How to convert .bgl dictionaries to mobi format?

[mobiuser]

Hi everyone,
I have some .bgl (babylon) dictionaries that I'd like to convert for my kindle pw. I tried to figure out the conversion process, but without any luck and this after I searched around on both forums and net. So it'd appear I'm a total newbie on the matter . I use win 7, by the way. Thanks for any help!

[Doitsu]

Originally Posted by mobiuser

Hi everyone,
I have some .bgl (babylon) dictionaries that I'd like to convert for my kindle pw.

BGL dictionaries are binary files that need to be converted to text files before further processing. You can convert them to tab delimited files or to .gls text files with PyGlossary. However, neither format is compatible with the Mobipocket dictionary input format.

I.e., you'd need to edit the text file in a text editor and wrap terms and definitions in the required MobiPocket dictionary tags, create an .opf file and compile it with KindleGen. (You cannot use Calibre.)

Alternatively, if you somehow managed to re-format the text file so that each line contains just one headword followed by a tab and the definition, you could follow the instructions on this website to generate a Kindle dictionary.

There are also a number of ready-made free bilingual Kindle dictionaries available for download on the Internet. For example, Michael Sheldon's dictionaries.

[mobiuser]

Well, I came across PyGlossary, but I didn't find any detailed tutorial explaining how to use it, so I got stuck. Is there any tutorial for it?

[Doitsu]

Originally Posted by mobiuser

Well, I came across PyGlossary, but I didn't find any detailed tutorial explaining how to use it, so I got stuck. Is there any tutorial for it?

It's fairly easy, you download the PyGlossary archive, unzip it with 7Zip and then double-click pyglossary.bat to start the GUI.
(PyGlossary requires Python. If you don't have Python installed, you can install ActivePython.)

In the PyGlossary GUI, click Browse to select source and target files and formats. Then click Load and Convert.

[mobiuser]

Ok, it works now . Apparently I used the 64-bit version of the program (Python/ActivePython). I'll do some tests tomorrow.

[mobiuser]

Uhmm, the conversion process doesn't seem to consider the html tags (color, font style...), there are specific symbols like this "●" that turn in this "9679", some weird symbols for the accented characters and other small things. Is there anything I can do to improve the result? I think I can resolve the symbols and the accented characters problems with a "search and replace" in notepad++, but for the html code I don't think I can do anything.

[Doitsu]

Originally Posted by mobiuser

Uhmm, the conversion process doesn't seem to consider the html tags (color, font style...)

IIRC, Babylon uses a couple of custom HTML tags that you need to delete, but all other HTML code (fort example <span>'s or HTML entities) can be left in the source files since the Mobipocket format was originally based on HTML 3.2.
(The latest Kindle format also partially supports HTML5, but not for dictionaries.)

Originally Posted by mobiuser

Is there anything I can do to improve the result?

I'm afraid not, unless you're familiar with a scripting language and/or regular expressions.

Originally Posted by mobiuser

I think I can resolve the symbols and the accented characters problems with a "search and replace" in notepad++, but for the html code I don't think I can do anything.

As I've already mentioned, valid HTML 3.2 markup doesn't need to be deleted.

IMHO, the whole project probably isn't worth the effort, because Amazon has lately added some very reasonably-priced Kindle dictionaries by traditional dictionary publishers with a good coverage.
Also, even if you manage to completely reformat the input file, the dictionary won't contain inflections and will be of limited use, unless the dictionary language isn't heavily inflected or you're an advanced user, who can automatically identify inflected word forms.

[mobiuser]

Ok... there is no hope for me then . I'll try some other bgl dictionary, though. I'm curious to see the results . Thanks!