To use the languages data generated from this tool you can use the bin/export-plural-rules command.

export-plural-rules supports the following options:

• --us-ascii If specified, the output will contain only US-ASCII characters. If not specified, the output charset is UTF-8.
• --languages=<LanguageId>[,<LanguageId>,...]] --language=<LanguageId>[,<LanguageId>,...]] Export only the specified language codes. Separate languages with commas; you can also use this argument more than once; it's case insensitive and accepts both '_' and '-' as locale chunks separator (eg we accept it_IT as well as it-it). If this option is not specified, the result will contain all the available languages.
• --reduce=yes|no If set to yes the output won't contain languages with the same base language and rules. For instance nl_BE (Flemish) will be omitted because it's the same as nl (Dutch). Defaults to no if --languages is specified, to yes otherwise.
• --output=<file name> If specified, the output will be saved to . If not specified we'll output to standard output.

export-plural-rules can generate data in the following formats:

• json: compressed JSON data

  export-plural-rules json

• prettyjson: uncompressed JSON data

  export-plural-rules prettyjson

• html: html table (see the result)

  export-plural-rules html

• php: build a php file that can be included

  export-plural-rules php > yourfile.php

  Then you can use that generated file in your php scripts:

  $languages = include 'yourfile.php';

• xml: generate an XML document (here you can find the xsd XML schema)

  export-plural-rules xml

• po: generate the gettext .po headers for a single language

  export-plural-rules po --language=YourLanguageCode

You can use Composer to include this tool in your project. Simply launch composer require gettext/languages or add "gettext/languages": "*" to the "require" section of your composer.json file.

If you don't use composer in your project, you can download this package in a directory of your project and include the autoloader file:

require_once 'path/to/src/autoloader.php';

The most useful functions of this tools are the following

$allLanguages = Gettext\Languages\Language::getAll();
...
$oneLanguage = Gettext\Languages\Language::getById('en_US');
...

getAll returns a list of Gettext\Languages\Language instances, getById returns a single Gettext\Languages\Language instance (or null if the specified language identifier is not valid).

The main properties of the Gettext\Languages\Language instances are:

• id: the normalized language ID (for instance en_US)
• name: the language name (for instance American English for en_US)
• supersededBy: the code of a language that supersedes this language code (for instance, jw is superseded by jv to represent the Javanese language)
• script: the script name (for instance, for zh_Hans - Simplified Chinese - the script is Simplified Han)
• territory: the name of the territory (for instance United States for en_US)
• baseLanguage: the name of the base language (for instance English for en_US)
• formula: the gettext formula to distinguish between different plural rules. For instance n != 1
• categories: the plural cases applicable for this language. It's an array of Gettext\Languages\Category instances. Each instance has these properties:
  • id: can be (in this order) one of zero, one, two, few, many or other. The other case is always present.
  • examples: a representation of some values for which this plural case is valid (examples are simple numbers like 1 or complex ranges like 0, 2~16, 100, 1000, 10000, 100000, 1000000, …)

Yes - as far as you trust the Unicode CLDR project.

The conversion from CLDR to gettext includes also a lot of tests to check the results. And all passes 😉.

The CLDR specifications define the following variables to be used in the CLDR plural formulas:

• n: absolute value of the source number (integer and decimals) (eg: 9.870 => 9.87)
• i: integer digits of n (eg: 9.870 => 9)
• v: number of visible fraction digits in n, with trailing zeros (eg: 9.870 => 3)
• w: number of visible fraction digits in n, without trailing zeros (eg: 9.870 => 2)
• f: visible fractional digits in n, with trailing zeros (eg: 9.870 => 870)
• t: visible fractional digits in n, without trailing zeros (eg: 9.870 => 87)

The gettext specifications define the following variables to be used in the gettext plural formulas:

• n: unsigned long int

The generated gettext formulas contain some extra parenthesis, in order to avoid problems in some programming language. For instance, let's assume we have this formula: (0 == 0) ? 0 : (0 == 1) ? 1 : 2

• in C it evaluates to 0 since is the same as (0 == 0) ? 0 : ((0 == 1) ? 1 : 2)
• in Java it evaluates to 0 since is the same as (0 == 0) ? 0 : ((0 == 1) ? 1 : 2)
• in JavaScript it evaluates to 0 since is the same as (0 == 0) ? 0 : ((0 == 1) ? 1 : 2)
• in PHP it evaluates to 2 since is the same as ((0 == 0) ? 0 : (0 == 1)) ? 1 : 2

So, in order to avoid problems, instead of a simple a ? 0 : b ? 1 : 2 the resulting formulas will be in this format: 'a ? 0 : (b ? 1 : 2)