round 3

gistfile1.txt

from string import ascii_lowercase
from collections import defaultdict
from itertools import zip_longest
from pprint import pprint
from random import random
import bisect
 

class Matrix(object):
    def __init__(self, words = None):
        self.top = defaultdict(int)
        self.m = defaultdict(lambda: defaultdict(int))
        self.sto = None
        if words is None:
            return
        for word in words:
            self.add(word)
    
    def add(self, word):
        # add in the first letter into the top matrix
        self.top[word[0]] += 1

        # now add sequences
        # - marks end of word for more natural endings
        for a,b in zip_longest(word, word[1:], fillvalue = '-'):
            self.m[a][b] += 1

    def build_part_sto(self, subm):
        total = sum([v for v in subm.values()])
        matrix = {l:c/total for l,c in subm.items()}
        return matrix

    def build_sto(self):
        self.top_sto = self.build_part_sto(self.top)
        self.sto = {}
        for letter in self.m:
            self.sto[letter] = self.build_part_sto(self.m[letter])

    def pick_from_sto(self, sto):
        options = sorted(sto.items(), key = lambda x:x[1])
        x = random()
        total = 0
        for c, item in enumerate(options):
            total += item[1]
            if x < total:
                break
        return options[c][0]


    def pick_letter(self, predecessor = None):
        if predecessor is None:
            return self.pick_from_sto(self.top_sto)
        return self.pick_from_sto(self.sto[predecessor])

    def random_word(self, length = 5, minimum = 3):
        '''
        create a random word with specified max and min lenghts
        words ends naturally based on '-'s
        note: first letter is picked randomly based on precidence
        '''
        last = None
        result = ''
        for count in range(1, length):
            letter = self.pick_letter(last)
            if letter == '-':
                if len(result)<minimum:
                    continue
                break
            result += letter
            last = letter
        return result


def main():
    m = Matrix()
    with open('dictionary.txt', mode='r') as fh:
        for word in fh:
            word = word.strip()
            m.add(word)
    m.build_sto()

    for i in range(10):
        print(m.random_word())


    return

if __name__ == '__main__':
    main()