Nevermind, I was able to make it work. You have to run the files in this order (for a topaz prc) ...
1. TopazExtract_Kindle4PC
2. TopazFiles2SVG
3. TopazFiles2html
The resulting book.html file was easily exported into
Calibre with a metadata TOC intact. On a quick inspection, there may be only minor cleanups needed which I can do in
Sigil.
Edit: It wasn't as clean as I thought it would be. There are typical errors you get from OCR books, as this uses an OCR process to convert the books. Topaz is a crappy format, and even on
K4PC the book's text doesn't look that great. One problem is that
Amazon doesn't specify on its kindle books whether they are topaz format or not. So it's risky buying from them unless you ONLY want to read the book on a kindle device. In this case, for me, I could not find this title in an ebook version anyplace else. So it was worth it for me. But if I have a choice, I will always opt for a non-Amazon ebook first.