The issue is most likely cloudflare blocking most the best scraping methods. If the site can be pulled down with eg. wget or curl without a bunch of options that you definitely aren't writing by hand, pandoc can just be used to directly make an epub.
Fanfiction.net is trivial... apart from it having Cloudflare bot blocking turned up to aggressive levels. I've not seen an approach that works, other than using headless browsers to fetch the content.