[Webdev] Negative lookahead assertions in Perl regex

[Dermot McGahon]

Hi,
 
I have a regex that strips out HTML comments from HTML documents. It
looks like this:
 
  $document =~ s{<!--.*?-->}{}gm;
 
However, when a browser doesn't know how to process the <SCRIPT
LANGUAGE="Javascript"> tag it often tries to display the javascript as
text so a common ploy to avoid this is to wrap the javascript with
HTML comments so that the browser displays nothing rather than the
javascript. Maybe an example will help:
 
<script language="JavaScript">
 
<!--
 
imBanner1  = new Image ();
imBanner1.src = "images/banner_1.gif";
sBanner1Link  = "http://www.forbes.com/asap/00/0403/84b.htm"; 
 
 
//-->
 
</script>
 
Non supporting browsers will now display nothing (well probably the
HTML below) rather than mistakingly displaying the javascript.
 
Anyway, I now need to modify the commment stripping regex so that it
only strips comments when they are not contained with <SCRIPT> and
</SCRIPT>. I thought that negative lookahead assertion might be the
way to go so I tried:
 
  $document =~ s{(<!--.*?-->)(?!\s*</script>)}{}gim;
 
and
 
  $document =~ s{<!--.*?-->\s*(?!</script>)}{}gim;
 
and a few other combinations but I can't get it to work as I'd
like. The first regex is matching past the </script> tag, it seems to
be matching until the next "-->" that doesn't have a </script> tag
after it. Of course, that's what I asked it to do :) but the behaviour
that I want is less greedy and I'm not sure how to get that behaviour.
 
The second regex matches exactly the same as the original one. It
strips out anything between "<!--" and "-->" and doesn't seem to pay a
blind bit of notice to the </script> tag.
 
Can anyone edify me as to what I'm doing wrong ?
 
Dermot.
--