Simple Python script to manifest a list of potential domain names from a keyword phrase/string. Script will also validate the domain.

manifest_domain.py

#!/usr/bin/python
# -*- coding: utf-8 -*-
import sys

stopwords = " a to the ltd is for of on will with pty" # not definitive
legalchars = "abcdefghijklmnopqrstuvwxyx1234567890"
replace = (
    (("Ã","Å","Ä","À","Á","Â","å","å","ä","à","á","â"),"a"),
    (("Ç","Č","ç","č"),"c"),
    (("É","È","Ê","Ë","Ĕ","è","ê","ë","ĕ","é"),"e"),
    (("Ğ","Ģ","ģ","ğ"),"g"),
    (("Ï","Î","Í","Ì","ï","î","í","ì"),"i"),
    (("Ñ","ñ"),"n"),
    (("Ö","Ô","Ō","Ò","Ó","Ø","ö","ô","ō","ò","ó","ø"),"o"),
    (("Ŝ","Ş","ŝ","ş"),"s"),
    (("Ü","Ū","Û","Ù","Ú","ü","ū","û","ù","ú"),"u"),
    (("Ÿ","ÿ"),"y"),
    )
domains = (
    ".co.nz",
    ".net.nz",
    ".org.nz",
    ".ac.nz",
    ".geek.nz",
    ".gen.nz",
    ".iwi.nz",
    ".maori.nz",
    ".school.nz",
)

def remove_blanks(lst):
    newlst = []
    for l in lst:
        if l:
            newlst.append(l)
    return newlst

def replace_chars(subject):
    """ swap out gliphs for acceptible replacement """
    for r in replace:
        for c in r[0]:
            subject = subject.replace(c,r[1])
    return subject

def remove_chars(subject):
    """ Replace all chars that arent in allowed list """
    for c in subject:
        if c != " " and c not in legalchars:
            subject=subject.replace(c,"")
    return subject

def remove_stopwords(subjects):
    """ removes any stop words in the provided list """
    new_subject = []
    sw = []
    stops = stopwords.split(" ")
    for s in stops:
        if s:
            sw.append(s)

    for w in subjects:
        if w not in sw:
            new_subject.append(w)
    return new_subject

def wordvariants(name):
    """ returns a list of word variants (doesn't append domain name)"""
    name = name.lower() 
    name = replace_chars(name)
    name = remove_chars(name) 
    words = remove_blanks(name.split(" "))
    
    variants = []
    variants.append("".join(words)) #all spaces removed 
    variants.append("-".join(words)) #words joined by dashes

    words_no_stopwords = remove_blanks(remove_stopwords(words))
    nsw = "".join(words_no_stopwords) #words with stop words removed spaces removed
    # no point in processing any dups
    if nsw not in variants:
        variants.append(nsw) #words with stop words removed joined by dashes
    nsw = "-".join(words_no_stopwords) 
    if nsw not in variants:
        variants.append(nsw) #words with stop words removed spaces removed

    for w in words_no_stopwords:
        variants.append(w) #try just single words 
    
    return variants

def manifest_domains(subject):
    words = wordvariants(subject)
    domain_options = []
    for w in words:
        for d in domains:
            domain_options.append(w + d)
    return domain_options

def test_domains(domains):
    import socket
    passed = []
    failed = []
    for d in domains:
        try:
            ip = socket.gethostbyname(d)
            if ip:
                passed.append({ d:ip })
                print d,ip
            else:
                failed.append(d)
        except socket.gaierror: 
            # [Errno -5] No address associated with hostname
            failed.append(d)
    return passed, failed

def manifest_and_test(subject):
    domains = manifest_domains(subject)
    return test_domains(domains)

if __name__ == "__main__":
    if len(sys.argv) >= 2:
        subject = sys.argv[1]  
        passed, failed = manifest_and_test(subject) 
        print "Available:"
        for d in failed:
            print d
        print "Domains in use:"
        for d in passed:
            #print "%s (%s)" % (d,passed[d])
            print d

    else:
        print "Usage: python manifest_url.py \"A name to check\""