Koreader running on Kobos (all of them, hopefully!)

[giorgio130]

Originally Posted by Markismus

Both 561 and 566 have a serious defect in dictionary lookup.

I've seen your bug report, however I have no clue on solving it since I've never worked on that aspect. I could try though...

[Markismus]

Great! Chrox is doing some work on it, but it is not working like build 545, yet. I hope it gets fixed before I try to enhance koreader with support for ancient Greek dictionary lookup.

It is in that line that I tried dictionary support for german Fraktur. It already has a dictionary, so it only puts a strain on the OCR side. Once I get the ancient Greek dicitonary in utf-8 format done, we'll see whether the lookup for non standard symbols holds up.

Would you already have an idea why the Aura HD's get the permission errors with suspend.sh?

[giorgio130]

Originally Posted by Markismus

Would you already have an idea why the Aura HD's get the permission errors with suspend.sh?

That's another thing that is bugging me. I have no idea and looking up the error online does not return anything useful.

[paola]

sorry guys, do you get dictionary support for pdf in any build?

[jlynton]

New build 570 available.

[stasys]

What's new in it?

[jlynton]

Originally Posted by stasys

What's new in it?

I have just installed, and can't see any obvious changes.

[Markismus]

Originally Posted by paola

sorry guys, do you get dictionary support for pdf in any build?

I get dictionary support for english, german, Fraktur (gothic german) and I am working on getting support for ancient greek.

The highlighting works (545 at least).

Only notes remain, if I remember your request correctly...


Of course you have to fill the dict directory with Stardict dictionaries and the tessdata directory with TesseractOCR data. If you do that you get lookup for text pdf's both with/without reflow mode and scanned pdf's in reflow mode.

In 1-Step-Hack I included the English TesseractOCR data and Wordnet dictionary.

[hawhill]

Originally Posted by jlynton

I have just installed, and can't see any obvious changes.

That's actually a good sign :-) Only some implementation details changed - i.e. no interface parts. Good to know that nothing got really broken.

This is the commit log which might give an idea what happened in the codebase:
https://github.com/koreader/koreader/commits/master

When there is even minor development activity, new nightlies will be build. They might be irrelevant from a user's perspective.

[paola]

Originally Posted by Markismus

I get dictionary support for english, german, Fraktur (gothic german) and I am working on getting support for ancient greek.

The highlighting works (545 at least).

Only notes remain, if I remember your request correctly...


Of course you have to fill the dict directory with Stardict dictionaries and the tessdata directory with TesseractOCR data. If you do that you get lookup for text pdf's both with/without reflow mode and scanned pdf's in reflow mode.

In 1-Step-Hack I included the English TesseractOCR data and Wordnet dictionary.

mmh, interesting - I do have all the dictionaries, but installed in a separate directory to avoid having to update it at every build. Maybe I've done something wrong linking there, I'll check, but thanks a lot, at least now I know it's me :-)

EDIT: got it, forgot to amend the koreader_kobo.sh file, thanks (I had reinstalled filemonitor, and forget to tell it to go look for this file in the .kobo directory)!

[Markismus]

Originally Posted by paola

mmh, interesting - I do have all the dictionaries, but installed in a separate directory to avoid having to update it at every build. Maybe I've done something wrong linking there, I'll check, but thanks a lot, at least now I know it's me :-)

EDIT: got it, forgot to amend the koreader_kobo.sh file, thanks (I had reinstalled filemonitor, and forget to tell it to go look for this file in the .kobo directory)!

Ancient Greek is underway. The dictionary is here. The lookup crashes build 545, though. In the next build special characters are represented correctly, too.

[paola]

that's very kind marksimus!

[jlynton]

New build 583, margins are correctly restored when re-opening book.

[Markismus]

Ancient Greek dictionary support is working rudimentary! It works for non-reflow text, but not yet in reflow mode.

1. the code needed will be in the next nightly (after build 583).
2. the dictionary is Liddell-Scott-Jones 126k words

For now it doesn't seem to work in reflow-mode. But when it will, you'll need

1. OCR-data for Ancient Greek. (Decompress and copy to koreader/data/tessdata.)
2. Change the following lines in default.lua to:
   -- document languages for OCR
   DKOPTREADER_CONFIG_DOC_LANGS_TEXT = {"English", "Chinese_S", "Ancient Greek"}
   DKOPTREADER_CONFIG_DOC_LANGS_CODE = {"eng", "chi_sim", "grc"}

A small Latin-English dictionary can be downloaded here. (Remember to decompress the dict.dz-file, before copying the ifo-, idx- and dict-files to the koreader/data/dict directory)

[stasys]

Can somebody, please, move bottom settings panel to center and make bottom touch area turning page forward?
Latest builds are cosmetic adjustments for dictionaries and margins, but usability of Koreader is somehow left aside