Word Generator

word_generator.py

#!/usr/bin/env python
#-*- coding: utf-8 -*-

"""This script was used to generate random test files for the course
of Computer Networks 2014 at University of Birmingham, Computer Science.

Author: Antonio Lima
License: WTFPL

"""

from __future__ import print_function

import collections
import itertools
import random
import string
import sys

def default_lengths(min_length=3, max_length=10):
    while True:
        yield random.randint(min_length, max_length)

def random_word_length(alphabet, length):
    return ''.join((random.choice(alphabet) for x in  xrange(length)))

def random_words(alphabet, lengths=default_lengths()):
    for length in lengths:
        yield random_word_length(alphabet, length)

def random_words(alphabet=string.letters, lengths=default_lengths()):
    random.seed("networks2014")
    rwl = random_word_length
    words = (rwl(alphabet, length) for length in lengths)
    for word in words:
        yield word

def random_words_repetition(repetition_probability, words):
    seen = collections.deque(maxlen=200000)
    for word in words:
        seen.append(word)
        yield word
        if random.random() < repetition_probability:
            yield seen.popleft()

def size_capper(total_size, words):
    size = 0
    for word in words:
        yield word
        size += len(word.encode('utf-8')) + 1 # Considers EOL
        if size > total_size:
            break

def main():
    usage = """Usage:
        wordgenerator <bytes> <repetition_probability>

    The script outputs to STDOUT. Repetition probability must be between
    0 and 1.

    """
    try:
        size = int(sys.argv[1])
        prob = float(sys.argv[2])
    except (IndexError, ValueError):
        print(usage, file=sys.stderr)
        return 1
    wordlist = random_words()
    wordlist_with_repetition = random_words_repetition(prob, wordlist)
    capped_wordlist = size_capper(size, wordlist_with_repetition)
    for word in capped_wordlist:
        print(word, file=sys.stdout)

if __name__ == "__main__":
    sys.exit(main())