Epub to KF8, unchanged?

[Barb-B]

Is there a way to convert an epub to KF8 without ANY changes other than wrapping it in the KF8 package? i.e. html file names are unchanged, font-sizes remain the same, etc... in other words if I then unpacked the kf8, the resulting epub would be indistinguishable from the original?

yeah, maybe that's a little OCD...

[KevinH]

Hi,
In short no. In longer form, a KF8 file is "compiled" as are all mobi ebooks. That means all css is cut away and stored at the end. The text itself is split in separate sections of nearly fixed lengths and is compressed. All separate files disappear. All images files disappear and are stored in sections in the mobi, all links disappear and are replaced with either text body offset or base32 encoded links. All svg is stripped out and stored after the text. Even any font files are removed and the contents are compressed and stored in numbered sections. Even the text itself is split into pieces called fragments of about 8000 chars in length and are stored out of sync with the pages html and body tags.

If you ask for the rawml using Kindleunpack, and examine the section map you can see all of this. Kindleunpack and Calibre has to undo all of this compilation nonsense to rebuild the epub without knowledge of any original file, or link names.

Sorry but in truth mobis are nothing like epubs, although KF8s are closer, they still need a lot of work to create something epub-like.

[Barb-B]

Hmmm, maybe I confused KF8 with azw3, thinking they were the same thing. When I "unwrap" an azw3 file, one of the things that comes out is a lovely fully formed epub file. Sorry if I wasted time by saying the wrong thing.

[KevinH]

Hi Barb,
No KF8 and AZW3 are the same things. When you "unpack" an azw3, the only reason you get a "lovely fully formed epub file" is because Calibre and the KindleUnpack program do their very best to undo all of the changes that happen when the ebook is compiled into a mobi format. But as I was trying to explain, all filenames are lost because they simply do not exist in the compiled mobi, which instead stores the contents of the files in sequential sections.

So no mobi is like an epub where all you have to do is unzip/open it. You actually have to rebuild it into something the user can work with, and for KF8/azw3 files we chose an epub.

Sorry, to burst your bubble, but this is what is going on inside.

Hope this helps,

KevinH

[KevinH]

Hi

If it helps, if you use Kindlegen to create your mobi, it will actually store the original epub file that was used to generate it inside a SRC section so that it can be kept. When you submit a kindlegen mobi to Amazon you are actually giving Amazon both the mobi and a complete copy of your original epub source! I am betting they are keeping that original epub source someplace for their own later use, but when they ship a mobi to a reader they strip off the original epub source.

Take care,

KevinH

[Barb-B]

Kevin,
Thanks so much for taking the time to explain all that, and being patient with a second go-round.

I'm guessing maybe that original epub source you mention is what sometimes appears in the unpack folder, and made me think what I was asking could be a possibility. I think I'll go check out Kindlegen and see if I can do what I want with it.

[edit] OK, so now I've done that, and it was quick and easy and accomplishes exactly what I was hoping for. So you tipped me in the right direction - thanks!