		if (type === '**') type = 'h3'
		if (type == '***') type = 'h4'
		if (type ==  null) type = 'p' 
		if (type === '*') (type = 'li',
			inList = true) // ;3

		if (inList) {
			lastList = lastList || html.appendChild(elm('ul'))
			lastList.appendChild(elm('li'))
			.innerHTML = htmlify(text)
		} else {
			html.appendChild(elm(type))
			.innerHTML = htmlify(text)
		}
	}
}

function htmlify(text) {
	return text
	.replace(/`(.+?)`/, '<code>$1</code')
	.replace(/\[([^[\]]+?)\]\((.+?)\)/, '<a href="$2">$1</a>')
	// gee this regex is complex... good thing the question marks magically fix it somehow! :)
	.replace(/(?<=\s)(https?:\/\/[^\s]+?)(?=[,.;:!?)]?(?:\s|$))/, '<a href="$1">$1</a>')
	// TODO: "tokenized" parsing to prevent corrupting e.g. `[why://would](*anyone* do http://this?)`
	/* .replace(/\b([`*]+)(.+?)\1\b/g, function(match, type, text) {
		if (type == '`') type = 'code'
		if (type == '*') type = 'em' // is that right? idk
	}) */
}

note:
currently (as of revision 1), this is a bit buggy as it doesn't handle list items with linebreaks
(not to mention all the empty <p/>s generated throughout)

also, it doesn't handle inline formatting. or html.
(but i plan to solve the latter by asking jetpack to go back to using inline markdown, not inline html ;)