markov.py

import random;

CORPUS_FILENAME = "/home/theblackparrot/corpus.txt";
corpus = {};

def addToCorpus(line):
	words = line.split();

	if len(words) < 2:
		return;

	phrase = words[0] + " " + words[1];

	if len(words) == 2:
		if phrase not in corpus:
			corpus[phrase] = [];

		return;

	nextWord = words[2];

	for i in range(1, len(words)):
		if phrase not in corpus:
			corpus[phrase] = [];

		if not nextWord:
			continue;

		corpus[phrase].append(nextWord);

		phrase = words[i] + " " + words[i+1];

		if i+2 < len(words):
			nextWord = words[i+2];
		else:
			nextWord = "";


def generateMarkovChain():
	phrase = random.choice(list(corpus.keys()));
	output = phrase;

	while corpus[phrase] and len(corpus[phrase]) > 0 and len(output) < 1000:
		parts = phrase.split();

		nextWord = corpus[phrase][random.randint(0, len(corpus[phrase])-1)];

		output += (" " + nextWord)

		phrase = parts[1] + " " + nextWord;

	return output.strip();


with open(CORPUS_FILENAME, 'r') as CORPUS_FILE:
	for line in CORPUS_FILE:
		if line:
			addToCorpus(line);
	print(generateMarkovChain());