Pocketbook dictionary format revisted

[Markismus]

Originally Posted by stef

Is there a way to keep backup an installed official dictionary?
I can't found it inside /system.

I have the pocketbook era 700 and the dictionary is the "English - Greek".

Thank you very much

Hi Stef,
If I remember correctly, the Pocketbook system allows for multiple dictionaries for the same language. So you don't have to replace an official one as we had to do with Kobo dictionaries in the past (---or maybe even the present. Don't work much with Kobo Nickel. It's start menu and Koreader for me.)
So if you don't overwrite a file with an identical name by copying a new dictionary, you're good.

[rkomar]

Originally Posted by stef

Is there a way to keep backup an installed official dictionary?
I can't found it inside /system.

I have the pocketbook era 700 and the dictionary is the "English - Greek".

Thank you very much

Official dictionaries are stored in a separate filesystem and mounted under "/mnt/secure". The default user account does not have access to that filesystem. You would need to be root or user "sreader" to be able to read and backup files in there. I believe that PocketBook license the dictionaries installed there from others, and it is likely to be part of the agreement that they do not allow the files to be copied and used elsewhere. So, they are hidden as much as possible.

[stef]

Originally Posted by Markismus

Hi Stef,
If I remember correctly, the Pocketbook system allows for multiple dictionaries for the same language. So you don't have to replace an official one as we had to do with Kobo dictionaries in the past (---or maybe even the present. Don't work much with Kobo Nickel. It's start menu and Koreader for me.)
So if you don't overwrite a file with an identical name by copying a new dictionary, you're good.

Thank you for your answer.
I know, I have multiple dictionaries, BUT this is the first time that there is an official dictionary for en-el for my era. It has double or triple the amount of words than the custom dic (babylon) and I want to have it as a backup.

I'm using koreader too but a bigger dictionary is the way to read and learn better English.
I found links to download the official kobo dictionaries, but there is not for my language. I was hoping that there is something like this for pocketbook.
Maybe it's not possible

Thank you anyway.

[stef]

Originally Posted by rkomar

Official dictionaries are stored in a separate filesystem and mounted under "/mnt/secure". The default user account does not have access to that filesystem. You would need to be root or user "sreader" to be able to read and backup files in there.

Oh, I don't have the knowledge to do all these

I believe that PocketBook license the dictionaries installed there from others, and it is likely to be part of the agreement that they do not allow the files to be copied and used elsewhere. So, they are hidden as much as possible.

The dictionary appeared out of the blue a few weeks ago. My country is not a big "market" and I am afraid that maybe it disappears with the same way. I don't want to use it somewhere else. I have it, I want it just for backup.



Thank you for your time

[Moonbase59]

Looks much nicer!

Originally Posted by Markismus

However, your version seems quite nice. Nicer, in fact. Care to share? Maybe via PM?

I made a conversion with pyglossary for mine in the meantime. Not sure how it’ll display on the small KOReader window, but most of the time I work on the PC anyway and use GoldenDict-ng. Btw, check your PM. (No attachments or unlawful stuff included.)

[Markismus]

@Moonbase59 I've ran the script for your Premium EN-DE version and put in on pCloud.. Could you check whether the dic-file works?

[Moonbase59]

Originally Posted by Markismus

Could you check whether the dic-file works?

For others (and my future self): Copied the .dic file to system/dictionaries on the PB700.

Here are some notes and screenshots:

• It appears with its name under "OTHER" in the dictionaries list. Maybe it needs the supported language pair set.
• When selecting it, the PB prompts "Do you want to download Morpholgy for OTHER?" (no .aff file, I’m supposing) I wonder what the naming conventions are, because I have both EN and DE morphologies already installed.
• Display looks a bit "glued together", but it is readable and I guess OK for the native reader app, because of the little screen estate we get.
• The info is empty. Maybe language and description can be set somewhere.
• The speaker button (read ord with TTS) is active and changes when clicking, but no sound to be heard. (PB dictionaries usually play the word using TTS)

So I’d say it works. You could probably put both EN-DE and DE-EN up, for others to benefit, what do you think?

