.html() corrupts attribute values containing things that look like self-closing tags

Reported by:	roan.kattouw@gmail.com	Owned by:
Priority:	undecided	Milestone:	None
Component:	unfiled	Version:	1.9.1
Keywords:		Cc:
Blocked by:		Blocking:

Description

h = '<span data-foo="foo<bar />baz">quux</span>';
d = $( '<div>' ).html( h );
d[0].innerHTML;
    "<span data-foo="foo<bar ></span>baz">quux</span>"

This happens because rxhtmlTag mistakenly believes my <span> is self-closed, because it sees <span followed by />, and "helpfully" tries to close it for me.

In Chrome, this is slightly less of an issue than in Firefox. If you have a DOM that looks like the one in my example, Chrome's .innerHTML will have the < and > escaped as < and > (which doesn't match rxhtmlTag); while Firefox's .innerHTML will have them as literals (which does match rxhtmlTag).

I wonder why rxhtmlTag exists in the first place. It's not documented in the source code at all, and it breaks things. Is it because IE's behavior with self-closing tags?

Attachments (0)

Change History (4)

Changed April 25, 2013 04:22PM UTC by roan.kattouw@gmail.com comment:1

jsfiddle demonstrating the bug: http://jsfiddle.net/FZs9P/ (breaks in Firefox, works in Chrome)

Changed April 25, 2013 04:35PM UTC by anonymous comment:2

Replying to [comment:1 roan.kattouw@…]:

jsfiddle demonstrating the bug: http://jsfiddle.net/FZs9P/ (breaks in Firefox, works in Chrome)

URL that actually works: http://jsfiddle.net/FZs9P/2/

Changed April 26, 2013 01:01AM UTC by gibson042 comment:3

resolution:	→ wontfix
status:	new → closed

Amazingly enough, it ''is'' possible to fix this concisely. But I'll give you a hint of the mess we're likely to leave behind by doing so:

rxhtmlTag = new RegExp( "<(?!area|br|col|embed|hr|img|input|link|meta|param)(([\\\\w:]+)(?:" +
  jQuery.expr.match.ATTR.source.replace(/^.*?\\[|(\\\\\\\\)\\.|(?=\\\\)(\\d)|.]$/g, function( str, backslash, capture ) {
    return +capture + 2 || backslash || "";
  }) +
  ")*\\\\s*)/>",
"gi" )

Better to let sleeping dogs lie, methinks.

Changed April 26, 2013 01:58AM UTC by dmethvin comment:4

As far as I can tell, this patch was put waaaaay back (before 2007) so that people could use invalid markup like <div id="haha" /> and after that much time I have no doubt it's widespread enough that we couldn't ever contemplate pulling it out.

Side navigation