Blacktango wrote: ↑10 January 2025, 00:24
What I mean is that, even with these libraries, you need to understand how to truncate your texts to make them translatable in any language.
For example, if you want to translate "n object(s)", you may use five different texts:
- 0 object
- 1 object
- 2 objects
- 3 objects
- x objects
In English, you need just two cases "object" and "objects", in Arab I think you need three.
What about the other languages? Maybe some objects have different gender too, then you need even more forms?
As a developer, how do you know how many different cases you need to use?
I understand what you are talking about. This is exactly what gettext's
ngettext() function is about. If a developer uses gettext, they
always use only two forms in their source code, like this:
Code: Select all
printf(ngettext("%d object", "%d objects", n), n);
Here, the inner call to
ngettext() takes three arguments: the two (always two!) plural forms for English singular and plural, and the number
n, which is used to select one of as many as necessary plural forms
according to target locale rules. Then, once
ngettext() returns the appropriate translation,
printf() (or
sprintf() or whatever other replacement method the developer uses) replaces the placeholder in the translated string with the number (hence the second
, n at the end), producing the final string.
From the localizer's side, each gettext translation file, if it supports plurals at all, has a "Plural-Forms" header, which specifies the number of plural forms and the logic of selecting the appropriate one. Here's an example for Lithuanian:
Code: Select all
"Plural-Forms: nplurals=3; plural=(n%10==1 && n%100!=11 ? 0 : n%10>=2 && (n%100<10 || n%100>=20) ? 1 : 2);\n"
With that header in place, my translation for the above line would look like this:
Code: Select all
msgid "%d object"
msgid_plural "%d objects"
msgstr[0] "%d objektas" # 1 objektas
msgstr[1] "%d objektai" # 2 objektai
msgstr[2] "%d objektų" # 10 objektų
The developer doesn't need to know or implement any language-dependent logic in their code (it would be a nightmare if everyone had to do that on their own). They only need to make use of the call itself.
Most popular gettext authoring tools (like
Poedit) already have rules for most of the languages accounted for, so, once the localizer selects the language they're working on, the
Plural-Forms header will be prefilled correctly. The localizer only need to concentrate on actual proper localization (and of course remember the correct order of plural forms) from there on. Some tools go even further and provide examples of translated string with the numbers in place on the fly.
If this is all new to you, I really suggest you to click on the two links in my previous comment.
