See output @ http://poeticalbot.tumblr.com/
National Poetry Generation Month 2016
LOCAL - node test/manualrunners/writepoem.js
REMOTE - heroku run node index.js
TODO: index should take option to NOT publish TODO: some things log, some things don't - it's erratic and the logs can be confusing
npm t
git push heroku master
TODO: look into automatic github hooks - https://devcenter.heroku.com/articles/github-integration
These config vars are used for local running. On Heroku itself, they will be config vars.
You can gt them from the heroku applications config, or from Tumblr
consumer_key=<OAuth consumer key>
consumer_secret=<OAuth consumer secret>
token=foo
token_secret=foo
- queaneau-buckets
- jgnoetry (headless)
- custom templates
- TODO: on-the-fly generated templates
- TODO: templates can have pre-populated text and spacing ?
- TODO: rewire for multi-pass with saved-text (and post sequences)
- Harvard Sentences drone
- TODO: the drone structure seems like it would work for other generators, if they output sentences/lines.
- random leading spaces
- sort (ascending/descending)
- mispelr
- phonetic
- rhyme appender
- more proof-of-concept than anything.
- existing implementation is sub-optimal
- first/last/random line
- random-selection from most common words in poem
- summary sentence (summary algorithm picks sentence)
- fails poorly when there aren't enough sentences
- lots of texts
- sorted into folders
- select with regex
- a number of pre-selected combinations, plus random collections
- randomize percentages for the jGnoetry model
- Hybridizer
- heijinian leading spaces
- mesostics
- news-text importer (one of the original ideas)
- (optionally) replace the syllable-detection algorithm in jgnoetry
- at a minimum, extract it for unit-testing
Boringly, I continue to work with unit-tests and code-coverage, and other dull things instead of the "cool" poetry generation all the time. So sue me.
It usually pays off in the long run, when I return to a project after a while not remembering how it works -- boom, the tests document usage! Also they run through so many scenarios I know when I do or do not break stuff (depending upon coverage).
- https://github.com/rossgoodwin/poetry-solver
- https://github.com/rossgoodwin/poetizer
- https://github.com/rossgoodwin/sonnetizer
- https://github.com/rossgoodwin/lyricist
- https://github.com/rossgoodwin/wikipoet (30 minutes to gen a poem!!!)
- https://www.npmjs.com/package/syllable
- https://github.com/nlp-compromise - what else can it do that would be... interesting?
- better text cleanup - see ebook_ebooks - a few features to improve readability, such as chapter numbering, paragraph breaks, and parenthesis/quotation mark balancing
- https://github.com/matthewsklar/PoetryBot
- aparrish's linear-systems poetry Hard to get good results, butwith practice, weird things can emerge.
- topic analysis? I played around with a lib, but the results were not promising. these "poems" are too weird to be coherent for topics, usually. And not sure what to do with the output.
- I tried using nlp-compromise's simple-english module, but it didn't do much, very often
- more meta-data on the poems/words/etc. So transforms can be done with more granularity?
- at the very least, try to keep re-processing the texts and poems into sentences and words multiple times.
So, I've been looking at the Lexeduct code that Chris Pressey started last year. I didn't look into it enough at the time, and my work with it was at cross-currents to its ideology (my work last year was in the gh-pages
branch, and "worked", even though it doesn't fit the main model in the master branch).
I think wrangling that understanding and applying it to want I currently want to do will be too time-consuming (although profitable).
So, I'm going to do the Simplest Thing That Could Possibly Work.
- static text generator posts to Tumblr
- text generator becomes non-static
- elaborate and iterate on step 2
- end-goal includes ingestion of source material from online news
SO 1-3 HAPPENED THAT'S GOOD And 3 continues to happen....