Plural forms support in Locale Kit

[KapiX]

As in topic, it would be nice if we had support for plurals in Locale Kit.

Example: In English there are only two forms: plural and plurals. In Polish there are three: liczba, liczby, liczb.

[pulkomandy]

A more complete example of polish wording to use as a test would be nice. How would a sentence such as: "1 object selected" "2 objects selected" translate?

[Wizard]

Wybrano 1 obiekt. Wybrano 2, 3, 4 obiekty. Wybrano 5, 6, 7.. 21 obiektów. Wybrano 22, 23, 24 obiekty. Wybrano 25, 26.. 31 obiektów. Wybrano 32, 33, 34 obiekty. Wybrano 35 obiektów. And so forth. 111-120 looks like 11-20, so Wybrano 113 obiektów.

[Wizard]

Simmilar rules apply in other slavic languages.

[KapiX]

​http://localization-guide.readthedocs.org/en/latest/l10n/pluralforms.html ​http://www.unicode.org/cldr/charts/latest/supplemental/language_plural_rules.html

[pulkomandy]

ICU manages this:

• ​http://www.icu-project.org/apiref/icu4c/classicu_1_1MessageFormat.html
• ​http://www.icu-project.org/apiref/icu4c/classicu_1_1PluralFormat.html

They use a pattern system which allows to "fold" all the rules into a string:

      UnicodeString msgPatEn = "{0,plural, one{# dog} other{# dogs}}";
      UnicodeString msgPatSl = "{0,plural, one{# pes} two{# psa} few{# psi} other{# psov}}";

the # is replaced with the value. The rules are defined by KapiX second link for all languages. Polish uses "one", "few", "many", and "other".

Does that format sound right for translators?

[beos_zealot]

Similar rules applies to Lithuanian language:

Pažymėtas 1 objektas Pažymėti 2-9 objektai Pažymėta 10-20 objektų Pažymėtas 21 objektas Pažymėti 22-29 objektai Pažymėta 30 objektų Pažymėtas 31 objektas Pažymėti 32-39 objektai Pažymėta 40 objektų ...

nplurals=3 *

plural=0 means ending '-as' plural=1 means ending '-ai' plural=2 means ending '-ų'

(*) ​http://localization-guide.readthedocs.org/en/latest/l10n/pluralforms.html lt / Lithuanian / nplurals=3; plural=(n%10==1 && n%100!=11 ? 0 : n%10>=2 && (n%100<10 or n%100>=20) ? 1 : 2);

UPDATE:

Lithuanian uses categories "one", "few", "many", and "other". (*2)

From previous example these rules covers additional case: many - for floating-point values

example: Pažymėta 0.1~0.9, 1.1~1.7, 10.1, 100.1, 1000.1, … objekto

one (-as, same as plural=0 in previous example) few (-ai, same as plural=1 in previous example) many (-o, for floating-point values, rule f != 0) other (-ų, same as plural=2 in previous example)

(*2) ​http://www.unicode.org/cldr/charts/latest/supplemental/language_plural_rules.html Name: Lithuanian / Code: lt

[KapiX]

Does that format sound right for translators?

Yes, it is OK, but how it would be used in translations?

Wybrano %n {0,plural, one{# obiekt} few{# obiekty} many{# obiektów} other{# obiektu}}.

Like that?

[pulkomandy]

You don't need the %n, the "0" is the identifier for each number in the message. So "10 files in 3 folders" would be (for english):

 "{fileCount, plural, one{# file} other{# files}} in {folderCount, plural, one{# folder} other {# folders}}"

Then this can be replaced with the rules for each translation. Of course it seems a good idea to have a "comment" with the expected rendering of the string in english, to help people understand what to do with this.

[bonefish]

FWIW, multiple plural support is not the only thing we aren't supporting: ​http://site.icu-project.org/design/formatting/select

[pulkomandy]

A few places which need to be fixed:

• ​http://cgit.haiku-os.org/haiku/tree/src/apps/aboutsystem/AboutSystem.cpp#n444 AboutSystem processor count
• ​http://cgit.haiku-os.org/haiku/tree/src/kits/tracker/CountView.cpp#n232 Tracker item count
• ​http://i18n.haiku-os.org/pootle/en/haiku/translate.html#search=%25num&sfields=source,target various cases where the generic "%num" marker is used in our locale strings.

[pulkomandy]

Implemented in hrev47971. All places in the code where this is needed have to be updatd to make use of the new system. Let me know if you find more of those (I fixed the ones mentionned above already).

All translations have to be updated using the syntax mentionned above.