N.B.: The short story I’m currently reading didn’t have the word "consciousness", so I manually typed it in (although "youngest" is selected).

P.S.: Looks to me the PB uses the rather standard Linux myspell/hunspell dictionary format (.dic & .aff morphology files), could that be true?

EDIT: Disregard the P.S. Just found that they use a proprietary format, apparently holding the dict and the index in one .dic file. Plus, you’d need a Windows "coverter.exe" transforming .xdxf dictionaries into their format. Morphologies I’m not sure about. Oh well, KOReader is much better with dictionaries anyway…

[Markismus]

The scripts conversion workflow for KoreaderOptimized is stardict binary->stardict xml->xdxf->reconstructed xdxf->reconstructed xml->KoreaderOptimized stardict binary.
The scripts conversion workflow for Pocketbook binary is stardict binary->stardict xml->xdxf->reconstructed xdxf->Pocketbook reconstructed binary.

Looking at the code for consciousness the starting xml is

<article>
<key>consciousness</key>

<definition type="h"><definition type="h">
<![CDATA[<p style="padding-left:0.3em;margin:0"><font color="navy"><b>con·scious·ness</b></font></p><p style="padding-left:1em;margin:0"><font color="teal">[ˈkɒn(t)ʃəsnəs</font>, <font color="indigo">AM</font> <font color="teal">ˈkɑ:n-]</font></p><p style="padding-left:1em;margin:0"><font color="fuchsia"><i>n</font> <font color="darkgray">no pl</i></font></p><p style="padding-left:1em;margin:0"><font color="firebrick"><b>1.</b></font> <font color="indigo">MED</font> <font color="mediumseagreen"><i>(sentience)</i></font> Bewusstsein <font color="darkgray"><i>nt</i></font></p><p style="padding-left:1em;margin:0"><font color="darkblue"><b>to lose ~</b></font> das Bewusstsein verlieren, bewusstlos werden</p><p style="padding-left:1em;margin:0"><font color="darkblue"><b>to recover</b> [<i>or</i> <b>regain</b>] <b>~</b></font> das Bewusstsein wiedererlangen, wieder zu Bewusstsein kommen</p><p style="padding-left:1em;margin:0"><font color="firebrick"><b>2.</b></font> <font color="mediumseagreen"><i>(perception)</i></font></p><p style="padding-left:1em;margin:0"><font color="cyan">▪</font> <font color="darkblue"><b>sb's ~ of sth</b></font> jds Wissen um etw <i>akk</i> </p><p style="padding-left:1em;margin:0"><font color="darkblue"><b>to enter into</b> [<i>or</i> <b>impinge on</b>] <b>sb's ~</b></font> in jds Bewusstsein eindringen, jds Bewusstsein erreichen</p><p style="padding-left:1em;margin:0"><font color="cyan">▪</font> <font color="darkblue"><b>sb's ~ that ...</b></font> das Bewusstsein, dass ...</p><p style="padding-left:1em;margin:0"><font color="firebrick"><b>3.</b></font> <font color="mediumseagreen"><i>(awareness) political, social</i></font> Bewusstsein <font color="darkgray"><i>nt</i></font></p><p style="padding-left:1em;margin:0"><font color="darkblue"><b>health ~</b></font> Bewusstsein <font color="darkgray"><i>nt</i></font> für die eigene Gesundheit</p><p style="padding-left:1em;margin:0"><font color="darkblue"><b>to raise sb's political ~</b></font> jds politisches Bewusstsein erweitern</p>]]>
</definition>
</article>

