CKEditor shows no content anymore when character '<' is used in html comment

[Michel]

Steps to reproduce

1. Go to the ckeditor demo: ​http://ckeditor.com/demo#full
2. Click button Broncode or Source code.
3. Paste the following code before anything else in the texteditor.

<style type="text/css">

/* Test <9 */ </style>

Expected result

Original text is still shown

Actual result

Nothing is shown

Other details (browser, OS, CKEditor version, installed plugins)

Browser used: IE11.

If there is a space between < and 9 everything works. /* Test <9 */

[Piotrek Koszuliński]

Confirmed. Using "<" in a comment in CSS is valid, so it should not break the editor. We'll check if it's possible to fix our parser or whatever else fails

[Szymon Cofalik]

This is caused by a bug in htmlparser engine. Unfortunately it is too complicated to be fixed in version 4.5.4.

Here is a description why the bug happens:

1. In htmlparser we use a pretty sophisticated regular expression that finds tag openings.
2. When the algorithm finds <style> tag, it looks further into code, looking for closing tag, but it finds <9 </style> instead.
3. Algorithm never finds closing tag for <style>.

If <9 was a part of normal html code this won't be a bug but just invalid html code. Unfortunately, in <style> tag, in a comment, it is valid.

For further reference: to fix it, we would have to change how we look for closing tag when "cdata mode" is active (cdata variable is an array, not undefined). This sounds easy at the first glance but we have to keep in mind that current regexp is complicated right now and it is a part of a loop which reads html code part after part so making an exception might be difficult. Also we have to think about possible regressions.

[Piotrek Koszuliński]

DUP reported #13718.