Guten Haiku was created as I had a severe need to extract as many legit Haiku poems from the whole corpus for Gutenberg.
It didn’t mean a lot to me if it was the “fastest kid in town” but it sure needed to be pretty ;-)
The journey:
0. I write in python3
1. I went for space as nlp lib of choice (they are so niiiiiice peoples)
2. I needed to be able to extract syllables but space didn’t know how to do that
3. I therefore created another lib called https://github.com/sloev/spacy-syllables to annotate tokens with their syllable counts
4. I now had a syllable counting pipeline built with space
5. Now I needed to make an 80's aesthetics inspired cli app for my pipeline
6. I started with creating the whole colourful look with colorama and click
7. I created the default command for extracting haiku from text
8. I created another command for downloading fancy AI models ... uuuuh fancy!
9. Boring test implementation... yawn... zzz zz z
10. Even more boring packaging with poetry
11. Even even more boring time with travis, I don't love you right now travis
12. Release to the internets on GitHub and pypi
13. Battle with asciinema and asciicast2gif to get a pretty README up and going
14. Write hacker news about by personal triumph :-)
I wanna shout out to everybody and say that you all should make something silly and pretty today. I value those things highly and if you haven't visited https://theuselessweb.com/ yet, please do so cause TIM HOLMAN is a really nice guy with a great imagination.
Peace out and take care during the virus thingy!