#!/usr/bin/python3
"""Generate words using a FSM model, or really a first-order Markov chain.

Words should be on stdin, one per line.  Characters α and ω are reserved.
"""
import argparse
import random
import sys


def model(words):
    model = {}
    for word in words:
        for a, b in pairs('α' + word.strip() + 'ω'):
            if a not in model:
                model[a] = []
            model[a].append(b)
    return model

def pairs(seq):
    seq = iter(seq)
    last = seq.__next__()
    for item in seq:
        yield last, item
        last = item

def generate(model):
    state = 'α'
    while True:
        state = random.choice(model[state])
        if state == 'ω':
            break
        yield state
        
def graphviz(model):
    yield 'digraph grammar {\n'
    nodes = dict((letter, i)
                 for i, letter in enumerate(set(model)
                                            | set(b for a in model
                                                  for b in model[a])))
    for letter, i in nodes.items():
        yield ' %d [label="%s"];\n' % (i, letter)
        for dest in set(model.get(letter, ())):
            yield ' %d -> %d;\n' % (i, nodes[dest])

    yield '}\n'

def main():
    p = argparse.ArgumentParser(description=__doc__)
    p.add_argument('-w', '--words', type=int,
                   help='generate some number of words')
    args = p.parse_args()

    m = model(sys.stdin)
    if args.words is None:
        sys.stdout.writelines(graphviz(m))
    else:
        for i in range(args.words):
            print(''.join(generate(m)))

if __name__ == '__main__':
    main()