Extracting URLs, not perfect but "good enough"

Detecting URLs in text is harder than you would expect. In fact, it can't be done with a set of hard rules, but we can get very close by using a single .NET regular expression. Purists will protest, but what's wrong with "good enough"?