Hyphenation not according Dutch rules

[cramoisi]

Originally Posted by outlandos

On all other books, when i do this, it says: No results (er zijn geen overeenkomstige schrijfregels gevonden).

Am I doing something wrong or is this okay?

It’s okay. Not all the books have specific hyphenation lines in their CSS. But if they have it, they need to be removed. If it’s no result, you just have to continue the process

[outlandos]

Originally Posted by cramoisi

It’s okay. Not all the books have specific hyphenation lines in their CSS. But if they have it, they need to be removed. If it’s no result, you just have to continue the process

None of my other books got any specific lines in their CSS.

Anyway, bad news.. It did not fix the problem, so im afraid the Dutch hyphenation dictionary used my Kobo is just incorect.

I do not know if you will got a Glo HD at your disposal, but Ill mail you the e-book, and im 99% sure you will have the same faulty hyphenation there.

http://static.tweakers.net/ext/f/8Ka...sY9UU/full.jpg

ontvoer-d. ???

If you can reproduce the 'bug' should we report this somewhere? I cant believe Dutch hyphenation works incorrectly, it seems to me like such a key (and basic) feature for a (130 euro) e-reader..

[cramoisi]

Hi ! Is the kobo_extra.css still there ? Is the custom CSS of the plugin correctly checked ? Have you ‘hyphanate this’ the book before sending it to the device ?

I’m wondering that for kepubs we probably need to force text-align:justify; on the text and deactivate all alignement in the Aa menu of the book. Is your book setting on full justify (it’s not, I checked it - so it need one more step : change the text alignment with calibre for a full justify one and disable the justification in the book’s Aa menu on the device) ?

I added soft hyphen myself and the plugin added them correctly :

no soft hyphen before ‘d’ :
https://www.dropbox.com/s/qkrig3ngb4...51.46.png?dl=0

soft hyphen at the right place :
https://www.dropbox.com/s/hl2v6t1gxu...52.08.png?dl=0

I’ll try on a gHD as soon as possible.

[JSWolf]

Originally Posted by cramoisi

Hi ! Is the kobo_extra.css still there ? Is the custom CSS of the plugin correctly checked ? Have you ‘hyphanate this’ the book before sending it to the device ?

I don't think you understand what is going on. The problem with Dutch hyphenation is the hyphenation dictionary has errors. Using Hyphenate This! would still have the same hyphenation issues as in kepub. The only way to fix Dutch hyphenation is every time you find an incorrect hyphen, edit the hyphenation dictionary to fix it.

[cramoisi]

Originally Posted by JSWolf

I don't think you understand what is going on. The problem with Dutch hyphenation is the hyphenation dictionary has errors. Using Hyphenate This! would still have the same hyphenation issues as in kepub. The only way to fix Dutch hyphenation is every time you find an incorrect hyphen, edit the hyphenation dictionary to fix it.

Hi !

I just proved that it is not true.

1) The dutch dic is correct and the soft hyphen has been correctly added by the plugin as it can be seen in the screenshots above. If the dic was wrong a soft hyphen would be added just before the d. There is none as shown by the 1st screenshot. (the dic in the plugin is of course the very same than in the device)

2) because I think I can read a hyphen.dic and the rules for the word which is problematic here are correct, as the first point already shows it anyway.

(but if I’m wrong, you can point out the different rules applied on the word ‘ontvoerd’ and show me where I missed something)

It’s just that when full justification is selected, the kepub renderer takes the upper hand on the style and set the internal stylesheet at 'auto' and the two behaviors mixes up. I got the same problem with french books. Deactivating the justification in option and set a full justification in the book’s css via calibre will solve the issue.

[outlandos]

I did indeed select full justification in Nickle itself.

So I should undo that and then delete all books and select justification in Calibre?

[outlandos]

Could anyone please point me out where exactly in the calibre options i can select full justification? Im searching for half an hour now without luck.

[cramoisi]

Originally Posted by outlandos

Could anyone please point me out where exactly in the calibre options i can select full justification? Im searching for half an hour now without luck.

Done by PM, sorry for the delay !

[outlandos]

Originally Posted by cramoisi

Hi !

I just proved that it is not true.

1) The dutch dic is correct and the soft hyphen has been correctly added by the plugin as it can be seen in the screenshots above. If the dic was wrong a soft hyphen would be added just before the d. There is none as shown by the 1st screenshot. (the dic in the plugin is of course the very same than in the device)

2) because I think I can read a hyphen.dic and the rules for the word which is problematic here are correct, as the first point already shows it anyway.

