HTMLParser doesn’t preserve all whitespace. #49

michaelrhodes · 2014-09-24T08:16:18Z

First off, thank you so much for putting this module together; it’s exactly what I wanted jsdom to be! My one pedantic little issue is that when given a full document, the parser seems to discard the whitespace of its outermost elements. It also replaces the doctype declaration.

For example, if you pass this document…

<!doctype html>
<html>
  <head>
    <title>A title</title>
  </head>
  <body>
    <h1>A heading</h1>
  </body>
</html>

…into domino.createDocument…

var read = require('fs').readFileSync
var document = require('domino').createDocument
var html = read('./input.html')

console.log(document(html).outerHTML)

…the output is:

<!DOCTYPE html><html><head>
    <title>A title</title>
  </head>
  <body>
    <h1>A heading</h1>


</body></html>

I think it would be ideal if the output matched the input.

cscott · 2014-09-24T14:15:18Z

So, a couple of standards-related issues here:

document.outerHTML isn't an official interface. It's a domino-specific extension, because it's useful. But keep that in mind. document.documentElement.outerHTML is an standard API, but that only gives you the <html> element.
I need to check that the standard actually allows DOM Text Nodes in the places you want. My experiments seem to indicate that document.documentElement.outerHTML in browsers deletes the whitespace around the <body> and <head> elements. Can you find a browser where this preserves whitespace?
In your example it looks like there is extra whitespace at the end of the </body>, after the </h1>. Is that a typo in your report, or is that really happening? That's a separate bug if so.

michaelrhodes · 2014-09-24T23:14:29Z

Yeah, you’re right, Chrome 37, Firefox 30, Safari 6.1, and Opera 12.16 all remove the whitespace around <head> and <body>.

As for the extra whitespace before </body>, that is really happening, but it also seems to be the behaviour of browsers (at least Chrome). Try comparing the source of github.com with document.documentElement.outerHTML’s output for example.

cscott · 2014-09-25T01:56:04Z

So, all that said, lots of people use domino to parse, mutate, then serialize, so I wouldn't object to adding a new domino-specific api (maybe even tweaking document.outerHTML) to use hidden domino-specific properties to preserve more whitespace. But I think we need to keep the standard DOM APIs browser-compatible.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

HTMLParser doesn’t preserve all whitespace. #49

HTMLParser doesn’t preserve all whitespace. #49

michaelrhodes commented Sep 24, 2014

cscott commented Sep 24, 2014

michaelrhodes commented Sep 24, 2014

cscott commented Sep 25, 2014

HTMLParser doesn’t preserve all whitespace. #49

HTMLParser doesn’t preserve all whitespace. #49

Comments

michaelrhodes commented Sep 24, 2014

cscott commented Sep 24, 2014

michaelrhodes commented Sep 24, 2014

cscott commented Sep 25, 2014