auto_pometizer, a Markov chain poetry generator

After gathering all the poetry available from PoetryDB, I turn it into a text file as one long string (with lines separated by \n).

I then create a dictionary with each word in the string as a key and a list of the words that immediately follow each of those key words in the string. I had to do some switcharoo maneuvering to include newline (\n) and tab (\t) characters among the keys and within the values of the poem dictionary.

Within the auto-pometizer function, you can then input a word count (1000 or fewer words, note: newline and tab characters are included in the count), as well as whether or not to use a rhyming function (of which there are three varieties).

Best of all, it is now an app!

To use the app locally, run the following in Terminal (after cloning the repo):

streamlit run app.py

How it works

The poetry generator randomly chooses its first word from the dictionary's keys, then a random word within that key's value, which becomes the next word and the next key; it then chooses a random word within that second word's value, which becomes the next word and the next key, and on and on until the word count is reached.

Within the auto_pometizer function, the optional to_rhyme argument employs the rhyming function from Allison Parrish's Pronouncing to change words at the end of every line (to_rhyme='endline'), randomly throughout the generated text (to_rhyme='random'), and all the words within the generated text (to_rhyme='all').

List of files

• scrap_files folder - backups, old workbooks, old texts/jsons, and other scraps.
  • includes NLTK-tokenized version of dictionary with some punctuation (poetic apostrophes and hyphens)
    • lacks some proper segmentation but not as oversegmented as the wordninja version
• .gitignore - list of files to ignore.
• Procfile - requirement for Heroku deploy.
• README.md - this very file!
• app.py - file with app layout and auto_pometizer function (reconfigured for use with Streamlit).
• auto_pometizer.ipynb - the main workbook, polished, from beginning to end. One can start by opening from the Markov dictionary.
• functions.py - text file with functions, from API calls to the final poetry generator.
• poems_dictionary.json - the hero poetry dictionary.
• poems_raw.txt - the initial string compiled from PoetryDB.
• requirements.txt - requirement for Heroku deploy.
• scraping_sandbox.ipynb - notebook with an example of scraping the work of one particular poet.
• setup.sh - requirement for Heroku deploy.