qt: Some translations have leading / trailing whitespace characters #16814

I know translations are managed on Transifex, but I noticed an oddity: some translations contain untrimmed whitespace.

• https://github.com/bitcoin/bitcoin/blob/5e202382a987351a59d5cda98ea9f2aee99f61a5/src/qt/locale/bitcoin_ar.ts#L2729
• https://github.com/bitcoin/bitcoin/blob/5e202382a987351a59d5cda98ea9f2aee99f61a5/src/qt/locale/bitcoin_ro_RO.ts#L2315

Do we need a linter / script to handle these cases?

to handle these cases?

Handle how?

to handle these cases?

Handle how?

By trimming them. It could happen in Transifex, or when copied to the repo.

Not sure if this is something that needs to be fixed or not. @laanwj is probably the right person to ask. In either case, any sort of trimming would probably be added to the update-translations.py script.

Honestly I'd leave it alone unless it causes actual problems for users. Resist the temptation to fix what isn't broken and muck around with spaces.

The script could look for spaces in first/last characters and remove them, I think that would be a good idea, and it's also not complex to implement. Willing to do work if it'll get implemented - but I also have a PR open for the script already which fixes common errors with format specifiers.

Closing this, I don't think this is really an issue.

I think this should be fixed, but on transifex via https://docs.transifex.com/setup/setting-translation-checks

FWIW whitespaces were manually fixed on transifex by me (for a whopping 8kB of disk space gained) but I agree with Marco that we should set up translation checks on transifex.

Note that some source strings have trailing whitespace, so removing it will cause alignment issues for the translations.

@MarcoFalke I found exactly one occurence in bitcoin_en.ts, and I'm thinking it's a typo. Alignments are done by qt, I believe?

I found more than 50 instances:

git grep ' </source>'|wc -l
54

I checked again and there are 3 trailing whitespaces in bitcoin_en.ts in branch 0.19, as opposed to 1 trailing space in master. Your number 54 corresponds to all trailing spaces from all translation files in src/qt/locale in master.

Strings in question:

src/qt/locale/bitcoin_en.ts:        <source>Your wallet is about to be encrypted. </source>
src/qt/locale/bitcoin_en.ts:        <source>Your wallet is now encrypted. </source>
src/qt/locale/bitcoin_en.ts:        <source>Wallet: </source>

I see absolutely no reason to add a trailing whitespace to the first 2 strings, and as for wallet - you'll notice there are actually 2 strings being translated: wallet and wallet_. So again, I'm guessing they're all typos.

Trailing whitespaces should never be translated and we should not rely on them for alignment - this could create multiple problems and is an overall bad idea, and I doubt anyone had this idea in mind when appending those 3 whitespaces. Also, translators have no way of seeing trailing whitespaces on transifex (apart from selecting the source string with the mouse manually), which means noone was ever "translating" them.

They are not typos, but oversights of putting trailing whitespace used for formatting into the source message:

$ git grep 'Your wallet is about to be encrypted. "'
src/qt/askpassphrasedialog.cpp:                                         tr("Your wallet is about to be encrypted. ") + encryption_reminder +

There are two fixes needed here:

• A transifex check to enforce the trailing whitespace is preserved
• (optional) remove the trailing whitespace from the translation source message