Removal of whitespace-only text nodes

[dmfrancisco]

Thank you for creating and maintaining this useful library 🙌

I noticed that whitespace was not being preserved between tags. After reading the code I realize that this is on purpose and done by this code (in the browser version):

else if (node.nodeType === NodeTypes.TEXT) {
  return node.textContent.trim() === '' ? null : unescape(node.textContent);
}

Personally I found this unexpected. For example:

convert("<span>hello</span> <span>world!</span>")

Renders to:

helloworld!

But I was expecting:

hello world!

Another example I noticed on a code block that had been syntax highlighted:

convert("<pre><span>Hello</span>\n<span>World</span></pre>")

Renders to:

HelloWorld

But I was expecting:

Hello
World

For my application I created a fork where I'm simply not trimming the whitespace. Let me know if you agree with the expectations and think this behavior should change. Thanks again 👍

[pveyes]

You're right, I forgot the exact reason why I decided to trim the whitespace for text nodes but I think it make sense to not remove it. Now that you bring it up I think I'll add pre-processing for text node (so we can still trim before rendering)

Would you create PR for this changes? I've seen your fork but I'd suggest some changes in the test: remove the whitespace caused by template literal for existing test (so the existing snapshot doesn't need to change)