Markov Poet

---

red hearts flying
deep with the chatter of day

walking home to the photo
the sound
upon red rocks

organ reedy as the highway
in twisted tapers with a winter solstice moon
under a broken bone of night sky
dreams of friends reunited

---

Markov Poet is a Markov chain-based poetry generator chain that simulates poetic line breaks by including a new line token in its model. It also includes the option of introducing an element of randomness, where by each step in walking through the chain has a chance of moving to any other state before continuing.

Usage

markov = Markov()

# This expects a plain text file.
markov.add_file('sample_corpus.txt')

haiku = """the piano room
pure ivory keys
under a layer of dust"""
markov.add_poem(haiku)

lines = ['old pond', 'frog leaping', 'splash']
markov.add_lines(lines)

# Randomness is optional.
randomness = 0.2
generator = markov.generator(randomness)
generator.generate_line()

number_of_lines = 7
print(generator.generate_formatted(number_of_lines))

To run tests:

$ pwd
/path/to/markov-poet
$ python3 -m tests.test

Demo

Also included is a demo driver.

usage: demo.py [-h] [--filename FILENAME] [--order ORDER]
               [--randomness RANDOMNESS] [--number NUMBER]

Poem generator command line tools.

optional arguments:
  -h, --help            show this help message and exit
  --filename FILENAME, -f FILENAME
                        specify a file to read from
  --order ORDER, -o ORDER
                        specify the order of the Markov chain
  --randomness RANDOMNESS, -r RANDOMNESS
                        introduce some randomness (between 0.0 and 1.0)
  --number NUMBER, -n NUMBER
                        Print more than one haiku

Demo Options

• --filename/-f : specify the name of a file to read as the reference corpus. The expected format is of plain text. The model tokenizes words and line breaks but will ignore other white space.
• --randomness/-r (between 0.0 and 1.0): introduce an element of randomness into the generation. At 1.0, the chain is ignored and every word is randomly selected from the language. The default value is 0.0.
• --number/-n : specify the number of poems to output. The default is 3.
• --order/-o : specify the size of a state in the chain. The default value is 1.

Example:

$ pwd
/path/to/markov-poet
$ python3 -m sample.demo -f sample/haiku.txt
$ python3 -m sample.demo -f sample/kanye.txt --number 3 --order 2

Credits

• Sample haiku corpus from the Haiku Society of America
• Sample Kanye lyrics from Kanye West