For the reconstructed xml it becomes

<article>
<key>consciousness</key>

<definition type="h">
<![CDATA[<p style="padding-left:0.3em;margin:0"><span style="color:#000080"><b>con·scious·ness</b></span></p><p style="padding-left:1em;margin:0"><span style="color:#008080">[ˈkɒn(t)ʃəsnəs</span>, <span style="color:#4B0082">AM</span> <span style="color:#008080">ˈkɑ:n-]</span></p><p style="padding-left:1em;margin:0"><span style="color:#FF00FF"><i>n</span> <span style="color:#A9A9A9">no pl</i></span></p><p style="padding-left:1em;margin:0"><span style="color:#B22222"><b>1.</b></span> <span style="color:#4B0082">MED</span> <span style="color:#3CB371"><i>(sentience)</i></span> Bewusstsein <span style="color:#A9A9A9"><i>nt</i></span></p><p style="padding-left:1em;margin:0"><span style="color:#00008B"><b>to lose ~</b></span> das Bewusstsein verlieren, bewusstlos werden</p><p style="padding-left:1em;margin:0"><span style="color:#00008B"><b>to recover</b> [<i>or</i> <b>regain</b>] <b>~</b></span> das Bewusstsein wiedererlangen, wieder zu Bewusstsein kommen</p><p style="padding-left:1em;margin:0"><span style="color:#B22222"><b>2.</b></span> <span style="color:#3CB371"><i>(perception)</i></span></p><p style="padding-left:1em;margin:0"><span style="color:#00FFFF">▪</span> <span style="color:#00008B"><b>sb&apos;s ~ of sth</b></span> jds Wissen um etw <i>akk</i> </p><p style="padding-left:1em;margin:0"><span style="color:#00008B"><b>to enter into</b> [<i>or</i> <b>impinge on</b>] <b>sb&apos;s ~</b></span> in jds Bewusstsein eindringen, jds Bewusstsein erreichen</p><p style="padding-left:1em;margin:0"><span style="color:#00FFFF">▪</span> <span style="color:#00008B"><b>sb&apos;s ~ that ...</b></span> das Bewusstsein, dass ...</p><p style="padding-left:1em;margin:0"><span style="color:#B22222"><b>3.</b></span> <span style="color:#3CB371"><i>(awareness) political, social</i></span> Bewusstsein <span style="color:#A9A9A9"><i>nt</i></span></p><p style="padding-left:1em;margin:0"><span style="color:#00008B"><b>health ~</b></span> Bewusstsein <span style="color:#A9A9A9"><i>nt</i></span> für die eigene Gesundheit</p><p style="padding-left:1em;margin:0"><span style="color:#00008B"><b>to raise sb&apos;s political ~</b></span> jds politisches Bewusstsein erweitern</p>]]>
</definition>
</article>

So the colors are replaced by their hexadecimal values and placed in span-blocks instead of font-blocks. Also some symbols are escaped, e.g. ' becomes '. This gives exactly the same results in Goldendict as the source. In Koreader, we regain colors the colors that are normally not shown, though. That means that the intermediary xdxf_reconstructed has that info, too.

The reconstructed xdxf file for Pocketbook looks different though

<ar>
<head><k>consciousness</k></head><def><b>con·scious·ness</b>[ˈkɒn(t)ʃəsnəs, AM ˈkɑ:n-]<i>n no pl</i><b>1.</b> MED <i>(sentience)</i> Bewusstsein <i>nt</i><b>to lose ~</b> das Bewusstsein verlieren, bewusstlos werden<b>to recover</b> [<i>or</i> <b>regain</b>] <b>~</b> das Bewusstsein wiedererlangen, wieder zu Bewusstsein kommen<b>2.</b> <i>(perception)</i>▪ <b>sb's ~ of sth</b> jds Wissen um etw <i>akk</i> <b>to enter into</b> [<i>or</i> <b>impinge on</b>] <b>sb's ~</b> in jds Bewusstsein eindringen, jds Bewusstsein erreichen▪ <b>sb's ~ that ...</b> das Bewusstsein, dass ...<b>3.</b> <i>(awareness) political, social</i> Bewusstsein <i>nt</i><b>health ~</b> Bewusstsein <i>nt</i> für die eigene Gesundheit<b>to raise sb's political ~</b> jds politisches Bewusstsein erweitern</def>
</ar>

