that's why I was thinking along the lines of taking the data that the program (Deffnetizer) does to generate its own, stripped-down html files for the epub, or the html files it outputs into the folder would have, I dunno, tags at the right places for the program to strip.
I have had bad problems converting older stories using calibre, and I have theories as to why but I'm no programmer. And I can't create a "Stripped-down" template and re-download those stories as they're not on FFN anymore, and trawling with the html code and changing them is a pain in the ass.
More so than taking a story with a running interlude threaded through the story, stripping those out and making it a pre-se-qual story (I Am What I Am, by MMcGregor, a Buffy The Vampire Slayer story, on TTH. Xander goes back in time, the interludes are events from the first timeline and jump around abit, I was trying to take the interludes and sort them into the right order. what a Pain in the ass. But not as much.)
For Epubs, the reader programs usually have their own rules on how to display, and really, that means the only data you need is the story itself, the author data, any extra that it'll recognise (for instance, if they decide to add "show story summaries" to it, or Word-counts.) and a Chapter List for the TOC. one program read an epub that didn't have it, and so went by the html files' filename, alphabetically. Another went by one of the dates on the file's metadata, so the earliest chapter would have an earlier date right? but of course, when I tried to see if that meant updated chapters got pushed to the end... yep.
I've since switched to Akido or something on my Nexus 7, as the most stable but often it doesn't override font color from black, but I had to factory reset it after the last attempt so no more trying from me until something works properly, and not haphazardly.