[Conversion Plugin] KFX Input

[jhowell]

Version 1.13.0 - 02 Feb 2019

Handle unexpected data found in some print replica textbooks. (Fixes "TypeError: %d format: a number is required, not IonStruct")

Ignore unexpected control characters found in some print replica textbooks.

Support KPF files produced by Kindle Create version 1.20. (Fixes "$538 has extra data: ('yj.semantics.page_number_begin': ...")

Support additional KFX features. (reflow-language-2, reflow-language-expansion-1, reflow-style-12, Ion symbols $136 $137)

[twokatmew]

Hello, I've been using Calibre and DeDRM tools to convert Kindle books for several years now. I'm running the latest DeDRM version (6.6.1), and I've updated all plugins. (Kfx-Input is v.1.13.0) I'm running v1.25.1 of Kindle for PC and Calibre 3.39.1. Kindle books are still importing as KFX-zip. I've reinstalled DeDRM tools and updated other plugins. Has something changed that I'm not aware of? Thanks!

[shamanNS]

@twokatmew:

Import cannot currently be performed on KFX books downloaded using Kindle for PC/Mac version 1.25 or greater due to new DRM.

You will need to install older version of Kindle for PC, and then disable updates in its settings.
Luckily you're not using Kindle for Mac...

Direct download link for Kindle for PC v1.24.3 (51068).

[DRM_hater]

Originally Posted by twokatmew

Hello, I've been using Calibre and DeDRM tools to convert Kindle books for several years now. I'm running the latest DeDRM version (6.6.1), and I've updated all plugins. (Kfx-Input is v.1.13.0) I'm running v1.25.1 of Kindle for PC and Calibre 3.39.1. Kindle books are still importing as KFX-zip. I've reinstalled DeDRM tools and updated other plugins. Has something changed that I'm not aware of? Thanks!

I was having the same problem just now.

In my case I'm using Mac. I found an old version online (1.17, LINK HERE).

I use App Cleaner (free app, LINK HERE) to uninstall Kindle completely and install the old version. Before putting your Amazon credentials, go to preferences and disable auto update.

[zionius]

I downloaded 2 books today, A was published in Nov. 2018, B was published in Oct. 2014, but recently had a update (so probably changed to a latest format). Kindle for PC version is 1.24.51068. The plugin worked perfectly for A. But didn't work for B. Calibre can read the book metadata correctly and creat a 204 MB kfx file. But when I convert it to epub/zip, it reports "out of memory". Here's the convert log:

calibre-debug -r "KFX Input" -- the.kfx
Software versions: KFX Input 1.13.0, calibre 3.39.1, Windows-7-6.1.7601-SP1
KFX Input plugin help is available at https://www.mobileread.com/forums/sh...d.php?t=291290
Converting the.kfx
Processing container: the.kfx
kfxgen version: 7.191.1.0/
Features: CanonicalFormat-2, kfxgen.pidMapWithOffset-1, kfxgen.textBlock-1, max_id-779, reflow-section-size-2, reflow-style-9
Metadata: ASIN=***, asset_id=***, author="***", book_id=***, cde_content_type=EBOK, content_id=***, cover_image=717x950, creator_version=2.15.0, file_creator=YJConversionTools, is_sample=False, issue_date=***, language=en, override_kindle_font=True, publisher="***", reading_orders=1, selection=enabled, title="***"
Converting book to EPUB 2.0
Format is EBOK
WARNING: Missing font family names: Anavio
Present referenced font family names: Archetype, Bodonian, Bodonian6, reads-Shift Light
Python function terminated unexpectedly
out of memory (Error Code: 1)

The kfx file is probably corrupted too, since it couldn't be opened in Kindle for PC. But idk how to generate log for that.

[jhowell]

Originally Posted by zionius

Calibre can read the book metadata correctly and creat a 204 MB kfx file. But when I convert it to epub/zip, it reports "out of memory".

A "memory error" occurs when converting a large book (204 MB in this case) that requires more memory than is available on your system. The 32 bit version of calibre is limited in the size of books that it can process.

You can switch to the 64 bit version of calibre if your system can handle that. If not then try one of the methods in this post for avoiding KFX: Dealing with Kindle for PC/Mac 1.19 (and later) and KFX in calibre

[zionius]

Originally Posted by jhowell

You can switch to the 64 bit version of calibre

It works! Thanks!

[twokatmew]

Originally Posted by shamanNS

@twokatmew:


You will need to install older version of Kindle for PC, and then disable updates in its settings.
Luckily you're not using Kindle for Mac...

