#!/usr/bin/env python

from operator import itemgetter
d = {}

# Limit the number of letters
limit = 4
# Alternatively, select words to exclude
removethese = ["the", "and", "i", "a", "is", "of", "in", "to", "we", "he", "with", "--", "at", "are", "you", "it", "but", "for"]

f = open('obamamccain.txt')
for line in f:
    oneline = line.strip().split()
    if len(oneline) > 0:

		# Ignore the all caps speaker identifier
        if (oneline[0][-1] == ":"):
            for word in range(1, len(oneline)):
                if len(oneline[word]) > limit:
                    d[oneline[word].lower()] = d.get(oneline[word].lower(), 0) + 1

		# Ignore the moderator's questions
        elif oneline[0] == "LEHRER:" or oneline[0] == "IFILL:":
            pass

        else:
            for word in range(len(oneline)):
                if len(oneline[word]) > limit:
                    d[oneline[word].lower()] = d.get(oneline[word].lower(), 0) + 1

"""
# Uncomment this region to remove particular words

for word in removethese:
    del d[word]
"""
posses = d.items()
posses.sort(key=itemgetter(1), reverse=True)

counter = 0
for key, value in posses:
    if counter < 30:
        print "%s\t\t%s" % (key, value)
        counter += 1