The author exploited this by adding a space to the URL so it no longer started with http:// rather (space)http:// but it looks like the call to Replace would be ineffective if the URL started with HTTP:// as well.
There were a bunch of ways to bypass the check. For example another way would be to use "http:\\" which wouldn't get detected either. The new version isn't vulnerable.