Add a tool to check printf style format string in translations

[BrettDong]

Summary

Infrastructure "Add a tool to check printf style format string in translations"

Purpose of change

Some printf style format string mismatches exist in translations, for example, %d something %s gets translated to %s XXX %d, and currently we lack an automated tool to check for such kind of errors.

Describe the solution

Write a Python script that uses polib library to read PO files and check for inconsistencies of printf format strings between original string and translated string.

Alternative

My first attempt: Run printf test frompofilter tool from translate-toolkit software package to check for printf format string errors, and filter its output by a Python script to get rid of false positives.

Update: There are too many false positives, so I just went to write my own printf checker.

Testing

Errors are detected, here is a portion of the output:

Checking ja.po => 1 error(s) detected:
original  :<npcname> doesn't have limbs that require splinting.
translated:%1$sに固定が必要な四肢はありません。


Checking ko.po => 2 error(s) detected:
original  :You avoid an incoming projectile!
translated:%s을(를) 피했다!

original  :<npcname> avoids an incoming projectile.
translated:<npcname>이(가) %s을(를) 피했다.


Checking pl.po => 3 error(s) detected:
original  :%1$s butts %2$s with their antlers
translated:%1$s łomocze %s swoim porożem

original  :It should take %d minutes to finish washing items in the %s.
translated:Wypranie rzeczy w %s powinno zająć %d minut.

original  :Insects begin to emerge from <npcname>'s skin!
translated:Owady rozrywają ciało %s i wychodzą na zewnątrz.


Checking pt_BR.po => 1 error(s) detected:
original  :The %s recovers from its skid.
translated:% se recupera de sua derrapagem.


Checking ru.po => 3 error(s) detected:
original  :You tear into %s with your serrated teeth
translated:Вы разрываете противника (%2$s) своими зазубренными зубами

original  :%1$s bashes %2$s with their powerful wings
translated:%1$s бьёт противника (%s) своими могучими крыльями

original  :Do you wish to sell the crop of %d %s for a profit of $%d?
translated:Вы хотите продать урожай (%s - %d), получив прибыль в размере $%d?


Checking zh_CN.po => 3 error(s) detected:
original  :You wrap your scarf tighter.
translated:你裹紧你的 %s。

original  :You wrap your scarf a bit tighter.
translated:你稍稍裹紧你的 %s。

Additional context

[jbytheway]

Since you're raised the subject of printf format strings and translations: I've been thinking about writing a clang-tidy check to require that every translated string with multiple format specifiers use the numbered form (1$, 2$, etc.). Do you think that would be worthwhile? In the short term it would invalidate a lot of translations, which is why I was thinking it would be good to do shortly after the stable release; in the long term it would save us scrutinizing PRs for that particular i18n issue.

[BrettDong]

The impact may be mitigated by building a tool that automatically migrates non positional format strings to positional format strings in translations.

[actual-nh]

This one appears to need positional parameters?

[actual-nh]

Suggested change

	msgstr "parte para inspeccionar terrenos..."
	msgstr "parte para inspeccionar terrenos..."

[kevingranade]

Unless I'm missing something, this %% was necessary in order to print a % to the final string.

[BrettDong]

I guess this is a plain unformatted string, because in the original English string there is only a single %

[kevingranade]

This %% was also necessary.

However, now that I'm looking at it, why does "Intermediate Akido" translate to this long phrase?

[BrettDong]

Probably a translation error. lang/po/es_ES.po in master branch reads

#: lang/json/martial_art_from_json.py
msgid "Intermediate Aikido"
msgstr "Aikido intermedio"

[BrettDong]

So I'd better revert the *.po changes and only add the checking script in this pull request.