All p- and span-blocks and colors are removed. Reason for that is that @ezdiy posted that only these tags are retained when converting to pocketbook binary:

    # Str = "?xml";
    # v23 = "xdxf";
    # v24 = "full_name";
    # v25 = "description";
    # v26 = "ar"; // this one for each definition entry
    # v27 = "k";
    # v29 = "i";
    # v28 = "b" // maybe this is for <br> too, due shared prefix?

So I'll add the span tag to be retained in the script, so you can test whether it's different, but I don't think it will work. Another thing to try is add <br/> tags after every span-block or p-block, before removing them. This is not what's normally wanted with those small screens, though.

I've uploaded 3 test-files in the ENG-DEU directory:

1. Retaining span-blocks
2. Replacing Span-blocks with a break
3. Replacing p-blocks with a break

If someone would be so kind to test and post the results? Maybe we can upgrade the layout of the Pocketbook dictionaries.

[Moonbase59]

Originally Posted by Markismus

This is not what's normally wanted with those small screens, though.

Yep, true. Scrolling in that small area sucks, too, and E-Ink displays will usually refresh.

[stef]

@Markismus

If I have a .txt file with the right format to produce, with pyglossary, a working (ugly but better than nothing) koreader dictionary, is there a way to produce a .dic dictionary for pocketbook?
Some settings in pyglossary, a few commands (I use linux), some other way?

Thank you for your time.

[Markismus]

The first post in this thread explains it, @Stef. For further reading there are links to the GitHub pages.

You realize that all dictionaries are either txt-files or binary files. So you will have to figure out, what format is used within the text file. E.g. it could be a comma separated value, a CSV-file. Or it could follow a meta language syntax. E.g. XML, HTML or XDXF.

[stef]

@Markismus Thank you.

I RTFM all these many times, I 'm using your dics for years, but I was looking for a "linux-way".
It's ok, maybe I'll try wine if I don't find another solution.

[Markismus]

The perl script explicitly uses wine for the convert-executable. We have found no alternative for the conversion. Still, wine is quite Linux, though.

[Moonbase59]

@Markismus:

Could only test one dic after another. If you copy all three to the device, they appear under one name only, and you don’t know which it has taken.

They also all appear under »Other« (»Andere« in German), instead of under »English → German«.

I had to compress all screenshots into one ZIP (upload limits here), and also include a screenshot from the original Pocketbook English → German dictionary.

On the Pocketbook, you have an app »Dictionary« as well as the (small) lookup area when reading a book. I enclose screenshots for both. For some reason, they all look alike.

PONS Premium EN-DE_reconstructed_replacingEndTagParagraphWithBreak .dic

Full: [ICODE]scr0022.bmp[/ICODE]Book: [ICODE]scr0023.bmp[/ICODE]

PONS Premium EN-DE_reconstructed_replacingEndTagSpanWithBreak.dic

Full: [ICODE]scr0024.bmp[/ICODE]Book: [ICODE]scr0025.bmp[/ICODE]

PONS Premium EN-DE_reconstructed_retainingspan.dic

Full: [ICODE]scr0026.bmp[/ICODE]Book: [ICODE]scr0027.bmp[/ICODE]

Pocketbook English → German

Full: [ICODE]scr0028.bmp[/ICODE]

[Markismus]

@Moonbase59 Well, at least we can now be certain the <br/> tag is stripped by convert.exe.

If you find other converted dictionaries in binary dic-format, that do show line breaks, could you sent a screenshot? Maybe we work back from that code to implement a linebreak that survives conversion.

It would be nice to create a tool to do the conversion ourselves. However, I have no idea how to reverse engineer a binary format.