(but if I’m wrong, you can point out the different rules applied on the word ‘ontvoerd’ and show me where I missed something)

It’s just that when full justification is selected, the kepub renderer takes the upper hand on the style and set the internal stylesheet at 'auto' and the two behaviors mixes up. I got the same problem with french books. Deactivating the justification in option and set a full justification in the book’s css via calibre will solve the issue.

cramaisi seems to be right; i select justification in calibre, and in kobo i selected ''off'' and this is the result:

http://static.tweakers.net/ext/f/WuA...kCtKg/full.jpg

[cramoisi]

Originally Posted by outlandos

cramaisi seems to be right; i select justification in calibre, and in kobo i selected ''off'' and this is the result:

http://static.tweakers.net/ext/f/WuA...kCtKg/full.jpg

Yup. I reproduced the bug on my colleague glo HD this morning. It was the very same than with french and is due to the full justification option in kepub renderer which set the internal css to auto when selected and overrode the rule added in the book by kobo_extra CSS. With ‘auto' enabled, the renderer should select preferably soft or hard hyphens when they are present but it can cut whenever it wants according its algorithm (which is the guilty one here), so the error you got.

To summarize, the steps to get correct hyphens on kepubs with this method are :

0. adding kobo_extra.css at the root of the device
1. removing mention of -webkit-hyphens in the book’s css
2. justifying left and right in the book’s css (text-align:justify; )
3. hyphenate this ! book
4. sending the book via KoboTouchExtended plugin with custom css checked and options to add hyphens unchecked.
5. disable justification in the aA menu of the book
6. reading the book cover to cover

I should add that there is no problem with hyphenation in epubs anymore (since 3.16.0). If you find working on your books time-consuming, you could just read epubs

[JSWolf]

Originally Posted by cramoisi

Hi !

I just proved that it is not true.

1) The dutch dic is correct and the soft hyphen has been correctly added by the plugin as it can be seen in the screenshots above. If the dic was wrong a soft hyphen would be added just before the d. There is none as shown by the 1st screenshot. (the dic in the plugin is of course the very same than in the device)

2) because I think I can read a hyphen.dic and the rules for the word which is problematic here are correct, as the first point already shows it anyway.

(but if I’m wrong, you can point out the different rules applied on the word ‘ontvoerd’ and show me where I missed something)

It’s just that when full justification is selected, the kepub renderer takes the upper hand on the style and set the internal stylesheet at 'auto' and the two behaviors mixes up. I got the same problem with french books. Deactivating the justification in option and set a full justification in the book’s css via calibre will solve the issue.

The left/right values could be different between the two and that would make the hyphenation results different. If I recall, it was discovered that the kepub renderer was off by 1 in the left/right values.

[outlandos]

Originally Posted by JSWolf

The left/right values could be different between the two and that would make the hyphenation results different. If I recall, it was discovered that the kepub renderer was off by 1 in the left/right values.

So you think it's not fixed and that it's just a matter of time before I find any new errors in the hyphenation?

[davidfor]

Originally Posted by outlandos

So you think it's not fixed and that it's just a matter of time before I find any new errors in the hyphenation?

From watching the thread, that would be my conclusion. The instructions cramoisi posted are a workaround to prevent the kepub renderer from doing the hyphenation.

The problem is that I don't have a clear understanding of how it is wrong. I haven't seen a clear test case that would make it easy for a developer to see and hopefully fix the problem. Especially a non-Dutch speaking developer. If I have a chance over the weekend, I might see if I can work it out, but, being mono-lingual and not really understanding English hyphenation rules might make that a bit hard.

[cramoisi]

Originally Posted by JSWolf

The left/right values could be different between the two and that would make the hyphenation results different. If I recall, it was discovered that the kepub renderer was off by 1 in the left/right values.

Originally Posted by davidfor

From watching the thread, that would be my conclusion. The instructions cramoisi posted are a workaround to prevent the kepub renderer from doing the hyphenation.

It’s exactly what I’ve done here : disabled the kepub renderer and forced the hyphenation to applied only on the soft hyphens correctly set by the plugin. Even if the kepub renderer «was off by 1 in the left/right values» (which is right) it will change nothing since it is shut off. I’ve yet not seen any error in french or english with this method and i’ve read 30 books since i applied it on my device.

[davidfor]

Originally Posted by cramoisi

It’s exactly what I’ve done here : disabled the kepub renderer and forced the hyphenation to applied only on the soft hyphens correctly set by the plugin. Even if the kepub renderer «was off by 1 in the left/right values» (witch is right) it will change nothing since it is shut off. I’ve yet not seen any error in french or english with this method and i’ve read 30 books since i applied it on my device.

That is fine. But, wouldn't it be better to get the renderer fixed? That way you don't have to play with the books. And anyone who buys and downloads directly from Kobo will have a better experience. Plus the search and dictionary lookup will work. If we supply a clear statement of how the renderer is wrong, preferably with test cases, Kobo will look at it. They might still not fix it, but that will be depending on time and importance of this compared to whatever else they are doing.