JBPatch

[elRicharD]

I have just joined the Spanish effort. It is the least I can do to help out. Ixtab and the rest of the kt devs, you deserve to be paid by Amazon, with your code, the good stuff gets even better! Keep it up!

[techiemonkey]

hey ixtab and others,thanks for jbpatch.it is really awesome.here are some ideas for patches.if you dont want to do it then just ignore this:
1. graphical homescreen that actually shows book covers(like nook) not like the prototype patch that is somewhere buried in this forums.
2.a scrollbar in the pop up dictionary so if we dont find the meaning we want,we wont have to tap 'click for full definition button.'
3.make mp3 player better i.e show all songs in one go,add play where you left features,etc.
4.epub support(i dont know if this can be actually possible with jbpatch)
5.customising page turn areas

[calvin]

Originally Posted by bhaak

http://xkcd.com/1015/

nice comment, I think it depends on the fonts you are using. I have just switched my kindle reading font to Linux Libertine and I think the kerning is much better now.

[kbaerwald]

I tried several mobi/azw3 with that hyphenation patch - some do work, some don't. Most of the self-made azw/mobi (text imported from epub into Kindle Previewer then converted into mobi and directly transferred to my K5) do not show hyphenation. Most Amazon ebooks do!

System is Touch 5.1.2 with patch from yesterday.

What is my problem? Please find attached an old German text from Nietzsche which shows no hyphenation on my Kindle.

Klaus

[ixtab]

Originally Posted by kbaerwald

I tried several mobi/azw3 with that hyphenation patch - some do work, some don't. Most of the self-made azw/mobi (text imported from epub into Kindle Previewer then converted into mobi and directly transferred to my K5) do not show hyphenation. Most Amazon ebooks do!

System is Touch 5.1.2 with patch from yesterday.

What is my problem? Please find attached an old German text from Nietzsche which shows no hyphenation on my Kindle.

Klaus

I'm relatively sure that .azw3 uses a completely different reader software, so the patch will not affect those books ("relatively", because I didn't get my hands on an .azw3 book yet). For the problem with .mobi/.azw files, I'll look into it, but it may take a week or two.

Edit: do the files where hyphenation doesn't work get block-aligned properly? If so, then it may be a problem with the book's language - it may be set incorrectly. You can check the log file from the JBPatch UI (System tab), it logs which language it is looking for.
Edit2: I verified the problem with the book that you attached. For some reason, the system doesn't request hyphenations for that book.

[elRicharD]

Originally Posted by elRicharD

I have just joined the Spanish effort. It is the least I can do to help out. Ixtab and the rest of the kt devs, you deserve to be paid by Amazon, with your code, the good stuff gets even better! Keep it up!

I am not sure what hyphenation is, but for the rest, Spanish is in place

[kbaerwald]

Thanks for the quick response - my log says:

Loading hyphenations for language ID 7 (de)
Hyphenation definitions loaded from /mnt/us/opt/j...
(here no hyphenation! - this is the Nietzsche text)

Loading hyphenations for language ID 1031 (de)
Hyphenation definitions loaded from /mnt/us/opt/j...
com.mobileread.ixtab.patch.tts applied to com....
(here hyphenation! - this is an ebook downloaded from Amazon)

Looks like this confirms your findings ...

[ixtab]

@kbaerwald:

The book you attached was in some very weird format, probably some old (or extremely new?) .mobi version. I simply imported it into calibre, let it convert (yes, input format=mobi, output format=mobi), and verified that the language was correctly set. That produced a file which is about 1/3 of the size of the file you posted, and which is hyphenated properly. I'm attaching the working one.

You may want to give this method a try.

[ixtab]

Originally Posted by elRicharD

I am not sure what hyphenation is, but for the rest, Spanish is in place

This is hy-phe-na-tion (rather: syllabification). And this is an example of wrong h-yp-henati-on. It's used for splitting words at the end of the line.

Muchas gracias por la traduccion!

[kbaerwald]

Originally Posted by ixtab

@kbaerwald:

The book you attached was in some very weird format, probably some old (or extremely new?) .mobi version. I simply imported it into calibre, let it convert (yes, input format=mobi, output format=mobi), and verified that the language was correctly set. That produced a file which is about 1/3 of the size of the file you posted, and which is hyphenated properly. I'm attaching the working one.

