Store references.xhtml in bibtex format
The resulting references.bib is then to be transformed to html via bib2xhtml or bibbase. Why: bibtex is more machine-parseable.
Edit: add why
Behold, I have compiled a references.bib based on the original html version with urls added from the html5 version. The .bib can be readily converted to html with bib2xhtml as to preserve the data representation used in the book, e.g. https://www.spinellis.gr/sw/textproc/bib2xhtml/eg/paragraph-u.html, and can be stored in bibjson format as well (less verbose, and also that json parsers can be found at virtually everywhere).
Good work! I'll think about how to incorporate bibtex references into the project. Sorry for the delay in answering, I've been travelling without a computer.