Something

#!/usr/bin/env python

# generate .mpc config

import sys, re

def parse_hostnames(hosts_file):
    """Read from stdin or a file given as the first command line argument
    and parse through a list of hosts that use the hostname convention
    typenum-site
    """

    constellation = {}
    types = {}


    for hostname in hosts_file:
        hostname = hostname.strip()
        host_site = hostname.split('-', 1)

        if len(host_site) != 2:
            continue

        host, site = host_site
        type_num_pattern = re.compile(r"([A-Za-z_]+)([0-9]+)")
        type_num = type_num_pattern.match(host)
        type, num = type_num.groups()

        print site, type, num

        # how should this work?
        #types[type] = num
        #constellation[site] = types

if __name__ == '__main__':
    if len(sys.argv) == 1:
        parse_hostnames(sys.stdin.readlines())
    else:
        parse_hostnames(open(sys.argv[1]).readlines())



Run this against a file or stdin with this input...

dist1-cn
dist2-cn
jnm1-cn
jnm2-cn
jns1-cn
jns2-cn
jns3-cn
jns4-cn
nm1-cn
nm2-cn
nr1-cn
ns1-cn
ns2-cn
pe1-cn