gistfile1.txt

# This submission is From Will Weatherford

from collections import Counter
import re
SEP = r'[\*]{10,999}'


def make_corpae(open_text_file):
  text = open_text_file.read()
  corpae = {corp: make_trigram_counts(corp) for corp in re.split(SEP, text)}
  return corpae


def make_trigram_counts(corp):
  trigrams = {}
  words = corp.split()
  
  for index, word in enumerate(words[2:]):
    bigram = tuple(words[index -2: index])
    counter = trigrams.setdefault(bigram, Counter())
    counter[word] += 1
  
  return trigrams
  
  
def trigram_probability(trigram, corpus):
  """Return probability that given trigram occurs in corpus."""
  counter = corpus[trigram.keys()[0]]
  total = sum([counter[ele] for ele in counter.elements()])
  return counter[trigram.items()[0]] / float(total)