KFX dictionary

[PoP]

I would like to create a NADSAT KFX dictionary (just as a proof of concept).

As the wiki says, dictionaries for new languages are now formatted as an SQLite data base.

and,

Originally Posted by Doitsu

Unfortunately, Amazon hasn't released an official KFX dictionary converter. You're stuck with KindleGen and MOBI dictionaries for the time being.

I have been looking at the "Gujarati into Gujarati dictionary_B06X6HVPX1.kfx" dictionary

Spoiler:

Has anyone else tried to reverse engineer the format? There are not many tables, and the few fragment properties are possibly indicative of a simple structure. I am hoping that replacing the encrypted DRM blobs with properly structured clear text would produce a valid KFX dictionary.

[jhowell]

I looked into this awhile ago.

Kindle Previewer 3 and Kindle Create produce a variant of KFX format known as KPF (Kindle Package Format). The main component of this format is an SQLite 3 database containing the book's metadata and content split into pieces known as fragments or entities. These are highly structured and coded in Amazon's binary Ion format.

The KFX dictionaries appear to be in the same format with some exceptions: The fragments that make up the book's content are encrypted in DRMION format. There is an added table that maps words in the dictionary in-language to positions within the book's content, presumably where their definitions are located.

Inflection processing appears to be hard coded in the reader software for the small set of languages that are supported by KFX dictionaries.

The KFXGenApp, which is part of Kindle Previewer 3 and is used by my KFX Output plugin for calibre, lacks the ability to convert dictionaries. If you bypass the check preventing conversion you find that it is missing language-specific DLLs that are needed for that process.

[coolice]

Originally Posted by PoP

I would like to create a NADSAT KFX dictionary (just as a proof of concept).

As the wiki says, dictionaries for new languages are now formatted as an SQLite data base.

and,


I have been looking at the "Gujarati into Gujarati dictionary_B06X6HVPX1.kfx" dictionary

Spoiler:

Attachment 170410

Has anyone else tried to reverse engineer the format? There are not many tables, and the few fragment properties are possibly indicative of a simple structure. I am hoping that replacing the encrypted DRM blobs with properly structured clear text would produce a valid KFX dictionary.

Hi, I know it was a while, but may I ask if you succeeded ?

[PoP]

Sadly I didn't. Lack of documentation, and proprietary KFXGenApp, halted my bragging pretentions.