Hi @eos87 ,
This pr is about how sweet lxml is :)

for elem in body.xpath("//*"):

I would expect lxml walk through objects inside body (I hope you also expected it), but it does not. It walks starting from the root (html element), which is wrong for us. Check a screenshot from my debugger:

Here you can see what result gives body.xpath("//*") in our context.

parent.remove(elem)

Here we see another lxml's feature. When we execute this command, lxml removes element, let's say an empty <a> and a content after it which belongs to parent. See a screenshot.

In this context elem is a reference to an empty <a> tag which is inside <p>, lxml removes an empty tag and rest of the content after it in context of parent's <p>.

Please, test this pr with cases ContentCleanUpPlugin was developed for 🍺 🍺 🍺