# -*- encoding: utf-8 -*-
import codecs
import re
def get_words(filepath):
f = codecs.open('sample3.txt', 'r', 'UTF-16')
return re.sub(r'\s+', ' ', f.read())
words_to_string = get_words('sample3.txt')
words_to_list = words_to_string.split()
word_dic = {}
#Getting words into dic and...