[Safari][Windows].html() with a text longer than 49934

[Joachim]

Hi,

I'm having a issue under Safari for windows version 3.1.2(525.21).

When i'm using thee following code

var data2 = ; var nb = 49934; for (cpt =0;cpt <nb;++cpt) {

data2 += 'a';

}

data2 = '<div><div id="website-background">taste</div><div>' + data2 + '</div></div>'; var currentSection = $('#website-background', data2).html(); console.log('Trace : ' + currentSection);

my currentSection var is filled whereas it's null if I use 49935 as nb

[flesler]

Can you please provide a concise test case ? That is, a small html file with the minimum html/css/js necessary to reproduce the problem.

Also, does it happen for any number equal or higher than 49935 ?

[Joachim]

Here is a sample test file to reproduce the behaviour

[Joachim]

I uploaded the file I used to reproduce the error.

I made a mistake in my first error description The length is 49935 + rhe length of <div><div id="website-background">taste</div><div></div></div>

[nathanhammon]

It is this line that it is choking on in jQuery.fn.init(): var match = quickExpr.exec( selector );

Process:

1. Calls jQuery.fn.init(html, context);
2. jQuery.fn.init() decides to not do anything on the first time through and try it again as jQuery.fn.init(context).find(selector);
3. It gets to the line mentioned above and fails to match.

That triggers the problem.

It looks like this might be a browser imposed limit to regex string length?

From here we need to:

1. Test this in other versions of Safari Win (who has a different version?).
2. Test this on Safari Mac (all versions supported).

[nathanhammon]

The total length limit appears to be 49996. Not even divisible by 8!

[nathanhammon]

It isn't us. ​https://bugs.webkit.org/show_bug.cgi?id=14873

IE: 67108863 Opera: 2097151 Safari: 49996 FF: ???

Not even in the same ballpark as the others.

[nathanhammond]

Updates test case that would have previously failed in Safari.

[nathanhammond]

Patch for sanity check. Includes patch for Test Suite.

[dmethvin]

Bug #4665 was related. The Webkit bug is now fixed according to their bug tracker above, but Nathan's patch seems like a good deal (even with a lower limit than 40,000 characters). Not execing a regexp on a monster string has got to be faster.

[dmethvin]

The regexp optimizations applied here should have fixed the performance issue: ​http://github.com/jquery/jquery/commit/9dc6e0c572b9c809a3a4c123071d96d48a01dd1c#L2L20