Regular expressions are just tools - parsing a URL could be done with or without them, the only requirement is that the implementation is complete and correct. Personally I'd use my spirit-like parser template library, but even without it the job wouldn't be too difficult (the PHP implementation looks more complicated than necessary, IMO). It just boils down to what you're willing to do, I guess.
I can parse URL's fine. The specific guideline's required to parse URL's 100% correctly requires that regular expression (others will result in incorrect parsing sooner or later), and that regular expression does not work with Boost::Regex (at least without some obscure attribute I have to try and find). Hence what I stated. So your point is?
My point is, you are basically re-building a library, that probably isn't as accurate (without putting some extra time into it).