Creating Japanese epub3 and kindle books from aozora bunko texts

[mike2046]

Hi,

I recently came across a bit of software AozoraEpub3 which can convert Aozora .txt files into epub3 format files. See the following link for more information

http://blog.preining.info/2013/01/10...nts-to-kindle/

it can produce either vertical and horizontal format files, depending on settings, and can also be set to exclude images. It also seems to handle ruby text when present. As the interface is in Japanese, the above link is useful for explaining how to use it - although as icons are used a lot some functions can be guessed at. Once an epub3 format file is produced, this can be run through kindlegen to produce a version for a Kindle.

Using this I have been able to produce vertical text books for my Kindle White. These also include the ruby text. Pages turn from right to left as expected when going "forward" through the book.

The other bonus of this is that by looking at the contents of the epub file one gains an idea of how to create a vertical text format book - which can be applied to one's own material.

[PoP]

Quite interesting. Just to report my findings on the Kindle Keyboard with firmware 3.4

Ruby works fine! Top to bottom, Right to Left not too good.

Following are (correct) Calibre vs (incorrect) Kindle rendering comparison

First page:


Last page:


...IMHO

[Doitsu]

I looked into displaying vertical text some time ago and stumbled upon Ryan Stenhouse's website.
Stenhouse came up with an ingenious way to approximate vertically rendered Japanese text that works even on a K3.

[PoP]

Good hack to know.

Previously I achieved a similar style in the Vulcan guide accompanying my NADSAT dictionary here, by using -webkit-transform:rotate(90deg); and a font with -90deg sideways glyphs.

[Doitsu]

Originally Posted by PoP

Previously I achieved a similar style in the Vulcan guide accompanying my NADSAT dictionary here, by using -webkit-transform:rotate(90deg); and a font with -90deg sideways glyphs.

Also good to know.

IIRC, the official Amazon style guide doesn't mention that KF8 supports CSS3 2D Transforms in general and -webkit extensions in particular.

[13xforever]

Guys, I might be necroposting, but I just wanted to say thank you, like a million times. You saved me a lot of time with this thread.

[99nyorituryo]

I'm not the developer of AozoraEpub 3. I modified AozoraEpub 3.
It's interesting to see AozoraEpub 3 users overseas.

https://github.com/kyukyunyorituryo/AozoraEpub3

[Quoth]

Use epub2 for ebooks, HTML for interactive / more advance. The epub3 is an almost pointless spec for ereaders, you need an app on a tablet, in which case HTML etc is a better more portable solution!

If what you want doesn't work on ePub2, then use HTML.

[JSWolf]

Originally Posted by FrustratedReader

Use epub2 for ebooks, HTML for interactive / more advance. The epub3 is an almost pointless spec for ereaders, you need an app on a tablet, in which case HTML etc is a better more portable solution!

If what you want doesn't work on ePub2, then use HTML.

This is for a Kindle using KF8. KF8 does support vertical text.

[Turtle91]

Originally Posted by FrustratedReader

Use epub2 for ebooks, HTML for interactive / more advance. The epub3 is an almost pointless spec for ereaders, you need an app on a tablet, in which case HTML etc is a better more portable solution!

If what you want doesn't work on ePub2, then use HTML.

I'm confused... ePub2 and ePub3 both use HTML... ePub3 simply allows some of the multimedia features, etc.

[jhowell]

Originally Posted by FrustratedReader

If what you want doesn't work on ePub2, then use HTML.

Ruby is an important part of Japanese text. EPUB 2 does not support ruby, at least according to EPUBCHECK.

[JSWolf]

Originally Posted by jhowell

Ruby is an important part of Japanese text. EPUB 2 does not support ruby, at least according to EPUBCHECK.

ePub2 does not support vertical text.

[Quoth]

The ePub2 uses a subset of the HTML a browser can use. Lots of HTML that works in a browser will not work in an eink Kindle (even PW3) or eink ePub based ereader.

I blame the Western Roman/Latin centric nature of the design of the original mobi and epub.

Some ereaders still in use can't do Hebrew, Arabic, Korean, Cyrillic, Horizontal Japanese, Horizontal Chinese and most other writing systems, basically only Roman-Latin and some Greek (really for maths & science, not Greek language). Linux supported all these before the first Amazon Kindle came out, which used Linux.

Hardly anything other than a few apps support epub3. Also some apps only partially support it.

So for non-mainstream western languages, it's only newer ereaders. For vertical text, an app or browser on an Android or Apple Tablet, not an eink ereader as it might not work.

[JSWolf]

Originally Posted by FrustratedReader

The ePub2 uses a subset of the HTML a browser can use. Lots of HTML that works in a browser will not work in an eink Kindle (even PW3) or eink ePub based ereader.

I blame the Western Roman/Latin centric nature of the design of the original mobi and epub.

Some ereaders still in use can't do Hebrew, Arabic, Korean, Cyrillic, Horizontal Japanese, Horizontal Chinese and most other writing systems, basically only Roman-Latin and some Greek (really for maths & science, not Greek language). Linux supported all these before the first Amazon Kindle came out, which used Linux.

Hardly anything other than a few apps support epub3. Also some apps only partially support it.

So for non-mainstream western languages, it's only newer ereaders. For vertical text, an app or browser on an Android or Apple Tablet, not an eink ereader as it might not work.

For non-mainstream Readers, it's just a matter of embedding a proper font for some languages. And if you want text to read from left-to-right, just reverse the letters. It can be done even if it is a kludge.

[Doitsu]

Originally Posted by JSWolf

And if you want text to read from left-to-right, just reverse the letters. It can be done even if it is a kludge.

That method would only work for RTL languages such as Hebrew that don't have different letter forms for initial, medial, final and isolated letters. (In Hebrew, five letters actually have different final forms that would need to be hard-coded.)