markovgenerator

A (WIP) Markov text generator written in Python.

The MarkovGenerator class takes a input text to generate sentences from, the length of the output you want (in characters), and an optional ngram length (the default is 2).

Many thanks to @jamak for help with the math to choose a good next word and @davoclavo for help making it pip installable!

installation

using pip:

pip install markovgenerator

or clone the repo and install:

python setup.py install

usage

Example: Generate 200-character sentences from Mary Shelley's Frankenstein based on an ngram length of 3

import markovgenerator

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

markov_gen = markovgenerator.MarkovGenerator(frankenstein, 200, 3)

markov_gen.generate_words()

returns: "The weather was fine; it was about the commission of his crimes, and I did right in refusing, to create a companion for the first creature. He showed unparalleled malignity and selfishness in evil."

Using an ngram of 2 (the default), the results get a bit more jibberish-y: "Clerval had never yet been able to perform it myself! I could banish disease from the first lesson; most of all the voyages made for purposes of discovery, he ceased to fear or to bend beneath words."

TODO:

• fix ending punctuation, don't add a period if there's already something (maybe works now)
• figure out what's going on with `` marks
• fix spaces before apostrophes (regex)
• make a setup.py file
• make pip installable
• pickle input?
• write tests