transmorgify.py

#!/usr/bin/env python
#

"""
$ ./transmorgify.py
input: http://yahoo.com www.blah.com 13 monkey's @username blah
output: ['monkeys', 'blah']
input: Super impressive http://tr.im/lWb7 (via @jcsalterego)
output: ['Super', 'impressive', 'via']
input: 
output: []
"""

import re
import sys

inputs = ("http://yahoo.com www.blah.com 13 monkey's @username blah",
          "Super impressive http://tr.im/lWb7 (via @jcsalterego)",
          "")

# remove digits, @words, urls, remove all the 's, & turn into an array?
def transmorgify(input):
    destroy = ('0-9', '\'', '(', ')')
    destroy_re = re.compile("[%s]" % "".join(destroy))

    fails = ('@', 'http://', 'www\.')
    fails_re = re.compile("|".join(fails))

    print("input: %s" % input)

    output = destroy_re.sub("", input)
    output = [word for word in output.split(" ")
              if word and not fails_re.match(word)]

    print("output: %s" % output)

def main():
    for input in inputs:
        transmorgify(input)

if __name__ == '__main__':
    sys.exit(main())