Kobo Touch and Arabic?

[readingglasses]

went back to indigo, tested microsdhc. went in all the way, flush with machine edge (the one I used before must've been a little blocked and so the card was just never read by the machine). All while avoiding a certain somebody I never thought I'd see again, right in the store.

it had firmware 1.9.5 (june 27th or stg).

epub, html, various text formats (utf8, unicode, ansi). NO pdf testing, I'm just assuming pdf's are fine.
anc. greek from gutenberg project looked nice except for occasional question marks. pehs they were compound characters?

english control group files were displayed perfectly.

arabic, hebrew and kanji a total question mark. just question marks instead of proper characters. I can't even tell if it's in the right direction because it's just q-marks and it's all centered.

perseus project with its greek texts in the web browser, all the greek was just replaced by a few dashes. the english all around it displayed fine.
I wonder how to set the font in the kobo browser. Would be nice if they keep updating the browser, so stg like changing character encoding, as you can in mozilla, would be useful for non-latin alphabet websites.

I wonder though, if this isn't also an issue with what fonts are loaded on the kobo? Maybe if one added fonts that were meant for any of these languages to the fonts folder, would it help?

Metamutator, might I ask, did you add any special font in order to see telugu in its basic form,even without the compound letters displaying properly?

This issue of compound characters is a bit confusing. whether telugu compound digraphs or classical greek normal letters compounded with accents, these are all just glyphs that must be included in a font to display. the machine is not "interpreting the language" and then drawing the compounds, it's just referencing glyphs in whatever font is in use and the font either has them or it doesn't. and some fonts have a subset of glyphs for any language. some common fonts have greek alphabet but not ALL the glyphs for classical greek. Some arabic/hebrew fonts are very limited, containing no vocalized or specially pointed glyphs.

perhaps there is a very involved mechanism for displaying compound characters in telugu, or another similar type of compound alphabet, korean hangul. but it stretches credulity, that so much artificial intelligence would be used when referencing glyphs is the obvious choice. and with arabic and hebrew (and especially chinese), there are almost no compounds, only different forms (vocalization is a kind of compound that could be ignored and left as unvocalized letter or could be displayed as a q-mark, like perhaps classical greek accented letters, and with arabic, it's the letter in its place, with different connective lines depending on position in a word).
I don't know how to add fonts while using the kobo, from an sd card or if it can be done. I only know how to add fonts from a pc when its connected.
maybe adding appropriate glyph-including fonts would solve it/help it.

and it really is just the cutest little device. that e-paper is absolutely like reading a printed page, so easy on the eyes. and it has a very solid yet unassumingly cheap material feel and relaxed yet business like square shape. light as a feather. a real shame if it there's no way to make foreign writing systems work on reflowable formats (epub, txt, html).

[HarryT]

Originally Posted by readingglasses

and it really is just the cutest little device. that e-paper is absolutely like reading a printed page, so easy on the eyes.

For the sake of accuracy, it's not "e-paper" but "eInk". "e-paper" is a very silly marketing term that some companies have started using to describe LCD (not eInk) screens.

[FJames]

Originally Posted by HarryT

For the sake of accuracy, it's not "e-paper" but "eInk". "e-paper" is a very silly marketing term that some companies have started using to describe LCD (not eInk) screens.

Actually e-paper is the more general term, while E-Ink is a brand name. See http://en.wikipedia.org/wiki/Electronic_Paper for some history and alternative approaches, and http://www.eink.com/display_products_pearl.html for brand information.

I recall reading about early versions 20 years ago, and they were called e-paper at that time. I guess it was a solution in search of a problem for a long time!

[cutterjohn42]

@readingglasses
might just be a case of installing a font that includes those characters. Either that or their text rendering engine is only capable of romance language character rendering.

[MetaMutator]

Originally Posted by readingglasses

I wonder though, if this isn't also an issue with what fonts are loaded on the kobo? Maybe if one added fonts that were meant for any of these languages to the fonts folder, would it help?

Metamutator, might I ask, did you add any special font in order to see telugu in its basic form,even without the compound letters displaying properly?

Yes, I had to add a Telugu font to get the letters displayed. Without that, you only see boxes.

Originally Posted by readingglasses

This issue of compound characters is a bit confusing. whether telugu compound digraphs or classical greek normal letters compounded with accents, these are all just glyphs that must be included in a font to display. [...] perhaps there is a very involved mechanism for displaying compound characters in telugu, or another similar type of compound alphabet, korean hangul. but it stretches credulity, that so much artificial intelligence would be used when referencing glyphs is the obvious choice

Actually, Unicode only displays characters, not glyphs. As I said, Indic scripts have conjunct vowels and conjunct consonants; to type "india" in Telugu, for instance, you have to type the letters "i" (I'm transliterating here; the actual character is ఇ) + the vowel "am" ("ం") + 'Da' ("డ") + "i" (" ి") + "ya" ("య") + "aa" (" ా") to form ఇండియా.