You may want to give this method a try.

The patch log still says "ID 7" but now shows hyphenation. Funny! The mobi was made with the latest Kindle Previewer version directly downloaded from Amazon. I wanted to avoid any mix up with calibre . Now it looks like I need calibre instead .

Thank you for the mobi/mobi conversion hint - I will give it a try.

[ixtab]

Originally Posted by kbaerwald

The patch log still says "ID 7" but now shows hyphenation. Funny! The mobi was made with the latest Kindle Previewer version directly downloaded from Amazon. I wanted to avoid any mix up with calibre . Now it looks like I need calibre instead .

Thank you for the mobi/mobi conversion hint - I will give it a try.

There is a slight caveat here, when making sense of the log file: The "loading hyphenation" thing is only shown when a *new* language ID is encountered. In other words: if you're loading 3 books in a row, which all have language ID 7 set, the message will only be shown once (for the first book). Once you load an english book (for instance), then you will again see the message.
(In your case, the "ID 7" that you saw previously was from *another* book that you loaded before the Nietzsche one. )

The way that I tried it was this:

Attempt 1:
- english book (message shown)
- Nietzsche (no message)
- german book (message shown)
Attempt 2:
- german book (message shown)
- Nietzsche (no message)
- english book (message shown)

This confirmed that the Nietzsche was neither german, nor english. In fact, it confirmed that the system didn't even try to look up any rules for that book

Edit: I just re-read your first post. If all you want to do is import epub books, then you can probably skip the Kindle previewer step completely, and simply import the books directly into calibre. It will automatically convert them to a (working) .mobi format when they're sent to the device. Library management with calibre really works like a charm!

[NiLuJe]

@ixtab/kbaerwald: The file attached was indeed a KF8/AZW3/mobi7/mobi8 (pick your favorite name ^^), don't be fooled by the .mobi extension . KindleGen/KindlePreviewer generates dual .mobi files since v2, with a mobi7 part (KF8), and a legacy mobi6 part, in the same file. The mobi7 part is used on devices that supports it (Fire/PC/Mac/iOS/Android/Touch 5.1/K4 4.1), while the legacy one is used by older devices.

And, yeah, I'm pretty sure it uses another reader entierly (com.amazon.ebook.booklet.htmlreader). No idea if there's a hint of hyphenation engine in there, but it does obey the shy (soft hyphen) html entity, and supports the language detection for the dictionary.

Since my device came with 5.1, I've been mostly using mobi7 files, except for old stuff purchased on Amazon that's still in mobi6/topaz...

[Antoinekamel]

Originally Posted by ixtab

Thank you! portuguese is now 96% translated: https://www.transifex.com/projects/p/jbpatch/

100% now.

[eureka]

@ixtab, .azw3 (as NiLuJe said) is a dual .mobi with KF8 content included. On KT, Java UI for viewing KF8-formatted content recieves from rendering process BMP image with some metadata (AFAIR, raw text and locations of words on screen for selecting/translating purposes). Rendering process is the webreader (/usr/bin/webreader) process. As KF8 format is a "restricted" HTML5-like format, webreader renders it with the help of WebKit engine and sends rendered image to Java UI.

webreader could be controlled through HTTP requests (it listens on 127.0.0.1:20450, you could find simple debug page by making HTTP request to it and control commands are represented as various URL-paths with query strings). But I believe, it's impossible to intercept rendering process without binary patch to webreader (I could be wrong, though).

[ixtab]

@eureka: Oh my. Seems like they really like to make things as complicated as they can. So yeah, this means no hyphenation for KF8 (at least not with JBPatch).

There seems to be a workaround by modifying the book's content itself: https://www.mobileread.com/forums/sho...d.php?t=179736. So it seems like we need someone to write a calibre plugin to add "shy" entities all over the place - any volunteers?

If you want to see the method that the Kindle uses for displaying KF8 for yourself, you can connect to your kindle like this:

ssh -L 20450:127.0.0.1:20450 root@kindle

Then open a KF8 book on the Kindle, and point the browser (on your PC) to http://localhost:20450/