Has anyone ever tried creating an api to convert notebooks to epub3?

Epubs use html, and I know there is some success in notebook to html conversion.

I have an idea for a notebook converter that would, go cell by cell, and turn text into html, and create pdfs of input, and separate pdfs of output, and the converter would stack the material appropriately. It would probably create a large file, but that isn't terrible.

Of course, once the file was created, you'd have to tinker with it in a different program, like Calibre, but so long as the bulk of the work could be done programmatically, i think this could be a really useful thing for the world.

So 1) have any of you ever made such an attempt? 2) Suggestions on html/xml conversions (or notebooks you have that make that conversion a tad easier 3) Ever run programmatic individual instances of input/output cell pdfing?

The reason for this application is that many epub platforms do not have the capacity to sell "fixed-image" publications. Which is what simply making an epub out of a pdf creates.