Pocketbook dictionary

[Markismus]

Yes, those are indeed the wrong ones.

Sorry, my kid and me were having a sick day. I'll look into it later.

[tropoy]

Originally Posted by EastEriq

... but it seems they are all still there. Are you sure you uploaded your last conversion?

I checked and the first conversion and the second one seems to be the same, indeed. Same date and size...

[tropoy]

Originally Posted by Markismus

Yes, those are indeed the wrong ones.

Sorry, my kid and me were having a sick day. I'll look into it later.

Ok, thanks, take care of yourselves.

[Markismus]

Allright, I just wrote the conversion from xdxf-format to Stardict text-format. Now I can generate the Stardict files without having to use Linguae. (I returned my pocketbook Inkpad 3 Pro, so I am using koreader on my Kobo Aura H2O again.)

Here is the link to the dictionaries. All other links now point to them, too.


Tested in koreader:

[EastEriq]

Uhm, appreciate, but still no joy. The stock Dictionary.app displays the codes literally, inside pbreader they are simply dropped. In CR3 I can't open the dictionary anymore (not necessarily because of this file).

[Markismus]

Seems that your dictionary app can't handle html entities at all.

The last thing you could do is convert the numerical symbols to characters and expect modern day apps to be able to handle unicode directly.
When I posted the DOCTYPE within a code block, I found that some codes were automatically interpreted. So you'll want a low level parser, that only interprets the &#xXXXX; parts.

I had expected the lookup in reader to be working, though? Should expect to be better even because the <f>-tags are retained now. But apparently it can handle entity names, but not numbers?

[EastEriq]

Originally Posted by Markismus

Seems that your dictionary app can't handle html entities at all.

Seems so. That's why I wrote "replace with unicode" Anyway, not tonight... Now what buggers me is why CR3 can't see dictionaries anymore (which most likely is unrelated to your file, now removed from PB)

[tropoy]

Originally Posted by EastEriq

Uhm, appreciate, but still no joy. The stock Dictionary.app displays the codes literally, inside pbreader they are simply dropped.

On my Inkpad 3 too, of course...

Thanks you both a lot anyway.

[Markismus]

So the unicode conversion is a tough one in Perl. Learned a lot...

For your testing purposes: Here are your dictionary files.

I tested it in koreader:


I also redirected _all_ previous links to files.

[EastEriq]

Try this.

How done:

1. Took Markismus last xdxf
2. substituted xml entities with unicode characters with this script:
   

   cat Nouveau\ Littre\ 2011\ \(from\ Bookeen\ by\ Penelope\)_reconstructed.xdxf |\
     sed -e "s/ / /g" \
         -e "s/'/'/g" \
         -e "s/<[/]*f>//g" \
         | perl -CS -pe 's/&#x([\dA-Fa-f]{3,4});/chr(hex($1))/eg' \
         | perl -CS -pe 's/&#(\d{3,4});/chr($1)/eg' \
   > N1.xdxf

3. converted using DictionaryConverter-neu 171109, augmented with the locales found in this post (rather than using the converter Markismus used - for no reason, just ran into the others first).

[EastEriq]

ETA: crossed mid-air....

[Markismus]

You just missed me! Sorry, didn't realize you were also tinkering.
How do the results compare?

Why did you remove the <f> tags?

BTW thanks for the link to more locales!

[EastEriq]

Originally Posted by Markismus

How do the results compare?

Uh, I don't know, byte-checking? At first sight on a few entries (on my one) the content seems about ok, I wouldn't go for an extensive check beyond that.

The appearance in the three (I manage to resurrect CR3 in the meantime) dictionary viewers I have varies a bit, each one has its formatting quirks, like adding spacing, using a font which misses some of the graphisms. None of the three dict viewers has really an impressing usability by itself, I once was aware of some configuration files to fiddle a bit with fonts in the one or the other, but that's all it is. This one .dic seems as good as others now.

[EastEriq]

Originally Posted by Markismus

Why did you remove the <f> tags?

because they were showing up in the dictionary app. But actually that maybe was because they were always tagging nonrecognised entities? Will try yours, if you kept them...

[Markismus]

Comparing the two, I think I like the Stardict with sametypesetting=m and unicoded text (post #294) better than the one with sametypesetting=h and html-handling of the text (post #289).

Though that could just mean that we should do some work on Koreader's html-handling.