beautiful soup post processing exporters, testing and building

[tonyfast]

on 10/25, i shared a notebook that explores html modifications to html notebook documents. these changes support this document.

todo

• [x] integrate beautiful soup into the library
• [x] integrate beautiful soup into the template
• [x] use hatch for the build/test system
• [x] <main>, <article> and <section> tags defined for the notebooks, cells, and outputs respectively.
• [x] a test example that replaces prompts with <aside>s
• [x] fix ci

[tonyfast]

this pull request is probably in a good place for review.

we focus on being about the post process documents exported by nbconvert. this allows for remediations to be reversed engineered for manual tests. for future work, we'll create a system that codifies out learning in formal jinja2 templates. the test notebook demostrates the features by using the nbconvert apis interactively and at the command line.

please help with language too. im not confident in writing formal technical content.

[isabela-pf]

I gave this a look, but I think it's too far outside my knowledge base for me to review. Please let me know if there's any feedback you'd like me to give.

[smythp]

Going to pull it in and review. Will ikely have questions. :)

[smythp]

Tony, pulled it in and got the tests running. (After taking a little while to realize I needed to reinstall the html5 module.)

This seems like a good basis to build on. I can't say I fully understand the invocation at the end (the command line command.) It seems you're adding the function written in the jupyter_nbconvert_config.py file to run at the end of the nbconvert process. It seems you're not passing a list of postprocess functions...how would you apply multiple transformations? Or is that a bad idea?

I'm going to merge this in. I'm pretty OK with BS4 as I do a lot of scraping, and might try creating a config file that does something different to see if I'm getting how the postprocessing works.

[tonyfast]

Tony, pulled it in and got the tests running. (After taking a little while to realize I needed to reinstall the html5 module.)

hopefully the next pull doesn't require that. i changed the build system and should have been more transparent.

This seems like a good basis to build on. I can't say I fully understand the invocation at the end (the command line command.) It seems you're adding the function written in the jupyter_nbconvert_config.py file to run at the end of the nbconvert process. It seems you're not passing a list of postprocess functions...how would you apply multiple transformations? Or is that a bad idea?

we can totally apply multiple transformations with either a change in the current configuration or encapsulating multiple functions in a one function.

I'm going to merge this in. I'm pretty OK with BS4 as I do a lot of scraping, and might try creating a config file that does something different to see if I'm getting how the postprocessing works.

that would be great to know how malleable the interface is.