#!/usr/bin/env python3

import json

START = "__START"
END = "__END"

with open('corpus.txt') as f:
    corpus = f.read().split()

markov = {}

def markov_add(cur, nxt):
    if cur not in markov:
        markov[cur] = []
    markov[cur].append(nxt)

cur = START
for word in corpus:
    if len(word) == 0:
        continue
    markov_add(cur, word)
    if word[-1] in '.?!':
        markov_add(word, END)
        cur = START
    else:
        cur = word

with open('markov.json', 'w') as f:
    json.dump(markov, f)