以通用方式使用html.ParseFragment

Question

Using the experimental code.google.com/p/go.net/html package, we can use ParseFragment to parse some sub-section of an HTML document.

Like this:

var s = `
    <option id="foo">first</option>
    <option Class="tester">second</option>
    <option>third</option>
`
doc, err := html.ParseFragment(strings.NewReader(s), &html.Node{
    Type: html.ElementNode,
    Data: "body",
    DataAtom: atom.Body,
})

This works fine for most elements. But it doesn't seem to work when certain elements are at the root position of the HTML, like tbody, tr, and td (and perhaps others, not sure). It simply ignores the tags and only gives the text content.

This can be remedied by providing the semantically correct parent instead of atom.Body, but that requires that we know in advance what the HTML will be.

I'd hoped there was a generic root like atom.DocumentFragment, but I don't see that. So is there some way to use this in such a manner that it'll work with any arbitrary HTML fragment?

Answer 1

ParseFragment is always context-sensitive because it follows the HTML5 fragment-parsing algorithm. That algorithm is designed for implementing the DOM innerHTML property, and the correct tree to generate from a given innerHTML string depends on the surrounding context (especially whether the context is in a table or not).

So the html package has no way to parse an HTML fragment independently of its context.

If you need more information about how the parsing depends on the context, see http://www.whatwg.org/specs/web-apps/current-work/multipage/parsing.html#reset-the-insertion-mode-appropriately