Skip to main content

Pluralization

MoonBuggy handles plural forms using CLDR rules for all Unicode locales. You write plural blocks in a concise pipe-separated syntax, and the source generator emits inline arithmetic for each language's plural categories.

Basic plurals (one | other)​

Wrap plural forms in $...|...$ with #var# as the selector:

_t("You have $#x# book|#x# books$", new { x })

The pipe separates the "one" and "other" forms. #x# both selects the form and renders the count in the output. The extractor produces this ICU MessageFormat in the PO file:

msgid "You have {x, plural, one {# book} other {# books}}"
msgstr ""

At runtime:

  • x = 1 renders "You have 1 book"
  • x = 3 renders "You have 3 books"

Integer types only​

The selector variable must be an integer type: int, long, byte, short, and their unsigned variants. Floating-point types (float, double, decimal) produce compiler error MB0009. This matches ICU's integer-based plural selection.

Three forms with zero (=0 | one | other)​

Use #var=0# to add an explicit zero case:

_t("$#x=0#No items|#x# item|#x# items$", new { x })

PO output:

msgid "{x, plural, =0 {No items} one {# item} other {# items}}"

At runtime:

  • x = 0 renders "No items"
  • x = 1 renders "1 item"
  • x = 5 renders "5 items"

The =0 form is an exact-value match — it takes priority over any CLDR category.

Hidden selectors​

Sometimes you want the plural form but not the number itself. Use #~var# to select without rendering:

_t("$#~x#one new message|#x# new messages$", new { x })

PO output:

msgid "{x, plural, one {one new message} other {# new messages}}"

At runtime:

  • x = 1 renders "one new message" (no number shown)
  • x = 5 renders "5 new messages"

Combine hidden selector with zero form using #~var=0#:

_t("$#~x=0#no messages|one new message|#x# new messages$", new { x })

CLDR plural categories​

English has two categories: one and other. But many languages have more. Russian has one, few, many, and other. Arabic has zero, one, two, few, many, and other.

You only write the English forms in source code — the translator provides the correct number of forms for their language. For example, a translator working with Russian would write:

msgid "You have {x, plural, one {# book} other {# books}}"
msgstr "У вас {x, plural, one {# книга} few {# книги} many {# книг} other {# книг}}"

MoonBuggy's source generator reads the CLDR plural rules for each target locale and emits the correct branching logic. The rules are baked in as integer arithmetic — no ICU runtime library is needed.

Multiple plural blocks​

A single message can contain multiple plural blocks:

_t("$#x# file|#x# files$ and $#y# folder|#y# folders$", new { x, y })

Each block has its own selector variable.

Variables inside plural branches​

Regular $var$ substitutions work inside plural branches:

_t("$#x# book by $author$|#x# books by $author$$", new { x, author })

PO output:

msgid "{x, plural, one {# book by {author}} other {# books by {author}}}"

Mixing variables and plurals​

Variables and plural blocks combine naturally:

_t("Hi $name$, you have $#x# book|#x# books$", new { name, x })

PO output:

msgid "Hi {name}, you have {x, plural, one {# book} other {# books}}"

What translators see​

Translators work exclusively with ICU MessageFormat. They never encounter MoonBuggy's $...|...$ syntax. This means:

  • Standard PO editors (Poedit, Crowdin, Weblate) understand the format natively
  • Translators can add plural categories required by their language
  • The # symbol in ICU represents the count value — translators can place it anywhere within a plural branch