Direct download link for Kindle for PC v1.24.3 (51068).

Thank you!! Solved my problem.

[traktor]

Hello,

I have a problem after the recent Kindle software update (Paperwhite 2).

The plugin spits out KFX-ZIP. I updated Calibre to 3.39.1 and KFX Input to 1.13.0. I've tried re-installing the plugin. Didn't help.

I don't know if the Kindle software update has something to do with it but I thought I'll mention it.

What can I do to make it work?

Thanks.

[jhowell]

Originally Posted by traktor

I have a problem after the recent Kindle software update (Paperwhite 2)

The KFX Input plugin will not work with books from recent Kindle firmware releases. You are best off using the “Download & transfer via USB” function from Amazon’s website to obtain books in a usable format. (The details of how to do that are not an appropriate topic for this thread.)

[rumplestiltskin]

Not sure why this happened but using Calibre in Windows 10 to convert a KFX file to ePub resulted in some really ugly body copy, almost like a font is missing and the substitution is just plain wrong. I can't tell you what font it was, only that it was like some bizarre scaling of an old bit-mapped font.

On my Mac, the conversion worked perfectly. I examined the file in Sigil's (and Calibre's) code-view; there are 38 "classes" in the "stylesheet.css" file. I seem to recall reading that there should be no more than 10.

What's really bizarre is that [I]every paragraph of body copy[I] references "class11" which seems pretty excessive. Shouldn't there be only one reference that encompasses all the body copy in the chapter (assuming there are no there are no other styles needed within the body copy - like a callout or the like)?

Has anyone else seen such strangeness and, if so, should I just leave it alone or maybe someone has a recipe for correcting such an excessive amount of redundant code?

Thank you

[jhowell]

Originally Posted by rumplestiltskin

Not sure why this happened but using Calibre in Windows 10 to convert a KFX file to ePub resulted in some really ugly body copy, almost like a font is missing and the substitution is just plain wrong. I can't tell you what font it was, only that it was like some bizarre scaling of an old bit-mapped font.

On my Mac, the conversion worked perfectly.

That is odd. Perhaps you have different default conversion settings on the two calibre installations.

If you let me know the ASIN of the book I can check out the free sample of it on Amazon see if there is anything unusual about that particular book.

Originally Posted by rumplestiltskin

I examined the file in Sigil's (and Calibre's) code-view; there are 38 "classes" in the "stylesheet.css" file. I seem to recall reading that there should be no more than 10.

What's really bizarre is that [I]every paragraph of body copy[I] references "class11" which seems pretty excessive. Shouldn't there be only one reference that encompasses all the body copy in the chapter (assuming there are no there are no other styles needed within the body copy - like a callout or the like)?

The plugin tries to convert the KFX coding to the equivalent HTML, but the results are often not pretty. KFX format is heavily processed and its content differs significantly from what the book publisher provided to Amazon.

The class names used are arbitrary and 38 classes is reasonable. I have never heard of a limit of 10 classes before. Most books use many more than that.

Originally Posted by rumplestiltskin

Has anyone else seen such strangeness and, if so, should I just leave it alone or maybe someone has a recipe for correcting such an excessive amount of redundant code?

It's up to you. If the internal coding matters to you then I suggest using one of the documented methods of avoiding KFX format. Books delivered by Amazon in KF8 (azw3) format retain the original HTML coding provided by the publisher.

[BetterRed]

@rumplestiltskin - I know SFA about KFX, but this might be worth trying:

• save your Windows config folder somewhere safe.
• export your config folder from your Mac to a USB
• import that config folder into your Windows setup
• try the conversion

If conversion on Windows now delivers the same result as you get on your Mac, then your problem is the difference in conversion settings. Otherwise I refer to jhowell's suggestions.

BR

[rumplestiltskin]

Here's the ASIN: B07N7JRRSY

I'll try the export/import of the settings.

Thanks to all for the replies.

[rumplestiltskin]

I took the "ugly" ePub converted from KFX > ePub (and even tried it KFX > AZW3 > ePub), attached it to an eMail, sent it to my Mac, and opened it using Calibre's ePub reader. Looks fine. So could this simply be a case of a font missing from my PC?

I've uploaded the ePub to a cloud share so anyone with the inclination to look at the ePub's css (and the knowledge to interpret the results) might be able to determine what's going on here. The body copy at issue is "class10" and "class11" - both spec'd as "Roman" font. They look identical in the ePub (just fine on my Mac but ugly on my PC).
http://cloud2.zoolz.com/s-66YClaZr