I've been using Python Sphinx for a number of years and mostly happy with it. The major limitation for me is it doesn't support "parts of a book." For whatever reason the maintainers have not been interested and ignored pull requests for it for years, saying it needed to be designed first. Then not responding to requests for feedback.
Not sure if moving a book to asciidoc is practical, but I've been considering it.
In case it's of interest, we wrote a system for converting Jupyter Notebooks to asciidoc, which we used to publish an O'Reilly book.
This is the book: https://www.amazon.com/Deep-Learning-Coders-fastai-PyTorch/d...
This is the nb->asciidoc convertor: https://github.com/fastai/fastdoc
These are the source notebooks: https://github.com/fastai/fastbook/