You should use HTML.parser and focus on the conversion to markdown. The way you parse HTML in the convert function is very inefficient and can easily produce incorrect results with valid HTML (e.g. <p class="some>stuff">some text</p> )