Element.cssSelector stop returning value #root after upgrading to 1.17 from 1.14

[smileLilith]

Hi team,

After upgrading from v1.14.3 to 1.17.1 I ran into an issue where Element.cssSelector stopped returning value #root for element with tag name #root. I have found nothing relevant in the release notes indicating the change of behavior.

Example

Document doc = Jsoup.parse("<p>hello</p>");
System.out.println(doc.cssSelector());

Output v1.14.3

#root

Output v1.17.1

(empty string)

[smileLilith]

It seems the behavior was broken in this commit 9170b1d where cssSelectorComponent isn't executed for element of instance Document where previously it was.

cssSelectorComponent contains the line

jsoup/src/main/java/org/jsoup/nodes/Element.java

Line 924 in 8eecef3

if (parent() == null || parent() instanceof Document) // don't add Document to selector, as will always have a html node

, however, it's not reachable due to

jsoup/src/main/java/org/jsoup/nodes/Element.java

Line 905 in 8eecef3

while (el != null && !(el instanceof Document)) {

@jhy was there any reason behind that?

[jhy]

@jhy was there any reason behind that?

Yes, as noted in the comment if you're getting a selector for an element it is going to descend from html and so it doesn't need the #root selector. And the #root nodename is not really documented, or supported outside of jsoup. I felt it was wrong to be leaking in to the selector.

Can you tell me more about your use case - what do need a Document selector generated from a Document for?

Perhaps we could change the cssSelector method to return #root if it's called on a Document directly, or there's no other contents to the selector.

[jhy]

Also - returning #root is just wrong as it won't match anything - it would look for elements with an ID of root, which won't match.

So I'm confused that you're hitting an issue here; would be great to get some more details.