• easy to use
• strong support of plural forms
• small (1.8 kB minified and gzipped)

1. Insert the script in your web page. The script adds an object in the global scope: gt.
2. Add a link element per translations you have, with rel and lang attributes :

<link rel='gettext' lang='en' href='translations/en.json' />

3. Set the language in a lang attribute in the html tag. Note that even without this library, this is recommended for accessibility to switch the language in screen readers.
4. Call gt.init. This will load the appropriate language file depending on the configured language. init can take some options, please see below for the available options.
5. Call gt.gettext or gt._ to translate a string.

Call this method to initialize the library. This will find the language to load using the lang attribute on the html node, and the language file to load using the link with the appropriate lang attribute as well. Then it will load and parse the language file.

The method takes an option parameter with the following possible options:

• sync (false by default). If true, the loading is synchronous: the call blocks until the file is downloaded, parsed, and the library is ready. Anyway, if this is true or false, the various listeners will be called when the library is ready.
• callback (empty by default). This function will be called when the loading finishes. See also addInitListener below.
• createShortcuts (false by default). If true, the library will create the top-level functions _, _n, _o, strfmt. See below for their documentation.

This registers a function to be called as soon as the language file is loaded. The function will be passed the gettext object.

If the language file is already loaded, the function will be called immediately (and synchronously).

This returns the translated string, or the passed string itself if there is no translation.

This methods makes it easier to translate plural forms. The three parameters are :

• string is the singular form to translate; this is the key that will be used to find the translation in the translation file.
• plural_string is the plural forms that will be returned if string is not found in the translation file and if number is greater than 1.
• number is the parameter that defines which plural form will be returned.

Please see below for more informations about plural forms.

This is a nice addition to the original gettext API. It lets you translate a whole object in one call, iterating through the object and translating each of its members.

For each pair of (key, value) in the object, it replaces the value with the result of gettext(value). It modifies the object that is passed and returns it as well.

This is only useful for non-contextual texts, without plural forms. You can use this in the callback function called by addInitListener.

var translations = {
  text1: "This is my text.",
  text2: "This is another text."
};

gt.addInitListener(onTranslationInit);

function onTranslationInit(gt) {
  gt._o(translations);
  // now, translations contains the translations of the texts.
}

This returns the locale used for translations in this session. Basically, this will be the value stored in the lang attribute of the html element.

This takes as argument a string pattern and some arguments.

The pattern contains some ordered parameters like {0} {1} that gets replaced by the arguments. As you may guess, the number between the curly braces determines which argument to pick. This allows you to put very generic strings inside your language files, and even reorder the replacement strings!

strfmt("{0} sheeps are in {1}", 5, "the sheepfold");

This is very useful for plural forms, for example we can also use it like that :

strfmt("One sheep is in {1}", 5, "la bergerie");

Please note that the first parameter is silently discarded, so that you can use the very same command and only change the pattern, which is returned by _n.

The language files are normal JSON files. As the files are parsed using JavaScript's eval you don't have to adhere to the stricter JSON rules: you can left your keys unquoted (although it's easier to always quote them), you can use string concatenation to make your file more readable, and you can use comments.

Here are the basic rules to follow :

• Each key to translate is a key in the JavaScript object.
• You must have an empty key. This empty key has several lines which are only informational except the line defining the plural forms. YOu can either use a concatened string with new lined or an object (see below).
• Plural forms are specified using normal JavaScript arrays.

{
    // all of this is not really needed but it's good
    // to provide informations to your translators
    "": "Project-Id-Version: gettext-example\n" +
        "Report-Msgid-Bugs-To: \n" +
        "PO-Revision-Date: 2011-11-29\n" +
        "Last-Translator: Julien Wajsberg <[email protected]>\n" +
        "Language-Team: France Telecom\n" +
        "MIME-Version: 1.0\n" +
        "Content-Type: text/plain; charset=UTF-8\n" +
        "Content-Transfer-Encoding: 8bit\n" +
        "Plural-Forms: nplurals=2; plural=n != 1;",

    // the translations
    "$$text1$$": "This is a really good text.",
    "$$text2$$": "This is an awesome text.",
    "$$text3$$": "This text is very bad, very naughty.",
    "$$sheeps-in-sheepfold$$": [
		"One sheep is in the sheepfold.",
		"{0} sheeps are in the sheepfold."
	]
}

The header can be specified using a normal JavaScript object as well :

{
    // all of this is not really needed but it's good
    // to provide informations to your translators
    "": {
        "Project-Id-Version": "gettext-example",
        "PO-Revision-Date": "2011-11-29",
        "Last-Translator": "Julien Wajsberg <[email protected]>",
        "Language-Team": "France Telecom",
        "MIME-Version": "1.0",
        "Content-Type": "text/plain; charset=UTF-8",
        "Content-Transfer-Encoding": "8bit",
        "Plural-Forms": "nplurals=2; plural=n > 1;"
    },
    ...
}

As a best practice, I use keys that are prefixed and suffixed with $$ so that I can then easily see what's not translated.

gettext.js has a strong support for plural forms. As you may know, different languages have different rules for plural forms. Some languages have more than 1 form of plural. Most existing libraries only support the english plural form, which is obviously not sufficient. For example, do you know that Polish has 3 forms ?

Here is how to use this support :

• first, you must have a correct Plural-Forms line in your language file. The gettext manual page about plural forms gathers lots of Plural-Forms examples you can just copy and paste. Another reference can be found on the Translate Toolkit & Pootle Website.
• then, when you want a key with plural forms, you must use a normal JavaScript array as a value, with as many elements as there are forms for this language (as in the $$sheeps-in-sheepfold$$ key above).
• and at last, use the ngettext method as described before.

There are very good alternatives to this library:

• Jed has similar features than gettext.js. It uses a quite different syntax and uses a grammar to parse the Plural Forms line (whereas gettext.js uses eval which we plan to change eventually).
• MessageFormat by the same author. This is an implementation of Java's MessageFormat. This is both more complicated and more featureful than gettext.js and Jed.
• Mozilla B2G's localization library is an attempt to get a simple l10n library for modern browsers. They're experimenting with a tentative standard API and they have different goals. As far as I know, they're only targeting modern browsers (I mean, no IE support).
• jsgettext is quite old but is a quite comprehensive port of the original GNU Gettext library. gettext.js uses jsgettext parsing for Plural-Forms and that's why we use LGPL for now. We also use the same format for our data files. Basically this library inspired gettext.js a lot so thanks to Joshua I. Miller for his work.