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 acceptit_IT
as well asit-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 instancenl_BE
(Flemish
) will be omitted because it's the same asnl
(Dutch
). Defaults tono
if--languages
is specified, toyes
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 dataexport-plural-rules json
-
prettyjson
: uncompressed JSON dataexport-plural-rules prettyjson
-
html
: html table (see the result)export-plural-rules html
-
php
: build a php file that can be includedexport-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 languageexport-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 instanceen_US
)name
: the language name (for instanceAmerican English
foren_US
)supersededBy
: the code of a language that supersedes this language code (for instance,jw
is superseded byjv
to represent the Javanese language)script
: the script name (for instance, forzh_Hans
-Simplified Chinese
- the script isSimplified Han
)territory
: the name of the territory (for instanceUnited States
foren_US
)baseLanguage
: the name of the base language (for instanceEnglish
foren_US
)formula
: the gettext formula to distinguish between different plural rules. For instancen != 1
categories
: the plural cases applicable for this language. It's an array ofGettext\Languages\Category
instances. Each instance has these properties:id
: can be (in this order) one ofzero
,one
,two
,few
,many
orother
. Theother
case is always present.examples
: a representation of some values for which this plural case is valid (examples are simple numbers like1
or complex ranges like0, 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
CLDR variable | gettext equivalent |
---|---|
n |
n |
i |
n |
v |
0 |
w |
0 |
f |
empty |
t |
empty |
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)