some_random_words.py

#!/usr/bin/env python3
""" some_random_words.py

    Usage:

        ./some_random_words.py [letters_start_w_key_letter]

    Prints out a bunch of random words that use these letters,
    and which include the key letter.
"""

import random
import sys

from collections import Counter


CHEAT_MODE = True
WORD_LEN = 20


def load_bigram_model(letters):
    with open('/usr/share/dict/words') as f:
        words = [line.strip().lower() for line in f]

    if CHEAT_MODE:
        words = [
            word
            for word in words
            if (all(let in letters for let in word) and letters[0] in word)
        ]

    bigram_counter = Counter()
    print('Analyzing all words .. ', end='', flush=True)
    for word in words:
        word_ = f'_{word}_'
        for a, b in zip(word_[:-1], word_[1:]):
            bigram_counter[a + b] += 1

    alpha = 'abcdefghijklmnopqrstuvwxyz_'

    # For each letter L, next_let_hist[L] will be a list of the form:
    # [(M, prob_of_M_given_L), ...]
    next_let_hist = {}
    for let in alpha:
        next_set = [
            (let2, bigram_counter[let + let2] + 1)
            for let2 in alpha
        ]
        total = sum(pair[1] for pair in next_set)
        next_set = [
            (pair[0], pair[1] / total)
            for i, pair in enumerate(next_set)
        ]
        next_let_hist[let] = next_set

    print('done!')

    return next_let_hist

def get_random_next_letter(this_letter, next_set):

    u = random.random()
    index, total = 0, next_set[0][1]
    while total < u:
        index += 1
        total += next_set[index][1]
    return next_set[index][0]

def filter_to_ok_letters(ok_letters, next_set):
    filtered_set = [pair for pair in next_set if pair[0] in ok_letters]
    total = sum(pair[1] for pair in filtered_set)
    return [(pair[0], pair[1] / total) for pair in filtered_set]

def generate_random_word(ok_letters, next_let_hist):
    letters = ['_']
    while True:
        next_set = next_let_hist[letters[-1]]
        ok = ok_letters if len(letters) == 1 else ok_letters + '_'
        next_set = filter_to_ok_letters(ok, next_set)
        letters.append(get_random_next_letter(letters[-1], next_set))
        if letters[-1] == '_':
            break
    word = ''.join(letters)[1:-1]
    if word == '':
        word = generate_random_word(next_let_hist)
    return word

if __name__ == '__main__':

    if len(sys.argv) < 2:
        print(__doc__)
        sys.exit(0)

    letters = sys.argv[1]
    key_letter = letters[0]

    next_let_hist = load_bigram_model(letters)

    for _ in range(500):

        word = ''
        while key_letter not in word:
            word = generate_random_word(letters, next_let_hist)
        print(word)

        if False:
            word = []
            while key_letter not in word:
                word = [random.choice(letters) for _ in range(20)]
            print(''.join(word))