ePub 3 to kindlegen

[AliceWonder]

Operating System: x86_64 CentOS 7 Linux (supports i386 too)

kindlegen tarball: kindlegen_linux_2.6_i386_v2_9.tar.gz

According to all the docs I read, it's suppose to produce a single file that has both mobi and kf8 in it. That's not what I'm seeing.

Source ePub 3 (which passes validation) is 7.4 MB (much of that is fonts)

When running the kindlegen binary on it, TWO files are produced - one with a .mobi extension and one with a .mobi8 extension.

The .mobi file is 14.7 MB so maybe it does have two versions in it?

The .mobi8 file is 106.6 kB meaning it clearly doesn't have much of anything in it, guessing none of the fonts. It would fit on the old 400k floppies my classic mac used

Can anyone describe what these output files are since they differ from all documentation I've seen on what is suppose to be produced (single output file)

I don't yet have a kindle to see how the conversion went, should be arriving shortly.

Seems the preview app doesn't work on Linux so I can't preview until I get a kindle (I don't have Windows or MacOS/OS X)

[Doitsu]

Originally Posted by AliceWonder

When running the kindlegen binary on it, TWO files are produced - one with a .mobi extension and one with a .mobi8 extension.

The .mobi and .mobi8 are temporary files. Usually, you only see them when KindleGen processes rather large files. Youl'll have to wait until KindleGen is done and displays:

Info(prcgen):I15000:  Approximate Standard Mobi Deliverable file size : 000xxxKB
Info(prcgen):I15001:  Approximate KF8 Deliverable file size :   000xxxKB

[AliceWonder]

All it is generating is the .mobi and .mobi8 files.

It is giving a bunch of warnings about abbr tag:

Warning(inputpreprocessor):W29007: Rejected unknown tag: <abbr>
      in file: /tmp/mobi-VUdMwo/OEBPS/About.xhtml     line: 0000125

I may need a shell script to change those to span tags before converting if that's a problem (abbr tag is very useful for a11y screen readers, at least in standard html, so they belong in the ePub)

But if those are just temporary files, then the program is exiting before completing w/o giving an error message as to why. But it does exit with status 0 which should indicate a successful exit.

I'll have to see if I can borrow a Mac and see if it behaves differently there.

[AliceWonder]

Okay in verbose this is how program ends -

Info(prcgen):I15000:  Approximate Standard Mobi Deliverable file size :   4184073KB
Info(prcgen):I15001:  Approximate KF8 Deliverable file size :   0003591KB
Info(prcgen):I1036: Mobi file built successfully

I'm assuming that it just is simply not trying to combine them into a single file, either because there is a binary it wants not present or because there is a bug in the Linux version.

It's not uncommon for commercial companies with closed source software to not really care about their linux builds.

[Doitsu]

Originally Posted by AliceWonder

I'm assuming that it just is simply not trying to combine them into a single file, either because there is a binary it wants not present or because there is a bug in the Linux version.

AFAIK, Kindlegen is a standalone binary without unusual dependencies. The Linux version works just fine on my Arch Linux version. If it doesn't work on your Linux machine, it's most likely a distro/security issue. (Maybe your OS prevents Kindlegen from merging the temp files or your file manager simply doesn't refresh the folder display after Kindlegen is done.)

Try compiling a blank epub created with Sigil. If you still end up with temporary files, it's definitely an OS issue.

[AliceWonder]

I have a sneaky suspicion that the .mobi8 file is a garbage file it forgets to delete and the .mobi is what I want.

[JSWolf]

Originally Posted by AliceWonder

I have a sneaky suspicion that the .mobi8 file is a garbage file it forgets to delete and the .mobi is what I want.

Yes, the file you want does end in .mobi.

[odamizu]

If KindleUnpack works on Linux, you can use it to unpack the .mobi file and see if there's a KF8 and standard Mobi inside.

In addition to the python scripts there is a KindleUnpack Calibre plug-in