How should I sort on write?

[anezih]

When verifying the integrity of a StarDict dictionary via stardict-editor it gives the warning of 'Wrong word order'.
Error occurs when using the glos.write() method as glos_syn.write(f"{outname}", format="Stardict") but it also happens when it is used as

named = namedSortKeyList[3] # -> name="stardict"
glos_syn._data.setSortKey(namedSortKey=named, sortEncoding="utf-8", writeOptions={})
glos_syn.write(f"{outname}", format="Stardict", dictzip=False)

I don't know whether it is the proper way of forcing sort on write but the second method works when it is used here. (Here, the headwords are mostly ascii characters with Turkish characters put at the back of the idx file but that is not a problem as long as stardict-editor is concerned.)

However, it does not work when sorting is applied on Japanese characters. See the attached screenshot.

Here is the script that I'm having the problem with.
I'm also attaching the source file for the aforementioned Japanese-Turkish dictionary.
vaneralper_jp_tr.zip

[ilius]

Does it give error when you normally convery using command line?

Can you use glos.convert instead of read and write?

If you are using glos.write, I would assume you are adding entries yourself (rathering reading a supported format), so you should sort them properly before adding to glossary. You should avoid accessing glos._data, because that's internal to pyglossary.

[ilius]

I honestly didn't think of this use case (reading from one glossary object into another one while sorting) when designing glossary_v2.

I think you have 3 choices:

• Go back to using the old pyglossary.glossary module and use glos.read() and glos.sortWords().
• Convert your original glossary to a temp .txt file with glos.convert() then convert than to StarDict with glos.convert().
• Use a code like this PR, which I have not tested BTW (might contain a typo).

The second one is probably easier.
But it's up to you.