[pLog-svn] ugh. r6488 breaks our good old friends <, >, &lt; in post text

[Jon Daley]

 	I had an idea about this.  What do people think about having a 
checkbox for plain text vs. HTML when commenting?  Then, if the person 
chooses (the default) plain-text, then the filtering is easy, because we 
can run html_entities() or whatever on everything.
 	And then how stuff currently works would happen when the HTML box 
is checked.
 	I was also wondering about not doing HTML at all, but forcing a 
bbcode or something other pseudo-HTML like that.

On Tue, 17 Feb 2009, Jon Daley wrote:

> No answer?  I'll just keep coding then.
>
> I'm having trouble with allowing HTML and a '<' in a comment.  I am not sure 
> how to get the bad-code-filtering and the XHTML checker to allow it. Right 
> now, '<' and '>' are removed, since they are bad HTML.  If we didn't allow 
> HTML, but instead used a bbcode-variant, that would fix it, that then I could 
> escape all HTML.
>
> On Fri, 13 Feb 2009, Jon Daley wrote:
>
>> 	I just noticed that due to the htmlDecode call in filterJavascript 
>> for the postText, it (now, as of rev. 6488) converts &lt; to < (in order to 
>> catch the tricky javascript coders) and then the < is removed by the 
>> xhtmlize() call later.
>> 	I first thought I could put a htmlentities() or htmlspecialchars() at 
>> the end of filterJavascript(), but that causes all html entities to be 
>> saved as non-html, ie. all < are converted to &lt;
>> 	Maybe we do need to switchover to a new filter?
>> 
>> And, please check out the following bug, which is likely the same issue as 
>> what I just found.  I see that no one has been testing 1.2.9?  Are you all 
>> still using insecure versions of LT (ie. 1.2.8 or 2.0?)
>> 
>> http://bugs.lifetype.net/view.php?id=1579
>> 
>> 
>> 
>> 
>
>

-- 
Jon Daley
http://jon.limedaley.com
~~
(Sung to the tune of
yellow submarine)
We all live in room 5419,
room 5419, room 5419
-- Mike Schuresko (presumably referring to 15-127 students)