How can I parse a host:port pair in Python

Well, this is Python, with batteries included. You have mention that the format is the standard one used in URIs, so how about urllib.parse?

import urllib.parse

def parse_hostport(hp):
    # urlparse() and urlsplit() insists on absolute URLs starting with "//"
    result = urllib.parse.urlsplit('//' + hp)
    return result.hostname, result.port

This should handle any valid host:port you can throw at it.

This should handle the whole parse in a single regex

regex = re.compile(r'''
(                            # first capture group = Addr
  \[                         # literal open bracket                       IPv6
    [:a-fA-F0-9]+            # one or more of these characters
  \]                         # literal close bracket
  |                          # ALTERNATELY
  (?:                        #                                            IPv4
    \d{1,3}\.                # one to three digits followed by a period
  ){3}                       # ...repeated three times
  \d{1,3}                    # followed by one to three digits
  |                          # ALTERNATELY
  [-a-zA-Z0-9.]+              # one or more hostname chars ([-\w\d\.])      Hostname
)                            # end first capture group
(?:                          
  :                          # a literal :
  (                          # second capture group = PORT
    \d+                      # one or more digits
  )                          # end second capture group
 )?                          # ...or not.''', re.X)

All that's needed then is to cast the second group to int.

def parse_hostport(hp):
    # regex from above should be defined here.
    m = regex.match(hp)
    addr, port = m.group(1, 2)
    try:
        return (addr, int(port))
    except TypeError:
        # port is None
        return (addr, None)

def split_host_port(string):
    if not string.rsplit(':', 1)[-1].isdigit():
        return (string, None)

    string = string.rsplit(':', 1)

    host = string[0]  # 1st index is always host
    port = int(string[1])

    return (host, port)

Actually confused on whether this is what you wanted, but I rewrote it up a bit and it still seems to follow the ideal output:

>>>> split_host_port("localhost")
('localhost', None)
>>>> split_host_port("example.com:1234")
('example.com', 1234)
>>>> split_host_port("1.2.3.4")
('1.2.3.4', None)
>>>> split_host_port("[0abc:1def::1234]")
('[0abc:1def::1234]', None)
>>>> 

As on the first line I didn't really like the chained function calls e.g. getattr(getattr(getattr(string, 'rsplit')(':', 1), '__getitem__')(-1), 'isdigit')() for the expanded version and then it's repeated again two lines after, perhaps I should make it a variable instead so there's no need for all the calls.

But I'm nitpicking here so feel free to call me out on that, heh.

Came up with a dead simple regexp that seems to work in most cases:

def get_host_pair(value):
    return re.search(r'^(.*?)(?::(\d+))?$', value).groups()

get_host_pair('localhost')
get_host_pair('localhost:80')
get_host_pair('[::1]')
get_host_pair('[::1]:8080')

It probably doesn't work when the base input is invalid however

Here's my attempt at this so far:

def parse_hostport(hp):
    """ parse a host:port pair
    """
    # start by special-casing the ipv6 literal case
    x = re.match('^(\[[0-9a-fA-F:]+\])(:(\d+))?$', hp)
    if x is not None:
        return x.group(1, 3)

    # otherwise, just split at the (hopefully only) colon
    splits = hp.split(':')

    if len(splits) == 1:
        return splits + [None,]
    elif len(splits) == 2:
        return splits

    raise ValueError("Invalid host:port input '%s'" % hp)

Here's a terser implementation which relies on attempting to parse the last component as an int:

def parse_hostport(s):
    out = s.rsplit(":", 1)
    try:
        out[1] = int(out[1])
    except (IndexError, ValueError):
        # couldn't parse the last component as a port, so let's
        # assume there isn't a port.
        out = (s, None)
    return out