That is,
ఇండియా (iMDiyaa) = i ("ఇ") + 'am' ("ం") + 'Da' ("డ") + "i" (" ి") + "ya" ("య") + "aa" (" ా")

Note that the first and the third keys I typed are for the same _letter_, ఇ. As far as the character-set is concerned, it's exactly the same character. As far as rendering goes though, they need to be treated differently, as the vowel ఇ and as the _conjunct_ vowel ి. When you "mix" the consonant డ (pronounced 'Da') with the conjunct vowel ి, you get the _letter_ డి, which is pronounced as 'Di'.

This isn't AI at all; this is how the scripts work. Here's a quick primer on rendering scripts in Wikipedia. Here's a bit more on how Windows does it.

Might stretch your credulity in believing that how you order characters matter in many of the world's scripts, but with all due respects, that's more of a comment about your cultural awareness than technical and cultural realities. Trust me when I say that you need a rendering engine in addition to a font. I've been doing this (this being linguistic computing) since 1998.

and it really is just the cutest little device. that e-paper is absolutely like reading a printed page, so easy on the eyes. and it has a very solid yet unassumingly cheap material feel and relaxed yet business like square shape. light as a feather. a real shame if it there's no way to make foreign writing systems work on reflowable formats (epub, txt, html).

Can't deny this.

[NooN]

oh that was a rich thread really ^^ .. Thanks all for your valuable replies ..
I will read it very carefully in order to understand it.

[carmelra]

Hi NooN,

As far as I know, Arabic ePub files are not readable by the Kobo touch reader. I used ePub with embedded fonts, the characters displayed properly but in the wrong order (left to right instead of right to left).

PDF might be fine.

I know that the Nook Simple Touch reader can be rooted and become an Android (e-ink) reader. As an Android reader, it can be installed with reading applications such as Moon+, OverDrive Media Console and Aldiko (previous versions) that can read RTL languages properly.

Good luck!

[joQ]

I've read Gurmukhi (a Punjabi alphabet) just fine using the Sony prs350. However, this required pdf. The reflow feature combined with landscape mode (when necessary) made this a very useable feature. I noticed that some fonts reflowed just fine, while other fonts, for some reason, did not.

[readingglasses]

Thanks for pointing this out. Good to know !

[Bedys]

I can't believe it but a simple addition of Arabic fonts to Kobo Glo makes it show Arabic script on website pages and most importantly show table of contents in Arabic! I'm stilll working on making it perfect but I am very satisfied with the result so far.

[baggins]

Hi NooN

I've been waiting for Kobo to add RTL support for a long time now. They still don't have it.
You can easily add a font with Arabic characters to the device, however, if you try to display an epub in Arabic, but it will be displayed from left to right instead of from right to left.

PDFs should be fine, but as pointed before me, you won't be able to zoom the characters, and will have to format the PDFs for small paper size, otherwise it would be very uncomfortable to read.

I haven't tried it yet, but html should probably work fine. I know web pages in Arabic are displayed correctly in the browser, so an html file should probably be displayed correctly as well.

[davidfor]

Baggins,

Try putting the ePub on the device with the extension ".kepub.epub". This will force it to be opened in a different reader app. I don't know if it will display differently, but it is worth the try.

[HarryT]

Originally Posted by davidfor

Baggins,

Try putting the ePub on the device with the extension ".kepub.epub". This will force it to be opened in a different reader app. I don't know if it will display differently, but it is worth the try.

The Kobo, like most ePub devices, licenses "Adobe Reader Mobile" software to display ePub books. It's the Adobe software itself that doesn't currently support right-to-left languages.

[baggins]

Originally Posted by HarryT

The Kobo, like most ePub devices, licenses "Adobe Reader Mobile" software to display ePub books. It's the Adobe software itself that doesn't currently support right-to-left languages.

According to DataLogics, the Adobe software does support this:
"RMSDK renders EPUB with support for multiple graphic formats including SVG; embedded fonts; right-to-left and mixed-direction text layout support; right-to-left reading order; Japanese text; and support for stylesheets and templates."

Perhaps Kobo are using an older version.
Regardless, don't they have a different reader for epub and kepub? I'll try renaming the epub file .kepub and see if it makes any difference.

[davidfor]

Originally Posted by HarryT

The Kobo, like most ePub devices, licenses "Adobe Reader Mobile" software to display ePub books. It's the Adobe software itself that doesn't currently support right-to-left languages.

True, but as of firmware 2.0.0, Kobo have a second application for reading "Kobo ePubs" or kepubs. These are basically epubs with a couple of Kobo extensions and a different DRM. They are downloaded from the Kobo server to the device or through a Kobo desktop app. The app is either named "ACCESS" or comes from a company of that name. And I believe this app supports epub3.

You can get a sideloaded epub to be opened by this application by using the double extension ".kepub.epub". I believe this application supports Japanese written vertically. If that is the case, it might support RTL languages. In any case, it won't hurt to try.