How to convert .mobi dictionary to stardict?

[AnimalOfArt]

@Markismus

I know KindleUnpack is not your code, but maybe you can help me:

I tried to unpack a mobi dictionary I converted from K4PC azw into mobi.

It always tells me "Unpacking failed"

Do you probably know what the problem is?

EDIT: Okay, nevermind. I used the Calibre plugin now and it worked flawlessly Now I'm going to setup (Strawberry) Perl and try your converter

[Markismus]

Kindle unpack generates a lot of data. Most interesting are the resulting xml-files. One is raw and the other is interpreted. If the interpretation of the raw xml fails, you get this error from Kindle Unpack.
Unfortunately, my code starts from the interpreted xml. Wish I had done that differently!

Also, the reason why you find so few mobi dictionary converters, is that the format is very loose. So whether any converter works, depends on if it has seen such xml-code before. it is hit and miss for new dictionaries.
If you find a workaround that is able to fix the format of a Kindle or mobi-dictionary into one fixed order, I am willing to write an extra importer for it. However, I haven’t got the time to write a generic importer based on heuristics.

[AnimalOfArt]

For now, I converted the mobi into book.html and I want to convert it into xdxf for stardict conversion.

When I set it up i get this strange output

Local path is C:/Users/Animal/Downloads/PocketbookDic/mobi7. (This is where the book.html is)
Full path is C:/Users/Animal/Downloads/PocketbookDic/C:/Users/Animal/Downloads/PocketbookDic/mobi7

and then if fails with this message

Can't find C:/Users/Animal/Downloads/PocketbookDic/C:/Users/Animal/Downloads/PocketbookDic/mobi7/Images/image15902.gif. Quitting.
Died at pocketbookdic.pl line 738.

Why does the Full path is so strange and what do I need to that to make it right?


EDIT: Ah nevermind. The command path parameter in the command line should also be relative

But now I get this error:

Unzulässiger Parameter - /Users
Cannot open C:/Users/Animal/Downloads/PocketbookDic/mobi7/Images/image15902.gif.png: No such file or directory
Died at pocketbookdic.pl line 1000.

The script attaches an .png extension to an existing gif file, image15902.gif. And this causes the script to end. Should I add png to all the gif files now?

[Markismus]

No. I would rather toggle the conversion of gif-files off. Search for the toggle governing it, please.
I is dependent on imagemagick, which only works on Linux.

[AnimalOfArt]

Ok that problemd is out of the way. It seemed to work until this message came up:

Cannot open C:/Users/Animal/Downloads/PocketbookDic/mobi7/: No such file or directory
Died at pocketbookdic.pl line 1000.

But I can tell your for sure, this directory exists

[Markismus]

https://github.com/Markismus/PocketB...kdic.pl#L1000:

open( FILE, "$FileName" )
   || (warn "Cannot open $FileName: $!\n" and die);
  }

It's part of the basic subroutine to open file. Open filehandle or warn. So it can't find the file.
Oddly enough, the message should have generated the whole warning, not just the '$!' part.

If this is an systemic error, than no files can be loaded anywhere in Windows, which seems odd: This is a very basic operation. Could you have given the wrong filename?

Maybe use https://www.cygwin.com/ to see whether it is Linux/Windows related or your input?

[AnimalOfArt]

Before I try it with cygwin:

What is $FileName supposed to be?

[Markismus]

It’s supposed to be the filename in the current path or an relative path/filename or an absolute path/filename.
It’s what is given to the subroutine as its first argument:
https://github.com/Markismus/PocketB...ookdic.pl#L989

[AnimalOfArt]

I use this command to start the script

perl pocketbookdic.pl mobi7/book.html de

My base dir is set to "C:/Users/Animal/Downloads/PocketbookDic"

The script seems to find the book.html inside C:/Users/Animal/Downloads/PocketbookDic/mobi7 because it creates a copy of it named as test_html_indexentries.html inside the base dir.

There is also a $FileName variable in the config on line 31. Is this for the output dictionary file and is this related to $FileName at line 989?

[Markismus]

The filename at the beginning of the script is used when none is given at the command line.

The test is just a test to see whether the script works. You can toggle its creation off in the start section.

I think that your workaround KindleUnpack isn’t working.
So you provide a filename. The script uses it to run KindleUnpack and tries to open its result. However, since KindleUnpack fails the file is not created and as the script tries to open it and can’t, it generates a warning.

[AnimalOfArt]

I have no KindleUnpack workaround in the script.
I put the mobi file into Calibre and then KindleUnpack it with it.
I do not set KindleUnpack in the Script. It is still your directory and because of that I get this

Can't find 'kindleunpack.py' in /home/mark/git/KindleUnpack/lib. Cannot handle mobi dictionaries.

Also I get this message:

Converting indexentries from HTML to XDXF. This will take some time. 20200514 08:37:21

So KindleUnpack is just for extracting mobi into book.html and images. As I have done that already and directly pointed the script to the book.html, KindleUnpack should be out of the way.



EDIT: Ok, I tried it with cygwin, but that is worse.

I'm getting this error

Can't find home/Animal/PocketbookDic/dict/mobi7/Images/image15902.gif. Quitting.
Died at pocketbookdic.pl line 738.

Although this path and file exists.

[Markismus]

Why don’t you send me a link for the dictionary? Maybe, using Dropbox? I’ll try and convert it. If it’s an error in the code, I’ll fix it. If it’s an unknown format of mobi dictionary, I’ll add warnings that make it explicit.
Who knows, maybe we’ll get lucky and conversion on Linux goes without a hitch.

[AnimalOfArt]

Thank you, I send you a PM

[Markismus]

The Duden 7. Auflage was already converted and uploaded here.

The PONS EN-DE dictionary is from 2016 Kindle ( based on wordlists from 2011) converted to mobi? I have the original PONS EN-DE mobi dictionaries from 2007.

Looking at your conversion, it has a lot of missing glyphs: Calibre isn't the way to convert it! It also garbled the code, resulting in non-conform XML.

Although, it's interesting to check why the code didn't filter that out, it's also a surefire way to get a crappy Stardict dictionary. Could you send me a link to the Kindle version?

[AnimalOfArt]

I guess you won't be able to open the original azw file because they have DRM bound to my account. This is the purest mobi file I can give you. I took the azw from my kindle content folder, put it into calibre, and it immediately DeDRMed and converted it to mobi.

However, I can send you the amazon link to the dictionary, if it's that what you have asked for:

https://www.amazon.de/W%C3%B6rterbuc...dp/B01NA9LE9T/


What was the source of the Duden you converted? Was it from Kindle?