I'd love to use the so elegant Jupyter-Book to construct documentation for my packages. I've tried quite a lot, and here is an example I'm finalizing those days :
- the package is available on a gitlab repository https://framagit.org/nlp/iamtokenizing
- the associated documentation is constructed (almost) automatically by calling
jupyter-book build documentation/
on CI/CL-deployment of the master branch of the repository, and visible there : https://nlp.frama.io/iamtokenizing
One can see I use
- jupyter-notebook to generate user guides or tutorials,
- markdown to generate simple static pages without codes, and
- sphinx autodoc to generate api documentation. Note I'm fairly newbie using sphinx and autodoc.
I'd like to enhance the publication style of the auto-generated api documentation. As one can see, e.g. on this page, there is no right sidebar in the API documentation, whereas one may want to write the names of the class methods, or module functions there.
Also, the parameters list might be better presented in indented lists instead of being queued in parenthesis, and the name of the class might be shorter, for instance only the name of the class, not the path to it might appear. In short, I'd prefer having
class BaseTokenizer(string: str = '',
subtoksep: chr = ' ',
intervals: Union[None, EvenSizedSortedSet] = None,
tokens: list = [])
instead of the monolithic presentation
class iamtokenizing.base_tokenizer.BaseTokenizer(string: str = '', subtoksep:
chr = ' ', intervals: Union[None,
extractionstring.even_sized_sorted_set.EvenSizedSortedSet] = None, tokens: list = [])
Is there a way to do all of that, or should I consider raising an issue instead ?