Xavier, what is the name of this feature? I always just call 
it "clever parsing" :p

The following items are on my usual testing page if you want 
to look there. (I was pleased to see redirect pages working 
like v3.20 again!  Intentional? ;)




Sites created with Adobe GoLive often have javascript 
rollover code with the following format:

CSInit[CSInit.length] = new Array(CSILoad,/*CMP*/'button6',
/*URL*/'/images/dotblack.gif',/*URL*/"../../images/dotwhite.
gif",/*URL*/'','Link title');

v3.30-rc15 is unable to find the images.  It is important 
that HTTrack can detect images within this code.





Second thing is just for consideration.  It's a "would be 
nice, but if it's too hard then don't bother.

Word documents saved as web page contain frustratingly crap 
HTML, including:

<!--[if gte vml 100]>
<v:imagedata src="/images/dotwhite.gif" o:title="sport strip 
2"/>
<![endif]-->

This is a HTML hack for Internet Explorer.  Because the 
image is within comments tags, HTTrack won't discover it, 
and IE will present a broken image.