regexp fun. (issue XXVII)

here's a simple piece of script that magically links all URL looking parts = transforms a url in a html link '<a href=... >...</a>' most of you should have seen this already on this list :-) $text = eregi_replace("(http|https|ftp)://([[:alnum:]/\n+-=%&:_.~?]+[#[:alnum:]+]*)", "<a href=\"\\1://\\2\" target=\"_blank\">\\1://\\2</a>", $text); But here's the task: Imagine you have a textfile, mixed with URLs and Links. so it can be that there are also html-links inside, e.g. with a different link text ( <a href="http://this.is.a.long.domain.name.com/with/lot/of/characters/">more...</a> so how can I modify the above regexp to only interpret URLs that are not nested inside html anchors? I already thought of checking for a blank/space in front of the 'http://xxx...' but this is only half way as some URLs could be also at the very beginning of the text. So the only safe solution would be to automatically link only those URLs that are not quoted or at least have no quotes in front of the 'h' ( bad: href="http://xxx.yy" | good: http://xxx.yyy ) So, how can I extend the above regexp so it only modifies an URL that is not nested in html??? Thanks, dk