MC Ride Trigrams. Enjoy the soothing verse of MC Ride from the comfort of your own computer.

markov_ride.py

from nltk.util import ngrams
from nltk.model import NgramModel
from nltk.probability import LidstoneProbDist

# Source: should be formatted on multiple lines for each sentence.
SOURCE_TEXT = 'Mystery.txt'

LINE_START = "<line>"
LINE_END = "</line>"

raw_text = open(SOURCE_TEXT).read()

filtered_lines = []

for line in raw_text.splitlines():
    if len(line) > 0:
        line_fixed = line.lower()
        line_fixed = line_fixed.replace(',', ' , ')
        line_fixed = ' '.join([LINE_START, line_fixed, LINE_END])
        filtered_lines += [line_fixed]

tokens = []

for line in filtered_lines:
    tokens += line.split()

grams3 = ngrams(tokens, 3)

estimator = lambda fdist, bins: LidstoneProbDist(fdist, 0.2)
lm = NgramModel(3, grams3, estimator=estimator)

generated_words = lm.generate(200, [LINE_START])
generated_lines = ' '.join(generated_words).split(LINE_START)
for line in generated_lines:
    line = line.replace(LINE_END, '')
    print line