Updated Brazilian .po files

[MichaelChirico]

Thanks in advance!

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 98.64%. Comparing base (ca2c377) to head (ecf6879).
Report is 1 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #6776   +/-   ##
=======================================
  Coverage   98.64%   98.64%           
=======================================
  Files          79       79           
  Lines       14642    14642           
=======================================
  Hits        14444    14444           
  Misses        198      198           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[rffontenelle]

Hi @MichaelChirico! Can you please remind me how can I update the source strings of a PO file and get the number of chars per line wrapped as is in the data.table repository? The translation editor software re-wrapped so the git diff looks polluted.

[MichaelChirico]

I update the .po files with potools::po_update(), does that help?

[rffontenelle]

Sadly no. I don't exactly how po_update works, but it didn't update R-/pt_BR.po files. I even replaced POT-Creation-Date to some previous date, but it didn't touch them.

[MichaelChirico]

Thanks. Feel free to just leave the .po update in the format your editor has in this PR, and I'll be sure to get it into the nicer format to reduce the diff before squash+merge.

[rffontenelle]

Done, PO file updated. Translation by @leofontenelle, proofread by me.

Chirico, I'd love to know how to adjust the line wrap, if you can check it later and share will me. Thanks in advance!

[MichaelChirico]

Hi @rffontenelle, I ran po_update() again to incorporate the recent changes to .pot (#6793 -- 3e900c9). For some reason your translations weren't copied over well so I tried to manually restore them.

I also made one correction: there are two messages for "too-many" vs. "too-few", I saw "muito poucos" for both in Portuguese, looked wrong. I think the correction is just "muitos" for the "too-many" case.

For the wrapping, AFAICT it's properly wrapped (I looked at the profile of nchar() and it looks similar to the other languages).

PTAL.

[rffontenelle]

I also made one correction: there are two messages for "too-many" vs. "too-few", I saw "muito poucos" for both in Portuguese, looked wrong. I think the correction is just "muitos" for the "too-many" case.

Indeed that's not correct.

I'll take a look a them.

Just for the record:

$ LC_ALL=C msgfmt pt_BR.po 
pt_BR.po:3365: number of format specifications in 'msgid' and 'msgstr' does not match
pt_BR.po:3412: number of format specifications in 'msgid' and 'msgstr' does not match
msgfmt: found 2 fatal errors
647 translated messages, 6 fuzzy translations, 14 untranslated messages.
$ LC_ALL=C msgfmt R-pt_BR.po 
468 translated messages, 1 fuzzy translation, 100 untranslated messages.

[rffontenelle]

@leofontenelle pt_BR.po translated, feel free to review. Now I'm working on R-pt_BR.po

[rffontenelle]

For the wrapping, AFAICT it's properly wrapped (I looked at the profile of nchar() and it looks similar to the other languages).

@MichaelChirico Looks like the line-wrapping problem is not the number of chars, but where to wrap. I'm using Gtranslator and it seems to not break as.name in different lines when line-wrapping: while Gtranslator put them together, potools leaves 'as.' in one line and name in the other.

The resulting string is the same, but the commit diff gets a bit polluted with these different line-wrapping rules.

For the last commit, I manually undid the wrapping-only changes done by my editor.

[rffontenelle]

R-pt_BR translated. 2 strings fuzzy because I was not sure how to translate it. Pending review

[leofontenelle]

Pedi dois ou três ajustes nos comentários

[rffontenelle]

@MichaelChirico ready to merge!

[MichaelChirico]

I'm using Gtranslator and it seems to not break as.name in different lines when line-wrapping: while Gtranslator put them together, potools leaves 'as.' in one line and name in the other.

Ah, interesting. Are you aware of a standard on this? Possible I have some wrong logic in {potools}... IIRC I based my own rule on what would leave the r-devel .po files with minimal diff, so that might be something else to check (i.e. does Gtranslator mess extensively with the .po files of r-devel).

[MichaelChirico]

Thanks! I deleted the deprecated (#~) messages and compiled po->mo.

I noticed there are two remaining fuzzy translations reported:

msgfmt -c --statistics -o inst/po/pt_BR/LC_MESSAGES/data.table.mo po/pt_BR.po
# 667 translated messages.
msgfmt -c --statistics -o inst/po/pt_BR/LC_MESSAGES/R-data.table.mo po/R-pt_BR.po
# 567 translated messages, 2 fuzzy translations.

It looks like it's these:

   message_source     type fuzzy                                msgid
           <char>   <char> <int>                               <char>
1:              R singular     1     Optimized subsetting with key %s
2:              R singular     1 Optimized subsetting with index '%s'
                                  msgstr msgid_plural msgstr_plural
                                  <char>       <list>        <list>
1:    Subconjunto otimizado com chave %s       [NULL]        [NULL]
2: Subconjunto otimizado com índice '%s'       [NULL]        [NULL]

Since it's only two I'll go ahead and merge here since I want to be sure this is included in 1.17.0. Feel free to file a quick follow-up & it still might make the release.