Pocketbook dictionary format revisted

[filipwaw]

Originally Posted by Markismus

There is a mirror of the library made by nhedgehog. The link is in this thread.
PCloud reaches it monthly limit quite often. I reckon it's because people download the entire library instead of just the ones they will actually use.

my bad, I didn't see it. Perfect, thanks a lot!

[Markismus]

Running system command:"WINEDEBUG=-all wine converter.exe "/home/mark/Downloads/PocketbookDic/dict/Le Grand Robert de la langue française (Dictionnaires Le Robert) (Z-Library)/mobi7/Le Grand Robert de la langue française (2005)_reconstructed.xdxf" fr"
Loading collates...
Loading morphems...
Loading keyboard...
Loading dictionary file...
83568 words loaded
Sorting dictionary...
Searching for equal words...
Packing dictionary...
Creating index...
Writing output file...
OK

Total words: 83568
Total containers: 11825 (55..18415 bytes)
Words in container: 1..41
Dictionary file size: 45889390 bytes

@BHUpasamo The only thing that I can change on my end is the language files used by converter.exe. So there are now two dic-files: One made with the fr-language directory and one made with the fre-directory. Success!

[EastEriq]

Originally Posted by Markismus

Could someone else test the Le Grande Robert dic-file?

They are both nonfunctional for me in the dictionary application. With the .fr I can get a keyboard and can type, but get blank screens as search results. With the .fre, I only see some oddity instead of the keyboard, can't input characters and can't search; the only interaction I allowed with the app is about changing dictionary. Attached a screenshot of the latter, and for comparison the Littré which is ok.

In KOreader, the .dict looks fine.

[Markismus]

This is a new kind of error. Wonder how it can be tackled.

Things I can think of is

• to strip all unnecessary tags out of the dictionary (basically doing converter.exe work for it.)
• to chop the dictionary in parts and hope that converter.exe doesn't choke on a smaller dictionary.

[Markismus]

I've uploaded Le Grand Robert 2005 to pCloud. The version of the dic-file of April 19th2025 has all tags stripped, that are not needed later in the script and are stripped by converter.exe as posted by @ezdiy in post #15 in this thread.

i've also made a pCloud transfer link in case the traffic is past its monthly limit.

So the stripped binary-file is a tad smaller than than unstripped one. Apparently, converter.exe retains a bit more than stripping with the script does.

Looking at La Grande Larousse, it's not quite enough if the dictionary is just too big. La Grande Larousse[/i] is 24MB versus 44MB for Le Grande Robert. So splitting it in 2 parts could do the trick.

[Markismus]

I've uploaded Le Grande Robert (2005) to pCloud in 2 halves.

I've also created a pCloud transfer link in case the traffic is past its monthly limit.

@EastEriq and @BhUpasamo would you be so kind to test these versions?

[BhUpasamo]

I've just tried your last three uploads, @Markismus, and I'm afraid it's still not working. Interestingly, the two halves show as only one dictionary in my ereader.

Other than that, I added the .dict to my GoldenDict yesterday, and it's a thing of beauty! I spent several hours reading through it and having great fun.

By the way, not that it matters very much, but it's Grand Robert and Grand Larousse without an e

[Markismus]

Ah, yes, I found it an oddity. Didn't think it through, though.

I did give both dictionaries the same start, though, including the title-block. So that's why the can be distinguished. Maybe, just install one of the half dic-file and see whether that works?

Size was the last impediment, I could think of. So I am out of ideas. Based on teh format and content there is no reason it should not work. The problem is with the dictionary program, not with the formatting or the conversion.

[BhUpasamo]

Originally Posted by Markismus

Maybe, just install one of the half dic-file and see whether that works?

Nope, it doesn't. Well fought anyway!

[EastEriq]

Originally Posted by Markismus

I've uploaded Le Grande Robert (2005) to pCloud in 2 halves.

@EastEriq and @BhUpasamo would you be so kind to test these versions?

Ditto for me, no juice. I've tried with the big one alone installed,with the three of them together, with both halves and with the first half alone on the SD.
I wouldn't rush to blame the size. I see that I have on the SD the "old" Littré, littre_pocketbook.dic (probably the one coming from ebookgratuits), which is 33.7MB and works.

[Markismus]

@BHUpasamo and @EastEriq: There are 2 new test dictionaries. Spaces surrounded the keywords in the script for manual cut of the dictionairy. Maybe that was the culprit.

[EastEriq]

I've tried the first one (i.e. the not strippedtags), and it appears to work perfectly in the dict app and in the reader. Great. Any reason I should look also at the other one?

[Markismus]

@EastEriq Only to see whether the stripping of tags introduces errors or garbles the lay-out.
Very happy to have found the error, though!

[EastEriq]

Also the stripped works for me. The unstripped renders a bit more spaced, see the shots (the first one is strippedtags); I can't say whether this is because the conversion added spaces, or because of missing font support. I note the annoying [] in this and other dictionaries; I never bothered to search whether with my old firmware the font used by the dictionary app is configurable somewhere, so to choose one with additional dingbats, whatever they could be.

[Markismus]

In this case the not displayed characters are UTF-8 characters, but they are availiable as ASCII characters.

❖: U+2756 : BLACK DIAMOND MINUS WHITE X
♦: U+2666 : BLACK DIAMOND SUIT -> ASCII value: 9830

But you're right that they're in the dingbats block.

So I am stipping too much. Seems like an identation difference. Could be that blockquotes are not stripped.