class Shaney: def __init__(self, N=3, out=100): """Sets Ngram's self.N value and number of words to output""" self.N = N self.out = out self.wordlist = {} def train(self, name): """Train Ngram dictionary off of text file input""" words = file(name).read() """Substitute \self.N and \r in text with spaces""" import re regex = re.compile('(\self.N|\r)') words = regex.sub(' ', words) words = words.split() """Turn words into Ngram keys""" for i in xrange(len(words)-self.N): if words[i] not in self.wordlist: self.wordlist[words[i]] = [] temp = [] for num in xrange(1,self.N+1): temp.append( words[i+num] ) self.wordlist[words[i]].append( temp ) def output(self, startword='The'): """Create randomised Mark V Shaney output starting with startword""" newtext = [] newtext.append(startword) import random a = 1 i = random.randint( 0, len( self.wordlist[newtext[-1]] )-1 ) while a: for num in xrange(0, self.N): newtext.append( self.wordlist[ newtext[-1-num] ][i][num] ) i = random.randint( 0, len( self.wordlist[ newtext[-1] ] )-1 ) if len(newtext) > self.out and newtext[-1][-1] == '.': a = 0 return ' '.join(newtext)"""Go through all the text files in the directory"""import globfiles = glob.glob('*.txt')S = Shaney()for name in files: S.train(name)print S.output()