Looks complete to me.

I think this makes sense. Two questions:

1. Is it possible to catch self-closing script tags? Like, <script src="./blah.js" />. This is more a hypothetical question. Such a tag is considered not valid, but many times in my life I've made the mistake (rather than closing with </script>).
2. Could I convince you to do another PR (or maybe within this PR?) for style tags?

1. I’ll try it. I hope Nokogiri can distinguish self-closing from empty content.

2. What do you want to check in the style tag? How could it be invalid?

   <style>
   body { color: black; background: white; }
   em { font-style: normal; color: red; }
   </style>
   

What do you want to check in the style tag?

Ugh, sorry, I promise I know how to computer. I meant

<link rel="stylesheet" type="text/css" href="styles.css">

Which is already covered.

Nokogiri can not distinguish self-closing node from empty one. This element <script src="./blah.js" /> is already parsed to <script src="./blah.js"></script>. You don’t have the original state of element on this stage. That’s why I would not check for self-closing scripts.

You could use regexp for each script node to see if it's self-closing.

@parkr I can’t figure out how because the node is already parsed:

node.to_s == '<script src="./blah.js"></script>'

You can add you solution on top of my PR, if you know how. Keep in mind this Stack Overflow answer.

@gjtorikian, you said in #30 (comment)

I don't think valid HTML is something I'm interested in here.

The self-closing check is HTML validation, so we could spare this check.

Rebased.