Some more notes. 
a) RFC3986 explicitly states that the presented regex (which you use)
   """ is the regular expression for breaking-down a *well-formed* URI
reference into its components. """ (Emphasis added). I am not sure this
is a particularly good starting point for parsing potentially
security-critical data.

b) The parser fails on URI's containing numerical IPv6 addresses (e.g.
"http://[::1]:88/path"). Specifically, the following code in
split_authority is broken:

    if hostport and ':' in hostport:
        host, port = hostport.split(':', 1)

Clearly, if the authority may contain a ":" in the host's IP field, you
cannot simply split() off the port part.

Again, I am afraid I have no simple solution. Hate to sound so negative.

Kind regards,
v.